Sexual transmission of urogenital bacteria: whole metagenome sequencing evidence from a sexual network study

ABSTRACT Sexual transmission of the urogenital microbiota may contribute to adverse sexual and reproductive health outcomes. The extent of sexual transmission of the urogenital microbiota is unclear as prior studies largely investigated specific pathogens. We used epidemiologic data and whole metagenome sequencing to characterize urogenital microbiota strain concordance between participants of a sexual network study. Individuals who screened positive for genital Chlamydia trachomatis were enrolled and referred their sexual contacts from the prior 60–180 days. Snowball recruitment of sexual contacts continued for up to four waves. Vaginal swabs and penile urethral swabs were collected for whole metagenome sequencing. We evaluated bacterial strain concordance using inStrain and network analysis. We defined concordance as ≥99.99% average nucleotide identity over ≥50% shared coverage; we defined putative sexual transmission as concordance between sexual contacts with <5 single-nucleotide polymorphisms per megabase. Of 138 participants, 74 (54%) were female; 120 (87%) had genital chlamydia; and 43 (31%) were recruited contacts. We identified 115 strain-concordance events among 54 participants representing 25 bacterial species. Seven events (6%) were between sexual contacts including putative heterosexual transmission of Fannyhessea vaginae, Gardnerella leopoldii, Prevotella amnii, Sneathia sanguinegens, and Sneathia vaginalis (one strain each), and putative sexual transmission of Lactobacillus iners between female contacts. Most concordance events (108, 94%) were between non-contacts, including eight female participants connected through 18 Lactobacillus crispatus and 3 Lactobacillus jensenii concordant strains, and 14 female and 2 male participants densely interconnected through 52 Gardnerella swidsinskii concordance events. IMPORTANCE Epidemiologic evidence consistently indicates bacterial vaginosis (BV) is sexually associated and may be sexually transmitted, though sexual transmission remains subject to debate. This study is not capable of demonstrating BV sexual transmission; however, we do provide strain-level metagenomic evidence that strongly supports heterosexual transmission of BV-associated species. These findings strengthen the evidence base that supports ongoing investigations of concurrent male partner treatment for reducing BV recurrence. Our data suggest that measuring the impact of male partner treatment on F. vaginae, G. leopoldii, P. amnii, S. sanguinegens, and S. vaginalis may provide insight into why a regimen does or does not perform well. We also observed a high degree of strain concordance between non-sexual-contact female participants. We posit that this may reflect limited dispersal capacity of vaginal bacteria coupled with individuals’ comembership in regional transmission networks where transmission may occur between parent and child at birth, cohabiting individuals, or sexual contacts.

. To address these limitations, we used genome-resolved metagenom ics to identify concordant strains in the vaginal and penile microbiomes of individuals enrolled in the Sexually Transmitted Infection Network Groups (STING) study.

RESULTS
Participants were recruited from seven clinics providing genital Chlamydia trachomatis (CT) and Neisseria gonorrhoeae (NG) screening in the Baltimore, MD, USA, metropolitan area: four clinics serving the general population and three pop-up clinics serving large universities.Female and male individuals were eligible if they screened positive for genital CT by nucleic acid amplification test (NAAT), were 16-40 years old and were willing to provide names and contact information of all sexual contacts from the prior 60 or 180 days (switched from 60 to 180 partway through recruitment).These participants will be referred to as wave 1. Sexual contacts of wave 1 participants were then invited to enroll.Contacts were eligible if they were 16-40 years old, willing to undergo CT/NG screening, and willing to provide names and contact information of sexual contacts from the prior 60 or 180 days.These participants will be referred to as wave 2. Recruiting sexual contacts via this snowball approach continued for up to four waves (Fig. 1).
All participants underwent a reproductive tract examination, during which vaginal or penile urethral swabs were collected.Swabs were used for whole metagenome sequencing, and metagenomes were successfully sequenced for 138 of 139 participants (median 48,744,779 read pairs; range: 132,491-179,383,783).Of these 138 participants, half were female, two-thirds reported Black race, and the median age was 21 (Table 1).Most participants (87%) had CT; two had NG.Ninety-three participants were enrolled in wave 1, 63 of which (46% of total) had no sexual contacts enrolled.The remaining 33 wave 1 participants (24%) had 40 sexual contacts enrolled (wave 2, 29%).Wave 2 participants had two sexual contacts enrolled (wave 3), and wave 3 participants had one sexual contact enrolled (wave 4).Genital NG (NAAT) (Continued on next page)

