Comparative genomic analysis of Chinese human leptospirosis vaccine strain and circulating isolate

ABSTRACT

Leptospira interrogans

serogroup Canicola is one of the most important pathogens causing leptospirosis and is used as a vaccine strain of the current Chinese human leptospirosis vaccine. To characterize leptospiral pathogens, L. interrogans serogroup Canicola vaccine strain 611 and circulating isolate LJ178 from different hosts at different periods were sequenced using a combined strategy of Illumina X10 and PacBio technologies, and a comprehensive comparative analysis with other published Leptospira strains was conducted in this study. High levels of genomic similarities were observed between vaccine strain 611 and circulating isolate LJ178; both had two circular chromosomes and two circular extrachromosomal replicons. Compared with the strain 611 genome, 132 single nucleotide polymorphisms and 92 indels were found in strain LJ178. The larger lipopolysaccharide biosynthesis locus of serogroup Canicola was identified in both genomes. The phylogenetic analysis based on whole-genome sequences revealed that serogroup Canicola was not restricted to a specific host or geographic location, suggesting adaptive evolution associated with the ecologic diversity. In summary, our findings provide insights into a better molecular understanding of the component strains of human leptospirosis vaccine in China. Furthermore, these data detail the genetic composition and evolutionary relatedness of Leptospira strains that pose a health risk to humans.

Introduction

Leptospirosis, caused by pathogenic species in the genus Leptospira, is a serious zoonotic disease.¹ Approximately 1.03 million human leptospirosis cases and 58,900 deaths occur worldwide each year, making it a leading zoonotic cause of morbidity and mortality.² Most cases occur in developing and underdeveloped countries, but international travel and global warming have led to an apparent surge in incidence in industrialized countries. In China, the earliest leptospirosis case can be traced to the 1920s,³ and more than 2.5 million cases and over 20,000 deaths have been reported in the country to date. Ten large outbreaks of leptospirosis, with an incidence of more than ten cases per 100,000 people, have occurred in China in the past 60 y.⁴ Therefore, a better molecular understanding of the pathogens associated with this disease would be of value.

Vaccination of at-risk populations remains the most effective approach to controlling leptospirosis. An inactivated whole-cell vaccine was first used in the 1920s,⁵ and it is still used in humans and other animals in some countries.^6-10 In China, a leptospirosis vaccine was successfully developed in 1958, and production of this vaccine gradually improved with biotechnological advances. Currently, a multivalent, inactivated leptospirosis vaccine containing seven major circulating L. interrogans serogroups, namely Icterohaemorrhagiae, Canicola, Grippotyphosa, Autumnalis, Pomona, Australis, and Hebdomadis, is recommended for vaccination of high-risk persons aged 7–60 y in major Leptospira-epidemic regions.¹¹ Because inactivated vaccines induce serogroup-specific immunity, the epidemiology of Leptospira must be monitored to guide the production of appropriate vaccines. Although the predominant serogroups have been consistent in China and old vaccine strains are still in use,^12,13 whether current circulating strains belong to the same serogroup as the vaccine strain isolated in the 1950s is unknown.

Canicola is one of the most important pathogenic serogroups in the genus Leptospira.³ Dogs are considered the primary reservoir hosts of L. interrogans serogroup Canicola, although this serogroup has also been found worldwide in humans, swine and cattle.¹⁴ Clinical cases in dogs and humans have reemerged in several countries,^15–17 implicating transmission between dogs and humans, which highlights the need to investigate this pathogen in dogs and review relevant prevention strategies. Similarly, leptospirosis cases associated with serogroup Canicola infections are increasing in China.^18,19 In the 1990s, of the 176 Leptospira isolates identified from eight provinces, which 10 (5.7%) belonged to the Canicola serogroup,¹⁸ and, in 2002–2015, the most prevalent Leptospira serogroups in various hosts in southern China were Icterohaemorrhagiae (61.1%), followed by Javanica (19.2%), Australis (9.7%), and Canicola (3.0%).¹² Most importantly, a multilocus variable-number tandem repeat analysis (MLVA) found that Leptospira isolates from different hosts were distinct from those of the seven vaccine strains used in China,¹³ indicating the need to select new vaccine strains. Therefore, it is important to understand the genetic composition and evolutionary relatedness of the vaccine strain and circulating isolates, which can provide critical insights for the development of whole-cell and recombinant leptospirosis vaccines. In the present study, we compared the whole genomes of a L. interrogans serogroup Canicola vaccine strain and circulating strain isolated from different hosts at different times. In addition, the molecular evolution of L. interrogans serogroup Canicola was investigated.

