Comparative Genomics Insights into a Novel Biocontrol Agent Paenibacillus peoriae Strain ZF390 against Bacterial Soft Rot

Simple Summary Bacterial soft rot, attributed to Pectobacterium brasiliense infection, has caused destructive impacts and colossal economic losses to China’s agricultural industry. Chemical control, which was ubiquitously used, cannot manage this disease as expected, so biocontrol has been followed with interest to date. In this study, we found a Paenibacillus peoriae strain ZF390 that had a potent control efficiency over cucumber plants against Pectobacterium brasiliense, and the comparative genomic analysis revealed biocontrol mechanisms might be involved in the strain ZF390. Abstract Bacterial soft rot, caused by Pectobacterium brasiliense, can infect several economically important horticultural crops. However, the management strategies available to control this disease are limited. Plant-growth-promoting rhizobacteria (PGPR) have been considered to be promising biocontrol agents. With the aim of obtaining a strain suitable for agricultural applications, 161 strains were isolated from the rhizosphere soil of healthy cucumber plants and screened through plate bioassays and greenhouse tests. Paenibacillus peoriae ZF390 exhibited an eminent control effect against soft rot disease and a broad antagonistic activity spectrum in vitro. Moreover, ZF390 showed good activities of cellulase, protease, and phosphatase and a tolerance of heavy metal. Whole-genome sequencing was performed and annotated to explore the underlying biocontrol mechanisms. Strain ZF390 consists of one 6,193,667 bp circular chromosome and three plasmids. Comparative genome analysis revealed that ZF390 involves ten gene clusters responsible for secondary metabolite antibiotic synthesis, matching its excellent biocontrol activity. Plenty of genes related to plant growth promotion, biofilm formation, and induced systemic resistance were mined to reveal the biocontrol mechanisms that might consist in strain ZF390. Overall, these findings suggest that strain ZF390 could be a potential biocontrol agent in bacterial-soft-rot management, as well as a source of antimicrobial mechanisms for further exploitation.


Introduction
Pectobacterium brasiliense (Pbr), considered one of the most virulent species among the Pectobacteriaceae, has a wide host range and causes soft rot on miscellaneous growing plants or harvested crops, including important food crops and ornamentals [1,2]. In addition to blackleg disease on potatoes, Pbr causes a devastating soft rot of cucumber in China, with an incidence that varies from 15 to 50% in different fields, causing 20 to 30% yield losses in five provinces during 2014-2015 [2][3][4]. The deficiency of the effective resistance present in commercial cultivars and practical chemical control agents makes the management of soft rot Pectobacteriaceae (SRP) challenging.
In recent decades, biocontrol strategies, which are more novel, cost-effective, and environmentally and human-safe than other control strategies, have become promising tools to reduce losses of vegetables [2,5]. Similar to Bacillus, the Paenibacillus species can be applied in agriculture as phytostimulators, biofertilizers, and bioagents. Paenibacillus

Bacterial Isolation, Growth Conditions, Antagonism Assays, and Biocontrol Assays
Strain ZF390 was derived from the healthy rhizosphere soil of cucumber plantations in Harbin, China, in August 2019. According to a previous study [24], 10 g of the soil sample was subjected to 10,000-fold dilution with sterile water, plated on LB plates, and incubated at 28 • C for 48 h. Bacillus-like and Paenibacillus-like strains were retained for further investigations. All the strains were purified on LB plates and cultivated in LB broth at 28 • C with shaking for 24 h.
The antagonistic activity of selected microorganisms against Pbr was estimated through plate bioassays. The suspensions of all the isolated strains were diluted to an OD 600 = 0.6 with sterile LB broth. Then, 5 µL of the suspensions were inoculated in the center of PDA plates and cultivated for 24 h, respectively. A total of 3 mL of chloroform was added to each upside-down dish for fumigation to inactivate the bacteria colony. Subsequently, 100 µL of the suspension of Pbr was equably added into 5 mL WA medium, and the mixture was poured on top of the PDA plates. If the isolated strain possessed the ability to against Pbr, the plate would change from white to transparent. The suppressive activity test of strain ZF390 against different kinds of phytopathogenic bacteria was similar to the aforesaid method. The transparent aperture appeared on the plate if the strain ZF390 had the capacity to defend against the pathogen. In addition, the inhibitory effect of strain ZF390 against various plant pathogenic fungi was assessed through plate assays. The fungus dishes with diameters of 5 mm were inoculated on the center of PDA plates, and 5 µL of ZF390 suspension was diluted to an OD 600 = 0.6 with sterile LB broth that was inoculated at the spots 3.5 cm away from the dishes. Fungal hyphae were inhibited after the plates were cultured at 28 • C for 7-10 d if strain ZF390 could defend against the phytopathogenic fungi.
The control efficacy of antagonistic strains against cucumber bacterial soft rot disease was assessed by greenhouse experiments with potted cucumbers. Each treatment contained five healthy cucumber seedings sprayed with 200 mL suspension of Pbr diluted to an OD 600 = 0.8 with sterile NB broth. The greenhouse was kept at 28 • C and 75% relative humidity for 24 h. Whereafter, 3 mL antagonistic strain suspension was subjected to 100-fold dilution with sterile water and spray inoculated on the cucumber seedings. The chemical pesticide group was treated with 300 mL of 3% Zhongshengmycin wettable powder 650-fold dilution, and the 300 mL sterile water served as a control group. The disease grade was divided into five grades: absence of spots represented grade 0; grade 1 was classified as a spot area less than 5% of leaf area; the proportion of spot area from 6% to 25% was grade 2; the proportion of spot area from 26% to 50% was grade 3; the proportion from 51% to 75% was grade 4; and the spot area larger than 76% was grade 5. Disease index = 100 × ∑ (the number of leaves in certain disease grade × relative disease grade)/(total number of leaves × the highest grade of disease). Control effect (%) = 100 × (disease index of control group-disease index of treatment group)/disease index of control group.
In addition, the activities of cellulase, protease, and phosphatase were measured using previously reported methods [25,26]. If cellulase was produced, the CMC agar plate would change from red to yellow after staining with 0.1% (w/v) Congo red solution for 30 min and washing with 1 M NaCl solution for 15 min. While, if protease and phosphatase were produced, haloes in plates became visible without further treatment. The heavymetal tolerance of strain ZF390 was analyzed using a standard 10-fold dilution plating assay. Different concentrations of ZF390 suspension were inoculated on the LB plates supplemented with 0.4 mM CdCl 2 ·2 1 / 2 H 2 O, 1 mM CoCl 2 ·6H 2 O, 3 mM MnSO 4 ·H 2 O, 5 mM PbCO 3 , and 3 mM ZnCl 2 , respectively. Additionally, the plates were incubated at 28 • C for 48 h. All of the above experiments were repeated three times.