Strain-concordance summary
We used inStrain to identify concordant strains between participants (68).We used an all-vs-all strategy to build inStrain profiles for each metagenome-MAG pair by mapping sequence reads against MAGs using Bowtie2 (69).To identify strain concordance, we applied stringent filters requiring ≥99.99% sequence similarity [population average nucleotide identity (popANI)] over ≥50% shared coverage.
We identified 115 concordant strains among 54 participants representing 25 taxa (Fig. 4).For 17 taxa (68%), we observed one concordance event (Tables 2 and 3).The remaining concordant taxa were Gardnerella leopoldii (n = 8 events), Gardnerella swidsinskii (n = 52), L. crispatus (n = 18), L. iners (n = 6), L. jensenii (n = 3), Lactobacillus mulieris (formerly L. jensenii [70]) (n = 3), Prevotella spp.(n = 4), and Prevotella amnii (n = 4).Several taxa were concordant exclusively between sexual contacts (Fannyhessea vaginae, Gardnerella vaginalis A, Sneathia sanguinegens, Sneathia vaginalis ([formerly a Continuous data are presented as median and interquartile range. b Only female participants were asked about current contraceptive use, so these data are only presented for female participants.c Wave and partnership counts overall and for male participants sum to 1 greater than the total because one male participant was included in waves 2 and 3. Percentages for overall and male participants are out of the total number of unique participants (i.e., 138 overall and 64 males).d Enrollment site was recorded only for wave 1 participants, not for sexual contacts.e CSTs were assigned using VALENCIA; CST, community state type.f PID, vaginal and cervical discharge, mucopurulent cervicitis, and vulvovaginal candidiasis were evaluated only among female participants.Penile discharge, urethritis, and circumcision status were evaluated only among male participants.g PID, pelvic inflammatory disease.

Strain concordance between sexual contacts
Seven concordance events (6%) were between sexual contacts in four dyads.Strain concordance between contacts does not necessarily reflect sexual transmission within the dyad, so we applied additional criteria to determine which events likely repre sent recent sexual transmission.We considered strains that were concordant between contacts with <5 popSNP/Mbp to be sexually transmitted.Of seven concordance events between contacts, six met this criterion.Heterosexual contacts (STING106 and STING224) enrolled 7 days apart had concord ant F. vaginae (4.5 popSNP/Mbp), G. vaginalis A (30 popSNP/Mbp), P. amnii (1.3 popSNP/ Mbp), and S. sanguinegens (0 popSNP/Mbp; Fig. 4, right middle).We consider these F. vaginae, P. amnii, and S. sanguinegens strains to be heterosexually transmitted.Both contacts were assigned to CST IV-B (Fig. 6).The female contact had CT and reported current contraceptive implant, current bloody vaginal discharge, and no BV history.The male contact was circumcised, screened negative for CT, and reported no symptoms (clinician noted scant urethral discharge).Both contacts reported being regular partners.Their most recent sex was 7-14 days previously, and they had sex three to five times in the prior 60 days with no condom use.
Heterosexual contacts (STING154 and STING190) enrolled on the same day had concordant G. leopoldii (1.0 popSNP/Mbp; Fig. 4, bottom middle), which we consider heterosexually transmitted.Both contacts were assigned to CST IV-B and reported no symptoms.The female contact had CT and reported no BV history.The male contact was circumcised and screened negative for CT.Both contacts reported being occasional partners; their last sex was 1 month previously, and they had sex two to three times in the prior 60 days with no condom use.
Heterosexual contacts (STING168 and STING201) enrolled on the same day had concordant S. vaginalis (0 popSNP/Mbp; Fig. 4, bottom left), which we consider heterosexually transmitted.Both contacts were assigned to CST IV-B and had CT.The female contact reported current abnormal vaginal discharge (clinician reported no BV findings), current copper intrauterine device, and no BV history.The male contact was circumcised and reported no symptoms.These contacts reported discrepant characteris tics of their sexual relationship.The female contact reported being regular partners; their last sex was 4 days previously, and they had sex four times in the prior 60 days with intermittent condom use.The male contact reported being new partners; their last sex was 0 days previously (within the prior 24 hours), and they had sex once with no condom use.
Female sexual contacts (STING117 and STING118) enrolled two days apart had concordant L. iners (0 popSNP/Mbp; Fig. 4, left middle), which we consider sexually transmitted.This was the only Lactobacillus strain concordant between contacts (no Lactobacillus concordance between heterosexual contacts).One contact each was assigned to CST III and CST IV-B.Both contacts had CT and reported no symptoms and no BV history; one contact reported current oral contraceptive use.They reported being

