Depot medroxyprogesterone acetate (DMPA) enhances susceptibility and increases the window of vulnerability to HIV-1 in humanized mice

The progestin-based hormonal contraceptive Depot Medroxyprogesterone Acetate (DMPA) is widely used in sub-Saharan Africa, where HIV-1 is endemic. Meta-analyses have shown that women using DMPA are 40% more likely than women not using hormonal contraceptives to acquire Human Immunodeficiency Virus (HIV-1). Therefore understanding how DMPA increases susceptibility to HIV-1 is an important public health issue. Using C57BL/6 mice and our previously optimized humanized mouse model (NOD-Rag1tm1Mom Il2rgtm1Wjl transplanted with hCD34-enriched hematopoietic stem cells; Hu-mice) where peripheral blood and tissues are reconstituted by human immune cells, we assessed how DMPA affected mucosal barrier function, HIV-1 susceptibility, viral titres, and target cells compared to mice in the diestrus phase of the estrous cycle, when endogenous progesterone is highest. We found that DMPA enhanced FITC-dextran dye leakage from the vaginal tract into the systemic circulation, enhanced target cells (hCD68+ macrophages, hCD4+ T cells) in the vaginal tract and peripheral blood (hCD45+hCD3+hCD4+hCCR5+ T cells), increased the rate of intravaginal HIV-1 infection, extended the window of vulnerability, and lowered vaginal viral titres following infection. These findings suggest DMPA may enhance susceptibility to HIV-1 in Hu-mice by impairing the vaginal epithelial barrier, increasing vaginal target cells (including macrophages), and extending the period of time during which Hu-mice are susceptible to infection; mechanisms that might also affect HIV-1 susceptibility in women.


DMPA impairs integrity of the murine vaginal epithelial barrier. The vaginal epithelial barrier pro-
vides the first line of defense against STIs like HIV-1. We therefore first sought to determine if DMPA negatively impacted the vaginal epithelial barrier in C57BL/6 mice, as has been reported in primates [16][17][18] , and another study in mice 19 , which could potentially enhance viral infection 20 . Our experimental designs are presented in Supplemental Fig. 1. In our first experiment C57BL/6 mice treated with DMPA were found to have decreased immunofluorescent staining for the cell-cell adhesion molecule desmoglein-1α in the vaginal epithelium as compared to mice in estrus (high endogenous estrogen) or diestrus (high endogenous progesterone) (N = 4/group) ( Fig. 1A-C). We were also able to assess the integrity of the murine vaginal barrier in a separate group of C57BL/6 mice by administering fluorescein isothiocyanate-dextran (FITC-dextran) dye directly into the vaginal lumen and measuring the fluorescence in blood samples from mice 4 h later; an indicator of how much dye leaked across the vaginal barrier into peripheral circulation. Using this assay, the vaginal barrier function was compared at estrus, diestrus, 1 week following treatment with DMPA (N = 4/group) to determine if DMPA enhanced leakage of the dye into the peripheral circulation, above what could be observed during the estrogen-high phase of the murine estrous cycle (estrus), and the progesterone-high phase of the estrous cycle (diestrus). Significantly more FITCdextran dye was quantified in the serum of mice that had received DMPA than those sacrificed during estrus or diestrus (Fig. 1D). Taken together, these two experiments demonstrate that DMPA negatively impacts vaginal cell-adhesion molecules and enhances leakage from the vaginal lumen into the submucosa, ultimately impairing the integrity of the murine vaginal epithelial barrier.

DMPA increases HIV-1 infection in humanized mice.
In addition to its effect on the vaginal barrier, DMPA may enhance viral infection through other mechanisms 13 . In our Hu-mouse model, as in nonhuman primates (NHP) [21][22][23] , and women 10,11 , we observed increased HIV-1 acquisition following intravaginal exposure 14 . Additionally, our previous work 15 found that target cells and inoculation dose are determinants of HIV-1 susceptibility. Herein, we thus set out to see if we could determine how DMPA affects intravaginal HIV-1 infection in this Hu-mouse model. Our experimental designs are presented in Supplemental Fig. 1.
