Hormonal Contraception and HIV-1 Infection: Medroxyprogesterone Acetate Suppresses Innate and Adaptive Immune Mechanisms

Abstract

Recent observational studies indicate an association between the use of hormonal contraceptives and acquisition and transmission of HIV-1. The biological and immunological mechanisms underlying the observed association are unknown. Depot medroxyprogesterone acetate (DMPA) is a progestin-only injectable contraceptive that is commonly used in regions with high HIV-1 prevalence. Here we show that medroxyprogesterone acetate (MPA) suppresses the production of key regulators of cellular and humoral immunity involved in orchestrating the immune response to invading pathogens. MPA inhibited the production of interferon (IFN)-γ, IL-2, IL-4, IL-6, IL-12, TNFα, macrophage inflammatory protein-1α (MIP-1α), and other cytokines and chemokines by peripheral blood cells and activated T cells and reduced the production of IFNα and TNFα by plasmacytoid dendritic cells in response to Toll-like receptor-7, -8, and -9 ligands. Women using DMPA displayed lower levels of IFNα in plasma and genital secretions compared with controls with no hormonal contraception. In addition, MPA prevented the down-regulation of HIV-1 coreceptors CXCR4 and CCR5 on the surface of T cells after activation and increased HIV-1 replication in activated peripheral blood mononuclear cell cultures. The presented results suggest that MPA suppresses both innate and adaptive arms of the immune system resulting in a reduction of host resistance to invading pathogens.

Safe and effective methods of contraception represent a critical component of preventive health care. Contraception provides women with a control over their reproductive health, reduces the number of unwanted pregnancies, decreases maternal and infant mortality and morbidity, and lowers the risk of mother-to-child transmission of HIV-1 (1). Injectable hormonal contraception is effective, affordable, and increasingly popular especially in countries with limited resources. Depot medroxyprogesterone acetate (DMPA) (Depo-Provera), a progestin-only–based contraceptive typically administered as a 3-monthly intramuscular (I.M.) injection, is one of the most commonly used contraceptives in sub-Saharan Africa and other areas with high HIV-1 prevalence. It is estimated that over 14 million women worldwide use DMPA; in some countries, DMPA is the method of choice for over 50% of women using modern methods of contraception (2, 3). Unfortunately, multiple observational studies suggest an association between the use of hormonal contraception and increased risk of HIV-1 acquisition and transmission (4–11). In most studies, the adjusted hazard ratio of HIV-1 acquisition associated with the use of injectable contraception, in particular DMPA, was higher than that associated with oral contraception (5–10, 12). Hormonal contraception was also shown to be associated with increased rate of HIV-1 replication (13, 14), accelerated disease progression (15), increased cervicovaginal shedding (16, 17), and elevated risk of acquisition of cervical candidiasis, chlamydial, gonococcal, and Mycoplasma genitalium infections (4, 6, 18–21). Studies using nonhuman primate models of infection demonstrated that administration of DMPA enhanced the risk of simian immunodeficiency virus (SIV) acquisition via vaginal exposure, increased viral levels in the acute phase of infection, and reduced the protective effect of prior immunization (22–25). However, the results of observational studies in humans have been inconsistent with some studies reporting no effect of hormonal contraception on HIV-1 acquisition or disease progression (26–30) or finding increased risk only in subgroups of subjects differing in age and herpes simplex virus 2 status (31–33). It was argued that the elevated risk of HIV-1 acquisition observed in hormonal contraceptive users may be due to confounding behavioral factors that are difficult to be accounted for due to methodological limitations (34, 35). It is therefore critical to establish whether the observed associations reflect a direct biological effect of hormonal contraceptives on immune system and determine the mechanisms exerted by specific hormones and dosages.

The potential biological mechanisms of the effect of MPA on viral and bacterial infections are unknown. In addition to binding to the progesterone (P4) receptor (PR), MPA binds with high affinity and activates the glucocorticoid receptor (GR) expressed at high levels by multiple immune cell types (36–40). The affinity of MPA for GR is significantly higher than that of its endogenous ligand cortisol (38). The MPA-GR complex suppresses the transcription of GR-regulated genes, including IL-2, IL-6, and IL-8, via transrepression (36, 42, 43). Due to its glucocorticoid activity, it is feasible that MPA exhibits significant immune regulatory effects that may impact the host's susceptibility to pathogens. Recently, it was reported that the use of DMPA is associated with decreased systemic responses to Mycobacterium bovis Bacillus Calmette–Guérin in household contacts of active tuberculosis patients (44).

In this study, we characterized the effect of MPA on innate and adaptive immune mechanisms and compared it with the effect of the natural hormones P4 and estrogen and a commonly used synthetic glucocorticoid, dexamethasone. We show that MPA inhibits the secretion of key regulators of cellular and humoral immunity by activated T cells and plasmacytoid dendritic cells (pDCs). Furthermore, MPA prevents the down-regulation of HIV coreceptors CXCR4 and CCR5 on activated T cells and increases the rate of HIV-1 replication in peripheral blood cell culture.

