Genomic revisitation and reclassification of the genus Providencia

ABSTRACT

Members of Providencia, although typically opportunistic, can cause severe infections in immunocompromised hosts. Recent advances in genome sequencing provide an opportunity for more precise study of this genus. In this study, we first identified and characterized a novel species named Providencia zhijiangensis sp. nov. It has ≤88.23% average nucleotide identity (ANI) and ≤31.8% in silico DNA-DNA hybridization (dDDH) values with all known Providencia species, which fall significantly below the species-defining thresholds. Interestingly, we found that Providencia stuartii and Providencia thailandensis actually fall under the same species, evidenced by an ANI of 98.59% and a dDDH value of 90.4%. By fusing ANI with phylogeny, we have reclassified 545 genomes within this genus into 20 species, including seven unnamed taxa (provisionally titled Taxon 1–7), which can be further subdivided into 23 lineages. Pangenomic analysis identified 1,550 genus-core genes in Providencia, with coenzymes being the predominant category at 10.56%, suggesting significant intermediate metabolism activity. Resistance analysis revealed that most lineages of the genus (82.61%, 19/23) carry a high number of antibiotic-resistance genes (ARGs) and display diverse resistance profiles. Notably, the majority of ARGs are located on plasmids, underscoring the significant role of plasmids in the resistance evolution within this genus. Three species or lineages (P. stuartii, Taxon 3, and Providencia hangzhouensis L12) that possess the highest number of carbapenem-resistance genes suggest their potential influence on clinical treatment. These findings underscore the need for continued surveillance and study of this genus, particularly due to their role in harboring antibiotic-resistance genes.

IMPORTANCE

The Providencia genus, known to harbor opportunistic pathogens, has been a subject of interest due to its potential to cause severe infections, particularly in vulnerable individuals. Our research offers groundbreaking insights into this genus, unveiling a novel species, Providencia zhijiangensis sp. nov., and highlighting the need for a re-evaluation of existing classifications. Our comprehensive genomic assessment offers a detailed classification of 545 genomes into distinct species and lineages, revealing the rich biodiversity and intricate species diversity within the genus. The substantial presence of antibiotic-resistance genes in the Providencia genus underscores potential challenges for public health and clinical treatments. Our study highlights the pressing need for increased surveillance and research, enriching our understanding of antibiotic resistance in this realm.

INTRODUCTION

Providencia, Gram-negative bacteria from the Enterobacteriales order, are widespread across diverse environments like water, soil, and animals (1). These bacteria, as opportunistic pathogens, can cause severe human infections, including urinary tract infections and meningitis (2–5). The genus currently encompasses 11 validly named species: Providencia stuartii, Providencia thailandensis, Providencia manganoxydans, Providencia burhodogranariea, Providencia sneebia, Providencia rettgeri, Providencia huaxiensis, Providencia alcalifaciens, Providencia heimbachae, Providencia vermicola, Providencia rustigianii (6–14), and three species that are believed to be not validly named: Providencia entomophila, Providencia wenzhouensis and Providencia hangzhouensis (15–17). Among these, P. stuartii, P. rettgeri, and the newly discovered P. hangzhouensis are predominantly implicated in human infections.

While prior studies have focused on one or several specific species within the Providencia genus (18–20), a comprehensive taxonomic reassessment and deep genomic insight into the entire genus remain elusive. Particularly concerning are the recurrent misclassifications, as evidenced by the incorrect attribution of strains to P. rettgeri that in fact belong to P. hangzhouensis (17). Such accurate species categorization is foundational to our grasp of bacterial habitat, epidemiology, pathogenesis, and microbiological traits, with vast repercussions across health, industry, and scientific research (21). Thus, there’s an ever-pressing need for up-to-date, meticulously curated taxonomic classifications. Although 16S rRNA gene sequence analysis is a popular tool for bacterial identification, its precision often falls short for accurate species differentiation (22). Owing to the rapid advancement in sequencing technologies, whole genome-based analyses have become more accessible, gaining prominence in species identification due to their high resolution (23). Average nucleotide identity (ANI), using cutoff values of ≥96%, and in silico digital DNA-DNA hybridization (dDDH) with a cutoff of ≥70.0% are commonly utilized for precise species identification (24–26).

Despite several studies (17–19) highlighting the challenges in Providencia classification, a comprehensive, genome-based investigation into these ambiguities remains absent. To address this deficiency, our study aims to demystify Providencia taxonomy, presenting an updated classification and introducing a novel species, Providencia zhijiangensis sp. nov. Leveraging the power of genomics, we established a robust genome-based phylogenetic framework to unravel the intricate diversity within the Providencia genus. Augmenting our primary focus, we also delved into the exploration of genus-specific core genes, and investigated the landscape of antibiotic resistance and plasmids, offering insights into the multifaceted nature of this genus.

