wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 0 5 0e5 0 5 8

Avai lab le at www.sc iencedi rect .com

journa l homepage : www.e lsev ier . com/ loca te /wat res

Bacteria from drinking water supply and their fatein gastrointestinal tracts of germ-free mice: A phylogeneticcomparison study

J. Lee a,b,*, C.S. Lee a, K.M. Hugunin c, C.J. Maute c, R.C. Dysko c

aCollege of Public Health, Division of Environmental Health Sciences, The Ohio State University, Columbus, OH 43210, USAbDepartment of Food Science & Technology, The Ohio State University, Columbus, OH 43210, USAcUnit for Laboratory Animal Medicine, University of Michigan Medical School, Ann Arbor, MI 48109, USA

a r t i c l e i n f o

Article history:

Received 30 March 2010

Received in revised form

26 June 2010

Accepted 9 July 2010

Available online 21 July 2010

Keywords:

Bacteria

Drinking water

Biofilm

Colonization

Gastrointestinal tracts

Abiotic mice

* Corresponding author. College of Public HeaThe Ohio State University, Columbus, OH 43

E-mail address: jlee@cph.osu.edu (J. Lee).0043-1354/$ e see front matter ª 2010 Elsevdoi:10.1016/j.watres.2010.07.027

a b s t r a c t

Microorganisms in drinking water sources may colonize in gastrointestinal (GI) tracts and

this phenomenon may pose a potential health risk especially to immunocompromised

population. The microbial community diversity of the drinking water was compared with

the GI tracts of the mice using phylogenetic and statistical analyses of 16S rRNA gene

sequences. A group of germ-free mice were fed with drinking water from public water

supply that passed through an automated watering system with documented biofilm

accumulation. From drinking water and GI tracts of the germ-free mice, 179 bacteria were

isolated and 75 unique 16S rRNA gene phylotypes were sequenced as operational taxo-

nomic unit (OTU, >97% similarity). Three major groups of the genus Acidovorax (21%),

Variovorax (42%) and Sphingopyxis (15%) were found in drinking water. Three major groups

of the genus Ralstonia (24%), Staphylococcus (20%) and Bosea (22%) were found in GI tracts.

Ralstonia (6%, 24%), Sphingopyxis (15%, 2%), Bacillus (3%, 5%), Escherichia coli (3%, 2%) and

Mesorhizobium (3%, 5%) were found in both sources e drinking water and GI tract. A lineage-

per-time plot shows that the both bacterial communities have convex shape lines, sug-

gesting an excess of closely related ecotypes. A significant FST test (0.00000e0.00901)

coupled with an insignificant P test (0.07e0.46) implies that the tree contained several

clades of closely related bacteria. Both phylogenetic and statistical results suggest

a correlation between the bacterial communities originating in the drinking water and

those associated with the GI tracts. The GI tract showed a higher genetic diversity than the

drinking water, but a similar lineage-per-time plot was obtained overall. It means a sudden

evolutionary transformation and colonization occurred with high selective forces.

ª 2010 Elsevier Ltd. All rights reserved.

1. Introduction water is managed strictly, there still remain risks of water-

Drinking water is distributed through complicated piping

systems after the water has been treated at water treatment

plants until it arrives at consumers’ tap. Even though drinking

lth, Division of Environm210, USA. Tel.: þ1 614 29

ier Ltd. All rights reserve

borne illness originating from the water systems in developed

countries (Reynolds et al., 2008). Many reported that outbreaks

of gastrointestinal illnesshavebeenattributed to consumption

of drinking water meeting conventional coliform standards.

ental Health Sciences, Department of Food Science & Technology,2 5546; fax: þ1 614 293 7710.

d.

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 0 5 0e5 0 5 8 5051

These have been observed in all population subgroups

(Payment et al., 1991). The age and elaborate nature of distri-

bution systems have increased the chances for contamination

events and waterborne disease that are not attributed to defi-

ciencies in water treatment. Although over the years tremen-

dousknowledgehasbeenaccumulatedregarding thedetection

of indicator organisms, pathogenic bacteria and viruses, and

microbial community structure in water distribution systems

(Lechevallier, 1990; Rogers and Keevil, 1992), questions about

the public health risks due to the microbial populations in

biofilm and bulk water are still remaining (Williams et al.,

2004).

As the water stands inside of the distribution system for an

extended period, even if a disinfectant residual is present and

the environment is oligotrophic, formation of microbial bio-

film can easily occur (LeChevallier et al., 1987; Pedersen, 1990;

Ridgway and Olson, 1981). Biofilm is a complicated mixture of

microbes, organic, and inorganic material which together

form a polymer matrix and attach to the inner surface of the

distribution system (U.S. Environmental Protection Agency,

2002). Biofilm found in water distribution systems are

known to cause public health concerns (Martiny et al., 2003),

such as protecting and supporting pathogenic microorgan-

isms (Cooper and Hanlon, 2010; Edwards, 1993), bacterial

regrowth (Lechevallier, 1990; Stach et al., 2003), and depletion

of disinfection agents (Payment et al., 1994). Previous studies

have shown the identity and diversity of these bacteria in

biofilms (Lappinscott and Costerton, 1990; Martiny et al., 2005;

Payment et al., 1994). Microorganisms in drinking water

source, either planktonic or biofilm biota on pipes, may colo-

nize in gastrointestinal (GI) tracts. This phenomenon may

pose a health risk especially to a population with deficient

immune systems. Biofilm bacteria on pipes can enhance

adhesion and proliferation of other pathogens into the biofilm

complex by providing adequate biological niches through

affording diverse adhesion sites and antibiotic resistance

(Lysenko et al., 2010). Cellecell signaling among the biofilm

members may trigger virulence determinants and expression

of pathogenic behaviors under a certain condition (Lysenko

et al., 2010). However, very little is known about the fate of

these bacteria in water distribution system once they enter

into a digestive system, especially, whether the bacteria

originate from water and the biofilm subsequently colonizes

in gastrointestinal tracts. Gnotobiotic animal models (e.g.