Each edge (line between nodes) represents a strain-concordance event between connected nodes. Thin edges indicate concordance between non-contacts.
Thick edges indicate concordance between sexual contacts.Putative sexual transmission is indicated by colored asterisks adjacent to and colored the same as edges representing sexual transmission.Concordant strains were required to have ≥99.99%popANI across ≥50% of shared coverage.Sexual transmission was defined as strain concordance between sexual contacts with <5 popSNP/Mbp.popANI is estimated by inStrain based on popSNPs, which are called when two samples share no alleles at a given site.F, female; M, male.regular partners; their last sex was 0 days prior to the first contact enrolling (within the prior 24 hours), and they had sex >20 times in the prior 60 days.

Strain concordance between non-contacts
Most concordance events (108, 94%) were between non-contacts, including 91 (84% of non-contact concordance) between female participants, 16 (15%) between female and male participants, and 1 (1%) between male participants.Eight female participants were interconnected through 18 L. crispatus and 3 L. jensenii concordant strains (Fig. 4 top right).Seventeen of these concordance events were between participants assigned to CST I; four were between participants assigned to CST I and CST V (Fig. S2).Median popSNP/Mbp between non-contacts was 35 for L. crispatus and 2.9 for L. jensenii; each L. jensenii concordance event had <5 popSNP/Mbp.Participants concordant for L. crispatus were enrolled a median of 5 weeks apart; those concordant for L. jensenii were enrolled a median of 27 weeks apart.
Fourteen female and two male participants were densely interconnected through 52 G. swidsinskii concordance events (Fig. 4 left middle).Twenty-eight of these events were between participants assigned to CST IV-B, 22 between participants assigned to CST IV-A and CST IV-B, and 2 between participants assigned to CST IV-A (Fig. S3).All G. swidsinskii concordant strains showed >5 popSNP/Mbp (median = 59).Participants concordant for G. swidsinskii were enrolled a median of 77 weeks apart.

Comparing concordance between contacts and non-contacts
Because nearly half of participants did not have a contact enrolled and few concordance events were between contacts, we examined the prevalence of concordance/transmis sion among contacts and non-contacts.For non-contacts, we defined the prevalence denominator as the total number of unique participant pairs minus the number of contact dyads, which represents all possible opportunities for non-contact concordance.The aim here is to describe the relative frequency at which we observed concordance between contacts vs between non-contacts, not to generate precise population-level prevalence estimates.
Considering putative heterosexual transmission, there were 74 female participants and 64 male, giving 4,736 unique female-male pairs.Thirty-nine were contact dyads, leaving 4,697 unique non-contact female-male pairs.The prevalence of concordance/    transmission was 7.7% between female-male contact dyads (3 of 39) and 0.5% between non-contact female-male pairs (22 of 4,697).Considering putative transmission between women, there were 2,701 unique female-female pairs.Two were contact dyads, leaving 2,699 unique non-contact female-female pairs.The prevalence of concordance/transmission was 50% between female-female contact dyads (1 of 2) and 3.2% between non-contact female-female pairs (87 of 2,699).

