Post-coital dynamics of the penile and cervico-vaginal genital microbiome

Abstract

Background

The vaginal and penile coronal sulcus microbiome influence reproductive health outcomes and susceptibility to multiple sexually transmitted infections including HIV. There is evidence that genital bacteria are shared between heterosexual partners during sex, but the dynamics of this microbiota exchange remain poorly understood.

Results

Using microbiome characterization from established heterosexual couples, we found that condomless penile-vaginal sex dramatically altered the coronal sulcus microbiome, with transient dominance by Lactobacillus spp. Conversely, condom-protected penile-vaginal sex did not result in significant shifts in overall composition (p = 0.63). Significant changes were observed in the female partner’s genital microbiome, including increased Corynebacterium spp. and increased abundance of Bacteria Associated with Seroconversion, Inflammation, and Cells (BASICs) (Prevotella bivia, Peptostreptococcus anaerobius, Dialister micraerophilus, Prevotella disiens, Dialister propionicifaciens, Dialister succinatiphilus) in individuals with a colonized male partner. By 72 h post-coitus L. iners cell-normalized abundance remained elevated in the penile microbiome, but other taxa returned to baseline levels. Causal mediation analysis indicated a pH-mediated increase in vaginal Gardnerella at 72 h.

Conclusions

Condom-protected penile-vaginal sex was associated with minimal genital microbiome changes, whereas condomless penile-vaginal sex among established couples led to extensive exchange of genital microbiota. Most disruptions to the microbiome resolved within 2–3 days, although changes in vaginal pH were associated with longer-lasting increases in Gardnerella abundance.

Download video file ^{(51.1MB, mp4)}

Video Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s40168-026-02364-2.

Introduction

The genital microbiome is unique for its marked differences between males and females, highlighting the potential for bacteria exchange during heterosexual activity. Furthermore, the clinical associations of the genital microbiome mean that such bacterial exchange has the potential to modulate a number of important health outcomes in a sexual partner. In particular, the makeup of both the penile and vaginal microbiome influences local epithelial integrity and the immune microenvironment [1–3], which influence the partners’ susceptibility to sexually transmitted infections and other negative health outcomes [4–6].

The vaginal microbiome has been implicated in various gynecologic and obstetric conditions. A Lactobacillus-dominated vaginal microbiome is generally associated with reduced risk, with a Lactobacillus crispatus-dominated microbiome hypothesized to be the most stable and optimal [7]. In contrast, a vaginal microbiome characterized by increased diversity and a higher abundance of anaerobic species such as Gardnerella, Prevotella, Atopobium, Megasphera and Sneathia reflects a non-optimal microbial state often associated with bacterial vaginosis (BV). This condition has been associated with adverse health outcomes including premature birth and increased susceptibility to HIV and other sexually transmitted infections (STIs) [4]. The penile (specifically coronal sulcus) microbiome has also been associated with increased inflammation and HIV susceptibility. Increased densities of six penile coronal sulcus anaerobic species (Peptostreptococcus anaerobius, Prevotella bivia, Prevotella disiens, Dialister propionicifaciens, Dialister micraerophilus, Dialister succinatiphilus), also known as Bacteria Associated with HIV Seroconversion, Inflammation and immune Cells (BASIC) species, have been linked to inflammation of the foreskin and increased susceptibility to HIV infection [8]. Medical male circumcision has been shown to both significantly reduce penile anaerobic bacteria and reduce HIV susceptibility in males and BV in the female sexual partner [1, 9–11], and antimicrobials reducing the coronal sulcus density of BASIC species also reduce foreskin inflammation and HIV susceptibility [12].

Even though genital bacteria exchange between sexual partners has been reported, our knowledge of the immediate dynamics of this exchange during sexual activity remains limited. Understanding these dynamics could yield valuable insights into the factors determining the acquisition of non-optimal genital bacteria associated with adverse health outcomes for both male and female partners. Previous research has highlighted the impact of sexual behavior on the genital microbiome composition and identified similarities in the genital microbiomes of heterosexual partners [13–18]. In addition, findings from a case report and a pilot study that analyzed pre- and post-coital partner microbiome data also support the presence of genital bacterial exchange [19, 20].

To assess the immediate dynamics of genital bacteria exchange from sexual activity, we compared the genital microbiome of established, STI-free heterosexual couples prior to coitus and at 1–2 h, 7–8 h and 48–72 h after penile-vaginal sex using self-collected coronal sulcus swabs and cervicovaginal secretions.