Materials and Methods

Study participants and sample collection

All procedures involving the use of human subjects were approved by the Institutional Review Board of the University of Alabama at Birmingham. Informed consent was obtained from all participants. Acid citrate dextrose-treated blood (100 ml) was collected from healthy premenopausal women by personnel trained in phlebotomy. Women volunteers were of a good general health, age 19–40 years, mixed race, not pregnant, not using steroids, and with no signs of vaginitis, sores, or ulcers at the time of collection. History of vaginal infections, medication, and menstrual cycle information was collected. Women volunteers using hormonal contraception were using either DMPA delivered by 3-monthly im injections (150 mg; Depo-Provera) for a median of 7 months (range, 1–96 months) or several different forms of combined oral contraception for a median of 3.5 months (range, 1-17 months). Cervicovaginal lavage (CVL) was obtained from women volunteers, by vigorous flushing of the cervix and vagina with 5 ml sterile saline; the sample was mixed with protease inhibitors (5 μg/ml aprotinin, 1 μg/ml leupeptin, 1 μg/ml antipain, 1 μg/ml pepstatin, 200 μg/ml sodium azide, and 1mM phenylmethylsulfonyl fluoride; all from Sigma, St Louis, Missouri) and subsequently cleared by centrifugation at 14,000g for 5 minutes before storage at −80°C. Vaginal tissue was obtained from postmenopausal women undergoing pelvic reconstructive surgery as anonymous remnant surgical material.

Materials

All cell culture reagents were obtained from Mediatech Inc (Manassas, Virginia), unless indicated otherwise. Chemicals, hormones, and enzymes were purchased from Sigma (St Louis, Missouri). Antibodies, beads, and columns for cell purification were obtained from Miltenyi Biotec (Auburn, California). Antibodies for flow cytometry were purchased from eBioscience (San Diego, California), unless listed otherwise. 17ß-Estradiol (E2), P4, MPA, dexamethasone (DEX), and mifepristone (RU486) (all from Sigma) were dissolved in 200-proof ethanol at a concentration of 10mM and stored at −80°C before use.

Mice

Female C57BL/6 (B6) mice and ovalbumin (OVA)-specific T cell receptor (TCR)-transgenic OT2 mice on a B6 background were obtained from The Jackson Laboratory (Bar Harbor, Maine). All experimental procedures involving animals were approved by the University of Alabama Institutional Animal Care and Use Committee.

Isolation of peripheral blood mononuclear cells, T cells, and monocytes

Peripheral blood mononuclear cells (PBMCs) were isolated by density gradient centrifugation using lymphocyte separation medium (MP Biomedicals, Solon, Ohio). T cells were purified from PBMCs by positive selection for CD3⁺ cells on a MACS column according to manufacturer's directions (Miltenyi). Macrophages were purified by positive selection for CD14⁺ cells from the CD3-depleted population. PBMCs and macrophages were cultured in phenol red-free RPMI containing 10% heat-inactivated charcoal dextran-scrubbed fetal bovine serum (FBS) (Atlanta Biologicals, Lawrenceville, Georgia), 2mM GlutaMAX (Invitrogen, Carlsbad California), 100 IU/ml penicillin, and 100 μg/ml streptomycin. Purified T cells were cultured in identical media except that regular FBS (Hyclone, Logan, Utah) was used.

Isolation of human vaginal mucosal mononuclear cells

Vaginal mucosal mononuclear cells (VMMCs) were isolated from remnant vaginal tissue from volunteers undergoing pelvic reconstructive surgery. In brief, vaginal tissue was collected immediately after the surgery in Hank's Balanced Salt Solution supplemented with 25mM HEPES, 2mM GlutaMAX, 2mM pyruvate, 200 IU/ml penicillin, 200 μg/ml streptomycin, 62.5 pg/ml Fungizone, and 50 μg/ml gentamicin (solution 1). The mucosa was mechanically removed from the vaginal tissue, dissected into 3 × 3 mm blocks and incubated in solution 1 supplemented with 1mM EDTA and 0.2 mg/ml dithiothreitol for 20 minutes with gentle agitation at 37°C to remove mucus. After two consecutive 30-minute incubations in solution 1 with 1mM EDTA, the tissue was digested with 0.6 mg/ml collagenase (type IV from C. histolyticum; Sigma) and 0.1 mg/ml deoxyribonuclease in phenol red-free RPMI supplemented with 25mM HEPES, 2mM GlutaMAX, 2mM pyruvate, 200 IU/ml penicillin, 200 μg/ml streptomycin, 62.5 pg/ml Fungizone, and 50 μg/ml gentamicin for 30 minutes with agitation at 37°C. Media were collected, and the tissue was minced and incubated for an additional 45 minutes in collagenase/deoxyribonuclease-containing media. Mononuclear cells were purified by density gradient centrifugation. The yield of VMMCs obtained from remnant vaginal tissue typically ranged from 4 to 15 × 10⁶.

