at SciVerse ScienceDirect

Anaerobe 20 (2013) 20e26

Contents lists available

Anaerobe

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

Clinical microbiology

MALDI-TOF MS fingerprinting facilitates rapid discrimination of phylotypes I, IIand III of Propionibacterium acnes

Elisabeth Nagy a,*, Edit Urbán a, Simone Becker b, Markus Kostrzewa b, Andrea Vörös c, Judit Hunyadkürti d

, István Nagy e

aHungarian Anaerobe Reference Laboratory, Institute of Clinical Microbiology, University of Szeged, 6725 Szeged, Semmelweis u 6, HungarybBruker Daltonik GmbH, 28359 Bremen, Fahrenheitstrasse 4, GermanycBay Zoltán Nonprofit Ltd., 6726 Szeged, Derkovits fasor 2, HungarydDepartment of General and Environmental Microbiology, University of Pécs, 7624 Pécs, Ifjúság útja 6, Hungarye Institute of Biochemistry, Biological Research Centre of the Hungarian Academy of Sciences, 6726 Szeged, Temesvári krt. 62, Hungary

a r t i c l e i n f o

Article history:Received 10 September 2012Received in revised form16 January 2013Accepted 29 January 2013Available online 26 February 2013

Keywords:Propionibacterium acnesTypingMALDI-TOF MSMLST

* Corresponding author.E-mail addresses: nagy.erzsebet@med.u-szeged.

med.u-szeged.hu (E. Urbán), simone.beckemarkus.kostrzewa@bdal.de (M. Kostrzewa), vorohunyadkurtij@baygen.hu (J. Hunyadkürti), nagyi@bay

1075-9964/$ e see front matter � 2013 Elsevier Ltd.http://dx.doi.org/10.1016/j.anaerobe.2013.01.007

a b s t r a c t

Matrix assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) is widelyused today for species determination of bacteria and fungi in routine microbiological laboratories, andcan also be used for subtyping of bacteria, such as Bacteroides fragilis. Propionibacterium acnes isfrequently referred to as an anaerobic skin commensal of relatively low pathogenicity. In addition to itsaccepted pathogenic role in acne, P. acnes is now emerging as an important opportunistic pathogen inmany other clinical situations, including late-stage prosthetic joint infections, osteomyelitis, endocarditis,endophthalmitis, post-neurosurgical infections and possibly prostate cancer. At the population geneticlevel, P. acnes can be differentiated into a number of distinct phylogroups, known as types IA1, IA2, IB, IC, IIand III, which may be associated with different types of infections and clinical conditions. The aim of thepresent study was to evaluate MS-based typing for resolution of these genetic groups after routineidentification by MALDI-TOF MS (Bruker MALDI Biotyper). The software package ClinProTools 2.2 wasused to analyze the protein based mass spectra of reference strains belonging to types IA, IB, IC, II and III.Phylogroup-specific peaks and peak shifts were then identified visually. In addition, peak variationsbetween the different types of P. acnes were investigated by using FlexAnalysis 3.3 software (Bruker). Adifferentiating library was created, which was used to type further 48 clinical isolates of P. acnes. Typingdata obtained by MALDI-TOF MS were then compared with the results from Multilocus Sequence Typing(MLST). Most of the clinical isolates (n ¼ 19) belonged to the type IA grouping according to MALDI-TOFMS. By MLST, all isolates were identified as type IA1. Twenty-one clinical isolates belonged to the type IBcluster based on both MALDI-TOF MS and MLST typing. Eight clinical isolates were identified as type IIstrains by both typing methods and all the type III reference strains could be distinguished by thepresence of a unique type III-specific peak (7238 Da) by the MALDI-TOF MS. Our study demonstrates thatMALDI-TOF MS is a reliable and powerful tool for rapid identification and typing of P. acnes strains fromthe main genetic divisions of the species.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Propionibacterium acnes (P. acnes) is a Gram-positive non-spor-eforming, anaerobic rod, which belongs to the normal flora of theskin, mucosa of the anterior nose, conjunctiva, mouth, upper

hu (E. Nagy), urban.edit@r@bdal.de (S. Becker),sa@baygen.hu (A. Vörös),gen.hu (I. Nagy).

All rights reserved.

respiratory tract, as well as the large intestine [1,2]. Beside thewidely investigated role of this bacterium in the pathogenesis ofacne vulgaris, P. acnes is now considered as an important oppor-tunistic pathogen in many other clinical situations, such as late-stage prosthetic joint infections, osteomyelitis, endocarditis,endophthalmitis, prostate cancer and post-neurosurgical infections[2e5]. Its ability to form a biofilm, may lead to its involvement inchronic periodontal diseases, as well as to be an under diagnosedcausative agent of late orthopedic implants infections [5,6].

