Herpesviruses and HIV-1 drug resistance mutations influence the virologic and immunologic milieu of the male genital tract

Abstract

Objectives

To further understand the role that chronic viral infections of the male genital tract play on HIV-1 dynamics and replication.

Design

Retrospective, observational study including236 paired semen and blood samples collected from 115 recently HIV-1 infected antiretroviral naïve men who have sex with men (MSM).

Methods

In this study, we evaluated the association ofseminal HIV-1 shedding to coinfections with seven herpesviruses, blood plasma HIV-1 RNA levels, CD4+ T-cell counts, presence of transmitted drug resistance mutations (DRM) in HIV-1 pol, participants’ age and stage of HIV-infection using multivariate generalized estimating equation methods (GEE). Associations between herpesvirus shedding, seminal HIV-1 levels, number and immune activation of seminal T-cells was also investigated (Mann Whitney).

Results

Seminal herpesvirus shedding was observed in 75.7% of subjects. Blood HIV-1 RNA levels (p<0.01), and seminal CMV and HHV-8 levels (p<0.05) were independent predictors of detectable seminal HIV-1 RNA, and higher seminal HIV-1 levelswere associated with CMV and EBV seminal shedding,and absence of DRM (p < 0.05). CMV and EBV seminal shedding was associated with higher number of seminalT-lymphocytes, but onlypresence of seminal CMV DNA was associated withincreasedimmune activation of T-lymphocytes in semen and blood.

Conclusions

Despite high median CD4cells number, we found a high frequency of herpesviruses seminal shedding in our cohort. Shedding of CMV, EBV and HHV-8 and absence of DRM were associated with increased frequency of HIV-1 shedding and/or higher levels of HIV-1 RNA in semen, which are likely important co-factors for HIV-1 transmission.

Introduction

In HIV-infected individuals, levels of human immunodeficiency virus-1 (HIV-1) RNA in blood [1,2] and semen [3,4] correlate with the risk of sexual transmission[5–7]. Although HIV-1 RNA levels in blood correlate with levels in seminal plasma[3,8–10], local genital factors, particularly concomitant bacterial sexual transmitted infections (STI), can increase HIV-1 shedding in semen [10–12]. Not only bacterial STI but also chronic viral co-infections, e.g.herpes simplex virus type 2 (HSV-2) and cytomegalovirus (CMV),likely influence HIV-1 pathogenesis, transmission and dynamics within the genital tract[13–16].

Previous works have demonstrated that some herpesviruses are likely important for HIV-1 transmission. For example, levels of CMV [9,16,17] and HSV-2 [18–20] DNAin semenareassociated with higher HIV-1 RNA seminal levels, and HSV-2 seropositivity of source partners is associated with HIV-transmission among men who have sex with men (MSM) [3]. However, treatment of HSV in HIV-infected potential source partners with acyclovir does not reduce HIV transmission [21], suggesting that other factors remain. Although levels of HIV-1, HSV-2 and CMV in the genital tract are well documented, they are not the only viruses that replicate in the male genital tract, and other members of the herpesfamily are sexually transmitted and are common worldwide [22,23]. These viruses often co-infect the same host and likely influence each other’svirologic dynamics and host immune response [13,22–24]. One study evaluated 50 semen samples collected from HIV-infected heterosexual men in India with a median of 281CD4+ T-cells/ ml and found that concomitant EBV and CMV seminal shedding was associated with increased HIV-1 RNA levels in semen [17]. Although CMV and EBV shedding was high in this cohort (70% and 56% respectively), shedding of HSV HHV-6, −7 and −8 was likely too low (8%, 2%, 12% and 6% respectively) to interpret associations with HIV-1 shedding.

To further understand the role that chronic viral infections of the male genital tract play on HIV-1 dynamics and replication, this study measured viral levels of seven different herpesviruses in 236semensamples from 115 HIV-infected antiretroviral-na¨ıve MSM, and examined the relationships between levels of these viruses and the presence of HIV-1 RNA within the genital tract.In addition, we evaluated the impact of HIV-1 transmitted drug resistance mutations (DRM) on HIV-1 seminal shedding and,on a subset of 36 subjects [13] we investigated the associations betweenthese herpesviruses and the number and immune activation status of T-cells in semen and blood.

