Genetic and functional enrichments associated with Enterococcus faecalis isolated from the urinary tract

ABSTRACT Enterococcus faecalis is the leading Gram-positive bacterial species implicated in urinary tract infection (UTI). An opportunistic pathogen, E. faecalis is a commensal of the human gastrointestinal tract (GIT), and its presence in the GIT is a predisposing factor for UTI. How E. faecalis colonizes and survives in the urinary tract (UT) is poorly understood, especially in uncomplicated or recurrent UTI. The UT is distinct from the GIT and is characterized by a sparse nutrient landscape and unique environmental stressors. In this study, we isolated and sequenced a collection of 37 clinical E. faecalis strains from the urine of primarily postmenopausal women. We generated 33 closed genome assemblies and 4 highly contiguous draft assemblies and conducted a comparative genomics analysis study to identify genetic features enriched in urinary E. faecalis with respect to E. faecalis isolated from the human GIT and blood. Phylogenetic analysis revealed high diversity among urinary strains and a closer relatedness between urine and gut isolates than blood isolates. Plasmid replicon (rep) typing further underscored possible UT-GIT interconnection, identifying nine shared rep types between urine and gut E. faecalis. Both genotypic and phenotypic analyses of antimicrobial resistance among urinary E. faecalis revealed infrequent resistance to the front-line UTI antibiotics nitrofurantoin and fluoroquinolones and no vancomycin resistance. Finally, comparing gene presence and absence among urinary and gut strains, we identified 19 candidate genes enriched among urinary strains. These genes are involved in the core processes of sugar transport, cobalamin import, glucose metabolism, and post-transcriptional regulation of gene expression. IMPORTANCE Urinary tract infection (UTI) is a global health issue that imposes a substantial burden on healthcare systems. Women are disproportionately affected by UTI, with >60% of women experiencing at least one UTI in their lifetime. UTIs can recur, particularly in postmenopausal women, leading to diminished quality of life and potentially life-threatening complications. Understanding how pathogens colonize and survive in the urinary tract is necessary to identify new therapeutic targets that are urgently needed due to rising rates of antimicrobial resistance. How Enterococcus faecalis, a bacterium commonly associated with UTI, adapts to the urinary tract remains understudied. Here, we generated a collection of high-quality closed genome assemblies of clinical urinary E. faecalis isolated from the urine of postmenopausal women that we used alongside detailed clinical metadata to perform a robust comparative genomic investigation of genetic factors that may be involved in E. faecalis survival in the urinary tract.

E nterococcus faecalis, a Gram-positive bacterium, is an increasingly frequent cause of urinary tract infection (UTI), especially in complicated or recurrent cases (1)(2)(3)(4)(5).While a strong body of work has elucidated virulence mechanisms associated with complicated enterococcal UTI, namely, catheter-associated UTI, little is known about how enterococci cause uncomplicated and recurrent UTI (6)(7)(8)(9)(10)(11). E. faecalis natively inhabits the human gastrointestinal tract (GIT) but is an opportunistic pathogen (12,13).E. faecalis GIT colonization is proposed to be a predisposing factor to UTI since periurethral contamina tion by gut bacteria is an important route of infection (14)(15)(16).This idea is supported by a reported association between gut enterococci abundance and Enterococcus UTI (14).In addition, premenopausal recurrent UTI (rUTI) patients were found to have a greater frequency of infection with endogenous gut microbes, specifically Escherichia coli and E. faecalis (17)(18)(19).Despite the association of E. faecalis with rUTI, little is known about the mechanisms by which E. faecalis colonizes the urinary tract in the absence of a catheter and the genetic determinants that enable its persistence.
Enterococci demonstrate great adaptability to thrive in stressful environments (13,20).Urine, as compared to the gut environment, is a nutrient-limited medium char acterized by high osmolarity, limited nitrogen and carbohydrate availability, moder ate oxygenation, and low pH (21)(22)(23).Urine is also antimicrobial, composed of high concentrations of urea and antimicrobial proteins (23).Survival in the urinary tract despite environmental pressures and antibiotic intervention suggests that E. faecalis urine isolates may be specialized genetically or phenotypically.However, understanding the genetic factors necessary for E. faecalis growth in the urinary environment is limited.
To date, studies of E. faecalis urinary fitness are limited and focus on OG1RF (oral isolate) or V583 (blood isolate) (24).Previous studies of E. faecalis urinary fitness were conducted using well-studied E. faecalis strains OG1RF, V583, and MMH594 grown in pooled urine (25)(26)(27)(28).These identified differential expression of genes encoding a sucrose phosphotransferase system (PTS), the lutABC operon for L-lactate metabolism, an amino acid ABC transporter, and efaCBA for manganese scavenging during growth in urine (28).However, no comprehensive study of a collection of clinical urinary E. faecalis isolates has been published.Although various virulence factors have been proposed as pathogenicity markers, a virulence genotype to distinguish uropathogenic E. faecalis from other E. faecalis strains has yet to be proposed (12).
To gain a deeper understanding of the genetics of urinary E. faecalis strains, we generated a collection of high-quality genomic sequences of clinical urinary E. faeca lis isolates and compared them to isolates from the human gut and blood-distinct anatomical sites that pose unique evolutionary pressures on E. faecalis.Our findings show that urinary strains are diverse, possessing a wide range of plasmid replicon types and exhibiting low antimicrobial resistance rates, as determined both by geno typic predictions and phenotypic assays.We find that vancomycin resistance, which is commonly studied in this genus, was absent from the urine group, and intermedi ate fluoroquinolone resistance was common but did not correlate with any known chromosomal mutations or resistance genes.Finally, we identified genes involved in carbohydrate transport and metabolism, as well as vitamin B12 import and post-tran scriptional regulation of gene expression, as enriched among urinary isolates.Together, this work provides a resource of high-quality and well-curated urinary E. faecalis genomes and identifies candidate pathways that may be important for E. faecalis urinary tract colonization.