As early as 1972 Johnson and Cummins [7] first revealed twodistinct phenotypes of P. acnes, designated as types I and II, which

E. Nagy et al. / Anaerobe 20 (2013) 20e26 21

were distinguished based on serological agglutination tests and cellwall sugar analysis. Several other methods for discriminating thetwo serotypes have been developed based on bacteriophage [8] andfermentation typing [9,10], as well as using monoclonal antibodiesand immunofluorescence microscopy [11]. By comparison of recA,and tly sequences, P. acnes isolates may be subdivided into phylo-types IA, IB, II and III [11,12]. Further subdivision came from theMultilocus Sequence Typing (MLST) approach, resulting in theidentification of over 100 sequence types [13e15]. Finally,enhanced MLST together with whole genome sequencings identi-fied unique signatures, such as the expression of dermatan-sulfateadhesins or the production of putative CAMP factor proteins, thatfurther subdivided type I isolates into IA1, IA2, IB and IC clusters[15e17]. Importantly, there are clear associations between someSTs and certain pathologic conditions, such as the association oftype IA1 clone ST1 with acne vulgaris; in contrast, other clones,such as type IB clone ST5 appear to be less pathogenic [14,15].

Mass spectrometry approaches for the identification of bacteriaand yeasts on species level are now gaining more and moreattention due to their accurate, inexpensive and rapid performance[18]. Thus, the aims of the present study were to evaluate thepossibility to use MALDI-TOF MS for rapid identification anddiscrimination of phylotypes I, II and III of P. acnes and to compareresults with MLST fingerprinting carried out blindly on 48 recentclinical isolates.

2. Materials and methods

2.1. Bacterial strains

A total of 61 P. acnes isolates were examined in this study. Mostof the strains (49) were fresh clinical isolates from the Institute ofClinical Microbiology of the University of Szeged, obtained fromvarious clinical samples disregardedwhether it was considered as areal pathogen or blood culture contamination. One of these isolates(76618 HU), obtained from a blood culture, was investigated at thevery beginning of this study byMLST typing and it proved to belongto the type II P. acnes strains. Twelve reference strains (Table 1,highlighted strains) with known typing results published earlier,including ATCC11828 (type II) from the American Type CultureCollection (ATCC) were included. Table 1 summarizes all the dataregarding the P. acnes isolates analyzed in this study. Clinical iso-lates of other Propionibacterium species (such as Propionibacteriumavidum, Propionibacterium granulosum and Propionibacterium pro-pionicum) were also included during the routine MALDI-TOF MSidentification, but were excluded from further study.

2.2. Bacterial culture

All bacterial strains were maintained at �80 �C in Brain HeartInfusion (BHI) broth (Oxoid, UK), containing 12% (v/v) glycerol, tillpending analysis. Organisms were cultured on Columbia agar base(Oxoid, UK) supplemented with 5% (v/v) bovine blood, hemin(1 mg ml�1) and vitamin K1 (5 mg ml�1) for 24 or 48 h at 37 �C in ananaerobic chamber (Concept 400; Ruskinn Technology, UK), underan atmosphere of 10% H2, 10% CO2, 80% N2 prior to analysis.Phenotypic identification was carried out according to the Wads-worth manual [19].

2.3Multilocus sequence typing (MLST) analysis

Genomic DNA was prepared as follows: 5 mg lysozyme (Sigma)was added to 2 ml of BHI broth culture and incubated at 37 �C for30 min. The suspension was centrifuged (13,000� g, 10 min atroom temperature), the pellet was resuspended in 200 ml of

AquaGenomic Kit (Aquaplasmid) and incubated at 90 �C for 30 min.The suspension was centrifuged as above and equal volume ofisopropanol was added to the supernatant. After vortexing andcentrifugation the pellet was washed with 70% (v/v) ethanol, driedat 40 �C for 15 min, resuspended in 30 ml PCR-grade water andallowed to rehydrate overnight at 4 �C. DNA concentration wasdetermined by Qubit (Life Technologies).

Partial sequences of the housekeeping gene loci aroE (424 bp),atpD (453 bp), gmk (400 bp), guaA (493 bp), lepA (452 bp), sodA(450 bp) and complete gene sequences from the putative virulencedeterminants tly (777 bp) and camp2 (804 bp) were amplified usingprimer pairs and amplification conditions previously described[15]. PCR products were purified with PCR fragment extraction kit(Geneaid) and sequenced on 3500 Series Genetic Analyzer (LifeTechnologies). Generated sequences were analyzed with GenomicsWorkbench 5.5 (CLC Bio); novel alleles for each locus were assigneda new allele number and distinct allelic profiles assigned a new STnumber. All allele sequences are available at (http://pubmlst.org/pacnes) [14,15].

