In July 2022, symptoms of a novel disease similar to pear fire blight were observed and recorded in a pear orchard in Shandan County, Zhangye City, Gansu Province (longitude 100° 41′ to 101° 42′ E; latitude 37° 50′ to 39° 03′ N), China. Ten sets of diseased leaves from three different pear trees were brought back to the laboratory for strains isolation. The diseased leaves were stored in kraft paper bags and stored at −4°C.

The pathogens were isolated following the method proposed by Han et al. (2022), briefly, to create bacteria suspensions, diseased leaves were cut into 1 cm² pieces, surface-disinfected with 75% alcohol for 30 s and 1% sodium hypochlorite for 60 s, washed three times with 5 mL sterile water, placed in a sterilized mortar, ground with sterile water, and left to stand for 30 min. Three replicates were set for the pathogen isolation test. An inoculation loop was used to transfer a small amount of each pathogen suspension into NA (Nutrient Agar: Beef extract 3 g, Peptone 10 g, Glucose 2.5 g, Sodium chloride 5 g, Agar 15 g, Distilled water 1,000 mL, pH value 7.0 ± 0.1) culture medium. The NA culture medium was then placed in an incubator at 28°C with a 12/12 h light/dark cycle for 24–48 h. When colonies appeared on the NA culture medium, they were purified using the streak plate method. The represented bacterial isolates were designated DE1-DE8. All eight bacterial isolates underwent pathogenicity testing.

Pathogenicity testing of the eight bacterial isolates was conducted on healthy, young detached leaves and fruits (Pyrus pyrifolia var. Sinensis (Lindley)Y. Teng et K. Tanabe, Pyrus ussuriensis Maxim, Pyrus brestschneideri Rehd.) collected in Gansu Agricultural University that were surface-disinfected with 75% alcohol and rinsed three times with sterile water. Six inoculated leaves and fruits per isolate, and four replicate treats. An equal volume of sterile water was used as the control treatment.

Each of the eight bacterial isolates was added to tryptic soy broth (TSB) and cultured at 25°C for 24 h, centrifuged to obtain a pellet containing the bacteria, and adjusted with sterile water to an optical density at 600 nm (OD₆₀₀) of 1.0 (to give a bacterial concentration of approximately 10⁸ CFU/mL). The abovementioned leaves and fruits were placed on the germination bed of a sterile growth bottle (Yao et al., 2021). A sterile inoculation needle was used to create three wounds per leaf, and a small piece (about 2 × 2 cm) of sterilized skimmed cotton was placed over the wounds. Next, 2 mL bacterial solution (10⁸ CFU/mL) was added to the cotton. A sterile scalpel was used to create a triangular wound on the fruit, measuring 2 cm in length, and the bacterial suspension was sprayed into the wound. A control experiment employed sterile water as a substitute for the bacterial suspension. Finally, 100 mL sterile water was added to the bottom of the bottle for moisturization. The bottle was placed at 23°C and 45% humidity in a group culture room with an 18/6 h light/dark cycle for incubation. Symptom onset on the inoculated leaves (spots similar to the disease symptoms in the field) was observed and recorded from 7 d. The morphology and 16S rRNA gene sequencing of the re-isolated and inoculated bacteria was verified to be the same, thereby verifying Koch’s postulates (Berraf-Tebbal et al., 2020).

After verification of Koch’s postulates, isolate DE2 was transferred to 100 mL NB (nutrient broth) medium for culture at 30°C and 160 rpm for 24 h. Thereafter, the OD₆₀₀ was adjusted with sterile water to 1.0. The solution was then centrifuged (10,000 rpm for 1 min) to obtain the bacteria, which was washed with sterile water. A bacterial genomic DNA extraction kit (TIANGEN, Beijing, China) was used to extract the genomic DNA. Next, the 16S rRNA gene was PCR amplified (Schedule 1 provides the reaction mixture details) using universal primers 27F (5′-GAGTTTATCCTGGTCAG-3′) and 1492R (5′-AAGGGGTGGATCCAGCCGCA-3′) (Clarridge, 2004; Hassan, 2017) (Supplementary Table S1). The PCR products were separated by 1% agarose gel electrophoresis and then sequenced by Sanger method at Sangon Biotechnology Company (Shanghai, China).

