Identification and Subtyping of Clinically Relevant Human ... · nant mycoplasma (sub)species were included in a mycoplasma MSP database to complete the Bruker MALDI Biotyper database.

Identification and Subtyping of Clinically Relevant Human andRuminant Mycoplasmas by Use of Matrix-Assisted Laser DesorptionIonization–Time of Flight Mass Spectrometry

S. Pereyre,a,b,c F. Tardy,d,e H. Renaudin,a,b,c E. Cauvin,f L. Del Prá Netto Machado,a,g A. Tricot,e,d F. Benoit,f M. Treilles,f* C. Bébéara,b,c

Université de Bordeaux, USC EA371 Infections Humaines à Mycoplasmes et Chlamydiae, Bordeaux, Francea; INRA, USC Infections Humaines à Mycoplasmes etChlamydiae, Bordeaux, Franceb; Centre Hospitalier Universitaire de Bordeaux, Laboratoire de Bactériologie, Bordeaux, Francec; Anses, Laboratoire de Lyon, UMRMycoplasmoses des Ruminants, Lyon, Franced; Université de Lyon, VetAgro Sup, UMR Mycoplasmoses des Ruminants, Marcy L’Etoile, Francee; Laboratoire Départementalde la Manche, Service Santé Animale, Saint-Lô, Francef; Programa de Pós Graduação em Farmácia, Universidade Federal de Santa Catarina, Florianópolis, Brazilg

Matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) recently emerged as a technol-ogy for the identification of bacteria. In this study, we aimed to evaluate its applicability to human and ruminant mycoplasmalidentification, which can be demanding and time-consuming when using phenotypic or molecular methods. In addition,MALDI-TOF MS was tested as a subtyping tool for certain species. A total of 29 main spectra (MSP) from 10 human and 13 rumi-nant mycoplasma (sub)species were included in a mycoplasma MSP database to complete the Bruker MALDI Biotyper database.After broth culture and protein extraction, MALDI-TOF MS was applied for the identification of 119 human and 143 ruminantclinical isolates that were previously identified by antigenic or molecular methods and for subcultures of 73 ruminant clinicalspecimens that potentially contained several mycoplasma species. MALDI-TOF MS resulted in accurate (sub)species-level iden-tification with a score of >1.700 for 96% (251/262) of the isolates. The phylogenetically closest (sub)species were unequivocallydistinguished. Although mixtures of the strains were reliably detected up to a certain cellular ratio, only the predominant specieswas identified from the cultures of polymicrobial clinical specimens. For typing purposes, MALDI-TOF MS proved to clusterMycoplasma bovis and Mycoplasma agalactiae isolates by their year of isolation and genome profiles, respectively, and Myco-plasma pneumoniae isolates by their adhesin P1 type. In conclusion, MALDI-TOF MS is a rapid, reliable, and cost-effectivemethod for the routine identification of high-density growing mycoplasmal species and shows promising prospects for its capac-ity for strain typing.

In recent years, matrix-assisted laser desorption ionization–timeof flight mass spectrometry (MALDI-TOF MS) has emerged as a

new technology for species identification. MALDI-TOF MS is arapid, inexpensive, and accurate approach for species-level iden-tification of Gram-negative and Gram-positive bacteria and yeasts(1). It has been recently optimized for routine use in clinical mi-crobiology laboratories (2, 3). This method has shown to be suf-ficiently robust to be applicable with fastidious bacteria, such asanaerobic bacteria, Legionella, Bartonella, or mycobacteria (4).

Mycoplasmas are wall-less bacteria with minimal genomesbut are nonetheless important pathogens in both human andveterinary medicine (5). The family Mycoplasmataceae of theclass Mollicutes regroups two major genera, Mycoplasma andUreaplasma (6), which will be indicated here by their trivialname, mycoplasmas.

The identification of human and ruminant mycoplasmas ischallenging, and phenotypic methods cannot always achieve iden-tification to the species level. Furthermore, in contrast to moreclassical bacteria, mycoplasmas express few biochemical traits thatare useful for diagnosis. Nonetheless, several biochemical proce-dures have been standardized, such as the hydrolysis of urea andarginine, fermentation of glucose (7), and phosphatase activity(8), yet these may be time-consuming and not always discriminat-ing. For example, it is not possible to distinguish Ureaplasma par-vum from Ureaplasma urealyticum in culture because they bothhydrolyze urea; it is likewise not possible to distinguish Myco-plasma pneumoniae from Mycoplasma amphoriforme in culturebecause they both grow in Hayflick modified broth medium sup-

plemented with glucose (HG medium). Consequently, to identifythese species easily, important efforts have been made over thepast decades to develop diagnostic tools that rely on the antigenic(9, 10) and molecular identification of isolates, mainly based onPCR assays (7, 11–14). Nevertheless, the identification of certainmycoplasmas, such as subspecies belonging to the Mycoplasmamycoides cluster, comprising several important ruminant patho-gens that are closely related both antigenically and genetically, hasremained problematic (15, 16). Although genes encoding 16SrRNA have long been the usual targets for universal PCRs fol-lowed by sequence or amplicon analysis (17, 18), species-specificPCRs targeting 16S rRNA or other housekeeping genes are oftenfavored because the techniques are more rapid and do not requirea specialized laboratory (for reviews, see references 12 and 7).However, with �100 mycoplasmas that are currently recognized

Received 17 June 2013 Returned for modification 7 July 2013Accepted 25 July 2013

Published ahead of print 31 July 2013

Address correspondence to S. Pereyre, [email protected], orF. Tardy, [email protected].

* Present address: M. Treilles, Laboratoire d’Analyses de Sèvres Atlantique, Niort,France.

S.P. and F.T. contributed equally to this work.

Copyright © 2013, American Society for Microbiology. All Rights Reserved.

doi:10.1128/JCM.01573-13

3314 jcm.asm.org Journal of Clinical Microbiology p. 3314–3323 October 2013 Volume 51 Number 10

on July 6, 2020 by guesthttp://jcm

.asm.org/

Dow

nloaded from

as pathogens in humans and animals, individual PCR-based sys-tems that were developed as species-specific tests might be diffi-cult to handle in a laboratory; thus, the quest for a single generictest remains pressing. MALDI-TOF MS might be an interestinguniversal alternative for mycoplasma (sub)species-level identifi-cation, though its use has been hindered thus far by a lack ofdatabases of mycoplasmal MALDI-TOF reference spectra. Thetechnique, per se, has been assessed only once for the identifica-tion of Mycoplasma pulmonis, a rodent mycoplasma species, usinga limited database of three MALDI-TOF spectra from three rodentpathogen species (19). Indeed, no ruminant or human myco-plasma identification using MALDI-TOF MS has been reported todate, even though these bacteria represent important pathogensfor these hosts (5).