2.4. Sample preparation for the MALDI-TOF MS measurement

One colony of each P. acnes strain, which had been subculturedfor 24 or 48 h in an anaerobic environment, was transferred into anEppendorf vial and carefully suspended in 300 ml of bidistilledwater and 96% (v/v) ethanol (900 ml) was added to the suspension,and mixed well. For MALDI-TOF MS analyzes the stabilized sampleswere sent from the Anaerobe Reference Laboratory in Hungary tothe Bruker laboratory in Bremen (Germany). Further samplepreparation was carried out as described previously [20]. In briefthe samples were centrifuged (13,000� g for 2 min), the superna-tants were removed and the pellets were dried at room tempera-ture and re-suspended in 50 ml of 70% aqueous formic acid and 50 mlof acetonitrile. After centrifugation (13,000� g for 2 min) 1 ml ofthe supernatant was placed onto the MALDI target plate and driedat room temperature, followed by adding 1 ml of MALDI matrix(solution of 10 mg/ml a-cyano-4-hydroxycinnamic acid in 50%acetonitrile/2.5% trifluoro-acetic acid).

2.5. MALDI-TOF MS measurement

All measurements were performed on a Microflex LT MALDI-TOF mass spectrometer (Bruker Daltonik, Bremen, Germany) inthe linear positive ion mode, with a laser frequency of 60 Hz. Themass rangewas 2000e20,000 Da. For each sample, the sum spectraof 240 single spectra were acquired, in portions of 40 single spectrafrom 6 different positions on a spot. For each sample, at least 20separate sum spectra were collected.

2.6. MALDI-TOF MS data processing

The mass spectra generated for the analyzed bacteria wereidentified as P. acnes strains (except a few other Propionibacteriumisolates which were excluded from the further study) by using theMALDI Biotyper 3.0 software and database (version 3.2.1.1., 4110entries, Bruker Daltonik, Bremen, Germany). Identification wasperformed with the standard pattern matching approach as rec-ommended by the manufacturer, applying log(score) � 2.0 forspecies identification, while a log(score) > 1.7 was indicative of aclose relationship at the genus level. The Bruker database containedmain spectra for 13 Propionibacterium spp. at the time of the in-vestigations. Clinical isolates belonging to other species thanP. acnes were clearly differentiated. For a subset of P. acnes strainsthe log(score)s obtained after 24 and 48 h incubation in an anaer-obic environment were compared.

Table 1Summary of the data of 61 P. acnes isolates analyzed in this study.