Materials and methods

Leptospira strains and cultures

L. interrogans serogroup Canicola vaccine strain 611, which was isolated from a human in the 1950s, was used by Wuhan Institute of Biological Products Co., Ltd., to generate a multiple-valent phenol-inactivated leptospiral vaccine. The vaccine strain is passaged in guinea pigs prior to production to preserve virulence, which is closely associated with efficacy.¹¹ The predominant circulating serogroup Canicola strain LJ178 was isolated from a dog in 2009 in Jiangxi, China.¹³ Both leptospiral strains were cultivated at 28°C on Leptospira Protein Medium Base Ellinghausen McCullough Johnson Harris (EMJH) supplemented with 10% rabbit serum. Genomic DNA was extracted using a Wizard Genomic DNA Purification Kit (Promega, Southampton, UK) following the manufacturer’s instructions. Integrity and quality of the extracted DNA were analyzed by agarose gel electrophoresis, and then the DNA was stored at – 70°C.

Genome sequencing and annotation

Genomes were sequenced using an Illumina X10 platform with a paired-end sequencing protocol and a Pacific BioSciences (PacBio) RSII platform with a HGAP SMRT Portal protocol. Assemblies based on Illumina data were polished in Pilon.²⁰ Both genomes were assembled de novo into two circular chromosomes and two plasmids, and then deposited in GenBank under accession numbers CP044513-6 and CP044509-12, respectively.

Comparative analysis

Putative protein-coding sequences were determined based on predictions in GeneMark,²¹ and functional annotations of coding sequences were obtained from the NCBI nonredundant protein database. Clusters of orthologous groups (COGs) and metabolic pathways were assigned in RPS-BLAST using the NCBI CDD library and Kyoto Encyclopedia of Genes and Genomes database, respectively.²² Putative protein subcellular localization was predicted using the SignalP 4.1 Server (http://www.cbs.dtu.dk/services/SignalP/), LipoP 1.0 Server (http://www.cbs.dtu.dk/services/LipoP/), and TMHMM Server 2.0 (http://www.cbs.dtu.dk/services/TMHMM/), respectively. Insertion sequence (IS) elements were identified using the ISfinder database (http://www-is.biotoul.fr/). To gain greater insight into the evolution of L. interrogans serogroup Canicola, published genomes of other L. interrogans serogroup Canicola strains and Leptospira species were included in the phylogenetic analysis (Supplementary Table 1).^23-30 Whole genomes were aligned using Mauve software.³¹ A phylogenetic tree was constructed by concatenating protein sequences of orthologs in PHYML as previously described.^32,33 The orthologs were genes having a minimum coverage of 80% and minimum identity of 80% with other strains in this study.

Results

Genomic features

Whole-genome sequencing of strains 611 and LJ178 was performed using Illumina X10 and PacBio technology platforms. PacBio RSII sequencing yielded 200,080 and 189,870 adapter-trimmed reads (subreads) with an average length of 8,701 bp and 8,556 bp, which corresponded to 363- and 339-fold coverage for strains 611 and LJ178, respectively. The subreads of each strain were assembled into four circular contigs. Then, 1.53-Gb Illumina reads for strain 611 and 1.18-Gb Illumina reads for LJ178 were mapped to contigs, and sequencing errors were corrected. The completed genome of each strain consisted of two circular chromosomes and two circular extrachromosomal replicons (Figure 1). Characteristics of the two genomes are summarized in Table 1. Chromosome I of strain 611 and LJ178 contained 4,255,595 bp and 4,259,066 bp, respectively, and chromosome II contained 357,285 bp and 357,037 bp. These genomes were approximately 1 Mb larger than the smallest Leptospira genome, that of L. borgpetersenii, which is 3.88 Mb.²⁵ The average G + C contents of the two genomes were 35.1% and 35.1%, respectively. Gene prediction revealed that 85.8% of each genome comprised coding sequence, containing 4,424 genes with an average size of 826 bp in strain 611 and 4,413 genes with an average size of 828 bp in strain LJ178. Among protein-coding genes in strains 611 and LJ178, 52.1% and 52.2%, respectively, were assigned to a COG functional category (Supplementary Tables 2 and 3). Further, each genome contained 37 tRNAs genes and three ribosomal RNA operons (Table 1).

