Erwinia piriflorinigrans is a newly described pathogen causing necrosis of pear blossoms. Complete sequencing of the 37-kb plasmid pEPIR37 common to 27 E. piriflorinigrans strains revealed homology to sequences of the ubiquitous plasmids pEA29 of the fire blight pathogen E. amylovora, plasmid pEP36 of E. pyrifoliae, plasmid pEJ30 of Erwinia sp. from Japan, and genomic regions of the related Rosaceae epiphytic Erwinia species E. tasmaniensis and E. billingiae. A second 5·5-kb cryptic plasmid pEPIR5, found in 12 E. piriflorinigrans strains, was also sequenced revealing mobilization and replication proteins with similarities to many small ColE1-type plasmids in Erwinia spp. and other enterobacteria. Functional analyses of pEPIR37 introduced into a strain of E. amylovora cured of pEA29 plasmid, which has a reduced virulence, showed a role in increasing symptom development similar to that observed in E. amylovora carrying plasmid pEA29.
Several Erwinia species have recently been described that have restricted geographic distribution, host range and/or virulence expression on European pear (Pyrus communis) and/or Asian pear (syn. Nashi pear; P. pyrifolia), relative to the broader-host-range fire blight pathogen of pear and other Rosaceae, Erwinia amylovora (Palacio-Bielsa et al., 2012). Among them, E. pyrifoliae causes blight of Asian pear and has thus far been reported only from Korea and Japan (Kim et al., 1999; Geider et al., 2009), Erwinia sp. strain Ejp617 causes pear shoot blight (Park et al., 2011) and E. uzenensis causes bacterial black shoot disease of pear (Matsuura et al., 2012). Erwinia tasmaniensis is a Rosaceae epiphyte, not known to be pathogenic to pear (Geider et al., 2006), but having genomic indications of potential phytopathogenicity to unknown hosts (Kube et al., 2008), and E. billingiae (formerly Pantoea agglomerans/Enterobacter agglomerans) is a cosmopolitan epiphyte with no indications of phytopathogenicity (Mergaert et al., 1999; Kube et al., 2010). Erwinia piriflorinigrans is one of the newest members of this cast, causing necrosis of pear flowers but no advancing symptoms into shoots. It does not affect other Rosaceae host plants and thus far it has only been reported in Spain (López et al., 2011).
Comparative genomic analysis of E. amylovora, E. pyrifoliae, E. tasmaniensis and E. billingiae has shown a high similarity in gene content, organization and virulence and/or ecological fitness factors (Smits et al., 2010b, 2011). Pathogenic species have hallmark genetic determinants for virulence and fitness on pome fruit hosts, including type IV, type III, and type VI secretion systems and effectors, exopolysaccharides, sucrose and sorbitol metabolism and desferroxamine siderophores (Smits et al., 2010a; De Maayer et al., 2011; Powney et al., 2011; Smits & Duffy, 2011; Kamber et al., 2012). Variations of this repertoire, however, occur in pome fruit non-pathogenic and epiphytic species, indicating that some auxiliary elements may contribute to full manifestation of specific pathogenic phenotypes (Sundin, 2007).
All these Erwinia species harbour a variety of plasmids and the presence of one plasmid in the range of 30 kb is a common characteristic they all have. Some of these plasmids have been studied, showing a role in host–pathogen interactions (Maxson-Stein et al., 2003). The best characterized is plasmid pEA29, which is essentially ubiquitous in E. amylovora (Laurent et al., 1989; McGhee & Jones, 2000) and it has been implicated in symptom development (Laurent et al., 1989; McGhee & Sundin, 2008; Llop et al., 2012). The presence of comparably sized plasmids (c. 30–40 kb), homology of plasmid-encoded genes (e.g. thiamine biosynthetic genes, msrA; Llop et al., 2012) and low virulence retention in strains devoid of pEA29 (Laurent et al., 1989) highlights the genetic versatility in Erwinia spp., but also the importance of characterizing novel plasmids.
