Sequence Diversity in the Dickeya fliC Gene: Phylogeny of the Dickeya Genus and TaqMan® PCR for 'D. solani', New Biovar 3 Variant on Potato in Europe

Worldwide, Dickeya (formerly Erwinia chrysanthemi) is causing soft rot diseases on a large diversity of crops and ornamental plants. Strains affecting potato are mainly found in D. dadantii, D. dianthicola and D. zeae, which appear to have a marked geographical distribution. Furthermore, a few Dickeya isolates from potato are attributed to D. chrysanthemi and D. dieffenbachiae. In Europe, isolates of Erwinia chrysanthemi biovar 1 and biovar 7 from potato are now classified in D. dianthicola. However, in the past few years, a new Dickeya biovar 3 variant, tentatively named ‘Dickeya solani’, has emerged as a common major threat, in particular in seed potatoes. Sequences of a fliC gene fragment were used to generate a phylogeny of Dickeya reference strains from culture collections and with this reference backbone, to classify pectinolytic isolates, i.e. Dickeya spp. from potato and ornamental plants. The reference strains of the currently recognized Dickeya species and ‘D. solani’ were unambiguously delineated in the fliC phylogram. D. dadantii, D. dianthicola and ‘D. solani’ displayed unbranched clades, while D. chrysanthemi, D. zeae and D. dieffenbachiae branched into subclades and lineages. Moreover, Dickeya isolates from diagnostic samples, in particular biovar 3 isolates from greenhouse ornamentals, formed several new lineages. Most of these isolates were positioned between the clade of ‘D. solani’ and D. dadantii as transition variants. New lineages also appeared in D. dieffenbachiae and in D. zeae. The strains and isolates of D. dianthicola and ‘D. solani’ were differentiated by a fliC sequence useful for barcode identification. A fliC TaqMan®real-time PCR was developed for ‘D. solani’ and the assay was provisionally evaluated in direct analysis of diagnostic potato samples. This molecular tool can support the efforts to control this particular phytopathogen in seed potato certification.


Introduction
The genus Dickeya was established by the reclassification of Pectobacterium (Erwinia) chrysanthemi and Brenneria paradisiaca as D. chrysanthemi and D. paradisiaca, respectively and for the accommodation of four new species, i.e. D. dadantii, D. dianthicola, D. dieffenbachiae and D. zeae; based on analysis of 16S rRNA gene sequences, DNA:DNA reassociation kinetics and phenotypic features including biochemical and serological reactions [1]. Multi Locus Sequence Analysis underpinned that Dickeya constitutes a distinct genetic clade in the soft rot Enterobacteriaceae [2]. Dickeya species are broad host range phytopathogens which principal disease symptom is maceration of plant tissues due to pectinolytic activity [2,3]. Strains affecting potato are mainly found in three Dickeya species, i.e. D. dadantii (biovar 3), D. dianthicola (biovars 1 and 7) and D. zeae (biovar 3). A few strains are assigned to D. chrysanthemi (biovar 5 and 6) and to D. dieffenbachiae (biovar 2) [4]. Erwinia chrysanthemi is known in potato production in some European countries for over 40 years and is associated with slow wilt and internal stem necrosis. These strains are now assigned to D. dianthicola [3,5]. However, in the past few years a new Dickeya biovar 3 strain, tentatively named 'Dickeya solani', has emerged as a common major threat, in particular on seed potatoes [5]. Across wide environmental conditions it causes extensive maceration of the seed tuber, rapid wilting and blackleg-like symptoms in the stem with decomposition of the pith. Both D. dianthicola and 'D. solani' are disseminated by infected or contaminated seed tubers. Seed certification generally implements a zero tolerance for blackleg in field inspections of high grade seed.