Figure 1.

Genomic maps of L. interrogans serogroup Canicola vaccine strain 611 and circulating isolate LJ178. A-B and E-F: Moving toward the inside, the first and second circles separately represent the coding genes in the plus and minus strands of chromosome I and II. Functional COG categories are delineated by default color. The third and fourth circles represented tRNA (red) and rRNA (blue) of the plus and minus strands. The fifth and sixth circles represented GC content and GC skew, respectively. C-D and G-H: the first and second circles separately represent the predicted protein-coding regions (forward and reverse strands, respectively). Functional COG categories are delineated by default color. GC content and GC skew are depicted in circles 3 and 4, respectively.

Table 1.

General features of the L. interrogans serogroup Canicola vaccine strain 611 and circulating isolate LJ178.

	Strain 611				Strain LJ178
Genome features	CI	CII	lbp1	lbp2	CI	CII	lbp1	lbp2
Genome size (bp)	4,255,595	357,285	75,928	66,534	4,259,066	357,037	75,855	66,530
G + C content (%)	35.05	35.08	34.66	33.37	35.05	35.07	34.66	33.37
Protein coding (%)	76.85	77.76	74.97	74.4	76.84	77.34	73.93	75.71
Total CDSs	3,954	332	64	74	3,952	329	63	69
CDSs with assigned function	1,947	169	25	13	1,935	166	26	13
CDSs without assigned function	2,007	163	39	61	2,017	163	37	56
Average CDS length (bp)	827	837	889	669	828	839	890	730
Transfer RNA	37	0	0	0	37	0	0	0
23S Ribosomal RNA	2	0	0	0	2	0	0	0
16S Ribosomal RNA	2	0	0	0	2	0	0	0
5S Ribosomal RNA	1	0	0	0	1	0	0	0

CI represents chromosome I;CII represents chromosome II; lpb1 represents plasmid1; lpb 2 represents plasmid 2.

Chromosomes of the two strains showed high synteny and 99% identity at the DNA level. The sequences of strain 611 and LJ178 chromosomes were extensively collinear, disrupted by few rearrangements; in contrast, L. interrogans serovar Lai did not show this collinearity with the two strains (Figure 2). The strain 611 genome shared 4034 orthologs with strain LJ178 and 3213 orthologs with L. interrogans serovar Lai; a total of 3208 orthologs were identified in the three L. interrogans strains (Figure 2). It was also noted that 140 and 130 had no orthologs in strain 611 and LJ178 compared with other genomes. Although most genes were predicted to encode hypothetical proteins, and the putative functions of some genes were related to metabolism (data not shown). Furthermore, compared with the strain 611 genome, many more polymorphic sites were found in strain LJ178, including 132 single nucleotide polymorphisms (SNPs) and 92 small insertions or deletions (indels). Of the SNPs, 53 were located in open reading frames and 75 were in intergenic regions. Forty-one were non-synonymous SNPs, resulting in amino acid changes in putative proteins; the other 12 SNPs were synonymous. The 53 SNPs in coding regions corresponded to 16 different COGs, although 28.3% were not assigned to any COG. The highest frequency COGs were related to signal transduction mechanisms (13.2%) (Figure 3). In addition, one non-synonymous SNP was located in the gene (strain 611_gene 3774), which encodes a thermolysin homolog precursor. Further analysis showed that several SNPs in this gene in strain 611 have been found in other strains such as L. interrogans serovar Linhai str. 56609 and serovar Copenhageni str. Fiocruz L1-130.

Figure 2.

Comparison of replicons and the shared and unique orthologs of L. interrogans. (a) The replicons of L. interrogans serovar Lai strain 56601, and serogroup Canicola vaccine strains 611 and circulating isolate LJ178 were aligned by BLASTn. Colored lines drawn between two adjacent linearized chromosomes (horizontal gray lines) showed the location of homologous regions, and indicated regions of shared similarity in the same (red) or opposite (blue) direction. (b) Venn diagram showing the distribution of shared and unique proteins by L. interrogans species.

Figure 3.

Clusters of orthologous genes associated with the specific SNPs between L. interrogans serogroup Canicola vaccine strain 611 and circulating isolate LJ178 genomes.

