Estradiol inhibits HIV-1BaL infection and induces CFL1 expression in peripheral blood mononuclear cells and endocervical mucosa

An inhibitory effect of estradiol (E2) on HIV-1 infection was suggested by several reports. We previously identified increased gene expression of actin-binding protein cofilin 1 (CFL1) in endocervix in the E2-dominated proliferative phase of the menstrual cycle. Actin cytoskeleton has an integral role in establishing and spreading HIV-1 infection. Herein, we studied in vitro effects of E2 on HIV-1 infection and on CFL1 expression to gain insight into the mechanism of HIV-1 inhibition by E2. E2 dose-dependently inhibited HIV-1BaL infection in peripheral blood mononuclear cells (PBMCs) and endocervix. In PBMCs and endocervix, E2 increased protein expression of total CFL1 and phosphorylated CFL1 (pCFL1) and pCFL1/CFL1 ratios. LIMKi3, a LIM kinase 1 and 2 inhibitor, abrogated the phenotype and restored infection in both PBMCs and endocervix; inhibited E2-induced expression of total CFL1, pCFL1; and decreased pCFL1/CFL1 ratios. Knockdown of CFL1 in PBMCs also abrogated the phenotype and partially restored infection. Additional analysis of soluble mediators revealed decreased concentrations of pro-inflammatory chemokines CXCL10 and CCL5 in infected tissues incubated with E2. Our results suggest a link between E2-mediated anti-HIV-1 activity and expression of CFL1 in PBMCs and endocervical mucosa. The data support exploration of cytoskeletal signaling pathway targets for the development of prevention strategies against HIV-1.


Effect of CFL1 knockdown on HIV-1 BaL infection in PBMCs.
PBMCs were plated in 24-well plate (10 6 cells/ml) in cRPMI 24 h prior to transfection. The cells were transfected with CFL1 small interfering RNA (siRNA) (Thermo Fisher), Trilencer-27 Universal scrambled negative control siRNA duplex (Origene, Rockville, MA) and Trilencer-27 HPRT Positive control siRNA duplex (Origene), using Viromer Green transfection reagent (Origene) following the manufacturer's protocol for suspension culture cells. Media was changed after overnight incubation. 48 h post-transfection, PBMCs were either incubated with E2 10,000 pg/ml for additional 48 h or left untreated. 96 h post-transfection, the cells were challenged with HIV-1 BaL 1000 TCID 50 /10 6 cells and cultured for 14 days. The supernatants were collected on days 0, 3, 7, 11 and 14 for HIV-1 gag qRT-PCR. E2 was added to the cultures on days 3, 7 and 11.
HIV-1 gag qRT-PCR. The infection was monitored in the supernatants using HIV-1 gag quantitative reverse transcription PCR (qRT-PCR) (lower limit of quantification [LLOQ] 2000 copies/ml; values below LLOQ were set to 2000 1/√2 = 215.86), using SYBR FAST One-Step qRT-PCR (Kapa Biosystems, Wilmington, MA). The primers used were forward primer 5′ GGT GCG AGA GCG TCA GTA TTAAG 3′ and reverse primer 5′ AGC TCC CTG CTT GCC CAT A 3′. The cycling conditions were 1 cycle at 42 °C for 5 min, 1 cycle at 95 °C for 5 min, and 40 cycles at 95 °C for 3 s and 60 °C for 20 s. HIV-1 gag qRT-PCR was performed using ViiA 7 realtime PCR system (Applied Biosystems, Carlsbad, CA) and data were analyzed using ViiA 7 software (Applied Biosystems). Dissociation curves were generated to verify the absence of nonspecific amplification. Results were analyzed with the standard curve method using pNL(AD8) 10 7 -10 copies/μl (11346; NIH AIDS Research and Reference Reagent Program, Germantown, MD) 14 . SOFT and CUM endpoint analyses of infection level based on HIV-1 gag copies on days 3-14 of the culture were performed as previously described 14 . SOFT is an estimation of HIV-1 gag copies at the start of the stationary phase of virus growth and CUM is the cumulative number of HIV-1 gag copies 14 www.nature.com/scientificreports/ RNA isolation and gene expression qPCR. RNA isolation from PBMCs and explants were done according to the manufacturer's protocol using Qiagen RNeasy mini and RNeasy Fibrous Tissue kit (Qiagen, Hilden, Germany), respectively. Primers for CFL1 were purchased from Bio-Rad (Hercules, CA). 200 ng of RNA was used to synthesize cDNA using High-capacity RNA-to-cDNA kit (Applied Biosystems). Gene expression was monitored by quantitative real time-PCR (qPCR) using SYBR FAST qPCR kit (Kapa Biosystems) and ViiA 7 real-time PCR system.