A phylogenetic analysis using concatenated atpD, carA and recA loci resolved the relatedness in the plant pathogenic Enterobacteriaceae [6]. Although Dickeya formed a contiguous clade with Pectobacterium, Brenneria and Samsonia, the gene amplicon sequences were sufficiently distinct to support the generic status of the taxon. Diagnostic identification of bacterial isolates as Dickeya is principally done by testing the production of indigoidin on specific culture medium [7], maceration of potato tuber tissue and the production of the 420 bp amplicon in a PCR based on the pelADE gene cluster [8]. However, robust tools for the differentiation of Dickeya species were not available until fairly recently. In the past few years, several gene loci have been found to reliably differentiate species in several taxa of plant pathogenic bacteria, e.g. Ralstonia solanacearum [9], Pseudomonas syringae [10] and Xanthomonas [11]. The recA locus was used for the first phylogenetic analysis of all species within the genus Dickeya, extending previous studies based on 16S rDNA [3]. It displayed new genomic clades and, in particular, a clonal delineation of an emerging biovar 3 variant isolated from potato in the past decade in Europe [12]. This new genetic clade was later on validated in a polyphasic analysis using dnaX sequence data and genomic fingerprinting [5]. These studies suggest that this new variant may represent a new species for which the name 'Dickeya solani' is provisionally used, but it is not formally accepted yet [4]. Further evidence for the taxonomic discrimination of this separate biological unit may be derived from sequence information of genes involved in pathogenesis and virulence.
Many phytopathogenic bacteria are motile by means of flagella and flagellar genes contribute to virulence [13,14,15] and to hostpathogen interactions, i.e. for pectinolytic Enterobacteriaceae [16,17,18]. The flagellar filament is composed of a single protein, flagellin, which is encoded by the fliC gene. The flagellin proteins contribute to antigenic variation [19] that is also displayed in Dickeya [20,21,22]. More than 10 different serogroups have been identified [23] and differences were found among Dickeya isolates from potato [24]. Sequence variability of the fliC gene has been used to differentiate among isolates of several clinical bacterial species [25,26,27] and the bacterial phytopathogen R. solanacearum [28], for molecular typing and phylogenetic analysis [29], for taxonomic applications [30] and it showed potential as a biomarker for phylogenetic and epidemiological studies [31].
This paper reports on the application of fliC sequences to differentiate Dickeya strains at the species and infraspecific level and to specifically diagnose 'D. solani' with a fliC barcode or TaqManH real-time PCR.

FliC phylogeny of the reference strains
The fliC-1/fliC-2 primers were used to produce the reference PCR amplicon of the fliC gene. The results for the individual strains are presented in Table S1. A single amplicon of approximately 650 bp was produced for all strains tested of D. chrysanthemi, D. dadantii, D. dianthicola, 'D. solani' and D. zeae. The amplicon sequences correspond to the fliC ORF region of D. dadantii 3937 strain (GenBank accession CP002038.1). A single amplicon of approximately 900 bp was obtained for the strains of D. paradisiaca. It shows homology with flagellin gene sequences of Dickeya strain Ech703 (complete genome sequence in GenBank accession CP001654). It did not reveal, however, a significant homology with fliC sequences in other available Dickeya genomes, i.e. D. dadantii 3937, Dickeya strain Ech586 and Dickeya strain Ech1591. D. paradisiaca has a limited biological and geographical distribution. Apparently strains have not been isolated over the past 30 years. Furthermore, the strains tested did not exhibit indigoidin production on NGM nor a genuine maceration activity in potato. Multiple amplicons were produced for the strains of D. dieffenbachiae. These did not share a valid homology with the fliC sequence of D. dadantii 3937. Ultimately, global sequence alignment was performed with 621 bp consensus sequences for all Dickeya reference strains, except for those of D. dieffenbachiae and D. paradisiaca. The customised phylogenetic relatedness is displayed in Figure 1. A separate clade with a clonal structure and a single sequevar is formed by the strains of 'D. solani', which are isolates from potato in Europe and Israel and one isolate from hyacinth in The Netherlands. The D. dianthicola reference strains also form a single clade with a single sequevar, regardless their different biological and geographical origin. D. dianthicola shows a very close relationship with 'D. solani'. Both are most related to the strains of D. dadantii which form a third clade. Although biologically and geographically quite diverse, D. dadantii strains also constituted a single sequevar. The D. zeae reference strains are attributed to two sub-clades and to a separate branch. The first sub-clade, phylotype 1 (P1), represents a single sequevar and contains strains isolated on the American and European continent. The second sub-clade, phylotype 2 (P2), also represents a single sequevar and consists of strains isolated on the Asian and Australian continent. Strain LMG 2497 isolated from sweet corn in the USA is attributed to a detached lineage of D. zeae and is considered a separate sequevar. The D. chrysanthemi reference strains form an aggregate clade with the type strain, other strains from chrysanthemum and strains from euphorbia, sunflower and carrot in a large sub-clade containing two sequevars. A biovar 6 strain from Parthenium and a strain from potato, both isolated in the USA, form a dichotomous branch in the D. chrysanthemi clade. An aggregate clade of Pectobacterium was formed containing P. betavasculorum, P. atrosepticum, P. carotovorum ssp. odoriferum and P. carotovorum ssp. brasiliensis. A fliC amplicon was not produced for the type strain of Pectobacterium carotovorum ssp. carotovorum and Pectobacterium wasabiae, nor for the potato associated bacteria tested.

