Phenotypic and genomic characterization of Paenibacillus salivarius sp. nov., isolated from the human oral cavity

Abstract

A novel Gram-stain-negative, facultatively anaerobic, endospore-forming bacterium, designated DYY-L-2^T, was isolated from the saliva of a 69-year-old patient with chronic periodontitis in Hengyang, China. Phylogenetic analysis based on the 16S rRNA gene sequences revealed that strain DYY-L-2^T belonged to the genus Paenibacillus, with the highest similarity to Paenibacillus konsidensis LBY^T (98.6%), followed by Paenibacillus vini LAM0504^T (97.2%). Whole-genome sequencing yielded a complete circular chromosome of 5,642,305 bp with a genomic DNA G + C content of 50.8%. Overall genome relatedness index analysis indicated low average nucleotide identity (ANI < 80%) and in silico DNA–DNA hybridization (isDDH < 25%) values between strain DYY-L-2^T and its closely related type strains, supporting its status as a novel genospecies. The strain grew optimally at 37 °C, in pH 7.0, and tolerated up to 5% (w/v) NaCl. Major cellular fatty acids were C_16:0 and anteiso-C_15:0. The genomic functional annotation identified unique regions enriched with mobile genetic elements and Type I restriction-modification systems (e.g. hsdR and hsdM), suggesting adaptive mechanisms for genomic stability in the oral environment. Based on polyphasic taxonomic data, strain DYY-L-2^T represents a novel species within the genus Paenibacillus, for which the name Paenibacillus salivarius sp. nov. is proposed. The type strain is DYY-L-2^T (= GDMCC 1.6010^T = KCTC 43851^T). This study expands the understanding of the Paenibacillus diversity in host-associated environments and provides insights into its ecological role in the human oral microbiome.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12866-025-04522-2.

​Introduction​​

The genus Paenibacillus, first reclassified from Bacillus in 1993 [1], comprises facultative anaerobic, endospore-forming bacteria distinguished by exceptional ecological versatility and metabolic diversity. The genus name, derived from the Latin paene (“almost”), reflects its phylogenetic proximity to Bacillus, yet its distinct evolutionary and functional traits have established it as a subject of increasing interest in the area of plant-microbe interactions, environmental biotechnology, and pathogenesis [1, 2]. These bacteria thrive in diverse environments, including soil, rhizospheres, aquatic systems, plant matter, glaciers, and even clinical specimens (blood, cerebrospinal fluid, urine, sputum, wound, and dental plaque) [3–6]. Their adaptability is further evidenced by their presence in extreme habitats, such as antarctic ecosystems [7]. While primarily recognized for their agricultural applications in nitrogen fixation, phosphate solubilization, and biocontrol activities [8], certain species are recognized as animal pathogens. For example, Paenibacillus larvae is a notorious bee pathogen causing American foulbrood disease and threatens beekeeping [9]. Moreover, some Paenibacillus species have emerged as opportunistic human pathogens. Recent reports highlight the pathogenic potential of Paenibacillus in vulnerable populations (e.g., infants, old or immunocompromised populations). Neonatal infections caused by Paenibacillus thiaminolyticus are associated with severe neurological sequelae including meningitis, cerebral abscesses, and hydrocephalus [6]. Adult infections, though less frequent, manifest as bacteremia, endocarditis, and localized infections often associated with medical devices [6, 10]. The genus demonstrated concerning antimicrobial resistance patterns, with intrinsic resistance to β-lactams and variable susceptibility to vancomycin [6, 11], suggesting that its antibiotic resistance profiles may vary at different species or strains and not apply uniformly across the genus. While their presence in fecal microbiota has been documented [8], the adaptation mechanisms enabling intestinal colonization remain poorly characterized. Moreover, genomic studies of human-sourced Paenibacillus are limited, the intra-species and inter-species diversity in niche-specific genomic evolution, antimicrobial resistance and metabolic capabilities remain largely unexplored.

Here, a Paenibacillus strain, designated DYY-L-2^T, was isolated from human saliva, comprehensive analysis was conducted to describe its phenotypic and genotypic traits. Specifically, (1) the polyphasic taxonomic approach was used to characterize its phylogenetic and phenotypic traits to determine the taxonomic position; (2) whole-genome sequencing and comparative genomic analysis were performed to characterize its genomic features associated with host adaptation and virulence. This study determines the taxonomy of this strain and describes the novel species it represented.