Briefly, Hu-mice were reconstituted with hCD34 hematopoietic stem cells as described in Methods and "humanization" (%hCD45+ cells in peripheral blood) was assessed 90-120 days later by flow cytometry (Supplemental Fig. 2; Supplemental File 1). Because our previous work 15 found that target cells were the primary determinant of HIV-1 susceptibility in these Hu-mice, Hu-mice with > 10% hCD45+ cells in peripheral blood and < 10% hCD45 hCD45+ cells in peripheral blood were evenly distributed between our experimental groups (DMPA and diestrus), and intravaginally challenged with the HIV-1 R5 strain (NLR4.3-Bal-Env). Reconstitution levels ("humanization" levels, %hCD45 in peripheral blood 90-120 days following intrahepatic injection of CD34 enriched hematopoietic stem cells) of the DMPA and diestrus Hu-mice can be found in Supplemental File 1. The overall proportion of HIV-1 infection following intravaginal exposure in Hu-mice was 0/4 (0%) when Hu-mice were challenged during the estrus (estrogen high) phase of the estrous cycle, 15 www.nature.com/scientificreports/ during the diestrus (progesterone high) phase of the estrous cycle, and 33/55 (60.0%) when Hu-mice were challenged 1 week after receiving a subcutaneous dose of 2 mg of DMPA ( Fig. 2A). The number of infections was significantly affected by cycle/hormonal status (P = 0.013, Chi-square). Plasma viral titres were compared in Hu-mice that had been intravaginally challenged with HIV-1 during estrus (N = 4), diestrus (N = 11), and DMPA (N = 11), 3 weeks following exposure (Fig. 2B). Hu-mice challenged at estrus did not become infected and had no detectable viral titers. Hu-mice infected at diestrus or following administration of DMPA did not show any significant differences in viral shedding (6.6 × 10 4 ± 7.5 × 10 3 vs. 6.5 × 10 4 ± 2.3 × 10 4 HIV-1 copies/mL respectively; P = 0.12, Mann-Whitney). From this we concluded that DMPA treated Hu-mice have increased susceptibility to intravaginal (IVAG) HIV-1, but similar levels of viral replication in peripheral blood as Hu-mice infected during diestrus, at 3 weeks post-infection.
In order to confirm that DMPA increased the odds of infection following IVAG exposure to HIV-1, the association between HIV-1 infection in Hu-mice following HIV-1 exposure and covariables was estimated using a univariate (unadjusted) and multivariate (adjusted) generalized linear model (Table 1). In the univariate analysis Hu-mice treated with DMPA (OR 2.50, 95% CI 1.09-5.88; P = 0.032), or with > 10% circulating hCD45+ cells (OR 2.92, 95% CI 1.25-7.11; P = 0.015), or challenged with high viral dose (10 5 ) (OR 2.94, 95% CI 1.29-6.89; P = 0.011) were more likely to become infected following IVAG HIV-1 challenge. After adjusting for CD45 and viral dose, DMPA-treated Hu-mice were 3.25 times more likely to become infected with HIV-1 following IVAG exposure than those inoculated during diestrus (3.25 aOR, 1.28-8.86 CI; P = 0.016; N = 95; 55 DMPA, 40 diestrus) ( Table 1). The frequency of hCD45+ target cells in the plasma as well as the viral inoculation dose had already been demonstrated as a determinant of HIV-1 infection in our previous study (P = 0.01537) 15 . To ensure that peripheral blood target cell frequency was equivalent between the Hu-mice challenged during diestrus and those . Vaginal tissues were stained