Effect of Hormonal Contraception on the Function of Plasmacytoid Dendritic Cells and Distribution of Immune Cell Populations in the Female Reproductive Tract

Abstract

Objective: 

Epidemiological evidence suggests an association between the use of hormonal contraception and an increased risk of acquiring sexually transmitted diseases including HIV-1. We sought to elucidate the biological mechanisms underlying the effect of hormonal contraception on the immune system.

Design: 

Cross-sectional study.

Methods: 

To delineate the biological mechanisms underlying the effect of hormonal contraceptives on the immune system, we analyzed the functional capacity of circulating plasmacytoid dendritic cells (pDCs), the distribution of vaginal immune cell populations, and the systemic and genital levels of immune mediators in women using depot medroxyprogesterone acetate (DMPA), NuvaRing, or combined oral contraceptives (COC).

Results: 

The use of DMPA or NuvaRing was associated with reduced capacity of circulating pDCs to produce interferon (IFN)-α and tumor necrosis (TNF-α) in response to TLR-9 stimulation. Systemic levels of IFN-α and cervicovaginal fluid levels of IFN-α, CXCL10, monocyte chemotactic protein-1, and granulocyte-colony stimulating factor were significantly lower in DMPA users compared to control volunteers not using hormonal contraception. The density of CD207⁺ Langerhans cells in the vaginal epithelium was reduced in NuvaRing and combined oral contraceptive users but not in DMPA users.

Conclusions: 

The presented evidence suggests that the use of some types of hormonal contraception is associated with reduced functional capacity of circulating pDCs and altered immune environment in the female reproductive tract.

INTRODUCTION

Access to safe, effective, and affordable methods of contraception is a critical component of preventive health care. Contraception provides women with a means of control over their reproductive health, reduces the number of abortions and unintended pregnancies, and, in effect, decreases maternal and infant morbidity and mortality. Injectable contraceptives, including depot medroxyprogesterone acetate (DMPA), are increasingly popular because of their ease of use, effectiveness, and affordability.^1,2 Areas with high use of injectables often overlap with areas of high HIV-1 incidence, particularly in eastern and southern Africa.³ Unfortunately, several studies have suggested that the use of hormonal contraceptives, particularly DMPA, is associated with an increased risk of acquiring and transmitting HIV-1.^4–10 A recent observational study among HIV-1 serodiscordant couples indicated that the risk of acquiring HIV-1 was increased 2- to 3.9-fold in women using DMPA compared to nonhormonal contraception users.^6,11 It has been estimated that injectable contraceptives may be responsible for 27,000–130,000 HIV-1 infections per year globally.³ However, the issue remains controversial as several studies failed to observe an overall effect of hormonal contraception on the incidence of HIV-1 infection or disease progression.⁹ Importantly, it has been estimated that removal of injectables from use in African countries without the majority of women switching to an equally effective contraception would lead to an increase in maternal deaths, outnumbering the frequency of potential of HIV-1 infections averted.^3,12 The World Health Organization has called for further research on possible links between hormonal contraceptives and HIV-1 acquisition and transmission.⁴

Multiple studies have sought to understand the biological mechanisms underlying the effect of hormonal contraception on HIV-1 acquisition (reviewed in^7,13,14). Hormonal contraception use has been linked to changes in the frequencies of immune cells in genital mucosae^15–21 and altered levels of cytokines, chemokines, and antiviral factors in cervicovaginal fluid.^22,23 Hormonal contraception use is associated with increased acquisition of cervical candidiasis, Chlamydia trachomatis, Neisseria gonorrhoeae, and Mycoplasma genitalium infections, which may independently facilitate HIV-1 acquisition and transmission.^10,24–27 Studies using non-human primate models demonstrated that DMPA enhances the risk of simian immunodeficiency virus acquisition by vaginal exposure (reviewed in^7,13). Medroxyprogesterone acetate (MPA), the progestin component of DMPA, was shown to suppress systemic regulators of cellular and humoral immunity, reduce cytokine production by plasmacytoid dendritic cells (pDCs), and increase the replication of HIV-1 in activated peripheral blood mononuclear cells (PBMCs) in vitro.^28–32