Collection of urinary E. faecalis strains isolated from the urine of postmeno pausal women
The role of E. faecalis in uncomplicated UTIs or in the microbiome of asymptomatic women is not well understood.Nevertheless, E. faecalis is commonly isolated from the female urogenital tract (29)(30)(31).Indeed, secondary analysis of our recently published metagenomic study of the urinary microbiome of postmenopausal women revealed that E. faecalis was present in 57.3% of samples (32).In patients with active rUTI at the time of urine collection, E. faecalis was detected in 46.2% of samples.In women without UTI but with a history of rUTI, E. faecalis was detected in 62.5% of urinary microbiomes (32).These observations highlight the need to understand how E. faecalis adapts to this unique environment.
To begin to understand genetic features associated with urinary E. faecalis strains, we sequenced 37 unique E. faecalis strains isolated from the urine of postmenopausal women using both Illumina and Nanopore sequencing for the generation of closed or highly contiguous high-quality genome assemblies (Table S1).E. faecalis strains were isolated from the urine of consenting postmenopausal women who were recruited from a tertiary care center in Dallas, TX, USA, between October 2017 and April 2019 (Fig. 1A; Table S2).The 37 women were stratified into four cohorts based on their history of rUTI, UTI symptoms, and urine culture: never UTI (no clinical history of symptomatic UTI, n = 4), sporadic (no history of rUTI, current symptomatic UTI, n = 3), remission  (33,34).c Isolates from Van Tyne et al. 's study (24).Genome assembly criteria used for the selection of comparator genomes.*A couple of isolates reported in Table S3 do not meet one criterion but were included as they were reported in previous literature and met all other criteria.(C) Counts of complete and incomplete genome assemblies in the urine (pink), gut (blue), and blood (gray) isolation groups.
Among the 37 urinary genomes, 33 were closed, while 4 were highly contiguous with <21 contigs (Table S1).As of March 2023, a total of 131 complete genome assemblies of E. faecalis isolated from the human host were publicly available on the National Center for Biotechnology Information (NCBI).Of the 131 complete genomes, only 9 were clearly identified as associated with human urine.This collection, therefore, presents a focused group of high-quality urinary E. faecalis genomes that, coupled with clinical metadata, provides a valuable resource for the field of E. faecalis and E. faecalis UTI.

Complete urinary genomes allow comparative analysis of urinary E. faecalis to gut and blood isolates
We hypothesized that specific genetic factors and phenotypes would be associated with urinary tract-colonizing E. faecalis strains when compared to E. faecalis strains colonizing different host niches.To test this hypothesis, we curated comparator genomes of E. faecalis strains isolated from the urinary tract (n = 50), gastrointestinal tract (n = 49), and blood (n = 46), which we term "isolation groups" (Fig. 1B; Table S3).Each isolation group is composed of publicly available genome assemblies obtained from NCBI that meet the inclusion criteria described in Materials and Methods (Fig. 1C).The majority of urinary strains (74%) were sequenced as part of this study (Fig. 1B); 31% of gut isolates were obtained from Dallas, TX, fecal surveillance (33,34); and 41% of blood isolates were obtained from a Wisconsin hospital outbreak in the 1980s (24).Proportions of complete assemblies available in the gut (31%) and blood (9%) isolation groups are substantially lower than those in the urine group (70%) due to the genome availability at the time of curation (Fig. 1C).