germ-free mice) have been used to examine the colonization

of bacteria entering the GI tracts to assess the roles of specific

bacterial types and to determine the bacterial effect on the

development of the GI system (Hudault et al., 2001). The

distinction between indigenous and nonindigenous microbes

is crucial to an ecological understanding of colonization after

the interaction has taken place between the intestinal

microbes and their host (Mackie et al., 1999). A wealth of

information on the ecology and phylogenetic diversity of

biofilm populations in water distribution systems has been

accumulated (Bischofberger et al., 1990; Martiny et al., 2005;

Tokajian et al., 2005; Williams et al., 2004; Yan et al., 2007),

but an accurate understanding has not yet been achieved

regarding the bacterial colonization in GI tracts. The aim of

this study was to provide insight into the potential of coloni-

zation of bacteria in drinking water to GI tracts. In order to

answer this question, we investigated the microbial commu-

nity and compared their diversity in drinking water and the GI

tracts of abiotic mice using phylogenetic and statistical anal-

yses of 16S rRNA gene sequences. A group of germ-free mice

were fed drinkingwater from the public water supply that had

passed through automated animal watering systems with

documented biofilm development.

2. Materials and methods

2.1. Preparation of samples and isolation of bacteria

Samples were prepared by the research group from the

University of Michigan (Ann Arbor, Michigan), at which

automated watering systems (AWS) for research mice had

operated for 1 year and biofilm was developed naturally by

passing drinking water through housing rack manifolds over

time (Hugunin et al., 2008, 2009). The biofilm buildup was

periodically confirmed using swab sampling and growing on

R2A plates (data not shown). The City of Ann Arbor uses ozone

as a primary disinfectant and monochloramine as a distribu-

tion system residual. The average of monochloramine

concentration of AnnArbor drinkingwaterwas 2.4e2.9 mg L�1

(http://www.a2gov.org/government/publicservices/water_

treatment/Documents/ccr.pdf). The drinking water was

collected from the drain outlet of the automated watering

manifold on an animal housing rack, and provided to germ-

free mice in a sterile water bottle. After 7 days of consuming

the water, the abiotic mice were euthanized and each

jejunum, cecum, and colon was sampled for bacterial growth.

Bacterial strains were recovered using blood agar and R2A

under both aerobic and anaerobic conditions using an

anaerobic jar and AnaeroPack system (Mitsubishi Gas Chem-

ical, Tokyo, Japan) at 37 �C (blood agar) or room temperature

23� 3 �C (R2A). Bacteria in the drinkingwaterwere obtained in

one of three ways: (1) collected from the manifold drain e

whichwas the samewater given to themicee passed through

a 0.2-mmfilter, with the filter swabbed and plated; (2) collected

from the water bottle at the end of the 7-day feeding trial with

a sterile swab; and (3) collected with a sterile swab from the

inside of the piping of all rackmanifolds in the animal housing

roomvia removal of adjacent segments of the pipe. For further

experiments including DNA extraction, isolated bacteria were

propagated on either blood agar or R2A with 0.1% (w/v) yeast

extract by incubating at 37 �C or room temperature for 24e48 h

depending upon their original growing media and condition.

Anaerobic jar with AnaeroPack system was used for the

anaerobic isolates. Stock bacterial cultures were stored in 40%

glycerol with liquid media at �80 �C for long-term storage.

2.2. DNA extraction, PCR amplification and sequencing

Genomic DNA was extracted using a DNeasy Blood & Tissue

kit (Qiagen, Valencia, CA) and the DNA concentration was

determined using a Nanodrop spectrophotometer (Nano-

drop Technologies, Wilmington, DE). From the extracted

DNA, 16S rRNA genes were amplified from the isolated

bacteria by using a Taq PCR core kit (Qiagen, Valencia, CA)

with a primer set of 27F (50AGAGTTTGATCMTGGCTCAG30)

Sphingopyxis

15%

Ralstonia

6%

Novosphigobium

6%

Mesorhizobium

3%

E. coli

3%

Bacillus

3%

Acidovorax

21%

Variovorax

42%

Fig. 1 e Bacterial diversity in drinking water. Pie chart

shows the relative diversity of each genus identified by 16S

rRNA gene sequencing.

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 0 5 0e5 0 5 85052

and 1525R (50AAGGAGGTGWTCCARCC30) (Lane, 1991). The

PCR conditions included 30 cycles of 94 �C (1 min), 52 �C(30 s), 72 �C (2 min), and one additional cycle with a final

10 min for chain elongation. Subsequently, the PCR prod-

ucts were purified with a QIAquick PCR purification kit

(Qiagen, Valencia, CA) and the sequences of 16S rRNA

genes were determined using ABI Prism 3730 DNA analyzer

(Applied Biosystems, Foster City, CA) at the Plant-Microbe

Genomics Facility of The Ohio State University (http://pmgf.

biosci.ohio-state.edu/) by using the same primer.

2.3. Phylogenetic analysis

Sequenceswere analyzed by comparing themwith known 16S

rRNA sequences using the BLAST algorithm (http://blast.ncbi.

nlm.nih.gov/Blast.cgi) to find the closest match in GenBank,

EMBL, DDBJ, and PDB sequence data. Most similar type species

with 97% similarity (<3% diversity) to the sequences of

isolates were designated as the same species. The 16S rRNA

sequences were aligned by Clustal X (Larkin et al., 2007) and

then were edited using BioEdit (Hall, 1999). The distances for

each 16S rRNA were calculated by the neighbor-joining

method (Tamura et al., 2004) and phylogenetic trees were

created by using MEGA (Tamura et al., 2007). The evolutionary

distances for each 16S rRNA were calculated by the neighbor-

joining method with Maximum Composite Likelihood model

by 1000 replicates (Tamura et al., 2004). All positions con-

taining gaps and missing data were eliminated from the data

set using complete deletion option.

2.4. Statistical analysis of microbial populationstructure

After the sequencing analysis, the genetic diversity of bacte-

rial population in each sample source was statistically esti-

mated using a lineage-per-time plot (Martin, 2002), Arlequin

v3.0 (Excoffier et al., 2005), and web-based UniFrac (http://

bmf2.colorado.edu/unifrac/) methods (Lozupone et al., 2006).