Materials and methods

Specimen collection and bacteria isolation

Saliva was collected from a 69-year-old female with a several-year history of chronic periodontitis at August 10th, 2024, at the Department of Gastroenterology and Hepatology, the First Affiliated Hospital, Hengyang Medical School, Hengyang, China. Following a minimum 1-hour fasting period and abstinence from oral hygiene practices, the oral cavity was rinsed three times with 10 mL of sterile water to remove transient contaminants. Subsequently, 2–5 mL of unstimulated saliva was passively drooled into a pre-chilled sterile tube and immediately transported for bacterial culture. Briefly, 100 µL of saliva was spread onto plates of Brain Heart Infusion (BHI) agar (Huankai Microbial, Guangzhou, China; Cat No. 028361), R2A agar (Huankai Microbial, Guangzhou, China; Cat No. 022029), or Luria-Bertani agar (LBA) (Huankai Microbial, Guangzhou, China; Cat No. 028334), respectively. Plates were incubated at 37 °C for 48 h under separate aerobic and anaerobic conditions. Anaerobic conditions were established by immediately transferring inoculated plates into a GENbag anaer system (bioMérieux, France) following plating, facilitating cultivation in an oxygen-deprived environment.

16S rRNA gene sequence-based Taxonomic Identification

Bacterial colonies were isolated using a sterile toothpick, suspended in 20 µL of distilled deionized water, and thoroughly mixed. Cells were lysed by heat denaturation (95 °C for 5 min), followed by brief centrifugation to pellet cellular debris. Two microliters (µL) of the resulting supernatant served as the DNA template and was added directly to a PCR master mix (TaqMan™ Universal PCR Master Mix, Thermo Fisher Scientific, USA; Cat No. 4304437). Amplification of the 16S rRNA gene was performed via polymerase chain reaction (PCR) using universal primer pairs 8 F and 1492R, generating an almost-full-length 16S rRNA product [12]. The PCR products were sequenced using the Sanger sequencing by BGI Genomics (BGI-Shenzhen, China). To determine the phylogenetic relationships between strain DYY-L-2^T and its closely related species, the 16S rRNA gene sequence of strain DYY-L-2^T was submitted to the EzBioCloud database to retrieve validly published 16S rRNA gene sequences with high similarities. These sequences were then aligned using the CLUSTAL W to generate a multiple sequence alignment [13]. A phylogenetic tree based on the 16S rRNA gene was reconstructed with the neighbor-joining method in the MEGA v11.0, employing the general time-reversible model and bootstrap analysis with 1000 replicates [14].

Whole-genome sequencing and genome assembly

Bacterial cells were cultivated overnight at 37 °C in Luria-Bertani (LB) broth (Huankai Microbial, Guangzhou, China; Cat. No. 028324). Subsequently, 200 µL of the overnight culture was evenly spread onto an LB agar (LBA) plate using a sterile spreader. Following another overnight incubation at 37 °C, a confluent bacterial lawn was formed. This lawn was thoroughly scraped and rinsed with 1 mL of sterile phosphate-buffered saline (PBS; pH 7.4; Thermo Fisher Scientific, USA; Cat. No. 10010023). The resulting bacterial suspension was transferred into a 1.5 mL sterile microcentrifuge tube using a pipette. Finally, bacterial cells were pelleted by centrifugation at 10,000 × g for 5 min. Genomic DNA was extracted immediately using the TaKaRa MiniBEST Bacteria Genomic DNA Extraction Kit (Takara Bio Inc., Shiga, Japan, Cat. No. 9763) according to the manufacturer’s instructions. Genomic DNA quality was assessed using a Synergy HTX Multi-Mode Reader (BioTek Instruments, Inc., Winooski, VT, USA). Whole-genome sequencing was performed on two platforms: the Nanopore PromethION platform (MAGIGENE, Guangzhou, China) for long-read sequencing and the Illumina NovaSeq platform (MAGIGENE, Guangzhou, China) for short-read sequencing. For long-read sequencing, libraries were prepared by using the 1D Ligation Kit (SQK-LSK109, Oxford Nanopore Technologies, Oxford, UK) and the Native Barcoding Expansion Kit (EXP-NBD196, Oxford Nanopore Technologies, Oxford, UK) for multiplexing, following the manufacturer’s protocol. Sequencing was subsequently conducted on an R9.4.1 flow cell (FLO-MIN106D). Base calling was performed using MinKNOW software (v1.15.4, Oxford Nanopore Technologies) with the high-accuracy model (dna_r9.4.1_450bps_hac.cfg), and reads with Q scores < 7 were discarded, which generated a total of 191,406 reads and 1,139,766,328 bases, with a reads N50 length of 16,098 bp. The short-read library was prepared using the ALFA-SEQ DNA Library Prep Kit (Finorop, Guangzhou, China, Cat. No. NDI001E-01) with a 350 bp insert size. Sequencing was conducted on a NovaSeq XPlus 25B chip using a paired-end 150 bp (PE150) strategy. Samples were individually barcoded and pooled with others. Raw reads were demultiplexed by the service provider (MAGIGENE, Guangzhou, China). Further quality control including adapter trimming and reads filtering were performed using the fastp v0.20.1 (Q30 >95) [15]. De novo genome assembly was performed using the Unicycler (version 0.4.9b) with its default hybrid assembly pipeline [16].