Materials and Methods

Participants, samples and clinical laboratory tests

A total of 236 semen samples from 115 recently HIV-infected antiretroviral-naïve participants from the San Diego Primary Infection Cohort [25] were included. Semen was collected by masturbation,as describedpreviously[13,26,27]. At baseline, Neisseria gonorrhoeae and Chlamydia trachomatis were ruled-out by PCR of urine samples and syphilis infection was excluded by rapid plasma reagin titers (RPR). In blood, CD4+ T lymphocyte subsets were measured by flow-cytometry (LabCorp) and HIV-1 RNA was quantified (Amplicor HIV Monitor Test, Roche Molecular Systems Inc.). HHV-8 serology was determined using the ‘‘HHV-8 IgG Antibody ELISA Kit’’ (Advanced Biotechnologies Inc.). Clinical data were collected, and estimated duration of infection (EDI) was calculated using established algorithm [25,28]; HIV-1 subtype was determined for 113 subjects using HIV-1 pol sequence data generated by Viroseq 2.0 (Applied Biosystems) using SCUEAL[29]. For 2 subjects sequences were not obtained because of blood plasma viral load (VL)<500copies/ml. DRM were analyzed using the Genotypic Resistance Interpretation Algorithm on the Stanford HIV drug resistance database [30]. The studies were conducted with appropriate written subject consent and were approved by the Human Research Protections Program at the University of California, San Diego.

RNA Extraction from seminal plasma, and HIV RNA quantification

HIV-1 RNA levels were measured in seminal plasma by first concentrating HIV-1 RNA from 500µL of seminal plasma with high-speed centrifugation (23,500×g at 4°C for 1 hour), as describedpreviously[13]. Concentrated RNA was extracted using High-Pure Viral RNA Kit (Roche) and HIV-1 cDNA was generated using the SuperScript III First-Strand Synthesis Kit (Invitrogen) with specific primer mf302 [31]. HIV RNA in seminal plasma was quantified by real-time PCR in an ABI 7900HT thermocycler (Applied Biosystems) with 0.005 µM ROX aspassive reference. A total reaction volume of 50 µl was added to each well consisting of 5 µl of cDNA, TaqMan Environmental Mastermix 2.0 (Applied Biosystems), PCR primers mf302 and mf299 [31] (1µM), and probe mf348 [31] (0.3 µM) (supplementary Table 1). The PCR conditions were 2min at 50°C, 10 min at 95°C and 60 cycles of 15s at 95°C and 60s at 60°C. HIV RNA quantification standard was obtained from the DAIDS Virology Quality Assurance (VQA) Program [32].

DNA Extraction from seminal plasma, and herpesvirus DNA quantification

Viral DNA was extracted from 200µl of seminal plasma using QIAamp DNA Mini Kit (Qiagen) per manufacturer’s protocol. EDTA (50 µM) was added to seminal plasma to inhibit DNase activity. Levels of different herpesviruses in semen were measured by real-time PCR in an ABI 7900HT thermocycler (Applied Biosystems, CA) with 0.005µM ROX as passive reference. A total reaction volume of 50µl was added to each well consisting of 10µl DNA, TaqMan Environmental Mastermix 2.0 (Applied Biosystems), virus specific PCR primers (1µM each) and probe (0.3 µM) (supplementary Table 1). Quantification standards for the different herpesviruses were obtained using plasmid preparations with known concentrations.

Multiparameter flow cytometry analysis (FACS)

For a subset of 36 subjects, seminal cell sample were analyzed by flowcytometryonadual-laser, 6-color Becton Dickinson FACS Canto using Diva (6.1) or FlowJo (9.0) software, as describedpreviously[13]. Cells were stained using CD45-PerCP/Cy5.5 (clone 2D1), CD45RA-PE (clone HI100), CD4-FITC (Leu 3a/3b multiclone), CD38-PE/Cy7 (clone HB7), CD3-APC (clone SK7) and CD8-APC/Cy7 (clone SK1) (BD Biosciences).

Statistics

Statistical analyses were performed using Graph-Pad Prism 5.0 software (GraphPad Software) and SAS (version 9.2).

