Transient association between semen exposure and biomarkers of genital inflammation in South African women at risk of HIV infection

Abstract Introduction Semen induces mucosal changes in the female reproductive tract to improve pregnancy outcomes. Since semen‐induced alterations are likely short‐lived and genital inflammation is linked to HIV acquisition in women, we investigated the contribution of recent semen exposure on biomarkers of genital inflammation in women at high HIV risk and the persistence of these associations. Methods We assessed stored genital specimens from 152 HIV‐negative KwaZulu‐Natal women who participated in the CAPRISA 008 trial between November 2012 and October 2014. During the two‐year study period, 651 vaginal specimens were collected biannually (mean five samples per woman). Cervicovaginal lavage (CVL) was screened for prostate‐specific antigen (PSA) by ELISA, whereas Y‐chromosome DNA (YcDNA) detection and quantification were conducted by RT‐PCR, representing semen exposure within 48 hours (PSA+YcDNA+) and semen exposure within three to fifteen days (PSA−YcDNA+). Soluble protein concentrations were measured in CVLs by multiplexed ELISA. T‐cell frequencies were assessed in cytobrushes by flow‐cytometry, and vulvovaginal swabs were used to detect common vaginal microbes by PCR. Linear mixed models adjusting for factors associated with genital inflammation and HIV risk were used to assess the impact of semen exposure on biomarkers of inflammation over multiple visits. Results Here, 19% (125/651) of CVLs were PSA+YcDNA+, 14% (93/651) were PSA−YcDNA+ and 67% (433/651) were PSA−YcDNA−. Semen exposure was associated with how often women saw their partners, the frequency of vaginal sex in the past month, HSV‐2 antibody detection, current gonorrhoea infection and Nugent Score. Both PSA detection (PSA+YcDNA+) and higher cervicovaginal YcDNA concentrations predicted increases in several cytokines, barrier‐related proteins (MMP‐2, TIMP‐1 and TIMP‐4) and activated CD4+CCR5+HLA‐DR+ T cells (β = 0.050; CI 0.001 to 0.098; p = 0.046) and CD4+HLA‐DR+ T cells (β = 0.177; CI 0.016 to 0.339; p = 0.032) respectively. PSA detection was specifically associated with raised pro‐inflammatory cytokines (including IL‐6, TNF‐α, IP‐10 and RANTES), and with the detection of BVAB2 (OR = 1.755; CI 1.116 to 2.760; p = 0.015), P. bivia (OR = 1.886; CI 1.102 to 3.228; p = 0.021) and Gardnerella vaginalis (OR = 1.815; CI 1.093 to 3.015; p = 0.021). Conclusions More recent semen exposure was associated with raised levels of inflammatory biomarkers and the detection of BV‐associated microbes, which declined by three to fifteen days of post‐exposure. Although transient, semen‐induced alterations may have implications for HIV susceptibility in women.