The objective of this study was to investigate the plasmids present in strains of the recently described species E. piriflorinigrans: plasmid pEPIR37, common to all analysed strains, and pEPIR5, found in some strains. Both plasmids were sequenced and a comparative genomic analysis of pEPIR37 conducted to identify similarities and differences with other characterized Erwinia plasmids and chromosomes. The contribution of pEPIR37 to virulence was investigated using heterologous transfer to E. amylovora.
Materials and methods
Bacterial strains and growth media
Bacterial strains used in this study are listed in Table 1. Plasmid extractions on 27 strains of E. piriflorinigrans were performed using the Real Miniprep Turbo Kit (Durviz). Escherichia coli strain JM109 was used for cloning experiments. All strains were grown on LB medium, with kanamycin (50 μg mL−1) added when appropriate.
Table 1. Bacterial strains and plasmids used in this study
Origin/year of isolation
Strain CFBP 1430 cured of plasmid pEA29.
According to Geider et al. (2009), these strains are now E. pyrifoliae.
A plasmid, named pEPIR37, was isolated from the type strain E. piriflorinigrans (CFBP 5888T) and one of smaller size named pEPIR5 from strain CFBP 5887, from an overnight suspension in LB at 26°C (Table 1). The extraction was analysed by electrophoresis in 0·8% agarose. Plasmids were extracted from gels using the QIAEX II kit (QIAGEN) and purified with a phenol–chloroform step and resuspended in distilled water. Restriction analyses with BamHI and EcoRI were performed to confirm the identity of the plasmids. To carry out the in vitro transposition labelling, the EZ-Tn5 <KAN-2> Insertion Kit with kanamycin resistance (Epicenter) was used, following the manufacturer's instructions. Briefly, 0·2 μg plasmid DNA was included in a 10-μL reaction volume with 1 × reaction buffer, 1 μL transposase and a volume molar equivalent of transposon EZ-Tn5 <KAN-2> (corresponding to 0·08 pmoles). This was then incubated at 37°C for 2 h and transformed into E. coli strain JM109, resulting in plasmids pEPIR37::Tn5 and pEPIR5::Tn5.
Labelled plasmids were transformed separately into E. coli strain JM109 by electroporation. Transformation was performed at 2·5 kV, 200 ohms in a Gene Pulser X Cell (Bio-Rad) using 0·2-cm electroporation cuvettes. After incubation for 1 h at 37°C in LB medium, suspensions were plated on LB amended with kanamycin (50 μg mL−1) and incubated for 24 h at 37°C. Transformants were checked by plasmid extraction, digested with EcoRI and BamHI and analysed by electrophoresis in 0·8% agarose to confirm the labelling of plasmids. Labelled plasmid pEPIR37 was purified and sequenced following a shotgun cloning and Sanger sequencing procedure (Macrogen). The coverage was 11·8 × with 440 reads (average read length 702 bp) in a single contig. Plasmid pEPIR5 was sequenced through direct sequencing from transposon Tn5 with overlapping primers at the sequencing facility of the Instituto de Biología Molecular y Celular de Plantas (IBMCP), Valencia, Spain.
Genes on pEPIR37 were predicted using a combined strategy as described in Llop et al. (2011). The resulting plasmid annotation was manually curated. Routine sequence manipulations were completed using several subroutines of the lasergene package (DNASTAR). Plasmid pEPIR5 reads were assembled into one contig using vectorNTI v. 8 (Invitrogen) and annotated using the approach described above for pEPIR37.
Comparison of pEPIR37 with plasmids pEA29, pEP36 and pEJ30
The common features among plasmids pEA29 of E. amylovora, pEP36 of E. pyrifoliae and pEJ30 from Japanese Erwinia strain Ejp556 (now identified as E. pyrifoliae by Geider et al., 2009), reported by Maxson-Stein et al. (2003), were investigated in case they were also present in plasmid pEPIR37. Coding sequences (CDS) found in pEPIR37 were analysed and gene content compared against the other plasmids.
