Erwinia Species Identification Using Matrix-Assisted Laser Desorption Ionization-Time of Flight Mass Spectrometry

Journal of Plant Pathology (2017), 99 (Special issue), 121-129 Edizioni ETS Pisa, 2017 121

ERWINIA SPECIES IDENTIFICATION USING MATRIX-ASSISTED LASER DESORPTION IONIZATION-TIME OF FLIGHT MASS SPECTROMETRY

F. Rezzonico1,2, B. Duffy1,2, T.H.M. Smits1,2 and J.F. Pothier1,2

1 Environmental Genomics and System Biology Research Group, Institute of Natural Resource Sciences, Zürich University of Applied Sciences (ZHAW), Campus Grüental, 8820 Wädenswil, Switzerland 2 Agroscope Changins-Wädenswil ACW, Plant Protection Division, Schloss, 8820 Wädenswil, Switzerland

SUMMARY INTRODUCTION

Rapid and reliable identification of plant pathogenic Erwinia (Winslow et al., 1920) is a genus of the Entero- bacteria is critical for effective implementation of phytos- bacteriaceae family that was originally created to unite anitary measures. The genus Erwinia includes a number of all Gram-negative, nonsporulating, fermentative, peritri- economically important plant pathogens such as fire blight chously flagellated plant-pathogenic bacteria (Kwon et al., agent Erwinia amylovora or Asian pear pathogen Erwinia 1997). Since its inception, the genus has undergone several pyrifoliae, together with closely related plant epiphytes of taxonomical rearrangements following refined phenotypic unknown pathogenicity or even with a potential use for characterization or DNA sequence analysis, with many biological control like Erwinia tasmaniensis or Erwinia bill- species being transferred to other genera such as Pecto- ingiae, respectively. Current laboratory methods to achieve bacterium (Hauben et al., 1998), Dickeya (Samson et al., satisfactory identification and discrimination between spe- 2005) (both genera containing phytopathogenic pectino- cies within the Erwinia genus are based on the isolation lytic bacteria causing soft rot diseases), Pantoea (Brady et on semi-selective media, serology, specific PCR and gene al., 2010), Brenneria (Hauben et al., 1998), Lonsdalea (Brady locus sequencing: these approaches are complicated and et al., 2012), or Enterobacter (Brenner et al., 1986; Dickey time-consuming, often requiring a priori assumptions over and Zumoff, 1988). In addition, a number of new species the identity of the isolates. Here we present a streamlined were recently described and approved within the genus approach based on whole-cell Matrix-Assisted Laser De- including Erwinia piriflorinigrans (López et al., 2011) and sorption Ionization–Time of Flight Mass Spectrometry Erwinia uzenensis (Matsuura et al., 2012), two novel patho- (MALDI-TOF MS) based on the AXIMA mass spectrom- gens that affect European pear trees; Erwinia typographica, eter of Shimadzu-Biotech Corp that demonstrates the po- isolated from the gut of healthy bark beetles (Skrodenytė- tential of this technology for quick species identification Arbačiauskienė et al., 2012); Erwinia gerundensis, a cos- in plant diagnostics within the genus Erwinia. mopolitan epiphyte originally isolated from pome fruit trees (Rezzonico et al., 2016); and “Candidatus Erwinia Keywords: bacterial identification, MALDI-TOF MS, dacicola”, an olive fly endosymbiont (Estes et al., 2009). As taxonomy, fire blight. of January 2017, 19 species are listed in the genus. Quick and reliable identification of plant pathogens and discrimination from closely related epiphytes is a critical requirement for disease control and implementation of the appropriate control measures (Montesinos, 2003). Several standardized procedures are available for the detection, identification and quantitation of the different Erwinia spp. in diagnostic laboratories (Bühlmann et al., 2013; Gottsberger, 2010; Kim and Jock, 2001; López et al., 2004; Pirc et al., 2009). Most of these methods habitually target one single species and thus require some a priori assump- tion on the identity of the organism under investigation. Yet, not all Erwinia spp. are covered by these protocols, which are as well obviously unsuited for the discovery of new taxa. While sequencing of 16S rRNA does not offer resolution at species level, single housekeeping genes have previously been employed with variable success to classify Corresponding author: F. Rezzonico E-mail: [email protected] * Supplementary materials can be downloaded at: http://bit.ly/2AmJB3f 122 MALDI-TOF identification of Erwinia spp. Journal of Plant Pathology (2017), 99 (Special issue), 121-129 the different species within Erwinia, although relation- a validated and comprehensive database of reference pep- ships between the different species are often inconsistent tide-mass fingerprints and commercial databases are still when comparing phylogenetic trees obtained with differ- currently heavily skewed towards clinical taxa and rarely ent target genes (Palmer et al., 2017). To overcome this include environmental bacterial species relevant to plant, problem, laborious and time-consuming methodologies animal and public health (Rezzonico et al., 2010). A proof such as multi-locus sequence analysis (MLSA) are nec- of principle that MALDI-TOF MS can be employed for essary. Using the latter approach, Erwinia constitutes a the identification of species within the genus Erwinia was monophyletic clade mostly composed by phytopathogenic, provided on the Bruker Daltonics platform (Sauer et al., non-pectinolytic bacteria that cause wilts and dry necro- 2008), yet only few species were considered and several of ses, but also includes non-pathogenic organisms, such as them were represented only by their type isolate, thus not epiphytes and insect symbionts (Rezzonico et al., 2016). embracing the full taxonomic diversity within each single In the last two decades, whole cell Matrix-Assisted species (Sauer et al., 2008; Wensing et al., 2012). Further- Laser Desorption Ionization-Time of Flight Mass Spec- more, since the two databases are not compatible, the data trometry (MALDI-TOF MS) has emerged as a reliable tool obtained through the Bruker system cannot directly be for rapid microbial identification in clinical diagnostics imported and used for Erwinia species identification using (Dierig et al., 2015; Patel, 2013). Identification is based on the Shimadzu platform. unique mass/charge ratio (m/z) fingerprints of proteins, The aim of this work was to expand the existing com- which are ionized using short laser pulses directed to sub- mercial SARAMIS™ database through new independent colony amounts of whole (“intact”) bacterial cells embed- measurements to include the maximum available number ded in a matrix. After desorption, ions are accelerated in of Erwinia species and to compare, at different taxonomic vacuum by a strong electric potential and separated on levels, the clustering of the strains generated by the analy- the basis of their time of flight to the detector, which is sis of MALDI-TOF MS spectra to the results obtained proportional to the mass-to-charge ratio. The main benefit using MLSA, the current gold standard for phylogenetic of this technique with respect to conventional molecular analysis. biological tools resides in the straightforward preparation of the samples, which can commonly be completed in less than a minute directly from a single bacterial colony, and MATERIALS AND METHODS in the fact that it does not require any a priori knowledge of the organism to be investigated. This technique has Bacterial strains and biological sample preparation proven itself reliable across a broad range of conditions, variables. A total of 109 strains representing 15 available being relatively unaffected by factors such as the cell- Erwinia species (Rezzonico et al., 2010, 2009), plus six so- growth phase or the composition of the culture medium, called “Pinking” isolates (Born et al., 2016; Stockwell et al., thus displaying limited variability in mass-peak signatures 2008) were selected for clustering analysis and generation within the typically designated mass range between 2,000 of MALDI-TOF MS SuperSpectra. Among the Erwinia and 20,000 Daltons (Lay, 2001; Maier et al., 2006; Rezzo- strains, 35 belonged to the species of fire blight causative nico et al., 2010). These features make whole cell MALDI- agent Erwinia amylovora, which