This study addresses the systemic and genital immune milieu in premenopausal women using DMPA, NuvaRing, or combined oral contraceptives (COCs). Circulating pDCs isolated from DMPA or NuvaRing users display decreased capacity to produce interferon (IFN)-α and tumor necrosis factor (TNF)-α following TLR-9 but not TLR-7/8 stimulation. Furthermore, DMPA and NuvaRing use is associated with altered levels of immune mediators both systemically and in the cervicovaginal fluid. The presented data demonstrate that the use of hormonal contraceptives is linked to significant alterations of systemic and local immune environment; however, the implications of the observed changes for the transmission of HIV-1 warrant further investigation.

METHODS

Volunteers and Study Design

A total of 84 female volunteers were recruited for this study from May 2011 to May 2012. Eligibility criteria included the following: age 19–40 years, using Depo Provera (intramuscular injection, 150 mg of DMPA), COCs (ethinyl estradiol and variable progestin component), NuvaRing (vaginal ring, 0.12 mg of etonogestrel and 0.015 mg of ethinyl estradiol release daily, on average), or not using any form of hormonal contraceptive (control group). Within the control group, 12 volunteers were assessed at the follicular phase and 12 at the luteal phase of the estrous cycle based on self-reported time since last menstrual period and 17β-estradiol (E2) and progesterone (P4) levels in plasma; 1 was not determined. Exclusion criteria included abnormal vaginal discharge, significant macroscopic vaginal pathology, any sign of bacterial vaginosis or infection, pregnancy, or current usage of an intrauterine device. The University of Alabama at Birmingham Institutional Review Board approved the study protocol and all subjects signed informed consent forms. Demographic, contraceptive, and sexually transmitted infection histories were collected on standardized forms at enrollment. In total, 20 mL of acid citrate dextrose (ACD)–treated blood was collected. A color pHast indicator strip (EMD Chemical, Gibbstown, NJ) was applied to the vaginal wall to determine intravaginal pH. Cervicovaginal lavage (CVL) was obtained by vigorous flushing of the cervix and vagina with 5 mL of sterile saline. The lavage fluid was subsequently mixed with protease inhibitors (5 μg/mL of aprotinin, 1 μg/mL of leupeptin, 1 μg/mL of antipain, 1 μg/mL of pepstain, 200 μg/mL of sodium azide, and 1 mM phenylmethylsulfonyl flouride; Sigma, St. Louis, MO) and cleared by centrifugation at 14,000g for 5 minutes before cryopreservation.

Analysis of Vaginal Biopsies

Lidocaine injection (1 mL) was applied and a full-thickness biopsy was obtained from the lateral vaginal wall with Acu-Punch 8 mm biopsy punch tool (Acuderm, Ft. Lauderdale, FL). Biopsies were immediately washed with Hanks' balanced salt solution and orientated in optimal cutting temperature compound (Tissue-Tek; Sakura, Zoeterwoude, the Netherlands). Samples were snap-frozen by immersion in precooled isopentane in an acetone bath encased with dry ice. Flash freezing of tissue used in this study has been noted to retain antigenicity, reduce tissue distortion, and preserve overall architecture.³³ Samples were stored at −80°C until use. All sectioning was performed on a Leica CM1900 UV cryostat using precleaned slides.

