This study was approved by the Committee for the Protection of Human Subjects (CPHS) and by the Health Sciences Institutional Review Board at Tufts University.

Genital tract tissue samples were obtained after surgery from HIV-negative women undergoing hysterectomies at Tufts Medical Center (Boston, MA, USA). Surgery was performed to treat benign conditions including fibroids, prolapse and menorrhagia. Trained pathologists selected tissue samples from ectocervix (ECX), endocervix (END), and endometrium (EM), free of pathological lesions and distant from the sites of pathology. Women were classified in this study as premenopausal (n=33; 24-55 years old; age (average)=42) and postmenopausal (n=9; 49-77 years old; age (average)=62). Menopausal status was self-identified.

Human blood samples were obtained from healthy female donors through Research Blood Components, LLC (Watertown, MA, USA). Women signed an Institutional Review Board (IRB) approved consent form where they gave permission to Research Blood Components, LLC to collect and sell their blood for research purposes. The only information provided about the donors was their age and menopausal status. Women were classified in this study as premenopausal women (n=42; 18-51 years old; age (average)=34) and as postmenopausal women (n=29; 48-72 years old; age (average)=60).

Tissues were processed to obtain a stromal cell suspension as described previously (16, 37, 39–41), with modifications. Enzymatic digestion was performed in gentleMACS™ Dissociator (Miltenyi Biotec, Auburn, CA, USA) using Tumor Dissociation Kit, human (Miltenyi Biotec) and 0.01% DNAse (Worthington Biochemical, Lakewood, NJ, USA). After filtering through a 100 μm, 70 μm, and 30 μm filters (MACS^® SmartStrainers; Miltenyi Biotec) to separate epithelial cells from stromal cells, cell suspensions underwent red blood cell removal (CD235a (Glycophorin A) MicroBeads, human; (Miltenyi Biotec)). Mixed cell suspensions were used for NET release experiments as indicated or underwent further processing to purify neutrophils. To isolate genital neutrophils, mixed cell suspensions were incubated first with CD66b Antibody, anti-human, PE, REAfinity™ (Miltenyi Biotec) followed by anti-PE MicroBeads UltraPure (Miltenyi Biotec) to be magnetically selected following the manufacturer’s instructions (Miltenyi Biotec) as described (37, 42).

Venous blood from healthy women was collected into 10 mL EDTA tubes. Neutrophil isolation was performed as described (37, 38) by positive selection using CD15 MicroBeads (Miltenyi Biotec) and a whole blood column kit following the manufacturer’s instructions (Miltenyi Biotec).

Human purified neutrophils and mixed cell suspensions from genital tissues were resuspended in Hanks’ Balanced Salt Solution (HBSS) culture medium (Gibco; Waltham, MA, USA), plated in a 96-well plate (Corning Inc.; Corning, NY, USA) and pre-incubated at room temperature with the following inhibitors: TLR8 inhibitor at 5 µM for 1 h (CU-CPT9a; InvivoGen, San Diego, CA, USA), TBK1/IKKϵ inhibitor at 100 nM for 1 h (BX795, InvivoGen), TLR9 inhibitor at 150 nM for 30 min (ODN TTAGGG (A151), InvivoGen), TLR7 inhibitor at 150 nM for 30 min (ODN 2088 Control (ODN 20958), Miltenyi Biotec), TLR7/9 inhibitor at 150 nM for 30 min (Dual iODN; EnzoLifeSciences, Farmingdale, NY, USA), Dynamin inhibitor (Dynasore; Abcam, Waltham, MA, USA) at 200 µM for 20 min. Every inhibitor had its own control condition matching the media used to resuspend the inhibitors: DMSO, buffer solutions supplied with inhibitors or HBSS.

Following specific treatments, purified neutrophils from blood or genital tissues and mixed cell suspension were stimulated with 5000 GFP-labeled HIV-viral like particles (HIV-VLPs) per cell or calcium ionophore at 25 µM (A23187; Sigma-Aldrich, Saint Louis, MO, USA).

Extracellular DNA was quantified to determine the NET-HIV area with the IncuCyte software as described (37, 38). Cytotox red reagent at 250 nM (Essen Bioscience; Ann Arbor, MI, USA) was used to label DNA as described and annexin V dye at 1:200 dilution (IncuCyte^® Annexin V Dye for Apoptosis; Sartorius, Bohemia, NY, USA) was used to detect annexin V+ NETs. Images were collected every 3–5 min at 37°C using a 10x objective with the IncuCyte S3 (Sartorius) and NETs quantified as described (37, 38). Briefly, a mask was applied to determine colocalization of extracellular red fluorescent signal (DNA) and green fluorescent signal (HIV-VLPs). Intracellular signal within round small objects was excluded from the NET analysis. Because images were taken with a 10x magnification objective, single individual HIV-VLPs were undetectable. HIV-VLP fluorescent signal became detectable when multiple particles accumulated in NETs (37).