DISCUSSION
We present strain-level metagenomic evidence that strongly supports heterosexual transmission of BV-associated bacteria and sexual transmission of L. iners between female contacts.These findings are consistent with prior molecular biology studies of strain-level urogenital bacterial concordance (50)(51)(52)(53)(54)(55)(56) and metataxonomic stud ies of species-level urogenital bacterial concordance (26,28,49,57,58) between vaginal and penile microbiomes of recent sexual contacts, providing strong support for sexual transmission of several BV-associated species and L. iners.
Whether BV can be sexually transmitted is debated (35)(36)(37)(38).This study cannot demonstrate heterosexual BV transmission; however, our observations do provide support for possible BV transmission.We identified putative heterosexual transmission of F. vaginae, G. leopoldii, P. amnii, S. sanguinegens, and S. vaginalis.Prior epidemiologic studies reported associations between these taxa and specific combinations of Amsel criteria; these associations were largely consistent between populations of Kenyan and American women (73,74).In vitro and omics studies indicate these taxa may contribute to BV signs and symptoms via production of biogenic amines (75)(76)(77) or extracellular enzymes that degrade cervicovaginal mucus (78)(79)(80).Here, one of three female contacts with putative transmission of BV-associated taxa reported abnormal vaginal discharge (none reported odor).There were no clinical BV findings for these female contacts.
A primary argument against BV sexual transmission is that early male partner treatment trials showed little to no benefit for reducing BV recurrence (39)(40)(41)(42)(43)(44)(45)(46).These trials are subject to major limitations including insufficient randomization methods, lack of power calculations, suboptimal BV treatment regimens for women, and lack of adherence data (39)(40)(41)(42)(43)(44)(45)(46); that these trials did not demonstrate a reduction in BV recurrence among dyads randomized to male partner treatment should not be interpreted as evidence against heterosexual BV transmission.Recent years have seen a resurgence of interest in BV male partner treatment with multiple studies and trials completed and several under way (28,(47)(48)(49).Identifying putative hetero sexual transmission of BV-associated taxa strengthens the evidence base supporting these ongoing investigations.While we cannot comment on potential transmission of additional BV-associated bacteria, measuring the impact of male partner treatment on F. vaginae, G. leopoldii, P. amnii, S. sanguinegens, and S. vaginalis may provide insight into a regimen's performance.Two recent pilot studies evaluated a combined regimen of oral metronidazole and topical clindamycin for male partners, which showed reductions in penile Prevotella spp.(28,49).No other taxa we observed to be heterosexually transmit ted were significantly reduced.
Another major area of ongoing research aims to rigorously design and test Lactoba cillus live biotherapeutic products (LBPs) to achieve/maintain optimal vaginal micro biota composition and vaginal health.Our observation of putative L. iners transmission between female contacts suggests Lactobacillus LBP may be particularly beneficial for WSW as the strain(s) may be transmitted within dyads and WSW sexual networks.Given high BV concordance between female sex partners (31,32), transmission of beneficial lactobacilli through WSW sexual networks could yield substantial gains in individual and community health.That said, Lactobacillus LBP research largely focuses on L. crispatus (81,82), which shows substantial genomic, metabolic, and epidemiologic differences to L. iners (8-10, 83, 84).Without identifying putative transmission of other lactobacilli, it is unclear how generalizable the observation of L. iners transmission is to possible transmission of Lactobacillus LBP strains.
While our data do not provide evidence of L. crispatus sexual transmission, they also cannot rule it out.Importantly, only 2 of 43 contact dyads included two women, limiting our insight into sexual transmission of vaginal bacteria between women.The minimum relative abundance of any putatively transmitted strain was 3% (S. vaginalis).L. crispatus relative abundances among female non-contacts with L. crispatus concordance ranged from 29% to 99%, with five of eight ≥89%.Conversely, L. crispatus relative abundances ranged from 0.02% to 0.10% among contacts in WSW dyads, and the maximum L. crispatus relative abundance among male participants in contact dyads was 17%, with most <1% (Fig. S4).These values suggest that if any contact dyad had sexually transmit ted L. crispatus, our analyses likely are not sensitive enough to detect this concordance.