RESULTS

Identification and characterization of novel species Providencia zhijiangensis

Strain D4759, preliminarily identified as Providencia alcalifaciens using the matrix-assisted laser desorption/ionization - time of flight mass spectrometry (MALDI-TOF MS). technique (Vitek MS system, bioMérieux, France), was isolated from a clinical patient’s bile sample. The patient, an 84-year-old man, admitted for acute obstructive suppurative cholangitis and septic shock, exhibited symptoms including poor appetite, fever, and impaired consciousness. Clinical findings upon admission included a temperature of 38.1°C, pulse rate of 131 beats per minute, respiratory rate of 28 breaths per minute, and blood pressure of 104/56 mmHg. Laboratory tests revealed a leukocyte count of 26.98 × 10⁹/L with 90.3% neutrophils, platelet count of 151 × 10⁹/L, and markedly elevated C-reactive protein levels at 260.76 mg/L. The patient also showed signs of acute kidney and liver injury, type 2 diabetes, hypertension, and coagulation abnormalities. Given the rarity of its isolation site and the patient’s severe clinical condition, we conducted whole-genome sequencing to ascertain its precise species classification.

The D4759 complete genome comprises a chromosome spanning 3,796,440 bp with a G + C content of 42.91% and two plasmids, pD4759_2_1 and pD4759_2_2, measuring 2,683 bp and 2,428 bp in length, respectively. Utilizing the 16S rRNA gene sequences from other Providencia species available in the EzBioCloud database (27) and a subsequently constructed phylogenetic tree (Fig. 1A), it was confirmed that strain D4759 aligns within the Providencia genus. Specifically, D4759 exhibited pronounced 16S rRNA gene sequence similarity to P. rustigianii DSM 4542 (99.74%; AM040489) and P. alcalifaciens DSM 30120 (99.73%; ABXW01000071). Given the previously stated insufficient resolution of 16S rRNA phylogenies for Providencia species delineation (9, 13, 19), we opted for a whole genome-based phylogenetic analysis approach, comparing D4759 with type strains from the Providencia genus. This phylogenetic analysis further substantiated D4759’s placement within the Providencia genus, particularly emphasizing its closest association with P. alcalifaciens DSM 30120 (Fig. 1B). Furthermore, upon comparing the ANI (79.81%–88.28%) and dDDH (20.9%–31.80%) values of strain D4759 with those of type strains from the Providencia species, we found them to be significantly below the thresholds defining species classification (ANI >96% and dDDH >70%; Table 1).

Fig 1.

Maximum-likelihood phylogenetic trees of the Providencia genus, constructed from (A) 16S rRNA gene sequences and (B) whole-genome sequences. The trees are rooted using Proteus mirabilis ATCC 29906 as the outgroup. Branch nodes with bootstrap values >0.8, obtained from 1,000 resamplings, are highlighted. The strain D4759 used in this study is indicated with a red star.

TABLE 1.

ANI and dDDH values between strains D4759 and the type strains of Providencia species

Taxonomic assignment	Type strain	Assembly accession	ANI (%)	dDDH (%)
Providencia stuartii	ATCC 25827	GCA_000154865.1	80.20	21.2
Providencia manganoxydans	LLDRA6	GCA_016618195.1	80.43	21.6
Providencia burhodogranariea	DSM 19968	GCA_000314855.2	79.81	21.1
Providencia sneebia	DSM 19967	GCA_000314895.2	79.90	21.2
Providencia hangzhouensis	PR-310	GCA_029193595.2	81.33	21.8
Providencia thailandensis	KCTC 23281	GCA_014652175.1	79.97	20.9
Providencia rettgeri	NCTC11801	GCA_900455085.1	81.31	22.0
Providencia huaxiensis	WCHPr000369	GCA_002843235.3	81.28	21.9
Providencia wenzhouensis	R33	GCA_019343475.1	81.54	21.8
Providencia alcalifaciens	DSM 30120	GCA_000173415.1	88.23	31.8
Providencia heimbachae	ATCC 35613	GCA_001655055.1	80.85	21.4
Providencia vermicola	DSM 17385	GCA_020381325.1	81.08	21.3
Providencia rustigianii	DSM 4541	GCA_000156395.1	83.59	24.0

The biochemical characteristics of strain D4759 and all other known Providencia species are detailed in Table 2. D4759 grows between 15°C and 42°C, exhibiting optimal growth at 35°C and 37°C. Morphologically, it consists of Gram-negative, motile, facultatively anaerobic rods. On nutrient agar incubated for 24 hours at 37°C, colonies appeared circular, raised, yellow, opaque, and with a smooth texture. Notably, D4759 lacks oxidase activity. D4759 can utilize acid from D-glucose, sucrose, and glycerol, but not from D-mannitol, inositol, D-sorbitol, L-rhamnose, melibiose, amygdalin, D-arabinose, aesculin, arbutin, cellobiose, 2-ketogluconate, D-lyxose, salicin, or D-xylose. A positive reaction is observed for deaminase, while negative reactions are found for β-galactosidase, arginine dihydrolase, ornithine decarboxylase, lysine decarboxylase, and gelatinase. The Voges-Proskauer reaction and phenylalanine deaminase test for D4759 are positive, but it shows negative results for urease activity and indole production. D4759 can utilize citrate but does not produce H₂S. D4759 exhibits distinct biochemical characteristics compared to other Providencia species. The strain D4759 stands out due to its positive response to acetoin production tests and its lack of indole production. When contrasting D4759 with clinically more prevalent Providencia species, specifically P. stuartii, P. rettgeri, and P. hangzhouensis, it is noteworthy that D4759 does not utilize inositol and indole. This distinctive biochemical profile could serve as a rapid diagnostic criterion to differentiate D4759 from its close relatives in clinical settings. The antimicrobial resistance susceptibility tests shows that D4759 is sensitive to all tested antibiotics except tigecycline (Table S1). The total fatty acid profile shows that there are two predominant fatty acids (>10%), i.e., C_16:0 (35.03%) and C_14:0 (11.41%).

