Xanthomonas arboricola pv. fragariae: what's in a name?

Authors


E-mail: martine.maes@ilvo.vlaanderen.be

Abstract

Polyphasic analysis exposed important heterogeneity between bacterial strains catalogued as Xanthomonas arboricola pv. fragariae (Xaf) from different culture collections. Two draft whole-genome sequences revealed pathogenicity related genes of the type-three secretion system in strain LMG 19146, whereas none were found in the Xaf pathotype strain LMG 19145. Also, considerable sequence divergence was observed in the phylogenetic marker genes gyrB, rpoD, dnaK and fyuA. Further study of 16 Xaf culture-collection strains showed that co-classification is not justified. Partial 16S rRNA gene and gyrB sequencing demonstrated that 12 strains belonged to X. arboricola, but that they did not form one homogeneous group within the species. The four remaining strains were identified as Xanthomonas fragariae and Xanthomonas sp. All sequence-based identifications were confirmed by MALDI-TOF MS fingerprinting. Also, the pathogenicity genes hrcQ and avrBs2 were detected in only three of the 12 analysed X. arboricola strains. The X. arboricola and Xanthomonas sp. strains showed pectolytic activity, and upon inoculation in strawberry none of the strains reproduced the leaf blight symptoms reported for Xaf. This study demonstrates that (i) no clear criteria exist for the identification of strains as Xaf, (ii) the name Xaf is currently used for a genetically diverse assortment of strains, and (iii) the species X. arboricola holds many undetermined plant-associated bacteria besides the described pathovars.

Introduction

Xanthomonas fragariae (Kennedy & King, 1962) was long considered the only bacterial pathogen on strawberry. In 1993, a new disease called ‘bacterial leaf blight’ was observed on strawberry plants in northern Italy (Scortichini, 1996), and a new pathogen, Xanthomonas arboricola pv. fragariae (Xaf), was reported as the causal agent (Janse et al., 2001). In contrast to symptoms typical for X. fragariae, the initial leaf lesions described for Xaf were not water-soaked and the later stages of the disease were characterized by a complete yellowing and withering of the infected leaves. The description of the new pathovar was based on the characteristics of three strains: two Italian strains (LMG 19144 and the pathotype strain LMG 19145) and a French strain (LMG 19146). The latter strain was isolated from strawberry leaf spots in 1986, deposited as X. fragariae in the French Collection for Plant-associated Bacteria (CFBP), and reclassified as Xaf upon description of the new pathovar (Janse et al., 2001). Although Xaf seemed restricted to Italy at the time of its first report, renaming of several other southern European X. fragariae strains in the CFBP as Xaf suggested that the disease was more prolific than initially assessed. The observation that Xaf and X. fragariae can co-occur in strawberry tissue, as they could be co-isolated from either angular leaf spot or bacterial leaf blight symptoms (Scortichini & Rossi, 2003), increased the concern for an unrecognized spread of the new pathogen. In 2002, the European and Mediterranean Plant Protection Organization (EPPO) listed Xaf as an alert organism in recognition of this concern (Anonymous, 2002), although it was removed from this list in 2007 (EPPO, 2007). Since the first report of the bacterium, additional Xaf isolations from strawberry plants with symptoms outside Italy were reported only once (Turkey; Ustun et al., 2007). Recently, X. arboricola and X. fragariae were co-isolated from angular leaf spot symptoms during a monitoring campaign for X. fragariae in Belgian nurseries (J. Vandroemme, M. Maes, unpublished). In the attempt to determine whether the X. arboricola isolate could be designated as pathovar fragariae, it became apparent that clear classification criteria for Xaf were lacking. The original description of Xaf was primarily based on phenotypic characteristics, for example pectolytic activity on potato tubers (Janse et al., 2001). Although this feature may still be useful for differentiation from X. fragariae or other X. arboricola pathovars, it is not a unique characteristic within the species: recently, a group of X. arboricola strains associated with a new grapevine leaf spot disease also exhibited pectolytic activity on potato tubers (Sawada et al., 2011b. Furthermore, the reported pathogenicity of Xaf upon artificial inoculation on strawberry has often been ambiguous: while the original pathovar description mentioned extensive vascular discoloration and wilting leaves within 45–60 days after vein inoculations (Janse et al., 2001), another study reported that symptoms were hard to reproduce in greenhouse experiments and that there was considerable virulence diversity among tested strains (Scortichini & Rossi, 2003). The most recent report of Xaf mentioned leaf spots, not wilting, as the most apparent symptom in pathogenicity tests (Ustun et al., 2007). Also, none of the reported Xaf pathogenicity tests included any of the renamed south European CFBP strains. Reliable genotypic criteria for classification of strains in Xaf are equally lacking: two fingerprinting methods, rep-PCR (Scortichini & Rossi, 2003) and integron gene cassette arrays (Barionovi & Scortichini, 2006), were developed with only a restricted set of exclusively Italian reference strains and also did not include any of the renamed south European CFBP strains. Although a real-time PCR detection method for Xaf was developed, it was unable to distinguish Xaf from other X. arboricola pathovars (Weller et al., 2007). The problematic classification criteria for Xaf have prompted this study into the uniformity of 16 culture-collection strains catalogued as Xaf, using a polyphasic approach. The whole genome sequence was drafted for two strains, the Italian pathotype strain LMG 19145 and the French strain LMG 19146, and their pathogenic and phylogenic marker genes were compared. The phylogenic background of the complete set of 16 Xaf-catalogued strains was then studied, and they were screened for the presence of two pathogenicity genes and tested for symptom production on strawberry plants and pectolytic activity.

