Post on 07-Apr-2018
transcript
Environmental Microbiology (2005)
7
(12) 1883ndash1895 doi101111j1462-2920200500762x
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd
Blackwell Science LtdOxford UKEMIEnvironmental Microbiology 1462-2912Blackwell Publishing Ltd 20057
1218831895
Original Article
Metagenomic analysis of freshwater bacteriaM T Cottrell L A Waidner L Yu and D L Kirchman
Received 9 March 2004 revised 26 July 2004 accepted 27 July2004 For correspondence E-mail Kirchmancmsudeledu Tel(
+
1) 302 6454375 Fax (
+
1) 302 6454028
Bacterial diversity of metagenomic and PCR libraries from the Delaware River
Matthew T Cottrell Lisa A Waidner Liying Yu and David L Kirchman
University of Delaware College of Marine Studies 700 Pilottown Road Lewes DE 19958 USA
Summary
To determine whether metagenomic libraries sampleadequately the dominant bacteria in aquatic environ-ments we examined the phylogenetic make-up of alarge insert metagenomic library constructed withbacterial DNA from the Delaware River a polymerasechain reaction (PCR) library of 16S rRNA genes andcommunity structure determined by fluorescence
insitu
hybridization (FISH) The composition of thelibraries and community structure determined byFISH differed for the major bacterial groups in theriver which included
Actinobacteria
beta-
proteobac-teria
and
Cytophaga
-like bacteria Beta-
proteobacte-ria
were underrepresented in the metagenomic librarycompared with the PCR library and FISH while
Cytophaga
-like bacteria were more abundant in themetagenomic library than in the PCR library and inthe actual community according to FISH The Dela-ware River libraries contained bacteria belonging toseveral widespread freshwater clusters includingclusters of
Polynucleobacter necessarius
Rhodof-erax
sp Bal47 and LD28 beta-
proteobacteria
theACK-m1 and STA2-30 clusters of
Actinobacteria
andthe PRD01a001B
Cytophaga
-like bacteria clusterCoverage of bacteria with
gt
97 sequence identitywas 65 and 50 for the metagenomic and PCRlibraries respectively Rarefaction analysis of repli-cate PCR libraries and of a library constructed withre-conditioned amplicons indicated that heteroduplexformation did not substantially impact the composi-tion of the PCR library This study suggests thatalthough it may miss some bacterial groups themetagenomic approach can sample other groups(eg
Cytophaga
-like bacteria) that are potentiallyunderrepresented by other culture-independentapproaches
Introduction
Metagenomic clone libraries of environmental DNAenable the exploration of the phylogenetic and metabolicdiversity of microbes in the environment without cultivationor polymerase chain reaction (PCR) New aspects ofmicrobial metabolism in uncultured microbes have beenuncovered using this metagenomic approach as firstshown in a study of marine
Bacteria
and
Archaea
(Stein
et al
1996) and then later in soils (Rondon
et al
2000)Intriguing examples include studies revealing differenttypes of phototrophy in the ocean (Beja
et al
2000a2001 Sabehi
et al
2003) Other studies have examinedheterotrophic metabolism including one study that identi-fied chitinase genes of uncultured marine microbes (Cot-trell
et al
1999) In soils different types of metabolismhave been targeted such as those involving alcohol oxi-doreductases (Knietsch
et al
2003) lipases amylasesand nucleases (Rondon
et al
2000) Soil clones havealso been obtained that produce broad-spectrum antibiot-ics (Gillespie
et al
2002) while other clones havehemolytic properties (Rondon
et al
2000)The data on the metabolic diversity revealed by metage-
nomic libraries are most valuable when viewed in thecontext of community structure Phylogenetic informationis needed together with estimates of metabolic potentialin order to link specific members of the community tobiogeochemical processes However the phylogeneticinformation present in metagenomic libraries has receivedlittle attention (Beja
et al
2000b Liles
et al
2003) Oneapproach to obtaining this phylogenetic information is toscreen metagenomic libraries for 16S rRNA genes inorder to identify clones that then can be used to explorethe metabolic potential of targeted bacterial groups (Beja
et al
2001 Quaiser
et al
2002 Liles
et al
2003) It isalso important to compare the phylogenetic make-up ofmetagenomic libraries to other measures of communitystructure such as 16S rDNA clone libraries and fluores-cence
in situ
hybridization (FISH) in order to determine ifany of the dominant groups of bacteria are missingFinally the phylogenetic information in metagenomiclibraries provides another view of community structurewithout the biases of PCR-dependent approaches (vonWintzingerode
et al
1997)Only one study examined the similarity between the
phylogenetic composition of a metagenomic library andpublished surveys of marine bacterial diversity (Beja
et al
1884
M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd
Environmental Microbiology
7
1883ndash1895
2000b) Phylogenetic analysis of metagenomic clonesbearing 16S rRNA genes from surface water of the PacificOcean identified several groups of
Archaea
and
Bacteria
that are common in marine bacterial communities includ-ing a
Euryarchaeota
clone and clones related to SAR86SAR116 and SAR11 bacteria
Roseobacter
spp and
Cytophaga
-like bacteria (Giovannoni and Rappeacute 2000)However no study has directly compared the phylogeneticcomposition of a metagenomic library and the microbialcommunity
in situ
It may be especially important to exam-ined this issue in freshwaters where Gram-positive bacte-ria can be abundant (Gloumlckner
et al
2000 Zwart
et al
2002) These bacteria could be underrepresented inmetagenomic libraries if DNA extraction from Gram-posi-tive bacteria is less efficient
Bacteria in rivers and other freshwater ecosystemshave not been examined previously using metagenomiclibraries Bacteria in rivers play an important role in theconsumption of organic matter transported from the ter-restrial environment to the ocean (Amon and Benner1996 Opsahl
et al
1999 Hernes and Benner 2002)and the make up of those communities likely has animpact on organic matter consumption (Covert andMoran 2001) The diversity of bacteria in rivers may beespecially high as they may contain mixtures of aquaticand terrestrial bacteria including phylogenetically andmetabolically diverse soil taxa (Liles
et al
2003 Topp2003) Microbes from soil may be introduced into water-ways by runoff (Zaitlin
et al
2003) and mix with freshwa-ter bacteria (Zwart
et al
2002) Less is known aboutriverine bacteria than their counterparts in other aquaticsystems A GenBank search in November 2003 for river-ine bacterial and archaeal 16S rRNA genes returnedapproximately 1000 hits from studies in 14 rivers com-pared with more than 17 000 sequences from marineenvironments and lakes
In this study we determined the coverage and compo-sition of metagenomic and PCR libraries from the Dela-ware River We also contrasted the composition of theselibraries with community structure determined by FISH Inorder to identify why the composition of libraries mightdiffer we tested the hypothesis that heteroduplex artefactsof the PCR step lead to overestimates of bacterialdiversity (Thompson
et al
2002) We found that themetagenomic analysis and PCR libraries identified
Actinobacteria
Cytophaga
-like bacteria and beta-proteobacteria which are potentially important membersof the bacterial community in the Delaware River
Results
Overview of the libraries
The metagenomic library was comprised of 4608 cloneswith an average insert size of 405
plusmn
31 kb (
n
=
15) Thelibrary contained approximately 180 Mb of DNA which isequivalent to 90 bacterial genomes assuming two Mbpper bacterial genome (Button and Robertson 2001)Eighty clones tested positive for 16S rRNA genes whichis consistent with the expected number of rRNA genes inthe library assuming a single 16S rRNA gene per bacte-rial genome Bacteria isolated from aquatic and low-nutri-ent environments typically have only one or two rRNAoperons (Button
et al
1998 Fegatella
et al
1998 Fogel
et al
1999 Klappenbach
et al
2000)The compositions of the metagenomic library and PCR
libraries were not significantly different at a high phyloge-netic level based on analyses of coverage calculatedusing the LIBSHUFF tool
BLAST
analysis was consistentwith this result revealing essentially the same majorgroups of bacteria in the two libraries (Table 1)
Cytoph-aga
-like bacteria beta-
proteobacteria
and
Actinobacteria
Table 1
Phylogenetic composition of the metagenomic and PCR libraries and structure of the actual community determined by FISH
Group of metagenomicclones of PCR clones
of total prokaryotesdetected by FISH
Alpha-
proteobacteria
3 12 6
plusmn
2Beta-
proteobacteria
17 50 25
plusmn
5Gamma-
proteobacteria
4 0 3
plusmn
2Epsilon-
proteobacteria
0 2 ND
Cytophaga-
like 54 13 17
plusmn
4
Actinobacteria
14 16 26
plusmn
5
Firmicutes
3 0 ND
Bdellovibrio
0 4 ND
Verrucomicrobiales
3 2 ND
Spirochaetaceae
1 0 NDndash ndash ndash
Total 100 100 77
Composition of the metagenomic library is based on nucleotide sequences from 72 clones bearing 16S rRNA genes Fifty-eight clones from thePCR library were sequenced and 500ndash1000 DAPI-positive prokaryotes were analysed with each FISH probeND not determined
Metagenomic analysis of freshwater bacteria
1885
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd
Environmental Microbiology
7
1883ndash1895
comprised the largest fractions of the libraries while alpha-
proteobacteria
gamma-
proteobacteria
Firmicutes
Verru-comicrobiales
and
Spirochaetaceae
were minor mem-bers However closer examination of the relativeabundance of different bacterial groups revealed notewor-thy differences between the metagenomic and PCRlibraries
Cytophaga-
like bacteria in clone libraries
The metagenomic and PCR libraries differed in the num-ber and kinds of
Cytophaga
-like bacteria One obviousdifference was the percentage of
Cytophaga
-like clonesOnly 13 of the clones in the PCR library belonged to the
Cytophaga
-like group compared with 54 for themetagenomic library (Table 1) In addition the variety of
Cytophaga
-like bacteria was greater in the metagenomiclibrary than in the PCR library More
Cytophaga
-like clus-ters contained fosmid sequences than PCR sequences inthe phylogenetic tree of
Cytophaga
-like bacteria (Fig 1)The metagenomic library but not the PCR library con-tained
Cytophaga
-like clones in the
Chitinophaga
and
Myroides
groups (Fig 1) One
Cytophaga
-like clone in thePCR library (Sta2ndash97) was not similar to any of themetagenomic clones
Seventy percent of the
Cytophaga
-like bacteria wereassociated with the
Flavobacteriales
group (Fig 1) andthese clones could be grouped into three clusters (AndashC)A fourth cluster (D) was in the
Flectobacillus
group(Fig 1) Clusters A and B were related to different kindsof
Flavobacterium
spp and included some cultured bac-teria Cluster A included an isolate from the Elbe Riverand cluster B included
Flavobacterium xinjiangensi
Theaverage sequence identity in these two clusters with cul-tured members was 966
plusmn
34 and 945
plusmn
61respectively Clusters C and D were associated with the
Myroides
and
Flectobacillus
groups respectively andincluded bacteria with a higher degree of sequence simi-larity (985
plusmn
05 and 995
plusmn
04 respectively) butthese clusters had no cultured members
Some of the
Cytophaga
-like bacteria were representedin both libraries Half of the
Cytophaga
-like clones in thePCR library were highly similar to clones in the metage-nomic library and every
Cytophaga
-like group with clonesfrom the PCR library also had representatives from themetagenomic library For example cluster D was com-prised of three PCR clones and four metagenomic clones(Fig 2) Cluster A included three metagenomic clonesand a highly similar clone from the PCR library Howeversimilarity between
Cytophaga
-like bacteria in the twolibraries was restricted to these two clusters
Beta-
proteobacteria
in clone libraries
The metage-nomic and PCR libraries also differed in the number andkinds of beta-
proteobacteria The PCR library had more
than twice as many clones in this group than the metage-nomic library (50 versus 17 respectively) (Table 1)The beta-proteobacteria clones in the two librariesbelonged to five clusters (AndashE) Clusters D and E werecomprised exclusively of clones from the PCR library(Fig 2) The degree of relatedness in the clusters madeup of only PCR clones and in cluster A was low rangingfrom 89 to 95 sequence identity In contrast some ofthe beta-proteobacteria sampled by the two libraries weresimilar Clusters B and C which included clones from bothlibraries were comprised of very closely related bacteriawith similarities of 964 plusmn 16 and 975 plusmn 31respectively
Other bacteria in the clone libraries
In contrast to the differences outlined above for theCytophaga-like and beta-proteobacteria groups Actino-bacteria in the metagenomic and PCR libraries were quitesimilar The libraries had equivalent percentages of Acti-nobacteria (about 15) and there was considerable over-lap in the types of Actinobacteria in the two libraries Ofthe three clusters of Actinobacteria clones (AndashC) eachincluded clones from both libraries (Fig 3) Cluster Bincluded the most closely related clones with averagesimilarities of 964 plusmn 22 The Actinobacteria clones inclusters A and C were less closely related Sequencesimilarities in these clusters were 932 plusmn 41 and929 plusmn 54 respectively
Alpha-proteobacteria comprised a minor component ofthe libraries accounting for only 3 and 12 of theclones in the metagenomic and PCR clones respectively(Table 1) The alpha-proteobacteria clones in the twolibraries were closely related to each other and to somecultured alpha-proteobacteria (results not shown) Threeclones from the PCR library and two from the metage-nomic library belonged to a cluster that included culturedSphingomonas spp One alpha-proteobacteria clone fromthe PCR library belonged to the Rhodospirillales group
The libraries also included clones belonging to fivegroups of bacteria that can be present in aquatic sys-tems but are typically minor members of the communityThese groups included Firmicutes and Spirochaetaceaethat occurred in only the metagenomic library and epsi-lon-proteobacteria and Bdellovibrio that were foundexclusively in the PCR library (Table 1) Clones belong-ing to the Verrucomicrobiales group occurred in bothlibraries
Library coverage and diversity
Library coverage increased with the genetic distance asexpected At an evolutionary distance of 003 coveragesof the metagenomic and PCR libraries were 65 and 51
1886 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Fig 1 Phylogenetic relationships of selected Cytophaga-like bacteria as depicted in the ARB database tree constructed of sequences longer than 1000 bp (June 2002) Delaware River Cytophaga-like bacteria were added using the ARB parsimony quick add tool Clones in the library of 16S rDNA amplicons are labelled PCR and 16S rDNA-bearing metagenomic DNA clones are labelled Fosmid Values on the trapezoids indicate the number of sequences in the group Labels A B C and D refer to clusters of closely related bacteria that are discussed in the text The scale bar represents 01 substitutions per site
Metagenomic analysis of freshwater bacteria 1887
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Fig 2 Phylogenetic relationships of selected beta-proteobacteria as depicted in the ARB database tree constructed of sequences longer than 1000 bp (June 2002) Delaware River beta-proteobacteria were added using the ARB parsimony quick add tool Clones in the library of 16S rDNA amplicons are labelled PCR and 16S rDNA-bearing metagenomic DNA clones are labelled Fosmid Values on the trapezoids indicate the number of sequences in the group Labels A B C D and E refer to clusters of closely related bacteria that are discussed in the text The scale bar represents 01 substitutions per site
1888 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Fig 3 Phylogenetic relationships of selected Actinobacteria as depicted in the ARB database tree constructed of sequences longer than 1000 bp (June 2002) Delaware River Actinobacteria were added using the ARB parsimony quick add tool Clones in the library of 16S rDNA amplicons are labelled PCR and 16S rDNA-bearing metagenomic DNA clones are labelled Fosmid Values on the trapezoids indicate the number of sequences in the group Labels A B and C refer to clusters of closely related bacteria that are discussed in the text The scale bar represents 01 substitutions per site
Metagenomic analysis of freshwater bacteria 1889
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
respectively (Fig 4) This distance is approximately equalto 97 sequence identity which is one approximation ofspecies-level identity (Stackebrandt and Goebel 1994)Coverage of the metagenomic library was higher thancoverage of the PCR library at all levels of genetic dis-tance with the greatest differences occurring at smallgenetic distance Coverage decreased rapidly withdecreasing genetic distance and at a genetic distance of001 the coverage of the metagenomic library was 46while the coverage of the PCR library decreased evenmore to just 16
Indices of diversity were higher for the metagenomiclibrary than the PCR library (Table 2) The Simpsonindex was 134 for the metagenomic library comparedwith 87 for the PCR library The difference was not assubstantial for the ShannonndashWeaver index (Table 2)Similarly measures of evenness were higher for themetagenomic library than the PCR library Evenness cal-culated using the ShannonndashWeaver index was 073 and066 for the metagenomic library and PCR libraryrespectively while the Simpson index of evenness was026 and 017 for the metagenomic and PCR librariesrespectively
We examined the possibility that diversity based onPCR clone library composition is overestimated becauseof artefacts of the PCR Re-conditioned PCR was used totest for potential artefacts introduced when heterologousPCR products re-anneal and form heteroduplexes that arethen subjected to mismatch repair upon cloning intoE coli (Thompson et al 2002) Re-conditioning of thePCR amplicons reduced the number of RFLP typesexpected in a library of 80 clones from 50 in the standardPCR library to 40 in the library produced with re-condi-tioned PCR (Fig 5) There was no difference betweenrarefaction curves of replicate PCR libraries produced withstandard PCR (Fig 5)
Comparison of FISH and clone libraries
Fluorescence in situ hybridization was used to assess thecorrespondence between clone library composition andbacterial community structure Microscopic enumerationof bacteria hybridized with fluorescently labelled 16SrRNA directed probes revealed a community dominatedby beta-proteobacteria Actinobacteria and Cytophaga-like bacteria (Table 1) As expected for a freshwater sys-tem Actinobacteria and beta-proteobacteria were themost abundant groups making up about 25 of the com-munity Cytophaga-like bacteria were also prominent
Fig 4 Coverage of the metagenomic and 16S rDNA libraries calcu-lated using the LIBSHUFF tool
Fig 5 Rarefaction curves of 16S rDNA clone libraries constructed with and without re-conditioned PCR PCR 1 and PCR 2 are clone libraries constructed from replicate PCR reactions without re-condi-tioning The error bars are one standard deviation
Table 2 Diversity of 16S rRNA genes in metagenomic and PCR clone libraries from the Delaware River
Library
ShannonndashWeaver Simpson
CoverageDiversity Evenness Diversity Evenness
Metagenomic 29 073 134 026 065PCR 26 066 87 017 052
Diversity indices were calculated using equations taken from Dunbar and colleagues (1999)
1890 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
accounting for 17 plusmn 4 of the community In contrastalpha-proteobacteria and gamma-proteobacteria wereconsiderably less abundant (less than 10) Seventy-seven percent of all prokaryotes were identified with agroup probe and 69 plusmn 11 were detected with the gen-eral bacterial probe Eub338
Bacterial community structure determined by FISH andinferred from the PCR library was surprisingly similar Theabundance of Cytophaga-like bacteria in the PCR librarywas within 25 of the abundance determined by FISHthe abundance of gamma-proteobacteria differed by 3or less and Actinobacteria differed by 38 according tothe two analyses (Table 1) In contrast the abundance ofbeta-proteobacteria clones in the PCR library was twicethe abundance of bacteria detected with the beta-proteo-bacteria FISH probe (Table 1) Three types of bacteriaincluding epsilon-proteobacteria Bdellovibrio and bacte-ria belonging to the candidate division TM6 were presentin the PCR library and would not have been detected withany of the FISH probes used in this study Although wehave no estimate of the abundance of bacteria in thesegroups they likely accounted for some of the 23 ofbacteria not detected by FISH (Table 1)
Discussion
It is necessary to examine whether the phylogenic com-position of metagenomic libraries reflects that of the orig-inal microbial composition in order to effectively linkcommunity structure and function revealed by metage-nomic analysis If metagenomic libraries represent abiased sample of microbial diversity they will not yield anaccurate assessment of metabolic diversity Assumingthey are unbiased large insert metagenomic libraries maybe particularly powerful in examining the relationshipbetween community structure and metabolic capacitiesLarge inserts increase the likelihood of obtaining a clonecontaining both lsquofunctionalrsquo genes (coding for metabolicfunction) and genes with phylogenetic information eg16S rRNA genes Therefore it is necessary to know howthe diversity of 16S rRNA genes sampled by large insertmetagenomic libraries corresponds to actual communitystructure Our main finding was that the composition ofthe metagenomic library at a high phylogenetic level wasnot statistically different from the PCR library but closerexamination revealed noteworthy differences between thelibraries and the community structure determined byFISH The metagenomic library sampled a broader diver-sity of Cytophaga-like bacteria than the PCR library but itmissed some groups of potentially important beta-proteo-bacteria The metagenomic and PCR libraries sampledthe same groups of Actinobacteria but FISH analysissuggested that both types of libraries may have failed toadequately sample Actinobacteria
Metagenomic analysis of Delaware River bacteria pro-vided a view of Cytophaga-like bacterial diversity that wasnot possible with the PCR library and FISH analyses Thenumber of clones representing Cytophaga-like bacteria inthe metagenomic library was threefold higher than theabundance determined by FISH and fourfold higher thanin the PCR library It was not surprising that the metage-nomic library sampled a larger diversity of Cytophaga-likebacteria than the PCR library because previous studieshad suggested that Cytophaga-like bacteria are underrep-resented in 16S rDNA PCR libraries relative to the abun-dance of these bacteria detected by FISH (Cottrell andKirchman 2000 Kirchman et al 2003) It was also notsurprising that the metagenomic library had more clonesfrom Cytophaga-like bacteria than would have beenexpected based on the FISH counts because the generalFISH probe CF319a for Cytophaga-like bacteria does notrecognize all Cytophaga-like bacteria (Weller et al 2000)The abundance of Cytophaga-like bacteria according tothe general FISH probe CFB560 for Cytophaga-like bac-teria (OrsquoSullivan et al 2002) was only 15 higher thanthat determined with CF319a The denaturing gradientelectrophoresis (DGGE) PCR primers used in this studyappear to adequately sample Cytophaga-like bacteria atleast compared with FISH analysis (Castle and Kirchman2004) and these primers apparently recognize a broaderdiversity of Cytophaga-like bacteria than the EubA andEubB primers as some Cytophaga-like bacteria detectedby DGGE in the metagenomic library were not sampledby the PCR library
The Cytophaga-like bacterial group includes a highlydiverse collection of bacteria but not all of this diversitywas present in our Delaware River sample In fact mostof the Cytophaga-like bacteria were most closely relatedto just two genera of Flavobacteriales (Flavobacteriumand Myroides) and a cluster in the Flectobacillus groupThis latter group is distinct from the CytophagandashFlavobac-terium cluster which is synonymous with the Flavobacte-riales (Kirchman 2002) The Delaware River bacteria inthe Flectobacillus group belong to the proposedPRD01a001B cluster of Cytophaga-like bacteria identifiedby Zwart and colleagues (2002) Nine of the 12 DelawareRiver sequences in the Flectobacillus group were greaterthan 98 similar to the PRD01a001B clone from theParker River Massachusetts The proposedPRD01a001B cluster which is currently comprised solelyof uncultivated bacteria has also been found in Adiron-dack lakes Lake Baikal and other freshwater environ-ments (Zwart et al 2002) However this group ofCytophaga-like bacteria has not been detected in allfreshwater environments for example it was not found inthe Columbia River (Zwart et al 2002)
The Delaware River beta-proteobacteria belonged toclusters widely distributed in freshwater systems Three
Metagenomic analysis of freshwater bacteria 1891
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
groups of Delaware River beta-proteobacteria clones (BD and E) were highly similar to bacteria in the freshwaterbeta-Proteobacteria clusters proposed by Zwart and col-leagues (2002) Sequences in Delaware River beta-pro-teobacteria group B were 95ndash99 similar to theRodoferax sp in the Rodoferax sp BAL47 cluster GroupD of Delaware Bay beta-proteobacteria corresponded tothe Polynucleobacter necessarius cluster (88ndash99sequence similarity to P necessarius) and sequences ingroup E probably belong to the proposed LD28 clusterSequences in group D were 89ndash97 similarity to theLD28 clone from Lake Loosdrech in the Netherlands(Zwart et al 1998)
The clusters of freshwater beta-proteobacteria pro-posed by Zwart and colleagues (2002) were more preva-lent in the PCR library than in the metagenomic librarysuggesting that they may represent bacteria that are pref-erentially sampled by PCR libraries Delaware River beta-proteobacteria group B included clones from both themetagenomic library and the PCR library but DelawareRiver groups D and E were comprised of only clones fromthe PCR library Different scenarios could lead to theoverrepresentation of bacterial groups in PCR librariescompared with libraries constructed without a PCR stepincluding selective PCR amplification with general bacte-rial primers differences in rRNA gene copy number orvariation in DNA extraction efficiency (von Wintzingerodeet al 1997) Fluorescence in situ hybridization probes areneeded to determine the abundance of these beta-proteo-bacteria which appear to be in a variety of freshwaterenvironments
There was a high degree of similarity in the Actinobac-teria sampled by the metagenomic and PCR libraries Allthree groups of Actinobacteria identified in this studyincluded clones from the metagenomic and PCR librariesOur results indicate that the presumed thicker cell wallsof these G+ bacteria do not interfere with the extraction ofhigh molecular weight DNA from these bacteria Howeverthe overall efficiency of DNA extraction may be low asActinobacteria were underrepresented in the librariescompared with their abundance determined by FISHAlthough the libraries may be missing some Actinobacte-ria they did recover clusters that are apparentlywidespread in freshwater systems Delaware River Acti-nobacteria groups B and C which correspond to the ACK-MI and STA-30 clusters identified by Zwart and colleagues(2002) were identified in the metagenomic and PCRlibraries Clones in cluster B were 92ndash99 similar to theACK-M1 clone and clones in cluster C were 92ndash97similar to the STA-30 clone Delaware River Actinobacte-ria group A which included clones from both libraries didnot overlap with previously described clusters of freshwa-ter Actinobacteria None of the G+ bacterial groups wedetected represented a terrestrial bacterial signal
because of transport of soils into the river Zwart andcolleagues (2002) also did not find evidence for an impor-tant contribution of soil bacteria to bacterial communitiesin rivers
Coverage in the metagenomic library was consistentlyhigher than in the PCR library over a broad range ofevolutionary distances although substantially higher cov-erage by the metagenomic library was restricted to evo-lutionary distances less than 005 Differences incoverage were not the result of different sequencelengths we analysed 1400 bp for the PCR clones but onlythe highly variable V3 region of the 16S rRNA gene forthe metagenomic clones LIBSHUFF analysis conductedwith the V3 region alone also revealed higher coverage bythe metagenomic library (data not shown) Differences incoverage may reflect more comprehensive sampling ofbacterial diversity by the metagenomic library
Lower coverage by the PCR library was consistent withour hypothesis that PCR artefacts lead to overestimatesof diversity Polymerase chain reaction re-conditioning flat-tened the rarefaction curve but the effect was only 20not as pronounced as would be expected if heterodu-plexes were a substantial source of diversity in the PCRlibraries The higher diversity in the metagenomic librarycompared with the PCR library is also consistent with theidea that the PCR libraries do not overestimate bacterialdiversity In fact they probably underestimate diversitybecause they miss certain groups such as Cytophaga-like bacteria (Cottrell and Kirchman 2000) Additional fac-tors such as cloning biases variation in genome size(Fogel et al 1999) rRNA gene copy number (Klappen-bach et al 2000) and ribosome content (Fegatella et al1998) could have resulted in larger differences than weobserved between the two library approaches and FISHGiven the potential sources of differences the degree ofsimilarity between the different library and FISHapproaches was higher than expected
These results and other suggest that the identities ofprokaryotes in aquatic environments are no longer com-pletely hidden in the microbial lsquoblack boxrsquo With the aid ofvarious culture-independent approaches it is now clearthat various proteobacterial groups Cytophaga-like bac-teria and Actinobacteria are the dominant groups of bac-teria in surface waters of aquatic systems (Giovannoniand Rappeacute 2000 Rappeacute and Giovannoni 2003)Archaea make a substantial contribution to prokaryoticdiversity in the ocean depths well below the euphotic zone(Massana et al 1998 Karner et al 2001) Althoughpotentially these groups could contain many subgroupsat least in marine waters it appears that only a few cladesdominate including alpha-proteobacteria belonging to theSAR11 group (Morris et al 2002 Rappeacute et al 2002) andRoseobacter spp (Gonzalez and Moran 1997) Cytoph-aga-like bacteria in Delaware cluster 2 (Kirchman et al
1892 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
2003) the marine Actinobacteria group (Rappeacute et al1999) and Archaea in group I and group II (Fuhrman et al1992 Massana et al 2000 Karner et al 2001)
Dominant groups in freshwater are also becomingapparent Our study revealed several widespread clustersof freshwater bacteria in the Delaware River that wereidentified in the metagenomic and PCR libraries Theseclusters are obvious candidates for further investigationincluding additional metagenomic analysis to exploretheir metabolic potential It is unclear if clusters such asthe P necessarius and LD28 beta-proteobacteria clus-ters which occurred in the PCR library but not themetagenomic library are selectively amplified by the PCRor for some reason are missed by the metagenomic clon-ing Additional examination by FISH could help resolvethe numerical importance of these freshwater bacteriaStudies such as these will further advance our under-standing of aquatic bacterial diversity as it evolves from aview of apparently intractable diversity towards whatappears to be a manageable number of dominant bacte-rial groups
Experimental procedures
Sample collection
A water sample was collected from a depth of 1 m in theDelaware River near Trenton New Jersey (40infin77cent-N74infin493cent-W) in December 2001 and subsampled for the anal-yses performed in this study
Fluorescence in situ hybridization
The sample for FISH analysis was fixed with 2 paraformal-dehyde at 4infinC for 12ndash18 h Bacteria were then collected on02 mm polycarbonate filters rinsed twice with deionizedwater and the filters were stored at -20infinC Bacteria weredetected with FISH probe Eub338 (Amann et al 1990) Wealso tried a mixture of general probes for Bacteria (Daimset al 1999) and found that it detected 70 plusmn 5 of DAPIpositive bacteria compared with 69 plusmn 11 for Eub338 aloneThe FISH probes for alpha- beta- and gamma-proteobacte-ria were Alf968 (Neef 1997 Gloumlckner et al 1999) Bet42aand Gam42a (Manz et al 1992) respectively Cytophaga-likebacteria were assayed with probe CF319a (Manz et al1996) and CFB560 (OrsquoSullivan et al 2002) and Actinobac-teria were enumerated using probe HGC69a (Roller et al1994) Non-specific binding was assessed using a negativecontrol probe (Karner and Fuhrman 1997) Hybridization ofCy3-labelled oligonucleotide probes was performed with por-tions of polycarbonate filters incubated in a hybridizationbuffer containing formamide to control the stringency ofhybridization (Cottrell and Kirchman 2000) Unbound probewas removed by washing the sample with a wash buffer at48infinC The filter piece was then rinsed in water and 70ethanol The air-dried sample was mounted on a glass slidewith a coverslip using an antifade mountant containing2 mg ml-1 DAPI Probe-positive bacteria (500ndash1000 bacteria
for each FISH probe) were enumerated using semiautomatedfluorescence microscopy (Cottrell and Kirchman 2003)
Preparation of the bacterial size fraction
Environmental DNA was extracted from the bacterial sizefraction obtained by pumping river water (500 l) sequentiallythrough a 1 mm nominal pore-size polypropylene string-wound filter (Cole Parmer) and a 08 mm polycarbonate filter(Nuclepore) Bacteria were collected from the filtrate by tan-gential flow filtration using a 01 mm hollow fibre filter (AGTechnology) and an Amicon DC10 gear pump The samplewas concentrated to 2 l rinsed with a buffer (05 M NaCl01 M EDTA 10 mM Tris pH 80) and stored frozen
Library construction using a fosmid vector
Bacteria were collected from thawed bacterial concentrate bycentrifugation in an SS34 rotor at 16 000 rpm for 30 minThe bacteria were resuspended in 1 ml of the supernatantand then mixed with 1 Sea Plaque agarose in deionizedwater at 55infinC (Stein et al 1996) The molten mixture wasdrawn into a 1 ml syringe and allowed to solidify The samplewas equilibrated with buffer (10 mM Tris pH 8 50 mM NaCl01 M EDTA 1 Sarkosyl) and digested with 1 mg ml-1
lysozyme at 37infinC for 4 h Buffer and lysozyme were replacedwith 1 Sarkosyl in 05 M EDTA and the sample wasdigested with 1 mg ml-1 proteinase K at 55infinC for 18 h Fol-lowing the enzyme treatments the sample was electrophore-sed for 8 h at 23 V through 03 SeaKem Gold agarose inTAE buffer DNA entering the gel but larger than the 48 kbsize standard was recovered from the gel by electroelutionand concentrated by ethanol precipitation The DNA was thensheared to 40 kb by 150 passages through a 200 ml pipettetip The DNA was prepared for ligation using a blunt endrepair kit (Epicentre Technologies) following the manufac-turerrsquos protocol The insert DNA was then size selected byelectrophoresis on 1 Sea Plaque GTG agarose DNA wasrecovered from the gel following treatment with Gelaseenzyme (Epicentre Technologies) digestion and ethanol pre-cipitation The DNA was then ligated to the pCC1FOS fosmidvector (Epicentre Technologies) and packaged using phageprotein extract following the manufacturerrsquos recommendedprocedure EPI300 host E coli infected with recombinantphage were plated on LB media containing chloramphenicolClones were picked and sorted into 96-well microtitre plates
PCR library construction
Three PCR libraries were constructed with the DNA isolatedfrom the Delaware River Two replicate libraries were con-structed by cloning amplicons generated from separate PCRreactions using the general bacterial primers EubA and EubB(Lane 1991) A third library was constructed using re-condi-tioned amplicons from the PCR reaction used to generate thefirst library Re-conditioning was performed with three roundsof thermal cycling applied to amplicons diluted 10-fold infresh PCR reagents (Thompson et al 2002) The 25 ml PCRreactions included 05 ml 12 ml and 2 ml additions of 10 mMdeoxynucloside triphosphate 25 mM MgCl2 10 mM primer
Metagenomic analysis of freshwater bacteria 1893
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
stocks respectively Bovine serum albumin was added to afinal concentration of 02 mg ml-1 Two and a half units of Taqpolymerase (Promega) and a one-tenth volume of 10yen bufferwere added to each reaction The thermal cycling conditionsconsisted of a touchdown series in which the annealing tem-perature decreased from 65infinC to 55infinC by 1infinC per cyclefollowed by 15 cycles at 55infinC each for 1 min Each cycleincluded a 1 min denaturation step at 95infinC and an extensionstep of 25 min at 72infinC Amplicons were cloned with a TOPO-TA cloning kit with pCR21-TOPO vector (Invitrogen) accord-ing to the manufacturerrsquos instructions
Screening PCR and metagenomic libraries
Clones were screened for insert size using M13 forward andreverse PCR primers which flank the cloned insert andthose with full-length inserts were screened by restrictionfragment length polymorphism (RFLP) analysis by digestingthe M13 PCR amplicons with a mixture of HhaI and RsaI(New England Biolabs) Restriction patterns on 2 Meta-phore agarose (FMC) agarose gels were compared manuallyClones with identical restriction patterns were groupedtogether into RFLP types The 16S rRNA genes from onerepresentative of each RFLP type in the first library werecompletely sequenced
Pools of 96 fosmid clones were screened for 16S rRNAgenes by DGGE of PCR amplicons generated with primersGC358F and 517R (Muyzer et al 1995) Selected bandsresolved on an 8 polyacrylamide gel containing a gradientof 25ndash55 denaturant (138ndash22 formamide and 105ndash23urea) were re-amplified and sequenced Phylogenetic classi-fication was determined using BLAST and the ARB sequenceanalysis tool as described below Bands with equal electro-phoretic mobility were assigned the same phylogeneticclassification
Sequence and library analysis
Nucleotide sequences were analysed using BLAST [version21 National Center for Biotechnology Information (httpwwwncbinlmnihgovBLAST) and the ARB package (Lud-wig et al 2004)] Sequences were aligned with the ARB FastAligner version 103 adjusted manually with secondary-structure criteria Phylogenetic relationships were deter-mined using the Quick Add Parsimony tool and the1000_pub_may02 phylogenetic tree in the ARB database(June 2002 release)
Sequence similarities were determined using the tools inARB and rarefaction curves were calculated using AnalyticRarefaction 13 (httpwwwugaedustratasoftware) Librar-ies were compared using the LIBSHUFF analysis tool(httpwwwarchesugaedu~whitmanlibshuffhtml) (Single-ton et al 2001) Diversity indices were calculated usingequations taken from Dunbar and colleagues (1999)
Statistical analysis
Libraries were compared using LIBSHUFF analysis (Single-ton et al 2001) This approach compares libraries based ontheir coverage and the number of sequences in one library
that are not found in the second library This so-called lsquohet-erologous coveragersquo is calculated over a range of evolution-ary distances to obtain a coverage curve The differencebetween the lsquoheterologous coveragersquo curve and the coveragecurve of the first library is calculated using the Crameacuter-vonMises test statistic (Pettit 1982) and compared with the cov-erage curve calculated after shuffling sequences between thetwo libraries
Nucleotide sequence accession numbers
The nucleotide sequence data reported in this work aredeposited in GenBank under accession numbers AY562245ndashAY562366
Acknowledgements
This work was supported by the US Department of EnergyMicrobial Genomes Program and the National ScienceFoundation
References
Amann RI Binder BJ Olson RJ Chisholm SWDevereux R and Stahl DA (1990) Combination of 16Sribosomal-RNA-targeted oligonucleotide probes with flowcytometry for analyzing mixed microbial populations ApplEnviron Microbiol 56 1919ndash1925
Amon RMW and Benner R (1996) Photochemical andmicrobial consumption of dissolved organic carbon anddissolved oxygen in the Amazon River system GeochimCosmochim Ac 60 1783ndash1792
Beja O Aravind L Koonin EV Suzuki MT Hadd ANguyen LP et al (2000a) Bacterial rhodopsin evidencefor a new type of phototrophy in the sea Science 2891902ndash1906
Beja O Suzuki MT Koonin EV Aravind L Hadd ANguyen LP et al (2000b) Construction and analysis ofbacterial artificial chromosome libraries from a marinemicrobial assemblage Environ Microbiol 2 516ndash529
Beja O Spudich EN Spudich JL Leclerc M andDeLong EF (2001) Proteorhodopsin phototrophy in theocean Nature 411 786ndash789
Button DK and Robertson BR (2001) Determination ofDNA content of aquatic bacteria by flow cytometry ApplEnviron Microbiol 67 1636ndash1645
Button DK Robertson BR Lepp PW and SchmidtTM (1998) A small dilute-cytoplasm high-affinity novelbacterium isolated by extinction culture and having kineticconstants compatible with growth at ambient concentra-tions of dissolved nutrients in seawater Appl EnvironMicrobiol 64 4467ndash4476
Castle D and Kirchman DL (2004) Composition of estu-arine bacterial communities assessed by denaturing gradi-ent gel electrophoresis and fluorescence in situhybridization Limnol Oceanogr 2 303ndash314
Cottrell MT and Kirchman DL (2000) Community compo-sition of marine bacterioplankton determined by 16S rRNAgene clone libraries and fluorescence in situ hybridizationAppl Environ Microbiol 66 5116ndash5122
1894 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Cottrell MT and Kirchman DL (2003) Contribution ofmajor bacterial groups to bacterial biomass production(thymidine and leucine incorporation) in the Delaware estu-ary Limnol Oceanogr 48 168ndash178
Cottrell MT Moore JA and Kirchman DL (1999) Chiti-nases from uncultured marine microorganisms Appl Envi-ron Microbiol 65 2553ndash2557
Covert JS and Moran MA (2001) Molecular characteriza-tion of estuarine bacterial communities that use high- andlow-molecular weight fractions of dissolved organic carbonAquat Microb Ecol 25 127ndash139
Daims H Bruhl A Amann R Schleifer KH and WagnerM (1999) The domain-specific probe EUB338 is insuffi-cient for the detection of all Bacteria Development andevaluation of a more comprehensive probe set Syst ApplMicrobiol 22 434ndash444
Dunbar J Takala S Barns SM Davis JA and KuskeCR (1999) Levels of bacterial community diversity in fourarid soils compared by cultivation and 16S rRNA genecloning Appl Environ Microbiol 65 1662ndash1669
Fegatella F Lim J Kjelleberg S and Cavicchioli R(1998) Implications of rRNA operon copy number and ribo-some content in the marine oligotrophic ultramicrobacte-rium Sphingomonas sp strain RB2256 Appl EnvironMicrobiol 64 4433ndash4438
Fogel GB Collins CR Li J and Brunk CF (1999)Prokaryotic genome size and SSU rDNA copy numberEstimation of microbial relative abundance from a mixedpopulation Microb Ecol 38 93ndash113
Fuhrman JA McCallum K and Davis AA (1992) Novelmajor Archaebacterial group from marine plankton Nature356 148ndash149
Gillespie DE Brady SF Bettermann AD CianciottoNP Liles MR Rondon MR et al (2002) Isolation ofantibiotics turbomycin A and B from a metagenomic libraryof soil microbial DNA Appl Environ Microbiol 68 4301ndash4306
Giovannoni SJ and Rappeacute MS (2000) Evolution diversityand molecular ecology of marine prokaryotes In MicrobialEcology of the Oceans Kirchman DL (ed) New YorkUSA Wiley-Liss pp 47ndash84
Gloumlckner FO Fuchs BM and Amann R (1999) Bacteri-oplankton compositions of lakes and oceans a first com-parison based on fluorescence in situ hybridization ApplEnviron Microbiol 65 3721ndash3726
Gloumlckner FO Zaichikov E Belkova N Denissova LPernthaler J Pernthaler A and Amann R (2000) Com-parative 16S rRNA analysis of lake bacterioplanktonreveals globally distributed phylogenetic clusters includingan abundant group of actinobacteria Appl Environ Micro-biol 66 5053ndash5065
Gonzalez JM and Moran MA (1997) Numerical domi-nance of a group of marine bacteria in the alpha- subclassof the class Proteobacteria in coastal seawater Appl Envi-ron Microbiol 63 4237ndash4242
Hernes PJ and Benner R (2002) Transport and diagene-sis of dissolved and particulate terrigenous organic matterin the North Pacific Ocean Deep-Sea Res Part II ndash TopStud Oceanogr 49 2119ndash2132
Karner M and Fuhrman JA (1997) Determination of activemarine bacterioplankton a comparison of universal 16S
rRNA probes autoradiography and nucleoid staining ApplEnviron Microbiol 63 1208ndash1213
Karner MB DeLong EF and Karl DM (2001) Archaealdominance in the mesopelagic zone of the Pacific OceanNature 409 507ndash510
Kirchman DL (2002) The ecology of CytophagandashFlavobac-teria in aquatic environments FEMS Microbiol Ecol 3991ndash100
Kirchman DL Yu LY and Cottrell MT (2003) Diversityand abundance of uncultured Cytophaga-like bacteria inthe Delaware Estuary Appl Environ Microbiol 69 6587ndash6596
Klappenbach JA Dunbar JM and Schmidt TM (2000)rRNA operon copy number reflects ecological strategies ofbacteria Appl Environ Microbiol 66 1328ndash1333
Knietsch A Waschkowitz T Bowien S Henne A andDaniel R (2003) Metagenomes of complex microbial con-sortia derived from different soils as sources for novelgenes conferring formation of carbonyls from short-chainpolyols on Escherichia coli J Mol Microbiol Biotechnol 546ndash56
Lane DJ (1991) 16S23S rRNA sequencing In Nucleic AcidTechniques in Bacterial Systematics Stackebrandt E andGoodfellow M (eds) Chichester UK Wiley and Sonspp 115ndash175
Liles MR Manske BF Bintrim SB Handelsman J andGoodman RM (2003) A census of rRNA genes and linkedgenomic sequences within a soil metagenomic libraryAppl Environ Microbiol 69 2684ndash2691
Ludwig W Strunk O Westram R Richter L Meier HYadhukumar et al (2004) ARB a software environmentfor sequence data Nucleic Acids Res 32 1363ndash1371
Manz W Amann R Ludwig W Wagner M and Schlei-fer KH (1992) Phylogenetic oligodeoxynucleotide probesfor the major subclasses of Proteobacteria ndash problems andsolutions Syst Appl Microbiol 15 593ndash600
Manz W Amann R Ludwig W Vancanneyt M andSchleifer KH (1996) Application of a suite of 16S rRNA-specific oligonucleotide probes designed to investigatebacteria of the phylum CytophagandashFlavobacterndashBacteroi-des in the natural environment Microbiology 142 1097ndash1106
Massana R Taylor LJ Murray AE Wu KY JeffreyWH and DeLong EF (1998) Vertical distribution andtemporal variation of marine planktonic archaea in theGerlache Strait Antarctica during early spring LimnolOceanogr 43 607ndash617
Massana R DeLong EF and Pedros-Alio C (2000) A fewcosmopolitan phylotypes dominate planktonic archaealassemblages in widely different oceanic provinces ApplEnviron Microbiol 66 1777ndash1787
Morris RM Rappeacute MS Connon SA Vergin KL Sie-bold WA Carlson CA and Giovannoni SJ (2002)SAR11 clade dominates ocean surface bacterioplanktoncommunities Nature 420 806ndash810
Muyzer G Teske A Wirsen CO and Jannasch HW(1995) Phylogenetic-relationships of Thiomicrospira spe-cies and their identification in deep-sea hydrothermal ventsamples by denaturing gradient gel-electrophoresis of 16SrDNA fragments Arch Microbiol 164 165ndash172
Neef A (1997) Anwendung der in situ-Einzelzell-
Metagenomic analysis of freshwater bacteria 1895
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Identifizierung von Bakterien zur Populationsanalyse inKomplexen Mikrobiellen Biozonosen Munich GermanyTechnische Universitat Munchen
Opsahl S Benner R and Amon RMW (1999) Major fluxof terrigenous dissolved organic matter through the ArcticOcean Limnol Oceanogr 44 2017ndash2023
OrsquoSullivan LA Weightman AJ and Fry JC (2002) Newdegenerate Cytophaga-Flexibacter-Bacteroides-specific16S ribosomal DNA-targeted oligonucleotide probes revealhigh bacterial diversity in River Taff epilithon Appl EnvironMicrobiol 68 201ndash210
Pettit AN (1982) Cramer-von Misess statistic In Encyclo-pedia of Statistical Sciences Kotz S Johnson NL andRead CB (eds) New York USA Wiley pp 220ndash221
Quaiser A Ochsenreiter T Klenk HP Kletzin ATreusch AH Meurer G et al (2002) First insight intothe genome of an uncultivated crenarchaeote from soilEnviron Microbiol 4 603ndash611
Rappeacute MS and Giovannoni SJ (2003) The unculturedmicrobial majority Annu Rev Microbiol 57 369ndash394
Rappeacute MS Gordon DA Vergin KL and GiovannoniSJ (1999) Phylogeny of Actinobacteria small subunit(SSU) rRNA gene clones recovered from marine bacteri-oplankton Syst Appl Microbiol 22 106ndash112
Rappeacute MS Connon SA Vergin KL and GiovannoniSJ (2002) Cultivation of the ubiquitous SAR11 marinebacterioplankton clade Nature 418 630ndash633