Phylogenetic and pangenome analyses reveal high diversity among urinary E. faecalis strains
Comparative analysis revealed that urinary strains are diverse (Fig. 2A).Urinary strains did not form unique clusters in a core gene phylogenetic tree.A large cluster of closely related ST6 blood isolates and a large mixed gut and urine cluster of ST179 and ST16 isolates were observed (Fig. 2A).Small urine-gut clusters as well as blood-gut clusters appear throughout the tree, suggesting closer relatedness between the urinary and gut strains than urinary and blood.The phylogeny largely follows the evolutionary estimates provided by multilocus sequence typing (MLST) (Fig. 2A).Urinary isolates are characterized by 28 different sequence types (STs) (Fig. 2B), with ST179 being the most common (14%).Among gut isolates, 27 different STs are present, with ST16 being the most common (20%).Interestingly, ST179 and ST16 are single locus variants of each other, differing by a synonymous variation in xpt.Blood isolates represent 17 STs and have a strong lineage bias for ST6, with 22 of the 46 isolates belonging to an ST6 hospital outbreak lineage (24).Despite this bias, this collection is representative of the genomic data available for bloodstream E. faecalis.We observed that urine isolates had the most unique STs (18) not found among isolates of the other groups, followed by gut ( 14) and blood isolates (9), and that urine and blood isolates do not share any STs that are absent from the gut group (Fig. 2B).
Pangenome analysis indicated that E. faecalis has an open pangenome consisting of a large proportion of accessory genes (soft core, shell, and cloud) and a small proportion of core genes (25.5%).In total, 8,297 genes make up the pangenome (Fig. 2C).Isolation source-specific pangenome analysis revealed a similar trend emphasizing the diversity among urinary isolates (Fig. S1).Analysis of genome size distribution showed that genomes from blood isolates are on average significantly larger than those from urine and gut (Fig. 2D).Conversely, no significant difference in average genome size was detected between urine and gut isolates.Furthermore, genome size was not significantly different between isolates of Dallas and non-Dallas origins (Fig. S1).These data indicate that urinary isolates are evolutionarily diverse, encompassing many STs and a wide range of genome sizes, and support the hypothesis that urine isolates originate from gut reservoirs (14,15,19,35).

Plasmid replicon typing suggests similarities in plasmid carriage between urinary and gut isolates
Plasmids are drivers of evolution and virulence in bacteria (36).We hypothesized that urinary isolates possess characteristic conserved plasmid replicon (rep) types due to within-niche specialization.PlasmidFinder identified 18 unique rep types among all E. faecalis strains, 8 of which were present in all groups (repUS43, rep9b, rep9a, rep9c, repUS11, rep2, rep1, rep7a) (Fig. 3A; Table S4).A circular extrachromosomal element within EfsPF36 was not typeable by PlasmidFinder or NCBI PGAP.Prophage analysis determined that the extrachromosomal element, named here EfsPF36_phage01, is an intact prophage, phiFL2A (Table S4).The total number of replicons per isolate did not significantly differ between Dallas and non-Dallas isolates (Fig. S2).
Rep6 was the only rep present in urine and gut isolates, but not in blood.Rep7b and repUS12 were only present in urine and blood isolates but not in the gut.Of the rep types that were unique to a single isolation group, rep17 was only identified among gut isolates; rep11b and repUS42 were identified only in urine isolates; and repUS56, rep21, rep13, and rep15 were only identified in blood isolates (Fig. 3A).We found that the blood group possessed the total highest number of replicons (n = 134), followed by urine (n = 92) and gut (n = 82) (Fig. 3A).Additionally, the blood group had the highest median number of replicons per isolate (Fig. 3B).
The most common rep type was repUS43, found in 64% of urine, 53% of gut, and 30% of blood isolates (Fig. 3A).Analysis of the complete, closed genomes generated in this study revealed that repUS43 is chromosomally integrated and is proximal to a tetracy cline resistance gene (tetM) (Table S4).The following two most prevalent replicons were rep9a and rep9b, which are associated with pheromone-responsive plasmids.rep9a is associated with pAD1 lineage plasmids known to encode the virulence factor cytolysin (37)(38)(39).Additionally, rep9b is associated with pMG2200, reported to encode vancomycin resistance (vanHBX) and a Bac41-type bacteriocin (40).We found that the identified rep9a plasmids were diverse in size, ranging from 149 kb to 42 kb, and only 47% encoded the cytolysin operon (Fig. 3C and D).Alignments of complete rep9a plasmids and the DS16 pAD1 reference, which was assembled as part of this study, further suggest that their genomic content varies widely (Fig. 3E; Fig. S3).Rep9b plasmids also have a broad size range from 106 to 44 kb; however, 40.9% of complete rep9b plasmids are approximately 68 kb (Fig. 3C).Alignments of complete rep9b plasmids and the pMG2200 reference indicate that vanHBX is only encoded in pMG2200, while 72.7% of rep9b plasmids encode the bac41 operon (Fig. 3D and E).