Immunofluorescence staining (IF). Following culture, PBMCs were plated on poly-l-Lysine coated slides (Polysciences Inc. Warrington, PA) for 1 h, washed with DPBS (Corning), fixed (15 min) with 4% paraformaldehyde (PFA), permeabilized using permeabilization buffer (Thermo Fisher Scientific) and blocked with 2% BSA. This was followed by incubation with primary CFL1 (0.108 µg/µl) and HIV-1-p24 -AF488 (0.2 µg/ µl) Abs in 2% BSA overnight at 4 °C and then by incubation with secondary Ab in 2% BSA and DAPI (1X) for 90 min. After DPBS washes and air-drying, cells were mounted in Prolong mounting medium (Invitrogen) and sealed with coverslips. The images were acquired on Nikon Eclipse Ti2 at 20X magnification and analyzed using NIS Elements v5.31.01. Total fluorescence intensity of CFL1 and p24 staining (3 fields per condition) was normalized relative to fluorescence intensity of IgG control. Following culture, explants were washed in DPBS and placed in the Histosette biopsy cassettes (Thermo Fisher Scientific), fixed overnight at 4 °C in 4% PFA followed by 10 min wash under running water. The cassettes were placed in 70% ethanol before paraffin embedding and sectioning at Molecular Cytology Core, MSKCC, New York. Slides were heated at 60 °C for 30 min followed by × 3 washes in Histo-Clear II (National Diagnostics, Atlanta, GA) for 10 min each, × 2 washes in absolute alcohol for 10 min each, 5 min washes each in 95%, 70% and 50% ethanol. The slides were rinsed in deionized water and rehydrated in DPBS for 10 min. Then slides were placed in pre-heated (92-95 °C) antigen retrieval buffer (Millipore, Burlington, MA) for 5 min and cooled down for 5 min at room temperature followed by water and DPBS rinses. Next, sections were permeabilized for 10 min, blocked with 5% BSA for 30 min and incubated with primary CFL1 Ab (0.108 µg/µl), and CD3-AF488 (0.2 µg/µl) Ab in 2% BSA overnight followed by three 10 min DPBS washes, and secondary Ab and DAPI (1X) staining for 90 min. After three 10 min DPBS washes, sections were air-dried and mounted in Prolong mounting medium. The images were recorded on Nikon Eclipse 90i at 10X magnification and analyzed using ImageJ (NIH). Total fluorescence intensity of CFL1 and p24 staining (3 fields per section) was normalized relative to fluorescence intensity of IgG control. CFL1 and pCFL1 Western Blot (WB) and densitometry. PBMCs. PBMCs used for WB were cultured as described above. PBMCs were incubated with E2 for 48 h or pre-treated with LIMKi3 for 3 h followed by 48 h incubation with E2. The cells were then either collected for WB analysis or challenged with 1000 TCID 50 HIV-1 BaL . PBMCs challenged with HIV-1 BaL were cultured for 7 or 14 days and collected for WB analysis.
In siRNA experiments, transfected PBMCs were collected 48 h post-transfection for WB analysis or incubated with or without E2 for additional 48 h. The cells were then collected for WB or challenged with 1000 TCID 50 HIV-1 BaL. PBMCs challenged with HIV-1 BaL were cultured for 3 days and collected for WB analysis.
