ORIGINAL RESEARCH

Characterization of Clavibacter michiganensis subsp. michiganensisstrains from recent outbreaks of bacterial wilt and canker in Serbia

Svetlana Milijašević-Marčić &

Karl-Heinz Gartemann & Jonas Frohwitter &

Rudolf Eichenlaub & Biljana Todorović &

Emil Rekanović & Ivana Potočnik

Accepted: 18 July 2012 /Published online: 8 August 2012# KNPV 2012

Abstract Sixty-eight Clavibacter michiganensis subsp.michiganensis (Cmm) strains from recent outbreaks ofbacterial wilt and canker in Serbia were collected fromseveral tomato growing regions during a three-year pe-riod. The pathogen was identified based on bacteriolog-ical characteristics and pathogenicity tests and theidentity of strains was confirmed by DAS ELISA andPCR amplification using primers CMM5/6 and PSA4/R.The strains showed homogeneity in biochemical andphysiological properties. However, pathogenicity testsrevealed differences in virulence that are presumably

due to a loss of the pat-1 gene. Further strain character-ization using DNA-based methods revealed a high di-versity of the Serbian Cmm strains. Based on multi-locussequence typing (MLST) analyses of five genes, Cmmstrains were divided into seven groups. The pulsed-fieldgel electrophoresis (PFGE) pattern of a selection ofstrains supported the groupings based on trees of thekdpA/sdhA sequences. On the other hand, groupingsmade according to PFGE and MLSTwere not correlatedto plasmid content in all cases. This study suggested thathigh genetic variability of the Serbian Cmm strains wasdetected both in MLST and PFGE analyses, and couldhave resulted either from new Cmm strains being intro-duced by seeds from different origins or as a conse-quence of an intraspecific hybridization process. Inaddition, this study proposed MLST as an efficient toolin epidemiological studies, population biology investi-gations and tracking the routes of transmission of patho-gens. Four of the five house-keeping genes (kdpA, sdhA,ligA and gyrB) selected to characterize Cmm strainsproved to be suitable for the MLST analysis. This is thefirst study carried out on the characterization of Cmmusing MLST.

Keywords Bacteriological characteristics . Bacterialwilt and canker .MLST. PFGE . Plasmids . Tomato

AbbreviationsCmm Clavibacter michiganensis subsp.

michiganensis

Eur J Plant Pathol (2012) 134:697–711DOI 10.1007/s10658-012-0046-x

Electronic supplementary material The online version of thisarticle (doi:10.1007/s10658-012-0046-x) containssupplementary material, which is available to authorized users.

S. Milijašević-Marčić (*) :B. Todorović : E. Rekanović :I. PotočnikInstitute of Pesticides and Environmental Protection,Banatska 31b, P.O. Box 163, 11080 Belgrade, Serbiae-mail: svetlana.milijasevic@pesting.org.rs

K.-H. Gartemann : J. Frohwitter : R. EichenlaubDepartment of Genetechnology/Microbiology,Faculty of Biology, University of Bielefeld,33501 Bielefeld, Germany

K.-H. Gartemanne-mail: kgartema@cebitec.uni-bielefeld.de

J. Frohwittere-mail: jonas.frohwitter@uni-bielefeld.de

R. Eichenlaube-mail: eichenlaub@uni-bielefeld.de

DASELISA

Double Antibody Sandwich ELISA

MLST Multi-locus sequence typingNJ Neighbor joiningPFGE Pulsed-field gel electrophoresis

Introduction

Bacterial wilt and canker of tomato (Solanum lycoper-sicum), caused by the Gram-positive bacteriumClavibacter michiganensis subsp. michiganensis(Cmm), is the most important bacterial disease oftomato. It can drastically reduce tomato yield andquality, thus causing substantial economic losses bothin greenhouses and in open-field production (Gitaitiset al. 1991; Chang et al. 1992). The pathogen is seed-borne and persists in plant debris in soil and on con-taminated greenhouse structures. Its long distance dis-semination seems to be carried out by contaminatedseed (de Leon et al. 2009; Kawaguchi et al. 2010).Infected seed is often considered to be the primarysource of inoculum, as well as a major source of Cmminfection outbreaks (Tsiantos 1987). Additional sour-ces are plant debris in infested soil and possibly otherplant species acting as natural reservoirs (Kleitman etal. 2008).

So far, many investigations have been carriedout to find adequate methods for the control ofCmm, but none has been found to be completelyeffective (Gleason et al. 1993; de Leon et al. 2011).Preventive cultural management recommendations,which include the use of pathogen-free seeds andtransplants, greenhouse disinfection, plant debris re-moval or ploughdown and rotation with non-solanaceous plants for at least two years (Gleason etal. 1993), still remain indispensable in the pathogencontrol. In Serbia, Cmm was first described in the latefifties (Šutić 1957), but had not posed a serious threatfor more than 50 years. However, severe outbreaks ofbacterial wilt and canker occurred in 2006 in severaltomato growing regions, mainly in greenhouse pro-duction (Milijašević et al. 2006, 2009). Detailed inves-tigations in tomato production regions during 2006–2008 showed the presence of canker and wilt symp-toms in greenhouses on five sites and on a single sitein open-field production.

A number of methods have been developed for theidentification and characterization of Clavibacter.Besides conventional isolation, reinfection and bio-chemical characterization, faster methods relying onPCR, ELISA and BIOLOG are nowadays commonlyused (Anonymous 2005; Kaneshiro et al. 2006). Mostof these methods aim only at the detection of thepathogen for appropriate diagnostics, but some meth-ods allow investigation of the population structure andeven epidemiological studies. The most recent re-search results have shown that among DNA-basedtyping procedures, PFGE (Kleitman et al. 2008),Rep-PCR using spatial pattern by Morisita-index(Kawaguchi et al. 2010) and ISSR-PCR fingerprints(Baysal et al. 2011) are suitable for epidemiologicalstudies such as strain tracking, discovering of thesources of contamination and transmission, and mon-itoring of the distribution and spread of Cmm.

However, knowledge about the ecology ofClavibacteris still scarce. Many species of solanaceous plants canbe infected by Cmm artificially (Strider 1969), but itis uncertain whether such plants also constitute natu-ral reservoirs for the bacteria. Generally, it is assumedthat most outbreaks are caused by already infectedseeds or seedlings distributed by seed companies. Itremains unknown whether virulent strains spread toadjacent areas or reinfections from plant material leftin the soil commonly occur.