FliC-based identification of Dickeya isolates
From the fifty isolates obtained from diagnostic samples, thirtyeight were attributed to Dickeya and twelve to Pectobacterium on the basis of indigoidin production on NGM, maceration of potato tuber tissue and a PCR amplicon produced with pelADE or pelY primers respectively. Subsequently, fliC amplicons were obtained and sequenced. The fliC phylogeny of the reference strains was used as backbone to position the isolates. All isolates preliminary identified as Dickeya with the above mentioned methods, were validated by their position in the phylogenetic fliC tree. The results are displayed in Figure 2 and in Table S2. The Dickeya isolates from potato were either classified in the D. dianthicola clade or in the 'D. solani' clade. Furthermore, the fliC sequences were identical for all strains tested of D. dianthicola and 'D. solani'. The consensus sequences exhibit twenty-five different signature positions which provide reliable barcodes to allocate isolates to one of these clades. Dickeya strain LMG 2918, isolated from Phalaenopsis orchids, is attributed to a separate branch, which is considered as an unassigned Dickeya lineage (UDL-1). Thirteen Dickeya biovar 3 isolates from greenhouse ornamentals exhibited substantial sequence variation. Six of those isolates are classified in two additional unassigned Dickeya lineages (UDL-2 and UDL-3). Four isolates are classified in the D. dadantii clade and one isolate is assigned to the D. zeae phylotype 1 sub-clade. Furthermore, another Dickeya biovar 3 isolate from Phalaenopsis orchids constitutes a fourth unassigned lineage (UDL-4) and one from Freesia clusters in the separate lineage with strain LMG 2497 from sweet corn which is now specified as UDL-5. A Dickeya biovar 3 isolate from corn in Belgium is assigned to the D. zeae phylotype 2 sub-clade and, finally, a Dickeya biovar 3 isolate from Belgian lettuce is placed in yet another unassigned lineage (UDL-6). These fifteen Dickeya biovar 3 isolates represent seven additional sequevars. The eighteen Dickeya fliC sequevars determined in this study are listed with their associated GenBank accession numbers in Table S3. Eight Pectobacterium isolates are assigned to the aggregate cluster containing P. carotovorum ssp. odoriferum and P. carotovorum ssp. brasiliensis. The fliC amplicon was not produced for four Pectobacterium isolates from potato.

FliC-based identification of Dickeya dieffenbachiae
A second primer set (fliC-for/fliC-rev) was used for D. dieffenbachiae to produce a single fliC amplicon (,370 bp) located inside the ,650 bp fliC amplicon. Comparative sequence analysis was done on 353 bp consensus sequences which were used to position the D. dieffenbachiae strains and isolates in the background of the eighteen Dickeya sequevars identified for the reference fliC fragment. The phylogram of fliC sequences trimmed at the shorter fragment is displayed in Figure 3. D. dieffenbachiae displayed an aggregate clade with two sub-clades. One contains the strains isolated from Dieffenbachia and a Dutch isolate from potato and exhibits an almost clonal structure. Another isolate from potato was attributed to a second sub-clade together with a Belgian isolate from Dieffenbachia sp. The D. dieffenbachiae clade showed a high degree of relatedness to D. dadantii. FliC TaqManH PCR for identification of 'Dickeya solani' The TaqManH real-time PCR for 'D. solani' was designed to amplify a 112 bp stretch of the fliC amplicon ( Table 1). The assay was applied on the fifty-six reference strains and the fifty diagnostic isolates. Positive reactions, attested by C t -values #25, were only obtained for the nine reference strains and the seven diagnostic isolates of 'D. solani'. Negative results, demonstrated by the absence of a C t -value after 40 PCR cycles, were obtained for all other bacterial cultures tested. The qualitative results of the fliC TaqManH PCR are given in Tables S1 and S2.
