at SciVerse ScienceDirect

Anaerobe 18 (2012) 392e399

Contents lists available

Anaerobe

journal homepage: www.elsevier .com/locate/anaerobe

Clinical microbiology

Multilocus sequence typing and repetitive-sequence-based PCR (DiversiLab)for molecular epidemiological characterization of Propionibacterium acnesisolates of heterogeneous origin

Sabina Davidsson a,b,*, Bo Söderquist b,c, Fredrik Elgh d, Jan Olsson d, Ove Andrén a,b, Magnus Unemo b,c,Paula Mölling c

aDepartment of Urology, Örebro University Hospital, SE-701 85 Örebro, Swedenb School of Health and Medical Science, Örebro University, Örebro, SwedencDepartment of Laboratory Medicine, Clinical Microbiology, Örebro University Hospital, SE-701 85 Örebro, SwedendDepartment of Clinical Microbiology, Umeå University, Umeå, Sweden

a r t i c l e i n f o

Article history:Received 2 March 2012Received in revised form25 April 2012Accepted 30 April 2012Available online 15 May 2012

Keywords:Propionibacterium acnesMultilocus sequence typing (MLST)DiversiLabDiscrimination indexMolecular epidemiology

* Corresponding author. Department of Urology,SE-701 85 Örebro, Sweden. Tel.: þ46 196026604; fa

E-mail address: sabina.davidsson@orebroll.se (S. D

1075-9964/$ e see front matter � 2012 Elsevier Ltd.doi:10.1016/j.anaerobe.2012.04.015

a b s t r a c t

Propionibacterium acnes is a gram-positive bacillus predominantly found on the skin. Although it is consideredan opportunistic pathogen it is also been associatedwith severe infections. Some specific P. acnes subtypes arehypothesized to be more prone to cause infection than others. Thus, the aim of the present study was toinvestigate the ability to discriminate between P. acnes isolates of a refinedmultilocus sequence typing (MLST)method and a genotyping method, DiversiLab, based on repetitive-sequence-PCR technology.

The MLST and DiversiLab analysis were performed on 29 P. acnes isolates of diverse origins; orthopedicimplant infections, deep infections following cardiothoracic surgery, skin, and isolates from perioperativetissue samples from prostate cancer. Subtyping was based on recA, tly, and Tc12S sequences.

The MLST analysis identified 23 sequence types and displayed a superior ability to discriminate P. acnesisolates compared to DiversiLab and the subtyping. The highest discriminatory index was found whenusing seven genes. DiversiLab was better able to differentiate the isolates compared to the MLST clonalcomplexes of sequence types.

Our results suggest that DiversiLab can be useful as a rapid typing tool for initial discrimination of P. acnesisolates. When better discrimination is required, such as for investigations of the heterogeneity of P. acnesisolates and its involvement in different pathogenic processes, the present MLST protocol is valuable.

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

Propionibacterium acnes is part of the normal human skin floraand is commonly found in thenares, conjunctivae, oral cavity, vaginaas well as in the upper respiratory and intestinal tracts. P. acnes ismainly recognized in thepathogenesis of acnevulgaris [1]. However,P. acnes has also the ability to act as an opportunistic pathogen inother locations [2]. There are several reports indicating that P. acnesare the etiological agent of various low-grade infections, especiallyin implant-related infections including joint prostheses [2], pros-thetic valve endocarditis [3], sternal wound infections after cardio-thoracic surgery [4,5], and shunt-associated central nervous systeminfections [6]. In recent reports, P. acnes has also been frequentlyidentified in prostate tissue from patients with benign prostatic

Örebro University Hospital,x: þ46 196026650.avidsson).

All rights reserved.

hyperplasia and prostate cancer [7e9], indicating that a P. acnesinfection could contribute to chronic inflammation and, morespeculatively, prostate cancer initiation. Specific P. acnes subtypesmay be more prone to cause inflammation and infection [10e12],but further studies regarding this issue are needed.

To investigate the genetic heterogeneity among isolates of Pro-pionibacterium species various DNA-based typing methods havebeen developed [13,14]. Analysis of randomly amplified poly-morphic DNA [15] and pulsed-field gel electrophoresis (PFGE)[5,16,17] have proven useful for short term or local epidemiologyinvestigations, but these methods lack standardization as well asobjective interpretation. Alternatively, it is possible to sequence the16S rRNA gene, the recA gene, the putative hemolysin gene tly [18]and the transcarboxylase 12S gene (Tc12S) [8]. Sequencing of recAand tly has yielded four distinct P. acnes subtypes (IA, IB, II, III)[18e20]. Multilocus sequence typing (MLST) is based on analysis ofallelic variation in multiple housekeeping genes and thesequencing of internal fragments of these genes to identify clonal

Table 1Primers used for PCR and sequencing of recA, tly, Tc12S, and MLST for 29 Propioni-bacterium acnes isolates of diverse origin.

