In this study, we report the sex-specific effects of E2 and DHT on primary cells in 2D culture and in 3D bulk tissue vasculogenesis models using multiple types of primary human endothelial cells and fibroblasts. Our data shows that treating female cells with E2 and male cells with DHT significantly increases cellular bioenergetics and proliferation, whereas treatment of female cells with DHT and male cells with E2 conversely exerts inhibitory effects in all cell types tested. Similar sex-specific hormone effects were measured in a 3D tissue vasculogenesis assay. These findings demonstrate the importance of developing sex-based methods for standard cell culture systems and tissue engineered models to improve translational relevance and accelerate the discovery of sex-based precision therapies.
For 2D immunofluorescence assays and RT-qPCR studies, 4 donors per sex were used for HUVEC and HLF studies while 3 donors per sex were used for HRMVEC and HOF. All donors were utilized individually without pooling. In vasculogenesis assays, HLF and HUVEC were randomly assigned into sex-matched pairs, and 3 pairs were used. See individual experiment sections for replication information. See Supplemental Fig. 1 for cell donor information. All cells were sexed via RT-qPCR for SRY gene to confirm sex was as listed by supplier, and all cells were analyzed via RT-qPCR to ensure expression of sex hormone receptors (Figure S1). Fibroblasts were grown in FibroLife fibroblast complete media kit (Lifeline Cell Technologies, LL-0011) and endothelial cells were grown in VascuLife VEGF endothelial medium complete kit (Lifeline Cell Technologies, LL-0003). Charcoal-stripped hormone depleted media was formulated using the same base Lifeline kits with phenol red and fetal bovine serum (FBS) omitted. Instead, 2% charcoal stripped serum (Gibco, 12676029, Lots #76-029, 74-229) was added. Sex hormones were dissolved in in molecular grade ethanol (Sigma Aldrich, E7023).
Primary human lung fibroblasts (HLF), human ocular fibroblasts (HOF), human umbilical vein endothelial cells (HUVEC), and human retinal microvascular endothelial cells (HRMVEC) from both XX and XY donors were sourced for use in these studies (ages 20-45). HLF and HOF were expanded in FibroLife fibroblast complete media kit (Lifeline Cell Technologies, LL-0011) and HUVEC were grown in VascuLife VEGF endothelial medium complete kit (Lifeline Cell Technologies, LL-0003). Cells were expanded until passage 3, and cells from passages 3 through 7 were used in all experiments.
XX and XY HUVEC, HRMVEC, HOF and HLF were seeded at a density of 10,000 cells per well in 24-well tissue culture plates and grown in cell type-specific Lifeline expansion medium as outlined above for 48 hours. After 48 hours, cells were switched to the same media formulated to be phenol red-free and supplemented with 2% charcoal stripped serum for 24 h to ensure complete hormone starvation. Following hormone starvation, cells were exposed to media supplemented with 0.1 nM, 1 nM, or 10 nM E2 or 0.01 nM, 0.1 nM, 1 nM or 10 nM DHT dissolved in ethanol for 48 h. Ethanol vehicle was added to the medium for all hormone-free controls. 48 h was selected for the duration of hormone exposure based on the results of RT-qPCR assays that showed significant gene expression changes at 48 h while minimizing the influence of trophic effects at longer timepoints. After 48 h of culture, monolayers were fixed with 4% paraformaldehyde at room temperature for 15 min. Monolayers were then gently washed with Dulbecco's Phosphate Buffered Saline (PBS, Gibco, 14190144) 3 times. Fixed well plates were stored covered in PBS at 4 °C until staining. Samples were blocked and permeabilized with 1% bovine serum albumin (BSA, Sigma Aldrich, A4737) and 0.1% Triton-X (Sigma Aldrich, X100) in PBS at room temperature for 20 min. Rabbit anti-Ki67 antibody (Abcam, ab16667, Lot # 1090780-56) was added at 10 μL/mL in 0.1% BSA in PBS. Plates were rocked for 2 hours protected from light at a gentle rock and then washed with PBS 4 times. Secondary antibody conjugated with Alexa Fluor 594 (Abcam, ab150092) at 4 μL/mL, 4 μL/mL 4',6-diamidino-2-phenylindole (DAPI, Invitrogen, H21486) to label nuclei, and 4 μL/mL Alexa Fluor 488 Phalloidin (Thermo Fisher, A12379) to label actin were added in 0.1% BSA in PBS for 40 min in a dark, gentle rocking. Plates were then washed again in PBS and stored covered in PBS at 4 °C protected from light until imaging. 4 donors per sex were used for HLF and HUVEC, and 3 donors per sex were used for HOF and HRMVEC. 3 wells per condition per donor were utilized. 3 images per well were analyzed.
