Gp

JMa

b

c

a

ARRA

KCBMgM

I

rtap((iarvcft

0d

Systematic and Applied Microbiology 34 (2011) 400– 407

Contents lists available at ScienceDirect

Systematic and Applied Microbiology

j ourna l ho mepage: www.elsev ier .de /syapm

yrB sequence analysis and MALDI-TOF MS as identification tools for plantathogenic Clavibacter

oanna Zalugaa,∗, Kim Heylena, Koenraad Van Hoordea,c, Bart Hostea, Johan Van Vaerenberghb, Martineaesb, Paul De Vosa

Laboratory of Microbiology, Department of Biochemistry and Microbiology, Ghent University, K.L. Ledeganckstraat 35, B-9000 Ghent, BelgiumPlant-Crop Protection, Institute for Agricultural and Fisheries Research – ILVO, Burg. Van Gansberghelaan 96, B-9820 Merelbeke, BelgiumFaculty of Applied Engineering Sciences, University College Ghent, Schoonmeersstraat 52, B-9000 Ghent, Belgium

r t i c l e i n f o

rticle history:eceived 6 January 2011eceived in revised form 2 May 2011ccepted 6 May 2011

eywords:lavibacteracterial identificationass spectrometry

yrBALDI-TOF MS

a b s t r a c t

The bacterial genus Clavibacter has only one species, Clavibacter michiganensis, containing five subspecies.All five are plant pathogens, among which three are recognized as quarantine pests (mentioned on theEPPO A2 list). Prevention of their introduction and epidemic outbreaks requires a reliable and accurateidentification. Currently, identification of these bacteria is time consuming and often problematic, mainlybecause of cross-reactions with other plant-associated bacteria in immunological tests and false-negativeresults in PCR detection methods. Furthermore, distinguishing closely related subspecies is not straight-forward. This study aimed at evaluating the use of matrix-assisted laser desorption ionization-time offlight mass spectrometry (MALDI-TOF MS) and a fragment of the gyrB sequence for the reliable and fastidentification of the Clavibacter subspecies. Amplification and sequencing of gyrB using a single primer sethad sufficient resolution and specificity to identify each subspecies based on both sequence similarities incluster analyses and specific signatures within the sequences. All five subspecies also generated distinct

and reproducible MALDI-TOF MS profiles, with unique and specific ion peaks for each subspecies, whichcould be used as biomarkers for identification. Results from both methods were in agreement and wereable to distinguish the five Clavibacter subspecies from each other and from representatives of closelyrelated Rathayibacter, Leifsonia or Curtobacterium species. Our study suggests that proteomic analysisusing MALDI-TOF MS and gyrB sequence are powerful diagnostic tools for the accurate identification ofClavibacter plant pathogens.

ntroduction

Clavibacter michiganensis harbours aerobic, non-sporulatingods that belong to the GC rich subgroup of Gram-positive bac-eria. It constitutes the only species within the genus Clavibacternd currently consists of five subspecies, all of which are plantathogens: C. m. subsp. insidiosus (Cmi), C. m. subsp. michiganensisCmm), C. m. subsp. nebraskensis (Cmn), C. m. subsp. sepedonicusCms) and C. m. subsp. tesselarius (Cmt) [12]. They are harmful onmportant agricultural crops such as alfalfa, tomato, corn, potatond wheat. The host range of each subspecies is limited, mostlyestricted to one plant species, and most of them are pathogens ofascular plants, causing wilt as main symptom. Based on cell wall

omposition, menaquinones and additional phenotypic markers,ormer Clavibacter species were reclassified to the genus Curtobac-erium, Leifsonia and Rathayibacter [4,13,45]. Cmm, Cms and Cmi

∗ Corresponding author. Tel.: +32 92645101; fax: +32 92645092.E-mail address: Joanna.Zaluga@ugent.be (J. Zaluga).

723-2020/$ – see front matter © 2011 Elsevier GmbH. All rights reserved.oi:10.1016/j.syapm.2011.05.001

© 2011 Elsevier GmbH. All rights reserved.

have quarantine status in plant health legislation in Europe [2].Every year they cause serious economic damage due to actual yieldreduction [7] but also due to statutory measures taken to eliminatethe pathogen from seed stocks [44]. Transport of certain plants andplant products within the EU is regulated, as well as introduction ofcertain commodities from third countries into the EU. Plant healthand seed certifications are mandatory to prevent the introduction,spread and establishment of specific plant pests and, if detected, toimplement control strategies with the aim of eradication [2].

Diagnostic procedures for phytopathogenic C. michiganensissubspecies are available from either the European Union or Euro-pean Plant Protection Organization (EPPO) [9–11]. Fundamental inthe diagnostic process is the rapid, reliable and accurate identifi-cation of the phytopathogen to confirm its presence and identity,which for C. michiganensis subspecies can be done by pheno-typic properties, serological tests, fatty acid or protein profiles and

molecular assays. Final confirmation is obtained in a pathogenic-ity test. Usually, a serological test or fatty acid profile is combinedwith a molecular test such as PCR, real-time PCR or a fingerprintPCR, e.g., BOX-PCR. PCR identification of the Clavibacter michiga-

plied

noTcmiaog[sriaaafi

Mop[tiirtplMow[

tMlosAwst

M

B

tNtCCwfsftRRabw

J. Zaluga et al. / Systematic and Ap

ensis subspecies targets the 16S–23S rRNA spacer region [29],ther chromosomal regions [27] or plasmid encoded genes [21,28].his diversity of tests and targets, each with its intrinsic specificity,an trigger inconsistent or conflicting results in identification of C.ichiganensis subspecies. PCR using a uniform target region present

n all C. michiganensis subspecies will allow complete differenti-tion within the whole species, greatly simplifying identificationf these important plant pathogens. Sequencing of the 16S rRNAene is the most common identification tool in bacterial taxonomy17], but its resolution is not sufficient for accurate delineation atubspecies level. Other targets, such as housekeeping genes atpA,poB, gyrB and others, have already proven to be useful for reliabledentification in genera such as Ensifer [25], Microbacterium [36]nd others. Sequence analysis of the gyrase B gene (gyrB), deservesttention as potential identification tool for the genus Clavibacters it was already proposed as a possible useful phylogenetic markeror members of the family Microbacteriacea, of which Clavibacters a member [35].