Roller C Wagner M Amann R Ludwig W and SchleiferKH (1994) In-situ probing of Gram-positive bacteria withhigh DNA G+C content using 23S-ribosomal-RNA-targetedoligonucleotides Microbiology 140 2849ndash2858
Rondon MR August PR Bettermann AD Brady SFGrossman TH Liles MR et al (2000) Cloning the soilmetagenome a strategy for accessing the genetic andfunctional diversity of uncultured microorganisms ApplEnviron Microbiol 66 2541ndash2547
Sabehi G Massana R Bielawski JP Rosenberg MDelong EF and Beja O (2003) Novel proteorhodopsinvariants from the Mediterranean and Red Seas EnvironMicrobiol 5 842ndash849
Singleton DR Furlong MA Rathbun SL and WhitmanWB (2001) Quantitative comparisons of 16S rRNA genesequence libraries from environmental samples Appl Envi-ron Microbiol 67 4374ndash4376
Stackebrandt E and Goebel BM (1994) A place for DNAndashDNA reassociation and 16S ribosomal-RNA sequence-analysis in the present species definition in bacteriologyInt J Syst Bacteriol 44 846ndash849
Stein JL Marsh TL Wu KY Shizuya H and DeLongEF (1996) Characterization of uncultivated prokaryotesIsolation and analysis of a 40-kilobase-pair genome frag-ment from a planktonic marine archaeon J Bacteriol 178591ndash599
Thompson JR Marcelino LA and Polz MF (2002) Het-eroduplexes in mixed-template amplifications formationconsequence and elimination by lsquoreconditioning PCRrsquoNucleic Acids Res 30 2083ndash2088
Topp E (2003) Bacteria in agricultural soils Diversity roleand future perspectives Can J Soil Sci 83 303ndash309
Weller R Gloumlckner FO and Amann R (2000) 16S rRNA-targeted oligonucleotide probes for the in situ detectionof members of the phylum Cytophaga-Flavobacterium-Bacteroides Syst Appl Microbiol 23 107ndash114
von Wintzingerode F Gobel UB and Stackebrandt E(1997) Determination of microbial diversity in environmen-tal samples pitfalls of PCR-based rRNA analysis FEMSMicrobiol Rev 21 213ndash229
Zaitlin B Watson SB Dixon J and Steel D (2003)Actinomycetes in the Elbow River Basin Alberta CanWater Quality Res J Can 38 115ndash125
Zwart G Hiorns WD Methe BA Van Agterveld MPHuismans R Nold SC et al (1998) Nearly identical 16SrRNA sequences recovered from lakes in North Americaand Europe indicate the existence of clades of globallydistributed freshwater bacteria Syst Appl Microbiol 21546ndash556
Zwart G Crump BC Agterveld M Hagen F and HanSK (2002) Typical freshwater bacteria an analysis ofavailable 16S rRNA gene sequences from plankton oflakes and rivers Aquat Microb Ecol 28 141ndash155
1884
M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd
Environmental Microbiology
7
1883ndash1895
2000b) Phylogenetic analysis of metagenomic clonesbearing 16S rRNA genes from surface water of the PacificOcean identified several groups of
Archaea
and
Bacteria
that are common in marine bacterial communities includ-ing a
Euryarchaeota
clone and clones related to SAR86SAR116 and SAR11 bacteria
Roseobacter
spp and
Cytophaga
-like bacteria (Giovannoni and Rappeacute 2000)However no study has directly compared the phylogeneticcomposition of a metagenomic library and the microbialcommunity
in situ
It may be especially important to exam-ined this issue in freshwaters where Gram-positive bacte-ria can be abundant (Gloumlckner
et al
2000 Zwart
et al
2002) These bacteria could be underrepresented inmetagenomic libraries if DNA extraction from Gram-posi-tive bacteria is less efficient
Bacteria in rivers and other freshwater ecosystemshave not been examined previously using metagenomiclibraries Bacteria in rivers play an important role in theconsumption of organic matter transported from the ter-restrial environment to the ocean (Amon and Benner1996 Opsahl
et al
1999 Hernes and Benner 2002)and the make up of those communities likely has animpact on organic matter consumption (Covert andMoran 2001) The diversity of bacteria in rivers may beespecially high as they may contain mixtures of aquaticand terrestrial bacteria including phylogenetically andmetabolically diverse soil taxa (Liles
et al
2003 Topp2003) Microbes from soil may be introduced into water-ways by runoff (Zaitlin
et al
2003) and mix with freshwa-ter bacteria (Zwart
et al
2002) Less is known aboutriverine bacteria than their counterparts in other aquaticsystems A GenBank search in November 2003 for river-ine bacterial and archaeal 16S rRNA genes returnedapproximately 1000 hits from studies in 14 rivers com-pared with more than 17 000 sequences from marineenvironments and lakes
In this study we determined the coverage and compo-sition of metagenomic and PCR libraries from the Dela-ware River We also contrasted the composition of theselibraries with community structure determined by FISH Inorder to identify why the composition of libraries mightdiffer we tested the hypothesis that heteroduplex artefactsof the PCR step lead to overestimates of bacterialdiversity (Thompson
et al
2002) We found that themetagenomic analysis and PCR libraries identified
Actinobacteria
Cytophaga
-like bacteria and beta-proteobacteria which are potentially important membersof the bacterial community in the Delaware River
Results
Overview of the libraries
The metagenomic library was comprised of 4608 cloneswith an average insert size of 405
plusmn
31 kb (
n
=
15) Thelibrary contained approximately 180 Mb of DNA which isequivalent to 90 bacterial genomes assuming two Mbpper bacterial genome (Button and Robertson 2001)Eighty clones tested positive for 16S rRNA genes whichis consistent with the expected number of rRNA genes inthe library assuming a single 16S rRNA gene per bacte-rial genome Bacteria isolated from aquatic and low-nutri-ent environments typically have only one or two rRNAoperons (Button
et al
1998 Fegatella
et al
1998 Fogel
et al
1999 Klappenbach
et al
2000)The compositions of the metagenomic library and PCR
libraries were not significantly different at a high phyloge-netic level based on analyses of coverage calculatedusing the LIBSHUFF tool
BLAST
analysis was consistentwith this result revealing essentially the same majorgroups of bacteria in the two libraries (Table 1)
Cytoph-aga
-like bacteria beta-
proteobacteria
and
Actinobacteria
Table 1
Phylogenetic composition of the metagenomic and PCR libraries and structure of the actual community determined by FISH
Group of metagenomicclones of PCR clones
of total prokaryotesdetected by FISH
Alpha-
proteobacteria
3 12 6
plusmn
2Beta-
proteobacteria
17 50 25
plusmn
5Gamma-
proteobacteria
4 0 3
plusmn
2Epsilon-
proteobacteria
0 2 ND
Cytophaga-
like 54 13 17
plusmn
4
Actinobacteria
14 16 26
plusmn
5
Firmicutes
3 0 ND
Bdellovibrio
0 4 ND
Verrucomicrobiales
3 2 ND
Spirochaetaceae
1 0 NDndash ndash ndash
Total 100 100 77
Composition of the metagenomic library is based on nucleotide sequences from 72 clones bearing 16S rRNA genes Fifty-eight clones from thePCR library were sequenced and 500ndash1000 DAPI-positive prokaryotes were analysed with each FISH probeND not determined
Metagenomic analysis of freshwater bacteria
1885
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd
Environmental Microbiology
7
1883ndash1895
comprised the largest fractions of the libraries while alpha-
proteobacteria
gamma-
proteobacteria
Firmicutes
Verru-comicrobiales
and
Spirochaetaceae
were minor mem-bers However closer examination of the relativeabundance of different bacterial groups revealed notewor-thy differences between the metagenomic and PCRlibraries
Cytophaga-
like bacteria in clone libraries
The metagenomic and PCR libraries differed in the num-ber and kinds of
Cytophaga
-like bacteria One obviousdifference was the percentage of
Cytophaga
-like clonesOnly 13 of the clones in the PCR library belonged to the
Cytophaga
-like group compared with 54 for themetagenomic library (Table 1) In addition the variety of
Cytophaga
-like bacteria was greater in the metagenomiclibrary than in the PCR library More
Cytophaga
-like clus-ters contained fosmid sequences than PCR sequences inthe phylogenetic tree of
Cytophaga
-like bacteria (Fig 1)The metagenomic library but not the PCR library con-tained
Cytophaga
-like clones in the
Chitinophaga
and
Myroides
groups (Fig 1) One
Cytophaga
-like clone in thePCR library (Sta2ndash97) was not similar to any of themetagenomic clones
Seventy percent of the
Cytophaga
-like bacteria wereassociated with the
Flavobacteriales
group (Fig 1) andthese clones could be grouped into three clusters (AndashC)A fourth cluster (D) was in the
Flectobacillus
group(Fig 1) Clusters A and B were related to different kindsof
Flavobacterium
spp and included some cultured bac-teria Cluster A included an isolate from the Elbe Riverand cluster B included
Flavobacterium xinjiangensi
Theaverage sequence identity in these two clusters with cul-tured members was 966
plusmn
34 and 945
plusmn
61respectively Clusters C and D were associated with the
Myroides
and
Flectobacillus
groups respectively andincluded bacteria with a higher degree of sequence simi-larity (985
plusmn
05 and 995
plusmn
04 respectively) butthese clusters had no cultured members
Some of the
Cytophaga
-like bacteria were representedin both libraries Half of the
Cytophaga
-like clones in thePCR library were highly similar to clones in the metage-nomic library and every
Cytophaga
-like group with clonesfrom the PCR library also had representatives from themetagenomic library For example cluster D was com-prised of three PCR clones and four metagenomic clones(Fig 2) Cluster A included three metagenomic clonesand a highly similar clone from the PCR library Howeversimilarity between
Cytophaga
-like bacteria in the twolibraries was restricted to these two clusters
Beta-
proteobacteria
in clone libraries
The metage-nomic and PCR libraries also differed in the number andkinds of beta-
proteobacteria The PCR library had more
than twice as many clones in this group than the metage-nomic library (50 versus 17 respectively) (Table 1)The beta-proteobacteria clones in the two librariesbelonged to five clusters (AndashE) Clusters D and E werecomprised exclusively of clones from the PCR library(Fig 2) The degree of relatedness in the clusters madeup of only PCR clones and in cluster A was low rangingfrom 89 to 95 sequence identity In contrast some ofthe beta-proteobacteria sampled by the two libraries weresimilar Clusters B and C which included clones from bothlibraries were comprised of very closely related bacteriawith similarities of 964 plusmn 16 and 975 plusmn 31respectively
Other bacteria in the clone libraries
In contrast to the differences outlined above for theCytophaga-like and beta-proteobacteria groups Actino-bacteria in the metagenomic and PCR libraries were quitesimilar The libraries had equivalent percentages of Acti-nobacteria (about 15) and there was considerable over-lap in the types of Actinobacteria in the two libraries Ofthe three clusters of Actinobacteria clones (AndashC) eachincluded clones from both libraries (Fig 3) Cluster Bincluded the most closely related clones with averagesimilarities of 964 plusmn 22 The Actinobacteria clones inclusters A and C were less closely related Sequencesimilarities in these clusters were 932 plusmn 41 and929 plusmn 54 respectively
Alpha-proteobacteria comprised a minor component ofthe libraries accounting for only 3 and 12 of theclones in the metagenomic and PCR clones respectively(Table 1) The alpha-proteobacteria clones in the twolibraries were closely related to each other and to somecultured alpha-proteobacteria (results not shown) Threeclones from the PCR library and two from the metage-nomic library belonged to a cluster that included culturedSphingomonas spp One alpha-proteobacteria clone fromthe PCR library belonged to the Rhodospirillales group
The libraries also included clones belonging to fivegroups of bacteria that can be present in aquatic sys-tems but are typically minor members of the communityThese groups included Firmicutes and Spirochaetaceaethat occurred in only the metagenomic library and epsi-lon-proteobacteria and Bdellovibrio that were foundexclusively in the PCR library (Table 1) Clones belong-ing to the Verrucomicrobiales group occurred in bothlibraries
Library coverage and diversity
Library coverage increased with the genetic distance asexpected At an evolutionary distance of 003 coveragesof the metagenomic and PCR libraries were 65 and 51
1886 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Fig 1 Phylogenetic relationships of selected Cytophaga-like bacteria as depicted in the ARB database tree constructed of sequences longer than 1000 bp (June 2002) Delaware River Cytophaga-like bacteria were added using the ARB parsimony quick add tool Clones in the library of 16S rDNA amplicons are labelled PCR and 16S rDNA-bearing metagenomic DNA clones are labelled Fosmid Values on the trapezoids indicate the number of sequences in the group Labels A B C and D refer to clusters of closely related bacteria that are discussed in the text The scale bar represents 01 substitutions per site
Metagenomic analysis of freshwater bacteria 1887
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Fig 2 Phylogenetic relationships of selected beta-proteobacteria as depicted in the ARB database tree constructed of sequences longer than 1000 bp (June 2002) Delaware River beta-proteobacteria were added using the ARB parsimony quick add tool Clones in the library of 16S rDNA amplicons are labelled PCR and 16S rDNA-bearing metagenomic DNA clones are labelled Fosmid Values on the trapezoids indicate the number of sequences in the group Labels A B C D and E refer to clusters of closely related bacteria that are discussed in the text The scale bar represents 01 substitutions per site
1888 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Fig 3 Phylogenetic relationships of selected Actinobacteria as depicted in the ARB database tree constructed of sequences longer than 1000 bp (June 2002) Delaware River Actinobacteria were added using the ARB parsimony quick add tool Clones in the library of 16S rDNA amplicons are labelled PCR and 16S rDNA-bearing metagenomic DNA clones are labelled Fosmid Values on the trapezoids indicate the number of sequences in the group Labels A B and C refer to clusters of closely related bacteria that are discussed in the text The scale bar represents 01 substitutions per site
Metagenomic analysis of freshwater bacteria 1889
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
respectively (Fig 4) This distance is approximately equalto 97 sequence identity which is one approximation ofspecies-level identity (Stackebrandt and Goebel 1994)Coverage of the metagenomic library was higher thancoverage of the PCR library at all levels of genetic dis-tance with the greatest differences occurring at smallgenetic distance Coverage decreased rapidly withdecreasing genetic distance and at a genetic distance of001 the coverage of the metagenomic library was 46while the coverage of the PCR library decreased evenmore to just 16
Indices of diversity were higher for the metagenomiclibrary than the PCR library (Table 2) The Simpsonindex was 134 for the metagenomic library comparedwith 87 for the PCR library The difference was not assubstantial for the ShannonndashWeaver index (Table 2)Similarly measures of evenness were higher for themetagenomic library than the PCR library Evenness cal-culated using the ShannonndashWeaver index was 073 and066 for the metagenomic library and PCR libraryrespectively while the Simpson index of evenness was026 and 017 for the metagenomic and PCR librariesrespectively
We examined the possibility that diversity based onPCR clone library composition is overestimated becauseof artefacts of the PCR Re-conditioned PCR was used totest for potential artefacts introduced when heterologousPCR products re-anneal and form heteroduplexes that arethen subjected to mismatch repair upon cloning intoE coli (Thompson et al 2002) Re-conditioning of thePCR amplicons reduced the number of RFLP typesexpected in a library of 80 clones from 50 in the standardPCR library to 40 in the library produced with re-condi-tioned PCR (Fig 5) There was no difference betweenrarefaction curves of replicate PCR libraries produced withstandard PCR (Fig 5)
Comparison of FISH and clone libraries
Fluorescence in situ hybridization was used to assess thecorrespondence between clone library composition andbacterial community structure Microscopic enumerationof bacteria hybridized with fluorescently labelled 16SrRNA directed probes revealed a community dominatedby beta-proteobacteria Actinobacteria and Cytophaga-like bacteria (Table 1) As expected for a freshwater sys-tem Actinobacteria and beta-proteobacteria were themost abundant groups making up about 25 of the com-munity Cytophaga-like bacteria were also prominent
Fig 4 Coverage of the metagenomic and 16S rDNA libraries calcu-lated using the LIBSHUFF tool
Fig 5 Rarefaction curves of 16S rDNA clone libraries constructed with and without re-conditioned PCR PCR 1 and PCR 2 are clone libraries constructed from replicate PCR reactions without re-condi-tioning The error bars are one standard deviation
Table 2 Diversity of 16S rRNA genes in metagenomic and PCR clone libraries from the Delaware River
Library
ShannonndashWeaver Simpson
CoverageDiversity Evenness Diversity Evenness
Metagenomic 29 073 134 026 065PCR 26 066 87 017 052
Diversity indices were calculated using equations taken from Dunbar and colleagues (1999)
1890 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
accounting for 17 plusmn 4 of the community In contrastalpha-proteobacteria and gamma-proteobacteria wereconsiderably less abundant (less than 10) Seventy-seven percent of all prokaryotes were identified with agroup probe and 69 plusmn 11 were detected with the gen-eral bacterial probe Eub338
Bacterial community structure determined by FISH andinferred from the PCR library was surprisingly similar Theabundance of Cytophaga-like bacteria in the PCR librarywas within 25 of the abundance determined by FISHthe abundance of gamma-proteobacteria differed by 3or less and Actinobacteria differed by 38 according tothe two analyses (Table 1) In contrast the abundance ofbeta-proteobacteria clones in the PCR library was twicethe abundance of bacteria detected with the beta-proteo-bacteria FISH probe (Table 1) Three types of bacteriaincluding epsilon-proteobacteria Bdellovibrio and bacte-ria belonging to the candidate division TM6 were presentin the PCR library and would not have been detected withany of the FISH probes used in this study Although wehave no estimate of the abundance of bacteria in thesegroups they likely accounted for some of the 23 ofbacteria not detected by FISH (Table 1)
Discussion
It is necessary to examine whether the phylogenic com-position of metagenomic libraries reflects that of the orig-inal microbial composition in order to effectively linkcommunity structure and function revealed by metage-nomic analysis If metagenomic libraries represent abiased sample of microbial diversity they will not yield anaccurate assessment of metabolic diversity Assumingthey are unbiased large insert metagenomic libraries maybe particularly powerful in examining the relationshipbetween community structure and metabolic capacitiesLarge inserts increase the likelihood of obtaining a clonecontaining both lsquofunctionalrsquo genes (coding for metabolicfunction) and genes with phylogenetic information eg16S rRNA genes Therefore it is necessary to know howthe diversity of 16S rRNA genes sampled by large insertmetagenomic libraries corresponds to actual communitystructure Our main finding was that the composition ofthe metagenomic library at a high phylogenetic level wasnot statistically different from the PCR library but closerexamination revealed noteworthy differences between thelibraries and the community structure determined byFISH The metagenomic library sampled a broader diver-sity of Cytophaga-like bacteria than the PCR library but itmissed some groups of potentially important beta-proteo-bacteria The metagenomic and PCR libraries sampledthe same groups of Actinobacteria but FISH analysissuggested that both types of libraries may have failed toadequately sample Actinobacteria
Metagenomic analysis of Delaware River bacteria pro-vided a view of Cytophaga-like bacterial diversity that wasnot possible with the PCR library and FISH analyses Thenumber of clones representing Cytophaga-like bacteria inthe metagenomic library was threefold higher than theabundance determined by FISH and fourfold higher thanin the PCR library It was not surprising that the metage-nomic library sampled a larger diversity of Cytophaga-likebacteria than the PCR library because previous studieshad suggested that Cytophaga-like bacteria are underrep-resented in 16S rDNA PCR libraries relative to the abun-dance of these bacteria detected by FISH (Cottrell andKirchman 2000 Kirchman et al 2003) It was also notsurprising that the metagenomic library had more clonesfrom Cytophaga-like bacteria than would have beenexpected based on the FISH counts because the generalFISH probe CF319a for Cytophaga-like bacteria does notrecognize all Cytophaga-like bacteria (Weller et al 2000)The abundance of Cytophaga-like bacteria according tothe general FISH probe CFB560 for Cytophaga-like bac-teria (OrsquoSullivan et al 2002) was only 15 higher thanthat determined with CF319a The denaturing gradientelectrophoresis (DGGE) PCR primers used in this studyappear to adequately sample Cytophaga-like bacteria atleast compared with FISH analysis (Castle and Kirchman2004) and these primers apparently recognize a broaderdiversity of Cytophaga-like bacteria than the EubA andEubB primers as some Cytophaga-like bacteria detectedby DGGE in the metagenomic library were not sampledby the PCR library
The Cytophaga-like bacterial group includes a highlydiverse collection of bacteria but not all of this diversitywas present in our Delaware River sample In fact mostof the Cytophaga-like bacteria were most closely relatedto just two genera of Flavobacteriales (Flavobacteriumand Myroides) and a cluster in the Flectobacillus groupThis latter group is distinct from the CytophagandashFlavobac-terium cluster which is synonymous with the Flavobacte-riales (Kirchman 2002) The Delaware River bacteria inthe Flectobacillus group belong to the proposedPRD01a001B cluster of Cytophaga-like bacteria identifiedby Zwart and colleagues (2002) Nine of the 12 DelawareRiver sequences in the Flectobacillus group were greaterthan 98 similar to the PRD01a001B clone from theParker River Massachusetts The proposedPRD01a001B cluster which is currently comprised solelyof uncultivated bacteria has also been found in Adiron-dack lakes Lake Baikal and other freshwater environ-ments (Zwart et al 2002) However this group ofCytophaga-like bacteria has not been detected in allfreshwater environments for example it was not found inthe Columbia River (Zwart et al 2002)
The Delaware River beta-proteobacteria belonged toclusters widely distributed in freshwater systems Three
Metagenomic analysis of freshwater bacteria 1891
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
groups of Delaware River beta-proteobacteria clones (BD and E) were highly similar to bacteria in the freshwaterbeta-Proteobacteria clusters proposed by Zwart and col-leagues (2002) Sequences in Delaware River beta-pro-teobacteria group B were 95ndash99 similar to theRodoferax sp in the Rodoferax sp BAL47 cluster GroupD of Delaware Bay beta-proteobacteria corresponded tothe Polynucleobacter necessarius cluster (88ndash99sequence similarity to P necessarius) and sequences ingroup E probably belong to the proposed LD28 clusterSequences in group D were 89ndash97 similarity to theLD28 clone from Lake Loosdrech in the Netherlands(Zwart et al 1998)
The clusters of freshwater beta-proteobacteria pro-posed by Zwart and colleagues (2002) were more preva-lent in the PCR library than in the metagenomic librarysuggesting that they may represent bacteria that are pref-erentially sampled by PCR libraries Delaware River beta-proteobacteria group B included clones from both themetagenomic library and the PCR library but DelawareRiver groups D and E were comprised of only clones fromthe PCR library Different scenarios could lead to theoverrepresentation of bacterial groups in PCR librariescompared with libraries constructed without a PCR stepincluding selective PCR amplification with general bacte-rial primers differences in rRNA gene copy number orvariation in DNA extraction efficiency (von Wintzingerodeet al 1997) Fluorescence in situ hybridization probes areneeded to determine the abundance of these beta-proteo-bacteria which appear to be in a variety of freshwaterenvironments
There was a high degree of similarity in the Actinobac-teria sampled by the metagenomic and PCR libraries Allthree groups of Actinobacteria identified in this studyincluded clones from the metagenomic and PCR librariesOur results indicate that the presumed thicker cell wallsof these G+ bacteria do not interfere with the extraction ofhigh molecular weight DNA from these bacteria Howeverthe overall efficiency of DNA extraction may be low asActinobacteria were underrepresented in the librariescompared with their abundance determined by FISHAlthough the libraries may be missing some Actinobacte-ria they did recover clusters that are apparentlywidespread in freshwater systems Delaware River Acti-nobacteria groups B and C which correspond to the ACK-MI and STA-30 clusters identified by Zwart and colleagues(2002) were identified in the metagenomic and PCRlibraries Clones in cluster B were 92ndash99 similar to theACK-M1 clone and clones in cluster C were 92ndash97similar to the STA-30 clone Delaware River Actinobacte-ria group A which included clones from both libraries didnot overlap with previously described clusters of freshwa-ter Actinobacteria None of the G+ bacterial groups wedetected represented a terrestrial bacterial signal
because of transport of soils into the river Zwart andcolleagues (2002) also did not find evidence for an impor-tant contribution of soil bacteria to bacterial communitiesin rivers
Coverage in the metagenomic library was consistentlyhigher than in the PCR library over a broad range ofevolutionary distances although substantially higher cov-erage by the metagenomic library was restricted to evo-lutionary distances less than 005 Differences incoverage were not the result of different sequencelengths we analysed 1400 bp for the PCR clones but onlythe highly variable V3 region of the 16S rRNA gene forthe metagenomic clones LIBSHUFF analysis conductedwith the V3 region alone also revealed higher coverage bythe metagenomic library (data not shown) Differences incoverage may reflect more comprehensive sampling ofbacterial diversity by the metagenomic library
Lower coverage by the PCR library was consistent withour hypothesis that PCR artefacts lead to overestimatesof diversity Polymerase chain reaction re-conditioning flat-tened the rarefaction curve but the effect was only 20not as pronounced as would be expected if heterodu-plexes were a substantial source of diversity in the PCRlibraries The higher diversity in the metagenomic librarycompared with the PCR library is also consistent with theidea that the PCR libraries do not overestimate bacterialdiversity In fact they probably underestimate diversitybecause they miss certain groups such as Cytophaga-like bacteria (Cottrell and Kirchman 2000) Additional fac-tors such as cloning biases variation in genome size(Fogel et al 1999) rRNA gene copy number (Klappen-bach et al 2000) and ribosome content (Fegatella et al1998) could have resulted in larger differences than weobserved between the two library approaches and FISHGiven the potential sources of differences the degree ofsimilarity between the different library and FISHapproaches was higher than expected
These results and other suggest that the identities ofprokaryotes in aquatic environments are no longer com-pletely hidden in the microbial lsquoblack boxrsquo With the aid ofvarious culture-independent approaches it is now clearthat various proteobacterial groups Cytophaga-like bac-teria and Actinobacteria are the dominant groups of bac-teria in surface waters of aquatic systems (Giovannoniand Rappeacute 2000 Rappeacute and Giovannoni 2003)Archaea make a substantial contribution to prokaryoticdiversity in the ocean depths well below the euphotic zone(Massana et al 1998 Karner et al 2001) Althoughpotentially these groups could contain many subgroupsat least in marine waters it appears that only a few cladesdominate including alpha-proteobacteria belonging to theSAR11 group (Morris et al 2002 Rappeacute et al 2002) andRoseobacter spp (Gonzalez and Moran 1997) Cytoph-aga-like bacteria in Delaware cluster 2 (Kirchman et al
1892 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
2003) the marine Actinobacteria group (Rappeacute et al1999) and Archaea in group I and group II (Fuhrman et al1992 Massana et al 2000 Karner et al 2001)
Dominant groups in freshwater are also becomingapparent Our study revealed several widespread clustersof freshwater bacteria in the Delaware River that wereidentified in the metagenomic and PCR libraries Theseclusters are obvious candidates for further investigationincluding additional metagenomic analysis to exploretheir metabolic potential It is unclear if clusters such asthe P necessarius and LD28 beta-proteobacteria clus-ters which occurred in the PCR library but not themetagenomic library are selectively amplified by the PCRor for some reason are missed by the metagenomic clon-ing Additional examination by FISH could help resolvethe numerical importance of these freshwater bacteriaStudies such as these will further advance our under-standing of aquatic bacterial diversity as it evolves from aview of apparently intractable diversity towards whatappears to be a manageable number of dominant bacte-rial groups
Experimental procedures
Sample collection
A water sample was collected from a depth of 1 m in theDelaware River near Trenton New Jersey (40infin77cent-N74infin493cent-W) in December 2001 and subsampled for the anal-yses performed in this study
Fluorescence in situ hybridization
The sample for FISH analysis was fixed with 2 paraformal-dehyde at 4infinC for 12ndash18 h Bacteria were then collected on02 mm polycarbonate filters rinsed twice with deionizedwater and the filters were stored at -20infinC Bacteria weredetected with FISH probe Eub338 (Amann et al 1990) Wealso tried a mixture of general probes for Bacteria (Daimset al 1999) and found that it detected 70 plusmn 5 of DAPIpositive bacteria compared with 69 plusmn 11 for Eub338 aloneThe FISH probes for alpha- beta- and gamma-proteobacte-ria were Alf968 (Neef 1997 Gloumlckner et al 1999) Bet42aand Gam42a (Manz et al 1992) respectively Cytophaga-likebacteria were assayed with probe CF319a (Manz et al1996) and CFB560 (OrsquoSullivan et al 2002) and Actinobac-teria were enumerated using probe HGC69a (Roller et al1994) Non-specific binding was assessed using a negativecontrol probe (Karner and Fuhrman 1997) Hybridization ofCy3-labelled oligonucleotide probes was performed with por-tions of polycarbonate filters incubated in a hybridizationbuffer containing formamide to control the stringency ofhybridization (Cottrell and Kirchman 2000) Unbound probewas removed by washing the sample with a wash buffer at48infinC The filter piece was then rinsed in water and 70ethanol The air-dried sample was mounted on a glass slidewith a coverslip using an antifade mountant containing2 mg ml-1 DAPI Probe-positive bacteria (500ndash1000 bacteria
for each FISH probe) were enumerated using semiautomatedfluorescence microscopy (Cottrell and Kirchman 2003)
Preparation of the bacterial size fraction
Environmental DNA was extracted from the bacterial sizefraction obtained by pumping river water (500 l) sequentiallythrough a 1 mm nominal pore-size polypropylene string-wound filter (Cole Parmer) and a 08 mm polycarbonate filter(Nuclepore) Bacteria were collected from the filtrate by tan-gential flow filtration using a 01 mm hollow fibre filter (AGTechnology) and an Amicon DC10 gear pump The samplewas concentrated to 2 l rinsed with a buffer (05 M NaCl01 M EDTA 10 mM Tris pH 80) and stored frozen
Library construction using a fosmid vector
Bacteria were collected from thawed bacterial concentrate bycentrifugation in an SS34 rotor at 16 000 rpm for 30 minThe bacteria were resuspended in 1 ml of the supernatantand then mixed with 1 Sea Plaque agarose in deionizedwater at 55infinC (Stein et al 1996) The molten mixture wasdrawn into a 1 ml syringe and allowed to solidify The samplewas equilibrated with buffer (10 mM Tris pH 8 50 mM NaCl01 M EDTA 1 Sarkosyl) and digested with 1 mg ml-1
lysozyme at 37infinC for 4 h Buffer and lysozyme were replacedwith 1 Sarkosyl in 05 M EDTA and the sample wasdigested with 1 mg ml-1 proteinase K at 55infinC for 18 h Fol-lowing the enzyme treatments the sample was electrophore-sed for 8 h at 23 V through 03 SeaKem Gold agarose inTAE buffer DNA entering the gel but larger than the 48 kbsize standard was recovered from the gel by electroelutionand concentrated by ethanol precipitation The DNA was thensheared to 40 kb by 150 passages through a 200 ml pipettetip The DNA was prepared for ligation using a blunt endrepair kit (Epicentre Technologies) following the manufac-turerrsquos protocol The insert DNA was then size selected byelectrophoresis on 1 Sea Plaque GTG agarose DNA wasrecovered from the gel following treatment with Gelaseenzyme (Epicentre Technologies) digestion and ethanol pre-cipitation The DNA was then ligated to the pCC1FOS fosmidvector (Epicentre Technologies) and packaged using phageprotein extract following the manufacturerrsquos recommendedprocedure EPI300 host E coli infected with recombinantphage were plated on LB media containing chloramphenicolClones were picked and sorted into 96-well microtitre plates
PCR library construction
Three PCR libraries were constructed with the DNA isolatedfrom the Delaware River Two replicate libraries were con-structed by cloning amplicons generated from separate PCRreactions using the general bacterial primers EubA and EubB(Lane 1991) A third library was constructed using re-condi-tioned amplicons from the PCR reaction used to generate thefirst library Re-conditioning was performed with three roundsof thermal cycling applied to amplicons diluted 10-fold infresh PCR reagents (Thompson et al 2002) The 25 ml PCRreactions included 05 ml 12 ml and 2 ml additions of 10 mMdeoxynucloside triphosphate 25 mM MgCl2 10 mM primer
Metagenomic analysis of freshwater bacteria 1893
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
stocks respectively Bovine serum albumin was added to afinal concentration of 02 mg ml-1 Two and a half units of Taqpolymerase (Promega) and a one-tenth volume of 10yen bufferwere added to each reaction The thermal cycling conditionsconsisted of a touchdown series in which the annealing tem-perature decreased from 65infinC to 55infinC by 1infinC per cyclefollowed by 15 cycles at 55infinC each for 1 min Each cycleincluded a 1 min denaturation step at 95infinC and an extensionstep of 25 min at 72infinC Amplicons were cloned with a TOPO-TA cloning kit with pCR21-TOPO vector (Invitrogen) accord-ing to the manufacturerrsquos instructions
Screening PCR and metagenomic libraries
Clones were screened for insert size using M13 forward andreverse PCR primers which flank the cloned insert andthose with full-length inserts were screened by restrictionfragment length polymorphism (RFLP) analysis by digestingthe M13 PCR amplicons with a mixture of HhaI and RsaI(New England Biolabs) Restriction patterns on 2 Meta-phore agarose (FMC) agarose gels were compared manuallyClones with identical restriction patterns were groupedtogether into RFLP types The 16S rRNA genes from onerepresentative of each RFLP type in the first library werecompletely sequenced
Pools of 96 fosmid clones were screened for 16S rRNAgenes by DGGE of PCR amplicons generated with primersGC358F and 517R (Muyzer et al 1995) Selected bandsresolved on an 8 polyacrylamide gel containing a gradientof 25ndash55 denaturant (138ndash22 formamide and 105ndash23urea) were re-amplified and sequenced Phylogenetic classi-fication was determined using BLAST and the ARB sequenceanalysis tool as described below Bands with equal electro-phoretic mobility were assigned the same phylogeneticclassification
Sequence and library analysis
Nucleotide sequences were analysed using BLAST [version21 National Center for Biotechnology Information (httpwwwncbinlmnihgovBLAST) and the ARB package (Lud-wig et al 2004)] Sequences were aligned with the ARB FastAligner version 103 adjusted manually with secondary-structure criteria Phylogenetic relationships were deter-mined using the Quick Add Parsimony tool and the1000_pub_may02 phylogenetic tree in the ARB database(June 2002 release)
Sequence similarities were determined using the tools inARB and rarefaction curves were calculated using AnalyticRarefaction 13 (httpwwwugaedustratasoftware) Librar-ies were compared using the LIBSHUFF analysis tool(httpwwwarchesugaedu~whitmanlibshuffhtml) (Single-ton et al 2001) Diversity indices were calculated usingequations taken from Dunbar and colleagues (1999)
Statistical analysis
Libraries were compared using LIBSHUFF analysis (Single-ton et al 2001) This approach compares libraries based ontheir coverage and the number of sequences in one library
that are not found in the second library This so-called lsquohet-erologous coveragersquo is calculated over a range of evolution-ary distances to obtain a coverage curve The differencebetween the lsquoheterologous coveragersquo curve and the coveragecurve of the first library is calculated using the Crameacuter-vonMises test statistic (Pettit 1982) and compared with the cov-erage curve calculated after shuffling sequences between thetwo libraries
Nucleotide sequence accession numbers
The nucleotide sequence data reported in this work aredeposited in GenBank under accession numbers AY562245ndashAY562366
Acknowledgements
This work was supported by the US Department of EnergyMicrobial Genomes Program and the National ScienceFoundation
References
Amann RI Binder BJ Olson RJ Chisholm SWDevereux R and Stahl DA (1990) Combination of 16Sribosomal-RNA-targeted oligonucleotide probes with flowcytometry for analyzing mixed microbial populations ApplEnviron Microbiol 56 1919ndash1925
Amon RMW and Benner R (1996) Photochemical andmicrobial consumption of dissolved organic carbon anddissolved oxygen in the Amazon River system GeochimCosmochim Ac 60 1783ndash1792
Beja O Aravind L Koonin EV Suzuki MT Hadd ANguyen LP et al (2000a) Bacterial rhodopsin evidencefor a new type of phototrophy in the sea Science 2891902ndash1906
Beja O Suzuki MT Koonin EV Aravind L Hadd ANguyen LP et al (2000b) Construction and analysis ofbacterial artificial chromosome libraries from a marinemicrobial assemblage Environ Microbiol 2 516ndash529
Beja O Spudich EN Spudich JL Leclerc M andDeLong EF (2001) Proteorhodopsin phototrophy in theocean Nature 411 786ndash789
Button DK and Robertson BR (2001) Determination ofDNA content of aquatic bacteria by flow cytometry ApplEnviron Microbiol 67 1636ndash1645
Button DK Robertson BR Lepp PW and SchmidtTM (1998) A small dilute-cytoplasm high-affinity novelbacterium isolated by extinction culture and having kineticconstants compatible with growth at ambient concentra-tions of dissolved nutrients in seawater Appl EnvironMicrobiol 64 4467ndash4476
Castle D and Kirchman DL (2004) Composition of estu-arine bacterial communities assessed by denaturing gradi-ent gel electrophoresis and fluorescence in situhybridization Limnol Oceanogr 2 303ndash314
Cottrell MT and Kirchman DL (2000) Community compo-sition of marine bacterioplankton determined by 16S rRNAgene clone libraries and fluorescence in situ hybridizationAppl Environ Microbiol 66 5116ndash5122
1894 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Cottrell MT and Kirchman DL (2003) Contribution ofmajor bacterial groups to bacterial biomass production(thymidine and leucine incorporation) in the Delaware estu-ary Limnol Oceanogr 48 168ndash178
Cottrell MT Moore JA and Kirchman DL (1999) Chiti-nases from uncultured marine microorganisms Appl Envi-ron Microbiol 65 2553ndash2557
Covert JS and Moran MA (2001) Molecular characteriza-tion of estuarine bacterial communities that use high- andlow-molecular weight fractions of dissolved organic carbonAquat Microb Ecol 25 127ndash139
Daims H Bruhl A Amann R Schleifer KH and WagnerM (1999) The domain-specific probe EUB338 is insuffi-cient for the detection of all Bacteria Development andevaluation of a more comprehensive probe set Syst ApplMicrobiol 22 434ndash444
Dunbar J Takala S Barns SM Davis JA and KuskeCR (1999) Levels of bacterial community diversity in fourarid soils compared by cultivation and 16S rRNA genecloning Appl Environ Microbiol 65 1662ndash1669
Fegatella F Lim J Kjelleberg S and Cavicchioli R(1998) Implications of rRNA operon copy number and ribo-some content in the marine oligotrophic ultramicrobacte-rium Sphingomonas sp strain RB2256 Appl EnvironMicrobiol 64 4433ndash4438
Fogel GB Collins CR Li J and Brunk CF (1999)Prokaryotic genome size and SSU rDNA copy numberEstimation of microbial relative abundance from a mixedpopulation Microb Ecol 38 93ndash113
Fuhrman JA McCallum K and Davis AA (1992) Novelmajor Archaebacterial group from marine plankton Nature356 148ndash149
Gillespie DE Brady SF Bettermann AD CianciottoNP Liles MR Rondon MR et al (2002) Isolation ofantibiotics turbomycin A and B from a metagenomic libraryof soil microbial DNA Appl Environ Microbiol 68 4301ndash4306
Giovannoni SJ and Rappeacute MS (2000) Evolution diversityand molecular ecology of marine prokaryotes In MicrobialEcology of the Oceans Kirchman DL (ed) New YorkUSA Wiley-Liss pp 47ndash84
Gloumlckner FO Fuchs BM and Amann R (1999) Bacteri-oplankton compositions of lakes and oceans a first com-parison based on fluorescence in situ hybridization ApplEnviron Microbiol 65 3721ndash3726
Gloumlckner FO Zaichikov E Belkova N Denissova LPernthaler J Pernthaler A and Amann R (2000) Com-parative 16S rRNA analysis of lake bacterioplanktonreveals globally distributed phylogenetic clusters includingan abundant group of actinobacteria Appl Environ Micro-biol 66 5053ndash5065
Gonzalez JM and Moran MA (1997) Numerical domi-nance of a group of marine bacteria in the alpha- subclassof the class Proteobacteria in coastal seawater Appl Envi-ron Microbiol 63 4237ndash4242
Hernes PJ and Benner R (2002) Transport and diagene-sis of dissolved and particulate terrigenous organic matterin the North Pacific Ocean Deep-Sea Res Part II ndash TopStud Oceanogr 49 2119ndash2132
Karner M and Fuhrman JA (1997) Determination of activemarine bacterioplankton a comparison of universal 16S
rRNA probes autoradiography and nucleoid staining ApplEnviron Microbiol 63 1208ndash1213
Karner MB DeLong EF and Karl DM (2001) Archaealdominance in the mesopelagic zone of the Pacific OceanNature 409 507ndash510
Kirchman DL (2002) The ecology of CytophagandashFlavobac-teria in aquatic environments FEMS Microbiol Ecol 3991ndash100
Kirchman DL Yu LY and Cottrell MT (2003) Diversityand abundance of uncultured Cytophaga-like bacteria inthe Delaware Estuary Appl Environ Microbiol 69 6587ndash6596
Klappenbach JA Dunbar JM and Schmidt TM (2000)rRNA operon copy number reflects ecological strategies ofbacteria Appl Environ Microbiol 66 1328ndash1333
Knietsch A Waschkowitz T Bowien S Henne A andDaniel R (2003) Metagenomes of complex microbial con-sortia derived from different soils as sources for novelgenes conferring formation of carbonyls from short-chainpolyols on Escherichia coli J Mol Microbiol Biotechnol 546ndash56
Lane DJ (1991) 16S23S rRNA sequencing In Nucleic AcidTechniques in Bacterial Systematics Stackebrandt E andGoodfellow M (eds) Chichester UK Wiley and Sonspp 115ndash175
Liles MR Manske BF Bintrim SB Handelsman J andGoodman RM (2003) A census of rRNA genes and linkedgenomic sequences within a soil metagenomic libraryAppl Environ Microbiol 69 2684ndash2691
Ludwig W Strunk O Westram R Richter L Meier HYadhukumar et al (2004) ARB a software environmentfor sequence data Nucleic Acids Res 32 1363ndash1371
Manz W Amann R Ludwig W Wagner M and Schlei-fer KH (1992) Phylogenetic oligodeoxynucleotide probesfor the major subclasses of Proteobacteria ndash problems andsolutions Syst Appl Microbiol 15 593ndash600
Manz W Amann R Ludwig W Vancanneyt M andSchleifer KH (1996) Application of a suite of 16S rRNA-specific oligonucleotide probes designed to investigatebacteria of the phylum CytophagandashFlavobacterndashBacteroi-des in the natural environment Microbiology 142 1097ndash1106
Massana R Taylor LJ Murray AE Wu KY JeffreyWH and DeLong EF (1998) Vertical distribution andtemporal variation of marine planktonic archaea in theGerlache Strait Antarctica during early spring LimnolOceanogr 43 607ndash617
Massana R DeLong EF and Pedros-Alio C (2000) A fewcosmopolitan phylotypes dominate planktonic archaealassemblages in widely different oceanic provinces ApplEnviron Microbiol 66 1777ndash1787
Morris RM Rappeacute MS Connon SA Vergin KL Sie-bold WA Carlson CA and Giovannoni SJ (2002)SAR11 clade dominates ocean surface bacterioplanktoncommunities Nature 420 806ndash810
Muyzer G Teske A Wirsen CO and Jannasch HW(1995) Phylogenetic-relationships of Thiomicrospira spe-cies and their identification in deep-sea hydrothermal ventsamples by denaturing gradient gel-electrophoresis of 16SrDNA fragments Arch Microbiol 164 165ndash172
Neef A (1997) Anwendung der in situ-Einzelzell-
Metagenomic analysis of freshwater bacteria 1895
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Identifizierung von Bakterien zur Populationsanalyse inKomplexen Mikrobiellen Biozonosen Munich GermanyTechnische Universitat Munchen
Opsahl S Benner R and Amon RMW (1999) Major fluxof terrigenous dissolved organic matter through the ArcticOcean Limnol Oceanogr 44 2017ndash2023
OrsquoSullivan LA Weightman AJ and Fry JC (2002) Newdegenerate Cytophaga-Flexibacter-Bacteroides-specific16S ribosomal DNA-targeted oligonucleotide probes revealhigh bacterial diversity in River Taff epilithon Appl EnvironMicrobiol 68 201ndash210
Pettit AN (1982) Cramer-von Misess statistic In Encyclo-pedia of Statistical Sciences Kotz S Johnson NL andRead CB (eds) New York USA Wiley pp 220ndash221
Quaiser A Ochsenreiter T Klenk HP Kletzin ATreusch AH Meurer G et al (2002) First insight intothe genome of an uncultivated crenarchaeote from soilEnviron Microbiol 4 603ndash611
Rappeacute MS and Giovannoni SJ (2003) The unculturedmicrobial majority Annu Rev Microbiol 57 369ndash394
Rappeacute MS Gordon DA Vergin KL and GiovannoniSJ (1999) Phylogeny of Actinobacteria small subunit(SSU) rRNA gene clones recovered from marine bacteri-oplankton Syst Appl Microbiol 22 106ndash112
Rappeacute MS Connon SA Vergin KL and GiovannoniSJ (2002) Cultivation of the ubiquitous SAR11 marinebacterioplankton clade Nature 418 630ndash633
Roller C Wagner M Amann R Ludwig W and SchleiferKH (1994) In-situ probing of Gram-positive bacteria withhigh DNA G+C content using 23S-ribosomal-RNA-targetedoligonucleotides Microbiology 140 2849ndash2858
Rondon MR August PR Bettermann AD Brady SFGrossman TH Liles MR et al (2000) Cloning the soilmetagenome a strategy for accessing the genetic andfunctional diversity of uncultured microorganisms ApplEnviron Microbiol 66 2541ndash2547
Sabehi G Massana R Bielawski JP Rosenberg MDelong EF and Beja O (2003) Novel proteorhodopsinvariants from the Mediterranean and Red Seas EnvironMicrobiol 5 842ndash849
Singleton DR Furlong MA Rathbun SL and WhitmanWB (2001) Quantitative comparisons of 16S rRNA genesequence libraries from environmental samples Appl Envi-ron Microbiol 67 4374ndash4376
Stackebrandt E and Goebel BM (1994) A place for DNAndashDNA reassociation and 16S ribosomal-RNA sequence-analysis in the present species definition in bacteriologyInt J Syst Bacteriol 44 846ndash849
Stein JL Marsh TL Wu KY Shizuya H and DeLongEF (1996) Characterization of uncultivated prokaryotesIsolation and analysis of a 40-kilobase-pair genome frag-ment from a planktonic marine archaeon J Bacteriol 178591ndash599
Thompson JR Marcelino LA and Polz MF (2002) Het-eroduplexes in mixed-template amplifications formationconsequence and elimination by lsquoreconditioning PCRrsquoNucleic Acids Res 30 2083ndash2088
Topp E (2003) Bacteria in agricultural soils Diversity roleand future perspectives Can J Soil Sci 83 303ndash309
Weller R Gloumlckner FO and Amann R (2000) 16S rRNA-targeted oligonucleotide probes for the in situ detectionof members of the phylum Cytophaga-Flavobacterium-Bacteroides Syst Appl Microbiol 23 107ndash114
von Wintzingerode F Gobel UB and Stackebrandt E(1997) Determination of microbial diversity in environmen-tal samples pitfalls of PCR-based rRNA analysis FEMSMicrobiol Rev 21 213ndash229
Zaitlin B Watson SB Dixon J and Steel D (2003)Actinomycetes in the Elbow River Basin Alberta CanWater Quality Res J Can 38 115ndash125
Zwart G Hiorns WD Methe BA Van Agterveld MPHuismans R Nold SC et al (1998) Nearly identical 16SrRNA sequences recovered from lakes in North Americaand Europe indicate the existence of clades of globallydistributed freshwater bacteria Syst Appl Microbiol 21546ndash556