Pseudolysogenic extrachromosomal linear phages are enriched among geographically proximal isolates
Complete genome assemblies of the urinary collection revealed a prevalent pseudolyso genic extrachromosomal linear phage (41).The phage, EF62phi, originally identified in E. faecalis 62 (SAMN02603509) isolated in Norway, encodes common structural phage components as well as a toxin-antitoxin system (Fig. 4A).Because EF62phi was identified in 13 (26%) of the urinary isolates but only in 18.4% of gut and 8.7% of blood isolates, we first hypothesized that EF62phi may be enriched in urine isolates (Fig. 4B; Table S4).However, EF62phi was commonly found in the gut strains also obtained from Dallas, TX.Thus, we hypothesized that EF62phi was instead associated with an isolate of geographi cal origin.Analysis utilizing available geographical metadata revealed that a greater percentage (36.5%) of E. faecalis isolated from the Dallas Metroplex in Texas, USA, harbored EF62phi than non-Dallas isolates (8.97%) (Fig. 4C).

Antimicrobial resistance is not widespread among urinary E. faecalis isolates
Antimicrobial resistance gene (ARG) analysis was used to predict E. faecalis resistance and assess differences between the isolation groups.We hypothesized that urinary isolates would be commonly resistant to UTI front-line therapies.Strikingly, the median number of ARGs, including known point mutations, per isolate was lowest in the urine group (Fig. 5A).To determine if this enrichment may be due to geographic location, we analyzed ARG counts between Dallas and non-Dallas isolates and found that the total number of ARGs did not significantly differ between Dallas and non-Dallas isolates (Fig. S2).Resistance was predicted for 11 drug classes and attributed to 31 unique ARGs or mutations in E. faecalis from all three isolation groups (Fig. 5B; Table S5).Several Numbers correspond to the presence of an ARG or mutation.Ubiquitous efflux pumps and intrinsic resistance genes are not depicted.
ubiquitous genes confer intrinsic resistance to antimicrobials in the species.These genes include lsaABE encoding multidrug resistance ABC-F subfamily efflux pumps, dfrE encoding a dihydrofolate reductase responsible for trimethoprim resistance, and genes encoding efflux pump components, efrAB and emeA (Table S5) (42)(43)(44)(45)(46).The most prevalent tetracycline resistance gene was tetM, which was found in 67% of gut, 56% of urine, and 46% of blood isolates.The presence of the tetM gene is directly correlated with the chromosomally integrated repUS43 locus (Spearman r = 0.7271, P < 0.0001) (Fig. 5B).For aminoglycosides, fluoroquinolones, and vancomycin, urine isolates were predicted to be least frequently resistant, whereas blood isolates were predicted to be most frequently resistant.Aminoglycoside resistance was more common in blood isolates (72%) than in gut (35%) and urine (20%) (Fig. 5B).Fluoroquinolone resistance, as predicted by the presence of either an ARG or known gyrA and/or parC mutations, was similarly prevalent in gut (20%) and blood (24%) isolates but less frequent in urine isolates (4%) (Fig. 5B).Finally, vancomycin resistance was absent in the urine group but found in similar frequency in gut (20%) and blood (24%) isolates (Fig. 5B).However, the selection criteria for sequenced E. faecalis blood strains-mostly from hospital outbreaks and surveillance (24, 33)-may, in part, bias the number of vancomycin-resistant blood strains.
Although gene predictions offer insight, phenotypic assessments are imperative to confirm resistance.Phenotypic antibiotic resistance analysis of a large collection of urinary isolates has not been previously conducted; therefore, we analyzed the resistance phenotypes of 38 strains to 8 clinically relevant antibiotics from 7 drug classes (Fig. 5C).OG1RF and V583 were included as representatives of well-studied E. faecalis model strains.Resistant phenotypes among urinary isolates largely converged with gene predictions where specific resistance genes were present, whereas ubiquitous efflux pumps were not associated with resistance in all strains (Fig. 5C; Table S5).
Thirty resistant phenotypes were observed, of which 29 (96.6%) were attributed to an encoded ARG or known chromosomal mutations (Fig. 5C).Resistance to erythro mycin was most widespread, with 42% of isolates demonstrating resistance, and was well-predicted by the ermB gene-15/16 resistant isolates encoded ermB.Gentamicin resistance was the second most prevalent with eight resistant isolates, and the presence of the aac(6′)-aph(2″) was associated with the most resistant phenotypes.Gentamicin resistance in EfsC85 was not explained by gene prediction, and the mechanism of resistance remains unclear (Fig. 5C).
The third most prevalent resistance was to doxycycline in the tetracycline class.The presence of tetM was not uniquely associated with resistance.In isolates encoding only tetM, isolate was resistant, 14 were intermediate, and 11 were susceptible.The co-occur rence of tetM and tetL was a more accurate predictor of doxycycline resistance (Fig. 5C).Chloramphenicol resistance was only predicted in one isolate and was associated with the cat gene.Finally, only one isolate was resistant to ciprofloxacin and levofloxacin due to the presence of known mutations in the quinolone resistance-determining regions (QRDRs) of the gyrA and parC genes (47) (Fig. 5C).
A total of 61 intermediate phenotypes were observed, of which 15 (39.4%) were attributed to the presence of a predicted ARG (Fig. 5C).The remaining 46 were detec ted for erythromycin (20), fluoroquinolones (ciprofloxacin, 14; levofloxacin, 9), and less commonly for chloramphenicol (2) and nitrofurantoin (1).Because of their clinical relevance, we further analyzed intermediate resistance to fluoroquinolones.We validated ciprofloxacin intermediate resistance by minimum inhibitory concentration (MIC) assay (Table S6) and confirmed that reported intermediate strains had ciprofloxacin MICs ≥2 and <4 μg/mL.Interestingly, no known point mutations or fluoroquinolone resistance ARGs were associated with the validated intermediate phenotypes (Fig. 5C).