Cell membrane steadiness was assessed with Cytotox red reagent (Essen Bioscience). Positive cells were quantified using a mask that specifically selects round and small objects (cells).

Rhod-3 AM (Rhod-3 Calcium Imaging Kit; Thermo Fisher, Waltham, MA, USA) was used to stain cytosolic calcium in red following manufacturer’s instructions. As an internal control of the positive signal, we used a calcium chelator (BAPTA, AM; Thermo Fisher) at 10 µM for 30 min at room temperature. To quantify the intracellular calcium area, we applied a mask to select intracellular red staining.

CellROX (CellROX™ Deep Red Reagent, Thermo Fisher) was used to stain in red cellular oxidative stress by reacting with ROS. CellROX dye is non-fluorescent when it is reduced but exhibits fluorescence once it is oxidated by ROS. As an internal control of the positive signal, we used a NADPH oxidase inhibitor (DPI; Sigma-Aldrich) at 10 µM for 30 min at room temperature. To quantify the intracellular ROS area, we applied a mask to select intracellular red staining.

Neutrophils were stimulated with CCR5-tropic GFP-labeled viral-like particles (VLPs) as described before (37, 38). GFP-labeled VLPs were generated using a modified pNL43 provirus-based plasmid for expression of GFP as described previously (43). The enhanced GFP (EGFP) coding sequence was expressed in the frame at the 3’end of the gag, replacing the protease and most of the reverse transcriptase coding region. The Ψ-signal on the RNA and the complete gag open reading frame (ORF) remained intact. To obtain CCR5 tropic strains (R5 strains), a plasmid with an inactivated Env ORF, resulting in no expression of functional Env protein (referred to as pNL4GagGSGFPDelta-env/K806), was derived from K795 for pseudotyping and then complemented with pBaL.26 Env expression plasmid (NIH AIDS Reagent program, catalog number 11,446, contributed by Dr. John Mascola) (44), generating VLPs with HIV-BaL envelope proteins (R5). For some experiments, VLP lacking envelope proteins (Delta-env) were used as indicated. Non-infectious, EGFP-labelled VLPs were produced by transfection, concentrated by ultracentrifugation, and enumerated essentially as described (43).

Cells were resuspended in HBSS and plated on coverslips pre-coated with poly-D-lysine (Gibco) for 20 min at room temperature. GFP-labeled HIV-VLPs (5,000-10,000 VLP/cell) were used to stimulate NET release for 5 min, 15 min, 30 min, 1h or 2 h at room temperature and then fixed with 4% paraformaldehyde. Unstimulated neutrophils incubated in HBSS were used as controls. Cells were stained for 2 h at room temperature with AlexaFluor 647-H3 (1B1-B2) (BioLegend) AlexaFluor 647-Rab5 (D-11) (Santa Cruz Biotechnology, Dallas, TX, USA), AlexaFluor 647-Lactoferrin (LF5-1D2) (Novusbio, Englewood, CO, USA), and/or PE-Annexin V (BioLegend, San Diego, CA, USA). Cells were carefully washed with PBS, and samples were mounted in Pro-Long Diamond anti-fade mounting with DAPI (Thermo Fisher) as described (37). Samples were imaged using a Leica SP8 confocal microscope in combination with Leica LAS X software.

To determine neutrophil abundance in the FGT, mixed cell suspensions from genital tissues were stained for surface markers with combinations of the following antibodies: CD45-vioblue450, CD11b-PE (Tonbo, San Diego, CA), CD3-VioGreen, CD15-FITC, CD66b-APC (Miltenyi biotec), CD11c-PerCp-Cy5.5, CD3-APC (Biolegend), CD14-e780, (eBiosciences, San Diego, CA). Dead cells were excluded with 7AAD (Southern Biotech, Birmingham, AL, USA) or zombie dye yellow staining (Biolegend). Analysis was performed on 8-color MACSQuant 10 (Miltenyi biotec) or Gallios (Beckman Coulter, Indianapolis, IN, USA) flow cytometers and data analyzed with FlowJo software (Tree Star, Inc. Ashland, OR, USA). Expression of surface markers is shown as percentage of positive cells. Fluorescence minus one (FMO) strategy was used to establish appropriate gates. Cells were gated as described before (37).