by immunofluorescence for the cell-cell adhesion molecule desmoglein-1 (green), and counterstained with DAPI (blue). Mice treated with DMPA appeared to have less desmoglein-1α staining as compared to those in estrus or diestrus. (D) We also assessed the murine vaginal barrier in a separate cohort of mice (N = 4/group) by administering a 4 kDa fluorescein isothiocyanate-dextran (FITC-dextran) dye into the vaginal lumen and measuring the fluorescence within the peripheral circulation at sacrifice. Significantly more FITC-dextran dye was quantified in the serum of mice that had received DMPA than those sacrificed during estrus or diestrus. (One-way ANOVA). Taken together DMPA negatively impacts vaginal celladhesion molecules, and enhances leakage from the vaginal lumen into the submucosa, ultimately impairing the integrity of the murine vaginal epithelial barrier. **P ≤ 0.01. ***P ≤ 0.001. Magnifications: 100× and inset is 400×. Data in D is presented as mean ± SEM. DMPA Depot-medroxyprogesterone acetate, L vaginal lumen. www.nature.com/scientificreports/ challenged post-DMPA the average reconstitution (circulating hCD45+ cells) was statistically compared between groups and was not found to be significantly different (10.8 ± 1.7% DMPA versus 12.5 ± 2.1% diestrus; P = 0.5223 Mann-Whitney). Furthermore, we observed the classical loss of HIV-1 target cells (hCD45+hCD3+) over time in our infected Hu-mice. This was observed in Hu-mice regardless of whether the Hu-mice had or had not received DMPA, and in those that received a high (10 5 ) or low (10 3 ) viral dose at challenge (Supplemental File 2). This decrease in frequency of hCD45+hCD3+ cells was not seen in Hu-mice that did not become infected following intravaginal HIV-1 challenge (Supplemental File 2), where in some mice an increase in frequency of these cells over time was observed, likely because of an increase in reconstitution in the absence of HIV-1 infection. Thus, in a direct comparison between Hu-mice under the influence of the synthetic progestin DMPA versus Hu-mice in the diestrus/progesterone high phase of the estrous cycle, DMPA was found to significantly enhance HIV-1 infection in humanized mice following IVAG exposure (3.25 aOR). This suggests that DMPA increases the risk of HIV-1 acquisition in Hu-mice even more than the endogenous progesterone in the estrous cycle does.   24 . Therefore, we next sought to determine if one of the mechanisms by which DMPA increased susceptibility to HIV-1 was by extending the period of susceptibility to infection. As above, Hu-mice were reconstituted with hCD34 hematopoietic stem cells as described in Methods and "humanization" (%hCD45+ cells in peripheral blood) was assessed 90-120 days later by flow cytometry (Supplemental Fig. 2; Supplemental File 1). We administered 2 mg subcutaneous DMPA and exposed Hu-mice IVAG to HIV-1 (NLR4.3-Bal-Env) at 2 (N = 4), 3 (N = 8), or 4 (N = 6) weeks post-administration of DMPA (Supplemental Fig. 1) to determine the length of time the mice remained susceptible to HIV-1. To ensure that the experiments were physiologically relevant with respect to hormone levels in serum, we quantified serum MPA, the active ingredient in DMPA, in the Hu-mice at the time of IVAG challenge (2, 3, and 4 weeks post-DMPA administration) (Fig. 3A), as in our prior study 14 .