Genes involved in sugar transport, metabolism, and post-transcriptional stress responses are enriched in urinary E. faecalis
We next sought to identify gene enrichments within urinary versus gut strains of E. faecalis.Following inclusion cutoffs of >70% frequency in urine isolates and ³20% higher prevalence in urine than in the gut group, gene enrichment analysis identified 19 candidate genes as enriched in urinary isolates (Fig. 6A; Table S7).Among the 19 genes, 6 encode a PTS system operon predicted to be responsible for the transport of mannose and fructose (manR, fryA, manP_1, manP_2, fruA, alsE).Eight genes are located within a syntenic region of prophage phage 04 of V583 (locus EF1988-EF2043) (48,49).These eight candidates include four hypothetical proteins (group_887, group_1388, group_2882, group_3220); two phage components, the ArpU family transcriptional regulator and a recombinase (xerC); and two intriguing, annotated candidates: csp encoding a cold-shock protein and hemH encoding a ferrochelatase (Fig. 6A and B).Csp is a unique member of the cold shock protein (CSP) family.There are seven csp orthologs in the E. faecalis pangenome, four of which are present in all E. faecalis strains.Among the csp genes that are not ubiquitous, the enriched csp candidate (EF1991) was found to be a unique cold shock family member present at higher frequency (86%) within urinary isolates (Fig. 6B; Figure S4A and B).The location of these genes within a prophage region suggests that they may have been acquired by horizontal gene transfer through a phage integration event.To determine if the enrichment of these eight prophage-associated genes was specific and not just a result of enrichment of the entire prophage, we aligned the phage region and performed blastn queries of candidate genes (Fig. 6C).These data, along with the presence/absence of data, suggest that the candidate genes are enriched independently of the intact phage because the phage structural genes, for example, are found at much lower frequencies among urine isolates than the eight enriched genes (Fig. 6C; Table S7).
The remaining five enrichment candidates include dgaF, natA, and btuD and two hypothetical proteins (group_1511, group_2248), one of which is directly down stream of natA (Fig. 6A and B).DgaF is an aldehyde lyase used in the Entner-Doudoroff (ED) pathway (50,51).NatA is a sodium ABC transporter ATP-binding protein, while BtuD is a vitamin B12 import ATP-binding protein.In the E. faecalis pangenome, there are up to 19 distinct btuD orthologs, some of which are present in all isolates.A single E. faecalis may possess between 6 and 10 of the btuD orthologs, ranging in length from 768 to 2,340 bp.The enriched btuD candidate encodes a 297aa protein.Its length as well as sequence composition suggest that it is a unique family member (Fig. S4C).To test if the enrichment analysis may be confounded by isolation geography, we assessed differences in gene enrichment of representatives of the urinary strain-enriched genes between Dallas and non-Dallas isolates (Fig. S5).All candidates, with the exception of btuD and natA, showed no significant difference in enrichment between Dallas and non-Dallas isolates.Interest ingly, both btuD and natA were significantly enriched (P = 0.0012 and P = 0.0084, respectively) among the Dallas isolates, suggesting that the observed urinary enrichment of these genes may indeed be confounded by isolation geography.
All hypothetical candidates were analyzed using the NCBI conserved domains database, and none were found to have conserved domains.However, blastp analysis found that candidate group_2248 was 100% identical to a T7SS effector LXG polymor phic toxin (accession numbers CP091889 and WP_238463980.1),and group_1511 downstream of natA was 99.1% identical to an annotated ABC transporter, permease protein (accession number HF558530).This, coupled with the respective lengths of natA and group_1511, suggests that the latter is natB (52).Candidate group_2882 of the phage04 region was 100% identical to an anti-repressor (accession number CP014949) (Table S7).plasmid sequences was conducted using Easyfig v2.2.2 with tblastx at default parameters (72).Reference plasmids representing rep9a (pAD1 from strain DS16, SAMN00809239) and rep9b (pMG2200, AB374546.1)were included in the analysis for comparison.Further alignment and visualization were conducted using BLAST Ring Image Generator v0.95 with blastn at default thresholds.