Two plasmids were found in both strains 611 and LJ178 genomes. The plasmid sizes were 75,928 bp and 75,855 bp for lbp1, and 66,534 bp and 66,530 bp for lbp2, with overall GC contents of 34.7% and 33.4% for strains 611 and LJ178, respectively. Plasmids lbp1 and lbp2 comprised 64 and 63 (strain 611), and 69 and 74 (strain LJ178) genes (Supplementary Table 4), respectively. These were more than 10 Kb larger than plasmid lbp1 (65 Kb) and lbp2 (59 Kb), containing 39 and 35 genes, respectively, in the L. borgpetersenii serovar Ballum genome.³⁴ There were high similarities (>99% identity at the DNA level) between the plasmids in the two strains. The average percentages of coding sequence in lbp1 and lbp2 in these strains were 74.4% and 75.1%, respectively. Although most plasmid genes were found to encode hypothetical proteins (57.8%), genes related to DNA replication, recombination, and repair, and transcription were also identified in lbp1 and lbp2. Sequence comparisons showed that all genes in published plasmids Gui1 (74,981 bp) and Gui 2 (66,851 bp) in the L. interrogans serovar Canicola strain Gui44 genome³⁵ had orthologs of genes in both the lbp1 and lbp2 plasmids identified in this study.

Lipopolysaccharide (LPS) biosynthetic system

LPS plays an important role in generating immunity against leptospiral pathogens, as well as the serological classification of Leptospira.^36-39 Different from the LPS rfb biosynthetic loci in L. interrogans serovar Lai (36 Kb) and L. borgpetersenii serovar Hardjo (36.7 Kb),^27,38,40 the larger rfb loci in the strain 611 and LJ178 genomes were found to be located in chromosome I with approximate sizes of 89,019 and 88,812 bp and containing 92 and 93 genes, respectively (Supplementary Table 5). Comparison of rfb locus nucleotide sequences in strains 611 and LJ178 showed >99% identity, including 92 pairs of orthologous genes (Supplementary Table 5). One unique gene (LJ178_gene 1540) in the rfb locus of strain LJ178 encoded a type 11 methyltransferase (Supplementary Table 5).

Further analysis showed that the rfb locus of strain LJ178 shares 28 orthologs with L. interrogans serovar Lai. These orthologs encode putative products that include enzymes involved in the biosynthesis of activated sugars and glycosyltransferases for assembly of the O-antigen subunit, as well as integral membrane proteins for the transport of O-antigen subunits through cell membranes and assembly into LPS. The putative functions of these different genes were determined to be mainly related to cell envelope biogenesis and carbohydrate transport and metabolism, although most genes were predicted to be hypothetical proteins (Supplementary Table 5).

Putative protein subcellular localization and virulence factors

To further understand the function of putative proteins from L. interrogans serogroup Canicola vaccine strain 611 and circulating isolate LJ178, protein sequences of both strains were analyzed, and subcellular localizations of these proteins were predicted. Putative proteins (85.2% for strain 611 and 84.6% for strain LJ178) of these strains were predicted to be located within the cytoplasm (Supplementary Figure 1). A total of 89 and 115 putative proteins were found to be extracellular proteins, including some known Leptospira outer surface proteins (Supplementary Table 2). For example, Lig B (strain 611_gene 0524 and strain LJ178_gene 0529) and Lig C (strain 611_gene 2822 and strain LJ178_gene 2812), containing 12 and 13 Ig-like domains, respectively, were predicted to be adhesion molecules involved in bacterial pathogenicity.⁴¹ We also identified 29 and 32 putative lipoproteins in strains 611 and LJ178, respectively. LipL41 (strain 611_gene 3367 and strain LJ178_gene 3367), which is highly conserved among pathogenic Leptospira species, is often used as a serodiagnostic antigen for the detection of Leptospira infection.⁴² Almost all outer surface proteins and lipoproteins that were different between both strains 611 and LJ178 were found to encode hypothetical proteins.