To determine HIV receptor expression on neutrophils, purified blood neutrophils and mixed cell suspension from genital tissues were stained with the following anti-human antibodies: CD45-BUV395 (BD Biosciences), CD66b-Pacific blue (BioLegend), CD15-FITC (BioLegend), CD4-BUV805 (BD Biosciences), CXCR4-BV785 (BioLegend) and CCR5-PE-Cy5 (clone: 2D7/CCR5) (BD Biosciences). Cell viability was assessed with a live/dead fixable blue dead cell stain kit (Thermo Fisher) used for 10 min in a 1:200 dilution. Cells were incubated with FcR Blocking Reagent (Miltenyi biotec) for 10 min (5µl/1M cells) to block Fc receptors. Analysis was performed on Aurora cytometer (Cytek Biosciences; Fremont, CA, USA) and assessed with OMIQ (www.omiq.ai).

Invitrogen™ Estradiol Human ELISA Kit (Thermo Fisher) was used to determine estradiol levels in plasma from blood donors following manufacturer’s instructions. Plasma samples were obtained from blood (1 ml) after centrifugation at 500g/15min/RT.

Human blood neutrophils were processed using RNeasy Plus Mini Kit (Qiagen; Hilden, Germany) to extract and purify RNA. Next, high-quality RNA was obtained by removing contaminants using RNeasy cleanup kit (Qiagen). TLR7, TLR8, TLR9, and RPL13a (as internal control gene) genes were amplified using PowerSYBR™ Green PCR Master Mix (Thermo Fisher) and the following primer pairs: TLR7, hFw 5’tttcccagagcatacagcttag3’ and hRv 5’gcctctgatgggacaagtaaa3’; TLR8, hFw 5’gctacaggtctctttccacatc3’ and hRv 5’gtggtagcgcagctcattta3’; TLR9, hFw 5’gctagacctgtcccacaataag3’ and hRv 5’aaagggctggctgttgtag3’; RPL13a, hFw 5’gccctacgacaagaaaaagcg3’ and hRv 5’tacttccagccaacctcgtga3’. Samples were run on the QuantStudio 6 Flex System. All kits were used according to manufacturer’s instructions.

Data analysis was performed using the GraphPad Prism 9 software. Data are represented as median ± interquartile range. One and two-sided p-value ≤ 0.05 were considered statistically significant as indicated. Non-parametric Mann–Whitney U test or Wilcoxon’s matched pair test was used for comparison of two groups. One sample t test was used to determine whether an unknown population is different from a hypothetical value, assuming sampling from a Gaussian distribution. Non-parametric Spearman’s correlation was used for correlation analysis.

Previously, we and others have demonstrated that neutrophils from blood and genital tissues release NETs in response to HIV (36–38). Here we wanted to investigate the impact of aging on HIV-induced NET release in blood and genital tissues. To do this, we stimulated blood and genital neutrophils from women (18-77 years of age) with GFP-labeled HIV viral like particles (VLPs) expressing CCR5-tropic envelope proteins (HIV-BaL) to visualize and quantify NET-release using time-lapse microscopy as described previously (37, 38).

Consistent with our previous findings (37, 38), we detected NET release after stimulation of blood neutrophils with GFP-labeled HIV-VLPs compared to control condition ( Figure 1A ). To further confirm the presence of HIV-NET complexes by an alternative method, HIV-NETs were visualized using confocal microscopy. Neutrophils were plated on glass slides and stimulated with GFP-labeled HIV-VLPs to induce the release of NETs, followed by staining of DNA and histone 3 (H3), a NET-associated protein (20). As shown in Figure 1B , we observed colocalization of HIV, extracellular DNA and H3, indicating HIV entrapment in NETs. Interestingly, a reduction in NET formation was visualized in postmenopausal when compared to premenopausal women ( Figure 1A ). Kinetic quantification of NET release by blood neutrophils overtime from premenopausal women identified a rapid initial peak of NET release (early NET release; 0-15min), followed by sustained release over the next 2h (late NET release; 15min-2h) ( Figure 1C ; black circles). In contrast, the kinetics of NET release in postmenopausal women revealed the absence of an initial peak of NET release following HIV stimulation which progressively increased over time ( Figure 1C ; white circles). Quantification of total NET release in multiple women (n=37) at 2h after HIV stimulation did not detect any differences between pre and postmenopausal women ( Figure 1D ; 0-2h); however, we detected significantly lower release of NETs in postmenopausal compared to premenopausal women at early time points ( Figure 1D ; 0-15min, p=0.02), with no significant differences at later time-points after HIV stimulation ( Figure 1D ; 15-2h). These results suggest a delay in the response of blood neutrophils to the virus in postmenopausal women compared to premenopausal women. Recognizing that menopause is a critical point in the aging process for women that marks the end of sex hormones cycles (18), we next investigated whether the observed effects in early NET release were influenced by aging following menopause. As seen in Figure 1E , during the premenopausal years, early NET release was maintained as women aged ( Figure 1E ; black circles), but significantly declined with aging following menopause ( Figure 1E ; white circles, r=-0.53, p=0.02).

