Bacterial strains were isolated from the bleeding cankers of Platanus x acerifolia (London plane) at St James Park, London and Tilia x europaea (common lime) at Tidworth Garrison, Wiltshire. Swab samples of the weeping exudate were collected from the London plane in May and October, 2020. A small bark panel (3 × 3 cm) was taken from the bleeding common lime in August 2019.

Swabs were rehydrated in sterile 1/4 Ringers, the swab spread on Luria-Bertani (LB) agar (Oxoid) and Eosin Methylene Blue (EMB) agar (Merck) which were initially incubated at 35°C under anaerobic conditions for 4 days to promote growth of facultatively anaerobic bacteria. Bark tissue was removed from the panel with a scalpel to expose the developing lesion. Wood chips were cut around the advancing lesion front, plated on LB agar and incubated at 25°C for 7 days. Single colonies were obtained through re-streaking on LB agar and incubation at 25°C.

The strains used in this study are listed in Supplementary Table 1.

Genomic DNA for genotypic characterisation was extracted using an alkali lysis method (Niemann et al., 1997) and stored at −20°C. Strains were initially screened and tentatively identified based on partial 16S rRNA gene sequences (using primer *pD), and the almost complete 16S rRNA gene sequence determined for the proposed type strain using the primers published by Coenye et al. (1999). Standard PCR conditions and cycles were used with an annealing temperature of 55°C. Multilocus sequence analysis (MLSA) based on partial sequences of four housekeeping genes (gyrB, rpoB, infB, and atpD) was performed on the strains tentatively identified as Erwinia sp. by partial 16S rRNA gene sequencing. MLSA PCR conditions, cycles and primers were applied as previously described (Brady et al., 2008), although with an updated version of the gyrB amplification primers: gyrB 01.2F (5′ TAARTT YGAYGAYAACTCBTAYAAACT 3′) and gyrB 02.2R (5′ CMCCYTCCACHARGTASAKTTC 3′). Consensus sequences for the 16S rRNA and housekeeping genes were generated in UGENE v38.1 (Okonechnikov et al., 2012) and aligned and trimmed in MEGA X (Kumar et al., 2018) to the following lengths: 16S rRNA—1346 bp, gyrB—742, rpoB—637, infB—615 bp, and atpD—642 bp. Maximum likelihood phylogenetic tree construction was performed on both datasets in PhyML 3.0 (Guindon et al., 2010) with 1,000 bootstrap replicates and automatic model selection by Smart Model Selection (Lefort et al., 2017).

To examine the genetic diversity between strains of the potential novel species, BOX PCR was performed using the primer and conditions previously published (Versalovic et al., 1994). Amplification products were separated on 1.5% agarose at 50 V for ∼3 h.

The whole genomes of strains BAC 15a-03b^T and Til 1, isolated from London plane and common lime respectively, were sequenced by MicrobesNG (Birmingham, UK). DNA was extracted by enzymatic cell lysis and purified with SPRI (Solid Phase Reversible Immobilisation) beads, before sequencing using the Illumina HiSeq platform. Adapters were trimmed from reads using Trimmomatic 0.30 with a sliding window quality cut-off of Q15 (Bolger et al., 2014), while de novo assembly was performed using SPAdes version 3.11.1 (Bankevich et al., 2012) and the resulting contigs annotated in the prokaryotic genome annotation pipeline (PGAP) (Tatusova et al., 2016). Average nucleotide identity (ANI) and average amino acid identity (AAI) were performed on the novel strains, along with validly published Erwinia species and “E. beijingensis,” using FastANI (Jain et al., 2018) and the AAI-matrix calculator from the Kostas Lab (Rodriguez-R and Konstantinidis, 2016), respectively. Additionally, pairwise percentage of conserved proteins (POCP) were calculated using the script pocp.rb (Hoelzer, 2020) which follows the approach by Qin et al. (2014).