A BLAST search in the Virulence Factors Database showed 476 and 475 potential virulence factors in strains 611 and LJ178, respectively, including various adhesions, outer membrane proteins, hemolysins, and lipoproteins with demonstrated virulence in pathogenic Leptospira species and that are homologous to virulence factors in other pathogens (Supplementary Table 6). Furthermore, some major Leptospira virulence factors such OmpA-like protein Loa22 (strain 611_gene 0212 and strain LJ178_gene 0213), lipoproteins LipL32 (strain 611_gene 2431 and strain LJ178_gene 2433), LipL41, and outer membrane protein OmpA (strain 611_gene 4249 and strain LJ178_gene 4246) were found to have 100% identity with factors in pathogenic L. interrogans serovar Lai strain 56601. A potential virulence gene in strain 611 genome, which encodes a putative long-chain-fatty-acid-CoA ligase, was found to be one more than that in strain LJ178 (Supplementary Table 6). In addition, some known Leptospira antigenic lipoproteins were among the virulence factors identified. For example, LipL32 is highly immunogenic, inducing a strong antibody response during natural infection and in animal models.⁴³

IS elements and the CRISPR-Cas system

IS and mobile genetic elements have played important roles in the evolution of bacteria. IS analysis revealed 85 and 83 IS elements in strains 611 and LJ178, respectively, comprising eight types of IS elements (Table 2). Most IS elements (88.2% and 88.0%) were found in chromosome I in both strains, whereas ISLin2, IS1500A, and IS1500B were located in chromosome I, II, and plasmid 1, respectively. No whole prophage was identified in either strain using Phage_Finder software.

Table 2.

IS element identified in L. interrogans serogroup Canicola vaccine strain 611 and circulating isolate LJ178.

IS elements	Region of location	Number in strain 611	Number in strain LJ178
ISLin1	CI	9	9
ISLin2	CI	11	11
ISLin2	CII	2	2
ISLin2	lpb1	1	1
ISLbp4	CI	12	11
IS1533	CI	2	2
IS1500A	CI	11	11
IS1500A	CII	2	2
IS1500A	lpb1	1	1
IS1500B	CI	11	11
IS1500B	CII	2	2
IS1500B	lpb1	1	1
IS1501	CI	6	6
IS1501	lpb1	1	1
IS1502	CI	13	12
Total		85	83

CI represents chromosome I;CII represents chromosome II; lpb1 represents plasmid1.

Previous bacterial genome sequencing studies have identified clustered regularly interspaced short palindromic repeats (CRISPRs), which are believed to be associated with invasion and evasion of host immune responses.⁴⁴ CRISPR repeat arrays consist of a leader sequence followed by numerous conserved direct repeat sequences. In the strain 611 genome, three CRISPR repeat arrays were detected, with lengths of 303 bp, 899 bp, and 320 bp, respectively, and each containing between four and 12 spacer sequences (Supplementary Table 7). In contrast, four CRISPR-Cas arrays were identified in strain LJ178, with three of these highly conserved (>95% nucleotide identity) between strains 611 and LJI178. A unique CRISPR-Cas array with a length of 1120 bp and containing 15 spacer sequences was found in strain LJ178.

Phylogenomic analysis

Phylogenetic analysis revealed high genomic similarities (86.3–99.3%) between different L. interrogans serogroup Canicola genomes (Supplementary Table 8). For example, genomic identity between L. interrogans serogroup Canicola strains LJ178 and 56666 was 86.3%. Differences in genomic similarities by host and source of isolates were not observed (Supplementary Table 8). For example, strain LJ178 from a dog showed 99.2% similarity with L. interrogans serovar Canicola strain Fiocruz LV133, isolated from a Brazilian patient, and strain 611 from a human showed 97.1% similarity with L. interrogans serovar Canicola strain L0-3 from a sow. Pairwise comparisons of strains 611 and LJ178 with other L. interrogans serogroup strains indicated genomic similarities that ranged from 86.0% to 87.8% (Supplementary Table 8). In contrast, genomic diversity was high among pathogenic species, with sequence similarities ranging from 20.5% to 54.1% in pairwise comparisons between strains 611 and LJ178 and other pathogenic species (Supplementary Table 9). For example, strain LJ178 showed genomic similarities with strains L. borgpetersenii and L. kirschneri of 22.0% and 54.1%, respectively (Supplementary Table 9).