Fast fliC TaqMan PCR diagnosis of 'Dickeya solani' in symptomatic potato tissue Thirty diagnostic samples from seed potato production in Flanders were analysed in this study. They consisted of either wilting potato stems with blackleg symptoms, necrosis and maceration of the stem pith or wilting stems without blackleg but with a macerated mother tuber. A fast diagnostic procedure for 'Dickeya solani' was performed by applying the fliC TaqManH PCR without prior enrichment of the sample extracts. The test results were compared with the conventional diagnostic protocol in which the bacteria were cultured from the sample extract by dilution plating and further characterization of the isolates in phenotypic and molecular tests. The results are summarized in Table 2. Definite positive results in the fliC TaqManH PCR, attested by 18.1, C t ,28.1, were obtained for twelve out of the thirty sample extracts, diagnosing the presence of 'D. solani'. In the conventional diagnostic protocol, 'D. solani' was indeed isolated as a distinct colony morphotype from these samples and sequence analysis of the fliC amplicon produced with primers fliC-1/fliC-2 confirmed the identity. The 'D. solani' morphotype was not isolated from the other eighteen sample extracts, which underpinned the negative reactions in the fliC TaqManH PCR on the extracts. Furthermore, seven of these samples were found to contain a different Dickeya variant producing indigoidin on NGM medium, maceration of potato tuber tissue and the pelADE amplicon in PCR. These isolates were identified as D. dianthicola by sequence analysis of the fliC amplicon. Pectobacterium spp. was diagnosed in the remaining eleven samples, as confirmed by maceration of potato tuber tissue, the absence of indigoidin production on NGM medium, a positive reaction in the pelY PCR and a negative reaction in the pelADE PCR. More detail of the diagnostic tests are presented in Table S4.

Discussion
In less than a few years, a biovar 3 of Dickeya, provisionally named 'Dickeya solani', has become the predominate cause of wilting, blackleg and tuber maceration of potato in several European countries. The new form has been described as more aggressive, more likely to infect at lower cell densities and to spread from plant to plant along and even across plant rows, and causing damage in a wider range of conditions than observed for the 'traditional' blackleg (Pectobacterium atrosepticum) or for Dickeya dianthicola which is known for over 40 years in potato in some European countries [4]. Consequently, the blackleg tolerance in crops of basic and certified seed potatoes has been substantially reduced. Some countries even implement protective measures to prevent the introduction of 'D. solani' with seed potatoes imported from countries where the pathogen is known to occur. Diagnostic tests should provide rapid and reliable identification of 'D. solani' and should discriminate it from Pectobacterium and other Dickeya taxa, mainly D. dianthicola, which can cause blackleg and potato tuber maceration as well. This paper provides a method that is based on the fliC gene which codes for the flagellar subunit protein flagellin. The fliC gene can exhibit considerable intra-species sequence variation that can be used for identification at an infraspecies or even strain level [31]. Phylogenetic analysis of the fliC amplicons unambiguously differentiated five of the six currently recognized Dickeya species [1] and 'D. solani', with most branches supported by high bootstrap values. In this respect, fliC sequence comparison confirmed that 'D. solani' is clearly a new, separate and clonal clade within the genus. The short phylogenetic branches indicate the relatively close relatedness of D. dianthicola, D. dadantii and 'D. solani', suggesting that taxonomically these taxa may well be delineated at subspecies level. The fliC phylogeny does not extensively support that 'D. solani' deserves a separate species status according to the current bacterial species concept [32]. Larger fliC sequence variation exist within D. zeae with two phylotypes and within D. chrysanthemi as demonstrated in the recA phylogeny [3]. A strain isolated from sweet corn is assigned to a lineage detached of D. zeae. This strain was classified as D. zeae phylotype 1 in the recA phylogram [3]. All other corn strains of D. zeae in the fliC clade were isolated from maize varieties used for livestock fodder or processing and thus the contrasting position in the fliC phylogram may reflect the association of fliC to host designation. The D. chrysanthemi strains clustered in an aggregate clade with a large subclade identified as D. chrysanthemi biovar chrysanthemi and a detached D. chrysanthemi biovar parthenii lineage according to the species description [1]. The D. chrysanthemi strain isolated from potato in the USA is more related to the 'parthenii' biovar than to the 'chrysanthemi' biovar. The assignment of D. dieffenbachiae in the fliC phylogeny was performed with an amplicon internal to the regular 650 bp amplicon. The grouping of Dickeya taxa and sequevars remained stable. The strains and isolates of D. dieffenbachiae are closely related to D. dadantii which underpins the recent reclassification of D. dieffenbachiae as D. dadantii ssp. dieffenbachiae [33]. The robustness of the fliC phylogeny is further underpinned by analogies in phylogenies based on recA [3] and dnaX [5] loci, presented in Table S5.