Characteristics, Microscopic Analysis and Genomic DNA Extraction
The growth characterization and physiological and biochemical characteristics were determined on the basis of reported approaches [27,28]. The strain morphologies were observed by transmission electron microscopy (TEM) and scanning electron microscopy (SEM). Genomic DNA was extracted from the cultured ZF390 cells using the TIANamp Bacteria DNA Kit (TIANGEN BIOTECH, Beijing, China).

Phylogenetic Analysis and Genome Comparisons
The taxonomic position of strain ZF390 was determined using multilocus gene sequence analysis (MLSA) based on eight housekeeping genes (gyrB, rpoD, rplA, atpD, holB, rho, dnaA and rpsA). The gene sequences were aligned using MUSCLE and trimmed to remove ambiguously aligned regions (the final sequence lengthes were 9910 bp). Subse-Biology 2022, 11, 1172 4 of 21 quently, the phylogenetic tree was constructed using the maximum likelihood method in MEGA 7.0 [42]. Other available gene sequences of the Paenibacillus strains to be used for phylogenetic tree construction were downloaded from the NCBI database. According to the phylogenetic analysis, four closely related Paenibacillus strains with released complete genome sequences, including P. peoriae HS311 (GenBank Accession No. CP011512.1), P. polymyxa HY96-2 (GenBank Accession No. CP025957.1), P. polymyxa SQR-21 (GenBank Accession No. CP006872.1), and P. kribbensis PS04 (GenBank Accession No. 041731.1), were selected for genome comparison with ZF390. Average nucleotide identities (ANI) and in silico DNA-DNA hybridization (isDDH) analyses were performed using the OrthoANIu algorithm and the Genome-to-Genome Distance Calculator (GGDC), respectively [43]. Moreover, the pairwise alignment of genomes was generated using the Mauve v2.3.1 (SourceForge, San Diego, CA, USA) and Venn diagrams were constructed using the R package [44].

Analysis of Gene Clusters for Biosynthesis of Secondary Metabolites
The secondary metabolite gene cluster prediction was performed using antiSMASH 5.0 on the authors' Web servers using the default parameters [45]. Comparative analyses of secondary metabolite gene clusters identified that, among strain ZF390, P. peoriae HS311, P. polymyxa HY96-2, P. polymyxa SQR-21, and P. kribbensis PS04 were performed based on the Kyoto Encyclopedia of Genes and Genomes database (KEGG) [35] and the GenBank database.

Genome Mining for Genes Encoding Plant Beneficial Traits
The functional genes involved in plant-growth promotion or plant-bacterial interactions, such as the genes responsible for plant-growth promotion, biofilm formation, and induced systemic resistance, were searched in the NCBI and KEGG databases. The identities of different functional genes at the amino acid level were compared between strain ZF390, P. peoriae HS311, P. polymyxa HY96-2, P. polymyxa SQR-21, and P. kribbensis PS04 using BLAST (Basic Local Alignment Search Tool).

Statistical Analysis
All data were analyzed using the analysis of variance in SPSS v24.0 (IBM, Armonk, NY, USA). The significant differences between means were compared by using the LSD test (Fisher's protected least significant differences test) at p = 0.05. A p value < 0.05 was considered significant.