and increases HIV infection risk, even by less fit viruses [4,8]. Biomarkers of genital inflammation include elevated cervicovaginal cytokines, immune cell recruitment, alterations in barrier-related proteins and increased microbial diversity [4][5][6][7]. Semen contains several bioactive molecules and a diverse array of microbial communities [9][10][11][12], which may alter HIV susceptibility in women by promoting genital inflammation. A better understanding of the female immune response during condomless sex and the semen properties that promote genital inflammation may aid in designing effective biomedical HIV prevention strategies in women.
Biomarkers that detect semen within vaginal specimens may be beneficial to characterize the effects of semen exposure on female genital inflammation and HIV risk. Prostate-specific antigen (PSA) and Y-chromosome deoxyribonucleic acid (YcDNA) detection are well-characterized biomarkers of semen exposure [13][14][15][16][17][18]. PSA is produced by the prostate gland, and detection in vaginal fluids at concentrations ≥1 ng/ mL can be used to assess semen exposure at the female genital tract (FGT) even from vasectomized males [was used to detect PSA in [17][18][19]. In women, PSA is present for a short duration following condomless sex, and detection indicates semen exposure within 48 hours of cervicovaginal sampling [14,18]. Alternatively, the measurement of YcDNA can be used as a more stable marker of semen exposure [15,16]. YcDNA detection involves polymerase chain reaction (PCR) amplification of the testis-specific protein Y-encoded (TSPY) gene region and the sex-determining region Y (SRY) gene region in the Y-chromosome [15,20]. YcDNA can be detected in vaginal fluid up to 15 days after condomless sex [15,16,21]. Furthermore, since YcDNA is detectable in the presence of spermatozoa, YcDNA quantities at the FGT may indicate sperm count and seminal protein concentrations.
Considering the transience of semen-associated immune alterations in the FGT [1,22], a pro-inflammatory immune response to semen exposure may be better characterized using a biomarker of recent condomless sex. Here, we hypothesized that more recent semen exposure and higher cervicovaginal YcDNA concentrations would be associated with the inflammatory environment related to HIV risk in women. To test this hypothesis, we compared immune and microbial markers of genital inflammation among women with evidence of semen exposure within 48 hours (PSA+YcDNA+), three to fifteen days (PSAÀYcDNA+) and no semen exposure within 15 days prior to genital sampling (PSAÀYcDNAÀ). Additionally, we investigated the association between markers of genital inflammation and cervicovaginal concentrations of YcDNA, which may also reflect more recent sex and male protein concentrations at the FGT.

| Study population and design
We enrolled 152 HIV-negative women between 20 and 44 years of age from KwaZulu-Natal who participated in the CAPRISA 008 study [23][24][25]. Women participated in the CAPRISA 008 study over two years, between November 2012 and October 2014, and were followed up for an average of 22 months [23,25]. Demographic data were assessed at baseline, and vaginal specimens (including cervicovaginal lavage (CVL), cytobrushes and vulvovaginal swabs) were collected at enrolment and biannually at months 6,12,18,24 and study exit (average 5 AE 1 visit; 651 genital specimens) as previously reported [24][25][26][27]. Semen biomarkers were measured in vaginal specimens collected at each of the multiple study visits per participant. The detection of semen biomarkers indicated whether semen exposure occurred between zero to two and three to fifteen days before specimen collections at the respective visits. Participants provided informed consent for the specimen storage and use in future studies (BFC237/010). This study protocol was approved by the University of KwaZulu-Natal Biomedical Research Ethics Committee (BE258/19).

| YcDNA detection and quantification
DNA extraction was conducted on CVL pellets using the Mag-NAPure LC DNA Isolation Kit I (Roche Applied Science, Indianapolis, IN, USA), as instructed by the manufacturer. The Human Y-chromosome Quantification Kit (PrimerDesign Ltd, Chandler's Ford, UK) was used to detect a TSPY1 gene region present on the Y-chromosome within the extracted DNA. The Y-chromosome primer/probe mix and the PrecisionFAST TM Mastermix were used according to the manufacturer's instructions (PrimerDesign Ltd). Amplification was conducted on the Applied Biosystems â QuantStudio TM 5 real-time (RT)-PCR System (Thermo Fisher Scientific, Waltham, MA, USA). Detection of YcDNA within vaginal specimens indicated semen exposure within 15 days before genital sampling [15,16]. The Quantifiler TM Trio DNA Quantification Kit from Applied Biosystems TM (Thermo Fisher Scientific) was used to quantify YcDNA concentrations in YcDNA+ CVL specimens as outlined in the manufacturer's protocol. The assay simultaneously quantified the total amount of amplifiable human DNA and human male DNA in 10 µL of the sample.