Finally, we observed frequent strain concordance between non-contacts, primarily G. swidsinskii and L. crispatus between female participants.Because concordance/transmis sion was 9-16 times more prevalent among contact dyads compared to non-contact pairs, strain concordance between non-contacts should not be taken as evidence against putative sexual transmission.Frequent strain concordance between female non-contacts is consistent with a prior analysis of vaginal bacterial strain concordance within and between families in the Baltimore area.While some reproductive-age mothers and their reproductive-age daughters had concordant strains, participants were more likely to have a concordant strain with someone they were not related to than someone in their family (85).Substantial strain concordance between female non-contacts is also consistent with a recent person-level meta-analysis of 7,646 stool and 2,069 saliva metagenomes from 20 countries that showed non-negligible strain concordance in these microbiomes between non-cohabitating individuals in the same community (86).
We hypothesize that a high degree of strain concordance between unrelated and/or non-sexual-contact women may reflect limited dispersal capacity of vaginal bacteria coupled with individuals' in regional transmission networks (85).Here, we are considering regions the size of a metropolitan area because STING recruited participants from the Baltimore metropolitan area.However, our hypotheses may apply to any area that encompasses a sexual and/or urogenital bacteria transmission network.Because vaginal lactobacilli and Gardnerella spp.are rarely detected in the environment or human body sites other than the genitals and many vaginal bacteria are anaerobic, we posit that direct person-to-person transmission (e.g., between parent and child at birth, cohabitating individuals, or sexual contacts) may represent a primary means of geographic dispersal of vaginal bacteria (85).Under this assumption, we would expect sexual network characteristics to influence vaginal bacterial dispersal.In sexual networks, small groups of high-contact individuals, termed core groups, play an integral role in maintaining bacterial STI endemicity/outbreaks (87).This can largely be attributed to assortative mixing and partner concurrency: core group members disproportionately partner with other core group members and are more likely to have concurrent partners than non-core-group members (87), which sustains STI transmission within core groups.When core group members partner with non-core-group members, this disseminates the STI in the general population.Similar processes may drive maintenance and dispersal of network-specific consortia of urogenital bacterial strains.Network characteristics appear to influence pathogen phylogenies (88); evaluating whether this is true for all or specific members of the vaginal microbiome could vastly improve our understanding of vaginal microbial ecology and biogeography as well as BV etiology.The sexual networks enrolled in STING were small and isolated (Fig. S5), so we were not able to test these hypotheses.
This study should be interpreted in the context of its limitations.While these metagenomic data provide strong evidence supporting sexual transmission of urogenital bacteria, the study design is insufficient to prove sexual transmission or identify the direction of putative transmission.We cannot determine whether puta tively transmitted strains engrafted in "recipient" contacts' urogenital microbiomes.It is possible strains failed to engraft, but sufficient DNA was present at sampling to detect the transient concordant strain.
Finally, STING participants were enrolled in the Baltimore area, had high CT preva lence, and were relatively young.It is unclear how generalizable these findings are to other geographic regions, less-densely populated settings, older populations, or populations with lower STI burden.
By integrating genome-resolved metagenomics in a sexual network study, we provide insight into the sexual transmissibility of the urogenital microbiota and vaginal bacterial biogeography.Our findings may have substantial implications for achieving and maintaining vaginal health among active individuals, particularly via novel interventions including BV partner treatment and LBP.We believe these data represent only "the tip of the iceberg" when it comes to urogenital bacterial transmission: we likely under-detected transmission in heterosexual and WSW dyads, and we cannot com ment on how representative our findings are.Additional studies in various geographic contexts and sexual networks with various characteristics are warranted to broaden our understanding of sexual transmission of the urogenital microbiota, vaginal bacterial biogeography, and whether these vary across populations.