was studied more in TOF MS an easy, rapid and cost-effective technique, which detail. To allow comparison with previously published is extraordinarily suited for high-throughput routine anal- genotypic data, we chose both Rubus- and Spiraeoideae- ysis (Seng et al., 2009). infecting strains taking care to select the latter within all A small number of MALDI-TOF MS platforms cur- major CRISPR genotypes (Mann et al., 2012; Rezzonico rently share the commercial market, the two most com- et al., 2011, 2012) in order to cover the maximum avail- mon being those from Bruker Daltonics, which includes a able diversity. Nine strains of species belonging to related mass spectrometer along with the Biotyper software and genera, some of which were previously included in the ge- database, and the Shimadzu Axima mass spectrometer nus Erwinia, were selected as outgroups. Reference strains with the Launchpad software and the SARAMIS™ (Spec- were mainly obtained from commercial culture collections tral ARchive And Microbial Identification System, Anag- (Supplementary Table S1). For standardized spectral acqui- nosTec GmbH) database created by AnagnosTec GmbH. sition and generation of SuperSpectra™ in SARAMIS™, The latter system was recently acquired by bioMérieux bacteria were routinely streaked on LB agar and grown at and is currently being redeveloped under the commercial 28°C for 16 to 24 h. name VITEK® MS. Both platforms have demonstrated an excellent accuracy in the identification of the isolates MALDI-TOF MS spectrum acquisition. Intact cells routinely scrutinized in medical diagnostic laboratories, from individual colonies were scraped from agar plates us- with a percentage of high-confidence identifications of ing a disposable loop and smeared onto a target spot of a about 94% for the Bruker system compared to about 89% polished ground steel 48-position MALDI sample target for Shimadzu (Cherkaoui et al., 2010). However, reliable (Industrietechnik mab AG, Basel, Switzerland). For each identification ultimately depends upon comparison with strain, four distinct single bacterial colonies were spotted in Journal of Plant Pathology (2017), 99 (Special issue), 121-129 Rezzonico et al. 123 order to compare the reproducibility of masses within repli- specific biomarkers. SuperSpectrum generation was based cates. Cells were overlaid with 1 μl of a saturated solution of on recovered mass signal markers with an absolute inten- sinapinic acid (Sigma-Aldrich, Buchs, Switzerland) in 60% sity of at least 200 mV included in the 2,000- to 20,000-Da acetonitrile (Sigma-Aldrich, Buchs, Switzerland) – 0.3% tri- mass range. To create the SuperSpectrum of a species, the fluoroacetic acid (Sigma-Aldrich, Buchs, Switzerland) and protein mass fingerprints of the strains clustering with the air-dried for a few minutes at room temperature. Protein respective type strain in the MLSA tree were used. Using mass fingerprints were obtained using an AXIMA Confi- the SARAMIS™ SuperSpectrum tool, a subset of protein dence Linear/Reflectron MALDI-TOF Mass Spectrometer masses found in at least 90% of the strains of one species (Shimadzu-Biotech Corp., Kyoto, Japan), with detection in were selected and tested for their discriminatory power by the linear, positive mode at a laser frequency of 50 Hz and comparing them to all of the database entries. Dependent within a mass range of 2,000 to 20,000 Da. Acceleration on the amount of remaining species-identifying marker voltage was 20 kV and the extraction delay time was 200 ns. masses, each was given a numeric value in order to get a A minimum of 20 laser shots per sample was used to gener- maximum total number of points not higher than 1,250. ate each ion spectrum. For each bacterial sample, a total of A set of 20 to 40 marker masses is normally sufficient to 50 protein mass fingerprints were averaged and processed obtain a specific identification to the species level. The using the Launchpad version 2.8 software (Shimadzu- identification results obtained using MALDI-TOF MS Biotech Corp.). For peak acquisition, the average smooth- were compared to those obtained using DNA sequencing ing method was chosen, with a smoothing filtering width combined with a BLAST similarity search. of 50 channels. Peak detection was performed with the threshold-apex peak detection method using the adaptive Multilocus sequence analysis of housekeeping genes. voltage threshold which roughly follows the signal noise Partial sequences of housekeeping genes gyrB (742 bp), level, and subtraction of the baseline was set with a baseline rpoB (637 bp), atpD (642 bp), and infB (615 bp) were se- subtraction filter width of 500 channels. For each sample, lected for multilocus sequence analysis in a subset of iso- a list of the significant spectrum peaks was generated that lates composed by at least one strain per species. Wherever included the m/z values of the peaks, their mass deviations, available, sequences were retrieved from NCBI and, if in- and signal intensity. Every target plate was calibrated indi- cluded in our study, the type strain of the species was se- vidually at the beginning of each plate acquisition using lected. For novel strains and in case of missing data, ampli- spectra of the reference strain Escherichia coli K-12 DSMZ fication and sequencing of the aforementioned genes were 1576 (GM48 genotype) and a second measurement was per- performed as described previously (Brady et al., 2008) by formed at the end of the run as a control. means of the HotStarTaq master mix kit (Qiagen, Basel, Switzerland) and the ABI PRISM BigDye Terminators ver- MALDI-TOF MS spectrum analysis and SuperSpec- sion 1.1 cycle sequencing kit (Applied Biosystems, Foster trum generation. Generated protein mass fingerprints City, CA), respectively. All 16S rRNA gene sequences were were imported in SARAMIS™ and analyzed using the obtained from GenBank. DNA sequences were aligned following presetting parameters: mass range, from 2,000 with ClustalW (Thompson et al., 1994). Sites presenting to 20,000 Da; allowed mass deviation, 800 ppm. For den- alignment gaps were excluded from analysis. The phyloge- drogram generation, the complete list of detected masses netic trees were generated on the basis of the concatenated was exported in Microsoft Excel as comma-separated val- sequences of the four housekeeping genes fragments or ues (csv) file. An Excel macro (J. Pothier, personal com- over 1162 positions of the 16S rRNA gene. The Molecular munication) was executed to convert the list of masses into Evolutionary Genetics Analysis (MEGA) program, version a binary matrix indicating the presence/absence of each 7 (Kumar et al., 2016), was used to calculate evolutionary individual mass signal. For each isolate, only the 150 stron- distances and to infer a tree based on the neighbor-joining gest mass signals present in at least 50% of the replicate (NJ) or minimum evolution (ME) method using the maxi- spectra were retained. The resulting file was implemented mum composite likelihood (MCL) model. Nodal robust- for cluster analysis in the PAST (Paleontological Statis- ness of the inferred tree was assessed by 1,000 bootstrap tics, http://folk.uio.no/ohammer/past/) software package replicates. GenBank accession numbers for sequences used version 2.11 (Hammer et al., 2001) using the PG (paired- in this work are listed in Supplementary Table S2. group) algorithm with 500 bootstrap replicates and visual- ized by means of the phylogenetic trees graphical viewer FigTree version 1.3.1 (http://tree.bio.ed.ac.uk/software/fig- RESULTS AND DISCUSSION tree/. Clustering of the different isolates in the tree based on MALDI-TOF MS data was compared to that obtained Confirmation of all Erwinia spp. The identity of all in the neighbor-joining phylogenetic tree constructed us- strains analyzed in this work was verified both using ing MLSA sequences. 16S rRNA gene sequences as well as at least one of the As reference spectra for a rapid identification, SARA- MLSA genes. Phylogenetic analysis confirms that the MIS™ uses so-called “SuperSpectra” consisting of taxon- genus Erwinia is not monophyletic with respect to this 124 MALDI-TOF identification of Erwinia spp. Journal of Plant Pathology (2017), 99 (Special issue), 121-129