Based on our titre data, we speculated that viral escape from the initial point of infection in the vaginal tract into the systemic circulation and peripheral tissues might be delayed in DMPA-treated Hu-mice, compared to mice infected during diestrus. As plasma titres were not significantly different between groups, we hypothesized that HIV-1 might escape the vaginal tract and disseminate more rapidly in DMPA-treated Hu-mice, perhaps due to increased HIV-1 target cells in blood. Therefore, we performed flow cytometry on the peripheral blood of uninfected Hu-mice to determine the effect of DMPA on activated, circulating HIV-1 target cells  www.nature.com/scientificreports/ (hCD45+hCD3+hCD4+hCCR5+) (Fig. 4E). When target cells in the peripheral blood of uninfected Hu-mice in diestrus (N = 6) were compared to uninfected Hu-mice 1 week (corresponds to the time we typically challenge Hu-mice with HIV-1; N = 6) and 4 weeks (corresponds to week 3 of infection, if the Hu-mice had been infected; N = 7) following DMPA, there were significantly more hCD45+hCD3+hCD4+hCCR5+ target cells in the peripheral blood of DMPA-treated Hu-mice at 4 weeks than 1 week post-DMPA or diestrus (66.9 ± 5.8% 4 weeks post-DMPA vs. 24.5 ± 6.6% 1 week post-DMPA versus 20.1 ± 4.0% diestrus; P < 0.0001 one-way ANOVA). This suggests that in the absence of HIV-1 infection, DMPA enhances circulating HIV-1 target cells in the peripheral blood 4 weeks after administration. In the presence of HIV-1 infection, we observed the classical decline in the proportion of hCD45+hCD3+hCD4+hCCR5+ HIV-1 target cells in the vagina and peripheral blood following HIV-1 infection (Supplemental Fig. 3). The proportion of hCD45+hCD3+hCD4+hCCR5+ cells in the vagina and peripheral blood were quantified by flow cytometry, and Hu-mice that had been intravaginally infected with HIV-1 had significantly less frequency of hCD45+hCD3+hCD4+hCCR5+ cells in the vagina and peripheral blood cells than uninfected Hu-mice (vagina: 12.0 ± 1. In order to further support our hypothesis that there is initially a delay in HIV-1 escape from vaginal tract into the systemic circulation in DMPA-treated Hu-mice versus those infected during diestrus we examined the extent of viral dissemination. Clinical real-time PCR for HIV-1 RNA was performed on homogenates (organ or tissue), and body fluids collected from Hu-mice infected during diestrus (N = 4) or following treatment with DMPA (N = 4) at 1 and 5 weeks post-infection, as described in Materials and Methods (Supplemental Fig. 1). Humice infected during diestrus had more extensive viral dissemination 1 week after infection than DMPA-treated Hu-mice (Supplemental Fig. 4A,B). However, by 5 weeks post-infection DMPA-treated Hu-mice had similar viral dissemination as Hu-mice infected during diestrus (Supplemental Fig. 4C,D), possibly due to the increased hCD45+hCD3+hCD4+hCCR5+ target cells we observed in the peripheral blood (Fig. 4E). Taken together, our results suggest that DMPA initially suppresses local HIV-1 titres in the vaginal tract, and delays systemic viral spread. However, in the peripheral blood, DMPA appears to increase activated target cells over time, which may allow for comparable viral levels to those Hu-mice infected during diestrus, by 5 weeks post-infection. www.nature.com/scientificreports/ 4 researchers who were blinded to treatment status. Researcher scores were translated into numerical scores according to the scoring continuum (Fig. 6E). DMPA treatment enhanced hCD3 ( Fig. 6A; P = 0.0001, one-way ANOVA), hCD4 ( Fig. 6B; P = 0.0036, oneway ANOVA), and hCD68 populations ( Fig. 6D; P = 0.0003, one-way ANOVA) by 4 weeks post-DMPA in the uninfected Hu-mouse vaginal tract, while hCD11c remained constant ( Fig. 6C; P = 0.0959, one-way ANOVA). However, the only HIV-1 target cell population that was significantly different between Hu-mice in diestrus and 1 week post-DMPA (time point corresponding to when we would typically challenge Hu-mice with HIV-1) was hCD68+ cells. This suggests that DMPA-treated Hu-mice may have been more susceptible to intravaginal infection with HIV-1 due to increased macrophages present in the vaginal mucosa at 1 week post-DMPA. The hCD3 and hCD4 populations are enhanced by DMPA, but this only occurs at 4 weeks post-DMPA (time point that would correspond to '3 weeks post-infection' if the Hu-mice had been infected). Thus, enhanced vaginal macrophages might explain why DMPA-treated mice are more susceptible to HIV-1 (1 week post-DMPA) as compared to Hu-mice in diestrus ( Fig. 2A, Table 1). Taken together we propose DMPA-treated Hu-mice may initially be more susceptible to HIV-1 as a result of an increase in the vaginal macrophage population.