Phylogenetic analyses with the amino acid sequences of thiamine biosynthetic genes (thiF, thiG, thiS and thiO) of related pathogenic and non-pathogenic Erwinia spp. (E. piriflorinigrans, E. amylovora, E. pyrifoliae and E. tasmaniensis) and some Pantoea species were performed using neighbour joining. Citreicella sp. SE45 was chosen as the out-group. Significance of nodes was evaluated by bootstrap analysis with 1000 replicates. Analyses of amino acid distance were based on the Dayhoff matrix-based model (Schwarz & Dayhoff, 1979) and all positions containing gaps and missing data were eliminated. The analyses were performed using mega v. 5 (Tamura et al., 2011). Erwinia billingiae proteins were not included in the analyses because this species harbours a different thiamine biosynthesis pathway.
Aggressiveness assays with pEPIR37 introduced into E. amylovora strains
Plasmid pEPIR37::Tn5, introduced into E. coli strain JM109, was extracted for transforming E. amylovora strain PMV 6014, which is strain CFBP 1430 of E. amylovora cured of pEA29 and shows low aggressiveness. Transformation was performed by electroporation as described previously, but the strains were incubated for 2 h at 26°C in LB medium before plating. The transformants were incubated on LB agar with 50 μg mL−1 kanamycin for 48 h at 26°C.
The possible function of plasmid pEPIR37 in symptom development was analysed in E. amylovora strain PMV 6014. This strain was tested formerly with the immature pear assay described previously (Llop et al., 2011). Strains were grown overnight at 26°C on LB agar with kanamycin. Cell suspensions were prepared and concentrations adjusted at 600 nm by serial dilutions in 0·5 × phosphate-buffered saline (PBS; 0·36% NaCl, 0·018% NaH2PO4.2H2O, 0·12% Na2HPO4.12H2O). Suspensions of 103 colony-forming units (CFU) mL−1 were used for inoculation of pear cv. Williams fruits as described in Llop et al. (2011) to observe if there was an increase in aggressiveness when plasmid pEPIR37::Tn5 was introduced. Plant material was prepared according to the procedure used by Cabrefiga & Montesinos (2005). Four wounds were made on immature pear fruits of 3–4 cm diameter using a pipette tip, and 3 μL suspension introduced into each wound. Fruits were then incubated at 26°C under controlled relative humidity conditions. Aggressiveness was evaluated daily until the 10th day. Wounds were considered infected when either drops of bacterial exudates or necrosis appeared in and around the inoculation site of the pathogen. The experimental design consisted of three repetitions of three pears (total of nine pears) for each strain and concentration. The incidence of infected wounds (%) for each repetition was assessed and severity was evaluated by means of a visual scale (from 0 to 3). The scale was based on necrosis progression or exudate production, with 0 = no symptoms, 1 = exudates at the inoculation point, 2 = necrosis around the wound area, and 3 = necrosis expanding through the fruit. Disease severity (S) was calculated using the formula:
where SI is the corresponding severity index in an inoculated wound, i is the wound number, n the total wounds inoculated and three the maximum severity index.
Erwinia amylovora strain CFBP 1430 was employed as a positive control, and PBS buffer and E. piriflorinigrans CFBP 5888T were used as negative controls in all assays. The inoculation experiments were performed twice.
anova was performed to analyse the effect of each strain treatment on infection incidence and means were separated by Tukey's test at P ≤ 0·05. The analyses were done with the GLM procedure of the pc-statistical analysis system v. 8.2 (SAS Institute Inc.).