| Investigation of soluble biomarkers of genital inflammation
The levels of genital cytokines and barrier-related proteins were assessed in CVL supernatants using multiplexed ELISA assays. Cytokine concentrations were measured using the Bio-Plex Pro TM Human Cytokine 21-Plex and 27-Plex kits (Bio-Rad Laboratory, Hercules, CA, USA) as previously described (Table S1) [28]. Concentrations of matrix metalloproteinases (MMPs) and tissue inhibitors of metalloproteinases (TIMPs) were quantified using the MMP 9-Plex, and TIMP 4-Plex kits (Bio-Rad Laboratory), respectively, as instructed by the manufacturer. MMP/TIMP measurements were performed on baseline samples (n = 136; Figure S1). All analyte concentrations were measured using the Bio-Plex â 200 system (Bio-Rad Laboratory).

| Investigation of immune cell frequencies
Multiparametric flow cytometry was used to assess the expression of activation markers (CD38+ or HLA-DR+), the maker of proliferation (Ki67+) and the HIV co-receptor (CCR5+) on CD4+ T cells in cervical cytobrush-derived specimens. The viability of cervical mononuclear cells (CMCs) was determined using the LIVE/DEAD TM Fixable Dead Cell Staining Kit (Invitrogen Life Technologies, Carlsbad, CA, USA) as outlined in the manufacturer's protocol. CMCs were treated with antibody-conjugated fluorophores, washed and fixed (Table S2). Data were acquired on an LSRII flow cytometer (BD Immunocytometry Systems, San Jose, CA, USA) and analysed with FlowJo TM Software version 9.9 (Tree Star, Inc., US). The gating strategy was previously published [24].

| Detection of sexually transmitted infections (STIs) and vaginal microbes
Genital swabs were used to detect common STI pathogens and vaginal microbes, as previously described (Table S3) [29]. Multiplex PCR amplification was conducted to detect STI pathogens using the Fast-track Diagnostics STD9 detection kit, as outlined in the manufacturer's protocol. Common vaginal microbes were detected using Applied Biosystems TM TaqMan â assays [29]. All reactions were conducted using the Applied Biosystems 7500 RT-PCR machine (Thermo Fisher Scientific). Data on STI detection in vaginal specimens were available for all visits (n = 648), whereas data on bacterial vaginosis (BV)-associated bacteria were generated for all visits except baseline (n = 515; Figure S1). BV was diagnosed using Nugent scoring by Gram stain microscopy [30].

| Statistical considerations
Statistical analyses were performed using GraphPad Prism version 8.4.3 (GraphPad Software, San Diego, CA, USA), STATA version 15.0 (StataCorp., College Station, TX, USA) and SAS version 9.4 (SAS Institute Inc., Cary, NC, USA). Continuous variables and proportions were compared at baseline using the Kruskal-Wallis test and the chi-square or Fisher's exact test respectively. Longitudinal analyses included one model with YcDNA concentrations as the main exposure variable and a complementary model with PSA/YcDNA categories as the main exposure variable. CVL specimens were classified in terms of PSA and YcDNA detection as PSAÀYcDNAÀ, PSA+YcDNA+ and PSAÀYcDNA+. The Mann-Whitney test was used to determine whether YcDNA concentrations differed significantly between PSA+YcDNA+ and PSAÀYcDNA+ specimens at baseline. Receiver operating characteristic (ROC) curve analysis was used to plot the lowest cut-off YcDNA concentration that predicted PSA positivity. Linear regression models were used to determine the impact of PSA+YcDNA+ and PSA-YcDNA+ events on MMP/ TIMP concentrations compared to PSAÀYcDNAÀ events at baseline. Spearman's Rank correlation was used to determine the relationship between concentrations of YcDNA and MMPs/TIMPs at baseline. Linear mixed models (LMMs) adjusting for repeated measures were used to assess the impact of PSA+YcDNA+ and PSAÀYcDNA+ events on cytokine concentrations and immune cell frequencies compared to PSAÀYcDNAÀ events over multiple visits. Generalized estimating equation (GEE) models using a logit link were used to compare microbe presence between PSA+YcDNA+, PSAÀYcDNA+ and PSAÀYcDNAÀ groups over multiple visits. Multivariable LMMs adjusted for the contribution of study arm (CAPRISA or family planning clinics), time in the study, Nugent Score, participant age, STIs, the frequency of vaginal sex in the past month, and genital inflammation status [4,28]. The Benjamini-Hochberg method was used to calculate the false discovery rate (FDR).