Recently, a number of studies demonstrated the usefulness ofALDI-TOF MS (matrix assisted laser desorption/ionization time

f flight mass spectrometry) for rapid identification of clinicalathogens [3,14], lactic acid bacteria [41], nonfermenting bacteria26], environmental bacteria [37], etc. In this type of mass spec-rometry, samples are prepared by embedding analyte moleculesn a crystal matrix of small acidic molecules. A brief laser pulserradiates the sample and the matrix absorbs the laser energyesulting in ablation of a small volume of matrix and desorption ofhe embedded analyte molecules which are ionized. Subsequently,redominantly single charged analyte ions can be detected and ana-

yzed [23]. Characterization and/or identification using MALDI-TOFS is based on differences in mass to charge ratio (m/z) fingerprints

f whole cell proteins, mainly representing ribosomal proteinshich are most abundantly expressed under all growth conditions

38].The objective of this study was to develop a robust identifica-

ion method applicable to all C. michiganensis subspecies. GyrB andALDI-TOF MS profiling were evaluated as identification tools on a

arge set of strains of all C. michiganensis subspecies retrieved fromutbreaks worldwide, complemented with a collection of referencetrains and representatives of phylogenetically closely related taxa.pplicability of both techniques for correct subspecies assignmentas evaluated in two ways: by cluster analysis and use of specific

ignature positions in case of sequence data and by cluster analysisogether with biomarkers for spectral data.

aterials and methods

acterial strains and growth conditions

A total of 173 strains were received from the BCCM/LMG Bac-eria Collection (Ghent, Belgium), PD collection (Wageningen, Theetherlands) GBBC collection (ILVO, Merelbeke, Belgium) and Insti-

uto Canario de Investigaciones Agrarias (Tenerife, Spain). Thislavibacter strain subset consisted of 67 Cmm, 65 Cms, 13 Cmi, 9mn, 4 Cmt and 7 unidentified Clavibacter sp. of which four strainsere described as Cmm look-alikes. Cmm look-alikes were isolated

rom infected tomato seeds, positively identified as belonging to theubspecies Cmm using the current diagnostic protocols [9], thoughailed the pathogenicity test in a bioassay (personal communica-ion PD collection). The following outgroup strains were included,athayibacter rathayi LMG 3717, Rathayibacter tritici LMG 3726,

athayibacter iranicus LMG 3677T, Leifsonia aquatica LMG 18699,nd four Curtobacterium strains. All strains are listed in Table S1. Theacteria were grown aerobically at 25 ◦C for 24–48 h. Stock culturesere stored at −20 ◦C in MicrobankTM beads (Pro-Lab Diagnostics,

Microbiology 34 (2011) 400– 407 401

Canada) until analysis. For MALDI-TOF MS analysis, bacteria weregrown on MTNA medium without antibiotics added [18]. To assessthe influence of different growth media on species recognitionusing MALDI-TOF MS, two additional media recommended by theBCCM/LMG bacteria collection (http://bccm.belspo.be/index.php),M6 (glucose 10 g, yeast extract 5 g, peptone 5 g, agar 15 g, distilledwater 1 L, pH 7.0) and M39 (medium M6 supplemented with 0.1 gcasein hydrolysate per liter) were used. These two media were alsoused to obtain cultures for DNA extraction.

DNA extraction and sequencing

Total genomic DNA was extracted according to the guanidium-thiocyanate–EDTA–sarkosyl method described by Pitcher [31]which was adapted for Gram-positive bacteria with an additionallysozyme step as described before 16S rRNA gene amplificationwas performed as described previously with primers pA-forwardand pH-reverse [16]. Subsequently, PCR products were purifiedand partially sequenced as described below using the primersBKL1-reverse (5′-ACTGCTGCCTCCCGTAGGAG-3′, position 536–516in the Escherichia coli numbering system) and gamma-reverse (5′-ACTGCTGCCTCCCGTAGGAG-3′, position 358–339) [16].

A fragment of the gyrB was amplified by PCR with the previouslydescribed primers gyrB-2F and gyrB-4R [35]. The PCR mixture had atotal volume of 25 �l, containing 1× PCR buffer (100 mM Tris–HCl,15 mM MgCl2, 500 mM KCl [pH 8.3]), dNTP’s 0.2 mM each, 0.6 �M ofeach primer, 0.5 U AmpliTaq DNA polymerase, and 50–60 ng tem-plate DNA. The PCR conditions were as described previously [35].The expected amplicon size was about 550 bp. Resulting ampliconswere purified using the Nucleofast®96 PCR clean up membranesystem (Macherey-Nagel, Germany). Sequencing reactions wereperformed in a total volume of 10 �l with 3 �l of purified ampli-con, 0.286 �l of BigDyeTM mixture (Terminator Cycle SequencingKit version 3.1, Applied Biosystems), 1× sequencing buffer and1.2 �M of each of the amplification primers (gyrB 2F and gyrB 4R).The thermal program consisted of 30 cycles (96 ◦C for 15 s, 35 ◦C for1 s, 60 ◦C for 4 min). Subsequently, the sequencing products werepurified using the BigDye XTerminator Kit (Applied Biosystems)and analyzed on a 3130xl Genetic Analyzer (Applied Biosystems).

Sequence analysis

The 16S rRNA gene and gyrB sequences were assembled withBioNumerics version 5.1 (Applied Maths, Belgium) and alignedusing ClustalW [42]. The identity assigned to each strain by thethree culture collections was verified with an nBLAST of the 16SrRNA gene sequence of each strain and a neighbor-joining clusteranalysis with all type strains of all species of all genera mentioned inthe twenty highest BLAST hits. The strains were assigned to a genusbased on the obtained pairwise 16S rRNA gene sequence similarity(>98% gene sequence similarity).