Results

Participant characteristics and sample collection

Thirty-eight established, HIV-negative and STI-uninfected couples were included in the analysis after excluding three couples with detectable prostate-specific antigen at baseline. The average age was 23.2 years among the female partners and 24.8 years among the male partners (Table 1). Half of the male partners (n = 19/38, 50%) were uncircumcised. Six female partners (16.2%) had a non-optimal microbiome based on Nugent score (4–10), including one case of bacterial vaginosis (Nugent score 7–10). Most participants were either Asian (Male n = 16/38; Female n = 15/37) or White (Male n = 15/38; Female n = 12/37). Additional demographic and sexual behavior data can be found in Table 1.

Table 1.

Study participant demographics and characteristics

Characteristic	Female Partners	Male Partners
Total individuals	38	38
Age in years, mean (SD)	23.2 (3.9)	24.8 (6.5)
Ethnicity, n (col %)
African/Caribbean	2 (5.3)	1 (2.6)
Asian	15 (42.1)	16 (42.1)
Middle Eastern	1 (2.6)	3 (7.9)
Latin American	1 (2.6)	2 (5.3)
White	12 (31.6)	15 (39.5)
Other	6 (15.8)	1 (2.6)
Time points with sample availability, n
Baseline	36	36
1–2 h post-coital	37	37
7–8 h post-coital	35	37
48–72 h post-coital	32	33
Sex type, n (col %)
Condomless, intra-vaginal ejaculation	18 (48.7)	17 (44.7)
Condomless, no intra-vaginal ejaculation	11 (29.7)	13 (34.2)
Condom-protected	8 (21.6)	8 (21.1)
Male circumcision status, n (col %)
Circumcised	-	19 (50.0)
Nugent score at baseline, n (col %)
Normal (Score 0–3)	31 (83.8)	-
Intermediate (score 4–6)	5 (13.5)	-
Bacterial vaginosis (score 7–10)	1 (2.7)	-

Prior to the pre-coital baseline visit, the average duration of sexual abstinence was 4 days (range, 2–23 days). The self-reported time elapsed between last penile-vaginal sex and sample collection was 1.2 h (SD = 0.5) for the 1–2 h post-coital visit, 7.2 h (SD = 0.5) for the 7–8 h post-coital visit and 67.5 h (SD = 15.8) for the 48–72 h post-coital visit.

Among partners reporting condomless sex, intravaginal ejaculation was more common among couples with circumcised male partners (11/14, 79%) than uncircumcised male partners (5/15, 33%; p = 0.02).

The pre-coital genital microbiomes

The genital microbiota differed substantially between the male and female partners at the pre-coital baseline visit. Among male partners, coronal sulcus bacterial density and microbiome composition differed significantly by circumcision status. Uncircumcised males had significantly higher coronal sulcus bacterial density (median: 7.2 vs 5.7 log₁₀ 16S rRNA gene copies/swab, p < 0.001) and greater prevalence and abundance of the BASIC species, including D. micraerophilus, D. propionicifaciens, D. succinatiphilus, and P. bivia, compared with circumcised males (Table 2).

Table 2.

Prevalence and proportional abundance of selected microbiome genera and species at baseline, by circumcision status and sex

	Prevalence, percent				Proportional Abundance, mean (SD)
	A. Circumcised male N = 18	B. Uncircumcised male N = 18	C. Female cervicovaginal secretion N = 36	A v. B p-value	D. Circumcised male N = 18	E. Uncircumcised male N = 18	F. Female cervicovaginal secretion N = 36	D v. E p-value
D. micraerophilus	1 (5.6%)	11 (61.1%)	2 (5.6%)	< 0.001	0.000 (0.000)	0.007 (0.007)	0.000 (0.000)	< 0.001
D. propionicifaciens	1 (5.6%)	9 (50.0%)	0 (-)	0.003	0.000 (0.000)	0.007 (0.010)	0 (NA)	0.001
D. succinatiphilus	2 (11.1%)	8 (44.4%)	5 (13.9%)	0.026	0.001 (0.003)	0.012 (0.025)	0.000 (0.002)	0.025
H. timonensis	3 (16.7%)	15 (83.3%)	6 (16.7%)	< 0.001	0.001 (0.003)	0.135 (0.160)	0.000 (0.000)	0.007
P. anaerobius	4 (22.2%)	2 (11.1%)	3 (8.3%)	0.37	0.000 (0.001)	0.001 (0.003)	0.001 (0.002)	0.48
P. bivia	1 (5.6%)	7 (38.9%)	5 (13.9%)	0.016	0.000 (0.000)	0.017 (0.050)	0.009 (0.036)	0.018
P. disiens	1 (5.6%)	4 (22.2%)	4 (11.1%)	0.15	0.003 (0.011)	0.001 (0.003)	0.001 (0.002)	0.21
L. crispatus	1 (5.6%)	2 (11.1%)	18 (50.0%)	0.55	0.002 (0.009)	0.003 (0.011)	0.412 (0.461)	0.58
L. gasseri	1 (5.6%)	0 (-)	9 (25.0%)	1.00	0.000 (0.001)	0 (NA)	0.037 (0.145)	0.35
L. iners	6 (33.3%)	6 (33.3%)	17 (47.2%)	1.00	0.016 (0.064)	0.006 (0.021)	0.231 (0.382)	1.00
L. jensenii	4 (22.2%)	1 (5.6%)	14 (38.9%)	0.15	0.000 (0.001)	0.001 (0.005)	0.099 (0.278)	0.21
Gardnerella	6 (33.3%)	2 (11.1%)	11 (30.6%)	0.11	0.003 (0.008)	0.001 (0.003)	0.082 (0.178)	0.17