TABLE 2.

Biochemical characteristics of strain D4759 and type strains of other Providencia species^a

Characteristic	D4759	1	2	3	4	5	6	7	8	9	10	11	12	13	14
API 20E tests:
β-Galactosidase	−	−	−	−	−	−	−	−	−	−	+	−	−	−	ND
L-arginine dihydrolase	−	−	−	−	−	−	−	−	−	−	+	−	−	−	ND
Lysine decarboxylase	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−
L-ornithine decarboxylase	−	−	−	−	−	−	−	−	−	−	+	−	−	−	ND
Citrate utilization	+	+	−	−	+	+	−	−	−	+	+	−	+	+	+
H2S production	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−
Urea hydrolysis	−	+	−	−	−	+	+	−	+	−	−	−	+	+	−
Deaminase	+	+	+	+	+	+	+	+	+	+	+	+	+	+	ND
Indole production	−	+	+	+	−	+	+	+	+	+	−	+	+	+	−
Acetoin production	+	−	−	−	−	−	+	−	−	−	+	−	+	−	−
Gelatinase	−	−	−	−	−	−	−	−	−	−	+	−	−	−	ND
D-glucose	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
D-mannitol	+	+	−	+	−	+	−	−	+	−	+	+	+	+	+
Inositol	−	+	−	+	−	+	+	−	−	+	+	+	+	+	−
D-sorbitol	−	−	−	−	−	−	−	−	+	−	+	−	−	−	ND
L-rhamnose	−	−	−	−	−	−	+	−	−	−	+	−	+	−	ND
Sucrose	+	+	−	−	−	−	−	−	−	+	+	−	−	−	ND
Melibiose	−	−	−	−	−	−	−	−	−	−	+	−	−	−	ND
Amygdalin	−	−	−	−	−	+	+	−	+	−	−	−	+	+	ND
L-arabinose	−	−	−	+	−	−	−	−	−	−	+	+	−	−	ND
API 50CHE tests:
Aesculin	−	−	+	+	+	+	+	+	+	−	+	−	ND	ND	ND
Arbutin	−	−	−	+	+	+	−	−	+	−	+	−	ND	ND	ND
Cellobiose	−	−	−	−	−	−	−	−	−	−	+	−	ND	ND	ND
Glycerol	+	+	−	−	−	−	+	−	−	+	+	−	ND	ND	−
2-Ketogluconate	−	−	−	−	−	+	−	−	−	−	+	+	ND	ND	ND
D-lyxose	−	+	−	−	−	−	−	−	−	+	−	−	ND	ND	ND
Salicin	−	−	−	+	−	+	−	−	+	−	+	−	ND	ND	+
D-xylose	−	−	−	−	−	−	−	−	+	−	+	−	ND	ND	ND

^a Strains: 1, P. manganoxydans LLDRA6; 2, P. alcalifaciens DSM 30120; 3, P. burhodogranariea DSM 19968; 4, P. heimbachae DSM 3591; 5, P. huaxiensis KCTC 62577; 6, P. rettgeri DSM4542; 7, P. rustigianii DSM 4541; 8, P. sneebia DSM 19967; 9, P. stuartii DSM 4539; 10, P. thailandensis KCTC 23281; 11, P. vermicola DSM 17385; 12, P. hangzhouensis PR-310; 13, P. entomophila IO-23; 14, P. wenzhouensis R33. Data for species other than P. zhijiangensis D4759 are from references (8, 17). (+), 90% to 100% positive reaction; (−), 0% to 10% positive reaction; ND, not determined.

Based on both genotypic and phenotypic characterizations, we propose that strain D4759 be recognized as a novel species, for which we suggest the name Providencia zhijiangensis sp. nov. (type strain D4759^T = CCTCC AB 2023263^T = NBRC 116614^T; zhi.jiang.en’sis. N.L. masc. adj. zhijiangensis, referring to the Xihu District of Hangzhou City, Zhejiang Province, China).

Curation of Providencia genomes with the updated taxonomy

Following the novel species identification and considering the noted inaccuracies in previous classifications (17), a re-evaluation is paramount. From our comprehensive collection approach, we sequenced other three recently collected Providencia clinical strains (two from urine and one from tissue fluid). Additionally, after quality assessment, 541 out of the 735 Providencia genomes from GenBank were included for a thorough species identification analysis.