A lineage-per-time plot is a tool used to estimate genetic

diversity in which they summarize a phylogeny as a cumula-

tive function of the number of lineages relative to arbitrary

time for the comparison of microbial diversity (Martin, 2002).

After phylogenetic trees were reconstructed as branch lengths

from the root to all the terminal species became identical by

assuming equal evolutionary rates in all lineages (Takezaki

et al., 1995), a plot was constructed by counting the number

of lineages at each time interval with the aim of illustrating

microbial diversity (Nee et al., 1994).

Bacterial genetic differentiation was determined by using

Arlequin v3.0 software and UniFrac. These can be used for

assessing the degree of differentiation between microbial

communities by calculating the diversity index. This is

expressed using the following equation; FST¼ (ӨteӨw)/Өt,

where Өt is the genetic diversity for all samples and Өw is the

genetic diversity within each community averaged over all

the communities being compared (Edwards, 1993; Slatkin,

1991). Next, the statistical significance of FST (F statistics)

was calculated by randomly assigning sequences to pop-

ulations and calculating the FST for 1000 permutations by

UniFrac (Oakley et al., 2010). The P test was used to examine

whether the communities exhibited covariation with

phylogeny (Martin, 2002).

3. Results

3.1. Isolation and identification of bacteria in drinkingwater and GI tracts

From the samples of GI tracts and drinkingwater, a total of 179

bacteria were purely isolated by a repeated subculture on R2A

and a blood agar plate under each aerobic and anaerobic

condition. From the GI tract samples, 60 aerobic and 62

anaerobic bacteria were isolated from the total 122 bacteria.

From the drinking water samples, 43 aerobic and 14 anaerobic

bacteria were isolated from the 57 bacteria. Gram staining was

performed for the isolates. Eighty-seven isolates were Gram

positive and 67were Gramnegative, while 25 isolateswere not

clearly determined.

DNA from the isolates was extracted using a DNeasy Blood

& Tissue kit (Qiagen, Valencia, CA) and made into 200 ml of

elution buffer as described in themanufacturer’s instructions.

Their DNA concentrations were in the range of 5e20 ng ml�1

after measuring with Nanodrop (Nanodrop Technologies,

Wilmington, DE). Their 16S rRNA genes were amplified by

using PCR inwhich the DNA amountwas adjusted as 10 ng per

20 ml of the PCR tube. Almost all the 16S rRNA genes were

amplified with a 27F/1525R primer set and they were taken

over to the Plant-Microbe Genomics Facility for sequencing

after purification. Partial segments of the 16S rRNA geneswere

sequenced and the final 42 and 33 sequences were analyzed

from the amplicons of 16S rRNA gene, which were extracted

from the isolates of GI tracts and drinking water, respectively.

In the drinking water samples, major bacterial groups were

identified as the genera Acidovorax, Variovorax and Sphingo-

pyxis (Fig. 1). Among the genera of Acidovorax and Variovorax,

all the isolates fell in the same species, either Acidovorax

delafieldii or Variovorax paradoxus, with the similarity of >97%

which is regarded as a standard value in 16S rRNA gene for

operational taxonomic unit (OTU). The genus Acidovorax was

Pseudomonas

2%

Pseudolabrys

2%

Mesorhizobium

5%

E. coli

2%

Bartonella

2%

Sporolactobacillu

s

2%

Staphylococcus

20%

Streptococcus

2%

Zooshikella

2%Bacillus

5%

Bosea

22%

Ralstonia

24%

Rickettsia

2%

Salinibacillus

2%

Sphingopyxis

2%

Fig. 2 e Bacterial diversity in GI tracts. Pie chart shows the

relative diversity of each genus detected in GI tracts

identified by 16S rRNA gene sequencing.

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 0 5 0e5 0 5 8 5053

separated as a new taxonomical group from the previously

known Pseudomonas facilis, Pseudomonas delafieldii and two

groups of clinical isolates (E. Falsen (EF) group 13 and EF group

16) (Willems et al., 1990; Falsen, 1983). These clinical isolates

created new taxonomical group of A. delafieldii as a novel

species (Willems et al., 1990). V. paradoxus was reclassified

fromAlcaligenes paradoxus and has two biotypes in its ability to

grow autotrophically in the presence or absence of H2

(Willems et al., 1991; Davis et al., 1969). Phylogenetic analyses

based on 16S rRNA gene sequences showed that the genus

Variovorax falls within the family Comamonadaceae of the

b-proteobacteria (Anzai et al., 2000). It commonly exists in

ubiquitous environment, such as biofilm (Rodrigues et al.,

2008), contaminated aquifer (Rooney-Varga et al., 1999), soil

(Smith et al., 2005) and human mouth (Anesti et al., 2005). In

case of the genus Sphingopyxis, three species were identified:

Sphingopyxis marina, Sphingopyxis panaciterrae, and Sphingopyxis

taejonensis. The genus Sphingopyxis was reclassified as a new

genus from one cluster of Sphingomonas on the basis of

phylogenetic and chemotaxonomic analyses (Takeuchi et al.,

2001). Among the three species, S. panaciterrae was a major

component in the drinking water samples. Escherichia coli,

Bacillus amyloliquefaciens, Mesorhizobium huakuii, Novos-

phingobium resinovorum, Novosphingobium naphthalenivorans

and Ralstonia pickettii were also found in the drinking water

samples.