A method to address these questions is the multi-locus sequence typing (MLST). This means of typingbacteria is a helpful tool in molecular epidemiologicalstudies, population biology and evolutionary analyses,and widespread now for human pathogens. A fewMLST schemes have also been developed for severalplant pathogens (www.pamdb.org; Almeida et al. 2010).The method has the advantage of enabling direct trans-fer and comparison of data generated for strains betweendifferent labs all over the world once the strain databasehas been built, while time-consuming controls alwayshave to be conducted for many other PCR-based meth-ods (BOX, AFLP, ERIC). The data generated shouldallow the identification and phylogenetic positioning ofCmm strains. Also, newly isolated strains can be easilygrouped into an existing framework of strains. For theMLST analysis, 5–10 house-keeping genes are ampli-fied by PCR and sequenced (Maiden 2006). House-keeping genes are under stabilizing selection and arepresent in all strains of a relevant genus/species to beanalyzed. Mutations in these essential genes are fixed

698 Eur J Plant Pathol (2012) 134:697–711

only rarely, allowing the construction of phylogenetictrees on the strain level both for individual genes and forthe strains themselves.

Since the genome sequences of Cmm NCPPB382and Clavibacter michiganensis subsp. sepedonicus(Cms) ATCC33113 became available (Bentley et al.2008; Gartemann et al. 2008) characterization ofClavibacter strains by MLST has become possible.Using MLST, strains can be tracked for their geneticdifferences, and their distribution and spread can bemonitored. In this study a first approach to developingan MLST scheme for Clavibacter is described.

Since there are no data relating to the origin, dis-semination and genetic diversity of Cmm strains inSerbia, the aim of this study was to characterizeCmm strains from recent outbreaks of bacterial wiltand canker using both phenotypic characteristics andDNA-based means of typing bacteria (PFGE, MLSTand plasmid content).

Materials and methods

Bacterial strains and growth conditions

The Cmm strains used in the present study were iso-lated from different parts of diseased tomato plants(leaves, stems, petioles, fruits, seeds) collected fromseveral tomato growing regions in Serbia during2006–2008 (Table 1). All strains except those fromStara Pazova were isolated from greenhouse crops.Serial dilutions of macerated plant tissue or seedssoaked in PBS saline (phosphate buffer saline) wereplated on mSCM, D2ANX, mCNS and NBY (Hadas etal. 2005; Schaad et al. 2001; Mortensen 1999). Plateswere incubated at 26 °C and examined after four to11 days. Presumptive colonies were purified by sub-culturing on Nutrient Glucose Agar (NGA) and YeastPeptone Glucose Agar (YPGA). Single cell colonieswere transferred on to King’s B medium (KBM) slants(Lelliott and Stead 1987) and stored at 4 °C for furtherstudies. For long-term storage they were preserved at−80 °C in liquid KBM with a final concentration of20 % (v/v) glycerol in the culture collection of theInstitute of Pesticides and Environmental Protection,Belgrade. The strains were compared with the typestrain of C. michiganensis subsp. michiganensisCFBP 4999 (equivalent strain designation0NCPPB2979) obtained from Collection Française des

Bactéries Phytopathogènes (CFBP). Additionally, thesequenced Cmm strain NCPPB382 was used as areference for some experiments. For the isolation ofplasmid DNA and DNA for pulsed-field gel electro-phoresis (PFGE) the strains were grown in TBY richmedium (10 g l-1 tryptone, 5 g l−1 yeast extract, 5 g l−1

NaCl, pH 7.5) at 26 °C for 6 days.

Pathogenicity test

Pathogenicity of the isolated strains was tested using atomato seedling test (Lelliott and Stead 1987). Theinoculum was prepared from the isolated strains andthe type strain (NCPPB 2979) by suspending a24-h old single colony in 100 μl of sterile distilledwater. Tomato seedlings cv. Saint Pierre were grown insterile plant growing substrate “B medium course”(Floragard, Germany) at 26 °C and >70 % relativehumidity. Seedlings were inoculated at the second trueleaf stage for each strain, including the reference one,by injection into the stem at the cotyledons and keptunder plastic bags for 48 h. Four plants were inocu-lated for each strain and the experiment was conductedtwice. Mock-infected plants injected with water servedas a negative control. Beginning on the fifth day, plantswere observed for wilting and stem canker for another20 days. The bacterium was isolated from wilting plantsby removing a 1-cm stem section from 2 cm above theinoculation point. Suspect colonies were subculturedand identified using Gram reaction (Lelliott and Stead1987) enzyme-linked immunosorbent assay (ELISA)(Clark and Adams 1977) and PCR with CMM5/6(Dreier et al. 1995) and PSA4/R (Pastrik and Rainey1999) sets of primers. Strains were also subjected tohypersensitive reaction using four-o’clock plants(Mirabilis jalapa) (Gitaitis 1990; Carlton et al. 1998).

Pathogen identification

The Cmm strains used in the study were identifiedbased on biochemical characteristics, DAS ELISAand PCR.

Biochemical pathogen characterization was carriedout by determining the following set of phenotypicproperties: Gram reaction, metabolism of glucose, cat-alase activity, Kovac’s oxidase test, levan formation,aesculin hydrolysis, starch hydrolysis, casein hydroly-sis, H2S production from peptone, acid productionfrom mannose and mannitol, and the use of sodium

Eur J Plant Pathol (2012) 134:697–711 699

acetate as carbon source (Lelliott and Stead 1987;Schaad et al. 2001).

ELISA test was carried out with DAS ELISA com-plete kit (no. 07063C/096) purchased from LOEWEBiochemica GmbH, Germany.

PCR tests for identification were carried out withprimers CMM5/6 and PSA4/R (Dreier et al. 1995;Pastrik and Rainey 1995). Primers CMM5/6 amplifythe pat-1 gene borne on the 70-kb plasmid pCM2 inthe Cmm strain NCPPB382, while PSA4/R amplifythe intergenic spacer between 16S and 23S rRNAgenes. Primers CMM5/6 allow the detection of viru-lent Cmm strains, but some non-virulent strains aremissed by the test (Kleitman et al. 2008).

Characterization of Cmm strains by DNA-basedmethods

Aiming to get a better insight into population struc-tures of the isolates, a further characterization by PCR

was conducted for five chromosomal house-keepinggenes that should be present in all Cmm strainsregardless of their virulence and two genes located inthe chromosomal pathogenicity island of CmmNCPPB382 (Table 2). Sequences of the five house-keeping genes for Cmm, Cms and Clavibacter michi-ganensis subsp. nebraskensis (Cmn) were extractedfrom the genome sequences (Gartemann et al. 2008;Bentley et al. 2008; Gartemann and Eichenlaub, un-published data). Primers were generated based on analignment of the orthologous genes in such a way thatamplification of a portion of the genes should bepossible for strains of all three Clavibacter subspecies.The primers used in the study are listed in Table 3 andamplicon length in Table 4.