Biocontrol Effect of Strain ZF390 against Pbr
To identify good biocontrol agents against cucumber soft rot disease caused by Pbr, 161 bacterial strains were segregated from the rhizosphere of cucumber plants. There were 14 strains that showed antagonistic activity against Pbr through plate bioassays (Supplementary Figure S1). The largest transparent halo was generated around the colony of strain ZF428, while only a small ring was produced around the colony of strain ZF390 (Supplementary Figure S1). The control effects of these 14 strains against Pbr were performed by greenhouse experiments on cucumber seedlings. Among these bacteria, the strain ZF390 indicated superb antagonistic activity in greenhouse tests, exhibiting the highest control efficiency of 77.47%, which was much higher than the efficacy of zhongshengmycin (47.89%) [46], which is a ubiquitous aminoglycoside antibiotic against bacterial phytopathogens (Table 1 and Figure 1). strain ZF390 indicated superb antagonistic activity in greenhouse tests, exhibiting the highest control efficiency of 77.47%, which was much higher than the efficacy of zhongshengmycin (47.89%) [46], which is a ubiquitous aminoglycoside antibiotic against bacterial phytopathogens (Table 1 and Figure 1).  The strain ZF390 also demonstrated prominent antagonistic activities against various plant bacterial and fungal pathogens, including Pbr, X. campestris pv. Campestris, Pseudomonas syringae pv. Tomato, Clavibacter michiganensis subsp. Michiganensis, P. syringae pv. Lachrymans, C. michiganensis subsp. Sepedonicum, Agrobacterium vitis, Acidovorax citrulli, Corynespora cassiicola, F. oxysporum, Colletotrichum capsici, P. capsici, Stemphylium solani, Ascochyta citrulline, Botrytis cinerea, and Rhizoctonia solani in the plate bioassays ( Figure 2). In addition, yellow and transparent halos were observed in extracellular enzyme assays, indicating that strain ZF390 produced cellulase, protease, and phosphatase (Supplementary Figure S2A-C), which might be one of its biocontrol mechanisms. Moreover, the strain ZF390 revealed varying degrees of tolerance to Cd 2+ , Co 2+ , Mn 2+ , Pb 2+ , and Zn 2+ (Supplementary Figure S2D). Strain ZF390 exhibited a superior colonization ability in the The strain ZF390 also demonstrated prominent antagonistic activities against various plant bacterial and fungal pathogens, including Pbr, X. campestris pv. Campestris, Pseudomonas syringae pv. Tomato, Clavibacter michiganensis subsp. Michiganensis, P. syringae pv. Lachrymans, C. michiganensis subsp. Sepedonicum, Agrobacterium vitis, Acidovorax citrulli, Corynespora cassiicola, F. oxysporum, Colletotrichum capsici, P. capsici, Stemphylium solani, Ascochyta citrulline, Botrytis cinerea, and Rhizoctonia solani in the plate bioassays ( Figure 2). In addition, yellow and transparent halos were observed in extracellular enzyme assays, indicating that strain ZF390 produced cellulase, protease, and phosphatase (Supplementary Figure S2A-C), which might be one of its biocontrol mechanisms. Moreover, the strain ZF390 revealed varying degrees of tolerance to Cd 2+ , Co 2+ , Mn 2+ , Pb 2+ , and Zn 2+ (Supplementary Figure S2D). Strain ZF390 exhibited a superior colonization ability in the rhizospheres of cucumbers cultivated in fermentable substrate after 15 days of inoculation (1.36 × 10 8 CFU/g) (Supplementary Figure S3). In summary, due to the remarkable biocontrol traits of strain ZF390, it was selected for further research.  Figure S3). In summary, due to the remarkable bioco trol traits of strain ZF390, it was selected for further research.

Organism Information
Strain ZF390 was demonstrated to be a motile, Gram-positive, and facultative anae obic bacterium belonging to the Paenibacillaceae family. Strain ZF390 grew readily on L plates at 28 °C and produced light yellow sticky colonies with irregular margins after 3 h of incubation (Supplementary Figure S4A), and the strain produced rod-shaped cel with a length of 2-4 μm and a diameter of 0.5-1 μm (Figure 3 and Supplementary Figu S4B). As indicated in the physiological and biochemical assays, casein was hydrolyzed b strain ZF390, but arginine dihydrolase and indole were not produced, and the nitrate r duction reaction was negative. Interestingly, the capability of strain ZF390 to liquefy ge atin was negative, which was the complete opposite of that of P. peoriae KCTC 3763 T [28 Acid was produced by ZF390 from the following compounds: glycerol, D-arabinose, arabinose, ribose, D-xylose, adonitol, methylxyloside, D-fructose, mannitol, methyl-D mannoside, salicin, maltose, D-trehalose, gentiobiose, 2-keto-D-gluconate, 5-keto-D-gl conate, rhamnose, D-turanose, and N-acetylglucosamine, and its capacity of utilizing D turanose was different from that of KCTC 3763 T (Supplementary Table S1).

Organism Information
Strain ZF390 was demonstrated to be a motile, Gram-positive, and facultative anaerobic bacterium belonging to the Paenibacillaceae family. Strain ZF390 grew readily on LB plates at 28 • C and produced light yellow sticky colonies with irregular margins after 36 h of incubation (Supplementary Figure S4A), and the strain produced rod-shaped cells with a length of 2-4 µm and a diameter of 0.5-1 µm (Figure 3 and Supplementary Figure S4B). As indicated in the physiological and biochemical assays, casein was hydrolyzed by strain ZF390, but arginine dihydrolase and indole were not produced, and the nitrate reduction reaction was negative. Interestingly, the capability of strain ZF390 to liquefy gelatin was negative, which was the complete opposite of that of P. peoriae KCTC 3763 T [28]. Acid was produced by ZF390 from the following compounds: glycerol, D-arabinose, L-arabinose, ribose, D-xylose, adonitol, methylxyloside, D-fructose, mannitol, methyl-D-mannoside, salicin, maltose, D-trehalose, gentiobiose, 2-keto-D-gluconate, 5-keto-D-gluconate, rhamnose, D-turanose, and N-acetylglucosamine, and its capacity of utilizing D-turanose was different from that of KCTC 3763 T (Supplementary Table S1).

General Genomic Features of P. peoriae ZF390
The complete genome of P. peoriae ZF390 was composed of a circular 6,193,667 bp chromosome with three additional plasmids, pPlas1 (137,220 bp), pPlas2 (