To determine if this decline in NET release was directly influenced by sex hormones, we measured estrogen (E2) levels in plasma samples from 24 premenopausal and 25 postmenopausal women, including 28 samples used for NET release experiments ( Figure S1 ). As expected, premenopausal women showed significantly higher levels of E2 in plasma compared to postmenopausal women ( Figure S1A ). Next, we analyzed pre and postmenopausal women groups independently to identify correlations between E2 levels and age in our cohort. For premenopausal women, we observed no significant correlation between the age of premenopausal women and their E2 levels ( Figure S1B ). This is consistent with the known high variability between women, and between cycles for the same woman as reported previously (11, 45–48). Additionally, we noticed a broad variability in the E2 levels in the 40-50-year-old range, with lower levels after 45. In the postmenopausal group, there was also no significant correlation between age and plasma E2 levels, which remained low and constant ( Figure S1B ). Importantly, when we correlated E2 plasma levels from every donor with HIV-NET release by purified blood neutrophils, there was no significant correlation ( Figure S1C ), indicating that the decrease in E2 levels in postmenopausal women is not directly influencing the release of HIV-induced NETs.

Next, we investigated NET release by genital neutrophils ( Figure 1F ). Similar to blood neutrophils, we visualized a general reduction in NET release between younger and older women. However, in contrast to our results with blood neutrophils, when NET release from multiple women was quantified, we did not detect a significant difference in NET release between pre and postmenopausal women, whether total, early or late NET release ( Figure 1G ). However, when NET release was correlated with age, genital neutrophils from postmenopausal women significantly decreased the early NET release progressively as women aged ( Figure 1H , white circles; r=-0.66, p=0.002). To determine if aging differentially affected neutrophils from different anatomical parts of the FGT, we stratified the results in Figure 1H according to FGT site (endometrium and cervix) in pre and postmenopausal women. As seen in Figure 1I , endometrium and cervix neutrophils displayed similar correlation between early NET release and aging in women ( Figure 1I ). Unexpectedly, NET release by endometrial neutrophils from premenopausal women significantly increased with the women’s age until they reached the perimenopausal years ( Figure 1I ; left graph), and the same trend was observed in the cervix, likely contributing to the lack of differences detected between pre and postmenopausal women when considered as a whole ( Figure 1G ). To determine if the observed correlation between reduced NET release and increasing age was associated with a decrease in neutrophil presence in the FGT, we measured the number of neutrophils and percentage of neutrophils in genital tissues by flow cytometry as described before (37). As seen in Figure 1J , neutrophil presence and percentage remained constant with increasing age in the different anatomical sites of the FGT.

Overall, these results demonstrate that aging impacts the ability of neutrophils from blood and the FGT to release NETs in response to HIV.

Since neutrophils from postmenopausal women demonstrated a defect in the early NET release response following HIV stimulation, we next investigated HIV recognition by neutrophils to determine if this mechanism was impaired with aging.

HIV ssRNA is recognized by TLR7/8 (49) and stimulation of these TLRs results in neutrophil activation and NET release (36, 37, 50). Therefore, we investigated the role of TLR7/8 in HIV recognition by blood and genital neutrophils and their involvement in early NET release.