VL variables were transformed to the base ten logarithm values. Non-normal data were either dichotomized or ordinalized. Comparison of viral seminal shedding between groups was performed using Fisher exact test.Genital herpesviruses, blood HIV-1 levels, CD4+ T-cell number, presence of DRM, age, or stage of HIVinfection (cutoff at 90 days after EDI) were associated with detectable seminal HIV-1 shedding (as a binomial variable, i.e. with seminal levels above/below 50 copies/ ml) using univariate and multivariate generalized estimating equation (GEE) to account for repeated measurements using a exchangeable correlation structure (entry criterion for multivariate analysis:p = 0.35; level of significance: p=0.05). CMVand EBV were categorized in 3 groups, i.e. undetectable, low and high shedding (cutoff: mean viral level among detectable samples). HIV1 genital levels in CMVand EBV positive versus negative samples and in samples with and without DRM, as well as differences in seminal T-cell numbers and immune activation were compared using Mann Whitney test.

Results

Study participants

HIV-infected participants (n = 115) were antiretroviral naïve MSM with a median age of 33 years (Table 1). All but threeparticipantswith available HIV-1polsequence (n = 113)were infected with HIV-1 subtype B virus (the exceptions being two B/D and one B/F1 recombinant). From these individuals, 236single or longitudinal semen samples (median of 2 time-points for each patient, range 1–8) were collected. For 60 subjects with repeated samplingthere was a median follow-up of 65 days (IQR:29–172). For all subjects at baseline, the median CD4 count was 519 cells/ml (IQR: 413–712cells/ml),theirmedian EDI was 97 days (IQR: 79–152), and theirmedian blood plasma HIV-1 levelwa-s4.83HIV-1 RNA log₁₀ copies/ml (IQR: 3.92–5.27).At baseline,ten participants had positive syphilis screening tests by RPR, three were positive for urethral Chlamydia by urine PCR, another three were positive for both Chlamydia and Neisseria gonorrhoeae by urine PCR, and one participant was positive for all three (Syphilis, Chlamydia and N. gonorrhea). Allsubjects with identified bacterial STI were treated for these infections and included in the study. Fifty-five subjects (48%) were positive for HHV-8 IgG antibody.

Table 1.

Demographics at baseline.

Characteristics
Participants; n	115
Samples; n	236
Time points/subject; median (range)	2 (1–8)
Male sex; n (%)	115 (100)
Age (years); median (IQR)	33 (27–39)
MSM HIV risk factor, n (%)	115 (100)
Caucasian/Non-Hispanic, n (%)	69 (60)
Antiretroviral naïve; n (%)	115 (100)
HIV-1 subtype B; n (%)	110 (97)
Drug resistance mutation; n(%)	19 (16.8)
EDI days; median (IQR)	97 (79–152)
CD4+ cell counts/ml; median (IQR)	519 (413–712)
Log10 HIV RNA copies; median (IQR)	4.83 (3.92–5.27)

n: number of participants/ejaculates included in the study; Time points: number of paired samples for each subject; IQR: interquartile range, MSM: men who have sex with men; EDI: time from estimated date of infection at baseline; subtype and DRM information included for 113 subjects with available pol sequence.

Shedding frequency and levels of HIV and herpesvirusesin semen

Despite high CD4+ cell count (519 CD4+ T-cells/ml) and early stage disease (median 97 days EDI at baseline),75.7% of the subjects had at least one herpesvirus detected in one or more of their seminal plasma samples, and70.3% of all samples had at least one detectable herpesvirus. These includedCMV (frequency 51.3%, medianpeak viral load 4.52 log₁₀ DNA copies/ml), EBV (40.9%, 2.40 log₁₀), HHV-8 (11.3%, 2.72 log₁₀), HSV-1 or 2 (10.4%, 2.83 log₁₀), HHV-6 (7.0%, 2.82 log₁₀), and HHV-7 (14.8%, 3.42 log₁₀)(Table 2 and Fig. 1).Fifty of the subjects (43%) were found to shed more than one herpesvirus either simultaneously or over time.While 12 participants had continuous detectable herpesvirus DNA in repeated tested time-points over many months and sometimes years (median 56 days [IQR:28–228]), most of the subjects with longitudinal sampling demonstrated intermittent herpesirus shedding, i.e. had seminal samples with detectable and undetectable levels of the same virus over a few weeks or months. We were able to discriminate between HSV-1 and-2byallele-specific PCR for11outof 12 participants with detectable HSV in their semen, and found that five of these (45%) were HSV-1 and six were HSV-2.