GyrB sequences were checked by amino acid translation withTransseq (http://www.ebi.ac.uk/Tools/emboss/transeq/) and pres-ence of the gyrB protein domain was confirmed with pBlast [1].Molecular Evolutionary Genetics Analysis software (Mega 4.1) [40]was used to calculate evolutionary distances based on a 500 bp (thisequal length was used for all strains) gyrB fragment of the gyrBamplicon, and to infer Maximum Parsimony (MP), Maximum Like-lihood (ML) and Neighbor-Joining (NJ) trees using the maximumcomposite likelihood (MCL) model.

TaxonGap software, version 2.4.1 [39], was used to define the

discriminatory power of both sequence markers, 16S rRNA andgyrB, and to identify sequence characters using strains of Clavibac-ter subspecies with established name and confirmed pathogenicity.To establish the positions within the gyrB and 16S rRNA gene, the

4 plied

we

M

tm(9aq[cnawCa

4elaiEpTwoiIgmwo(md

tsmdsocw(m

sa(aftiatmtthte

02 J. Zaluga et al. / Systematic and Ap

hole Cmm genome sequence (AM711867) was used as a refer-nce.

ALDI-TOF MS

Preparation of cell extracts. The formic acid–acetonitrile extrac-ion as developed by Bruker Daltonics [24] was optimized to retain

ostly ribosomal proteins. For this, a small amount of bacterial cellsyellow loop) was suspended in 300 �l of distilled water after which00 �l of absolute ethanol was added. This mixture was centrifugedt 18,000 × g for 3 min and the supernatant was discarded. Subse-uently, the pellet was resuspended in 50 �l of formic acid (70%v/v]) after which 50 �l of acetonitrile (ACN) was added. Followingentrifugation at 18,000 × g for 3 min, one microliter of the super-atant was spotted on a steel target plate (AB Sciex, Belgium) andir dried at room temperature. Finally, the sample spot was overlaidith 1 �l of a 0.5% (w/v) �-cyano-4-hydroxy cinnamic acid (�-HCA) solution in ACN-MilliQ water-trifluoroacetic (50:48:2 v/v/v)nd air dried.

MALDI-TOF MS analysis. Measurements were performed on a800 Plus MALDI TOF/TOFTM Analyzer (AB Sciex, Belgium) in lin-ar, positive-ion mode with a 200-Hz frequency tripled UV Nd:YAGaser operating at a wavelength of 355 nm. Generated ions wereccelerated at 20 kV through a grid at 19.2 kV and separated accord-ng to their m/z ratio in a 1.5 m long linear, field-free drift region.ach generated spectrum resulted from 40 laser shots at 50 randomositions within the measuring spot (2000 spectra/spot). MALDI-OF mass spectra were generated in the range 2–20 kDa. Calibrationas performed using an external protein calibration mix composed

f the Protein Calibration Standard I developed by Bruker Dalton-cs (Bruker) [composition: insulin ([M+H]+, m/z 5734.6), ubiquitin

([M+H]+, m/z 8565.9), cytochrome C ([M+H]+, m/z 12361.5), myo-lobin ([M+H]+, m/z 16952.3)] and an ACTH clip 18–39 ([M+H]+,/z 2466.7) (Sigma–Aldrich), resulting in five peaks spread over thehole fingerprint spectrum. The error allowed to match the peak

f the proteins was set at 200 ppm. Bruker Bacterial Test StandardBruker) of E. coli DH5� was included with every set of measure-

ents as a positive control. Each sample was spotted at least inuplicate to check the reproducibility.

Analysis of the spectral data. Mass spectra were generated in2d format and converted to txt files using ABI Data Explorer 4.0oftware (AB Sciex), after which they were imported into BioNu-erics 5.1 software (Applied Maths, Belgium). To obtain reliable

ata analysis, the spectra with extensive noise and/or insufficientignal intensity were excluded from further analysis. High levelsf noise signals, possibly due to interference of remaining polysac-harides or salts in the cell extracts at the beginning of the profileere left out of the analysis as excluding the first 5% of the profile

shifting the m/z range from 2 kDa to 3 kDa) was enough to obtainore reliable profiles.The similarity between the spectra was expressed using Pear-

on’s product moment correlation coefficient, a curve basednalysis, and the spectra were clustered using the UPGMAunweighted pair group method with arithmetic mean) clusteringlgorithm. Next to cluster analysis, spectral data were investigatedor the presence of biomarkers characteristic for each Clavibac-er subspecies. After visual inspection and comparison, the mostntensive and predominantly present protein peaks were selectednd screened in representatives of each subspecies. Becausehe in-house cell extraction protocol was designed to retain

ainly ribosomal proteins, selected biomarkers could be assignedo specific ribosomal proteins by comparing protein masses

o the Rapid Microorganism Identification Database (RMIDb)ttp://www.rmidb.org/cgi-bin/index.pl. This database was createdo identify bacteria and viruses using existing mass spectrom-try protocols. RMIDb contains all bacterial, virus, plasmid and

Microbiology 34 (2011) 400– 407

environmental protein sequences from the Ventor Institute’s Com-prehensive Microbial Resource (CMR), UniProt’s Swiss-Prot andTrEMBL, Genbank’s Protein, and RefSeq’s Protein and Genome [30].

Nucleotide accession numbers

The Clavibacter sequences have been deposited in the EMBLDatabase with accession numbers FR728257–FR728381 andFR727975–FR728147 for the 16S rRNA gene and gyrB sequencesrespectively.