Immunofluorescence staining was performed to determine the location and density of immune cells in the vaginal epithelium. Sections were allowed to thaw for 30 seconds and blocked with a bovine serum albumin-containing saponin buffer (2.5 g of BSA, 0.25 g of saponin per 250 mL of phosphate buffered solution) for 1 hour. Thirty-micrometer sections from each biopsy were stained for Langerin (CD207 PE; Beckman Coulter, Brea, CA) and CD3 (rabbit anti-human C7930; Sigma). Twelve-micrometer sections were stained for CD3 and CD4 (mouse anti-human NCL-CD4-368; Leica Microsystems, Milton Keynes, United Kingdom) or CD3 and CD8 (mouse anti-human NCL-CD8-4B11; Leica Microsystems). Slides were incubated with primary antibodies at 4°C overnight, washed, and incubated with secondary antibody (goat anti-rabbit Alexa 488 for CD3 or goat anti-mouse Alexa 555 for CD4/CD8; Invitrogen, Carlsbad, CA) for 2 hours at room temperature, with subsequent fixation in 4% paraformaldehyde for 10 minutes and 5 minutes in Hoechst nuclear stain (Fisher Scientific, Pittsburgh, PA) and mounting in Fluoro-Gel with TES buffer (EMS, Hatfield, PA). Images were taken on a Zeiss A1 Confocal microscope with Z stack analysis at ×10 magnification. A minimum of 6 images at ×10 magnification were analyzed per biopsy per stain. Reference points were plotted on each image along the apical side of the epithelium at 30 μm intervals. Coordinates for each Langerin⁺, CD3⁺, CD3⁺CD4⁺, or CD3⁺CD8⁺ cell were compared against the apical epithelium reference points to determine the shortest distance using a specifically designed algorithm (http://shortdistance.imagejs.googlecode.com/git/shortdistance.html). Image analysis was performed with ImageJ64 open source software.

Cytokine and Immunoglobulin Analysis

Concentrations of 26 cytokines and chemokines in CVL and plasma were determined using the 26-plex MILLIPLEX Human Cytokine/Chemokine Panel kit (Millipore, Billerica, MA). Samples were analyzed on a Bioplex 100 system with Bioplex Manager Software 5.0 (Bio-Rad, Hercules, CA). IFN-α levels in controls, DMPA, and COC users reported in this study partially overlap with a previously published data set.²⁸ Concentrations of total IgA, IgM, and IgG in CVL were determined with sandwich enzyme-linked immunosorbent assay. In brief, 96-well plates were coated with 1 μg/mL of polyclonal goat F(ab′)² fragments of anti-human IgG, IgA, or IgM (Jackson Immunoresearch, West Grove, PA) in phosphate-buffered saline, 100 μL per well, overnight at 4°C. Plates were washed with phosphate-buffered saline containing 0.05% Tween-20 (PBST) and blocked with 5% goat serum in PBST for IgA and IgM or with 1% BSA for IgG, 200 μL per well, for 2 hours at room temperature. Samples and standards (Microgenetics, Fremont, CA) were prepared at appropriate dilutions in 2% goat serum in PBST for IgA and IgM or 1% BSA in PBST for IgG. Blocking buffer was removed and sample and standard were added, 100 μL per well, and incubated overnight at 4°C. After washing, biotinylated F(ab′)² fragments of goat anti-human IgA (1:5000; KPL, Gaithersburg, MD), IgM (1:5000; KPL), or IgG (1:8000; Invitrogen, Camarillo, CA) were added, 100 μL per well, and incubated for 1 hour at room temperature. Plates were washed followed by the addition of Streptavidin-conjugated peroxidase (1:5000; Invitrogen, Carlsbad, CA), 100 μL per well, and incubated at room temperature for 30 minutes. After the final wash, SureBlue TMB peroxidase substrate (KPL) was added, 100 μL per well, incubated for 15 minutes at room temperature, after which the reaction was stopped by the addition of 100 μL of 1N HCl. Optical density values were determined at 450 nm.

