Introduction

Pectobacterium carotovorum subsp. carotovorum (syn. Erwinia carotovorum subsp. carotovorum) and Pectobacterium atrosepticum (syn. E. carotovorum subsp. atrosepticum) are the primary enterobacteria responsible for soft rotting of potato in temperate climates. Pectobacterium carotovorum subsp. carotovorum has a broad host range, likely due to its ubiquity in natural environments (Avrova et al. 2002), whereas P. atrosepticum is found almost exclusively on potatoes, usually associated with an aerial stem disease called blackleg (Pérombelon 2002). More recently, the highly virulent enterobacteria Pectobacterium carotovorum subsp. brasiliensis (syn. Erwinia carotovorum subsp. brasiliensis) was found to cause soft rotting and blackleg on potato crops in Brazil. Genetic and biochemical analyses differentiated strains belonging to this subsp. from other Pectobacterium (Duarte et al. 2004). Pectobacterium carotovorum subsp. brasiliensis was subsequently isolated from potatoes grown in other areas of the world (Ma et al. 2007; J. van der Waals, personal communication).

Pectobacterium was previously classified in the genus Erwinia, which was divided into numerous species and subsp. on the basis of molecular, biochemical and host range differences. Subsequent taxonomic studies led to the reclassification of many Erwinias and to the elevation of several subsp. to distinct species of Pectobacterium (Gardan et al. 2003). A number of the redefined species were originally isolated from hosts other than potato including Pectobacterium betavasculorum (Thomson et al. 1981), Pectobacterium carotovorum subsp. oderiferum (Gallois et al. 1992), and P. wasabiae (Goto and Matsumoto 1987), which cause diseases of sugar beet, chicory and Japanese horseradish, respectively. Another enterobacterial species, Erwinia chrysanthemi, is responsible for aerial soft rot and wilt on potato. Taxonomic revision divided E. chrysanthemi into six genomic species belonging to a novel genus, Dickeya. As a result, strains of Dickeya infecting potato were transferred to several of the six newly designated species (Samson et al. 2005).

The diversity of enterobacteria and their different host ranges imply that multiple mechanisms may be used to infect potato and other hosts. Secreted cell wall-degrading enzymes such as pectinases, cellulases and proteases that macerate plant tissue are the primary determinants of virulence (Thomson et al. 1999). However, very little is known about other factors involved in disease development or that confer host specificity. Genome sequencing of P. atrosepticum SCRI1043 uncovered six secretion systems that are likely to contribute to virulence, including a type III secretion system (T3SS) (Bell et al. 2004). The T3SS is encoded by hrp (hypersensitive response and pathogenicity) and hrc (hypersensitive response conserved) genes, which encode a molecular syringe that injects virulence (effector) proteins into host cells (Galán and Collmer 1999).

A functional T3SS has been identified in several strains of Pectobacterium and Dickeya, and the hrp genes have been shown to be important for virulence. In P. atrosepticum, mutations in hrpN, encoding a helper protein secreted via the T3SS, as well as structural genes hrcC and hrcV involved in the formation of the T3SS result in significantly reduced virulence on potato (Holeva et al. 2004). HrpN also produces a rapid hypersensitive response (HR) in a non-host, tobacco, indicating that the hrp genes may also be involved in host specificity (Yang et al. 2002). A disease-specific effector, DspE, secreted by the T3SS, has been shown to play a role in necrogenic disease development in Erwinia amylovora (Bogdanove et al. 1998) and in soft rot pathogenesis in P. atrosepticum (Holeva et al. 2004).

Type I secretion systems, like the alpha-haemolysin secretion system consisting of the secreted adhesion protein (HecA) and the associated activator (HecB), are present in several species of phytopathogenic enterobacteria, including P. atrosepticum and Dickeya. The role of these genes in the development of disease on potato has not been confirmed, but the hec cluster in D. chrysanthemi is linked to attachment, aggregation and epidermal cell killing in Nicotiana subsp. (Rojas et al. 2002).