Discussion
Herein we demonstrated that DMPA, a synthetic progestin, alters the integrity of the vaginal epithelium, and increases the proportion of Hu-mice infected with HIV-1 compared to Hu-mice challenged during diestrus (progesterone high phase of the estrous cycle). We also show that DMPA treatment increases the window of HIV-1 susceptibility since Hu-mice are continuously susceptible up to 4 weeks post-DMPA, while untreated Humice are only susceptible during diestrus, which represents about 50% of the murine estrous cycle. Interestingly, we found that at that 1 week post-DMPA; the time we would typically challenge Hu-mice with HIV-1, CD68+  www.nature.com/scientificreports/ with HIV-1 following DMPA and those infected during diestrus by 5 weeks post-infection. A tabular summary of our study results can be found in Table 2, and our experimental designs are presented in Supplemental Fig. 1. The female sex hormones estradiol and progesterone, and their synthetic derivatives, found in hormonal contraceptives, are associated with changes in susceptibility to STIs including HIV-1, likely via a variety of biological mechanisms (reviewed in 13 ). Typically, estradiol is associated with protection from HIV-1 in women, while progesterone and DMPA have been thought to increase the risk of infection 9,10,[25][26][27] , and similar phenomena are reported in non-human primate (NHP) model systems 5,6,8,28 . One of the mechanisms by which the female sex hormones are thought to modify the risk of HIV-1 infection is via the modification of target cells. Indeed, estradiol has been shown to reduce susceptibility of human peripheral blood CD4+ T cells and macrophages to HIV-1 29 in vitro. Interestingly, in the same study the protective effect of estradiol against HIV-1 infection was more pronounced in macrophages than in CD4+ T cells. In a humanized mouse model (NOD-scid-IL-2Rgc−/− mice with human peripheral blood mononuclear cells-no tissue reconstitution), different from the Hu-mouse model used in the present study (where we have tissue reconstitution with human immune cells 15 ), Quispe Calla et al. 30 demonstrated that when control mice were challenged during estrus (when estradiol is high) they were not susceptible to IVAG HIV-1, similar to our findings ( Fig. 2A). Their study also demonstrated that Hu-mice treated with DMPA were susceptible to IVAG infection, but could be protected when concurrently treated with estradiol 30 , suggesting that the presence of estradiol is necessary to protect against HIV-1 infection in Hu-mice. Additionally, our previous paper also demonstrated that DMPA enhanced the risk of HIV-1 infection in Hu-mice 14 , and herein we assessed the mechanisms by which this occurs. As DMPA is potently anti-estrogenic; women on DMPA are hypo-estrogenic (reviewed in 13 ), it is perhaps this lack of estrogen that increases the susceptibility of vaginal macrophages in our DMPA-treated mice to HIV-1 and results in an increased infection rate as compared to Hu-mice challenged during diestrus (where there would still be some estradiol in the circulation). Furthermore, we found that the integrity of the vaginal epithelial barrier was greatest in mice in estrus (when estradiol is highest), compared with those in diestrus or treated with DMPA (when estradiol is lower) (Fig. 1). This suggests that estradiol likely protects Hu-mice against HIV-1 by multiple mechanisms.