Fig. 1. Taxonomy of representative strains of the Erwinia species considered in this study based on 16S rRNA sequences (A) and the concatenated sequences of housekeeping genes gyrB-rpoB-atpD-infB (B) The trees were constructed with the Neighbor-Join- ing method and nodal supports were assessed by 1000-bootstrap replicates. Only bootstrap values greater than 50% are shown. The scale bar represents the number of base substitutions per site. Wherever possible sequences were obtained from the NCBI databases (http://www.ncbi.nlm.nih.gov), else the corresponding gene were sequenced according a protocol described previously for Enterobacteriaceae (Brady et al., 2008). gene (Rezzonico et al., 2016) with its species that can be the respective type strain when using the MALDI-TOF assigned to several clusters based on the 16S rRNA gene data to build the cladogram, thus demonstrating the ef- sequences (Fig. 1A): Cluster I represents the pathoadapted ficacy of this approach for rapid identification at species group of Erwinia (Kamber et al., 2012; Smits et al., 2013) level (Fig. 2). In terms of species grouping, an outcome and includes the type species of the genus, E. amylovora, similar to that obtained using MLSA was reached when together with E. pyrifoliae, E. piriflorinigrans and E. tas- cluster analysis was performed on the basis of protein maniensis; Cluster II comprises E. rhapontici, E. persicina mass fingerprint patterns derived from MALDI-TOF MS and E. aphidicola; and Cluster III encompasses E. papayae, measurements (Fig. 2). Although the topology between the E. mallotivora, E. psidii and E. tracheiphila. E. oleae and different Erwinia species within the single clusters was not E. gerundensis are joined in a fourth cluster that contains identical to the one obtained with the analysis of the con- also outgroup P. agglomerans, whereas E. billingiae and E. catenated housekeeping genes, the branching among the toletana formed single deep-branching taxa not directly different clusters was similar to the one obtained using related to any of the above clusters (Rezzonico et al., 2010, the MLSA approach, with Clusters I-III of Erwinia species 2009) (Fig. 1A). The presence of Clusters I-III, but of not being conserved in both trees. Both MALDI-TOF MS and of the fourth cluster containing P. agglomerans LMG the MLSA trees were concordant in locating E. billingiae 1286T, could be confirmed by MLSA using concatenated next to Cluster I+II as well as E. oleae and E. toletana next sequences of the housekeeping genes gyrB, rpoB, atpD and to Cluster III on individual branches within the genus. On infB, which allowed to place all investigated Erwinia spe- the other hand, E. gerundensis showed a somewhat diver- cies within a single monophyletic group that is distinct gent peak pattern, thereby clustering outside the genus, from the other Enterobacteriaceae genera evaluated as which is thus not monophyletic in the MALDI-TOF MS outgroups (Fig. 1B), thereby confirming previous findings tree (Fig. 2). Even if a direct comparison of the respec- (Palmer et al., 2017; Rezzonico et al., 2016). The periph- tive MALDI-TOF MS datasets is not possible due to the eral position of P. agglomerans with respect to the Erwinia different chemical matrices used in the two studies, it is group is an artefact due to the imbalance in the number interesting to notice that the protein mass fingerprints of of taxa between the two genera. E. gerundensis showed quite divergent patterns from both MALDI-TOF MS analysis was able to discriminate those of Erwinia and Pantoea placing this new taxon at a strains within Erwinia and to segregate them into sepa- distance from both genera in the individual trees based on rate species with the same level of accuracy as MLSA. MALDI-TOF MS data (Fig. 2 and Rezzonico et al., 2010), Strains belonging to the same species examined in this respectively. This, together with the marginal position of work (Supplemental Table S1) clustered together and with the species in the genus Erwinia according to both MLSA Journal of Plant Pathology (2017), 99 (Special issue), 121-129 Rezzonico et al. 125 and phylogenomic analysis (Rezzonico et al., 2010), suggests that E. gerundensis may be a member of a novel genus situated between Erwinia and Pantoea. All three analyses placed strains Cu 3298, Cu 3299T and Cu 3300, received and described in the literature as ‘Erwinia lupinicola’ (Büh- lmann et al., 2013; Poza Carrion et al., 2008), into an individual branch within the former clade IV of necrogenic Brenneria species, thus confirming the results previously obtained on the basis of glyceraldehyde-3-phosphate de- hydrogenase gene sequences (Brown et al., 2000). These pathogenic strains, causing the drippy pod of white lupine, were later suggested to belong to a pathovar of Brenneria quercina (i.e., B. quercina pv. lupinicola) (Lu and Gross, 2010), a species now reclassified into the novel genus Lon- sdalea (Brady et al., 2012).

Two distinct pathotypes within E. amylovora. A proof- of-principle of the ability of MALDI-TOF MS analysis to allow discrimination at sub-species level was provided by an in-depth investigation performed on 35 isolates belonging to the causative agent of fire blight E. amylovora, a destructive disease that affects rosaceous plants worldwide (Bonn and van der Zwet, 2000). Although E. amylovora is mainly recognized as a pathogen of Malus and Pyrus spp., the disease has been described for other taxa of the Spiraeoideae subfamily as well as for members of the Ro- soideae subfamily belonging to the genus Rubus, such as raspberry or blackberry (Rezzonico et al., 2012). Cross- infectivity between Spiraeoideae- and Rubus-infecting isolates of E. amylovora is limited and the two pathotypes can be distinguished molecularly in several ways, e.g., by characterization of their lipopolysaccharide biosynthesis genes (Rezzonico et al., 2012), of their CRISPR repeat regions (Rezzonico et al., 2011), or by molecular fingerprinting techniques such as rep-PCR and PCR-ribotyping (McManus and Jones, 1995). Fig. 2. Dendrogram of the strains belonging to the genus Er- A total of 26 Spiraeoideae-infecting and 9 Rubus-infect- winia based on MALDI-TOF MS protein mass fingerprints. The tree was constructed using the PG algorithm with 500 ing isolates were analyzed in this study, whereby the clus- bootstrap replicates. All the branches containing strains be- tering obtained using MALDI-TOF MS closely matched longing to the same species are collapsed, whereby their num- the one previously achieved using rpoB gene sequences ber is indicated between parentheses. Based on the obtained (Rezzonico et al., 2012), with a single clade containing all mass fingerprints within each cluster SuperSpectra for each Spiraeoideae-infecting isolates (S) and Rubus-infecting species were created. Species E. amylovora and E. rhapontici strains divided into three separate branches, one of which are further subdivided into two clusters, demonstrating the (R2) clustering basal to the Spiraeoideae clade (Fig. 3). De- ability of MALDI-TOF MS to allow discrimination at subspecies level (see Fig. 3 and Fig. 4) spite the ability of MALDI-TOF MS to distinguish E. amylovora pathotypes, it was not always possible to recognize single strains for this species as demonstrated for P. agglo- to scatter along the SARAMIS tree, albeit remaining merans, where repeated independent measurements of the within the same pathotype, suggests that the measurable same isolate typically clustered together in the tree gener- diversity within this species is smaller than the noise that ated by SARAMIS (Rezzonico et al., 2010). The current is implicit in each MALDI-TOF MS measurement, thus distribution pattern of E. amylovora diversity is believed to placing strain typing for E. amylovora out of reach of our have originated from two evolutionary bottleneck events, current protocol. resulting in a genetically highly homogeneous population in Spiraeoideae-infecting strains, especially those isolated Species determination of isolates displaying the in Europe (Rezzonico et al., 2011). The fact that replicate “pinking” phenotype. Iron (II) chelator proferrorosa- measurements of the same E. amylovora isolate were found mine A (L-2-(2-pyridyl)-1-pyrroline-5-carboxylic acid) is 126 MALDI-TOF identification of Erwinia spp. Journal of Plant Pathology (2017), 99 (Special issue), 121-129