Table 2.

Viral frequencies and viral levels in semen for each participant (n = 115).

Virus	Detectable frequency n (%)	Median Log10 viral load (IQR)
HIV-1	89 (77.4)	3.03 (2.41–3.85)
CMV	59 (51.3)	4.52 (3.48–5.23)
EBV	47 (40.9)	2.40 (1.76–3.27)
HSV-1 or HSV-2	12 (10.4)	2.83 (2.37–3.30)
HHV-6	8 (7.0)	2.82 (2.04–3.63)
HHV-7	17 (14.8)	3.42 (2.87–3.85)
HHV-8	13 (11.3)	2.72 (2.50–3.25)

HIV-1: human immunodeficiency virus type 1; HSV-1 or HSV-2: herpes simplex virus type 1 or 2; CMV: cytomegalovirus; EBV: Epstein-Barr virus; HHV-6, -7 or -8; human herpesviruses type 6, 7 or 8; absolute number (N) and frequency (%) of subjects with at least one positive semen sample (N = 115); peak VL: log₁₀ viral load (median/IQR (interquartile range) among detectable samples)

Fig. 1. Viral frequencies in semen for each HIV-1 infected subject(n=115).

Frequency (%) of patients(n = 115) having at least one semen sample positive for any herpesvirusDNA and for each individual tested virus separately.

Differences in viral levels by HIV stage andCD4+ T-cell count

To clarify other factors that may be associated with viral shedding, we investigated if there was a difference in detection of herpesvirusesin semen collected before and after 90 days EDI and among those with lower and higher CD4+T-cell counts (≥500 vs. <500 and ≥350 vs. <350 cells/ml). Overall, there wasno clear relationship between EDI andseminal herpesvirus shedding, but there was a trend towardsincreased sheddingfor HHV-8 (8.2% vs. 1.5% p = 0.07) andEBV (38.5% vs. 25.8% p = 0.07) in later collected samples. There was a propensity for participants with detectable seminal HIV-1 levels to be more frequently sampled within the first 90 days of their infection (p = 0.12), probably as a consequence of higher VLsobserved in the early-stage group (p<0.01). While HIV-1 RNAseminal shedding was associated with lower CD4+ counts (<500 cells/ml; p=0.04), HSVseminal shedding was associated with higher CD4+ counts(>500 cells/ml; p=0.01). At a CD4+ T-cell threshold of 350 cells/ml, only CMV showed differential shedding with CMV genital shedding being twice as frequentat lower counts (50% shedding ifCD4+ cell < 350 cells/mlversus 24.1% if CD4+ cell>350 cells/ml). No participants had a CD4+count <200.

Predictors of HIV-1 RNA shedding in semen

Multiple factors likely influence seminal shedding of HIV-1 RNA. In this study we evaluated the association of seminal HIV-1 shedding to herpesviruses DNA, blood HIV-1 RNA levels, CD4+ T-cell count, presence of transmitted HIV-1 DRM, subjects’age and stage of infection(Table 3). In univariate analysis, detectable seminal HIV-1 RNA was marginally associated withhigher seminal levels of CMV DNA (p = 0.06),higher HIV-1 RNA levels in blood plasma (p < 0.01), and lower numbers of CD4+ T-cells in blood (p < 0.05).There was also a weak trend towards increased HIV-1 shedding for samples withEBV at levels >2.7 log₁₀copies/ ml(p = 0.17).In univariate analysis, there were no associations between presence ofHIV-1 seminal shedding and presence of transmitted DRM,subject’s age, and presence of other herpesviruses (i.e. HSV HHV-6, −7 and −8). In multivariate analysis, blood HIV-1 RNA levels (p<0.01), seminal CMV and HHV-8 levels (p < 0.05) were independent predictors of HIV-1 RNA seminal shedding. Interestingly, only high levels of CMV DNA in semen (>4.4 log₁₀ copies/ml) were associated with detectable HIV-1 RNA in semen. In regard to HHV-8 co-infection, 87% of the semen samples with detectable HHV-8 DNA also had HIV-1RNA shedding compared to 67% HIV-1 shedding in samples when HHV-8 was not present (Fisher’s Exact p = 0.16). When considering only subjects with HHV-8 positive serology, this difference was even greater 87% versus 56% (p = 0.04). Similarly, in our multivariable GEE model using the same covariates displayed in Table 3 but including only subjects with positive HHV-8 serology (n = 56) the association of HHV-8 shedding with HIV-1 shedding in semen remained statistically significant (p = 0.02).