Cytokine production in vitro

PBMCs, VMMCs, or purified CD3⁺ T cells at 5 × 10⁵ cells/ml or purified CD14⁺ monocytes at 2.5 × 10⁵ cells/ml were incubated for 24 hours in 200 μl medium containing hormones. Medium containing the final percentage of ethanol of the highest hormone concentration (0.1%) was used as vehicle control. PBMCs, VMMCs, and purified T cells were activated with the T cell activation/expansion kit MACSi beads coated with antibodies against CD2/CD3/CD28 (Miltenyi) at a bead to cell ratio of 1:2 (PBMCs and VMMCs) or 1:1 (T cells). Blood-derived monocytes were stimulated with 100 ng/ml lipopolysaccharide (LPS) (Escherichia coli 0111:B4; Sigma). Cells were activated for 24 hours, cell-free media was collected and filtered through 0.45-μm polyvinylidene difluoride filter plates (Millipore, Billerica, Massachusetts). Concentrations of interferon (IFN)-γ and TNFα were determined with Ready-Set-Go ELISA kits (eBioscience). Concentrations of 39 cytokines and chemokines were determined using the 39-plex MILLIPLEX human cytokine/chemokine panel kit (Millipore), and samples were analyzed undiluted on a Bioplex 100 system with Bioplex Manager Software version 5.0 (Bio-Rad, Hercules, California). GRO stands for CXCL1, -2, and -3; eotaxin stands for CCL-11. Heat maps were created using matrix2png (http://www.bioinformatics.ubc.ca/matrix2png/).

Antigen presentation and T cell proliferation assay

Bone marrow-derived dendritic cells (DCs) were prepared from bone marrow cells harvested from B6 mice by 8 days incubation in the presence of 50 ng/ml granulocyte-macrophage colony-stimulating factor (Peprotech, Rocky Hill, New Jersey) in phenol red-free RPMI containing 10% FBS, 2mM GlutaMAX, 100 IU/ml penicillin, and 100 μg/ml streptomycin as described (45). Purified DCs were incubated with OVA protein (100 μg/ml) in the absence or presence of hormones for 24 hours followed by irradiation (3000 rads). Cells were washed with media without antigen or hormones before co-incubation with T cells. OVA-specific TCR-transgenic T lymphocytes were purified from spleens and lymph nodes of OT-2 mice using the Pan-T cell kit (Miltenyi). OT-2 T cells (5 × 10⁴) were co-incubated with sequential dilutions of DCs in the presence of 100 μg/ml OVA and indicated hormonal regulators for 96 hours; T cell proliferation was determined by [³H]thymidine incorporation during the last 24 hours of culture.

pDC activation and intracellular cytokine staining assay

PBMCs (1.5 × 10⁶ cells/ml) were preincubated for 6 hours in the presence or absence of hormones before the stimulation with respective Toll-like receptor (TLR) ligands (5 μg/ml TLR7/8 ligand R848 [Invivogen, San Diego, California] or 2μM TLR9 ligand CpG ODN2216 [Hycult Biotech, Uden, The Netherlands]). Immediately after the addition of the stimulant, 1 μl of GolgiPlug (BD Biosciences, San Diego, California) was added. After 20 hours, cells were collected and stained for pDC markers (CD123-PE-Cy7 [eBioscience] and CD303-APC [BCDA-2; Miltenyi]). Cells were permeabilized using the Cytofix/Cytoperm kit (BD Biosciences) and stained intracellularly with IFNα-PE (BD Biosciences) and TNFα-fluorescein isothiocyanate monoclonal antibodies. Samples were analyzed on the LSR-II flow cytometer (BD Biosciences), and data analysis was performed using the FACSDiva (BD Biosciences) and Flowjo (Treestar, Inc, Ashland, Oregon) software.

Analysis of coreceptor expression on T cells

PBMCs (1.25 × 10⁶ cells/ml) were incubated for 24 hours in media in the absence or presence of hormonal regulators at the indicated concentrations. Cells were activated with MACSi beads at a 1:2 bead to cell ratio for an additional 24 hours. Nonadherent cells were collected and stained with anti-CD3-eFluor450, CD4-APC-eFluor780, CD8-PerCP-CY5.5, CD27-PE-Cy7, CD45RO-PE (eBioscience), CCR5-AlexaFluor 647, and CXCR4-APC antibodies (Biolegend, San Diego, California). Appropriate isotype controls were used to determine the percentages of cells expressing the respective HIV-1 coreceptors.

Virus infection in vitro

The construction of the Renilla luciferase-expressing, replication-competent HIV-1 CXCR4 tropic virus NL-LucR-T2A and CCR5 tropic virus NL-LucR-T2A–Bal.ecto was described previously (46). Viruses were kindly provided by the University of Alabama Center for AIDS Research virology core facility directed by Dr. J. Kappes. Target cells were cultured in media supplemented with 20 U/ml IL-2 (National Institutes of Health AIDS Research and Reference Reagent Program; Germantown, Maryland). 5 × 10⁵ cells/ml of CD8-depleted PBMCs or purified CD4⁺ T cells (depleted of CD14⁺ cells using CD14-specific beads; Miltenyi) were incubated for 1 or 5 days in media containing hormones. Cells were activated with either MACSi beads at a bead to cell ratio of 1:2 for CD8-depleted PBMCs and 1:1 for purified CD4⁺ T cells or with 2 μg/ml phytohaemagglutinin-L (PHA-L). Virus was added 24 hours after activation at a multiplicity of infection of 0.5. At day 4 after infection, fresh media were supplemented. At day 7, the cells were lysed in 1× lysis buffer (Renilla luciferase kit; Promega, Madison, Wisconsin). Lysates were subjected to 1 freeze-thaw cycle to aid the lysis and stored at −80°C. The level of viral infection and replication was determined as relative light units using Renilla luciferase assay and Victor ³V luminometer (PerkinElmer, Waltham, Massachusetts).