Results

Strain set

A total of 165 Clavibacter strains and eight outgroup strainswere included in this study, covering a wide geographical andtemporal spread (Table S1). Information received from the cul-ture collections on subspecies allocation of strains was initiallyregarded as correct. However, partial 16S rRNA gene sequence anal-ysis demonstrated that not all strains belonged to Clavibacter (datanot shown). Neighbor joining analysis of 16S rRNA gene sequencesdistinguished two separate groups of Clavibacter, one containingthe majority of clavibacters and the second containing four strains,three of which were originally assigned to the Cmm and one tothe subspecies sepedonicus. However, BLAST analysis of membersof this second group showed 100% 16S rRNA gene sequence sim-ilarity to members of the genus Curtobacterium (Curtobacteriumflaccumfaciens, accession number GU586309.1, Curtobacterium sp.,accession number GU120656.1 and Curtobacterium flaccumfacienspv. flaccumfaciens, accession number AM410688.1). These resultsindicated that these strains had been incorrectly identified asClavibacter, which was confirmed by pairwise sequence similar-ity analysis with type strains of genus Curtobacterium. The separateposition of these four Curtobacterium strains was later also observedin gyrB sequence and MALDI-TOF MS analyses. Because of therelevance to diagnostic purposes, Cmm look-alikes from the PDcollection were also included in the strain set. These bacteria wereisolated from infected tomato seeds and had been positively iden-tified as Cmm although they could not be confirmed as pathogenicwith the bioassay test. Both MALDI-TOF MS and gyrB sequence anal-ysis grouped these Cmm look-alikes in separate clusters. Moreover,both gyrB sequence analysis and MALDI-TOF MS were able to assignthree unidentified Clavibacter spp., with strain numbers PD 5715,PD 5716 and PD 5714, to subsp. nebraskensis, subsp. michiganen-sis and subsp. tesselarius, respectively. In addition, the outgroupstrains Rathayibacter; Curtobacterium and Leifsonia were includedto evaluate the sensitivity and robustness of MALDI-TOF MS andgyrB sequence analysis.

GyrB sequence analysis

GyrB amplicons were retrieved from all strains and sequenceswere translated into amino acids as a quality check. Sequencedata were subsequently analyzed with two different approaches,namely by phylogenetic analysis and by identification of signaturenucleotides specific for each subspecies. Cluster analysis clearlygrouped all gyrB sequences from Clavibacter spp. separately fromoutgroups and Curtobacterium spp. Also, sequences from each sub-species formed distinct clusters, supported with high bootstrapvalues (Fig. 1). In Cmn and Cmi the sequences were identical,whereas in Cms sequences similarity ranged from 97.1 to 100%,

in Cmm 98.2% to 100% and in Cmt 96.2 to 99.3%. Interestingly,the gyrB sequences from the Cmm look-alikes grouped outsidethe Cmm cluster, although these strains were positively identi-fied as Cmm using the current diagnostic procedure. As indicated

J. Zaluga et al. / Systematic and Applied Microbiology 34 (2011) 400– 407 403

subsp . michiganensisC. michiganensis

Cmm look-alikes. tesselarius C. michiganensis subsp

subsp . sepedonicusC. michiganensis

nebraskensis C. michiganensis subsp.

insidiosus C. michiganensis subsp.

r sp. RathayibacteLMG18699 Leifsonia aquatica

sp. Curtobacterium

100

9970

100

99

91

95

82

100

98

100

0.02Number of base substitutions per site

Fig. 1. Phylogenetic analysis of gyrB sequences. Rooted neighbor-joining tree was based on partial gyrB sequence (position 646–1143 of Cmm NCPPB 382 accession numberA replici equen

bvaga1ttps1tAwadso

gndBefdwt

M

p

the technique itself and through the use of different growth condi-tions together with its high reproducibility provide evidence for therobustness of the MALDI-TOF MS for analysis of Clavibacter strains.

Table 1Overview of signatures in gyrB gene sequences, which are unique combinationsof characters for individual subspecies of Clavibacter. Specific nucleotide presentat each discriminative position (position in reference gyrB sequence of Cmm NCPPB382 accession nr AM711867) is given. A, adenine; G, guanine; C, cytosine; T, thymine.

Name (# strains) Positions of specificnucleotide

Number of uniquesignatures

Cmm (67) 702G, 1119G 2Cms (65) 687T, 728A, 786G, 1035T,

1056G, 1059T, 1080C,1089T, 1113G

9

Cmn (9) 1041G, 1053C, 1077G 3

M711867) of 173 Clavibacter strains. Bootstrap values were generated from 1000

n accordance with maximum parsimony and maximum likelihood analysis. GyrB s

y branch lengths in the phylogenetic tree (Fig. 1), gyrB sequenceariation between strains of the same subspecies was very limitednd even absent within Cms, Cmi and Cmn. This limited diver-ence in gyrB sequence was translated in only a few changes inmino acid composition. Therefore, the suitability of the gyrB and6S rRNA sequences as phylogenetic markers was investigated fur-her using the TaxonGap 2.4.1 software (Fig. S2) for visualization ofhe intra- and intersubspecies sequence heterogeneity and com-arison of the discriminative power of sequence markers for aet of taxonomic units. Analysis was performed on both gyrB and6S rRNA gene sequences of strains that were assigned to a cer-ain subspecies (only strains with an established name were used.ll Cmm look-alike strains and strains received as Clavibacter spp.ere excluded from this analysis). For all subspecies, gyrB clearly

llowed distinct higher sequence divergence between sequences ofifferent subspecies than within the subspecies. For Cmt, the gyrBequence heterogeneity within the subspecies was higher than inther subspecies.

In addition to the standard cluster analysis approach, alignedyrB sequences were manually searched for the presence of sig-ature positions specific for each subspecies. In total, 39 positionsiffered among subspecies throughout the gyrB fragment of 500 bp.ased on these positions; unique signatures could be selected forach subspecies (Fig. S3). Numbers of unique nucleotides rangedrom two for Cmm to nine for Cms (Table 1). Almost all sequenceivergences at these positions were the result of silent mutationshereas only four changes in amino acid sequence were found

hroughout all strains of C. michiganensis.