By comparing the pairwise ANI of the genomes, we categorized all assemblies into 20 phylogroups using a threshold of 96%. The pairwise dDDH between each phylogroup was found to be less than 70%. Interestingly, when we further determined the species name of each phylogroup by Type Strain Genome Server (TYGS), we found that phylogroup 1 corresponded to both Providencia stuartii (dDDH: 94.70%) and Providencia thailandensis (dDDH: 94.20%). A comparison of type strains of P. thailandensis [KCTC 23281 (GCA_014652175.1)] and P. stuartii [ATCC 25827 (GCA_000154865.1)] revealed that their dDDH and ANI values were 94.10% and 99.11%, respectively, suggesting that they indeed constitute a single species. Following the principles outlined by the International Code of Nomenclature of Bacteria (ICNP) (28), P. stuartii (6) holds priority over P. thailandensis (7) for the species name. Consequently, we propose that P. thailandensis should be considered a later heterotypic synonym of P. stuartii. These results suggested the taxonomy of the Providencia genus could be updated to comprise 20 species, encompassing seven unnamed taxa (Taxon 1–7; Table 3).

TABLE 3.

Updated genomic species affiliated to genus Providencia^a

Phylogroup	Taxonomic assignmentb	Type strainc	G + C mol%d	Assembly accession	No. of genomes
1	Providencia stuartii	ATCC 25827	41.47	GCA_000154865.1	87
2	Taxon 7	PROV260	40.92	GCA_028477965.1	25
3	Providencia manganoxydans	LLDRA6	40.12	GCA_016618195.1	6
4	Providencia burhodogranariea	DSM 19968	39.08	GCA_000314855.2	1
5	Taxon 1	BIGb0506	39.86	GCA_025961275.1	1
6	Providencia sneebia	DSM 19967	38.06	GCA_000314895.2	1
7	Providencia hangzhouensise	PR-310	40.43	GCA_029193595.2	137
8	Providencia zhijiangensis	D4759	42.85	GCA_030315915.1	5
9	Taxon 2	PR_162	40.92	GCA_003936755.2	4
10	Providencia rettgeri	NCTC11801	40.35	GCA_900455085.1	103
11	Taxon 3	PROV167	40.27	GCA_028480775.1	80
12	Providencia huaxiensis	WCHPr000369	40.83	GCA_002843235.3	16
13	Providencia wenzhouensis e	R33	41.47	GCA_019343475.1	6
14	Taxon 4	PROV227	41.32	GCA_028479185.1	9
15	Providencia alcalifaciens	DSM 30120	41.86	GCA_000173415.1	31
16	Providencia heimbachae	ATCC 35613	40.41	GCA_001655055.1	5
17	Providencia vermicola	DSM 17385	41.17	GCA_020381325.1	2
18	Providencia rustigianii	DSM 4541	41.29	GCA_000156395.1	9
19	Taxon 5	JGM172	42.55	GCA_018257405.1	16
20	Taxon 6	2019-04-29291-1-1	42.55	GCA_915403165.1	1

^a Owing to the unavailability of a genome for Providencia entomophila, this species was excluded from the updated classification.

^b Taxa identified in this study are marked in bold.

^c The type strain is selected based on the earliest time of release or the deposited type strain in the NCBI GenBank database.

^d Average GC content within species.

^e The species name is not validly published.

The majority of strains of Providencia (n = 137, 25.14%) were identified as P. hangzhouensis. This was followed by P. rettgeri (n = 103, 18.90%) and P. stuartii (n = 87, 15.96%; Table 3). Notably, a substantial portion of genomes (n = 136, 24.95%) could not be classified under existing species names. We have tentatively named these as Taxon 1–7, with Taxon 3 being the most prevalent (n = 80, 58.82%; Table 3). Further characterization of Taxon 1 through 7 using phenotypic methods is necessary to establish their species status and assign proper species names, in accordance with the current ICNP (28).

However, the increase in newly recognized species presents a new question: does this genus stand as a separate entity, or is it a conglomerate of multiple genera? To further scrutinize the Providencia genus, we calculated the pairwise average amino acid identity (AAI) values among all genomes within this genus. The values were found to be greater than or equal to 81.26%, surpassing the acknowledged genus-level threshold, which typically ranges between 65% and 72% AAI (29). This means that Providencia presently is indeed a distinct genus comprising 20 species.

Genome diversity within the Providencia genus

Variations in genome size, protein-coding sequences, and GC content both within and between species indicate the presence of significant genetic diversity within the Providencia genus (Table 3; Table S2; Fig. S1). To gain insight into the genome diversity of the Providencia genus, a maximum likelihood (ML) phylogeny was constructed from a concatenated core-gene alignment comprised of 2,077 loci present (covering 2.23 M nucleotides) in 95% of the 545 Providencia strains collected in our study. Deep divisions within this phylogeny, as depicted in Fig. 2, correspond to both the updated genus taxonomy described above and species-level grouping calculated using ANI with a cutoff of 96% for clustering. Applying a hierarchical Bayesian analysis of population structure (BAPS) to the core single-nucleotide polymorphism results, we clustered the 545 genomes into 23 monophyletic lineages, designated L1–L23.

Fig 2.

The ML phylogeny of the Providencia genus based on the core-genome alignment of 545 isolates. The branch colors correspond to phylogroups, which are clustered by ANI, while the shading of clades represents lineages as divided by fastbaps. Unless otherwise specified, the color schemes for all species and lineages in this study are in accordance with those established in Fig. 2.