Statistical analysis

Data were analyzed using Student's t test, Mann-Whitney rank sum test, and repeated-measures ANOVA test as appropriate. Correlations were performed using Spearman rank order test. A standard level of statistical significance α = .05 was used; all reported P values are two-sided. SigmaStat version 3.1 (SPSS, Chicago, Illinois) and GraphPad Prism version 5 (GraphPad Software Inc, La Jolla, California) statistical and graphing software packages were used. The adjustments for multiple comparisons of cytokine expression analysis were not applied because that would result in an increase in the frequency of the null hypothesis being valid (type II error) when the association in the data is not a result of chance (47). The correction for the type I error applies only to the universal null hypothesis, that is, that the two groups are identical on all variables analyzed (48). The concentration of IFNγ at each assay condition was determined by independent assays (ELISA and Luminex), therefore, a valid association that is not null. Furthermore, there is no empirical justification for a hypothesis that all the associations observed are unpredictable manifestations of random processes (47). MPA causes systemic interdependent effects on cytokine and chemokine expression and is likely to simultaneously inhibit the production of multiple factors. Therefore, the null hypothesis does not apply, and an application of Bonferroni adjustment would result in an amplification of the type II error (47).

Results

MPA inhibits cytokine production by activated T cells

To address whether MPA affects cytokine production by activated T cells, PBMCs from healthy women volunteers were incubated with T cell-activating microbeads coated with antibodies against CD2/CD3/CD28 (MACSi beads) in the presence of increasing concentrations of MPA, E2, P4, or DEX. As demonstrated in Figure 1A and B, MPA exerts a significant inhibitory effect on the production of IFNγ by CD2/CD3/CD28-stimulated PBMCs and VMMCs at concentrations 10⁻⁷M and higher. Inhibition of IFNγ production was observed after stimulation of purified blood CD3⁺ cells, suggesting that MPA acts directly on T cells (Figure 1C). The suppressive effect of MPA is not restricted to T cells because MPA also reduced the production of TNFα by purified CD14⁺ monocytes after stimulation with TLR4 ligand bacterial LPS (Figure 1D). In contrast, E2 and P4 did not have a significant inhibitory effect on T cells or monocytes at concentrations up to 10⁻⁶M. DEX displayed an inhibitory effect on cytokine production by PBMCs, VMMCs, and monocytes at concentration as low as 10⁻⁹M. To minimize the potential influence of hormones present in serum and to avoid estrogen receptor activation by phenol red in culture media, all experiments were performed in phenol red-free medium supplemented with charcoal-dextran–treated FBS. However, the inhibition by MPA was independent of the type of serum or medium, and a similar effect was observed in the presence of normal human serum (Supplemental Figure 1, A and B, published on The Endocrine Society's Journals Online web site at http://endo.endojournals.org). An immunosuppressive effect of MPA but not P4 was also observed in experiments using antigen-specific stimulation with HIV-1 Gag peptide pool and PBMCs from HIV-1-infected individuals (Supplemental Figure 1C). The MPA-mediated inhibition of cytokine production by T cells and monocytes was fully reversed in the presence of equimolar concentrations of RU486, an antagonist of GR and progesterone receptor (PR) (Figure 1, E–H). In contrast, the inhibition mediated by DEX was only partially reversed by RU486.

Figure 1.

The effect of estrogen (E2), P4, and MPA on cytokine production by activated mononuclear cells. A–D, PBMCs (A), VMMCs (B), purified CD3⁺ T cells (C), or CD14⁺ monocytes (D) obtained from healthy female volunteers were incubated for 24 hours in the presence of a vehicle (Veh) or indicated concentrations of E2, P4, MPA, or DEX and activated for an additional 24 hours with microbeads (MACSi) coated with antibodies against CD2/CD3/CD28 (A–C) or with 100 ng/ml LPS (D). Graphs represent the concentration of IFNγ or TNFα in the culture media. Data are normalized to vehicle control (Veh); mean ± SEM of 5 (A) or 3 (B–D) donors is shown. *, Statistically significant difference of MPA- and DEX-treated samples compared with the P4-treated samples. E–H, Inhibitory effect of MPA is reversed by RU486 (RU). Experimental conditions and analysis are as described in A except that the cells were incubated with MPA or DEX in the presence of RU486 (all reagents at 10⁻⁶M). Mean ± SEM of normalized data is presented. I, Effect of sex steroid hormones and MPA on the production of cytokines by CD2/CD3/CD28-stimulated PBMCs. Data are presented as normalized values relative to control of 5 independent experiments; error bars depict SEM. Statistically significant differences compared with vehicle control are indicated: * P < .05; ** P < .01; *** P < .001.

Importantly, the inhibitory effect of MPA on cytokine production is not limited to IFNγ and TNFα. As shown on specific examples in Figure 1I and in the heat map representation in Figure 2, a significant reduction of the production of IL-2, IL-3, IL-4, IL-5, IL-9, IL-12 (p40 and p70), IL-13, sIL-2ra, and CD154 (sCD40L) by PBMCs incubated with T cell-activating microbeads was observed in the presence of MPA at concentration 10⁻⁷M and higher. Other cytokines and chemokines including MIP-1α and -β, IL-1α and -β, and IL-6 were inhibited in the presence of 10⁻⁶M MPA. Incubation with P4 at 10⁻⁷M and higher led to a reduced production of IL-4, IL-12 (p70), interferon gamma-induced protein 10 (IP-10; CXCL10), and monocyte-derived cytokine (MDC), whereas estrogen did not exert any discernible effect. MPA inhibited the production of IL-2, IL-13, and other cytokines by VMMCs; however, a greater variability was observed in assays employing the cells isolated from genital tissues (Figure 2).

