Comparative Genomics of Lactobacillus crispatus from the Gut and Vagina Reveals Genetic Diversity and Lifestyle Adaptation.

Lactobacillus crispatus colonizes the human feces, human vagina, and the crops and ceca of chicken. To explore the genetic characteristics and evolutionary relationships of L. crispatus isolated from different niches, we selected 37 strains isolated from the human vagina (n = 17), human feces (n = 11), and chicken feces (n = 9), and used comparative genomics to explore the genetic information of L. crispatus from the feces and vagina. No significant difference was found in the three sources of genomic features such as genome size, GC content, and number of protein coding sequences (CDS). However, in a phylogenetic tree constructed based on core genes, vagina-derived L. crispatus and feces-derived strains were each clustered separately. Therefore, the niche exerted an important impact on the evolution of L. crispatus. According to gene annotation, the L. crispatus derived from the vagina possessed a high abundance of genes related to acid tolerance, redox reactions, pullulanase, and carbohydrate-binding modules (CBMs). These genes helped L. crispatus to better adapt to the acidic environment of the vagina and obtain more nutrients, maintaining its dominance in the vagina in competition with other strains. In feces-derived bacteria, more genes encoding CRISPR/Cas system, glycoside hydrolases (GHs) family, and tetracycline/lincomycin resistance genes were found to adapt to the complex intestinal environment. This study highlights the evolutionary relationship of L. crispatus strains isolated from the vagina and feces, and the adaptation of L. crispatus to the host environment.


Introduction
Lactobacillus crispatus is an obligately homofermentative, catalase-negative, Gram-positive bacterium [1]. Numerous studies have shown the considerable probiotic potential of L. crispatus, such as modulating the host's immune system [2], reducing allergic symptoms in mice [3], down-regulating The study (Ethical No. KS202006) was approved by the clinical new technology and scientific research ethics committee of the Wuxi People's Hospital, Wuxi, Jiangsu province, China.

Genome Sequencing, Assembly, and Annotation
The draft genomes of L. crispatus were sequenced using Illumina Hiseq×10 platform (Majorbio BioTech Co, Shanghai, China). Each sample provided an amount of raw sequencing data with no less than 100× coverage depth of the genome. The genome sequences were assembled using the software SOAP denovo2 (v2.0) [21]. The protein-coding sequences (CDS) were predicted using Glimmer according to Delcher et al. [22]. The tRNAscan-SE (v2.0) [23] and Barrnap were used to predict the tRNA and rRNA of genomes, respectively. The CDS of predicted genes were annotated across the Kyoto Encyclopedia of Genes and Genomes (KEGG) database to obtain functional information [24].

Average Nucleotide Identity Calculation
The average nucleotide identity (ANI) value between any two genomes was calculated using ANI perl script. The resulting matrices were clustered and visualized using R-packages' heat map software (v1.0.8).

Pangenome and Core Genome Analysis
The pangenome and core genome of the 37 L. crispatus genomes were calculated using the PGAP1.21 [25], and the pangenome model was analyzed according to Heap's law.

Phylogenetic Analysis
The protein sequences of the 37 L. crispatus were analyzed by OrthoMCL1.4 [26]. The protein families with the same function were then clustered using the Markov cluster algorithm (MCL). Core genes and specific genes were shown in a Venn plot. The core genes were aligned using MAFFT-7.313 [27], clustering analysis was performed using the neighbor-joining (NJ) method [28], and phylogenetic trees were constructed using PHYLIP (v3.6).

Carbohydrate Active Enzyme Analysis
The strain's carbohydrate metabolism genes were annotated using the Carbohydrate Active Enzyme Database (CAZy) [29].

Statistical Analysis
Difference between two groups was evaluated using an independent samples t-test (SPSS 19.0), and a p value of < 0.05 was considered to indicate a significant difference.