Primer Sequence (50 - 30) Ampliconsize (bp)

Ref.

lac up GCCGCAGCCTTGGGACTCT [11]lac dn GAAATGCTGTCGCCCCGTG 469 [11]oxc up GTGCTGCCGGAAAAGTCG [11]oxc dn CACCGGCGTCAGGATTGT 426 [11]fba up AGGACCCGCTATTTCAACTCTCA [11]fba dn ACGCGGGTCGTACATCTTCTT 532 [11]coa up GCGGGAATCGAGGGTGCTA [11]coa dn AGGGCCGCCGCTAGATAAGTA 541 [11]zno up CGCCGGCATCACCACCTATT [11]zno dn TCTCACATCGCCCGCAACC 507 [11]zno new dna CTCGGCGCAGAAATCGAGCA This studygms-gls up CCGCCTCACCGTCCAGCA [11]gms-gls dn CACATCGAGAACCGCATCACTC 538 [11]pak up CGACGCCTCCAATAACTTCC [11]pak dn GTCGGCCTCCTCAGCATC 454 [11]cel up GCCGACGTTTTCTACAGTGAGC [11]cel dn GGCGGTGAGGGTCCATTCA 432 [11]cob up CATCTCTGGCTCGCGAAGG This studycob dn TAGAACAGCACATGGGCCAC 637 This studytly- PAT-1 CAGGACGTGAGTGCAATGCGA [18]tly-PAT-2 TCGTTCACAAGACCACAGTAGC 830 [18]recA-PAR1 AGCTCGGTGGGGTTCTCTCATC [18]recA-PAR2 GCTTCCTCATACCACTGGTCATC 1030 [18]Tc12S-MMF CGGGTACGGCAAGATGTTCT [8]Tc12S-MMR GGAATATTGAACGAGTCGCAGA 562 [8]

a Used only in sequencing.

S. Davidsson et al. / Anaerobe 18 (2012) 392e399 393

lineages [21,22]. One MLST protocol specifically for P. acnes, mainlyemploying isolates from skin, has recently been published byLomholt and Kilian [11]. Another MLST protocol utilizing additionalgenes was presented recently by McDowell et al. [23].

DiversiLab (DL) system (bioMérieux, Marcy �lEtoile, France) isa commercially available repetitive-sequence-based PCR thatutilizes primers targeting noncoding repetitive sequences inter-spersed throughout the bacterial genome [24]. After amplification,the DNA fragments are separated by gel electrophoresis; theresulting genomic fingerprint then allows discrimination betweenisolates [25]. To our knowledge no previous study has evaluated DLfor molecular epidemiological investigations of P. acnes.

The aim of the present studywas to investigate the ability of twotechniques to discriminate P. acnes isolates isolates of diverseorigin: an MLST protocol and a repetitive-sequence-based PCR, DL.

2. Materials and methods

2.1. Bacterial isolates

Twenty-five clinical isolates and four reference strains of P. acneswere examined. The clinical isolates, collected in Sweden, wereselected to represent infections affecting different locations of thehuman body. We included prosthetic infections of the knee joint(n ¼ 3), hip joint (n ¼ 1), and femur implant (n ¼ 1); deep sternalinfections following cardiothoracic surgery (n ¼ 10); as well asisolates from perioperative tissue samples from the prostate(n ¼ 4). Skin isolates (n ¼ 6) and four P. acnes reference strainsisolated from different locations were also used; hip prosthesis(CCUG 35749), facial acnes (CCUG 1794), human pustule (CCUG36609) and one with unknown location (CCUG 38584).

2.2. Culture diagnostics

The culture and species verification of P. acneswas performed inaccordance with routine diagnostic procedures. Briefly, the isolateswere cultured, in an anaerobic atmosphere on FAA plates (4.6% LAB90 Fastidious Anaerobe Agar, LAB M, Lancashire, United Kingdom)supplementedwith 5% horse blood. The isolates were characterizedby colony morphology, Gram-staining, catalase and indole tests. Allsuspected P. acnes isolates were further confirmed to species levelby API 20 A (bioMérieux, Marcy �lEtoile, France).

2.3. MLST and typing based on recA, tly, and Tc12S sequencing

Isolates were sub-cultured on FAA plates, and two colonies weresuspended in 180 ml sterile distilled water including 100 units ofMutanolysin (SigmaeAldrich, Stockholm, Sweden) and incubatedfor 30 min at 37 �C. Thereafter 200 ml lysis buffer and 20 mlProteinase K (20 mg/ml) were added and the samples were incu-bated for 30 min at 56 �C. DNAwas subsequently isolated using theQIAamp DNA Mini Kit (Qiagen, Hilden, Germany), according to themanufacturer’s instructions. The DNA preparations were stored at4 �C prior to PCR.