PBMCs were gently washed with ice cold DPBS, lysed using RIPA lysis and extraction buffer (Thermo Fisher Scientific) supplemented with protease inhibitor cocktail (Sigma-Aldrich) according to manufacturer's instructions. Whole cell extracts (WCEs) were stored at − 20 °C. Protein concentrations were measured by Bradford method (Bio-Rad). 60-120 µg of protein was mixed with NuPAGE LDS sample buffer (Thermo Fisher Scientific) and β-Mercaptoethanol (Thermo Fisher Scientific) and heated for 10-15 min at 95 °C. WCEs were resolved on standard NuPAGE 4-12% Bis-Tris SDS-PAGE in MOPS SDS buffer (Thermo Fisher Scientific) and transferred to PVDF membrane using Transblot (Bio-Rad). Membranes were blocked with standard 5% non-fat dry milk (Sigma-Aldrich) dissolved in DPBST (0.1% Tween-20, Sigma-Aldrich). Blots were incubated with CFL1, pCFL1, HPRT1 and β-Actin Abs in DPBST overnight at 4 °C followed by HRP-conjugated secondary Ab incubation for 1 h at room temperature. The membranes were developed using the ECL system (Thermo Fisher Scientific) and analyzed using Amersham Imager 600 (Cytiva).
Endocervical tissue. Endocervical tissue explants were incubated with E2 (100 and 10,000 pg/ml) for 48 h. The explants were then either collected for WB analysis or challenged with 500 TCID 50 HIV-1 BaL . Explants challenged with HIV-1 BaL were cultured for 7 days and collected for WB analysis.
The explants were washed with ice cold DPBS, snap frozen in liquid nitrogen, cut, and lysed in RIPA lysis and extraction buffer supplemented with protease inhibitor cocktail to prepare WCEs. Bradford assay was used to measure protein concentration. 60-120 µg of protein was mixed with NuPAGE LDS sample buffer and β-Mercaptoethanol and heated at 95 °C for 10-15 min. WCEs were resolved on NuPAGE 8-16% Tris Glycine gel in Tris Glycine buffer (Thermo Fisher scientific). The gel/blot transfer and Ab probing were done as described above.

Flow cytometry.
Uninfected and HIV-1 BaL infected PBMCs were incubated with E2, Raloxifene or LIMKi3 as described above. Untreated PBMCs and 16% PFA treated cells were included as controls. The cells were stained with FVS780, fixed with 4% PFA, and acquired on a BD LSR II (BD Biosciences) and analyzed by FLOWJO 8.8.6 software.

E2 inhibits HIV-1 BaL infection in PBMCs and endocervix in a dose-dependent manner.
To determine the effect of E2 on HIV-1 BaL infection in PBMCs and endocervical mucosa, PBMCs and endocervical explants were challenged with HIV-1 BaL and cultured in the presence of E2. In selected experiments, Raloxifene and 3TC control were included. E2 at non-toxic concentrations dose-dependently inhibited HIV-1 BaL infection in both PBMCs and endocervix (Fig. 1a,b; Supplementary Fig. S1 a,b). Viral growth kinetics in individual experiments are presented in Supplementary Fig. S2a,b. Expectedly, higher infection level was observed in PBMCs than in endocervix, likely due to higher challenge dose and higher target cell numbers. Consistent with published data 19 , Raloxifene, a SERM that blocks Estrogen receptor (ERα), reverted E2-mediated HIV-1 BaL inhibition at the non-toxic 1 µM concentration ( Supplementary Fig. S1 a,b).
Effect of E2 on CFL1 mRNA expression in PBMCs and endocervix. We analyzed effect of E2 on CFL1 mRNA expression in uninfected and HIV-1 BaL infected PBMCs and in uninfected endocervical tissues. www.nature.com/scientificreports/ CFL1 expression post-E2 treatment tended to be higher than in untreated controls ( Supplementary Fig. S3). However, no significant changes in CFL1 expression were observed after 48 h of E2 treatment in uninfected PBMCs and endocervix ( Supplementary Fig. S3a). An increase in CFL1 mRNA expression in HIV-1 BaL infected PBMCs post-E2 (1000 pg/ml) treatment was observed on day 3 of the culture. No changes in CFL1 gene expression in infected PBMCs were observed on day 7 of the culture ( Supplementary Fig. S3b).