Materials and methods

Strains and culture conditions

Xanthomonas arboricola pv. fragariae culture-collection strains were selected for their heterogeneous isolation date and geographic origin (Table 1). Strains were received from CFBP (Beaucouzé, France), BCCM-LMG (Belgian Co-ordinated Collections of Microorganisms – Laboratory of Microbiology Ghent, Ghent, Belgium) and NCPPB (National Collection of Plant Pathogenic Bacteria, York, UK). Two Turkish isolates, S-Tr-1 and S-Tr-2 (Ustun et al., 2007), were received from the Plant Protection Research Institute (Izmir, Turkey) and then deposited as LMG 26912 and LMG 26913 in BCCM-LMG. The Belgian strain LMG 26906 was isolated together with Xfragariae from strawberry leaves exhibiting clear angular leaf spot symptoms in 2011 and was identified as X. arboricola based on 16S rRNA gene and gyrB sequencing. Xanthomonas fragariae strain LMG 25863, previously designated GBBC-Xf 920 (Vandroemme et al., 2008), was isolated from a typical angular leaf spot on strawberry and extensively characterized at the Institute for Agricultural and Fisheries Research, also by whole-genome sequencing (J. Vandroemme, B. Cottyn, M. Maes, unpublished data). This X. fragariae reference strain was included in all molecular and plant infection experiments in this study. For matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) standardized spectral acquisition, bacteria were grown on peptone yeast extract glycerol agar medium (NYGA) at 28°C for 48 h. For all other purposes, the strains were grown on Wilbrink's agar medium with nitrate (Wilbrink-N; Koike,1965) at 28°C.

Table 1. Overview of bacterial strains used in this study, together with their gyrB-sequence-based identification and results of the avrBs2- and hrcQ-targeted PCRs, strawberry leaf inoculation experiments and potato rot tests
StrainSynonymsaReceived asIsolationyearLocationgyrB IDavrBs2 PCRhrcQ PCRAngularleaf spotBacterial leaf blightPotato rot test
  1. a

    Synonymous strain numbers are provided for cross-referencing with earlier publications and GenBank entries.

  2. b

    Pathotype strain.

  3. c

    Strawberry inoculation tests repeated twice.