NCBI BLASTN¹ and EZ BioCloud 16S-based ID² tools were used for online comparison of isolate DE2 with known bacterial strains. The 16S rRNA sequences of 18 bacterial strains with high sequence similarity to isolate DE2 were selected to construct a phylogenetic tree using MEGA 11. Using the Neighbor-Joining method to construct the phylogenetic tree, with the Bootstrap value set to 1,000.

The physiological and biochemical characteristics of isolate DE2 were determined using an API 20E kit (BioMèrieux, France) and a GEN III Microplate (Biolog, United States) according to the manufacturers’ instructions.

Fresh bacteria were scraped onto tryptic soy agar (TSA) medium and incubated for 24 h, transferred to a 15 mL Eppendorf tube, freeze-dried, and subjected to fatty acid analysis using the Sherlock Microbial Identification System (MIDI, United States) to analyze the fatty acid composition of isolate DE2.

Isolate DE2 underwent whole-genome sequencing by Majorbio Biotech (Shanghai, China). First, to ensure that the DNA samples met the required standards for whole-genome sequencing, NanoDrop 2000 was employed to assess the DNA purity, and a Quantus Fluorometer (using PicoGreen reagent) was used to assess the DNA concentration.

Genomic DNA was sequenced using a combination of PacBio RS II Single Molecule Real Time (SMRT) and Illumina sequencing platforms. The Illumina data was used to evaluate the complexity of the genome.

For Illumina sequencing, at least 1 μg genomic DNA per strain was used for library construction. The DNA samples were sheared into 400–500 bp fragments using a Covaris M220 Focused Acoustic Shearer following the manufacturer’s protocol. Illumina sequencing libraries were prepared from the sheared fragments using a NEXTflex^™ Rapid DNA-Seq Kit. Briefly, 5′ ends were first end-repaired and phosphorylated. Next, the 3′ ends were A-tailed and ligated to sequencing adapters. Thereafter, the adapter-ligated products were enriched by PCR. The prepared libraries were then used for paired-end Illumina sequencing (2 × 150 bp) on an Illumina HiSeq X Ten machine.

For PacBio sequencing, 15 μg genomic DNA was spun in a Covaris g-TUBE (Covaris, MA, United States) at 6,000 rpm for 60 s using an Eppendorf 5424 centrifuge (Eppendorf, NY, United States). The DNA fragments were then purified, end-repaired, and ligated with SMRTbell sequencing adapters (Pacific Biosciences, CA, United States) following the manufacturer’s recommendations. The resulting sequencing library was purified three times using 0.45× volumes of Agencourt AMPure XP beads (Beckman Coulter Genomics, MA, United States) following the manufacturer’s recommendations. Next, a ~ 10-kb insert library was prepared and sequenced on one SMRT cell using standard methods.

The PacBio and Illumina data were used for bioinformatics analysis using the Majorbio Cloud Platform³ (Shanghai Majorbio Bio-pharm Technology Co., Ltd.). The detailed procedures are as follows. The complete genome sequence was assembled using both the PacBio and Illumina reads. The original image data were transferred via base calling into sequence data, defined as raw data or raw reads and saved as FASTQ files. The FASTQ files (containing read sequences and quality information) are the original data provided for users. A statistic indicating quality information was used for quality trimming, during which low-quality data were removed to generate clean data. The reads were then assembled into a contig using a hierarchical genome assembly process (HGAP) and canu (Koren et al., 2017). The last circular step was checked and finished manually, generating a complete genome with seamless chromosomes and plasmids (Stothard and Wishart, 2005). Finally, error correction of the PacBio assembly was performed based on the Illumina reads using Pilon (Liu et al., 2020). Glimmer (Delcher et al., 2007) was used for coding sequence (CDS) prediction, tRNAscan-SE (Borodovsky and Mcininch, 1993) was used for tRNA prediction, and Barrnap was used for rRNA prediction. Using CGview website⁴ to draw a whole genome loop diagram of the assembled genome sequence.