Fig. 3. Intraspecific clustering of the isolates belonging to E. amylovora based on MALDI-TOF MS protein mass fingerprints (A) and comparison with clustering formerly obtained on the basis of the rpoB gene (B) (Rezzonico et al., 2012). All Spiraeoideae- infecting strains are regrouped within a single clade (S), whereas the Rubus-infecting strains can be divided into three groups: a major cluster (R1) encompassing all Canadian and most of the US isolates, plus two single-strain groups: the first (R2) is represented by strain ATCC BAA-2158 (IL-5), previously found to be closely related to Spiraeoideae-infecting strains along with isolates PD103 and Ea515 (which were not included in this work), whereas the second single-strain group (R3) is constituted by strain MR-1, which shows the largest divergence in both trees. a metabolite produced by E. persicina and E. rhapontici four Pinking group II isolates clustered together with E. that causes pink colony coloration in culture media con- rhapontici strains LMG 2645 and LMG 2648 (Fig. 4). This taining iron (Born et al., 2016; Feistner et al., 1997; Stock- subspecies division within E. rhapontici is not apparent in well et al., 2008). Both E. persicina and E. rhapontici were studies based on 16S rRNA gene sequences (Kwon et al., reported as regular co-isolates during fire blight surveys 1997; Thapa et al., 2012), but confirms previous indica- and uptake of proferrorosamine A by colonies of E. amy- tions obtained on the basis of MLSA data (Brady et al., lovora can lead to false identifications of the latter spe- 2012). cies in the diagnostics (Stockwell et al., 2008). We tested here a number of “pinking” isolates routinely identified on King’s B agar from plant samples during fire blight CONCLUSIONS surveys performed by Agroscope (Wädenswil, Switzer- land) and used MALDI-TOF MS to assess whether they This study demonstrates the accurateness and the ease belong to E. persicina or E. rhapontici. The results were of MALDI-TOF MS for the identification of isolates be- then compared with those obtained using gyrB and atpD longing to the genus Erwinia. Identification at species level sequencing-based identification. Both sequence as well was achieved on the basis of the protein profiles generated as MALDI-TOF analysis revealed that, unlike the result with minimal labor, time and materials directly from colo- obtained in a previous study with North American iso- nies grown on agar. This offers an attractive alternative lates (Stockwell et al., 2008), none of the six Swiss “pink- to the relatively high investment required for single-locus ing” isolates tested belonged to E. persicina. Two isolates validation, PCR amplification, and sequencing. The invest- (ACW 44286 and ACW 41072) were found to cluster with ment in a MALDI-TOF mass spectrometer is comparable E. rhapontici type strain NCPPB 1578T (Pinking group I), to the one needed for a 16-capillary DNA sequencing ma- whereas the other four (ACW 44214, ACW 43997, ACW chine, but it requires a fraction of the operating costs and 41558 and ACW 40945) formed a discrete group (Pinking consumables. In some cases, such as for fire blight caus- group II) not associated to any other E. rhapontici strain ative agent E. amylovora or proferrorosamine-producing included in this study (Fig. 2). However, the same segrega- bacterium E. rhapontici, it is even possible to discern the tion of the “pinking” isolates into two groups within E. individual pathotypes or the intraspecific groups, respec- rhapontici was found also when analyzing the trees based tively, that compose the species and that are not evident on the gyrB and atpD gene sequences, whereby the latter based on 16S rRNA gene sequencing. Journal of Plant Pathology (2017), 99 (Special issue), 121-129 Rezzonico et al. 127

Fig. 4. Clustering of the isolates displaying the “pinking” phenotype (*) within the species E. rhapontici and E. persicina on the basis of gyrB, atpD gene sequences and MALDI-TOF MS. All tested “pinking” isolates-were found to belong to E. rhapontici. The gyrB and atpD trees were constructed with the Neighbor-Joining method and nodal supports were assessed by 1000-bootstrap replicates. Only bootstrap values greater than 50% are shown. The scale bar represents the number of base substitutions per site. The MALDI-TOF tree was exported from SARAMIS, the scale bar represents the percentage of divergence between the spectra. E. aphidicola was used as outgroup to root the trees.

MALDI-TOF MS offers thus an attractive alternative (BLW/FOAG Project ACHILLES) as part of the Agro- for identification of Erwinia isolates at species or subspe- scope Research Programme ProfiCrops, the Swiss Federal cies level compared to the relatively high workload re- Office for Professional Education and Technology, Inno- quired for multi-locus PCR and sequencing. vation and Promotion Agency (CTI 11225.2 PFLS-LS), the European Union FP-7 EUPHRESCO ERA-Net pilot project PhytFire, and the Life Science and Facility Man- ACKNOWLEDGEMENTS agement Department at ZHAW. The work was conducted within the European Science Foundation funded research The authors are indebted to S.V. Beer (Cornell Univer- networks COST Action 864 and Action 873. sity, USA), T. Dreo (NIB, Slovenia), K. Geider (JKI, Ger- many), M.M. López (IVIA, Spain), R. Mann (La Trobe University, Australia), E. Montesinos (University of Gi- REFERENCES rona, Spain), Chiaraluce Moretti (Università di Perugia, Italy), Bea Schoch (Agroscope, Switzerland), V. Stockwell Bonn W.G., van der Zwet T., 2000. Distribution and economic (OSU, USA) and G. Sundin (MSU, USA) for providing importance of fire blight. In: Vanneste J.L. (ed.). Fire blight: bacterial strains. We thank V. Pflüger (MABRITEC AG, the Disease and its Causative Agent, Erwinia amylovora, pp. Riehen, Switzerland) for technical support with MALDI- 37-53. CAB International, Wallingford, UK. TOF MS analysis. Financial support was provided by the Born Y., Remus-Emsermann M., Bieri M., Kamber T., Piel J., 2 Swiss Federal Office of Agriculture ACHILLES project Pelludat C., 2016. The Fe + chelator Proferrorosamine A: a 128 MALDI-TOF identification of Erwinia spp. Journal of Plant Pathology (2017), 99 (Special issue), 121-129