While estradiol is often found to protect against STIs, progesterone, and in particular DMPA, is generally thought to enhance the risk of HIV-1 acquisition. Indeed, meta-analyses have found women on DMPA are 40% more likely to acquire HIV-1 than women not on hormonal contraceptives 11 . Studies in NHPs show that DMPA enhances the risk of vaginal acquisition of SIV and increases viral titres during early infection [21][22][23] . Similarly, NHPs exposed to HIV-1 during the luteal phase of their menstrual cycle (progesterone high) are more susceptible to vaginal HIV-1 infection than those challenged during the follicular phase, when estrogen is high 5,31 . Here, we were interested in directly comparing and determining if Hu-mice treated with a synthetic progestin (DMPA) were more susceptible to HIV-1 as compared to Hu-mice in diestrus, with naturally elevated progesterone. Diestrus is the stage of the murine reproductive cycle when endogenous progesterone is highest. We found that indeed, Hu-mice challenged 1 week after treatment with 2 mg of DMPA were 3.25 times more susceptible to intravaginal infection with HIV-1 as compared to those challenged during diestrus. This study demonstrates for Table 2. A summary of the study results. An upward arrow (↑) indicates the factor of interest was significantly greater in that group (Diestrus vs. DMPA), while an equals sign (=) indicates groups were similar, and a downward arrow (↓) indicates the factor of interest was significantly lower.

HIV-1 infection ↓ ↑
Window of HIV-1 susceptibility ↓ ↑ www.nature.com/scientificreports/ the first time that DMPA increases susceptibility to HIV-1 in Hu-mice above the naturally occurring window of susceptibility present during the progesterone-high diestrus phase of the estrous cycle (50% of the estrous cycle). In order to determine the mechanisms by which DMPA enhances susceptibility to HIV-1 we examined the length of the window of vulnerability and prevalence of vaginal HIV-1 target cells. First, Hu-mice treated with DMPA were susceptible to intravaginal infection up to 4 weeks following a single 2 mg dose, while untreated Hu-mice were only susceptible to infection during diestrus. This suggests that one of the mechanisms by which DMPA enhances the risk of acquiring HIV-1 is because it causes an extended state of perpetual viral susceptibility. As compared to Hu-mice in diestrus, which were only susceptible during this phase of the estrous cycle (representing approximately 50% of the length of the mouse estrous cycle), Hu-mice treated with DMPA could be infected following viral exposure at any point in the 4 weeks following treatment. Second, DMPAtreated Hu-mice had a greater abundance of CD68+ macrophages, one of the types of HIV-1 target cells, in the vaginal mucosa as compared to Hu-mice in diestrus, suggesting that DMPA may initially enhance the vaginal macrophage population. This increase in macrophages might explain why DMPA-treated Hu-mice were 3.25× more likely to be infected following intravaginal challenge as compared to Hu-mice in diestrus. In fact, DMPA has previously been reported to enhance macrophage populations in the human vagina 32 and endometrium 33 .

HIV-1 titres
Similarly, DMPA has been demonstrated to significantly increase CD68+ macrophages in the ectocervix and vaginal mucosa of NHPs, even above levels observed during the luteal phase (progesterone high) of the menstrual cycle 8 , suggesting this may be the reason DMPA-treated Hu-mice are more susceptible to HIV-1 infection than those challenged during diestrus (progesterone high).