E2 increases expression of total CFL1, pCFL1 and pCFL1/CFL1 ratios in PBMCs and endocervix. In agreement with IF results, WB analysis revealed an increase in CFL1 and pCFL1 expression in uninfected and HIV-1 BaL infected PBMCs and endocervix following E2 treatment, with more variation detected in infected endocervix (Fig. 6a-d). Quantification of bands by densitometry supported this observation and showed an increase in CFL1 and pCFL1 post-E2 treatment (Fig. 6e). The analysis also revealed higher pCFL1/ CFL1 ratios following E2 treatment than in untreated controls ( Fig. 6f; Supplementary Fig. S4a-d). However, the changes in total CFL1, pCFL1 and ratios were not statistically significant.
LIMKi3 reverts E2-mediated anti-HIV-1 BaL activity in PBMCs and endocervix and decreases expression of total CFL1, pCFL1 and pCFL1/CFL1 ratios in PBMCs. To explore the role of pCFL1 in  Fig. S5). LIMKi3 reduced E2-mediated anti-HIV-1 activity in PBMCs and in endocervix (Fig. 7a) and decreased E2-induced phosphorylation of CFL1 at all analyzed time points in uninfected (48 h) and infected (D7 and D14) PBMCs ( Fig. 7b; Supplementary Fig. S6a-c). LIMKi3 did not change expression of total CFL1 as compared to untreated control. However, a reduction of E2-induced total CFL1 expression in the presence of LIMKi3 was detected ( Fig. 7b; Supplementary Fig. S6a-c). Densitometry analysis supported these observations and demonstrated decrease in CFL1, pCFL1 expression and in pCFL1/ CFL1 ratios in LIMKi3 + E2 conditions as compared to E2 only conditions (Fig. 7c,d). These results were not statistically significant. The specificity of LIMKi3 was checked using AKT and pAKT Abs (Supplementary Fig. S7).

Knockdown of CFL1 reduces total CFL1 and pCFL1 expression in PBMCs and reverts E2-mediated anti-HIV-1 BaL activity.
To confirm involvement of CFL1 in E2-mediated HIV-1 inhibition, we conducted CFL1 knockdown experiments. Transfection of PBMCs with CFL1 siRNA resulted in decreased total CFL1 and, expectedly, decreased pCFL1 protein level 48 h post-transfection as compared to control conditions (untransfected cells and cells transfected with control siRNAs) ( Fig. 8a; Supplementary Fig. S8). PBMCs transfected with CFL1 siRNA were incubated with E2 for additional 48 h and then challenged with HIV-1 BaL and

Effect of E2 on CC/CK concentrations in HIV-1 BaL infected endocervical tissue cultures.
To further explore the effects of E2 in mucosa, we analyzed tissue CC/CK profile following E2 treatment. CC/CK concentrations in untreated and E2-treated HIV-1 BaL infected tissue supernatants collected on day 3 of the culture were similar. Expectedly, concentrations of several CC/CK (IL1RA, IL7, CXCL8, CXCL10 and GM-CSF) were higher in untreated condition on day 7 vs. day 3. CXCL10 and CCL5 concentrations in day 7 E2-treated culture supernatants were significantly lower than in untreated supernatants ( Supplementary Fig. S9).

Discussion
This study demonstrates a link between E2-mediated inhibition of HIV-1 BaL infection and increased expression of total CFL1, pCFL1 as well as increased pCFL1/CFL1 ratios in PBMCs and endocervix.
Our experimental approach focused on repeated addition of E2 to PBMCs and endocervix before viral challenge and during culture. This approach was rationalized to mimic continuous exposure to endogenous E2 or repeated exposure to vaginal products containing E2 in vivo. The E2 concentrations range used in this study covers concentrations detected systemically in premenopausal women (up to ~ 500 pg/ml) as well as higher concentrations (up to 10,000 pg/ml), which are relevant considering intravaginally applied E2-containing products [20][21][22][23] .