Zwart G Crump BC Agterveld M Hagen F and HanSK (2002) Typical freshwater bacteria an analysis ofavailable 16S rRNA gene sequences from plankton oflakes and rivers Aquat Microb Ecol 28 141ndash155
Metagenomic analysis of freshwater bacteria
1885
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd
Environmental Microbiology
7
1883ndash1895
comprised the largest fractions of the libraries while alpha-
proteobacteria
gamma-
proteobacteria
Firmicutes
Verru-comicrobiales
and
Spirochaetaceae
were minor mem-bers However closer examination of the relativeabundance of different bacterial groups revealed notewor-thy differences between the metagenomic and PCRlibraries
Cytophaga-
like bacteria in clone libraries
The metagenomic and PCR libraries differed in the num-ber and kinds of
Cytophaga
-like bacteria One obviousdifference was the percentage of
Cytophaga
-like clonesOnly 13 of the clones in the PCR library belonged to the
Cytophaga
-like group compared with 54 for themetagenomic library (Table 1) In addition the variety of
Cytophaga
-like bacteria was greater in the metagenomiclibrary than in the PCR library More
Cytophaga
-like clus-ters contained fosmid sequences than PCR sequences inthe phylogenetic tree of
Cytophaga
-like bacteria (Fig 1)The metagenomic library but not the PCR library con-tained
Cytophaga
-like clones in the
Chitinophaga
and
Myroides
groups (Fig 1) One
Cytophaga
-like clone in thePCR library (Sta2ndash97) was not similar to any of themetagenomic clones
Seventy percent of the
Cytophaga
-like bacteria wereassociated with the
Flavobacteriales
group (Fig 1) andthese clones could be grouped into three clusters (AndashC)A fourth cluster (D) was in the
Flectobacillus
group(Fig 1) Clusters A and B were related to different kindsof
Flavobacterium
spp and included some cultured bac-teria Cluster A included an isolate from the Elbe Riverand cluster B included
Flavobacterium xinjiangensi
Theaverage sequence identity in these two clusters with cul-tured members was 966
plusmn
34 and 945
plusmn
61respectively Clusters C and D were associated with the
Myroides
and
Flectobacillus
groups respectively andincluded bacteria with a higher degree of sequence simi-larity (985
plusmn
05 and 995
plusmn
04 respectively) butthese clusters had no cultured members
Some of the
Cytophaga
-like bacteria were representedin both libraries Half of the
Cytophaga
-like clones in thePCR library were highly similar to clones in the metage-nomic library and every
Cytophaga
-like group with clonesfrom the PCR library also had representatives from themetagenomic library For example cluster D was com-prised of three PCR clones and four metagenomic clones(Fig 2) Cluster A included three metagenomic clonesand a highly similar clone from the PCR library Howeversimilarity between
Cytophaga
-like bacteria in the twolibraries was restricted to these two clusters
Beta-
proteobacteria
in clone libraries
The metage-nomic and PCR libraries also differed in the number andkinds of beta-
proteobacteria The PCR library had more
than twice as many clones in this group than the metage-nomic library (50 versus 17 respectively) (Table 1)The beta-proteobacteria clones in the two librariesbelonged to five clusters (AndashE) Clusters D and E werecomprised exclusively of clones from the PCR library(Fig 2) The degree of relatedness in the clusters madeup of only PCR clones and in cluster A was low rangingfrom 89 to 95 sequence identity In contrast some ofthe beta-proteobacteria sampled by the two libraries weresimilar Clusters B and C which included clones from bothlibraries were comprised of very closely related bacteriawith similarities of 964 plusmn 16 and 975 plusmn 31respectively
Other bacteria in the clone libraries
In contrast to the differences outlined above for theCytophaga-like and beta-proteobacteria groups Actino-bacteria in the metagenomic and PCR libraries were quitesimilar The libraries had equivalent percentages of Acti-nobacteria (about 15) and there was considerable over-lap in the types of Actinobacteria in the two libraries Ofthe three clusters of Actinobacteria clones (AndashC) eachincluded clones from both libraries (Fig 3) Cluster Bincluded the most closely related clones with averagesimilarities of 964 plusmn 22 The Actinobacteria clones inclusters A and C were less closely related Sequencesimilarities in these clusters were 932 plusmn 41 and929 plusmn 54 respectively
Alpha-proteobacteria comprised a minor component ofthe libraries accounting for only 3 and 12 of theclones in the metagenomic and PCR clones respectively(Table 1) The alpha-proteobacteria clones in the twolibraries were closely related to each other and to somecultured alpha-proteobacteria (results not shown) Threeclones from the PCR library and two from the metage-nomic library belonged to a cluster that included culturedSphingomonas spp One alpha-proteobacteria clone fromthe PCR library belonged to the Rhodospirillales group
The libraries also included clones belonging to fivegroups of bacteria that can be present in aquatic sys-tems but are typically minor members of the communityThese groups included Firmicutes and Spirochaetaceaethat occurred in only the metagenomic library and epsi-lon-proteobacteria and Bdellovibrio that were foundexclusively in the PCR library (Table 1) Clones belong-ing to the Verrucomicrobiales group occurred in bothlibraries
Library coverage and diversity
Library coverage increased with the genetic distance asexpected At an evolutionary distance of 003 coveragesof the metagenomic and PCR libraries were 65 and 51
1886 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Fig 1 Phylogenetic relationships of selected Cytophaga-like bacteria as depicted in the ARB database tree constructed of sequences longer than 1000 bp (June 2002) Delaware River Cytophaga-like bacteria were added using the ARB parsimony quick add tool Clones in the library of 16S rDNA amplicons are labelled PCR and 16S rDNA-bearing metagenomic DNA clones are labelled Fosmid Values on the trapezoids indicate the number of sequences in the group Labels A B C and D refer to clusters of closely related bacteria that are discussed in the text The scale bar represents 01 substitutions per site
Metagenomic analysis of freshwater bacteria 1887
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Fig 2 Phylogenetic relationships of selected beta-proteobacteria as depicted in the ARB database tree constructed of sequences longer than 1000 bp (June 2002) Delaware River beta-proteobacteria were added using the ARB parsimony quick add tool Clones in the library of 16S rDNA amplicons are labelled PCR and 16S rDNA-bearing metagenomic DNA clones are labelled Fosmid Values on the trapezoids indicate the number of sequences in the group Labels A B C D and E refer to clusters of closely related bacteria that are discussed in the text The scale bar represents 01 substitutions per site
1888 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Fig 3 Phylogenetic relationships of selected Actinobacteria as depicted in the ARB database tree constructed of sequences longer than 1000 bp (June 2002) Delaware River Actinobacteria were added using the ARB parsimony quick add tool Clones in the library of 16S rDNA amplicons are labelled PCR and 16S rDNA-bearing metagenomic DNA clones are labelled Fosmid Values on the trapezoids indicate the number of sequences in the group Labels A B and C refer to clusters of closely related bacteria that are discussed in the text The scale bar represents 01 substitutions per site
Metagenomic analysis of freshwater bacteria 1889
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
respectively (Fig 4) This distance is approximately equalto 97 sequence identity which is one approximation ofspecies-level identity (Stackebrandt and Goebel 1994)Coverage of the metagenomic library was higher thancoverage of the PCR library at all levels of genetic dis-tance with the greatest differences occurring at smallgenetic distance Coverage decreased rapidly withdecreasing genetic distance and at a genetic distance of001 the coverage of the metagenomic library was 46while the coverage of the PCR library decreased evenmore to just 16
Indices of diversity were higher for the metagenomiclibrary than the PCR library (Table 2) The Simpsonindex was 134 for the metagenomic library comparedwith 87 for the PCR library The difference was not assubstantial for the ShannonndashWeaver index (Table 2)Similarly measures of evenness were higher for themetagenomic library than the PCR library Evenness cal-culated using the ShannonndashWeaver index was 073 and066 for the metagenomic library and PCR libraryrespectively while the Simpson index of evenness was026 and 017 for the metagenomic and PCR librariesrespectively
We examined the possibility that diversity based onPCR clone library composition is overestimated becauseof artefacts of the PCR Re-conditioned PCR was used totest for potential artefacts introduced when heterologousPCR products re-anneal and form heteroduplexes that arethen subjected to mismatch repair upon cloning intoE coli (Thompson et al 2002) Re-conditioning of thePCR amplicons reduced the number of RFLP typesexpected in a library of 80 clones from 50 in the standardPCR library to 40 in the library produced with re-condi-tioned PCR (Fig 5) There was no difference betweenrarefaction curves of replicate PCR libraries produced withstandard PCR (Fig 5)
Comparison of FISH and clone libraries
Fluorescence in situ hybridization was used to assess thecorrespondence between clone library composition andbacterial community structure Microscopic enumerationof bacteria hybridized with fluorescently labelled 16SrRNA directed probes revealed a community dominatedby beta-proteobacteria Actinobacteria and Cytophaga-like bacteria (Table 1) As expected for a freshwater sys-tem Actinobacteria and beta-proteobacteria were themost abundant groups making up about 25 of the com-munity Cytophaga-like bacteria were also prominent
Fig 4 Coverage of the metagenomic and 16S rDNA libraries calcu-lated using the LIBSHUFF tool
Fig 5 Rarefaction curves of 16S rDNA clone libraries constructed with and without re-conditioned PCR PCR 1 and PCR 2 are clone libraries constructed from replicate PCR reactions without re-condi-tioning The error bars are one standard deviation
Table 2 Diversity of 16S rRNA genes in metagenomic and PCR clone libraries from the Delaware River
Library
ShannonndashWeaver Simpson
CoverageDiversity Evenness Diversity Evenness
Metagenomic 29 073 134 026 065PCR 26 066 87 017 052
Diversity indices were calculated using equations taken from Dunbar and colleagues (1999)
1890 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
accounting for 17 plusmn 4 of the community In contrastalpha-proteobacteria and gamma-proteobacteria wereconsiderably less abundant (less than 10) Seventy-seven percent of all prokaryotes were identified with agroup probe and 69 plusmn 11 were detected with the gen-eral bacterial probe Eub338
Bacterial community structure determined by FISH andinferred from the PCR library was surprisingly similar Theabundance of Cytophaga-like bacteria in the PCR librarywas within 25 of the abundance determined by FISHthe abundance of gamma-proteobacteria differed by 3or less and Actinobacteria differed by 38 according tothe two analyses (Table 1) In contrast the abundance ofbeta-proteobacteria clones in the PCR library was twicethe abundance of bacteria detected with the beta-proteo-bacteria FISH probe (Table 1) Three types of bacteriaincluding epsilon-proteobacteria Bdellovibrio and bacte-ria belonging to the candidate division TM6 were presentin the PCR library and would not have been detected withany of the FISH probes used in this study Although wehave no estimate of the abundance of bacteria in thesegroups they likely accounted for some of the 23 ofbacteria not detected by FISH (Table 1)
Discussion
It is necessary to examine whether the phylogenic com-position of metagenomic libraries reflects that of the orig-inal microbial composition in order to effectively linkcommunity structure and function revealed by metage-nomic analysis If metagenomic libraries represent abiased sample of microbial diversity they will not yield anaccurate assessment of metabolic diversity Assumingthey are unbiased large insert metagenomic libraries maybe particularly powerful in examining the relationshipbetween community structure and metabolic capacitiesLarge inserts increase the likelihood of obtaining a clonecontaining both lsquofunctionalrsquo genes (coding for metabolicfunction) and genes with phylogenetic information eg16S rRNA genes Therefore it is necessary to know howthe diversity of 16S rRNA genes sampled by large insertmetagenomic libraries corresponds to actual communitystructure Our main finding was that the composition ofthe metagenomic library at a high phylogenetic level wasnot statistically different from the PCR library but closerexamination revealed noteworthy differences between thelibraries and the community structure determined byFISH The metagenomic library sampled a broader diver-sity of Cytophaga-like bacteria than the PCR library but itmissed some groups of potentially important beta-proteo-bacteria The metagenomic and PCR libraries sampledthe same groups of Actinobacteria but FISH analysissuggested that both types of libraries may have failed toadequately sample Actinobacteria
Metagenomic analysis of Delaware River bacteria pro-vided a view of Cytophaga-like bacterial diversity that wasnot possible with the PCR library and FISH analyses Thenumber of clones representing Cytophaga-like bacteria inthe metagenomic library was threefold higher than theabundance determined by FISH and fourfold higher thanin the PCR library It was not surprising that the metage-nomic library sampled a larger diversity of Cytophaga-likebacteria than the PCR library because previous studieshad suggested that Cytophaga-like bacteria are underrep-resented in 16S rDNA PCR libraries relative to the abun-dance of these bacteria detected by FISH (Cottrell andKirchman 2000 Kirchman et al 2003) It was also notsurprising that the metagenomic library had more clonesfrom Cytophaga-like bacteria than would have beenexpected based on the FISH counts because the generalFISH probe CF319a for Cytophaga-like bacteria does notrecognize all Cytophaga-like bacteria (Weller et al 2000)The abundance of Cytophaga-like bacteria according tothe general FISH probe CFB560 for Cytophaga-like bac-teria (OrsquoSullivan et al 2002) was only 15 higher thanthat determined with CF319a The denaturing gradientelectrophoresis (DGGE) PCR primers used in this studyappear to adequately sample Cytophaga-like bacteria atleast compared with FISH analysis (Castle and Kirchman2004) and these primers apparently recognize a broaderdiversity of Cytophaga-like bacteria than the EubA andEubB primers as some Cytophaga-like bacteria detectedby DGGE in the metagenomic library were not sampledby the PCR library
The Cytophaga-like bacterial group includes a highlydiverse collection of bacteria but not all of this diversitywas present in our Delaware River sample In fact mostof the Cytophaga-like bacteria were most closely relatedto just two genera of Flavobacteriales (Flavobacteriumand Myroides) and a cluster in the Flectobacillus groupThis latter group is distinct from the CytophagandashFlavobac-terium cluster which is synonymous with the Flavobacte-riales (Kirchman 2002) The Delaware River bacteria inthe Flectobacillus group belong to the proposedPRD01a001B cluster of Cytophaga-like bacteria identifiedby Zwart and colleagues (2002) Nine of the 12 DelawareRiver sequences in the Flectobacillus group were greaterthan 98 similar to the PRD01a001B clone from theParker River Massachusetts The proposedPRD01a001B cluster which is currently comprised solelyof uncultivated bacteria has also been found in Adiron-dack lakes Lake Baikal and other freshwater environ-ments (Zwart et al 2002) However this group ofCytophaga-like bacteria has not been detected in allfreshwater environments for example it was not found inthe Columbia River (Zwart et al 2002)
The Delaware River beta-proteobacteria belonged toclusters widely distributed in freshwater systems Three
Metagenomic analysis of freshwater bacteria 1891
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
groups of Delaware River beta-proteobacteria clones (BD and E) were highly similar to bacteria in the freshwaterbeta-Proteobacteria clusters proposed by Zwart and col-leagues (2002) Sequences in Delaware River beta-pro-teobacteria group B were 95ndash99 similar to theRodoferax sp in the Rodoferax sp BAL47 cluster GroupD of Delaware Bay beta-proteobacteria corresponded tothe Polynucleobacter necessarius cluster (88ndash99sequence similarity to P necessarius) and sequences ingroup E probably belong to the proposed LD28 clusterSequences in group D were 89ndash97 similarity to theLD28 clone from Lake Loosdrech in the Netherlands(Zwart et al 1998)
The clusters of freshwater beta-proteobacteria pro-posed by Zwart and colleagues (2002) were more preva-lent in the PCR library than in the metagenomic librarysuggesting that they may represent bacteria that are pref-erentially sampled by PCR libraries Delaware River beta-proteobacteria group B included clones from both themetagenomic library and the PCR library but DelawareRiver groups D and E were comprised of only clones fromthe PCR library Different scenarios could lead to theoverrepresentation of bacterial groups in PCR librariescompared with libraries constructed without a PCR stepincluding selective PCR amplification with general bacte-rial primers differences in rRNA gene copy number orvariation in DNA extraction efficiency (von Wintzingerodeet al 1997) Fluorescence in situ hybridization probes areneeded to determine the abundance of these beta-proteo-bacteria which appear to be in a variety of freshwaterenvironments
There was a high degree of similarity in the Actinobac-teria sampled by the metagenomic and PCR libraries Allthree groups of Actinobacteria identified in this studyincluded clones from the metagenomic and PCR librariesOur results indicate that the presumed thicker cell wallsof these G+ bacteria do not interfere with the extraction ofhigh molecular weight DNA from these bacteria Howeverthe overall efficiency of DNA extraction may be low asActinobacteria were underrepresented in the librariescompared with their abundance determined by FISHAlthough the libraries may be missing some Actinobacte-ria they did recover clusters that are apparentlywidespread in freshwater systems Delaware River Acti-nobacteria groups B and C which correspond to the ACK-MI and STA-30 clusters identified by Zwart and colleagues(2002) were identified in the metagenomic and PCRlibraries Clones in cluster B were 92ndash99 similar to theACK-M1 clone and clones in cluster C were 92ndash97similar to the STA-30 clone Delaware River Actinobacte-ria group A which included clones from both libraries didnot overlap with previously described clusters of freshwa-ter Actinobacteria None of the G+ bacterial groups wedetected represented a terrestrial bacterial signal
because of transport of soils into the river Zwart andcolleagues (2002) also did not find evidence for an impor-tant contribution of soil bacteria to bacterial communitiesin rivers
Coverage in the metagenomic library was consistentlyhigher than in the PCR library over a broad range ofevolutionary distances although substantially higher cov-erage by the metagenomic library was restricted to evo-lutionary distances less than 005 Differences incoverage were not the result of different sequencelengths we analysed 1400 bp for the PCR clones but onlythe highly variable V3 region of the 16S rRNA gene forthe metagenomic clones LIBSHUFF analysis conductedwith the V3 region alone also revealed higher coverage bythe metagenomic library (data not shown) Differences incoverage may reflect more comprehensive sampling ofbacterial diversity by the metagenomic library
Lower coverage by the PCR library was consistent withour hypothesis that PCR artefacts lead to overestimatesof diversity Polymerase chain reaction re-conditioning flat-tened the rarefaction curve but the effect was only 20not as pronounced as would be expected if heterodu-plexes were a substantial source of diversity in the PCRlibraries The higher diversity in the metagenomic librarycompared with the PCR library is also consistent with theidea that the PCR libraries do not overestimate bacterialdiversity In fact they probably underestimate diversitybecause they miss certain groups such as Cytophaga-like bacteria (Cottrell and Kirchman 2000) Additional fac-tors such as cloning biases variation in genome size(Fogel et al 1999) rRNA gene copy number (Klappen-bach et al 2000) and ribosome content (Fegatella et al1998) could have resulted in larger differences than weobserved between the two library approaches and FISHGiven the potential sources of differences the degree ofsimilarity between the different library and FISHapproaches was higher than expected
These results and other suggest that the identities ofprokaryotes in aquatic environments are no longer com-pletely hidden in the microbial lsquoblack boxrsquo With the aid ofvarious culture-independent approaches it is now clearthat various proteobacterial groups Cytophaga-like bac-teria and Actinobacteria are the dominant groups of bac-teria in surface waters of aquatic systems (Giovannoniand Rappeacute 2000 Rappeacute and Giovannoni 2003)Archaea make a substantial contribution to prokaryoticdiversity in the ocean depths well below the euphotic zone(Massana et al 1998 Karner et al 2001) Althoughpotentially these groups could contain many subgroupsat least in marine waters it appears that only a few cladesdominate including alpha-proteobacteria belonging to theSAR11 group (Morris et al 2002 Rappeacute et al 2002) andRoseobacter spp (Gonzalez and Moran 1997) Cytoph-aga-like bacteria in Delaware cluster 2 (Kirchman et al
1892 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
2003) the marine Actinobacteria group (Rappeacute et al1999) and Archaea in group I and group II (Fuhrman et al1992 Massana et al 2000 Karner et al 2001)
Dominant groups in freshwater are also becomingapparent Our study revealed several widespread clustersof freshwater bacteria in the Delaware River that wereidentified in the metagenomic and PCR libraries Theseclusters are obvious candidates for further investigationincluding additional metagenomic analysis to exploretheir metabolic potential It is unclear if clusters such asthe P necessarius and LD28 beta-proteobacteria clus-ters which occurred in the PCR library but not themetagenomic library are selectively amplified by the PCRor for some reason are missed by the metagenomic clon-ing Additional examination by FISH could help resolvethe numerical importance of these freshwater bacteriaStudies such as these will further advance our under-standing of aquatic bacterial diversity as it evolves from aview of apparently intractable diversity towards whatappears to be a manageable number of dominant bacte-rial groups
Experimental procedures
Sample collection
A water sample was collected from a depth of 1 m in theDelaware River near Trenton New Jersey (40infin77cent-N74infin493cent-W) in December 2001 and subsampled for the anal-yses performed in this study
Fluorescence in situ hybridization
The sample for FISH analysis was fixed with 2 paraformal-dehyde at 4infinC for 12ndash18 h Bacteria were then collected on02 mm polycarbonate filters rinsed twice with deionizedwater and the filters were stored at -20infinC Bacteria weredetected with FISH probe Eub338 (Amann et al 1990) Wealso tried a mixture of general probes for Bacteria (Daimset al 1999) and found that it detected 70 plusmn 5 of DAPIpositive bacteria compared with 69 plusmn 11 for Eub338 aloneThe FISH probes for alpha- beta- and gamma-proteobacte-ria were Alf968 (Neef 1997 Gloumlckner et al 1999) Bet42aand Gam42a (Manz et al 1992) respectively Cytophaga-likebacteria were assayed with probe CF319a (Manz et al1996) and CFB560 (OrsquoSullivan et al 2002) and Actinobac-teria were enumerated using probe HGC69a (Roller et al1994) Non-specific binding was assessed using a negativecontrol probe (Karner and Fuhrman 1997) Hybridization ofCy3-labelled oligonucleotide probes was performed with por-tions of polycarbonate filters incubated in a hybridizationbuffer containing formamide to control the stringency ofhybridization (Cottrell and Kirchman 2000) Unbound probewas removed by washing the sample with a wash buffer at48infinC The filter piece was then rinsed in water and 70ethanol The air-dried sample was mounted on a glass slidewith a coverslip using an antifade mountant containing2 mg ml-1 DAPI Probe-positive bacteria (500ndash1000 bacteria
for each FISH probe) were enumerated using semiautomatedfluorescence microscopy (Cottrell and Kirchman 2003)
Preparation of the bacterial size fraction
Environmental DNA was extracted from the bacterial sizefraction obtained by pumping river water (500 l) sequentiallythrough a 1 mm nominal pore-size polypropylene string-wound filter (Cole Parmer) and a 08 mm polycarbonate filter(Nuclepore) Bacteria were collected from the filtrate by tan-gential flow filtration using a 01 mm hollow fibre filter (AGTechnology) and an Amicon DC10 gear pump The samplewas concentrated to 2 l rinsed with a buffer (05 M NaCl01 M EDTA 10 mM Tris pH 80) and stored frozen
Library construction using a fosmid vector
Bacteria were collected from thawed bacterial concentrate bycentrifugation in an SS34 rotor at 16 000 rpm for 30 minThe bacteria were resuspended in 1 ml of the supernatantand then mixed with 1 Sea Plaque agarose in deionizedwater at 55infinC (Stein et al 1996) The molten mixture wasdrawn into a 1 ml syringe and allowed to solidify The samplewas equilibrated with buffer (10 mM Tris pH 8 50 mM NaCl01 M EDTA 1 Sarkosyl) and digested with 1 mg ml-1
lysozyme at 37infinC for 4 h Buffer and lysozyme were replacedwith 1 Sarkosyl in 05 M EDTA and the sample wasdigested with 1 mg ml-1 proteinase K at 55infinC for 18 h Fol-lowing the enzyme treatments the sample was electrophore-sed for 8 h at 23 V through 03 SeaKem Gold agarose inTAE buffer DNA entering the gel but larger than the 48 kbsize standard was recovered from the gel by electroelutionand concentrated by ethanol precipitation The DNA was thensheared to 40 kb by 150 passages through a 200 ml pipettetip The DNA was prepared for ligation using a blunt endrepair kit (Epicentre Technologies) following the manufac-turerrsquos protocol The insert DNA was then size selected byelectrophoresis on 1 Sea Plaque GTG agarose DNA wasrecovered from the gel following treatment with Gelaseenzyme (Epicentre Technologies) digestion and ethanol pre-cipitation The DNA was then ligated to the pCC1FOS fosmidvector (Epicentre Technologies) and packaged using phageprotein extract following the manufacturerrsquos recommendedprocedure EPI300 host E coli infected with recombinantphage were plated on LB media containing chloramphenicolClones were picked and sorted into 96-well microtitre plates
PCR library construction
Three PCR libraries were constructed with the DNA isolatedfrom the Delaware River Two replicate libraries were con-structed by cloning amplicons generated from separate PCRreactions using the general bacterial primers EubA and EubB(Lane 1991) A third library was constructed using re-condi-tioned amplicons from the PCR reaction used to generate thefirst library Re-conditioning was performed with three roundsof thermal cycling applied to amplicons diluted 10-fold infresh PCR reagents (Thompson et al 2002) The 25 ml PCRreactions included 05 ml 12 ml and 2 ml additions of 10 mMdeoxynucloside triphosphate 25 mM MgCl2 10 mM primer
Metagenomic analysis of freshwater bacteria 1893
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
stocks respectively Bovine serum albumin was added to afinal concentration of 02 mg ml-1 Two and a half units of Taqpolymerase (Promega) and a one-tenth volume of 10yen bufferwere added to each reaction The thermal cycling conditionsconsisted of a touchdown series in which the annealing tem-perature decreased from 65infinC to 55infinC by 1infinC per cyclefollowed by 15 cycles at 55infinC each for 1 min Each cycleincluded a 1 min denaturation step at 95infinC and an extensionstep of 25 min at 72infinC Amplicons were cloned with a TOPO-TA cloning kit with pCR21-TOPO vector (Invitrogen) accord-ing to the manufacturerrsquos instructions
Screening PCR and metagenomic libraries
Clones were screened for insert size using M13 forward andreverse PCR primers which flank the cloned insert andthose with full-length inserts were screened by restrictionfragment length polymorphism (RFLP) analysis by digestingthe M13 PCR amplicons with a mixture of HhaI and RsaI(New England Biolabs) Restriction patterns on 2 Meta-phore agarose (FMC) agarose gels were compared manuallyClones with identical restriction patterns were groupedtogether into RFLP types The 16S rRNA genes from onerepresentative of each RFLP type in the first library werecompletely sequenced
Pools of 96 fosmid clones were screened for 16S rRNAgenes by DGGE of PCR amplicons generated with primersGC358F and 517R (Muyzer et al 1995) Selected bandsresolved on an 8 polyacrylamide gel containing a gradientof 25ndash55 denaturant (138ndash22 formamide and 105ndash23urea) were re-amplified and sequenced Phylogenetic classi-fication was determined using BLAST and the ARB sequenceanalysis tool as described below Bands with equal electro-phoretic mobility were assigned the same phylogeneticclassification
Sequence and library analysis
Nucleotide sequences were analysed using BLAST [version21 National Center for Biotechnology Information (httpwwwncbinlmnihgovBLAST) and the ARB package (Lud-wig et al 2004)] Sequences were aligned with the ARB FastAligner version 103 adjusted manually with secondary-structure criteria Phylogenetic relationships were deter-mined using the Quick Add Parsimony tool and the1000_pub_may02 phylogenetic tree in the ARB database(June 2002 release)
Sequence similarities were determined using the tools inARB and rarefaction curves were calculated using AnalyticRarefaction 13 (httpwwwugaedustratasoftware) Librar-ies were compared using the LIBSHUFF analysis tool(httpwwwarchesugaedu~whitmanlibshuffhtml) (Single-ton et al 2001) Diversity indices were calculated usingequations taken from Dunbar and colleagues (1999)
Statistical analysis
Libraries were compared using LIBSHUFF analysis (Single-ton et al 2001) This approach compares libraries based ontheir coverage and the number of sequences in one library
that are not found in the second library This so-called lsquohet-erologous coveragersquo is calculated over a range of evolution-ary distances to obtain a coverage curve The differencebetween the lsquoheterologous coveragersquo curve and the coveragecurve of the first library is calculated using the Crameacuter-vonMises test statistic (Pettit 1982) and compared with the cov-erage curve calculated after shuffling sequences between thetwo libraries
Nucleotide sequence accession numbers
The nucleotide sequence data reported in this work aredeposited in GenBank under accession numbers AY562245ndashAY562366
Acknowledgements
This work was supported by the US Department of EnergyMicrobial Genomes Program and the National ScienceFoundation
References
Amann RI Binder BJ Olson RJ Chisholm SWDevereux R and Stahl DA (1990) Combination of 16Sribosomal-RNA-targeted oligonucleotide probes with flowcytometry for analyzing mixed microbial populations ApplEnviron Microbiol 56 1919ndash1925
Amon RMW and Benner R (1996) Photochemical andmicrobial consumption of dissolved organic carbon anddissolved oxygen in the Amazon River system GeochimCosmochim Ac 60 1783ndash1792
Beja O Aravind L Koonin EV Suzuki MT Hadd ANguyen LP et al (2000a) Bacterial rhodopsin evidencefor a new type of phototrophy in the sea Science 2891902ndash1906
Beja O Suzuki MT Koonin EV Aravind L Hadd ANguyen LP et al (2000b) Construction and analysis ofbacterial artificial chromosome libraries from a marinemicrobial assemblage Environ Microbiol 2 516ndash529
Beja O Spudich EN Spudich JL Leclerc M andDeLong EF (2001) Proteorhodopsin phototrophy in theocean Nature 411 786ndash789
Button DK and Robertson BR (2001) Determination ofDNA content of aquatic bacteria by flow cytometry ApplEnviron Microbiol 67 1636ndash1645
Button DK Robertson BR Lepp PW and SchmidtTM (1998) A small dilute-cytoplasm high-affinity novelbacterium isolated by extinction culture and having kineticconstants compatible with growth at ambient concentra-tions of dissolved nutrients in seawater Appl EnvironMicrobiol 64 4467ndash4476
Castle D and Kirchman DL (2004) Composition of estu-arine bacterial communities assessed by denaturing gradi-ent gel electrophoresis and fluorescence in situhybridization Limnol Oceanogr 2 303ndash314
Cottrell MT and Kirchman DL (2000) Community compo-sition of marine bacterioplankton determined by 16S rRNAgene clone libraries and fluorescence in situ hybridizationAppl Environ Microbiol 66 5116ndash5122
1894 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Cottrell MT and Kirchman DL (2003) Contribution ofmajor bacterial groups to bacterial biomass production(thymidine and leucine incorporation) in the Delaware estu-ary Limnol Oceanogr 48 168ndash178
Cottrell MT Moore JA and Kirchman DL (1999) Chiti-nases from uncultured marine microorganisms Appl Envi-ron Microbiol 65 2553ndash2557
Covert JS and Moran MA (2001) Molecular characteriza-tion of estuarine bacterial communities that use high- andlow-molecular weight fractions of dissolved organic carbonAquat Microb Ecol 25 127ndash139
Daims H Bruhl A Amann R Schleifer KH and WagnerM (1999) The domain-specific probe EUB338 is insuffi-cient for the detection of all Bacteria Development andevaluation of a more comprehensive probe set Syst ApplMicrobiol 22 434ndash444
Dunbar J Takala S Barns SM Davis JA and KuskeCR (1999) Levels of bacterial community diversity in fourarid soils compared by cultivation and 16S rRNA genecloning Appl Environ Microbiol 65 1662ndash1669
Fegatella F Lim J Kjelleberg S and Cavicchioli R(1998) Implications of rRNA operon copy number and ribo-some content in the marine oligotrophic ultramicrobacte-rium Sphingomonas sp strain RB2256 Appl EnvironMicrobiol 64 4433ndash4438
Fogel GB Collins CR Li J and Brunk CF (1999)Prokaryotic genome size and SSU rDNA copy numberEstimation of microbial relative abundance from a mixedpopulation Microb Ecol 38 93ndash113
Fuhrman JA McCallum K and Davis AA (1992) Novelmajor Archaebacterial group from marine plankton Nature356 148ndash149
Gillespie DE Brady SF Bettermann AD CianciottoNP Liles MR Rondon MR et al (2002) Isolation ofantibiotics turbomycin A and B from a metagenomic libraryof soil microbial DNA Appl Environ Microbiol 68 4301ndash4306
Giovannoni SJ and Rappeacute MS (2000) Evolution diversityand molecular ecology of marine prokaryotes In MicrobialEcology of the Oceans Kirchman DL (ed) New YorkUSA Wiley-Liss pp 47ndash84
Gloumlckner FO Fuchs BM and Amann R (1999) Bacteri-oplankton compositions of lakes and oceans a first com-parison based on fluorescence in situ hybridization ApplEnviron Microbiol 65 3721ndash3726
Gloumlckner FO Zaichikov E Belkova N Denissova LPernthaler J Pernthaler A and Amann R (2000) Com-parative 16S rRNA analysis of lake bacterioplanktonreveals globally distributed phylogenetic clusters includingan abundant group of actinobacteria Appl Environ Micro-biol 66 5053ndash5065
Gonzalez JM and Moran MA (1997) Numerical domi-nance of a group of marine bacteria in the alpha- subclassof the class Proteobacteria in coastal seawater Appl Envi-ron Microbiol 63 4237ndash4242
Hernes PJ and Benner R (2002) Transport and diagene-sis of dissolved and particulate terrigenous organic matterin the North Pacific Ocean Deep-Sea Res Part II ndash TopStud Oceanogr 49 2119ndash2132
Karner M and Fuhrman JA (1997) Determination of activemarine bacterioplankton a comparison of universal 16S
rRNA probes autoradiography and nucleoid staining ApplEnviron Microbiol 63 1208ndash1213
Karner MB DeLong EF and Karl DM (2001) Archaealdominance in the mesopelagic zone of the Pacific OceanNature 409 507ndash510
Kirchman DL (2002) The ecology of CytophagandashFlavobac-teria in aquatic environments FEMS Microbiol Ecol 3991ndash100
Kirchman DL Yu LY and Cottrell MT (2003) Diversityand abundance of uncultured Cytophaga-like bacteria inthe Delaware Estuary Appl Environ Microbiol 69 6587ndash6596
Klappenbach JA Dunbar JM and Schmidt TM (2000)rRNA operon copy number reflects ecological strategies ofbacteria Appl Environ Microbiol 66 1328ndash1333
Knietsch A Waschkowitz T Bowien S Henne A andDaniel R (2003) Metagenomes of complex microbial con-sortia derived from different soils as sources for novelgenes conferring formation of carbonyls from short-chainpolyols on Escherichia coli J Mol Microbiol Biotechnol 546ndash56
Lane DJ (1991) 16S23S rRNA sequencing In Nucleic AcidTechniques in Bacterial Systematics Stackebrandt E andGoodfellow M (eds) Chichester UK Wiley and Sonspp 115ndash175
Liles MR Manske BF Bintrim SB Handelsman J andGoodman RM (2003) A census of rRNA genes and linkedgenomic sequences within a soil metagenomic libraryAppl Environ Microbiol 69 2684ndash2691
Ludwig W Strunk O Westram R Richter L Meier HYadhukumar et al (2004) ARB a software environmentfor sequence data Nucleic Acids Res 32 1363ndash1371
Manz W Amann R Ludwig W Wagner M and Schlei-fer KH (1992) Phylogenetic oligodeoxynucleotide probesfor the major subclasses of Proteobacteria ndash problems andsolutions Syst Appl Microbiol 15 593ndash600
Manz W Amann R Ludwig W Vancanneyt M andSchleifer KH (1996) Application of a suite of 16S rRNA-specific oligonucleotide probes designed to investigatebacteria of the phylum CytophagandashFlavobacterndashBacteroi-des in the natural environment Microbiology 142 1097ndash1106
Massana R Taylor LJ Murray AE Wu KY JeffreyWH and DeLong EF (1998) Vertical distribution andtemporal variation of marine planktonic archaea in theGerlache Strait Antarctica during early spring LimnolOceanogr 43 607ndash617
Massana R DeLong EF and Pedros-Alio C (2000) A fewcosmopolitan phylotypes dominate planktonic archaealassemblages in widely different oceanic provinces ApplEnviron Microbiol 66 1777ndash1787
Morris RM Rappeacute MS Connon SA Vergin KL Sie-bold WA Carlson CA and Giovannoni SJ (2002)SAR11 clade dominates ocean surface bacterioplanktoncommunities Nature 420 806ndash810
Muyzer G Teske A Wirsen CO and Jannasch HW(1995) Phylogenetic-relationships of Thiomicrospira spe-cies and their identification in deep-sea hydrothermal ventsamples by denaturing gradient gel-electrophoresis of 16SrDNA fragments Arch Microbiol 164 165ndash172
Neef A (1997) Anwendung der in situ-Einzelzell-
Metagenomic analysis of freshwater bacteria 1895
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Identifizierung von Bakterien zur Populationsanalyse inKomplexen Mikrobiellen Biozonosen Munich GermanyTechnische Universitat Munchen
Opsahl S Benner R and Amon RMW (1999) Major fluxof terrigenous dissolved organic matter through the ArcticOcean Limnol Oceanogr 44 2017ndash2023
OrsquoSullivan LA Weightman AJ and Fry JC (2002) Newdegenerate Cytophaga-Flexibacter-Bacteroides-specific16S ribosomal DNA-targeted oligonucleotide probes revealhigh bacterial diversity in River Taff epilithon Appl EnvironMicrobiol 68 201ndash210
Pettit AN (1982) Cramer-von Misess statistic In Encyclo-pedia of Statistical Sciences Kotz S Johnson NL andRead CB (eds) New York USA Wiley pp 220ndash221
Quaiser A Ochsenreiter T Klenk HP Kletzin ATreusch AH Meurer G et al (2002) First insight intothe genome of an uncultivated crenarchaeote from soilEnviron Microbiol 4 603ndash611
Rappeacute MS and Giovannoni SJ (2003) The unculturedmicrobial majority Annu Rev Microbiol 57 369ndash394
Rappeacute MS Gordon DA Vergin KL and GiovannoniSJ (1999) Phylogeny of Actinobacteria small subunit(SSU) rRNA gene clones recovered from marine bacteri-oplankton Syst Appl Microbiol 22 106ndash112
Rappeacute MS Connon SA Vergin KL and GiovannoniSJ (2002) Cultivation of the ubiquitous SAR11 marinebacterioplankton clade Nature 418 630ndash633
Roller C Wagner M Amann R Ludwig W and SchleiferKH (1994) In-situ probing of Gram-positive bacteria withhigh DNA G+C content using 23S-ribosomal-RNA-targetedoligonucleotides Microbiology 140 2849ndash2858
Rondon MR August PR Bettermann AD Brady SFGrossman TH Liles MR et al (2000) Cloning the soilmetagenome a strategy for accessing the genetic andfunctional diversity of uncultured microorganisms ApplEnviron Microbiol 66 2541ndash2547
Sabehi G Massana R Bielawski JP Rosenberg MDelong EF and Beja O (2003) Novel proteorhodopsinvariants from the Mediterranean and Red Seas EnvironMicrobiol 5 842ndash849
Singleton DR Furlong MA Rathbun SL and WhitmanWB (2001) Quantitative comparisons of 16S rRNA genesequence libraries from environmental samples Appl Envi-ron Microbiol 67 4374ndash4376
Stackebrandt E and Goebel BM (1994) A place for DNAndashDNA reassociation and 16S ribosomal-RNA sequence-analysis in the present species definition in bacteriologyInt J Syst Bacteriol 44 846ndash849
Stein JL Marsh TL Wu KY Shizuya H and DeLongEF (1996) Characterization of uncultivated prokaryotesIsolation and analysis of a 40-kilobase-pair genome frag-ment from a planktonic marine archaeon J Bacteriol 178591ndash599
Thompson JR Marcelino LA and Polz MF (2002) Het-eroduplexes in mixed-template amplifications formationconsequence and elimination by lsquoreconditioning PCRrsquoNucleic Acids Res 30 2083ndash2088
Topp E (2003) Bacteria in agricultural soils Diversity roleand future perspectives Can J Soil Sci 83 303ndash309
Weller R Gloumlckner FO and Amann R (2000) 16S rRNA-targeted oligonucleotide probes for the in situ detectionof members of the phylum Cytophaga-Flavobacterium-Bacteroides Syst Appl Microbiol 23 107ndash114
von Wintzingerode F Gobel UB and Stackebrandt E(1997) Determination of microbial diversity in environmen-tal samples pitfalls of PCR-based rRNA analysis FEMSMicrobiol Rev 21 213ndash229
Zaitlin B Watson SB Dixon J and Steel D (2003)Actinomycetes in the Elbow River Basin Alberta CanWater Quality Res J Can 38 115ndash125
Zwart G Hiorns WD Methe BA Van Agterveld MPHuismans R Nold SC et al (1998) Nearly identical 16SrRNA sequences recovered from lakes in North Americaand Europe indicate the existence of clades of globallydistributed freshwater bacteria Syst Appl Microbiol 21546ndash556
Zwart G Crump BC Agterveld M Hagen F and HanSK (2002) Typical freshwater bacteria an analysis ofavailable 16S rRNA gene sequences from plankton oflakes and rivers Aquat Microb Ecol 28 141ndash155
1886 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Fig 1 Phylogenetic relationships of selected Cytophaga-like bacteria as depicted in the ARB database tree constructed of sequences longer than 1000 bp (June 2002) Delaware River Cytophaga-like bacteria were added using the ARB parsimony quick add tool Clones in the library of 16S rDNA amplicons are labelled PCR and 16S rDNA-bearing metagenomic DNA clones are labelled Fosmid Values on the trapezoids indicate the number of sequences in the group Labels A B C and D refer to clusters of closely related bacteria that are discussed in the text The scale bar represents 01 substitutions per site
Metagenomic analysis of freshwater bacteria 1887
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Fig 2 Phylogenetic relationships of selected beta-proteobacteria as depicted in the ARB database tree constructed of sequences longer than 1000 bp (June 2002) Delaware River beta-proteobacteria were added using the ARB parsimony quick add tool Clones in the library of 16S rDNA amplicons are labelled PCR and 16S rDNA-bearing metagenomic DNA clones are labelled Fosmid Values on the trapezoids indicate the number of sequences in the group Labels A B C D and E refer to clusters of closely related bacteria that are discussed in the text The scale bar represents 01 substitutions per site
1888 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Fig 3 Phylogenetic relationships of selected Actinobacteria as depicted in the ARB database tree constructed of sequences longer than 1000 bp (June 2002) Delaware River Actinobacteria were added using the ARB parsimony quick add tool Clones in the library of 16S rDNA amplicons are labelled PCR and 16S rDNA-bearing metagenomic DNA clones are labelled Fosmid Values on the trapezoids indicate the number of sequences in the group Labels A B and C refer to clusters of closely related bacteria that are discussed in the text The scale bar represents 01 substitutions per site
Metagenomic analysis of freshwater bacteria 1889
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
respectively (Fig 4) This distance is approximately equalto 97 sequence identity which is one approximation ofspecies-level identity (Stackebrandt and Goebel 1994)Coverage of the metagenomic library was higher thancoverage of the PCR library at all levels of genetic dis-tance with the greatest differences occurring at smallgenetic distance Coverage decreased rapidly withdecreasing genetic distance and at a genetic distance of001 the coverage of the metagenomic library was 46while the coverage of the PCR library decreased evenmore to just 16
Indices of diversity were higher for the metagenomiclibrary than the PCR library (Table 2) The Simpsonindex was 134 for the metagenomic library comparedwith 87 for the PCR library The difference was not assubstantial for the ShannonndashWeaver index (Table 2)Similarly measures of evenness were higher for themetagenomic library than the PCR library Evenness cal-culated using the ShannonndashWeaver index was 073 and066 for the metagenomic library and PCR libraryrespectively while the Simpson index of evenness was026 and 017 for the metagenomic and PCR librariesrespectively
We examined the possibility that diversity based onPCR clone library composition is overestimated becauseof artefacts of the PCR Re-conditioned PCR was used totest for potential artefacts introduced when heterologousPCR products re-anneal and form heteroduplexes that arethen subjected to mismatch repair upon cloning intoE coli (Thompson et al 2002) Re-conditioning of thePCR amplicons reduced the number of RFLP typesexpected in a library of 80 clones from 50 in the standardPCR library to 40 in the library produced with re-condi-tioned PCR (Fig 5) There was no difference betweenrarefaction curves of replicate PCR libraries produced withstandard PCR (Fig 5)
Comparison of FISH and clone libraries
Fluorescence in situ hybridization was used to assess thecorrespondence between clone library composition andbacterial community structure Microscopic enumerationof bacteria hybridized with fluorescently labelled 16SrRNA directed probes revealed a community dominatedby beta-proteobacteria Actinobacteria and Cytophaga-like bacteria (Table 1) As expected for a freshwater sys-tem Actinobacteria and beta-proteobacteria were themost abundant groups making up about 25 of the com-munity Cytophaga-like bacteria were also prominent
Fig 4 Coverage of the metagenomic and 16S rDNA libraries calcu-lated using the LIBSHUFF tool
Fig 5 Rarefaction curves of 16S rDNA clone libraries constructed with and without re-conditioned PCR PCR 1 and PCR 2 are clone libraries constructed from replicate PCR reactions without re-condi-tioning The error bars are one standard deviation
Table 2 Diversity of 16S rRNA genes in metagenomic and PCR clone libraries from the Delaware River
Library
ShannonndashWeaver Simpson
CoverageDiversity Evenness Diversity Evenness
Metagenomic 29 073 134 026 065PCR 26 066 87 017 052
Diversity indices were calculated using equations taken from Dunbar and colleagues (1999)
1890 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
accounting for 17 plusmn 4 of the community In contrastalpha-proteobacteria and gamma-proteobacteria wereconsiderably less abundant (less than 10) Seventy-seven percent of all prokaryotes were identified with agroup probe and 69 plusmn 11 were detected with the gen-eral bacterial probe Eub338
Bacterial community structure determined by FISH andinferred from the PCR library was surprisingly similar Theabundance of Cytophaga-like bacteria in the PCR librarywas within 25 of the abundance determined by FISHthe abundance of gamma-proteobacteria differed by 3or less and Actinobacteria differed by 38 according tothe two analyses (Table 1) In contrast the abundance ofbeta-proteobacteria clones in the PCR library was twicethe abundance of bacteria detected with the beta-proteo-bacteria FISH probe (Table 1) Three types of bacteriaincluding epsilon-proteobacteria Bdellovibrio and bacte-ria belonging to the candidate division TM6 were presentin the PCR library and would not have been detected withany of the FISH probes used in this study Although wehave no estimate of the abundance of bacteria in thesegroups they likely accounted for some of the 23 ofbacteria not detected by FISH (Table 1)
Discussion