General Genomic Features of P. peoriae ZF390
The complete genome of P. peoriae ZF390 was composed of a circular 6,193,667 bp chromosome with three additional plasmids, pPlas1 (137,220 bp), pPlas2 (16,869 bp), and pPlas3 (36,233 bp), and the average GC contents were 44.96%, 45.58%, 45.12%, and 41.66%, respectively (Supplementary Table S2). In total, 5890 open reading frames (ORFs) were forecasted in the genome of ZF390. In addition to 5574 protein-coding genes (CDSs), the genome constituted 145 RNA genes, containing 40 rRNA operons, 101 tRNA genes, 4 other RNAs, and 171 pseudogenes (Supplementary Table S3). In addition, the ZF390 genome included 17 genomic islands, 5 prophage regions, and 11 CRISPR-associated genes (Supplementary Table S2). A total of 4001 functional genes of the ZF390 genome were categorized using the Clusters of Orthologous Groups of proteins (COG) database. Most of the genes were related to carbohydrate transport and metabolism (505 genes), followed by transcription (482 genes), general function prediction only (451 genes), amino acid transport and metabolism (355 genes), and signal transduction mechanisms (304 genes). However, 439 genes were not predicted by the COG database, and their specific functions need to be further verified (Supplementary Table S4). The graphical circular genomic maps exhibiting the genome structure and functions of P. peoriae ZF390 are presented in Figure 4. Moreover, 5402 genes were annotated to 204 different KEGG pathway maps, the majority of which were associated with metabolic pathways, biosynthesis of secondary metabolites, ABC transporters, the biosynthesis of antibiotics, microbial metabolism in diverse environments, two-component systems, the biosynthesis of amino acids, and quorum sensing. In addition, 3754, 5645, 481, 3754, 2296, and 342 ORFs were compared to diverse pathways or domains using the GO, NR, TCDB, Pfam, Swiss-Prot, and CAZy databases (Supplementary Table S2).

Comparative Genomics Analysis of P. peoriae ZF390 with Other Paenibacillus Strains
To comprehend the genetic relationship of P. peoriae ZF390 with other Paenibacillus species, sixteen publicly available complete genome sequences of Paenibacillus-type strains were selected to establish a phylogenetic tree based on eight housekeeping genes (gyrB, rpoD, rplA, atpD, holB, rho, dnaA, and rpsA) (Supplementary Table S5). As expected, all of the strains were prominently corroborated by bootstrap values. According to the genetic distance, strain ZF390 was closest to P. peoriae KCTC 3763 T , followed by P. polymyxa DSM 36 T and P. kribbensis AM49 T ( Figure 5). The phylogenetic analysis determined an accurate systematic position of ZF390 as P. peoriae.  Table S5. The red star marks the location of strain ZF390.
ANI and isDDH are widely applied analyses that appraise the similarities between bacterial strains by comparing whole genomic sequence data. Pairwise compared strain with ANI values ≥ 96% and isDDH values ≥ 70% are typically considered the same specie [47,48]. In this study, the ANI and isDDH values between strains ZF390 and HS311 were 96.61% and 71.80%, respectively (Supplementary Figure S5). Simultaneously, the ANI val ues were over 96%, and the isDDH values were over 70%, when comparing strain ZF390 with P. polymyxa strains ATCC 15970, CR1, E681, J and YC0573. Nevertheless, lower AN and isDDH values were obtained when other P. polymyxa strains were used as reference genomes. These findings may suggest inaccurate identities of partial P. polymyxa strains In addition, the ANI values that contrasted strain ZF390 with P. kribbensis strains AM49 and PS04 were 87.46% and 87.41%, respectively, and the isDDH values were 33.70% and 33.80%, respectively. Altogether, these results revealed that P. polymyxa and P. kribbensi were in more appressed relationships with P. peoriae than other Paenibacillus species. Strain ZF390 was closely related to P. peoriae HS311 and could be assigned to the same taxonomi position (Supplementary Figure S5).
Genomic information was compared between P. peoriae ZF390 and four other acces sible complete genomes of Paenibacillus strains (P. peoriae HS311, P. polymyxa SQR-21, P polymyxa HY96-2 and P. kribbensis PS04) (Supplementary Table S3  ANI and isDDH are widely applied analyses that appraise the similarities between bacterial strains by comparing whole genomic sequence data. Pairwise compared strains with ANI values ≥ 96% and isDDH values ≥ 70% are typically considered the same species [47,48]. In this study, the ANI and isDDH values between strains ZF390 and HS311 were 96.61% and 71.80%, respectively (Supplementary Figure S5). Simultaneously, the ANI values were over 96%, and the isDDH values were over 70%, when comparing strain ZF390 with P. polymyxa strains ATCC 15970, CR1, E681, J and YC0573. Nevertheless, lower ANI and isDDH values were obtained when other P. polymyxa strains were used as reference genomes. These findings may suggest inaccurate identities of partial P. polymyxa strains. In addition, the ANI values that contrasted strain ZF390 with P. kribbensis strains AM49 T and PS04 were 87.46% and 87.41%, respectively, and the isDDH values were 33.70% and 33.80%, respectively. Altogether, these results revealed that P. polymyxa and P. kribbensis were in more appressed relationships with P. peoriae than other Paenibacillus species. Strain ZF390 was closely related to P. peoriae HS311 and could be assigned to the same taxonomic position (Supplementary Figure S5).
Genomic information was compared between P. peoriae ZF390 and four other accessible complete genomes of Paenibacillus strains (P. peoriae HS311, P. polymyxa SQR-21, P. polymyxa HY96-2 and P. kribbensis PS04) (Supplementary Table S3). The genome size of strain ZF390 (6.38 Mb) was conspicuously large compared to the other four strains, HS311 (6.22 Mb), SQR-21 (5.83 Mb), HY96-2 (5.75 Mb), and PS04 (5.74 Mb). These results also illustrate a maximum number of ORFs and CDSs in the ZF390 genome. Nevertheless, the order of average GC content of these five selected strains was antipodal with the order of genome size results; strain ZF390 had the lowest GC content (44.99%), under 45%. Furthermore, P. peoriae ZF390 and HS311 comprised three plasmids and one plasmid, respectively, and the other three Paenibacillus strains merely contained a single chromosome without a plasmid (Supplementary Table S3).
To estimate the evolutionary distance among these five Paenibacillus strains, their whole genome sequences were compared using Mauve ( Figure 6A). The alignments between ZF390 and HS311 showed that several gene inversions and a large region of deletion were detectable in P. peoriae ZF390. Compared to SQR-21 and HY96-2, a number of gene insertions or deletions and local collinear block (LCB) inversions were also detected in strain ZF390. These results show that large LCB inversions occurred among strain PS04 and the other four strains, and the ZF390 genome is much more similar to that of HS311, supporting the relationship described above ( Figure 6A). Whole genome sequences of these five Paenibacillus strains were also compared to identify specific genes in P. peoriae ZF390 ( Figure 6B). As shown in Figure 6B, there were 3372 core genes present in ZF390 that were shared with strains HS311, SQR-21, HY96-2, and PS04. It was speculated that these orthologous protein-coding genes were relatively conserved in Paenibacillus. Additionally, 4122 orthologous genes were shared between ZF390 and HS311, while ZF390 shared 3912 and 3889 orthologous genes with SQR-21 and HY96-2, respectively. Moreover, strain ZF390 shared merely 3652 orthologous genes with PS04, whereas it indicated a higher level of similarity with strain HS311. Furthermore, only two and three unique genes were identified in the genomes of strains ZF390 and PS04, which revealed close relationships among the five selected strains.