gene cluster of Erwinia rhapontici P45 involved in its synthe- of flight mass spectrometry (MALDI-TOF MS). The Pediat- sis and its impact on growth of Erwinia amylovora ric Infectious Disease Journal 34: 97-99. CFBP1430. Microbiology 162: 236-245. Estes A.M., Hearn D.J., Bronstein J.L., Pierson E.A., 2009. The Brady C., Cleenwerck I., Venter S.N., Vancanneyt M., Swings olive fly endosymbiont, Candidatus“ Erwinia dacicola”, J., Coutinho T.A., 2008. Phylogeny and identification of Pan- switches from an intracellular existence to an extracellular toea species associated with plants, humans and the natural existence during host insect development. Applied and En- environment based on multilocus sequence analysis (MLSA). vironmental Microbiology 75: 7097-7106. Systematic and Applied Microbiology 31: 447-460. Feistner G.J., Mavridis A., Rudolph K., 1997. Proferrorosamines Brady C. L., Cleenwerck I., Venter S.N., Engelbeen K., de Vos and phytopathogenicity in Erwinia spp. BioMetals 10: 1-10. P., Coutinho T.A., 2010. Emended description of the genus Gottsberger R.A., 2010. Development and evaluation of a real- Pantoea and description of four novel species from human time PCR assay targeting chromosomal DNA of Erwinia clinical samples, Pantoea septica sp. nov., Pantoea eucrina sp. amylovora. Letters in Applied Microbiology 51: 285-292. nov., Pantoea brenneri sp. nov. and Pantoea conspicua sp. Hammer Ø., Harper D.A.T., Ryan P.D., 2001. PAST: Paleonto- nov., and transfer of Pectobacterium cypripedii (Hori 1911) logical statistics software package for education and data Brenner et al., 1973 emend. Hauben et al., 1998 to the genus analysis. Palaeontologia Electronica 4: 9. Pantoea emend. as Pantoea cypripedii comb. nov. Interna- Hauben L., Moore E.R.B., Vauterin L., Steenackers M., Mer- tional Journal of Systematic and Evolutionary Microbiology gaert J., Verdonck L., Swings J., 1998. Phylogenetic position : 2430-2440. 60 of phytopathogens within the Enterobacteriaceae. Systematic Brady C.L., Cleenwerck I., Denman S., Venter S.N., Rodríguez- and Applied Microbiology 21: 384-397. Palenzuela P., Coutinho T.A., De Vos P., 2012. Proposal to Kamber T., Smits T.H.M., Rezzonico F., Duffy B., 2012. Ge- reclassify Brenneria quercina (Hildebrand and Schroth, 1967) nomics and current genetic understanding of Erwinia amy- Hauben et al. 1999 into a new genus, Lonsdalea gen. nov., as lovora and the fire blight antagonist Pantoea vagans. Trees - Lonsdalea quercina comb. nov., descriptions of Lonsdalea quer- Structure and Function 26: 227-238. cina subsp. quercina comb. nov., Lonsdalea quercina subsp. iberica subsp. nov. and Lonsdalea quercina subsp. britannica Kim S., Jock S., 2001. Molecular detection and differentiation subsp. nov., emendation of the description of the genus of Erwinia pyrifoliae and host range analysis of the Asian pear pathogen. Plant Disease : 1183-1188. Brenneria, reclassification of Dickeya dieffenbachiae as Dickeya 85 dadantii subsp. dieffenbachiae comb. nov., and emendation of Kumar S., Stecher G., Tamura K., 2016. MEGA7: Molecular the description of Dickeya dadantii. International Journal of Evolutionary Genetics Analysis version 7.0 for bigger datas- Systematic and Evolutionary Microbiology 62: 1592-1602. ets. Molecular Biology and Evolution 33: 1870-1874. Brenner D.J., McWhorter A.C., Kai A., Steigerwalt A.G., Farm- Kwon S.-W., Go S.-J., Kang H.-W., Ryu J.-C., Jo J.-K., 1997. Phy- er III J.J., 1986. Enterobacter asburiae sp. nov., a new species logenetic analysis of Erwinia species based on 16S rRNA found in clinical specimens, and reassignment of Erwinia gene sequences. International Journal of Systematic Bacteriol- dissolvens and Erwinia nimipressuralis to the genus Entero- ogy 47: 1061-1067. bacter as Enterobacter dissolvens comb. nov. and Enterobacter Lay J.O., 2001. MALDI-TOF mass spectrometry of bacteria. nimipressuralis comb. nov. Journal of Clinical Microbiology Mass Spectrometry Reviews 20: 172-194. 23: 1114-1120. López M.M., Roselló M., LLop P., Ferrer S., Christen R., Gar- Brown E.W., Davis R.M., Gouk C., van der Zwet T., 2000. Phy- dan L., 2011. Erwinia piriflorinigrans sp. nov., a novel patho- logenetic relationships of necrogenic Erwinia and Brenneria gen that causes necrosis of pear blossoms. International Jour- species as revealed by glyceraldehyde-3-phosphate dehydro- nal of Systematic and Evolutionary Microbiology 61: 561-567. genase gene sequences. International Journal of Systematic López M.M., Keck M., Llop P., Gorris M.T., Peñalver J., Donat and Evolutionary Microbiology 50: 2057-2068. V., Cambra M., 2004. Diagnostic protocols for regulated Bühlmann A., Pothier J.F., Rezzonico F., Smits T.H.M., An- pests: Erwinia amylovora. OEPP/EPPO Bulletin, EPPO Stan- dreou M., Boonham N., Duffy B., Frey, J.E., 2013. Erwinia dards 34: 159-171. amylovora loop-mediated isothermal amplification (LAMP) Lu S.E., Gross D.C., 2010. Drippy pod of white lupine: a new assay for rapid pathogen detection and on-site diagnosis of bacterial disease caused by a pathovar of Brenneria quercina. fire blight. Journal of Microbiological Methods 92: 332-339. Plant Disease 94: 1431-1440. Cherkaoui A., Hibbs J., Emonet S., Tangomo M., Girard M., Maier T., Klepel S., Renner U., Kostrzewa M., 2006. Fast and Francois P., Schrenzel J., 2010. Comparison of two Matrix- reliable MALDI-TOF MS–based microorganism identifica- Assisted Laser Desorption Ionization-Time of Flight Mass tion. Nature Methods 3: doi:10.1038/nmeth1870. Spectrometry methods with conventional phenotypic iden- Mann R.A., Blom J., Bühlmann A., Plummer K.M., Beer S.V., tification for routine identification of bacteria to the species Luck J.E., Goesmann A., Frey J.E., Rodoni B.C. et al., 2012. level. Journal of Clinical Microbiology 48: 1169-1175. Comparative analysis of the integrative conjugative element Dickey R.S., Zumoff C.H., 1988. Emended description of En- of Rubus- and Spiraeoideae-infecting Erwinia amylovora terobacter cancerogenus comb. nov. (formerly Erwinia can- strains gives indications to genome reduction during species cerogena. International Journal of Systematic and Evolutionary evolution. Gene 504: 6-12. Microbiology 38: 371-374. Matsuura T., Mizuno A., Tsukamoto T., Shimizu Y., Saito N., Dierig A., Frei R., Egli A., 2015. The fast route to microbe iden- Sato S., Kikuchi S., Uzuki T., Azegami K. et al., 2012. Er- tification: matrix assisted laser desorption/ionization-time winia uzenensis sp. nov., a novel pathogen that affects Euro- Journal of Plant Pathology (2017), 99 (Special issue), 121-129 Rezzonico et al. 129