General Genome Features of the L. crispatus Strains
The 37 L. crispatus genomes ranged in size from 1.87 Mb to 2.26 Mb, with an average length of 2.09 Mb (Figure 1). Genomic sizes of vaginal-source bacteria (range 1.87-2.22 Mb,~2.08 Mb) were similar to those of strains isolated from chicken (range 1.97-2.22 Mb,~2.03 Mb). The genomes of the human-fecal-origin L. crispatus (range 2.03-2.26 Mb,~2.16 Mb) were the largest. The number of CDS varied from 1869 to 2325, with an average of 2110 (Table 1). The GC content of the genomes was on average 36.76%, and no significant difference was found among the three sources ( Figure 1). The ANI value of the 37 strains compared with the reference strain L. crispatus ST1 from NCBI and between any two strains was 96.27% to 99.96% (Figure 2), i.e., greater than 96%, indicating that they belonged to the same species [30]. As shown in Figure 2, the genomes of vaginal L. crispatus strains clustered separately from the feces isolates based on hierarchal clustering of ANI values. However, the human fecal-origin L. crispatus did not form a separate cluster from the chicken isolates. The feces-derived strains had a closer genetic relationship.  Table 1). The GC content of the genomes was on average 36.76%, and no significant difference was found among the three sources ( Figure 1).The ANI value of the 37 strains compared with the reference strain L. crispatus ST1 from NCBI and between any two strains was 96.27% to 99.96% (Figure 2), i.e., greater than 96%, indicating that they belonged to the same species [30]. As shown in Figure 2, the genomes of vaginal L. crispatus strains clustered separately from the feces isolates based on hierarchal clustering of ANI values. However, the human fecal-origin L. crispatus did not form a separate cluster from the chicken isolates. The feces-derived strains had a closer genetic relationship.

Pan-genome and Core Genome Analysis of L. crispatus
The pan-genome of 37 strains contained 5014 genes. As the number of strains increased, the pangenome curve gradually converged. The derived mathematical function had an index value of less than 0.5, indicating that L. crispatus has a closed pan-genome [16]. However, the core genome only consisted of 1250 genes. The number of core genes was saturated and the gene curve trend was stable after the first 30 strains were analyzed ( Figure 3a). The results showed that the 37 strains of L. crispatus in this study fully represented this species. According to the result of MCL clustering, the core genes, accessory genes, and specific genes of 37 strains were obtained. An accessory gene is defined as a gene shared by at least two strains. A specific gene is a gene that only exists in a certain strain. The number of specific genes in L. crispatus varied from 4 to 134 (Figure 3b). The average number of specific genes of L. crispatus isolated from chicken and human feces was 61 and 57 respectively. However, L. crispatus strains isolated from vaginal samples had an average number of specific genes of only 21.
Accessory genes contribute to species diversity. They are not important for bacterial growth, but confer selective advantages such as adaptation to different niches, drug resistance, or colonization of new hosts [31]. The distribution of the 2497 accessory genes was different in the 37 strains, as visualized by heat map (Figure 3c). It was shown that the 37 L. crispatus strains were divided into two large branches; one branch was vaginal-derived strains and the other was strains from feces. This indicated that the accessory genes were affected by the habitat. Meanwhile, strains isolated from chicken or human feces were similar in the composition of their accessory genes.

Pan-genome and Core Genome Analysis of L. crispatus
The pan-genome of 37 strains contained 5014 genes. As the number of strains increased, the pan-genome curve gradually converged. The derived mathematical function had an index value of less than 0.5, indicating that L. crispatus has a closed pan-genome [16]. However, the core genome only consisted of 1250 genes. The number of core genes was saturated and the gene curve trend was stable after the first 30 strains were analyzed ( Figure 3a). The results showed that the 37 strains of L. crispatus in this study fully represented this species. According to the result of MCL clustering, the core genes, accessory genes, and specific genes of 37 strains were obtained. An accessory gene is defined as a gene shared by at least two strains. A specific gene is a gene that only exists in a certain strain. The number of specific genes in L. crispatus varied from 4 to 134 (Figure 3b). The average number of specific genes of L. crispatus isolated from chicken and human feces was 61 and 57 respectively. However, L. crispatus strains isolated from vaginal samples had an average number of specific genes of only 21.
confer selective advantages such as adaptation to different niches, drug resistance, or colonization of new hosts [31]. The distribution of the 2497 accessory genes was different in the 37 strains, as visualized by heat map (Figure 3c). It was shown that the 37 L. crispatus strains were divided into two large branches; one branch was vaginal-derived strains and the other was strains from feces. This indicated that the accessory genes were affected by the habitat. Meanwhile, strains isolated from chicken or human feces were similar in the composition of their accessory genes.  Accessory genes contribute to species diversity. They are not important for bacterial growth, but confer selective advantages such as adaptation to different niches, drug resistance, or colonization of new hosts [31]. The distribution of the 2497 accessory genes was different in the 37 strains, as visualized by heat map (Figure 3c). It was shown that the 37 L. crispatus strains were divided into two large branches; one branch was vaginal-derived strains and the other was strains from feces. This indicated that the accessory genes were affected by the habitat. Meanwhile, strains isolated from chicken or human feces were similar in the composition of their accessory genes.