K-mer analysis involves utilizing reads derived from second-generation sequencing, selecting high-quality sequencing regions, and estimating the genome size by analyzing individual K-mers at a base-by-base level. Genome coverage analysis employs two approaches: (1) K-mer-based approach: By leveraging the statistical patterns of K-mer distribution, the genome size is estimated, and this estimate is then juxtaposed with the actual genome assembly size to infer the genome’s coverage. This gives: Coverage = Assembled genome length / Estimated genome length based on K-mer analysis; (2) Reads-based approach: By constructing a library from the assembly results and performing soap analysis on the set of reads used for assembly, the length of the alignment is determined, and this length is then used to infer the genome’s coverage. This gives: Coverage = Sum of lengths of alignments/Length of assembly result.

As different genomes have different GC contents, the GC content can be used to determine whether there is any contamination from other species’ genomes in the samples. SOAP software was used to compare the raw paired-end library reads to the assembled sequence to obtain the base depth. A length (1, 3, 5, 8, 10 kb) was selected as a window, and the sequence was advanced base by base (or advanced without repeats), and the mean depth of each window was calculated along with the GC content to obtain a GC-depth distribution.

To further determine the classification of strain DE2, JSpeciesWS⁵ and the Genome-to-Genome Distance Calculator (GGDC)⁶ were used to calculate the average nucleotide identity (ANI) and digital DNA–DNA hybridization (dDDH) values between the complete genome of isolate DE2 and Erwinia strains with high similarity (from the NCBI genome database⁷). In generally, a dDDH value ≥70% or an ANI value ≥95–96% indicates that the strains are members of the same species. Conversely, ANI values <95–96% or dDDH values <70% indicate that the strain does not belong to any known species. A heatmap based on ANI values was generated using TBtools software (Yao et al., 2022).

To understand the direct genetic differences between strain DE2 and other Erwinia species, genome sequences of closely related Erwinia species were downloaded from the NCBI database. The complete genome sequences of strain DE2 and these other Erwinia species were then uploaded together to the Majorbio Cloud Platform.⁸ The pan-genomic analysis of strain DE2 is completed by this platform. The Pan-Genome Analysis Pipeline (PGAP) was used for pan-genome analysis (Ren et al., 2022).

The Virulence Factor Database (VFDB)⁹ was used to annotate the virulence factors encoded by strain DE2 (Chen et al., 2005).

The main symptom of the disease was multiple disease spots on diseased leaves. The infected fruit turned black and completely rotted, withering and leading to a dead state. These caused the leaves to progressively change from yellow to a dark brown, withering but remaining attached to the branches. Black spots were produced in the phloem and xylem of infected branches, which expanded and produced bacterial pus (Figure 1). The symptoms were similar to those of pear fire blight.

Based on the morphological characteristics (color, size, and shape) of the isolates from the diseased leaves, the isolates were classified into 8 types, coded DE1–DE8. It was then purified, cultured, and subjected to pathogenicity testing. The isolate of DE2 caused black lesions in detached leaves and fruits on day 7. It was noteworthy that DE2 showed different levels of pathogenicity after infecting pear fruits of different varieties. Among them, the symptoms of fruits (Pyrus pyrifolia var. Sinensis (Lindley)Y. Teng et K.) were the most severe, with the fastest infection rate and the largest area. However, the symptoms of fruits (Pyrus ussuriensis Maxim.) were the mildest. The symptoms of fruits (Pyrus brestschneideri Rehd.) were intermediate between the two, accompanied by the production of bacterial beads. This was consistent with the symptoms observed in the field. No disease symptoms were observed in the control group (Figure 2) and other seven isolates. The pathogenic bacteria were re-isolated from the diseased site of the detached leaves and fruits that had been inoculated with DE2, and the purified isolate was found to be identical to DE2. Thus, Koch’s postulates were verified, demonstrating that isolate DE2 was the disease pathogen.

Isolate DE2 colonies on NA medium were initially milky white, while the later colonies were light yellow, sticky, dome-shaped, with a smooth surface and neat edges (Figure 3A). Isolate DE2 colonies on TSA medium were yellow (light yellow at the center, with a light yellow transparent ring at the periphery), with circular protrusions, a smooth, glossy surface, and neat edges (Figure 3B).