It is necessary to examine whether the phylogenic com-position of metagenomic libraries reflects that of the orig-inal microbial composition in order to effectively linkcommunity structure and function revealed by metage-nomic analysis If metagenomic libraries represent abiased sample of microbial diversity they will not yield anaccurate assessment of metabolic diversity Assumingthey are unbiased large insert metagenomic libraries maybe particularly powerful in examining the relationshipbetween community structure and metabolic capacitiesLarge inserts increase the likelihood of obtaining a clonecontaining both lsquofunctionalrsquo genes (coding for metabolicfunction) and genes with phylogenetic information eg16S rRNA genes Therefore it is necessary to know howthe diversity of 16S rRNA genes sampled by large insertmetagenomic libraries corresponds to actual communitystructure Our main finding was that the composition ofthe metagenomic library at a high phylogenetic level wasnot statistically different from the PCR library but closerexamination revealed noteworthy differences between thelibraries and the community structure determined byFISH The metagenomic library sampled a broader diver-sity of Cytophaga-like bacteria than the PCR library but itmissed some groups of potentially important beta-proteo-bacteria The metagenomic and PCR libraries sampledthe same groups of Actinobacteria but FISH analysissuggested that both types of libraries may have failed toadequately sample Actinobacteria
Metagenomic analysis of Delaware River bacteria pro-vided a view of Cytophaga-like bacterial diversity that wasnot possible with the PCR library and FISH analyses Thenumber of clones representing Cytophaga-like bacteria inthe metagenomic library was threefold higher than theabundance determined by FISH and fourfold higher thanin the PCR library It was not surprising that the metage-nomic library sampled a larger diversity of Cytophaga-likebacteria than the PCR library because previous studieshad suggested that Cytophaga-like bacteria are underrep-resented in 16S rDNA PCR libraries relative to the abun-dance of these bacteria detected by FISH (Cottrell andKirchman 2000 Kirchman et al 2003) It was also notsurprising that the metagenomic library had more clonesfrom Cytophaga-like bacteria than would have beenexpected based on the FISH counts because the generalFISH probe CF319a for Cytophaga-like bacteria does notrecognize all Cytophaga-like bacteria (Weller et al 2000)The abundance of Cytophaga-like bacteria according tothe general FISH probe CFB560 for Cytophaga-like bac-teria (OrsquoSullivan et al 2002) was only 15 higher thanthat determined with CF319a The denaturing gradientelectrophoresis (DGGE) PCR primers used in this studyappear to adequately sample Cytophaga-like bacteria atleast compared with FISH analysis (Castle and Kirchman2004) and these primers apparently recognize a broaderdiversity of Cytophaga-like bacteria than the EubA andEubB primers as some Cytophaga-like bacteria detectedby DGGE in the metagenomic library were not sampledby the PCR library
The Cytophaga-like bacterial group includes a highlydiverse collection of bacteria but not all of this diversitywas present in our Delaware River sample In fact mostof the Cytophaga-like bacteria were most closely relatedto just two genera of Flavobacteriales (Flavobacteriumand Myroides) and a cluster in the Flectobacillus groupThis latter group is distinct from the CytophagandashFlavobac-terium cluster which is synonymous with the Flavobacte-riales (Kirchman 2002) The Delaware River bacteria inthe Flectobacillus group belong to the proposedPRD01a001B cluster of Cytophaga-like bacteria identifiedby Zwart and colleagues (2002) Nine of the 12 DelawareRiver sequences in the Flectobacillus group were greaterthan 98 similar to the PRD01a001B clone from theParker River Massachusetts The proposedPRD01a001B cluster which is currently comprised solelyof uncultivated bacteria has also been found in Adiron-dack lakes Lake Baikal and other freshwater environ-ments (Zwart et al 2002) However this group ofCytophaga-like bacteria has not been detected in allfreshwater environments for example it was not found inthe Columbia River (Zwart et al 2002)
The Delaware River beta-proteobacteria belonged toclusters widely distributed in freshwater systems Three
Metagenomic analysis of freshwater bacteria 1891
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
groups of Delaware River beta-proteobacteria clones (BD and E) were highly similar to bacteria in the freshwaterbeta-Proteobacteria clusters proposed by Zwart and col-leagues (2002) Sequences in Delaware River beta-pro-teobacteria group B were 95ndash99 similar to theRodoferax sp in the Rodoferax sp BAL47 cluster GroupD of Delaware Bay beta-proteobacteria corresponded tothe Polynucleobacter necessarius cluster (88ndash99sequence similarity to P necessarius) and sequences ingroup E probably belong to the proposed LD28 clusterSequences in group D were 89ndash97 similarity to theLD28 clone from Lake Loosdrech in the Netherlands(Zwart et al 1998)
The clusters of freshwater beta-proteobacteria pro-posed by Zwart and colleagues (2002) were more preva-lent in the PCR library than in the metagenomic librarysuggesting that they may represent bacteria that are pref-erentially sampled by PCR libraries Delaware River beta-proteobacteria group B included clones from both themetagenomic library and the PCR library but DelawareRiver groups D and E were comprised of only clones fromthe PCR library Different scenarios could lead to theoverrepresentation of bacterial groups in PCR librariescompared with libraries constructed without a PCR stepincluding selective PCR amplification with general bacte-rial primers differences in rRNA gene copy number orvariation in DNA extraction efficiency (von Wintzingerodeet al 1997) Fluorescence in situ hybridization probes areneeded to determine the abundance of these beta-proteo-bacteria which appear to be in a variety of freshwaterenvironments
There was a high degree of similarity in the Actinobac-teria sampled by the metagenomic and PCR libraries Allthree groups of Actinobacteria identified in this studyincluded clones from the metagenomic and PCR librariesOur results indicate that the presumed thicker cell wallsof these G+ bacteria do not interfere with the extraction ofhigh molecular weight DNA from these bacteria Howeverthe overall efficiency of DNA extraction may be low asActinobacteria were underrepresented in the librariescompared with their abundance determined by FISHAlthough the libraries may be missing some Actinobacte-ria they did recover clusters that are apparentlywidespread in freshwater systems Delaware River Acti-nobacteria groups B and C which correspond to the ACK-MI and STA-30 clusters identified by Zwart and colleagues(2002) were identified in the metagenomic and PCRlibraries Clones in cluster B were 92ndash99 similar to theACK-M1 clone and clones in cluster C were 92ndash97similar to the STA-30 clone Delaware River Actinobacte-ria group A which included clones from both libraries didnot overlap with previously described clusters of freshwa-ter Actinobacteria None of the G+ bacterial groups wedetected represented a terrestrial bacterial signal
because of transport of soils into the river Zwart andcolleagues (2002) also did not find evidence for an impor-tant contribution of soil bacteria to bacterial communitiesin rivers
Coverage in the metagenomic library was consistentlyhigher than in the PCR library over a broad range ofevolutionary distances although substantially higher cov-erage by the metagenomic library was restricted to evo-lutionary distances less than 005 Differences incoverage were not the result of different sequencelengths we analysed 1400 bp for the PCR clones but onlythe highly variable V3 region of the 16S rRNA gene forthe metagenomic clones LIBSHUFF analysis conductedwith the V3 region alone also revealed higher coverage bythe metagenomic library (data not shown) Differences incoverage may reflect more comprehensive sampling ofbacterial diversity by the metagenomic library
Lower coverage by the PCR library was consistent withour hypothesis that PCR artefacts lead to overestimatesof diversity Polymerase chain reaction re-conditioning flat-tened the rarefaction curve but the effect was only 20not as pronounced as would be expected if heterodu-plexes were a substantial source of diversity in the PCRlibraries The higher diversity in the metagenomic librarycompared with the PCR library is also consistent with theidea that the PCR libraries do not overestimate bacterialdiversity In fact they probably underestimate diversitybecause they miss certain groups such as Cytophaga-like bacteria (Cottrell and Kirchman 2000) Additional fac-tors such as cloning biases variation in genome size(Fogel et al 1999) rRNA gene copy number (Klappen-bach et al 2000) and ribosome content (Fegatella et al1998) could have resulted in larger differences than weobserved between the two library approaches and FISHGiven the potential sources of differences the degree ofsimilarity between the different library and FISHapproaches was higher than expected
These results and other suggest that the identities ofprokaryotes in aquatic environments are no longer com-pletely hidden in the microbial lsquoblack boxrsquo With the aid ofvarious culture-independent approaches it is now clearthat various proteobacterial groups Cytophaga-like bac-teria and Actinobacteria are the dominant groups of bac-teria in surface waters of aquatic systems (Giovannoniand Rappeacute 2000 Rappeacute and Giovannoni 2003)Archaea make a substantial contribution to prokaryoticdiversity in the ocean depths well below the euphotic zone(Massana et al 1998 Karner et al 2001) Althoughpotentially these groups could contain many subgroupsat least in marine waters it appears that only a few cladesdominate including alpha-proteobacteria belonging to theSAR11 group (Morris et al 2002 Rappeacute et al 2002) andRoseobacter spp (Gonzalez and Moran 1997) Cytoph-aga-like bacteria in Delaware cluster 2 (Kirchman et al
1892 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
2003) the marine Actinobacteria group (Rappeacute et al1999) and Archaea in group I and group II (Fuhrman et al1992 Massana et al 2000 Karner et al 2001)
Dominant groups in freshwater are also becomingapparent Our study revealed several widespread clustersof freshwater bacteria in the Delaware River that wereidentified in the metagenomic and PCR libraries Theseclusters are obvious candidates for further investigationincluding additional metagenomic analysis to exploretheir metabolic potential It is unclear if clusters such asthe P necessarius and LD28 beta-proteobacteria clus-ters which occurred in the PCR library but not themetagenomic library are selectively amplified by the PCRor for some reason are missed by the metagenomic clon-ing Additional examination by FISH could help resolvethe numerical importance of these freshwater bacteriaStudies such as these will further advance our under-standing of aquatic bacterial diversity as it evolves from aview of apparently intractable diversity towards whatappears to be a manageable number of dominant bacte-rial groups
Experimental procedures
Sample collection
A water sample was collected from a depth of 1 m in theDelaware River near Trenton New Jersey (40infin77cent-N74infin493cent-W) in December 2001 and subsampled for the anal-yses performed in this study
Fluorescence in situ hybridization
The sample for FISH analysis was fixed with 2 paraformal-dehyde at 4infinC for 12ndash18 h Bacteria were then collected on02 mm polycarbonate filters rinsed twice with deionizedwater and the filters were stored at -20infinC Bacteria weredetected with FISH probe Eub338 (Amann et al 1990) Wealso tried a mixture of general probes for Bacteria (Daimset al 1999) and found that it detected 70 plusmn 5 of DAPIpositive bacteria compared with 69 plusmn 11 for Eub338 aloneThe FISH probes for alpha- beta- and gamma-proteobacte-ria were Alf968 (Neef 1997 Gloumlckner et al 1999) Bet42aand Gam42a (Manz et al 1992) respectively Cytophaga-likebacteria were assayed with probe CF319a (Manz et al1996) and CFB560 (OrsquoSullivan et al 2002) and Actinobac-teria were enumerated using probe HGC69a (Roller et al1994) Non-specific binding was assessed using a negativecontrol probe (Karner and Fuhrman 1997) Hybridization ofCy3-labelled oligonucleotide probes was performed with por-tions of polycarbonate filters incubated in a hybridizationbuffer containing formamide to control the stringency ofhybridization (Cottrell and Kirchman 2000) Unbound probewas removed by washing the sample with a wash buffer at48infinC The filter piece was then rinsed in water and 70ethanol The air-dried sample was mounted on a glass slidewith a coverslip using an antifade mountant containing2 mg ml-1 DAPI Probe-positive bacteria (500ndash1000 bacteria
for each FISH probe) were enumerated using semiautomatedfluorescence microscopy (Cottrell and Kirchman 2003)
Preparation of the bacterial size fraction
Environmental DNA was extracted from the bacterial sizefraction obtained by pumping river water (500 l) sequentiallythrough a 1 mm nominal pore-size polypropylene string-wound filter (Cole Parmer) and a 08 mm polycarbonate filter(Nuclepore) Bacteria were collected from the filtrate by tan-gential flow filtration using a 01 mm hollow fibre filter (AGTechnology) and an Amicon DC10 gear pump The samplewas concentrated to 2 l rinsed with a buffer (05 M NaCl01 M EDTA 10 mM Tris pH 80) and stored frozen
Library construction using a fosmid vector
Bacteria were collected from thawed bacterial concentrate bycentrifugation in an SS34 rotor at 16 000 rpm for 30 minThe bacteria were resuspended in 1 ml of the supernatantand then mixed with 1 Sea Plaque agarose in deionizedwater at 55infinC (Stein et al 1996) The molten mixture wasdrawn into a 1 ml syringe and allowed to solidify The samplewas equilibrated with buffer (10 mM Tris pH 8 50 mM NaCl01 M EDTA 1 Sarkosyl) and digested with 1 mg ml-1
lysozyme at 37infinC for 4 h Buffer and lysozyme were replacedwith 1 Sarkosyl in 05 M EDTA and the sample wasdigested with 1 mg ml-1 proteinase K at 55infinC for 18 h Fol-lowing the enzyme treatments the sample was electrophore-sed for 8 h at 23 V through 03 SeaKem Gold agarose inTAE buffer DNA entering the gel but larger than the 48 kbsize standard was recovered from the gel by electroelutionand concentrated by ethanol precipitation The DNA was thensheared to 40 kb by 150 passages through a 200 ml pipettetip The DNA was prepared for ligation using a blunt endrepair kit (Epicentre Technologies) following the manufac-turerrsquos protocol The insert DNA was then size selected byelectrophoresis on 1 Sea Plaque GTG agarose DNA wasrecovered from the gel following treatment with Gelaseenzyme (Epicentre Technologies) digestion and ethanol pre-cipitation The DNA was then ligated to the pCC1FOS fosmidvector (Epicentre Technologies) and packaged using phageprotein extract following the manufacturerrsquos recommendedprocedure EPI300 host E coli infected with recombinantphage were plated on LB media containing chloramphenicolClones were picked and sorted into 96-well microtitre plates
PCR library construction
Three PCR libraries were constructed with the DNA isolatedfrom the Delaware River Two replicate libraries were con-structed by cloning amplicons generated from separate PCRreactions using the general bacterial primers EubA and EubB(Lane 1991) A third library was constructed using re-condi-tioned amplicons from the PCR reaction used to generate thefirst library Re-conditioning was performed with three roundsof thermal cycling applied to amplicons diluted 10-fold infresh PCR reagents (Thompson et al 2002) The 25 ml PCRreactions included 05 ml 12 ml and 2 ml additions of 10 mMdeoxynucloside triphosphate 25 mM MgCl2 10 mM primer
Metagenomic analysis of freshwater bacteria 1893
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
stocks respectively Bovine serum albumin was added to afinal concentration of 02 mg ml-1 Two and a half units of Taqpolymerase (Promega) and a one-tenth volume of 10yen bufferwere added to each reaction The thermal cycling conditionsconsisted of a touchdown series in which the annealing tem-perature decreased from 65infinC to 55infinC by 1infinC per cyclefollowed by 15 cycles at 55infinC each for 1 min Each cycleincluded a 1 min denaturation step at 95infinC and an extensionstep of 25 min at 72infinC Amplicons were cloned with a TOPO-TA cloning kit with pCR21-TOPO vector (Invitrogen) accord-ing to the manufacturerrsquos instructions
Screening PCR and metagenomic libraries
Clones were screened for insert size using M13 forward andreverse PCR primers which flank the cloned insert andthose with full-length inserts were screened by restrictionfragment length polymorphism (RFLP) analysis by digestingthe M13 PCR amplicons with a mixture of HhaI and RsaI(New England Biolabs) Restriction patterns on 2 Meta-phore agarose (FMC) agarose gels were compared manuallyClones with identical restriction patterns were groupedtogether into RFLP types The 16S rRNA genes from onerepresentative of each RFLP type in the first library werecompletely sequenced
Pools of 96 fosmid clones were screened for 16S rRNAgenes by DGGE of PCR amplicons generated with primersGC358F and 517R (Muyzer et al 1995) Selected bandsresolved on an 8 polyacrylamide gel containing a gradientof 25ndash55 denaturant (138ndash22 formamide and 105ndash23urea) were re-amplified and sequenced Phylogenetic classi-fication was determined using BLAST and the ARB sequenceanalysis tool as described below Bands with equal electro-phoretic mobility were assigned the same phylogeneticclassification
Sequence and library analysis
Nucleotide sequences were analysed using BLAST [version21 National Center for Biotechnology Information (httpwwwncbinlmnihgovBLAST) and the ARB package (Lud-wig et al 2004)] Sequences were aligned with the ARB FastAligner version 103 adjusted manually with secondary-structure criteria Phylogenetic relationships were deter-mined using the Quick Add Parsimony tool and the1000_pub_may02 phylogenetic tree in the ARB database(June 2002 release)
Sequence similarities were determined using the tools inARB and rarefaction curves were calculated using AnalyticRarefaction 13 (httpwwwugaedustratasoftware) Librar-ies were compared using the LIBSHUFF analysis tool(httpwwwarchesugaedu~whitmanlibshuffhtml) (Single-ton et al 2001) Diversity indices were calculated usingequations taken from Dunbar and colleagues (1999)
Statistical analysis
Libraries were compared using LIBSHUFF analysis (Single-ton et al 2001) This approach compares libraries based ontheir coverage and the number of sequences in one library
that are not found in the second library This so-called lsquohet-erologous coveragersquo is calculated over a range of evolution-ary distances to obtain a coverage curve The differencebetween the lsquoheterologous coveragersquo curve and the coveragecurve of the first library is calculated using the Crameacuter-vonMises test statistic (Pettit 1982) and compared with the cov-erage curve calculated after shuffling sequences between thetwo libraries
Nucleotide sequence accession numbers
The nucleotide sequence data reported in this work aredeposited in GenBank under accession numbers AY562245ndashAY562366
Acknowledgements
This work was supported by the US Department of EnergyMicrobial Genomes Program and the National ScienceFoundation
References
Amann RI Binder BJ Olson RJ Chisholm SWDevereux R and Stahl DA (1990) Combination of 16Sribosomal-RNA-targeted oligonucleotide probes with flowcytometry for analyzing mixed microbial populations ApplEnviron Microbiol 56 1919ndash1925
Amon RMW and Benner R (1996) Photochemical andmicrobial consumption of dissolved organic carbon anddissolved oxygen in the Amazon River system GeochimCosmochim Ac 60 1783ndash1792
Beja O Aravind L Koonin EV Suzuki MT Hadd ANguyen LP et al (2000a) Bacterial rhodopsin evidencefor a new type of phototrophy in the sea Science 2891902ndash1906
Beja O Suzuki MT Koonin EV Aravind L Hadd ANguyen LP et al (2000b) Construction and analysis ofbacterial artificial chromosome libraries from a marinemicrobial assemblage Environ Microbiol 2 516ndash529
Beja O Spudich EN Spudich JL Leclerc M andDeLong EF (2001) Proteorhodopsin phototrophy in theocean Nature 411 786ndash789
Button DK and Robertson BR (2001) Determination ofDNA content of aquatic bacteria by flow cytometry ApplEnviron Microbiol 67 1636ndash1645
Button DK Robertson BR Lepp PW and SchmidtTM (1998) A small dilute-cytoplasm high-affinity novelbacterium isolated by extinction culture and having kineticconstants compatible with growth at ambient concentra-tions of dissolved nutrients in seawater Appl EnvironMicrobiol 64 4467ndash4476
Castle D and Kirchman DL (2004) Composition of estu-arine bacterial communities assessed by denaturing gradi-ent gel electrophoresis and fluorescence in situhybridization Limnol Oceanogr 2 303ndash314
Cottrell MT and Kirchman DL (2000) Community compo-sition of marine bacterioplankton determined by 16S rRNAgene clone libraries and fluorescence in situ hybridizationAppl Environ Microbiol 66 5116ndash5122
1894 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Cottrell MT and Kirchman DL (2003) Contribution ofmajor bacterial groups to bacterial biomass production(thymidine and leucine incorporation) in the Delaware estu-ary Limnol Oceanogr 48 168ndash178
Cottrell MT Moore JA and Kirchman DL (1999) Chiti-nases from uncultured marine microorganisms Appl Envi-ron Microbiol 65 2553ndash2557
Covert JS and Moran MA (2001) Molecular characteriza-tion of estuarine bacterial communities that use high- andlow-molecular weight fractions of dissolved organic carbonAquat Microb Ecol 25 127ndash139
Daims H Bruhl A Amann R Schleifer KH and WagnerM (1999) The domain-specific probe EUB338 is insuffi-cient for the detection of all Bacteria Development andevaluation of a more comprehensive probe set Syst ApplMicrobiol 22 434ndash444
Dunbar J Takala S Barns SM Davis JA and KuskeCR (1999) Levels of bacterial community diversity in fourarid soils compared by cultivation and 16S rRNA genecloning Appl Environ Microbiol 65 1662ndash1669
Fegatella F Lim J Kjelleberg S and Cavicchioli R(1998) Implications of rRNA operon copy number and ribo-some content in the marine oligotrophic ultramicrobacte-rium Sphingomonas sp strain RB2256 Appl EnvironMicrobiol 64 4433ndash4438
Fogel GB Collins CR Li J and Brunk CF (1999)Prokaryotic genome size and SSU rDNA copy numberEstimation of microbial relative abundance from a mixedpopulation Microb Ecol 38 93ndash113
Fuhrman JA McCallum K and Davis AA (1992) Novelmajor Archaebacterial group from marine plankton Nature356 148ndash149
Gillespie DE Brady SF Bettermann AD CianciottoNP Liles MR Rondon MR et al (2002) Isolation ofantibiotics turbomycin A and B from a metagenomic libraryof soil microbial DNA Appl Environ Microbiol 68 4301ndash4306
Giovannoni SJ and Rappeacute MS (2000) Evolution diversityand molecular ecology of marine prokaryotes In MicrobialEcology of the Oceans Kirchman DL (ed) New YorkUSA Wiley-Liss pp 47ndash84
Gloumlckner FO Fuchs BM and Amann R (1999) Bacteri-oplankton compositions of lakes and oceans a first com-parison based on fluorescence in situ hybridization ApplEnviron Microbiol 65 3721ndash3726
Gloumlckner FO Zaichikov E Belkova N Denissova LPernthaler J Pernthaler A and Amann R (2000) Com-parative 16S rRNA analysis of lake bacterioplanktonreveals globally distributed phylogenetic clusters includingan abundant group of actinobacteria Appl Environ Micro-biol 66 5053ndash5065
Gonzalez JM and Moran MA (1997) Numerical domi-nance of a group of marine bacteria in the alpha- subclassof the class Proteobacteria in coastal seawater Appl Envi-ron Microbiol 63 4237ndash4242
Hernes PJ and Benner R (2002) Transport and diagene-sis of dissolved and particulate terrigenous organic matterin the North Pacific Ocean Deep-Sea Res Part II ndash TopStud Oceanogr 49 2119ndash2132
Karner M and Fuhrman JA (1997) Determination of activemarine bacterioplankton a comparison of universal 16S
rRNA probes autoradiography and nucleoid staining ApplEnviron Microbiol 63 1208ndash1213
Karner MB DeLong EF and Karl DM (2001) Archaealdominance in the mesopelagic zone of the Pacific OceanNature 409 507ndash510
Kirchman DL (2002) The ecology of CytophagandashFlavobac-teria in aquatic environments FEMS Microbiol Ecol 3991ndash100
Kirchman DL Yu LY and Cottrell MT (2003) Diversityand abundance of uncultured Cytophaga-like bacteria inthe Delaware Estuary Appl Environ Microbiol 69 6587ndash6596
Klappenbach JA Dunbar JM and Schmidt TM (2000)rRNA operon copy number reflects ecological strategies ofbacteria Appl Environ Microbiol 66 1328ndash1333
Knietsch A Waschkowitz T Bowien S Henne A andDaniel R (2003) Metagenomes of complex microbial con-sortia derived from different soils as sources for novelgenes conferring formation of carbonyls from short-chainpolyols on Escherichia coli J Mol Microbiol Biotechnol 546ndash56
Lane DJ (1991) 16S23S rRNA sequencing In Nucleic AcidTechniques in Bacterial Systematics Stackebrandt E andGoodfellow M (eds) Chichester UK Wiley and Sonspp 115ndash175
Liles MR Manske BF Bintrim SB Handelsman J andGoodman RM (2003) A census of rRNA genes and linkedgenomic sequences within a soil metagenomic libraryAppl Environ Microbiol 69 2684ndash2691
Ludwig W Strunk O Westram R Richter L Meier HYadhukumar et al (2004) ARB a software environmentfor sequence data Nucleic Acids Res 32 1363ndash1371
Manz W Amann R Ludwig W Wagner M and Schlei-fer KH (1992) Phylogenetic oligodeoxynucleotide probesfor the major subclasses of Proteobacteria ndash problems andsolutions Syst Appl Microbiol 15 593ndash600
Manz W Amann R Ludwig W Vancanneyt M andSchleifer KH (1996) Application of a suite of 16S rRNA-specific oligonucleotide probes designed to investigatebacteria of the phylum CytophagandashFlavobacterndashBacteroi-des in the natural environment Microbiology 142 1097ndash1106
Massana R Taylor LJ Murray AE Wu KY JeffreyWH and DeLong EF (1998) Vertical distribution andtemporal variation of marine planktonic archaea in theGerlache Strait Antarctica during early spring LimnolOceanogr 43 607ndash617
Massana R DeLong EF and Pedros-Alio C (2000) A fewcosmopolitan phylotypes dominate planktonic archaealassemblages in widely different oceanic provinces ApplEnviron Microbiol 66 1777ndash1787
Morris RM Rappeacute MS Connon SA Vergin KL Sie-bold WA Carlson CA and Giovannoni SJ (2002)SAR11 clade dominates ocean surface bacterioplanktoncommunities Nature 420 806ndash810
Muyzer G Teske A Wirsen CO and Jannasch HW(1995) Phylogenetic-relationships of Thiomicrospira spe-cies and their identification in deep-sea hydrothermal ventsamples by denaturing gradient gel-electrophoresis of 16SrDNA fragments Arch Microbiol 164 165ndash172
Neef A (1997) Anwendung der in situ-Einzelzell-
Metagenomic analysis of freshwater bacteria 1895
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Identifizierung von Bakterien zur Populationsanalyse inKomplexen Mikrobiellen Biozonosen Munich GermanyTechnische Universitat Munchen
Opsahl S Benner R and Amon RMW (1999) Major fluxof terrigenous dissolved organic matter through the ArcticOcean Limnol Oceanogr 44 2017ndash2023
OrsquoSullivan LA Weightman AJ and Fry JC (2002) Newdegenerate Cytophaga-Flexibacter-Bacteroides-specific16S ribosomal DNA-targeted oligonucleotide probes revealhigh bacterial diversity in River Taff epilithon Appl EnvironMicrobiol 68 201ndash210
Pettit AN (1982) Cramer-von Misess statistic In Encyclo-pedia of Statistical Sciences Kotz S Johnson NL andRead CB (eds) New York USA Wiley pp 220ndash221
Quaiser A Ochsenreiter T Klenk HP Kletzin ATreusch AH Meurer G et al (2002) First insight intothe genome of an uncultivated crenarchaeote from soilEnviron Microbiol 4 603ndash611
Rappeacute MS and Giovannoni SJ (2003) The unculturedmicrobial majority Annu Rev Microbiol 57 369ndash394
Rappeacute MS Gordon DA Vergin KL and GiovannoniSJ (1999) Phylogeny of Actinobacteria small subunit(SSU) rRNA gene clones recovered from marine bacteri-oplankton Syst Appl Microbiol 22 106ndash112
Rappeacute MS Connon SA Vergin KL and GiovannoniSJ (2002) Cultivation of the ubiquitous SAR11 marinebacterioplankton clade Nature 418 630ndash633
Roller C Wagner M Amann R Ludwig W and SchleiferKH (1994) In-situ probing of Gram-positive bacteria withhigh DNA G+C content using 23S-ribosomal-RNA-targetedoligonucleotides Microbiology 140 2849ndash2858
Rondon MR August PR Bettermann AD Brady SFGrossman TH Liles MR et al (2000) Cloning the soilmetagenome a strategy for accessing the genetic andfunctional diversity of uncultured microorganisms ApplEnviron Microbiol 66 2541ndash2547
Sabehi G Massana R Bielawski JP Rosenberg MDelong EF and Beja O (2003) Novel proteorhodopsinvariants from the Mediterranean and Red Seas EnvironMicrobiol 5 842ndash849
Singleton DR Furlong MA Rathbun SL and WhitmanWB (2001) Quantitative comparisons of 16S rRNA genesequence libraries from environmental samples Appl Envi-ron Microbiol 67 4374ndash4376
Stackebrandt E and Goebel BM (1994) A place for DNAndashDNA reassociation and 16S ribosomal-RNA sequence-analysis in the present species definition in bacteriologyInt J Syst Bacteriol 44 846ndash849
Stein JL Marsh TL Wu KY Shizuya H and DeLongEF (1996) Characterization of uncultivated prokaryotesIsolation and analysis of a 40-kilobase-pair genome frag-ment from a planktonic marine archaeon J Bacteriol 178591ndash599
Thompson JR Marcelino LA and Polz MF (2002) Het-eroduplexes in mixed-template amplifications formationconsequence and elimination by lsquoreconditioning PCRrsquoNucleic Acids Res 30 2083ndash2088
Topp E (2003) Bacteria in agricultural soils Diversity roleand future perspectives Can J Soil Sci 83 303ndash309
Weller R Gloumlckner FO and Amann R (2000) 16S rRNA-targeted oligonucleotide probes for the in situ detectionof members of the phylum Cytophaga-Flavobacterium-Bacteroides Syst Appl Microbiol 23 107ndash114
von Wintzingerode F Gobel UB and Stackebrandt E(1997) Determination of microbial diversity in environmen-tal samples pitfalls of PCR-based rRNA analysis FEMSMicrobiol Rev 21 213ndash229
Zaitlin B Watson SB Dixon J and Steel D (2003)Actinomycetes in the Elbow River Basin Alberta CanWater Quality Res J Can 38 115ndash125
Zwart G Hiorns WD Methe BA Van Agterveld MPHuismans R Nold SC et al (1998) Nearly identical 16SrRNA sequences recovered from lakes in North Americaand Europe indicate the existence of clades of globallydistributed freshwater bacteria Syst Appl Microbiol 21546ndash556
Zwart G Crump BC Agterveld M Hagen F and HanSK (2002) Typical freshwater bacteria an analysis ofavailable 16S rRNA gene sequences from plankton oflakes and rivers Aquat Microb Ecol 28 141ndash155
Metagenomic analysis of freshwater bacteria 1887
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Fig 2 Phylogenetic relationships of selected beta-proteobacteria as depicted in the ARB database tree constructed of sequences longer than 1000 bp (June 2002) Delaware River beta-proteobacteria were added using the ARB parsimony quick add tool Clones in the library of 16S rDNA amplicons are labelled PCR and 16S rDNA-bearing metagenomic DNA clones are labelled Fosmid Values on the trapezoids indicate the number of sequences in the group Labels A B C D and E refer to clusters of closely related bacteria that are discussed in the text The scale bar represents 01 substitutions per site
1888 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Fig 3 Phylogenetic relationships of selected Actinobacteria as depicted in the ARB database tree constructed of sequences longer than 1000 bp (June 2002) Delaware River Actinobacteria were added using the ARB parsimony quick add tool Clones in the library of 16S rDNA amplicons are labelled PCR and 16S rDNA-bearing metagenomic DNA clones are labelled Fosmid Values on the trapezoids indicate the number of sequences in the group Labels A B and C refer to clusters of closely related bacteria that are discussed in the text The scale bar represents 01 substitutions per site
Metagenomic analysis of freshwater bacteria 1889
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
respectively (Fig 4) This distance is approximately equalto 97 sequence identity which is one approximation ofspecies-level identity (Stackebrandt and Goebel 1994)Coverage of the metagenomic library was higher thancoverage of the PCR library at all levels of genetic dis-tance with the greatest differences occurring at smallgenetic distance Coverage decreased rapidly withdecreasing genetic distance and at a genetic distance of001 the coverage of the metagenomic library was 46while the coverage of the PCR library decreased evenmore to just 16
Indices of diversity were higher for the metagenomiclibrary than the PCR library (Table 2) The Simpsonindex was 134 for the metagenomic library comparedwith 87 for the PCR library The difference was not assubstantial for the ShannonndashWeaver index (Table 2)Similarly measures of evenness were higher for themetagenomic library than the PCR library Evenness cal-culated using the ShannonndashWeaver index was 073 and066 for the metagenomic library and PCR libraryrespectively while the Simpson index of evenness was026 and 017 for the metagenomic and PCR librariesrespectively
We examined the possibility that diversity based onPCR clone library composition is overestimated becauseof artefacts of the PCR Re-conditioned PCR was used totest for potential artefacts introduced when heterologousPCR products re-anneal and form heteroduplexes that arethen subjected to mismatch repair upon cloning intoE coli (Thompson et al 2002) Re-conditioning of thePCR amplicons reduced the number of RFLP typesexpected in a library of 80 clones from 50 in the standardPCR library to 40 in the library produced with re-condi-tioned PCR (Fig 5) There was no difference betweenrarefaction curves of replicate PCR libraries produced withstandard PCR (Fig 5)
Comparison of FISH and clone libraries
Fluorescence in situ hybridization was used to assess thecorrespondence between clone library composition andbacterial community structure Microscopic enumerationof bacteria hybridized with fluorescently labelled 16SrRNA directed probes revealed a community dominatedby beta-proteobacteria Actinobacteria and Cytophaga-like bacteria (Table 1) As expected for a freshwater sys-tem Actinobacteria and beta-proteobacteria were themost abundant groups making up about 25 of the com-munity Cytophaga-like bacteria were also prominent
Fig 4 Coverage of the metagenomic and 16S rDNA libraries calcu-lated using the LIBSHUFF tool
Fig 5 Rarefaction curves of 16S rDNA clone libraries constructed with and without re-conditioned PCR PCR 1 and PCR 2 are clone libraries constructed from replicate PCR reactions without re-condi-tioning The error bars are one standard deviation
Table 2 Diversity of 16S rRNA genes in metagenomic and PCR clone libraries from the Delaware River
Library
ShannonndashWeaver Simpson
CoverageDiversity Evenness Diversity Evenness
Metagenomic 29 073 134 026 065PCR 26 066 87 017 052
Diversity indices were calculated using equations taken from Dunbar and colleagues (1999)
1890 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
accounting for 17 plusmn 4 of the community In contrastalpha-proteobacteria and gamma-proteobacteria wereconsiderably less abundant (less than 10) Seventy-seven percent of all prokaryotes were identified with agroup probe and 69 plusmn 11 were detected with the gen-eral bacterial probe Eub338
Bacterial community structure determined by FISH andinferred from the PCR library was surprisingly similar Theabundance of Cytophaga-like bacteria in the PCR librarywas within 25 of the abundance determined by FISHthe abundance of gamma-proteobacteria differed by 3or less and Actinobacteria differed by 38 according tothe two analyses (Table 1) In contrast the abundance ofbeta-proteobacteria clones in the PCR library was twicethe abundance of bacteria detected with the beta-proteo-bacteria FISH probe (Table 1) Three types of bacteriaincluding epsilon-proteobacteria Bdellovibrio and bacte-ria belonging to the candidate division TM6 were presentin the PCR library and would not have been detected withany of the FISH probes used in this study Although wehave no estimate of the abundance of bacteria in thesegroups they likely accounted for some of the 23 ofbacteria not detected by FISH (Table 1)
Discussion
It is necessary to examine whether the phylogenic com-position of metagenomic libraries reflects that of the orig-inal microbial composition in order to effectively linkcommunity structure and function revealed by metage-nomic analysis If metagenomic libraries represent abiased sample of microbial diversity they will not yield anaccurate assessment of metabolic diversity Assumingthey are unbiased large insert metagenomic libraries maybe particularly powerful in examining the relationshipbetween community structure and metabolic capacitiesLarge inserts increase the likelihood of obtaining a clonecontaining both lsquofunctionalrsquo genes (coding for metabolicfunction) and genes with phylogenetic information eg16S rRNA genes Therefore it is necessary to know howthe diversity of 16S rRNA genes sampled by large insertmetagenomic libraries corresponds to actual communitystructure Our main finding was that the composition ofthe metagenomic library at a high phylogenetic level wasnot statistically different from the PCR library but closerexamination revealed noteworthy differences between thelibraries and the community structure determined byFISH The metagenomic library sampled a broader diver-sity of Cytophaga-like bacteria than the PCR library but itmissed some groups of potentially important beta-proteo-bacteria The metagenomic and PCR libraries sampledthe same groups of Actinobacteria but FISH analysissuggested that both types of libraries may have failed toadequately sample Actinobacteria
Metagenomic analysis of Delaware River bacteria pro-vided a view of Cytophaga-like bacterial diversity that wasnot possible with the PCR library and FISH analyses Thenumber of clones representing Cytophaga-like bacteria inthe metagenomic library was threefold higher than theabundance determined by FISH and fourfold higher thanin the PCR library It was not surprising that the metage-nomic library sampled a larger diversity of Cytophaga-likebacteria than the PCR library because previous studieshad suggested that Cytophaga-like bacteria are underrep-resented in 16S rDNA PCR libraries relative to the abun-dance of these bacteria detected by FISH (Cottrell andKirchman 2000 Kirchman et al 2003) It was also notsurprising that the metagenomic library had more clonesfrom Cytophaga-like bacteria than would have beenexpected based on the FISH counts because the generalFISH probe CF319a for Cytophaga-like bacteria does notrecognize all Cytophaga-like bacteria (Weller et al 2000)The abundance of Cytophaga-like bacteria according tothe general FISH probe CFB560 for Cytophaga-like bac-teria (OrsquoSullivan et al 2002) was only 15 higher thanthat determined with CF319a The denaturing gradientelectrophoresis (DGGE) PCR primers used in this studyappear to adequately sample Cytophaga-like bacteria atleast compared with FISH analysis (Castle and Kirchman2004) and these primers apparently recognize a broaderdiversity of Cytophaga-like bacteria than the EubA andEubB primers as some Cytophaga-like bacteria detectedby DGGE in the metagenomic library were not sampledby the PCR library
The Cytophaga-like bacterial group includes a highlydiverse collection of bacteria but not all of this diversitywas present in our Delaware River sample In fact mostof the Cytophaga-like bacteria were most closely relatedto just two genera of Flavobacteriales (Flavobacteriumand Myroides) and a cluster in the Flectobacillus groupThis latter group is distinct from the CytophagandashFlavobac-terium cluster which is synonymous with the Flavobacte-riales (Kirchman 2002) The Delaware River bacteria inthe Flectobacillus group belong to the proposedPRD01a001B cluster of Cytophaga-like bacteria identifiedby Zwart and colleagues (2002) Nine of the 12 DelawareRiver sequences in the Flectobacillus group were greaterthan 98 similar to the PRD01a001B clone from theParker River Massachusetts The proposedPRD01a001B cluster which is currently comprised solelyof uncultivated bacteria has also been found in Adiron-dack lakes Lake Baikal and other freshwater environ-ments (Zwart et al 2002) However this group ofCytophaga-like bacteria has not been detected in allfreshwater environments for example it was not found inthe Columbia River (Zwart et al 2002)
The Delaware River beta-proteobacteria belonged toclusters widely distributed in freshwater systems Three
Metagenomic analysis of freshwater bacteria 1891
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
groups of Delaware River beta-proteobacteria clones (BD and E) were highly similar to bacteria in the freshwaterbeta-Proteobacteria clusters proposed by Zwart and col-leagues (2002) Sequences in Delaware River beta-pro-teobacteria group B were 95ndash99 similar to theRodoferax sp in the Rodoferax sp BAL47 cluster GroupD of Delaware Bay beta-proteobacteria corresponded tothe Polynucleobacter necessarius cluster (88ndash99sequence similarity to P necessarius) and sequences ingroup E probably belong to the proposed LD28 clusterSequences in group D were 89ndash97 similarity to theLD28 clone from Lake Loosdrech in the Netherlands(Zwart et al 1998)
The clusters of freshwater beta-proteobacteria pro-posed by Zwart and colleagues (2002) were more preva-lent in the PCR library than in the metagenomic librarysuggesting that they may represent bacteria that are pref-erentially sampled by PCR libraries Delaware River beta-proteobacteria group B included clones from both themetagenomic library and the PCR library but DelawareRiver groups D and E were comprised of only clones fromthe PCR library Different scenarios could lead to theoverrepresentation of bacterial groups in PCR librariescompared with libraries constructed without a PCR stepincluding selective PCR amplification with general bacte-rial primers differences in rRNA gene copy number orvariation in DNA extraction efficiency (von Wintzingerodeet al 1997) Fluorescence in situ hybridization probes areneeded to determine the abundance of these beta-proteo-bacteria which appear to be in a variety of freshwaterenvironments
There was a high degree of similarity in the Actinobac-teria sampled by the metagenomic and PCR libraries Allthree groups of Actinobacteria identified in this studyincluded clones from the metagenomic and PCR librariesOur results indicate that the presumed thicker cell wallsof these G+ bacteria do not interfere with the extraction ofhigh molecular weight DNA from these bacteria Howeverthe overall efficiency of DNA extraction may be low asActinobacteria were underrepresented in the librariescompared with their abundance determined by FISHAlthough the libraries may be missing some Actinobacte-ria they did recover clusters that are apparentlywidespread in freshwater systems Delaware River Acti-nobacteria groups B and C which correspond to the ACK-MI and STA-30 clusters identified by Zwart and colleagues(2002) were identified in the metagenomic and PCRlibraries Clones in cluster B were 92ndash99 similar to theACK-M1 clone and clones in cluster C were 92ndash97similar to the STA-30 clone Delaware River Actinobacte-ria group A which included clones from both libraries didnot overlap with previously described clusters of freshwa-ter Actinobacteria None of the G+ bacterial groups wedetected represented a terrestrial bacterial signal
because of transport of soils into the river Zwart andcolleagues (2002) also did not find evidence for an impor-tant contribution of soil bacteria to bacterial communitiesin rivers
Coverage in the metagenomic library was consistentlyhigher than in the PCR library over a broad range ofevolutionary distances although substantially higher cov-erage by the metagenomic library was restricted to evo-lutionary distances less than 005 Differences incoverage were not the result of different sequencelengths we analysed 1400 bp for the PCR clones but onlythe highly variable V3 region of the 16S rRNA gene forthe metagenomic clones LIBSHUFF analysis conductedwith the V3 region alone also revealed higher coverage bythe metagenomic library (data not shown) Differences incoverage may reflect more comprehensive sampling ofbacterial diversity by the metagenomic library
Lower coverage by the PCR library was consistent withour hypothesis that PCR artefacts lead to overestimatesof diversity Polymerase chain reaction re-conditioning flat-tened the rarefaction curve but the effect was only 20not as pronounced as would be expected if heterodu-plexes were a substantial source of diversity in the PCRlibraries The higher diversity in the metagenomic librarycompared with the PCR library is also consistent with theidea that the PCR libraries do not overestimate bacterialdiversity In fact they probably underestimate diversitybecause they miss certain groups such as Cytophaga-like bacteria (Cottrell and Kirchman 2000) Additional fac-tors such as cloning biases variation in genome size(Fogel et al 1999) rRNA gene copy number (Klappen-bach et al 2000) and ribosome content (Fegatella et al1998) could have resulted in larger differences than weobserved between the two library approaches and FISHGiven the potential sources of differences the degree ofsimilarity between the different library and FISHapproaches was higher than expected
These results and other suggest that the identities ofprokaryotes in aquatic environments are no longer com-pletely hidden in the microbial lsquoblack boxrsquo With the aid ofvarious culture-independent approaches it is now clearthat various proteobacterial groups Cytophaga-like bac-teria and Actinobacteria are the dominant groups of bac-teria in surface waters of aquatic systems (Giovannoniand Rappeacute 2000 Rappeacute and Giovannoni 2003)Archaea make a substantial contribution to prokaryoticdiversity in the ocean depths well below the euphotic zone(Massana et al 1998 Karner et al 2001) Althoughpotentially these groups could contain many subgroupsat least in marine waters it appears that only a few cladesdominate including alpha-proteobacteria belonging to theSAR11 group (Morris et al 2002 Rappeacute et al 2002) andRoseobacter spp (Gonzalez and Moran 1997) Cytoph-aga-like bacteria in Delaware cluster 2 (Kirchman et al
1892 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
2003) the marine Actinobacteria group (Rappeacute et al1999) and Archaea in group I and group II (Fuhrman et al1992 Massana et al 2000 Karner et al 2001)
Dominant groups in freshwater are also becomingapparent Our study revealed several widespread clustersof freshwater bacteria in the Delaware River that wereidentified in the metagenomic and PCR libraries Theseclusters are obvious candidates for further investigationincluding additional metagenomic analysis to exploretheir metabolic potential It is unclear if clusters such asthe P necessarius and LD28 beta-proteobacteria clus-ters which occurred in the PCR library but not themetagenomic library are selectively amplified by the PCRor for some reason are missed by the metagenomic clon-ing Additional examination by FISH could help resolvethe numerical importance of these freshwater bacteriaStudies such as these will further advance our under-standing of aquatic bacterial diversity as it evolves from aview of apparently intractable diversity towards whatappears to be a manageable number of dominant bacte-rial groups
Experimental procedures
Sample collection
A water sample was collected from a depth of 1 m in theDelaware River near Trenton New Jersey (40infin77cent-N74infin493cent-W) in December 2001 and subsampled for the anal-yses performed in this study
Fluorescence in situ hybridization
The sample for FISH analysis was fixed with 2 paraformal-dehyde at 4infinC for 12ndash18 h Bacteria were then collected on02 mm polycarbonate filters rinsed twice with deionizedwater and the filters were stored at -20infinC Bacteria weredetected with FISH probe Eub338 (Amann et al 1990) Wealso tried a mixture of general probes for Bacteria (Daimset al 1999) and found that it detected 70 plusmn 5 of DAPIpositive bacteria compared with 69 plusmn 11 for Eub338 aloneThe FISH probes for alpha- beta- and gamma-proteobacte-ria were Alf968 (Neef 1997 Gloumlckner et al 1999) Bet42aand Gam42a (Manz et al 1992) respectively Cytophaga-likebacteria were assayed with probe CF319a (Manz et al1996) and CFB560 (OrsquoSullivan et al 2002) and Actinobac-teria were enumerated using probe HGC69a (Roller et al1994) Non-specific binding was assessed using a negativecontrol probe (Karner and Fuhrman 1997) Hybridization ofCy3-labelled oligonucleotide probes was performed with por-tions of polycarbonate filters incubated in a hybridizationbuffer containing formamide to control the stringency ofhybridization (Cottrell and Kirchman 2000) Unbound probewas removed by washing the sample with a wash buffer at48infinC The filter piece was then rinsed in water and 70ethanol The air-dried sample was mounted on a glass slidewith a coverslip using an antifade mountant containing2 mg ml-1 DAPI Probe-positive bacteria (500ndash1000 bacteria