Although DMPA-treated Hu-mice were more susceptible to HIV-1 infection, unexpectedly they had significantly lower vaginal HIV-1 titres and initial viral dissemination was limited as compared to Hu-mice infected during diestrus, even though plasma HIV-1 titres were comparable. In addition to its anti-estrogenic effect, DMPA is also reported to have immunosuppressive effects, particularly at high doses in humans; however, the threshold below which immunosuppression no longer occurs may vary by individual and/or may also occur at physiologically relevant circulating concentrations of MPA (active ingredient in DMPA) (reviewed in 13 ). Unlike progesterone, which only binds progesterone receptors, DMPA binds both progesterone receptors and glucocorticoid receptors (GRs). Although interactions are complex and not easily explained, it is thought that the immunosuppressive effects of DMPA occur due to its ability to bind GRs and repress a multitude of proinflammatory pathways (reviewed in 13 ). In our previous study 14 we found that a 2 mg subcutaneous dose of DMPA in Hu-mice (same dose employed in the present study) consistently led to circulating levels of MPA similar to those observed in women; including both a peak and plateau phase. Thus, as DMPA can be immunosuppressive in women at physiologically relevant concentrations, it is likely that these effects would also be seen in Hu-mice, and this might explain why DMPA-treated Hu-mice had significantly lower vaginal HIV-1 titres and initially limited viral dissemination as compared to Hu-mice infected during diestrus. Viral replication in the blood (Fig. 4B,D), and dissemination during late infection (5 weeks post-infection) (Supplemental Fig. 4C,D) was similar between Hu-mice infected with HIV-1 following DMPA and those infected during diestrus, which might be explained by the fact that vaginal T cells (hCD3+, hCD4+) (Fig. 6A,B) and activated circulating HIV-1 target cells (hCD45+hCD3+hCD4+hCCR5+) (Fig. 4E) were greater in DMPA-treated Hu-mice (4 weeks post-DMPA-in the absence of HIV-1 infection) than Hu-mice in diestrus. Enhanced vaginal recruitment of target cells and enhanced target cells in the peripheral blood would allow viral spread in the DMPA-treated Hu-mice to catch up to that observed in diestrus infected Hu-mice. Together our results suggest DMPA initially suppresses local HIV-1 titres in the vaginal tract, and delays viral spread, while enhanced target cells in vaginal tract and the peripheral blood by 5 weeks post-infection allows dissemination to then proceed quickly, rendering it comparable to Hu-mice infected during diestrus.
Other Hu-mouse models have found that administration of DMPA results in loss of genital mucosal barrier function 19 , and enhanced genital transmission of cell-associated HIV (as compared to estradiol treated Humice). NHP models have demonstrated that DMPA-treated macaques have a greater density of HIV-1 target cells (CD4+ and CD68+ cells) in the ectocervix and vagina as compared to untreated animals, and that macaques in the late luteal phase (high endogenous progesterone) also had more vaginal HIV-1 target cells than those in the follicular phase and mid-cycle 8 . Furthermore, increased viral entry into the cervix and decreased vaginal epithelial thickness during progesterone/progestin dominant states were reported in these macaques 8 , and thinned vaginal epithelium and changes in vaginal pH in pigtail macaques treated with DMPA 17 , offering several additional mechanisms by which DMPA may impact susceptibility to SHIV in the NHP model. A thinned vaginal epithelium was also seen in macaques given subcutaneous progesterone implants, and the vaginal transmission of SIV was enhanced 7.7 fold in these animals compared to placebo implants or macaques challenged during the follicular phase of the menstrual cycle 28 . Unlike in our study, Marx et al. found plasma SIV RNA was elevated in progestin-treated macaques for 3 months following infection, and several of the progestin-treated animals developed rapidly spreading disease. Similarly, in a SHIV macaque model, acute viremia was higher in DMPA-treated animals than naïve animals, and the genetic complexity of the replicating virus was greater too 21 . This study also reported dampened cellular immune responses and an immunosuppressive effect of DMPA 21 , similar to what we observed in our Hu-mice. Although we did not see a difference in acute viremia between Hu-mice infected during diestrus versus those infected following DMPA, perhaps this is a reflection of the fact that our control group also had high endogenous progesterone, whilst controls in other studies may not have had been from the progesterone high phase of the cycle, or it could be an inherent difference between Hu-mouse models and NHPs.