Gene Clusters Involved in the Synthesis of Secondary Metabolites
With the aim of identifying the antibiotics that may be produced by ZF390, with its broad antagonistic spectrum, the gene clusters for antibiotic biosynthesis were detected in the ZF390 genomic sequence using the antiSMASH database. Ten gene clusters with different biosynthesis pathways involved in secondary metabolite production were retrieved, including three clusters encoding NRPSs (nonribosomal peptide synthetases), one T1PKS (type I polyketide synthases) or NRPS, one TransAT-PKS (trans-acyl transferase polyketide synthetases) or NRPS, one lassopeptide, one siderophere, one lanthipeptide-class-i, one RiPP-like (ribosomally synthesized and post-translationally modified peptides) protein, and an NRPS-like protein ( Table 2). Gram-negative bacteria can be suppressed by brevicidine, tridecaptin, paeninodin, and octapeptin while paenilan, fusaricidin B, paeninodin, and octapeptin control Gram-positive bacteria [49][50][51][52][53][54][55]. Meanwhile, fusaricidin B, octapeptin, and paeninodin show an outstanding inhibitory effect against fungal pathogens [49,54,55]. The bioactive spectrum of paenilipoheptin and three unknown antibiotics found in the genome is still unclear ( Table 2). The comparison analysis of the above five strains revealed a deficiency of the paenilan synthesis gene cluster in strain HS311. An unknown antibiotic synthesized via the siderophore pathway was not retrieved in SQR-21 and HY96-2. The genome of PS04 did not contain the relevant clusters of brevicidine, octapeptin, and paenilan. Interestingly, however, the gene cluster of paenilipoheptin identified in strain PS04 showed poor similarity with the clusters retrieved in strains ZF390, HS311, SQR-21, and HY96-2 ( Figure 7). Furthermore, the gene cluster for the biosynthesis of octapeptin detected in the ZF390 genome had a larger core gene cluster (Figure 7).

Mining for Functional Genes Potentially Associated with Plant-Bacteria Interactions
The key genes related to plant-growth promotion, biofilm formation, and induced systemic resistance were retrieved from P. peoriae ZF390 based on the KEGG database, and the identities of these genes among HS311, SQR-21, HY96-2, and PS04 were compared according to the NCBI database. The identities of all homologous sequences between strain ZF390 and the other four strains were higher than 77%. The sequence identities of strain ZF390 exhibited very high similarities with those in strain HS311; concurrently, lower similarities were observed when comparing ZF390 and PS04 (Supplementary Tables S6-S8).
Five key genes related to phosphate solubilization (phnE, phnC, pstB, pstA, and pstC) and 18 other genes responsible for organic acid biosynthesis were all retrieved from the five Paenibacillus strains with high similarity (87.07-99.21%) (Supplementary Table S6). The results demonstrated a latent capacity of ZF390 to promote plant growth. As data shown in Supplementary Table S7 indicate, genes involved in biofilm formation, such as flagellar assembly, non-ribosomal peptide structures, bacterial chemotaxis, and quorum sensing were mined in the genome sequences of five Paenibacillus strains. Flagellar structure coding genes, including fli gene cluster (fliACEFGHIJKLMNOPQRY), flg gene cluster (flgBCDE-FGKLM), gene flhA and gene flhB, and flagellar motivation coding genes motA and motB, were located in the genome of strain ZF390. However, gene fliD and gene fliS were not found in ZF390 genomic sequence. Genes (fenA, fenB, and fenC) related to Fengycin production were also retrieved. The che gene cluster (cheABCDRWXY) and multiple copies of gene mcp were found in five genomes, and the sequence identity exceeded 84.03%. The key gene in AI-2 QS system, luxS, and the four genes (spo0A, spo0B, spo0F, and degU) involved in biofilm formation pathway were found in genome of strain ZF390, and the sequence identity between ZF390 and HS311 (97.64%) exhibited higher values than those between strain ZF390 with SQR-21 (92.42%), HY96-2 (92.42%), or PS04 (91.06%). Key genes related to the plant-resistance inducers in ZF390 were retrieved in light of previous studies, and the identities of these genes among HS311, SQR-21, HY96-2, and PS04 were compared [7,56]. The results demonstrate that the key genes associated with volatile organic compounds belong to systemic resistance inducers, such as 2,3-butanediol, methanethiol, and isoprene synthesis, were all detected in the genomes of the five strains, with sequence identities exceeding 85% (Supplementary Table S8).