Isolate Country Source MLST MALDI-TOF Reference

Allelic profile ST Phylotype Phylotype

aroE-atpD-gmk-guaA-lepA-sodA-tly-camp2

55835 HU Cystitis 1-1-1-3-1-1-1-1 ST-1 IA1 IA3632 HU Blood culture 1-1-1-3-1-1-1-1 ST-1 IA1 IA4579 HU Skin sample 1-1-1-3-1-1-1-1 ST-1 IA1 IA12604 HU Pleural aspirate 1-1-1-3-1-1-1-1 ST-1 IA1 IA16634 HU Acne 1-1-1-3-1-1-1-1 ST-1 IA1 IA50460 HU Acne 1-1-1-3-1-1-1-1 ST-1 IA1 IA63006 HU Wound from the face 1-1-1-3-1-1-1-1 ST-1 IA1 IA19107 HU Abscess after surgery 1-1-1-3-1-1-1-1 ST-1 IA1 IA20550 HU Intraabdominal sample 1-1-1-3-1-1-1-1 ST-1 IA1 IA44064 HU Superficial wound 1-1-1-3-1-1-1-1 ST-1 IA1 IA64745 HU Abscess 1-1-1-3-1-1-1-1 ST-1 IA1 IA15614 HU Blood culture 1-1-1-3-1-1-2-2 ST-3 IA1 IA26048 HU Blood culture 1-1-1-3-1-1-2-2 ST-3 IA1 IA28585 HU Blood culture 1-1-1-3-1-1-8-6 ST-4 IA1 IAPRP-60a UK Acne 5-1-1-3-1-1-1-1 ST-20 IA1 IA [15]76793 HU Intraoral sample 1-1-1-3-1-14-1-1 ST-101 IA1 IA19695 HU Wound 1-15-1-3-1-1-1-1 ST-103 IA1 IA44264 HU Blood culture 21-1-1-3-1-1-1-6 ST-104 IA1 IA44261 HU Blood culture 21-1-1-3-1-1-1-6 ST-104 IA1 IA56853 HU Heart implant 1-1-1-3-1-1-1-6 ST-105 IA1 IAP.acn33a France Aqueous humor 1-1-1-5-1-4-8-2 ST-2 IA2 IA [15,16]P.acn17a France Corneal scrape 1-1-1-5-3-5-8-7 ST-22 IA2 IA [15,16]P.acn31a France Aqueous humor 1-1-1-13-1-4-8-2 ST-36 IA2 IA [15,16]6609a HU Normal skin 1-1-1-4-1-4-8-6 ST-5 IB IB [15,32,33]89071 HU Superficial wound 1-1-1-4-1-4-8-6 ST-5 IB IB96088 HU Skin sample 1-1-1-4-1-4-8-6 ST-5 IB IB102778 HU Joint aspirate 1-1-1-4-1-4-8-6 ST-5 IB IB1945 HU Pleural aspirate 1-1-1-4-1-4-8-6 ST-5 IB IB4837 HU Surgical sample 1-1-1-4-1-4-8-6 ST-5 IB IB10738 HU Skin abscess from the face 1-1-1-4-1-4-8-6 ST-5 IB IB16703 HU Otitis media 1-1-1-4-1-4-8-6 ST-5 IB IB20527 HU Blood culture 1-1-1-4-1-4-8-6 ST-5 IB IB14179 HU Blood culture 1-1-1-4-1-4-8-6 ST-5 IB IB23766 HU Wound from the face 1-1-1-4-1-4-8-6 ST-5 IB IB24728 HU Wound sample after trauma 1-1-1-4-1-4-8-6 ST-5 IB IB45808 HU Intrauterine devise 1-1-1-4-1-4-8-6 ST-5 IB IB51056 HU Intraoral sample 1-1-1-4-1-4-8-6 ST-5 IB IB58828 HU Secretion of conjunctiva 1-1-1-4-1-4-8-6 ST-5 IB IB57826 HU Secretion of conjunctiva 1-1-1-4-1-4-8-6 ST-5 IB IB61489 HU Appendicitis 1-1-1-4-1-4-8-6 ST-5 IB IB80329 HU Abscess 1-1-1-4-1-1-8-31 ST-12 IB IB74873 HU Blood culture 1-1-1-4-1-4-8-21 ST-42 IB IB67060 HU Superficial wound 1-1-1-4-1-4-8-21 ST-42 IB IB101320 HU Surgical sample 1-1-1-4-1-4-8-34 ST-98 IB IB24763 HU Intraabdominal sample 1-1-1-4-1-4-26-6 ST-99 IB IBPRP-38a UK Acne 9-1-4-8-6-8-14-14 ST-70 IC IB(IC) [15,17]72995 HU Surgical sample 15-4-2-4-2-3-10-10 ST-7 II II44073 HU Blood culture 15-4-2-4-2-3-10-10 ST-7 II II65763 HU Blood culture 15-4-2-4-2-3-10-10 ST-7 II II65311 HU Blood culture 17-9-2-4-2-3-10-10 ST-25 II IIATCC11828a USA Subcutaneous abscess 17-4-2-4-9-12-10-13 ST-27 II II [15,31,33]76618 HU Blood culture 17-4-2-4-2-6-10-12 ST-30 II II6187 HU Blood culture 17-4-2-4-2-6-10-12 ST-30 II II12553 HU Blood culture 15-4-2-4-2-3-24-10 ST-96 II II63848 HU Secretion of conjunctiva 20-4-2-4-2-6-10-12 ST-100 II II46361 HU Acne 15-4-2-19-2-3-10-10 ST-106 II II12Sa Brazil Progressive macular hypomelanosis 7-6-3-7-5-9-12-15 ST-32 III III [15]Asn12a UK Spinal disc material 7-6-3-7-5-9-13-16 ST-33 III III [15]VA2/5a HU Normal skin 7-6-3-7-5-9-13-16 ST-33 III III [15]Asn13a UK Spinal disc material 7-6-3-11-5-9-13-16 ST-73 III III [15]Asn10a UK Prosthetic hip 7-6-7-7-5-9-12-16 ST-81 III III/1 [15]

a Reference strains used in this study are highlighted.

E. Nagy et al. / Anaerobe 20 (2013) 20e2622

For the detection of differences between the mass spectra ofP. acnes strains belonging to different types we followed our pre-vious strategy to differentiate divisions I and II Bacteroides fragilisstrains [21]. Mass spectrum sets of well-defined P. acnes type I (fourIA, one IB and one IC), type II (one) and type III (three) referencestrains and one clinical isolate (76618 HU) for which the MLST

typing was carried out at the beginning of this study and proved tobelong to type II, were imported in the ClinProTools 2.2 software(Bruker Daltonik, Bremen, Germany). Spectra were normalized andrecalibrated, using the respective functionalities of the software.Subsequently, group-characteristic peaks and peak shifts weresearched for by careful visual investigation of the spectrum sets.

E. Nagy et al. / Anaerobe 20 (2013) 20e26 23

In addition, peak variations between the three groups wereinvestigated by using the FlexAnalysis 3.3 software (Bruker Dalto-nik, Bremen, Germany).