Pairwise phylogenomic comparisons, between the novel species and current closest relatives from the genera Erwinia and Mixta, were carried out on the Type (Strain) Genome Server (TYGS) (Meier-Kolthoff and Göker, 2019) using Genome Blast Distance Phylogeny (GBDP). Accurate intergenomic distances were inferred under the algorithm “trimming” and distance formula d₅ (Meier-Kolthoff et al., 2013), with 100 distance replicates each. The resulting intergenomic distances were used to infer a balanced minimum evolution tree with branch support via FASTME 2.1.6.1 including SPR postprocessing (Lefort et al., 2015) and 100 pseudo-bootstrap replicates. The tree was rooted at the midpoint (Farris, 1972).

Colony morphology was observed following growth of strains on LB agar incubated at 30°C for 48 h. The temperature range for growth was determined on LB agar incubated at 4, 10, 25, 30, 37, and 41°C. Tolerance of the novel species to salt was tested in saline-free nutrient broth with NaCl added in increments of 1% w/v from 1 to 10%, while pH tolerance was tested in LB broth with the pH adjusted from 4 to 10 in increments of 1 as previously described (Brady et al., 2022). Catalase and oxidase activity were determined by bubble production in 3% v/v H₂O₂ and by staining with Kovács reagent (1% tetra-methyl-p-phenylenediamine dihydrochloride), respectively. Cell size, morphology and motility were examined using a light microscope and the CellSens software v 1.11 (Olympus Life Science, Tokyo, Japan) following growth in LB broth at 25°C for ∼6 h. The flagella arrangement was visualised by transmission electron microscopy (FEI Tecnai 12 120kV BioTwin Spirit TEM) following negative staining as previously published (Brady et al., 2022).

Commercial phenotypic assays, including API 20E, API 50 CHB/E (bioMérieux), and GEN III GN/GP microplates (Biolog), were performed on strains BAC 15a-03b^T, BAC 1-01-01, Til 1, and Til 5 according to the manufacturer’s instructions. The type strain of E. toletana LMG 24162^T was included as a positive control. The API 20E galleries were scored after 24 h incubation at 37°C, while the API 50 CHB/E galleries and GEN III microplates were scored after 24 h incubation at 30°C, and again after 48 h. The fatty acid methyl ester (FAME) profile of the novel species was determined by Fera Science Ltd. (York, UK). Following the protocol based on the Sherlock Microbial Identification System Version 6.4 (MIDI Inc.), strains were cultivated on TSA at 28°C for 24 h prior to fatty acid extraction. The results obtained were compared against the library RTSBA6 6.21.

The protein annotations generated through PGAP for the whole genome sequences of BAC 15a-03b^T, Til 1, E. toletana DAPP-PG 735, E. iniecta B120^T, and “E. beijingensis” JZB2120001^T were queried through two databases to assess the pathogenic potential of their proteome. DIAMOND version v2.0.11.149 (Buchfink et al., 2021) was used to query genomes with the BlastP command against the Virulence Factor Database (VFDB) (Liu et al., 2022), accessed October 3, 2022. A query cut-off of 97% and a percentage identity equal or greater than 50 were used to ensure high sequence alignment identification by DIAMOND against the VFDB (Doonan et al., 2019). To assess the microbe-plant interaction potential of the four species, the annotations were queried using BlastP + HMMER Aligner/Mapper against the “plant bacterial only interaction factors” (PIFAR-Pred) database through the PLant-associated BActeria web resource (PLaBAse) (Martínez-García et al., 2016; Patz et al., 2021). For the identification of type III secretion system effectors (T3SS), open reading frames (ORFs) were first identified using orfipy and exported in fasta format (Singh and Wurtele, 2021). ORFs were then individually uploaded to the Effectidor web server for the prediction of T3SS effectors (Wagner et al., 2022).

