Cytomegalovirus Replication in Semen Is Associated with Higher Levels of Proviral HIV DNA and CD4⁺ T Cell Activation during Antiretroviral Treatment

Participants, samples, and clinical laboratory tests.

Paired semen and blood samples collected from asymptomatic, chronically HIV-infected, sexually active men who have sex with men (MSM) who were prospectively enrolled in the California Collaborative Treatment Group (CCTG) 592 trial were included in this study (25). CCTG 592 is a study of an Internet-based behavioral intervention with MSM that included baseline collection of paired blood and semen from a total of 180 HIV-infected MSM who were on or off ART. For this study, we included baseline semen samples from a subset of 53 CMV-seropositive subjects who were receiving effective ART with blood plasma HIV RNA at <200 copies/ml within 3 months before the seminal-sample collection and who had peripheral blood mononuclear cell (PBMC) samples stored for further analysis. The individuals started ART during chronic HIV infection, but no information is available about the durations of HIV infection. Fifty subjects had HIV RNA at <50 copies/ml, while three had HIV RNA levels between 50 and 200 copies/ml. All included participants were CMV seropositive (26). Semen was collected and processed as previously described (27, 28). Blood CD4⁺ T lymphocyte absolute counts were measured by flow cytometry, and HIV RNA levels in blood plasma were quantified by the Amplicor HIV Monitor Test (Roche Molecular Systems Inc.).

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; the Los Angeles Biomedical Research Institute at Harbor-UCLA Medical Center; and the University of Southern California. Written informed consent was provided by all study participants.

Quantification of HIV DNA, 2-LTR circular HIV DNA, and herpesvirus DNA in PBMCs and semen.

DNA was extracted from 5 million PBMCs and 200 μl seminal plasma for each participant (QIAamp DNA minikit, Qiagen, CA). For PBMCs, total HIV DNA (Pol) and the 2-long-terminal-repeat (2-LTR) junction were quantified by droplet digital PCR (ddPCR) from extracted DNA (29). Briefly, 1,000 ng of DNA per replicate was digested with BSAJ1 enzyme (New England BioLabs) prior to ddPCR. Total HIV DNA (Pol) and 2-LTR PCRs were performed as a duplex assay with VIC (Pol)- and 6-carboxyfluorescein (FAM) (2-LTR)-labeled probes, respectively, with the following cycling conditions: 10 min at 95°C, 40 cycles consisting of a 30-s denaturation at 94°C followed by a 60°C extension for 60 s, and a final 10 min at 98°C. Total CMV and Epstein-Barr virus (EBV) DNA PCRs were performed as a duplex with FAM (CMV)- and HEX (EBV)-labeled probes, respectively, with the following cycling conditions: 10 min at 95°C, 40 cycles consisting of a 30-s denaturation at 94°C followed by a 54°C extension for 60 s, and a final 10 min at 98°C. A 1:10 dilution of the digested DNA was used for host cell ribonuclease P/MRP 30-kDa subunit (RPP30) PCR (probe VIC) and cycled with the same parameters as the Pol–2-LTR duplex. Copy numbers were calculated as the mean of replicate PCR measurements and normalized to 1 million CD4⁺ T cells as determined by RPP30 (total cell count) and flow cytometry (percentage of CD4 T cells). The levels of 7 herpesviruses (CMV; EBV; herpes simplex virus 1 [HSV-1] and HSV-2; and human herpesvirus 6 [HHV-6], HHV-7, and HHV-8) were also measured by real-time PCR in the DNA extracted from seminal plasma, as described previously (21).

Quantification of cellular HIV RNA in PBMCs.

Cellular HIV RNA was measured for the subset of 42 individuals with enough stored PBMCs to perform these analyses and completely suppressed HIV RNA levels in blood plasma (<50 copies/ml). The extracted RNA (500 ng) was reverse transcribed into 20 μl cDNA (iScript Advanced cDNA synthesis kit for RT-qPCR; Bio-Rad) using the manufacturer's protocol. The cDNA product (8 μl; approximately 200 ng) was added to the ddPCR reaction mixture. Unspliced HIV RNA (usGag) and multiply spliced (msTat-Rev) PCRs were performed as a duplex with HEX (usGag)- and FAM (msTat-Rev)-labeled probes, respectively, using primers and probes as previously described (30, 31). The cycling conditions for this PCR were as follows: 10 min at 95°C, 60 cycles consisting of a 30-s denaturation at 94°C followed by a 60°C extension for 60 s, and a final 10 min at 98°C. Copy numbers were calculated as the mean of replicate PCR measurements and normalized to total RNA as determined by the A₂₆₀/A₂₈₀ absorption ratio using a NanoDrop 2000 spectrophotometer (Thermo Scientific).