Figure 2.

Heat map representation of the effect of MPA and sex steroid hormones on the production of cytokines and chemokines by PBMCs activated with a T cell-activating stimulus. PBMCs or VMMCs were activated with anti–CD2/CD3/CD28-coated microbeads, and the accumulation of cytokines and chemokines in the medium was analyzed as described in Figure 1. The concentration of factors in the media was determined using 39-plex MILLIPLEX MAP assay. Data are presented as normalized values relative to vehicle; means of 5 and 3 independent experiments for PBMCs and VMMCs, respectively, are presented. *, Statistically significant difference compared with vehicle control.

MPA inhibits cytokine production by activated pDCs

Several reports have suggested that the activation of pDCs, a cell population playing a key role in the early recognition of viral and bacterial infections, is modulated by P4 and synthetic progestins (49–51). To characterize the effect of MPA on pDC function, intracellular cytokine staining by multiparameter flow cytometry was used to quantify the percentage of IFNα- and TNFα-producing pDCs after stimulation with TLR7/8 and TLR9 ligands. PBMCs were preincubated with E2, P4, MPA, or DEX for 6 hours before stimulation with TLR7/8-specific ligand R848 or TLR9-specific ligand CpG in the presence of brefeldin A as described (50). After 20 hours of stimulation, the intracellular production of IFNα and TNFα in pDCs defined as CD123⁺CD303⁺ cells was determined (Figure 3A). As demonstrated in Figure 3B, both MPA and DEX inhibit R848- and CpG-induced IFNα and TNFα production by stimulated pDCs. In contrast, P4 or E2 did not exert any effect on cytokine production by pDCs after TLR stimulation.

Figure 3.

MPA inhibits cytokine production by activated pDCs. A, Gating strategy for the analysis of intracellular production of IFNα and TNFα by pDCs that were either unstimulated (control [Ctrl]), stimulated with R848 (5 μg/ml) or a combination of R848 and MPA (10⁻⁶M). B, Effect of E2, P4, MPA, and DEX at 10⁻⁸M (black bars) or 10⁻⁶M (white bars) on intracellular production of IFNα and TNFα in pDCs. Data from 3 independent donors normalized to vehicle (Veh) control ± SEM are presented. C, IFNα levels in plasma and CVL of women without any form of hormonal contraception (Control), women using DMPA, and women using combined oral contraceptives (COC). D, Effect of E2, P4, MPA, and DEX on antigen-induced T cell proliferation. Purified mouse DCs were incubated with 100 μg/ml OVA in the presence of hormonal regulators for 24 hours followed by 4 days incubation with OVA and purified transgenic OVA-specific OT-2 T cells in the presence of hormonal regulators at the indicated DC to T cell ratios. Cell proliferation was determined by incorporation of [³H]thymidine during the last 24 hours of culture. Data are shown as mean ± SEM. Statistically significant differences compared with vehicle control are indicated: * P < .05; ** P < .01; *** P < .001 (analyzed using paired t test [B], Mann-Whitney U test [C], and unpaired t test [D]).

|DMPA use is associated with lower levels of IFNα in plasma and CVL

pDCs represent a major source of IFNα in vivo (52, 53). To assess whether the use of DMPA is associated with changes in the systemic and local levels of IFNα, plasma and CVL from 21 women not using any form of hormonal contraception, 19 women using DMPA, and 14 women using combined oral contraception were analyzed. As shown in Figure 3C, use of DMPA was associated with significantly lower levels of IFNα in plasma (P = .032) and CVL (P = .034). In the control group without any form of hormonal contraception, no significant difference in IFNα levels in plasma or CVL was detected between women in the follicular vs luteal phase of the ovarian cycle.

MPA suppresses T cell proliferation after antigen presentation

To address whether MPA or sex steroid hormones affect the capacity of antigen-presenting cells to induce antigen-specific T cell proliferation, a TCR-transgenic mouse model was employed. Bone marrow-derived DCs from B6 mice were preloaded with OVA in the presence of E2, P4, MPA, or DEX for 24 hours and then co-incubated with purified OVA-specific TCR-transgenic OT-2 T cells at indicated DC to T cell ratios in the presence or absence of hormonal regulators. Both MPA and DEX significantly inhibited T cell proliferation after antigen presentation by DCs (Figure 3D). Preincubation of DCs with MPA and DEX followed by careful washing and antigen presentation in the absence of hormonal regulators in the culture medium resulted in a lower but significant inhibition of T cell proliferation (Supplemental Figure 2). These results indicate that MPA inhibits the level of T cell proliferation after antigen presentation both directly and indirectly via an effect on antigen-presenting cells.