Among female partners, the vaginal microbiota was dominated by Lactobacillus spp. (prevalence 94.4%; mean proportional abundance = 0.816, SD = 0.339) (Additional file 1). Although Gardnerella, Finegoldia, Prevotella, Atopobium and Peptoniphilus were also prevalent, their proportional and cell-normalized abundance were much lower than Lactobacillus. The female partner’s vaginal microbiome composition was not associated with the male partner circumcision status.

Condomless sex causes immediate and short-term changes in genital bacterial density

Among male partners, condom use and male circumcision status significantly influenced the post-coital changes in coronal sulcus bacterial density. Condomless sex significantly increased bacterial density among circumcised male partners (p < 0.001 at all post-coital timepoints) but not among uncircumcised male partners (all p > 0.05; Fig. 1). No changes were observed after condom-protected sex (p = 0.63) (Additional file 1).

Fig. 1.

Impact of penile-vaginal sex on coronal sulcus and cervico-vaginal secretion bacterial load and proportional abundance. Legend: The effect of penile-vaginal sex on the total bacterial load in the coronal sulcus (panel A) and cervico-vaginal secretions (panel C) is shown for couples with complete data (n = 36) at baseline, + 1–2 h, + 7–8 h and + 48–72 h. Coronal sulcus data are shown separately for circumcised (n = 18) and uncircumcised males (n = 18), given the profound effect of circumcision status on CS bacterial load. In A and C, box and whisker plots visualize the median (box middle line), mean (circle), first quartile (lower end of box), third quartile (upper end of box), and outlier values within 1.5-times the inner quartile range (whiskers). Statistical comparisons conducted using a Wilcoxon Rank Sum test. Proportional abundance of bacterial genera shown as a stacked bar plot in B and D. Participant order is kept consistent across time and cervical secretion and coronal sulcus pairs are vertically aligned. Panels B and D exclude samples without a paired data available across cervical secretions and coronal sulcus samples

Among female partners, vaginal bacterial density also remained unchanged after condom-protected sex (p = 0.14). However, after condomless sex, vaginal bacterial density decreased at 1–2 h after condomless sex (9.2 to 8.7 log10 16S rRNA gene copies/mL p = 0.004) (Fig. 1C), returned to baseline levels by 7 h (p = 0.29), and increased significantly above baseline by 48–72 h (9.4 log10 copies/mL, p < 0.001) (Fig. 1C).

Rapid and reversible changes in coronal sulcus microbiome composition after condomless sex

Condomless sex induced extensive but transient shifts in the coronal sulcus microbiome, with significant shifts immediately post-coitus (PerMANOVA p < 0.01; Fig. S2), but returning to baseline by 48–72 h (p = 0.82). Sexual activity differentially impacted various genital bacteria, such as Lactobacillus, skin-associated bacteria (Corynebacterium and Staphylococcus), BV-associated bacteria (Prevotella and Gardnerella), and BASIC species.