Pairwise 16S rRNA sequence similarity of BAC 15a-03b^T, the proposed type strain of the novel species, in EzBioCloud (Yoon et al., 2019) showed greatest similarity to E. toletana (99.4%), Erwinia endophytica (98.2%), and E. billingiae (97.9%). These similarity values are reflected in the 16S rRNA gene phylogenetic tree (Supplementary Figure 1) where BAC 15a-03b^T was situated on a separate branch between E. toletana and E. endophytica, albeit with no bootstrap support. The proposed novel species was confirmed as a new taxon in the concatenated MLSA phylogenetic tree (Figure 1), as all strains formed a single cluster without any reference strains of Erwinia species with validly published names, with 100% bootstrap support. This cluster was in a well-supported clade (88%) with E. toletana and E. iniecta, corresponding to Subgroup 3 from Soutar and Stavrinides (2022) providing support for their transfer to a novel genus. “Erwinia beijingensis” was situated on a separate branch on the border of this clade, but without bootstrap support, suggesting that its position could change with the addition of further novel species.

The BOX PCR profiles demonstrated that there were slight variations between strains isolated from the same tree host species, although identical clones isolated from common lime were also observed (Supplementary Figure 2). Several shared bands were observed between strains isolated from London plane and common lime. However, these are clearly different hosts situated in different counties, suggesting the strains do not originate from a single clone and that a genetically divergent population exists in Great Britain.

Assembly of the sequences for BAC 15a-03b^T and Til 1 resulted in genomes of 5.23–5.31 Mbp, with the DNA G + C content ranging from 53.5–53.6 mol%. Genome sequences were submitted to GenBank under the BioProject number PRJNA878947. Assembly accession numbers and genome features are listed in Supplementary Table 2. ANI values (Table 1) confirmed the status of the strains from bleeding cankers as a single novel taxon, with 100% similarity observed between the strains. The ANI data reflects the phylogenetic relationship observed between the novel species and E. iniecta and E. toletana, with similarity values ranging from 83.8 to 91.7% between the three species, although these are well below the suggested cut-off value of 95–96% for species delimitation (Goris et al., 2007). The remaining Erwinia species tested demonstrated lower values of 79.1–80.6% when compared to the novel species, E. iniecta and E. toletana. “E. beijingensis” exhibited ANI values of 80.4–80.5% to the three species under examination, which are in the same region as the values shared by E. aphidicola and the three species (80.5–80.6%). The conclusions inferred from the ANI analysis were confirmed by in silico DNA-DNA hybridisation (Meier-Kolthoff et al., 2013), the results for which are also presented in Table 1. The AAI data (Supplementary Table 3) were congruent with those from ANI, with higher values observed between the novel species, E. iniecta and E. toletana (86–95%) and lower values to the remaining Erwinia species (76–78%) providing justification for their transfer to a novel genus. The AAI values between the proposed novel genus and “E. beijingensis” are 81–82%, which are slightly higher than those exhibited by other Erwinia species but significantly less than the values observed within the novel genus. Genus delineation criteria, based on AAI, have been suggested with species exhibiting > 76% AAI to the type strain of the type species of the genus, and all type strains within the genus sharing > 74% AAI to each other (Nicholson et al., 2020). These cut-off values appear too relaxed for the Erwiniaceae, and Soutar and Stavrinides (2022) suggested that 79% may be an appropriate threshold for delineating genera within this family. However, several Erwinia species exhibited AAI values which can be considered “borderline.” The POCP values (Supplementary Table 3) demonstrated a similar pattern with higher values (75–87%) observed between species of the proposed novel genus, and lower values (64–73%) to “E. beijingensis” and Erwinia species. It is worth noting that E. aphidicola, E. oleae, and E. persicina all demonstrated higher POCP values to the proposed novel genus than “E. beijingensis.”

The phylogenomic tree generated in TYGS clustered the novel species, E. toletana and E. iniecta in a single clade on a separate lineage with 100% support (Figure 2). “E. beijingensis” was situated on a separate branch on the border of this clade, further removed from the proposed novel genus but with 100% support. The phylogeny of the whole genome tree demonstrates the divergence within the genus Erwinia with three clear subgroups of species, as defined previously (Soutar and Stavrinides, 2022) and confirms the proposal that E. iniecta and E. toletana should be reclassified within a novel genus along with the novel species.