MPA prevents activation-induced down-regulation of surface expression of HIV-1 coreceptors CXCR4 and CCR5 on T cells

It was previously reported that the expression of HIV-1 coreceptor CCR5 on the surface of T cells is modulated by endogenous levels of P4 as well as by progestins delivered as a component of hormonal contraception (54, 55). To address whether MPA affects the levels of HIV-1 coreceptors on T cells, PBMCs were incubated with MPA or sex steroid hormones for 24 hours followed by incubation with T cell-activating anti–CD2/CD3/CD28-coated microbeads for an additional 24 hours. Expression of CCR5 and CXCR4 was analyzed on subsets of naive, central memory, effector memory, and late effector T cells defined by surface expression of CD27 and CD45RO (gating strategy is depicted in Supplemental Figure 3). As demonstrated in Figure 4 and previously reported (56, 57), activation of T cells via CD3 resulted in a significant down-regulation of CXCR4 on the surface of both CD4⁺ and CD8⁺ T cells. Importantly, the down-regulation was fully reversed in the presence of MPA during activation. A similar trend was observed for CCR5, although the effect of MPA on CCR5 expression after in vitro activation was less pronounced (Figure 4 and Supplemental Figure 4).

Figure 4.

MPA prevents activation-induced down-regulation of HIV-1 coreceptors on the surface of T cells. PBMCs were incubated for 24 hours with vehicle (Veh), P4, MPA, or DEX at 10⁻⁶M before activation with anti–CD2/CD3/CD28-coated microbeads at a bead to cell ratio 1:2 for 24 hours. A and B, Expression of HIV-1 coreceptors CXCR4 and CCR5 on central memory (Tcm) and effector memory (Tem) subsets of CD4⁺ (A) and CD8⁺ (B) T cells is shown. Histogram overlays indicate the staining with isotype control (shaded) and CXCR4 or CCR5 antibody (thick line) on unstimulated (unstim) T cells or T cells stimulated with anti–CD2/CD3/CD28-coated microbeads. Results of one of three similar experiments are shown.

MPA enhances HIV-1 proliferation in in vitro culture

Previous studies suggested that E2 and P4 regulate the rate of HIV-1 replication in peripheral blood cells (58) and that glucocorticoid agonists accelerate the rate of HIV-1 transcription (59–62). To address whether MPA exerts an effect on HIV-1 infection and replication in vitro, CD8-depleted PBMCs or purified CD4⁺ T cells depleted of CD14⁺ monocytes were incubated in the presence of steroid hormones or MPA for 1 or 5 days before activation with either anti–CD2/CD3/CD28-coated microbeads or PHA-L for 24 hours. Activated cells were infected with either a CCR5-tropic (BaL-LucR) or a CXCR4-tropic (NL4.3-LucR) recombinant HIV-1 virus encoding Renilla luciferase, a sensitive indicator of HIV-1 replication. Luciferase activity in cell lysate was determined at 4 and 7 days after infection. The presence of MPA at 10⁻⁶M results in a significant increase in luciferase activity in both CD8-depleted PBMCs and purified CD4⁺ T cells infected with CCR5- or CXCR4-tropic viruses (Figure 5 and Supplemental Figure 5). In contrast, treatment with E2 or P4 did not result in an altered rate of viral proliferation. Treatment with DEX increased HIV-1 proliferation at 10⁻⁸M but led to a decreased viral proliferation at 10⁻⁶M likely due to the suppression of T cell proliferation.

Figure 5.

MPA enhances the replication of HIV-1 in in vitro cell culture. A and B, CD8-depleted PBMCs (A) or purified CD4⁺ T cells depleted of CD14⁺ cells (B) were incubated in the presence of hormonal regulators at the indicated concentration for 5 days before activation with either anti–CD2/CD3/CD28-coated microbeads (MACSi; bead to cell ratio of 1:2 for PBMCs and 1:1 for CD4⁺ T cells) or 2 μg/ml PHA-L for 24 hours. CCR5-tropic (Bal) or CXCR4-tropic (NL4.3) virus encoding Renilla luciferase was added at a multiplicity of infection of 0.5, and the cells were cultured for an additional 7 days before analysis. Luciferase activity is expressed as relative light units (RLU). The results of a representative experiment of 4 similar experiments are shown. Error bars depict SD. Statistically significant difference compared with vehicle control are indicated: * P < .05; ** P < .01; *** P < .001 (analyzed using unpaired t test).

Discussion

The effect of hormonal contraception on HIV-1 acquisition and transmission represents an important global public health issue with profound implications for policies on family planning in countries with high HIV-1 risk. In this study, we addressed the mechanisms by which MPA may affect the acquisition of HIV-1 and other infections. Characterization of the effect of MPA on innate and adaptive immune mechanisms as a function of concentration indicated that MPA suppresses both arms of the immune system at a significantly lower concentration than unmodified P4. At a concentration of 10⁻⁷M (100nM) and higher, MPA inhibited the production of IFNγ, IL-2, IL-4, IL-12, IL-13, TNFα, soluble CD40 ligand, and other cytokines and chemokines by T cells activated via the TCR and the production of TNFα by monocytes activated by LPS via the CD14/MD-2/TLR4 complex (Figures 1 and 2). MPA suppressed the proliferation of T cells after antigen presentation and, importantly, reduced the ability of pDCs to respond to stimulation via TLR7/8 and TLR9 by production of IFNα and TNFα (Figure 3). The suppressive effect of MPA on IFNα production is further supported by the data demonstrating lower systemic and genital secretion levels of IFNα in women using DMPA (Figure 3C). pDCs are a key immune population functioning as a sentinel recognizing early viral and bacterial infections and inducing antiviral responses via a release of factors such as class-I IFNs (52, 53). IFNα-producing pDCs accumulate beneath the genital epithelium at 1 day after SIV infection in rhesus macaques (63). Active suppression of central immune mechanisms, including the function of pDCs, antigen presentation, and T cell effector mechanisms, may tip the balance between the proliferation of a founder viral population and immune control at an early stage of infection toward the benefit of the virus. This may be the underlying mechanism contributing to the observed increase in HIV-1 acquisition in women using high-dose DMPA contraception. Our observations are consistent with previous reports demonstrating that MPA suppresses the production of IL-1α, IL-12, IL-10, IL-13, and granulocyte colony-stimulating factor by Bacillus Calmette–Guérin antigen-stimulated human PBMCs (44) and that administration of DMPA to mice reduces IFNα production by pDCs (51).