Intriguingly, the phylogenetic analysis revealed that L19 are composed of three species, including P. zhijiangensis and Taxon 5–6 (Fig. 2). A comparison of pairwise ANI within these species has shown ANI values ranging between 94.25% and 94.63%. These values approached the threshold typically used to delineate species boundaries, implying a close relationship among these species.

Pangenome analysis of Providencia genus

Prior pangenomic studies on individual Providencia species have highlighted significant variability within the pangenome (18). After a comprehensive reclassification of the entire Providencia genus, we constructed its pangenome using a population structure-aware approach (30). The pangenome of the 545 Providencia strains comprises 34,087 gene families, with 2,077 of these designated as traditional core genes due to their presence in at least 99% of all genomes. However, when considering only those core genes present in more than 95% of isolates within each lineage, a total of 1,550 genes emerged as core to all lineages, thereby forming the collection core (Fig. 3A and B). Collectively, these 1,550 core genes represent approximately a third of genes in an individual isolate and constitute 4.55% of the total pangenome. The pangenome accumulation curves for species with multiple instances showed an open pangenome tendency in almost all species, excluding Taxon 4 (Fig. 3C). Our pangenome analysis also identified species or lineage-specific genes (Fig. 3B). For instance, in P. stuartii and Taxon 3, a distinct set of unique core genes was observed, contrasting with other species (Fig. S2 and S3). Conversely, a significant proportion of genes in Taxon 7 are also found in other species, primarily in P. stuartii (Fig. S4). Additionally, in P. hangzhouensis L13 and L14, most of the core lineage genes are spread across various lineages (Fig. S5 and S6).

Fig 3.

The pangenome analysis of Providencia genus. (A) Pangenome gene presence and absence heatmap along with the ML tree. The arrangement of genes in the heatmap adheres to the gene classification as defined by twilight (30). (B) A UpSetR plot of intersections of lineage-specific core genomes generated by UpSetR package (31). Each row symbolizes a distinct lineage, while each column indicates their intersections within the matrix, denoted by black dots. (C) Pangenome accumulation curves of each species (n > 1).

Functional characterization of the Providencia genus-core genes

Functional analysis of the 1,550 core genes of the Providencia genus was conducted using the Clusters of Orthologous Groups (COG) approach, aiming to shed light on the foundational functional characteristics and metabolic abilities of the genus. The analysis revealed that these core genes predominantly support essential cellular processes, encompassing coenzyme metabolism (32), inorganic ion transport, and DNA replication (Fig. 4A). Among these, coenzymes stood out as they made up the most substantial proportion at 10.56%, emphasizing a pronounced activity of the Providencia genus in intermediate metabolism. An additional 4.82% of genes associated with “replication, recombination and repair” could be indicative of potential horizontal gene transfer (HGT) events (33). Notably, a considerable fraction, 15.44%, of the core genes has yet to be assigned specific functions, suggesting uncharted territories in the functional landscape of the genus that beckon further exploration.

Fig 4.

The bar chart presents the functional annotation of 1,550 core genes, systematically arranged following the COG (A), Gene Ontology (GO) (B), and Kyoto Encyclopedia of Genes and Genomes (KEGG) (C) classification schemes.

Supplementary insights from Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses complemented our understanding. There was a marked enrichment in metabolic processes, unveiling key metabolic pathways in the genus, spanning organic substance, cellular, and nitrogen compound metabolism (Fig. 4B). Interestingly, the KEGG analysis also brought to light pathways related to “beta-Lactam resistance” (Fig. 4C), suggesting a potential tendency within the genus toward resistance against β-lactams, a critical class of antibiotics.

Resistome characteristics of Providencia genus

Antibiotic resistance has been documented in various degrees among Providencia species in clinical settings (34, 35). In an effort to understand the landscape of antibiotic-resistance genes (ARGs) in the Providencia genomes, we performed an analysis on 545 strains, primarily sourced from public databases. Of these, we identified 164 genes spanning 14 categories of antibiotic resistance, ranging from aminoglycosides to tigecycline (Fig. S7). Overall, the presence of ARGs was detected in 399 genomes (73.21%), encompassing the vast majority of lineages, with the exception of L6 (P. rustigianii), L19 (P. burhodogranariea and Taxon 1), L20 (P. wenzhouensis), and L23 (P. sneebia). In our attempt to find distinctive resistance genes unique to each species or lineage, no such specific genes were discovered. Despite this, P. stuartii L4 presented an interesting scenario. We noticed a high frequency of three specific genes—aac(2')-Ia, tet(B), and catA3—in this lineage. More precisely, aac(2')-Ia was found in 98.85% of the P. stuartii genomes, tet(B) in 96.55%, and catA3 in 94.25%. The prevalent presence of these genes in this lineage may provide a plausible explanation for the frequent categorization of P. stuartii as a multidrug-resistant organism.