To better understand the molecular evolution of L. interrogans serogroup Canicola, a maximum likelihood phylogenetic tree that included published sequences of other Leptospira and L. interrogans serogroup Canicola genomes was constructed (Figure 4). Intermediate (including L. wolffii, L. licerasiae, L. inadai, L. broomii, and L. fainei) and pathogenic species formed the deepest branches, which diverged from branches of saprophytic species (including L. biflexa, L. yanagawae, L. meyeri, and L. wolbachii), with pathogenic species grouped in three major clades (Figure 4). As expected, intra-species distances were much less than inter-species distances. However, three distinct sublineages of L. interrogans were clearly demarcated. Isolates from the same serogroup were distributed in different sublineages within a species. Serogroup Canicola strains 611 and LJ178 formed a single sublineage, whereas other Canicola strains from China (i.e., strains 56603, 56666, and 56647) clustered with strains from other serogroups (i.e., Icterohaemorrhagiae, Grippotyphosa, and Pyrogenes) into a major sublineage in the phylogenetic tree. Another sublineage included three serogroup Canicola strains from Brazil and strain P2655 from Portugal. Furthermore, strain 611, associated with human leptospirosis, clustered with canine isolate LJ187. Similarly, the remaining two sublineages contained serogroup Canicola strains obtained from humans and other animals (Figure 4 and Supplementary Table 1).

Figure 4.

Phylogenetic relationship among Leptospira species. The tree was constructed using the concatenated orthologous proteins of each strain. Scale bar indicated an evolutionary distance of 0.05 amino acid substitutions per position for phylogeny Leptospira species, and 0.001 amino acid substitutions per position for L. interrogans. Bootstrap values were shown for branches separating different species or serovars. Asterisk (*) represented serogroup Canicola analyzed.

Discussion

To characterize a circulating strain of L. interrogans serogroup Canicola, we used high-resolution whole-genome sequencing in this study. Although PacBio sequencing, which is high throughput, fast, and produces longer reads than other platforms, has been used in many genomic studies, the error rate is fairly high, leading to lower accuracy of assembly.^45,46 To reduce errors and improve the assembly, the genomes of a L. interrogans serogroup Canicola vaccine strain and a circulating isolate were sequenced using a combined strategy with both Illumina X10 (short read) and PacBio (long read) technologies. Complete genomes of the two strains were obtained after de novo assembly, suggesting that our hybrid approach substantially improved the quality of bacterial genome assembly compared to that from second-generation sequencing alone. The genome of LJ178 was approximately 3 Kb larger than that of vaccine strain 611, and orthologs in the genomes showed high identity, confirming the similarity of the two serogroup Canicola strains from different hosts and times. These findings may provide genomic evidence to support representativeness of the vaccine strain, even though strains 611 and LJ178 showed different MLVA profiles.¹³ Furthermore, the ratio of non-synonymous to synonymous SNPs was 3.4:1 in both genomes. The high proportion of non-synonymous to synonymous changes indicates that some genes in each strain undergo positive selection. A recent study showed that thermolysin family proteins are synergistically involved in inactivation of host immune effectors during Leptospira infection.⁴⁷ SNP sites in the gene encoding the thermolysin homolog precursor may be associated with the evolution for infectivity of Leptospira strains, although most genes with SNPs encode unknown products. Further studies should be conducted to determine potential functional changes with SNPs, especially for those key proteins associated with immunogenicity.

Plasmids are one of the most important mobile elements for transmission of genetic information between bacteria; they can facilitate the evolution and adaptation of pathogens.⁴⁸ Consistent with the results of previous Leptospira studies,³⁵ two plasmids were found in each genome. Zero and three plasmids, which greatly contribute to Leptospira diversity, have been found in different pathogenic Leptospira species and serogroups. The two plasmids in the L. interrogans serogroup Canicola strains characterized in this study contained many novel genes, although the functions of most of these genes are unknown. High similarities were found between the sequences of plasmids from published genomes of three L. interrogans serogroup Canicola strains and the two strains in this study, whereas the two strains and L. borgpetersenii serovar Ballum showed low similarities (<20%),^34,49 suggesting that these plasmids play important roles in host adaptation of strains in different serogroups. Plasmids with extrachromosomal self-replicating activities and those that code their own cell-to-cell conjugal transfer systems are particularly important contributors to horizontal gene transfer among bacteria.⁴⁸ Indeed, we recently found that horizontal gene transfer facilitated gene gain associated with the emergence of pathogenic Leptospira.³⁰ Knowing the manner in which these plasmids evolve is important to understanding the adaptive evolution of different L. interrogans serogroups.