for each FISH probe) were enumerated using semiautomatedfluorescence microscopy (Cottrell and Kirchman 2003)
Preparation of the bacterial size fraction
Environmental DNA was extracted from the bacterial sizefraction obtained by pumping river water (500 l) sequentiallythrough a 1 mm nominal pore-size polypropylene string-wound filter (Cole Parmer) and a 08 mm polycarbonate filter(Nuclepore) Bacteria were collected from the filtrate by tan-gential flow filtration using a 01 mm hollow fibre filter (AGTechnology) and an Amicon DC10 gear pump The samplewas concentrated to 2 l rinsed with a buffer (05 M NaCl01 M EDTA 10 mM Tris pH 80) and stored frozen
Library construction using a fosmid vector
Bacteria were collected from thawed bacterial concentrate bycentrifugation in an SS34 rotor at 16 000 rpm for 30 minThe bacteria were resuspended in 1 ml of the supernatantand then mixed with 1 Sea Plaque agarose in deionizedwater at 55infinC (Stein et al 1996) The molten mixture wasdrawn into a 1 ml syringe and allowed to solidify The samplewas equilibrated with buffer (10 mM Tris pH 8 50 mM NaCl01 M EDTA 1 Sarkosyl) and digested with 1 mg ml-1
lysozyme at 37infinC for 4 h Buffer and lysozyme were replacedwith 1 Sarkosyl in 05 M EDTA and the sample wasdigested with 1 mg ml-1 proteinase K at 55infinC for 18 h Fol-lowing the enzyme treatments the sample was electrophore-sed for 8 h at 23 V through 03 SeaKem Gold agarose inTAE buffer DNA entering the gel but larger than the 48 kbsize standard was recovered from the gel by electroelutionand concentrated by ethanol precipitation The DNA was thensheared to 40 kb by 150 passages through a 200 ml pipettetip The DNA was prepared for ligation using a blunt endrepair kit (Epicentre Technologies) following the manufac-turerrsquos protocol The insert DNA was then size selected byelectrophoresis on 1 Sea Plaque GTG agarose DNA wasrecovered from the gel following treatment with Gelaseenzyme (Epicentre Technologies) digestion and ethanol pre-cipitation The DNA was then ligated to the pCC1FOS fosmidvector (Epicentre Technologies) and packaged using phageprotein extract following the manufacturerrsquos recommendedprocedure EPI300 host E coli infected with recombinantphage were plated on LB media containing chloramphenicolClones were picked and sorted into 96-well microtitre plates
PCR library construction
Three PCR libraries were constructed with the DNA isolatedfrom the Delaware River Two replicate libraries were con-structed by cloning amplicons generated from separate PCRreactions using the general bacterial primers EubA and EubB(Lane 1991) A third library was constructed using re-condi-tioned amplicons from the PCR reaction used to generate thefirst library Re-conditioning was performed with three roundsof thermal cycling applied to amplicons diluted 10-fold infresh PCR reagents (Thompson et al 2002) The 25 ml PCRreactions included 05 ml 12 ml and 2 ml additions of 10 mMdeoxynucloside triphosphate 25 mM MgCl2 10 mM primer
Metagenomic analysis of freshwater bacteria 1893
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
stocks respectively Bovine serum albumin was added to afinal concentration of 02 mg ml-1 Two and a half units of Taqpolymerase (Promega) and a one-tenth volume of 10yen bufferwere added to each reaction The thermal cycling conditionsconsisted of a touchdown series in which the annealing tem-perature decreased from 65infinC to 55infinC by 1infinC per cyclefollowed by 15 cycles at 55infinC each for 1 min Each cycleincluded a 1 min denaturation step at 95infinC and an extensionstep of 25 min at 72infinC Amplicons were cloned with a TOPO-TA cloning kit with pCR21-TOPO vector (Invitrogen) accord-ing to the manufacturerrsquos instructions
Screening PCR and metagenomic libraries
Clones were screened for insert size using M13 forward andreverse PCR primers which flank the cloned insert andthose with full-length inserts were screened by restrictionfragment length polymorphism (RFLP) analysis by digestingthe M13 PCR amplicons with a mixture of HhaI and RsaI(New England Biolabs) Restriction patterns on 2 Meta-phore agarose (FMC) agarose gels were compared manuallyClones with identical restriction patterns were groupedtogether into RFLP types The 16S rRNA genes from onerepresentative of each RFLP type in the first library werecompletely sequenced
Pools of 96 fosmid clones were screened for 16S rRNAgenes by DGGE of PCR amplicons generated with primersGC358F and 517R (Muyzer et al 1995) Selected bandsresolved on an 8 polyacrylamide gel containing a gradientof 25ndash55 denaturant (138ndash22 formamide and 105ndash23urea) were re-amplified and sequenced Phylogenetic classi-fication was determined using BLAST and the ARB sequenceanalysis tool as described below Bands with equal electro-phoretic mobility were assigned the same phylogeneticclassification
Sequence and library analysis
Nucleotide sequences were analysed using BLAST [version21 National Center for Biotechnology Information (httpwwwncbinlmnihgovBLAST) and the ARB package (Lud-wig et al 2004)] Sequences were aligned with the ARB FastAligner version 103 adjusted manually with secondary-structure criteria Phylogenetic relationships were deter-mined using the Quick Add Parsimony tool and the1000_pub_may02 phylogenetic tree in the ARB database(June 2002 release)
Sequence similarities were determined using the tools inARB and rarefaction curves were calculated using AnalyticRarefaction 13 (httpwwwugaedustratasoftware) Librar-ies were compared using the LIBSHUFF analysis tool(httpwwwarchesugaedu~whitmanlibshuffhtml) (Single-ton et al 2001) Diversity indices were calculated usingequations taken from Dunbar and colleagues (1999)
Statistical analysis
Libraries were compared using LIBSHUFF analysis (Single-ton et al 2001) This approach compares libraries based ontheir coverage and the number of sequences in one library
that are not found in the second library This so-called lsquohet-erologous coveragersquo is calculated over a range of evolution-ary distances to obtain a coverage curve The differencebetween the lsquoheterologous coveragersquo curve and the coveragecurve of the first library is calculated using the Crameacuter-vonMises test statistic (Pettit 1982) and compared with the cov-erage curve calculated after shuffling sequences between thetwo libraries
Nucleotide sequence accession numbers
The nucleotide sequence data reported in this work aredeposited in GenBank under accession numbers AY562245ndashAY562366
Acknowledgements
This work was supported by the US Department of EnergyMicrobial Genomes Program and the National ScienceFoundation
References
Amann RI Binder BJ Olson RJ Chisholm SWDevereux R and Stahl DA (1990) Combination of 16Sribosomal-RNA-targeted oligonucleotide probes with flowcytometry for analyzing mixed microbial populations ApplEnviron Microbiol 56 1919ndash1925
Amon RMW and Benner R (1996) Photochemical andmicrobial consumption of dissolved organic carbon anddissolved oxygen in the Amazon River system GeochimCosmochim Ac 60 1783ndash1792
Beja O Aravind L Koonin EV Suzuki MT Hadd ANguyen LP et al (2000a) Bacterial rhodopsin evidencefor a new type of phototrophy in the sea Science 2891902ndash1906
Beja O Suzuki MT Koonin EV Aravind L Hadd ANguyen LP et al (2000b) Construction and analysis ofbacterial artificial chromosome libraries from a marinemicrobial assemblage Environ Microbiol 2 516ndash529
Beja O Spudich EN Spudich JL Leclerc M andDeLong EF (2001) Proteorhodopsin phototrophy in theocean Nature 411 786ndash789
Button DK and Robertson BR (2001) Determination ofDNA content of aquatic bacteria by flow cytometry ApplEnviron Microbiol 67 1636ndash1645
Button DK Robertson BR Lepp PW and SchmidtTM (1998) A small dilute-cytoplasm high-affinity novelbacterium isolated by extinction culture and having kineticconstants compatible with growth at ambient concentra-tions of dissolved nutrients in seawater Appl EnvironMicrobiol 64 4467ndash4476
Castle D and Kirchman DL (2004) Composition of estu-arine bacterial communities assessed by denaturing gradi-ent gel electrophoresis and fluorescence in situhybridization Limnol Oceanogr 2 303ndash314
Cottrell MT and Kirchman DL (2000) Community compo-sition of marine bacterioplankton determined by 16S rRNAgene clone libraries and fluorescence in situ hybridizationAppl Environ Microbiol 66 5116ndash5122
1894 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Cottrell MT and Kirchman DL (2003) Contribution ofmajor bacterial groups to bacterial biomass production(thymidine and leucine incorporation) in the Delaware estu-ary Limnol Oceanogr 48 168ndash178
Cottrell MT Moore JA and Kirchman DL (1999) Chiti-nases from uncultured marine microorganisms Appl Envi-ron Microbiol 65 2553ndash2557
Covert JS and Moran MA (2001) Molecular characteriza-tion of estuarine bacterial communities that use high- andlow-molecular weight fractions of dissolved organic carbonAquat Microb Ecol 25 127ndash139
Daims H Bruhl A Amann R Schleifer KH and WagnerM (1999) The domain-specific probe EUB338 is insuffi-cient for the detection of all Bacteria Development andevaluation of a more comprehensive probe set Syst ApplMicrobiol 22 434ndash444
Dunbar J Takala S Barns SM Davis JA and KuskeCR (1999) Levels of bacterial community diversity in fourarid soils compared by cultivation and 16S rRNA genecloning Appl Environ Microbiol 65 1662ndash1669
Fegatella F Lim J Kjelleberg S and Cavicchioli R(1998) Implications of rRNA operon copy number and ribo-some content in the marine oligotrophic ultramicrobacte-rium Sphingomonas sp strain RB2256 Appl EnvironMicrobiol 64 4433ndash4438
Fogel GB Collins CR Li J and Brunk CF (1999)Prokaryotic genome size and SSU rDNA copy numberEstimation of microbial relative abundance from a mixedpopulation Microb Ecol 38 93ndash113
Fuhrman JA McCallum K and Davis AA (1992) Novelmajor Archaebacterial group from marine plankton Nature356 148ndash149
Gillespie DE Brady SF Bettermann AD CianciottoNP Liles MR Rondon MR et al (2002) Isolation ofantibiotics turbomycin A and B from a metagenomic libraryof soil microbial DNA Appl Environ Microbiol 68 4301ndash4306
Giovannoni SJ and Rappeacute MS (2000) Evolution diversityand molecular ecology of marine prokaryotes In MicrobialEcology of the Oceans Kirchman DL (ed) New YorkUSA Wiley-Liss pp 47ndash84
Gloumlckner FO Fuchs BM and Amann R (1999) Bacteri-oplankton compositions of lakes and oceans a first com-parison based on fluorescence in situ hybridization ApplEnviron Microbiol 65 3721ndash3726
Gloumlckner FO Zaichikov E Belkova N Denissova LPernthaler J Pernthaler A and Amann R (2000) Com-parative 16S rRNA analysis of lake bacterioplanktonreveals globally distributed phylogenetic clusters includingan abundant group of actinobacteria Appl Environ Micro-biol 66 5053ndash5065
Gonzalez JM and Moran MA (1997) Numerical domi-nance of a group of marine bacteria in the alpha- subclassof the class Proteobacteria in coastal seawater Appl Envi-ron Microbiol 63 4237ndash4242
Hernes PJ and Benner R (2002) Transport and diagene-sis of dissolved and particulate terrigenous organic matterin the North Pacific Ocean Deep-Sea Res Part II ndash TopStud Oceanogr 49 2119ndash2132
Karner M and Fuhrman JA (1997) Determination of activemarine bacterioplankton a comparison of universal 16S
rRNA probes autoradiography and nucleoid staining ApplEnviron Microbiol 63 1208ndash1213
Karner MB DeLong EF and Karl DM (2001) Archaealdominance in the mesopelagic zone of the Pacific OceanNature 409 507ndash510
Kirchman DL (2002) The ecology of CytophagandashFlavobac-teria in aquatic environments FEMS Microbiol Ecol 3991ndash100
Kirchman DL Yu LY and Cottrell MT (2003) Diversityand abundance of uncultured Cytophaga-like bacteria inthe Delaware Estuary Appl Environ Microbiol 69 6587ndash6596
Klappenbach JA Dunbar JM and Schmidt TM (2000)rRNA operon copy number reflects ecological strategies ofbacteria Appl Environ Microbiol 66 1328ndash1333
Knietsch A Waschkowitz T Bowien S Henne A andDaniel R (2003) Metagenomes of complex microbial con-sortia derived from different soils as sources for novelgenes conferring formation of carbonyls from short-chainpolyols on Escherichia coli J Mol Microbiol Biotechnol 546ndash56
Lane DJ (1991) 16S23S rRNA sequencing In Nucleic AcidTechniques in Bacterial Systematics Stackebrandt E andGoodfellow M (eds) Chichester UK Wiley and Sonspp 115ndash175
Liles MR Manske BF Bintrim SB Handelsman J andGoodman RM (2003) A census of rRNA genes and linkedgenomic sequences within a soil metagenomic libraryAppl Environ Microbiol 69 2684ndash2691
Ludwig W Strunk O Westram R Richter L Meier HYadhukumar et al (2004) ARB a software environmentfor sequence data Nucleic Acids Res 32 1363ndash1371
Manz W Amann R Ludwig W Wagner M and Schlei-fer KH (1992) Phylogenetic oligodeoxynucleotide probesfor the major subclasses of Proteobacteria ndash problems andsolutions Syst Appl Microbiol 15 593ndash600
Manz W Amann R Ludwig W Vancanneyt M andSchleifer KH (1996) Application of a suite of 16S rRNA-specific oligonucleotide probes designed to investigatebacteria of the phylum CytophagandashFlavobacterndashBacteroi-des in the natural environment Microbiology 142 1097ndash1106
Massana R Taylor LJ Murray AE Wu KY JeffreyWH and DeLong EF (1998) Vertical distribution andtemporal variation of marine planktonic archaea in theGerlache Strait Antarctica during early spring LimnolOceanogr 43 607ndash617
Massana R DeLong EF and Pedros-Alio C (2000) A fewcosmopolitan phylotypes dominate planktonic archaealassemblages in widely different oceanic provinces ApplEnviron Microbiol 66 1777ndash1787
Morris RM Rappeacute MS Connon SA Vergin KL Sie-bold WA Carlson CA and Giovannoni SJ (2002)SAR11 clade dominates ocean surface bacterioplanktoncommunities Nature 420 806ndash810
Muyzer G Teske A Wirsen CO and Jannasch HW(1995) Phylogenetic-relationships of Thiomicrospira spe-cies and their identification in deep-sea hydrothermal ventsamples by denaturing gradient gel-electrophoresis of 16SrDNA fragments Arch Microbiol 164 165ndash172
Neef A (1997) Anwendung der in situ-Einzelzell-
Metagenomic analysis of freshwater bacteria 1895
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Identifizierung von Bakterien zur Populationsanalyse inKomplexen Mikrobiellen Biozonosen Munich GermanyTechnische Universitat Munchen
Opsahl S Benner R and Amon RMW (1999) Major fluxof terrigenous dissolved organic matter through the ArcticOcean Limnol Oceanogr 44 2017ndash2023
OrsquoSullivan LA Weightman AJ and Fry JC (2002) Newdegenerate Cytophaga-Flexibacter-Bacteroides-specific16S ribosomal DNA-targeted oligonucleotide probes revealhigh bacterial diversity in River Taff epilithon Appl EnvironMicrobiol 68 201ndash210
Pettit AN (1982) Cramer-von Misess statistic In Encyclo-pedia of Statistical Sciences Kotz S Johnson NL andRead CB (eds) New York USA Wiley pp 220ndash221
Quaiser A Ochsenreiter T Klenk HP Kletzin ATreusch AH Meurer G et al (2002) First insight intothe genome of an uncultivated crenarchaeote from soilEnviron Microbiol 4 603ndash611
Rappeacute MS and Giovannoni SJ (2003) The unculturedmicrobial majority Annu Rev Microbiol 57 369ndash394
Rappeacute MS Gordon DA Vergin KL and GiovannoniSJ (1999) Phylogeny of Actinobacteria small subunit(SSU) rRNA gene clones recovered from marine bacteri-oplankton Syst Appl Microbiol 22 106ndash112
Rappeacute MS Connon SA Vergin KL and GiovannoniSJ (2002) Cultivation of the ubiquitous SAR11 marinebacterioplankton clade Nature 418 630ndash633
Roller C Wagner M Amann R Ludwig W and SchleiferKH (1994) In-situ probing of Gram-positive bacteria withhigh DNA G+C content using 23S-ribosomal-RNA-targetedoligonucleotides Microbiology 140 2849ndash2858
Rondon MR August PR Bettermann AD Brady SFGrossman TH Liles MR et al (2000) Cloning the soilmetagenome a strategy for accessing the genetic andfunctional diversity of uncultured microorganisms ApplEnviron Microbiol 66 2541ndash2547
Sabehi G Massana R Bielawski JP Rosenberg MDelong EF and Beja O (2003) Novel proteorhodopsinvariants from the Mediterranean and Red Seas EnvironMicrobiol 5 842ndash849
Singleton DR Furlong MA Rathbun SL and WhitmanWB (2001) Quantitative comparisons of 16S rRNA genesequence libraries from environmental samples Appl Envi-ron Microbiol 67 4374ndash4376
Stackebrandt E and Goebel BM (1994) A place for DNAndashDNA reassociation and 16S ribosomal-RNA sequence-analysis in the present species definition in bacteriologyInt J Syst Bacteriol 44 846ndash849
Stein JL Marsh TL Wu KY Shizuya H and DeLongEF (1996) Characterization of uncultivated prokaryotesIsolation and analysis of a 40-kilobase-pair genome frag-ment from a planktonic marine archaeon J Bacteriol 178591ndash599
Thompson JR Marcelino LA and Polz MF (2002) Het-eroduplexes in mixed-template amplifications formationconsequence and elimination by lsquoreconditioning PCRrsquoNucleic Acids Res 30 2083ndash2088
Topp E (2003) Bacteria in agricultural soils Diversity roleand future perspectives Can J Soil Sci 83 303ndash309
Weller R Gloumlckner FO and Amann R (2000) 16S rRNA-targeted oligonucleotide probes for the in situ detectionof members of the phylum Cytophaga-Flavobacterium-Bacteroides Syst Appl Microbiol 23 107ndash114
von Wintzingerode F Gobel UB and Stackebrandt E(1997) Determination of microbial diversity in environmen-tal samples pitfalls of PCR-based rRNA analysis FEMSMicrobiol Rev 21 213ndash229
Zaitlin B Watson SB Dixon J and Steel D (2003)Actinomycetes in the Elbow River Basin Alberta CanWater Quality Res J Can 38 115ndash125
Zwart G Hiorns WD Methe BA Van Agterveld MPHuismans R Nold SC et al (1998) Nearly identical 16SrRNA sequences recovered from lakes in North Americaand Europe indicate the existence of clades of globallydistributed freshwater bacteria Syst Appl Microbiol 21546ndash556
Zwart G Crump BC Agterveld M Hagen F and HanSK (2002) Typical freshwater bacteria an analysis ofavailable 16S rRNA gene sequences from plankton oflakes and rivers Aquat Microb Ecol 28 141ndash155
1888 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Fig 3 Phylogenetic relationships of selected Actinobacteria as depicted in the ARB database tree constructed of sequences longer than 1000 bp (June 2002) Delaware River Actinobacteria were added using the ARB parsimony quick add tool Clones in the library of 16S rDNA amplicons are labelled PCR and 16S rDNA-bearing metagenomic DNA clones are labelled Fosmid Values on the trapezoids indicate the number of sequences in the group Labels A B and C refer to clusters of closely related bacteria that are discussed in the text The scale bar represents 01 substitutions per site
Metagenomic analysis of freshwater bacteria 1889
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
respectively (Fig 4) This distance is approximately equalto 97 sequence identity which is one approximation ofspecies-level identity (Stackebrandt and Goebel 1994)Coverage of the metagenomic library was higher thancoverage of the PCR library at all levels of genetic dis-tance with the greatest differences occurring at smallgenetic distance Coverage decreased rapidly withdecreasing genetic distance and at a genetic distance of001 the coverage of the metagenomic library was 46while the coverage of the PCR library decreased evenmore to just 16
Indices of diversity were higher for the metagenomiclibrary than the PCR library (Table 2) The Simpsonindex was 134 for the metagenomic library comparedwith 87 for the PCR library The difference was not assubstantial for the ShannonndashWeaver index (Table 2)Similarly measures of evenness were higher for themetagenomic library than the PCR library Evenness cal-culated using the ShannonndashWeaver index was 073 and066 for the metagenomic library and PCR libraryrespectively while the Simpson index of evenness was026 and 017 for the metagenomic and PCR librariesrespectively
We examined the possibility that diversity based onPCR clone library composition is overestimated becauseof artefacts of the PCR Re-conditioned PCR was used totest for potential artefacts introduced when heterologousPCR products re-anneal and form heteroduplexes that arethen subjected to mismatch repair upon cloning intoE coli (Thompson et al 2002) Re-conditioning of thePCR amplicons reduced the number of RFLP typesexpected in a library of 80 clones from 50 in the standardPCR library to 40 in the library produced with re-condi-tioned PCR (Fig 5) There was no difference betweenrarefaction curves of replicate PCR libraries produced withstandard PCR (Fig 5)
Comparison of FISH and clone libraries
Fluorescence in situ hybridization was used to assess thecorrespondence between clone library composition andbacterial community structure Microscopic enumerationof bacteria hybridized with fluorescently labelled 16SrRNA directed probes revealed a community dominatedby beta-proteobacteria Actinobacteria and Cytophaga-like bacteria (Table 1) As expected for a freshwater sys-tem Actinobacteria and beta-proteobacteria were themost abundant groups making up about 25 of the com-munity Cytophaga-like bacteria were also prominent
Fig 4 Coverage of the metagenomic and 16S rDNA libraries calcu-lated using the LIBSHUFF tool
Fig 5 Rarefaction curves of 16S rDNA clone libraries constructed with and without re-conditioned PCR PCR 1 and PCR 2 are clone libraries constructed from replicate PCR reactions without re-condi-tioning The error bars are one standard deviation
Table 2 Diversity of 16S rRNA genes in metagenomic and PCR clone libraries from the Delaware River
Library
ShannonndashWeaver Simpson
CoverageDiversity Evenness Diversity Evenness
Metagenomic 29 073 134 026 065PCR 26 066 87 017 052
Diversity indices were calculated using equations taken from Dunbar and colleagues (1999)
1890 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
accounting for 17 plusmn 4 of the community In contrastalpha-proteobacteria and gamma-proteobacteria wereconsiderably less abundant (less than 10) Seventy-seven percent of all prokaryotes were identified with agroup probe and 69 plusmn 11 were detected with the gen-eral bacterial probe Eub338
Bacterial community structure determined by FISH andinferred from the PCR library was surprisingly similar Theabundance of Cytophaga-like bacteria in the PCR librarywas within 25 of the abundance determined by FISHthe abundance of gamma-proteobacteria differed by 3or less and Actinobacteria differed by 38 according tothe two analyses (Table 1) In contrast the abundance ofbeta-proteobacteria clones in the PCR library was twicethe abundance of bacteria detected with the beta-proteo-bacteria FISH probe (Table 1) Three types of bacteriaincluding epsilon-proteobacteria Bdellovibrio and bacte-ria belonging to the candidate division TM6 were presentin the PCR library and would not have been detected withany of the FISH probes used in this study Although wehave no estimate of the abundance of bacteria in thesegroups they likely accounted for some of the 23 ofbacteria not detected by FISH (Table 1)
Discussion
It is necessary to examine whether the phylogenic com-position of metagenomic libraries reflects that of the orig-inal microbial composition in order to effectively linkcommunity structure and function revealed by metage-nomic analysis If metagenomic libraries represent abiased sample of microbial diversity they will not yield anaccurate assessment of metabolic diversity Assumingthey are unbiased large insert metagenomic libraries maybe particularly powerful in examining the relationshipbetween community structure and metabolic capacitiesLarge inserts increase the likelihood of obtaining a clonecontaining both lsquofunctionalrsquo genes (coding for metabolicfunction) and genes with phylogenetic information eg16S rRNA genes Therefore it is necessary to know howthe diversity of 16S rRNA genes sampled by large insertmetagenomic libraries corresponds to actual communitystructure Our main finding was that the composition ofthe metagenomic library at a high phylogenetic level wasnot statistically different from the PCR library but closerexamination revealed noteworthy differences between thelibraries and the community structure determined byFISH The metagenomic library sampled a broader diver-sity of Cytophaga-like bacteria than the PCR library but itmissed some groups of potentially important beta-proteo-bacteria The metagenomic and PCR libraries sampledthe same groups of Actinobacteria but FISH analysissuggested that both types of libraries may have failed toadequately sample Actinobacteria
Metagenomic analysis of Delaware River bacteria pro-vided a view of Cytophaga-like bacterial diversity that wasnot possible with the PCR library and FISH analyses Thenumber of clones representing Cytophaga-like bacteria inthe metagenomic library was threefold higher than theabundance determined by FISH and fourfold higher thanin the PCR library It was not surprising that the metage-nomic library sampled a larger diversity of Cytophaga-likebacteria than the PCR library because previous studieshad suggested that Cytophaga-like bacteria are underrep-resented in 16S rDNA PCR libraries relative to the abun-dance of these bacteria detected by FISH (Cottrell andKirchman 2000 Kirchman et al 2003) It was also notsurprising that the metagenomic library had more clonesfrom Cytophaga-like bacteria than would have beenexpected based on the FISH counts because the generalFISH probe CF319a for Cytophaga-like bacteria does notrecognize all Cytophaga-like bacteria (Weller et al 2000)The abundance of Cytophaga-like bacteria according tothe general FISH probe CFB560 for Cytophaga-like bac-teria (OrsquoSullivan et al 2002) was only 15 higher thanthat determined with CF319a The denaturing gradientelectrophoresis (DGGE) PCR primers used in this studyappear to adequately sample Cytophaga-like bacteria atleast compared with FISH analysis (Castle and Kirchman2004) and these primers apparently recognize a broaderdiversity of Cytophaga-like bacteria than the EubA andEubB primers as some Cytophaga-like bacteria detectedby DGGE in the metagenomic library were not sampledby the PCR library
The Cytophaga-like bacterial group includes a highlydiverse collection of bacteria but not all of this diversitywas present in our Delaware River sample In fact mostof the Cytophaga-like bacteria were most closely relatedto just two genera of Flavobacteriales (Flavobacteriumand Myroides) and a cluster in the Flectobacillus groupThis latter group is distinct from the CytophagandashFlavobac-terium cluster which is synonymous with the Flavobacte-riales (Kirchman 2002) The Delaware River bacteria inthe Flectobacillus group belong to the proposedPRD01a001B cluster of Cytophaga-like bacteria identifiedby Zwart and colleagues (2002) Nine of the 12 DelawareRiver sequences in the Flectobacillus group were greaterthan 98 similar to the PRD01a001B clone from theParker River Massachusetts The proposedPRD01a001B cluster which is currently comprised solelyof uncultivated bacteria has also been found in Adiron-dack lakes Lake Baikal and other freshwater environ-ments (Zwart et al 2002) However this group ofCytophaga-like bacteria has not been detected in allfreshwater environments for example it was not found inthe Columbia River (Zwart et al 2002)
The Delaware River beta-proteobacteria belonged toclusters widely distributed in freshwater systems Three
Metagenomic analysis of freshwater bacteria 1891
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
groups of Delaware River beta-proteobacteria clones (BD and E) were highly similar to bacteria in the freshwaterbeta-Proteobacteria clusters proposed by Zwart and col-leagues (2002) Sequences in Delaware River beta-pro-teobacteria group B were 95ndash99 similar to theRodoferax sp in the Rodoferax sp BAL47 cluster GroupD of Delaware Bay beta-proteobacteria corresponded tothe Polynucleobacter necessarius cluster (88ndash99sequence similarity to P necessarius) and sequences ingroup E probably belong to the proposed LD28 clusterSequences in group D were 89ndash97 similarity to theLD28 clone from Lake Loosdrech in the Netherlands(Zwart et al 1998)
The clusters of freshwater beta-proteobacteria pro-posed by Zwart and colleagues (2002) were more preva-lent in the PCR library than in the metagenomic librarysuggesting that they may represent bacteria that are pref-erentially sampled by PCR libraries Delaware River beta-proteobacteria group B included clones from both themetagenomic library and the PCR library but DelawareRiver groups D and E were comprised of only clones fromthe PCR library Different scenarios could lead to theoverrepresentation of bacterial groups in PCR librariescompared with libraries constructed without a PCR stepincluding selective PCR amplification with general bacte-rial primers differences in rRNA gene copy number orvariation in DNA extraction efficiency (von Wintzingerodeet al 1997) Fluorescence in situ hybridization probes areneeded to determine the abundance of these beta-proteo-bacteria which appear to be in a variety of freshwaterenvironments
There was a high degree of similarity in the Actinobac-teria sampled by the metagenomic and PCR libraries Allthree groups of Actinobacteria identified in this studyincluded clones from the metagenomic and PCR librariesOur results indicate that the presumed thicker cell wallsof these G+ bacteria do not interfere with the extraction ofhigh molecular weight DNA from these bacteria Howeverthe overall efficiency of DNA extraction may be low asActinobacteria were underrepresented in the librariescompared with their abundance determined by FISHAlthough the libraries may be missing some Actinobacte-ria they did recover clusters that are apparentlywidespread in freshwater systems Delaware River Acti-nobacteria groups B and C which correspond to the ACK-MI and STA-30 clusters identified by Zwart and colleagues(2002) were identified in the metagenomic and PCRlibraries Clones in cluster B were 92ndash99 similar to theACK-M1 clone and clones in cluster C were 92ndash97similar to the STA-30 clone Delaware River Actinobacte-ria group A which included clones from both libraries didnot overlap with previously described clusters of freshwa-ter Actinobacteria None of the G+ bacterial groups wedetected represented a terrestrial bacterial signal
because of transport of soils into the river Zwart andcolleagues (2002) also did not find evidence for an impor-tant contribution of soil bacteria to bacterial communitiesin rivers
Coverage in the metagenomic library was consistentlyhigher than in the PCR library over a broad range ofevolutionary distances although substantially higher cov-erage by the metagenomic library was restricted to evo-lutionary distances less than 005 Differences incoverage were not the result of different sequencelengths we analysed 1400 bp for the PCR clones but onlythe highly variable V3 region of the 16S rRNA gene forthe metagenomic clones LIBSHUFF analysis conductedwith the V3 region alone also revealed higher coverage bythe metagenomic library (data not shown) Differences incoverage may reflect more comprehensive sampling ofbacterial diversity by the metagenomic library
Lower coverage by the PCR library was consistent withour hypothesis that PCR artefacts lead to overestimatesof diversity Polymerase chain reaction re-conditioning flat-tened the rarefaction curve but the effect was only 20not as pronounced as would be expected if heterodu-plexes were a substantial source of diversity in the PCRlibraries The higher diversity in the metagenomic librarycompared with the PCR library is also consistent with theidea that the PCR libraries do not overestimate bacterialdiversity In fact they probably underestimate diversitybecause they miss certain groups such as Cytophaga-like bacteria (Cottrell and Kirchman 2000) Additional fac-tors such as cloning biases variation in genome size(Fogel et al 1999) rRNA gene copy number (Klappen-bach et al 2000) and ribosome content (Fegatella et al1998) could have resulted in larger differences than weobserved between the two library approaches and FISHGiven the potential sources of differences the degree ofsimilarity between the different library and FISHapproaches was higher than expected
These results and other suggest that the identities ofprokaryotes in aquatic environments are no longer com-pletely hidden in the microbial lsquoblack boxrsquo With the aid ofvarious culture-independent approaches it is now clearthat various proteobacterial groups Cytophaga-like bac-teria and Actinobacteria are the dominant groups of bac-teria in surface waters of aquatic systems (Giovannoniand Rappeacute 2000 Rappeacute and Giovannoni 2003)Archaea make a substantial contribution to prokaryoticdiversity in the ocean depths well below the euphotic zone(Massana et al 1998 Karner et al 2001) Althoughpotentially these groups could contain many subgroupsat least in marine waters it appears that only a few cladesdominate including alpha-proteobacteria belonging to theSAR11 group (Morris et al 2002 Rappeacute et al 2002) andRoseobacter spp (Gonzalez and Moran 1997) Cytoph-aga-like bacteria in Delaware cluster 2 (Kirchman et al
1892 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
2003) the marine Actinobacteria group (Rappeacute et al1999) and Archaea in group I and group II (Fuhrman et al1992 Massana et al 2000 Karner et al 2001)
Dominant groups in freshwater are also becomingapparent Our study revealed several widespread clustersof freshwater bacteria in the Delaware River that wereidentified in the metagenomic and PCR libraries Theseclusters are obvious candidates for further investigationincluding additional metagenomic analysis to exploretheir metabolic potential It is unclear if clusters such asthe P necessarius and LD28 beta-proteobacteria clus-ters which occurred in the PCR library but not themetagenomic library are selectively amplified by the PCRor for some reason are missed by the metagenomic clon-ing Additional examination by FISH could help resolvethe numerical importance of these freshwater bacteriaStudies such as these will further advance our under-standing of aquatic bacterial diversity as it evolves from aview of apparently intractable diversity towards whatappears to be a manageable number of dominant bacte-rial groups
Experimental procedures
Sample collection
A water sample was collected from a depth of 1 m in theDelaware River near Trenton New Jersey (40infin77cent-N74infin493cent-W) in December 2001 and subsampled for the anal-yses performed in this study
Fluorescence in situ hybridization
The sample for FISH analysis was fixed with 2 paraformal-dehyde at 4infinC for 12ndash18 h Bacteria were then collected on02 mm polycarbonate filters rinsed twice with deionizedwater and the filters were stored at -20infinC Bacteria weredetected with FISH probe Eub338 (Amann et al 1990) Wealso tried a mixture of general probes for Bacteria (Daimset al 1999) and found that it detected 70 plusmn 5 of DAPIpositive bacteria compared with 69 plusmn 11 for Eub338 aloneThe FISH probes for alpha- beta- and gamma-proteobacte-ria were Alf968 (Neef 1997 Gloumlckner et al 1999) Bet42aand Gam42a (Manz et al 1992) respectively Cytophaga-likebacteria were assayed with probe CF319a (Manz et al1996) and CFB560 (OrsquoSullivan et al 2002) and Actinobac-teria were enumerated using probe HGC69a (Roller et al1994) Non-specific binding was assessed using a negativecontrol probe (Karner and Fuhrman 1997) Hybridization ofCy3-labelled oligonucleotide probes was performed with por-tions of polycarbonate filters incubated in a hybridizationbuffer containing formamide to control the stringency ofhybridization (Cottrell and Kirchman 2000) Unbound probewas removed by washing the sample with a wash buffer at48infinC The filter piece was then rinsed in water and 70ethanol The air-dried sample was mounted on a glass slidewith a coverslip using an antifade mountant containing2 mg ml-1 DAPI Probe-positive bacteria (500ndash1000 bacteria
for each FISH probe) were enumerated using semiautomatedfluorescence microscopy (Cottrell and Kirchman 2003)
Preparation of the bacterial size fraction
Environmental DNA was extracted from the bacterial sizefraction obtained by pumping river water (500 l) sequentiallythrough a 1 mm nominal pore-size polypropylene string-wound filter (Cole Parmer) and a 08 mm polycarbonate filter(Nuclepore) Bacteria were collected from the filtrate by tan-gential flow filtration using a 01 mm hollow fibre filter (AGTechnology) and an Amicon DC10 gear pump The samplewas concentrated to 2 l rinsed with a buffer (05 M NaCl01 M EDTA 10 mM Tris pH 80) and stored frozen
Library construction using a fosmid vector
Bacteria were collected from thawed bacterial concentrate bycentrifugation in an SS34 rotor at 16 000 rpm for 30 minThe bacteria were resuspended in 1 ml of the supernatantand then mixed with 1 Sea Plaque agarose in deionizedwater at 55infinC (Stein et al 1996) The molten mixture wasdrawn into a 1 ml syringe and allowed to solidify The samplewas equilibrated with buffer (10 mM Tris pH 8 50 mM NaCl01 M EDTA 1 Sarkosyl) and digested with 1 mg ml-1
lysozyme at 37infinC for 4 h Buffer and lysozyme were replacedwith 1 Sarkosyl in 05 M EDTA and the sample wasdigested with 1 mg ml-1 proteinase K at 55infinC for 18 h Fol-lowing the enzyme treatments the sample was electrophore-sed for 8 h at 23 V through 03 SeaKem Gold agarose inTAE buffer DNA entering the gel but larger than the 48 kbsize standard was recovered from the gel by electroelutionand concentrated by ethanol precipitation The DNA was thensheared to 40 kb by 150 passages through a 200 ml pipettetip The DNA was prepared for ligation using a blunt endrepair kit (Epicentre Technologies) following the manufac-turerrsquos protocol The insert DNA was then size selected byelectrophoresis on 1 Sea Plaque GTG agarose DNA wasrecovered from the gel following treatment with Gelaseenzyme (Epicentre Technologies) digestion and ethanol pre-cipitation The DNA was then ligated to the pCC1FOS fosmidvector (Epicentre Technologies) and packaged using phageprotein extract following the manufacturerrsquos recommendedprocedure EPI300 host E coli infected with recombinantphage were plated on LB media containing chloramphenicolClones were picked and sorted into 96-well microtitre plates
PCR library construction
Three PCR libraries were constructed with the DNA isolatedfrom the Delaware River Two replicate libraries were con-structed by cloning amplicons generated from separate PCRreactions using the general bacterial primers EubA and EubB(Lane 1991) A third library was constructed using re-condi-tioned amplicons from the PCR reaction used to generate thefirst library Re-conditioning was performed with three roundsof thermal cycling applied to amplicons diluted 10-fold infresh PCR reagents (Thompson et al 2002) The 25 ml PCRreactions included 05 ml 12 ml and 2 ml additions of 10 mMdeoxynucloside triphosphate 25 mM MgCl2 10 mM primer
Metagenomic analysis of freshwater bacteria 1893
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
stocks respectively Bovine serum albumin was added to afinal concentration of 02 mg ml-1 Two and a half units of Taqpolymerase (Promega) and a one-tenth volume of 10yen bufferwere added to each reaction The thermal cycling conditionsconsisted of a touchdown series in which the annealing tem-perature decreased from 65infinC to 55infinC by 1infinC per cyclefollowed by 15 cycles at 55infinC each for 1 min Each cycleincluded a 1 min denaturation step at 95infinC and an extensionstep of 25 min at 72infinC Amplicons were cloned with a TOPO-TA cloning kit with pCR21-TOPO vector (Invitrogen) accord-ing to the manufacturerrsquos instructions
Screening PCR and metagenomic libraries
Clones were screened for insert size using M13 forward andreverse PCR primers which flank the cloned insert andthose with full-length inserts were screened by restrictionfragment length polymorphism (RFLP) analysis by digestingthe M13 PCR amplicons with a mixture of HhaI and RsaI(New England Biolabs) Restriction patterns on 2 Meta-phore agarose (FMC) agarose gels were compared manuallyClones with identical restriction patterns were groupedtogether into RFLP types The 16S rRNA genes from onerepresentative of each RFLP type in the first library werecompletely sequenced
Pools of 96 fosmid clones were screened for 16S rRNAgenes by DGGE of PCR amplicons generated with primersGC358F and 517R (Muyzer et al 1995) Selected bandsresolved on an 8 polyacrylamide gel containing a gradientof 25ndash55 denaturant (138ndash22 formamide and 105ndash23urea) were re-amplified and sequenced Phylogenetic classi-fication was determined using BLAST and the ARB sequenceanalysis tool as described below Bands with equal electro-phoretic mobility were assigned the same phylogeneticclassification
Sequence and library analysis
Nucleotide sequences were analysed using BLAST [version21 National Center for Biotechnology Information (httpwwwncbinlmnihgovBLAST) and the ARB package (Lud-wig et al 2004)] Sequences were aligned with the ARB FastAligner version 103 adjusted manually with secondary-structure criteria Phylogenetic relationships were deter-mined using the Quick Add Parsimony tool and the1000_pub_may02 phylogenetic tree in the ARB database(June 2002 release)
Sequence similarities were determined using the tools inARB and rarefaction curves were calculated using AnalyticRarefaction 13 (httpwwwugaedustratasoftware) Librar-ies were compared using the LIBSHUFF analysis tool(httpwwwarchesugaedu~whitmanlibshuffhtml) (Single-ton et al 2001) Diversity indices were calculated usingequations taken from Dunbar and colleagues (1999)
Statistical analysis
Libraries were compared using LIBSHUFF analysis (Single-ton et al 2001) This approach compares libraries based ontheir coverage and the number of sequences in one library
that are not found in the second library This so-called lsquohet-erologous coveragersquo is calculated over a range of evolution-ary distances to obtain a coverage curve The differencebetween the lsquoheterologous coveragersquo curve and the coveragecurve of the first library is calculated using the Crameacuter-vonMises test statistic (Pettit 1982) and compared with the cov-erage curve calculated after shuffling sequences between thetwo libraries
Nucleotide sequence accession numbers
The nucleotide sequence data reported in this work aredeposited in GenBank under accession numbers AY562245ndashAY562366
Acknowledgements
This work was supported by the US Department of EnergyMicrobial Genomes Program and the National ScienceFoundation
References
Amann RI Binder BJ Olson RJ Chisholm SWDevereux R and Stahl DA (1990) Combination of 16Sribosomal-RNA-targeted oligonucleotide probes with flowcytometry for analyzing mixed microbial populations ApplEnviron Microbiol 56 1919ndash1925
Amon RMW and Benner R (1996) Photochemical andmicrobial consumption of dissolved organic carbon anddissolved oxygen in the Amazon River system GeochimCosmochim Ac 60 1783ndash1792
Beja O Aravind L Koonin EV Suzuki MT Hadd ANguyen LP et al (2000a) Bacterial rhodopsin evidencefor a new type of phototrophy in the sea Science 2891902ndash1906
Beja O Suzuki MT Koonin EV Aravind L Hadd ANguyen LP et al (2000b) Construction and analysis ofbacterial artificial chromosome libraries from a marinemicrobial assemblage Environ Microbiol 2 516ndash529
Beja O Spudich EN Spudich JL Leclerc M andDeLong EF (2001) Proteorhodopsin phototrophy in theocean Nature 411 786ndash789
Button DK and Robertson BR (2001) Determination ofDNA content of aquatic bacteria by flow cytometry ApplEnviron Microbiol 67 1636ndash1645
Button DK Robertson BR Lepp PW and SchmidtTM (1998) A small dilute-cytoplasm high-affinity novelbacterium isolated by extinction culture and having kineticconstants compatible with growth at ambient concentra-tions of dissolved nutrients in seawater Appl EnvironMicrobiol 64 4467ndash4476
Castle D and Kirchman DL (2004) Composition of estu-arine bacterial communities assessed by denaturing gradi-ent gel electrophoresis and fluorescence in situhybridization Limnol Oceanogr 2 303ndash314
Cottrell MT and Kirchman DL (2000) Community compo-sition of marine bacterioplankton determined by 16S rRNAgene clone libraries and fluorescence in situ hybridizationAppl Environ Microbiol 66 5116ndash5122
1894 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Cottrell MT and Kirchman DL (2003) Contribution ofmajor bacterial groups to bacterial biomass production(thymidine and leucine incorporation) in the Delaware estu-ary Limnol Oceanogr 48 168ndash178
Cottrell MT Moore JA and Kirchman DL (1999) Chiti-nases from uncultured marine microorganisms Appl Envi-ron Microbiol 65 2553ndash2557
Covert JS and Moran MA (2001) Molecular characteriza-tion of estuarine bacterial communities that use high- andlow-molecular weight fractions of dissolved organic carbonAquat Microb Ecol 25 127ndash139
Daims H Bruhl A Amann R Schleifer KH and WagnerM (1999) The domain-specific probe EUB338 is insuffi-cient for the detection of all Bacteria Development andevaluation of a more comprehensive probe set Syst ApplMicrobiol 22 434ndash444
Dunbar J Takala S Barns SM Davis JA and KuskeCR (1999) Levels of bacterial community diversity in fourarid soils compared by cultivation and 16S rRNA genecloning Appl Environ Microbiol 65 1662ndash1669
Fegatella F Lim J Kjelleberg S and Cavicchioli R(1998) Implications of rRNA operon copy number and ribo-some content in the marine oligotrophic ultramicrobacte-rium Sphingomonas sp strain RB2256 Appl EnvironMicrobiol 64 4433ndash4438
Fogel GB Collins CR Li J and Brunk CF (1999)Prokaryotic genome size and SSU rDNA copy numberEstimation of microbial relative abundance from a mixedpopulation Microb Ecol 38 93ndash113
Fuhrman JA McCallum K and Davis AA (1992) Novelmajor Archaebacterial group from marine plankton Nature356 148ndash149
Gillespie DE Brady SF Bettermann AD CianciottoNP Liles MR Rondon MR et al (2002) Isolation ofantibiotics turbomycin A and B from a metagenomic libraryof soil microbial DNA Appl Environ Microbiol 68 4301ndash4306
Giovannoni SJ and Rappeacute MS (2000) Evolution diversityand molecular ecology of marine prokaryotes In MicrobialEcology of the Oceans Kirchman DL (ed) New YorkUSA Wiley-Liss pp 47ndash84
Gloumlckner FO Fuchs BM and Amann R (1999) Bacteri-oplankton compositions of lakes and oceans a first com-parison based on fluorescence in situ hybridization ApplEnviron Microbiol 65 3721ndash3726
Gloumlckner FO Zaichikov E Belkova N Denissova LPernthaler J Pernthaler A and Amann R (2000) Com-parative 16S rRNA analysis of lake bacterioplanktonreveals globally distributed phylogenetic clusters includingan abundant group of actinobacteria Appl Environ Micro-biol 66 5053ndash5065
Gonzalez JM and Moran MA (1997) Numerical domi-nance of a group of marine bacteria in the alpha- subclassof the class Proteobacteria in coastal seawater Appl Envi-ron Microbiol 63 4237ndash4242
Hernes PJ and Benner R (2002) Transport and diagene-sis of dissolved and particulate terrigenous organic matterin the North Pacific Ocean Deep-Sea Res Part II ndash TopStud Oceanogr 49 2119ndash2132
Karner M and Fuhrman JA (1997) Determination of activemarine bacterioplankton a comparison of universal 16S
rRNA probes autoradiography and nucleoid staining ApplEnviron Microbiol 63 1208ndash1213
Karner MB DeLong EF and Karl DM (2001) Archaealdominance in the mesopelagic zone of the Pacific OceanNature 409 507ndash510
Kirchman DL (2002) The ecology of CytophagandashFlavobac-teria in aquatic environments FEMS Microbiol Ecol 3991ndash100
Kirchman DL Yu LY and Cottrell MT (2003) Diversityand abundance of uncultured Cytophaga-like bacteria inthe Delaware Estuary Appl Environ Microbiol 69 6587ndash6596
Klappenbach JA Dunbar JM and Schmidt TM (2000)rRNA operon copy number reflects ecological strategies ofbacteria Appl Environ Microbiol 66 1328ndash1333
Knietsch A Waschkowitz T Bowien S Henne A andDaniel R (2003) Metagenomes of complex microbial con-sortia derived from different soils as sources for novelgenes conferring formation of carbonyls from short-chainpolyols on Escherichia coli J Mol Microbiol Biotechnol 546ndash56
Lane DJ (1991) 16S23S rRNA sequencing In Nucleic AcidTechniques in Bacterial Systematics Stackebrandt E andGoodfellow M (eds) Chichester UK Wiley and Sonspp 115ndash175
Liles MR Manske BF Bintrim SB Handelsman J andGoodman RM (2003) A census of rRNA genes and linkedgenomic sequences within a soil metagenomic libraryAppl Environ Microbiol 69 2684ndash2691
Ludwig W Strunk O Westram R Richter L Meier HYadhukumar et al (2004) ARB a software environmentfor sequence data Nucleic Acids Res 32 1363ndash1371
Manz W Amann R Ludwig W Wagner M and Schlei-fer KH (1992) Phylogenetic oligodeoxynucleotide probesfor the major subclasses of Proteobacteria ndash problems andsolutions Syst Appl Microbiol 15 593ndash600
Manz W Amann R Ludwig W Vancanneyt M andSchleifer KH (1996) Application of a suite of 16S rRNA-specific oligonucleotide probes designed to investigatebacteria of the phylum CytophagandashFlavobacterndashBacteroi-des in the natural environment Microbiology 142 1097ndash1106
Massana R Taylor LJ Murray AE Wu KY JeffreyWH and DeLong EF (1998) Vertical distribution andtemporal variation of marine planktonic archaea in theGerlache Strait Antarctica during early spring LimnolOceanogr 43 607ndash617