KdpA (CMM_2751), sdhA (CMM_0970), ligA(CMM_1404) and gyrB (CMM_0006) are house-keeping genes encoding subunits of a K+-dependantATPase, succinate dehydrogenase, DNA ligase anda gyrase subunit, respectively. The bipA gene

Table 1 Strains of Clavibacter michiganensis subsp. michiganensis used in this study

No Strain designationa Origin(Locality)

Plant organ/Tomato hybrid

Isolationdate

MLSTGroup

PFGEGroup

1–18 P-2, P-3, P-4, P-5, P-7, P-8, P-9, P-10, P-11, P-12, P-13, P-14, P-15,P-16, P-17, P-18, P-19, P-20

PadinskaSkela

Leaf/Delfine 2006 2 B

19–22 P-64, P-67, P-68, P-69 Lebane Stem/Belle 2006 3c D

23 P-70 Lebane Stem/Belle 2006 5 A

24–27 P-71, P-72, P-73, P-74 Leskovac Stem/Magnus 2006 5 A

28–35 P-100, P-101, P-102, P-103, P-104, P-108, P-109, P-110 Trstenik Petiole/Trogir 2007 5 A

36 P-115 Šabac Stem/Trogir 2007 1

37–39 P-117, P-118, P-119 Šabac Stem/Trogir 2007 5

40–43 P-121, P-122, P-123, P-124 Šabac Stem/Trogir 2007 1 C

44–47 P-125, P-126, P-127, P-128 Šabac Stem/Trogir 2007 5 A

48 P-137 StaraPazova

Fruit/Hector 2007 3a F

49–52 P-139, P-140, P-141, P-142 StaraPazova

Stem/Hector 2007 3b E

53 P-180 StaraPazova

Seed/Hector 2007 3b

54–60 P-501, P-502, P-503, P-504, P-505, P-507, P-510 PadinskaSkela

Stem/Magnus 2008 2 B

61–63,68

P-520, P-521, P-522, P-527 Šabac Stem/Trogir 2008 4 G

64–67 P-523, P-524, P-525, P-526 Šabac Stem/Trogir 2008 5 A

69 NCPPB2979 (CFBP4999) Hungary Solanumlycopersicum

1957

70 NCPPB382 U.K. Solanumlycopersicum

1956

a Strains highlighted in gray were selected for analysis of the VspI macrorestriction pattern by PFGE

700 Eur J Plant Pathol (2012) 134:697–711

Table 2 Characterization of Clavibacter michiganensis subsp. michiganensis strains

Character Strain designation

Pathogenicitya All the strains listed in Table 1 (P-137, P-180, P-501, P-502, P-503, P-504, P-505,P-507, P-510)

PCR with plasmid based primers pat-1b All the strains listed in Table 1 except P-137, P-180, P-501, P-502, P-503, P-504, P-505, P-507, P-510

PCR with primers based on intergenic spacerbetween 16S and 23S rRNA

All the strains listed in Table 1

House-keeping and virulence genesc

kdpA All the strains listed in Table 1

sdhA All the strains listed in Table 1

gyrB All the strains listed in Table 1 (P-10, P-64, P-70, P-73, P-100, P-121, P-125, P-137,P-140, P-501, P-520, P-523, NCPPB 2979)

ligA All the strains listed in Table 1 (P-10, P-64, P-70, P-73, P-100, P-121, P-125, P-137,P-140, P-501, P-520, P-523, NCPPB 2979)

bipA All the strains listed in Table 1 (P-10, P-64, P-70, P-73, P-100, P-121, P-125, P-137,P-140, P-501, P-520, P-523, NCPPB 2979)

tomA All the strains listed in Table 1 (P-10, P-64, P-70, P-73, P-100, P-121, P-125, P-137,P-140, P-501, P-520, P-523, NCPPB 2979)

chpC All the strains listed in Table 1 (P-10, P-64, P-70, P-73, P-100, P-121, P-125, P-137,P-140, P-501, P-520, P-523, NCPPB 2979)

Macrorestriction patternd P-10, P-64, P-70, P-73, P-100, P-121, P-125, P-137, P-140, P-501, P-520, P-523,NCPPB 2979, NCPPB382

a Pathogenicity tests on tomato plants were carried out as described in Materials and methods. Strains in parentheses induced weaksymptoms, showing an attenuated virulenceb Strains printed in bold letters were negative in PCRc PCR amplicates of expected size were obtained from all listed strains. All PCR products of kdpA and sdhAwere sequenced using thesame primers as for the PCR. For ligA, gyrB, bipA, tomA and chpC amplicons only selected products from one or two members of eachof the groups established by kdpA and sdhAwere sequenced (given in parentheses)d PFGE analysis of VspI digested DNA

Table 3 PCR primers used in this study

Primer Sequence Annealing temperature Amplicon (bp) Reference

kdpA_forward GTG CAG AAC TTC GTC TCG G 60 °C 693 this workkdpA_reverse GAG CAT CAT GTT GAT CAT CG

sdhA_forward CCT GGATGT TCG TGT ACC 58 °C 778 this worksdhA_reverse GAG GAC ATG GAG TTC TTC C

chpC_forward GCT CTT GGG CTA ATG GCC G 60 °C 639 Kleitman et al. 2008chpC_reverse GTC AGT TGT GGA AGATGC TG

gyrB_forward GAC ATC CAG ATC ACC ATG C 58 °C 606 this workgyrB_reverse GCT GAT CTT CTT GAC CGT G

ligA_forward GTT CGA CGA GCT GAATGC 56 °C 544 this workligA_reverse CTC GAC CTT CTC CAT GAC

bipA_forward GAT CTT CAC GTT CTT GAC G 58 °C 698 this workbipA_reverse GCA TGA TGG ACT CGA ACG

tomA_forward CGA ACT CGA CCA GGT TCT C 60 °C 529 Kleitman et al. 2008tomA_reverse GGT CTC ACG ATC GGA TCC

Eur J Plant Pathol (2012) 134:697–711 701

(CMM_2179) codes for a putative elongation factor, ahomolog of Ef-Tu, with an unknown function.Additionally, two genes involved in virulence wereselected: ChpC (CMM_0052), which encodes a se-creted serine protease located on the putative pathoge-nicity island that is involved in colonization (Stork etal. 2008) and tomA (CMM_0090) that codes for thetomatinase of Cmm (Kaup et al. 2005). For thesegenes, it is known that some isolates do not carry them(Kleitman et al. 2008), and they are absent from Cmsand Cmn, too.