ALDI-TOF MS

All strains were subjected to MALDI-TOF MS analysis. Fifteenercent of the strains were analyzed in duplicate to assess the

ates. Rathayibacter sp. and Leifsonia aquatica were included as outgroups. NJ tree isces had minimum length of 500 bp.

reproducibility (mean ± standard deviation; 91.98% ± 3.91) of thewhole MALDI-TOF MS procedure, starting from growth and cellextraction over spectrum generation to data analysis. In addition,extracts were always spotted twice on the target plate, to evaluatevariation (mean ± SD; 99.21 ± 0.95) introduced by the MALDI-TOFMS method. The standard protocol included MTNA growth mediumfor all strains of Clavibacter. However, in diagnostics, differentgrowth media are applied for isolation of different Clavibactersubspecies. Therefore, the influence of alternative growth media,namely M6 and M39, recommended by BCCM/LMG collection, onMALDI-TOF MS spectra was investigated for 54 strains (31% of allstrains). The choice of growth medium had no influence on the cor-rect subspecies grouping (Fig. S1). The low variations introduced by

Cmi (13) 725G, 726C, 789T, 969T,981G, 1092G

6

Cmt (4) 798T, 855G, 957C 3Cmm look-alikes (4) 654G, 743C, 825C 3

404 J. Zaluga et al. / Systematic and Applied Microbiology 34 (2011) 400– 407

Table 2Characteristic masses (in daltons) selected as possible biomarkers for identification of Clavibacter. Masses observed in 4 or all subspecies are marked in bold. Assigned proteinscalculated using RMIDb.

Cmm Cms Cmt Cmi Cmn Assigned ribosomal protein (mass in Da,protein subunit/name, origin organism)

3224a

3238 3239 3237 32373353 3353 3353 3350 33503525 3529 3529 35244343 4341 4343 4346-50S/L36 Cmm

4358a 4360-50S/L36 Cms4613a 4613-50S/L36 Cmm/Cms4772 4772 4773 4770 47715078 5078 5078 5076 5076 5078-50S/L34 Cmm/Cms

5724 5724 57245738a

5752 5754 5755 5751 57525823a

6044a

6068a

6145a

6454a 6456-50S/L33 Cms6482 6482 6479 6480 6483-50S/L33 Cmm6713 6713 6713 6710 6712 6714-50S/L30 Cmm/Cms7013a 7012-50S/L35 Cmm7058 7059 7060 7057

7070a 7069-50S/L35 Cms7500 7501 7501-50S/L32 Cms

7998a

8489a

8510a

8526 8525 8523 8525-50S/L28 Cmm/Cms8560 8560 8563 8559 8564-50S/L27 Cmm/Cms9004 9004 9005 9002 9002-30S/S20 Cmm9302 9300 9300 9300 9302-30S/S18 Cmm

9360a 9360-30S/S18 Cms

EasostNscosschmatfiewo

mbmtCb(ub

9549 9549 9550 9544

a Subspecies unique mass values marked in Fig. 2.

The MALDI-TOF MS data analyzed in BioNumerics 5.1 and Data-xplorer 4.0 software were in agreement with the gyrB sequencenalysis. Indeed, cluster analysis of the whole spectral profilehowed distinct clusters for each subspecies, Cmm look-alikes andutgroups (data not shown). However, Cmm strains formed twoeparate groups, one with only a few strains that closely clusteredo Cmn strains and the other with the majority of the Cmm strains.ext to cluster analyses, the obtained spectral profiles were also

creened for the presence of recurring peaks or biomarker ions spe-ific for the whole species or a subspecies. A selection of 20 profilesf representatives of all Clavibacter subspecies, including the typetrains, was visually screened for the most informative and inten-ive peaks. A typical MALDI-TOF MS spectrum of C. michiganensisontains about 50 ion peaks between 2000 and 20,000 Da, with theighest intensity peaks between 4000 and 10,000 Da. Table 2 sum-arizes 32 selected m/z values. Six m/z values were detected in

ll subspecies, making them characteristic for the genus Clavibac-er. Unique peaks for each subspecies ranged from one in Cmt tove in Cms (Fig. 2). In order to validate these peaks as biomark-rs, analysis of another 25 randomly selected Clavibacter profilesas performed and the assignment to the expected subspecies was

btained.The applied cell extraction protocol favors isolation of riboso-

al subunit proteins. Therefore, we tried to assign the selectediomarkers by comparing the experimentally obtained proteinasses with those of the ribosomal subunit proteins deduced from

wo fully sequenced Clavibacter genomes (Cmm NCPPB382 andms ATCC 33113T) using the RMIDb database. In this way, eleven

iomarkers could be identified as specific ribosomal proteinsTable 2). However, our assignment procedure left several massesnidentified, probably because of slight differences in protein massetween different subspecies (only genomes of representatives of

9547

two subspecies were available), inclusion of nonribosomal proteinsin the analyzed cell extracts, and possible occurrence of 2+ ions.

Discussion

This study demonstrated the suitability of gyrB sequencesand MALDI-TOF MS protein profiles for identification of plantpathogenic bacteria of the genus Clavibacter. The identificationcan be based on the generated data as a whole, or on signa-tures within these data that have enough discriminatory powerto distinguish between the different subspecies of C. michiganen-sis. Moreover, both the sequence and the protein method suggestedthat diversity within this Clavibacter species is higher than currentlyaccepted. This study reports for the first time on the use of gyrB asa phylogenetic marker in the genus Clavibacter. Molecular mark-ers for identification purposes must exhibit the smallest amountof heterogeneity within a taxonomic unit and should maximallydiscriminate between different species/subspecies [25], which wasdemonstrated in this study for gyrB sequences in the genus Clav-ibacter. In addition to cluster analysis, a selection of positions inthe gyrB contained nucleotide signatures characteristic for eachsubspecies and were shown to be decisive for subspecies assign-ment. This character-based identification approach was previouslyproposed as an alternative for tree building methods [33]. How-ever, in bacterial taxonomy, this approach is not as popular as inhuman medicine where SNPs (single nucleotide polymorphisms)are commonly used in identification of pathogens or loci associ-ated with complex diseases [5,22]. These SNPs have also been used

to design highly specific molecular tools for identification [43]. Fur-thermore, the fact that the gyrB fragment can be amplified andsequenced with the same protocol in all Clavibacter and closelyrelated strains tested, creates great potential for future application