Identification of Pectobacterium and Dickeya infecting potato relies predominantly on biochemical and genetic differentiation of isolates (e.g. De Boer and Sasser 1986; Toth et al. 2001). However, studies investigating the diversity of enterobacteria on potatoes have found numerous atypical strains that could not be classified using these diagnostic techniques (Oliveira et al. 2003; Yahiaoui-Zaidi et al. 2003). Amplified Fragment Length Poylmorphism (AFLP) analyses and multilocus sequence typing (MLST) have also demonstrated a greater diversity of pectolytic enterobacteria infecting potato than previously thought (Avrova et al. 2002; Yap et al. 2004).

Recently, we found a variety of atypical strains whilst using a polyphasic approach to study the diversity of enterobacteria causing soft rotting of potatoes in New Zealand. A number of strains, although identified as P. carotovorum subsp. carotovorum by carbon utilisation profiling, produced no PCR product with primers specific to P. carotovorum subsp. carotovorum and clustered closely with P. wasabiae in phylogenetic analyses using the 16S rRNA gene (Pitman et al. 2008). Several studies have demonstrated that the 16S rRNA gene provides only coarse resolution of enterobacterial phytopathogens due to relatively conserved rates of mutation (Young and Park 2007). As a result, we have further characterised the atypical P. wasabiae strains isolated from potato tubers using AFLPs and phylogenetic analyses of mdh and acnA, as both methods have previously been shown to robustly differentiate strains of Pectobacterium and Dickeya.

We also investigated the presence of key virulence genes in enterobacteria that have been associated with disease development in potatoes in order to establish the possible mechanisms used by P. wasabiae to infect potato. We present new evidence confirming the identity of P. wasabiae isolated from potato tubers in New Zealand and demonstrate that these strains may use novel mechanisms for pathogenicity on potato.

Materials and methods

Bacterial strains and media

The bacterial strains used in this study were isolated from potatoes showing soft rot symptoms in New Zealand (Pitman et al. 2008) or obtained from the International Collection of Micro-organisms from Plants (ICMP), Landcare Research, New Zealand (Table 1). All bacterial strains were routinely grown on Luria-Bertani medium (LB) at 28°C for 12–24 h and kept at −80°C for storage.

Table 1 Bacterial strains used in this study
Full size table

DNA sequencing and multilocus sequence analyses

Genomic DNA was extracted from overnight broth cultures using a DNeasy tissue kit (Qiagen). For each bacterial strain, partial acnA and mdh DNA sequences (approximately 500 bp) were then amplified as they have previously been shown to be phylogenetically informative at this taxonomic level (Yap et al. 2004). PCR reactions were performed in a total volume of 25 μl consisting of 10 ng of template DNA, 0.2 U of Taq polymerase (Roche Diagnostics), 2.5 μl of 10× PCR buffer, 1.5 mM MgCl2, 200 μM of each deoxynucleoside triphosphate (dNTP), and 0.2 μmol of each primer. PCR amplifications were carried out in a GeneAmpR PCR System 9700 thermocycler (Applied Biosciences) with the following steps: (i) initial denaturation at 95°C for 5 min; (ii) 35 amplification cycles of 94°C for 30 s, 52°C for 30 s, and 72°C for 60 s; and (iii) a terminal extension phase at 72°C for 7 min. Primers used to amplify the acnA and mdh genes are listed in Table 2.

Table 2 Primers used in this study
Full size table

PCR fragments were purified with a QIAquick PCR purification kit (Qiagen), and were sequenced using an Automatic Sequencer 3730× (Macrogen Inc., Korea). DNA sequences were then assembled and edited with Sequencher 4.5 (Gene Codes), and deposited in GenBank (Table 1). Partial acnA and mdh gene sequences were aligned using ClustalX 1.83 (Thompson et al. 1997) and both ends of the alignments were trimmed to produce the following DNA sequences: acnA, 412 positions; mdh, 435 positions. Partial 16S rRNA gene sequences for each strain used in this study were obtained from previously deposited GenBank accessions (Pitman et al. 2008). For phylogenetic comparison, partial mdh DNA sequences from enterobacteria associated with soft rot of potato overseas were also retrieved from the GenBank database (Table 1).

Phylogenetic analysis of aligned acnA, mdh and 16S rRNA gene sequences was completed using PAUP* (edition 4.0b10) (Swofford 2002). Firstly, the three individual data sets were compared statistically for incongruence using the partition homogeneity test. Neighbor-joining and Maximum parsimony analyses were then performed on the individual and combined data sets incorporating 21 taxa using Kimura 2-parameter distances and the heuristic search option, respectively. Published sequences of Yersinia pestis and Y. pseudotuberculosis were used to root the resulting gene trees by the outgroup method (Ma et al. 2007). Stability of clades was assessed by 1,000 parsimony bootstrap replicates implemented in PAUP.