For PCR, cells from 2–3 weeks old cultures weredirectly used (about one colony for each PCR reac-tion). The PCR reactions consisted of 2 μl DMSO,2 μl premixed primer pairs (25 pM of each primer),40 μl Taq Mastermix (Taq PCR Master Mix Kit,Qiagen, Germany), and 36 μl water. The PCR wasrun in a DNA Engine Dyad™ (MJ Research) using thefollowing program: 10 min at 94 °C, followed by 35cycles of 90 s at 94 °C, 2 min at annealing temperature(Table 3), and 2 min at 72 °C; and a final extension for10 min at 72 °C. PCR products were sequenced by theIIT Biotech GmbH (Bielefeld, Germany) with bothprimers also used in the PCR reaction.

Sequences of each strain were aligned separately(Lasergene, DNAStar) and manually corrected usingChromas 2.13 (www.technelysium.com.au) .Consensus sequences were used to generate Neighborjoining (NJ) trees using the default parameters ofClustalX1.83 (Thompson et al. 1997). For sdhA, only700 bp of overlapping sequence information wereobtained and used for the generation of the tree, whilefor the other genes overlapping sequences of the com-plete amplicon were obtained. The corresponding

sequences of Clavibacter michiganensis subspeciessepedonicus and nebraskensis were used as outgroups.Bootstrapping was conducted with 1000 repetitions.

For kdpA and sdhA all amplicons were sequencedand grouped according to the generated tree. Onlysome representatives of each group were analyzedfor the remaining genes after amplification and se-quencing (Table 2).

DNA isolation

Plasmid DNAwas isolated according to the method ofRamos-Gonzalez et al. (1991). The bacteria were grownin TBY liquid medium for 4–6 days at 26 °C. A 0.5 mlportion of the cultures was harvested, the resultingpellet was resuspended in 200 μl solution A containing8 mg/ml lysozyme and incubated at 37 °C for 45 min.An amount of 100 μl of solution B was added andcarefully mixed. After 5 min incubation at 55 °C theDNA solution was phenolized. Finally, the aqueousphase was directly separated in 0.8 % agarose gels.

DNA for PFGE was prepared as described byRedenbach et al. (2000) from 20 ml cultures grownfor 6 days at 26 °C. DNAwas digested with VspI andseparated in a Biorad CHEF-DR II PFGE unit (Biorad,München, Germany). The DNA was separated in 1 %agarose gels at 6 V/cm in 0.5 TBE buffer at 14 °C for24 h. The pulse time changed linearly from 10 to 60 s.These parameters allow a separation of fragments upto approximately 700 kb, larger fragments are notseparated. Because the PFGE was only used for diag-nostic purposes, only a comparison of fragment pat-tern was conducted, the size and number of all VspIfragments was not determined.

Table 4 Sequence variation of the genes

Gene Amplicon (bp) No. of strainssequenced

Sequencetypes

polymorphic sites(No., percentage)

silent mutations/amino acid exchangeb

kdpA 693 68 6 19 (2.74 %) 16 (3)

sdhA 700a 68 7 36 (5.14 %) 36 (0)

gyrB 606 13 4 4 (0.66 %) 2 (2)

ligA 544 13 4 5 (0.92 %) 3 (2)

bipA 698 13 1 0 –

tomA 529 13 3 2 (0.38 %) 2 (0)

chpC 639 13 1 0 –

a The amplicon has 778 bp, but only 700 bp of overlapping sequence information were obtainedb Amino acids exchanges compared to the sequence of the Cmm type strain

702 Eur J Plant Pathol (2012) 134:697–711

Nucleotide sequence accession numbers

The sequences reported here have been deposited inthe EMBL database (accession nos.: HE861572-HE861641, HE861642-HE861711, HE861531-HE861543, HE861544-HE861557, HE861558-HE861571, HE861505-HE861517, and HE861518-HE861530 for kdpA, sdhA, gyrB, ligA, bipA, tomAand chpC, respectively).

Results

Characteristics of Clavibacter michiganensis subsp.michiganensis strains

A survey of tomato growing regions in Serbia duringthree consecutive years (2006–2008) resulted in theisolation of sixty-eight Cmm strains (Table 1). Thestrains were subjected to pathogenicity tests and mostof them showed canker and wilt symptoms after15 days. Plants injected with sterile distilled waterremained healthy. However, nine Cmm strains origi-nating from Padinska Skela (P-501, P-502, P-503,P-504, P-505, P-507 and P-510) and Stara Pazova(P-137, P-180) induced only weak symptoms (leafcurling and stem canker), showing reduced virulence.Bacteria were isolated from the symptomatic plantsand identified. All Cmm strains also induced hyper-sensitive reaction (HR) on ‘four-o’clock’ plant(Mirabilis jalapa) leaves within 24 h of infiltrationwith bacterial cells.

The results of physiological and biochemical testsof the bacterium were as follows for all strains isolat-ed: Gram positive; oxidative metabolism of glucose;catalase positive; oxidase negative; levan negative;aesculin hydrolysis positive; starch hydrolysis posi-tive; casein hydrolysis negative; H2S produced frompeptone; acid produced aerobically from mannose butnot from mannitol; and sodium acetate used as acarbon source. All isolates had the same character-istics as the type strain NCPPB2979.

All Cmm strains reacted positively in the DASELISA test and the PCR test with the PSA4/R primerpair. All strains were positive for pat-1 except thestrains P501-510 (Padinska Skela, 2008) and strainsP137 and P180 (Stara Pazova, 2007).

Five genes were chosen for the development of anMLST scheme for Clavibacter, two of which are

involved in basic metabolism, two in DNA metabo-lism, and one probably involved in translation. For thehouse-keeping genes kdpA and sdhA PCR ampliconsof the expected size were obtained from all strains. Allamplicons were sequenced using the same primers asfor the PCR. The virulence gene chpC was also suc-cessfully amplified from all strains, but only somestrains were sequenced. For the other genes, onlyrepresentative strains were amplified and sequenced.