Discussion
Pbr has caused a devastating soft rot disease of cucumbers in China, and chemical agents did not effectively control it; so, biocontrol has been given higher expectations and attention [4]. Chemical methods are commonly used to control soft rot but are not desirable, due to increased resistance and environmental contamination [57,58]. The biocontrol of soft rot through biocontrol strains is one of the safe, effective, and sustainable methods and research hotspots used to control soft rot [59][60][61]. Abundant biocontrol agents were screened through in vitro inhibition tests, for instance, estimating the inhibition of a target pathogen on agar tablets; however, the bottom line for biocontrol is whether it works under production conditions, and numerous researchers have reported no correlation between in vitro inhibition tests and field performance [62].
In this study, P. peoriae ZF390 showed a conspicuous alleviation effect on cucumber plants infected with soft rot disease in a greenhouse, representing the good potential of biopesticides or microbial inoculants, even though its effect on the plate assays was not particularly noteworthy (Figure 1 and Supplementary Figure S1). In addition, the antimicrobial properties of strain ZF390 were investigated, demonstrating antagonistic activity against a variety of bacteria and fungi (Figure 2), and the results found were similar to those of previous studies [63,64]. Extracellular enzymes, such as protease (Prt) and cellulase (Cel), produced by strain ZF390 might result in antagonistic action against a wide range of microorganisms (Supplementary Figure S2A,B). Due to its superior antimicrobial activity, the genome of P. peoriae ZF390 was completely sequenced to provide insight into its biocontrol mechanisms. Metal-tolerant bacteria could survive in the tissue of plants grown on the metal-polluted soil and perform good adaptability to the environment [65]. Strain ZF390 might be used in a wide range of conditions due to its tolerance to Cd 2+ , Co 2+ , Mn 2+ , Pb 2+ , and Zn 2+ (Supplementary Figure S2D). P. peoriae was originally recognized as Bacillus peoriae and reclassified as a member of the genus Paenibacillus after performing an amplified ribosomal DNA restriction analysis, a fatty acid methyl ester analysis, sodium dodecyl sulfate-polyacrylamide gel electrophoresis of whole-cell proteins, API, and other routine phenotypic tests [28,66]. Due to its extreme uniformity, 16S rDNA is rarely recommended for examining the relationships between genera or even deeper taxonomic levels, while multilocus sequence typing (MLST) with a sample of between four-and eight-core gene loci is sufficient for the subdivision of most bacterial species [67]. However, there is a general consensus among taxonomists that all taxonomic information about a bacterium is incorporated into the complete nucleotide sequence of its genome [68]. The average nucleotide identity (ANI) and DNA-DNA hybridization (DDH) are considered significant criteria for species identification. The commonly accepted ANI criteria to set bounds on species is generally 95% [69,70], and the critical value of DDH is typically 70% [71]. Nevertheless, an ANI threshold between 93% and 96% can be considered a fuzzy zone where the intra-genus and interspecies boundary of may fall; the definition of the optimal ANI range for prokaryotic species is still controversial [72]. Previous studies corroborated that the ≥95% ANI cutoff alone is unable to resolve closely related species, and the stringent ≥96% ANI cutoff is more accurate with a better correlation with the 70% DDH cutoff [47,73]. It is highlighted that species assignment based on a single algorithm may not be robust enough.
In silico DNA-DNA hybridization (isDDH), determined using the Genome-to-Genome Distance Calculator (Formula 2), is particularly attractive, as it has been shown to be better correlated to the traditional DDH, which is a gold standard for species assignment, than to ANI [43]. Therefore, it is proposed that any genome-based species assignments should be determined by both ANI and isDDH, and the pairwise ANI with a ≥96% cutoff and isDDH with a ≥70% cutoff have been widely used for precise species identification [47,48]. Based on phenotypic, phylogenetic, and genotypic data, P. peoriae ZF390 was subjected to a taxonomic study by using a polyphasic approach. The phylogenetic tree based on eight housekeeping genes revealed an intimate but independent relationship among P. peoriae, P. polymyxa, and P. kribbensis ( Figure 5). The ANI and isDDH values for strains ZF390 and HS311 were 96.61% and 71.80%, respectively, which were considerably higher than the values for ZF390 with the selected P. polymyxa and P. kribbensis strains (Supplementary Figure S5).
In contrast to the other four strains, the genome of ZF390 was larger, comprising more plasmids, but with a slightly smaller GC content, and abundant CDSs were predicted (Supplementary Table S3). The analysis of the whole ZF390 genome revealed an impressive capability to produce a diverse spectrum of different secondary metabolites aimed at suppressing harmful phytopathogens. In total, 10 gene clusters (Table 2) representing more than 8.7% of the genome were devoted to synthesizing antimicrobial metabolites, approximately the same proportion as the distinguished biocontrol agent B. amyloliquefaciens FZB42 [74].
On the basis of antiSMASH database, comparisons of the genes and gene clusters related to antibiotic synthesis demonstrated differences among the five Paenibacillus strains (Table 2 and Figure 7). Regarding the control of bacteria, five antibacterial metabolites (brevicidine, tridecaptin, paenilan, fusaricidin B, paeninodin, and octapeptin) were predicted to be produced by P. peoriae ZF390. Brevicidine, a bacterial non-ribosomally-produced cyclic lipododecapeptide, which has low hemolytic activity and cytotoxicity, displays potent antimicrobial activity against Gram-negative pathogens [75]. Brevicidine B was discovered from the producer strain of brevicidine, Brevibacillus laterosporus DSM 25. Recently, researchers found that brevicidine B, compared with an earlier reported member of the brevicidine family, has a broadened antimicrobial spectrum caused by its increased membrane-disruptive capacity against Gram-positive pathogens [52]. The strong antibacte-rial activity of strain ZF390 might be associated with the production of brevicidine; however, its capability of producing brevicidine B should be further investigated. Tridecaptin is another type of non-ribosomal antibacterial peptide (NRAP) with effective activity against highly problematic strains of Gram-negative bacteria and low levels of resistance development with the potential for safe administration [76]. All five Paenibacillus strains were found to contain tridecaptin-related gene clusters, but more core biosynthetic genes were predicted in SQR-21 and HY96-2 than in the other three strains (Figure 7). With purification and structure analysis, paenilan was identified as a novel class-I antibiotic, a ribosomally synthesized peptide that exhibits antimicrobial activity against Gram-positive bacteria and is well conserved in different Paenibacillus sp. isolated from globally distant places [53]. The core genes epiB and epiC and biosynthesis cluster were observed in the genomes of ZF390, SQR-21, and HY96-2, while no genes or gene clusters involved in the synthesis of paenilan were found on the chromosomes of HS311 and PS04 ( Table 2). Fusaricidin B, paeninodin, and octapeptin not only possess the antibacterial activity, but also impede the growth of phytopathogenic fungi. According to former reports on its particularly antagonistic activity against Gram-positive bacteria and plant pathogenic fungi, fusaricidin, generated by a non-ribosomal peptide synthetase (NRPS), seems to have immense potential for industrial use [77,78]. Fusaricidin B that was possibly produced by strain ZF390 reported excellent germicidal activity against Candida albicans and Saccharomyces cerevisiae [54]. Octapeptins, which are structurally similar to polymyxins, are lipopeptide antibiotics possessing much broader antimicrobial activity against Gram-negative and Gram-positive bacteria, yeast, fungi, and even protozoa, whereas polymyxins are only active against Gram-negative bacteria [55]. The core gene region belonging to the synthesis gene cluster of octapeptin predicted in strain ZF390 was much larger than that in HS311, SQR-21, and HY96-2; the octapeptin cluster was not found in strain PS04 (Figure 7). Paeninodin, a new lasso peptide, belongs to the ribosomally synthesized and post-translationally modified peptides (RiPPs) that exhibit a wide range of antimicrobial or antiviral activities, and the synthesis cluster was found in all five strains [79]. Paenilipoheptin was recently identified via the genome mining of P. polymyxa E681, and its antimicrobial activity spectrum remains to be elucidated [23,80].