MATERIALS AND METHODS
We used data and samples from the Sexually Transmitted Infection Network Groups study to evaluate urogenital bacterial strain concordance between sexual contacts and non-contacts.We report our results according to Strengthening the Organization and Reporting of Microbiome Studies guidelines (Table S2) (89).

STING recruitment and procedures
Between February 2016 and March 2020, participants were recruited from seven clinics providing CT/NG screening in the Baltimore, MD, USA, metropolitan area.Individuals were eligible if they screened positive for genital CT by NAAT, were 16-40 years old, and were willing to provide the names and contact information of all sexual contacts from the prior 60 or 180 days.Individuals were ineligible if they were living with HIV, ever exchanged sex for money or drugs, were taking immunosuppressive medica tions, had a severe medical condition, or were non-English-speaking (all self-report).Female individuals were ineligible if they were pregnant (point-of-care human chorionic gonadotropin urine test).Male individuals were ineligible if they exclusively had sex with men.
Wave 1 participants provided names and contact information of sexual contacts.Trained staff notified contacts of their STI exposure and invited them to enroll.Sexual contacts were eligible if they were 16-40 years old, willing to undergo CT/NG screening, and willing to provide names and contact information of their sexual contacts from the prior 60 or 180 days.Pregnant female contacts and male contacts who exclusively had sex with men were eligible for waves 2-4; otherwise, exclusions were the same as for wave 1. Snowball recruitment of sexual contacts continued for up to four waves.All participants and sexual contacts who screened positive for CT/NG were treated free of charge, after sample collection, based on national guidelines (90).We constructed a sexual network diagram (Fig. S5).Sexual networks were small and isolated from each other, so we did not examine sexual network characteristics.
Demographic information and medical history were collected via in-person interview, and the participants completed social and behavioral history questionnaires (online or paper).Vaginal or penile urethral swabs were collected during reproductive tract examination, immediately stored at −18°C, and transferred daily on ice to the Institute for Genome Sciences at the University of Maryland School of Medicine (UMSOM), where they were stored at −80°C.

Whole metagenome sequencing
Shotgun metagenomic data were generated for 138 samples (74 vaginal and 64 penile) as described previously (84).Swabs were resuspended in 1 mL of Amies transport medium.Genomic DNA was extracted using 200 µL of resuspended material and the MagAttract PowerMicrobiome DNA/RNA Kit (Qiagen) with bead-beating using Tissue Lyser II.Sequence libraries were generated from extracted DNA using the Illumina Nextera XT Flex kit.Procedures were performed according to manufacturer instructions and automated on the Hamilton STAR robotic platform.Libraries were sequenced on an Illumina NovaSeq 6000 (150 bp, paired-end) at Maryland Genomics at UMSOM.Host reads were identified and removed using BMtagger and the GRCh38 human genome as reference (91).Non-host reads were processed using fastp (v.0.21) (92).Metagenome taxonomic compositions were determined via mapping to VIRGO (59), and CSTs were assigned using VALENCIA (2).VALENCIA is a supervised classifier for the vaginal microbiome.The penile CST assignments we observed are consistent with prior unsupervised clustering results (25,(27)(28)(29), so we feel it is reasonable to use VALENCIA with penile samples in STING.

Identification of strain concordance
We used inStrain to identify concordant strains between participants (68).We built inStrain profiles for each metagenome-MAG pair by mapping each metagenome's reads against all retained MAGs using Bowtie2 (69).We summarized profiles using the inStrain compare function with Ward linkage, which reports coverage overlap and sequence similarity between each metagenome and each MAG.To measure sequence similarity, inStrain estimates popANI based on popSNPs, which are called when metagenomes share no alleles (major or minor) at a given site.To identify concordant strains, we applied stringent filters requiring ≥99.99% sequence similarity (popANI) over ≥50% shared coverage.We considered strains concordant between sexual contacts with <5 popSNP/Mbp to be sexually transmitted.Illumina NovaSeq 6000 error rates are ~0.1% (93); inStrain accounts for an error rate of 0.1%, so sequencing errors should not produce artifactual concordance/transmission events.
Several participant pairs appeared to be concordant for multiple strains of the same species.We assumed these participants were concordant for a single strain that mapped sufficiently well to multiple MAGs to meet concordance criteria when compared to multiple MAGs.For redundant events, we retained the event with the greatest number of base pairs compared and discarded remaining redundant events (215 events in 60 participant pairs discarded).
To identify and discard strain-concordance events likely resulting from contamination, we built inStrain profiles for the negative control sample.We used the same criteria as above to identify concordance between metagenomes and the negative control.We considered concordance to be due to contamination when two participants and the negative control were all concordant for the same strain, and we discarded four G. leopoldii concordance events likely resulting from contamination.