Genomic taxonomy and phylogenetic analysis

Genome assembly quality was assessed using CheckM v1.0.12 with the lineage-specific workflow [17]. The ‘gtdbtk_wf’ workflow within GTDB-Tk software was utilized for further validation of taxonomic assignments [18]. The phylogenomic analysis was performed with the Type Strain Genome Server (TYGS) available at https://tygs.dsmz.de/ [19]. In silico DNA–DNA hybridization (isDDH) values between genomes of all strains were calculated using the Genome-to-Genome Distance Calculator (GGDC3.0, http://ggdc.dsmz.de/ggdc.php) provided by Meier-Kolthoff et al. [20]. The Similar Genome Finder tool integrated into PATRIC v3.6.10 was employed to identify publicly available genomes exhibiting high sequence similarity to the assembled genome [21]. Average Nucleotide Identity (ANI) values between the assembled genome and these similar genomes were calculated using FastANI [22]. A phylogenomic tree was reconstructed using the REALPHY webserver with default settings [23], by uploading the genome sequences to the site: https://realphy.unibas.ch/realphy/.

Genome annotation and comparative genomics

Comprehensive genome function annotation was conducted using the eggNOG-mapper v2.1.12, which integrates multiple databases including Carbohydrate-Active enZymes (CAZy), Cluster of orthologous Groups (COG), etc [24]. Virulence-associated genes and antimicrobial resistance genes within the genomes were predicted using ABRicate v1.00 (https://github.com/tseemann/abricate), querying the pre-computed NCBI AMRFinderPlus database [25] and the Virulence Factor Database (VFDB) [26].

Physiology and chemotaxonomy

The physiological and chemotaxonomic properties of the strain, encompassing growth characteristics, substrate utilization, acid production, and cellular fatty acid (CFA) composition, were determined following the methodologies outlined by Chen et al. [27]. Briefly, growth at temperatures of 4, 10, 20, 30, 37, 42, 50, and 55 °C was investigated in Tryptic Soy Broth (TSB) medium (Huankai Microbial, Guangzhou, China; Cat No. 024051). Growth at pH 3, 4, 5, 6, 7, 8, 9, 10, and 11 was evaluated in TSB medium, and the pH was adjusted according to the method described by Ruan et al. [28]. Salt tolerance was assessed in TSB medium prepared with varying NaCl concentrations (0, 0.5, 1, 2, 3, 5, 10, and 15% (w/v)). Optimal growth conditions were determined by monitoring the corresponding optical density of the culture at 600 nm (OD600) using a Synergy Mx Microplate Reader (BioTek, USA). Catalase activity was evaluated by observing bubble formation in 3% (v/v) hydrogen peroxide (H₂O₂). Oxidase activity was tested using tetramethyl-p-phenylenediamine. Additional biochemical profilings, including acid production from carbohydrates and enzyme activities, were performed using the API 50CH and API ZYM systems (bioMérieux, France) according to the manufacturer’s instructions. For cellular fatty acid analysis, strain DYY-L-2^T was cultivated in TSB at 30 °C for 48 h. Fatty acids were extracted, methylated, and analyzed using the Sherlock Microbial Identification System (Microbial ID, Newark, DE, USA) with the standard MIS Library Generation Software (version 6.0, database 4.0).

Results and discussion

Taxonomic assignment based on 16S rRNA gene

A total of 21 single colonies were isolated using BHI agar, R2A agar, and LBA agar plates. The predominant isolates belonged to the genus Streptococcus, comprising 13 strains (see Supplementary Table S1). NCBI BLAST analysis of the nearly complete 16S rRNA sequences revealed that 20 strains exhibited identities > 99% with known species. Only a strain, DYY-L-2^T, showed a 16S rRNA sequence identity of < 99% (see Supplementary Table S1). Subsequently, the 16S rRNA gene sequence of strain DYY-L-2^T was compared against the EzBioCloud database. The pairwise similarity analysis revealed the highest sequence identity (98.6%) with Paenibacillus konsidensis LBY^T, followed by Paenibacillus vini LAM0504^T (97.2%) and Paenibacillus faecis CIP 101,062^T (96.9%) (see Supplementary Table S2). A neighbor-joining phylogenetic tree reconstructed from 16S rRNA gene sequences confirmed that strain DYY-L-2^T forms a robust clade with P. konsidensis LBY^T, supported by high bootstrap values (Fig. 1). This clustering aligns with the high sequence similarity observed and indicated a close evolutionary relationship. The tree topology further distinguishes strain DYY-L-2^T from other closely related species, such as P. vini and P. faecis, as well as from the plant-associated species Paenibacillus rhizolycopersici (Fig. 1).