In the kdpA amplicon, nineteen positions (2.74 %)showed variations in the nucleotide sequence leadingto three amino acid exchanges, compared to kdpA ofNCPPB2979 (Table 2 and 4). The tree derived fromthe kdpA amplicons showed seven sequence typesfor the Serbian strains (Fig. 1). An alignment ofone representative of each group is shown inSupplementary Figure 1. Strains from the same loca-tion and the same year displaying the same sequencemay represent clonal isolates. Strain NCPPB382 orig-inating from the UK was isolated in the kdpA tree butwas more closely related to most of the Serbian strainsthan the type strain. All strains from Padinska Skela(group 2) displayed the same kdpA sequence regard-less of the date of isolation. Also, only one sequencetype was identified in Leskovac and Trstenik (group 5)and the same one was also found in Šabac. The Šabacstrains showed the greatest variability, they were dis-tributed over three groups. Five strains from Šabac(group 1) had the same sequence as the type strainNCPPB2979 isolated in Hungary in 1957, while theother Šabac strains isolated in the same year clusteredin group 5, indicating two different Cmm strains in thearea in 2007. In 2008, group 5 strains were still pres-ent, which implicated that these strains survived in thegreenhouses. Additionally, a new group 4 appeared,while no strains belonging to group 1 were isolatedanymore.

The alignment of the sdhA sequences revealed 36polymorphic sites (5.14 %) (Table 4). The internalgroupings of the Serbian strains were consistent inthe trees derived from kdpA and sdhA, with group 3being the only exception (Fig. 2; SupplementaryFig. 2). Only the relative positions of the groups andthe reference strains NCPPB2979 and 382 varied, e.g.group 4 strains clustered with group 5 in the kdpA tree,while they were more closely related to group 1 in thesdhA tree.

As for kdpA, group 5 contained strains from differ-ent locations: the strains isolated in Trstenik (2007,

Eur J Plant Pathol (2012) 134:697–711 703

Trogir) and Leskovac (2006, Magnus) had identicalsdhA sequences. They were also identical to one strainfrom Lebane (P70, 2006, Belle) and some of thestrains from Šabac (2007 and 2008, Trogir). The otherLebane strains (2006, Belle) clustered together withthe strains from Stara Pazova (2007, Hector, openfield) in the kdpA tree (group 3c), but they formed adistinct group related to group 2 in the sdhA tree. Inthe sdhA tree, one strain from Stara Pazova (P137)

clustered only with NCPPB382, while the five remain-ing strains were distantly related to group 5. Since theclustering of the Serbian strains is supported by bothtrees, only a few strains from each (sub)group thusdefined were analyzed for the remaining genes.

The two genes involved in DNA metabolism hadfewer sequence deviations, only five and four poly-morphic sites were found for ligA and gyrB, respec-tively (Table 4; Supplementary figures 3 and 4). Four

CmsCmn

NCPPB2979TP115P121P122P123P124

P520P521P522P527

P524P525P526

P128P523

P126P127

P117P118P119P125

P100P101

P104

P109P110

P102P103

P108

P70P71

P74P73P72

P137

P139

P180P142P141P140

P69

P64P67P68

P14

P11P10

P12

P15

P13

P9

P5P4

P2

P7P8

P16

P19

P17P18

P20P501

P504P503P502

P505P507P510

P3

NCPPB382

Padinska Skela

Lebane

LebaneLeskovac

LeskovacLeskovacLeskovac

TrstenikTrstenik

TrstenikTrstenikTrstenikTrstenikTrstenik

Trstenik

Sabac

Sabac

SabacGroup 1

Group 4

Group 5

Group 2

SabacSabac

Sabac

Sabac

SabacSabac

Sabac

SabacSabacSabac

Stara Pazova

Stara Pazova

993

994

620

619

222

867

466

271

10000.005

Group 3c

Group 3b

Fig. 1 NJ phylogenetic treefor the 693 bp-kdpA ampli-cons. The sequence groupsand origins of the Serbianstrains are indicated. Boot-strap values (1000 repeti-tions) are shown at thebranches

704 Eur J Plant Pathol (2012) 134:697–711

sequence types were found both for gyrB and ligA. Incase of gyrB, both reference strains formed groups oftheir own. The Serbian strains formed two clusters:one containing P137, P140 and P520, and all the otherstrains forming the second sequence type. For ligA,group 1 included NCPPB382 and P64, while groups 2and 3 contained only P121 and P520, respectively. Allother strains had the same nucleotide sequence as thetype strain NCPPB2979.

No nucleotide variations were found in the 698 bp-amplicon of the bipA gene, all sequences were identicalto the sequenced strain NCPPB382 (data not shown). Atree for the concatenated sequences (including all3,241 bp of the five amplicons) is shown in Fig. 3.

The genes potentially involved in virulence were notused for the construction of phylogenetic trees because itis known that they are absent at least from some non-virulent strains (Kleitman et al. 2008). Both genes

CmnCms

1000

1000

1000

354

995

992

842

814

980

675

620

459

590

0.005 NCPPB2979T

NCPPB382

P115P121P122P123P124

SabacGroup 1

P14

P11P10

P12

P15

P13

P9

P5P4

P2

P7P8

P16

P19

P17P18

P20P501

P504P503P502

P505P507P510

P3

Padinska SkelaGroup 2Group 2Group 2

P520P521P522P527

Group 4Sabac

P69

P64P67P68 Lebane

Group 3c

P139

P180P142P141P140

Stara PazovaGroup 3b

P137 Stara Pazova

P524P525P526

P128P523

P126P127

P117P118P119P125

P100P101

P104

P109P110

P102P103

P108

P70P71

P74P73P72

LebaneLeskovac

LeskovacLeskovacLeskovac

TrstenikTrstenik

TrstenikTrstenikTrstenikTrstenikTrstenik

Trstenik

Sabac

Sabac

Group 5

SabacSabac

Sabac

SabacSabac

Sabac

SabacSabacSabac

Fig. 2 NJ phylogenetic treefor the 700 bp of overlap-ping sequence of the sdhAamplicons. Sequence typesand localities of isolation ofthe strains are indicated.Bootstrap values (1000 rep-etitions) are shown at thebranches

Eur J Plant Pathol (2012) 134:697–711 705

showed only minimal sequence variation. In tomA onlytwo variable positions occurred. Both are silent mutationsin the third codon position not affecting the amino acidsequence. Three sequence types occurred, one containingonly Cmm NCPPB382, another one for P121 fromŠabac, while all other Serbian strains grouped with thetype strain NCPPB2979 (Supplementary figure 5). InchpC no polymorphisms were found (data not shown).