Pangenome and phylogenetic analyses
Pangenome analysis was conducted using Panaroo v1.2.10 with the merge-paralog option selected and core gene alignment conducted using mafft (73,74).Reference pan-genome sequences are available in Data S1.Core gene alignment output was then used to construct a phylogenetic tree with IQ-TREE using the GTR + F + I + I + R6 model with ultrarapid bootstrapping (1,000 inferences) (75)(76)(77).A model of nucleotide substitution was selected with the IQ-TREE model (77).The phylogenetic tree was rooted using min-VAR rooting with FastRoot v1.5 and visualized with iTOL (78,79).

Gene enrichment analysis and functional annotation
Gene enrichment analysis was conducted using Scoary v1.6.16(80).Genes with a P-value <0.05 were retained for further analysis (Table S7).Candidates were considered enriched if they were present in >70% of urine isolates and were at least 20% more prevalent in urine isolates than in the gut.Since the urine group is primarily composed of isolates sequenced in this study (74%), it was imperative to apply stringent cutoffs to ensure biological significance.Candidates were validated using blastn to confirm that paralogs were not erroneously split and account for highly conserved homologs.Hypothetical candidate genes were further assessed using NCBI Conserved Domain Search, PSORTb v3.0.3, blastn, and blastp queries to elucidate potential function (68,69,81).Allele sequence alignments were conducted using nucleotide MUSCLE v3.8.425 alignment at default parameters in Geneious Prime v2022.1.1 (82,83).

Antimicrobial resistance gene prediction
Antimicrobial resistance genes and point mutations were predicted using ResFinder v4.0 on CGE with the E. faecalis database at default parameters (84,85).ABRicate v1.0.1 was used to query the CARD database at 70% coverage and 70% identity thresholds to validate ResFinder findings and identify additional resistance determinants (86,87).

Antimicrobial resistance phenotype assessment
Resistance phenotypes were assessed by Kirby-Bauer disk diffusion, as described previously, on brain-heart infusion (BHI) agar and following the established breakpoints of the Clinical and Laboratory Standards Institute (CLSI) (88)(89)(90).All ciprofloxacin phenotypes and intermediate or resistant gentamicin and chloramphenicol phenotypes were further validated using the MIC microdilution assay in BHI broth per CLSI break points using the HT-MIC workflow as previously described (91).Further details can be found in the supplemental material.