LMG 19144CFBP 6770, PD 2696Xanthomonas arboricola pv. fragariae1993Italy Xanthomonas arboricola cc+
LMG 19145bCFBP 6771, PD 2780X. arboricola pv. fragariae1993Italy X. arboricola cc+
LMG 19146CFBP 3548, PD 3164X. arboricola pv. fragariae1986France X. arboricola ++cc+
CFBP 3544 X. arboricola pv. fragariae1985FranceXanthomonas sp.cc+
CFBP 3549PD 3160X. arboricola pv. fragariae1986France X. arboricola +
CFBP 5253 X. arboricola pv. fragariae1998France Xanthomonas fragariae +++cc
CFBP 5261 X. arboricola pv. fragariae1999France X. fragariae +++cc
CFBP 6762PD 2694X. arboricola pv. fragariae1993Italy X. arboricola +++
CFBP 6763PD 2697X. arboricola pv. fragariae1993Italy X. arboricola +++
CFBP 6772PD 2803X. arboricola pv. fragariae1993Italy X. arboricola +
CFBP 6773PD 3145Xarboricola pv. fragariaeUnknownSpain X. arboricola cc+
NCPPB 4182PD 2803X. arboricola pv. fragariae1993Italy X. arboricola +
NCPPB 4183PD 2806X. arboricola pv. fragariae1993Italy X. arboricola +
LMG 26912S-Tr-1, GBBC-2049X. arboricola pv. fragariae2004TurkeyXanthomonas sp.cc+
LMG 26913S-Tr-2, GBBC-2050X. arboricola pv. fragariae2004Turkey X. arboricola cc+
LMG 26906R-46653, GBBC-2097 X. arboricola 2011Belgium X. arboricola cc+
LMG 25863GBBC-Xf 920 X. fragariae 2002Belgium X. fragariae +++

Draft genome analysis

The genomes of LMG 19145 and LMG 19146 were sequenced with 50-bp paired-end reads of a shotgun library on an Illumina sequencer (BaseClear). Quality trimming of the raw sequencing reads and de novo whole-genome sequence assembly were performed using clc bio v. 4.0 (CLC bio). The resulting draft genomes were annotated with the rast v. 4.0 online annotation pipeline (Aziz et al., 2008). The seed viewer v. 2.0 web interface (Overbeek et al., 2005) was used to analyse the gene content of both genomes. Homologous partial gene sequences from the pathotype strains of X. arboricola pathovars celebensis, corylina, juglandis, populi and pruni (Bull et al., 2010) were retrieved from GenBank and the sequences of LMG 19145 and LMG 19146 were trimmed in silico to match their lengths. Partial sequences extracted from the draft genome of X. fragariae strain LMG 25863 (unpublished) were used to root the cluster analysis trees. Sequences of the phylogenetic marker genes gyrB, rpoD, dnaK and fyuA (Young et al., 2008) were compared using bionumerics v. 6.6 (Applied Maths). The full coding sequences were deposited in GenBank under the accession numbers: JQ596006, JQ596007 and JQ387612 (gyrB); JQ596004, JQ596005 and JQ680972 (rpoD); JQ595996, JQ595997 and JQ680970 (dnaK); and JQ596002, JQ596003 and JQ680971 (fyuA), for LMG 19145, LMG 19146 and LMG 25863, respectively. Both Xaf genomes were also searched for the presence of type-three secretion system (TTSS) structural genes and 53 TTSS effector genes currently identified in other Xanthomonas genomes (www.xanthomonas.org; White et al., 2009). The hrp operon gene hrcQ and all TTSS effector genes retrieved from the genome sequence of LMG 19146 were deposited in GenBank under the accession numbers JQ425035 (hrcQ), JQ425034 (avrBs2) and JQ595991JQ595995 (xopF1, xopA, hrpW, hpaA, xopR, respectively).

Bacterial DNA preparation

Genomic DNA used in this study was extracted using the Gentra Puregene Cell Kit (QIAGEN), according to the manufacturer's instructions. Quantity and quality of extracted DNA were checked using the NanoDrop 1000 spectrophotometer (Thermo Fisher Scientific). All DNA extracts were stored at −20°C and freshly diluted to the desired concentration before use in the different PCR reactions.

16S rRNA gene sequencing

Partial 16S rRNA gene sequencing was used to identify the Xaf-catalogued strains to the genus level (Hauben et al., 1997). Primers 16F27 and 16R1525 were used as described by the authors for the amplification of a 1500-bp fragment of the 16S rRNA gene. Cycle sequencing reactions were performed using the amplification primer 16F27 and the sequencing primer 16R519 to generate 493-bp sequences. The sequences were used as blastn queries against the nucleotide database of GenBank to retrieve an identification by best hit. All 16S rRNA gene sequences generated in this study were deposited at GenBank under the accession numbers JQ387619JQ387635.