Determination of Hormonal Concentration

Plasma (400 μL) and CVL samples (200 μL) from each patient were analyzed at the Obstetrics and Gynecology Laboratory at the University of Alabama at Birmingham to determine levels of 17β-estradiol (E2) and progesterone (P4). P4 and E2 were determined by the Access Progesterone Assay and the Access Estrogen Assay, respectively (Beckman Coulter).

pDC Activation and Intracellular Cytokine Staining Assay

pDC activation and cytokine production assays were performed as described.^28,30 In brief, PBMCs (1.5 × 10⁶ cells/mL) were stimulated with 5 μg/mL of TLR-7/8 ligand R848 (Invivogen, San Diego, CA) or 2 µM TLR-9 ligand CpG ODN2216 (Hycult Biotech, Uden, the Netherlands) in the presence of 1 μL/mL of GolgiPlug (BD Biosciences, San Diego, CA). After 20 hours, cells were collected and stained for pDC markers [CD123-PE-Cy7 (eBioscience, San Diego, CA) and CD303-APC (BCDA-2; Miltenyi, Auburn, CA)]. Cells were permeabilized using the Cytofix/Cytoperm kit (BD) and stained intracellularly with IFN-α-PE (BD) and TNF-α-FITC (eBioscience) monoclonal antibodies. Samples were analyzed on LSR-II flow cytometer (BD), and data analysis was performed using the FACSDiva software (BD).

Analysis of Memory and Effector T-Cell Subsets

Isolated PBMCs (0.5 × 10⁶) were stained with anti-CD3-eFluor450, CD4-APC-eFluor780, CD8-PerCP-CY5.5, CD27-APC, CD45RO-PE, CD57-FITC, and CD38-PE-Cy7 (eBioscience) to distinguish subsets. Appropriate isotype controls were used to determine the percentage of cells expressing the respective markers (Biolegend, San Diego, CA). Samples were analyzed on LSR-II flow cytometer (BD), and data analysis was performed using the FACSDiva software (BD).

Statistical Analysis

All data were stored in Microsoft Excel 2007 (Microsoft, Redmond, WA) and exported to GraphPad Prism (GraphPad Software, La Jolla, CA) for analysis. Statistical significance of the difference between experimental groups was determined using Mann–Whitney U test; correlations were determined using Spearman rank correlation test. A standard level of statistical significance (α = 0.05) was used; all reported P-values are 2-sided.

RESULTS

To study the effect of hormonal contraception on the systemic and genital immune environment, 22 female volunteers using DMPA, 17 NuvaRing users, 17 COC users, and 25 volunteers not using any form of hormonal contraceptives were recruited. Demographic data of the study population are listed in Table 1. The average age of the study population was 28 ± 5.6 years; average body mass index (BMI) was 31.1 ± 8.4 kg/m². Eleven (50%) of the women using DMPA had experienced amenorrhea for at least 1 month and up to 3 years before the date of sample collection. Women using any of the 3 forms of hormonal contraception displayed significantly reduced serum estrogen levels compared to controls (Table 1; P < 0.0001 for all groups). There were no significant correlations between collected immunological parameters and the demographic data.