Table 3.

Associations of herpesvirus and HIV shedding in semen.

Factor		HIV detected in semen (N = 162)	HIV not detected in semen (N = 74)	Univariate p-value	Multivariate p-value
Recent HIV-infection (≤90 days from EDI) (N = 236)		49 (30.3%)	17 (23.0%)	0.12	0.55
Median age of subject (IQR) (N = 236)		33 (25–39)	32 (26–37)	0.51	–
Median Log10 HIV RNA copies in blood (IQR) (N = 230)		4.87 (4.28–5.19)	4.01 (3.45–4.86)	<0.01	<0.01
MedianCD4+T cell number (IQR) (N = 236)		525 (412–680)	666(472–883)	0.03	0.36
Drug resistance mutation		22 (13.6)	12 (17.4)	0.39	–
Log10 CMV DNA shedding in semen (N = 236)	nd	81 (50.0%)	47 (63.5%)	Reference	Reference
	0.1–4.4	33(20.4%)	19 (25.7%)	0.86	0.98
	>4.4	48(29.6%)	8 (10.8%)	0.06	0.03
Log10 EBV DNA shedding in semen (N = 235)	nd	97 (59.9%)	56 (76.7%)	Reference	Reference
	0.1–2.7	37 (22.8%)	14 (19.2%)	0.76	0.68
	>2.7	28 (17.3%)	3(4.1%)	0.17	0.43
HSV-1 or HVS-2 shedding (N = 236)		17 (10.5%)	3 (4.1%)	0.31	0.19
HHV-6 shedding (N = 232)		7 (4.4%)	2 (2.7%)	0.76	–
HHV-7 shedding (N = 230)		22 (13.9%)	8 (11.1%)	0.95	–
HHV-8 shedding (N = 236)		13 (8.0%)	2 (2.7%)	0.24	0.03

HIV-1: human immunodeficiency virus type 1; HSV-1 or HSV-2: herpes simplex virus type 1 or 2; CMV; cytomegalovirus; EBV: Epstein-Barr virus; HHV-6, -7 or -8; human herpeses virus type 6, 7 ot 8; nd: not detectable;Univariate and multivariate p values were determined in a generalized estimating equation (GEE) regression model adjusting for repeated measures by subject. EBV and CVM were ordinalized using the respective mean values as threshold between low-level and high-level shedders. Final multivariate model includes all covariates with univariate p<0.35; IQR: inter quartile range; SD: standard deviation; samples with HIV-1 < 50 copy number/ml were considered undetectable. P < 0.05 in bold.

As anadditional endpoint, we investigated the relationships between detectable CMV or EBV DNA in semen (the most frequently detected viruses) and levels of seminal HIV-1 RNA. Seminal shedding of both herpesviruses was associated with higher levels of HIV-1 RNA within the same compartment (median 2.64 versus 2.18 log₁₀ copies/ml, p= 0.05 for CMV, and 2.79 versus 2.15 log₁₀ copies/ml, p < 0.01 for EBV, Fig. 2a and 2b). All of these observations remained true after excluding the 16 patients with concomitant bacterial STI.

Fig. 2. Comparison of HIV-1 RNA levels in semen samples with or without detectable CMV and EBV DNA and presence of transmitted drug resistance mutations.