pean pear trees (Pyrus communis L.). International Journal of Samson R., Legendre J.B., Christen R., Fischer-Le Saux M., Ach- Systematic and Evolutionary Microbiology 62: 1799-1803. ouak W., Gardan L., 2005. Transfer of Pectobacterium chry- McManus P.S., Jones A.L., 1995. Genetic fingerprinting of Er- santhemi (Burkholder et al., 1953) Brenner et al., 1973 and winia amylovora strains isolated from tree-fruit crops and Brenneria paradisiaca to the genus Dickeya gen. nov. as Dickeya Rubus spp. Phytopathology 85: 1547-1553. chrysanthemi comb. nov. and Dickeya paradisiaca comb. nov. Montesinos E., 2003. Plant-associated microorganisms: a view and delineation of four novel species, Dickeya dadantii sp. from the scope of microbiology. International Microbiology nov., Dickeya dianthicola sp. nov., Dickeya dieffenbachiae sp. nov. and Dickeya zeae sp. nov. International Journal of System- 6: 221-223. atic and Evolutionary Microbiology 55: 1415-1427. Palmer M., Steenkamp E.T., Coetzee M.P.A., Chan W.-Y., Van Zyl E., De Maayer P., Coutinho T.A., Blom J., Smits T.H.M. Sauer S., Freiwald A., Maier T., Kube M., Reinhardt R., et al., 2017. Phylogenomic resolution of the bacterial genus Kostrzewa M., Geider K., 2008. Classification and identifica- Pantoea and its relationship with Erwinia and Tatumella. tion of bacteria by mass spectrometry and computational Antonie van Leeuwenhoek, submitted. analysis. PLoS One 3: e2843. Patel R., 2013. Matrix-Assisted Laser Desorption Ionization– Seng P., Drancourt M., Gouriet F., La Scola B., Fournier P.-E., Time of Flight Mass Spectrometry in clinical microbiology. Rolain J. M., Raoult D., 2009. Ongoing revolution in bacteriology: routine identification of bacteria by Matrix-Assisted Clinical Infectious Diseases 57: 564-572. Laser Desorption Ionization Time-of-Flight Mass Spectrom- Pirc M., Ravnikar M., Tomlinson J., Dreo T., 2009. Improved etry. Clinical Infectious Diseases 49: 543-551. fireblight diagnostics using quantitative real-time PCR detection of Erwinia amylovora chromosomal DNA. Plant Pa- Skrodenytė-Arbačiauskienė V., Radžiutė S., Stunžėnas V., Būda V., 2012. Erwinia typographi sp.nov., isolated from bark bee- thology 58: 872-881. tle (Ips typographus) gut. International Journal of Systematic Poza Carrion C., Aguilar A., Gallego Rodriguez F.J., Nuñez and Evolutionary Microbiology 62: 942-948. Moreno Y., González Biosca E., González R., Lopez M.M., Rodriguez Palenzuela P., 2008. Brenneria quercina and Ser- Smits T.H.M., Rezzonico F., López M.M., Blom J., Goesmann ratia spp. isolated from Spanish oak trees: molecular charac- A., Frey J.E., Duffy B., 2013. Phylogenetic position and viru- lence apparatus of the pear flower necrosis pathogen Erwin- terization and PCR development. Plant Pathology 57: 308-319. ia piriflorinigrans CFBP 5888 T as assessed by comparative Rezzonico F., Smits T.H.M., Duffy B., 2011. Diversity and genomics. Systematic and Applied Microbiology 36: 449-456. functionality of CRISPR regions in fire blight pathogen Erwinia amylovora. Applied and Environmental Microbiol- Stockwell V.O., Hockett K., Marie C., Duffy B., 2008. Pink Er- winia amylovora: colony discoloration in diagnostic isolations ogy 77: 3819-3829. by co-cultured bacteria. Acta Horticulturae 793: 539-542. Rezzonico F., Vogel G., Duffy B., Tonolla M., 2010. Application of whole-cell Matrix-Assisted Laser Desorption Ionization– Thapa S.P., Cho S.Y., Hur J.H., Lim C.K., 2012. Phenotypic and Time of Flight Mass Spectrometry for rapid identification genetic characterization of Erwinia rhapontici isolated from and clustering analysis of Pantoea species. Applied and Envi- diseased Asian pear fruit trees. Phytoparasitica 40: 507-514. ronmental Microbiology 76: 4497-4509. Thompson J.D., Higgins D.G., Gibson T.J., 1994. CLUSTAL Rezzonico F., Smits T.H.M., Montesinos E., Frey J.E., Duffy B., W: improving the sensitivity of progressive multiple se- 2009. Genotypic comparison of Pantoea agglomerans plant quence alignment through sequence weighting, position- specific gap penalties and weight matrix choice. Nucleic and clinical strains. BMC Microbiology 9: 204. Acids Research 22: 4673-4680. Rezzonico F., Braun-Kiewnick A., Mann R.A., Rodoni B., Goessmann A., Duffy B., Smits T.H.M., 2012. Lipopolysac- Wensing A., Gernold M., Geider K., 2012. Detection of Er- charide biosynthesis genes discriminate between Rubus- and winia species from the apple and pear flora by mass spectros- Spiraeoideae- infective genotypes of Erwinia amylovora. copy of whole cells and with novel PCR primers. Journal of Applied Microbiology : 147-158. Molecular Plant Pathology 13: 975-984 112 Rezzonico F., Smits T.H.M., Born Y., Blom J., Frey J.E., Goes- Winslow C.E.A., Broadhurst J., Buchanan R.E., Krumwiede Jr. mann A., Cleenwerck I., de Vos P., Bonaterra A. et al., 2016. C., Rogers L.A., Smith G.H., 1920. The families and genera Erwinia gerundensis sp. nov., a cosmopolitan epiphyte origi- of the bacteria. Final report of the Committee of the Society nally isolated from pome fruit trees. International Journal of of American Bacteriologists on characterization and classification of bacterial types. Journal of Bacteriology : 191-229. Systematic and Evolutionary Microbiology 66: 1583-1592. 5

Received April 27, 2017 Accepted May 9, 2017

Journal of Plant Pathology (2017), 99 (Special issue), 121-129 Rezzonico et al. S1

ERWINIA SPECIES IDENTIFICATION USING MATRIX-ASSISTED LASER DESORPTION IONIZATION-TIME OF FLIGHT MASS SPECTROMETRY

F. Rezzonico, B. Duffy, T.H.M. Smits, and J.F. Pothier

SUPPLEMENTARY MATERIAL

Supplementary Table S1. Strains used in this study.