β-Lactams, commonly employed in clinical practice, are often central to the treatment of resistant infections, with carbapenems being the last resort. In light of this, we focused our study on carbapenem-resistance genes and identified 17 such genes. Among these, bla_IMP-27 and bla_NDM-1 were the most prevalent (Fig. S7 and S8). Our investigation into the diversity and abundance of these genes revealed that the clinical lineages L4 (P. stuartii) and L12 (P. hangzhouensis), primarily of human origin (>75%), exhibited a higher diversity of genes (47.06%, 8/17) compared to other lineages. This finding suggests that these lineages possess greater number of antibiotic-resistance genes, warranting further investigation. Additionally, despite the non-clinical lineage L3 (Taxon 3) having only 50% human origin, we identified a significant presence of carbapenem-resistance genes in it (Fig. S8 and S9). The preponderance of these resistance genes in this lineage underscores the need for heightened vigilance regarding the emergence of this species in the clinical setting.

Contribution of plasmids to the Providencia genus

To elucidate the potential impact of plasmids on the Providencia genus, we conducted a meticulous examination of plasmid contigs within 545 genomes. Our investigation led to the identification of 721 putative plasmids across 342 genomes spanning 16 distinct species, of which 294 (40.78%) are present in P. hangzhouensis. The plasmid data set displays a broad spectrum of sizes (~1–240 kb) and GC content (~28%–62%), highlighting their diversity within the Providencia species (Supplementary Data 2; Fig. 5B; Fig. S10B and C).

Fig 5.

Predicted plasmids across Providencia genus. (A) Distribution of the 71 plasmid clusters against the ML phylogeny. (B) Distribution map of plasmids carried by various species. Plasmid size is denoted on the horizontal axis, whereas the vertical axis corresponds to the GC content. Distinct shapes are employed to symbolize various predicted host sources.

Cluster analysis of the plasmids revealed that the 721 plasmids could be further categorized into 71 clusters (Fig. 5A). Intriguingly, cluster 47, identified as the Col3M plasmid, emerged as the most dominant, comprising 29.40% (212/721) of the total, and was found in a major fraction of species harboring plasmids (68.75%, 11/16). Further investigation into the average GC content of the ColM plasmid revealed a similarity with the average GC content of its genus, suggesting that this plasmid may be a genus-specific carrier (36, 37). Though only 24.41% of the identified plasmids were predicted to be either conjugative or mobilizable, their hosts were diverse, spanning across multiple genera and phyla (Fig. S10A). This observation could be attributable, in part, to an insufficient exploration of this genus’s plasmids and potentially to frequent HGT activity within the plasmid, leading to loss of the corresponding transfer element.

Given the significance of plasmids in the dissemination of ARGs, we focused our attention on the distribution of ARGs within these clusters. Of the 3,757 ARGs, 2,019 (54.27%) were identified on plasmids, indicating a heightened burden of ARGs carried by these entities. A standout observation centered on cluster 0. Not only did this cluster claim the highest count of resistance genes, encompassing an impressive 36.97% of the overall ARGs, but it also emerged as a primary reservoir for carbapenemase genes, with a notable 34 out of the 66 identified carbapenemase genes localized within it (Fig. S11). Intriguingly, cluster 0 predominantly consists of large MOBH conjugative plasmids that are directly associated with the globally disseminated multidrug resistance (MDR)-associated IncC type plasmids (38), further emphasizing the significant clinical implications of our findings.

Furthermore, we found that plasmids bore a higher load (43.79%, 645/1473) of insertion sequences (ISs) associated with HGT. Certain ISs demonstrated a strong association with plasmids, with the most robust correlation observed for IS26 (Fig. S12). Previous research has established that IS26 can instigate plasmid reorganization in clinical settings via replicative transposition (39), as well as amplify ARGs (40).

DISCUSSION

In this study, we first identified and characterized a novel species, Providencia zhijiangensis sp. nov. Intriguingly, upon combining phylogeny and overall genome relatedness indexes (OGRIs) (including ANI and dDDH), we found that P. stuartii and P. thailandensis actually belong to the same species, contradicting the results of the biochemical identification conducted by Khunthongpan et al. (7). This discrepancy could be attributed to variations in measurement conditions or intra-species differences. Through our analysis of OGRIs and phylogeny, we unearthed a remarkable finding: the Providencia genus, previously assumed to comprise only 14 species, in fact includes 20 species (Table 3). This updated count includes our previously identified P. hangzhouensis (17), along with seven unnamed taxa (Taxon 1–7) which warrant further phenotype-based characterizations. Additionally, upon precise identification, we found that the labels of a large number of genomes are inconsistent with their actual species (Fig. 2), indicating significant misclassification. This is especially evident in the case of P. rettgeri, where 166 strains initially labeled as P. rettgeri were in fact primarily composed of the species P. hangzhouensis and Taxon 3 (Fig. 2; Supplementary Data 1). These observations underscore the prevailing confusion in the current classification of Providencia, highlighting the importance of accurate reclassification of species within this genus as accomplished in this study.

The considerable variance in genomic characteristics among different species and lineage divisions suggests a diverse genome within the genus. It appears unusual that lineage 19 occupies the ecological niche of P. zhijiangensis and Taxon 5–6, although ANI values approaching the species threshold (94.25%–94.63%) indicate a high degree of similarity between these three species. Nonetheless, further internal classification requires a larger pool of genomes.

The significant diversity observed in the Providencia pangenome, as underscored by earlier research (18), suggests notable genomic plasticity within the Providencia species. This plasticity is further reflected in our findings, where a significant portion of the pangenome remains open. It is intriguing that, despite sharing a phylogenetic branch and a similar core-gene count with P. wenzhouensis, Taxon 4 exhibits a closed pangenome, whereas P. wenzhouensis has an open one. This could shed light on distinct evolutionary pathways or pressures faced by these lineages.