The critical question is whether the concentration at which MPA exerts its immunosuppressive properties in vitro is relevant to the physiological concentration in women using DMPA. Pharmacokinetic studies show that after administration of 150 mg DMPA by I.M. injection, the serum concentration of MPA reaches up to 6.5 × 10⁻⁸M (25 ng/ml) within days after injection and decreases to about 0.3 to 2 × 10⁻⁸M (1-9 ng/ml) in the following weeks (37, 64–67). The immunosuppressive effect of MPA in vitro is generally observed at concentrations equal and higher than the peak concentration detected in plasma of DMPA-using women (≥10⁻⁷M or 38 ng/ml); however, alteration of cytokine production was observed at concentrations as low as 10⁻⁸M in some individuals. Importantly, long-term exposure of immune cells and tissues to MPA in vivo is likely to exert a significant effect at lower concentrations (51). Furthermore, the effect of MPA may be enhanced locally in tissues such as the genital mucosa due to the differences in the expression of GR and PR and the presence of cell types with increased sensitivity to GR- and PR-mediated signaling (37). A potential limitation of the studies using VMMCs is that the cells were purified from the mucosa of postmenopausal women. Cell type frequencies and properties of mucosa-populating cells may differ between pre- and postmenopausal women. Because the data obtained in vitro cannot be directly translated to the situation in vivo due to the variances in the dosage, pharmacokinetics, effect of serum hormone-binding proteins, and length of exposure of immune cells, it is critical to study the effect of MPA and other progestins on the immune system in vivo. In this respect, the observation of reduced levels of IFNα in circulation and CVL of women using DMPA (Figure 3C) represents important evidence of an effect of DMPA on immune system in vivo.

It is likely that MPA exerts its immunosuppressive properties via the engagement of the GR (36–40, 42). Activated GR was shown to regulate the expression of a number of cytokines and proinflammatory factors via a direct interaction with nuclear factor κB (NFκB) and activator protein-1 (AP-1), modification of the basal transcriptional machinery, chromatin remodeling, and repression of the induction of NFκB inhibitor IκB resulting in transrepression (68–72). The GR-ligand complex decreases the activity of the Th1 transcription factor T-box expressed in T cells (T-bet) and Th2 transcription factor GATA-binding protein-3 (GATA-3) (73, 74). Interestingly, HIV-1-infected patients display glucocorticoid hypersensitivity associated with reduced T cell function (75). HIV-1 accessory protein Vpr binds to the steroid receptor coactivator motif LXXLL on the GR, leading to enhanced GR activation and suppression of production of IL-12 and other cytokines (72, 76). Thus, the sensitivity of immune cells to the effect of MPA may be significantly enhanced in the context of HIV-1 infection. In contrast to MPA, P4 does not efficiently activate the GR signaling cascade, and its effect on immune processes is less pronounced. However, it cannot be excluded that the effect of MPA is partially mediated via the PR pathway because MPA has higher affinity to PR than P4 (37). Previous studies have demonstrated that P4 affects T cell-mediated immune mechanisms by inhibition of cytokine production and cytolytic activity (4, 77–85) and promotion of the expression of Th2-type cytokines (86–93). Estrogen was shown to exert a stimulatory or inhibitory effect on the expression of a number of cytokines depending on the concentration and experimental system (4, 94–97). Here we show that P4 at concentration 10⁻⁷M reduces the levels of IL-4, IL-12 p70, IP-10 and MDC production after nonspecific stimulation of T cells. Treatment with estrogen at 10⁻⁶M resulted in reduced levels of IL-5, IL-12 p70, and TNFß; no effect was discernible at 10⁻⁷M. P4 but not estrogen caused a slight reduction of antigen-specific T cell proliferation at 1:8 DC to T cell ratio at 10⁻⁶M. Differences between our study and previously published studies may be caused by variances in experimental conditions and methods of T cell activation. Concurrent stimulation of various steroid receptors may result in a synergistic or antagonistic effect (58). In our experiments, the presence of a combination of E2 and P4 at midproliferative (10⁻¹⁰M and 10⁻⁹M) or midsecretory levels (10⁻⁹M and 10⁻⁷M) did not result in a change in the production of IFNγ by activated PBMCs or modify the suppressive effect of MPA (data not shown).