Phylogenetic Analysis of L. crispatus Strains
To explore the effects of different niches on the phylogeny of the L. crispatus genome, 1091 single-copy orthologous genes were used to construct a phylogenetic tree (Figure 4) against the 37 strains via the neighbor-joining method. The feces-derived strains and the vagina-derived strains were each clustered into different clusters. The phylogenetic tree was divided into four branches, including Branches A, B, C, and D. All 17 vagina-isolated strains were clustered into Branch D. However, chicken and human isolates were distributed in either Branch A or Branch B. Branch C contained only six L. crispatus strains, all from human feces. Therefore, different habitats might relate to differences in the genetic evolution of the strains.

Phylogenetic Analysis of L. crispatus Strains
To explore the effects of different niches on the phylogeny of the L. crispatus genome, 1091 singlecopy orthologous genes were used to construct a phylogenetic tree (Figure 4) against the 37 strains via the neighbor-joining method. The feces-derived strains and the vagina-derived strains were each clustered into different clusters. The phylogenetic tree was divided into four branches, including Branches A, B, C, and D. All 17 vagina-isolated strains were clustered into Branch D. However, chicken and human isolates were distributed in either Branch A or Branch B. Branch C contained only six L. crispatus strains, all from human feces. Therefore, different habitats might relate to differences in the genetic evolution of the strains. Based on 1091 single-copy core genes of 37 strains, the phylogenetic tree was constructed using the neighbor-joining method.

Evolution and Adaptation to Environment
To investigate whether L. crispatus showed some adaptability to different habitats in terms of gene function, we performed KEGG annotation for all 37 strains. However, based on the results of ANI value, phylogenetic tree, and the distribution of the accessory genes, we found that the fecesderived (human/chicken) strains had a high degree of similarity. Therefore, in terms of subsequent habitat adaptation analysis, we mainly focused on the adaptation of fecal-derived strains and vaginaderived strains to their respective niches. Across the two niches, genes associated with carbohydrate metabolism (~13%) and membrane transport (~11%) occupied a large proportion. The genes encoding transport and catabolism, cell motility, global and overview maps, xenobiotics biodegradation and metabolism, lipid metabolism, metabolism of cofactors and vitamins, metabolism of other amino acids, amino acid metabolism, glycan biosynthesis and metabolism, metabolism of terpenoids and polyketides, and drug resistance were annotated. Significant differences (p < 0.05) were observed between the two niches in genes involved with antimicrobial and environmental adaptation ( Figure  5).
To further explore genes that might be adaptive to the vaginal and intestinal environment, we analyzed the genes with significant differences (p < 0.05) in gene abundance of each KEGG orthology between the two sources. Meanwhile, if a gene was present in more than half of the fecal strains, but in fewer than half in the vaginal strains, this gene was considered adapted to the gut. The genes Based on 1091 single-copy core genes of 37 strains, the phylogenetic tree was constructed using the neighbor-joining method. According to the clustering results of the phylogenetic tree, the phylogenetic tree is divided into 4 branches, namely Branch A, B, C and D.

Evolution and Adaptation to Environment
To investigate whether L. crispatus showed some adaptability to different habitats in terms of gene function, we performed KEGG annotation for all 37 strains. However, based on the results of ANI value, phylogenetic tree, and the distribution of the accessory genes, we found that the feces-derived (human/chicken) strains had a high degree of similarity. Therefore, in terms of subsequent habitat adaptation analysis, we mainly focused on the adaptation of fecal-derived strains and vagina-derived strains to their respective niches. Across the two niches, genes associated with carbohydrate metabolism (~13%) and membrane transport (~11%) occupied a large proportion. The genes encoding transport and catabolism, cell motility, global and overview maps, xenobiotics biodegradation and metabolism, lipid metabolism, metabolism of cofactors and vitamins, metabolism of other amino acids, amino acid metabolism, glycan biosynthesis and metabolism, metabolism of terpenoids and polyketides, and drug resistance were annotated. Significant differences (p < 0.05) were observed between the two niches in genes involved with antimicrobial and environmental adaptation ( Figure 5).
19% to carbohydrate metabolism (deoxyribose-phosphate aldolase, sucrose phosphorylase, αgalactosidase). Other genes were linked to biosynthesis of other secondary metabolites and to amino acid metabolism. In the 38 genes of vagina-derived L. crispatus, 31% were related to carbohydrate metabolism (galactitol-specific IIC component, pullulanase, hexulose-6-phosphate isomerase). In addition, there were many genes involved in redox reactions and acid tolerance in the vagina, such as iron-sulfur protein and manganese ion transporters (Tables S1, S2).