Prevalence of Lactobacillus among male partners increased significantly from 46.4% at baseline to 96.4% at 1–2 h and 7–8 h (both p < 0.001) and reverted by 48–72 h (50.0%). Lactobacillus proportional abundance showed the same pattern (Pre-coital mean: 0.031; 1–2 h mean: 0.621; 7–8 h mean: 0.515; 48–27 h mean: 0.026). Lactobacillus cell-normalized abundance increased rapidly and remained slightly elevated at 48–72 h (Pre-coital cell-normalized abundance median: 5.7; 1–2 h median: 6,657.9; 7–8 h median: 3,422.0; 48–72 h median: 22.1; all p < 0.05 vs. baseline). While all major Lactobacillus species, including L. crispatus, L. iners, and L. gasseri increased early (all p < 0.01; Fig. 1B), only L. iners remained elevated at 48–72 h (p = 0.01).

The abundance of skin-associated bacteria also increased significantly after condomless sex. Corynebacterium and Staphylococcus rose significantly at 7–8 and 48–72 h (all p ≤ 0.048) (Additional file 1). However, abundance of BVAB or BASIC species did not change significantly, except for Prevotella and Gardnerella, which had significant increases. Gardnerella mean proportional abundance increased from 0.015 at baseline to 0.069 at 1–2 h (p = 0.008) and 0.053 at 7–8 h (p = 0.009), then reverted by 48–72 h (0.001, p = 0.15); these changes were mirrored in prevalence and cell-normalized abundance.

Condom use substantially reduced post-coital changes in the coronal sulcus microbiome. Bacterial density only increased moderately at 48–72 h (p = 0.046), but no changes were observed at 1–2 h or 7–8 h (Additional file 1), and the overall microbiome composition remained stable at 1–2 h and 48–72 h (PerMANOVA, p = 0.22 and p = 0.86, respectively) (Fig. S3). Lactobacillus showed small transient increases (Fig. S1), but skin-associated bacteria, BV-associated bacteria, and BASIC species were unchanged. These findings indicate that condom use effectively limits microbial transfer.

Male circumcision status modifies post-coital coronal sulcus microbiome response

The male partner’s circumcision status significantly shaped the coronal sulcus microbiome responses to condomless sex. Circumcised males showed transient increases in BASIC species (D. micraerophilus, D. propionicifaciens, P.bivia, and P. disiens) at 1–2 h and 7–8 h before returning to baseline levels. These changes were not observed in uncircumcised males.

In contrast, circumcision status did not alter the impact of condomless sex on Lactobacillus and BV-associated bacteria, indicating that circumcision status specifically impacted post-coital changes in BASIC species. Given that uncircumcised male partners had higher bacterial density and greater BASIC species carriage at baseline, the absence of detectable shifts of BASIC species in this group likely reflect masking by their high-density, anaerobe-dominated microbiome rather than a true lack of change.

Vaginal microbiome remains largely stable after condomless sex

In contrast to the marked perturbations observed in the male coronal sulcus, the vaginal microbiome remained largely stable following condomless sex. Although vaginal bacterial density showed changes—decreasing immediately after condomless sex (pre-coital v. 1–2 h median: 9.2 vs. 8.3 log10 copies/mL; p = 0.02) and with a small increase above baseline by 48–72 h (9.6 log10 copies/mL, p < 0.001)—vaginal microbiome composition remained similar across early and late time points (PerMANOVA 1–2 h: p = 0.50; 48–72 h: p = 0.99).

Lactobacillus abundance decreased moderately at 1–2 h after condomless sex in both proportional (0.81 to 0.74; p = 0.048) and cell-normalized abundance (9.2 to 8.8 log10 copies/mL; p = 0.06). By 48–72 h, while Lactobacillus proportional abundance returned to baseline, Lactobacillus cell-normalized abundance showed a small but significant increase (9.5 log10 copies/mL, p < 0.001; Additional file 1). This pattern was consistent across all major vaginal Lactobacillus species, including L. crispatus and L. iners (Fig. 2).

Fig. 2.

Bacterial cell-normalized abundance log response ratios at baseline and after sex for cervico-vaginal secretion and coronal sulcus samples. Legend: Mean log response ratio (LRR) of selected taxa cell-normalized abundance across baseline (T1), 1-2 h (T2), 7-8 h (T3), and 48-72 h (T4) post sex. Dots represent the mean LRR, with whiskers representing the 95% confidence intervals

Corynebacterium and Staphylococcus increased transiently after condomless sex in prevalence, proportional abundance, and cell-normalized abundance (all p < 0.01) but returned to baseline by 48–72 h (all p > 0.1; Fig. 2).