In the GI tracts samples, more diverse bacterial pop-

ulations were observed than in the water samples. Three

major groups were the genera of Ralstonia, Staphylococcus and

Bosea. All of the isolates identified as Ralstonia were of the

species of R. pickettii. Recently, R. pickettii has gained

substantial interest as a nosocomial infectious agent in water,

water system components, distilled facilities, and potable

water dispenser in international space station (Adiloglu et al.,

2004; Kendirli et al., 2004; Moreira et al., 2005; Ryan et al., 2006;

Szymanska, 2007; Wong et al., 2010). Various Staphylococcus

and Bosea species were found including Sphingopyxis epi-

dermidis, Sphingopyxis equorum subsp. equorum, Sphingopyxis

muscae, Sphingopyxis pasteuri, Sphingopyxis saccharolyticus,

Sphingopyxis thermophilus and Bosea eneae, B. minatitlanensis,

Bosea vestrisii (Fig. 2). The Staphylococcus group resides nor-

mally on skin and mucous membranes of human and other

organisms, but it is also a small constituent of soil microbial

flora in nature environment. B. eneae and B. verstrisii were

reported that they were isolated from a hospital water system

by co-cultivation with amoeba (La Scola et al., 2003). Besides

thesemajor groups, other diverse species were found in the GI

tracts, such as Bacillus pumilus, E. coli, Mesorhizobium huakuii,

Mesorhizobium temperatum, Pseudolabrys taiwanensis, Pseudo-

monas fulva, Salinibacillus aidingensis, Sporolactobacillus inulinus,

S. panaciterrae and Zooshikella ganghwensis (Fig. 3).

3.2. Phylogenetic analysis

We analyzed 16S rRNA sequences of isolates by constructing

phylogenetic trees. Fig. 4 shows the phylogenetic analysis of

the bacteria in both the drinking water and GI tracts. The

result showed that abundant phylotypes e Acidovorax, Vari-

ovorax and Sphingopyxise in the drinking water were clustered

with the related taxa. In the water samples, the most

abundant sequence types, representing Acidovorax and Vari-

ovorax, belonged to the same family of Comamonadaceae and

the class of b-proteobacteria. Ralstoniawas found in thewater as

a minor population and it showed a close association with

Acidovorax and Variovorax under the same order of Bur-

kholderiales. Sphingopyxis is a genus which belongs to the

family of Sphingomonadaceae and the class of a-Proteobac-

teria. Sphingomonas and Novosphingobium are the closely linked

genera in the same class.

In the GI tracts, the most abundant sequence types were

Bosea, Ralstonia, and Staphylococcus. One of the interesting

results was that even though the genus Staphylococcuswas not

isolated from the drinking water sample, many Staphylococcus

species were found in the GI tract samples. Ralstonia was not

a major population in the drinking water, but predominant in

the GI tracts. In contrast, Variovorax and Acidovorax were

abundant in the drinking water, but absent in the GI tracts.

Only small numbers of Sphingopyxis remained in existence in

the GI tracts. Bacillus, E. coli, andMesorhizobiumwere present in

both the drinking water and the GI tracts.

Previously, bacteria in the biofilm placed on the surfaces of

the rubber-coated drinking water valves have been investi-

gated by cloning library (Schmeisser et al., 2003). Proteobacteria

constituted 86% of the clones identified and represented the

majority of microbes within the bacterial community

(Schmeisser et al., 2003). The genera Acidovorax and Ralstonia

were also found to be abundant ecotypes in their study

(Schmeisser et al., 2003). In other study, a novel strain closely

related to the genus Variovoraxwas also found in the drinking

water biofilm (Kalmbach et al., 1997). They also noticed that as

much as two-thirds of the autochthonous drinking water

population could be viable but non-culturable (VBNC)

(Kalmbach et al., 1997). This might be a reason why some

colonized bacteria, such as Staphylococcus, in the GI tracts were

not found in the water sample. According to Zavarzin et al.

(1991), members of the alpha subclass of Proteobacteria are

considered as mostly oligotrophic species. Recently, various

Bosea spp. have been isolated from hospital water supplies

(La Scola et al., 2003) and they could be also observed in

Fig. 3 e Bacterial richness of each genus isolated from cecum, colon, jejunum, and mouth based on 16S rRNA gene

sequencing. High numbers of Bosea, Staphylococcus and Ralstonia were observed. The numbers in columns show the

frequency of each genus.

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 0 5 0e5 0 5 85054

drinking water systems. In our study, these oligotrophic

bacteria e Bosea spp. e were found to be colonized in the GI

tracts which might be under nutrient-rich conditions even

though Bosea spp. were not isolated from the drinking water

samples.

Fig. 4 e Phylogenetic locations of the bacteria isolated from

GI tracts and drinking water with related type species as

reference (B: reference species; C: GI tracts; :: drinking

water). The tree was created using Neighbor-Joining

method using Maximum Composite Likelihood model

based on 75 sequences of 16S rRNA genes with similarity

greater than 97% to the related 16S rRNA sequences of the

reference species. The bacteria from GI tracts were more

interspersed in the phylogenetic tree than those from

drinking water. Bar indicates 1 substitution per 10

nucleotide positions.

3.3. Statistical analysis of population structure

Firstly, phylogenetic diversity between the two communities

was compared using lineage-per-time plots (Fig. 5). Both

communities showed convex-shaped lines, but the plot of the

0

5

10

15

20

25

30

35

40

1 2 3 4 5 6 7 8 9 10 11Time since common ancestry (arbitrary)

Nu

mb

er o

f lin

eag

es

Water GI tracts

Fig. 5 e Lineage-per-time plots for comparison of the

community diversity from drinking water and GI tracts

(Drinking water C; GI tracts ,). The microbial community

from GI tracts shows higher diversity than the one from

drinking water. The dashed linear line represents the

hypothetical trend at the same evolutionary rate of birth

and extinction.

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 0 5 0e5 0 5 8 5055

GI tracts had less convex shape, indicating higher divergent in

population than the drinking water.

Secondly, the phylogenetic diversity within each commu-

nity was compared using F statistics with Arlequin software

and using phylogenetic grouping of taxa (P test) with UniFrac.