Fig. 1.

Neighbor-joining phylogenetic tree based on 16S rRNA gene sequences showing the position of strain DYY-L-2^T among related type strains of the genus Paenibacillus. Genbank accession numbers of the 16S rRNA gene sequences are given in parentheses. Evolutionary relationships are inferred using the neighbor-joining method. Bootstrap values (1000 replicates) greater than 50% are shown at branch nodes. Bacillus aryabhattai B8W22^T is used as an outgroup to root the tree. The bar represents 0.05 substitutions per nucleotide position. Strain DYY-L-2^T is in bold label in the tree

Genomic-based taxonomic assignment and general genomic features

Hybrid assembly of short- and long-reads from the genome of strain DYY-L-2^T produced a complete circular contig free of Ns. Quality assessment using the CheckM indicated high-quality assembly (completeness = 99.9%, contamination = 0.6%, strain heterogeneity = 0%). Briefly, strain DYY-L-2^T possesses a circular chromosome of 5,642,305 bp, containing 5,170 coding sequences (CDSs) and exhibiting a genomic DNA G + C content of 50.8%. Taxonomic assignment using GTDB-Tk classified strain DYY-L-2^T as a member of Fontibacillus sp., lacking reference genomes and ANI support, which is unreliable. This GTDB-Tk classification result was inconsistent with the 16S rRNA gene-based identification, suggesting that strain DYY-L-2^T represents a novel species. Furthermore, whole-genome-based taxonomic analysis conducted using the TYGS server identified 19 type-strain genomes closely related to strain DYY-L-2^T. The phylogenomic tree revealed that strain DYY-L-2^T formed a distinct branch separate from other type strains (Fig. S1). Among these type strains, the three closest relatives based on isDDH values were Paenibacillus popilliae ATCC 14,706^T (25.5%), Paenibacillus aceti L14^T (24.3%), and Paenibacillus aceti CGMCC 1.15420^T (23.9%). The corresponding ANI values for these genomes were 69.8%, 73.5%, and 77.2%, respectively (Supplementary Table S3). Both the isDDH and ANI values between these reference strains and strain DYY-L-2^T were significantly below the established thresholds for bacterial species delineation (70% for isDDH and 95% for ANI), suggesting that strain DYY-L-2^T represents a novel genospecies within the genus Paenibacillus.

Genomic and Ecological Niche Characterization of Paenibacillus Strains Close to strain DYY-L-2^T

Although comprehensive genomic analysis confirmed that strain DYY-L-2^T belongs to the genus Paenibacillus, the type-strain genomes identified by TYGS did not include sequences from the two species exhibiting the highest 16S rRNA gene similarity to strain DYY-L-2^T, namely P. konsidensis and P. vini. To better characterize strain DYY-L-2^T and its phylogenetic closest relatives, we further employed the Similarity Genome Finder, which identified 50 genome sequences from 15 species most closely related to strain DYY-L-2^T (see Supplementary Table S4). Regrettably, the genome of P. konsidensis LBY^T has not been sequenced. Among these 50 genomes, P. vini J42TS3 showed the highest ANI with strain DYY-L-2^T (79.6%). Furthermore, applying the species delineation threshold of ANI > 95% as proposed by Jain et al. (2018), we determined that the Paenibacillus sp. genomes analyzed in this study could be delineated into at least six novel species. Several strains also appeared to be misclassified: strains Paenibacillus thermophilus DSM 24,746^T and JCM 17,693^T should be reclassified as Paenibacillus macerans, while Paenibacillus antibioticophila J41TS12 and Paenibacillus apis J41TS4 likely belong to the same species (Fig. S2A). We further compared the general genomic features of four strains (Fontibacillus phaseoli CECT8333^T, P. vini J42TS3, Paenibacillus sp. Marseille-P2973, and P. vini CENA-BCM001) with the highest ANI to strain DYY-L-2^T. These strains exhibited relatively similar genome sizes, ranging from 5.5 Mb to 5.7 Mb, with strain Marseille-P2973 being the smallest and strain CENA-BCM001 the largest. The corresponding numbers of CDSs were 5241 and 4919, respectively. The genomic DNA G + C content of strain DYY-L-2^T (50.8%) was higher than those of these four strains (48.9% to 49.4%), while its coding density (86.8%) was lower (87.4% to 88.0%) (see Supplementary Table S5).

