Effect of depot medroxyprogesterone acetate on human β-defensin production and structural integrity of the human vaginal epithelium

Abstract

BackgroundRecent studies implicate the use of hormonal contraceptives, particularly depot medroxyprogesterone acetate (DMPA; Depo-Provera), with increased risk of HIV-1 acquisition. In this study, we recruited long-term users of hormonal contraceptives to further understand the biological mechanism behind this phenomenon. Methods83 healthy, premenopausal women using DMPA (22 participants), combined oral contraceptives (17), NuvaRing (17), or non-hormonal contraceptives (26) were recruited in Birmingham, AL, USA, for a vaginal biopsy, blood draw, and cervicovaginal lavage (CVL). Participants were excluded if they were pregnant or symptomatic for vaginitis or ulcers. Informed consent was obtained from all patients; the study was approved by the Institutional Review Board of University of Alabama at Birmingham. Vaginal epithelium was assessed with two distinct measurements: a mean thickness and shortest distance (from the lumen to the nearest lamina propria projection; minimum 100 points per biopsy). CVL antiviral content was assessed with ELISA and Luminex. RNA sequencing was performed on four DMPA and four control vaginal biopsies with use of laser capture microdissection to dissect the epithelium from the lamina propria. FindingsDMPA use was significantly associated with thinner vaginal epithelium (median DMPA 290·7 μm; median control 353·8 μm; p=0·01). DMPA and NuvaRing suppressed production of human β defensin-2 (HBD2) (control 13 198 pg/mL; DMPA 741·5 pg/mL, p=0·0002; NuvaRing 326·7 pg/mL, p=0·0003) and HBD3 (control 776·4 pg/mL; DMPA 50·2 pg/mL, p<0·0001; NuvaRing 41·9 pg/mL, p=0·0004) in CVL. HBD3 transcript levels were 330-fold reduced in the vaginal epithelium of DMPA users compared with controls (control fragments per kilobase of exon per million fragments mapped [FPKM] 69·7, DMPA FPKM 0·21; p=4·17 × 10−5; q=8·08 × 10−3). Immunofluorescence staining on DMPA biopsies confirmed the loss of HBD3 protein. DMPA users showed significant decreases in key structural genes including corneodesmosin (control FPKM 9·9; DMPA FPKM 0·26; p=1·36 × 10−6; q=5·86 × 10−4) and loricrin (control FPKM 10·2; DMPA FPKM 0·9; p=1·3 × 10−7; q=1·15 × 10−4). Markers of wound healing, including haemoglobin α2 (control FPKM 3·8; DMPA FPKM 40; p=1·6 × 10−5; q=3·7 × 10−3) and plasminogen activator (control FPKM 0·9; DMPA FPKM 12·7; p=2·4 × 10−8; q=2·8 × 10−5), were significantly upregulated in the vaginal epithelium of DMPA users compared with controls. InterpretationBecause HBDs are implicated in a specific anti-HIV response, loss of HBDs in the genital tract could partly account for the increased risk of infection in DMPA users. Decreased transcription of structural genes combined with increased expression of genes indicating wound healing suggests that DMPA weakens the natural barrier of the vaginal epithelium, which could lead to increased epithelial damage and exposure of target cells to mucosal pathogens such as HIV-1. FundingThis work was supported by NIH Grant P01 AI083027.