The aim of this study was to determine whether MALDI-TOFMS is a useful tool for the identification of ruminant and humanmycoplasmas at the species or subspecies level and as a subtypingmethod for certain species. A total of 29 reference peak lists, ormain spectra (MSPs), from 10 human and 13 ruminant myco-plasma (sub)species were first generated to construct a myco-

plasma spectral database and complete the database of the BrukerMALDI Biotyper platform. MALDI-TOF MS was then applied foridentification purposes to 119 human and 143 ruminant clinicalisolates that were previously identified to the species level usingphenotypic, antigenic, or molecular techniques and to 73 subcul-tures of clinical specimens known to potentially contain morethan one species of ruminant mycoplasmas. All analyses were op-timized to be compatible with the routine workflow of our labo-ratories. The accuracy of subtyping was assessed using clinicalstrains of M. pneumoniae, Mycoplasma bovis, and Mycoplasmaagalactiae.

MATERIALS AND METHODSReference strains, clinical isolates, and specimens. The reference strainsused to generate MSPs are listed in Table 1. All strains were grown at 37°Cin diverse media, and several strains were used as the reference strains. Forcomparison purposes, a maximum likelihood phylogenetic tree of the 16SrRNA sequences of these strains or species was computed with theMEGA5 software (http://www.megasoftware.net/). All the 16S rRNAsequences were downloaded from the Ribosomal Database project

TABLE 1 Reference strains used to generate the mycoplasma MSP database with their respective phylogenetic group, usual host, and culturemedium

Phylogenetic groupa and (sub)species Main host Strain Source or ATCC/NCTC no. Growth mediumc Other information

HominisMycoplasma agalactiae Ovine/caprine PG2 NCTC 10123 Modified PPLOM. agalactiae Ovine/caprine 5632 Nouvel et al. (30) Modified PPLOMycoplasma alkalescens Bovine PG51 ATCC 29103 Modified PPLOMycoplasma arginini Several animal hosts G230 ATCC 23838 Modified PPLOMycoplasma bovigenitalium Bovine PG11 ATCC 19852 Modified PPLOMycoplasma bovirhinis Bovine PG43 ATCC 27748 Modified PPLOMycoplasma bovis Bovine PG45 ATCC 25523 Modified PPLOM. bovis Bovine 1067 Hermeyer et al. (32) Modified PPLOMycoplasma canadense Bovine 275C ATCC 29418 Modified PPLOMycoplasma canis Canine/bovine PG14 ATCC 19525 Modified PPLOMycoplasma fermentans Human PG18 ATCC 19989 HG mediumMycoplasma hominis Human PG21 ATCC 23114 HA mediumM. hominis Human H34 ATCC 15056 HA mediumM. hominis Human M132 ATCC 43521 HA mediumMycoplasma orale Human CH 19299 ATCC 23714 HA mediumMycoplasma ovipneumoniae Ovine/caprine 14811 Strain being sequencedb Modified PPLOMycoplasma salivarium Human PG20 ATCC 23064 HA medium

SpiroplasmaMycoplasma capricolum subsp. capricolum Caprine California kid ATCC 27343 Modified PPLOMycoplasma mycoides subsp. capri Caprine GM12 ATCC 35297 Modified PPLOMycoplasma mycoides subsp. capri Caprine PG3 NCTC 10137 Modified PPLOM. putrefaciens Caprine KS1 ATCC 15718 Modified PPLOMycoplasma yeatsii Caprine GIH ATCC 51346 Modified PPLO

PneumoniaeMycoplasma amphoriforme Human MA-3663 Pereyre et al. (13) HG mediumMycoplasma genitalium Human G37 ATCC 33530 HG mediumMycoplasma penetrans Human GTU-54 ATCC 55252 HG mediumMycoplasma pneumoniae Human M129 ATCC 29342 HG medium Adhesin P1 type 1M. pneumoniae Human FH ATCC 15531 HG medium Adhesin P1 type 2Ureaplasma parvum Human 27 ATCC 27815 Shepard medium Serovar 3Ureaplasma urealyticum Human T960 ATCC 27618 Shepard medium Serovar 8

a Inferred from 16S rRNA gene sequences as defined in reference 25.b Bioproject accession no. PRJNA164759.c HA, Hayflick modified broth supplemented with arginine; HG, Hayflick modified broth supplemented with glucose. The compositions of the media are described in references 7and 9. Culture conditions were 37°C under a normal atmosphere for human species or 5% CO2 for ruminant species to the mid-log phase.

Identification of Mycoplasmas by MALDI-TOF MS

October 2013 Volume 51 Number 10 jcm.asm.org 3315

on July 6, 2020 by guesthttp://jcm

.asm.org/

Dow

nloaded from

(http://rdp.cme.msu.edu/index.jsp) or from the Nucleotide database atNCBI (http://www.ncbi.nlm.nih.gov/nuccore/).

Three clinical isolates of M. amphoriforme, one clinical isolate ofMycoplasma penetrans, and 50 clinical isolates of M. pneumoniae collectedat the Bordeaux University Hospital, France, between 2006 and 2011,together with four adhesin P1 variants 2a (MP-Ho, MP-Sta, MP-79809,and MP-79692) (20) and one variant 1 (MP-4817) (21) of M. pneumoniae,were used in the study. These isolates were previously identified using aspecific real-time PCR assay targeting the adhesin P1 gene (22) for M.pneumoniae and the 16S rRNA gene for M. amphoriforme (13) and using16S rRNA gene sequencing for M. penetrans. The isolates and variantswere cultured at 37°C in HG medium (7). Fifty-five clinical isolates of M.hominis collected at the Bordeaux University Hospital in 2009 and 2010that were previously identified by culture and confirmed by a specificreal-time PCR assay targeting the yidC gene (11) were cultured at 37°C inHayflick modified broth supplemented with arginine (HA medium) (7).Six clinical isolates of U. parvum and five clinical isolates of U. urealyticumcollected at the Bordeaux University Hospital and that were previouslyidentified by a real-time PCR assay targeting the urease gene (14) werecultured at 37°C in Shepard broth (7).

The clinical isolates from ruminants were grown in PPLO broth(Difco, France), supplemented as previously described (9), at 37°C and in5% CO2. The isolates were first identified using dot immunobinding onmembrane filtration (MF-dot) (9), the current method for routine strainidentification at the Anses Lyon Laboratory (16) within the framework ofthe epidemiosurveillance network VIGIlance to MYCoplasmoses of ru-minants (VIGIMYC) (23). The specimens consisted of (i) 48 M. agalactiaeisolates collected between 1951 and 2009 in France and Europe fromdomestic small ruminants and wild ungulates with contagious agalactia,(ii) 48 M. bovis isolates collected between 1986 and 2010 worldwide fromcattle with pneumonia or, to a lesser extent, mastitis (n � 5) or arthritis(n � 1), and (iii) 47 isolates from the M. mycoides cluster, mainly from M.mycoides subsp. capri (n � 19) and Mycoplasma capricolum subsp. capri-colum (n � 19), with a few Mycoplasma putrefaciens (n � 4) and Myco-plasma yeatsii (n � 5) isolates, all collected from ruminants with mastitis,arthritis, or pneumonia, or in the case of the M. yeatsii strains, fromasymptomatic animals.