Internal fragments of the housekeeping genes; lac (L-lactatedehydrogenase), oxc (Cytochrome c oxidase subunit II), fba (Fruc-tose bisphosphatealdolase), coa (O-succinylbenzoate-CoA), zno(Zn-dependant alcohol dehydrogenase), gms (Glutamyl-tRNAsynthetase), pak (Pantothenate kinase), cob (Cobalamin biosyn-thesis CobD/CbiB protein), and cel (Transcription regulator CelR)were amplified by PCR and sequenced. These genes were initiallyrecommended for genetic analysis of P. acnes but recently cob hasbeen replaced by recA at the MLST internet site (www.mlst.net).Lomholt and Kilian used the same genes except for cob, which theysubstituted for recA. Furthermore, the recA, tly and Tc12S genes

were also amplified by PCR and sequenced. All primers used for PCRamplifications and sequencing reactions were obtained fromScandinavian Gene Synthesis AB (Köping, Sweden) with theexception of for recA, tly and Tc12S (SigmaeAldrich), (see Table 1).The PCRs for all genes, except cob, were performed in a real-timeLight Cycler system (Roche Diagnostics, Mannheim, Germany)using SYBR Green I fluorescence melting curve analysis for detec-tion of specific amplicon. Each reaction mixture (20 ml) contained:2 ml Light Cycler FastStart DNA Master SYBR Green I (Roche Diag-nostics), 3 mM MgCl2, 0.5 mM primers and 2 ml DNA template. ThePCR programs started with a pre-incubation at 95 �C for 10 min,followed by 40 cycles of 95 �C for 10 s, annealing at differenttemperatures (55 �C for lac, fba, pak and cel; 58 �C for oxc, coa, gms,zno, tly and Tc12S; 66 �C for recA) for 10 s, and 72 �C for 37 s.

The PCR including melting curve analysis for the cob gene wasperformed in the real-time Rotorgene system (Qiagen). Each reac-tion mixture (20 ml) contained: 1X Rotor-Gene SYBR Green PCR Kit(Qiagen), 0.5 mM of each primer and 2 ml template. The thermalcycling conditions included an initial pre-incubation at 95 �C for10 min, and 40 cycles of 95 �C for 10 s and 60 �C for 30 s.

Prior to sequencing, the PCR products were purified usingMultiScreen PCRm96 plate (Millipore, Molsheim, France), accordingto the manufacturer’s instructions. One microliter of the purifiedPCR products were then cycle sequenced using 2 ml Big DyeTerminator v3.1, according to the manufacturer’s instructions(Applied Biosystems, Bleiswijk, Netherland). The cycle sequencingPCRs consisted of 25 cycles of 96 �C for 10 s, annealing at differenttemperatures (55 �C for lac, fba, pak and cel; 58 �C for oxc, coa, gms,zno, tly and Tc12S; 60 �C for cob and recA) for 5 s and 60 �C for 4 min.Subsequently, the reactions were purified using ethanol-sodiumacetate precipitation. The nucleotide sequences were determinedby capillary electrophoreses using an ABI PRISM 3130XL geneticanalyzer (Applied Biosystems). The amplicon sizes are presented inTable 1.

Multiple-sequence alignments of the nucleotide sequences of allloci were performed using BioEdit (version 7.0.0) software. Poly-morphism ratios, i.e. number of polymorphisms/100 bp� 100, were

S. Davidsson et al. / Anaerobe 18 (2012) 392e399394

calculated in the sequence alignment for each gene fragment. Eachdistinct gene sequence was assigned an allele number or allele type(AT) and each unique combination of the nine genes ATs wasassigned a new sequence type (ST). The most common ST wasassigned the lowest ST number.

P. acnes tly sequences were also compared to GenBanksequences with accession number AY527219 (Type IA), AY644408(Type IB), and AY644409 (Type II). P. acnes recA sequences werecompared to AY642055 (Type IA), EU687255 (Type IB), AY642061(Type II), and DQ672252 (Type III).

To identify different sequence variants phylogenetic analysiswasperformed with the software TREECON (version 1.3b) with the Jinand Nei substitution model, the Kimura evolutionary model, an a-valueof 0.5 and theneighbor-joiningmethod [26,27]. The sequencesof lac, fba, pak, cel, oxc, coa, gms, zno, and cob of the genomesequenced P. acnes strains SK137 (CP001977, Type IA) andKPA171202 (AE017283, Type IB) were downloaded from GenBankandalignedwith the examined sequences (single locus aswell as theconcatenated sequences of the nine examined loci). The sequencesofKPA171202wereused to root the respective tree. To identifyclonalcomplexes and their founders, the eBURSTv3 program (http://eburst.mlst.net) was used [28]. This web-based program analyzesthe samples regarding similarityof allelic profiles anddivides theSTsinto clonal complexes based on these similarities.

2.4. Repetitive-sequence-based PCR e DiversiLab

The DNA from the isolates was extracted with the UltraCleanMicrobial DNA Isolation kit (bioMérieux, Marcy �lEtoile, France)following the manufacturer’s instructions. ND-1000 spectropho-tometer (NanoDrop Technologies Inc, Wilmington, USA) was usedfor DNA quantification and all the samples were adjusted to containapproximately 25e50 ng/ml.