Species-specific genes frequently delineate the unique ecological niche of individual organisms, a notion reflected in our pangenome analysis which uncovers species- or lineage-specific genes (Fig. 3B). Our analysis reaffirms this notion by identifying lineage-specific genes, suggesting distinct ecological or evolutionary pressures on different lineages. The observed gene specificity in P. stuartii and Taxon 3, for example, hints at possible adaptive strategies unique to these taxa. Conversely, the gene flow observed in Taxon 7 and P. hangzhouensis suggests potential interactions or shared evolutionary histories with other lineages. Furthermore, pangenome analysis revealed noticeable core-gene flow among different species, promoting the shaping of species diversity. The primary metabolic pathways and functional characteristics of the genus’s core genes were also discerned. The pathway associated with β-lactam resistance could be related to the frequent acquisition of antibiotic-resistance genes within this genus, highlighting potential adaptive strategies in the face of antibiotic pressures.

The considerable presence of antibiotic-resistance genes, coupled with their wide spectrum, highlights a potential frequent acquisition of such genes within species of this genus, typically ascribed to HGT events. Prior research has demonstrated the presence of multiple types of integron structures within the genus to carry resistance genes (18). In line with this, our study discovered that plasmids play a larger role in carrying resistance genes, with nearly half of these genes found to be plasmid-borne. Particularly, plasmid cluster 0 emerged as a significant locus, not only accounting for a remarkable proportion of the overall ARGs but also establishing itself as a primary reservoir for carbapenemase genes. This dominance suggests that cluster 0 may play a critical role in the dissemination of resistance genes, particularly carbapenem-resistance genes, through plasmid-mediated transfer.

Upon delving into the species-specific resistance genes, we singled out three high-frequency resistance genes in P. stuartii, indicating they might be intrinsic to this species. Regarding the clinically significant carbapenem-resistance genes, Taxon 3, P. stuartii, and P. hangzhouensis L12 displayed the greatest number of such genes along with diverse resistance genotypes. While this observation hints at a potential heightened clinical concern related to these species or lineages, it is important to note the limitations due to potential database biases and highlight the need for more comprehensive studies and continuous monitoring.

While our study provides valuable insights into the taxonomy and antibiotic resistance of the Providencia genus, certain limitations must be acknowledged. A substantial portion of our data was sourced from public databases. Although these repositories offer a wealth of genomic data, they can introduce biases. Specifically, strains exhibiting unique phenotypes, heightened resistances, or of particular clinical or scientific interest might be overrepresented due to targeted sequencing efforts. These selection biases can influence the observed distribution of resistance genes or specific genomic features. For instance, there is a possibility that resistant strains are sequenced more frequently, leading to an overrepresentation of resistance genes in publicly available data sets. Given these considerations, while our findings shed light on notable patterns and trends within the Providencia genus, the derived conclusions, especially regarding antibiotic resistance, should be interpreted with caution. Future studies aiming for a more holistic view of this genus would benefit from systematic, unbiased sampling, coupled with experimental validation and detailed epidemiological data.

In conclusion, this study addresses the previously perplexing taxonomical classification of species within the Providencia genus. It not only identifies a novel species—Providencia zhijiangensis sp. nov.—but also uncovers a synonymous pair, P. stuartii and P. thailandensis, in addition to expanding the total number of species in the genus number to 20 (Table 3). These findings significantly broaden the genomic landscape of Providencia, pointing out critical gaps in our understanding of the phylogenetic space it occupies. Through the investigation of core genes, we unveiled the primary metabolic pathways within the genus. Further explorations into antibiotic-resistance genes and plasmids underscored the critical role of plasmids in carrying resistance genes. Additionally, the diversity of carbapenem-resistance genes within three specific species or lineages signals the need for enhanced surveillance of these organisms in the future.

MATERIALS AND METHODS

Strain collection and whole-genome sequencing

To achieve a comprehensive evaluation of the Providencia genus, we adopted a holistic approach. Specifically, given the rarity of Providencia strains in our hospital, we obtained and sequenced four unique human-derived strains in 2022, representing the entirety of the samples available to us at that time. These strains were isolated from diverse clinical sources: two from urine, one from bile, and one from tissue fluid, reflecting a broad spectrum of clinical contexts.

Simultaneously, to ensure a comprehensive assessment of the Providencia genus, we extracted all available Providencia genomes from the NCBI’s GenBank database as of 1 December 2022, seeking to capture the broadest genomic diversity possible. This preliminary count indicated a total of 735 genomes. Recognizing the importance of data quality, we subjected each genome to a meticulous quality control assessment using QUAST v5.2.0 (41) and CheckM v1.2.2 (42). Genomes that failed to meet our quality standards (contig counts <300, N50 >50 kb, completeness >90%, contamination <2%) were excluded from subsequent analyses (Supplementary Data 1).