Statistical analysis
We performed two differential abundance analyses using the ZicoSeq and ZicoSeq.plotfunctions of the GUniFrac package (v.1.7)in R (v.4.2.1, throughout) (67).We com pared vaginal and penile metagenome relative abundances, and we compared penile metagenome relative abundances by CT status.ZicoSeq uses a permutation-based approach for multiple comparisons adjustment, and we controlled FWER at 1% using 999 permutations.
We used Kruskal-Wallis tests to evaluate whether popSNP/Mbp between concordant strains differed across species or when strains were concordant between contacts vs non-contacts, whether vaginal and penile non-host reads differed, and whether per-MAG comparable sites differed between vaginal and penile metagenomes.We considered hypothesis tests significant at P < 0.05 (two-tailed, no multiple comparisons adjustment).

FIG 3
FIG 3 Differential abundance of taxa between vaginal and penile metagenomes.R 2 is the proportion of variance in metagenome composition (across vaginal and penile samples) explained by a given taxon.The vertical dashed line at R 2 = 0 indicates no difference in abundance between vaginal and penile metagenomes, negative R 2 values indicate enrichment in vaginal metagenomes, and positive R 2 values indicate enrichment in penile metagenomes.We controlled the FWER at 1%, as indicated by the horizontal dashed line at log 10 (FWER − adjusted P) = 2.All taxa that are differentially abundant at this FWER fall above the horizontal line and are labeled.Taxa that are presented in other figures are also labeled, regardless of significance.

FIG 4
FIG4 Network diagram of strain-concordance events between sexual contacts and non-contacts.Each node (circle or diamond) represents one participant.

a
Relative abundance data are not presented for Prevotella spp. or Prevotella sp000758925 because there was no clear comparator taxon.Genus-level relative abundance data are presented for Fannyhessea spp., all Gardnerella spp., Megasphaera lornae, Mobiluncus curtisii, Staphylococcus epidermidis, and all Streptococcus spp.For Lactobacillus mulieris, relative abundance data from Lactobacillus jensenii are presented due to reclassification.For Limosilactobacillus portuensis, relative abundance data from Lactobacillus vaginalis are presented due to reclassification.

TABLE 1
Participant characteristics overall and stratified by gender

TABLE 1
Participant characteristics overall and stratified by gender (Continued)

TABLE 2
Interpersonal characteristics of strain-concordance events between sexual contacts

Dyad members Taxon N concordance events within a : Same CST b Weeks between enrollment Relative abundances in dyad Heterosexual dyad Woman-woman dyad Minimum Maximum
Strain concordance defined as ≥99.99% popANI over ≥50% of the length of the genomes.inStrain defines popANI based on popSNPs, which are called when two samples share no alleles at a given site.For G. vaginalis A and G. leopoldii concordance events, relative abundances are presented for Gardnerella spp. at the genus level.
a b CSTs assigned using VALENCIA.c

TABLE 3
Interpersonal characteristics of strain-concordance events between non-contacts

Taxon a N concord ance events b Concordant between: Relative abundances in non-contact pairs with concordance Relative abundance ratio between non-contacts with concordance Female participants Female and male participants Male participants Same CST c Weeks between enrollment N Row % N Row % N Row % N Row % Med d IQR d Minimum Maximum Med d Range
Fannyhessea spp.