After 48 h growth on LB agar, colonies were round, 2–3 mm in diameter, smooth with entire edges, slightly convex, translucent, and cream to pale yellow in colour. Strains grew readily at temperatures of 10–30°C after incubation for 3 days, with weak growth observed at 4 and 37°C. Salt was tolerated by strains in broth supplemented with up to 7% NaCl, while growth in a narrow pH range of 6–8 was observed, with weak growth at pH 9. The proposed novel species is catalase positive and oxidase negative. Cells are straight rods with rounded ends, typically 0.6–1.3 × 1.5–3.5 μm in size, can occur singly, and in pairs or chains and motile by peritrichous flagella (Supplementary Figure 3).

The novel species could be differentiated from its current closest phylogenetic relatives by several phenotypic characteristics including reactions to acetoin production, tryptophan deaminase activity, aspartic acid and pectin (differentiation from E. iniecta) and L-rhamnose, amygdalin, β-methyl-D-glucoside and D-serine (differentiation from E. toletana). Differentiating phenotypic characteristics are presented in Table 2, and positive phenotypic characteristics shared by all current members of the proposed novel genus are listed in Supplementary Table 4. The fatty acid profile of the novel species was most similar to its current closest phylogenetic relative, E. toletana, with C_16:0, C_18:1 ω7c, and summed features 2 and 3 composing the major fatty acids. The complete fatty acid profiles of the novel Erwinia species and current closest phylogenetic relatives are listed in Table 3. “E. beijingensis” exhibited a distinctive fatty acid profile when compared to species of the proposed novel genus, with a high percentage of the cyclopropane fatty acid C_19:0 ω8c and only a minor amount of summed feature 3, which is a major fatty acid of the novel genus. Phenotypic data available from literature for the genera Erwinia and Mixta (Liu et al., 2013; Hauben and Swings, 2015; Ramírez-Bahena et al., 2016; Rezzonico et al., 2016; Mavima et al., 2022) were examined for characteristics to support the differentiation of the proposed novel genus. Although complete phenotypic information for species of Erwinia and Mixta is not widely published (Palmer et al., 2018), some useful phenotypic traits are listed in Table 4.

The BlastP alignment results from the VFDB using the query parameters revealed the presence of 101 virulence genes for BAC 15a-03b^T, Til 1 and E. toletana DAPP-PG 735 while 121 and 105 alignments were identified in E. iniecta B120^T and “E. beijingensis” JZB2120001^T, respectively. Within each comparison 11–23 type III secretion effector delivery system proteins such as SpaQ and Spa9 were identified, excluding “E. beijingensis.” This is most likely due to the stringent search parameters used, which when relaxed revealed a number of T3SS related genes which were previously identified (Xu et al., 2021). The other major protein roles identified were related to motility, immune modulation, invasion, and adherence. The presence of a T3SS has long been used to implicate bacteria as phytopathogens. Within the order Enterobacterales, it is usually an indicator of plant pathogens such as the soft rot pathogen Pectobacterium atrosepticum or the blight pathogens E. amylovora and E. pyrifoliae (Toth et al., 2006; Kube et al., 2010). Due to the importance of the secretion system, Effectidor was used to identify T3SS proteins and effectors from ORFs for each genome. Each species was found to contain 26 core TS33 proteins and related effectors, some of which appear to be novel.

The PIFAR-predictions also demonstrated similar implications for the novel species, with the identified “plant only interaction factors” being “genes that are related to plant interaction and virulence.” Nine different plant cell wall degrading enzymes (PCWDE) were identified in each genome. Many genes relating to detoxification and adhesion were identified amongst other plant interaction factors, such as hormone regulation and exopolysaccharides. Notable potential virulence genes are listed in Supplementary Table 5. The identified genes from both the VFDB and PIFAR comparisons indicate that all four species investigated contain genes used in invasion, adhesion and breakdown of plant tissue, indicating a strong pathogenic potential. Given the point of isolation of the novel species was from bleeding cankers of broadleaf hosts, further investigation of its role as novel pathogen is warranted.