Simpson’s discriminatory index (D) values and adjusted Randvalues were estimated using the online tool at http://www.comparingpartitions.info.

3. Results

Eleven P. acnes strains: four types IA1 and IA2 (PRP-60, P.acn33,P.acn17, and P.acn31), one type IB (6609) one type IC (PRP-38), twotype II (ATCC11828 and a clinical isolatewith knownMLST type fromthe beginning of this study: 76618 HU) and three type III (12S, Asn12and Asn10) with known MLST types were selected for the firstmeasurement (Table 1, highlighted isolates). After the routine mea-surements with the MALDI Biotyper system, which differentiatedclearly all P. acnes strains from other species of Propionibacteriatested, with high log(score)s (mean score 2.34), differences in themass spectra were looked for. Overall, the spectra of the differentP. acnes strains exhibited a high degree of similarity with only a fewvariations in peak positions. However a pair of peaks showed a sig-nificant variation in masses, which were in line with the differentP. acnes types. Fig. 1 shows characteristic differences in the interval6900e7300 Da, visualized by the gel/stack view in ClinProToolssoftware between the types I, II and III P. acnes strains. Even differ-ences between the subtypes of IA and IB could be clearly seen (Fig.1).Table 2 shows the characteristic peaks enabling the discrimination ofthe different types of P. acnes. Type IC, a recently described newsubtype of P. acnes [14,15] gave a similar MS profile as type IB duringthe first evaluation in this mass range, however a further distinctpeak could be visualized in the region of 9900 and 10,000 Da (Fig. 2).As no further reference strain belonging to this subtypewas availableat the time of this screening and no further clinical isolate showedthis pattern we could not confirm the discriminatory power of thisfinding. Therefore, the specificity fordiscriminationof types IB and ICby this peak has to be proven with further strains in the future.

To investigate the possibility of predicting types and subtypes ofP. acnes strains based on the peak pattern determined in the first setof strains further 48 fresh clinical P. acnes isolates and 2 furtherreference strains belonging to the type III group of P. acnes ac-cording to the previousMLST typing (VA2/5 and Asn13) were tested

Fig. 1. Gel-like overview of mass spectra of types I, II and III P. acn

blindly. The MLST typing and MALDI-TOF measurements werecarried out independently. Altogether 19 clinical isolates werefound to belong to the type IA group of P. acnes using the MALDI-TOF measurement, however the IA1 and IA2 subgroups could notbe differentiated. By the MLST typing all of them proved to be IA1subtype isolates. No P. acnes clinical isolate was found in this study,which belonged to the type IA2 subtype by MLST typing (Table 1).Twenty-one clinical isolates were characterized as type IB both bythe MALDI-TOF MS and MLST. Eight clinical isolates proved tobelong to the type II group of P. acnes strains by both typingmethods. The two further reference strains from previous studiesbelonging to type III group of P. acnes strains with known MLSTtyping results gave the same pattern as the previously studied typeIII strains with the MALDI-TOF MS. Out of the five type III P. acnesstrains investigated by the MALDI-TOF MS one strain (Asn10) had,beside the common (and unique) peak for the type III isolates(7238 Da), a slightly different MS pattern with a peak 7063 insteadof all other type III strains which have a 7004 Da peak. We desig-nated it as type III/1 (Table 2, Fig. 3).

Using the MALDI-TOF approach, all 61 isolates could be differ-entiated into six distinct partitions that corresponded to the phy-lotypes IA, IB, IC, II, III and III-1, with an overallD value of 71% (95% CI65.0e76.6). Using themore discriminatoryMLSTapproach, a total of28 partitions that corresponded to distinct STswere resolved, with aD value of 90% (95% CI 83.4e94.9) (p < 0.001) based on this partic-ular isolate collection. The overall concordance between the twomethods as measured by the adjusted Rand valuewas 45.6% (95% CI26.6e65.5). While theMLSTapproachwas, as expected, muchmorediscriminatory, there was a good correlation between the twomethods when they were compared for phylogroup classificationonly, and not strain discrimination; although, strains identified astype IA by MALD-TOF could be differentiated into phylogroups IA1and IA2 byMLST. On that basis, the overall concordance between thetwo methods for phylogroup differentiation as measured by theadjusted Rand value was 91% (95% CI 81.9e100).

We carried out a pilot study to determine whether the incuba-tion time influences the MALDI Biotyper log(score)s for identifica-tion of P. acnes strains and the type specific peak pattern. It isknown that P. acnes produces pit point colonies after 24 h incuba-tion in an anaerobic chamber even if a suitable rich anaerobicmedia is used. Using the MALDI-TOF MS Biotyper identification

es reference strains; zoom into m/z region of differentiation.

Table 2Characteristic MS peaks for different types of P. acnes with previously known MLSTtyping results.