Genomic map analysis revealed four large unique regions (R1-R4) specific to the genome of strain DYY-L-2^T. These regions are enriched in Mobile Genetic Elements (MGEs) and restriction-modification (R-M) genes and exhibit significant deviations in the genomic DNA G + C content from the average. For instance, R1 contained the hsdR and hsdM genes, and R4 contained the hsdM gene. These genes encoded the Type I restriction enzyme subunits (HSDR proteins), which are widespread in prokaryotes and protect the host from foreign DNA uptake. They assemble into an enzyme complex (R2M2S1) that modifies hemimethylated DNA and restricts unmethylated DNA [29]. This mechanism may contribute to adaptation in the human oral cavity, a niche abundant in foreign DNA, by providing defense against horizontal acquisition of genetic material, thereby promoting genomic stability. Concurrently, unique regions R2 and R3 in the genome of strain DYY-L-2^T carried numerous transposase genes (e.g., tnpA, tnpB, tnpR) from the Tnp family, and these genes are also distributed in the shared genomic regions with other four strains (Fig. S2B). This suggested that the genomes of strain DYY-L-2^T and its close relatives possess considerable plasticity, potentially acquiring new functions via Horizontal Gene Transfer (HGT) events mediated by transposases. We hypothesized that the presence of a Type I R-M system suggests a retained capacity for genomic plasticity and potential for intra-species HGT, which could be facilitated among strains with homologous R-M systems. The combination of R-M system and transposases may represent an adaptive strategy balancing genome stability with plasticity. However, conclusions regarding genomic plasticity and stability between the strain DYY-L-2^T represented species and P. vini require more genomic data for full validation. Additionally, all strains possessed oxidative stress response genes (e.g., xre, mazG, and rex), aiding their adaptation to oxygen exposure [30–32]. Strains DYY-L-2^T, CECT8333 ^T, Marseille-P2973, and CENA-BCM001 also possessed various cas genes, which are absent in the genomes of strains J42TS3 and I6 (Fig. S2B). All compared strains harbored vancomycin resistance genes (e.g., vanY, vanT) and a lincosamide/clindamycin resistance gene (llmA, encoding 23 S ribosomal RNA methyltransferase) (Fig. S2B). These specific resistance genes are also commonly found in other Paenibacillus species genomes [33, 34]. Although the Proksee webserver did not identify genes encoding β-lactamases in the genome of strain DYY-L-2^T, integrated annotations from the NCBI database and eggNOG-mapper suggested the potential presence of β-lactamase-encoding genes. These are predicted to be primarily represented by enzymes belonging to the serine hydrolase and metallo-β-lactamase superfamilies (see Supplementary Table S6). Moreover, virulence factor prediction using the VFDB database did not detected virulence-associated genes in the genomes of strain DYY-L-2^T or any other strains examined in this study, revealing that strain DYY-L-2^T and its close relatives should be non-pathogenic.

Phylogenomic tree reconstructed from these genomes showed that strain DYY-L-2^T formed a distinct branch from the close 50 strains, indicating its status as a putative novel Paenibacillus genospecies. Its closest relative was P. vini (Fig. 2), consistent with our prior whole-genome ANI analysis. Among these genomes used to reconstruct the tree, only P. macerans I6 and strain DYY-L-2^T had ‘complete’ assembly levels; most of the remaining genomes were assembled at the ‘contig’ levels. All included genomes were of high assembly quality, exhibiting >98% completeness and < 2% contamination (Fig. 2). The phylogenomic tree clustered into two primary clades. Clade I comprised 12 strains, predominantly Paenibacillus ihbetae, which were phylogenetically distant from strain DYY-L-2^T (Fig. 2). P. ihbetae was a species and its type strain was isolated and characterized from a high-altitude lake [35]. While its functional traits have not been extensively described, several closely related Paenibacillus species, isolated from soil environments, have been reported to possess plant growth-promoting capabilities [3, 36]. Clade II consisted of multiple Paenibacillus species and was further subdivided into two subclades. Subclade IIa contained 25 strains, mainly represented by P. macerans, along with novel and unidentified Paenibacillus species. Subclade IIb comprised 14 strains with greater taxonomic diversity, including P. vini, Paenibacillus woosongensis, P. apis, P. aceti, and strain DYY-L-2^T, suggesting unexplored interspecies diversity within these taxa. The genome sizes of strains within the subclade IIb ranged from 5.4 to 5.9 Mbp (mean 5.7 Mbp), significantly lower than those in the clade I (5.8–6.9 Mbp, mean 6.9 Mbp) and the subclade IIa P. macerans strains (7.1–7.4 Mbp, mean 7.2 Mbp), but higher than other species within the subclade IIa (4.6–5.5 Mbp, mean 5.2 Mbp). The number of predicted protein-coding genes in subclade IIb followed the same trend. Furthermore, the genomic DNA G + C contents of strains in the subclade IIb were lower than those of strains in the subclade IIa and clade I.