Massana R DeLong EF and Pedros-Alio C (2000) A fewcosmopolitan phylotypes dominate planktonic archaealassemblages in widely different oceanic provinces ApplEnviron Microbiol 66 1777ndash1787
Morris RM Rappeacute MS Connon SA Vergin KL Sie-bold WA Carlson CA and Giovannoni SJ (2002)SAR11 clade dominates ocean surface bacterioplanktoncommunities Nature 420 806ndash810
Muyzer G Teske A Wirsen CO and Jannasch HW(1995) Phylogenetic-relationships of Thiomicrospira spe-cies and their identification in deep-sea hydrothermal ventsamples by denaturing gradient gel-electrophoresis of 16SrDNA fragments Arch Microbiol 164 165ndash172
Neef A (1997) Anwendung der in situ-Einzelzell-
Metagenomic analysis of freshwater bacteria 1895
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Identifizierung von Bakterien zur Populationsanalyse inKomplexen Mikrobiellen Biozonosen Munich GermanyTechnische Universitat Munchen
Opsahl S Benner R and Amon RMW (1999) Major fluxof terrigenous dissolved organic matter through the ArcticOcean Limnol Oceanogr 44 2017ndash2023
OrsquoSullivan LA Weightman AJ and Fry JC (2002) Newdegenerate Cytophaga-Flexibacter-Bacteroides-specific16S ribosomal DNA-targeted oligonucleotide probes revealhigh bacterial diversity in River Taff epilithon Appl EnvironMicrobiol 68 201ndash210
Pettit AN (1982) Cramer-von Misess statistic In Encyclo-pedia of Statistical Sciences Kotz S Johnson NL andRead CB (eds) New York USA Wiley pp 220ndash221
Quaiser A Ochsenreiter T Klenk HP Kletzin ATreusch AH Meurer G et al (2002) First insight intothe genome of an uncultivated crenarchaeote from soilEnviron Microbiol 4 603ndash611
Rappeacute MS and Giovannoni SJ (2003) The unculturedmicrobial majority Annu Rev Microbiol 57 369ndash394
Rappeacute MS Gordon DA Vergin KL and GiovannoniSJ (1999) Phylogeny of Actinobacteria small subunit(SSU) rRNA gene clones recovered from marine bacteri-oplankton Syst Appl Microbiol 22 106ndash112
Rappeacute MS Connon SA Vergin KL and GiovannoniSJ (2002) Cultivation of the ubiquitous SAR11 marinebacterioplankton clade Nature 418 630ndash633
Roller C Wagner M Amann R Ludwig W and SchleiferKH (1994) In-situ probing of Gram-positive bacteria withhigh DNA G+C content using 23S-ribosomal-RNA-targetedoligonucleotides Microbiology 140 2849ndash2858
Rondon MR August PR Bettermann AD Brady SFGrossman TH Liles MR et al (2000) Cloning the soilmetagenome a strategy for accessing the genetic andfunctional diversity of uncultured microorganisms ApplEnviron Microbiol 66 2541ndash2547
Sabehi G Massana R Bielawski JP Rosenberg MDelong EF and Beja O (2003) Novel proteorhodopsinvariants from the Mediterranean and Red Seas EnvironMicrobiol 5 842ndash849
Singleton DR Furlong MA Rathbun SL and WhitmanWB (2001) Quantitative comparisons of 16S rRNA genesequence libraries from environmental samples Appl Envi-ron Microbiol 67 4374ndash4376
Stackebrandt E and Goebel BM (1994) A place for DNAndashDNA reassociation and 16S ribosomal-RNA sequence-analysis in the present species definition in bacteriologyInt J Syst Bacteriol 44 846ndash849
Stein JL Marsh TL Wu KY Shizuya H and DeLongEF (1996) Characterization of uncultivated prokaryotesIsolation and analysis of a 40-kilobase-pair genome frag-ment from a planktonic marine archaeon J Bacteriol 178591ndash599
Thompson JR Marcelino LA and Polz MF (2002) Het-eroduplexes in mixed-template amplifications formationconsequence and elimination by lsquoreconditioning PCRrsquoNucleic Acids Res 30 2083ndash2088
Topp E (2003) Bacteria in agricultural soils Diversity roleand future perspectives Can J Soil Sci 83 303ndash309
Weller R Gloumlckner FO and Amann R (2000) 16S rRNA-targeted oligonucleotide probes for the in situ detectionof members of the phylum Cytophaga-Flavobacterium-Bacteroides Syst Appl Microbiol 23 107ndash114
von Wintzingerode F Gobel UB and Stackebrandt E(1997) Determination of microbial diversity in environmen-tal samples pitfalls of PCR-based rRNA analysis FEMSMicrobiol Rev 21 213ndash229
Zaitlin B Watson SB Dixon J and Steel D (2003)Actinomycetes in the Elbow River Basin Alberta CanWater Quality Res J Can 38 115ndash125
Zwart G Hiorns WD Methe BA Van Agterveld MPHuismans R Nold SC et al (1998) Nearly identical 16SrRNA sequences recovered from lakes in North Americaand Europe indicate the existence of clades of globallydistributed freshwater bacteria Syst Appl Microbiol 21546ndash556
Zwart G Crump BC Agterveld M Hagen F and HanSK (2002) Typical freshwater bacteria an analysis ofavailable 16S rRNA gene sequences from plankton oflakes and rivers Aquat Microb Ecol 28 141ndash155
Metagenomic analysis of freshwater bacteria 1889
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
respectively (Fig 4) This distance is approximately equalto 97 sequence identity which is one approximation ofspecies-level identity (Stackebrandt and Goebel 1994)Coverage of the metagenomic library was higher thancoverage of the PCR library at all levels of genetic dis-tance with the greatest differences occurring at smallgenetic distance Coverage decreased rapidly withdecreasing genetic distance and at a genetic distance of001 the coverage of the metagenomic library was 46while the coverage of the PCR library decreased evenmore to just 16
Indices of diversity were higher for the metagenomiclibrary than the PCR library (Table 2) The Simpsonindex was 134 for the metagenomic library comparedwith 87 for the PCR library The difference was not assubstantial for the ShannonndashWeaver index (Table 2)Similarly measures of evenness were higher for themetagenomic library than the PCR library Evenness cal-culated using the ShannonndashWeaver index was 073 and066 for the metagenomic library and PCR libraryrespectively while the Simpson index of evenness was026 and 017 for the metagenomic and PCR librariesrespectively
We examined the possibility that diversity based onPCR clone library composition is overestimated becauseof artefacts of the PCR Re-conditioned PCR was used totest for potential artefacts introduced when heterologousPCR products re-anneal and form heteroduplexes that arethen subjected to mismatch repair upon cloning intoE coli (Thompson et al 2002) Re-conditioning of thePCR amplicons reduced the number of RFLP typesexpected in a library of 80 clones from 50 in the standardPCR library to 40 in the library produced with re-condi-tioned PCR (Fig 5) There was no difference betweenrarefaction curves of replicate PCR libraries produced withstandard PCR (Fig 5)
Comparison of FISH and clone libraries
Fluorescence in situ hybridization was used to assess thecorrespondence between clone library composition andbacterial community structure Microscopic enumerationof bacteria hybridized with fluorescently labelled 16SrRNA directed probes revealed a community dominatedby beta-proteobacteria Actinobacteria and Cytophaga-like bacteria (Table 1) As expected for a freshwater sys-tem Actinobacteria and beta-proteobacteria were themost abundant groups making up about 25 of the com-munity Cytophaga-like bacteria were also prominent
Fig 4 Coverage of the metagenomic and 16S rDNA libraries calcu-lated using the LIBSHUFF tool
Fig 5 Rarefaction curves of 16S rDNA clone libraries constructed with and without re-conditioned PCR PCR 1 and PCR 2 are clone libraries constructed from replicate PCR reactions without re-condi-tioning The error bars are one standard deviation
Table 2 Diversity of 16S rRNA genes in metagenomic and PCR clone libraries from the Delaware River
Library
ShannonndashWeaver Simpson
CoverageDiversity Evenness Diversity Evenness
Metagenomic 29 073 134 026 065PCR 26 066 87 017 052
Diversity indices were calculated using equations taken from Dunbar and colleagues (1999)
1890 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
accounting for 17 plusmn 4 of the community In contrastalpha-proteobacteria and gamma-proteobacteria wereconsiderably less abundant (less than 10) Seventy-seven percent of all prokaryotes were identified with agroup probe and 69 plusmn 11 were detected with the gen-eral bacterial probe Eub338
Bacterial community structure determined by FISH andinferred from the PCR library was surprisingly similar Theabundance of Cytophaga-like bacteria in the PCR librarywas within 25 of the abundance determined by FISHthe abundance of gamma-proteobacteria differed by 3or less and Actinobacteria differed by 38 according tothe two analyses (Table 1) In contrast the abundance ofbeta-proteobacteria clones in the PCR library was twicethe abundance of bacteria detected with the beta-proteo-bacteria FISH probe (Table 1) Three types of bacteriaincluding epsilon-proteobacteria Bdellovibrio and bacte-ria belonging to the candidate division TM6 were presentin the PCR library and would not have been detected withany of the FISH probes used in this study Although wehave no estimate of the abundance of bacteria in thesegroups they likely accounted for some of the 23 ofbacteria not detected by FISH (Table 1)
Discussion
It is necessary to examine whether the phylogenic com-position of metagenomic libraries reflects that of the orig-inal microbial composition in order to effectively linkcommunity structure and function revealed by metage-nomic analysis If metagenomic libraries represent abiased sample of microbial diversity they will not yield anaccurate assessment of metabolic diversity Assumingthey are unbiased large insert metagenomic libraries maybe particularly powerful in examining the relationshipbetween community structure and metabolic capacitiesLarge inserts increase the likelihood of obtaining a clonecontaining both lsquofunctionalrsquo genes (coding for metabolicfunction) and genes with phylogenetic information eg16S rRNA genes Therefore it is necessary to know howthe diversity of 16S rRNA genes sampled by large insertmetagenomic libraries corresponds to actual communitystructure Our main finding was that the composition ofthe metagenomic library at a high phylogenetic level wasnot statistically different from the PCR library but closerexamination revealed noteworthy differences between thelibraries and the community structure determined byFISH The metagenomic library sampled a broader diver-sity of Cytophaga-like bacteria than the PCR library but itmissed some groups of potentially important beta-proteo-bacteria The metagenomic and PCR libraries sampledthe same groups of Actinobacteria but FISH analysissuggested that both types of libraries may have failed toadequately sample Actinobacteria
Metagenomic analysis of Delaware River bacteria pro-vided a view of Cytophaga-like bacterial diversity that wasnot possible with the PCR library and FISH analyses Thenumber of clones representing Cytophaga-like bacteria inthe metagenomic library was threefold higher than theabundance determined by FISH and fourfold higher thanin the PCR library It was not surprising that the metage-nomic library sampled a larger diversity of Cytophaga-likebacteria than the PCR library because previous studieshad suggested that Cytophaga-like bacteria are underrep-resented in 16S rDNA PCR libraries relative to the abun-dance of these bacteria detected by FISH (Cottrell andKirchman 2000 Kirchman et al 2003) It was also notsurprising that the metagenomic library had more clonesfrom Cytophaga-like bacteria than would have beenexpected based on the FISH counts because the generalFISH probe CF319a for Cytophaga-like bacteria does notrecognize all Cytophaga-like bacteria (Weller et al 2000)The abundance of Cytophaga-like bacteria according tothe general FISH probe CFB560 for Cytophaga-like bac-teria (OrsquoSullivan et al 2002) was only 15 higher thanthat determined with CF319a The denaturing gradientelectrophoresis (DGGE) PCR primers used in this studyappear to adequately sample Cytophaga-like bacteria atleast compared with FISH analysis (Castle and Kirchman2004) and these primers apparently recognize a broaderdiversity of Cytophaga-like bacteria than the EubA andEubB primers as some Cytophaga-like bacteria detectedby DGGE in the metagenomic library were not sampledby the PCR library
The Cytophaga-like bacterial group includes a highlydiverse collection of bacteria but not all of this diversitywas present in our Delaware River sample In fact mostof the Cytophaga-like bacteria were most closely relatedto just two genera of Flavobacteriales (Flavobacteriumand Myroides) and a cluster in the Flectobacillus groupThis latter group is distinct from the CytophagandashFlavobac-terium cluster which is synonymous with the Flavobacte-riales (Kirchman 2002) The Delaware River bacteria inthe Flectobacillus group belong to the proposedPRD01a001B cluster of Cytophaga-like bacteria identifiedby Zwart and colleagues (2002) Nine of the 12 DelawareRiver sequences in the Flectobacillus group were greaterthan 98 similar to the PRD01a001B clone from theParker River Massachusetts The proposedPRD01a001B cluster which is currently comprised solelyof uncultivated bacteria has also been found in Adiron-dack lakes Lake Baikal and other freshwater environ-ments (Zwart et al 2002) However this group ofCytophaga-like bacteria has not been detected in allfreshwater environments for example it was not found inthe Columbia River (Zwart et al 2002)
The Delaware River beta-proteobacteria belonged toclusters widely distributed in freshwater systems Three
Metagenomic analysis of freshwater bacteria 1891
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
groups of Delaware River beta-proteobacteria clones (BD and E) were highly similar to bacteria in the freshwaterbeta-Proteobacteria clusters proposed by Zwart and col-leagues (2002) Sequences in Delaware River beta-pro-teobacteria group B were 95ndash99 similar to theRodoferax sp in the Rodoferax sp BAL47 cluster GroupD of Delaware Bay beta-proteobacteria corresponded tothe Polynucleobacter necessarius cluster (88ndash99sequence similarity to P necessarius) and sequences ingroup E probably belong to the proposed LD28 clusterSequences in group D were 89ndash97 similarity to theLD28 clone from Lake Loosdrech in the Netherlands(Zwart et al 1998)
The clusters of freshwater beta-proteobacteria pro-posed by Zwart and colleagues (2002) were more preva-lent in the PCR library than in the metagenomic librarysuggesting that they may represent bacteria that are pref-erentially sampled by PCR libraries Delaware River beta-proteobacteria group B included clones from both themetagenomic library and the PCR library but DelawareRiver groups D and E were comprised of only clones fromthe PCR library Different scenarios could lead to theoverrepresentation of bacterial groups in PCR librariescompared with libraries constructed without a PCR stepincluding selective PCR amplification with general bacte-rial primers differences in rRNA gene copy number orvariation in DNA extraction efficiency (von Wintzingerodeet al 1997) Fluorescence in situ hybridization probes areneeded to determine the abundance of these beta-proteo-bacteria which appear to be in a variety of freshwaterenvironments
There was a high degree of similarity in the Actinobac-teria sampled by the metagenomic and PCR libraries Allthree groups of Actinobacteria identified in this studyincluded clones from the metagenomic and PCR librariesOur results indicate that the presumed thicker cell wallsof these G+ bacteria do not interfere with the extraction ofhigh molecular weight DNA from these bacteria Howeverthe overall efficiency of DNA extraction may be low asActinobacteria were underrepresented in the librariescompared with their abundance determined by FISHAlthough the libraries may be missing some Actinobacte-ria they did recover clusters that are apparentlywidespread in freshwater systems Delaware River Acti-nobacteria groups B and C which correspond to the ACK-MI and STA-30 clusters identified by Zwart and colleagues(2002) were identified in the metagenomic and PCRlibraries Clones in cluster B were 92ndash99 similar to theACK-M1 clone and clones in cluster C were 92ndash97similar to the STA-30 clone Delaware River Actinobacte-ria group A which included clones from both libraries didnot overlap with previously described clusters of freshwa-ter Actinobacteria None of the G+ bacterial groups wedetected represented a terrestrial bacterial signal
because of transport of soils into the river Zwart andcolleagues (2002) also did not find evidence for an impor-tant contribution of soil bacteria to bacterial communitiesin rivers
Coverage in the metagenomic library was consistentlyhigher than in the PCR library over a broad range ofevolutionary distances although substantially higher cov-erage by the metagenomic library was restricted to evo-lutionary distances less than 005 Differences incoverage were not the result of different sequencelengths we analysed 1400 bp for the PCR clones but onlythe highly variable V3 region of the 16S rRNA gene forthe metagenomic clones LIBSHUFF analysis conductedwith the V3 region alone also revealed higher coverage bythe metagenomic library (data not shown) Differences incoverage may reflect more comprehensive sampling ofbacterial diversity by the metagenomic library
Lower coverage by the PCR library was consistent withour hypothesis that PCR artefacts lead to overestimatesof diversity Polymerase chain reaction re-conditioning flat-tened the rarefaction curve but the effect was only 20not as pronounced as would be expected if heterodu-plexes were a substantial source of diversity in the PCRlibraries The higher diversity in the metagenomic librarycompared with the PCR library is also consistent with theidea that the PCR libraries do not overestimate bacterialdiversity In fact they probably underestimate diversitybecause they miss certain groups such as Cytophaga-like bacteria (Cottrell and Kirchman 2000) Additional fac-tors such as cloning biases variation in genome size(Fogel et al 1999) rRNA gene copy number (Klappen-bach et al 2000) and ribosome content (Fegatella et al1998) could have resulted in larger differences than weobserved between the two library approaches and FISHGiven the potential sources of differences the degree ofsimilarity between the different library and FISHapproaches was higher than expected
These results and other suggest that the identities ofprokaryotes in aquatic environments are no longer com-pletely hidden in the microbial lsquoblack boxrsquo With the aid ofvarious culture-independent approaches it is now clearthat various proteobacterial groups Cytophaga-like bac-teria and Actinobacteria are the dominant groups of bac-teria in surface waters of aquatic systems (Giovannoniand Rappeacute 2000 Rappeacute and Giovannoni 2003)Archaea make a substantial contribution to prokaryoticdiversity in the ocean depths well below the euphotic zone(Massana et al 1998 Karner et al 2001) Althoughpotentially these groups could contain many subgroupsat least in marine waters it appears that only a few cladesdominate including alpha-proteobacteria belonging to theSAR11 group (Morris et al 2002 Rappeacute et al 2002) andRoseobacter spp (Gonzalez and Moran 1997) Cytoph-aga-like bacteria in Delaware cluster 2 (Kirchman et al
1892 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
2003) the marine Actinobacteria group (Rappeacute et al1999) and Archaea in group I and group II (Fuhrman et al1992 Massana et al 2000 Karner et al 2001)
Dominant groups in freshwater are also becomingapparent Our study revealed several widespread clustersof freshwater bacteria in the Delaware River that wereidentified in the metagenomic and PCR libraries Theseclusters are obvious candidates for further investigationincluding additional metagenomic analysis to exploretheir metabolic potential It is unclear if clusters such asthe P necessarius and LD28 beta-proteobacteria clus-ters which occurred in the PCR library but not themetagenomic library are selectively amplified by the PCRor for some reason are missed by the metagenomic clon-ing Additional examination by FISH could help resolvethe numerical importance of these freshwater bacteriaStudies such as these will further advance our under-standing of aquatic bacterial diversity as it evolves from aview of apparently intractable diversity towards whatappears to be a manageable number of dominant bacte-rial groups
Experimental procedures
Sample collection
A water sample was collected from a depth of 1 m in theDelaware River near Trenton New Jersey (40infin77cent-N74infin493cent-W) in December 2001 and subsampled for the anal-yses performed in this study
Fluorescence in situ hybridization
The sample for FISH analysis was fixed with 2 paraformal-dehyde at 4infinC for 12ndash18 h Bacteria were then collected on02 mm polycarbonate filters rinsed twice with deionizedwater and the filters were stored at -20infinC Bacteria weredetected with FISH probe Eub338 (Amann et al 1990) Wealso tried a mixture of general probes for Bacteria (Daimset al 1999) and found that it detected 70 plusmn 5 of DAPIpositive bacteria compared with 69 plusmn 11 for Eub338 aloneThe FISH probes for alpha- beta- and gamma-proteobacte-ria were Alf968 (Neef 1997 Gloumlckner et al 1999) Bet42aand Gam42a (Manz et al 1992) respectively Cytophaga-likebacteria were assayed with probe CF319a (Manz et al1996) and CFB560 (OrsquoSullivan et al 2002) and Actinobac-teria were enumerated using probe HGC69a (Roller et al1994) Non-specific binding was assessed using a negativecontrol probe (Karner and Fuhrman 1997) Hybridization ofCy3-labelled oligonucleotide probes was performed with por-tions of polycarbonate filters incubated in a hybridizationbuffer containing formamide to control the stringency ofhybridization (Cottrell and Kirchman 2000) Unbound probewas removed by washing the sample with a wash buffer at48infinC The filter piece was then rinsed in water and 70ethanol The air-dried sample was mounted on a glass slidewith a coverslip using an antifade mountant containing2 mg ml-1 DAPI Probe-positive bacteria (500ndash1000 bacteria
for each FISH probe) were enumerated using semiautomatedfluorescence microscopy (Cottrell and Kirchman 2003)
Preparation of the bacterial size fraction
Environmental DNA was extracted from the bacterial sizefraction obtained by pumping river water (500 l) sequentiallythrough a 1 mm nominal pore-size polypropylene string-wound filter (Cole Parmer) and a 08 mm polycarbonate filter(Nuclepore) Bacteria were collected from the filtrate by tan-gential flow filtration using a 01 mm hollow fibre filter (AGTechnology) and an Amicon DC10 gear pump The samplewas concentrated to 2 l rinsed with a buffer (05 M NaCl01 M EDTA 10 mM Tris pH 80) and stored frozen
Library construction using a fosmid vector
Bacteria were collected from thawed bacterial concentrate bycentrifugation in an SS34 rotor at 16 000 rpm for 30 minThe bacteria were resuspended in 1 ml of the supernatantand then mixed with 1 Sea Plaque agarose in deionizedwater at 55infinC (Stein et al 1996) The molten mixture wasdrawn into a 1 ml syringe and allowed to solidify The samplewas equilibrated with buffer (10 mM Tris pH 8 50 mM NaCl01 M EDTA 1 Sarkosyl) and digested with 1 mg ml-1
lysozyme at 37infinC for 4 h Buffer and lysozyme were replacedwith 1 Sarkosyl in 05 M EDTA and the sample wasdigested with 1 mg ml-1 proteinase K at 55infinC for 18 h Fol-lowing the enzyme treatments the sample was electrophore-sed for 8 h at 23 V through 03 SeaKem Gold agarose inTAE buffer DNA entering the gel but larger than the 48 kbsize standard was recovered from the gel by electroelutionand concentrated by ethanol precipitation The DNA was thensheared to 40 kb by 150 passages through a 200 ml pipettetip The DNA was prepared for ligation using a blunt endrepair kit (Epicentre Technologies) following the manufac-turerrsquos protocol The insert DNA was then size selected byelectrophoresis on 1 Sea Plaque GTG agarose DNA wasrecovered from the gel following treatment with Gelaseenzyme (Epicentre Technologies) digestion and ethanol pre-cipitation The DNA was then ligated to the pCC1FOS fosmidvector (Epicentre Technologies) and packaged using phageprotein extract following the manufacturerrsquos recommendedprocedure EPI300 host E coli infected with recombinantphage were plated on LB media containing chloramphenicolClones were picked and sorted into 96-well microtitre plates
PCR library construction
Three PCR libraries were constructed with the DNA isolatedfrom the Delaware River Two replicate libraries were con-structed by cloning amplicons generated from separate PCRreactions using the general bacterial primers EubA and EubB(Lane 1991) A third library was constructed using re-condi-tioned amplicons from the PCR reaction used to generate thefirst library Re-conditioning was performed with three roundsof thermal cycling applied to amplicons diluted 10-fold infresh PCR reagents (Thompson et al 2002) The 25 ml PCRreactions included 05 ml 12 ml and 2 ml additions of 10 mMdeoxynucloside triphosphate 25 mM MgCl2 10 mM primer
Metagenomic analysis of freshwater bacteria 1893
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
stocks respectively Bovine serum albumin was added to afinal concentration of 02 mg ml-1 Two and a half units of Taqpolymerase (Promega) and a one-tenth volume of 10yen bufferwere added to each reaction The thermal cycling conditionsconsisted of a touchdown series in which the annealing tem-perature decreased from 65infinC to 55infinC by 1infinC per cyclefollowed by 15 cycles at 55infinC each for 1 min Each cycleincluded a 1 min denaturation step at 95infinC and an extensionstep of 25 min at 72infinC Amplicons were cloned with a TOPO-TA cloning kit with pCR21-TOPO vector (Invitrogen) accord-ing to the manufacturerrsquos instructions
Screening PCR and metagenomic libraries
Clones were screened for insert size using M13 forward andreverse PCR primers which flank the cloned insert andthose with full-length inserts were screened by restrictionfragment length polymorphism (RFLP) analysis by digestingthe M13 PCR amplicons with a mixture of HhaI and RsaI(New England Biolabs) Restriction patterns on 2 Meta-phore agarose (FMC) agarose gels were compared manuallyClones with identical restriction patterns were groupedtogether into RFLP types The 16S rRNA genes from onerepresentative of each RFLP type in the first library werecompletely sequenced
Pools of 96 fosmid clones were screened for 16S rRNAgenes by DGGE of PCR amplicons generated with primersGC358F and 517R (Muyzer et al 1995) Selected bandsresolved on an 8 polyacrylamide gel containing a gradientof 25ndash55 denaturant (138ndash22 formamide and 105ndash23urea) were re-amplified and sequenced Phylogenetic classi-fication was determined using BLAST and the ARB sequenceanalysis tool as described below Bands with equal electro-phoretic mobility were assigned the same phylogeneticclassification
Sequence and library analysis
Nucleotide sequences were analysed using BLAST [version21 National Center for Biotechnology Information (httpwwwncbinlmnihgovBLAST) and the ARB package (Lud-wig et al 2004)] Sequences were aligned with the ARB FastAligner version 103 adjusted manually with secondary-structure criteria Phylogenetic relationships were deter-mined using the Quick Add Parsimony tool and the1000_pub_may02 phylogenetic tree in the ARB database(June 2002 release)
Sequence similarities were determined using the tools inARB and rarefaction curves were calculated using AnalyticRarefaction 13 (httpwwwugaedustratasoftware) Librar-ies were compared using the LIBSHUFF analysis tool(httpwwwarchesugaedu~whitmanlibshuffhtml) (Single-ton et al 2001) Diversity indices were calculated usingequations taken from Dunbar and colleagues (1999)
Statistical analysis
Libraries were compared using LIBSHUFF analysis (Single-ton et al 2001) This approach compares libraries based ontheir coverage and the number of sequences in one library
that are not found in the second library This so-called lsquohet-erologous coveragersquo is calculated over a range of evolution-ary distances to obtain a coverage curve The differencebetween the lsquoheterologous coveragersquo curve and the coveragecurve of the first library is calculated using the Crameacuter-vonMises test statistic (Pettit 1982) and compared with the cov-erage curve calculated after shuffling sequences between thetwo libraries
Nucleotide sequence accession numbers
The nucleotide sequence data reported in this work aredeposited in GenBank under accession numbers AY562245ndashAY562366
Acknowledgements
This work was supported by the US Department of EnergyMicrobial Genomes Program and the National ScienceFoundation
References
Amann RI Binder BJ Olson RJ Chisholm SWDevereux R and Stahl DA (1990) Combination of 16Sribosomal-RNA-targeted oligonucleotide probes with flowcytometry for analyzing mixed microbial populations ApplEnviron Microbiol 56 1919ndash1925
Amon RMW and Benner R (1996) Photochemical andmicrobial consumption of dissolved organic carbon anddissolved oxygen in the Amazon River system GeochimCosmochim Ac 60 1783ndash1792
Beja O Aravind L Koonin EV Suzuki MT Hadd ANguyen LP et al (2000a) Bacterial rhodopsin evidencefor a new type of phototrophy in the sea Science 2891902ndash1906
Beja O Suzuki MT Koonin EV Aravind L Hadd ANguyen LP et al (2000b) Construction and analysis ofbacterial artificial chromosome libraries from a marinemicrobial assemblage Environ Microbiol 2 516ndash529
Beja O Spudich EN Spudich JL Leclerc M andDeLong EF (2001) Proteorhodopsin phototrophy in theocean Nature 411 786ndash789
Button DK and Robertson BR (2001) Determination ofDNA content of aquatic bacteria by flow cytometry ApplEnviron Microbiol 67 1636ndash1645
Button DK Robertson BR Lepp PW and SchmidtTM (1998) A small dilute-cytoplasm high-affinity novelbacterium isolated by extinction culture and having kineticconstants compatible with growth at ambient concentra-tions of dissolved nutrients in seawater Appl EnvironMicrobiol 64 4467ndash4476
Castle D and Kirchman DL (2004) Composition of estu-arine bacterial communities assessed by denaturing gradi-ent gel electrophoresis and fluorescence in situhybridization Limnol Oceanogr 2 303ndash314
Cottrell MT and Kirchman DL (2000) Community compo-sition of marine bacterioplankton determined by 16S rRNAgene clone libraries and fluorescence in situ hybridizationAppl Environ Microbiol 66 5116ndash5122
1894 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Cottrell MT and Kirchman DL (2003) Contribution ofmajor bacterial groups to bacterial biomass production(thymidine and leucine incorporation) in the Delaware estu-ary Limnol Oceanogr 48 168ndash178
Cottrell MT Moore JA and Kirchman DL (1999) Chiti-nases from uncultured marine microorganisms Appl Envi-ron Microbiol 65 2553ndash2557
Covert JS and Moran MA (2001) Molecular characteriza-tion of estuarine bacterial communities that use high- andlow-molecular weight fractions of dissolved organic carbonAquat Microb Ecol 25 127ndash139
Daims H Bruhl A Amann R Schleifer KH and WagnerM (1999) The domain-specific probe EUB338 is insuffi-cient for the detection of all Bacteria Development andevaluation of a more comprehensive probe set Syst ApplMicrobiol 22 434ndash444
Dunbar J Takala S Barns SM Davis JA and KuskeCR (1999) Levels of bacterial community diversity in fourarid soils compared by cultivation and 16S rRNA genecloning Appl Environ Microbiol 65 1662ndash1669
Fegatella F Lim J Kjelleberg S and Cavicchioli R(1998) Implications of rRNA operon copy number and ribo-some content in the marine oligotrophic ultramicrobacte-rium Sphingomonas sp strain RB2256 Appl EnvironMicrobiol 64 4433ndash4438
Fogel GB Collins CR Li J and Brunk CF (1999)Prokaryotic genome size and SSU rDNA copy numberEstimation of microbial relative abundance from a mixedpopulation Microb Ecol 38 93ndash113
Fuhrman JA McCallum K and Davis AA (1992) Novelmajor Archaebacterial group from marine plankton Nature356 148ndash149
Gillespie DE Brady SF Bettermann AD CianciottoNP Liles MR Rondon MR et al (2002) Isolation ofantibiotics turbomycin A and B from a metagenomic libraryof soil microbial DNA Appl Environ Microbiol 68 4301ndash4306
Giovannoni SJ and Rappeacute MS (2000) Evolution diversityand molecular ecology of marine prokaryotes In MicrobialEcology of the Oceans Kirchman DL (ed) New YorkUSA Wiley-Liss pp 47ndash84
Gloumlckner FO Fuchs BM and Amann R (1999) Bacteri-oplankton compositions of lakes and oceans a first com-parison based on fluorescence in situ hybridization ApplEnviron Microbiol 65 3721ndash3726
Gloumlckner FO Zaichikov E Belkova N Denissova LPernthaler J Pernthaler A and Amann R (2000) Com-parative 16S rRNA analysis of lake bacterioplanktonreveals globally distributed phylogenetic clusters includingan abundant group of actinobacteria Appl Environ Micro-biol 66 5053ndash5065
Gonzalez JM and Moran MA (1997) Numerical domi-nance of a group of marine bacteria in the alpha- subclassof the class Proteobacteria in coastal seawater Appl Envi-ron Microbiol 63 4237ndash4242
Hernes PJ and Benner R (2002) Transport and diagene-sis of dissolved and particulate terrigenous organic matterin the North Pacific Ocean Deep-Sea Res Part II ndash TopStud Oceanogr 49 2119ndash2132
Karner M and Fuhrman JA (1997) Determination of activemarine bacterioplankton a comparison of universal 16S
rRNA probes autoradiography and nucleoid staining ApplEnviron Microbiol 63 1208ndash1213
Karner MB DeLong EF and Karl DM (2001) Archaealdominance in the mesopelagic zone of the Pacific OceanNature 409 507ndash510
Kirchman DL (2002) The ecology of CytophagandashFlavobac-teria in aquatic environments FEMS Microbiol Ecol 3991ndash100
Kirchman DL Yu LY and Cottrell MT (2003) Diversityand abundance of uncultured Cytophaga-like bacteria inthe Delaware Estuary Appl Environ Microbiol 69 6587ndash6596
Klappenbach JA Dunbar JM and Schmidt TM (2000)rRNA operon copy number reflects ecological strategies ofbacteria Appl Environ Microbiol 66 1328ndash1333
Knietsch A Waschkowitz T Bowien S Henne A andDaniel R (2003) Metagenomes of complex microbial con-sortia derived from different soils as sources for novelgenes conferring formation of carbonyls from short-chainpolyols on Escherichia coli J Mol Microbiol Biotechnol 546ndash56
Lane DJ (1991) 16S23S rRNA sequencing In Nucleic AcidTechniques in Bacterial Systematics Stackebrandt E andGoodfellow M (eds) Chichester UK Wiley and Sonspp 115ndash175
Liles MR Manske BF Bintrim SB Handelsman J andGoodman RM (2003) A census of rRNA genes and linkedgenomic sequences within a soil metagenomic libraryAppl Environ Microbiol 69 2684ndash2691
Ludwig W Strunk O Westram R Richter L Meier HYadhukumar et al (2004) ARB a software environmentfor sequence data Nucleic Acids Res 32 1363ndash1371
Manz W Amann R Ludwig W Wagner M and Schlei-fer KH (1992) Phylogenetic oligodeoxynucleotide probesfor the major subclasses of Proteobacteria ndash problems andsolutions Syst Appl Microbiol 15 593ndash600
Manz W Amann R Ludwig W Vancanneyt M andSchleifer KH (1996) Application of a suite of 16S rRNA-specific oligonucleotide probes designed to investigatebacteria of the phylum CytophagandashFlavobacterndashBacteroi-des in the natural environment Microbiology 142 1097ndash1106
Massana R Taylor LJ Murray AE Wu KY JeffreyWH and DeLong EF (1998) Vertical distribution andtemporal variation of marine planktonic archaea in theGerlache Strait Antarctica during early spring LimnolOceanogr 43 607ndash617
Massana R DeLong EF and Pedros-Alio C (2000) A fewcosmopolitan phylotypes dominate planktonic archaealassemblages in widely different oceanic provinces ApplEnviron Microbiol 66 1777ndash1787
Morris RM Rappeacute MS Connon SA Vergin KL Sie-bold WA Carlson CA and Giovannoni SJ (2002)SAR11 clade dominates ocean surface bacterioplanktoncommunities Nature 420 806ndash810
Muyzer G Teske A Wirsen CO and Jannasch HW(1995) Phylogenetic-relationships of Thiomicrospira spe-cies and their identification in deep-sea hydrothermal ventsamples by denaturing gradient gel-electrophoresis of 16SrDNA fragments Arch Microbiol 164 165ndash172
Neef A (1997) Anwendung der in situ-Einzelzell-
Metagenomic analysis of freshwater bacteria 1895
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Identifizierung von Bakterien zur Populationsanalyse inKomplexen Mikrobiellen Biozonosen Munich GermanyTechnische Universitat Munchen
Opsahl S Benner R and Amon RMW (1999) Major fluxof terrigenous dissolved organic matter through the ArcticOcean Limnol Oceanogr 44 2017ndash2023
OrsquoSullivan LA Weightman AJ and Fry JC (2002) Newdegenerate Cytophaga-Flexibacter-Bacteroides-specific16S ribosomal DNA-targeted oligonucleotide probes revealhigh bacterial diversity in River Taff epilithon Appl EnvironMicrobiol 68 201ndash210
Pettit AN (1982) Cramer-von Misess statistic In Encyclo-pedia of Statistical Sciences Kotz S Johnson NL andRead CB (eds) New York USA Wiley pp 220ndash221
Quaiser A Ochsenreiter T Klenk HP Kletzin ATreusch AH Meurer G et al (2002) First insight intothe genome of an uncultivated crenarchaeote from soilEnviron Microbiol 4 603ndash611
Rappeacute MS and Giovannoni SJ (2003) The unculturedmicrobial majority Annu Rev Microbiol 57 369ndash394
Rappeacute MS Gordon DA Vergin KL and GiovannoniSJ (1999) Phylogeny of Actinobacteria small subunit(SSU) rRNA gene clones recovered from marine bacteri-oplankton Syst Appl Microbiol 22 106ndash112
Rappeacute MS Connon SA Vergin KL and GiovannoniSJ (2002) Cultivation of the ubiquitous SAR11 marinebacterioplankton clade Nature 418 630ndash633
Roller C Wagner M Amann R Ludwig W and SchleiferKH (1994) In-situ probing of Gram-positive bacteria withhigh DNA G+C content using 23S-ribosomal-RNA-targetedoligonucleotides Microbiology 140 2849ndash2858
Rondon MR August PR Bettermann AD Brady SFGrossman TH Liles MR et al (2000) Cloning the soilmetagenome a strategy for accessing the genetic andfunctional diversity of uncultured microorganisms ApplEnviron Microbiol 66 2541ndash2547
Sabehi G Massana R Bielawski JP Rosenberg MDelong EF and Beja O (2003) Novel proteorhodopsinvariants from the Mediterranean and Red Seas EnvironMicrobiol 5 842ndash849
Singleton DR Furlong MA Rathbun SL and WhitmanWB (2001) Quantitative comparisons of 16S rRNA genesequence libraries from environmental samples Appl Envi-ron Microbiol 67 4374ndash4376
Stackebrandt E and Goebel BM (1994) A place for DNAndashDNA reassociation and 16S ribosomal-RNA sequence-analysis in the present species definition in bacteriologyInt J Syst Bacteriol 44 846ndash849
Stein JL Marsh TL Wu KY Shizuya H and DeLongEF (1996) Characterization of uncultivated prokaryotesIsolation and analysis of a 40-kilobase-pair genome frag-ment from a planktonic marine archaeon J Bacteriol 178591ndash599
Thompson JR Marcelino LA and Polz MF (2002) Het-eroduplexes in mixed-template amplifications formationconsequence and elimination by lsquoreconditioning PCRrsquoNucleic Acids Res 30 2083ndash2088
Topp E (2003) Bacteria in agricultural soils Diversity roleand future perspectives Can J Soil Sci 83 303ndash309
Weller R Gloumlckner FO and Amann R (2000) 16S rRNA-targeted oligonucleotide probes for the in situ detectionof members of the phylum Cytophaga-Flavobacterium-Bacteroides Syst Appl Microbiol 23 107ndash114
von Wintzingerode F Gobel UB and Stackebrandt E(1997) Determination of microbial diversity in environmen-tal samples pitfalls of PCR-based rRNA analysis FEMSMicrobiol Rev 21 213ndash229
Zaitlin B Watson SB Dixon J and Steel D (2003)Actinomycetes in the Elbow River Basin Alberta CanWater Quality Res J Can 38 115ndash125
Zwart G Hiorns WD Methe BA Van Agterveld MPHuismans R Nold SC et al (1998) Nearly identical 16SrRNA sequences recovered from lakes in North Americaand Europe indicate the existence of clades of globallydistributed freshwater bacteria Syst Appl Microbiol 21546ndash556
Zwart G Crump BC Agterveld M Hagen F and HanSK (2002) Typical freshwater bacteria an analysis ofavailable 16S rRNA gene sequences from plankton oflakes and rivers Aquat Microb Ecol 28 141ndash155
1890 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
accounting for 17 plusmn 4 of the community In contrastalpha-proteobacteria and gamma-proteobacteria wereconsiderably less abundant (less than 10) Seventy-seven percent of all prokaryotes were identified with agroup probe and 69 plusmn 11 were detected with the gen-eral bacterial probe Eub338
Bacterial community structure determined by FISH andinferred from the PCR library was surprisingly similar Theabundance of Cytophaga-like bacteria in the PCR librarywas within 25 of the abundance determined by FISHthe abundance of gamma-proteobacteria differed by 3or less and Actinobacteria differed by 38 according tothe two analyses (Table 1) In contrast the abundance ofbeta-proteobacteria clones in the PCR library was twicethe abundance of bacteria detected with the beta-proteo-bacteria FISH probe (Table 1) Three types of bacteriaincluding epsilon-proteobacteria Bdellovibrio and bacte-ria belonging to the candidate division TM6 were presentin the PCR library and would not have been detected withany of the FISH probes used in this study Although wehave no estimate of the abundance of bacteria in thesegroups they likely accounted for some of the 23 ofbacteria not detected by FISH (Table 1)
Discussion
It is necessary to examine whether the phylogenic com-position of metagenomic libraries reflects that of the orig-inal microbial composition in order to effectively linkcommunity structure and function revealed by metage-nomic analysis If metagenomic libraries represent abiased sample of microbial diversity they will not yield anaccurate assessment of metabolic diversity Assumingthey are unbiased large insert metagenomic libraries maybe particularly powerful in examining the relationshipbetween community structure and metabolic capacitiesLarge inserts increase the likelihood of obtaining a clonecontaining both lsquofunctionalrsquo genes (coding for metabolicfunction) and genes with phylogenetic information eg16S rRNA genes Therefore it is necessary to know howthe diversity of 16S rRNA genes sampled by large insertmetagenomic libraries corresponds to actual communitystructure Our main finding was that the composition ofthe metagenomic library at a high phylogenetic level wasnot statistically different from the PCR library but closerexamination revealed noteworthy differences between thelibraries and the community structure determined byFISH The metagenomic library sampled a broader diver-sity of Cytophaga-like bacteria than the PCR library but itmissed some groups of potentially important beta-proteo-bacteria The metagenomic and PCR libraries sampledthe same groups of Actinobacteria but FISH analysissuggested that both types of libraries may have failed toadequately sample Actinobacteria
Metagenomic analysis of Delaware River bacteria pro-vided a view of Cytophaga-like bacterial diversity that wasnot possible with the PCR library and FISH analyses Thenumber of clones representing Cytophaga-like bacteria inthe metagenomic library was threefold higher than theabundance determined by FISH and fourfold higher thanin the PCR library It was not surprising that the metage-nomic library sampled a larger diversity of Cytophaga-likebacteria than the PCR library because previous studieshad suggested that Cytophaga-like bacteria are underrep-resented in 16S rDNA PCR libraries relative to the abun-dance of these bacteria detected by FISH (Cottrell andKirchman 2000 Kirchman et al 2003) It was also notsurprising that the metagenomic library had more clonesfrom Cytophaga-like bacteria than would have beenexpected based on the FISH counts because the generalFISH probe CF319a for Cytophaga-like bacteria does notrecognize all Cytophaga-like bacteria (Weller et al 2000)The abundance of Cytophaga-like bacteria according tothe general FISH probe CFB560 for Cytophaga-like bac-teria (OrsquoSullivan et al 2002) was only 15 higher thanthat determined with CF319a The denaturing gradientelectrophoresis (DGGE) PCR primers used in this studyappear to adequately sample Cytophaga-like bacteria atleast compared with FISH analysis (Castle and Kirchman2004) and these primers apparently recognize a broaderdiversity of Cytophaga-like bacteria than the EubA andEubB primers as some Cytophaga-like bacteria detectedby DGGE in the metagenomic library were not sampledby the PCR library
The Cytophaga-like bacterial group includes a highlydiverse collection of bacteria but not all of this diversitywas present in our Delaware River sample In fact mostof the Cytophaga-like bacteria were most closely relatedto just two genera of Flavobacteriales (Flavobacteriumand Myroides) and a cluster in the Flectobacillus groupThis latter group is distinct from the CytophagandashFlavobac-terium cluster which is synonymous with the Flavobacte-riales (Kirchman 2002) The Delaware River bacteria inthe Flectobacillus group belong to the proposedPRD01a001B cluster of Cytophaga-like bacteria identifiedby Zwart and colleagues (2002) Nine of the 12 DelawareRiver sequences in the Flectobacillus group were greaterthan 98 similar to the PRD01a001B clone from theParker River Massachusetts The proposedPRD01a001B cluster which is currently comprised solelyof uncultivated bacteria has also been found in Adiron-dack lakes Lake Baikal and other freshwater environ-ments (Zwart et al 2002) However this group ofCytophaga-like bacteria has not been detected in allfreshwater environments for example it was not found inthe Columbia River (Zwart et al 2002)
The Delaware River beta-proteobacteria belonged toclusters widely distributed in freshwater systems Three
Metagenomic analysis of freshwater bacteria 1891
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
groups of Delaware River beta-proteobacteria clones (BD and E) were highly similar to bacteria in the freshwaterbeta-Proteobacteria clusters proposed by Zwart and col-leagues (2002) Sequences in Delaware River beta-pro-teobacteria group B were 95ndash99 similar to theRodoferax sp in the Rodoferax sp BAL47 cluster GroupD of Delaware Bay beta-proteobacteria corresponded tothe Polynucleobacter necessarius cluster (88ndash99sequence similarity to P necessarius) and sequences ingroup E probably belong to the proposed LD28 clusterSequences in group D were 89ndash97 similarity to theLD28 clone from Lake Loosdrech in the Netherlands(Zwart et al 1998)
The clusters of freshwater beta-proteobacteria pro-posed by Zwart and colleagues (2002) were more preva-lent in the PCR library than in the metagenomic librarysuggesting that they may represent bacteria that are pref-erentially sampled by PCR libraries Delaware River beta-proteobacteria group B included clones from both themetagenomic library and the PCR library but DelawareRiver groups D and E were comprised of only clones fromthe PCR library Different scenarios could lead to theoverrepresentation of bacterial groups in PCR librariescompared with libraries constructed without a PCR stepincluding selective PCR amplification with general bacte-rial primers differences in rRNA gene copy number orvariation in DNA extraction efficiency (von Wintzingerodeet al 1997) Fluorescence in situ hybridization probes areneeded to determine the abundance of these beta-proteo-bacteria which appear to be in a variety of freshwaterenvironments
There was a high degree of similarity in the Actinobac-teria sampled by the metagenomic and PCR libraries Allthree groups of Actinobacteria identified in this studyincluded clones from the metagenomic and PCR librariesOur results indicate that the presumed thicker cell wallsof these G+ bacteria do not interfere with the extraction ofhigh molecular weight DNA from these bacteria Howeverthe overall efficiency of DNA extraction may be low asActinobacteria were underrepresented in the librariescompared with their abundance determined by FISHAlthough the libraries may be missing some Actinobacte-ria they did recover clusters that are apparentlywidespread in freshwater systems Delaware River Acti-nobacteria groups B and C which correspond to the ACK-MI and STA-30 clusters identified by Zwart and colleagues(2002) were identified in the metagenomic and PCRlibraries Clones in cluster B were 92ndash99 similar to theACK-M1 clone and clones in cluster C were 92ndash97similar to the STA-30 clone Delaware River Actinobacte-ria group A which included clones from both libraries didnot overlap with previously described clusters of freshwa-ter Actinobacteria None of the G+ bacterial groups wedetected represented a terrestrial bacterial signal
because of transport of soils into the river Zwart andcolleagues (2002) also did not find evidence for an impor-tant contribution of soil bacteria to bacterial communitiesin rivers
Coverage in the metagenomic library was consistentlyhigher than in the PCR library over a broad range ofevolutionary distances although substantially higher cov-erage by the metagenomic library was restricted to evo-lutionary distances less than 005 Differences incoverage were not the result of different sequencelengths we analysed 1400 bp for the PCR clones but onlythe highly variable V3 region of the 16S rRNA gene forthe metagenomic clones LIBSHUFF analysis conductedwith the V3 region alone also revealed higher coverage bythe metagenomic library (data not shown) Differences incoverage may reflect more comprehensive sampling ofbacterial diversity by the metagenomic library