TABLE 1: Demographic Data of Study Participants

Previous data from our laboratory have implicated MPA and etonogestrel, the progestin components of DMPA and NuvaRing, respectively, in impairing the cytokine response of pDCs to TLR-7/8 and TLR-9 stimulation in vitro.^28,30 To assess pDC function in hormonal contraception users, PBMCs were isolated and stimulated with TLR-9 (CpG) or TLR-7/8 (R848) ligand for 20 hours. Intracellular production of IFN-α and TNF-α in CD123⁺ CD303⁺ pDC population was determined using polychromatic flow cytometry (Fig. 1A). pDCs from DMPA and NuvaRing users displayed impaired capacity to produce IFN-α and TNF-α following TLR-9 stimulation (Figs. 1B, C). In contrast, hormonal contraceptive use did not affect pDC cytokine production following TLR-7/8–specific stimulation (Figs. 1D, E). Within the control group, no statistically significant difference in pDC response was detected between volunteers in the follicular versus luteal phases of the menstrual cycle. No statistically significant correlation between 17β-estradiol (E2) and progesterone (P4) levels in plasma and pDC responses was detected in any of the experimental groups. Within the COC group, direct correlations between BMI and pDC production of IFN-α (R = 0.67; P = 0.02) and TNF-α (R = 0.6; P = 0.04) were observed; no correlations between BMI and pDC responses were observed in any other experimental group. Because pDCs represent the predominant immune cell producers of IFN-α,³⁴ we investigated whether the use of hormonal contraception alters plasma levels of IFN-α or other chemokines and cytokines. The plasma levels of IFN-α (Fig. 2A; P = 0.02) and interleukin (IL)-8 (Fig. 2B; P = 0.01) were significantly reduced in DMPA users. A marginally significant decrease in plasma levels of IL-6 was observed (data not shown); no alterations in plasma levels of IL-1β, IL-2, IL-10, IL-12, TNF-α, IFN-γ, granulocyte-colony stimulating factor (G-CSF), CXCL10, or monocyte chemotactic protein(MCP)-1 were detected. Within the DMPA group, direct correlations between the plasma levels of IL-8 and E2 (R = 0.5; P = 0.02) and between plasma levels of IL-8 and time since DMPA injection (0–90 days; R = 0.48; P = 0.03) were detected; no other correlations between plasma E2 and P4 and plasma cytokines levels in any of the experimental groups were observed. A direct correlation between the overall length of DMPA treatment (1 week to 36 months) and plasma GM-CSF was observed (R = 0.56; P = 0.03); no other cytokine has been correlated to the overall length of DMPA treatment. Hormonal contraception was not associated with any significant alterations in the distribution between naive, effector, central memory, and effector memory subsets within the CD4⁺ and CD8⁺ T-cell populations (see Figure S1, Supplemental Digital Content, https://links.lww.com/QAI/A624). NuvaRing use was associated with an increased expression of activation marker CD38 on central memory (P = 0.006), effector memory (P = 0.006), and effector (P = 0.006) CD8⁺ T-cell subsets; no alteration in the expression of T-cell senescence marker CD57 was observed in any of the experimental groups (data not shown).

FIGURE 1: pDCs from NuvaRing and DMPA users display reduced ability to produce IFN-α and TNF-α in response to stimulation by TLR-9. PBMCs were isolated from 20 control volunteers not using any form of hormonal contraception, 17 DMPA users, 14 NuvaRing users, and 13 COC users, and stimulated with 2 μM TLR-9 ligand CpG ODN2216 or 5 μg/mL of TLR-7/8 ligand R848. A, Gating strategy for the analysis of intracellular production of IFN-α and TNF-α in pDCs identified as CD123⁺ CD303⁺ population. B–E, Intracellular cytokine staining for IFN-α (B, D) or TNF-α (C, E) in pDCs was performed in PBMCs stimulated with CpG (B, C) or R848 (D, E). Values are corrected for background staining of unstimulated cells. Bars indicate median values; statistical significance was determined using the Mann–Whitney U test.

FIGURE 2: DMPA use is associated with lower levels of plasma IFN-α and IL-8. Plasma from 21 control volunteers, 19 DMPA, 14 NuvaRing, and 14 COC users were analyzed for the concentrations of IFN-α (A) and IL-8 (B). Bars indicate median values; significance was determined using the Mann–Whitney U test.