J. Zaluga et al. / Systematic and Applied Microbiology 34 (2011) 400– 407 405

Fig. 2. MALDI-TOF MS protein mass fingerprints of type strains of genus Clavibacter. Similar and different marker masses for the identification of Clavibacter subspecies arel nd thm

atmE

ioabtfiotiifitascccowit

isted in Table 2. Relative intensities of ions (percentages) are shown on the y axis aass-to-charge ratios. *Unique peaks positions for each of subspecies.

s a new diagnostic tool. Based on the signature nucleotide posi-ions, new methods for specific PCR and Q-PCR detection in plant

aterial may also be developed, which is especially relevant for theU-quarantine pathogens Cmm, Cms and Cmi.

Because of its simplicity and applicability, MALDI-TOF MSs already widely used for identification and characterizationf diverse microorganisms [15,32,34]. Currently, spectral datare commonly analyzed using two different approaches. Specificiomarker ions are identified based on additional information onheoretical protein masses (from e.g., genome projects). New pro-les of unknown strains are then screened for presence/absencef these unique biomarker ions to subsequently assign to a certainaxonomic group [30]. This approach is more suitable for discrim-nation between closely related species as well as for accuratedentification [6]. Another workflow uses a database with referencengerprinting profiles obtained as a consensus profile from a set ofype strains. New spectra are then compared with database entriesnd the closest spectrum is found [19]. Widely used commercialystems like SARAMIS (Spectral ARchiving and Microbial Identifi-ation System, AnagnosTec) and Maldi Biotyper (Bruker Daltonics)ontain average spectra together with peak lists of the most dis-riminating biomarker ions which are scored according to their

ccurrence and specificity. Because we wanted to use both thehole spectral profile and biomarker ions, we used BioNumer-

cs and DataExplorer 4.0 (AB Sciex) respectively, as an alternativeo commercially available database systems. Both software appli-

e masses (in Daltons) of the ions are shown on the x axis. The m/z values represent

cations allowed us to look at the spectra in detail, perform ourown quality check and analyze the data in different ways. More-over, BioNumerics software allowed the coupling of additional datafrom more than one experiment type (e.g., sequencing information)making the analysis easier, more comprehensive and straightfor-ward. However, the use of non-specialized software also had thedisadvantages that (i) no specific experiment type was available forspectral profiles, and (ii) biomarker ions detection had to be per-formed manually because appropriate tools were lacking. There isstill debate on the discriminatory level of MALDI-TOF MS, which issaid to be accurate at genus, species [34] or even below species level[3]. Nevertheless, a very good discrimination between Clavibactersubspecies was consistently obtained. As MALDI-TOF MS is gainingmore popularity as a quick and robust identification tool, requiredsoftware tools will probably be available in the near future.

To evaluate the robustness and discriminatory power of gyrBsequencing and MALDI-TOF MS analysis, the closely related Clav-ibacter strains were analyzed together with Cmm look-alikes.Discrimination of these closely related non-pathogenic Clavibac-ter by MALDI-TOF MS and gyrB sequence analysis indicates thatthese Cmm look-alikes may be used to study specific mechanismsof plant disease development caused by pathogenic Cmm. Cmm

look-alikes are an especially interesting group of organisms thatpossibly form a new group of Clavibacter since there is almost noevidence of wild-type non-pathogenic bacteria within this genus[46]. Both techniques applied in this study were discriminative

4 plied

ebbtooisldt[c

atWiisrf

A

RSfi0RpN

A

t

R

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

06 J. Zaluga et al. / Systematic and Ap

nough to differentiate strains of the closely related genus Curto-acterium. Misidentification of Curtobacterium as Clavibacter coulde attributed to their high morphological similarities with Clavibac-er or contamination during the handling. Most of the membersf the genus Curtobacterium have been isolated from plants butnly one species, Curtobacterium flaccumfaciens pv. flaccumfaciens,s regarded as a plant pathogen. Because of its importance, thispecies is also classified as a quarantine organism on the A2 EPPOist and in the Council Directive 2000/29/EC [2,8]. Problems in theiscrimination of Curtobacterium from Clavibacter were encoun-ered previously in a numerical classification study of coryneforms20], where members of both groups were allocated in the sameluster.

The results obtained with the genetic- and protein-basedpproach used in our study were in perfect agreement and showedheir value as stand-alone diagnostic tools for the genus Clavibacter.

hereas the sequence-based approach is more time consuming, its surely a valuable tool when MALDI-TOF MS is not available. But,ncreasing applicability of MALDI-TOF MS will make it more acces-ible for the end users. Besides, low costs of sample preparation andapid measurement procedure are of great value for unbiased andast diagnostics.

cknowledgements

We thank the PD, GBBC and BCCM/LMG collections and Anaodríguez Pérez (Instituto Canario de Investigaciones Agrarias,pain) for providing necessary strains. We also thank An Coorevitsor assistance with TaxonGap software. This work was performedn the Seventh Framework Programme of project KBBE-2008-1-4-1 (QBOL) nr 226482 funded by the European Commission. Theesearch Fund University College Ghent is acknowledged for theostdoctoral fellowship of Koenraad Van Hoorde, and the BelgianPPO (FAVV) for partially financing ILVO-research.

ppendix A. Supplementary data

Supplementary data associated with this article can be found, inhe online version, at doi:10.1016/j.syapm.2011.05.001.

eferences

[1] Altschul, S.F., Gish, W., Miller, W., Myers, E.W., Lipman, D.J. (1990) Basic localalignment search tool. J. Mol. Biol. 215, 403–410.