(a) Seminal HIV-1 RNA levels compared between samples with and without detectable CMV DNA (Mann Whitney, p = 0.05), boxplots display whiskers with min and max values. (b) Seminal HIV-1 RNA levels compared between sample with and without detectable EBV DNA (Mann Whitney, p < 0.01), boxplots display whiskers with min and max values. (c) Seminal HIV-1 RNA levels compared between samples with and without transmitted drug resistance mutations (DRM) (Mann Whitney, p = 0.02), boxplots display whiskers with min and max values

Impact of transmitted HIV-1 DRM on seminal HIV-1 levels

Since transmitted DRM in the HIV-1 polgene can influence HIV-1 levels [33,34], we also evaluated the presence of DRM in our antiretroviral-naive cohort. Nineteen participants were found to have relevant DRM in their baseline HIV-1 pol sequences (non-nucleoside reverse transcriptase inhibitor (NNRTI): K101P (1), K103N (11), Y181C (1), P225H (1); nucleoside reverse transcriptase inhibitor (NRTI): M41L (5), D67N/G (4), T69-insert (1), K70R (1), L74V (1), M184V (2), L210W (1), T215S/Y (7), K219E/Q (3); protease inhibitor (PI): M46I (1), I54V (2), V82A (1), I84V (1), L90 M (2)).Overall, the presence of transmitted DRM was associated with a lower VL in blood (4.31 versus 4.62 log₁₀ copies/ml, p=0.05), and with a more pronounced difference in semen (median of 2.09 versus 2.49 log₁₀ copies/ml, p = 0.02, Fig. 2c).

Associations between seminal shedding of herpesviruses and absolute numbers and activation levels of lymphocytes in semen and blood

A recent paper from our group [13] reported that detectable CMV levels in semen were positively correlated with the absolute numbers of seminal T-cells and with their activation status in semen and blood. On the same subset of participants, we now investigated the associations between shedding of other herpesviruses with the numbers and immune activation status (RA-CD38+) of CD4+ and CD8+ T-cells. Similar to what we previously found with CMV, the presence of seminal EBV replication was associated with higher absolute numbers of CD4+ (p<0.01) and CD8+ (p<0.05) Tcells in semen. None of the herpesviruses detected in the genital tract (other than CMV) significantly impacted the observed immune activation in semen or blood.

Discussion

Understanding the viral and immunologic dynamics in the seminal compartment and characterizing the contribution of co-factors that increase the risk of sexual transmission could provide important information for the development of effective prevention strategies.Since HIV-1 in male genital secretions account for more than half of all HIV-transmissions among MSM in particular[35], we investigated a cohort of HIV-infected MSM for interactions between seminal replication of sevenherpesviruses and HIV-1shedding.

Despite relatively preserved immunity and early stage disease, a high proportion of participants (75.7%) had at least one genital sample with shedding of herpesviruses. Similar to previous reports [17,36], CMVand EBV were frequently detected in about half of all analyzed seminal samples, while HSVand HHV-6, −7, and −8 were found at lower frequencies, between 7–14.8%. Interestingly 45% of HSV-DNA found in the semen was HSV-1 rather than HSV-2, suggesting that future studies involving genital HSV replication should discern each type.

While subjects with detectable HIV-1 levels in semen were more often identified earlier during HIV-infection, there were trends towards increased shedding>90 days EDI for HHV-8 (p=0.07) and for EBV (p=0.07), while there was no difference for the other herpesviruses. This suggests that individuals within the first 90 days of infectionare more likely to shed seminal HIV-1 independentlyof herpesviruses co-infections,probablyas a consequence of higher blood HIV-1 VL. On the other hand, seminal shedding of herpesviruses may have a greater impact on HIV-1 shedding after acute HIV-infection has passed, i.e. >90 days EDI.Further, immune status, as measured by CD4+ T-cell count, also influenced HIV-1 and herpesvirus shedding. While HIV-1 (p+0.04) and CMV (p<0.01)seminal sheddingwere associated with lower CD4+T-cell counts, consistent with previous reports[16,22], there was strong evidence that HSVshedding occurred at CD4+counts >500 cells/ ml (P = 0.01). These results contrast with a recent report[37],documentingborderline increased HSV-shedding in subjects with CD4+<500 cells/ml (p=0.08). On the other hand, the subjects in our study had relatively intact immune systems and did not include severely immunocompromised persons in whom the association between herpesviruses and HIV-1 shedding may be stronger. Also, since our cohort included only asymptomatic HIV-infected individuals, it is possible that HSVshedders with low CD4+ cells and ulcerative genital disease were not included in this study, but this should be pursued in future investigations.