Although the overall BVAB group did not change significantly over time, Gardnerella cell-normalized abundance increased by 48–72 h (4.3 × 10⁵ to 7.3 × 10⁵ copies/mL; p = 0.012), while its prevalence and proportional abundance remained stable (all p > 0.05). The BASIC species showed no significant changes following condomless sex. Condom use prevented detectable alterations in vaginal microbiome composition (PerMANOVA 1–2 h: p = 0.95; 48–72 h: p = 0.90) and in all genital bacterial groups.

BASIC species carriage in couples

BASIC species carriage at baseline was lower among couples with circumcised male partners. Approximately half of the couples with circumcised male partners had no detectable BASICs (n = 8/17, 47.1%). Most commonly, BASICs were detected in only one partner (circumcised male n = 4; female n = 4), with only one couple having both colonized. In contrast, BASICs were detected in 82.4% of uncircumcised male partners (n = 14/17) and in two of the female partners, with only three couples having no detectable BASICs when the male partner is uncircumcised.

Post-coital acquisition patterns differed by male circumcision status. In couples with circumcised male partners (n = 12), BASIC were acquired at 1–2 h by 33.3% of female partners (4/12) and 50.0% of the circumcised male partners (6/12) (Additional file 1). At 48–72 h, BASICs persisted in 62.5% of female partners who acquired them (n = 5/8), but only 28.6% of the circumcised male partners (n = 2/7).

Among couples with uncircumcised male partners (n = 14), 28.6% of female partners acquired BASICs (n = 4/14), while the uncircumcised male partners generally retained their baseline status (n = 9/14 retained BASICs, n = 3/14 lost BASICs), and by 48–72-h, all but one couple returned to their baseline BASIC carriage status.

Vaginal pH as a mediator of Gardnerella abundance after condomless sex

Post-coital vaginal pH changes significantly influenced the vaginal bacterial composition (PERMANOVA R² = 0.04, p < 0.001 change vs. no change in pH). Vaginal pH increased significantly 1–2 h after condomless sex (pH + 1.5; one-sided t-test p < 0.001), with larger increases among couples who practiced intravaginal ejaculation (pH + 2.2 vs + 0.5; p < 0.001).

Changes in vaginal pH partially mediated the association between intravaginal ejaculation and Gardnerella abundance. Intravaginal ejaculation was directly associated with a significant decrease in Gardnerella cell-normalized abundance (Direct Effect: –2.3, p = 0.029). However, intravaginal ejaculation also increased vaginal pH, which was in turn also associated with increases in Gardnerella cell-normalized abundance (Indirect Effect: 1.8, p = 0.025), indicating countervailing mediation (Supplementary Table S1).

Each 2-unit rise in vaginal pH corresponded to an increase of 2.4 × 10⁵ Gardnerella cell-normalized abundance, after adjusting for Lactobacillus abundance and male circumcision status (Supplementary Table S2). In contrast, Gardnerella abundance remained stable or decreased among women who did not have increases in vaginal pH (Fig. S1).

Discussion

Our study of heterosexual couples showed that condomless sex extensively altered the male coronal sulcus microbiome at 1–2 and 7–8 h, though these changes were largely reversed by 48–72 h. Circumcision status of the male partners shaped these responses: circumcised men experienced transient increases in BASIC species, while uncircumcised men showed minimal detectable change, likely reflecting their higher baseline bacterial density. In contrast, the vaginal microbiome remained largely stable, with only modest shifts in bacterial density and select taxa. Post-coital increase in vaginal pH is associated with increases in Gardnerella abundance, whereas no pH change corresponded with stable or decreased Gardnerella. These findings underscore the dynamic yet resilient nature of genital microbiomes and the factors influencing microbial transfer.

This study highlights the dynamic nature of the genital microbiome and its disruption during condomless penile-vaginal sex. In women, bacterial vaginosis (BV) is a condition linked to vaginal inflammation, increased susceptibility to HIV and other STIs, and adverse obstetric outcomes such as prematurity and low birth weight [21]. BV is characterized by increased diversity and higher cell-normalized abundances of vaginal bacteria including Gardnerella. In males, higher cell-normalized abundances of BASIC species in the coronal sulcus are associated with genital inflammation and HIV acquisition risk [8], with urethral G. vaginalis and Prevotella also linked to inflammation [1, 22]. Epidemiologic studies strongly suggest that genital bacteria can be shared between partners [9, 13, 15, 16, 23–26] and recent penile-vaginal sex has been associated with changes in the vaginal [21] and penile [16] microbiota, but the immediate and short-term dynamics of this sharing are not well defined. In the current study we demonstrate significant but distinct microbiota changes in both sexes following condomless sex.