Active Carbohydrate Enzymes
The carbohydrate-utilization-related genes occupied the largest proportion in each genome by KEGG annotation. Therefore, we further compared the number of genes encoding carbohydrateactive enzymes in strains from feces and vagina.
The 37 strains had genes encoding 33 glycoside hydrolases (GHs) families. The GH family was the most important type of carbohydrate enzyme in L. crispatus, accounting for 55.9%. The abundance of GH1, GH13_18, GH2, GH20, GH25, GH3, GH36, GH43_14, GH73, GH78, and GH92 differed significantly (p < 0.05) between strains isolated from the feces and vagina. L. crispatus strains of fecal origin contained more types of carbohydrate enzyme than did L. crispatus of vaginal origin. Twelve GH families were only present in strains derived from feces. The GH140 and GH43_4 only existed in human feces isolates, whereas GH105 were only present in strains isolated from chicken feces ( Figure  6a).
Additionally, nine glycosyl transferases (GTs), three auxiliary activities (AAs), six carbohydratebinding modules (CBMs), six carbohydrate esterases (CEs), and two polysaccharide lyases (PL) were detected in 37 strains. The abundances of GT14, GT4, GT8, CE10, and CBM families were significantly different between the two groups. The CBM4 and PL15_1 were present only in vaginal strains ( Figure  6b, c). To further explore genes that might be adaptive to the vaginal and intestinal environment, we analyzed the genes with significant differences (p < 0.05) in gene abundance of each KEGG orthology between the two sources. Meanwhile, if a gene was present in more than half of the fecal strains, but in fewer than half in the vaginal strains, this gene was considered adapted to the gut. The genes associated with vaginal adaptation were also analyzed according to the same method. Based on this criterion, 69 important genes were found. Among the feces-derived L. crispatus isolates, 31 genes were analyzed, including 11 fecal-specific genes, (i.e., only present in feces). Of 38 genes found in the vagina-derived strains, 7 were vagina-specific genes.
Of the 31 genes in the feces-derived strains, 22% were related to the CRISPR-Cas system, and 19% to carbohydrate metabolism (deoxyribose-phosphate aldolase, sucrose phosphorylase, α-galactosidase). Other genes were linked to biosynthesis of other secondary metabolites and to amino acid metabolism. In the 38 genes of vagina-derived L. crispatus, 31% were related to carbohydrate metabolism (galactitol-specific IIC component, pullulanase, hexulose-6-phosphate isomerase). In addition, there were many genes involved in redox reactions and acid tolerance in the vagina, such as iron-sulfur protein and manganese ion transporters (Tables S1 and S2).

Active Carbohydrate Enzymes
The carbohydrate-utilization-related genes occupied the largest proportion in each genome by KEGG annotation. Therefore, we further compared the number of genes encoding carbohydrate-active enzymes in strains from feces and vagina.
The 37 strains had genes encoding 33 glycoside hydrolases (GHs) families. The GH family was the most important type of carbohydrate enzyme in L. crispatus, accounting for 55.9%. The abundance of GH1, GH13_18, GH2, GH20, GH25, GH3, GH36, GH43_14, GH73, GH78, and GH92 differed significantly (p < 0.05) between strains isolated from the feces and vagina. L. crispatus strains of fecal origin contained more types of carbohydrate enzyme than did L. crispatus of vaginal origin. Twelve GH families were only present in strains derived from feces. The GH140 and GH43_4 only existed in human feces isolates, whereas GH105 were only present in strains isolated from chicken feces (Figure 6a).