The phylogenetic tree based on 16S rRNA sequences showed that isolate DE2 was clustered with species in the genus Erwinia (Figure 4), indicating that the classification of isolate DE2 at the genus level is a species belonging to the genus Erwinia. The NCBI comparison results showed that the sequence similarity between isolate DE2 and strain Erwinia billingiae Eb661 was the highest, reaching 98.70% (Supplementary Table S2). The EZ BioCloud comparison results showed that the sequence similarity between isolate DE2 and strain Erwinia billingiae CIP 106121 was the highest, reaching 98.77% (Supplementary Table S3). Based on 16S rRNA analysis, the other isolates DE1, DE3-DE8 were all identified as Bacillus species (Supplementary Figure S1).

According to the API 20E and GEN III results, isolate DE2 was negative for oxidase, and did not produce indole, urease, gelatinase, tryptophan dehydrogenase, arginine hydrolase, lysine decarboxylase, guanate decarboxylase, or β-galactosidase. Citric acid utilization and the Voges–Prokauer (VP) reaction were negative. Acid production could be achieved by utilizing glucose, mannitol, inositol, sorbitol, rhamnose, amygdalin, and arabinose. The carbon sources that could be utilized were α-D-glucose, D-sorbitol, D-mannose, D-mannitol, aminoacetyl-L-proline, D-galacturonic acid, methyl pyruvate, D-maltose, melibiose, D-fructose, D-arabitol, L-alanine, L-galacturonolactone, D-trehalose, β-formyl-D-glucoside, D-galactose, inositol, D-gluconic acid, L-lactic acid, D-cellobiose, D-salicylic acid, glycerol, L-aspartic acid, D-glucuronic acid, citric acid, N-acetyl-D-glucosamine, D-glucose-6-phosphate, L-glutamic acid, glucuronamide amide, α-Ketone glutaric acid, acetoacetic acid, N-acetyl-β-D-mannosamine, D-fructose-6-phosphate, L-histamine, mucic acid, L-rhamnose, quinic acid, L-malic acid, acetic acid, inosine, L-serine, gluconic acid, bromosuccinic acid, and formic acid. The carbon sources that could not be utilized were D-pine disaccharide, stachyose, trisaccharide α-D-lactose, L-arginine γ-aminobutyric acid, L-butyric acid, α-ketobutyric acid, and propionic acid. Isolate DE2 could be distinguished by 15 physiological indicators from its relatives, Erwinia billingiae Eb661, Erwinia toletana WS4403, Pantoea wallisii LMG 26277, and Erwinia persicina NBRC 102418 (Supplementary Table S4).

The main features of the fatty acid profile of isolate DE2 (Supplementary Figure S2) were as follows: C16:0 accounted for 56.01%, C18:1n9c for 10.86%, C22:1n9 for 10.26%, C16:1 for 6.26%, C18:2n6c for 6.21%, and C14:0 for 5.11% (Supplementary Table S5). Isolate DE2 could be distinguished from related species based on the cellular fatty acid profile (Supplementary Table S6).

Whole-genome sequencing revealed that isolate DE2 contains 1 circular chromosome and 5 plasmids (Figure 5). The circular chromosome was 4,523,022 bp, with a GC content of 54.54%. Plasmid 1 was 182,370 bp, with a GC content of 50.02%. Plasmid 2 was 40,780 bp, with a GC content of 44.00%. Plasmid 3 was 5,809 bp, with a GC content of 54.30%. Plasmid 4 was 3,577 bp, with a GC content of 50.55%. Plasmid 5 was 2,872 bp, with a GC content of 46.20%. The genome coverage statistics for strain DE2 based on K-mer analysis stand at 92.51%. The coverage statistics derived from second-generation sequencing data aligning reads reach 99.30%, and those from the third-generation sequencing data aligning reads are 99.53%.

The circular chromosome contained 4,155 CDS, 77 tRNA, and 22 rRNA genes. Plasmid 1 contained 199 CDS. Plasmid 2 contained 43 CDS. Plasmid 3 contained 8 CDS. Plasmid 4 contained 9 CDS. Plasmid 5 contained 2 CDS (Supplementary Table S7).