The capacity of MALDI-TOF MS to simultaneously identify twostrains in mixed culture was also assessed by analyzing (i) artificially con-structed mixtures of strains and (ii) a total of 73 subcultures of bovinepneumonia clinical specimens known to often contain more than onemycoplasma species. The artificial mixtures were prepared by mixing var-ious volumes of culture in HA medium and Shepard medium for thehuman mycoplasmas or by mixing equal volumes of individual straincultures after cell counting using a “most probable number” method forthe ruminant species (24). The specimen subcultures had been routinelyanalyzed by MF-dot within the VIGIMYC network and kept frozen at�20°C prior to further analysis by MALDI-TOF MS in a unique run.

All analyses of the ruminant mycoplasmas were performed using asingle-blind protocol. The blinding was performed in Lyon, France, wherethe strains were grown, and the specimens were then sent to Saint-Lô,France, for protein extraction and MALDI-TOF MS analysis. Culturingand analysis of the human mycoplasmas were performed in Bordeaux,France.

Protein extraction for MALDI-TOF MS analysis. Mycoplasma pel-lets were obtained after centrifugation of various volumes of culture, de-pending on the growth capacities of the individual species. For the humanmycoplasmas, 30 ml of culture of reference strain was used to generate theMSP library, with the exception for Ureaplasma spp., for which 100 ml wasrequired. The culture volumes were the same for the identification of thehuman clinical isolates using MALDI-TOF MS, i.e., 30 ml for the M.pneumoniae clinical isolates and 100 ml for the Ureaplasma sp. isolates; theexception was for the M. hominis clinical isolates, for which a 1-ml culturewas sufficient. For the ruminant mycoplasmas, 1 ml of culture was usedfor both the MSP construction and identification of clinical isolates. The

mycoplasma pellets were washed once for the human mycoplasmas andtwice for the ruminant mycoplasmas in 1� phosphate-buffered saline(PBS) prior to being suspended in 300 �l of PBS or water, and 900 �l ofabsolute ethanol was added to precipitate the proteins. After centrifuga-tion at 20,000 � g for 2 min, the supernatant was removed and the pelletwas dissolved in an equal volume (20 �l or 50 �l) of 70% formic acid andacetonitrile before centrifugation for 2 min at 20,000 � g. One microliterof the supernatant was spotted onto a MALDI MSP 96 target polished steelplate (Bruker Daltonics, Bremen, Germany). After air-drying at roomtemperature, each sample was overlaid with 1 �l of matrix solution com-posed of 10 mg/ml �-cyano-4-hydroxycinnamic acid in 50% acetonitrileand 2.5% trifluoroacetic acid and air dried prior to the MALDI-TOF MSmeasurement.

MALDI-TOF mass spectrometry. Peptide mass fingerprint production spectra were generated using a Microflex LT Biotyper operating sys-tem (Bruker Daltonics, Bremen, Germany). The data were analyzed in theautomatic mode using the Bruker Biotyper 3.0 software and both taxon-omy libraries, i.e., that from Bruker and that generated in the presentstudy. A bacterial test standard (BTS) (part no. 255343; Bruker Daltonics,Inc.) was used in each run as a calibrator and for quality control.

Mycoplasma MSP database construction and isolate identificationby MALDI-TOF MS. Reference peak lists were assigned to different ref-erence strains (Table 1) to construct the mycoplasma MSP database,which was used in addition to the Bruker MALDI Biotyper MSP database3.1.2.0 (January 2011) for species-level identification of the clinical sam-ples. The MSPs were created according to the manufacturer’s recommen-dations by using the automated MSP creation functionality of the MALDIBiotyper software package (version 3.0). For each reference strain, 20individual mass spectrum measurements from 20 different spots of pro-tein extracts from human mycoplasma species or three independent massspectrum measurements of eight different spots of protein extracts fromruminant mycoplasma were acquired. The quality of individual raw spec-trum measurements was carefully checked (for absence of intrusive peaks,absence of flat-line spectra, low matrix background signal) and a mini-mum of 18 spectra of high quality were selected for MSP creation. Aftersmoothing, baseline correction, and peak picking, the resulting peak listswere used by the program to calculate and store an MSP containing theinformation about mean peak masses, mean peak intensities, and peakfrequencies.

For the identification of clinical isolates, the protein extracts werespotted once, twice, or 10 times, according to the species to be analyzed,and were classified by matching MSPs. The degree of spectral concor-dance was expressed as a logarithmic identification score ranging from 0to 3 and was interpreted according to the manufacturer’s instructions,with a modification of the score that was acceptable for probable species-level identification, which was lowered from �2.000 to �1.700.

The dendrograms were generated using the Biotyper 3.0 software.They were used to infer the relationships of mycoplasma species or strains,the closeness of which is reflected by an arbitrary distance level calculatedby the software. This distance level is a relative value and thus cannot becompared between dendrograms.

RESULTSHuman and ruminant mycoplasma MSP database construc-tion. To create the human and ruminant mycoplasma MSP data-base, 29 strains (listed in Table 1) were cultured in an appropriatemedium prior to protein extraction, MALDI-TOF MS, and MSPgeneration. These strains represent 10 human and 13 pathogenicor commensal mycoplasma (sub)species belonging to differentphylogenetic groups based on 16S rRNA sequences, as illustratedin Fig. 1A. To control the quality of this database, the MSPs gen-erated were further used to perform a hierarchical clustering ofstrains using the MSP dendrogram tool of the MALDI Biotypersoftware. The resulting score-oriented dendrogram (Fig. 1B)

Pereyre et al.

3316 jcm.asm.org Journal of Clinical Microbiology

on July 6, 2020 by guesthttp://jcm

.asm.org/

Dow

nloaded from

showed an overall topology that was comparable to that of thephylogenetic 16S rRNA tree defined by Weisburg et al. (25) (Fig.1A). When using an arbitrary distance level of 900 as a cutoff,all the strains were correctly assigned to their respective phyloge-

netic groups in the dendrogram. The pneumoniae, spiroplasma,and hominis groups yielded unequivocally separated branches,whereas both the hominis and pneumoniae groups were furthersplit into two branches. This was particularly intriguing for M.

FIG 1 Clustering of Mycoplasma and Ureaplasma strains as a function of their 16S rRNA phylogeny (A) or their MSP dendrogram (B). (A) Phylogenetic treeshowing positions of the strains used in this study. *, Strains for which a complete 16S rRNA gene sequence was not available were replaced with other strains ofthe same species. Only partial sequences were available for strains M. hominis H34 and M132, and these strains were consequently not included in the tree. Theevolutionary history of the strains was inferred using the maximum likelihood method. The tree with the highest likelihood is shown, and Acholeplasma laidlawiiPG8 was used as an outgroup. A total of 1,407 nucleotides from the 16S rRNA gene were used in the analysis. The data set was resampled 1,000 times, and thebootstrap percentage values (�50%) are given at the nodes. Phylogenetic groups as defined by Weisburg et al. (25) are also presented. (B) Score-orienteddendrogram of MSPs from human and ruminant mycoplasma species used to construct the mycoplasma database. The dendrogram was generated with thedefault settings in MALDI Biotyper (distance measure by correlation and linkage by average).