Accumulative sequence variation in each group of the

community population, drinking water and GI tracts, was

computed and defined as “FST” (Holsinger andWeir, 2009). The

grouping of populations into drinking water and GI tracts

allowed the analysis of molecular variance (AMOVA) at two

levels: between drinking water and each part of the GI tracts,

and between drinking water and total GI tracts (Table 1). All

calculations were conducted with Arlequin software,

including random-permutation procedures to assess statis-

tical significance (Excoffier et al., 1992, 2005). Each 33 and 42

numbers of distinct sequences were found in drinking water

and GI tracts, respectively (Table 2). From these, they could be

further grouped into 10 (drinking water) and 21 (GI tracts)

distinct species. Gene diversity, nucleotide diversity and theta

value of the drinking water were 0.78� 0.06, 0.23� 0.11 and

35.36� 17.57, respectively. Gene diversity, nucleotide diversity

and theta value of the GI tracts were 0.95� 0.03, 0.31� 0.16,

52.12� 25.54, respectively. All three of these values indicated

that the bacterial community in the drinking water was less

divergent than that of the GI tracts.

4. Discussion

Drinking water systems are a distinctive habitat for micro-

organisms. Biofilm bacteria are important in water distribu-

tion system because drinking water is considered as

oligotrophic environment. Limitation or starvation with

respect to one or more nutrients is common to most bacteria

in drinking water (Lappinscott and Costerton, 1990). It is

known that biofilm in a sessile state in low-nutrient drinking

water is favored over free-living state for their growth

(Szewzyk et al., 1994). In order to investigate the fate of

introducing bacteria originated from biofilm-developed

drinking water system into GI tracts, two microbial commu-

nities were compared from the drinking water and the GI

tracts of abiotic mice. For the drinking water, both swab

samples of biofilm from the inner surface of water pipes and

water samples were collected. We collectively called both

samples as ‘drinking water’ in this study. After 16S rRNA

gene amplification, their identities were obtained from the

sequencing results and phylogenetic trees were investigated

with related type strain bacteria. Sequencing results revealed

Table 1 e Summary of FST and P tests for comparison ofmicrobial diversity between drinking water and GI tracts.

Group comparison P Value

FST Test P test

Drinking water vs. GI tracts 0.00000 0.07

Drinking water vs. Jejunum 0.00901 0.46

Drinking water vs. Cecum 0.00000 0.17

Drinking water vs. Colon 0.00901 0.33

that each three major groups of the genus, Acidovorax, Vari-

ovorax, Sphingopyxis, and Ralstonia, Staphylococcus and Bosea,

were found in the drinking water and the GI tracts, respec-

tively. Ralstonia, Sphingopyxis, Bacillus, E. coli and Meso-

rhizobiumwere found in both sources e drinking water and GI

tracts. It is noteworthy that E. coli was found both in the

drinking water and the GI tracts. E. coli is an indicator of

human and animal fecal contamination and has been used as

the biological indicator of water treatment safety (Edberg

et al., 2000), thus the presence of E. coli in the water

samples may suggest a possible fecal contamination origi-

nated from the source water (LeChevallier, 1990). Previously

well-known waterborne bacteria, such as Aeromonas and

Legionella, were not found in this study. Among the enteric

bacteria of major importance, such as Salmonella and E. coli,

which are not regarded as highly competitive microorgan-

isms in oligotrophic water ecosystems (Leclerc and Moreau,

2002), only E. coli was found in our study. An interesting

observation was that E. coli was isolated both from the

drinking water and the GI tract samples in this study. It has

been known that E. coli is one of the first bacterial genera,

along with Streptococcus, to colonize the intestine of animal

and human (Mackie et al., 1999) and E. coli exerts a barrier

effect against other Enterobacteriaceae group to colonize

further in a gut (Hudault et al., 2001). Therefore, it could be

inferred that the E. coli in the water might have colonized as

a first member in the GI tract and then their presence in the

gut could have influenced on the colonization of other

bacterial types.

Some species were isolated in the GI tract but not from the

water. This disagreement may be attributed to the possible

presence of VBNC bacteria in our AWS, in which biofilm had

been formed by operating for a long period of time in low-

nutrient water. All the major bacterial groups found in this

study show strong association with biofilm in drinking water

system. This biofilm could act as a place of harboring VBNC

bacteria (Juhna et al., 2007) and the VBNC bacteria may

resuscitate later and colonize in GI tracts. This phenomenon

could explain the discrepancy of the bacterial types between

the water and the GI tracts (Juhna et al., 2007).

A lineage-per-time plot shows that the both bacterial

communities had convex-shaped lines, suggesting an excess

of closely related ecotypes (Fig. 5). Generally, a convex line

implies that sudden evolutionary transformation has

occurred, not gradually, but quickly with highly selective

forces, and then only closely related species remained in the

community (Martin, 2002). This phenomenon may also

suggest that once colonization occurred successfully, then the

bacteria became tight ‘gnotobiotes’ in the GI tracts of the

germ-free mice. One of the selective forces could be nutrient

level in this case (low in water and high in GI tracts).

The gene diversity, nucleotide diversity and theta value in

the drinking water and the GI tracts were very similar to or

slightly less than other bacterial communities found in other

studies. For instance, nucleotide diversity and theta value of

sea sediment communities were 0.11e0.5 and 32.4e63.9,

respectively (Stach et al., 2003). Gene diversity, nucleotide

diversity and theta value for bacterial community of the

humanmouthwere 0.96, 0.3, and 341.7, respectively. In case of

the human gut, gene diversity, nucleotide diversity, and theta

Table 2 e Comparison of standard ecological and molecular estimates of sequence diversity for bacterial communitiesisolated from drinking water and GI tracts.

Community No. of distinct sequences No. of distinct species Gene diversity Nucleotide diversity Theta (Pi)

Drinking water 33 10 0.78� 0.06 0.23� 0.11 35.36� 17.57

GI tracts 42 21 0.95� 0.03 0.32� 0.16 52.12� 25.54

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 0 5 0e5 0 5 85056

value were 0.96, 0.30, and 154.8, respectively (Martin, 2002).

These are very similar to our own findings. Overall, our results

showed a similar range of the three values that were found in

the references. In general, if the population has abundant OTU

in a bacterial community analysis, it would likely result in

many different species (Stach et al., 2003). Thus, gene diversity

and nucleotide diversity would show high values. In our

study, the drinkingwater population had less genetic diversity

compared to other references, but the GI tracts showed almost

the same genetic diversity compared to other references.