Fig. 2.

Phylogenomic tree and isolation source features of the top 50 Paenibacillus strains close to strain DYY-L-2^T. The tree is reconstructed using the RECOPHY server (https://recophy.unibas.ch/recophy/) under default parameters. Branch lengths represent the number of substitutions per site, with a scale bar (0.01 substitutions per site) provided for reference. Statistical branch support values are not included, as the RECOPHY pipeline does not compute them. The coloration of branches corresponds to the isolation host of each strain, as detailed in the legend. Tip labels denote strain names, while full genome assembly accession numbers are comprehensively listed in Supplementary Table S4

The 51 strains (including strain DYY-L-2^T) were predominantly isolated from the United States (13), Japan (8), and China (9). Primary isolation sources were related to human (15 strains), environmental (17), and food (8). Specifically, strains within the clade I were mainly isolated from unspecified environmental sources in the USA. Strains in the clade II were frequently associated with human sources. For example, among the 14 strains in the subclade IIb, 2 were isolated from plant roots and 12 from human-associated sources (e.g., food, oral cavity, human gut). Of these 12 human-associated strains, 4 originated from the human gut and 4 from honey (Fig. 2). Overall, the clear clustering pattern was not observed relating to isolation sources. A 2025 systematic review documented 179 cases of human infection caused by Paenibacillus, comprising 38 cases in adults and 141 in infants. The 38 adult infections were attributed to 23 different Paenibacillus species. Among infant infections, P. thiaminolyticus was predominant (112/141, 79%), followed by Paenibacillus alvei (2/141, 1%), Paenibacillus dendritiformis (2/141, 1%), and unidentified Paenibacillus species (including some cases of multispecies infection) (27/141, 19%). All infected infants presented with sepsis syndrome or meningitis, complicated by extensive cerebral destruction and hydrocephalus [6]. Those Paenibacillus species reported to cause human infections are phylogenetically distant from strain DYY-L-2^T and the other 50 strains, exhibiting low genomic similarity (67% < ANI < 70%, aligned nucleotides < 22%; calculated using the JSpecies webserver: https://jspecies.ribohost.com/jspeciesws/#analyse) (See supplementary Table S7). Only a historical case involved a 52-year-old healthy French man who developed a brain abscess due to P. macerans and a Clostridium species following an orbital injury sustained from falling into a bush. Despite drainage procedures and antibiotic therapy (amoxicillin-clavulanate, followed by amoxicillin and metronidazole), the patient still died [37]. However, this report was outdated, and literatures describing human infection or mortality solely or jointly attributed to P. macerans were not identified. These results indicated that although Paenibacillus species closely related to strain DYY-L-2^T (particularly within subclade IIb) are frequently associated with human, they are unlikely to be significant human pathogens. The complete genome sequence of strain DYY-L-2^T is the first completely assembled genome (i.e., assembled into one single circular sequence) within the subclade IIb; other strains in this clade are draft assemblies (contig-level). This completed genome sequence provides a high-quality reference genomic resource for subsequent investigations into the related species and strains.

Morphological, physiological and biochemical characteristics

Cells were observed to be rod-shaped, facultatively anaerobic, and capable of forming endospores. After 24 h of incubation at 37 °C on TSB agar under aerobic condition, colonies appeared circular with intact but non-smooth margins, measuring 0.7–1.2 mm in diameter. The colonies were convex, and shiny in appearance. Physiological tests indicated that strain DYY-L-2^T is catalase- and oxidase-positive (Table 1). Growth occurred at temperatures between 20 °C and 50 °C, with an optimum at 37 °C. The strain tolerated NaCl concentrations up to 5% (w/v), with optimal growth at 2%. The pH range for growth was 5.0–9.0, with an optimum at pH 7.0.

Table 1.