Sequence analyses and plasmid annotation of pEPIR37
A total of 27 E. piriflorinigrans strains, isolated from 1999 to 2011 in Valencia, Spain (Table 1), were analysed. Using plasmid extraction analysis followed by digestion with BamHI and EcoRI (data not shown), all strains were found to harbour the same plasmid, pEPIR37, and 12 of them also contained the plasmid pEPIR5 (Fig. 1). Plasmid pEPIR37 was determined to be 37 376 bp with a G + C content of 49·8%. The annotation of the sequence identified 42 coding sequences (CDS) (Fig. 2; Table 2), for which the best blast hits were one CDS common to plasmid pEA29, two CDS common to pEJ01, seven CDS common to pEP36, four CDS to pET46 and one CDS to pEB102. Another 12 CDS showed homology to putative proteins from the chromosome of E. pyrifoliae (two CDS), E. tasmaniensis (six CDS) and E. billingiae (four CDS). Another six CDS were homologues of those observed in different Enterobacteriaceae species (i.e. Serratia odorifera, Edwarsiella ictaluri, Dickeya dadantii, E. coli, Pantoea spp. and Pectobacterium wasabiae) and one Clostridiaceae (Clostridium sp.) (Table 2).
Table 2. Predicted coding sequences (CDS) of proteins of Erwinia piriflorinigrans plasmid pEPIR37 in the GenBank nonredundant database
Description from GenDB
Locus tags of orthologues in pEA29 of Erwinia amylovora CFBP 1430
Locus tags of orthologues in pJE01 of Erwinia sp. Ejp617
Locus tags of orthologues in pEP36 of Erwinia pyrifoliae DSM 12163
Locus tags of orthologues in pET46 of Erwinia tasmaniensis Et1/99
Locus tags of orthologues in chromosome of Erwinia tasmaniensis Et1/99
Locus tags of orthologues in genome of Erwinia billingiae Eb661
Sequence identity (%)
Bold type indicates genes that showed the highest sequence identity to genes in pEPIR37 by maximum percentage identity, query coverage, E-value and/or biological significance.
plasmid pEB102 E. billingiae.
chromosome E. pyrifoliae.
Insertion element IS2A uncharacterized 48·2-kDa protein
The replication origin was found to have high homology with RepE proteins and two partitioning proteins (ParA and ParB) in related bacteria, e.g. E. tasmaniensis pET46 (Kube et al., 2008) or P. vagans pPag3 (Smits et al., 2010c) (Table 2). A 120-bp fragment of RepA protein (94% sequence identity), similar to the replication protein present in pEA29, pEJ30 and pEP36 plasmids, was also present in pEPIR37 (Fig. 3). Remnants of an insertion sequence (IS) (CDS 18) and one Tn3-like and one Tn2-like transposon (CDS 14 and 15 respectively), similar to others present in related enterobacteria (Yersiniaenterocolitica and E. coli) were also found in pEPIR37, but they were different from those found in pEP36 and pEJ30. Analyses of pEPIR37 also showed that the genes for thiamine biosynthesis (thiO, thiS, thiG and thiF; CDS 25–28) were similar to genes present in E. amylovora, Erwinia sp., E. pyrifoliae and E. tasmaniensis. These genes had a protein identity of 90–94% with the proteins observed in pEA29, pEP36 and pEJ01, and also in the genome of E. tasmaniensis (Table 2). The phylogenetic analysis of the concatenated ThiOSGF proteins in the different Erwinia species studied is shown in Figure 4 and for separate proteins in Figure S1. The dendrogram shows that the E. piriflorinigrans plasmid proteins for thiamine biosynthesis are closely related to the other compared Erwinia spp.