[2] Anonymous (2000) Council Directive 2000/29/EC of 8 May 2000 on protectivemeasures against the introduction into the Community of organisms harmfulto plants or plant products and against their spread within the Community.Office for Official Publications of the European Communities. Consolidated Text.1–139.

[3] Barbuddhe, S.B., Maier, T., Schwarz, G., Kostrzewa, M., Hof, H., Domann, E.,Chakraborty, T., Hain, T. (2008) Rapid identification and typing of Listeriaspecies by matrix-assisted laser desorption ionization-time of flight mass spec-trometry. Appl. Environ. Microbiol. 74, 5402–5407.

[4] Collins, M., Jones, D. (1983) Reclassification of Corynebacterium flaccumfaciens,Corynebacterium betae, Corynebacterium oortii and Corynebacterium poinset-tiae in the genus Curtobacterium, as Curtobacterium flaccumfaciens comb nov.Microbiology 129, 3545.

[5] Degefu, Y. Molecular diagnostic technologies and end user applications: poten-tials and challenges. Published at www.smts.fi.

[6] Dieckmann, R., Strauch, E., Alter, T. (2010) Rapid identification and characteri-zation of Vibrio species using whole-cell MALDI-TOF mass spectrometry. J. Appl.Microbiol. 109, 199–211.

[7] Easton, G.D. (1979) The biology and epidemiology of potato ring rot Am. PotatoJ. 56, 459–460.

[8] EPPO, 2005. Curtobacterium flaccumfaciens pv. flaccumfaciens OEPP/EPPO DataSheets on Quarantine Pests.

[9] EPPO, 2005. PM7/42: diagnostic for Clavibacter michiganensis subsp. michiga-

nensis. OEPP/EPPO Bulletin 35, 273–285.

10] EPPO, 2006. PM7/59: diagnostic for Clavibacter michiganensis subsp. sepe-donicus. OEPP/EPPO Bulletin 36, 99–109.

11] EPPO, 2010. PM7/99: diagnostic for Clavibacter michiganensis subsp. insidiosusOEPP/EPPO Bulletin 40, 353–364.

[

Microbiology 34 (2011) 400– 407

12] Evtushenko, L., Taekeuchi, M. (2006) The family Microbacteriaceae. Prokaryotes3, 1020–1098.

13] Evtushenko, L.I., Dorofeeva, L.V., Subbotin, S.A., Cole, J.R., Tiedje, J.M. (2000)Leifsonia poae gen nov., sp nov., isolated from nematode galls on Poa annua,and reclassification of ‘Corynebacterium aquaticum’ Leifson 1962 as Leifsoniaaquatica (ex Leifson 1962) gen. nov., nom. rev., comb. nov and Clavibacter xyliDavis et al. 1984 with two subspecies as Leifsonia xyli (Davis et al. 1984) gen.nov., comb. nov. Int. J. Syst. Evol. Microbiol. 50, 371–380.

14] Ferreira, L., Sanchez-Juanes, F., Gonzalez-Avila, M., Cembrero-Fucinos, D.,Herrero-Hernandez, A., Gonzalez-Buitrago, J., Munoz-Bellido, J. (2010) Directidentification of urinary tract pathogens from urine samples by matrix-assistedlaser desorption ionization-time of flight mass spectrometry. J. Clin. Microbiol.48, 2110.

15] Giebel, R., Worden, C., Rust, S., Kleinheinz, G., Robbins, M., Sandrin, T. (2010)Microbial fingerprinting using matrix-assisted laser desorption ionizationtime-of-flight mass spectrometry (MALDI-TOF MS): applications and chal-lenges. Adv. Appl. Microbiol. 14, 9–184.

16] Heyrman, J., Swings, J. (2001) 16S rDNA sequence analysis of bacterial iso-lates from biodeteriorated mural paintings in the servilia tomb (Necropolisof Carmona, Seville, Spain). Syst. Appl. Microbiol. 24, 417–422.

17] Janda, J., Abbott, S. (2007) 16S rRNA gene sequencing for bacterial identificationin the diagnostic laboratory: pluses, perils, and pitfalls. J. Clin. Microbiol. 45,2761.

18] Jansing, H., Rudolph, K. (1998) Physiological capabilities of Clavibacter michi-ganensis subsp. sepedonicus and development of a semi-selective medium. Z.Pflanzenk. Pflanzen. 105, 590–601.

19] Jarman, K., Cebula, S., Saenz, A., Petersen, C., Valentine, N., Kingsley, M.,Wahl, K. (2000) An algorithm for automated bacterial identification usingmatrix-assisted laser desorption/ionization mass spectrometry. Anal. Chem.72, 1217–1223.

20] Kämpfer, P., Seiler, H., Dott, W. (1993) Numerical classification of coryneformbacteria and related taxa. J. Gen. Appl. Microbiol. 39, 135–214.

21] Kleitman, F., Barash, I., Burger, A., Iraki, N., Falah, Y., Sessa, G., Weinthal, D.,Chalupowicz, L., Gartemann, K., Eichenlaub, R. (2008) Characterization of aClavibacter michiganensis subsp. michiganensis population in Israel. Eur. J. PlantPathol. 121, 463–475.

22] Lai, E. (2001) Application of SNP technologies in medicine: lessons learned andfuture challenges. Genome Res. 11, 927.

23] Lay, J. (2001) MALDI-TOF mass spectrometry of bacteria. Mass Spectrom. Rev.20, 172–194.

24] Maier, T., Schwarz, G., Kostrzewa, M., 2008. Microorganism Identification andClassification Based on MALDI-TOF MS Fingerprinting with MALDI Biotyper InApplication Note # MT-80.

25] Martens, M., Dawyndt, P., Coopman, R., Gillis, M., De Vos, P., Willems, A. (2008)Advantages of multilocus sequence analysis for taxonomic studies: a case studyusing 10 housekeeping genes in the genus Ensifer (including former Sinorhizo-bium). Int. J. Syst. Evol. Microbiol. 58, 200.