Lower coverage by the PCR library was consistent withour hypothesis that PCR artefacts lead to overestimatesof diversity Polymerase chain reaction re-conditioning flat-tened the rarefaction curve but the effect was only 20not as pronounced as would be expected if heterodu-plexes were a substantial source of diversity in the PCRlibraries The higher diversity in the metagenomic librarycompared with the PCR library is also consistent with theidea that the PCR libraries do not overestimate bacterialdiversity In fact they probably underestimate diversitybecause they miss certain groups such as Cytophaga-like bacteria (Cottrell and Kirchman 2000) Additional fac-tors such as cloning biases variation in genome size(Fogel et al 1999) rRNA gene copy number (Klappen-bach et al 2000) and ribosome content (Fegatella et al1998) could have resulted in larger differences than weobserved between the two library approaches and FISHGiven the potential sources of differences the degree ofsimilarity between the different library and FISHapproaches was higher than expected
These results and other suggest that the identities ofprokaryotes in aquatic environments are no longer com-pletely hidden in the microbial lsquoblack boxrsquo With the aid ofvarious culture-independent approaches it is now clearthat various proteobacterial groups Cytophaga-like bac-teria and Actinobacteria are the dominant groups of bac-teria in surface waters of aquatic systems (Giovannoniand Rappeacute 2000 Rappeacute and Giovannoni 2003)Archaea make a substantial contribution to prokaryoticdiversity in the ocean depths well below the euphotic zone(Massana et al 1998 Karner et al 2001) Althoughpotentially these groups could contain many subgroupsat least in marine waters it appears that only a few cladesdominate including alpha-proteobacteria belonging to theSAR11 group (Morris et al 2002 Rappeacute et al 2002) andRoseobacter spp (Gonzalez and Moran 1997) Cytoph-aga-like bacteria in Delaware cluster 2 (Kirchman et al
1892 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
2003) the marine Actinobacteria group (Rappeacute et al1999) and Archaea in group I and group II (Fuhrman et al1992 Massana et al 2000 Karner et al 2001)
Dominant groups in freshwater are also becomingapparent Our study revealed several widespread clustersof freshwater bacteria in the Delaware River that wereidentified in the metagenomic and PCR libraries Theseclusters are obvious candidates for further investigationincluding additional metagenomic analysis to exploretheir metabolic potential It is unclear if clusters such asthe P necessarius and LD28 beta-proteobacteria clus-ters which occurred in the PCR library but not themetagenomic library are selectively amplified by the PCRor for some reason are missed by the metagenomic clon-ing Additional examination by FISH could help resolvethe numerical importance of these freshwater bacteriaStudies such as these will further advance our under-standing of aquatic bacterial diversity as it evolves from aview of apparently intractable diversity towards whatappears to be a manageable number of dominant bacte-rial groups
Experimental procedures
Sample collection
A water sample was collected from a depth of 1 m in theDelaware River near Trenton New Jersey (40infin77cent-N74infin493cent-W) in December 2001 and subsampled for the anal-yses performed in this study
Fluorescence in situ hybridization
The sample for FISH analysis was fixed with 2 paraformal-dehyde at 4infinC for 12ndash18 h Bacteria were then collected on02 mm polycarbonate filters rinsed twice with deionizedwater and the filters were stored at -20infinC Bacteria weredetected with FISH probe Eub338 (Amann et al 1990) Wealso tried a mixture of general probes for Bacteria (Daimset al 1999) and found that it detected 70 plusmn 5 of DAPIpositive bacteria compared with 69 plusmn 11 for Eub338 aloneThe FISH probes for alpha- beta- and gamma-proteobacte-ria were Alf968 (Neef 1997 Gloumlckner et al 1999) Bet42aand Gam42a (Manz et al 1992) respectively Cytophaga-likebacteria were assayed with probe CF319a (Manz et al1996) and CFB560 (OrsquoSullivan et al 2002) and Actinobac-teria were enumerated using probe HGC69a (Roller et al1994) Non-specific binding was assessed using a negativecontrol probe (Karner and Fuhrman 1997) Hybridization ofCy3-labelled oligonucleotide probes was performed with por-tions of polycarbonate filters incubated in a hybridizationbuffer containing formamide to control the stringency ofhybridization (Cottrell and Kirchman 2000) Unbound probewas removed by washing the sample with a wash buffer at48infinC The filter piece was then rinsed in water and 70ethanol The air-dried sample was mounted on a glass slidewith a coverslip using an antifade mountant containing2 mg ml-1 DAPI Probe-positive bacteria (500ndash1000 bacteria
for each FISH probe) were enumerated using semiautomatedfluorescence microscopy (Cottrell and Kirchman 2003)
Preparation of the bacterial size fraction
Environmental DNA was extracted from the bacterial sizefraction obtained by pumping river water (500 l) sequentiallythrough a 1 mm nominal pore-size polypropylene string-wound filter (Cole Parmer) and a 08 mm polycarbonate filter(Nuclepore) Bacteria were collected from the filtrate by tan-gential flow filtration using a 01 mm hollow fibre filter (AGTechnology) and an Amicon DC10 gear pump The samplewas concentrated to 2 l rinsed with a buffer (05 M NaCl01 M EDTA 10 mM Tris pH 80) and stored frozen
Library construction using a fosmid vector
Bacteria were collected from thawed bacterial concentrate bycentrifugation in an SS34 rotor at 16 000 rpm for 30 minThe bacteria were resuspended in 1 ml of the supernatantand then mixed with 1 Sea Plaque agarose in deionizedwater at 55infinC (Stein et al 1996) The molten mixture wasdrawn into a 1 ml syringe and allowed to solidify The samplewas equilibrated with buffer (10 mM Tris pH 8 50 mM NaCl01 M EDTA 1 Sarkosyl) and digested with 1 mg ml-1
lysozyme at 37infinC for 4 h Buffer and lysozyme were replacedwith 1 Sarkosyl in 05 M EDTA and the sample wasdigested with 1 mg ml-1 proteinase K at 55infinC for 18 h Fol-lowing the enzyme treatments the sample was electrophore-sed for 8 h at 23 V through 03 SeaKem Gold agarose inTAE buffer DNA entering the gel but larger than the 48 kbsize standard was recovered from the gel by electroelutionand concentrated by ethanol precipitation The DNA was thensheared to 40 kb by 150 passages through a 200 ml pipettetip The DNA was prepared for ligation using a blunt endrepair kit (Epicentre Technologies) following the manufac-turerrsquos protocol The insert DNA was then size selected byelectrophoresis on 1 Sea Plaque GTG agarose DNA wasrecovered from the gel following treatment with Gelaseenzyme (Epicentre Technologies) digestion and ethanol pre-cipitation The DNA was then ligated to the pCC1FOS fosmidvector (Epicentre Technologies) and packaged using phageprotein extract following the manufacturerrsquos recommendedprocedure EPI300 host E coli infected with recombinantphage were plated on LB media containing chloramphenicolClones were picked and sorted into 96-well microtitre plates
PCR library construction
Three PCR libraries were constructed with the DNA isolatedfrom the Delaware River Two replicate libraries were con-structed by cloning amplicons generated from separate PCRreactions using the general bacterial primers EubA and EubB(Lane 1991) A third library was constructed using re-condi-tioned amplicons from the PCR reaction used to generate thefirst library Re-conditioning was performed with three roundsof thermal cycling applied to amplicons diluted 10-fold infresh PCR reagents (Thompson et al 2002) The 25 ml PCRreactions included 05 ml 12 ml and 2 ml additions of 10 mMdeoxynucloside triphosphate 25 mM MgCl2 10 mM primer
Metagenomic analysis of freshwater bacteria 1893
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
stocks respectively Bovine serum albumin was added to afinal concentration of 02 mg ml-1 Two and a half units of Taqpolymerase (Promega) and a one-tenth volume of 10yen bufferwere added to each reaction The thermal cycling conditionsconsisted of a touchdown series in which the annealing tem-perature decreased from 65infinC to 55infinC by 1infinC per cyclefollowed by 15 cycles at 55infinC each for 1 min Each cycleincluded a 1 min denaturation step at 95infinC and an extensionstep of 25 min at 72infinC Amplicons were cloned with a TOPO-TA cloning kit with pCR21-TOPO vector (Invitrogen) accord-ing to the manufacturerrsquos instructions
Screening PCR and metagenomic libraries
Clones were screened for insert size using M13 forward andreverse PCR primers which flank the cloned insert andthose with full-length inserts were screened by restrictionfragment length polymorphism (RFLP) analysis by digestingthe M13 PCR amplicons with a mixture of HhaI and RsaI(New England Biolabs) Restriction patterns on 2 Meta-phore agarose (FMC) agarose gels were compared manuallyClones with identical restriction patterns were groupedtogether into RFLP types The 16S rRNA genes from onerepresentative of each RFLP type in the first library werecompletely sequenced
Pools of 96 fosmid clones were screened for 16S rRNAgenes by DGGE of PCR amplicons generated with primersGC358F and 517R (Muyzer et al 1995) Selected bandsresolved on an 8 polyacrylamide gel containing a gradientof 25ndash55 denaturant (138ndash22 formamide and 105ndash23urea) were re-amplified and sequenced Phylogenetic classi-fication was determined using BLAST and the ARB sequenceanalysis tool as described below Bands with equal electro-phoretic mobility were assigned the same phylogeneticclassification
Sequence and library analysis
Nucleotide sequences were analysed using BLAST [version21 National Center for Biotechnology Information (httpwwwncbinlmnihgovBLAST) and the ARB package (Lud-wig et al 2004)] Sequences were aligned with the ARB FastAligner version 103 adjusted manually with secondary-structure criteria Phylogenetic relationships were deter-mined using the Quick Add Parsimony tool and the1000_pub_may02 phylogenetic tree in the ARB database(June 2002 release)
Sequence similarities were determined using the tools inARB and rarefaction curves were calculated using AnalyticRarefaction 13 (httpwwwugaedustratasoftware) Librar-ies were compared using the LIBSHUFF analysis tool(httpwwwarchesugaedu~whitmanlibshuffhtml) (Single-ton et al 2001) Diversity indices were calculated usingequations taken from Dunbar and colleagues (1999)
Statistical analysis
Libraries were compared using LIBSHUFF analysis (Single-ton et al 2001) This approach compares libraries based ontheir coverage and the number of sequences in one library
that are not found in the second library This so-called lsquohet-erologous coveragersquo is calculated over a range of evolution-ary distances to obtain a coverage curve The differencebetween the lsquoheterologous coveragersquo curve and the coveragecurve of the first library is calculated using the Crameacuter-vonMises test statistic (Pettit 1982) and compared with the cov-erage curve calculated after shuffling sequences between thetwo libraries
Nucleotide sequence accession numbers
The nucleotide sequence data reported in this work aredeposited in GenBank under accession numbers AY562245ndashAY562366
Acknowledgements
This work was supported by the US Department of EnergyMicrobial Genomes Program and the National ScienceFoundation
References
Amann RI Binder BJ Olson RJ Chisholm SWDevereux R and Stahl DA (1990) Combination of 16Sribosomal-RNA-targeted oligonucleotide probes with flowcytometry for analyzing mixed microbial populations ApplEnviron Microbiol 56 1919ndash1925
Amon RMW and Benner R (1996) Photochemical andmicrobial consumption of dissolved organic carbon anddissolved oxygen in the Amazon River system GeochimCosmochim Ac 60 1783ndash1792
Beja O Aravind L Koonin EV Suzuki MT Hadd ANguyen LP et al (2000a) Bacterial rhodopsin evidencefor a new type of phototrophy in the sea Science 2891902ndash1906
Beja O Suzuki MT Koonin EV Aravind L Hadd ANguyen LP et al (2000b) Construction and analysis ofbacterial artificial chromosome libraries from a marinemicrobial assemblage Environ Microbiol 2 516ndash529
Beja O Spudich EN Spudich JL Leclerc M andDeLong EF (2001) Proteorhodopsin phototrophy in theocean Nature 411 786ndash789
Button DK and Robertson BR (2001) Determination ofDNA content of aquatic bacteria by flow cytometry ApplEnviron Microbiol 67 1636ndash1645
Button DK Robertson BR Lepp PW and SchmidtTM (1998) A small dilute-cytoplasm high-affinity novelbacterium isolated by extinction culture and having kineticconstants compatible with growth at ambient concentra-tions of dissolved nutrients in seawater Appl EnvironMicrobiol 64 4467ndash4476
Castle D and Kirchman DL (2004) Composition of estu-arine bacterial communities assessed by denaturing gradi-ent gel electrophoresis and fluorescence in situhybridization Limnol Oceanogr 2 303ndash314
Cottrell MT and Kirchman DL (2000) Community compo-sition of marine bacterioplankton determined by 16S rRNAgene clone libraries and fluorescence in situ hybridizationAppl Environ Microbiol 66 5116ndash5122
1894 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Cottrell MT and Kirchman DL (2003) Contribution ofmajor bacterial groups to bacterial biomass production(thymidine and leucine incorporation) in the Delaware estu-ary Limnol Oceanogr 48 168ndash178
Cottrell MT Moore JA and Kirchman DL (1999) Chiti-nases from uncultured marine microorganisms Appl Envi-ron Microbiol 65 2553ndash2557
Covert JS and Moran MA (2001) Molecular characteriza-tion of estuarine bacterial communities that use high- andlow-molecular weight fractions of dissolved organic carbonAquat Microb Ecol 25 127ndash139
Daims H Bruhl A Amann R Schleifer KH and WagnerM (1999) The domain-specific probe EUB338 is insuffi-cient for the detection of all Bacteria Development andevaluation of a more comprehensive probe set Syst ApplMicrobiol 22 434ndash444
Dunbar J Takala S Barns SM Davis JA and KuskeCR (1999) Levels of bacterial community diversity in fourarid soils compared by cultivation and 16S rRNA genecloning Appl Environ Microbiol 65 1662ndash1669
Fegatella F Lim J Kjelleberg S and Cavicchioli R(1998) Implications of rRNA operon copy number and ribo-some content in the marine oligotrophic ultramicrobacte-rium Sphingomonas sp strain RB2256 Appl EnvironMicrobiol 64 4433ndash4438
Fogel GB Collins CR Li J and Brunk CF (1999)Prokaryotic genome size and SSU rDNA copy numberEstimation of microbial relative abundance from a mixedpopulation Microb Ecol 38 93ndash113
Fuhrman JA McCallum K and Davis AA (1992) Novelmajor Archaebacterial group from marine plankton Nature356 148ndash149
Gillespie DE Brady SF Bettermann AD CianciottoNP Liles MR Rondon MR et al (2002) Isolation ofantibiotics turbomycin A and B from a metagenomic libraryof soil microbial DNA Appl Environ Microbiol 68 4301ndash4306
Giovannoni SJ and Rappeacute MS (2000) Evolution diversityand molecular ecology of marine prokaryotes In MicrobialEcology of the Oceans Kirchman DL (ed) New YorkUSA Wiley-Liss pp 47ndash84
Gloumlckner FO Fuchs BM and Amann R (1999) Bacteri-oplankton compositions of lakes and oceans a first com-parison based on fluorescence in situ hybridization ApplEnviron Microbiol 65 3721ndash3726
Gloumlckner FO Zaichikov E Belkova N Denissova LPernthaler J Pernthaler A and Amann R (2000) Com-parative 16S rRNA analysis of lake bacterioplanktonreveals globally distributed phylogenetic clusters includingan abundant group of actinobacteria Appl Environ Micro-biol 66 5053ndash5065
Gonzalez JM and Moran MA (1997) Numerical domi-nance of a group of marine bacteria in the alpha- subclassof the class Proteobacteria in coastal seawater Appl Envi-ron Microbiol 63 4237ndash4242
Hernes PJ and Benner R (2002) Transport and diagene-sis of dissolved and particulate terrigenous organic matterin the North Pacific Ocean Deep-Sea Res Part II ndash TopStud Oceanogr 49 2119ndash2132
Karner M and Fuhrman JA (1997) Determination of activemarine bacterioplankton a comparison of universal 16S
rRNA probes autoradiography and nucleoid staining ApplEnviron Microbiol 63 1208ndash1213
Karner MB DeLong EF and Karl DM (2001) Archaealdominance in the mesopelagic zone of the Pacific OceanNature 409 507ndash510
Kirchman DL (2002) The ecology of CytophagandashFlavobac-teria in aquatic environments FEMS Microbiol Ecol 3991ndash100
Kirchman DL Yu LY and Cottrell MT (2003) Diversityand abundance of uncultured Cytophaga-like bacteria inthe Delaware Estuary Appl Environ Microbiol 69 6587ndash6596
Klappenbach JA Dunbar JM and Schmidt TM (2000)rRNA operon copy number reflects ecological strategies ofbacteria Appl Environ Microbiol 66 1328ndash1333
Knietsch A Waschkowitz T Bowien S Henne A andDaniel R (2003) Metagenomes of complex microbial con-sortia derived from different soils as sources for novelgenes conferring formation of carbonyls from short-chainpolyols on Escherichia coli J Mol Microbiol Biotechnol 546ndash56
Lane DJ (1991) 16S23S rRNA sequencing In Nucleic AcidTechniques in Bacterial Systematics Stackebrandt E andGoodfellow M (eds) Chichester UK Wiley and Sonspp 115ndash175
Liles MR Manske BF Bintrim SB Handelsman J andGoodman RM (2003) A census of rRNA genes and linkedgenomic sequences within a soil metagenomic libraryAppl Environ Microbiol 69 2684ndash2691
Ludwig W Strunk O Westram R Richter L Meier HYadhukumar et al (2004) ARB a software environmentfor sequence data Nucleic Acids Res 32 1363ndash1371
Manz W Amann R Ludwig W Wagner M and Schlei-fer KH (1992) Phylogenetic oligodeoxynucleotide probesfor the major subclasses of Proteobacteria ndash problems andsolutions Syst Appl Microbiol 15 593ndash600
Manz W Amann R Ludwig W Vancanneyt M andSchleifer KH (1996) Application of a suite of 16S rRNA-specific oligonucleotide probes designed to investigatebacteria of the phylum CytophagandashFlavobacterndashBacteroi-des in the natural environment Microbiology 142 1097ndash1106
Massana R Taylor LJ Murray AE Wu KY JeffreyWH and DeLong EF (1998) Vertical distribution andtemporal variation of marine planktonic archaea in theGerlache Strait Antarctica during early spring LimnolOceanogr 43 607ndash617
Massana R DeLong EF and Pedros-Alio C (2000) A fewcosmopolitan phylotypes dominate planktonic archaealassemblages in widely different oceanic provinces ApplEnviron Microbiol 66 1777ndash1787
Morris RM Rappeacute MS Connon SA Vergin KL Sie-bold WA Carlson CA and Giovannoni SJ (2002)SAR11 clade dominates ocean surface bacterioplanktoncommunities Nature 420 806ndash810
Muyzer G Teske A Wirsen CO and Jannasch HW(1995) Phylogenetic-relationships of Thiomicrospira spe-cies and their identification in deep-sea hydrothermal ventsamples by denaturing gradient gel-electrophoresis of 16SrDNA fragments Arch Microbiol 164 165ndash172
Neef A (1997) Anwendung der in situ-Einzelzell-
Metagenomic analysis of freshwater bacteria 1895
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Identifizierung von Bakterien zur Populationsanalyse inKomplexen Mikrobiellen Biozonosen Munich GermanyTechnische Universitat Munchen
Opsahl S Benner R and Amon RMW (1999) Major fluxof terrigenous dissolved organic matter through the ArcticOcean Limnol Oceanogr 44 2017ndash2023
OrsquoSullivan LA Weightman AJ and Fry JC (2002) Newdegenerate Cytophaga-Flexibacter-Bacteroides-specific16S ribosomal DNA-targeted oligonucleotide probes revealhigh bacterial diversity in River Taff epilithon Appl EnvironMicrobiol 68 201ndash210
Pettit AN (1982) Cramer-von Misess statistic In Encyclo-pedia of Statistical Sciences Kotz S Johnson NL andRead CB (eds) New York USA Wiley pp 220ndash221
Quaiser A Ochsenreiter T Klenk HP Kletzin ATreusch AH Meurer G et al (2002) First insight intothe genome of an uncultivated crenarchaeote from soilEnviron Microbiol 4 603ndash611
Rappeacute MS and Giovannoni SJ (2003) The unculturedmicrobial majority Annu Rev Microbiol 57 369ndash394
Rappeacute MS Gordon DA Vergin KL and GiovannoniSJ (1999) Phylogeny of Actinobacteria small subunit(SSU) rRNA gene clones recovered from marine bacteri-oplankton Syst Appl Microbiol 22 106ndash112
Rappeacute MS Connon SA Vergin KL and GiovannoniSJ (2002) Cultivation of the ubiquitous SAR11 marinebacterioplankton clade Nature 418 630ndash633
Roller C Wagner M Amann R Ludwig W and SchleiferKH (1994) In-situ probing of Gram-positive bacteria withhigh DNA G+C content using 23S-ribosomal-RNA-targetedoligonucleotides Microbiology 140 2849ndash2858
Rondon MR August PR Bettermann AD Brady SFGrossman TH Liles MR et al (2000) Cloning the soilmetagenome a strategy for accessing the genetic andfunctional diversity of uncultured microorganisms ApplEnviron Microbiol 66 2541ndash2547
Sabehi G Massana R Bielawski JP Rosenberg MDelong EF and Beja O (2003) Novel proteorhodopsinvariants from the Mediterranean and Red Seas EnvironMicrobiol 5 842ndash849
Singleton DR Furlong MA Rathbun SL and WhitmanWB (2001) Quantitative comparisons of 16S rRNA genesequence libraries from environmental samples Appl Envi-ron Microbiol 67 4374ndash4376
Stackebrandt E and Goebel BM (1994) A place for DNAndashDNA reassociation and 16S ribosomal-RNA sequence-analysis in the present species definition in bacteriologyInt J Syst Bacteriol 44 846ndash849
Stein JL Marsh TL Wu KY Shizuya H and DeLongEF (1996) Characterization of uncultivated prokaryotesIsolation and analysis of a 40-kilobase-pair genome frag-ment from a planktonic marine archaeon J Bacteriol 178591ndash599
Thompson JR Marcelino LA and Polz MF (2002) Het-eroduplexes in mixed-template amplifications formationconsequence and elimination by lsquoreconditioning PCRrsquoNucleic Acids Res 30 2083ndash2088
Topp E (2003) Bacteria in agricultural soils Diversity roleand future perspectives Can J Soil Sci 83 303ndash309
Weller R Gloumlckner FO and Amann R (2000) 16S rRNA-targeted oligonucleotide probes for the in situ detectionof members of the phylum Cytophaga-Flavobacterium-Bacteroides Syst Appl Microbiol 23 107ndash114
von Wintzingerode F Gobel UB and Stackebrandt E(1997) Determination of microbial diversity in environmen-tal samples pitfalls of PCR-based rRNA analysis FEMSMicrobiol Rev 21 213ndash229
Zaitlin B Watson SB Dixon J and Steel D (2003)Actinomycetes in the Elbow River Basin Alberta CanWater Quality Res J Can 38 115ndash125
Zwart G Hiorns WD Methe BA Van Agterveld MPHuismans R Nold SC et al (1998) Nearly identical 16SrRNA sequences recovered from lakes in North Americaand Europe indicate the existence of clades of globallydistributed freshwater bacteria Syst Appl Microbiol 21546ndash556
Zwart G Crump BC Agterveld M Hagen F and HanSK (2002) Typical freshwater bacteria an analysis ofavailable 16S rRNA gene sequences from plankton oflakes and rivers Aquat Microb Ecol 28 141ndash155
Metagenomic analysis of freshwater bacteria 1891
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
groups of Delaware River beta-proteobacteria clones (BD and E) were highly similar to bacteria in the freshwaterbeta-Proteobacteria clusters proposed by Zwart and col-leagues (2002) Sequences in Delaware River beta-pro-teobacteria group B were 95ndash99 similar to theRodoferax sp in the Rodoferax sp BAL47 cluster GroupD of Delaware Bay beta-proteobacteria corresponded tothe Polynucleobacter necessarius cluster (88ndash99sequence similarity to P necessarius) and sequences ingroup E probably belong to the proposed LD28 clusterSequences in group D were 89ndash97 similarity to theLD28 clone from Lake Loosdrech in the Netherlands(Zwart et al 1998)
The clusters of freshwater beta-proteobacteria pro-posed by Zwart and colleagues (2002) were more preva-lent in the PCR library than in the metagenomic librarysuggesting that they may represent bacteria that are pref-erentially sampled by PCR libraries Delaware River beta-proteobacteria group B included clones from both themetagenomic library and the PCR library but DelawareRiver groups D and E were comprised of only clones fromthe PCR library Different scenarios could lead to theoverrepresentation of bacterial groups in PCR librariescompared with libraries constructed without a PCR stepincluding selective PCR amplification with general bacte-rial primers differences in rRNA gene copy number orvariation in DNA extraction efficiency (von Wintzingerodeet al 1997) Fluorescence in situ hybridization probes areneeded to determine the abundance of these beta-proteo-bacteria which appear to be in a variety of freshwaterenvironments
There was a high degree of similarity in the Actinobac-teria sampled by the metagenomic and PCR libraries Allthree groups of Actinobacteria identified in this studyincluded clones from the metagenomic and PCR librariesOur results indicate that the presumed thicker cell wallsof these G+ bacteria do not interfere with the extraction ofhigh molecular weight DNA from these bacteria Howeverthe overall efficiency of DNA extraction may be low asActinobacteria were underrepresented in the librariescompared with their abundance determined by FISHAlthough the libraries may be missing some Actinobacte-ria they did recover clusters that are apparentlywidespread in freshwater systems Delaware River Acti-nobacteria groups B and C which correspond to the ACK-MI and STA-30 clusters identified by Zwart and colleagues(2002) were identified in the metagenomic and PCRlibraries Clones in cluster B were 92ndash99 similar to theACK-M1 clone and clones in cluster C were 92ndash97similar to the STA-30 clone Delaware River Actinobacte-ria group A which included clones from both libraries didnot overlap with previously described clusters of freshwa-ter Actinobacteria None of the G+ bacterial groups wedetected represented a terrestrial bacterial signal
because of transport of soils into the river Zwart andcolleagues (2002) also did not find evidence for an impor-tant contribution of soil bacteria to bacterial communitiesin rivers
Coverage in the metagenomic library was consistentlyhigher than in the PCR library over a broad range ofevolutionary distances although substantially higher cov-erage by the metagenomic library was restricted to evo-lutionary distances less than 005 Differences incoverage were not the result of different sequencelengths we analysed 1400 bp for the PCR clones but onlythe highly variable V3 region of the 16S rRNA gene forthe metagenomic clones LIBSHUFF analysis conductedwith the V3 region alone also revealed higher coverage bythe metagenomic library (data not shown) Differences incoverage may reflect more comprehensive sampling ofbacterial diversity by the metagenomic library
Lower coverage by the PCR library was consistent withour hypothesis that PCR artefacts lead to overestimatesof diversity Polymerase chain reaction re-conditioning flat-tened the rarefaction curve but the effect was only 20not as pronounced as would be expected if heterodu-plexes were a substantial source of diversity in the PCRlibraries The higher diversity in the metagenomic librarycompared with the PCR library is also consistent with theidea that the PCR libraries do not overestimate bacterialdiversity In fact they probably underestimate diversitybecause they miss certain groups such as Cytophaga-like bacteria (Cottrell and Kirchman 2000) Additional fac-tors such as cloning biases variation in genome size(Fogel et al 1999) rRNA gene copy number (Klappen-bach et al 2000) and ribosome content (Fegatella et al1998) could have resulted in larger differences than weobserved between the two library approaches and FISHGiven the potential sources of differences the degree ofsimilarity between the different library and FISHapproaches was higher than expected
These results and other suggest that the identities ofprokaryotes in aquatic environments are no longer com-pletely hidden in the microbial lsquoblack boxrsquo With the aid ofvarious culture-independent approaches it is now clearthat various proteobacterial groups Cytophaga-like bac-teria and Actinobacteria are the dominant groups of bac-teria in surface waters of aquatic systems (Giovannoniand Rappeacute 2000 Rappeacute and Giovannoni 2003)Archaea make a substantial contribution to prokaryoticdiversity in the ocean depths well below the euphotic zone(Massana et al 1998 Karner et al 2001) Althoughpotentially these groups could contain many subgroupsat least in marine waters it appears that only a few cladesdominate including alpha-proteobacteria belonging to theSAR11 group (Morris et al 2002 Rappeacute et al 2002) andRoseobacter spp (Gonzalez and Moran 1997) Cytoph-aga-like bacteria in Delaware cluster 2 (Kirchman et al
1892 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
2003) the marine Actinobacteria group (Rappeacute et al1999) and Archaea in group I and group II (Fuhrman et al1992 Massana et al 2000 Karner et al 2001)
Dominant groups in freshwater are also becomingapparent Our study revealed several widespread clustersof freshwater bacteria in the Delaware River that wereidentified in the metagenomic and PCR libraries Theseclusters are obvious candidates for further investigationincluding additional metagenomic analysis to exploretheir metabolic potential It is unclear if clusters such asthe P necessarius and LD28 beta-proteobacteria clus-ters which occurred in the PCR library but not themetagenomic library are selectively amplified by the PCRor for some reason are missed by the metagenomic clon-ing Additional examination by FISH could help resolvethe numerical importance of these freshwater bacteriaStudies such as these will further advance our under-standing of aquatic bacterial diversity as it evolves from aview of apparently intractable diversity towards whatappears to be a manageable number of dominant bacte-rial groups
Experimental procedures
Sample collection
A water sample was collected from a depth of 1 m in theDelaware River near Trenton New Jersey (40infin77cent-N74infin493cent-W) in December 2001 and subsampled for the anal-yses performed in this study
Fluorescence in situ hybridization
The sample for FISH analysis was fixed with 2 paraformal-dehyde at 4infinC for 12ndash18 h Bacteria were then collected on02 mm polycarbonate filters rinsed twice with deionizedwater and the filters were stored at -20infinC Bacteria weredetected with FISH probe Eub338 (Amann et al 1990) Wealso tried a mixture of general probes for Bacteria (Daimset al 1999) and found that it detected 70 plusmn 5 of DAPIpositive bacteria compared with 69 plusmn 11 for Eub338 aloneThe FISH probes for alpha- beta- and gamma-proteobacte-ria were Alf968 (Neef 1997 Gloumlckner et al 1999) Bet42aand Gam42a (Manz et al 1992) respectively Cytophaga-likebacteria were assayed with probe CF319a (Manz et al1996) and CFB560 (OrsquoSullivan et al 2002) and Actinobac-teria were enumerated using probe HGC69a (Roller et al1994) Non-specific binding was assessed using a negativecontrol probe (Karner and Fuhrman 1997) Hybridization ofCy3-labelled oligonucleotide probes was performed with por-tions of polycarbonate filters incubated in a hybridizationbuffer containing formamide to control the stringency ofhybridization (Cottrell and Kirchman 2000) Unbound probewas removed by washing the sample with a wash buffer at48infinC The filter piece was then rinsed in water and 70ethanol The air-dried sample was mounted on a glass slidewith a coverslip using an antifade mountant containing2 mg ml-1 DAPI Probe-positive bacteria (500ndash1000 bacteria
for each FISH probe) were enumerated using semiautomatedfluorescence microscopy (Cottrell and Kirchman 2003)
Preparation of the bacterial size fraction
Environmental DNA was extracted from the bacterial sizefraction obtained by pumping river water (500 l) sequentiallythrough a 1 mm nominal pore-size polypropylene string-wound filter (Cole Parmer) and a 08 mm polycarbonate filter(Nuclepore) Bacteria were collected from the filtrate by tan-gential flow filtration using a 01 mm hollow fibre filter (AGTechnology) and an Amicon DC10 gear pump The samplewas concentrated to 2 l rinsed with a buffer (05 M NaCl01 M EDTA 10 mM Tris pH 80) and stored frozen
Library construction using a fosmid vector
Bacteria were collected from thawed bacterial concentrate bycentrifugation in an SS34 rotor at 16 000 rpm for 30 minThe bacteria were resuspended in 1 ml of the supernatantand then mixed with 1 Sea Plaque agarose in deionizedwater at 55infinC (Stein et al 1996) The molten mixture wasdrawn into a 1 ml syringe and allowed to solidify The samplewas equilibrated with buffer (10 mM Tris pH 8 50 mM NaCl01 M EDTA 1 Sarkosyl) and digested with 1 mg ml-1
lysozyme at 37infinC for 4 h Buffer and lysozyme were replacedwith 1 Sarkosyl in 05 M EDTA and the sample wasdigested with 1 mg ml-1 proteinase K at 55infinC for 18 h Fol-lowing the enzyme treatments the sample was electrophore-sed for 8 h at 23 V through 03 SeaKem Gold agarose inTAE buffer DNA entering the gel but larger than the 48 kbsize standard was recovered from the gel by electroelutionand concentrated by ethanol precipitation The DNA was thensheared to 40 kb by 150 passages through a 200 ml pipettetip The DNA was prepared for ligation using a blunt endrepair kit (Epicentre Technologies) following the manufac-turerrsquos protocol The insert DNA was then size selected byelectrophoresis on 1 Sea Plaque GTG agarose DNA wasrecovered from the gel following treatment with Gelaseenzyme (Epicentre Technologies) digestion and ethanol pre-cipitation The DNA was then ligated to the pCC1FOS fosmidvector (Epicentre Technologies) and packaged using phageprotein extract following the manufacturerrsquos recommendedprocedure EPI300 host E coli infected with recombinantphage were plated on LB media containing chloramphenicolClones were picked and sorted into 96-well microtitre plates
PCR library construction
Three PCR libraries were constructed with the DNA isolatedfrom the Delaware River Two replicate libraries were con-structed by cloning amplicons generated from separate PCRreactions using the general bacterial primers EubA and EubB(Lane 1991) A third library was constructed using re-condi-tioned amplicons from the PCR reaction used to generate thefirst library Re-conditioning was performed with three roundsof thermal cycling applied to amplicons diluted 10-fold infresh PCR reagents (Thompson et al 2002) The 25 ml PCRreactions included 05 ml 12 ml and 2 ml additions of 10 mMdeoxynucloside triphosphate 25 mM MgCl2 10 mM primer
Metagenomic analysis of freshwater bacteria 1893
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
stocks respectively Bovine serum albumin was added to afinal concentration of 02 mg ml-1 Two and a half units of Taqpolymerase (Promega) and a one-tenth volume of 10yen bufferwere added to each reaction The thermal cycling conditionsconsisted of a touchdown series in which the annealing tem-perature decreased from 65infinC to 55infinC by 1infinC per cyclefollowed by 15 cycles at 55infinC each for 1 min Each cycleincluded a 1 min denaturation step at 95infinC and an extensionstep of 25 min at 72infinC Amplicons were cloned with a TOPO-TA cloning kit with pCR21-TOPO vector (Invitrogen) accord-ing to the manufacturerrsquos instructions
Screening PCR and metagenomic libraries
Clones were screened for insert size using M13 forward andreverse PCR primers which flank the cloned insert andthose with full-length inserts were screened by restrictionfragment length polymorphism (RFLP) analysis by digestingthe M13 PCR amplicons with a mixture of HhaI and RsaI(New England Biolabs) Restriction patterns on 2 Meta-phore agarose (FMC) agarose gels were compared manuallyClones with identical restriction patterns were groupedtogether into RFLP types The 16S rRNA genes from onerepresentative of each RFLP type in the first library werecompletely sequenced
Pools of 96 fosmid clones were screened for 16S rRNAgenes by DGGE of PCR amplicons generated with primersGC358F and 517R (Muyzer et al 1995) Selected bandsresolved on an 8 polyacrylamide gel containing a gradientof 25ndash55 denaturant (138ndash22 formamide and 105ndash23urea) were re-amplified and sequenced Phylogenetic classi-fication was determined using BLAST and the ARB sequenceanalysis tool as described below Bands with equal electro-phoretic mobility were assigned the same phylogeneticclassification
Sequence and library analysis
Nucleotide sequences were analysed using BLAST [version21 National Center for Biotechnology Information (httpwwwncbinlmnihgovBLAST) and the ARB package (Lud-wig et al 2004)] Sequences were aligned with the ARB FastAligner version 103 adjusted manually with secondary-structure criteria Phylogenetic relationships were deter-mined using the Quick Add Parsimony tool and the1000_pub_may02 phylogenetic tree in the ARB database(June 2002 release)
Sequence similarities were determined using the tools inARB and rarefaction curves were calculated using AnalyticRarefaction 13 (httpwwwugaedustratasoftware) Librar-ies were compared using the LIBSHUFF analysis tool(httpwwwarchesugaedu~whitmanlibshuffhtml) (Single-ton et al 2001) Diversity indices were calculated usingequations taken from Dunbar and colleagues (1999)
Statistical analysis
Libraries were compared using LIBSHUFF analysis (Single-ton et al 2001) This approach compares libraries based ontheir coverage and the number of sequences in one library
that are not found in the second library This so-called lsquohet-erologous coveragersquo is calculated over a range of evolution-ary distances to obtain a coverage curve The differencebetween the lsquoheterologous coveragersquo curve and the coveragecurve of the first library is calculated using the Crameacuter-vonMises test statistic (Pettit 1982) and compared with the cov-erage curve calculated after shuffling sequences between thetwo libraries
Nucleotide sequence accession numbers
The nucleotide sequence data reported in this work aredeposited in GenBank under accession numbers AY562245ndashAY562366
Acknowledgements
This work was supported by the US Department of EnergyMicrobial Genomes Program and the National ScienceFoundation
References
Amann RI Binder BJ Olson RJ Chisholm SWDevereux R and Stahl DA (1990) Combination of 16Sribosomal-RNA-targeted oligonucleotide probes with flowcytometry for analyzing mixed microbial populations ApplEnviron Microbiol 56 1919ndash1925
Amon RMW and Benner R (1996) Photochemical andmicrobial consumption of dissolved organic carbon anddissolved oxygen in the Amazon River system GeochimCosmochim Ac 60 1783ndash1792
Beja O Aravind L Koonin EV Suzuki MT Hadd ANguyen LP et al (2000a) Bacterial rhodopsin evidencefor a new type of phototrophy in the sea Science 2891902ndash1906
Beja O Suzuki MT Koonin EV Aravind L Hadd ANguyen LP et al (2000b) Construction and analysis ofbacterial artificial chromosome libraries from a marinemicrobial assemblage Environ Microbiol 2 516ndash529
Beja O Spudich EN Spudich JL Leclerc M andDeLong EF (2001) Proteorhodopsin phototrophy in theocean Nature 411 786ndash789
Button DK and Robertson BR (2001) Determination ofDNA content of aquatic bacteria by flow cytometry ApplEnviron Microbiol 67 1636ndash1645
Button DK Robertson BR Lepp PW and SchmidtTM (1998) A small dilute-cytoplasm high-affinity novelbacterium isolated by extinction culture and having kineticconstants compatible with growth at ambient concentra-tions of dissolved nutrients in seawater Appl EnvironMicrobiol 64 4467ndash4476
Castle D and Kirchman DL (2004) Composition of estu-arine bacterial communities assessed by denaturing gradi-ent gel electrophoresis and fluorescence in situhybridization Limnol Oceanogr 2 303ndash314
Cottrell MT and Kirchman DL (2000) Community compo-sition of marine bacterioplankton determined by 16S rRNAgene clone libraries and fluorescence in situ hybridizationAppl Environ Microbiol 66 5116ndash5122
1894 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Cottrell MT and Kirchman DL (2003) Contribution ofmajor bacterial groups to bacterial biomass production(thymidine and leucine incorporation) in the Delaware estu-ary Limnol Oceanogr 48 168ndash178
Cottrell MT Moore JA and Kirchman DL (1999) Chiti-nases from uncultured marine microorganisms Appl Envi-ron Microbiol 65 2553ndash2557
Covert JS and Moran MA (2001) Molecular characteriza-tion of estuarine bacterial communities that use high- andlow-molecular weight fractions of dissolved organic carbonAquat Microb Ecol 25 127ndash139
Daims H Bruhl A Amann R Schleifer KH and WagnerM (1999) The domain-specific probe EUB338 is insuffi-cient for the detection of all Bacteria Development andevaluation of a more comprehensive probe set Syst ApplMicrobiol 22 434ndash444
Dunbar J Takala S Barns SM Davis JA and KuskeCR (1999) Levels of bacterial community diversity in fourarid soils compared by cultivation and 16S rRNA genecloning Appl Environ Microbiol 65 1662ndash1669
Fegatella F Lim J Kjelleberg S and Cavicchioli R(1998) Implications of rRNA operon copy number and ribo-some content in the marine oligotrophic ultramicrobacte-rium Sphingomonas sp strain RB2256 Appl EnvironMicrobiol 64 4433ndash4438
Fogel GB Collins CR Li J and Brunk CF (1999)Prokaryotic genome size and SSU rDNA copy numberEstimation of microbial relative abundance from a mixedpopulation Microb Ecol 38 93ndash113
Fuhrman JA McCallum K and Davis AA (1992) Novelmajor Archaebacterial group from marine plankton Nature356 148ndash149
Gillespie DE Brady SF Bettermann AD CianciottoNP Liles MR Rondon MR et al (2002) Isolation ofantibiotics turbomycin A and B from a metagenomic libraryof soil microbial DNA Appl Environ Microbiol 68 4301ndash4306
Giovannoni SJ and Rappeacute MS (2000) Evolution diversityand molecular ecology of marine prokaryotes In MicrobialEcology of the Oceans Kirchman DL (ed) New YorkUSA Wiley-Liss pp 47ndash84
Gloumlckner FO Fuchs BM and Amann R (1999) Bacteri-oplankton compositions of lakes and oceans a first com-parison based on fluorescence in situ hybridization ApplEnviron Microbiol 65 3721ndash3726
Gloumlckner FO Zaichikov E Belkova N Denissova LPernthaler J Pernthaler A and Amann R (2000) Com-parative 16S rRNA analysis of lake bacterioplanktonreveals globally distributed phylogenetic clusters includingan abundant group of actinobacteria Appl Environ Micro-biol 66 5053ndash5065
Gonzalez JM and Moran MA (1997) Numerical domi-nance of a group of marine bacteria in the alpha- subclassof the class Proteobacteria in coastal seawater Appl Envi-ron Microbiol 63 4237ndash4242
Hernes PJ and Benner R (2002) Transport and diagene-sis of dissolved and particulate terrigenous organic matterin the North Pacific Ocean Deep-Sea Res Part II ndash TopStud Oceanogr 49 2119ndash2132
Karner M and Fuhrman JA (1997) Determination of activemarine bacterioplankton a comparison of universal 16S
rRNA probes autoradiography and nucleoid staining ApplEnviron Microbiol 63 1208ndash1213
Karner MB DeLong EF and Karl DM (2001) Archaealdominance in the mesopelagic zone of the Pacific OceanNature 409 507ndash510
Kirchman DL (2002) The ecology of CytophagandashFlavobac-teria in aquatic environments FEMS Microbiol Ecol 3991ndash100
Kirchman DL Yu LY and Cottrell MT (2003) Diversityand abundance of uncultured Cytophaga-like bacteria inthe Delaware Estuary Appl Environ Microbiol 69 6587ndash6596
Klappenbach JA Dunbar JM and Schmidt TM (2000)rRNA operon copy number reflects ecological strategies ofbacteria Appl Environ Microbiol 66 1328ndash1333
Knietsch A Waschkowitz T Bowien S Henne A andDaniel R (2003) Metagenomes of complex microbial con-sortia derived from different soils as sources for novelgenes conferring formation of carbonyls from short-chainpolyols on Escherichia coli J Mol Microbiol Biotechnol 546ndash56
Lane DJ (1991) 16S23S rRNA sequencing In Nucleic AcidTechniques in Bacterial Systematics Stackebrandt E andGoodfellow M (eds) Chichester UK Wiley and Sonspp 115ndash175
Liles MR Manske BF Bintrim SB Handelsman J andGoodman RM (2003) A census of rRNA genes and linkedgenomic sequences within a soil metagenomic libraryAppl Environ Microbiol 69 2684ndash2691
Ludwig W Strunk O Westram R Richter L Meier HYadhukumar et al (2004) ARB a software environmentfor sequence data Nucleic Acids Res 32 1363ndash1371
Manz W Amann R Ludwig W Wagner M and Schlei-fer KH (1992) Phylogenetic oligodeoxynucleotide probesfor the major subclasses of Proteobacteria ndash problems andsolutions Syst Appl Microbiol 15 593ndash600
Manz W Amann R Ludwig W Vancanneyt M andSchleifer KH (1996) Application of a suite of 16S rRNA-specific oligonucleotide probes designed to investigatebacteria of the phylum CytophagandashFlavobacterndashBacteroi-des in the natural environment Microbiology 142 1097ndash1106
Massana R Taylor LJ Murray AE Wu KY JeffreyWH and DeLong EF (1998) Vertical distribution andtemporal variation of marine planktonic archaea in theGerlache Strait Antarctica during early spring LimnolOceanogr 43 607ndash617
Massana R DeLong EF and Pedros-Alio C (2000) A fewcosmopolitan phylotypes dominate planktonic archaealassemblages in widely different oceanic provinces ApplEnviron Microbiol 66 1777ndash1787
Morris RM Rappeacute MS Connon SA Vergin KL Sie-bold WA Carlson CA and Giovannoni SJ (2002)SAR11 clade dominates ocean surface bacterioplanktoncommunities Nature 420 806ndash810
Muyzer G Teske A Wirsen CO and Jannasch HW(1995) Phylogenetic-relationships of Thiomicrospira spe-cies and their identification in deep-sea hydrothermal ventsamples by denaturing gradient gel-electrophoresis of 16SrDNA fragments Arch Microbiol 164 165ndash172
Neef A (1997) Anwendung der in situ-Einzelzell-
Metagenomic analysis of freshwater bacteria 1895
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Identifizierung von Bakterien zur Populationsanalyse inKomplexen Mikrobiellen Biozonosen Munich GermanyTechnische Universitat Munchen
Opsahl S Benner R and Amon RMW (1999) Major fluxof terrigenous dissolved organic matter through the ArcticOcean Limnol Oceanogr 44 2017ndash2023
OrsquoSullivan LA Weightman AJ and Fry JC (2002) Newdegenerate Cytophaga-Flexibacter-Bacteroides-specific16S ribosomal DNA-targeted oligonucleotide probes revealhigh bacterial diversity in River Taff epilithon Appl EnvironMicrobiol 68 201ndash210
Pettit AN (1982) Cramer-von Misess statistic In Encyclo-pedia of Statistical Sciences Kotz S Johnson NL andRead CB (eds) New York USA Wiley pp 220ndash221
Quaiser A Ochsenreiter T Klenk HP Kletzin ATreusch AH Meurer G et al (2002) First insight intothe genome of an uncultivated crenarchaeote from soilEnviron Microbiol 4 603ndash611
Rappeacute MS and Giovannoni SJ (2003) The unculturedmicrobial majority Annu Rev Microbiol 57 369ndash394
Rappeacute MS Gordon DA Vergin KL and GiovannoniSJ (1999) Phylogeny of Actinobacteria small subunit(SSU) rRNA gene clones recovered from marine bacteri-oplankton Syst Appl Microbiol 22 106ndash112
Rappeacute MS Connon SA Vergin KL and GiovannoniSJ (2002) Cultivation of the ubiquitous SAR11 marinebacterioplankton clade Nature 418 630ndash633
Roller C Wagner M Amann R Ludwig W and SchleiferKH (1994) In-situ probing of Gram-positive bacteria withhigh DNA G+C content using 23S-ribosomal-RNA-targetedoligonucleotides Microbiology 140 2849ndash2858
Rondon MR August PR Bettermann AD Brady SFGrossman TH Liles MR et al (2000) Cloning the soilmetagenome a strategy for accessing the genetic andfunctional diversity of uncultured microorganisms ApplEnviron Microbiol 66 2541ndash2547
Sabehi G Massana R Bielawski JP Rosenberg MDelong EF and Beja O (2003) Novel proteorhodopsinvariants from the Mediterranean and Red Seas EnvironMicrobiol 5 842ndash849
Singleton DR Furlong MA Rathbun SL and WhitmanWB (2001) Quantitative comparisons of 16S rRNA genesequence libraries from environmental samples Appl Envi-ron Microbiol 67 4374ndash4376
Stackebrandt E and Goebel BM (1994) A place for DNAndashDNA reassociation and 16S ribosomal-RNA sequence-analysis in the present species definition in bacteriologyInt J Syst Bacteriol 44 846ndash849
Stein JL Marsh TL Wu KY Shizuya H and DeLongEF (1996) Characterization of uncultivated prokaryotesIsolation and analysis of a 40-kilobase-pair genome frag-ment from a planktonic marine archaeon J Bacteriol 178591ndash599
Thompson JR Marcelino LA and Polz MF (2002) Het-eroduplexes in mixed-template amplifications formationconsequence and elimination by lsquoreconditioning PCRrsquoNucleic Acids Res 30 2083ndash2088
Topp E (2003) Bacteria in agricultural soils Diversity roleand future perspectives Can J Soil Sci 83 303ndash309
Weller R Gloumlckner FO and Amann R (2000) 16S rRNA-targeted oligonucleotide probes for the in situ detectionof members of the phylum Cytophaga-Flavobacterium-Bacteroides Syst Appl Microbiol 23 107ndash114
von Wintzingerode F Gobel UB and Stackebrandt E(1997) Determination of microbial diversity in environmen-tal samples pitfalls of PCR-based rRNA analysis FEMSMicrobiol Rev 21 213ndash229
Zaitlin B Watson SB Dixon J and Steel D (2003)Actinomycetes in the Elbow River Basin Alberta CanWater Quality Res J Can 38 115ndash125
Zwart G Hiorns WD Methe BA Van Agterveld MPHuismans R Nold SC et al (1998) Nearly identical 16SrRNA sequences recovered from lakes in North Americaand Europe indicate the existence of clades of globallydistributed freshwater bacteria Syst Appl Microbiol 21546ndash556
Zwart G Crump BC Agterveld M Hagen F and HanSK (2002) Typical freshwater bacteria an analysis ofavailable 16S rRNA gene sequences from plankton oflakes and rivers Aquat Microb Ecol 28 141ndash155
1892 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