Gyrase B gene sequencing

Partial sequencing of the phylogenetic marker gene gyrB was used to designate the Xaf-catalogued strains to a Xanthomonas species (Parkinson et al., 2007). Primers XgyrPCR2F and Xgyrrsp1 were used as described by the authors for the amplification of a 700-bp fragment and the same primers were used in the cycle sequencing reactions. Because not all sequences were of equal length, they were trimmed in silico at positions corresponding to the 5′ end ATCACCGGCG and 3′ end GAGCAGCTGTG in X. fragariae to generate 530-bp-long sequences and blastn-queried against the nucleotide database of GenBank. Cluster analysis was performed using bionumerics software with the 530-bp trimmed sequences produced here and from GenBank. The gyrB sequences generated in this study are available at GenBank under the accession numbers JQ387602JQ387618.

MALDI-TOF MS analysis

The 16S rRNA gene and gyrB sequence-based identification of the Xaf-catalogued strains was challenged with MALDI-TOF MS. Protein mass fingerprints were obtained using an AXIMA Confidence MALDI-TOF mass spectrometer (Shimadzu-biotech), with detection in the linear, positive mode at a laser frequency of 50 Hz and within a mass range of 2–30 kDa. The acceleration voltage was 20 kV, and the extraction delay time was 200 ns (Rezzonico et al., 2010). All MALDI-TOF MS spectra used in this study were averages of at least four replicate measurements, at two different evaluation times, per strain. During each replicate measurement, the protein mass fingerprint of a sample was determined at 100 different spots with at least 10 laser shots per spot. The 100 fingerprints were averaged and processed to a raw spectrum using launchpad v. 2.8 software (Shimadzu-biotech). This software was also used for peak processing of all raw spectra with the following settings: the Advanced Scenario was chosen from the Parent Peak Cleanup menu, Peak Width was set to 80 channels, Smoothing Filter Width to 50 channels, Baseline Filter Width to 500 channels and Threshold Apex was chosen as the peak detection method. For the Threshold Apex Peak Detection, the Threshold Type was set to Dynamic, the Threshold Offset to 0·020 mV and the Threshold Response Factor to 1·2. Each target plate was externally calibrated using spectra of reference strain Escherichia coli DH5α. The processed spectra were reduced to a maximum of 100 most intensive peaks with an ad hoc PERL script (J. F. Pothier, V. Pflüger, B. Duffy, unpublished) and exported as peak lists with m/z values for each peak and signal intensity in the ASCII format. A binary matrix was then generated using the saramis (Spectral ARchive And Microbial Identification System, AnagnosTec) SuperSpectrum tool. Peak lists were trimmed to a mass range of 2–20 kDa. Peak lists were binned and average masses were calculated using the saramis SuperSpectrum tool with an error of 800 ppm. Final consensus spectra for each strain were generated with a customized VBA script (J. F. Pothier, V. Pflüger, B. Duffy, unpublished) by eliminating masses present within fewer than half of the replicate measurements. Multivariate cluster analysis using the UPGMA algorithm with Dice coefficient was performed in past v. 2.14 (Hammer et al., 2001) and the resulting dendrogram was visualized with figtree v. 1.3.1.

avrBs2 and hrcQ amplification and sequencing

All strains in this study were screened for presence of the TTSS by PCR amplification of two broadly retained genes: avrBs2, a ubiquitous TTSS effector protein in Xanthomonas (White et al., 2009) and hrcQ, one of the conserved hrp genes that encode for the core components of the TTSS apparatus (Fadouloglou et al., 2004). Primers 5′-TGCCGGTGTTGATGCACGA-3′ and 5′-TCGGTCAGCAGGCTTTC-3′ (Hajri et al., 2012) were applied for the amplification of an 848-bp fragment of avrBs2, and primers RST21 and RST22 (Leite et al., 1994) were used for the amplification of a 939-bp fragment of hrcQ. Both PCR reactions were performed as described by the authors, and the amplicons were cycle-sequenced using the respective amplification primers. All avrBs2 and hrcQ sequences generated in this study are available at GenBank under the accession numbers JQ387637JQ387642 (avrBs2) and JQ387643JQ387648 (hrcQ).