In summary, we demonstrate the striking ability of the synthetic progestin DMPA to significantly increase intravaginal HIV-1 infection 3.25× beyond what was observed for endogenous progesterone in Hu-mice. We also found that the window of HIV-1 susceptibility was extended in DMPA-treated Hu-mice. Furthermore, vaginal HIV-1 target cells, in particular vaginal macrophages (hCD68+) are enhanced in DMPA-treated Hu-mice, and may thus play a previously underappreciated but important role in the vaginal acquisition of HIV-1. We know www.nature.com/scientificreports/ endometrial macrophages are increased in women on DMPA 33 , however the specific effect of DMPA on vaginal macrophages in women is lacking, but warranted based on our study in Hu-mice. We also found that DMPA initially suppresses local HIV-1 titres in the vaginal tract, and delays viral spread, while enhanced activated target cells in the peripheral blood may allow for viral dissemination to catch up by 5 weeks post-infection. Taken together, results suggest DMPA may enhance susceptibility to HIV-1 in Hu-mice by increasing vaginal target cells (including macrophages), and extending the period of time during which Hu-mice were susceptible to infection; mechanisms that might also affect HIV-1 susceptibility in women.  were generated as previously described 15 . Briefly, umbilical cord blood from healthy newborns born at McMaster University Medical Centre was obtained following parental consent, processed by RosetteSep Human Cord Blood Progenitor Enrichment Kit (Stem Cell Technologies, Vancouver, BC, Canada), and cryopreserved until required. Four day old NOD-Rag1 tm1Mom Il2rg tm1Wjl (NRG) mice were sub-lethally irradiated (γ-ray, 300 cGy) and given an intrahepatic injection with 1 × 10 6 to 2 × 10 6 CD34-enriched hematopoietic stem cells. The percent human immune cell reconstitution was evaluated by flow cytometry 90-120 days following injection by quantifying expression of hCD45, hCD3, hCD4, as well as mouse CD45 in peripheral blood mononuclear cells (PBMCs), as in Nguyen et al., 2017. These Hu-mice have engraftment of human cells in peripheral blood, and body tissues 15 . The average human immune cell reconstitution of the Hu-mice used in the experiments described herein was 12% hCD45+ cells in the peripheral blood (as shown in the first panel of gating strategy in Supplemental Fig. 1, with % hCD45 shown in the first gate). The Hu-mice used in the experiments herein were between 3 and 6 months of age. The reconstitution levels, HIV-1 dose, peripheral blood (PB) titres, vaginal lavage (VL) titres, and infection status for Hu-mice included in Table 1  The Bond Polymer Refine kit (Leica Biosystems) was employed for CD3 (post-primary step excluded) and CD4 staining on the Bond RX automated slide stainer (Leica Biosystems), while rabbit and mouse Vectastain ABC kits (Vector Labs, Burlington, ON, Canada) were employed for CD11c and CD68. All slides were stained with DAB as a chromogen, and counterstained with hematoxylin. Images were visualized using an Olympus IX81 inverted microscope (Olympus, Richmond Hill, ON, Canada), and captured using the Infinity camera (Lumenera Corp., Ottawa, ON, Canada).

Statistical analyses.
To assess the relationship between DMPA and HIV-1 susceptibility in humanized mice ( www.nature.com/scientificreports/ stats and lme4 package in R version 3.2.3 41 . In the univariate model, the odds ratio (OR) and 95% confidence interval (CI) were calculated for each covariate (CD45 > 10%, and viral dose (high)). In the multivariate model, the adjusted odds ratios (aOR) and 95% CIs were estimated after adjusting for covariates. In the multivariate model the adjusted regression coefficients (β) and 95% CIs were calculated. For all statistical models, a P < 0.05 was considered statistically significant. Data were analyzed and graphed using GraphPad Prism 6 (GraphPad Software, La Jolla, CA). Data were considered statistically significant if the P values obtained with a t-test, analysis variance of the mean (ANOVA), or the equivalent non-parametric tests if the data was not normally distributed, were < 0.05. Significant differences are noted as *P < 0.05, **P < 0.01, ***P < 0.001, or ****P < 0.0001.