Macrorestriction patterns of the Serbian Cmm strains

An analysis of the VspI macrorestriction pattern byPFGE was conducted for the strains selected by thekdpA/sdhA trees (Table 2, Fig. 4). The pattern

supported the groupings made: the strains from group5 (P70, P73, P100, P125 and P523) showed identicalrestriction fragments. Also the two strains of group 2(P10 and P501) seem to be identical, although neitherhad VspI fragments in the size range separated at all.Only one representative was analyzed from each of theother groups. Each had a distinct pattern and none wasclosely related to NCPPB2979 or NCPPB382. Thegroup 3 strains (3a P137; 3b P140; 3c P64) are partic-ularly notable for having no common VspI fragmentsin the size range analyzed, and the Šabac strain P121from group 1 showed a completely different patternthan the Šabac strains from groups 4 (P520) and 5(P125 and P523).

1000

751

305

1000

P520

947

661

972

595

406

1000

0.005

Cms

Cmn

NCPPB2979T

NCPPB382

P137(Stara Pazova)

P64

P10

P501 (Padinska Skela)

(Lebane)

P520Group 4(Sabac)

Group 3c

Group 3a

P121 (Sabac)Group 1

P140 (Stara Pazova)Group 3b

P125

P100

P70

P73(Lebane,Leskovac,Trstenik,Sabac)

P523

Group 5

Group 2

Fig. 3 a NJ phylogenetic tree for the concatenated sequences ofthe five genes kdpA, sdhA, gyrB, ligA and bipA used in the MLSTanalysis of the Cmm strains. Bootstrap values (1000 repetitions)

are shown at the branches. b Distribution of Cmm groups asdetermined byMLST (numbers) and PFGE (letters in parenthesis)in Serbia

706 Eur J Plant Pathol (2012) 134:697–711

Plasmid content

Plasmids were isolated from some strains in each groupestablished by the MLST analysis. Additionally, plas-mids were isolated from the strains negative in the PCRwith CMM5/6 since this may be caused by the loss of apat-1 carrying plasmid as in the attenuated strainCMM101, a derivative of NCPPB382 lacking the largecircular plasmid pCM2 which carries pat-1 (Meletzus etal. 1993). The pat-1-negative strains P137 and P180from Stara Pazova were clearly distinct from all otherStara Pazova strains, they lack the pat-1 gene and as aconsequence have a reduced virulence. However, onlystrain P137 grouped differently in both the MLST andthe PFGE analysis, while strain P180 was identical to

the other strains in the MLST analysis. No PFGE wasconducted for P180. This may be caused by differencesin plasmid status in these strains which are not reflectedby PFGE (there are no VspI sites in the known plasmidspCM1 and pCM2 from NCPPB382) and MLST. Allstrains isolated from Padinska Skela belonged to group2, a group containing no strains from other areas, inde-pendent of the time of isolation or the cultivar, so thatthe corresponding Cmm strains should have survived inthe greenhouses. However, all isolates from PadinskaSkela in the second year (2008) seemed to have lost thepat-1 gene, which led to a reduced virulence.

The strains from Padinska Skela that showedreduced virulence (P501-P510) had only one plas-mid slightly larger than pCM1 of NCPPB382;

Fig. 3 continued.

Eur J Plant Pathol (2012) 134:697–711 707

while the fully virulent strain P10 of the sameorigin contained a second, larger plasmid that in-dicated a loss of the plasmid carrying pat-1(Fig. 5a). Considering the Stara Pazova strains,strain P137 showed only one plasmid band, similarin size to pCM1 from NCPPB382. Strain P140from the virulent group 3b apparently containedthis pCM1-sized plasmid and the second, largerone, while the smaller plasmid was lacking in thenon-virulent strain P180 (Fig. 5b). Most plasmidsin the strains analyzed do not match the sizes ofpCM1 and pCM2 from NCPPB382.

Discussion

This study showed that all Cmm strains from Serbiahad the same response as the reference strain in testsof biochemical and physiological characteristics.However, pathogenicity tests revealed differences invirulence between Serbian strains. Strains P501-510(Padinska Skela, 2008) and strains P137 and P180(Stara Pazova, 2007) showed an attenuated virulence.As they were also negative in the PCR test for pat-1gene, the reason seems to be the loss of pat-1 andpossibly of the plasmid carrying the gene. Moreover,DNA-based methods revealed a high diversity of theSerbian Cmm strains. Based on MLST analyses, the

Serbian Cmm strains were divided into seven groups.The PFGE pattern supported the groupings madebased on trees of the kdpA/sdhA sequences.

This study demonstrated a genetic variability exist-ing in different tomato growing regions (Fig. 3).Group 2, for example, was only detected in strainsfrom Padinska Skela. Group 5, being the most wide-spread, seemed to be dominant since it was detected infour different growing regions throughout the threeconsecutive years. Our study implied that strains ingroup 2 from the same location (Padinska Skela) iso-lated in two different years survived in greenhouses. Asimilar scenario occurred in Šabac with the group 5strains being also repeatedly isolated despite the con-trol measures undertaken. These findings can be veryuseful to tomato breeders.

It is important that samples from both of theselocalities were taken from large growing facilities. InSerbia, tomatoes are grown once or twice a year ingreenhouses along with the standard open field pro-duction but such practice depends on whether a heat-ing system is available. In most cases, small-scalegrowers only produce early greenhouse tomatoes inthe spring. In these small facilities, phytosanitarymeasures are usually poor and the practice of disin-fection is mostly neglected, so that infested substrate,pots and tools may allow the pathogen to survive andthe inoculum to augment from year to year. On theother hand, large growers have two harvests a year, inthe spring and autumn. Although these facilities havebetter sanitary management (disinfection, spraying),growing tomatoes twice a year might be the reasonfor pathogen survival. Moreover, Cmm is able to sur-vive in plant debris in soil and in greenhouse struc-tures for several years (Fatmi and Schaad, 2002;Gleason et al. 1993), while the known bactericidescan reduce the population size and spread of the bac-terium (Werner et al. 2002) but ultimately fail tocontrol the pathogen fully. However, the pathogenhad not been recorded in Serbia for more than 50 years(Milijašević et al. 2009).