To do this, we incubated blood neutrophils with specific TLR inhibitors prior to HIV stimulation and quantified NET formation overtime. In blood neutrophils, as shown in a representative example, TLR8 inhibition (CU-CPT9a) delayed the kinetics of NET release compared to HIV alone ( Figure 2A ; Supplementary Movie 1 ), while inhibition of TLR7 and TLR9 (TLR7/9: Dual iODN) impacted NET release at later time points ( Figure 2A ; Supplementary Movie 2 ). Quantification of HIV-NET release by blood neutrophils from multiple women after blocking TLR8 with CU-CPT9a demonstrated significant reduction (45% median reduction; p=0.03) at early time points (0-15 min), with no significant changes at later time points (15min-2h) ( Figure 2B , left panel). In contrast, TLR7/9 inhibition with Dual iODN trended to reduce the release of NETs at late time points ( Figure 2B , right panel). Because Dual iODN inhibits TLR7 and TLR9, to determine whether the observed effect was due to TLR7 or TLR9 signaling, we used specific TLR7 (ODN 20958) or TLR9 (ODN TTAGGG) inhibitors and observed a significant reduction in late NET release after TLR7 inhibition ( Figure 2C ; 59% median reduction; p=0.007), but no effect with TLR9 inhibition ( Figure 2C ). We also tested the contribution of additional pattern recognition receptors (PRRs) able to detect viral RNA by inhibiting TANK-binding kinase 1 (TBK1) and I-kappa-B kinase epsilon (IKKϵ) before HIV stimulation, two IκB kinase homologs which are indispensable for the signaling downstream of RIG-I-like receptors (RLRs) (51). However, we did not detect significant differences in the release of NETs ( Figure 2C , right panel). Our findings indicate that blood neutrophil recognition of HIV through TLR8 mediates early NET release, while TLR7 preferentially mediates late NET release.

To evaluate the effects of menopause and aging on HIV recognition and NET release by blood neutrophils, we stratified the data in Figure 2B into pre and postmenopausal women. We hypothesized that HIV recognition would be impaired in postmenopausal women as a result of reduced TLR signaling and this would be responsible for the delay in HIV-induced NET release (Figure1). Unexpectedly, inhibition of TLR8 and TLR7/9 significantly reduced NET release in the postmenopausal group, with no significant effect in premenopausal women ( Figure 2D ). Furthermore, normalization of HIV-induced NET release in the presence of the TLR inhibitor versus in the presence of HIV only (control condition) revealed that the percentage of NET reduction was enhanced in postmenopausal compared to premenopausal women ( Figure 2E ). We next assessed TLR7, TLR8 and TLR9 expression to determine if the observed differences in endosomal TLR signaling between pre and postmenopausal women were due to age-dependent changes in TLR expression. However, we did not detect differences between pre and postmenopausal women, and TLR8 was expressed at higher levels than the other TLRs ( Figure S2 ), consistent with prior publications (52).

Overall, these data indicate that blood neutrophil recognition of HIV through TLR8 mediates early NET release, while TLR7 preferentially mediates late NET release. Further, TLR8 and TLR7 signaling preferentially mediates HIV-induced NET-release after menopause, suggesting that premenopausal women possess alternative mechanisms responsible for triggering NET release in response to HIV stimulation.

Next, we analyzed genital neutrophils. TLR8 inhibition significantly reduced early NET release (28.2% median reduction; p=0.005) ( Figure 2F , left panel) but to a lesser extent than in blood neutrophils ( Figure 2B , left panel). TLR7/9 and TLR7 inhibition reduced late NET release (48% median reduction; p=0.002) ( Figure 2F ). In contrast to blood neutrophils, when pre and postmenopausal women were analyzed separately, genital neutrophils from premenopausal women reduced the release of NETs in response to HIV when TLR8 was inhibited, but no effect of inhibition was detected in postmenopausal women ( Figure 2G , left panel), potentially due to the already low response to HIV alone. Similarly, NET reduction was detected after TLR7/9 inhibition only in premenopausal women ( Figure 2G , right panel). Normalization of HIV-induced NET release in the presence of the TLR inhibitor respect to HIV only demonstrated a reduction in the effect of TLR8 inhibition as women aged, while TLR7 inhibition remained equally effective in pre and postmenopausal women treated with Dual iODN ( Figure 2H ). These results suggest that HIV-induced NET release in genital neutrophils is preferentially induced through TLR8 and TLR7 in a sequential manner, and that TLR8 signaling mechanisms are impaired as women age.

Overall, these data demonstrate that HIV triggers NET release in blood and genital neutrophils through differential mechanisms and that aging specifically impairs TLR8 signaling in the FGT.

The lack of NET reduction after inhibition of TLR8 and TLR7 signaling in HIV-stimulated blood neutrophils from younger women suggested the presence of alternative NET-triggering mechanisms. Furthermore, although TLR8 and TLR7 inhibition resulted in a significant reduction of early and late NET release in blood neutrophils from older women, NET formation was not completely abrogated, suggesting the existence of additional activation mechanisms. Thus, we explored the potential role of intracellular calcium, which has been described as a rapid mediator of NET release in response to bacteria (33–35).