The most striking microbiota change in male partners induced by penile-vaginal sex was the transient dominance of Lactobacillus spp. in the coronal sulcus, which persisted for at least 8 h after sex. This enrichment reflects the baseline Lactobacillus-dominant vaginal microbiota of female partners. The transient nature of this shift may be due to microenvironmental differences in oxygen tension and nutrient availability in the coronal sulcus compared to the vagina [27]. Whether there are similar patterns of post-sex colonization by vagina-derived species in the penile urethra will be an interesting area for future research, particularly since G. vaginalis dominance of the urethral microbiome has been linked to recent vaginal sex [22, 24]. Transient increases in BASIC species were observed among circumcised males, likely reflecting their lower baseline abundance compared to uncircumcised males, where these taxa are already prevalent. These short-term shifts were generally not sustained.

In women, changes in the vaginal microbiome were less pronounced but more persistent, which may have implications for BV pathogenesis. The most notable finding was the delayed increase in Gardnerella cell-normalized abundance, first observed at 7–8 h and persisting for at least 72 h. This increase was only evident only under a mediation framework, where a near-zero total effect masked opposing associations between sex, post-coital pH changes, and Gardnerella abundance. These countervailing pathways may help explain why BV risk can increase after sex with a new partner or recur with repeated exposures, particularly in contexts where BV-associated species are present. The pH increase at 1–2 h may partly reflect residual seminal fluid after intravaginal ejaculation; however, this does not change our interpretation of pH mediating Gardnerella expansion. Importantly, Gardnerella expansion occurred even when vaginal Lactobacillus predominated at baseline, suggesting that sex-induced pH changes could promote expansion of low-abundance species. The male partner’s urethral microbiome may also serve as a reservoir for Gardnerella and other genital bacteria after condomless sex.

Our findings confirm that bacterial exchange occurs during condomless penile-vaginal sex, with both partners potentially serving as reservoirs for non-optimal bacteria. However, it is less clear the degree to which the bacteria exchange drives the window of genital inflammation induced by penile-vaginal sex [28, 29]. Prior analyses using this study showed that vaginal inflammation occurred after both condom-protected and condomless sex, preceded the expansion of Gardnerella, and was associated with biomarkers of epithelial disruption [28]. Together, this suggests that inflammatory effects may be primarily driven by epithelial trauma rather than by microbiome transfer or alteration. However, this hypothesis should be confirmed through larger studies, particularly in populations where non-optimal genital bacteria are more prevalent, such as in sub-Saharan Africa. In these settings, BASIC species have been associated with increased HIV acquisition risk among uncircumcised males [8].

This study has several limitations. The inclusion of both circumcised and uncircumcised male partners and both condom-protected and condomless sex increased heterogeneity of the study population and the small sample size limited subanalyses. Urethral swabs were not collected by male partners for practical reasons but should be included in future studies to further examine determinants of Gardnerella colonization and transfer. Although couples with PSA detected at baseline were excluded, sex in the preceding days could still influence the baseline microbiome. In addition, most female partners demonstrated a L. crispatus dominated vaginal microbiome at baseline, with BV and L. iners being much less common. Thus, future larger couples’ studies are needed, particularly with a focus on condomless sex and in communities where BV and L. iners are more prevalent in female partners and BASIC species in male partners, including groups in North America [30, 31] and sub-Saharan Africa [32, 33]. Future studies should incorporate long-read sequencing or complementary approaches to better resolve species-level associations, particularly for taxa that are challenging to differentiate using 16S V3–V4 sequencing.

Our study provides novel insights into genital microbiome disruption and exchange during condomless sex, emphasizing its relevance to BV pathogenesis and the potential for partners to act as reservoirs for non-optimal bacteria. Future research in diverse populations is essential to build on these findings and further elucidate the links between microbiome disruption, inflammation, and sexual health outcomes.

Methods

Sex as a biological variable

The Sex, Couples and Science (SECS) study is a longitudinal study to define the impact of penile-vaginal sex on the microbiome and immunology of the penile coronal sulcus and vagina. Analyses involved both biological male and biological female partners within established couples; additional data regarding the gender identity of participants were not collected.