Strawberry leaf inoculation assays

To test the ability to induce symptoms in strawberry plants (bacterial leaf blight or angular leaf spot), new and fully expanded leaves of young strawberry plants (3–4 weeks old, cv. Elsanta) were inoculated with bacterial suspensions of 108 colony-forming units (CFU) mL−1 in sterile water, with water as a negative control, and a suspension of X. fragariae LMG 25863 as the angular leaf spot pathogen control. Two inoculation methods were used: for induction of angular leaf spot, the inoculum was sprayed on the lower leaf surface using a spray gun, resulting in mild water-soaking of the leaf tissue (Hildebrand et al., 2005); for the induction of bacterial leaf blight, the major vein of each of the three leaflets on a leaf was punctured with a sterile needle and a droplet of inoculum was placed on the puncture wound and allowed to dry (Janse et al., 2001; Scortichini & Rossi, 2003; Ustun et al., 2007). Each strain was tested in triplicate, on leaves of three different plants. After inoculation the plants were covered with plastic bags for the first 3 days and then kept in a growth chamber at 20/15°C (day/night) with >85% humidity and examined weekly for disease symptoms for a period of 2 months. The experiment was repeated with 10 out of the 16 Xaf-catalogued strains (Table 1).

Potato rot test

The pectolytic activity of all strains was tested by the potato soft rot test (Lelliott & Stead, 1987). Seed potato tubers (cv. Bintje) were washed, surface-sterilized by wiping with 70% ethanol and cut into 0·5-cm-thick slices with a sterilized knife. The slices were placed on a moistened filter paper in a Petri dish and inoculated by spreading 500 μL suspension (10CFU mL−1) of each strain on the potato slice surface. The inoculated slices were incubated at 28°C and tissue rotting was evaluated after 48 h by probing the slices with a sterile inoculation needle.

Results

Draft genome analysis

The genomes of LMG 19145 and LMG 19146 were assembled to draft status (Table 2). The quality of the LMG 19146 Illumina data set was slightly less than that of LMG 19145, resulting in a lower average coverage and a higher number of small contigs with low coverage (925 contigs smaller than 10 kb; 19·2 average coverage). The largest parts of both draft genomes were of comparable quality, as indicated by the similar N50 contig sizes. In the genome of the Xaf-catalogued strain LMG 19146, an entire hrp operon coding for the structural elements of the TTSS injection system was found, together with six TTSS effector genes: avrBs2, xopF1, xopA, hrpW, hpaA and xopR. None of these TTSS genes were found in the genome of the Xaf pathotype strain LMG 19145. Neighbour-joining trees were constructed for sequences of four common phylogenetic marker genes, viz. the three housekeeping genes gyrB, rpoD and dnaK, and the siderophore receptor gene fyuA (Fig. 1). The whole-genome-derived sequences of these four marker genes were confirmed by Sanger sequencing (results not shown). The consensus sequences used for the four genes were 530, 873, 940 and 695 bp long, respectively. The trees confirmed that LMG 19145 and LMG 19146 belonged to the species X. arboricola, although the sequences of both strains differed for all four genes, with the sequences of other X. arboricola pathotype strains scattered between them. The grouping of the strains was different for each gene.

Table 2. Basic metrics of the draft genome sequences of Xanthomonas strains LMG 19145 and LMG 19146
 LMG 19145LMG 19146
  1. a

    Size of the smallest of all contigs comprising at least 50% of the total genome sequence.

  2. b

    Ratio between total mapped read sequence length and total contig sequence length.

Genome size (bp)4 861 6515 092 662
Contigs ≥ 200 bp4251087
% GC65·564·4
N50 contig sizea (bp)25 94022 638
Average coverageb74×48×
Figure 1.

Neighbour-joining trees of Xanthomonas arboricola pv. fragariae strains LMG 19145 and LMG 19146 (in bold), together with other X. arboricola pathotype strains based on partial sequences of the gyrB, rpoD, dnaK and fyuA genes. Similarity distances are given as percentage values on the axis and bootstrap values calculated for 100 reiterations are shown at cluster nodes when >50. The homologous partial sequence of X. fragariae strain LMG 25863 was used to root each tree. GenBank accession numbers are given between rectangular brackets.

16S rRNA gene and gyrase B sequencing

All 16 Xaf-catalogued strains were assigned to the genus Xanthomonas based on their 16S rRNA gene sequences by comparison to GenBank references using blastn. The sequences obtained for LMG 19145 and LMG 19146 were 100% identical to their whole-genome-derived 16S rRNA sequences. Only 14 strains showed 99% internal 16S rRNA sequence similarity. The Turkish strain LMG 26912 and the French strain CFBP 3544 had identical sequences but showed only 97% sequence similarity with the other 14 strains. In GenBank, these two aberrant strains showed closer relationships (99–100%) with the X  albilineans core than with the X. campestris core within the genus.