The isolates from diagnostic samples were introduced the fliC phylogram allowing their assignment to a Dickeya taxon. All isolates from potato were assigned to D. dianthicola or 'D. solani' and accentuated the clonal structure of these two clades. Dickeya biovar 3 isolates from various ornamentals exhibited significant sequence variability which resulted in six unassigned lineages and, typically, strains isolated from the same plant family (the Araceae, with Philodendron and Aglaonema) constituted one of these lineages (UDL-2). So, again not totally unexpected, fliC-based bacterial typing could be informative on different plant hosts of strains. Furthermore, three of these lineages are positioned in between 'D. solani' and D. dadantii, the latter being biologically and geographically the most diverse Dickeya taxon [3]. Although the phylogenetic significance of these up to now not formally classified lineages remains to be clarified, it is our opinion that the presented data allude to the origin of 'D. solani' as being from one of the variants existing on ornamentals which then spread clonally in potato. Alternatively, fliC sequence drift when residing on different plant hosts could be responsible for the existence of the unassigned lineages. The apparent pectinolytic activity of Dickeya spp. make them broad host range pathogens which increases the potential for genetic exchange as a result of adaptation to a different environment, i.e. a new plant host [34]. Furthermore, the short branch lengths in the fliC phylogeny tend to reveal that D. dianthicola, 'D. solani', D. dadantii and the unassigned lineages UDL-1, UDL-2, UDL-3 and UDL-4 are a species complex. Further analysis should clarify the taxonomic position of these taxa, i.e. the classification at subspecies level as already done for the D. dadantii -D. dieffenbachiae aggregate [33].
The apparent clonal structure of the 'D. solani' clade enabled the development of a TaqManH PCR to specifically identify this variant and for its direct diagnosis in symptomatic potato stems and tubers. The primers are positioned in the more variable stretches of the fliC gene amplicon and the TaqManH probe is situated in a region with two single nucleotide polymorphisms. Furthermore, the 39 MGB probe is more appropriate for single base mismatches, thus increasing the specificity of the assay. 'D.
solani' is differentiated from all other reference strains and isolates, i.e. from D. dianthicola isolates from potato and from the Dickeya biovar 3 isolates from greenhouse ornamentals which were attributed to unassigned lineages. However, it did not differentiate the 'D. solani' isolate from a hyacinthus bulb which sets focus on the relation of this variant with the cultivation of bulb-producing ornamentals. The specificity of the molecular assay for 'D. solani' was also demonstrated in direct analysis of potato samples. Such diagnostics are essential if legislation that imposes a zero tolerance in seed potatoes is to be effective. The assay has not yet been validated for detection of latent infections and is, pending the outcome of these tests, proposed here as a diagnostic tool.

Conclusions
The

Bacterial strains
The strains and isolates used are listed in Tables S1 and S2. The reference set (Table S1) consisted of strains from the six currently recognized Dickeya species, typed Dickeya isolates from the new clade of biovar 3 strains from potato ('D. solani'), strains from the Pectobacterium taxa and strains of Clavibacter michiganensis subsp. sepedonicus, Ralstonia solanacearum and Paenibacillus macerans. Most reference strains were acquired from public and certified culture collections, while the isolates of 'D. solani' were obtained in the framework of the European Dickeya consortium (Dickeya Research Network hosted by the James Hutton Institute, Dundee, Scotland, UK). The second set (Table S2) consisted of Dickeya and Pectobacterium isolates from diagnostic samples that were obtained from the Diagnostic Centre for Plants of ILVO (GBBC numbers) and from diagnostic culture collections in The Netherlands (PD and IPO/PRI numbers). These isolates were mainly recovered from symptomatic potato plants and tubers and from greenhouse ornamentals. All strains and isolates were archived in cryopreservation.