Discussion
Comparative genomic analysis of 37 L. crispatus strains from fecal and vaginal sources revealed the host's influence on the genetic characteristics and evolution of the strain, as well as the adaptability of the strain to the intestine and vagina.
The general genomic characteristics in terms of size, GC content, and number of CDS (Table 1) showed that there was no significant difference between the L. crispatus strains derived from the three sources. However, when compared with chicken feces and human vaginal isolates, the genome size of strains isolated from human feces was larger. A possible explanation is that the human gut placed more environmental pressure on the bacteria [32]. To survive in a complicated environment, the L. crispatus could have obtained additional genes through horizontal transfer [33].
The phylogenetic tree results showed that the fecal isolates and the vaginal isolates were each clustered separately (Figure 4). This distinction indicated that the niche exerted an important impact on the evolution of the L. crispatus [15]. This result provided the insight that we might be able to determine the niche of an unknown source L. crispatus based on the clustering results of strains in the phylogenetic tree. A previous study revealed that Bifidobacteria isolated from the vagina and feces had no tendency to host in the phylogenetic tree [34]. In this case, it might have been that Bifidobacteria were not the main strain in the vagina. Therefore, the effect of the source on the evolutionary relationship of Bifidobacteria was not obvious.
Compared with the gut, the vagina has a lower pH (4 ± 0.5) [35]. Low pH helps reduce the risk of gynecological diseases and maintain female reproductive health [36]. Therefore, L. crispatus from the vagina require more defense mechanisms to respond to the low pH. The gene encoding manganese transport protein was presented only in vagina-sourced strains. This protein can take up Mn 2+ and expel protons out of the cell to take part in the acid response, maintaining intracellular pH homeostasis [35]. Meanwhile, Mn 2+ can also help in protecting bacteria against oxidative stress [37].
Lactobacilli in the vagina produce high levels of hydrogen peroxide [38,39]. Hydrogen peroxide is an important antibacterial substance. It can inhibit such pathogens as Gardnerella vaginalis and Additionally, nine glycosyl transferases (GTs), three auxiliary activities (AAs), six carbohydrate-binding modules (CBMs), six carbohydrate esterases (CEs), and two polysaccharide lyases (PL) were detected in 37 strains. The abundances of GT14, GT4, GT8, CE10, and CBM families were significantly different between the two groups. The CBM4 and PL15_1 were present only in vaginal strains (Figure 6b,c).