Identification of Mycoplasmas by MALDI-TOF MS

October 2013 Volume 51 Number 10 jcm.asm.org 3317

on July 6, 2020 by guesthttp://jcm

.asm.org/

Dow

nloaded from

penetrans strain GTU-54, which was isolated in a specific branch.We further examined this topological particularity by generatinganother MSP using a clinical strain of the M. penetrans species,namely, strain 6414. In the dendrogram, M. penetrans GTU-54and 6414 remained grouped on a unique branch and were sepa-rated from each other by a very short distance, as would be ex-pected for strains belonging to the same species.

Identification of clinical isolates. The mycoplasma coloniesare so small and often inlaid in the agar, such that in our study,direct colony deposition often failed to result in species- or genus-level identification, with the exception of (sub)species that grew toa high density, such as M. mycoides subsp. capri and some isolatesof M. hominis. Moreover, the extraction step was preferred formycoplasma species identification because attempts to deposit themycoplasma pellet directly onto the target plate resulted in weakerscores than when the complete extraction procedure was applied.Thus, broth culture and protein extraction were performed in allcases in this study. The required culture volume depended on thespecies to be identified (see below).

As the broth media that are used to grow mycoplasmas containa large concentration of proteins, the nonseeded HA, HG,Shepard, and PPLO medium batches were analyzed by MALDI-TOF MS. These samples were treated like the clinical isolates, asdescribed in Materials and Methods, and the resulting spectrawere compared to those in the mycoplasma and MALDI Biotyperdatabases. No spectral concordances were obtained, suggestingthat the proteins contained in the different culture media had nosignificant influence on the identification of the clinical isolates.Moreover, we tested the SP4 medium (7) instead of the usual HGmedium to grow the M. pneumoniae M129 reference strain beforeperforming MALDI-TOF MS. The strain was accurately identifiedas M. pneumoniae M129 with a score of �2.000, confirming thatchanging the culture medium had no influence on the identifica-tion of this species using MALDI-TOF MS.

For the M. hominis species and other species growing in HAmedium, preliminary tests revealed that protein extraction from aculture volume of 1 ml was sufficient to achieve species-level iden-tification. Of the 55 M. hominis clinical isolate extracts that werespotted twice, 100% were identified to the species level, 95% (52/55) had a score value of �2.000 for both spots, and 5% (3/55) hadonly one score of �2.000 and the other score in the range 1.700 to1.999. For M. pneumoniae and other species growing in HG me-dium and for the Ureaplasma species growing in Shepard me-dium, culture volumes of 30 ml and 100 ml, respectively, wererequired before protein extraction. Ten spots from the same pro-tein extract were deposited onto the target plate. Fifty clinicalisolates of M. pneumoniae and three clinical isolates of M. am-phoriforme were analyzed. In all cases, at least eight spots out of10 yielded scores of �1.700, corresponding to an accurate spe-cies-level identification. There were no misidentifications, andMALDI-TOF MS was able to distinguish M. pneumoniae fromMycoplasma genitalium, the phylogenetically closest species.Eleven clinical isolates of Ureaplasma were analyzed, consistingof six clinical isolates of U. parvum and five clinical isolates ofU. urealyticum. In 100% of the cases, at least eight spots out of10 yielded scores of �1.700, which allowed for accurate iden-tification to the species level.

For the ruminant mycoplasmas, each clinical sample was onlyspotted once to develop a procedure that was compatible with aroutine workflow. Of the first set, 96% (46/48) of the M. agalactiae

clinical isolates were successfully identified by MALDI-TOF MS,with scores of �1.700. Thirty-three strains had scores of �2.000and 13 had scores between 1.700 and 1.999. The only identifica-tion failures were due to poor growth of the strains, one of whichresulted nonetheless in an accurate identification of M. agalactiae,with a score of 1.5.

For M. bovis, 98% (47/48) of the analyzed isolates were cor-rectly identified, with 36 yielding scores of �2.000 and 11 yieldingscores between 1.700 and 1.999. The only strain that yielded anunsatisfactory score of 1.6 was strain 8790, which is known byboth multilocus sequence typing (MLST) and 16S rRNA sequenc-ing to have an intermediate position, between M. agalactiae andM. bovis, two highly related species of the hominis group (26).This strain was nonetheless assigned by MALDI-TOF MS to M.bovis.

Forty-seven strains from the M. mycoides cluster were alsoanalyzed, including several species and subspecies. For both M.capricolum subsp. capricolum and M. mycoides subsp. capri,95% (18/19) of the strains were correctly assigned to their re-spective subspecies, with scores of �1.700. Interestingly, theonly strain among the M. mycoides subsp. capri strains thatyielded a score of 1.56 was not M. mycoides subsp. capri, as wassuggested by its MF-dot profile; indeed, this strain was laterassigned by housekeeping gene sequence analysis to the newlydescribed species Mycoplasma feriruminatoris sp. nov. (27).The identification of the M. putrefaciens strains was congruentwith the MF-dot results despite 2/4 yielding a score of �1.700.In contrast, the identification scores (�1.300) were consideredunacceptable for M. yeatsii (n � 5).

Analysis of clinical specimens containing more than onemycoplasma (sub)species. Because human and ruminant clinicalspecimens are often simultaneously contaminated by differentmycoplasma species, we assessed the capacity of MALDI-TOF MSto address mixtures of strains. When two species, M. bovis andMycoplasma bovirhinis, U. parvum and M. hominis, or U. urealyti-cum and M. hominis, were artificially mixed in equivalent cellularconcentrations, both species were detected by MALDI-TOF MSwith an approximately equivalent score for each species. However,when the cellular ratio was unbalanced (from 1/5 to 1/100, de-pending on the species), the predominant species had a score of�2.000, whereas the species at the lower concentration was notdetected or received a score of �1.300, which was considered to beuninterpretable.

A total of 73 subcultures of the bovine pneumonia clinicalspecimens were also analyzed, 57 of which had previously beenidentified as a single strain by MF-dot (M. bovis, M. bovirhinis,Mycoplasma arginini, and Mycoplasma bovigenitalium) and 16 as amixture of 2 to 3 strains from different species, including M. bovisand one or two other species, such as Mycoplasma alkalescens, M.bovirhinis, Mycoplasma canadense, M. bovigenitalium, and M.arginini. Of the 57 samples containing a single strain, 97% (55/57)were identified by MALDI-TOF MS, with scores of �1.700 and aspecies match that was consistent with that obtained by MF-dot.Two subcultures yielded different identification results when ex-amined by MF-dot prior to storage at �20°C (identification of M.arginini) and by MALDI-TOF MS after storage (identification ofM. bovis, with scores of �1.700). This discrepancy was not theresult of a wrong identification but of a modification to the ratio ofthe species during the storage of the strain mixtures. For the 16specimens containing 2 to 3 species, as detected by MF-dot, M.