Because alterations to the immune milieu of the female reproductive tract (FRT) may have a significant effect on HIV-1 acquisition and transmission,³⁵ the effects of hormonal contraception on the genital immune milieu were investigated. Analysis of cytokine and chemokine levels in the CVL of hormonal contraception users demonstrated that DMPA use is associated with significantly reduced levels of IFN-α (P = 0.03), CXCL10 (P = 0.008), MCP-1 (P = 0.0009), and G-CSF (P = 0.006; Fig. 3). Within the control group, no statistically significant difference in the levels of these cytokines was detected between volunteers in the follicular versus luteal phases of the menstrual cycle. The concentration of other immune mediators detected at high levels in the CVL, including IL-1β, IL-8, IL-6, and MIP-1β, did not differ among experimental groups (see Figure S2A, Supplemental Digital Content, https://links.lww.com/QAI/A624). DMPA use was associated with decreased CVL levels of IgG (P = 0.02); COC use was associated with decreased levels of IgA (P = 0.02) (see Figure S2B, Supplemental Digital Content, https://links.lww.com/QAI/A624). No statistically significant correlation between the FRT levels of immunoglobulins and E2 and P4 levels in plasma was detected. Within the control group, significant negative correlations were detected between the plasma level of P4 and CVL levels of G-CSF (R = −0.57; P = 0.01), GM-CSF (R = −0.51; P = 0.02), 1–12 p40 (R = −0.49; P = 0.03), IL-12 p70 (R = −0.47; P = 0.04), IL-15 (R = −0.7; P = 0.0008), IL-17 (R = −0.66; P = 0.01), MCP-1 (R = −0.55; P = 0.01), and TNF-α (R = −0.49; P = 0.03). No other statistically significant correlations between plasma levels of E2 and P4 and cytokine levels in CVL were observed in any of the experimental groups. A positive correlation between the BMI and IFN-α level in the CVL was observed in the control group (R = 0.64; P = 0.004). Within the DMPA group, CVL levels of MCP-1 correlated with the time since last DMPA injection (R = 0.47; P = 0.04); no other statistically significant correlation between the time from last DMPA injection or the overall length of DMPA treatment versus cytokine production by pDCs or cytokine levels in plasma and cervicovaginal fluid was observed.

FIGURE 3: DMPA use is associated with lower levels of IFN-α, CXCL10, MCP-1, G-CSF, TIMP-1, and TIMP-2 in the cervicovaginal fluid. CVLs from 21 controls, 20 DMPA, 14 NuvaRing, and 14 COC users were analyzed for chemokine, cytokine, MMP-7 and -9, and TIMP-1 and -2 concentrations. Bars indicate median values; significance was established using Mann–Whitney U test.

Matrix metalloproteinases (MMPs) and their tissue inhibitors (TIMPs) affect chemokine processing and epithelial tissue remodeling.³⁶ MMPs and TIMPs are highly expressed in reproductive tissues and are hormonally regulated throughout the menstrual cycle.³⁷ In CVLs of subjects of all experimental groups, the predominant metalloproteinases were MMP-9, MMP-7, MMP-2, MMP-12, and MMP-10, in descending order of concentration, (Figs. 3G, H, and data not shown). The use of DMPA was associated with reduced CVL levels of TIMP-1 (Fig. 3E, P = 0.02) and TIMP-2 (Fig. 3F, P = 0.02); NuvaRing users displayed reduced levels of TIMP-2 (Fig. 3F, P = 0.01). TIMP-1 and TIMP-2 display broad inhibitory activity against most MMPs³⁸; thus, a decrease in TIMP concentration is consistent with an increase in the overall activity of MMPs in the genital tract. Within the control group, no statistically significant correlations between the CVL levels of MMPs and TIMPs and plasma levels of E2 and P4 were detected. A positive correlation between the BMI and CVL TIMP-1 level was observed in the control group (R = 0.64; P = 0.01).

Langerhans cells (LCs) represent the main dendritic cell population of the vaginal epithelium.³⁹ Experimental evidence suggests that vaginal LCs pass infectious HIV virions to CD4⁺ T cells without being productively infected; however, the issue remains controversial.⁴⁰ To address the effect of hormonal contraception on immune cell populations in the lower FRT, the frequency and distribution of vaginal intraepithelial CD207⁺ LCs (Figs. 4A, C) and CD3⁺ T cells (Figs. 4B, D) were determined in vaginal biopsies of hormonal contraceptive users. The frequency of intraepithelial LCs was significantly decreased in both COC and NuvaRing users compared to either DMPA users or controls (Fig. 4A). However, no significant difference in LC density was detected between DMPA users and controls, consistent with previous studies.^19–21,41 A trend toward increased CD3⁺ cell density in DMPA users compared to controls was detected (Fig. 4B); however, the trend did not reach statistical significance. Specific analysis of the frequency of CD3⁺CD4⁺ and CD3⁺CD8⁺ T-cell subsets did not reveal any statistically significant differences between controls and the experimental groups (see Figure S3, Supplemental Digital Content, https://links.lww.com/QAI/A624). LCs and CD3⁺ cells were localized closer to the apical epithelial surface in women using DMPA compared to women using NuvaRing or COCs (Fig. 4C, D); the difference compared to control group did not reach statistical significance. Within the control group, no statistically significant correlations between the intraepithelial immune cell frequencies, plasma levels of E2 and P4, BMI, and age were detected. No differences were observed among volunteers in the follicular versus luteal phases of the menstrual cycle, in concordance with previous studies.^16,21