All the samples were amplified using the DiversiLab Propioni-bacterium Fingerprinting kit (bioMérieux, Marcy �lEtoile, France)following the manufacturer’s instructions. Each reaction mixture(25 ml) contained: 18 ml kit-supplied rep-PCRmastermix, 2 ml genomicDNA,0.5mlAmpliTaqpolymerase (AppliedBiosystems),2ml kit-specificprimer mix, and 2.5 ml of GeneAmp 10x PCR Buffer. The PCRs wereperformed on a thermal cycler (GeneAmpPCR System 9700, AppliedBiosystem). The thermal cycling conditions included an initial pre-incubation at 94 �C for 2 min, and 35 cycles of 94 �C for 30 s, 60 �Cfor 30 s, 70 �C for 90 s, and final extension of 70 �C for 3 min.

To detect the genomic fingerprints of each P. acnes isolate, theautomated microbial genotyping system (DiversiLab System) wasused. The amplified fragments were separated by electrophoresisperformed in microfluidics DNA LabChip and detected with anAgilent 2100 Bioanalyzer. One CCUG isolate served as positivecontrol in each round of analysis. The similarity between theisolates was analyzed by the DL software, version 3.3.40. In thisversion, the Kullback-Leiber method weighs the presence andabsence of peaks rather than peak intensities given that there areonly one or two large intensity peaks for P. acnes. The UnweightedPair Group Method with Arithmetic mean (UPGMA) was used asa clustering method and to create dendrograms and scatter plots[24]. The strain clustering was defined with a difference of up toone band in the dendrogram.

2.5. Discrimination index

To evaluate the discriminatory capacity of the typing methods,a discrimination index (D-index) [29] was calculated for combina-tions of theMLST genes and recA, tly, and Tc12S, as well as for the DLfingerprints. In short, the D-index represents the likelihood thattwo randomly picked strains from the sample are recognized as two

different strains using the evaluated method. The maximumpossible D-index is 1 (100%). Confidence intervals of 95% werecalculated as described by Grundmann et al. [30].

3. Results

3.1. recA, tly, and Tc12S subtyping

When compared with the sequences of recA and tly obtainedfrom GenBank, the P. acnes collection included all four previouslydescribed subtypes: IA (n ¼ 14), IB (n ¼ 7), II (n ¼ 7), and III (n ¼ 1).P. acnes subtypes IA and IB were found in isolates obtained fromskin, thoracic, prostatic, and prosthesis samples. P. acnes subtype II,however, could not be identified in any of the skin samplesanalyzed. The only subtype III isolate was found in a hip prosthesisinfection. Polymorphism ratios and the number of ATs in each geneare presented in Table 2. The collection was diversified into twoclusters when using Tc12S for typing.

3.2. MLST-typing

The MLST analysis based on the concatenated sequences of thenine genes (4536 bp) revealed 23 different STs among the 29isolates as shown in Table 3. The most common ST, ST1, was sharedby 6 isolates, all with subtype IA but with different origins. Oneisolate obtained from a thoracic surgery infection showed the sameST, ST10, as a prostatic isolate, both classified as subtype IB. Theremaining STs were represented by single isolates.

To further investigate and illustrate the genetic diversity amongthe P. acnes isolates, phylogenetic trees based on the individualMLST genes were constructed. Comparison of the trees showedconcordant phylogeny, in general, with clustering according to themajor subtypes I, II, and III (data not shown). Subsequently,a phylogenetic tree based on concatenated sequences of the nineMLST genes was constructed to illustrate the diversity observedwhen several genes are used for the analysis (data not shown). Animproved diversification was observed among subtype I isolateswhen using several genes compared with using a single gene. Theisolates were further separated into IA and IB clusters. The excep-tions were three IB isolates (#24, #25, #26), all with skin origin,which were related to 13 of 14 subtype IA isolates. The remaining IAisolate (#10) formed a distinct cluster with a single subtype IBisolate (#18). Furthermore, all subtype II isolates and the subtype IIIisolate clustered separately.

Calculated D-indices are presented in Table 2. The highest D-index for a single gene was found for fba (0.76), followed by gms(0.70), zno (0.65), and cel (0.62). By combining the seven MLSTgenes displaying the highest D-index; fba, gms, zno, cel, lac, cob, andcoa the maximum possible D-index was observed (0.96). Includingalso oxc and pak in the MLST protocol did not increase the D-indexfurther. In addition, replacing cob with recA (as in the previouslyproposed MLST protocol [11]) decreased the D-index to 0.94. Thephylogenetic tree based on the concatenated sequences of fba, gms,zno, cel, lac, cob, and coa (3656 bp) is shown in Fig. 1.

The eBURSTv3 analysis of the allele profile based on sevenalleles used to study the relationship between the different STsidentified four clusters. Fig. 2 shows STs in each cluster and thesingletons.

3.3. DiversiLab-typing

All of the P. acnes clinical isolates and the four reference strainswere successfully typed with the repetitive-sequence-based PCR.The dendrogram showed clustering according to the four majorsubtypes with a few differences (Table 3). All subtype II isolates

Table 2Polymorphism ratios, allele types, and discrimination index (D-index) based on recA, tly, Tc12S, and MLST genes of 29 Propionibacterium acnes isolates of diverse origin.