Furthermore, the P test indicates the relationship between

the communities (Table 1). If P value is below a defined

threshold (0.05), the samples are considered to be significantly

different (Lozupone et al., 2006). For example, the microbial

diversity of shallow and deep-sea sediments from the Cariaco

Basin was compared (Madrid et al., 2001) and examined by

a P test (Martin, 2002). While the diversity analysis revealed

differences between the shallow-water (500 m) and deeper-

water (1310 m) samples, an analysis of the genetic diversity

suggested that the microbial species present in communities

were organisms from the same pool of diversity with the

result of P¼ 0.093 (Martin, 2002). Our result of P test was

P¼ 0.07e0.46, which is greater than 0.05, implying that the

two communities are not significantly different. Thus, it may

suggest an existence of correlation between the two

communities. This result agrees well with the phylogenetic

tree containing several clades of common bacterial species

(R. pickettii, S. panaciterrae, M. huakuii, and E. coli) in both water

and GI tract communities, but these clades are interspersed

throughout the phylogenetic tree (data not shown). Collec-

tively, the microbial communities in drinking water and GI

tracts had a suggestive relationship.

A significant FST test (0.00000e0.00901) coupled with an

insignificant P test (0.07e0.46) implies that the phylogenetic

tree contained several clades of closely related bacteria.

Collectively, it means that the bacterial groups in the drinking

water distribution system can be colonized in GI tracts.

Although not all taxa inwater and GI tracts could be culturable

and there might have been a bias in the cultivability of

bacteria, themicrobial communities in the drinkingwater and

the GI tracts were discovered to have an evident relationship

to each other. Culturing techniques were considered inher-

ently skewed because only a small fraction of all microor-

ganisms are able to grow fairly and rapidly on agar plates.

However, compared to cloning analysis, our used protocol

generated an appropriately significant amount to show the

relationship between bacterial community in the drinking

water and the GI tracts. Furthermore, the results of this study

emphasized the understanding of microbial diversity and

their proportions comprising the communities originated

from biofilm-containing water supply system. Another

notable finding was that Bacilluswas colonized in the GI tracts

of the tested mice. It has been known that even a small

percentage (1e2%) of the genus Bacillus could have cytotox-

icity (Leclerc and Moreau, 2002; Leclerc, 2003) and they could

be colonized and adhered as ‘tight residents’ on GI tracts

originated from a biofilm-covered drinking water source.

5. Conclusions

Our results suggest that consumption of drinking water that

passed through biofilm-contained distribution system had

impact on the colonization of microbial populations in GI

tracts. Several bacteria were found in the drinking water

samples and all belonged to Proteobacteria, such as the genera

Acidovorax, Variovorax, Sphingopyxis, Ralstonia and Novos-

phingobium. E. coli and Bacilluswere also found as minor in the

drinking water. In contrast, various genera were found in the

GI tracts including the genus Bosea, Ralstonia, Staphylococcus,

Bacillus, and Mesorhizobium. Minor groups were Bartonella,

E. coli, Pseudolabrys, Pseudomonas, Rickettsia, Sporolactobacillus,

Salinibacillus, Sphingopyxis, Streptococcus, and Zooshikella.

Lineage-per-time plots showed that the bacterial communi-

ties in the drinking water and GI tracts had a step-wise

ascending shape, suggesting that they had similar gene

diversity and that evolution occurred suddenly by highly

selective forces in these environments. It also suggested that

once colonization occurred, bacteria became tight ‘gnoto-

biotes’ in the GI tracts of the germ-free mice. A significant

P value of the FST test (0.00000e0.00901) coupled with an

insignificant P value of P test (0.07e0.46) indicated that both

communities had a suggestive relationship between the

drinking water-originated and the GI tract-associated bacte-

rial communities. As a suggestion, further study focusing on

the genus Ralstonia is recommended, in particular the species

of R. pickettii, in order to elucidate the fate of drinking water-

originated bacteria and their colonization in GI tracts. For

R. pickettii has been found frombiofilm in drinkingwater (Ryan

et al., 2006; Szymanska, 2007) and its infection has been

reported from contaminated water (Kendirli et al., 2004;

Labarca et al., 1999; Maroye et al., 2000; Moreira et al., 2005),

it may be a good microbial candidate to study waterborne

bacterial flora and its colonization in a colon system. For the

eventual success of colonization, hostemicrobe immune

responses and their interaction may play an important role

during the final stage of colonization.

Acknowledgements

This study was supported by the start-up fund from the

College of Public Health, The Ohio State University. We thank

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 0 5 0e5 0 5 8 5057

Kaedra Wetzel for her help in bacterial identification. We also

express our gratitude to Dr. Hua Wang for lending us the

NanoDrop system.

r e f e r e n c e s

Adiloglu, A.K., et al., 2004. Case report: nosocomial Ralstoniapickettii infection in neonatal intensive care unit. MikrobiyolBul 38 (3), 257e260.

Anesti, V., et al., 2005. Isolation and molecular detection ofmethylotrophic bacteria occurring in the human mouth.Environ. Microbiol. 7 (8), 1227e1238.

Anzai, Y., et al., 2000. Phylogenetic affiliation of thepseudomonads based on 16S rRNA sequence. Int. J. Syst. Evol.Microbiol. 50 (Pt 4), 1563e1589.

Bischofberger, T., et al., 1990. The bacterial flora of non-carbonated, natural mineral water from the springs toreservoir and glass and plastic bottles. Int. J. Food Microbiol.11 (1), 51e71.

Cooper, I.R., Hanlon, G.W., 2010. Resistance of Legionellapneumophila serotype 1 biofilms to chlorine-based disinfection.J. Hosp. Infect. 74 (2), 152e159.

Davis, D.H., et al., 1969. Proposal to reject the genusHydrogenomonas: taxonomic implications. Int. J. Syst.Bacteriol. 19, 375e390.