Study design

Established heterosexual couples were recruited through community fliers and social media at the Women’s Health in Women’s Hands (WHIWH) Community Health Clinic in Toronto, Canada between July 2017 and November 2018. Study exclusion criteria included: HIV-1/2, T. pallidum, N. gonorrhoeae or C. trachomatis infection in either partner; genital ulceration or discharge in either partner during baseline clinical examination; either partner age < 16 years; irregular menses or pregnancy in the female partner; and a history of immunosuppressive medications or antibiotics within the previous 4 weeks in either partner. Participants with prostate-specific antigen detected at baseline were excluded from the analysis. After consenting, participants received pregnancy and BV/STI testing at an additional screening visit, including for bacterial vaginosis (BV) by Nugent Score, for Neisseria gonorrhoeae and Chlamydia trachomatis in urine (ProbeTech Assay, BD, Sparks, MD) and for HIV-1/2 and syphilis by chemiluminescent microparticle immunoassay (ARCHITECT System, Abbott GmbH & Co. KG) according to manufacturer instruction. Nugent scores were determined using Gram-stained vaginal smears based on the relative abundance of Lactobacillus, Gardnerella/Bacteroides, and Mobiluncus morphotypes. Scores were categorized as normal (0–3), intermediate (4–6), or bacterial vaginosis (7–10). All testing was performed at the Sinai Health Systems Microbiology Laboratory, Toronto, Canada.

Sample collection and timing

Eligible couples were sampled at four time points: one pre-coital visit prior to penile-vaginal sex and three post-coital visits at 1–2 h, 7–8 h, and 48–72 h after sex. Participants were asked to abstain from sex for > 48 h prior to the pre-coital visit, where participants completed a demographic/behavioural questionnaire. Following this, couples had either condomless or condom-protected penile-vaginal sex and were asked to both refrain from bathing or genital washing until after the 7–8 h post-coital visit and to refrain from additional sexual activity until after the 48–72 h post-coital visit. Among 41 enrolled couples, three couples were excluded from all analyses due to Prostate-Specific Antigen (PSA) detection in the female partner’s vaginal sample at the pre-coital (baseline) visit, and three additional couples were excluded from the 72-h analysis due to vaginal PSA detection at this visit.

At each timepoint, study coordinators guided sample collection in a location near the research facilities. Males self-collected 4 penile swabs using pre-moistened flocked swabs, one from each of the left and right side of the coronal sulcus/glans (with uncircumcised males instructed to retract foreskin), and one from each of the left and right side of the shaft. Two swabs were placed in 500 μl PBS, and 2 in dry cryovials for anaerobe culturing, with all swabs cryopreserved at −80˚C within 30 min. Female participants provided self-collected cervicovaginal secretion (CVS) samples at each visit using an Instead Softcup (Evofem, San Diego, CA, USA) that was self-inserted for 1 min; again, samples were transported on ice to the lab within 30 min of collection, where CVS was diluted 1:10 in sterile phosphate-buffered saline (PBS) and frozen in aliquots at −80C.

Prostate-specific antigen (PSA) (Serateac PSA Semiquant kit, Göttingen, Germany) testing was performed on vaginal samples as a marker of semen exposure to confirm abstinence prior to the pre-coital visit and to verify condomless sex using post-coital samples. Vaginal pH was assessed using provider-collected sterile swabs of the posterior fornix and nitrazine indicator pH paper.

Microbiome analysis

DNA was extracted from 80 μL of diluted swab eluent or 80 μL of cervicovaginal secretions using enzymatic and chemical lysis protocol as described previously [1]. The extracted DNA was analyzed by 16S rRNA gene-based broad-range qPCR (BactQuant) [34] and amplicon sequencing (V3-V4) using a modified protocol from Fadrosh et al. [35] with primers from Liu et al. [34], as described previously in Galiwango et al. [1]. Resultant data were processed as described earlier [1] using cutadapt v2.1, Trimmomatic v0.39, and DADA2 v1.16 [36] modules for primer removal, reads-filtering, chimera check, and inferred error models to identify Amplicon Sequence Variants (ASVs). ASV classification from the phylum to genus-level at 80% bootstrap confidence level was performed using the Naïve Bayesian Classifier (v.2.12)0.20 [37], with species level classification for selected curated and trained species (Dialister, Lactobacillus, Prevotella, Peptostreptococcus) using 100% sequence identity to reference strains [12, 38]. Additional details can be found at https://github.com/araclab/general/tree/main/microbiome/mb_analysis.

After excluding samples with < 1000 reads and removing rare taxa (< 0.0025%), we calculated taxon proportional abundance for each sample, defined as reads assigned to the taxon divided by the total reads in each sample. Vaginal samples were unavailable or failed sequencing in two participants; microbiota data were analyzed for 36 couples.