Blood neutrophils were pre-incubated with red fluorescent Rhod-3 AM dye to measure intracellular calcium levels with and without HIV stimulation. As seen in Figure 3A and Supplementary Movie 3 , intracellular calcium fluorescence signal was increased after HIV stimulation compared to control conditions. Kinetic quantification showed a very rapid increase in intracellular calcium within 5 min of HIV treatment, which peaked approximately 20 min after stimulation and was then maintained ( Figure 3B ). As an additional control to confirm that HIV was responsible for the intracellular calcium increase detected, we incubated blood neutrophils with BAPTA, an intracellular calcium chelator, prior to HIV stimulation. As seen in Figure 3C , BAPTA treatment significantly reduced intracellular calcium levels in the absence of HIV stimulation and abrogated the intracellular calcium response to HIV.

Next, we compared blood neutrophils from pre and postmenopausal women. Neutrophils from premenopausal women significantly upregulated intracellular calcium 5 min after stimulation and this difference remained significant at 15 min ( Figure 3D ). In contrast, neutrophils from postmenopausal women did not significantly upregulate intracellular calcium until 15 min following stimulation, with the overall amount of intracellular calcium lower in post compared to premenopausal women ( Figure 3D ). We then investigated whether intracellular calcium levels after HIV stimulation were affected by aging following menopause and detected a negative and significant correlation between the fold-change in intracellular calcium in response to HIV stimulation and women’s age ( Figure 3E ; r=-0.44; p=0.04).

To determine if intracellular calcium levels were associated with early NET release, we quantified NET release at 15 min and tested for correlation of NET-HIV area with the fold-change in intracellular calcium following HIV stimulation. As seen in Figure 3F , there was a positive and significant correlation between early NET release and intracellular calcium responses to HIV (r=0.54; p=0.03). This data suggests that intracellular calcium levels mediate early NET release in younger women. To test this, we quantified NET release in the presence of BAPTA in pre and postmenopausal women and detected a significant reduction in early NET release after HIV stimulation in premenopausal women ( Figure 3G ; 53% median reduction at 15 min), but a delayed effect in reduction of NET release in postmenopausal women ( Figure 3G ; 59% median reduction at 2h). These data demonstrate that early NET release in blood neutrophils from premenopausal women is triggered by intracellular calcium increase, but this response is delayed in postmenopausal women.

To determine if increased intracellular calcium levels in response to HIV stimulation were a result of HIV recognition by endosomal TLRs, cytosolic calcium levels were measured in blood neutrophils treated with endosomal TLR inhibitors prior to HIV stimulation. Interestingly, we did not observe any changes in intracellular calcium when TLR7/9 (Dual iODN), TLR7 alone (ODN 20958) or TLR8 (CU-CPT9a) were blocked prior to HIV stimulation ( Figure 3H ), indicating that calcium release was not triggered by endosomal viral recognition. We then explored whether the cytosolic calcium response was triggered by viral engagement of plasma membrane receptors by comparing HIV-VLPs with and without viral envelope (Env) proteins. As seen in Figure 3I , intracellular calcium was increased in response to HIV-VLPs with Env, and no changes were observed in the absence of Env proteins (Δenv), indicating that the intracellular calcium response was driven by the HIV Env, and not by the viral RNA genome.

Next, we investigated genital neutrophils. To evaluate calcium responses, we purified genital neutrophils as previously described (37) prior to HIV stimulation and quantification of intracellular calcium. As seen in Figure 3J and Supplementary Movie 4 , low levels of intracellular calcium were detected in genital neutrophils following HIV stimulation. Kinetic quantification demonstrated the lack of an intracellular calcium response compared to unstimulated control condition ( Figure 3K ). Analysis of multiple donors (n=4) did not detect intracellular calcium increase after HIV stimulation ( Figure 3L ). Importantly, although HIV stimulation did not increase calcium levels in genital neutrophils, these neutrophils were able to release NETs ( Figure 3J ; green), but no correlation was found between NET release and intracellular calcium levels ( Figure 3M ). Furthermore, BAPTA did not reduce the NET release ( Figure 3N ), indicating that calcium is not a mediator for the release of NETs by genital neutrophils in response to HIV stimulation.

Overall, these findings indicate fundamentally different mechanisms for NET release between blood and genital neutrophils in response to HIV, with increase of intracellular calcium triggered only in blood neutrophils after recognition of the HIV envelope and in an age-dependent manner.

We next investigated the production of ROS, which has been implicated in NET release in response to HIV stimulation (36, 37), as an additional possible mediator of the observed effects.