| Study population
Baseline clinical and demographic data are reported for 137/ 152 women with both PSA and YcDNA data available ( Table 1). The study participants' overall median age was 28 years (interquartile range (IQR) 25 to 33 years). Semen exposure was determined by YcDNA detection (YcDNA+), with PSA detection further stratifying participant groups into semen exposure within 48 hours of sampling (PSA+YcDNA+) and between three and fifteen days of sampling (PSAÀYcDNA+). At baseline, 28% (38/137) of women were PSA+YcDNA+, 14% (19/137) were PSAÀYcDNA+ and 58% (80/137) were PSAÀYcDNAÀ, suggesting no semen exposure within 15 days of sampling. Semen exposure was associated with how often women saw their partners (p = 0.038), the frequency of vaginal sex in the past month (p = 0.006), Herpes simplex virus (HSV)-2 detection (p = 0.047), current gonorrhoea infection (p = 0.018) and Nugent Score (p = 0.003). Higher median Nugent Scores were specifically driven by semen exposure within 48 hours compared to no semen exposure (median 3 (IQR 1 to 7) vs. median 1 (IQR 0 to 3), respectively, p = 0.003). Furthermore, women with evidence of semen exposure within three to fifteen days reported more frequent acts of vaginal sex in the last month than women with no evidence of semen exposure (median 8 (IQR 4 to 10) vs. median 4 (IQR 2 to 6), respectively, p = 0.017).
Since higher YcDNA concentrations were observed soonest after semen exposure, it suggests a potential for a threshold YcDNA concentration to serve as a proxy for the timing of semen exposure before genital sampling. ROC curve analysis suggests a YcDNA concentration cutoff of 0.005 ng/µL predicts PSA positivity with the greatest sensitivity [87.4%, confidence interval (CI) 81.2, 93.6] and with a specificity of 62.3% (CI 50.9, 73.8; Table S4; Figure S2).

| Semen-associated alterations to cervicovaginal cytokines
Since semen's impact on the FGT is likely short-lived [1,22], we hypothesized that higher YcDNA concentrations and semen exposure within 48 hours would be associated with greater alterations in cervicovaginal cytokines (Table S1). Multivariable LMMs were used to compare YcDNA  (3) The chi-square and Fisher's exact tests were used to compare proportions between the groups as deemed appropriate. Continuous data were assessed by Kruskal-Wallis tests to compare differences between no semen exposure (PSAÀYcDNAÀ), semen exposure within 48 hours (PSA+YcDNA+) and semen exposure within three to fifteen days (PSAÀYcDNA+), Dunn's post-testing was applied to adjust for multiple comparisons. Significant differences between no semen exposure and semen exposure within 48 hours are indicated by (a), whereas differences between no semen exposure and exposure within three to fifteen days are indicated by (b).

| Semen-associated alterations to endocervical immune cell frequencies
Since HIV must gain access to local target cells for infection, we investigated whether YcDNA concentrations and the timing of semen exposure were associated with alterations in endocervical immune cell frequencies. In multivariable LMMs, higher YcDNA concentrations were associated with increased frequencies of the activated CD4+HLA-DR+ T cell populations (n = 145 genital specimens; b = 0.177; CI 0.016, 0.339; p = 0.032; Figure 5a).