DISCUSSION
In this study, we analyzed the genomes of urinary, gut, and blood E. faecalis isolates to identify genetic features and functions associated with E. faecalis strains colonizing the urinary tract.A collection of 37 clinical urinary isolates was obtained from predominately postmenopausal women, sequenced, and hybrid-assembled to generate closed or highly contiguous genome assemblies.This collection provides a tool for further research in the fields of E. faecalis genomics and E. faecalis UTI.Our analysis of these and additional curated, publicly available E. faecalis genomes provides comprehensive insight into niche-specific population genetics, mobile genetic elements, antimicrobial resistance, and gene enrichments.
Here, we sought to test the hypothesis that urinary E. faecalis possesses gene enrichments that may be associated with colonization of the urinary environment.We postulated that urinary isolates should be genetically conserved, encode characteristic plasmids, be resistant to similar antimicrobials, and encode genes that may aid in survival within the urinary tract.Additionally, as the gut-bladder axis becomes further defined in UTI research, we estimated that urinary isolates will resemble gut isolates more than blood isolates (15,19,35).
Our findings reveal that urinary E. faecalis is diverse, and no lineage emerges as strongly associated with the urinary niche.The most frequent ST in the urinary group, ST179, is a single-locus variant of the most common ST in the gut group, ST16.Strains belonging to these two STs form a large phylogenetic cluster and may support the gastrointestinal origin of urinary isolates.Blood isolates demonstrate an expected lineage bias for ST6 (24).As genome availability is often a reflection of study selection criteria, ST6 blood isolates have been primarily associated with hospital outbreaks and represent a large collection of publicly available strains (24).We recognize that this introduces bias to the analysis and cautiously draw conclusions from urinary versus blood comparisons.
Plasmid replicon analysis did not identify strong associations between rep type and isolation source.With the exception of two rep types (rep11b, repUS41), which are both present in a single isolate, EfsC94, no rep types were found to be strictly associated with urinary isolates.Most rep types were represented in all three isolation groups, with pheromone-responsive rep9a, rep9b, and rep9c being among the most prevalent.RepUS43 was the most common rep type, and our analysis found it to be chromosomally integrated within all complete genome assemblies assessed-an insight not previously reported in the literature.Furthermore, repUS43 was positively correlated with the presence of the tetracycline resistance gene tetM, suggesting hat this resistance gene may be widespread due to a recombination event with a repUS43 plasmid in a common ancestor.Rep typing offered an insight into the plasmid content of urinary E. faecalis but not without limitations.Alignments of complete rep9a and rep9b plasmids revealed that plasmids of these rep types can vary widely in terms of size as genes encoded.Rep typing, thus, may be useful for certain conserved replicons but should be utilized with caution when considering plasmid function.
Antimicrobial resistance in silico predictions indicated that urinary E. faecalis encodes fewer ARGs than its gut or blood counterparts.In all but two drug classes, urinary isolates encoded ARGs less frequently than gut and blood isolates.Acquired ARGs or mutations predicted to confer fluoroquinolone or nitrofurantoin resistance were not prevalent in urine isolates.Phenotypic in vitro assessments of the antibiotic susceptibility of urinary isolates demonstrated a similar trend in which resistance was not widespread and intermediate phenotypes were more prevalent.Intermediate fluoroquinolone resistance was relatively common, but the mechanism remains unclear.Uncharacterized gyrA and parC mutations were present in some strains, although not within the QRDR regions.A limitation of this study is that the strains analyzed were isolated from the same clinic and are likely not fully representative of different locations, clinical practices, and demographic groups.
To address the hypothesis that urinary E. faecalis possesses gene enrichments that may be associated with urinary tract colonization, we conducted gene enrichment analysis comparing urinary and gut isolates.Applying stringent inclusion cutoffs, we identified 19 candidate genes enriched in the urinary group.We then further narrowed down candidates to those that did not demonstrate enrichment in Dallas versus non-Dallas isolates to control for a potential confounder of isolate geography.Of note, we found an enrichment of genes encoding a mannose/fructose PTS system.This may be relevant to the urinary niche because the glucose concentration in urine is often low (0.2-0.6 mM) (92), and therefore this operon may help E. faecalis utilize alternate carbon sources that are more abundant in urine, like mannose (28).Intriguingly, D-mannose supplements are routinely advised for UTI prophylaxis, as studies suggest that they may block colonization by uropathogenic E. coli, which adhere to the host urothelium via mannosylated cell surface proteins (93,94).
Additional candidates like dgaF, which functions as a 2-keto-3-deoxygluconate 6-phosphate aldolase to form glyceraldehyde-3-phosphate and pyruvate similar to the Eda enzyme in the ED pathway and is found in all urinary isolates, may be important in E. faecalis growth under the environmental pressures of the urinary niche (95).The ED pathway produces less ATP than glycolysis but offsets this metabolic cost by requiring fewer enzymes.While the ED pathway is often used by aerobic Gram-negative bacteria, E. faecalis is one of the few facultative anaerobic Gram-positive species that possess it (50,96).We postulate that E. faecalis may utilize the ED pathway during urinary colonization to conserve protein and therefore energy costs (97).
Finally, two enrichment candidates, hemH and csp, were localized within a prophage region.HemH is a ferrochelatase known to incorporate ferrous iron to form coproheme in Gram-positive bacteria (98).However, E. faecalis lacks the machinery to synthesize heme, and thus the role of this ferrochelatase remains unclear (99).Of particular interest is csp, encoding a CSP.CSPs are a common adaptive mechanism for stress and changes in environmental conditions.The first CSP, CspA, was identified in E. coli and later found to have nine homologs (CspA-CspI).These proteins, commonly expressed in response to temperature decreases, function as chaperones that prevent the formation of secondary structures in RNA transcripts, thereby allowing proper translation (100,101).However, some CSPs are known to be non-cold-inducible.Four of the E. coli CspA family members are not induced by cold shock.Such CSPs have been reported to be involved in osmotic, oxidative, starvation, pH, and ethanol stress tolerance as well as host cell invasion (100,101).An example of such adaptive CSP has been described by Michaux et al. (101), who suggested that the CspR of E. faecalis plays a role in virulence and persistence (101).The enriched csp gene highlighted by our analysis was shown previously to be upregulated in V583 grown in blood (27).However, the function of the CSP allele enriched in urinary isolates and its role in urinary fitness have yet to be characterized in E. faecalis.