Blood neutrophils were incubated with CellROX Deep Red, a cell-permeant dye that becomes fluorescent when oxidized by ROS, to visualize and measure intracellular ROS production with and without HIV stimulation. As seen in Figure 4A and in Supplementary Movie 5 , HIV stimulation enhanced intracellular ROS production in HIV-stimulated neutrophils compared to resting neutrophils. Next, we quantified intracellular ROS production in pre and postmenopausal women after stimulation with HIV-VLPs with and without Env. Quantification of intracellular ROS revealed increased production after 1 h of HIV stimulation, but only with HIV-VLPs containing Env glycoproteins ( Figures 4B, C ), and this response was impaired in postmenopausal women ( Figures 4B, C ). Since intracellular ROS production and calcium increase seemed to be dependent on engagement of viral Env proteins, we next investigated whether calcium increase could be triggering intracellular ROS production in response to HIV stimulation. Thus, we measured intracellular ROS in neutrophils that were pre-incubated with BAPTA, and we used DPI (a NADPH oxidase inhibitor) as a positive control to inhibit ROS production. Interestingly, we found that BAPTA significantly reduced intracellular ROS levels when neutrophils were stimulated with HIV, and this reduction was similar in magnitude to the reduction obtained after inhibition of NADPH oxidase using DPI ( Figure 4D ). This suggests that intracellular calcium increase triggers the production of ROS. Finally, to determine if HIV recognition through endosomal TLRs led to NADPH oxidase activation and ROS production, we measured intracellular ROS in blood neutrophils treated with TLR7 inhibitor (ODN20958) and TLR8 inhibitor (CU-CPT9a). Only TLR7 inhibition reduced ROS production in response to HIV stimulation in blood neutrophils from postmenopausal women, but no effect was detected in premenopausal women or in women of all ages after TLR8 blockade ( Figure 4E ). These results indicate that in blood neutrophils HIV-induced calcium increase triggers ROS production in a TLR-independent manner preferentially in premenopausal women, while in postmenopausal women ROS production is preferentially induced by HIV-induced signaling through TLR7.

Next, we investigated genital neutrophils. To evaluate ROS responses, we purified genital neutrophils as described (37) and we quantify intracellular ROS with and without HIV stimulation. As visualized in Figure 4F , genital neutrophils enhanced intracellular ROS production after HIV stimulation compared to resting neutrophils, and ROS production was initiated at 30 min ( Figure 4G ), earlier than the response detected for blood neutrophils ( Figure 4B ). Further, in contrast to blood neutrophils, ROS production was not dependent on viral envelope stimulation since no differences were found after stimulation with HIV-VLP with or without the envelope ( Figure 4G ), suggesting that ROS production was induced after endosomal TLR activation.

Overall, these findings further support differential signaling pathways after HIV stimulation in blood and tissue neutrophils, with blood neutrophils responding to HIV viral Env proteins with upregulation of calcium and ROS production in an age-dependent manner, and genital neutrophils responding with rapid ROS production in a calcium-independent and HIV-Env independent manner.

In light of our results demonstrating that HIV can activate membrane receptors and also endosomal TLRs in neutrophils, we next investigated potential mechanisms for viral internalization.

Neutrophils are not considered as HIV-target cells, although the presence of a subset of peripheral blood neutrophils that express CD4 has been reported in some individuals (53, 54). Thus, we first determined if blood and genital neutrophils express the HIV receptor CD4, and coreceptors CCR5 and CXCR4. As shown in a representative example in Figure 5A , and with multiple patients (blood: n=11; tissue samples: n=22) ( Figure 5B ), we did not detect CD4 expression in blood neutrophils, and CXCR4 and CCR5 expression were detected in less than 1% of cells. In contrast, in genital neutrophils ( Figures 5A, B ), HIV receptor (CD4) was detectable in a small percentage of cells (2.2%) and coreceptors (CCR5 and CXCR4) were expressed in 2.6 and 10.1% of the neutrophil population, respectively. This demonstrates that the expression profile for HIV receptors is different between blood and genital neutrophils, with higher expression in the genital mucosa.

Next, we investigated HIV internalization through endocytosis. We first measured the expression of Rab5 after HIV stimulation. Rab5 is involved in the formation of the early endosome (55), which has been reported to be important for HIV endocytosis in cells with low or absent expression of HIV receptors (56). Fluorescence microscopy revealed Rab5 expression in a small number of blood neutrophils, which co-stained with HIV-GFP ( Figures 5C, D ). We next visualized Rab5 in genital tissues and identified a small number of genital neutrophils expressing Rab5 ( Figure 5E ). Quantification of Rab5 positive cells in multiple fields from multiple patients (blood: n=6; FGT: n=6) revealed a significant increase in Rab5+ cells after HIV stimulation in blood neutrophils (at 5 min) and in genital mixed cell suspension (at 15 min) ( Figure 5F ).