The observed E2-mediated anti-HIV-1 activity in two model systems is consistent with previously reported (i) in vitro E2-mediated anti-HIV-1 activity in macrophages and CD4 + T cells (through induction of IFNα and β-catenin pathways) 19,24,25 ; (ii) our ex vivo data in cervical explants showing direct association between serum E2 concentrations and cervical infection level 1 ; (iii) data showing protective effect of vaginal E2-containing creams against cervical HIV-1 infection 26,27 ; and (iv) in vivo data in non-human primates demonstrating protective effects of E2-containing implants and estriol cream against vaginal SIV challenge 28,29 . E2 was also reported to inhibit TCR activation of HIV-1 transcription through ESR1 and is likely to limit viral emergence from latency 30 . It needs to be noted that the stimulatory effect of E2 on HIV-1 transcription was also reported 31 . www.nature.com/scientificreports/ In contrast to the initial hypothesis that was based on our RNAseq studies 2 , no consistent significant upregulation in CFL1 mRNA expression in PBMCs and endocervix was observed. This could have been due to selection of time points and/or effect of post-transcriptional regulation. We aimed to have parallel mRNA and protein expression data and selected time points starting from 48 h to days after E2 exposure. Analysis of earlier time points was not performed. As the analysis was done either in uninfected or HIV-1 BaL infected PBMCs, we cannot exclude effect of HIV-1 BaL infection on CFL1 mRNA expression.
It is well established that HIV-1 binding to blood CD4 + T cells leads to temporal changes in actin polymerization and depolymerization mediated by CFL1 inhibition through phosphorylation and activation through dephosphorylation, regulating HIV infection [4][5][6] . We speculate that the detected increase in total CFL1, pCFL1 and increased pCFL1/CFL ratio in PBMCs and endocervix following E2 exposure may impair the dynamic cytoskeletal treadmill by locking CFL1 in inactivated state in HIV target cells, and therefore, provide protection against infection. Future studies addressing E2-induced changes in CFL1 expression in specific HIV target cells are warranted.
Treatment of PBMCs and endocervical explants with LIMKi3 (specific inhibitor of CFL1 phosphorylation targeting LIMK1 and 2; no off-targets have been described 32 ) reverted E2-mediated anti-HIV-1 activity. LIMKi3 reverted E2-induced increase in total CFL1 expression, pCFL1 expression and an increase in pCFL1/CFL1 ratios. These data are consistent with published data demonstrating involvement of LIMK in regulation of E2-mediated CFL1 phosphorylation in hippocampal neurons 33 . The data further implicate changes in CFL1 protein expression and pCFL1/CFL1 ratio in E2-mediated anti-HIV-1 activity. Lack of E2-mediated increase in total CFL1 expression in the presence of LIMKi3 requires further investigation.
CFL1 knockdown experiments in PBMCs demonstrated partial reversal of E2-mediated anti-HIV-1 activity. These data confirm involvement of CFL1 in E2-induced signaling leading to inhibition of HIV-1 infection. The partial effect is likely attributed to transient nature of CFL1 knockdown.
Our results support recent data implicating pCFL1 as potential biomarker of HIV-1 susceptibility and disease progression 6,13 . Intriguingly, blocking of HIV-induced LIMK1 activity was previously shown to inhibit HIV-1 entry, nuclear migration, viral release, and cell-cell transmission 7,34,35   www.nature.com/scientificreports/ course of the culture period. A decrease in CXCL10 and CCL5 after E2 exposure suggests an anti-inflammatory effect, which could have contributed to the observed anti-HIV-1 activity. CXCL10 was previously identified as a potent marker of high HIV-1 acquisition risk as high genital concentrations in seronegative women were found to be associated with higher subsequent infection [36][37][38] . CCL5 is a proinflammatory chemokine which attracts HIV-1 target cells to the mucosal sites and is increased in vaginal secretions of women at the higher risk of infection [39][40][41] . High CCL5 in vaginal fluids was shown to correlate with high HIV-1 seroconversion rate 39 . Our study has several limitations. We did not explore effects of different E2 exposure regimens on HIV-1 BaL infection and CFL1 expression as we focused on one regimen mimicking continuous exposure to E2. We also did not explore if progesterone might impact observed E2-mediated activity. Given the dynamic changes in progesterone and E2 concentrations during menstrual cycle, potential impact of progesterone on E2-induced changes in CFL1 expression and anti-HIV-1 activity deserves further exploration.