26] Mellmann, A., Bimet, F., Bizet, C., Borovskaya, A., Drake, R., Eigner, U., Fahr, A.,He, Y., Ilina, E., Kostrzewa, M. (2009) High interlaboratory reproducibility ofmatrix-assisted laser desorption ionization-time of flight mass spectrometry-based species identification of nonfermenting bacteria. J. Clin. Microbiol. 47,3732.

27] Mills, D., Russell, B., Hanus, J. (1997) Specific detection of Clavibacter michi-ganensis subsp. sepedonicus by amplification of three unique DNA sequencesisolated by subtraction hybridization. Phytopathology 87, 853–861.

28] Ozdemir, Z. (2005) Development of a Multiplex PCR Assay for ConcurrentDetection of Clavibacter michiganensis ssp. michiganensis and Xanthomonasaxonopodis pv. vesicatoria. Plant Pathol. J. 4, 133–137.

29] Pastrik, K., Rainey, F. (1999) Identification and differentiation of Clavibactermichiganensis subspecies by polymerase chain reaction-based techniques. J.Phytopathol. 147, 687–693.

30] Pineda, F.J., Antoine, M.D., Demirev, P.A., Feldman, A.B., Jackman, J., Longe-necker, M., Lin, J.S. (2003) Microorganism identification by matrix-assistedlaser/desorption ionization mass spectrometry and model-derived ribosomalprotein biomarkers. Anal. Chem. 75, 3817–3822.

31] Pitcher, D., Saunders, N., Owen, R. (1989) Rapid extraction of bacterial genomicDNA with guanidium thiocyanate. Lett. Appl. Microbiol. 8, 151–156.

32] Qian, J., Cutler, J., Cole, R., Cai, Y. (2008) MALDI-TOF mass signatures for differ-entiation of yeast species, strain grouping and monitoring of morphogenesismarkers. Anal. Bioanal. Chem. 392, 439–449.

33] Rach, J., DeSalle, R., Sarkar, I., Schierwater, B., Hadrys, H. (2008) Character-basedDNA barcoding allows discrimination of genera, species and populations inOdonata. Proc. R. Soc. B: Biol. Sci. 275, 237.

34] Rezzonico, F., Vogel, G., Duffy, B., Tonolla, M. (2010) Whole cell MALDI-TOFmass spectrometry application for rapid identification and clustering analysisof Pantoea species. Appl. Environ. Microbiol.

35] Richert, K., Brambilla, E., Stackebrandt, E. (2005) Development of PCRprimers specific for the amplification and direct sequencing of gyrBgenes from microbacteria, order Actinomycetales. J. Microbiol. Methods 60,115–123.

36] Richert, K., Brambilla, E., Stackebrandt, E. (2007) The phylogenetic significance

of peptidoglycan types: molecular analysis of the genera Microbacterium andAureobacterium based upon sequence comparison of gyrB, rpoB, recA and ppkand 16SrRNA genes. Syst. Appl. Microbiol. 30, 102–108.

37] Ruelle, V., El Moualij, B., Zorzi, W., Ledent, P., De Pauw, E. (2004) Rapididentification of environmental bacterial strains by matrix-assisted laser des-

plied

[

[

[

[

[

[

[

[

grasses. Int. J. Syst. Evol. Microbiol. 43, 143.

J. Zaluga et al. / Systematic and Ap

orption/ionization time-of-flight mass spectrometry. Rapid Commun. MassSpectrom. 18, 2013–2019.

38] Ryzhov, V., Fenselau, C. (2001) Characterization of the protein subset desorbedby MALDI from whole bacterial cells. Anal. Chem. 73, 746–750.

39] Slabbinck, B., Dawyndt, P., Martens, M., De Vos, P., De Baets, B. (2008) Taxon-Gap: a visualization tool for intra-and inter-species variation among individualbiomarkers. Bioinformatics 24, 866.

40] Tamura, K., Dudley, J., Nei, M., Kumar, S. (2007) MEGA4: molecular evolutionarygenetics analysis (MEGA) software version 4.0. Mol. Biol. Evol. 24, 1596–1599.

41] Tanigawa, K., Kawabata, H., Watanabe, K. (2010) Identification and typing ofLactococcus lactis by matrix-assisted laser desorption ionization-time of flight

mass spectrometry. Appl. Environ. Microbiol. 76, 4055–4062.

42] Thompson, J., Higgins, D., Gibson, T. (1994) CLUSTAL W: improving the sensitiv-ity of progressive multiple sequence alignment through sequence weighting,position-specific gap penalties and weight matrix choice. Nucleic Acids Res. 22,4673.

[

Microbiology 34 (2011) 400– 407 407

43] Umeyama, T., Sano, A., Kamei, K., Niimi, M., Nishimura, K., Uehara, Y. (2006)Novel approach to designing primers for identification and distinction of thehuman pathogenic fungi Coccidioides immitis and Coccidioides posadasii by PCRamplification. J. Clin. Microbiol. 44, 1859.

44] Van der Wolf, J., Elphinstone, J., Stead, D., Metzler, M., Müller, P., Hukkanen, A.,Karjalainen, R. (2005) Epidemiology of Clavibacter michiganensis subsp. sepe-donicus in relation to control of bacterial ring rot. Plant Res. Int. Rep. 95.

45] Zgurskaya, H., Evtushenko, L., Akimov, V., Kalakoutskii, L. (1993) Rathayibactergen nov including the species Rathayibacter rathayi comb. nov., Rathayibactertritici comb. nov., Rathayibacter iranicus comb. nov., and six strains from annual

46] Zinniel, D.K., Lambrecht, P., Harris, N.B., Feng, Z., Kuczmarski, D., Higley, P.,Ishimaru, C.A., Arunakumari, A., Barletta, R.G., Vidaver, A.K. (2002) Isolationand characterization of endophytic colonizing bacteria from agronomic cropsand prairie plants. Appl. Environ. Microbiol. 68, 2198–2208.