Phenotypic characterization of Providencia zhijiangensis sp. nov

As described previously (43, 44), the Gram stain was performed and the biochemical characteristics were determined using the bioMérieux API 20E and API 50CH kits according to the manufacturer’s instructions. Oxidase activity was determined using oxidase reagent (bioMérieux). The whole-cell fatty acids of strain D459 were assessed by the Guangdong Institute of Microbiology (Guangzhou, Guangdong, China). Meanwhile, in vitro antimicrobial susceptibility tests were conducted using Vitek II through broth microdilution. Breakpoints were determined using Clinical and Laboratory Standards Institute (CLSI) criteria (45), with the exception of tigecycline, for which the European Committee on Antimicrobial Susceptibility Testing (http://www.eucast.org/) guidelines were adopted.

Genome assembly and OGRIs calculations

The genomic DNA of the isolates was extracted with the AxyPrep bacterial genomic DNA miniprep kit by Axygen Scientific, located in Union City, California, USA. The whole-genome sequencing was carried out using two platforms: the Illumina HiSeq 2500 system, which performed a paired-end run with 2 × 150 base pairs, and the Oxford Nanopore MinION platform. Raw sequencing reads underwent preprocessed with fastp v0.23.2 (46) for adapter sequences and low-quality bases trimming. Trimmed paired-end reads were subjected to assembly using shovill v1.1.0 (https://github.com/tseemann/shovill), with contigs shorter than 200 bp being filtered out. The complete D4759 genome was assembled by integrating the short-read data from Illumina sequencing with long-read data from Oxford Nanopore sequencing using the Unicycler hybrid assembly pipeline (47). The 16S rRNA gene sequences of D4759 were obtained using PCR, with the application of universal primers 27F and 1492R (48). The related gene sequences from other species were sourced from the EzBioCloud database (27). All these 16S rRNA gene sequences were then aligned for comparison using MAFFT v7.505 (49). Subsequently, a phylogenetic tree was constructed using FastTree (50) based on the maximum likelihood method. OGRIs including ANI and in silico dDDH were assessed using the fastANI v1.33 (51) and genome-to-genome distance calculator (formula 2) (25, 52), respectively. A cutoff of≥96%ANI (24) or≥70.0% in dDDH (24, 25) was employed to define a bacterial species. Further identification of the accurate bacterial species based on the results of TYGS database (53). Pairwise average AAI between genomes was computed using CompareM v0.1.2 (https://github.com/dparks1134/CompareM), with thresholds (65%–72%) for genus-level ranks (29, 54).

Pangenome analysis

All genomes were annotated with Bakta v1.7.0 (55). Pangenomes of all the isolates included in this study were analyzed using Panaroo v1.3.2 (56), with the annotated contigs in GFF3 format generated by Bakta as input. To generate the core-gene alignment, the parameters were set to a moderate model and a 95% threshold for protein sequence identity similarity and length difference cutoff as previously described (57). Additionally, MAFFT v7.505 (49) was specified with the “-a” flag to align the core genes present in at least 95% of the genomes.

A pangenome gene classification, which is attuned to the population structure and based on within-species distribution, is derived using twilight package (30). Groups are delineated by the lineages established via fastbaps (as detailed below). The minimum size genome of the group is assigned a value of 1 to include all species.

Phylogenetic analysis and population structure

An ML phylogeny was constructed based on the core-genome alignment generated by Panaroo using RaxML v8.2.12 (58) with a general time reversible (GTR) nucleotide substitution mode and 1,000 rapid bootstrap replications. The phylogeny was plotted and annotated using ggtree v3.7.1 (59). The core-genome alignment and the phylogeny were subjected to fastbaps v1.0.8 (60) to infer population structure.

Function classification

The COG functional categories of the gene sets for the Providencia genus were classified using eggNOG-mapper v2.1.9 (61). GO and KEGG ortholog assignment were annotated using InterProScan v5.56 (19) and KofamScan v1.3.0 (62), respectively.

Identification of ARGs and ISs

ARGs were identified using ABRicate v1.0.0 (https://github.com/tseemann/abricate) with a minimum identity threshold of 90% and a minimum coverage threshold of 90% based on NCBI AMRFinder Plus database. ISs were annotated via ISfinder (63).

Plasmid annotation and clustering

Plasmids from all assemblies were detected and reconstructed using the mob-recon tool from the MOB-suite toolkit v3.1.2 (64), employing default parameters. Plasmid clustering was undertaken using RabbitTClust v.2.2.1 (65) with a cutoff of a mash distance of 0.05.

DATA AVAILABILITY

The whole-genome sequences of four Providencia isolates from this study have been deposited in the GenBank database under BioProject accession PRJNA983084. The complete genome, including the chromosome and plasmids of strain D4759, has been submitted to NCBI GenBank under accession numbers CP135990-CP135992. All genome information used in this study can be found in Supplementary Data 1.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/msphere.00731-23.

Data S1. msphere.00731-23-s0001.xlsx.

Metadata for the isolates included in this study.

Data S2. msphere.00731-23-s0002.xlsx.

Details of plasmid sequences analyzed in this study.

Supplemental figures and tables. msphere.00731-23-s0003.pdf.

Fig. S1-S12; Tables S1 and S2.

Descriptions. msphere.00731-23-s0004.pdf.

Descriptions of Data S1 and S2.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.