Phylogenetic relationships between the partial mdh gene sequences obtained from isolates on potato in this study and those previously submitted to the GenBank database were analysed using PAUP. DNA sequences, representing 29 taxa, were aligned using ClustalX 1.83 and then trimmed to 313 characters (due to differences in the lengths of GenBank submissions). Uninformative characters were excluded from the dataset before Maximum likelihood and Maximum parsimony analyses were performed using the heuristic search option. Dickeya sp. NZEC151 was selected as the outgroup for rooting of the Maximum likelihood tree.

Fluorescent Amplified Length Polymorphism (fAFLP) analysis

Fluorescent AFLP analysis was performed as previously described (Vos et al. 1995), with minor modifications. Genomic DNA (750 ng) was digested with EcoRI and MseI prior to ligation with EcoRI and MseI double-stranded adaptors in a total volume of 35 µl for 3 h at 37°C. The DNA was then amplified by PCR using the non-selective EcoRI and MseI primers, E00 and M00, respectively. Each 25 µl reaction contained 2 µl of the ligation mixture, 1 U of AmpliTaq, 200 µM of each dNTP, 1.5 mM MgCl2, and 50 ng each of the EcoRI and MseI primers. PCR reactions were performed using 35 cycles of 30 s denaturing at 94°C, 30 s annealing at 60°C, and 1 min of extension at 72°C, followed by a final extension step of 72°C for 20 min. Following a 3-fold dilution, 2 µl of the non-selective PCR amplification product was used as template for a selective PCR with primers E00 and MseI-C. For selective PCR, the 11 µl reaction mixture contained 15 ng of the E00 primer, 3.5 ng of the MseI-C primer, 1.5 mM MgCl2, 200 µM dNTPs and 1 U of Taq polymerase. Thermal cycling was performed using: (i) 12 cycles of 94°C for 30 s, 65°C for 30 s, and 72°C for 1 min, decreasing the annealing temperature by 1.4°C every 2 cycles; and (ii) 23 cycles at 94°C for 30 s, 56°C for 30 s, and 72°C for 1 min with a final extension at 72°C for 40 min. Finally, AFLP products were visualised by a PRISM 310 Genetic Analyser (Applied Biosystems) using the GeneScan 500 size standard (Applied Biosystems). All primers used in the fAFLP reactions are listed in Table 2.

AFLP profiles were sized using GeneMarker (version 1.0.5.0, SoftGenetics LLC). A binary table was then obtained by scoring amplified fragments as present (1) or absent (0) and the data extrapolated to produce a similarity matrix using the Jaccard coefficient (Jc ij ) (Jaccard 1908). Finally, the data set of AFLP coefficients was used to construct a dendrogram by unweighted pair group method with arithmetic mean (UPMGA) cluster analysis using the MultiVariate Statistical Package (Kovach Computing Services).

Identification of enterobacterial virulence factors

Initially, degenerate primers (Table 2) were designed for PCR amplification of the hrpN, dspE and hecB genes using the consensus sequences from P. atrosepticum and P. carotovorum subsp. carotovorum. PCR reactions were then performed in a total volume of 25 μl consisting of 10 ng of template DNA, 0.2 U of Taq polymerase (Roche Diagnostics), 2.5 μl of 10× PCR buffer, 1.5 mM MgCl2, 200 μM dNTPs, and 0.2 μmol of each primer. Each reaction was carried out as follows: initial denaturation at 94°C for 5 min; two cycles of 20 s denaturing at 94°C, 20 s annealing at 60°C, and 2 min of extension at 72°C; two cycles of 20 s denaturing at 94°C, 20 s annealing at 57°C, and 2 min of extension at 72°C; two cycles of 20 s denaturing at 94°C, 20 s annealing at 55°C, and 2 min of extension at 72°C; 30 cycles of 20 s denaturing at 94°C, 20 s annealing at 50°C, and 2 min of extension at 72°C. Reactions were terminated by a final extension step at 72°C for 7 min. Amplicons of 771 bp, 1,120 bp and 361 bp were expected for hrpN, dspE and hecB, respectively.