Similar to previous studies[4,38], the strongest predictorof HIV-1 seminal shedding in our cohortwasthe level of HIV-1 RNA in blood; however, high levels of CMV DNA (>4.4 log₁₀ copy/ml) in semen werealso independently associated with HIV-1shedding[16]. The lack of association between seminal CMV and HIV-1 shedding when CMVwas present at lower levels, could suggest that a certain threshold of CMV replication is needed to influence HIV-1 RNA shedding or that an unidentified factor influencing both HIV-1 and CMV shedding could be present, like localized immune activation[13,36,39].This study also found that presence of seminal EBV replication was strongly associated with higher levels of HIV-1 RNA in semen and also with higher CD4+ T-cells counts in semen(both p<0.01), suggesting that also EBVplays a role in HIV-1dynamics. Perhaps most interesting, this study identified that the presence of HHV-8 was also an independent predictor for HIV-1 RNA seminal shedding. Unlike the previous Indian study of heterosexual individuals [22], the ability to detect this effect is likely a consequence of the higher rate of HHV-8 seminal shedders in our cohort (11.3% vs. 6%)[22,36]. Seminal shedding of HHV-8 has been reported more frequently in MSM [40–42], but a limitation is that we did not evaluate oral shedding [43– 46], and this should be pursued given these results. With the historical connections between Kaposi sarcoma and the male-to-male sex for HIV-1 and HHV-8 infections, the associations between HIV-1 and HHV-8 genital tract shedding should be explored further, and might be especially important among African populations where HHV-8 and Kaposi sarcoma prevalence is higher [47– 49].

We observed significantly lower HIV-1 RNA levels in blood plasma of participants in which the HIV population had a DRM (p = 0.05), and this difference was even more pronounced in semen (p = 0.02). Previous reports have demonstrated the effect of DRM on blood plasma VL[33,34], but while intuitive, this is the first report of this effect of seminal VL. The fitness cost of DRM seems to have a higher impact on replication capacity of HIV-1 in the genital compartment compared to wildtype virus. Since all of our cohort participants were antiretroviral-lnaïve, all of the observed DRM were likely transmitted, so no conclusions can be made on how failing antiretroviral therapy selecting for DRM can influence seminal shedding of HIV-1.

This study has a number of limitations. Because this was aretrospective, observational study, we cannot establish a causal relationship between the described correlations. Second, screening for STI was performed only at baseline and an unrecognized incurred bacterial STI couldconfoundthe observations. Further, although this is the largest study of its kind, the sample size still limited power in discerning a significant effect for some viruses, e.g. HSVor HHV-6. In any case, the effect of these relatively rarer herpesviruses on HIV-1 seminal shedding and thus on the potential for HIV-1 transmission in MSM risk group is likely relatively small. Lastly, since this study included only MSM from a relatively small geographic area, it is necessary to determineif the results can be generalized for other risk groupsor populations.

To our knowledge, this is the largest study of this kind (n = 236 samples,115 different subjects) investigating the simultaneous interactions between HIV-1 and seven differentherpesviruses in a cohort of MSM followed since primary HIV-infection[50].Thislarge dataset permitted the description of the shedding of herpesviruses and HIV-1 in different subsets of subjects, and a multivariate analysis taking in account a number of potentially confounding variables(age, stage of infections, blood-HIV-1 VL, and CD4+ counts). This study is also unique in that it evaluated the possible influence of DRM, and the associations between herpesviruses and lymphocytes subsets and immune activation in semen in a subgroup of participants.Consistent with previous studies [16,22],blood HIV-1 levels, seminal CMVand EBVwere the most significant associations for HIV-1 sheddingin semen.Uniquely interesting, HHV-8 was associated with increased seminal HIV-1 shedding. Moreover, we described for the first time how presence of transmitted DRM impacts HIV-1 replication in the genital tract. Taken together, these results provide a clearer picture of how the highly prevalent herpesviruses influence the dynamics of HIV-1 in the male genital tract and how they likely influence the risk of sexual transmission of HIV-1.Future studies should determine if these CMV correlations hold true in subjects on ART, and if this is the case, adding CMV suppressive therapy to standard ART in high risk subjects or in discordant couples may be clinically relevant.