The reference strains were first cultured on Difco LB agar (Miller's modification) and then subcultured on nutrient sucrose agar (NSA = Difco Nutrient Agar supplemented with 5% sucrose). Dickeya strains were verified with the Dickeya genus specific pelADE primers [8], used in colony PCR. A single colony from a 48 hr culture on NSA was suspended in 1 ml of sterile 10 mM phosphate buffer (PB) pH 7.2 and DNA was obtained by alkaline lysis [35]. After pulse centrifugation to sediment cell debris, two microliters of the supernatant were used in the PCR reactions.
The isolates from diagnostic samples were cultured on nutrient glycerol agar supplemented with manganese chloride (NGM) for production of indigoidin pigment, which is characteristic for Dickeya spp. [7]. The macerating properties of the isolates were determined on potato tubers 'Spunta' derived from minitubers. Cell suspensions with density of approximately 10 7 colony forming units per ml were prepared in sterile 10 mM PB. A conical tissue core was removed at the heel end of the tubers and 100 ml of the cell suspension was pipetted onto the cut surface. The tissue cone was reinstalled after the applied volume was absorbed and the cone was then tightened by parafilm tape. Three tubers were used for each isolate in one unreplicated assay. They were placed in moist sterilised white sand in an appropriate receptacle that was closed with a lid and aerobically incubated for 48 hours at 28uC. Macerative isolates producing the indigoidin pigment were further identified as Dickeya spp. with the pelADE PCR. Macerative isolates not producing the indigoidin pigment were further tested with the pelY PCR to identify Pectobacterium strains [36]. PCRs were performed on bacterial DNA prepared as described above for the collection strains.

Conventional fliC PCR and amplicon sequencing
Single bacterial colonies were transferred in 3 ml LB broth and grown in a shaking incubator (200 rpm) at 28uC. DNA was isolated from overnight broth cultures using the Qiagen DNeasy Blood & Tissue kit as described by the manufacturer, including the pre-treatment for gram-negative bacteria. DNA concentration and quality (according to A 260/280 and A 260/230 ratios) were assessed using a Nanodrop ND-1000 spectrophotometer. Isolated DNA was adjusted to approximately 50 ng/ml. The fliC gene fragment was amplified with PCR primers (Table 1) designed for the Dickeya dadantii 3937 strain. PCR with the fliC-1 and fliC-2 primers [37] was performed with 5 ml DNA template in 1 x Faststart High Fidelity reaction buffer (Roche Applied Science) with 2 mM MgCl 2 , 0.2 mM of each dNTP, 0.2 mM of each primer, 1 unit of FastStart Taq DNA polymerase (Roche Applied Science) and sterile molecular-grade water for a total volume of 50 ml. PCR was performed in a Bio-Rad Laboratories C1000 thermal cycler with initial denaturation at 95uC for 4 minutes, followed by 35 cycles of 95uC for 30 seconds, 55uC for 1 minute and 72uC for 45 seconds, and a terminal extension step of 7 minutes at 72uC and subsequent cooling to 12uC. PCR with the fliC-for and fliC-rev primers [38] was performed with 5 ml DNA template in 1 x OneTaq standard reaction buffer (New England Biolabs) with 2 mM MgCl2, 0.2 mM of each dNTP, 0.2 mM of each primer, 1,25 unit of OneTaq Hotstart polymerase (New England Biolabs) and sterile molecular-grade water for a total volume of 50 ml. PCR was performed in a Bio-Rad Laboratories C1000 thermal cycler with initial denaturation at 94uC for 30 seconds, followed by 30 cycles of 94uC for 30 seconds, 53uC for 1 minute and 68uC for 30 seconds, and a terminal extension step of 5 minutes at 68uC and subsequent cooling to 12uC. PCR amplicons were resolved by electrophoresis in a 1.5% agarose gel stained with ethidium bromide. PCR amplicons were extracted from gel with the Nucleospin Extract II kit (Macherey-Nä gel). DNA concentration and quality were assessed in a Nanodrop ND-1000 spectrophotometer. Purified PCR amplicons were sequenced in both directions by a commercial sequencing service (Macrogen Ltd, Korea), using the same primer set as for PCR amplification.  [39]. Sequence alignment of the trimmed sequences was done using the clustalW algorithm [40] in MEGA 5 and phylogenetic trees were generated using the neighbourjoining, maximum parsimony and maximum likelihood algorithm [41]. Distance estimation was calculated using the p-distance substitution model [42] with 1000 bootstrapping replications. Based on the sequence distances, fliC clades were differentiated by monophyletic clustering [43] with type strains and reference strains. Sequevars were designated within these clades on the basis of at least 1% sequence difference [9]. Sequevar sequences were submitted to GenBank (Table S3). The fliC sequence of Erwinia amylovora CFBP 1430 (GenBank accession AY743588) was used to root the phylogenies.