Southern hybridisations were performed using the recommended procedure of Roche Diagnostics. Firstly, the primers for hrpN, dspE and hecB were used to amplify three Digoxygenin (DIG)—labelled probes from P. atrosepticum strain ICMP1526 using the PCR DIG Probe Synthesis Kit (Roche). BglII-digested chromosomal DNA from selected enterobacteria was then separated by gel electrophoresis, transferred to Hybond nylon membranes (Amersham) and hybridised with one of three DIG-labelled probes. After hybridisation the membranes were washed at low stringency in a 2 × SSC, 0.1% SDS solution.

The partial nucleotide sequences of the hecB gene amplified from different isolates of Pectobacterium in this study were deposited in GenBank under the accession numbers GQ144361 to GQ144365.

HR assays

HR assays (four replicates) were performed using tobacco leaves (Nicotiana tabacum cv. `Xanthi’). Mid-log-phase cultures of enterobacterial strains were washed twice with sterile deionised water. The first fully expanded tobacco leaves were then infiltrated with a suspension containing 2 × 108 CFU of bacteria ml-1 and the inoculated plants incubated at room temperature for 24–48 h.

Results

Phylogenetic characterisation of P. wasabiae on potato

Partial DNA sequences were obtained for the acnA, mdh and 16S rRNA genes of Pectobacterium causing soft rot of potatoes in New Zealand as well as a selection of type strains and other well characterised isolates of soft rotting enterobacteria. Given the recorded fragility of the 16S rDNA sequence for phylogenetic classification of Pectobacterium, the congruence of the three genes was evaluated via a partition homogeneity test. The test revealed significant incongruence (P = 0.01), suggesting that it would be inappropriate to combine the datasets for the purposes of phylogenetic reconstruction. However, removal of the 16S rDNA sequence data identified significant homogeneity between acnA and mdh (P = 0.22) as previously shown by Yap et al. (2004). Based on the results of the partition homogeneity test and the accepted use of acnA and mdh in other studies of enterobacteria, sequence data from acnA and mdh were combined for the phylogenetic analyses of putative P. wasabiae strains.

Phylogenetic reconstruction using the concatenated acnA and mdh DNA sequences resolved the relationships between New Zealand isolates and other well characterised strains of soft rotting enterobacteria, dividing the Pectobacterium into four distinct and strongly supported clades (Fig. 1). All strains previously identified as P. carotovorum subsp. carotovorum clustered into a single clade (clade B) supported by a bootstrap value of 89%. Clade B was represented by type strain P. carotovorum subsp. carotovorum ICMP5702, but also included the type strain for P. carotovorum subsp. oderiferum, confirming its classification as a subsp. of P. carotovorum in other studies (Ma et al. 2007). Several isolates from potato tubers clustered closely with the type strain for P. wasabiae, ICMP9121, in Clade C, corroborating previous research that showed the presence of P. wasabiae strains infecting potato in New Zealand (Pitman et al. 2008). This relationship was supported by a 92% bootstrap value. The phylogenies also confirmed the existence of a novel group of strains, distinct, but related to the known species of phytopathogenic Pectobacterium. These strains, which previously formed a unique clade using 16S DNA sequencing, clustered within a separate clade (Clade A) using the acnA and mdh genes and are represented by the highly virulent strain NZEC211.

Fig. 1
figure 1

A Maximum parsimony phylogram showing the relationships between putative P. wasabiae isolates from New Zealand and other enterobacterial strains based on concatenated acnA and mdh DNA sequences (tree 1 of 8 equally parsimonious trees produced using heuristic searches). Numbers by nodes represent bootstrap support >50% from 1,000 replicates. T, Type strain

Full size image

The partial mdh DNA sequences from the New Zealand isolates obtained in this study were also compared to the mdh gene sequences from other closely related enterobacteria previously found on potato overseas (obtained from GenBank) using Maximum likelihood. Phylogenetic reconstruction grouped the isolates into four clades (A–D) (Fig. 2), similar to those found using the concatenated acnA and mdh DNA sequences. The New Zealand P. wasabiae strains clustered together, and with several isolates from the USA that had tentatively been described as P. wasabiae in previous studies (Ma et al. 2007). Bootstrap support for this clade (Clade C) was 81%, discriminating this group of isolates from P. atrosepticum (Clade D) and P. carotovorum subsp. carotovorum (Clade B). The P. wasabiae isolates were also distinguished from representative strains of P. carotovorum subsp. brasiliensis, which clustered together within Clade B, consistent with their status as a subsp. of P. carotovorum.