On the basis of their gyrB sequences, four of the 16 Xaf-catalogued strains did not belong to X. arboricola. The two strains already aberrant in 16S rRNA gene sequences, LMG 26912 and CFBP 3544, showed 99% similarity with each other, less than 85% similarity with any pathotype strain of X. arboricola, and only 91% similarity with the closest Xanthomonas relative in GenBank (X. translucens). These strains formed a separate gyrB clade that could not be allocated to any known Xanthomonas species, so were named Xanthomonas sp. Another two strains, CFBP 5253 and CFBP 5261, were 100% identical to X. fragariae and showed less than 90% similarity with any pathotype strain of X. arboricola pathovars, so were re-identified as X. fragariae. The 12 strains identified as X. arboricola all shared more than 98% gyrB sequence similarity with the pathotype strains of X. arboricola pathovars celebensis, corylina, juglandis and pruni. However, they did not form one coherent infraspecific group when clustered with the X. arboricola pathotype strains (Fig. 2). The gyrB sequences of some strains were highly similar or identical to those of several X. arboricola strains isolated from a diverse range of plant hosts (NCPPB 2978 from mahogany, NCPPB 2856 and NCPPB 2864 from chrysanthemum, NCPPB 2978 from Zantedeschia, ICMP 9894 from liquidambar) and to multiple members of a polyphyletic group of strains isolated from a new bacterial spot disease described on grapevine (all MAFF numbers; Sawada et al., 2011a.

Figure 2.

Partial gyrB sequence-based neighbour-joining tree for the 12 bacterial strains identified as Xanthomonas arboricola (in bold), their best blastn hits in GenBank and the X. arboricola pathotype strains (indicated with *). Similarity distances are given as percentage values on the axis and bootstrap values calculated for 100 reiterations are shown at cluster nodes when >50. The gyrB sequence of the X. arboricola pv. populi pathotype strain ICMP 8923 was used to root the tree. GenBank accession numbers are given between rectangular brackets.

MALDI-TOF MS analysis

MALDI-TOF MS analysis revealed three distinct groups with less than 30% similarity (Fig. 3). The largest group, with an internal similarity above 50%, contained the reference spectra of all X. arboricola pathotype strains. The 12 Xaf-catalogued strains identified as X. arboricola based on their gyrB sequences all belonged to this group, but were dispersed over two subgroups with less than 60% similarity and with four X. arboricola pathotype strains scattered between them. A second and third group were clearly differentiated from the X. arboricola cluster and from each other. The second group had an internal similarity of 70% and contained LMG 26912 and CFBP 3544, for which species identification was not possible. The third group had an internal similarity above 80% and contained the two strains CFBP 5253 and CFBP 5261 that clearly grouped with the X. fragariae reference strain LMG 25863.

Figure 3.

Dendrogram derived from whole-cell MALDI-TOF MS protein mass fingerprints of the 16 bacterial strains catalogued as Xanthomonas arboricola pv. fragariae (in bold), together with X. arboricola pathotype strains (indicated with *) and LMG 25863 as the X. fragariae reference. The tree was constructed using the unweighted pair-group average (UPGMA) clustering algorithm with Dice coefficient. Similarity distances are given as percentage values on the axis and bootstrap values calculated for 1000 reiterations are shown at cluster nodes when >50.

avrBs2 and hrcQ PCRs and sequencing

The avrBs2- and hrcQ-targeted PCRs only yielded amplicons with five out of the 16 tested strains. The X. fragariae identity of strains CFBP 5253 and CFBP 5261 was again confirmed, as both produced amplicons and the sequences were 100% identical to those of the X. fragariae reference strain LMG 25863. The X. arboricola strains CFBP 6762 and CFBP 6763 also produced amplicons for both genes, with 100% sequence similarity between both strains. The fifth strain producing both amplicons was the X. arboricola strain LMG 19146, and its sequences were related but not identical to those of the two former strains (Fig. 4). Both avrBs2 and hrcQ were indeed present in the draft whole-genome sequence of LMG 19146 and were 100% identical to the amplicon sequences obtained here with Sanger sequencing. Both PCRs were negative for the unidentified Xanthomonas sp. strains LMG 26912 and CFBP 3544 and for the nine other X. arboricola strains, among which was LMG 19145, which did not have those genes in its whole-genome draft sequence.