To estimate taxon abundance while accounting for variation in 16S rRNA operon copy number, we applied copy-number correction using rrnDB v5.7. For each ASV, we identified the median operon copy number of its assigned genus (C_g). Raw ASV read counts assigned to that genus (summed as R_g) were divided by these operon copy numbers to obtain copy-number–adjusted counts [39]. To scale these values to the sample’s total bacterial load, we multiplied each genus’s adjusted count by the ratio of the sample’s broad-range 16S qPCR gene copies (Q) to the sample’s total sequencing depth (R_T) [40]. Thus, estimated cell-normalized abundance for genus g was calculated as:

$$g=\left(\frac{\mathit{Rg}}{\mathit{Cg}}\right)x(\frac{Q}{R\mathrm{T}})$$

These values represent cell-normalized abundance estimates (i.e., cell-equivalent units), rather than true absolute cell counts. The full sequencing data set for this study can be accessed at SRA project number PRJNA1248021.

Statistical analysis

When comparing across groups within the same time period, differences in continuous variables (including proportional and cell-normalized abundance) were assessed using Wilcoxon rank-sum test, while differences in categorical variables (including prevalence) were assessed with a Chi-squared test. Changes in paired continuous variables across time were compared using the Wilcoxon signed rank test, and changes in categorical variables were compared using the McNemar-Bowker test. Median cell-normalized abundances are reported among those with any detection of the respective taxon. Log response ratios for cell-normalized abundance metrics were calculated as log10(1 + cell-normalized abundance per swab at T_i)/(1 + cell-normalized abundance per swab at T_j). Missing visits were excluded from comparisons.

Differences in overall microbiome composition were compared using a Permutational multivariate analysis of variance (PERMANOVA). PerMANOVA analyses were performed in R using the “adonis” function from the vegan package [41]. Bacteria Associated with Seroconversion, Inflammation, and Cells (BASICs) included Prevotella bivia, Peptostreptococcus anaerobius, Dialister micraerophilus, Prevotella disiens, Dialister propionicifaciens, Dialister succinatiphilus [8]. Bacterial vaginosis associated bacteria (BVAB) included Sneathia, Prevotella, Mobiluncus, Megasphaera, Fannyhessea, Gardnerella, Dialister, Porphyromonas, Peptostreptococcus, Parvimonas, and Bacteroides [42–46].

Changes in Gardnerella abundance across time were assessed using a mixed effects model with pH, condom use, and intravaginal ejaculation as predictors, controlled for individual level changes, and adjusted for Lactobacillus abundance and circumcision status. We conducted a causal mediation analysis (PROC CAUSALMED) to assess whether changes in vaginal pH mediated the association between intravaginal ejaculation and changes in Gardnerella abundance from T₁ to T₄. The exposure was intravaginal ejaculation (yes/no), the mediator was change in pH (T₂–T₁), and the outcome was change in Gardnerella abundance (T₄–T₁). Estimates of natural direct, indirect, and total effects were obtained using the counterfactual framework [47]. Statistical analyses were conducted in SAS, version 9.4 (SAS Institute Inc., Cary, NC, USA), and R, version 3.3.1 (R Core Team).

Authors’ contribution

All authors contributed to the study. AM and RK conceptualized and designed the study. AM, SB, AF, ET, SH, WT, and RK were involved in study planning, participant recruitment and sampling; DP, SN, JS, AO, JV, MA and CL planned and executed all aspects of 16S sequencing and analysis; DP and CL performed data analysis. DP, RK, AM, and CL drafted the first draft of the manuscript; all authors contributed to subsequent manuscript revisions.

Funding

Canadian Institutes of Health (CIHR; PJT-156123 and TMI-138656, RK); National Institutes of Health (R01A123002-01A1; CL). RK is supported by the Elisabeth Hofmann Chair in Translational Medical Research at the University of Toronto. The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data availability

Sequencing data and associated metadata for this study have been deposited in the NCBI SRA database and can be accessed under the BioProject accession number PRJNA1248021. The bioinformatics pipeline used to process the sequencing data can be found at https://github.com/araclab/general/tree/main/microbiome/mb_analysis.

Declarations

Ethics approval and consent to participate

The protocol was reviewed and approved by the HIV Research Ethics Board at the University of Toronto (RIS protocol #33381) and determined to be exempt by the George Washington University IRB (NCR235162). Written informed consent was obtained from all participants.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.