Fig. 2
figure 2

Maximum likelihood tree showing relationships of P. wasabiae collected from potatoes and closely related enterobacteria isolated from potato overseas based on comparative analysis of mdh DNA sequences. Numbers by nodes represent bootstrap support >50% from 1,000 replicates using heuristic searches. T, Type strain. *, Strains for which genome sequencing is publically available

Full size image

Cluster analyses using fAFLPs confirms the identity of P. wasabiae strains infecting potato in New Zealand

The identity of putative P. wasabiae strains causing soft rot of potatoes in New Zealand was confirmed by comparison with a selection of type strains and other well characterised strains of soft rotting Pectobacterium using fAFLP fingerprinting. The restriction enzymes EcoRI and MseI, together with the selective primers E00 and MseI-C generated 174 diverse fAFLP peaks in the set of 17 Pectobacterium strains within a range of 40–550 bp. Analysis of the resulting Jaccard coefficients identified four distinct clusters at the Jc ij level of 0.26 (Fig. 3), grouping the strains into the four clades (A–D) resolved using phylogenetic construction of the acnA and mdh DNA sequences.

Fig. 3
figure 3

A dendrogram constructed using UPGMA linkage of Jaccard’s coefficients derived from fAFLP banding patterns of Pectobacterium. A Jc ij  = 0.26 cut off value was used to divide the species of Pectobacterium into four distinct clusters. T, Type strain

Full size image

Cluster B included all the strains previously identified as P. carotovorum subsp. carotovorum using biochemical tests and 16S rDNA sequencing (Pitman et al. 2008). AFLP fingerprinting also aligned P. carotovorum subsp. oderiferum into cluster B, congruent with its status as a subsp. of P. carotovorum. Cluster C contained all the putative P. wasabiae strains from New Zealand. These strains were linked together at a Jc ij level of 0.70 and with the type strain for P. wasabiae, ICMP9121, at a Jc ij level of 0.4. According to the AFLP results, this cluster was the most distantly related of any of the clusters containing known species of Pectobacterium.

At the Jc ij level of 0.26, strains of P. atrosepticum and P. betavasculorum grouped together in cluster D, similar to the results from phylogenetic analyses of acnA and mdh DNA sequences. However, these isolates could be resolved further at a Jc ij level of 0.27 using fAFLPs, indicating that P. atrosepticum and P. betavasculorum are genetically distinct. Cluster A contained several New Zealand isolates different from all other known Pectobacterium in this study. Their alignment into a unique fAFLP cluster confirmed the results of the phylogenetic studies using the acnA and mdh genes.

Divergence of the T3SS and key virulence factors in Pectobacterium

PCR amplification of hrpN, dspE, and hecB produced DNA fragments of 771, 1,120 and 361 bp, respectively, in P. atrosepticum, P. carotovorum subsp. carotovorum and P. carotovorum subsp. odoriferum (Fig. 4). The hrpN and dspE genes were also amplified from P. betavasculorum. However, with the exception of NZEC12, all the P. wasabiae strains failed to produce a PCR product indicative of hrpN, dspE, or hecB, further differentiating the P. wasabiae strains found on potatoes from other closely related Pectobacterium. Southern blots, probed with a DNA fragment containing hrpN, dspE or hecB, confirmed that the virulence-associated genes were absent or highly divergent in P. wasabiae. As shown in Fig. 4, the hrpN, dspE, and hecB probes only hybridised to the chromosomal DNA of strains that could also be used to amplify the corresponding genes by PCR.