| Semen-associated alterations to vaginal microbes
Since vaginal microbial diversity is associated with HIV infection in women [6,7], we examined the contribution of YcDNA concentrations and the timing of semen exposure on vaginal microbe detection over multiple visits [6,7]. In adjusted GEE models, higher YcDNA concentrations were associated with reduced detection of Gardnerella vaginalis in vaginal specimens  Longitudinal comparison of cytokine concentrations between semen exposure within 48 hours (PSA+YcDNA+ specimens; n = 124) and three to fifteen days (PSAÀYcDNA+ specimens; n = 93) relative to no semen exposure (PSAÀYcDNAÀ specimens; n = 433). Cytokines are ordered according to their functions: pro-inflammatory (red circles), chemotactic (blue squares), growth/haematopoiesis (green triangles), adaptive response (purple diamonds) and regulatory (orange hexagons) cytokines. Grey shadings represent the cytokines previously associated with genital inflammation and in demonstrating its association with HIV risk in this cohort [4,28]. b-coefficients are depicted by shapes and error bars indicate the 95% CI. Filled shapes indicate significant p-values (p < 0.05), and significance after FDR adjustment is indicated by (*). Table S1 contains a list of abbreviations for the 48 cytokines measured in this study. CI, confidence interval; YcDNA, Y-chromosome DNA.

| DISCUSSION
We investigated the contribution of recent semen exposure to the inflammatory environment linked to HIV risk in women.
PSA detection and higher YcDNA concentrations, both suggesting more recent semen exposure, were associated with elevated cervicovaginal cytokines, barrier-related proteins, increased detection of BV-associated microbes and higher HIV target cell frequencies in vaginal specimens (Figure 7). These findings suggest that semen-associated immune responses at the FGT are transient and highlight the importance of considering the timing of genital sampling after condomless sex.
At baseline, more than a third of the women had detectable YcDNA, indicating semen exposure within a range of 15 days before genital sampling. The timing of semen exposure was specifically related to the number of vaginal sex acts reported and the BV Nugent Score. Semen exposure within three to fifteen days allowed for a longer range of semen detection and was therefore associated with a greater number of coital episodes in the last month. Semen exposure within 48 hours was associated with higher Nugent Scores, suggesting that condomless sex is linked to a short-term increased presence of BV-associated microbes at the FGT.
We assessed the relationship between YcDNA quantities and the timing of semen exposure. Concentrations of YcDNA at the FGT may indicate the amount of seminal proteins and sperm count in the ejaculate of the male partner during condomless sex. These data demonstrated that YcDNA concentrations are also related to the timing of semen exposure since PSA detection was associated with higher YcDNA concentrations. The Quantifiler TM Trio DNA Quantification Kit used in this study correctly predicted PSA positivity in vaginal specimens with even minimal levels of YcDNA. YcDNA concentrations could be useful in studies requiring the use of a single semen biomarker, which also indicates the timing of semen exposure. Further studies are needed to confirm the feasibility of using YcDNA concentrations to determine more recent semen exposure (i.e. PSA detection) before genital sampling.
Semen exposure within 48 hours and higher YcDNA concentrations were associated with raised levels of several cytokines, whereas semen exposure within three to fifteen days was associated with increased concentrations of only two cytokines (IL-3 and IP-10). Although transient, elevated concentrations of IL-6, TNF-a, IP-10, RANTES and IL-10 associated with more recent semen exposure have also been related to increased HIV risk in women [4,26,32]. Of particular importance are the chemokines, with raised IP-10 independently associated with HIV seroconversion and T-cell recruitment [4,33], and RANTES, a CCR5 ligand, involved in both the blocking of HIV binding to CCR5 on target cells and the recruitment of these target cells to the FGT [34]. Recent semen exposure was also associated with increased concentrations of MMP-2, TIMP-1 and TIMP-4. Increased MMPs/ TIMPs may signify wound healing since microabrasions are introduced at the FGT during coitus [35] and may also impact HIV risk in women through reduced mucosal barrier integrity and/or target cell recruitment [5].
Our findings confirmed that more recent semen exposure was also associated with increased frequencies of activated endocervical CD4 T cells (CD4+HLA-DR+ and CD4+CCR5+HLA-DR). Studies suggest that semen exposure at the FGT is associated with an initial pro-inflammatory response resulting in leucocyte recruitment to remove excess and abnormal sperm [1,36]. This may explain the increase in CD4+HLA-DR+ T cell frequencies in response to higher YcDNA concentrations observed in this study, which may also indicate higher sperm counts. Others have reported that this inflammatory response dissipates within 48 to 72 hours, and a regulatory T cell immune response is mounted to facilitate conception [2,3,37]. This may also explain why semen exposure within three to fifteen days was not related to immune cell alterations in this study. Activated endocervical CD4 T cells are putative HIV target cells, and increased frequencies of these cells may promote heterosexual transmission of HIV from an infected male partner.
Pathogens, STIs and commensal microbes present in semen and at the male genital tract [11,12,38,39] are transferred to the FGT during condomless sex and could alter the vaginal microbial composition [11,40]. Surprisingly, higher concentrations of cervicovaginal YcDNA were associated with reduced detection of G. vaginalis. Further studies are required to determine reasons for this association, for example, the contribution of semen-derived antimicrobial activities, the impact of semen on G. vaginalis growth dynamics, etc. Conversely, semen exposure within 48 hours was associated with increased detection of several BV-associated microbes (BVAB2, P. bivia and G. vaginalis) and reduced detection of L. jensenii, whereas women exposed to semen within three to fifteen days only had an increased detection of P. bivia in vaginal specimens. Studies have demonstrated that a diverse vaginal microbiome with increased Prevotella and reduced Lactobacillus crispatus is associated with a higher risk of HIV acquisition among women [6,7]. This is due to the ability of BV-associated microbes to induce cytokine production and the recruitment of activated CD4+ T cells, both of which were observed in this study [6,7].
A strength of this study was the use of recent semen biomarkers and the wealth of immune and microbial data to reliably assess the effect of semen exposure on the female genital mucosa over multiple visits. To our knowledge, this is the first study to assess the impact of YcDNA concentrations and the timing of semen exposure on markers of genital inflammation related to HIV risk in women. Given that semen itself contains several endogenously produced cytokines and CD4+ T cells from the male partner [10,22,41,42], a limitation of this study was the inability to determine whether the elevated immune responses detected were from semen itself or an immune response elicited at the FGT. However, Chen et al.
[43], demonstrated that endometrial epithelial cells and stromal fibroblasts treated with seminal plasma had raised levels of several cytokines after adjusting for endogenous seminal plasma cytokines. Similar to our findings, seminal plasma exposure was associated with elevated G-CSF, IL-6, TNF-a and VEGF among others. This study distinguished cytokines produced by the FGT from endogenous seminal plasma cytokines and suggests that the immune markers detected here were mounted by the FGT and not due to residual seminal components.