Anti-CMV IgG antibody levels.

As previously described (26), anti-CMV IgG antibody levels were measured in blood plasma. The number of units per ml were determined by interpolation from a standard curve of a known anti-CMV IgG solution (26, 32).

Multiparameter flow cytometry analysis.

For a subset of 49 patients, analysis of cellular activation and proliferation was performed by multicolor flow cytometry. Aliquots of 5 million PBMCs were quickly thawed at 37°C, resuspended in RPMI supplemented with 20% fetal bovine serum (FBS), incubated at room temperature for 20 min, centrifuged, and resuspended in 500 μl phosphate-buffered saline (PBS) staining buffer. Cell viability was assessed using the LIVE/DEAD Fixable Aqua Dead Cell Stain Kit (Life Technologies). Four subjects (out of the total 53 included subjects) were excluded because of <85% cell viability. Approximately 300,000 to 500,000 PBMCs per tube were stained for cell surface markers prior to fixation and, when required, followed with permeabilization for intracellular-protein detection (FOXP3/Transcription Factor Staining Buffer set; eBioscience). The stained cells were acquired on a FACSCanto (BD Biosciences). We used the following antibody combinations to evaluate immune activation and proliferation in T cells: (tube 1) HLA-DR–fluorescein isothiocyanate (FITC), CD45RA-phycoerythrin (PE), CD4–peridinin chlorophyll protein (PerCP)-Cy5.5, CD38–PE-Cy7, CD27-allophycocyanin (APC), CD3–APC-Cy7, and CD8-Pacific Blue; (tube 2) Ki67-FITC, CD45RA-PE, CD4–PerCP-Cy5.5, CCR7–PE-Cy7, CD27-APC, CD3–APC-Cy7, and CD8-Pacific Blue (BD Biosciences). The antibody combinations in tube 1 were used to assess naive (CD45RA⁺ CD27⁺), central memory (CD45RA⁻ CD27⁺), effector memory (CD45RA⁻ CD27⁻), and terminally differentiated (CD45RA⁺ CD27⁻) CD4 and CD8 T cell subsets. With tube 2, T cell subsets were defined as naive (CD45RA⁺ CD27⁺ CCR7⁺), central memory (CD45RA⁻ CD27⁺ CCR7⁺), transitional memory (CD45RA⁻ CD27⁺ CCR7⁻), and effector memory (CD45RA^+/− CD27⁻ CCR7⁻) in both the CD4⁺ and CD8⁺ subsets. Immune activation was defined by CD38⁺ HLA-DR⁺ expression and proliferation and by Ki67⁺ in total CD4⁺ and CD8⁺ T cell populations and subsets. To normalize the levels of HIV DNA and 2-LTR circles, the percentage of CD4⁺ T cells within total PBMCs was determined for each sample. All data analyses were performed using Flow Jo software (version 9.6.2).

Statistics.

Statistical analyses were performed using SAS (version 9.2). Each continuous variable was assessed for normal distribution. Viral-load variables were transformed to log₁₀ values, and CMV shedding and the presence of the LTR were dichotomized (undetectable/detectable). Comparisons were performed using the Fisher exact text (for sparse data), t test (for continuous, normally distributed variables), or Mann-Whitney U test (for continuous, non-normally distributed variables). Correlation analyses were performed using parametric (Pearson) or nonparametric (Spearman) correlation coefficients. A multivariate linear regression for the association of detectable seminal CMV DNA with HIV DNA in blood was performed, adjusting for possible confounders that were found to be significant in the univariate analysis (P < 0.05). Analyses were performed on the complete data set of 53 subjects (including three subjects experiencing an HIV RNA blip of <200 copies/ml in the 3 months before sample collection) and on the reduced data set of 50 subjects with fully suppressed HIV RNA in blood plasma (<50 copies/ml). To estimate the average viral transcriptional activity per single HIV DNA copy (Pol), we calculated the ratio between cellular HIV RNA and total HIV DNA for each sample, as described previously (31, 33). Because of the exploratory character of the study, we decided not to correct our results for false detection (multiple comparisons).