Strain designationa Geographic origin Host (subfamily/genus)b CRISPR groupc Year Notes (Received from/as) Erwinia amylovora 01SFR-BO Ravenna, Italy Sorbus sp. (S) Ia 1991 ACW 56400 FR, Switzerland Pyrus communis (S) Ia 2007 AFRS2 Alabama, USA Malus sp. (S) A 1994 Ea110 Michigan, USA Malus domestica ‘Jonathan’ (S) Ib 1975 Ea153 Oregon, USA Malus domestica (S) Ia 1989 Ea1/79 Germany Cotoneaster sp. I 1979 Ea222 Czech Republic Cotoneaster sp. I 1994 Ea263 Israel Cydonia oblonga (S) Ia 1997 Ea273 New York, USA Malus domestica (S) Ia 1971 Ea4/82 Egypt Pyrus communis (S) Ia 1982 Ea510 Alberta, Canada Rubus idaeus (R) R < 1996 Ea530 Maine, USA Rubus idaeus (R) R 1949 Ea644 Massachusetts, USA Rubus sp. (R) R 2003 Ea646 Quebec, Canada Rubus idaeus (R) R - Ea6-96r Canada Rubus sp. (R) R 1996 Ea7-96r Canada Rubus sp. (R) R 1996 CFBP 1376 Belgium Cotoneaster sp. (S) I 1972 CFBP 1430 France Crataegus sp. (hawthorn) (S) I 1972 CFBP 2582 Ireland Pyrus sp. (S) I 1986 CFBP 2586 Ireland Pyracantha sp. (S) I 1986 CFBP 3049 Ontario, Canada Malus domestica ‘Jonathan’ (S) Ib 1981 CFBP 3792 Idaho, USA Prunus salicina (plum) (S) II < 1996 CFBP 3860 Hungary Malus domestica ‘Idared’ (S) I 1996 IH3-1 Louisiana, USA Rhaphiolepis indica (Indian hawthorn) (S) IH 1998 ATCC BAA-2158 Illinois, USA Rubus sp. (R) R 1977 IL-5 JL1168 Washington, USA Pyrus communis (S) III 1988 G. Sundin JL1185 Washington, USA Pyrus communis (S) III 1988 G. Sundin JL1189 Washington, USA Pyrus communis (S) III 1988 G. Sundin LA036 Washington, USA - Other 1988 V. Stockwell LA096 Washington, USA - III 1988 V. Stockwell ME1 Ohio (?), USA Rubus sp. (R) R 2003 MR1 Michigan, USA Rubus idaeus (R) R - MO-E-101b Missouri, USA - III - SLAPL-3 California, USA Malus sp. (S) A 1995 UPN527 Navarra, Spain Malus domestica (S) Ia 1997 Erwinia aphidicola ATCC 27991 - Clinical (eye drainage) 1974 ATCC 27992 - Clinical (eye drainage) 1974 JCM 21239 Japan Acyrthosiphon pisum (pea aphid) 1993 JCM 21240 Japan Acyrthosiphon pisum (pea aphid) 1993 JCM 21241 Japan Acyrthosiphon pisum (pea aphid) 1993 JCM 21242 Japan Acyrthosiphon pisum (pea aphid) 1993 LMG 5341 USA Clinical 1969 Received as P. agglomerans Erwinia billingiae 23048B Australia Pyrus communis - R. Powney 23050A Australia Pyrus communis - R. Powney 38#14 Australia Pyrus communis - R. Powney BE66 Australia Chaenomeles sp. - R. Powney DAR72021 Australia Cotoneaster sp. - R. Powney LMG 2610 United Kingdom Crataegus laevigata 1958 LMG 2613T United Kingdom Pyrus communis 1959 LMG 2622 United Kingdom Prunus avium 1959 LMG 2623 United Kingdom Ulmus sp. 1960 S2 MALDI-TOF identification of Erwinia spp. Journal of Plant Pathology (2017), 99 (Special issue), 121-129

Supplementary Table S1 (continued). Strains used in this study.

Strain designationa Geographic origin Host (subfamily/genus)b CRISPR groupc Year Notes (Received from/as) Erwinia gerundensis EM477 Catalonia, Spain Pyrus malus ‘Royal Gala’ 1994 E. Montesinos CCOS 904 Catalonia, Spain Pyrus malus ‘Top red’ 1994 E. Montesinos EM573 Catalonia, Spain Pyrus malus ‘Golden’ 1994 E. Montesinos CCOS 903T Catalonia, Spain Pyrus communis ‘Winter Nellis’ 1994 E. Montesinos EM603 Catalonia, Spain Pyrus communis ‘General Lecler’ 1994 E. Montesinos Erwinia mallotivora CFBP 2503T Japan Mallotus japonicus 1975 LMG 1270 Japan Mallotus japonicus 1975 LMG 1271 Japan Mallotus japonicus 1975 Erwinia oleae CFBP 6632 Toledo, Spain Olea europaea L. (Olive knot) 1998 LMG 25321 Valenzano, Italy Olea europaea L. (Olive knot) 2007 C. Moretti LMG 25322T Scanzano, Italy Olea europaea L. (Olive knot) 2003 C. Moretti LMG 25323 Perugia, Italy Olea europaea L. (Olive knot) 2003 C. Moretti Erwinia papayae CFBP 5164 Grenada Carica papaya ‘Solo’ 1994 CFBP 5180 St. Thomas, Virgin Carica papaya ‘Local’ 1995 Islands CFBP 5189T Martinique, France Carica papaya ‘Local’ 1995 CFBP 5215 Trinidad and Tobago Carica papaya ‘Taiwan’ 1995 CFBP 5294 Guadeloupe, France Carica papaya ‘Local’ 1998 Erwinia persicina CFBP 3622T Japan Lycopersicon esculentum < 1990 NCPPB 3375 USA Phaseolus vulgaris 1981 NCPPB 3771 Hong Kong (?) - 1990 NCPPB 3775 Hong Kong (?) - 1990 Erwinia piriflorinigrans CFBP 5882 Spain Pyrus communis ‘Tendral’ 2000 M.M. López CFBP 5883 Spain Pyrus communis ‘Ercolini’ 2000 M.M. López CFBP 5884 Spain Pyrus communis ‘Ercolini’ 2000 M.M. López CFBP 5885 Spain Pyrus communis ‘Ercolini’ 2000 M.M. López CFBP 5886 Spain Pyrus communis ‘Ercolini’ 2000 M.M. López CFBP 5887 Spain Pyrus communis ‘Ercolini’ 2000 M.M. López CFBP 5888T Spain Pyrus communis ‘Ercolini’ < 2001 M.M. López Erwinia psidii CFBP 3627T Brasil Psidium guajava 1982 CFBP 5844 Brasil Psidium guajava ‘Paluma’ 2000 CFBP 5845 Brasil Psidium guajava 2000 LMG 7035 Valinhos, Brazil Psidium guajava 1983 LMG 7039 Valinhos, Brazil Psidium guajava 1982 Erwinia pyrifoliae CFBP 4171 South Korea Pyrus pyrifolius < 1997 CFBP 4173 South Korea Pyrus pyrifolius < 1997 CFBP 4174 South Korea Pyrus pyrifolius < 1997 CFBP 7155 Hokkaido, Japon Pyrus pyrifolius 1994 DSMZ 12162 South Korea Pyrus pyrifolia 1996 DSMZ 12163T Chuncheon, South Korea Pyrus pyrifolia Nakai 1996 Erwinia rhapontici CFBP 3618 United Kingdom Rheum rhabarbarum 1963 NCPPB 139 United Kingdom Rheum rhaponticum 1940 NCPPB 1578T United Kingdom Rheum rhaponticum 1963 NCPPB 1739 United Kingdom Rheum rhaponticum 1965 NCPPB 2549 France Triticum aestivum 1973 Received as E. persicina NCPPB 2860 United Kingdom Malus sylvestris 1969 Erwinia tasmaniensis BE57 Australia Prunus sp. R. Powney BE65 Australia Chaenomeles japonica R. Powney CFBP 7178 Victoria, Australia Pyrus communis L. 1998 Et4/99 Australia Malus domestica K. Geider LA540 Oregon, USA Malus domestica 1994 V. Stockwell LMG 25318T Tasmania, Australia Malus sp. 1999 K. Geider Journal of Plant Pathology (2017), 99 (Special issue), 121-129 Rezzonico et al. S3