Edberg, S.C., Rice, E.W., Karlin, R.J., Allen, M.J., 2000. Escherichiacoli: the best biological drinking water indicator for publichealth protection. J. Appl. Microbiol. 88 (Suppl), S106eS116.

Edwards, S.V., 1993. Mitochondrial gene genealogy and gene flowamong island and mainland populations of a sedentarysongbird, the grey-crowned babbler (Pomatostomus temporalis).Evolution 47, 1118e1137.

Excoffier, L., et al., 1992. Analysis of molecular variance inferredfrom metric distances among DNA haplotypes: application tohuman mitochondrial DNA restriction data. Genetics 131 (2),479e491.

Excoffier, L., et al., 2005. Arlequin (version 3.0): an integratedsoftware package for population genetics data analysis. Evol.Bioinform. Online 1, 47e50.

Falsen, E., 1983. Immunodiffusion as an aid in routineidentification of uncommon aerobic gram negative bacteria.In: Leclerc, H. (Ed.), Gram Negative Bacteria of Medical andPublic Health Importance: Taxonomy-identification-application. Les editions de l’Institut National de la Sante et dela Recherche Medicale, Paris, pp. 477e483.

Hall, T.A., 1999. BioEdit: a user-friendly biological sequencealignment editor and analysis program forWindows 95/98/NT.Nucl. Acids Symp. Ser. 41 98e98.

Holsinger, K.E., Weir, B.S., 2009. Genetics in geographicallystructured populations: defining, estimating and interpretingF(ST). Nat. Rev. Genet. 10 (9), 639e650.

Hudault, S., et al., 2001. Escherichia coli strains colonising thegastrointestinal tract protect germfree mice against Salmonellatyphimurium infection. Gut 49 (1), 47e55.

Hugunin, K.M., et al., 2008. Effect of reverse osmosis,deionization, and ultraviolet filtration on biofilm developmentin automated watering manifolds. J. Am. Assoc. Lab. Anim.Sci. 47 (5), 131.

Hugunin, K.M., et al., 2009. Colonization of abiotic mice by biofilmbacteria. J. Am. Assoc. Lab. Anim. Sci. 48 (5), 625.

Juhna, T., et al., 2007. Effect of Phosphorus on survival ofEscherichia coli in drinking water biofilms. Appl. Environ.Microbiol. 73 (11), 3755e3758.

Kalmbach, S., et al., 1997. Isolation of new bacterial species fromdrinking water biofilms and proof of their in situ dominance

with highly specific 16S rRNA probes. Appl. Environ. Microbiol.63 (11), 4164e4170.

Kendirli, T., et al., 2004. Ralstonia pickettii outbreak associated withcontaminated distilled water used for respiratory care ina paediatric intensive care unit. J. Hosp. Infect. 56 (1), 77e78.

La Scola, B., et al., 2003. Bosea eneae sp. nov., Bosea massiliensis sp.nov. and Bosea vestrisii sp. nov., isolated from hospital watersupplies, and emendation of the genus Bosea (Das et al. 1996).Int. J. Syst. Evol. Microbiol. 53 (Pt.1), 15e20.

Labarca, J.A., et al., 1999. A multistate nosocomial outbreak ofRalstonia pickettii colonization associated with an intrinsicallycontaminated respiratory care solution. Clin. Infect. Dis. 29 (5),1281e1286.

Lane, D.J., 1991. 16S/23S RRNA Sequencing. John Wiley and Sons,Chichester, United Kingdom, pp. 115e175.

Lappinscott, H.M., Costerton, J.W., 1990. Starvation andpenetration of bacteria in soils and rocks. Experientia 46 (8),807e812.

Larkin, M.A., et al., 2007. Clustal W and Clustal X version 2.0.Bioinformatics 23 (21), 2947e2948.

LeChevallier, M.W., et al., 1987. Examination and characterizationof distribution system biofilms. Appl. Environ. Microbiol.53 (12), 2714e2724.

LeChevallier, M.W., 1990. Coliform regrowth in drinking-water e

a review. J. Am. Water Work Assoc. 82 (11), 74e86.Leclerc, H., Moreau, A., 2002. Microbiological safety of natural

mineral water. FEMS Microbiol. Rev. 26 (2), 207e222.Leclerc, H., 2003. Relationships between common water bacteria

and pathogens in drinking-water. Heterotrophic Plate CountsDrinking-Water Saf., 80e118.

Lozupone, C., et al., 2006. UniFrac e an online tool for comparingmicrobial community diversity in a phylogenetic context. BMCBioinform. 7, 371.

Lysenko, E.S., et al., June 2010. Within-host competition drivesselection for the capsule virulence determinant ofStreptococcus pneumoniae. Curr. Biol. 17. doi:10.1016/j.cub.2010.05.051.

Mackie, R.I., et al., 1999. Developmental microbial ecology of theneonatal gastrointestinal tract. Am. J. Clin. Nutr. 69 (5),1035Se1045S.

Madrid, V.M., et al., 2001. Phylogenetic diversity of bacterial andarchaeal communities in the anoxic zone of the Cariaco Basin.Appl. Environ. Microbiol. 67 (4), 1663e1674.

Maroye, P., et al., 2000. Investigation of an outbreak of Ralstoniapickettii in a paediatric hospital by RAPD. J. Hosp. Infect. 44 (4),267e272.

Martin, A.P., 2002. Phylogenetic approaches for describing andcomparing the diversity of microbial communities. Appl.Environ. Microbiol. 68 (8), 3673e3682.

Martiny, A.C., et al., 2003. Long-term succession of structure anddiversity of a biofilm formed in a model drinking waterdistribution system. Appl. Environ. Microbiol. 69 (11),6899e6907.

Martiny, A.C., et al., 2005. Identification of bacteria in biofilm andbulk water samples from a nonchlorinated model drinkingwater distribution system: detection of a large nitrite-oxidizing population associated with Nitrospira spp. Appl.Environ. Microbiol. 71 (12), 8611e8617.

Moreira, B.M., et al., 2005. Ralstonia pickettii and Burkholderiacepacia complex bloodstream infections related to infusion ofcontaminated water for injection. J. Hosp. Infect. 60 (1), 51e55.