It is a well known fact that infected seeds may actas the primary source of inoculum, as well as a majorsource of Cmm infections (Tsiantos 1987). Recentoutbreaks in Serbia occurred in two major growingareas in 2006. Early production of tomatoes, mostlyintended for fresh market, has increased during thepast decade, especially in greenhouses. Since tomatoesare mainly produced either from imported seeds

Fig. 4 Analysis of the macrorestriction pattern of VspI digestedDNAbyPFGE. TheDNAwas run in a 1%agarose gel at 6V/cm in0.5 TBE buffer at 14 °C for 24 h. The pulse times changed linearlyfrom10 to 60 s.The sizes (in kb) of theλ concatemers are indicated.The type strain and the sequenced reference strain are underlinedand the PFGE groupings of the Serbian strains are given

708 Eur J Plant Pathol (2012) 134:697–711

obtained from international seed companies or fromseedlings grown by producers from imported seeds,many tomato seed varieties are introduced each sea-son. Introduction and transmission of the pathogen byseeds can be suggested. However, our study alsoshowed a genetic variability within the same regionas the strains from Šabac clustered into three differentgroups. Partially, this is due to the fact that sampleswere collected from several large greenhouses in thisarea but it also points to contaminated seeds as thenovel sources of infection. Moreover, the strains fromŠabac isolated in the same year (2007) were distribut-ed into two groups (group 1 and group 4). This studysuggested that high genetic variability of the SerbianCmm strains from recent outbreaks was detected bothby MLST and PFGE analyses, and could be either aresult of introduction of the new strains of the bacte-rium by seeds from different origins or a consequenceof an intraspecific hybridization process resulting innovel pathogenic behavior.

Employing molecular tools in epidemiologicalinvestigations is often useful to identify and under-stand transmission routes of microorganisms. Ourstudy confirmed PFGE as a suitable method forDNA typing in epidemiological studies of Cmm, aspreviously reported by Kleitman et al. (2008).However, the high difference that was observedamong the Serbian PFGE groups may reflect a varia-tion in the presence and location of a chromosomallow G+C island in the corresponding genomes as wasreported in other studies (Gartemann et al. 2008;Kleitman et al. 2008).

DNA sequence analysis of the two virulence genesshowed that they can be used for the identification of(virulent) Cmm strains but the virulence genes lack aresolving power to differentiate between the strains. Inaddition, the analysis of house-keeping genes byMLST also proved to be a successful tool for determi-nation of genetic diversity of Cmm strains. This is thefirst study carried out on the characterization of Cmmusing a MLST scheme. This method proved to have anumber of advantages over other DNA typing proce-dures. The most important ones are reproducibility, anability to detect changes at DNA level which is notpossible by phenotypic approaches, and a generateddatabase that can be shared between labs throughoutthe world. These features proposed MLST as an effi-cient tool in epidemiological studies, population biol-ogy investigations and tracking the routes of pathogentransmission (Maiden et al. 2006). Four of the genestested in this study seem to be suitable for MLSTanalysis. The genes involved in metabolism showeda relatively high number of polymorphic sites, whilethe genes ligA and gyrB from DNA replication had alower rate of sequence variation. The fifth gene,bipA, probably involved in translation, displayed nopolymorphisms at all, and thus may be excludedfrom future improvements of the MLST scheme.Furthermore, this gene may be not selectively neutralas a mutant with an inactivated bipA has shown anattenuated virulence phenotype (Gartemann et al.unpubl. data).

On the other hand, the groupings made according toPFGE and MLST were not correlated to plasmid

Fig. 5 Plasmid profiles of representativeClavibactermichiganen-sis subsp. michiganensis strains. Conditions were described inMaterials and methods. a Strains from MLST group 2 originating

fromPadinskaSkela.Theisolateof2006(P10)shows twoplasmids,while the 2008 isolates (P501-510) retain only the smaller plasmid.bPlasmidprofiles fromrepresentativesof theSerbianMLSTgroups

Eur J Plant Pathol (2012) 134:697–711 709

content in all cases, implying that plasmid profilesalone should not be used for diagnostic purposes, butare still important as they influence the degree ofpathogenicity. This had been reported previously byKleitman et al (2008), who suggested that the plasmidsand the chromosome evolved independently. Therefore,a complete diagnostic procedure for Clavibacter shouldinclude an analysis of both the chromosome (e.g. byMLST or PFGE) and the plasmid(s) (detection of pat-1and celA, investigation of the pCM1/2 replicons byPCR, and identification of further plasmids).

Acknowlegdements This research was carried out as a col-laboration of the Faculty of Biology, University of Bielefeld,Germany, Department of Genetechnology/Microbiology and theInstitute of Pesticides and Environmental Protection, Belgrade,Serbia, Department of Plant Pathology. A segment of this studywas conducted as a part of the Project TR 34043, which isfinancially supported by the Ministry of Education and Scienceof the Republic of Serbia. The sequencing of the PCR productswas conducted by IIT Biotech GmbH (Bielefeld, Germany).

References

Almeida, N. F., Yan, S., Cai, R., Clarke, C. R., Morris, C. E., etal. (2010). PamDB, a multilocus sequence typing andanalysis database and website for plant-associatedmicrobes. Phytopathology, 100, 208–215.

Anonymous. (2005). Clavibacter michiganensis subsp. michiga-nensis. PM 7/42. Bull. OEPP/EPPO Bulletin, 35, 275–283.

Baysal, O., Mercati, F., Ikten, H., Yildiz, R. C., Carimi, F.,Aysan, Y., & Teixeira da Silva, J. (2011). Clavibactermichiganensis subsp. michiganensis: Tracking strains us-ing their genetic differentiations by ISSR markers in South-ern Turkey. Physiological and Molecular Plant Pathology,75, 113–119.

Bentley, S. D., Corton, C., Brown, S. E., Barron, A., Clark, L., etal. (2008). Genome of the actinomycete plant pathogen Clav-ibacter michiganensis subsp. sepedonicus suggests recentniche adaptation. Journal of Bacteriology, 190, 2150–2160.

Carlton, W. M., Braun, E. J., & Gleason, M. L. (1998). Ingress ofClavibacter michiganensis subsp. michiganensis in to tomatoleaves through hydatodes. Phytopathology, 88, 525–529.

Chang, R. J., Ries, S. M., & Pataky, J. K. (1992). Reduction inyield of processing tomatoes and incidence of bacterialcanker. Plant Disease, 76, 805–809.