Figure 4.

Neighbour-joining trees based on partial sequences of the type-three secretion system genes avrBs2 and hrcQ, for Xanthomonas arboricola strains LMG 19146, CFBP 6762 and CFBP 6763. Similarity distances are given as percentage values on the axis and bootstrap values calculated for 100 reiterations are shown at cluster nodes when >50. The homologous partial sequences of X. fragariae strain LMG 25863 were used to root the trees.

Pathogenicity assays

All 16 Xaf-catalogued strains were inoculated onto strawberry plants. The strains that were identified as X. fragariae (CFBP 5253, CFBP 5261) based on gyrB sequences, produced clear angular leaf spots 10–14 days after spray inoculation, identical to those obtained with the X. fragariae reference strain LMG 25863. All other strains failed to produce symptoms by the spray inoculation method. After vein inoculation, slight discoloration of the puncture wound appeared on some leaves 6 weeks after inoculation, but this was not specific and was also seen on some water- and X. fragariae-inoculated controls. Distinct leaf blight or wilting was not observed for up to 2 months after inoculation, by which time most inoculated plants, including the water-inoculated controls, had wilted yellowish leaves. The experiment was repeated with 10 of the 16 strains, and yielded similar results (Table 1).

Potato rot test

All strains identified as X. arboricola, and also the two strains tentatively named Xanthomonas sp., macerated the potato slices within 48 h, which clearly typed them as pectolytic strains. Conversely, the potato slices inoculated with strains CFBP 5253 and CFBP 5261, renamed as X. fragariae, and the X. fragariae reference strain LMG 25863 remained firm.

Discussion

A representative set of 16 culture-collection strains catalogued as X. arboricola pv. fragariae was studied. No clear-cut grouping of the strains within X. arboricola could be confirmed, casting doubt upon the existence of a distinct pathovar fragariae. Reiterating the available literature on Xaf, this conclusion is not completely unexpected. The phylogenic and pathogenic basis for allocating bacterial strains to Xaf is unclear. Vague, mainly phenotypic criteria are offered in the original pathovar description, and are mostly intended for discrimination from the strawberry pathogen X. fragariae (Janse et al., 2001). Moreover, unclear pathogenicity among Xaf-catalogued strains has been reported (Scortichini & Rossi, 2003).

In this study, the genomes of the Italian Xaf pathotype strain LMG 19145 and the French Xaf strain LMG 19146 were sequenced to draft status and compared. Cluster analysis of phylogenetic marker genes showed that these two strains represent different genomovars within X. arboricola. Moreover, the TTSS genes identified in LMG 19146 were all absent in LMG 19145, suggesting a different pathogenic capacity.