Comparison of the phenotypic and physiological characteristics of strain DYY-L-2^T and its relatives. Strains: 1, DYY-L-2^T; 2, P. konsidensis JCM 14798^T; 3, P. vini LAM0504^T; 4, P. faecis 656.84^T; 5, F. phaseoli LMG 27589^T. Data in column 1 are from this study; data in column 2, 3, 5 are from Chen et al. (2015) [27]; data in column 4 are from Clermont et al.. (2015) [38]. Symbols: +, positive; -, negative; NA, data not available

Characteristics	1	2	3	4	5
Optimum growth temperature (°C)	37	30	30	30–37	37
Catalase	+	+	+	+	-
Oxidase	+	+	-	+	+
Acid production from: (API 50CH)
L-Arabinose	+	-	+	NA	-
D-Ribose	-	+	+	-	-
D-Xylose	+	+	+	+	-
Methyl β-D-xylopyranoside	-	+	+	NA	-
D-Fructose	+	+	-	NA	-
D-Mannose	-	+	+	+	-
L-Rhamnose	-	+	-	+	-
Amygdalin	+	+	+	NA	+
D-Turanose	+	+	-	+	-
D-Lyxose	-	-	-	NA	+
Enzyme activities: (API ZYM)
Alkaline phosphatase	-	-	-	-	+
Trypsin	-	-	-	-	-
α-Chymotrypsin	-	+	+	NA	-
Acid phosphatase	-	+	-	-	+
α-Glucosidase	+	+	+	+	+
N-Acetyl-β-glucosaminidase	-	-	+	-	-

Biochemical profiling of strain DYY-L-2^T was performed using API systems (i.e., API 50CH and API ZYM). The results indicated that strain DYY-L-2^T produces acid from a range of carbon sources, a characteristic that distinguishes it from closely related strains. Specifically, strain DYY-L-2^T utilized L-arabinose, D-xylose, and amygdalin for acid production, a profile shared with P. vini LMG 0504ᵀ. In contrast, P. konsidensis JCM 14798ᵀ did not produce acid from L-arabinose, and F. phaseoli LMG 27589ᵀ failed to utilize both L-arabinose and D-xylose. Furthermore, strain DYY-L-2^T was capable of acid production from D-fructose, whereas P. vini LMG 0504ᵀ was not. However, strain DYY-L-2^T could not metabolize D-mannose or L-rhamnose for acid production, unlike P. konsidensis JCM 14798ᵀ and P. faecis 656.84ᵀ, which yielded positive results for these substrates.

Enzyme activity assays revealed that strain DYY-L-2^T was positive for esterase (C4), esterase lipase (C8), naphthol-AS-BI-phosphohydrolase, α-glucosidase, α-galactosidase, and β-galactosidase. Negative reactions were observed for alkaline phosphatase, trypsin, α-chymotrypsin, acid phosphatase, and N-acetyl-β-glucosaminidase (Table 1). The enzymatic profile of strain DYY-L-2^T was similar to those of P. konsidensis JCM 14798ᵀ, P. vini LMG 0504ᵀ, and P. faecis 656.84ᵀ, but differed significantly from that of F. phaseoli LMG 27589ᵀ. The differences in the physiological and biochemical characteristics between strain DYY-L-2^T and its relatives are shown in Table 1.

Chemotaxonomic characterization

Chemotaxonomic analyses aligned strain DYY-L-2^T with members of the genus Paenibacillus. The major cellular fatty acids (> 10%) were identified as C_16:0 (34.6%) and anteiso-C_15:0 (27.2%) (Table 2). This profile was consistent with those of known Paenibacillus species, although the proportion of anteiso-C_15:0 was lower than in P. konsidensis JCM 14798^T (32.4%), P. faecis 656.84^T (39.7%), F. phaseoli LMG 27589^T (33.4%), and P. vini LAM0504^T (43.1%). The proportion of C_16:0 was higher than in P. konsidensis JCM 14798^T (22.8%), P. faecis 656.84^T (15.8%), F. phaseoli LMG 27589^T (22.3%) and P. vini LAM0504^T (15.9%) (Table 2). These chemotaxonomic features collectively support the placement of strain DYY-L-2^T within the genus Paenibacillus, while also highlighting distinct compositional differences from its closest relatives.

Table 2.