P. acnesisolates

CharacteristicMS peaks

MSphylotype

MLSTphylotype [15]

PRP-60 7034, 7180 IA IA1

P.acn33 7034, 7180 IA IA2

P.acn17 7034, 7180 IA IA2

P.acn31 7034, 7180 IA IA2

6609 7004, 7180 IB IBPRP-38 7004, 7180, 9950 IB/(IC) ICATCC11828 6985, 7525 II II76618a 6985, 7525 II II12S 7004, 7238 III IIIAsn12 7004, 7238 III IIIAsn10 7063, 7238 III/1 IIIVA2/5 7004, 7238 III IIIAsn10 7004, 7238 III III

Bold values represent possible discriminative peaks of MALDI-TOF MS for differentphylotypes.

a Recent blood culture isolate, the MLST was carried out at the beginning of thisstudy.

E. Nagy et al. / Anaerobe 20 (2013) 20e2624

system regularly it is recommended to carry out the measurementsafter 24 h of incubation time. By this time it is hardly possible to getenough sample to carry out the extraction procedure. We selected 7reference strains of P. acnes (belonging to type IA and type III) andmeasured the log(score)s after 24 and 48 h of incubation time usingthe extraction method described in the Materials and methods. Themean log(score) for the isolates after 24 h incubation time was2.302 (2.178e2.401) and it changed only slightly after 48 h incu-bation being 2.276 (1.965e2.472). The type specific peaks werealways clearly detectable, demonstrating that the incubation timedid not influence significantly MS based identification and typing ofP. acnes strains. In the case of all seven isolates the typical peaks todistinguishing type I and type III strains (Table 2) were identifiedafter both incubation times (data not shown).

4. Discussion

Since the introduction of MALDI-TOF MS in routine clinicalmicrobiological laboratories, it has developed froman experimental

Fig. 2. Gel-like overview of mass spectra of type IB and type IC reference s

tool to a rapid species identification technology with significantbenefit for routine aswell as for scientific investigations. It is appliedcurrently for the rapid species level identification of various bacteriaand fungi after a positive culture as it requires very limited time forsample preparation and measurement [18]. Less information isavailable today about the applicability of this technique for directidentification of pathogens from clinical samples or to use it fordetermination of the antibiotic resistance of isolated pathogens[22e24]. The differences in the MS peaks of isolates of the samespecies obtained by MALDI-TOF measurements give the possibilityof subtyping somemicroorganisms in a rapid and reproducibleway.Barbuddhe et al. [25] demonstrated that beside accurate identifi-cation of 146 Listeria strains (representing six species) at specieslevel, MALDI-TOF MS correctly subtyped all strains of Listeria mon-ocytogenes and the results corresponded to PFGE classification. TheMALDI-TOFmethodologywas also used for the subtyping of Yersiniaenterocolitica [26] and it yielded identical results compared to acharacterization by a combination of biotyping and serotyping. Thismethod was also used for the identification and typing of myco-bacteria and it was stated that MALDI-TOF MS has potential as arapid and reproducible platformbeside identification also for typingof Mycobacterium species [27]. In the field of anaerobic bacteria itwas possible to differentiate B. fragilis strains belonging to the di-vision I (which do not harbor the cfiA gene, responsible for thecarbapenem resistance) and division II (which always carry the cfiAgene) [21,28].

In the present study we confirmed that MALDI-TOF MS iscapable to discriminate the different, currently known main phy-lotypes of P. acnes (types IA, IB, IC, II and III) by selecting charac-teristic MS peaks for known type strains. The identification with alog(score) > 2.0 was always achieved for the P. acnes isolates andthe incubation time of 24 or 48 h did not influence the accuracy ofthe identification nor the determination of the types. By using ourdifferentiating scheme, 48 fresh clinical isolates of P. acnes could beput in the three main types after species determination and thetyping results were concordant with those obtained by the MLSTmethod. MALDI-TOF also identified type I subtypes placing them intwo groups: types IA and IB. Although we identified a peak at9950 Da possibly discriminating type IC, further strains are needed

trains of P. acnes showing the potentially differentiating 9950 Da peak.

Fig. 3. Gel-like overview of mass spectra of type III and type IA references strains of P. acnes, cultured for 48 or 24 h with the possible subtype of type III.

E. Nagy et al. / Anaerobe 20 (2013) 20e26 25

to confirm the discriminatory power of the selected peak patternfor type IC P. acnes isolates. For type III we found that some furthersub-differentiationmight be possible byMALDI-TOFMS, based on afurther specific peak shift.

Although MLST is considered as the gold standard for typing andgenetic population analysis in the case of most bacteria, it is a timeconsuming and costly typing method. Typing of P. acnes strains byMALDI-TOF MS using the extraction method needs less than 20 minincluding identification and uses the software showing the peakshifts. The same is for 96 samples may last for 2 h. The cost does notexceed the cost of the identification byMALDI-TOFMSwith<1 Eurofor consumables. It provides a possibility to carryout typing just afterthe identification in the routine clinical microbiological laboratories.