Clark, M. F., & Adams, A. N. (1977). Characterization of themicroplate method of enzyme linked immunosorbent assayfor detection of plant viruses. Journal of General Virology,34, 475–483.

de Leon, L., Rodriguez, A., Llop, P., Lopez, M. M., & Siverio,F. (2009). Comparative study of genetic diversity of Clav-ibacter michiganensis subsp. michiganensis isolates fromthe Canary Islands by RAPD-PCR, BOX-PCR and AFLP.Plant Pathology, 58, 862–871.

de Leon, L., Siverio, F., Lopez, M. M., & Rodriguez, A. (2011).Clavibacter michiganensis subsp. michiganensis, a seed-borne tomato pathogen: Healthy seeds are still the goal.Plant Disease, 95, 1328–1338.

Dreier, J., Bermpohl, A., & Eichenlaub, R. (1995). Southernhybridization and PCR for specific detection of phytopath-ogenic Clavibacter michiganensis subsp. michiganensis.Phytopathology, 85, 462–468.

Gartemann, K.-H., Abt, B., Bekel, T., Burger, A., Engemann, J.,et al. (2008). The genome sequence of the tomato-pathogenic actinomycete Clavibacter michiganensis subsp.michiganensis NCPPB382 reveals a large island involvedin pathogenicity. Journal of Bacteriology, 190, 2138–2149.

Gitaitis, R. D. (1990). Induction of a hypersensitive like reactionin four-o’clock by Clavibacter michiganensis subsp. mich-iganensis. Plant Disease, 74, 58–60.

Gitaitis, R. D., Beaver, R., & Voloudakis, A. (1991). Detection ofClavibacter michiganensis subsp. michiganensis in symp-tomless tomato transplants. Plant Disease, 75, 834–838.

Gleason, M. L., Gitaitis, R. D., & Ricker, M. D. (1993). Recentprogress in understanding and controlling bacterial cankerof tomato in eastern North America. Plant Disease, 77,1069–1076.

Hadas, R., Kritzman, G., Klietman, F., Gefen, T., & Manulis, S.(2005). Comparison of extraction procedures and determina-tion of the detection treshold for Clavibacter michiganensissubsp. michiganensis in tomato seeds. Plant Pathology, 54,643–649.

Kaneshiro, W. S., Mizumoto, C. Y., & Alvarez, A. M. (2006).Differentiation of Clavibacter michiganensis subsp michi-ganensis from seed-borne saprophytes using ELISA,Biolog and 16 S rDNA sequencing. European Journal ofPlant Pathology, 116, 45–56.

Kawaguchi, A., Tanina, K., & Inoue, K. (2010). Molecular typingand spread of Clavibacter michiganensis subsp.michiganen-sis in greenhouses in Japan. Plant Patholology, 59, 76–83.

Kleitman, F., Barash, I., Burger, A., Iraki, N., Falah, Y., Sessa, G.,Weinthal, D., Chalupowicz, L., Gartemann, K.-H., Eichen-laub, R., & Manulis-Sasson, S. (2008). Characterization of aClavibacter michiganensis subsp.michiganensis population inIsrael. European Journal of Plant Pathology, 121, 463–475.

Lelliott, R. A., & Stead, D. E. (1987).Methods for the diagnosisof bacterial diseases of plants. Oxford: Blackwell Scien-tific Publications.

Maiden, M. C. (2006). Multi locus sequence typing of bacteria.Annual Reviews of Microbiology, 60, 561–588.

Meletzus, D., Bermpohl, A., Dreier, J., & Eichenlaub, R. (1993).Evidence for plasmid-encoded virulence factors in the phy-topathogenic bacterium Clavibacter michiganensis subsp.michiganensis. Journal of Bacteriology, 175, 2131–2136.

Milijašević, S., Todorović, B., & Balaž, J. (2006). Clavibactermichiganensis subsp. michiganensis, bacterial canker oftomato: 1. Conventional and molecular identification. Pes-ticides and Phytomedicine, 21, 185–192.

Milijašević, S., Todorović, B., Rekanović, E., & Stepanović, M.(2009). Occurrence of bacterial canker of tomatoes in SouthernSerbia. (Paper presented at the Second International Sympo-sium on Tomato Diseases, Kusadasi,Turkey, pp 307–311).

Mortensen, C. N. (1999). Seed-borne Bacterial Diseases. DanishGovernment Institute of Seed Pathology for DevelopingCountries, Hellrup, Copenhagen, Denmark, pp. 86–92.

710 Eur J Plant Pathol (2012) 134:697–711

Pastrik, K. H., & Rainey, F. A. (1999). Identification and dif-ferentiation of Clavibacter michiganensis subspecies byPolymerase Chain Reaction-based techniques. Journal ofPhytopathology, 147, 687–693.

Ramos-Gonzalez, M.-I., Duque, E., & Ramos, J. L. (1991). Con-jugational transfer of recombinant DNA in cultures and insoils: host range of Pseudomonas putida TOL plasmids.Applied and Environmental Microbiology, 57, 3020–3027.

Redenbach, M., Scheel, J., & Schmidt, U. (2000). Chromosometopology and genome size of selected actinomycetes spe-cies. Antonie van Leeuwenhoek, 78, 227–235.

Schaad, N., Jones, J. B., & Chun, W. (2001). Laboratory guidefor identification of plant pathogenic bacteria. St. Paul,Minnesota, USA: APS Press.

Stork, I., Gartemann, K.-H., Burger, A., & Eichenlaub, R.(2008). A family of serine proteases of Clavibacter mich-iganensis subsp. michiganensis: chpC plays a role in col-onization of the host plant tomato. Molecular PlantPathology, 9, 599–608.

Strider, D. L. (1969). Bacterial canker of tomato caused byCorynebacterium michiganense: a literature review andbibliography. North Carolina Agricultural Experiment Sta-tion Technical Bulletin 193

Šutić, D. (1957). Bakterioze crvenog patlidžana. Institut zazaštitu bilja, Beograd. Posebna izdanja, doktorska diserta-cija, 1–67.

Thompson, J. D., Gibson, T. J., Plewniak, F., Jeanmougin, F., &Higgins, D. G. (1997). The ClustalX windows interface:flexible strategies for multiple sequence alignment aidedby quality analysis tools. Nucleic Acids Research, 24,4876–4882.

Tsiantos, J. (1987). Transmission of Corynebacterium michiga-nense pv.michiganense by seeds. Journal of Phytopathology,119, 142–146.

Werner, N. A., Fulbright, D.W., Podolsky, R., Bell, J., &Hausbeck,M. K. (2002). Limiting populations and spread of Clavibactermichiganensis subsp. michiganensis on seedling tomatoes inthe greenhouse. Plant Disease, 86, 535–542.

Eur J Plant Pathol (2012) 134:697–711 711