FIGURE 4: Density and shortest apical distance of vaginal intraepithelial Langerhans and CD3⁺ T cells are altered in users of hormonal contraception. Vaginal biopsies obtained from DMPA, NuvaRing, and COC users and control volunteers were sectioned at 30 μm and stained for CD207⁺ LCs (A, C) and CD3⁺ T cells (B, D). Intraepithelial density (A, B) and shortest distance to the apical epithelial surface (C, D) of Langerin⁺ (A, C) and CD3⁺ (B, D) cells were analyzed. Bars indicate median value for each group; significance was determined using Mann–Whitney U test.

DISCUSSION

In light of multiple studies implicating hormonal contraception use with increased susceptibility to HIV-1 infection,^5–11 identifying safe and effective contraceptives is a critical public health issue.⁴ This study examined alterations to systemic and local immune mediators associated with extended use of hormonal contraception to delineate biological mechanisms that may affect the susceptibility to HIV-1 and other sexually transmitted diseases. The study reveals 4 major findings: (1) the use of DMPA or NuvaRing is associated with reduced capacity of circulating pDCs to produce IFN-α and TNF-α in response to stimulation by TLR-9; (2) systemic and cervicovaginal levels of IFN-α are lower in DMPA users compared to controls not using hormonal contraception; (3) cervicovaginal levels of CXCL10, MCP-1, and G-CSF are lower in DMPA users; and (4) the frequency of CD207⁺ LCs in the vaginal epithelium is reduced in NuvaRing and COC users.

pDCs play a key role in the early recognition of viral and bacterial infections and represent a major source of IFN-α in vivo.^34,42,43 IFN-α-producing pDCs accumulate in the genital lamina propria 1 day following simian immunodeficiency virus infection in macaques.³⁴ Previous research from our laboratory demonstrated that in vitro incubation of pDCs with physiological concentrations of MPA results in an impaired response to stimulation by TLR-7/8 and TLR-9. In contrast, etonogestrel, the progestin component of NuvaRing, exerted inhibitory effect only at high concentrations.^28,30 Data presented here demonstrate that pDCs freshly isolated from DMPA or NuvaRing users display an impaired IFN-α or TNF-α response following TLR-9 but not TLR-7/8 stimulation. This observation, corroborated by lower plasma and CVL levels of IFN-α in DMPA users (Fig. 2A), indicates partial functional impairment of pDCs in DMPA and NuvaRing users. Suppression of pDC function may tip the balance between the proliferation of a founder viral population and early immune control during the acute phase of infection in favor of the transmitted virus. The reasons for the observed differences between the effect of various progestins on immune mechanisms in vitro and in vivo are unclear. Plausible causes include differences in serum concentration, bioavailability, and differences in affinities of binding to glucocorticoid receptor and other members of the steroid receptor family.^{7,28,30,44–46}

Levels of cervicovaginal CXCL10, G-CSF, MCP-1, TIMP-1, and TIMP-2 were reduced in DMPA users. CXCL10, G-CSF, and MCP-1 play a role in recruiting monocytes, neutrophils, and T lymphocytes to sites of infection and are produced in significant quantities in the uterine and endocervical tissues.⁴⁷ These data are consistent with the previous observation that MPA inhibits CXCL10 and MCP-1 production by activated PBMCs.³⁰ Reduced levels of these factors in the cervicovaginal fluid may result in an altered frequency and/or activation status of immune cells within the genital mucosae.