Gene Polymorphism ratio Number of allele types D-index 95% confidence interval

recA 1.6 4 0.67 0.574e0.771tly 4.0 4 0.67 0.574e0.771Tc12S 3.2 3 0.43 0.260e0.602recA tly 0.67 0.574e0.771reca tly Tc12S 0.67 0.574e0.771oxc 3.8 4 0.48 0.307e0.654pak 2.2 3 0.51 0.395e0.629coa 4.1 4 0.54 0.358e0.720cob 3.9 6 0.55 0.357e0.746lac 3.0 7 0.61 0.448e0.778cel 10.4 5 0.62 0.495e0.791zno 6.3 7 0.65 0.520e0.785gms 3.5 5 0.70 0.581e0.823fba 1.7 7 0.76 0.635e0.882fba gms 9 0.78 0.642e0.910fba gms zno 13 0.84 0.730e0.960fba gms zno cel 14 0.87 0.770e0.970fba gms zno cel lac 17 0.92 0.841e0.992fba gms zno cel lac cob 19 0.94 0.875e1.000fba gms zno cel lac cob coa 23 0.96 0.906e1.000fba gms zno cel lac cob coa pak 23 0.96 0.906e1.000fba gms zno cel lac cob coa pak oxc 23 0.96 0.906e1.000fba gms zno cel lac recA coa 21 0.94 0.877e1.000

*The polymorphism ratio: number of nucleotide polymorphisms/100 bp � 100.

S. Davidsson et al. / Anaerobe 18 (2012) 392e399 395

were clustered together with the exception of #17, which clusteredwith three out of seven subtype IB isolates (#2, #9, and #29). Theremaining four subtype IB isolates (#18, and #24e26) formeda cluster with all of the subtype IA isolates. In addition, DL

Table 3Dendrogram and typing results based recA, tly, DiversiLab, and multilocus sequence typi

differentiated the collection into six rep-PCR fingerprint patternscomprising more than one isolate (P1, P3, P5, P6, P7, P9) and fiveunique rep-PCR fingerprint patterns. The D-index of the DL methodwas 0,87. The repetitive-sequence-based PCR results are

ng of 29 Propionibacterium acnes isolates of diverse origin.

Fig. 1. Phylogenetic tree based on the concatenated sequences of fba, gms, zno, cel, lac, cob, and coa of 29 P. acnes isolates of diverse origins and the two genome sequenced P. acnesstrains, SK137 and KPA171202. Sk ¼ Skin, Thx ¼ Thoracic, PJI ¼ Prosthetic, Pro ¼ Prostate, CCUG ¼ Culture collection, University of Gothenburg, Sweden, followed by subtype.

S. Davidsson et al. / Anaerobe 18 (2012) 392e399396

summarized in Table 3. The reproducibility for duplicate DNAextracts amplified in the same PCR run on the same chip wasapproximately 99% for each isolate.

3.4. Comparison MLST and DiversiLab

MLST and DL clustered the 29 P. acnes isolates in a similar manneraccording to the four hitherto described subtypes (IA, IB, II, III). P1 wasdivided into threedifferent STs andP5andP7 into fourand sixdifferentSTs, respectively. Accordingly, the D-index of MLST was higher (0.96)than that of DL (0.87). The concordance of the twomethodswasmuchhigher when clustering based on clonal complexes of the STs wascompared to the rep-PCR fingerprint patterns. The only major differ-ence between the methods was observed in the eBURST cluster 1,which was divided into four different rep-PCR fingerprint patterns.Table 3 summarizes the results from ST, eBURST, and rep-PCR finger-print pattern analysis for the collection.

4. Discussion

Based on a collection of 29 P. acnes isolates we investigate thediscriminatory power of an MLST protocol compared to repetitive-sequence-based PCR (DL System). P. acnes subtypes IA and IB werefound in all examined anatomical locations, compared to subtype IIand III isolates that were absent in samples from the skin. Thesefindings are in concordance with previous studies, reportingsubtype I as the predominant phylotype on the skin, whereas typesII and III are rarely present at this site [11,31,32]. Both MLST and the

repetitive-sequence-based PCR method clustered four out of eightsubtype IB isolates with all the subtype IA isolates, indicating thatthey may genetically belong to subtype IA. This finding supportsthe theory of Lomholt and Kilian who demonstrated that somealleles used to identify subtype IB strains are shared with somesubtype IA strains [11].

MLST has been proven to be a reliable and applicable method tocharacterize many bacterial species due to its high discriminatorypower [22,33] and is usually based on seven genetic loci [21,22]. Inthe present study, an MLST approach based on internal fragmentsof nine loci were used in order to perform genetic characterizationand phylogenetic discrimination of 29 P. acnes isolates of diverseorigins. Among the 29 P. acnes isolates, 23 STs were identified. Thecalculated D-indices showed that high discriminatory power wasachieved when including only seven of the genes: fba, gms, zno, cel,lac, cob, and coa in the MLST protocol. When the two remaininggenes (pak and oxc) were added, no additional increase of the D-index was observed. A further observation was the relatively largedifference in discriminatory capacity between the by us proposedMLST protocol and recA sequencing with D-indices of 0.96 and 0.67,respectively. Given that we also found a decreased D-index (0.94)when replacing cobwith recA in the MLST protocol we believe thatthe MLST protocol presented in this study that includes the sevenhousekeeping genes can be used as a typing tool to study thephylogenetic relations among large populations of P. acnes isolates.Our findings are further supported by a study by Niazi et al. thatreported a lack of robustness of the recA phylotyping in their studyevaluating P. acnes in endodontic lesions [32].