Common and different features observed in plasmids pEA29, pEP36, pEJ01 and pEPIR37
The different plasmids in the range of 30–40 kb from pathogenic pome fruit Erwinia species showed several similarities, as well as some differences (Fig. 3). Plasmids pEP36 and pEJ30 were almost identical, except for the presence of a transposon (Tn5394) of approximately 6 kb in pEP36, which explains their difference in size (Maxson-Stein et al., 2003). Analysis of the plasmid sequence of pEPIR37 also showed that it lacked short sequence repeats (SSR) and a BamHI restriction site position, common features in pEA29, pEP36 and pEJ01 (Table 2); it also showed IS elements and transposons different to pEP36, but similar to six other genes observed in related enterobacteria, as indicated above. Twelve CDS were common to pEP36, pEJ30 and pEA29, whereas 12 CDS were similar to genes present in the genomes of E. pyrifoliae, E. tasmaniensis and E. billingiae. Comparisons of the four plasmids showed that plasmids pEPIR37 and pEA29 had 12 CDS in common, while plasmids pEP36 and pEJ01 showed 14 common CDS with pEPIR37 (Table 3). In addition, the organization of genes hns and msrA was identical to the gene cluster located on the chromosome of E. tasmaniensis.
Table 3. Common features observed in 30-kb plasmids from pathogenic Erwinia spp. compared to E. piriflorinigrans pEPIR37
Sequence analyses and plasmid annotation of pEPIR5
The cryptic ColE1-type plasmid pEPIR5 was 5·5 kb in size, with a G + C content of 54·3%, and nine CDS (Fig. 5; Table 4). This plasmid showed mobilization proteins similar to the ones present in plasmid pRK10 of Serratia marcescens, as well as in the chromosome of E. coli strains MS78-1 and E482, and also similar to a hypothetical protein present in plasmid pPAT9B02 of Pantoea sp. AT-9b. Four CDS showed no homology with sequences registered in the NCBI database.
Table 4. Predicted coding sequences (CDS) of proteins of Erwinia piriflorinigrans plasmid pEPIR5 in the GenBank nonredundant database
Description from GenBank
Sequence identity (%)
Hypothetical protein Pat9b_5020 (Pantoea sp. At-9b)
To test whether the plasmid is involved in virulence, the effect on incidence and severity of symptoms of the introduction of pEPIR37::Tn5 into the low-virulence E. amylovora strain PMV 6014 without plasmid pEA29 in immature pear fruits was studied (Fig. 6). As expected, the type strain of E. piriflorinigrans showed no symptoms in pear fruitlets and strain PMV 6014 showed significantly lower severity (19·1 ± 1·2) than strain CFBP 1430 (73·5 ± 1·2). However, severity increased significantly with strain PMV 6014 when pEPIR37::Tn5 was introduced when it was inoculated, in the two sets of assays (54·6 ± 5·4), even though Tn5 was inserted in gene thiF, affecting the ability to synthesize thiamine.
Regarding the symptoms observed after 6 days in immature pear assays, 30·5% of wounds inoculated with strain PMV 6014 showed symptoms of necrosis, in contrast to 69·4% of wounds inoculated with strains PMV 6014 + pEPIR37::Tn5 and 80·5% of those inoculated with CFBP 1430 (positive control) (Fig. 7).
This is the first report on the complete sequencing of two plasmids of the recently described species E. piriflorinigrans. The data here show that all analysed strains harbour a plasmid of 37 kb and many of them another of 5 kb. Relatedness among pathogenic bacteria of the genus Erwinia can be shown by various shared characteristics (Smits et al., 2011), because they have the same hosts and harbour plasmids of either similar sizes (in the range of 30–40 kb) and/or small cryptic plasmids of similar size (Llop et al., 2011). These species cause more or less similar symptomatology on pear, but appear to have decreasing disease severity: E. amylovora > E. pyrifoliae > Japanese Erwinia > E. piriflorinigrans. Information about the virulence of E. piriflorinigrans strains is scarce because the bacterium has only been reported in pear orchards in Valencia (Spain), and the damage caused by the disease seems to be limited to necrosis of pear blossoms (Roselló et al., 2006). However, comparisons of virulence among strains and species should take into account host and/or pathogen intraspecies genotype (e.g. cultivar; Norelli et al., 1984). Care should also be taken when pooling observations of different studies where experimental conditions that can influence disease severity typically vary (e.g. pathogen inoculum concentration, inoculation conditions; Cabrefiga & Montesinos, 2005; Billing, 2011).