Previous research on vaginal immune cell populations in hormonal contraception users has been conflicting, with some studies noting a decrease in CD1a⁺ LC density in DMPA users¹⁵ and other studies showing no change in CD1a⁺ LCs in COC or DMPA users.²¹ Topical progesterone was found to increase CD1a⁺ dendritic cell populations in the vaginal epithelium.¹⁸ Studies using S-100 as a marker for vaginal LCs have uniformly noted no difference between DMPA users and controls.^19,20,41 The data presented here demonstrate significant decreases in CD207⁺ LC density in NuvaRing and COC users but not in DMPA users. As it has been previously suggested that subsets of vaginal LCs differ in their expression of CD207,⁴⁰ differences in experimental approaches and population markers may account for the differences between studies. Conflicting results have been reported regarding the alterations to vaginal lymphocytes in hormonal contraception users, with some studies finding an increase in vaginal CD3⁺, CD8⁺, and CD4⁺CCR5⁺ cells¹⁶ and other reporting no change or decrease in the frequency of these populations.^15,19–21 Here we report no significant alterations to vaginal CD3⁺, CD3⁺CD8⁺, or CD3⁺CD4⁺ lymphocyte population density or localization in DMPA users. Thus, DMPA does not seem to increase the frequency of HIV-1 target cells in vaginal mucosa.

Some of the observed changes in immune parameters in DMPA users may be related to chronic anovulation and reduced concentration of endogenous estrogen rather than a direct effect of DMPA. Furthermore, immune mechanisms may be affected by MPA concentration that peaks at 1–2 weeks after DMPA injection and decreases at the end of the 3-month administration period. This study addressed the effect of the overall length of DMPA use (1 week to 96 months) and the effect of the time since last DMPA injection. We observed a direct correlation between the overall length of DMPA treatment and plasma levels of GM-CSF and a correlation between the time since last DMPA injection and CVL levels of MCP-1; no other statistically significant correlations between the overall length of DMPA treatment and the time from the last DMPA injection in relation to any other immune parameter including cytokine production by pDCs and immune cell population densities in vaginal mucosa were observed. A positive correlation between the plasma levels of E2 versus IL-8 was observed within the DMPA group. No other correlation between plasma E2 and plasma and CVL cytokines levels, pDC function, and intraepithelial layer cell frequencies were observed in any of the experimental groups.

This study is subject to limitations, including small sample size overall and within each contraceptive group. Volunteers were not eligible if macroscopic genital inflammation was detected; however, potential effects mediated by previous or ongoing genital infections cannot be excluded. Frequency or last instance of intercourse was not considered in the analysis; these factors may influence the localization and density of immune cells. There was a bias within DMPA users toward volunteers of African descent (82%). No significant correlations between race and immune parameters, including cell density, cell localization, or cytokine and chemokine levels, were observed.

In summary, the presented data suggest that hormonal contraceptives modulate pDC function and levels of systemic and genital immune mediators. This adds to the growing amount of evidence of the biological effect of progestins on the immune system and on the immune environment in lower FRT.^7,13,14 However, it remains to be established whether the contraceptive-mediated alterations to systemic and genital immune mechanisms described here and in other studies exert an effect on the susceptibility to HIV-1 and other infections. Additional research into the biological effect of hormonal contraception on the immune system is critical for evidence-based selection of safe forms of contraceptives for women at risk of HIV-1. Women using hormonal contraceptives should be advised to use male or female condoms and practice additional HIV prevention strategies as recommended by World Health Organization guidelines.⁴

ACKNOWLEDGMENTS

The authors thank Dr. Jonas Almeida at the University of Alabama at Birmingham for developing the macro used for image analysis of cell localization.