Fig. 2. The eBURST analysis on the allele profile, of 29 P. acnes isolates of diverse origin, in the seven genes included in the proposed MLST protocol identified four clusters and foursingletons. Size of dots represents number of isolates, for example ST1 (n ¼ 6) and ST10 (n ¼ 2) and the remaining STs (n ¼ 1).

S. Davidsson et al. / Anaerobe 18 (2012) 392e399 397

Repetitive-sequence-based PCR has been used to distinguishamong strains of several different bacterial species, such asmethicillin-resistant Staphylococcus aureus [34], Staphylococcusepidermidis [35], and vancomycin-resistant Enterococcus spp. [36].Furthermore, some studies have been performed to evaluate thediscriminatory ability of MLST compared to DL [37,38]. Thesestudies have concluded that the main value of DL is for rapidscreening of the source of infection, while MLST exhibit superiordiscriminatory capacity. The findings of the present study are in line

with these earlier reports. First, we showed that DL clustered theP. acnes isolates into the four major subtypes, including the fourisolates classified as subtype IB based on recA and tly similar toMLST. Second, compared to the DL technique the MLST displayeda superior capability to diversify the isolates.

Based on our results we propose that DL can be a useful first-linetool for rapid screening of P. acnes isolates. The method is less techni-cally demanding and time consuming compared to MLST. Further-more, the total turnaround time for DL is approximately 1e2 days

S. Davidsson et al. / Anaerobe 18 (2012) 392e399398

compared to 2e3 days for MLST. However, for more thorough geneticinvestigations studying the clonal heterogeneity of P. acnes isolates,MLST will provide higher and more reliable discrimination. Inconcordance with our results, Lomholt and Kilian proposed the sameapproach in their recent investigation regarding the role of P. acnes inthe pathogenesis of acne [11].Weused the sameMLSTprotocol as theyin our study with one exception, namely the inclusion of the cob geneinstead of the recA gene.Our results showedhowever a higherD-indexwhen using cob in the MLST analysis. Therefore we propose that cobshould be included in the protocol to achieve as high discriminatorycapacity as possible in the MLST.

In our collection of P. acnes isolates of diverse origins, 23 STswere identified. This shows that several minor DNA alterationswithin the major four subtypes exist. Furthermore, subtypes IA, IB,and II were isolated from distinct sampling sites, which supportsthe hypothesis that some subtypes participate in specific patho-genic processes. It has recently been proposed that predominantlyP. acnes subtype I is playing an etiologic role in acne [11]. In addi-tion, P. acnes subtype II and III were the most prevalent phylotypesobtained from endodontic infections [32], orthopedic implants[12,39,40], and from radical prostatectomy specimens frompatients with prostate cancer [7,8].

Moreover, proteins secreted by the four major subtypes ofP. acnes have recently been identified. These secretomes harborseveral proteins involved in host-tissue degradation and inflam-mation. Although the majority of these proteins were the same forthe four presently known P. acnes subtypes, some distinct subtypesdifferences between were observed, including reduced lysozymesecretion by type IA strains compared with type IB strains [41]. Thisis in line with the theory that specific subtypes can be more proneto cause infection and inflammation.

The present study shows that both the DiversiLab and the MLSTare valuable techniques for molecular epidemiological investiga-tions. A limitation is the small number of P. acnes isolates in thecollection, which may cause a bias of the D-index. A strength of ourstudy, on the other hand, is the diverse origins from which theisolateswereobtained. Inaddition, the suggestedMLSTprotocolwasfound to have superior capacity to discriminate the P. acnes isolatescompared with using recA, tly or Tc12S genes individually or incombination, regardless of the small number of isolates analyzed.

Further studies are required to provide insight in the pathoge-netic potential of P. acnes in general and more specific the role ofspecific subtypes. In order to investigate the molecular epidemi-ology of the possible role for P. acnes in cancer initiation andprogression we are currently employing DL as first-line screening,and subsequently MLST on a large collection of P. acnes isolatesobtained from prostate tissue samples.

5. Conclusions

In conclusion, we propose that DL is the typing method to beused for rapid screening of larger collections of P. acnes isolates inepidemiological studies. We also present a genetic typing methodbased on anMLST protocol including seven housekeeping genes formore thorough molecular genetic investigation of the clonalheterogeneity of the P. acnes. Furthermore, the MLST protocol couldbe used to identify specific subtypes of P. acnes and their associa-tions with various pathologic diagnoses.