This phylogenic and pathogenic disparity was also obvious in the entire set of 16 Xaf-catalogued strains. First of all, the strains clustered in three clearly distinct species groups on the basis of gyrB sequencing and MALDI-TOF MS profiling. The results of both identification methods were in perfect agreement: 12 strains were identified as X. arboricola, two as X. fragariae and two as belonging to an undefined Xanthomonas sp. Furthermore, the 12 strains identified as X. arboricola did not form one coherent group, neither on the basis of gyrB nor MALDI-TOF MS, while MALDI-TOF MS was earlier shown to be promising for infraspecific grouping of X. arboricola pathovars (J. F. Pothier, V. Pflüger, B. Duffy, unpublished; Pothier et al., 2011). Their heterogeneity was further confirmed by additional sequence data from the phylogenetic marker genes rpoD, dnaK and fyuA (results not shown). These 12 X. arboricola strains were also screened for the two TTSS pathogenicity genes hrcQ and avrBs2, which are broadly retained in xanthomonads harbouring a TTSS (Leite et al., 1994; Fadouloglou et al., 2004; White et al., 2009; Hajri et al., 2012). In nine of the 12 strains, these two genes were not detected, suggesting the lack of a TTSS. Draft whole-genome sequencing confirmed this for the pathotype strain LMG 19145. Both hrcQ and avrBs2 were present in the three other X. arboricola strains. Two of these, CFBP 6762 and CFBP 6763, had identical sequences in those genes but differed from the third one, LMG 19146. The same separation of these three strains was also reported by Hajri et al. (2012) based on the TTSS effector gene xopA, while only limited allelic variation in TTSS genes existed in other X. arboricola pathovars. Hajri et al. (2012) noted the limited effector repertoire (restricted to the ‘core set’ of six genes avrBs2, xopF1, xopA, xopW, hpaA and xopR) in these three strains (i.e. LMG 19146, CFBP 6762 and CFBP 6763). The LMG 19146 draft whole-genome sequence confirms this result, making it the smallest known TTSS effector repertoire found in any Xanthomonas whole-genome sequence so far (www.xanthomonas.org; White et al., 2009). The hrp-type TTSS is believed to be crucial for virulence and symptom development in phytopathogenic Xanthomonas (White et al., 2009) and the associated effector repertoire is an important determinant for host specialization (Hajri et al., 2012). It has been found missing in non-pathogenic or opportunistic strains (Leite et al., 1994; Gonzalez et al., 2002), but also in the sugarcane pathogen X. albilineans, where its absence was attributed to an exceptional (i.e. xylem limited) lifestyle (Pieretti et al., 2009). Here, the lack of the TTSS in most of the 12 Xaf-catalogued strains, and a restricted effector repertoire in the others, suggest that they do not share an equal pathogenic capacity. In addition, their reported pathogenicity on strawberry could not be confirmed here, despite the use of different inoculation methods. Of course, influence of the selected strawberry cultivar or experimental conditions on this result cannot be completely ruled out. Because the pathogenicity of the Xaf-catalogued strains was compared here with that of X. fragariae, optimal conditions for X. fragariae infection were selected. However, these conditions (20°C day, 15°C night, >85% RH) seemed to agree with the ‘humid mid-autumn weather’ mentioned as conducive for Xaf pathogenicity by Scortichini & Rossi (2003). Moreover, the current study is not the first to report troublesome pathogenicity testing with Xaf-catalogued strains: Scortichini & Rossi (2003) reported irreproducible pathogenicity tests with Xaf-catalogued strains, but ascribed this to ‘unknown environmental or edaphic conditions likely required for full phytopathogenic activity’. Whatever the cause, pathogenicity testing on strawberry is not reliable for identifying these strains as pathogens and classifying them in a single pathovar. A shared characteristic among the 12 Xaf-catalogued strains was their pectolytic activity, as shown in a potato tuber assay. The fact that related true pathogens with a clear host specificity, such as X. fragariae (this study) and X. arboricola pv. pruni (Sawada et al., 2011b, do not exhibit this pectolytic activity, may indicate that the 12 Xaf-named strains are adapted to a saprophytic association with plants, including strawberry. Occasional co-isolation of such X. arboricola strains from strawberry plants with a range of symptoms, even clear angular leaf spots caused by X. fragariae (Scortichini & Rossi, 2003), is therefore not surprising. Indeed, ubiquitous plant-associated non-pathogenic or opportunistic xanthomonads, several belonging to X. arboricola, were also reported by Vauterin et al. (1996).

Serious crop losses on strawberry as a result of Xaf-catalogued strains have never been reported, justifying the removal of the name from the EPPO alert list in 2007 (EPPO, 2007). Moreover, based on the data presented here, it has to be concluded that the current classification of strains as X. arboricola pv. fragariae is not relevant. At present, the name is applied to an assortment of phylogenetically diverse strains that do not constitute one biological unit. Comparable results of Hajri et al. (2012) with strains of X. arboricola pv. populi suggest that even more pathovars within the X. arboricola species may harbour an assortment of diverse strains. A recent study also reported on a group of phylogenetically diverse, pectolytic X. arboricola strains associated with bacterial spot on grapevine, and refrained from proposing a new pathovar because of the diversity and weak natural pathogenicity of the constituent strains (Sawada et al., 2011b. Some of the polyphyletic grapevine isolates reported by Sawada et al. (2011b showed 100% gyrB sequence identity with X. arboricola strains studied here. In fact, they may all be common plant residents. The advent of the genomics era may present further thorny issues in a genus where the infraspecific pathovar classification is often based on limited plant host tests.

Acknowledgements

The authors wish to thank Dr Ustun of the Plant Protection Research Institute (Izmir, Turkey) for providing strains S-Tr-1 and S-Tr-2, and for giving approval to deposit the strains in the BCCM-LMG culture collection.

Ancillary