Pereyre et al.

3318 jcm.asm.org Journal of Clinical Microbiology

on July 6, 2020 by guesthttp://jcm

.asm.org/

Dow

nloaded from

bovis was the only component of the mixture that was identified byMALDI-TOF MS.

Mycoplasma subtyping. The spectra acquired for speciesidentification purposes were further used to assess the subtypingcapacity of MALDI-TOF MS for M. pneumoniae strains, and alsofor M. bovis and M. agalactiae strains. Based on the analysis of thegene encoding the P1 protein, M. pneumoniae is known to com-prise two subtypes (type 1 and type 2) and a few variants related toeach subtype (28). Using MALDI-TOF MS, all 50 M. pneumoniaespectral profiles were accurately clustered into 2 separate groupscorresponding to M. pneumoniae adhesin P1 type 1 (referencestrain M. pneumoniae M129) and M. pneumoniae adhesin P1 type2 (reference strain M. pneumoniae FH) (Fig. 2). In addition, threeM. pneumoniae variant 2a strains and one variant 1 strain wereaccurately ranked among the type 2 and type 1 strains, respectively(Fig. 2). We also searched for an association between the MALDI-TOF MS results and the recently developed multilocus variable-number tandem-repeat (VNTR) analysis (MLVA) typing methodthat is based on an analysis of the number of tandem repeats pres-ent at five loci of the M. pneumoniae genome (20). Although noassociation could be drawn because the isolates of the same MLVA

type were present in different branches of the dendrogram (Fig. 2),all MLVA types J, P, E, U, X, 29, and 31 were clustered in theadhesin P1 type 1 strains, and all MLVA types B, T, G, V, M, O, S,and H were clustered in the adhesin P1 type 2 strains. This findingwas expected because a correlation between the MLVA typingresults and the type of adhesin P1 gene was previously reported(20), with the former MLVA types being related to adhesin P1 type1 and the latter MLVA types being related to adhesin P1 type 2.

M. bovis and M. agalactiae are two highly related mycoplasmaspecies that are important pathogens of cattle and small rumi-nants, respectively. An MLST approach was recently proposed asan unequivocal tool for strain differentiation, characterization,and molecular typing of these two species (26). To infer whetherMALDI-TOF MS might be useful to type strains within each spe-cies, we generated a dendrogram using 18 strains from each spe-cies, all of which, except for the reference strains, were collected inFrance in different years, and from different hosts in the case of M.agalactiae (Fig. 3). In the resulting dendrogram, the M. agalactiaestrains were separated into two branches that correspond to thetwo reference strains, namely, PG2 and 5632. For M. bovis, thestrains were regrouped into two main branches according to their

FIG 2 Dendrogram of the MALDI-TOF MS profiles generated using MALDI Biotyper 3.0 with 50 clinical isolates of M. pneumoniae, M. pneumoniae referencestrains FH and M129, four M. pneumoniae variants 2a and 1 M. pneumoniae variant 1; distance measure was set at correlation, and linkage was set at average. Theadhesin P1 type and the MLVA type are given in parentheses, respectively. The reference strains M129 and FH are in bold type, and the variants are italicized. MP,M. pneumoniae.

Identification of Mycoplasmas by MALDI-TOF MS

October 2013 Volume 51 Number 10 jcm.asm.org 3319

on July 6, 2020 by guesthttp://jcm

.asm.org/

Dow

nloaded from

years of isolation. In the dendrogram, we also observed that the M.agalactiae group has longer branches than the M. bovis branch,which indicates a more variable species, an observation that is inagreement with the MLST data (26).

DISCUSSION

To assess the feasibility of using MALDI-TOF MS for the identi-fication of clinically relevant mycoplasmas in both human andveterinary medicine, we first attempted to enrich an availablemycoplasma database that only contained (as of the beginning ofthis study) one porcine mycoplasma species, Mycoplasma hyorhi-nis (Bruker MALDI Biotyper MSP database 3.1.2.0). For this pur-pose, 29 MSPs corresponding to 23 mycoplasma (sub)specieswere constructed, representing most of the human and ruminantpathogenic and commensal species distributed in three phyloge-netic groups of the class Mollicutes, namely, spiroplasma, hominis,and pneumoniae. The inclusion of pathogenic and commensalspecies in this database had the two purposes of (i) being able tounequivocally exclude the presence of pathogenic species by theidentification of nonpathogenic species and (ii) addressing a mix-ture of species in clinical specimens. For example, in ruminantrespiratory specimens, M. bovis, a pathogenic mycoplasma, is fre-quently associated with M. bovirhinis, which is a commensal spe-cies.

The MSP dendrogram resulted in a clustering of strains thatwas overall congruent with the 16S rRNA phylogeny, thoughsome groups were further subdivided into several branches (Fig.1). This observation was an important quality control for furtherspecies identification using this mycoplasma database. To limitthe potential misidentification of atypical strains from species thatare known to be variable, we included several strains from the

same species in the database. For instance, three strains of M.hominis were included, as this species is known to be highly het-erogeneous (29). Both M. agalactiae PG2 type strain and 5632were acquired as well, as they are considered to be situated at eachend of the genetic spectrum encountered in M. agalactiae (30).This proved successful because, for instance, several strains fromwild ungulates (e.g., Capra ibex) that are known to be geneticallydifferent from domestic ruminant strains (31) were correctlyidentified as M. agalactiae. For M. bovis, a strain known to bepathogenic (strain 1067) (32) was used together with the typestrain PG45. Lastly, within the recently reclassified M. mycoidessubsp. capri taxon, the former type strain of the Mycoplasmamycoides subsp. mycoides large colony (MmmLC) biotype,namely, GM12, was included in addition to the PG3 type strain.Generating several MSPs per species is also a convenient step forsubtyping strains once they have been assigned to a species (seebelow). While this study was being conducted, Bruker released anupdate of their own database (MALDI Biotyper MSP database3.3.1.0, April 2012), which included a total of 10 species of theMycoplasma genus. However, only six ruminant mycoplasma spe-cies were added, none of which are from the M. mycoides cluster,and no human mycoplasma species or strains from the Urea-plasma genus were included.

The mycoplasmal cultivation and sample preparation were op-timized to be compatible with the laboratory diagnostic workflow.The direct deposition of a colony or a cell pellet on the target platewas shown to occasionally yield species-level identification,though with scores weaker than those obtained with the proteinextraction process. As a consequence, as was already suggested forother bacteria (1), we chose to apply the protein extraction proto-col to increase the rates of identification. This protocol relies on a

FIG 3 Dendrogram of MALDI-TOF MS profiles generated using MALDI Biotyper 3.0 with 36 clinical isolates of M. agalactiae and M. bovis. The M. agalactiaestrains are designated Maga, followed by the strain number, the year of isolation, and the host species. The M. bovis strains are designated Mbov, followed by thestrain and the year of isolation in parentheses.

Pereyre et al.