Major fatty acids of strain DYY-L-2^T and its relatives. Strains: 1, DYY-L-2^T; 2, P. konsidensis JCM 14798^T; 3, P. vini LAM0504^T; 4, P. faecis 656.84^T; 5, F. phaseoli LMG 27589^T. Data in column 1 are from this study; data in column 2, 3, 5 are from Chen et al.. (2015) [27]; data in column 4 are from Clermont et al.. (2015) [38]. Fatty acids present in amounts lower than 1% in all strains are not shown and major components (higher than 10%) in each strain are in bold. Symbols: +, positive; -, negative; NA, data not available

Fatty acids (%)	1	2	3	4	5
Unbranched
C14:0	2.2	2.4	3.5	NA	3.2
C16:0	34.6	22.8	15.9	15.8	22.3
iso-unbranched
iso-C14:0	NA	1.9	2.7	NA	NA
iso-C15:0	4.9	8.4	10	NA	5.1
iso-C16:0	4.6	11.4	9.6	13.6	9.7
iso-C17:0	5.4	6.9	4.9	NA	6.8
Anteiso-unbranched
anteiso-C15:0	27.2	32.4	43.1	39.7	33.4
anteiso-C17:0	9.1	8.4	6.1	11.7	12.3

Description of Paenibacillus salivariussp. nov.

Paenibacillus salivarius (sa.li.va’ri.us. L. masc. adj. salivarius, pertaining to saliva, from which the type strain was isolated).

Cells are Gram-stain-negative, facultatively anaerobic, endospore-forming rods. Colonies are circular, convex, shiny, with intact margins, and 0.7–1.2 mm in diameter. Growth occurs at temperatures between 20 and 50 °C (optimum 37 °C), at pH 5.0–9.0 (optimum pH 7.0), and in the presence of 0–5% (w/v) NaCl (optimum 2%). Positive for catalase and oxidase activities.

The major cellular fatty acids are C_16:0 and anteiso-C_15:0. The genomic DNA G + C content of the type strain is 50.8%.

The type strain, DYY-L-2ᵀ (= GDMCC 1.6010ᵀ = KCTC 43851ᵀ), was isolated from the saliva of a patient with chronic periodontitis in Hengyang, China. The GenBank accession numbers for the 16S rRNA gene and whole-genome sequences of strain DYY-L-2ᵀ are PX404122 and CP195592.1, respectively.

Conclusion

This study presents the polyphasic characterization of strain DYY-L-2ᵀ, representing a novel species within the genus Paenibacillus, for which the name Paenibacillus salivarius sp. nov. is proposed. The assignment was based on its distinct phylogenetic position, evidenced by low whole-genome Average Nucleotide Identity (ANI < 80%) and in silico DNA-DNA hybridization (isDDH < 25%) values against its close relatives. The genome of strain DYY-L-2ᵀ features unique regions enriched with mobile genetic elements and Type I restriction-modification systems (e.g., hsdR, hsdM), suggesting adaptive evolution for maintaining genomic stability within the competitive oral niche.

Notably, limitations arise from the scarcity of closely related isolates – with only a genome representing this novel species and few draft genomes available for most phylogenetically neighboring taxa. This restricts comprehensive interspecies functional genomic comparisons. Nevertheless, this study expands the understanding of the genetic and functional diversity of the genus Paenibacillus within host-associated environments and high-quality genome of strain DYY-L-2^T provides a foundational genomic resource for future studies investigating its ecological role and interactions within the human oral microbiome.

Authors’ contributions

XH performed the bacteria isolation and taxa identification. YP and XC performed the WGS-based analysis. XH, MC, and XC drafted the manuscript. PH, QZ, ZL, YW and DM reviewed and revised the manuscript. LX, YP and XC designed the whole study. All authors made substantial and direct contributions to the work, and read and approved the final version of the manuscript.

Funding

This work was supported by the Shenzhen Science and Technology Program (JCYJ20230808105208017), the University Research Projects of Anhui Provincial (2024AH051076), the Open Project of Key Lab. of Provincial Key Laboratory (Wxn202403), the Undergraduate Innovation and Entrepreneurship Training Program (202410372016, 202410372004S), the Industry Academia-Research Project of Demonstration Research on Fungal Genome Analysis and Technological Innovation in Microbial Agricultural Wastewater Treatment (H20250225), the Ministry of Education’s 2024 Themed Case Study Project (ZT-2410372012), the Major Science and Technology Projects of the Scientific and Technological Innovation Platform (202305a12020022) and the Anqing Normal University Graduate Student Industry-University Collaborative Practice Program Special Fund Grant (Industry-University Co-constructed Demonstration Course: Restoration Ecology, X2025xqkc007)​.

Data availability

The GenBank accession numbers for the 16S rRNA gene and whole-genome sequences of strain DYY-L-2ᵀ are PX404122 and CP195592.1, respectively.

Declarations

Ethics approval and consent to participate

This study was approved by the First Affiliated Hospital, Hengyang Medical School, University of South China Ethics and Scientific Committee. It was performed in accordance with the Declaration of Helsinki. Written informed consent was obtained from the participant prior to the enrollment.

Consent for publication

Not applicable.

Competing interests

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.