| CONCLUSIONS
This longitudinal study demonstrates that semen exposure induces a transient pro-inflammatory immune response at the FGT which may significantly impact HIV risk in women. Here, PSA detection and higher concentrations of YcDNA, both indicative of recent semen exposure, were associated with increases in cervicovaginal cytokines and barrier-related proteins, recruitment, and activation of endocervical CD4 T cells and alterations in vaginal microbes. These findings emphasize the need for studies of genital mucosal immunity to STIs such as HIV to consider the contribution of semen exposure to the FGT immune and microbial environments, preferably using a biomarker of recent semen exposure. This study also highlights the importance of consistent condom use, particularly in settings where women are at increased risk of acquiring HIV.

SUPPORTING INFORMATION
Additional information may be found under the Supporting Information tab for this article. Table S1. List of cytokines measured in cervicovaginal lavage supernatant specimens Table S2. Flow cytometry information on the antibody clones, fluorophores, and suppliers Table S3. List of common STI pathogens and other vaginal microbes measured in vulvovaginal swabs Table S4. Sensitivity and specificity for the YcDNA concentration cutoff value of 0.005 ng/µL Figure S1. Graphical representation of the data available at baseline and longitudinally for CAPRISA 008 study participants. Figure S2. ROC curve for all YcDNA concentration cutoff values.