3320 jcm.asm.org Journal of Clinical Microbiology

on July 6, 2020 by guesthttp://jcm

.asm.org/

Dow

nloaded from

series of centrifugation and resuspension steps that are, in ouropinion, compatible with routine usage. An increased number ofwashing steps performed on the mycoplasma pellet prior to ex-traction improved the spectral quality, and this was mainly due tothe removal of the strong protein background generated by theyeast extract and horse serum, two major components of myco-plasma growth media. Nonetheless, increasing the number ofwashing steps also resulted in a decrease in the quantity of theproteins that were contained in the final extract.

For clinical identification, a 1-ml culture was sufficient to ob-tain interpretable spectra for most species, except for those grow-ing in HG medium, such as M. pneumoniae, or in Shepard me-dium, such as Ureaplasma spp., for which 30 ml and 100 ml ofculture, respectively, were required. Handling such volumesmight not be compatible with a routine clinical workflow; how-ever, attempts to reduce these volumes while obtaining similarresults failed. These results are consistent with the cell densities ofthe individual cultures. For instance, M. hominis, one of the lessfastidious human mycoplasmas, typically reaches concentrationsof up to 108 to 109 color-changing units (CCU)/ml at the mid-logphase in HA medium, whereas M. pneumoniae and Ureaplasmaspp. only yield concentrations of 107 to 108 CCU/ml and 106 to 107

CCU/ml, respectively. Most of the ruminant species analyzed inthis study reached a concentration of 108 to 1010 CFU/ml at thestationary phase, and Stevenson et al. (33) determined that excel-lent spectra were obtained when a minimum of 106 CFU was spot-ted onto the target plate, which is approximately what we obtainedwhen starting from 1 ml of a 108-CFU/ml culture.

Bruker Daltonics recommends scores of �1.700 and �2.000for genus- and species-level identifications, respectively. How-ever, as has previously been demonstrated for other bacterial spe-cies (3, 34, 35, 36), our results strongly advocate a reduction of theacceptable scores for mycoplasma species identification to�1.700. With these parameters, the rate of species-level identifi-cation was 100% with no misidentifications, even for the low-cell-density species, i.e., Ureaplasma spp. and M. pneumoniae. Fur-thermore, several subspecies of the complex M. mycoides clusterthat are known to be closely related were unequivocally identified.Additionally, based on the high rate of identification to the (sub)species level, a reduction in the number of spots that were depos-ited on the target plate from the same isolate was considered whenappropriate. For species that grow in HA medium, such as M.hominis, one spot can be utilized for routine diagnosis because inour study, it yielded a score of �1.700 in all cases. In contrast, 10spots were utilized for more fastidious species that grow in HG orShepard medium. When considering the results that were ob-tained using the first three spots for the 50 M. pneumoniae isolatesand 11 Ureaplasma isolates, identification to the species level witha score of �1.700 for two out of these three spots was obtained in98% of cases (60/61), suggesting that three spots are sufficient forthese organisms. For the ruminant mycoplasmas, a single spot wasdeposited, leading to a correct identification in 95 to 98% of caseswith a score of �1.700. In cases of a misidentification or a score of�1.700, we recommend either relaunching the analysis with sev-eral spots or increasing the quantity of starting material.

We further demonstrated that MALDI-TOF MS can accom-modate in vitro-reconstituted mixtures of species up to a certainratio. Regarding human urogenital mycoplasma species, it mustbe noted that Ureaplasma spp. and M. hominis do not grow in thesame culture medium, which reduces the need to detect this mix-

ture of species by MALDI-TOF MS because the precultivation stepresults in a high predominance of the cultured species. However,such discrimination is of interest for ruminant species that grow inthe same medium, as we previously reported that �10% of clinicalspecimens might contain more than one mycoplasma (sub)spe-cies (23). However, in the 16 specimens containing two to threeruminant species, MALDI-TOF MS only detected M. bovis, theclinically relevant species, which may have been favored to thedetriment of other species during the storage phase at �20°C.Another potential explanation for these discrepancies betweenMF-dot and MALDI-TOF MS identification techniques might bebased on the slightly better sensitivity of MF-dot (from 2 � 104 to8 � 106 mycoplasma organisms per well [9]) versus MALDI-TOFMS (106 CFU per spot [33]), which can result in the codetection ofsome nonpredominant species when using MF-dot. In this study,different identification results were obtained for two clinical spec-imens that were not considered to be mixtures based on MF-dotresults, e.g., M. arginini by MF-dot and M. bovis by MALDI-TOFMS. These results might point toward another mixture of strainsin which M. bovis was overgrown by M. arginini when the MF-dotwas performed but for which the species ratio was modified by along period of storage at �20°C prior to MALDI-TOF MS iden-tification. In conclusion, as already shown by Stevenson et al. (33),not all organisms that are present in a polymicrobial sample can bereliably detected by MALDI-TOF MS.

MALDI-TOF MS allowed for the identification of Ureaplasmaspp. at the species level, which is generally not achieved with rou-tine laboratory diagnostic methods. Indeed, the former U. urea-lyticum species consisted of a heterogeneous species comprising 2biovars and 14 serovars. This taxon was split 10 years ago into twodistinct species, U. parvum and U. urealyticum, corresponding tothe former biovar 1 (including 4 serovars) and biovar 2 (including10 serovars), respectively (37). The distinction between the speciesrequires either antibody-based phenotyping methods, which areoften inconclusive because of multiple cross-reactions, or moreaccurate molecular techniques (14, 38). Although an additionalovernight incubation was necessary to obtain a 100-ml culturevolume, MALDI-TOF MS allowed for easy identification of Urea-plasma spp. to the species level, which is a benefit of the clinical useof MALDI-TOF MS because several studies have reported U. urea-lyticum to be more pathogenic than U. parvum (39, 40).

MALDI-TOF MS also proved to accurately distinguish speciesthat are known to be closely related, M. pneumoniae, M. genita-lium, and M. amphoriforme, which grow in the same HG medium,and M. bovis and M. agalactiae, which used to be classified as twosubspecies of the same species prior to being separated into twodifferent species based on serological, DNA-DNA reassociationexperiments (41), and 16S rRNA sequence data (42). Further-more, within individual species, the dendrograms generated fromthe MALDI-TOF MS spectra achieved epidemiologically relevantstrain clustering. For instance, the M. agalactiae strains were sep-arated into two branches that correspond to the two referencestrains, namely, PG2 and 5632, which are described as represent-ing each end of the genetic spectrum encountered in M. agalactiae(30). The strains from wild ungulates, namely, Capra ibex orchamois, interestingly were grouped together, as was previouslyreported with the use of partial sequencing of a housekeeping gene(31). For M. bovis, the strains were regrouped into two mainbranches that correspond to recent and old strains, a subclusteringthat is consistent with the recent data obtained in our group by

Identification of Mycoplasmas by MALDI-TOF MS

October 2013 Volume 51 Number 10 jcm.asm.org 3321

on July 6, 2020 by guesthttp://jcm

.asm.org/

Dow

nloaded from

MLST and MLVA (C. A. Becker, personal communication). Al-though these preliminary typing results are promising, furtherstudies that include a larger number of isolates are needed to con-firm these findings. Moreover, MALDI-TOF MS allowed for notonly the identification of M. pneumoniae isolates but also theiradhesin P1 typing in a single step, without the need of additionalmolecular typing techniques (28, 43, 44). However, no direct cor-relation was found between MALDI-TOF MS and MLVA typing.