Nee, S., et al., 1994. The reconstructed evolutionary process.Philos. Trans. R Soc. Lond B. Biol. Sci. 344 (1309), 305e311.

Oakley, B.B., et al., 2010. Evolutionary divergence andbiogeography of sympatric niche-differentiated bacterialpopulations. ISME J. 4 (4), 488e497.

Payment, P., et al., 1991. A randomized trial to evaluate the risk ofgastrointestinal disease due to consumption of drinking water

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 0 5 0e5 0 5 85058

meeting current microbiological standards. Am. J. PublicHealth 81 (6), 703e708.

Payment, P., et al., 1994. Blood agar to detect virulence factorsin tap water heterotrophic bacteria. Appl. Environ. Microbiol.60 (4), 1179e1183.

Pedersen, K., 1990. Biofilm development on stainless-steel andPVC surfaces in drinking water. Water Res. 24 (2), 239e243.

Reynolds, K.A., et al., 2008. Risk of waterborne illness viadrinking water in the United States. Rev. Environ. Contam.Toxicol. 192 (192), 117e158.

Ridgway, H.F., Olson, B.H., 1981. Scanning electron microscopeevidence for bacterial colonization of a drinking waterdistribution system. Appl. Environ. Microbiol. 41 (1), 274e287.

Rodrigues, A.L., et al., 2008. Characterization of biofilm formationon a humic material. J. Ind. Microbiol. Biotechnol. 35 (11),1269e1276.

Rogers, J., Keevil, C.W., 1992. Immunogold and fluoresceinimmunolabelling of Legionella pneumophila within an aquaticbiofilm visualized by using episcopic differential interferencecontrast microscopy. Appl. Environ. Microbiol. 58 (7),2326e2330.

Rooney-Varga, J.N., et al., 1999. Microbial communities associatedwith anaerobic benzene degradation in a petroleum-contaminated aquifer. Appl. Environ. Microbiol. 65 (7),3056e3063.

Ryan, M.P., et al., 2006. Ralstonia pickettii: a persistentgram-negative nosocomial infectious organism. J. Hosp.Infect. 62 (3), 278e284.

Schmeisser, C., et al., 2003. Metagenome survey of biofilms indrinking-water networks. Appl. Environ. Microbiol. 69 (12),7298e7309.

Slatkin, M., 1991. Inbreeding coefficients and coalescence times.Genet. Res. 58 (2), 167e175.

Smith, D., et al., 2005. Cooperative catabolic pathways within anatrazine-degrading enrichment culture isolated from soil.FEMS Microbiol. Ecol. 53 (2), 265e273.

Stach, J.E., et al., 2003. Statistical approaches for estimatingactinobacterial diversity in marine sediments. Appl. Environ.Microbiol. 69 (10), 6189e6200.

Szewzyk, U., et al., 1994. Growth and in-situ detection ofa pathogenic Escherichia coli in biofilms of a heterotrophicwater-bacterium by use of 16S-ribosomal-RNA-directed and23S-ribosomal-RNA-directed fluorescent oligonucleotideprobes. FEMS Microbiol. Ecol. 13 (3), 169e175.

Szymanska, J., 2007. Bacterial contamination of water in dentalunit reservoirs. Ann. Agric. Environ. Med. 14 (1), 137e140.

Takeuchi, M., et al., 2001. Proposal of the genus Sphingomonassensu stricto and three new genera, Sphingobium,Novosphingobium and Sphingopyxis, on the basis ofphylogenetic and chemotaxonomic analyses. Int. J. Syst. Evol.Microbiol. 51 (Pt. 4), 1405e1417.

Takezaki, N., et al., 1995. Phylogenetic test of the molecular clockand linearized trees. Mol. Biol. Evol. 12 (5), 823e833.

Tamura, K., et al., 2004. Prospects for inferring very largephylogenies by using the neighbor-joining method. Proc. Natl.Acad. Sci. U.S.A. 101 (30), 11030e11035.

Tamura, K., et al., 2007. MEGA4: molecular evolutionary geneticsanalysis (MEGA) software version 4.0. Mol. Biol. Evol. 24 (8),1596e1599.

Tokajian, S.T., et al., 2005. Phylogenetic assessment ofheterotrophic bacteria from a water distribution system using16S rDNA sequencing. Can. J. Microbiol. 51 (4), 325e335.

U.S. Environmental Protection Agency, 2002. Health Risks fromMicrobial Growth and Biofilms in Drinking Water DistributionSystems. D.C, Washington. http://www.epa.gov/safewater/tcr/pdf/biofilms.pdf.

Willems, A., et al., 1990. Acidovorax, a new genus for Pseudomonasfacilis,Pseudomonasdelafieldii, E. Falsen (EF) group13, EFgroup16,and several clinical isolates, with the species Acidovorax faciliscomb. nov., Acidovorax delafieldii comb. nov., and Acidovoraxtemperans sp. nov. Int. J. Syst. Bacteriol. 40 (4), 384e398.

Willems, A., et al., 1991. Comamonadaceae, a new familyencompassing the Acidovorans rRNA complex, includingVariovorax paradoxus gen. Nov., comb. Nov., for Alcaligenesparadoxus (Davis 1969). Int. J. Syst. Bacteriol. 41, 445e450.

Williams, M.M., et al., 2004. Phylogenetic diversity of drinkingwater bacteria in a distribution system simulator. J. Appl.Microbiol. 96 (5), 954e964.

Wong, W.C., et al., 2010. Efficacy of various chemical disinfectantson biofilms formed in spacecraft potable water systemcomponents. Biofouling 26 (5), 583e586.

Yan, X., et al., 2007. Cloning of environmental genomic fragmentsas physical markers for monitoring microbial populations incoking wastewater treatment system. Microb. Ecol. 53 (1),163e172.

Zavarzin, G.A., et al., 1991. A correlation of phylogenetic diversityin the Proteobacteria with the influences of ecological forces.Can. J. Microbiol. 37 (1), 1e6.