To determine if viral endocytosis was involved in NET release, we blocked dynamin-dependent internalization, previously described to be implicated in HIV endocytosis in human placental cells (56). Blood neutrophils and genital mixed cell suspensions were incubated with Dynasore, a dynamin inhibitor, prior to HIV stimulation and NET release quantification. Dynamin inhibition significantly reduced NET release in blood neutrophils (77% median reduction; p=0.03), and in genital neutrophils (31% median reduction: p=0.05) after HIV stimulation ( Figure 5G ).

These results indicate that dynamin-dependent endocytosis of HIV by neutrophils is partly implicated in NET release. Further, we demonstrate that a small proportion of neutrophils express HIV receptors, particularly in the FGT.

Two main mechanisms of NET formation have been described: lytic NET release, a NADPH ROS-dependent delayed process that implicates cell membrane rupture and death; and rapid non-lytic NET release, independent from ROS production and after which neutrophils remain viable (30, 31). Since we observed that HIV exposure induced early and late NET-release through different ROS-dependent and independent pathways, we next investigated whether cell death could be involved in late NET-release in response to HIV stimulation. The reagent we used to label extracellular DNA in NETs (cytotox) also stains intracellular DNA when the cell membrane is compromised, and it is generally used to identify dead or dying cells. Therefore, to quantify dead cells, we counted the number of cytotox+ neutrophils following HIV stimulation as round intracellular red staining and excluded NETs, which are large, irregular, extracellular objects ( Figure 6A ). We found no differences in the number of cytotox+ neutrophils cells in control and HIV conditions ( Figures 6A, B ).

To determine if there was a difference in the form of death in cytotox+ neutrophils from control and HIV-stimulated conditions, we extended our study to investigate apoptosis by adding annexin V to the media prior to HIV stimulation. We did not detect annexin V+ staining in neutrophils, even when they were cytotox⁺ ( Figure 6C ), indicating that neutrophils were not undergoing apoptosis. Interestingly, we observed differential annexin V incorporation into a subset of NETs in response to HIV stimulation ( Figure 6D ). Kinetic analysis of NET formation revealed that annexin V+ NETs were released at later time points compared to annexin V- NETs ( Figure 6E , Supplementary Movie 6 ). Quantification of the first annexin V+ and annexin V- NET released in blood neutrophils revealed a significant delay in the formation of annexin V+ NETs ( Figure 6F ). Because lytic NET formation implies the disruption of the cell membrane, while non-lytic NET release maintains cell membrane integrity (30), we hypothesized that annexin V was incorporated in NETs released through lytic mechanisms. As a control, we used calcium ionophore, known to induce non-lytic NET release through calcium mediated pathways (33). As seen in Figure 6G , calcium ionophore stimulation did not induce annexin V+ NETs, suggesting that annexin V+ is incorporated into NETs released after lytic mechanisms.

Next, we investigated whether annexin V+ NET release was dependent on TLR stimulation. Since annexin V+ NETs were released at later time points, we speculated that their induction could be dependent on TLR7 stimulation. As seen in Figure 6H , TLR8 inhibition (CU-CPT9a) significantly reduced early release of annexin V- NETs but had not effect on annexin V+ NET release. In contrast, TLR7 (ODN 20958) inhibition significantly reduced annexin V+ NETs, with no effect on annexin V- NETs ( Figure 6I ). These results suggest that TLR8 activates non-lytic NET release while TLR7 is involved in the lytic and ROS-dependent NET release pathways.

As part of these studies, we investigated the release of annexin V+ NETs by genital neutrophils. Interestingly, genital neutrophils rarely produced annexin V+ NETs (4 out 26 tissue samples) and this production was significantly lower compared to blood neutrophils (8 out 16 samples) ( Figures 6J, K ), suggesting preferential non-lytic NET release by tissue neutrophils.

Because lytic NET release has been described as proinflammatory, we explored potential functional differences between annexin V+ and annexin V- NETs by analyzing incorporation of lactoferrin (LTF), known for its proinflammatory and chemoattractant properties (57). We stimulated blood and genital neutrophils with HIV and determined incorporation of annexin V and LTF into NETs using confocal microscopy. In blood neutrophils we observed co-expression of annexin V and LTF in some NETs, while annexin V- NETs did not incorporate LTF ( Figure 6L ). In genital neutrophils, LTF was detected inside the neutrophils, but was not incorporated into NETs ( Figure 6M ). These data suggest a proinflammatory role for annexin V+ NETs.

Overall, these results demonstrate that blood and genital neutrophils release NETs with a different protein composition in response to HIV, and that annexin V and LTF can be used as markers to visualize NETs formed by lytic pathways.