In this study, we developed a mycoplasma spectral database ofhuman and ruminant mycoplasmas that proved to be reliable forthe identification of clinical isolates. A high concordance wasfound between the MALDI-TOF MS species identification andspecies identification using biochemical, antigenic, or molecularmethods. MALDI-TOF MS is a rapid, reliable, and cost-effectivemethod, particularly for the routine identification of M. hominisand ruminant mycoplasmas that grow to a high cell density, and itmight replace conventional identification methods in the future.Moreover, MALDI-TOF MS with no further optimization provedto be useful for the subtyping of several mycoplasma species andmay be promising for other typing developments.

ACKNOWLEDGMENTS

We thank Marc Bonneu and Jean-William Dupuy from the genomic plat-form of the University Bordeaux Segalen for their advice. We are gratefulto all VIGIMYC members, François Poumarat, and the technical staff inLyon responsible for managing the collection of the ruminant mycoplas-mas strains. We thank Marie Gardette for technical assistance.

REFERENCES1. Croxatto A, Prod’hom G, Greub G. 2011. Applications of MALDI-TOF

mass spectrometry in clinical diagnostic microbiology. FEMS Microbiol.Rev. 36:380 – 407.

2. Ford BA, Burnham CA. 2013. Optimization of routine identification ofclinically relevant Gram-negative bacteria by use of matrix-assisted laserdesorption ionization–time of flight mass spectrometry and the BrukerBiotyper. J. Clin. Microbiol. 51:1412–1420.

3. McElvania Tekippe E, Shuey S, Winkler DW, Butler MA, Burnham CA.2013. Optimizing identification of clinically relevant Gram-positive or-ganisms by use of the Bruker Biotyper matrix-assisted laser desorptionionization–time of flight mass spectrometry system. J. Clin. Microbiol.51:1421–1427.

4. Biswas S, Rolain JM. 2013. Use of MALDI-TOF mass spectrometry foridentification of bacteria that are difficult to culture. J. Microbiol. Meth-ods 92:14 –24.

5. Razin S, Yogev D, Naot Y. 1998. Molecular biology and pathogenicity ofmycoplasmas. Microbiol. Mol. Biol. Rev. 62:1094 –1156.

6. Johansson KE, Pettersson B. 2002. Taxonomy of Mollicutes, p 1–29. InRazin S, Herrmann R (ed), Molecular biology and pathogenicity ofmycoplasmas. Kluwer Academic/Plenum Publishers, New York, NY.

7. Waites KB, Bébéar CM, Robertson JA, Talkington DF, Kenny GE (ed).2001. Cumitech 34, Laboratory diagnosis of mycoplasmal infections.American Society for Microbiology, Washington, DC.

8. Poveda JB. 1998. Biochemical characteristics in mycoplasma identifica-tion, p 69 –70. In Miles RJ, Nicholas RAJ (ed), Methods in molecularbiology, vol 104: mycoplasma protocols. Humana Press, Inc., Totowa, NJ.

9. Poumarat F, Perrin B, Longchambon D. 1991. Identification of rumi-nant mycoplasmas by dot immunobinding on membrane filtration (MFdot). Vet. Microbiol. 29:329 –338.

10. Poveda JB, Nicholas R. 1998. Serological identification of mycoplasmasby growth and metabolic inhibition tests. Methods Mol. Biol. 104:105–111.

11. Férandon C, Peuchant O, Janis C, Benard A, Renaudin H, Pereyre S,Bébéar C. 2011. Development of a real-time PCR targeting the yidC genefor the detection of Mycoplasma hominis and comparison with quantita-tive culture. Clin. Microbiol. Infect. 17:155–159.

12. Markham PF, Noormohammadi AH. 2005. Diagnosis of mycoplasmosisin animals, p 355–382. In Blanchard A, Browning G (ed), Mycoplasmas:

molecular biology pathogenicity and strategies for control. Horizon Bio-science, Norfolk, United Kingdom.

13. Pereyre S, Renaudin H, Touati A, Charron A, Peuchant O, Hassen AB,Bébéar C, Bébéar CM. 2009. Detection and susceptibility testing of My-coplasma amphoriforme isolates from patients with respiratory tract infec-tions. Clin. Microbiol. Infect. 16:1007–1009.

14. Yi J, Yoon BH, Kim EC. 2005. Detection and biovar discrimination ofUreaplasma urealyticum by real-time PCR. Mol. Cell. Probes 19:255–260.

15. Le Grand D, Saras E, Blond D, Solsona M, Poumarat F. 2004. Assess-ment of PCR for routine identification of species of the Mycoplasmamycoides cluster in ruminants. Vet. Res. 35:635– 649.

16. Tardy F, Gaurivaud P, Tricot A, Maigre L, Poumarat F. 2009. Epide-miological surveillance of mycoplasmas belonging to the ‘Mycoplasmamycoides’ cluster: is DGGE fingerprinting of 16S rRNA genes suitable?Lett. Appl. Microbiol. 48:210 –217.

17. Johansson KE, Heldtander MU, Pettersson B. 1998. Characterization ofmycoplasmas by PCR and sequence analysis with universal 16S rDNAprimers, Methods Mol. Biol. 104:145–165.

18. McAuliffe L, Ellis RJ, Lawes JR, Ayling RD, Nicholas RA. 2005. 16SrDNA PCR and denaturing gradient gel electrophoresis; a single generictest for detecting and differentiating Mycoplasma species. J. Med. Micro-biol. 54:731–739.

19. Goto K, Yamamoto M, Asahara M, Tamura T, Matsumura M,Hayashimoto N, Makimura K. 2012. Rapid identification of Mycoplasmapulmonis isolated from laboratory mice and rats using matrix-assistedlaser desorption ionization time-of-flight mass spectrometry. J. Vet. Med.Sci. 74:1083–1086.

20. Dégrange S, Cazanave C, Charron A, Renaudin H, Bébéar C, BébéarCM. 2009. Development of multiple-locus variable-number tandem-repeat analysis for molecular typing of Mycoplasma pneumoniae. J. Clin.Microbiol. 47:914 –923.

21. Dorigo-Zetsma JW, Wilbrink B, Dankert J, Zaat SA. 2001. Mycoplasmapneumoniae P1 type 1- and type 2-specific sequences within the P1 cytad-hesin gene of individual strains. Infect. Immun. 69:5612–5618.

22. Touati A, Bénard A, Ben Hassen A, Bébéar CM, Pereyre S. 2009.Evaluation of five commercial real-time PCR assays for the detection ofMycoplasma pneumoniae in respiratory tract specimens. J. Clin. Micro-biol. 47:2269 –2271.