Our data show that estrogen or progesterone at concentrations of up to 10⁻⁶M had no effect on TLR-stimulated production of IFNα and TNFα by pDCs (Figure 3). This is consistent with a previous study demonstrating that P4 inhibits IFNα production by CpG-stimulated purified human pDCs at a concentration of 20 μg/ml (6.4 × 10⁻⁵M), whereas no inhibition was observed at 0.2 μg/ml (6.4 × 10⁻⁷M) (51). Interestingly, pDCs isolated from mice treated with DMPA produced lower levels of IFNα upon CpG stimulation and displayed lower levels of serum IFNα after vesicular stomatitis virus infection (51). These data are in concordance with the observation of decreased systemic and cervicovaginal levels of IFNα in women using DMPA described here. It has been reported that IFNα production upon stimulation with synthetic HIV-1–derived TLR7/8 ligand is higher in pDCs isolated from women than men (49, 50). Estrogen signaling was not responsible for this effect (49), and a positive correlation was observed between plasma P4 levels (up to 5 × 10⁻⁸M) and IFNα production by pDCs in women (50). We did not observe an effect of P4 on IFNα production by pDCs at these concentrations.

Previously, it was reported that the use of oral hormonal contraceptives is associated with increased expression of HIV-1 coreceptor CCR5 on intraepithelial endocervical CD4⁺ T lymphocytes (54) and that the increased level of P4 in pregnancy directly correlates with CCR5 expression on T cells in peripheral blood and genital tissue (55). We show that T cells activated in the presence of MPA display higher levels of surface HIV-1 coreceptors than either untreated or P4-treated cells (Figure 4). Our data are consistent with the observation that CXCR4 surface levels on T cells in cultured PBMCs are up-regulated under nonstimulating conditions within several hours of culture and significantly contract after T cell activation using anti-CD3 antibodies or PHA, leading to a decreased rate of replication of CXCR4-tropic virus (98). Down-modulation of surface CXCR4 after CD3 engagement is caused by endocytosis through clathrin-coated pits involving protein kinase C (PKC) signaling (56, 57). Down-modulation of CCR5 can occur via both clathrin-dependent and independent pathways (57, 99). The precise molecular mechanisms of the effect of MPA on the endocytosis and recycling of CXCR4 and CCR5 are unclear. The observed retention of surface coreceptor levels on activated T cells in the presence of MPA may explain the previous observation of increased CCR5 expression on CD4⁺ T cells in women using hormonal contraception (54).

MPA at 10⁻⁶M resulted in a significant increase in the rate of proliferation of CCR5- and CXCR4-tropic viruses in CD8⁺ T cell-depleted PBMCs and purified CD4⁺ T cells (Figure 5 and Supplemental Figure 5). The observed increase in HIV-1 proliferation may be caused by enhanced retention of HIV-1 coreceptors on the surface of CD4⁺ T cells as suggested by our data and reported previously (98), by inhibition of production of antiviral factors, or by a direct effect on HIV-1 transcription. In this respect, we show that MPA reduced the levels of MIP-1α and -ß (CCL3 and CCL4), major HIV-suppressive factors (100, 101), in activated PBMC cultures (Figure 2). It has been reported that MIP-1ß and other CCR5 ligands induce endocytosis of CCR5 and decrease infectability with CCR5-tropic HIV-1 (102–104). The enhancement of HIV-1 proliferation may be also mediated by direct binding of activated GR-MPA complex to glucocorticoid response elements located within the HIV-1 long-terminal repeat (105, 106). The data reported here are consistent with previous reports demonstrating that HIV-1 replication can be stimulated by various glucocorticoid agonists (59–62). Importantly, activation of the GR releases unstimulated PBMCs from an early block in HIV-1 replication via a mechanism requiring Vpr (107). The presence of a GR ligand results in a translocation of GR-Vpr complex into the nucleus and efficient proviral integration. This mechanism may be particularly important in the early stage of infection.

In summary, we show that MPA may constrain immune responses to HIV-1 by suppression of innate and adaptive immune responses, by retention of HIV-1 coreceptors on T cells after activation, and by increasing the rate of HIV-1 replication. These mechanisms may promote HIV-1 proliferation in the genital tissue in the early stage of infection and contribute to the increased shedding of HIV-1 in the genital tract of HIV-1–infected women (16, 17, 108). The results of this study, previous studies indicating immunosuppressive properties of DMPA (4, 22–24, 42–44, 51, 78, 87, 109), and epidemiological studies demonstrating an association between DMPA use and increased risk of HIV-1 and other infections (4–11, 18–21) strongly suggest that careful consideration should be used when selecting the preferred form of contraception in areas with high HIV-1 prevalence. We recognize that observational studies may be burdened with confounding factors and difficult to interpret (34, 35); however, the identification of biological mechanisms underlying the observed association constitutes a strong argument against the use of high-dose injectable DMPA in women at high risk of HIV-1 infection. Importantly, immediate withdrawal of DMPA from family planning programs without offering other effective forms of contraception is not warranted because it could result in a sharp increase in unwanted births and maternal and infant mortality (110). At the current time, the risk of DMPA discontinuation without a switch to a safe and effective method of contraception far outweighs the risk associated with continuous DMPA use because in some regions, up to 9 additional maternal deaths will occur for every case of HIV-1 averted (41). Women using DMPA or other forms of hormonal contraception should be strongly advised to use condoms, male or female, as recommended by recent World Health Organization guidelines (1). DMPA should be gradually replaced with alternative modern methods of contraception such as contraceptives with a lower impact on the immune system (41).