Several typing methods were developed during the past years,such as serotyping, immunofluorescence microscopy and differentmolecular genetic methods including the sequencing of the recA,and tly genes, MLST with the aim to differentiate normal floraP. acnes isolates from those, involved in severe infections such asacne, late-stage prosthetic joint infections, endocarditis, endoph-thalmitis, prostate cancer or post-neurosurgical infections [5e10,15e17]. Most of these studies were carried out separately fromthe routine clinical procedures on collected isolates. A recent studydemonstrated that types IA, IB and II of P. acnes strains werecommonly found on the healthy human skin of people originatingfrom Europe and from Japan [29], but this study also demonstratedthe global distribution of type III P. acnes strains on the skin ofpatients with atopic dermatitis and healthy population. In anotherstudy using the sequence polymorphism of recA gene, the distri-bution of types of P. acnes strains isolated from deep infections andfrom the skin of the forehead of healthy patients was compared andthey did not find significant differences between the prevalence ofIA, IB and II strains [30]. Using theMLST typing, McDowell et al. [15]found that acne was predominantly associated with type IA1 clonalcomplexes. In contrast type IA2 strains were recovered at a ratesimilar to types IB and II isolates. Ophthalmic infections werepredominantly associates with types IA1 and IA2 strains while typesIB and II were more frequently recovered from soft tissue andmedical device associated infections [15].

All these studies were carried out by time consuming moleculargenetic methods which are not suitable for routine laboratories andcannot be used to differentiate P. acnes isolates in real timewhetherthey are contaminants or real pathogens during an infectious

process. As MLST typing is not part of the routine clinical micro-biology procedures, the ability to use MALDI-TOFMS for typing andsubtyping of P. acnes after species identification in a reasonabletime, using the same equipment, may facilitate to find novel cor-relations between types/subtypes of this bacterium and patholog-ical conditions. It could be used for initial screening of P. acnesstrains directly in the clinical microbiological laboratories to selectisolates for more detailed MLST analyzes.

In conclusion our study demonstrates that MALDI Biotyper is areliable and powerful tool for rapid identification of P. acnes,furthermore using our discriminatory typing scheme, it allows thetimely discrimination of the main phylotypes I, II and III of P. acnesand further subtyping of types IA, IB and IC may also be possible.However, MLST provides a much deeper level of resolution withinthe main phylogroups of P. acnes identifying specific lineages orsequence types.

Potential conflict of interest

Simone Becker and Markus Kostrzewa are employees of theBruker Daltonik GmbH, the manufacturer of the mass spectrometryused in this study.

Acknowledgments

We thank Andrew McDowell for allocating new allele and STnumbers. This work was supported the French-Hungarian Associ-ated European Laboratory (LEA) SkinChroma OMFB-00272/2009and by TÁMOP-4.2.1.B-10/2/KONV-2010-0002.

References

[1] Perry A, Lambert P. Propionibacterium acne: infection beyond the skin. ExpertRev Anti Infect Ther 2011;9:1149e56.

[2] Eady EA, Ingham E. Propionibacterium acnes e friend or foe? Rev MedMicrobiol 1994;5:163e73.

[3] Nisbet M, Briggs S, Ellis-Pegler R, Thomas M, Holland D. Propionibacteriumacnes: an under-appreciated cause of post-neurosurgical infection.J Antimicrob Chemother 2007;60:1097e103.

[4] Brook I, Frazier EH. Infections caused by Propionibacterium species. Rev InfectDis 1991;13:819e22.

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

E. Nagy et al. / Anaerobe 20 (2013) 20e2626

[6] Ciardi JE, McCray GFA, Kolenbrander PE, Lua A. Cell-to-cell interaction ofStreptococcus sanguis and Propionibacterium acnes on saliva-coated hydroxy-apatite. Infect Immun 1987;55:1441e6.

[7] 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.

[8] Webster GF, Cummins CS. Use of bacteriophage typing to distinguish Pro-pionibacterium acnes types I and II. J Clin Microbiol 1978;7:84e90.

[9] Hugaki S, Kitagawa T, Kagoura M, Morohashi M, Yamagishi T. Correlationbetween Propionibacterium acnes biotypes, lipase activity and rash degree inacne patients. J Dermatol 2000;27:519e22.

[10] Kishishita M, Ushijima T, Ozaki Y, Ito Y. Biotyping of Propionibacterium acnesisolated from normal human facial skin. Appl Environ Microbiol 1979;38:585e9.

[11] 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.

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

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

[14] 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.