Fig. 4
figure 4

Detection of hrpN, dspE and hecB in various Pectobacterium using Southern hybridisation and gene-specific PCR. Lanes 1–10, ICMP1526, ICMP4226, ICMP9121, ICMP5702, ICMP9290, NZEC9, NZEC10, NZEC12, NZEC8974, and ICMP11533, respectively

Full size image

PCR and Southern hybridisation showed the presence of one of the virulence-associated genes tested in this study, hecB, in NZEC12. A partial hecB fragment of 321 bp was subsequently amplified from P. wasabiae NZEC12, P. atrosepticum ICMP1526, P. carotovorum subsp. carotovorum ICMP5702 and P. carotovorum subsp. odoriferum ICMP11533 using the degenerate primers hecB_F6 and hecB_R5 (Table 2). Cluster analysis of the resulting hecB DNA sequences showed that hecB from NZEC12 was divergent from those found in both P. atrosepticum and P. carotovorum (data not shown).

Pectobacterium wasabiae causes a mild HR response on tobacco

Infiltration of tobacco leaves with all strains of P. wasabiae used in this study resulted in the collapse of the plant cells at the point of infection, although NZEC10 and NZEC9 only produced a weak response (Fig. 5). The collapse of the plant tissue, characteristic of a HR, indicated that P. wasabiae strains infecting potato in New Zealand possess a functional T3SS and HrpN.

Fig. 5
figure 5

The HR in tobacco leaves upon inoculation with P. wasabiae strains a ICMP9121, b NZEC8974, c NZEC12, d NZEC9, and e NZEC10. Sterile deionised water was used as a negative control

Full size image

Discussion

The identity of P. wasabiae causing soft rotting of potato in New Zealand was confirmed by phylogenetic analyses of the acnA and mdh genes from a variety of phytopathogenic enterobacteria. All putative P. wasabiae isolates clustered closely together, along with the P. wasabiae type strain, ICMP9121, using a comparison of concatenated acnA and mdh DNA sequences. Phylogenetic analyses of the mdh DNA sequences by Maximum likelihood also clustered these isolates within the same strongly supported clade as strains collected from potato in the USA that were tentatively described as P. wasabiae using DNA fingerprinting and MLST (Yap et al. 2004; Ma et al. 2007). Yap et al. (2004) found isolates of P. wasabiae and other species of Pectobacterium collected from different plant hosts grouped together, suggesting that these species had a wide host range. The results of our study provide further evidence that P. wasabiae, which had previously been thought to be host-specific, has a broad host range similar to P. carotovorum.

Although the trees generated by the acnA and mdh genes were not entirely congruent with those created using partial 16S rDNA sequences, the major clades resolved in this phylogenetic study were in accordance with our previous results using the 16S rRNA gene (Pitman et al. 2008). Pectobacterium wasabiae (Clade C) was strongly differentiated from all the known Pectobacterium responsible for diseases on potato, including P. atrosepticum (Clade D), P. carotovorum subsp. carotovorum and P. carotovorum subsp. brasiliensis (Clade B). However, the relationships among several of these species varied from those observed using the 16S rRNA gene. In other studies, comparative analyses of 16S rDNA sequences from Pectobacterium have demonstrated that this gene only provides coarse resolution of phytopathogenic enterobacteria (Young and Park 2007) or that this gene may be relatively unrepresentative of the bacterial genome (Zeigler 2003). In contrast, the mdh gene has been used successfully to compare the relationships among enteric bacteria (e.g. Boyd et al. 1994). Ma et al. (2007) used mdh as well as acnA, gapA, icdA, mtlD, pgi and proA for multilocus sequence analysis of enterobacteria. As in our study using acnA and mdh, these genes produced congruent trees emphasising the importance of choosing genes representative of the bacterial genome and the benefit of a multilocus approach for classification of enterobacteria.

An alternative method used for investigating diversity is AFLP analysis—one of the most sensitive genomic techniques for distinguishing amongst bacterial strains (Vos et al. 1995). Avrova et al. (2002) successfully used AFLPs to study the genetic diversity amongst enterobacteria on potato. In this study, we adapted the AFLP procedure for use with fluorescent dye-labelled primers and an automated DNA sequencer for data capture to reduce the complexity of this technique. As observed previously by Avrova et al. (2002) using AFLPs, fAFLP distinguished all species and subspecies, the taxa clustering into four distinct groups (clusters A–D). These clusters were similar to the clades (A–D) derived by MLST. Avrova et al. (2002) showed that P. wasabiae was the most distantly related amongst Pectobacterium species using AFLPs. Strains of P. wasabiae from New Zealand were also found to group in a single cluster (Cluster C), and were more distantly related to other species than all remaining Pectobacterium. The differences in the relationships between P. wasabiae and other Pectobacterium, according to AFLP techniques and MLST, may be due to the restricted loci used in multilocus analysis compared to the genomic profile produced by AFLPs. FAFLP further differentiated the isolates of P. wasabiae, distinguishing those shown to have identical acnA and mdh genotypes (e.g. NZEC9 and NZEC8974). Thus, fAFLP fingerprinting is a powerful technique for studying the genetic variation in enterobacteria, providing a simpler alternative to standard AFLP techniques and higher resolution than MLST.