Recent genome comparison studies have shown that E. amylovora strains share more than 99·99% identity at the nucleotide level, indicating only minimal evolution since its geographical dispersal (Smits et al., 2010a; Kamber et al., 2012). Also, the genomes of two E. pyrifoliae strains from Korea (Ep1/96 and DSM 12163T) have shown they are almost identical (Zhao & Qi, 2011). Chromosomal colinearity of E. amylovora to the closely related E. pyrifoliae strains and to Erwinia sp. Ejp617 and E. tasmaniensis was observed despite the fact that large-scale chromosomal rearrangements were detected (Kamber et al., 2012).
The main difference among E. amylovora strains is reported to be their plasmid content, pEA29 being the plasmid most frequently reported. This plasmid has been described as playing a role in the increase of symptoms, because of the presence of genes involved in the biosynthesis of thiamine (thiO, thiS, thiG and thiF) (Laurent et al., 1989; McGhee & Jones, 2000; McGhee & Sundin, 2008). Thiamine genes are also located on plasmids pEP36 and pEJ30 in E. pyrifoliae and on the chromosome in E. tasmaniensis. The gene content of plasmids pEA29, pEP36 and pEJ30 is highly conserved, as well as the thi genes; other genes are also present in the same order and orientation (e.g. msrA, betT). This study has shown that all these genes are also present on E. piriflorinigrans pEPIR37. Phylogenetic analyses performed on the concatenated ThiOSGF proteins of the related pathogenic and non-pathogenic Erwinia spp. and Pantoea spp. revealed a close relationship between the proteins encoded on pEPIR37 and those from other Erwinia spp. For as far as known, the thiOSGF clusters in Pantoea spp. are present on plasmids as well (Smits et al., 2010c; P. De Maayer, FABI, University of Pretoria, South Africa, personal communication).
The fact that pEPIR37 has a RepE replication protein related to that of E. tasmaniensis plasmid pET46 and ABC transporter proteins similar to E. billingiae and Pantoea spp. also possibly links it with these two species of non-pathogenic Erwinia and Rosaceae epiphytes. This is also supported by the fact that E. piriflorinigrans is able to cause necrosis of pear blossoms, but the symptoms do not progress to other parts of the plant and it does not affect immature pear fruits. However, heterologous introduction of pEPIR37 into a plasmidless E. amylovora strain increased its virulence on immature fruits to a degree similar to that obtained with pEA29, despite the fact that gene thiF is mutated in the labelled plasmid, affecting the biosynthesis of thiamine.
This study also annotated a cryptic plasmid pEPIR5, with no clear evidence of what type of role it provides to its host. It can be grouped with genes from pEPIR37 that showed no homology to previously sequenced genes in being called orphan genes (Mira et al., 2002).
The sequencing of more plasmids from other strains of Erwinia species may confirm their genomic versatility. Saprophytic and non-pathogenic Erwinia have similar rosaceous plant hosts as several pathogenic species, and their geographic distribution could be much more widespread than is currently reported. They could serve as a reservoir for transmission of plasmids or horizontal gene transfer and a step in evolutionary adaptation to different environments or host genotypes. Whole-genome analyses of more strains of the species Erwinia that share fruit trees as habitat, such as the most recently described Erwinia species, E. uzenensis (Matsuura et al., 2012), or other uncharacterized Erwinia spp. from Japan (Palacio-Bielsa et al., 2012) for which nothing about plasmid content and genome composition is yet known, could give new clues to the origin of these plant pathogens and a better understanding of their life cycle.
This research was financed through Project CYCIT AGL2008-05723-C02-01/AGR from the Spanish Ministry of Science and Innovation, the Swiss Federal Office of Agriculture, an IVIA institutional grant to SB, and a COST short-term scientific mission grant to PL. We thank Luis Rubio for technical assistance in the phylogenetic analyses. This work was conducted within the Swiss ProfiCrops programme and the European research network COST Action 864.