[15] McDowell A, Barnard E, Nagy I, Gao A, Tomida S, Li H, et al. An expandedmultilocus sequence typing scheme for Propionibacterium acnes: investigationof ‘pathogenic’, ‘commensal’ and antibiotic resistant strains. PLoS One 2012a;7:e41480.

[16] Vörös A, Horváth B, Hunyadkürti J, McDowell A, Barnard E, Patrick S, et al.Complete genome sequence of three Propionibacterium acnes isolates from thetype IA2 cluster. J Bacteriol 2012;194:1621e2.

[17] McDowell A, Hunyadkürti J, Horváth B, Vörös A, Bernard E, Patrick S, et al.Draft genome sequence of an antibiotic resistant Propionibacterium acnesstrain PRP-38 from the novel type IC cluster. J Bacteriol 2012b;194:3260e1.

[18] Bizzini A, Greub G. Matrix-assisted laser desorption ionization time-of-flightmass spectrometry, a revolution in clinical microbial identification. ClinMicrobiol Infect 2010;16:1614e9.

[19] Jousimies-Somer HR, editor. Wadsworth-KTL anaerobic bacteriology manual.6th ed. California, USA: Star Publishing Company; 2002.

[20] Nagy E, Maier T, Urban E, Terhes G, Kostrzewa M. Species identification ofclinical isolates of Bacteroides by matrix-assisted laser-desorption/ionizationtime-of-flight mass spectrometry. Clin Microbiol Infect 2009;15:796e802.

[21] Nagy E, Becker S, Sóki J, Urbán E, Kostrzewa M. Differentiation of division I(cfiA-negative) and division II (cfiA-positive) Bacteroides fragilis strains bymatrix-assisted laser desorption/ionization time-of-flight mass spectrometry.J Med Microbiol 2011;60:1584e90.

[22] Borovskaya A, Ilina E, Govorun V, Morozova O, Ditmarova J, Kruglov A, et al.Bacterial species identification directly from urine samples by MALDI-TOF MSfingerprinting. 19th ECCMID 2009; Abstract P1065.

[23] Stevenson LG, Drake SK, Murray PR. Rapid identification of bacteria in positiveblood culture broths by matrix assisted laser desorption ionization time-of-flight mass spectrometry. J Clin Microbiol 2010;48:444e7.

[24] Hrabák K, Studentová V, Walkova R, Zemlickova H, Jakubu V, Chudackova E,et al. Detection of NDM-1, VIM-1, KPC, Oxa-48, and OXA-162 carbapenemasesby matrix-assisted laser desorption ionization-time of flight mass spectrom-etry. J Clin Microbiol 2012;50:2441e3.

[25] Barbuddhe SB, Maier T, Schwarz G, Kostrzewa M, Hof H, Domann E, et al.Rapid identification and typing of Listeria species by matrix-assisted laserdesorption, ionization-time of flight mass spectrometry. Appl EnvironMicrobiol 2008;74:5402e7.

[26] Stephan R, Cernela Z, Ziegler D, Plüger V, Tonolla M, Ravasi D, et al. Rapidspecies identification and subtyping of Yersinia enterocolitica by MALDI-TOFmass spectrometry. J Med Microbiol 2011;87:150e3.

[27] Shitikov E, Ilina E, Chernousova L, Borovskaya A, Rukin I, Afanas’ev M, et al.Mass spectrometry based methods for the discrimination and typing ofmycobacteria. Infect Genet Evol 2012;12:838e45.

[28] Wybo I, De Bel A, Soetens O, Eschahidi F, Vandoorslaer K, Van Cauwenbergh M,et al. Differentiation of cfiA-negative and cfiA-positive Bacteroides fragilis iso-lates by matrix-assisted laser desorption ionization time-of-flight mass spec-trometry. J Clin Microbiol 2011;49:1961e4.

[29] Dekio I, Rajedram D, Morita E, Garbia S, Shah HN. Genetic diversity of Pro-pionibacterium acnes strains isolated from human skin in Japan and compar-ison with their distribution in Europe. J Med Microbiol 2012;61:622e30.

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

[31] Horváth B, Hunyadkürti J, Vörös A, Fekete C, Urbán E, Kemény L, et al. Genomesequence of Propionibacterium acnes type II strain ATCC11828. J Bacteriol2012;194:202e3.

[32] Hunyadkürti J, Feltóti Zs, Horváth B, Nagymihály M, Vörös A, McDowell A,et al. Complete genome sequence of Propionibacterium acnes type IB strain6609. J Bacteriol 2011;193:4561e2.

[33] Nagy I, Pivarcsi A, Koreck A, Széll M, Urbán E, Kemény L. Distinct strains ofPropionibacterium acnes induces selective human b-defensin-2 andinterleukin-8 expression in human keratinocytes through toll-like receptors.J Invest Dermatol 2005;124:931e8.