Previous studies investigating the pathogenicity of putative P. wasabiae isolates on potato have demonstrated their ability to cause soft rot of tubers as well as stem lesions and wilting (Yap et al. 2004; Glasner et al. 2008; Pitman et al. 2008). Yet, based on PCR and DNA hybridisation results, Kim et al. (2009) found that the putative P. wasabiae isolates tested in their study, including strains WPP161, WPP163, WPP168, WPP172, and SCRI 488, lacked a T3SS. Two of the T3SS-deficient strains also lacked genes adjacent to the T3SS gene cluster. The T3SS is important in delivering virulence factors in a variety of Gram-negative pathogens. It has also been shown to be present in strains of P. atrosepticum and P. carotovorum infecting potato and to be involved in the pathogenicity of phytopathogenic enterobacteria (Rantakari et al. 2001; Holeva et al. 2004; Rojas et al. 2004). PCR and DNA hybridisation were unable to detect the hrpN and dspE genes in New Zealand isolates belonging to the same clade (Clade C). Given that hrpN and dspE are indicative of a functional T3SS, their absence suggests that, like similar strains isolated in the USA, P. wasabiae from New Zealand possess a divergent T3SS or that the entire cluster encoding the T3SS has undergone excision from the genome. Thus, the T3SS might not be important for virulence in P. wasabiae; isolates belonging to this species may use other mechanisms for infection of plant hosts instead.

As another means of investigating the presence of hrpN and a functional T3SS in P. wasabiae, each strain was tested on tobacco leaves to assess whether any of the isolates could elicit a HR. In contrast to the majority of the strains screened by Kim et al. (2009), we showed that P. wasabiae isolates infecting potato in New Zealand can elicit cell collapse. However, several strains were found to produce only a weak response. These results suggest the existence of a functional T3SS in some P. wasabiae strains in this study and that they carry a highly divergent T3SS, which cannot be detected by PCR and DNA hybridisation. Certainly, WPP19, tentatively identified as a P. wasabiae strain from potato, was shown to hybridise faintly to the T3SS genes only at low stringency and to elicit a weak HR in tobacco plants (Yap et al. 2004).

Three closely related soft rot pathogens of potato P. atrosepticum SCRI1043, P. carotovorum WPP14 and P. brasiliensis 1692 vary in host range and geographical distribution. Recent comparison of their genomes found that each strain possessed a T3SS cluster, including all the genes known to be required to synthesise a functional T3SS (Glasner et al. 2008). Interestingly, only two of the three sequenced strains, P. carotovorum WPP14 and P. brasiliensis 1692, elicited a type III-dependent HR on infiltration into tobacco leaves, implying that the HR is induced by host recognition of T3SS-secreted effectors. Perhaps the weak HR observed upon infiltration of tobacco leaves with P. wasabiae is due to a unique or divergent type III effector.

The hecB gene was found to be missing or divergent in P. wasabiae, an orthologue of hecB being detected only in NZEC12. As the hecAB genes are found to flank the hrp gene cluster in P. atrosepticum and P. carotovorum (Bell et al. 2002), this result provides further evidence of rearrangement of the hrp cluster and its flanking region in P. wasabiae. Furthermore, hecB is involved in attachment, aggregation and epidermal cell killing in Dickeya, suggesting that P. wasabiae strains may be compromised in their ability to attach to the host surface. It would be interesting to compare the epiphytic fitness of P. wasabiae and other Pectobacterium to determine whether hecAB is important for colonisation of potato.

Comparative genomics of P. wasabiae strains infecting potato and other hosts would likely help to identify differences in the T3SS in this species of Pectobacterium. Genome sequencing might also identify virulence factors other than hecB that are responsible for the virulence and host-specificity of these enterobacterial pathogens.