Bacteria belonging to the genus Paenibacillus have been isolated from a variety of environments, and the majority of them are found in soil, demonstrating benefits for plants with mutual interactions. They are considered predominant plant-growth-promoting rhizobacteria (PGPR) and are widely used for agricultural applications [81,82]. Phosphorus is a significant macronutrient for plant nutrition, but its unique characteristics of high fixation and poor solubilization in soil lead to the overuse of chemical phosphorus fertilizers applied to agricultural land, causing severe environmental damage [83,84]. By inoculating fields with phosphorus-solubilizing microorganisms, such as Paenibacillus, the application of chemical phosphorus fertilizer could be reduced, and the principal mechanism of microorganisms to make phosphorus available to plants could be performed by producing organic acids [85]. A differential gene expression study revealed an extremely intricate transcriptional response of P. sonchi SBR5 to soluble and insoluble phosphorus sources, suggesting that the production of organic acids might be the most vital strategy utilized by SBR5 to perform phosphorus solubilization [86]. An evident and clear dissolution halo can be observed around the P. polymyxa inoculum grown on a solid NBRIP medium, in which Ca 3 (PO 4 ) 2 is the sole phosphate source [87]. According to the comparative genomic analysis, 23 genes related to phosphorus solubilization were retrieved in the ZF390 genome, some of which encode organic acid production (Supplementary Table S5). However, the main genes (gcd and gad) responsible for the production of gluconic acid and its conversion were not detected in the genome of ZF390. Similarly, ZF390 does not carry the complete phosphonate gene cluster (phn), except for phnC and phnE, which were deemed a gene gain and loss event during evolution [88]. In addition, the ZF390 genome possesses the pst operon (pstC, pstA, and pstB) that integrates the phosphate-specific transport system. Taken together, P. peoriae ZF390 is a potential PGPR that may become an additive in biofertilizer.
Endospore-forming bacteria such as Paenibacillus have the capacity to survive for long periods in the soil under adverse environmental conditions, which is appropriate for field inoculation [85]. PGPR plays a crucial role in plant growth promotion by effectively colonizing plant roots. Bacteria in natural environments persist by forming biofilms, which are highly structured, surface-attached communities of cells encased in a self-produced extracellular matrix [89]. A series of genes related to flagellar assembly, non-ribosomal peptide structures, bacterial chemotaxis, and quorum sensing play increasingly important roles in biofilm formation. Paenibacillus performs two forms of flagellar-driven motility: swarming on a surface and swimming in liquid culture; swarming motility is crucial during root colonization or for rapid colonization of a specific tissue [20,90]. A threshold amount of power is needed for an organism to swarm, and hyper-flagellation, a dramatic augmentation in flagellum density, may contribute to a cumulative increase in power [90]. According to the KEGG database and comparative genomics results, a complete flagellar assembly gene cluster was revealed in the ZF390 genome, including, but not limited to, the fliC gene encoding the filament, the flgE, flgK, and flgL genes encoding the hook, and multitudinous genes encoding the basal structure [91,92]. Recent research reported that P. polymyxa generates energy through the mechanism of degrading phospholipids to create sn-glycerol 3-phosphate and feed it into the glycolysis/pentose phosphate pathway, which appears likely to be necessary for swarming motility, including flagellar rotation [90]. However, the applicability of this mechanism in ZF390 needs to be further explored. In addition to multicellular rafts of highly flagellated cells, the lipopeptide surfactants or non-ribosomal peptides secreted by bacteria act as wetting agents, providing a mechanism to reduce surface tension between the substrate and the bacterial cells, helping them spread rapidly over surfaces [90,[93][94][95]. Except for the non-ribosomal peptide biosynthesis gene clusters identified in the antiSMASH database, several genes belonging to fengycin synthesis were also detected in the genome of ZF390. The presence of these substances distinctly provides the essential requirement for biofilm swarms on plant roots.
Bacterial chemotaxis, a biased movement in chemical gradients, is a motility-based response that generally enhances the efficiency of environmental colonization by motile bacteria [96]. Chemoreceptors called methyl-accepting chemotaxis proteins (MCPs) detect various environmental signals to stimulate the chemotactic response [97]. The basic molecular mechanism of chemotaxis is generally conserved in bacteria, and it was systematically studied in Escherichia coli, revealing only a single chemotaxis operon in its genome [96,98,99]. Nevertheless, multiple parallel chemotaxis systems and a large number of chemoreceptors have been found in motile bacterial species, which is likely to provide a competitive advantage for surviving in the soil, as well as competition for root-surface colonization [100]. We found multiple copies of genes, especially the mcp gene, involved in chemotaxis in the genome of ZF390. Thus, we propose that chemotaxis may play a crucial role in the colonization of ZF390.
Moreover, bacterial biofilm formation is often regulated by quorum sensing (QS), which is a cell-communication mechanism that depends on population density [101]. To date, two typical QS systems using an autoinducing peptide (AIP) or an autoinducer-2 (AI-2) as a signal molecule have been reported in Gram-positive bacteria [102]. The luxS gene, a key regulatory gene of the AI-2-mediated QS system, has been confirmed to impact the biomass and morphology of biofilms formed by Paenibacillus species. In P. polymyxa HY96-2, luxS improved biocontrol efficacy by promoting its biofilm formation ability [103]. On the basis of recent studies, we speculate that luxS can positively regulate biofilm formation, but this assumption needs more experimental arguments [103,104]. The biofilm formation pathway genes spo0A, spo0B, apo0F, and degU were found in the ZF390 genome when searching for QS genes, suggesting that two pathways may be involved in ZF390 [56]. From the above results, it is possible to deduce that ZF390 could colonize well in the rhizosphere by forming biofilms, but the colonization abilities of the five strains might differ because the variety, amount, and expression levels of these genes related to biofilm formation are different.
Many beneficial rhizobacteria including members of Paenibacillus can trigger ISR when present in sufficiently high population densities [85]. Systemic resistance inducers, such as volatile organic compounds (VOCs), including methanethiol, isoprene, 2,3-butanediol and butyl acetate, are generated and released into the surrounding environment by biocontrol agents, offering protection against phytopathogens and insect herbivores [85]. P. polymyxa E681 was interpreted to use VOC elicitors to protect Arabidopsis thaliana plants against the bacterium P. syringae pv. maculicola via the primed transcription of salicylic acid, jasmonic acid, and ethylene-signaling genes [105]. In previous studies, key genes associated with VOC production were investigated in P. polymyxa strains HY96-2, ZF129, and ZF197 [56,88]. The genes responsible for inducers (2,3-butanediol, methanethiol and isoprene) in strains ZF390 and HS311 exhibited 93% identity, which is higher than the identity observed among strains SQR-21, HY96-2, and PS04. These results deduced a similar capability of systemic resistance in these five strains with varying effectiveness owing to the variations in the related genes.