23. Chazel M, Tardy F, Le Grand D, Calavas D, Poumarat F. 2010. Myco-plasmoses of ruminants in France: recent data from the national surveil-lance network. BMC Vet. Res. 6:32. doi:10.1186/1746-6148-6-32.

24. Cochran WG. 1950. Estimation of bacterial densities by means of the“most probable number.” Biometrics 6:105–116.

25. Weisburg WG, Tully JG, Rose DL, Petzel JP, Oyaizu H, Yang D,Mandelco L, Sechrest J, Lawrence TG, Van Etten J. 1989. A phylogeneticanalysis of the mycoplasmas: basis for their classification. J. Bacteriol. 171:6455– 6467.

26. Manso-Silván L, Dupuy V, Lysnyansky I, Ozdemir U, Thiaucourt F.2012. Phylogeny and molecular typing of Mycoplasma agalactiae andMycoplasma bovis by multilocus sequencing. Vet. Microbiol. 161:104 –112.

27. Fischer A, Santana-Cruz I, Giglio M, Nadendla S, Drabek E, Vilei EM,Frey J, Jores J. 2013. Genome sequence of Mycoplasma feriruminatoris sp.nov., a fast-growing Mycoplasma species. Genome Announc. 1:e00216-12.doi:10.1128/genomeA.00216-12.

28. Cousin-Allery A, Charron A, de Barbeyrac B, Fremy G, Skov Jensen J,Renaudin H, Bébéar C. 2000. Molecular typing of Mycoplasma pneu-moniae strains by PCR-based methods and pulsed-field gel electrophore-sis. Application to French and Danish isolates. Epidemiol. Infect. 124:103–111.

29. Ladefoged SA, Christiansen G. 1992. Physical and genetic mapping of thegenomes of five Mycoplasma hominis strains by pulsed-field gel electro-phoresis. J. Bacteriol. 174:2199 –2207.

30. Nouvel LX, Sirand-Pugnet P, Marenda MS, Sagné E, Barbe V,Mangenot S, Schenowitz C, Jacob D, Barré A, Claverol S, Blanchard A,Citti C. 2010. Comparative genomic and proteomic analyses of two My-coplasma agalactiae strains: clues to the macro- and micro-events that areshaping mycoplasma diversity. BMC Genomics 11:86. doi:10.1186/1471-2164-11-86.

31. Tardy F, Baranowski E, Nouvel LX, Mick V, Manso-Silvàn L, Thia-ucourt F, Thébault P, Breton M, Sirand-Pugnet P, Blanchard A, Gar-nier A, Gibert P, Game Y, Poumarat F, Citti C. 2012. Emergence ofatypical Mycoplasma agalactiae strains harboring a new prophage and as-

Pereyre et al.

3322 jcm.asm.org Journal of Clinical Microbiology

on July 6, 2020 by guesthttp://jcm

.asm.org/

Dow

nloaded from

sociated with an alpine wild ungulate mortality episode. Appl. Environ.Microbiol. 78:4659 – 4668.

32. Hermeyer K, Buchenau I, Thomasmeyer A, Baum B, Spergser J, Rosen-garten R, Hewicker-Trautwein M. 2012. Chronic pneumonia in calvesafter experimental infection with Mycoplasma bovis strain 1067: charac-terization of lung pathology, persistence of variable surface protein anti-gens and local immune response. Acta Vet. Scand. 54:9. doi:10.1186/1751-0147-54-9.

33. Stevenson LG, Drake SK, Murray PR. 2010. Rapid identification ofbacteria in positive blood culture broths by matrix-assisted laser desorp-tion ionization–time of flight mass spectrometry. J. Clin. Microbiol. 48:444 – 447.

34. Bizzini A, Jaton K, Romo D, Bille J, Prod’hom G, Greub G. 2011.Matrix-assisted laser desorption ionization–time of flight mass spectrom-etry as an alternative to 16S rRNA gene sequencing for identification ofdifficult-to-identify bacterial strains. J. Clin. Microbiol. 49:693– 696.

35. Alatoom AA, Cazanave CJ, Cunningham SA, Ihde SM, Patel R. 2012.Identification of non-diphtheriae Corynebacterium by use of matrix-assisted laser desorption ionization–time of flight mass spectrometry. J.Clin. Microbiol. 50:160 –163.

36. Alatoom AA, Cunningham SA, Ihde SM, Mandrekar J, Patel R. 2011.Comparison of direct colony method versus extraction method for iden-tification of gram-positive cocci by use of Bruker Biotyper matrix-assistedlaser desorption ionization–time of flight mass spectrometry. J. Clin. Mi-crobiol. 49:2868 –2873.

37. Robertson JA, Stemke GW, Davis JW, Jr, Harasawa R, Thirkell D, KongF, Shepard MC, Ford DK. 2002. Proposal of Ureaplasma parvum sp. nov.

and emended description of Ureaplasma urealyticum (Shepard et al. 1974)Robertson et al. 2001. Int. J. Syst. Evol. Microbiol. 52(Pt 2):587–597.

38. Xiao L, Paralanov V, Glass JI, Duffy LB, Robertson JA, Cassell GH,Chen Y, Waites KB. 2011. Extensive horizontal gene transfer in ureaplas-mas from humans questions the utility of serotyping for diagnostic pur-poses. J. Clin. Microbiol. 49:2818 –2826.

39. Deguchi T, Yoshida T, Miyazawa T, Yasuda M, Tamaki M, Ishiko H,Maeda SI. 2004. Association of Ureaplasma urealyticum (biovar 2) withnongonococcal urethritis. Sex. Transm. Dis. 31:192–195.

40. Povlsen K, Bjørnelius E, Lidbrink P, Lind I. 2002. Relationship ofUreaplasma urealyticum biovar 2 to nongonococcal urethritis. Eur. J. Clin.Microbiol. Infect. Dis. 21:97–101.

41. Askaa G, Erno H. 1976. Elevation of Mycoplasma agalactiae subsp. bovisto species rank: Mycoplasma bovis (Hale et al.) comb. nov. Int. J. Syst.Bacteriol. 323–325.

42. Pettersson B, Uhlén M, Johansson KE. 1996. Phylogeny of some myco-plasmas from ruminants based on 16S rRNA sequences and definition of anew cluster within the hominis group. Int. J. Syst. Bacteriol. 46:1093–1098.

43. Schwartz SB, Thurman KA, Mitchell SL, Wolff BJ, Winchell JM. 2009.Genotyping of Mycoplasma pneumoniae isolates using real-time PCR andhigh-resolution melt analysis. Clin. Microbiol. Infect. 15:756 –762.

44. Spuesens EB, Hoogenboezem T, Sluijter M, Hartwig NG, van RossumAM, Vink C. 2010. Macrolide resistance determination and moleculartyping of Mycoplasma pneumoniae by pyrosequencing. J. Microbiol.Methods 82:214 –222.

Identification of Mycoplasmas by MALDI-TOF MS

October 2013 Volume 51 Number 10 jcm.asm.org 3323

on July 6, 2020 by guesthttp://jcm

.asm.org/

Dow

nloaded from