Discussion
Comparative genomic analysis of 37 L. crispatus strains from fecal and vaginal sources revealed the host's influence on the genetic characteristics and evolution of the strain, as well as the adaptability of the strain to the intestine and vagina.
The general genomic characteristics in terms of size, GC content, and number of CDS (Table 1) showed that there was no significant difference between the L. crispatus strains derived from the three sources. However, when compared with chicken feces and human vaginal isolates, the genome size of strains isolated from human feces was larger. A possible explanation is that the human gut placed more environmental pressure on the bacteria [32]. To survive in a complicated environment, the L. crispatus could have obtained additional genes through horizontal transfer [33].
The phylogenetic tree results showed that the fecal isolates and the vaginal isolates were each clustered separately (Figure 4). This distinction indicated that the niche exerted an important impact on the evolution of the L. crispatus [15]. This result provided the insight that we might be able to determine the niche of an unknown source L. crispatus based on the clustering results of strains in the phylogenetic tree. A previous study revealed that Bifidobacteria isolated from the vagina and feces had no tendency to host in the phylogenetic tree [34]. In this case, it might have been that Bifidobacteria were not the main strain in the vagina. Therefore, the effect of the source on the evolutionary relationship of Bifidobacteria was not obvious.
Compared with the gut, the vagina has a lower pH (4 ± 0.5) [35]. Low pH helps reduce the risk of gynecological diseases and maintain female reproductive health [36]. Therefore, L. crispatus from the vagina require more defense mechanisms to respond to the low pH. The gene encoding manganese transport protein was presented only in vagina-sourced strains. This protein can take up Mn 2+ and expel protons out of the cell to take part in the acid response, maintaining intracellular pH homeostasis [35]. Meanwhile, Mn 2+ can also help in protecting bacteria against oxidative stress [37].
Lactobacilli in the vagina produce high levels of hydrogen peroxide [38,39]. Hydrogen peroxide is an important antibacterial substance. It can inhibit such pathogens as Gardnerella vaginalis and Neisseria gonorrhoeae in the vagina, and help maintain the normal flora of the vagina [40]. Meanwhile, hydrogen peroxide also causes oxidative stress in L. crispatus. Therefore, we observed that genes associated with oxidative stress were significantly enriched in vaginally derived bacteria, as described in Table S1. The two-component system is a broad signal transduction pathway in bacteria. It plays a major role in adapting to changing environmental conditions such as cell envelope stress response, phosphate regulation, and oxidative stress [37,41]. The OmpR family significantly increases the resistance of bacteria to hydrogen peroxide [42]. The genes encoding Fe-S cluster assembly protein were more abundant in the vagina (Table S1), and these genes are critical in the catalysis of electron transfer or metabolic support reactions [43]. Meanwhile, the Suf protein complex helps to assemble or repair oxygen-labile Fe-S clusters under oxidative stress [44,45]. In addition, hydrogen peroxide could further react with some Fe 2+ iron to result in high-activity oxidants through the Fenton reaction [46]. This highly active oxidant might have a stronger inhibitory effect on pathogens in the vagina and a more positive effect on reproductive health, but this requires further testing to verify.
Pullulanase is an enzyme that breaks down glycogen [14]. In the vagina, glycogen is the main source of carbohydrates for microorganisms [47]. Genes encoding pullulanase were more abundant in the vagina-derived L. crispatus strains than in strains of fecal origin (Table S1). L. crispatus uses this enzyme to break down glycogen and produce lactic acid, which could maintain the low pH and inhibit pathogenic bacteria in the vagina [48]. The genes coding clumping factor A and A GntR family transcription factor were only present in the vagina-derived L. crispatus strains (Table S1).
Due to the variety and high density of bacteria, the intestine is an ideal niche for phage survival [49]. The L. crispatus isolated from feces were more susceptible to phage infection. The CRISPR/Cas system is known as the prokaryotic immune system and resists the invasion of foreign genetic material such as phage viruses and foreign plasmids [50]. To protect against damage to cells by foreign DNA, more genes related to the CRISPR-Cas system were observed in feces-derived L. crispatus (Table S2).
Tetracycline and lincomycin resistance genes were present only in the fecal strains (Table S2). Antibiotics have been used extensively to treat diseases caused by bacterial infections in humans and animals. Correspondingly, bacteria have acquired certain resistance genes through horizontal transfer to improve antibiotic resistance and their own survival ability [51]. Gene transfer at the level of drug-resistance genes often occurs in the human gastrointestinal tract [52]. In addition, genes related to streptomycin biosynthesis, such as rfbA, rfbB, and rfbC, were significantly enriched in the feces-derived bacteria (Table S2). Meanwhile, some genes associated with galactose metabolism, cysteine and methionine metabolism, and arginine biosynthesis were found only in the fecal strains.
To increase our understanding of the ability of L. crispatus to use host-derived glycogen fermentation, the genomes of 37 L. crispatus were analyzed and compared using CAZymes. The feces-derived strain was enriched in various GH family enzymes, including GH140, GH20, GH3, GH43_14, GH78, and GH92, which are involved in the utilization of a variety of carbon sources (Figure 6a). This might be because the gut contains more types of carbohydrate than the vagina. L. crispatus would need more enzymes to break down these carbohydrates to provide them with energy. The abundance of GH2 (β-galactosidase) and GH 36 (α-galactosidase) isolates in human feces was significantly higher than human vagina and chicken feces. Milk and other dairy products are an important part of human diet. GH2 and GH36 can hydrolyze galactose in dairy products, which is beneficial to the growth of strains in the intestine [53]. However, the abundance of CBM12, CBM37, CBM4, CBM66, GH25, and GH73 family in the vagina was higher than in the gut. CBM is a non-catalytic member of the cellulolytic enzymes. It is thought to assist in the synergy of enzymes through proximity and targeting, and plays an important role in the degradation of insoluble substrates [54,55]. Cellulose is the main component of plant cell walls. Humans and chickens consume cellulose-containing vegetables, but the feces isolates had a lower CBM abundance than the vagina (Figure 6b). This might require more in-depth exploration of the specific role of CBM in vagina-derived L. crispatus. GH25 has a role in facilitating cell division and in defense, as a lysozyme involved in cell wall and peptidoglycan catabolism [56]. GH73 encodes β-N-acetylglucosaminidases that cleave the β-1,4 glycosidic linkage between the N-acetylglucosamine (GlcNAc) and N-acetylmuramic acid (MurNAc) residues of bacterial cell wall peptidoglycan (PG) [57]. These two GH family lysozymes might provide a bacterial strategy to improve the competitive advantage of L. crispatus by destroying the cell walls of other pathogenic bacteria in the vagina and lysing them [58]. In summary, the above results demonstrate that L. crispatus is adaptable to different environments to improve its ability to survive.

Conclusions
In this study, we selected the genomes of 37 L. crispatus strains isolated from the human vagina, human feces, and chicken feces. The results of comparative genomics demonstrated that niche affected the evolution of L. crispatus to a large extent. Moreover, the L. crispatus strains showed adaptability to different environments to improve their own survival ability during the evolution of the strain.