Participants, samples, and clinical laboratory tests.

The study participants (n = 53) were HIV-infected men with a median age of 46 (95% interquartile range [IQR], 38 to 51) years who were receiving ART and had ≤200 copies of HIV RNA/ml of blood plasma. Fifty subjects had <50 copies of HIV RNA/ml, while three had levels between 50 and 200 copies of HIV RNA/ml. The median time on ART at the time of sample collection was 3.8 (IQR, 1.8 to 5.3) years, and the median CD4 T cell count was 625 (IQR, 535 to 739) cells/μl. Eighty-two percent were on a regimen including tenofovir, 41% were on a regimen including a nonnucleoside reverse transcriptase inhibitor (NNRTI), 61% were on a regimen including a protease inhibitor (PI), and 16% were on a regimen also including an integrase inhibitor. Self-reported levels of ART adherence during the preceding month were >90% of daily doses for 90.4% of the subjects. The subjects' characteristics are summarized in Table 1.

Herpesviruses and viral reservoir characteristics.

Five participants (9.4%) were shedding HIV RNA, and 31 (58.5%) were shedding at least one herpesvirus in seminal plasma at the time of collection. Specifically, 23 (43.4%) were shedding CMV, 13 (24.5%) EBV, 2 (3.8%) HSV-2, 3 (5.7%) HHV-6, 4 (7.6%) HHV-7, and 2 (3.8%) HHV-8 (Table 1; see Fig. S1 in the supplemental material). None of the subjects were shedding HSV-1. Eleven subjects (21%) had detectable levels of CMV DNA in PBMCs (median, 4.2 copies/million cells among detectable samples; IQR, 2.6 to 5.8 copies), and 50 (94%) had detectable levels of EBV in PBMCs (median, 75 copies/million cells; IQR, 18.5 to 389 copies) (see Fig. S1 in the supplemental material).

Total levels of HIV DNA (Pol) as measured by ddPCR were detectable in 92% of PBMC samples (n = 49), with a median level of 2.54 log₁₀ copies per million CD4⁺ T cells (IQR, 2.0 to 3.1 copies), while 2-LTR circles were detectable in 43% of all PBMC samples (n = 23), with an average level of 1.42 log₁₀ copies per million CD4⁺ T cells among detectable samples (IQR, 1.19 to 1.89 copies). In the subset of 42 patients with completely undetectable HIV RNA levels and sufficient samples available, levels of unspliced HIV RNA were detectable in 95% of the samples (n = 40), while levels of multiply spliced HIV RNA encoding Tat and Rev were detectable in 64% of the samples (n = 27).

Associations between CMV replication, HIV DNA levels, immunological markers, and cellular HIV transcription. (i) Predictors of higher HIV DNA levels (n = 53).

Univariate analysis revealed significantly higher levels of HIV DNA in the CD4⁺ T cells of participants with detectable seminal CMV than in those without detectable CMV in semen (mean, 2.84 versus 2.11 log₁₀ copies per million CD4⁺ T cells; P < 0.05) (Fig. 1A). There was also a trend for increased levels of CMV IgG to be associated with higher levels of total log₁₀ HIV DNA in CD4⁺ T cells (P = 0.08). As might be expected, the presence of detectable 2-LTR circles was associated with higher levels of total HIV DNA (P < 0.01), and we also found a significant negative correlation between the nadir CD4⁺ T cell count and HIV DNA Pol copies per CD4⁺ T cells (r = −0.29; P = 0.03), as previously described (34). The presence of seminal shedding of HIV and herpesviruses other than CMV, the CD4⁺ T cell count, age, time on ART, and self-reported adherence to ART were not associated with levels of total HIV DNA (Table 2).