Supplementary Table S1 (continued). Strains used in this study.

Strain designationa Geographic origin Host (subfamily/genus)b CRISPR groupc Year Notes (Received from/as) Erwinia toletana CFBP 6630 Toledo, Spain Olea europaea L. 1998 CFBP 6639 Toledo, Spain Retama sphaerocarpa 1998 CFBP 6643 Toledo, Spain Olea europaea L. 1998 CFBP 6645 Toledo, Spain Olea europaea L. 1998 LMG 24162T Toledo, Spain Olea europaea L. 1998 Erwinia tracheiphila LMG 5020 USA Cucumis sativus 1951 LMG 5021 USA Cucumis sativus 1951 “Pinking” isolates ACW 40945 LU, Switzerland Malus sp. 2004 P45, Born et al. (2016) ACW 41072 SG, Switzerland n.a. 2004 B. Schoch ACW 41558 ZH, Switzerland Crataegus sp. 2004 B. Schoch ACW 43997 TG, Switzerland Pyrus sp. 2005 B. Schoch ACW 44214 SG, Switzerland Cydonia sp. 2005 B. Schoch ACW 44286 SG, Switzerland Malus sp. 2005 B. Schoch

Outgroups Species LMG 2709T USA Juglans regia Brenneria rubrifaciens 1963 CFBP 2048T USA Chrysanthemum morifolium Dickeya chrysantemi 1956 DSMZ 17580T Czechoslovakia Populus canadensis ‘Regenerata’ Enterobacter 1964 cancerogenus CU3298 Washington, USA Lupinus sulfurens Lonsdalea quercina 1988 Received as Erwinia lupinicola CU3299T Washington, USA Lupinus albus Lonsdalea quercina 1987 Received as Erwinia lupinicola CU3300 Washington, USA Lupinus albus Lonsdalea quercina 1986 Received as Erwinia lupinicola NCPPB 1852T USA Quercus sp. Lonsdalea quercina 1964 subsp. quercina LMG 1286T Zimbabwe Clinical (knee laceration) Pantoea agglomerans 1956 LMG 2794 France Drinking water Rahnella aquatilis 1981

a Acronyms of institutes and commercial culture collections associated to the strains: ACW: Agroscope Changins-Wädenswil; ATCC: American Type Culture Collection; CCOS: Culture Collection Of Switzerland; CFBP: Collection Française de Bactéries associées aux Plantes; DSMZ: Deutsche Sam- mlung von Mikroorganismen und Zellkulturen; JCM: Japan Collection of Microorganisms; LMG: Belgian Co-ordinated Collections of Micro-organisms, Laboratory for Microbiology of the Faculty of Sciences of the Ghent University; NCPPB: National Collection of Plant Pathogenic Bacteria; b Subfamily/genus of the host plant from which the strain was isolated: (S) Spiraeoideae, (R) Rubus; c E. amylovora CRISPR group according to (Rezzonico et al., 2011): I-III, Spiraeoideae CRISPR groups I-III; R, Rubus group; A, Spiraeoideae ancestral group. S4 MALDI-TOF identification of Erwinia spp. Journal of Plant Pathology (2017), 99 (Special issue), 121-129

Supplementary Table S2. NCBI accession numbers of the sequences used for MLSA. Accessions generated within this work are marked in bold.

Species GenBank accession numbers Strains 16S rRNA atpD gyrB infB rpoB B. rubrifaciens LMG 2709T FJ611885 JF311504 JF311617 JF311730 JF311842 D. chrysanthemi CFBP 2048T Z96093 JF311523 JF311636 JF311749 JF311861 E. cancerogenus DSM 17580T FJ611867 JX424850 JX424980 JX425109 JX425239 E. amylovora CFBP 1232T AJ233410 HQ393596 HQ393608 HQ393620 HQ393632 E. aphidicola JCM21238T FN547376 FN547378 FN547377 FN547373 FN547374 E. billingiae LMG 2613T JQ929658 EU145259 EU145275 EU145291 EU145307 E. gerundensis CCOS903T FJ611848 JN591389 FJ617413 KP070835 KP070836 E. mallotivora CFBP 2503T AJ233414 HQ393589 HQ393601 HQ393613 HQ393625 E. oleae LMG 25322T GU810925 GU991653 GU991654 GU991655 GU991656 E. papayae CFBP 5189T AY131237 HQ393588 HQ393600 HQ393612 HQ393624 E. persicina CFBP 3622T Z96086 HQ393598 HQ393610 HQ393622 HQ393634 ICMP 15602 – JF311460 JF311573 – – LMG 2689 – JF311459 JF311572 – – NCPPB 3375 – HQ393599 HQ393611 – – E. piriflorinigrans CFBP 5888T GQ405202 JF311469 JF311582 JF311695 JF311808 E. psidii CFBP 3627T JQ809696 HQ393594 HQ393606 HQ393618 HQ393630 E. pyrifoliae DSM 12163T EF122435 HQ393597 HQ393609 HQ393621 HQ393633 E. rhapontici NCPPB 1578T Z96087 EF988751 EF988838 EF988924 EF989010 LMG 2642 – JF311454 JF311567 – – LMG 2645 – JF311455 JF311568 – – LMG 2648 – JF311456 JF311569 – – ACW 40945 – KY563262 KY563254 – – ACW 41072 – KY563257 KY563250 – – ACW 41558 – KY563258 KY563251 – – ACW 43997 – KY563259 KY563252 – – ACW 44214 – KY563260 KY563253 – – ACW 44286 – KY563254 KY563254 – – E. tasmaniensis LMG 25318T AM055716 CU468135 CU468135 CU468135 CU468135 E. toletana LMG 24162T FN547375 EU145258 EU145274 EU145290 EU145306 E. tracheiphila LMG 2906T Y13250 HQ393591 HQ393603 HQ393615 HQ393627 L. quercina NCPPB 1852T AJ223469 JF311543 JF311656 JF311769 JF311881 CU 3299 EU490595 KY563256 KY563249 KY563264 KY563263 P. agglomerans LMG 1286T AJ251466 EF988711 EF988798 EF988884 EF988970 R. aquatilis LMG 2794T AJ233426 KF387629 KF387630 KF387631 KF387632

T = type strain