Acknowledgment

This research was supported by the Örebro County CouncilResearch Committee and the Foundation for Medical Research atÖrebro University Hospital, Sweden.

References

[1] Bojar RA, Holland KT. Acne and Propionibacterium acnes. Clin Dermatol 2004;22:375e9.

[2] Perry AL, Lambert PA. Propionibacterium acnes. Lett Appl Microbiol 2006;42:185e8.

[3] Mohsen AH, Price A, Ridgway E, West JN, Green S, McKendrick MW. Propio-nibacterium acnes endocarditis in a native valve complicated by intraven-tricular abscess: a case report and review. Scand J Infect Dis 2001;33:379e80.

[4] Friberg O, Svedjeholm R, Kallman J, Soderquist B. Incidence, microbiologicalfindings, and clinical presentation of sternal wound infections after cardiacsurgery with and without local gentamicin prophylaxis. Eur J Clin MicrobiolInfect Dis 2007;26:91e7.

[5] Unemo M, Friberg O, Enquist E, Kallman J, Soderquist B. Genetic homogeneity/heterogeneity of Propionibacterium acnes isolated from patients duringcardiothoracic reoperation. Anaerobe 2007;13:121e6.

[6] Viraraghavan R, Jantausch B, Campos J. Late-onset central nervous systemshunt infections with Propionibacterium acnes: diagnosis and management.Clin Pediatr (Phila) 2004;43:393e7.

[7] Alexeyev OA, Marklund I, Shannon B, Golovleva I, Olsson J, Andersson C, et al.Direct visualization of Propionibacterium acnes in prostate tissue by multicolorfluorescent in situ hybridization assay. J Clin Microbiol 2007;45:3721e8.

[8] Cohen RJ, Shannon BA, McNeal JE, Shannon T, Garrett KL. Propionibacteriumacnes associated with inflammation in radical prostatectomy specimens:a possible link to cancer evolution? J Urol 2005;173:1969e74.

[9] Fassi Fehri L, Mak TN, Laube B, Brinkmann V, Ogilvie LA, Mollenkopf H, et al.Prevalence of Propionibacterium acnes in diseased prostates and its inflam-matory and transforming activity on prostate epithelial cells. Int J MedMicrobiol 2011;301:69e78.

[10] Brzuszkiewicz E, Weiner J, Wollherr A, Thurmer A, Hupeden J, Lomholt HB,et al. Comparative genomics and transcriptomics of Propionibacterium acnes.PLoS One 2011;6:e21581.

[11] Lomholt HB, Kilian M. Population genetic analysis of Propionibacterium acnesidentifies a subpopulation and epidemic clones associated with acne. PLoSOne 2010;5:e12277.

[12] Sampedro MF, Piper KE, McDowell A, Patrick S, Mandrekar JN, Rouse MS, et al.Species of Propionibacterium and Propionibacterium acnes phylotypes asso-ciated with orthopedic implants. Diagn Microbiol Infect Dis 2009;64:138e45.

[13] de Carvalho AF, GautierM, Grimont F. Identification of diary Propionibacteriumspecies by rRNA gene restriction patterns. Res Microbiol 1994;145:667e76.

[14] Rossi F, Torriani S, Dellaglio F. Identification and clustering of dairy propio-nibacteria by RAPD-PCR and CGE-REA methods. J Appl Microbiol 1998;85:956e64.

[15] Perry AL, Worthington T, Hilton AC, Lambert PA, Stirling AJ, Elliott TS. Analysisof clinical isolates of Propionibacterium acnes by optimised RAPD. FEMSMicrobiol Lett 2003;228:51e5.

[16] Oprica C, Emtestam L, Lapins J, Borglund E, Nyberg F, Stenlund K, et al.Antibiotic-resistant Propionibacterium acnes on the skin of patients withmoderate to severe acne in Stockholm. Anaerobe 2004;10:155e64.

[17] Oprica C, Lofmark S, Lund B, Edlund C, Emtestam L, Nord CE. Genetic basis ofresistance in Propionibacterium acnes strains isolated from diverse types ofinfection in different European countries. Anaerobe 2005;11:137e43.

[18] McDowell A, Valanne S, Ramage G, Tunney MM, Glenn JV, McLorinan GC, et al.Propionibacterium acnes types I and II represent phylogenetically distinctgroups. J Clin Microbiol 2005;43:326e34.

[19] Johnson JL, Cummins CS. Cell wall composition and deoxyribonucleic acidsimilarities among the anaerobic coryneforms, classical propionibacteria, andstrains of Arachnia propionica. J Bacteriol 1972;109:1047e66.

[20] Valanne S, McDowell A, Ramage G, Tunney MM, Einarsson GG, O’Hagan S,et al. CAMP factor homologues in Propionibacterium acnes: a new proteinfamily differentially expressed by types I and II. Microbiology 2005;151:1369e79.

[21] Enright MC, Day NP, Davies CE, Peacock SJ, Spratt BG. Multilocus sequencetyping for characterization of methicillin-resistant and methicillin-susceptibleclones of Staphylococcus aureus. J Clin Microbiol 2000;38:1008e15.