Conclusions
P. peoriae ZF390, which was obtained from cucumber rhizosphere soil, exhibited an extraordinary effect on controlling cucumber soft rot caused by Pbr and exhibited broadspectrum antagonistic activities against 16 distinct bacterial and fungal phytopathogens. Based on the polyphasic analyses combined with phenotypic, phylogenetic, and genotypic data, strain ZF390 was identified as P. peoriae, which is closely related to P. polymyxa and P. kribbensis. According to the antiSMASH database, strain ZF390 possesses miscellaneous secondary metabolite biosynthesis clusters with biocontrol functions, including brevicidine, fusaricidin B, octapeptin, paenilan, paenilipoheptin, paeninodin, tridecaptin, and three other unknown antibiotics, demonstrating preeminent antimicrobial activity. Based on the comparative genomic analysis and phosphate solubilization assay, strain ZF390 may perform as a potential PGPR. In addition, many genes associated with colonization, such as flagellar assembly, bacterial chemotaxis, quorum sensing, and non-ribosomal peptide structures, were detected in the ZF390 genome and shared high homology with P. peoriae strain HS311. Overall, these properties indicate that ZF390 is a promising biocontrol agent for further application in agricultural fields.  Figure  S2: Plate assays for production of cellulase (A), protease (B) and phosphatase (C) by Paenibacillus peoriae ZF390 and heavy metal resistance of strain ZF390 (D). Figure S3: Colonization of the strain ZF390 in different types of soils. Figure S4: General characteristics of Paenibacillus peoriae ZF390. (A): Image of ZF390 colony morphology; (B): Image of ZF390 cells using transmission electron microscopy. Figure S5: Percentage of average nucleotide indentities (ANI) and in silico DNA-DNA hybridization (isDDH) among the selected Paenibacillus genomes. Table S1: Physiological and biochemical characteristics of strain ZF390. Table S2: Genome statistics. Table S3: Genomic features of P. peoriae ZF390 and other Paenibacillus strains. Table S4: Number of genes associated with general COG functional categories. Table S5: Strains and housekeeping gene locus tag information used for the phylogenetic tree construction in this study. Table S6: Comparison of genes involved in plant growth promotion between strains ZF390, HS311, SQR-21, HY96-2 and PS04. Table S7: Comparison of genes mostly involved in biofilm formation between strains ZF390, HS311, SQR-21, HY96-2, and PS04. Table S8: Comparison of genes involved in synthesis of resistance inducers between strains ZF390, HS311, SQR-21, HY96-2, and PS04.