As shown in the GC-depth distribution (Supplementary Figure S3), the upper and right histograms represent the sliding window frequency distributions of GC content and depth, respectively, showing that there was no obvious bias in the GC content of the sequencing sample; the multiple points in the middle were concentrated in a relatively small range, which indicated that there was no contamination of the DNA of isolate DE2.

dDDH and ANI are the widely accepted gold standards for identifying a bacterial isolate species. Through 16S rRNA gene sequencing, isolate DE2 was found to belong to the genus Erwinia. To clarify the taxonomic classification of isolate DE2, its complete genome was compared to the complete genomes of 29 reference strains of the genus Erwinia based on dDDH and ANI. The ANI values between isolate DE2 and the 29 Erwinia species were 76.42–81.80% (Figure 6). Similarly, the dDDH values were 21.3–26.2% (Figure 7). The ANI and dDDH thresholds for determining whether two strains are the same species are ≥95–96% and ≥70%, respectively. As the ANI and dDDH values between isolate DE2 and species of the genus Erwinia were lower than the threshold values, isolate DE2 is a potentially new species in the genus Erwinia.

The PGAP analysis showed that of the 6,781 genes in 10 Erwinia strains, 1,373 (20.25% of the pan-genome) were core genes shared by all 10. Each of the 10 strains had unique genes (134–1,126), and strain DE2 had 279 unique genes (Figure 8A). The equation for the relationship between the pan-genome size (y) and the number of genomes (x) is y = 4 , 106 + 389.5 x − 1 + 1 , 930 exp − 2 3.06 1 − exp − x − 1 3.06 1 − exp − 1 3.06 (Figure 8B). The equation for the relationship between the number of core genes (y) and the number of genomes (x) is y = 1 , 533 + 4 , 328 exp − x 1.85 (Figure 8C).

Comparing the complete genome sequence of strain DE2 with the VFDB database showed that strain DE2 has 548 putative virulence genes, comprising 534 on the chromosome and 14 in plasmids (Figure 9). These virulence genes can be divided into 14 categories, comprising Nutritional/metabolic factors (149 genes), Immune regulation (75 genes), Adhesion (67 genes), Exotoxins (56 genes), Invasion (49 genes), Regulation (39 genes), Mobility (35 genes), Effect delivery systems (34 genes), Biofilm (16 genes), Stress survival (12 genes), Antibiotic activity/competitive advantage (11 genes), Exoenzymes (3 genes), Post-translational modification (2 genes), and Others (1 gene) (Supplementary Table S8). Additionally, 1,064 genes were annotated using the Pathogen–Host Interactions (PHI)-base (Supplementary Table S9). Among them, plasmid 1 has 8 virulence factors, plasmid 2 has 6 virulence factors, and the remaining plasmids have no virulence factors.

Erwinia pyri (pyri. L. n. pyrus betulaefolia pear tree; L. n. pyrus betulaefolia leaves; L. part. Adj. blackened, darkened; N. L. part. Adj. Pyri withered pear leaves, from which the strain was isolated).

The strain is a Gram-negative, facultative aerobe that does not form spores. Oxidase reaction, citric acid utilization, and VP reaction are negative. Colonies grow on TSA medium after 48 h at 28°C; they are non-fluorescent, round, light yellow, slightly raised mucous colonies, 2–4 mm in diameter, with regular edges. Colonies grow on NA medium. Colonies grow in TSB nutrient solution at 28°C and 30°C but not at ≥32°C. Based on the API 20E system, indole, urease, gelatinase, tryptophan dehydrogenase, arginine hydrolase, lysine decarboxylase, β-galactosidase are negative, but acid production can be achieved by utilizing glucose, mannitol, inositol, sorbitol, rhamnose, melibiose, amygdalin, and arabinose. The main fatty acids were (in descending order): C16:0, C18:1n9c, C22:1n9, C16:1, C18:2n6c, C14:0, C18:0, C12:0, C20:3n3, C22:0, C18:3n3, C17:0, C24:1, C23:0, C24:0, C22:2, C20:0, C20:5n3, C20:1, C15:0, C13:0, and C22:6n3. The strain has a circular chromosome and five plasmids. The DNA G + C content is 54.54 mol%.