[22] Maiden MC, Bygraves JA, Feil E, Morelli G, Russell JE, Urwin R, et al. Multilocussequence typing: a portable approach to the identification of clones withinpopulations of pathogenic microorganisms. Proc Natl Acad Sci U S A 1998;95:3140e5.

[23] McDowell A, Gao A, Barnard E, Fink C, Murray PI, Dowson CG, et al. A novelmultilocus sequence typing scheme for the opportunistic pathogen Propio-nibacterium acnes and characterization of type I cell surface-associated anti-gens. Microbiology 2011;157:1990e2003.

[24] Healy M, Huong J, Bittner T, Lising M, Frye S, Raza S, et al. Microbial DNAtyping by automated repetitive-sequence-based PCR. J Clin Microbiol 2005;43:199e207.

[25] Versalovic J, Koeuth T, Lupski JR. Distribution of repetitive DNA sequences ineubacteria and application to fingerprinting of bacterial genomes. NucleicAcids Res 1991;19:6823e31.

[26] Saitou N, Nei M. The neighbor-joining method: a new method for recon-structing phylogenetic trees. Mol Biol Evol 1987;4:406e25.

[27] Van de Peer Y, De Wachter R. TREECON for Windows: a software package forthe construction and drawing of evolutionary trees for the MicrosoftWindows environment. Comput Appl Biosci 1994;10:569e70.

S. Davidsson et al. / Anaerobe 18 (2012) 392e399 399

[28] Feil EJ, Li BC, Aanensen DM, Hanage WP, Spratt BG. eBURST: inferring patternsof evolutionary descent among clusters of related bacterial genotypes frommultilocus sequence typing data. J Bacteriol 2004;186:1518e30.

[29] Hunter PR, Gaston MA. Numerical index of the discriminatory ability of typingsystems: an application of Simpson’s index of diversity. J Clin Microbiol 1988;26:2465e6.

[30] Grundmann H, Hori S, Tanner G. Determining confidence intervals whenmeasuring genetic diversity and the discriminatory abilities of typingmethods for microorganisms. J Clin Microbiol 2001;39:4190e2.

[31] McDowell A, Perry AL, Lambert PA, Patrick S. A new phylogenetic group ofPropionibacterium acnes. J Med Microbiol 2008;57:218e24.

[32] Niazi SA, Clarke D, Do T, Gilbert SC, Mannocci F, Beighton D. Propionibacteriumacnes and Staphylococcus epidermidis isolated from refractory endodonticlesions are opportunistic pathogens. J Clin Microbiol 2010;48:3859e69.

[33] Maiden MC. Multilocus sequence typing of bacteria. Annu Rev Microbiol2006;60:561e88.

[34] Shutt CK, Pounder JI, Page SR, Schaecher BJ, Woods GL. Clinical evaluation ofthe DiversiLab microbial typing system using repetitive-sequence-based PCRfor characterization of Staphylococcus aureus strains. J Clin Microbiol 2005;43:1187e92.

[35] Trevino M, Martinez-Lamas L, Romero-Jung PA, Giraldez JM, Alvarez-Escudero J, Regueiro BJ. Endemic linezolid-resistant Staphylococcus epi-dermidis in a critical care unit. Eur J Clin Microbiol Infect Dis 2009;28:527e33.

[36] Pounder JI, Shutt CK, Schaecher BJ, Woods GL. Clinical evaluation of repetitivesequence-based polymerase chain reaction using the Diversi-Lab System forstrain typing of vancomycin-resistant enterococci. Diagn Microbiol Infect Dis2006;54:183e7.

[37] Al Nakib M, Longo M, Tazi A, Billoet A, Raymond J, Trieu-Cuot P, et al.Comparison of the Diversilab(R) system with multi-locus sequencetyping and pulsed-field gel electrophoresis for the characterizationof Streptococcus agalactiae invasive strains. J Microbiol Methods 2011;85:137e42.

[38] Lau SH, Cheesborough J, Kaufmann ME, Woodford N, Dodgson AR,Dodgson KJ, et al. Rapid identification of uropathogenic Escherichia coli of theO25:H4-ST131 clonal lineage using the DiversiLab repetitive sequence-basedPCR system. Clin Microbiol Infect 2010;16:232e7.

[39] Holmberg A, Lood R, Morgelin M, Soderquist B, Holst E, Collin M, et al. Biofilmformation by Propionibacterium acnes is a characteristic of invasive isolates.Clin Microbiol Infect 2009;15:787e95.

[40] Soderquist B, Holmberg A, Unemo M. Propionibacterium acnes as an etiologicalagent of arthroplastic and osteosynthetic infectionsetwo cases with specificclinical presentation including formation of draining fistulae. Anaerobe 2010;16:304e6.

[41] Holland C, Mak TN, Zimny-Arndt U, Schmid M, Meyer TF, Jungblut PR, et al.Proteomic identification of secreted proteins of Propionibacterium acnes. BMCMicrobiol 2010;10:230.