2003) the marine Actinobacteria group (Rappeacute et al1999) and Archaea in group I and group II (Fuhrman et al1992 Massana et al 2000 Karner et al 2001)
Dominant groups in freshwater are also becomingapparent Our study revealed several widespread clustersof freshwater bacteria in the Delaware River that wereidentified in the metagenomic and PCR libraries Theseclusters are obvious candidates for further investigationincluding additional metagenomic analysis to exploretheir metabolic potential It is unclear if clusters such asthe P necessarius and LD28 beta-proteobacteria clus-ters which occurred in the PCR library but not themetagenomic library are selectively amplified by the PCRor for some reason are missed by the metagenomic clon-ing Additional examination by FISH could help resolvethe numerical importance of these freshwater bacteriaStudies such as these will further advance our under-standing of aquatic bacterial diversity as it evolves from aview of apparently intractable diversity towards whatappears to be a manageable number of dominant bacte-rial groups
Experimental procedures
Sample collection
A water sample was collected from a depth of 1 m in theDelaware River near Trenton New Jersey (40infin77cent-N74infin493cent-W) in December 2001 and subsampled for the anal-yses performed in this study
Fluorescence in situ hybridization
The sample for FISH analysis was fixed with 2 paraformal-dehyde at 4infinC for 12ndash18 h Bacteria were then collected on02 mm polycarbonate filters rinsed twice with deionizedwater and the filters were stored at -20infinC Bacteria weredetected with FISH probe Eub338 (Amann et al 1990) Wealso tried a mixture of general probes for Bacteria (Daimset al 1999) and found that it detected 70 plusmn 5 of DAPIpositive bacteria compared with 69 plusmn 11 for Eub338 aloneThe FISH probes for alpha- beta- and gamma-proteobacte-ria were Alf968 (Neef 1997 Gloumlckner et al 1999) Bet42aand Gam42a (Manz et al 1992) respectively Cytophaga-likebacteria were assayed with probe CF319a (Manz et al1996) and CFB560 (OrsquoSullivan et al 2002) and Actinobac-teria were enumerated using probe HGC69a (Roller et al1994) Non-specific binding was assessed using a negativecontrol probe (Karner and Fuhrman 1997) Hybridization ofCy3-labelled oligonucleotide probes was performed with por-tions of polycarbonate filters incubated in a hybridizationbuffer containing formamide to control the stringency ofhybridization (Cottrell and Kirchman 2000) Unbound probewas removed by washing the sample with a wash buffer at48infinC The filter piece was then rinsed in water and 70ethanol The air-dried sample was mounted on a glass slidewith a coverslip using an antifade mountant containing2 mg ml-1 DAPI Probe-positive bacteria (500ndash1000 bacteria
for each FISH probe) were enumerated using semiautomatedfluorescence microscopy (Cottrell and Kirchman 2003)
Preparation of the bacterial size fraction
Environmental DNA was extracted from the bacterial sizefraction obtained by pumping river water (500 l) sequentiallythrough a 1 mm nominal pore-size polypropylene string-wound filter (Cole Parmer) and a 08 mm polycarbonate filter(Nuclepore) Bacteria were collected from the filtrate by tan-gential flow filtration using a 01 mm hollow fibre filter (AGTechnology) and an Amicon DC10 gear pump The samplewas concentrated to 2 l rinsed with a buffer (05 M NaCl01 M EDTA 10 mM Tris pH 80) and stored frozen
Library construction using a fosmid vector
Bacteria were collected from thawed bacterial concentrate bycentrifugation in an SS34 rotor at 16 000 rpm for 30 minThe bacteria were resuspended in 1 ml of the supernatantand then mixed with 1 Sea Plaque agarose in deionizedwater at 55infinC (Stein et al 1996) The molten mixture wasdrawn into a 1 ml syringe and allowed to solidify The samplewas equilibrated with buffer (10 mM Tris pH 8 50 mM NaCl01 M EDTA 1 Sarkosyl) and digested with 1 mg ml-1
lysozyme at 37infinC for 4 h Buffer and lysozyme were replacedwith 1 Sarkosyl in 05 M EDTA and the sample wasdigested with 1 mg ml-1 proteinase K at 55infinC for 18 h Fol-lowing the enzyme treatments the sample was electrophore-sed for 8 h at 23 V through 03 SeaKem Gold agarose inTAE buffer DNA entering the gel but larger than the 48 kbsize standard was recovered from the gel by electroelutionand concentrated by ethanol precipitation The DNA was thensheared to 40 kb by 150 passages through a 200 ml pipettetip The DNA was prepared for ligation using a blunt endrepair kit (Epicentre Technologies) following the manufac-turerrsquos protocol The insert DNA was then size selected byelectrophoresis on 1 Sea Plaque GTG agarose DNA wasrecovered from the gel following treatment with Gelaseenzyme (Epicentre Technologies) digestion and ethanol pre-cipitation The DNA was then ligated to the pCC1FOS fosmidvector (Epicentre Technologies) and packaged using phageprotein extract following the manufacturerrsquos recommendedprocedure EPI300 host E coli infected with recombinantphage were plated on LB media containing chloramphenicolClones were picked and sorted into 96-well microtitre plates
PCR library construction
Three PCR libraries were constructed with the DNA isolatedfrom the Delaware River Two replicate libraries were con-structed by cloning amplicons generated from separate PCRreactions using the general bacterial primers EubA and EubB(Lane 1991) A third library was constructed using re-condi-tioned amplicons from the PCR reaction used to generate thefirst library Re-conditioning was performed with three roundsof thermal cycling applied to amplicons diluted 10-fold infresh PCR reagents (Thompson et al 2002) The 25 ml PCRreactions included 05 ml 12 ml and 2 ml additions of 10 mMdeoxynucloside triphosphate 25 mM MgCl2 10 mM primer
Metagenomic analysis of freshwater bacteria 1893
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
stocks respectively Bovine serum albumin was added to afinal concentration of 02 mg ml-1 Two and a half units of Taqpolymerase (Promega) and a one-tenth volume of 10yen bufferwere added to each reaction The thermal cycling conditionsconsisted of a touchdown series in which the annealing tem-perature decreased from 65infinC to 55infinC by 1infinC per cyclefollowed by 15 cycles at 55infinC each for 1 min Each cycleincluded a 1 min denaturation step at 95infinC and an extensionstep of 25 min at 72infinC Amplicons were cloned with a TOPO-TA cloning kit with pCR21-TOPO vector (Invitrogen) accord-ing to the manufacturerrsquos instructions
Screening PCR and metagenomic libraries
Clones were screened for insert size using M13 forward andreverse PCR primers which flank the cloned insert andthose with full-length inserts were screened by restrictionfragment length polymorphism (RFLP) analysis by digestingthe M13 PCR amplicons with a mixture of HhaI and RsaI(New England Biolabs) Restriction patterns on 2 Meta-phore agarose (FMC) agarose gels were compared manuallyClones with identical restriction patterns were groupedtogether into RFLP types The 16S rRNA genes from onerepresentative of each RFLP type in the first library werecompletely sequenced
Pools of 96 fosmid clones were screened for 16S rRNAgenes by DGGE of PCR amplicons generated with primersGC358F and 517R (Muyzer et al 1995) Selected bandsresolved on an 8 polyacrylamide gel containing a gradientof 25ndash55 denaturant (138ndash22 formamide and 105ndash23urea) were re-amplified and sequenced Phylogenetic classi-fication was determined using BLAST and the ARB sequenceanalysis tool as described below Bands with equal electro-phoretic mobility were assigned the same phylogeneticclassification
Sequence and library analysis
Nucleotide sequences were analysed using BLAST [version21 National Center for Biotechnology Information (httpwwwncbinlmnihgovBLAST) and the ARB package (Lud-wig et al 2004)] Sequences were aligned with the ARB FastAligner version 103 adjusted manually with secondary-structure criteria Phylogenetic relationships were deter-mined using the Quick Add Parsimony tool and the1000_pub_may02 phylogenetic tree in the ARB database(June 2002 release)
Sequence similarities were determined using the tools inARB and rarefaction curves were calculated using AnalyticRarefaction 13 (httpwwwugaedustratasoftware) Librar-ies were compared using the LIBSHUFF analysis tool(httpwwwarchesugaedu~whitmanlibshuffhtml) (Single-ton et al 2001) Diversity indices were calculated usingequations taken from Dunbar and colleagues (1999)
Statistical analysis
Libraries were compared using LIBSHUFF analysis (Single-ton et al 2001) This approach compares libraries based ontheir coverage and the number of sequences in one library
that are not found in the second library This so-called lsquohet-erologous coveragersquo is calculated over a range of evolution-ary distances to obtain a coverage curve The differencebetween the lsquoheterologous coveragersquo curve and the coveragecurve of the first library is calculated using the Crameacuter-vonMises test statistic (Pettit 1982) and compared with the cov-erage curve calculated after shuffling sequences between thetwo libraries
Nucleotide sequence accession numbers
The nucleotide sequence data reported in this work aredeposited in GenBank under accession numbers AY562245ndashAY562366
Acknowledgements
This work was supported by the US Department of EnergyMicrobial Genomes Program and the National ScienceFoundation
References
Amann RI Binder BJ Olson RJ Chisholm SWDevereux R and Stahl DA (1990) Combination of 16Sribosomal-RNA-targeted oligonucleotide probes with flowcytometry for analyzing mixed microbial populations ApplEnviron Microbiol 56 1919ndash1925
Amon RMW and Benner R (1996) Photochemical andmicrobial consumption of dissolved organic carbon anddissolved oxygen in the Amazon River system GeochimCosmochim Ac 60 1783ndash1792
Beja O Aravind L Koonin EV Suzuki MT Hadd ANguyen LP et al (2000a) Bacterial rhodopsin evidencefor a new type of phototrophy in the sea Science 2891902ndash1906
Beja O Suzuki MT Koonin EV Aravind L Hadd ANguyen LP et al (2000b) Construction and analysis ofbacterial artificial chromosome libraries from a marinemicrobial assemblage Environ Microbiol 2 516ndash529
Beja O Spudich EN Spudich JL Leclerc M andDeLong EF (2001) Proteorhodopsin phototrophy in theocean Nature 411 786ndash789
Button DK and Robertson BR (2001) Determination ofDNA content of aquatic bacteria by flow cytometry ApplEnviron Microbiol 67 1636ndash1645
Button DK Robertson BR Lepp PW and SchmidtTM (1998) A small dilute-cytoplasm high-affinity novelbacterium isolated by extinction culture and having kineticconstants compatible with growth at ambient concentra-tions of dissolved nutrients in seawater Appl EnvironMicrobiol 64 4467ndash4476
Castle D and Kirchman DL (2004) Composition of estu-arine bacterial communities assessed by denaturing gradi-ent gel electrophoresis and fluorescence in situhybridization Limnol Oceanogr 2 303ndash314
Cottrell MT and Kirchman DL (2000) Community compo-sition of marine bacterioplankton determined by 16S rRNAgene clone libraries and fluorescence in situ hybridizationAppl Environ Microbiol 66 5116ndash5122
1894 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Cottrell MT and Kirchman DL (2003) Contribution ofmajor bacterial groups to bacterial biomass production(thymidine and leucine incorporation) in the Delaware estu-ary Limnol Oceanogr 48 168ndash178
Cottrell MT Moore JA and Kirchman DL (1999) Chiti-nases from uncultured marine microorganisms Appl Envi-ron Microbiol 65 2553ndash2557
Covert JS and Moran MA (2001) Molecular characteriza-tion of estuarine bacterial communities that use high- andlow-molecular weight fractions of dissolved organic carbonAquat Microb Ecol 25 127ndash139
Daims H Bruhl A Amann R Schleifer KH and WagnerM (1999) The domain-specific probe EUB338 is insuffi-cient for the detection of all Bacteria Development andevaluation of a more comprehensive probe set Syst ApplMicrobiol 22 434ndash444
Dunbar J Takala S Barns SM Davis JA and KuskeCR (1999) Levels of bacterial community diversity in fourarid soils compared by cultivation and 16S rRNA genecloning Appl Environ Microbiol 65 1662ndash1669
Fegatella F Lim J Kjelleberg S and Cavicchioli R(1998) Implications of rRNA operon copy number and ribo-some content in the marine oligotrophic ultramicrobacte-rium Sphingomonas sp strain RB2256 Appl EnvironMicrobiol 64 4433ndash4438
Fogel GB Collins CR Li J and Brunk CF (1999)Prokaryotic genome size and SSU rDNA copy numberEstimation of microbial relative abundance from a mixedpopulation Microb Ecol 38 93ndash113
Fuhrman JA McCallum K and Davis AA (1992) Novelmajor Archaebacterial group from marine plankton Nature356 148ndash149
Gillespie DE Brady SF Bettermann AD CianciottoNP Liles MR Rondon MR et al (2002) Isolation ofantibiotics turbomycin A and B from a metagenomic libraryof soil microbial DNA Appl Environ Microbiol 68 4301ndash4306
Giovannoni SJ and Rappeacute MS (2000) Evolution diversityand molecular ecology of marine prokaryotes In MicrobialEcology of the Oceans Kirchman DL (ed) New YorkUSA Wiley-Liss pp 47ndash84
Gloumlckner FO Fuchs BM and Amann R (1999) Bacteri-oplankton compositions of lakes and oceans a first com-parison based on fluorescence in situ hybridization ApplEnviron Microbiol 65 3721ndash3726
Gloumlckner FO Zaichikov E Belkova N Denissova LPernthaler J Pernthaler A and Amann R (2000) Com-parative 16S rRNA analysis of lake bacterioplanktonreveals globally distributed phylogenetic clusters includingan abundant group of actinobacteria Appl Environ Micro-biol 66 5053ndash5065
Gonzalez JM and Moran MA (1997) Numerical domi-nance of a group of marine bacteria in the alpha- subclassof the class Proteobacteria in coastal seawater Appl Envi-ron Microbiol 63 4237ndash4242
Hernes PJ and Benner R (2002) Transport and diagene-sis of dissolved and particulate terrigenous organic matterin the North Pacific Ocean Deep-Sea Res Part II ndash TopStud Oceanogr 49 2119ndash2132
Karner M and Fuhrman JA (1997) Determination of activemarine bacterioplankton a comparison of universal 16S
rRNA probes autoradiography and nucleoid staining ApplEnviron Microbiol 63 1208ndash1213
Karner MB DeLong EF and Karl DM (2001) Archaealdominance in the mesopelagic zone of the Pacific OceanNature 409 507ndash510
Kirchman DL (2002) The ecology of CytophagandashFlavobac-teria in aquatic environments FEMS Microbiol Ecol 3991ndash100
Kirchman DL Yu LY and Cottrell MT (2003) Diversityand abundance of uncultured Cytophaga-like bacteria inthe Delaware Estuary Appl Environ Microbiol 69 6587ndash6596
Klappenbach JA Dunbar JM and Schmidt TM (2000)rRNA operon copy number reflects ecological strategies ofbacteria Appl Environ Microbiol 66 1328ndash1333
Knietsch A Waschkowitz T Bowien S Henne A andDaniel R (2003) Metagenomes of complex microbial con-sortia derived from different soils as sources for novelgenes conferring formation of carbonyls from short-chainpolyols on Escherichia coli J Mol Microbiol Biotechnol 546ndash56
Lane DJ (1991) 16S23S rRNA sequencing In Nucleic AcidTechniques in Bacterial Systematics Stackebrandt E andGoodfellow M (eds) Chichester UK Wiley and Sonspp 115ndash175
Liles MR Manske BF Bintrim SB Handelsman J andGoodman RM (2003) A census of rRNA genes and linkedgenomic sequences within a soil metagenomic libraryAppl Environ Microbiol 69 2684ndash2691
Ludwig W Strunk O Westram R Richter L Meier HYadhukumar et al (2004) ARB a software environmentfor sequence data Nucleic Acids Res 32 1363ndash1371
Manz W Amann R Ludwig W Wagner M and Schlei-fer KH (1992) Phylogenetic oligodeoxynucleotide probesfor the major subclasses of Proteobacteria ndash problems andsolutions Syst Appl Microbiol 15 593ndash600
Manz W Amann R Ludwig W Vancanneyt M andSchleifer KH (1996) Application of a suite of 16S rRNA-specific oligonucleotide probes designed to investigatebacteria of the phylum CytophagandashFlavobacterndashBacteroi-des in the natural environment Microbiology 142 1097ndash1106
Massana R Taylor LJ Murray AE Wu KY JeffreyWH and DeLong EF (1998) Vertical distribution andtemporal variation of marine planktonic archaea in theGerlache Strait Antarctica during early spring LimnolOceanogr 43 607ndash617
Massana R DeLong EF and Pedros-Alio C (2000) A fewcosmopolitan phylotypes dominate planktonic archaealassemblages in widely different oceanic provinces ApplEnviron Microbiol 66 1777ndash1787
Morris RM Rappeacute MS Connon SA Vergin KL Sie-bold WA Carlson CA and Giovannoni SJ (2002)SAR11 clade dominates ocean surface bacterioplanktoncommunities Nature 420 806ndash810
Muyzer G Teske A Wirsen CO and Jannasch HW(1995) Phylogenetic-relationships of Thiomicrospira spe-cies and their identification in deep-sea hydrothermal ventsamples by denaturing gradient gel-electrophoresis of 16SrDNA fragments Arch Microbiol 164 165ndash172
Neef A (1997) Anwendung der in situ-Einzelzell-
Metagenomic analysis of freshwater bacteria 1895
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Identifizierung von Bakterien zur Populationsanalyse inKomplexen Mikrobiellen Biozonosen Munich GermanyTechnische Universitat Munchen
Opsahl S Benner R and Amon RMW (1999) Major fluxof terrigenous dissolved organic matter through the ArcticOcean Limnol Oceanogr 44 2017ndash2023
OrsquoSullivan LA Weightman AJ and Fry JC (2002) Newdegenerate Cytophaga-Flexibacter-Bacteroides-specific16S ribosomal DNA-targeted oligonucleotide probes revealhigh bacterial diversity in River Taff epilithon Appl EnvironMicrobiol 68 201ndash210
Pettit AN (1982) Cramer-von Misess statistic In Encyclo-pedia of Statistical Sciences Kotz S Johnson NL andRead CB (eds) New York USA Wiley pp 220ndash221
Quaiser A Ochsenreiter T Klenk HP Kletzin ATreusch AH Meurer G et al (2002) First insight intothe genome of an uncultivated crenarchaeote from soilEnviron Microbiol 4 603ndash611
Rappeacute MS and Giovannoni SJ (2003) The unculturedmicrobial majority Annu Rev Microbiol 57 369ndash394
Rappeacute MS Gordon DA Vergin KL and GiovannoniSJ (1999) Phylogeny of Actinobacteria small subunit(SSU) rRNA gene clones recovered from marine bacteri-oplankton Syst Appl Microbiol 22 106ndash112
Rappeacute MS Connon SA Vergin KL and GiovannoniSJ (2002) Cultivation of the ubiquitous SAR11 marinebacterioplankton clade Nature 418 630ndash633
Roller C Wagner M Amann R Ludwig W and SchleiferKH (1994) In-situ probing of Gram-positive bacteria withhigh DNA G+C content using 23S-ribosomal-RNA-targetedoligonucleotides Microbiology 140 2849ndash2858
Rondon MR August PR Bettermann AD Brady SFGrossman TH Liles MR et al (2000) Cloning the soilmetagenome a strategy for accessing the genetic andfunctional diversity of uncultured microorganisms ApplEnviron Microbiol 66 2541ndash2547
Sabehi G Massana R Bielawski JP Rosenberg MDelong EF and Beja O (2003) Novel proteorhodopsinvariants from the Mediterranean and Red Seas EnvironMicrobiol 5 842ndash849
Singleton DR Furlong MA Rathbun SL and WhitmanWB (2001) Quantitative comparisons of 16S rRNA genesequence libraries from environmental samples Appl Envi-ron Microbiol 67 4374ndash4376
Stackebrandt E and Goebel BM (1994) A place for DNAndashDNA reassociation and 16S ribosomal-RNA sequence-analysis in the present species definition in bacteriologyInt J Syst Bacteriol 44 846ndash849
Stein JL Marsh TL Wu KY Shizuya H and DeLongEF (1996) Characterization of uncultivated prokaryotesIsolation and analysis of a 40-kilobase-pair genome frag-ment from a planktonic marine archaeon J Bacteriol 178591ndash599
Thompson JR Marcelino LA and Polz MF (2002) Het-eroduplexes in mixed-template amplifications formationconsequence and elimination by lsquoreconditioning PCRrsquoNucleic Acids Res 30 2083ndash2088
Topp E (2003) Bacteria in agricultural soils Diversity roleand future perspectives Can J Soil Sci 83 303ndash309
Weller R Gloumlckner FO and Amann R (2000) 16S rRNA-targeted oligonucleotide probes for the in situ detectionof members of the phylum Cytophaga-Flavobacterium-Bacteroides Syst Appl Microbiol 23 107ndash114
von Wintzingerode F Gobel UB and Stackebrandt E(1997) Determination of microbial diversity in environmen-tal samples pitfalls of PCR-based rRNA analysis FEMSMicrobiol Rev 21 213ndash229
Zaitlin B Watson SB Dixon J and Steel D (2003)Actinomycetes in the Elbow River Basin Alberta CanWater Quality Res J Can 38 115ndash125
Zwart G Hiorns WD Methe BA Van Agterveld MPHuismans R Nold SC et al (1998) Nearly identical 16SrRNA sequences recovered from lakes in North Americaand Europe indicate the existence of clades of globallydistributed freshwater bacteria Syst Appl Microbiol 21546ndash556
Zwart G Crump BC Agterveld M Hagen F and HanSK (2002) Typical freshwater bacteria an analysis ofavailable 16S rRNA gene sequences from plankton oflakes and rivers Aquat Microb Ecol 28 141ndash155
Metagenomic analysis of freshwater bacteria 1893
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
stocks respectively Bovine serum albumin was added to afinal concentration of 02 mg ml-1 Two and a half units of Taqpolymerase (Promega) and a one-tenth volume of 10yen bufferwere added to each reaction The thermal cycling conditionsconsisted of a touchdown series in which the annealing tem-perature decreased from 65infinC to 55infinC by 1infinC per cyclefollowed by 15 cycles at 55infinC each for 1 min Each cycleincluded a 1 min denaturation step at 95infinC and an extensionstep of 25 min at 72infinC Amplicons were cloned with a TOPO-TA cloning kit with pCR21-TOPO vector (Invitrogen) accord-ing to the manufacturerrsquos instructions
Screening PCR and metagenomic libraries
Clones were screened for insert size using M13 forward andreverse PCR primers which flank the cloned insert andthose with full-length inserts were screened by restrictionfragment length polymorphism (RFLP) analysis by digestingthe M13 PCR amplicons with a mixture of HhaI and RsaI(New England Biolabs) Restriction patterns on 2 Meta-phore agarose (FMC) agarose gels were compared manuallyClones with identical restriction patterns were groupedtogether into RFLP types The 16S rRNA genes from onerepresentative of each RFLP type in the first library werecompletely sequenced
Pools of 96 fosmid clones were screened for 16S rRNAgenes by DGGE of PCR amplicons generated with primersGC358F and 517R (Muyzer et al 1995) Selected bandsresolved on an 8 polyacrylamide gel containing a gradientof 25ndash55 denaturant (138ndash22 formamide and 105ndash23urea) were re-amplified and sequenced Phylogenetic classi-fication was determined using BLAST and the ARB sequenceanalysis tool as described below Bands with equal electro-phoretic mobility were assigned the same phylogeneticclassification
Sequence and library analysis
Nucleotide sequences were analysed using BLAST [version21 National Center for Biotechnology Information (httpwwwncbinlmnihgovBLAST) and the ARB package (Lud-wig et al 2004)] Sequences were aligned with the ARB FastAligner version 103 adjusted manually with secondary-structure criteria Phylogenetic relationships were deter-mined using the Quick Add Parsimony tool and the1000_pub_may02 phylogenetic tree in the ARB database(June 2002 release)
Sequence similarities were determined using the tools inARB and rarefaction curves were calculated using AnalyticRarefaction 13 (httpwwwugaedustratasoftware) Librar-ies were compared using the LIBSHUFF analysis tool(httpwwwarchesugaedu~whitmanlibshuffhtml) (Single-ton et al 2001) Diversity indices were calculated usingequations taken from Dunbar and colleagues (1999)
Statistical analysis
Libraries were compared using LIBSHUFF analysis (Single-ton et al 2001) This approach compares libraries based ontheir coverage and the number of sequences in one library
that are not found in the second library This so-called lsquohet-erologous coveragersquo is calculated over a range of evolution-ary distances to obtain a coverage curve The differencebetween the lsquoheterologous coveragersquo curve and the coveragecurve of the first library is calculated using the Crameacuter-vonMises test statistic (Pettit 1982) and compared with the cov-erage curve calculated after shuffling sequences between thetwo libraries
Nucleotide sequence accession numbers
The nucleotide sequence data reported in this work aredeposited in GenBank under accession numbers AY562245ndashAY562366
Acknowledgements
This work was supported by the US Department of EnergyMicrobial Genomes Program and the National ScienceFoundation
References
Amann RI Binder BJ Olson RJ Chisholm SWDevereux R and Stahl DA (1990) Combination of 16Sribosomal-RNA-targeted oligonucleotide probes with flowcytometry for analyzing mixed microbial populations ApplEnviron Microbiol 56 1919ndash1925
Amon RMW and Benner R (1996) Photochemical andmicrobial consumption of dissolved organic carbon anddissolved oxygen in the Amazon River system GeochimCosmochim Ac 60 1783ndash1792
Beja O Aravind L Koonin EV Suzuki MT Hadd ANguyen LP et al (2000a) Bacterial rhodopsin evidencefor a new type of phototrophy in the sea Science 2891902ndash1906
Beja O Suzuki MT Koonin EV Aravind L Hadd ANguyen LP et al (2000b) Construction and analysis ofbacterial artificial chromosome libraries from a marinemicrobial assemblage Environ Microbiol 2 516ndash529
Beja O Spudich EN Spudich JL Leclerc M andDeLong EF (2001) Proteorhodopsin phototrophy in theocean Nature 411 786ndash789
Button DK and Robertson BR (2001) Determination ofDNA content of aquatic bacteria by flow cytometry ApplEnviron Microbiol 67 1636ndash1645
Button DK Robertson BR Lepp PW and SchmidtTM (1998) A small dilute-cytoplasm high-affinity novelbacterium isolated by extinction culture and having kineticconstants compatible with growth at ambient concentra-tions of dissolved nutrients in seawater Appl EnvironMicrobiol 64 4467ndash4476
Castle D and Kirchman DL (2004) Composition of estu-arine bacterial communities assessed by denaturing gradi-ent gel electrophoresis and fluorescence in situhybridization Limnol Oceanogr 2 303ndash314
Cottrell MT and Kirchman DL (2000) Community compo-sition of marine bacterioplankton determined by 16S rRNAgene clone libraries and fluorescence in situ hybridizationAppl Environ Microbiol 66 5116ndash5122
1894 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Cottrell MT and Kirchman DL (2003) Contribution ofmajor bacterial groups to bacterial biomass production(thymidine and leucine incorporation) in the Delaware estu-ary Limnol Oceanogr 48 168ndash178
Cottrell MT Moore JA and Kirchman DL (1999) Chiti-nases from uncultured marine microorganisms Appl Envi-ron Microbiol 65 2553ndash2557
Covert JS and Moran MA (2001) Molecular characteriza-tion of estuarine bacterial communities that use high- andlow-molecular weight fractions of dissolved organic carbonAquat Microb Ecol 25 127ndash139
Daims H Bruhl A Amann R Schleifer KH and WagnerM (1999) The domain-specific probe EUB338 is insuffi-cient for the detection of all Bacteria Development andevaluation of a more comprehensive probe set Syst ApplMicrobiol 22 434ndash444
Dunbar J Takala S Barns SM Davis JA and KuskeCR (1999) Levels of bacterial community diversity in fourarid soils compared by cultivation and 16S rRNA genecloning Appl Environ Microbiol 65 1662ndash1669
Fegatella F Lim J Kjelleberg S and Cavicchioli R(1998) Implications of rRNA operon copy number and ribo-some content in the marine oligotrophic ultramicrobacte-rium Sphingomonas sp strain RB2256 Appl EnvironMicrobiol 64 4433ndash4438
Fogel GB Collins CR Li J and Brunk CF (1999)Prokaryotic genome size and SSU rDNA copy numberEstimation of microbial relative abundance from a mixedpopulation Microb Ecol 38 93ndash113
Fuhrman JA McCallum K and Davis AA (1992) Novelmajor Archaebacterial group from marine plankton Nature356 148ndash149
Gillespie DE Brady SF Bettermann AD CianciottoNP Liles MR Rondon MR et al (2002) Isolation ofantibiotics turbomycin A and B from a metagenomic libraryof soil microbial DNA Appl Environ Microbiol 68 4301ndash4306
Giovannoni SJ and Rappeacute MS (2000) Evolution diversityand molecular ecology of marine prokaryotes In MicrobialEcology of the Oceans Kirchman DL (ed) New YorkUSA Wiley-Liss pp 47ndash84
Gloumlckner FO Fuchs BM and Amann R (1999) Bacteri-oplankton compositions of lakes and oceans a first com-parison based on fluorescence in situ hybridization ApplEnviron Microbiol 65 3721ndash3726
Gloumlckner FO Zaichikov E Belkova N Denissova LPernthaler J Pernthaler A and Amann R (2000) Com-parative 16S rRNA analysis of lake bacterioplanktonreveals globally distributed phylogenetic clusters includingan abundant group of actinobacteria Appl Environ Micro-biol 66 5053ndash5065
Gonzalez JM and Moran MA (1997) Numerical domi-nance of a group of marine bacteria in the alpha- subclassof the class Proteobacteria in coastal seawater Appl Envi-ron Microbiol 63 4237ndash4242
Hernes PJ and Benner R (2002) Transport and diagene-sis of dissolved and particulate terrigenous organic matterin the North Pacific Ocean Deep-Sea Res Part II ndash TopStud Oceanogr 49 2119ndash2132
Karner M and Fuhrman JA (1997) Determination of activemarine bacterioplankton a comparison of universal 16S
rRNA probes autoradiography and nucleoid staining ApplEnviron Microbiol 63 1208ndash1213
Karner MB DeLong EF and Karl DM (2001) Archaealdominance in the mesopelagic zone of the Pacific OceanNature 409 507ndash510
Kirchman DL (2002) The ecology of CytophagandashFlavobac-teria in aquatic environments FEMS Microbiol Ecol 3991ndash100
Kirchman DL Yu LY and Cottrell MT (2003) Diversityand abundance of uncultured Cytophaga-like bacteria inthe Delaware Estuary Appl Environ Microbiol 69 6587ndash6596
Klappenbach JA Dunbar JM and Schmidt TM (2000)rRNA operon copy number reflects ecological strategies ofbacteria Appl Environ Microbiol 66 1328ndash1333
Knietsch A Waschkowitz T Bowien S Henne A andDaniel R (2003) Metagenomes of complex microbial con-sortia derived from different soils as sources for novelgenes conferring formation of carbonyls from short-chainpolyols on Escherichia coli J Mol Microbiol Biotechnol 546ndash56
Lane DJ (1991) 16S23S rRNA sequencing In Nucleic AcidTechniques in Bacterial Systematics Stackebrandt E andGoodfellow M (eds) Chichester UK Wiley and Sonspp 115ndash175
Liles MR Manske BF Bintrim SB Handelsman J andGoodman RM (2003) A census of rRNA genes and linkedgenomic sequences within a soil metagenomic libraryAppl Environ Microbiol 69 2684ndash2691
Ludwig W Strunk O Westram R Richter L Meier HYadhukumar et al (2004) ARB a software environmentfor sequence data Nucleic Acids Res 32 1363ndash1371
Manz W Amann R Ludwig W Wagner M and Schlei-fer KH (1992) Phylogenetic oligodeoxynucleotide probesfor the major subclasses of Proteobacteria ndash problems andsolutions Syst Appl Microbiol 15 593ndash600
Manz W Amann R Ludwig W Vancanneyt M andSchleifer KH (1996) Application of a suite of 16S rRNA-specific oligonucleotide probes designed to investigatebacteria of the phylum CytophagandashFlavobacterndashBacteroi-des in the natural environment Microbiology 142 1097ndash1106
Massana R Taylor LJ Murray AE Wu KY JeffreyWH and DeLong EF (1998) Vertical distribution andtemporal variation of marine planktonic archaea in theGerlache Strait Antarctica during early spring LimnolOceanogr 43 607ndash617
Massana R DeLong EF and Pedros-Alio C (2000) A fewcosmopolitan phylotypes dominate planktonic archaealassemblages in widely different oceanic provinces ApplEnviron Microbiol 66 1777ndash1787
Morris RM Rappeacute MS Connon SA Vergin KL Sie-bold WA Carlson CA and Giovannoni SJ (2002)SAR11 clade dominates ocean surface bacterioplanktoncommunities Nature 420 806ndash810
Muyzer G Teske A Wirsen CO and Jannasch HW(1995) Phylogenetic-relationships of Thiomicrospira spe-cies and their identification in deep-sea hydrothermal ventsamples by denaturing gradient gel-electrophoresis of 16SrDNA fragments Arch Microbiol 164 165ndash172
Neef A (1997) Anwendung der in situ-Einzelzell-
Metagenomic analysis of freshwater bacteria 1895
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Identifizierung von Bakterien zur Populationsanalyse inKomplexen Mikrobiellen Biozonosen Munich GermanyTechnische Universitat Munchen
Opsahl S Benner R and Amon RMW (1999) Major fluxof terrigenous dissolved organic matter through the ArcticOcean Limnol Oceanogr 44 2017ndash2023
OrsquoSullivan LA Weightman AJ and Fry JC (2002) Newdegenerate Cytophaga-Flexibacter-Bacteroides-specific16S ribosomal DNA-targeted oligonucleotide probes revealhigh bacterial diversity in River Taff epilithon Appl EnvironMicrobiol 68 201ndash210
Pettit AN (1982) Cramer-von Misess statistic In Encyclo-pedia of Statistical Sciences Kotz S Johnson NL andRead CB (eds) New York USA Wiley pp 220ndash221
Quaiser A Ochsenreiter T Klenk HP Kletzin ATreusch AH Meurer G et al (2002) First insight intothe genome of an uncultivated crenarchaeote from soilEnviron Microbiol 4 603ndash611
Rappeacute MS and Giovannoni SJ (2003) The unculturedmicrobial majority Annu Rev Microbiol 57 369ndash394
Rappeacute MS Gordon DA Vergin KL and GiovannoniSJ (1999) Phylogeny of Actinobacteria small subunit(SSU) rRNA gene clones recovered from marine bacteri-oplankton Syst Appl Microbiol 22 106ndash112
Rappeacute MS Connon SA Vergin KL and GiovannoniSJ (2002) Cultivation of the ubiquitous SAR11 marinebacterioplankton clade Nature 418 630ndash633
Roller C Wagner M Amann R Ludwig W and SchleiferKH (1994) In-situ probing of Gram-positive bacteria withhigh DNA G+C content using 23S-ribosomal-RNA-targetedoligonucleotides Microbiology 140 2849ndash2858
Rondon MR August PR Bettermann AD Brady SFGrossman TH Liles MR et al (2000) Cloning the soilmetagenome a strategy for accessing the genetic andfunctional diversity of uncultured microorganisms ApplEnviron Microbiol 66 2541ndash2547
Sabehi G Massana R Bielawski JP Rosenberg MDelong EF and Beja O (2003) Novel proteorhodopsinvariants from the Mediterranean and Red Seas EnvironMicrobiol 5 842ndash849
Singleton DR Furlong MA Rathbun SL and WhitmanWB (2001) Quantitative comparisons of 16S rRNA genesequence libraries from environmental samples Appl Envi-ron Microbiol 67 4374ndash4376
Stackebrandt E and Goebel BM (1994) A place for DNAndashDNA reassociation and 16S ribosomal-RNA sequence-analysis in the present species definition in bacteriologyInt J Syst Bacteriol 44 846ndash849
Stein JL Marsh TL Wu KY Shizuya H and DeLongEF (1996) Characterization of uncultivated prokaryotesIsolation and analysis of a 40-kilobase-pair genome frag-ment from a planktonic marine archaeon J Bacteriol 178591ndash599
Thompson JR Marcelino LA and Polz MF (2002) Het-eroduplexes in mixed-template amplifications formationconsequence and elimination by lsquoreconditioning PCRrsquoNucleic Acids Res 30 2083ndash2088
Topp E (2003) Bacteria in agricultural soils Diversity roleand future perspectives Can J Soil Sci 83 303ndash309
Weller R Gloumlckner FO and Amann R (2000) 16S rRNA-targeted oligonucleotide probes for the in situ detectionof members of the phylum Cytophaga-Flavobacterium-Bacteroides Syst Appl Microbiol 23 107ndash114
von Wintzingerode F Gobel UB and Stackebrandt E(1997) Determination of microbial diversity in environmen-tal samples pitfalls of PCR-based rRNA analysis FEMSMicrobiol Rev 21 213ndash229
Zaitlin B Watson SB Dixon J and Steel D (2003)Actinomycetes in the Elbow River Basin Alberta CanWater Quality Res J Can 38 115ndash125
Zwart G Hiorns WD Methe BA Van Agterveld MPHuismans R Nold SC et al (1998) Nearly identical 16SrRNA sequences recovered from lakes in North Americaand Europe indicate the existence of clades of globallydistributed freshwater bacteria Syst Appl Microbiol 21546ndash556
Zwart G Crump BC Agterveld M Hagen F and HanSK (2002) Typical freshwater bacteria an analysis ofavailable 16S rRNA gene sequences from plankton oflakes and rivers Aquat Microb Ecol 28 141ndash155
1894 M T Cottrell L A Waidner L Yu and D L Kirchman
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Cottrell MT and Kirchman DL (2003) Contribution ofmajor bacterial groups to bacterial biomass production(thymidine and leucine incorporation) in the Delaware estu-ary Limnol Oceanogr 48 168ndash178
Cottrell MT Moore JA and Kirchman DL (1999) Chiti-nases from uncultured marine microorganisms Appl Envi-ron Microbiol 65 2553ndash2557
Covert JS and Moran MA (2001) Molecular characteriza-tion of estuarine bacterial communities that use high- andlow-molecular weight fractions of dissolved organic carbonAquat Microb Ecol 25 127ndash139
Daims H Bruhl A Amann R Schleifer KH and WagnerM (1999) The domain-specific probe EUB338 is insuffi-cient for the detection of all Bacteria Development andevaluation of a more comprehensive probe set Syst ApplMicrobiol 22 434ndash444
Dunbar J Takala S Barns SM Davis JA and KuskeCR (1999) Levels of bacterial community diversity in fourarid soils compared by cultivation and 16S rRNA genecloning Appl Environ Microbiol 65 1662ndash1669
Fegatella F Lim J Kjelleberg S and Cavicchioli R(1998) Implications of rRNA operon copy number and ribo-some content in the marine oligotrophic ultramicrobacte-rium Sphingomonas sp strain RB2256 Appl EnvironMicrobiol 64 4433ndash4438
Fogel GB Collins CR Li J and Brunk CF (1999)Prokaryotic genome size and SSU rDNA copy numberEstimation of microbial relative abundance from a mixedpopulation Microb Ecol 38 93ndash113
Fuhrman JA McCallum K and Davis AA (1992) Novelmajor Archaebacterial group from marine plankton Nature356 148ndash149
Gillespie DE Brady SF Bettermann AD CianciottoNP Liles MR Rondon MR et al (2002) Isolation ofantibiotics turbomycin A and B from a metagenomic libraryof soil microbial DNA Appl Environ Microbiol 68 4301ndash4306
Giovannoni SJ and Rappeacute MS (2000) Evolution diversityand molecular ecology of marine prokaryotes In MicrobialEcology of the Oceans Kirchman DL (ed) New YorkUSA Wiley-Liss pp 47ndash84
Gloumlckner FO Fuchs BM and Amann R (1999) Bacteri-oplankton compositions of lakes and oceans a first com-parison based on fluorescence in situ hybridization ApplEnviron Microbiol 65 3721ndash3726
Gloumlckner FO Zaichikov E Belkova N Denissova LPernthaler J Pernthaler A and Amann R (2000) Com-parative 16S rRNA analysis of lake bacterioplanktonreveals globally distributed phylogenetic clusters includingan abundant group of actinobacteria Appl Environ Micro-biol 66 5053ndash5065
Gonzalez JM and Moran MA (1997) Numerical domi-nance of a group of marine bacteria in the alpha- subclassof the class Proteobacteria in coastal seawater Appl Envi-ron Microbiol 63 4237ndash4242
Hernes PJ and Benner R (2002) Transport and diagene-sis of dissolved and particulate terrigenous organic matterin the North Pacific Ocean Deep-Sea Res Part II ndash TopStud Oceanogr 49 2119ndash2132
Karner M and Fuhrman JA (1997) Determination of activemarine bacterioplankton a comparison of universal 16S
rRNA probes autoradiography and nucleoid staining ApplEnviron Microbiol 63 1208ndash1213
Karner MB DeLong EF and Karl DM (2001) Archaealdominance in the mesopelagic zone of the Pacific OceanNature 409 507ndash510
Kirchman DL (2002) The ecology of CytophagandashFlavobac-teria in aquatic environments FEMS Microbiol Ecol 3991ndash100
Kirchman DL Yu LY and Cottrell MT (2003) Diversityand abundance of uncultured Cytophaga-like bacteria inthe Delaware Estuary Appl Environ Microbiol 69 6587ndash6596
Klappenbach JA Dunbar JM and Schmidt TM (2000)rRNA operon copy number reflects ecological strategies ofbacteria Appl Environ Microbiol 66 1328ndash1333
Knietsch A Waschkowitz T Bowien S Henne A andDaniel R (2003) Metagenomes of complex microbial con-sortia derived from different soils as sources for novelgenes conferring formation of carbonyls from short-chainpolyols on Escherichia coli J Mol Microbiol Biotechnol 546ndash56
Lane DJ (1991) 16S23S rRNA sequencing In Nucleic AcidTechniques in Bacterial Systematics Stackebrandt E andGoodfellow M (eds) Chichester UK Wiley and Sonspp 115ndash175
Liles MR Manske BF Bintrim SB Handelsman J andGoodman RM (2003) A census of rRNA genes and linkedgenomic sequences within a soil metagenomic libraryAppl Environ Microbiol 69 2684ndash2691
Ludwig W Strunk O Westram R Richter L Meier HYadhukumar et al (2004) ARB a software environmentfor sequence data Nucleic Acids Res 32 1363ndash1371
Manz W Amann R Ludwig W Wagner M and Schlei-fer KH (1992) Phylogenetic oligodeoxynucleotide probesfor the major subclasses of Proteobacteria ndash problems andsolutions Syst Appl Microbiol 15 593ndash600
Manz W Amann R Ludwig W Vancanneyt M andSchleifer KH (1996) Application of a suite of 16S rRNA-specific oligonucleotide probes designed to investigatebacteria of the phylum CytophagandashFlavobacterndashBacteroi-des in the natural environment Microbiology 142 1097ndash1106
Massana R Taylor LJ Murray AE Wu KY JeffreyWH and DeLong EF (1998) Vertical distribution andtemporal variation of marine planktonic archaea in theGerlache Strait Antarctica during early spring LimnolOceanogr 43 607ndash617
Massana R DeLong EF and Pedros-Alio C (2000) A fewcosmopolitan phylotypes dominate planktonic archaealassemblages in widely different oceanic provinces ApplEnviron Microbiol 66 1777ndash1787
Morris RM Rappeacute MS Connon SA Vergin KL Sie-bold WA Carlson CA and Giovannoni SJ (2002)SAR11 clade dominates ocean surface bacterioplanktoncommunities Nature 420 806ndash810
Muyzer G Teske A Wirsen CO and Jannasch HW(1995) Phylogenetic-relationships of Thiomicrospira spe-cies and their identification in deep-sea hydrothermal ventsamples by denaturing gradient gel-electrophoresis of 16SrDNA fragments Arch Microbiol 164 165ndash172
Neef A (1997) Anwendung der in situ-Einzelzell-
Metagenomic analysis of freshwater bacteria 1895
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Identifizierung von Bakterien zur Populationsanalyse inKomplexen Mikrobiellen Biozonosen Munich GermanyTechnische Universitat Munchen
Opsahl S Benner R and Amon RMW (1999) Major fluxof terrigenous dissolved organic matter through the ArcticOcean Limnol Oceanogr 44 2017ndash2023
OrsquoSullivan LA Weightman AJ and Fry JC (2002) Newdegenerate Cytophaga-Flexibacter-Bacteroides-specific16S ribosomal DNA-targeted oligonucleotide probes revealhigh bacterial diversity in River Taff epilithon Appl EnvironMicrobiol 68 201ndash210
Pettit AN (1982) Cramer-von Misess statistic In Encyclo-pedia of Statistical Sciences Kotz S Johnson NL andRead CB (eds) New York USA Wiley pp 220ndash221
Quaiser A Ochsenreiter T Klenk HP Kletzin ATreusch AH Meurer G et al (2002) First insight intothe genome of an uncultivated crenarchaeote from soilEnviron Microbiol 4 603ndash611
Rappeacute MS and Giovannoni SJ (2003) The unculturedmicrobial majority Annu Rev Microbiol 57 369ndash394
Rappeacute MS Gordon DA Vergin KL and GiovannoniSJ (1999) Phylogeny of Actinobacteria small subunit(SSU) rRNA gene clones recovered from marine bacteri-oplankton Syst Appl Microbiol 22 106ndash112
Rappeacute MS Connon SA Vergin KL and GiovannoniSJ (2002) Cultivation of the ubiquitous SAR11 marinebacterioplankton clade Nature 418 630ndash633
Roller C Wagner M Amann R Ludwig W and SchleiferKH (1994) In-situ probing of Gram-positive bacteria withhigh DNA G+C content using 23S-ribosomal-RNA-targetedoligonucleotides Microbiology 140 2849ndash2858
Rondon MR August PR Bettermann AD Brady SFGrossman TH Liles MR et al (2000) Cloning the soilmetagenome a strategy for accessing the genetic andfunctional diversity of uncultured microorganisms ApplEnviron Microbiol 66 2541ndash2547
Sabehi G Massana R Bielawski JP Rosenberg MDelong EF and Beja O (2003) Novel proteorhodopsinvariants from the Mediterranean and Red Seas EnvironMicrobiol 5 842ndash849
Singleton DR Furlong MA Rathbun SL and WhitmanWB (2001) Quantitative comparisons of 16S rRNA genesequence libraries from environmental samples Appl Envi-ron Microbiol 67 4374ndash4376
Stackebrandt E and Goebel BM (1994) A place for DNAndashDNA reassociation and 16S ribosomal-RNA sequence-analysis in the present species definition in bacteriologyInt J Syst Bacteriol 44 846ndash849
Stein JL Marsh TL Wu KY Shizuya H and DeLongEF (1996) Characterization of uncultivated prokaryotesIsolation and analysis of a 40-kilobase-pair genome frag-ment from a planktonic marine archaeon J Bacteriol 178591ndash599
Thompson JR Marcelino LA and Polz MF (2002) Het-eroduplexes in mixed-template amplifications formationconsequence and elimination by lsquoreconditioning PCRrsquoNucleic Acids Res 30 2083ndash2088
Topp E (2003) Bacteria in agricultural soils Diversity roleand future perspectives Can J Soil Sci 83 303ndash309
Weller R Gloumlckner FO and Amann R (2000) 16S rRNA-targeted oligonucleotide probes for the in situ detectionof members of the phylum Cytophaga-Flavobacterium-Bacteroides Syst Appl Microbiol 23 107ndash114
von Wintzingerode F Gobel UB and Stackebrandt E(1997) Determination of microbial diversity in environmen-tal samples pitfalls of PCR-based rRNA analysis FEMSMicrobiol Rev 21 213ndash229
Zaitlin B Watson SB Dixon J and Steel D (2003)Actinomycetes in the Elbow River Basin Alberta CanWater Quality Res J Can 38 115ndash125
Zwart G Hiorns WD Methe BA Van Agterveld MPHuismans R Nold SC et al (1998) Nearly identical 16SrRNA sequences recovered from lakes in North Americaand Europe indicate the existence of clades of globallydistributed freshwater bacteria Syst Appl Microbiol 21546ndash556
Zwart G Crump BC Agterveld M Hagen F and HanSK (2002) Typical freshwater bacteria an analysis ofavailable 16S rRNA gene sequences from plankton oflakes and rivers Aquat Microb Ecol 28 141ndash155
Metagenomic analysis of freshwater bacteria 1895
copy 2005 Society for Applied Microbiology and Blackwell Publishing Ltd Environmental Microbiology 7 1883ndash1895
Identifizierung von Bakterien zur Populationsanalyse inKomplexen Mikrobiellen Biozonosen Munich GermanyTechnische Universitat Munchen
Opsahl S Benner R and Amon RMW (1999) Major fluxof terrigenous dissolved organic matter through the ArcticOcean Limnol Oceanogr 44 2017ndash2023
OrsquoSullivan LA Weightman AJ and Fry JC (2002) Newdegenerate Cytophaga-Flexibacter-Bacteroides-specific16S ribosomal DNA-targeted oligonucleotide probes revealhigh bacterial diversity in River Taff epilithon Appl EnvironMicrobiol 68 201ndash210
Pettit AN (1982) Cramer-von Misess statistic In Encyclo-pedia of Statistical Sciences Kotz S Johnson NL andRead CB (eds) New York USA Wiley pp 220ndash221
Quaiser A Ochsenreiter T Klenk HP Kletzin ATreusch AH Meurer G et al (2002) First insight intothe genome of an uncultivated crenarchaeote from soilEnviron Microbiol 4 603ndash611
Rappeacute MS and Giovannoni SJ (2003) The unculturedmicrobial majority Annu Rev Microbiol 57 369ndash394
Rappeacute MS Gordon DA Vergin KL and GiovannoniSJ (1999) Phylogeny of Actinobacteria small subunit(SSU) rRNA gene clones recovered from marine bacteri-oplankton Syst Appl Microbiol 22 106ndash112
Rappeacute MS Connon SA Vergin KL and GiovannoniSJ (2002) Cultivation of the ubiquitous SAR11 marinebacterioplankton clade Nature 418 630ndash633
Roller C Wagner M Amann R Ludwig W and SchleiferKH (1994) In-situ probing of Gram-positive bacteria withhigh DNA G+C content using 23S-ribosomal-RNA-targetedoligonucleotides Microbiology 140 2849ndash2858
Rondon MR August PR Bettermann AD Brady SFGrossman TH Liles MR et al (2000) Cloning the soilmetagenome a strategy for accessing the genetic andfunctional diversity of uncultured microorganisms ApplEnviron Microbiol 66 2541ndash2547
Sabehi G Massana R Bielawski JP Rosenberg MDelong EF and Beja O (2003) Novel proteorhodopsinvariants from the Mediterranean and Red Seas EnvironMicrobiol 5 842ndash849
Singleton DR Furlong MA Rathbun SL and WhitmanWB (2001) Quantitative comparisons of 16S rRNA genesequence libraries from environmental samples Appl Envi-ron Microbiol 67 4374ndash4376
Stackebrandt E and Goebel BM (1994) A place for DNAndashDNA reassociation and 16S ribosomal-RNA sequence-analysis in the present species definition in bacteriologyInt J Syst Bacteriol 44 846ndash849
Stein JL Marsh TL Wu KY Shizuya H and DeLongEF (1996) Characterization of uncultivated prokaryotesIsolation and analysis of a 40-kilobase-pair genome frag-ment from a planktonic marine archaeon J Bacteriol 178591ndash599
Thompson JR Marcelino LA and Polz MF (2002) Het-eroduplexes in mixed-template amplifications formationconsequence and elimination by lsquoreconditioning PCRrsquoNucleic Acids Res 30 2083ndash2088
Topp E (2003) Bacteria in agricultural soils Diversity roleand future perspectives Can J Soil Sci 83 303ndash309
Weller R Gloumlckner FO and Amann R (2000) 16S rRNA-targeted oligonucleotide probes for the in situ detectionof members of the phylum Cytophaga-Flavobacterium-Bacteroides Syst Appl Microbiol 23 107ndash114
von Wintzingerode F Gobel UB and Stackebrandt E(1997) Determination of microbial diversity in environmen-tal samples pitfalls of PCR-based rRNA analysis FEMSMicrobiol Rev 21 213ndash229
Zaitlin B Watson SB Dixon J and Steel D (2003)Actinomycetes in the Elbow River Basin Alberta CanWater Quality Res J Can 38 115ndash125
Zwart G Hiorns WD Methe BA Van Agterveld MPHuismans R Nold SC et al (1998) Nearly identical 16SrRNA sequences recovered from lakes in North Americaand Europe indicate the existence of clades of globallydistributed freshwater bacteria Syst Appl Microbiol 21546ndash556
Zwart G Crump BC Agterveld M Hagen F and HanSK (2002) Typical freshwater bacteria an analysis ofavailable 16S rRNA gene sequences from plankton oflakes and rivers Aquat Microb Ecol 28 141ndash155