FliC TaqManH real-time PCR for 'Dickeya solani'
Primers and TaqManH MGB (59-FAM/39-BHQ1) probe (Life Technologies) specific for the fliC amplicon of 'D. solani' (Table 1) were designed with Premier Biosoft's Allele ID version 7 software. The real-time PCR was performed in a 25 ml volume in a MicroAmp Optical 96 well reaction plate with Optical Caps (Life Technologies). Briefly, 2 ml DNA template was added to 12,5 ml Taqman Gene Expression master mix 2x, 0,5 ml of primers Dsf and Dsr (15 mM), 0,5 ml of probe Dsp (10 mM) and molecular-grade water up to a final volume of 25 ml. Amplification and signal detection was done in an ABI Prism 7900HT Sequence Detection System (Life Technologies). The cycling profile is consisted of 2 minutes at 50uC for UNG-activation, 10 minutes at 95uC followed by 40 cycles of 15 seconds at 95uC and 1 minute at 63uC. The specificity of the TaqMan assay was tested by colony PCR with all reference strains and diagnostic isolates. Finally, suspensions of about 10 6 colony forming units per ml in sterile 10 mM PB were tested as described for pelADE and pelY PCR using 2 ml of target per TaqMan PCR reaction.

Molecular diagnosis of 'Dickeya solani' in symptomatic potato samples
Thirty diagnostic samples were analysed. For classical diagnosis, pathogen identification was done by isolation and further characterisation of the dominant bacterial type cultured upon plating of serial decimal dilutions of the extract from the symptomatic tissue. Therefore, minute quantities of affected tissue were aseptically removed at the margin of disease development in the stem or in the tuber and transferred in 1 ml of sterile 10 mM PB in a microvial. After vortexing of the preparation, dilution plating was performed on potato dextrose agar (Oxoid PDA) supplemented with cycloheximide. Isolated bacterial colonies representative at the higher extract dilutions were cultured on NGM and macerative properties were assessed as described before. Presumptive identification of macerative isolates was done by conventional pelADE or pelY PCR for Dickeya spp. or Pectobacterium spp. respectively. Suspensions of single colonies were prepared in 1 ml of sterile 10 mM PB and bacterial cells were subjected to alkaline lysis as explained above. D. dianthicola isolates were identified by sequencing of the PCR amplicon with the fliC-1 and fliC-2 primers and 'D. solani' isolates were identified in fliC TaqManH real-time PCR as described.
The fliC TaqManH real-time PCR was also performed directly on the potato sample extracts. The extract was allowed to settle for 15 minutes and then 100 ml was transferred in a 1.5 ml microvial and centrifuged for 10 minutes at 13000 g. The supernatant was removed and the pellet was resuspended in 100 ml 1 mM Tris-HCl pH = 8. DNA isolation was performed with the QuickPick TM Plant DNA kit (Bio-Nobile) using the Pickpen-8M magnetic tool according to the manufacterer's protocol for 100 mg starting material. The TaqMan PCR was performed using the protocol described above using 2 ml of eluted DNA as template.

Supporting Information
Table S1 Reference strains of the recognized Dickeya taxa and 'D. solani', of Pectobacterium taxa and taxa of other phytopathogenic bacteria from potato, investigated in fliC phylogeny and fliC TaqmanH real-time PCR. 1 as identified in [1], [3] or [5]   Dickeya strains and isolates. The recA attribution was obtained from [11]. The dnaX attribution was obtained from [5]. The fliC attribution is obtained in this study. (XLS)