Multivariate analysis indicated that detectable seminal CMV was independently associated with higher levels of HIV DNA in CD4⁺ T cells (P = 0.04) after adjusting for the presence of 2-LTR circles and nadir CD4⁺ T cells. Of note, CMV IgG presented a trend toward an association with HIV DNA levels in univariate analysis (P = 0.08), but it became nonsignificant when added to the multivariate model (P = 0.69). Even after restricting the analysis to the 50 subjects with HIV RNA in blood plasma at <50 copies/ml, seminal CMV shedding remained independently associated with higher levels of HIV DNA (P = 0.04) (Table 2). Similarly, there was a significant but weak correlation between seminal CMV levels and CD4-associated HIV DNA levels in blood (P = 0.04; Pearson, r = 0.26), which remained significant after adjusting for the presence of 2-LTR circles and nadir CD4⁺ in a multivariate model. As previously reported (35), subjects with detectable CMV DNA showed a trend toward lower CD4⁺ T cell counts than non-CMV shedders (P = 0.039), but no difference was found for the nadir CD4⁺ T cell count (P = 0.24). The presence of CMV DNA in PBMCs was not associated with significantly higher HIV DNA levels (2.53 versus 2.26 log₁₀; P = 0.43) or with higher CD4⁺ T cell immune activation or proliferation. Similarly, the amount of EBV DNA in PBMCs did not correlate with HIV DNA levels (P = 0.26).

(ii) Immunological markers (n = 49).

To investigate whether increased immune activation and proliferation of T cells could be a possible mechanism connecting CMV replication with higher levels of proviral HIV DNA, we grouped patients according to whether CMV was detectable in seminal plasma (Table 3). We observed that subjects with detectable CMV in semen (n = 21) had lower total CD4 T cell counts than those with undetectable seminal CMV (n = 28; P = 0.039). Moreover, the CD4 T cell compositions were different in the two groups: subjects with detectable CMV in semen had a significantly higher frequency of transitional memory CD4⁺ (median, 13.1% versus 10.23%; Mann-Whitney, P = 0.04) and effector memory CD4⁺ (12.6% versus 7.3%; P = 0.05) T cells in blood than those with undetectable seminal CMV. Subjects with detectable seminal CMV DNA also had higher levels of proliferating (Ki67⁺) total CD4⁺ T cells (P = 0.01) (Fig. 1B) and effector memory CD4⁺ T cells (P = 0.09) (Table 4). There was also a trend toward higher levels of proliferating CD8⁺ T cells (P = 0.1) (Fig. 1C) in those with detectable seminal CMV than in those whose levels were undetectable. Among subjects with CMV DNA in semen, there was also a significant correlation between HIV DNA levels and frequency of activated (P < 0.01; Spearman, r = 0.6) (Fig. 2A) and proliferating (P < 0.01; r = 0.64) (Fig. 2B) CD4⁺ T cells. However, we did not find similar correlations between the frequency of proliferating (r = −0.29; P = 0.15) and activated (r = −0.04; P = 0.85) CD4⁺ T cells and HIV DNA among CMV nonshedders. The trend, if any, was a negative correlation, opposite to what we observed for CMV shedders. After including the entire population (CMV shedders and nonshedders), there was no association between CD4⁺ proliferation and HIV DNA (r = 0.10; P = 0.51) and a diminished association between CD4⁺ activation and HIV DNA (r = 0.35; P = 0.02). Subjects with detectable seminal CMV DNA also had significantly increased levels of immune activation (HLA-DR⁺ CD38⁺) of total CD4⁺ T cells (P < 0.01) (Fig. 1D) and effector memory CD4⁺ T cells (P = 0.02) (Table 4). No difference was observed in CD8⁺ immune activation of T cells between the two groups. All the described associations remained significant after excluding the three subjects with HIV RNA levels in blood plasma at 50 to 200 copies/ml.

(iii) Cellular HIV transcription (n = 42).

For a subset of 42 subjects with available PBMC aliquots and completely undetectable HIV RNA in blood plasma, we investigated if the presence of seminal CMV replication was associated with the frequency and levels of 2-LTR circles, cell-associated HIV RNA (unspliced and multiply spliced, encoding Tat-Rev), and increased average transcriptional activity (estimated as the cellular HIV RNA/HIV DNA ratio). Subjects with detectable CMV DNA in semen (n = 15) had higher levels of multiply spliced cell-associated HIV RNA (P = 0.017) and a trend toward increased average transcriptional activity (P = 0.068) compared to those with undetectable seminal CMV (n = 27). No association was observed between CMV shedding and unspliced HIV RNA (frequency or levels) or 2-LTR circles (Table 5).