Winslowiella (Win.slow.i.el’la. N.L. fem. n. Winslowiella, named in honour of C.E. Winslow who proposed the genus name Erwinia).

The description is based on the data from Rojas et al. (2004); Campillo et al. (2015), and this study. Cells are Gram-negative, short, straight rods with rounded ends (0.5–1.5 × 1.5–3.5 μm) and can occur singly or in pairs and sometimes chains. Motile by peritrichous flagella. Catalase positive and oxidase negative, facultatively anaerobic. On LB agar colonies are slightly convex, circular with smooth edges, translucent, glistening, non-pigmented and cream to yellow in colour with a diameter of 1–3 mm. Strains can grow at 10–37°C, optimally at 28–30°C but not above 39°C, within the pH range of 6–8 and can tolerate supplemented salt conditions of up to 4%, with some species able to grow in 7% NaCl. Negative for arginine dihydrolase, lysine decarboxylase, ornithine decarboxylase, citrate utilisation, H₂S production, urease, indole production, and gelatinase. β-galactosidase, tryptophan deaminase activity, acetoin production, and nitrate reduction are all variable.

Major fatty acids are C_16:0, C_18:1 ω7c, and summed feature 3 (C_16:1 ω7c and/or C_16:1 ω6c).

Winslowiella species are secondary invaders on diseased olive trees, and have been isolated from wheat aphids, bleeding cankers of broadleaf hosts and the wider environment.

The DNA G + C content ranges from 51.1 to 53.6 mol%. The type species is Winslowiella toletana.

Winslowiella toletana (to.le.ta.na. N.L. fem. adj. toletana, from Toletum, the Roman name for Toledo, the location from which the organisms were isolated).

Basonym: Erwinia toletana (Rojas et al., 2004).

The description is as given above for the genus and E. toletana in Rojas et al. (2004).

The DNA G + C content is 53.6 mol%.

The type strain is A37^T (=CFBP 6631^T = ATCC 700880^T = CECT 5263^T) and was isolated from olive knots in association with Pseudomonas savastanoi pv. savastanoi as secondary invaders on diseased plants.

Winslowiella iniecta (in.iec’ta. L. fem. part. adj. iniecta, thrust in, injected, referring to the fact that the first strains isolated from artificial media were introduced or “thrust in” via the aphid stylet).

Basonym: Erwinia iniecta (Campillo et al., 2015).

The description is as given above for the genus and E. iniecta in Campillo et al. (2015).

The DNA G + C content ranges from 51.1 to 52.2 mol%.

The type strain is B120^T (=CFBP 8182^T = NCCB 100485^T), and was isolated from artificial diets fed on by Diuraphis noxia, biotype 2.

Winslowiella arboricola (ar.bo.ri’cola. L. fem. n. arbor, a tree; L. masc./fem. n. suff. -cola, dweller; from L. masc./fem. n. incola, dweller; N.L. fem. n. arboricola, tree dweller).

The description is as given for the genus with the following additional characteristics. Strains can tolerate supplemented salt conditions of up to 7%. β-galactosidase activity is variable among strains, but tryptophan deaminase activity, acetoin production, and nitrate reduction are all negative. Acid is produced from L-arabinose, D-ribose, D-xylose, D-galactose, D-glucose, D-fructose, D-mannose, L-rhamnose, inositol, D-mannitol, N-acetyl glucosamine, arbutin, esculin ferric citrate, salicin, D-cellobiose, D-maltose, D-lactose, D-melibiose, D-trehalose, gentiobiose, and D-arabitol (API 50 CHB/E).

The following carbon sources are utilised in addition to those listed in Supplementary Table 4: β-methyl-D-glucoside, N-acetyl neuraminic acid, L-rhamnose, inosine, and mucic acid.

The DNA G + C content ranges from 53.5 to 53.6 mol%.

The type strain is BAC 15a-03b^T = LMG 32576^T = NCPPB 4696^T, and was isolated from the bleeding lesion of a Platanus x hispanica (London plane) tree in London, Great Britain.