FIG 1
FIG 1 Clinical cohorts and an isolated collection of Enterococcus faecalis.(A) Patient cohorts were stratified based on patient UTI history, symptoms, and urinalysis results at the time of specimen collection.n denotes the number of isolates.(B) Number of isolate genomes from each isolation group.In the urine group, 13 isolates have been obtained from National Center for Biotechnology Information, while the remaining isolates have been sequenced as part of this study.a Isolates sequenced as part of this study.b Isolates from Palacios Araya et al. 's study(33,34).c Isolates from Van Tyne et al. 's study(24).Genome assembly

FIG 3
FIG 3 Rep type analysis and in-depth comparison of rep9a and rep9b.(A) Frequency (raw counts normalized to total group size) of all reps identified.Isolate counts are listed at the bar top.(B) Distribution of the number of plasmid reps per isolate in each isolation group.Statistical significance was determined (Continued on next page)

FIG 3 FIG 4
FIG 3 (Continued) using the Kruskal-Wallis test with multiple comparisons.(C) Size in kilobases of each rep9a and rep9b plasmid within complete genome assemblies.Each dot represents a single plasmid and is color-coded by isolation group.Rep9a n = 17, Rep9b n = 22.(D) Number of plasmids possessing cytolysin, vancomycin resistance operon VanHBX, and bacteriocin Bac41 as identified by sequence alignments.Bars are colored by the plasmid rep type associated with the locus.(E)tblastx alignments of rep9a (left, n = 17) and rep9b (right, n = 22) complete plasmids.Arrows denote coding sequences and are color-coded by plasmid isolation source: urine (pink), gut (blue), blood (gray), and reference (black).The reference rep9a plasmid is DS16 pAD1.The reference rep9b plasmid is pMG2200.Shaded lines between plasmid sequences are colored based on sequence identity (%).The loci of the cytolysin operon, the VanHBX operon, and bac41 are highlighted in the reference plasmid by red blocks.

FIG 5
FIG 5 Antimicrobial resistance genes, chromosomal mutation predictions, and resistance phenotypes of urinary isolates.(A) Distribution of the total number of ARGs per strain in each isolation group.Each copy of a multi-copy ARG was counted.*Chromosomal mutations were counted as ARGs.Statistical significance was determined using the Kruskal-Wallis test with multiple comparisons.(B) Frequency of resistant isolates in each isolation source as predicted by ARG in silico analysis.Isolate counts are listed as bar ends.(C) Heatmap of resistance phenotypes assessed by disk diffusion and minimum inhibitory concentration assays.

FIG 6
FIG 6 Candidate genes enriched among urinary E. faecalis isolates.(A) Presence/absence map of 19 candidate genes.Columns represent a single isolate, and colored blocks correspond to the isolation source: urine (pink), gut (blue), and blood (gray).Genes were identified by comparing urine and gut isolates; their presence in blood isolates is provided for reference.(B) Frequency of candidate genes within each isolation source.Isolate counts are listed at bar tops.(C) BLASTRing Image Generator blastn alignment of urinary isolates (pink rings) phage04 integration region (eno to rhaS) to V583 phage04 reference (gray ring).Prophage annotations are listed, and enriched candidates are in bold.Isolate UMB0891 is not included in the alignment since the conserved region is not syntenic.