Differences in LPS rfb locus genes are directly associated with Leptospira serogroup and serovar diversity.^38,39 The whole rfb locus of serogroup Canicola was annotated in detail in this study. These loci were remarkably similar in the two L. interrogans serogroup Canicola strains, but they differed from those in other L. interrogans serogroup strains, reflecting the antigenic/serologic specificity of each serogroup.^38,39 Previously, Gerald and his colleagues showed that mutations affecting LPS can lead to the loss of virulence in pathogenic Leptospira species.⁵⁰ Furthermore, significant differences have been found in the LPS carbohydrates and lipids of pathogenic species versus intermediate pathogenicity and nonpathogenic species of Leptospira.⁵¹ Because of the large differences in rfb gene clusters among different serogroups, experimental investigations are needed to characterize and compare the LPS in different serogroup strains to better understand mechanisms of host adaptation and colonization.

Rodents and canids are important reservoirs of Leptospira isolates in China, and these hosts have played key roles in the transmission of Leptospira spp. to humans.^12,13 Leptospiral virulence factors such as various adhesions, lipoproteins, and outer membrane proteins have been extensively studied and reviewed.^1,52 Our results show that Canicola strains from both human and canids contain major leptospiral virulence factors, including lipoproteins LipL32 and LipL41 and OmpA-like protein Loa22, with 100% identity of these proteins in serogroup Canicola isolates from different hosts. These factors may contribute to machinery that allows animal strains to infect humans. On the other hand, unlike LPS, some virulence factors such as LipL32 are antigenically conserved among pathogenic Leptospira species regardless of the serovar or serogroup,^43,53 suggesting that these antigens may be used as components of a universal cross-reactive vaccine against Leptospira.

IS elements, the smallest and most numerous transposable elements have played important roles in shaping bacterial host genomes through transposition and recombination.⁵⁴ IS elements comprise approximately 3.7% of L. borgpetersenii genomes, and L. borgpetersenii is believed to have undergone an IS-mediated genome reduction that contributed to evolution of this bacterium from a host-specific pathogen.²⁵ Similar to other Leptospira strains,^20,25 numerous IS elements were present in the genomes of both L. interrogans serogroup Canicola strains. However, Leptospira species contain different copy numbers of IS elements, with 77 and 84 copies of IS 1533 found in L. borgpetersenii serovar Hardjo strains L550 and JB197, respectively.²⁵ In contrast, a low copy number (0–4) and variation in IS1533 were found in L. interrogans in this and previous studies,^20,55 perhaps reflecting a different mechanism of genome expression and evolution. Our findings show the value of comparing Leptospira IS copy numbers.²⁰

Our comparative genome analyses showed that inter-species genomic identities were much lower than intra-species identities, perhaps marking distinctions based on pathogenicity.^30,44 Recently, genomes of serogroup Icterohaemorrhagiae strains from distinct geographic locations were found to be highly conserved, suggesting that specific SNPs/indels play crucial roles in the survival of this serogroup in diverse habitats.⁵⁶ In agreement with these findings, high levels of genomic similarities were found among serogroup Canicola strains from different hosts and geographic sources, possibly implicating the sampled animals as reservoirs for human infections. Here, we showed the genetic relatedness of the two serogroup Canicola strains. In addition, our whole genomic phylogenetic analysis indicates that serogroup Canicola strains are not always restricted to specific hosts or geographic locations, suggesting adaptive evolution associated with the ecologic diversity.

The limitation of this study is that only one circulating strain from a dog was compared with the vaccine strain. Further comparative genomic studies should be performed with a large number of L. interrogans serogroup Canicola strains from different hosts and geographical regions in China.

Conclusions

In summary, we compared the whole genomes of a L. interrogans serogroup Canicola vaccine strain and circulating strain isolated from different hosts and at different times. The entire rfb locus in serogroup Canicola was annotated in detail in this study. High genomic similarities were observed between strains 611 and LJ178 from a human and a dog, although SNPs or indels were also identified. Functional studies will be required to understand whether these SNPs or indels are associated with phenotypes important for vaccine immunogenicity or safety. This whole-genome analysis provides high-resolution information on Leptospira genomes, contributing to a better understanding of Chinese human leptospirosis vaccines, and sublineages in the serogroup Canicola. In addition, our findings indicate that serogroup Canicola strains are not restricted by host or geography, which is encouraging for the current and continued effectiveness of current leptospirosis vaccines.

Funding Statement

This work was supported by the National Key R&D Program of China (No. [2018YFC1603900] and The National Natural Science Funds, China [81471968].

Disclosure of potential conflicts of interest

The authors report no conflict of interest.

Supplementary material

Supplemental data for this article can be accessed online at http://doi.org/10.1080/21645515.2020.1720439.