Cells were plated at 5000 cells per well in 6 well plates. At 50% confluence, cells were changed to phenol red-free media supplemented with charcoal stripped serum for a 24 h hormone starvation. After 48 h of starvation, we exposed cells to 0.1 nM, 1 nM or 10 nM E2 or 0.01 nM, 0.1 nM, 1 nM, and 10 nM DHT. E2 and DHT were dissolved in ethanol, and control groups were supplemented with the same volume of ethanol as a vehicle control. E2 and DHT were replenished every 24 h, and media was changed every 48 h. Cells were collected via trypsinization at 48 h, 72 h, and 168 h of exposure. If not isolated immediately, cell pellets were stored in RNAlater (Thermo Fisher) until RNA isolation. mRNA was isolated from collected samples via Qiagen RNEasy kit following the manufacturer instructions, and sample purity was validated by measurement of the 260/280 nM and 260/230 nM ratios using a Nanodrop spectrophotometer (Thermo Fisher). mRNA samples were stored at -80 °C for long term storage. cDNA was synthesized using qScript cDNA SuperMix (QuantaBio) according to the supplier recommended protocol, diluted 1:10 in UltraPure DNase and RNase free water (Thermo Fisher), and stored at -20 °C. Quantitative reverse transcription polymerase chain reaction (RT-qPCR) was used to measure the expression of genes associated with energetics and stress response in all cells. RT-qPCR was completed using SybrGreen PowerUp on the ThermoFisher StepOnePlus system (Thermo Fisher). Primers used are listed in Table 1. GAPDH was utilized as the housekeeping gene. RT-qPCR results were analyzed via the 2 method. 4 donors per sex were used for HLF and HUVEC, and 3 donors per sex were used for HRMVEC and HOF. 3 biological replicas per donor and 3 technical replicas per gene were analyzed.
Monolayers were cultured as described in Ki67 immunofluorescence. After 48 h of hormone culture, cells were exposed to 100 nM tetramethylrhodamine methyl ester (TMRM) dissolved in DMSO (Sigma Aldrich, D8418) for 30 min at 37 °C per manufacturer instructions (Thermo Fisher, T668). Monolayers were washed thoroughly with PBS and imaged immediately. 4 donors per sex were used for HLF and HUVEC, and 3 donors per sex were used for HOF and HRMVEC. For each end condition, 3 wells were analyzed for each time point using 3 images per well.
A 9-well milliscale tissue array was designed in SolidWorks (Dassault Systèmes) and transferred into PreForm 3D Printing Software (FormLabs). Molds were printed using a Form3 SLA 3D printer (FormLabs). After printing the molds were washed, cured, and flattened to generate acceptable molds for polydimethylsiloxane (PDMS) soft lithography. As PDMS is known to absorb small, hydrophobic molecules such as steroid hormones, molds were polyurethane coated to increase smoothness of PDMS and reduce surface area to minimize absorption as we have previously reported. Absorption of sex hormones was further mitigated by cross-mixing a polyethylene glycol (PEG) and PDMS self-segregating polymer (PDMS-PEG) into the PDMS mixture prior to PDMS curing. The ideal amount of PDMS-PEG for our purposes was found to be 0.25% by weight as this endowed the material with the surface properties we desired while maintaining the benefits of traditional PDMS. As such, PDMS was mixed at a 1:10 ratio of elastomer to fixative agent (Sylgard 184, Ellsworth Adhesives). The PDMS-PEG copolymer was then added to a total of 0.25% weight (Gelest, DBE-712). The polymer was then mixed thoroughly and degassed. PDMS was poured into molds, degassed again, and kept in an oven at 60 °C for at least 4 hours to ensure complete curing. Prior to use, we sanitized devices via UV light sterilization for 2 hours. Poor protein adsorption and cell seeding on untreated PDMS surfaces is well-documented. Thus, surfaces of the array wells were treated with a 5 mg/ml polydopamine (PDA) solution for 2 h in UV light to facilitate stable anchorage of ECM hydrogels. Devices were then stored in a dark space at room temperature until use.
Nile Red absorption assays were completed to test that PDMS-PEG copolymer microarrays would prevent hormone absorption. Nile Red, a fluorescent stain with similar size to E2, has been used in previous studies to assay PDMS absorption of small molecules. 1 μg/mL Nile Red (Thermo Fisher) was dissolved in ethanol and loaded into central wells of small PDMS copolymer devices and allowed to sit for 1 h. Devices were washed with PBS 3 times. PDMS with or without copolymer made on coated and non-coated molds were imaged under fluorescence. Images were quantified using the average gray intensity across the well as calculated in FiJi.
HUVEC and HLF were expanded using the same techniques used in 2D culture. Polydopamine (PDA)-treated PDMS microarrays were loaded with 2 × 10 cells/ml HLF and 2 × 10 cells/ml HUVEC in a hydrogel comprised of blended collagen type I (2.25 mg/ml collagen type I, Corning) and fibrin (5 mg/mL fibrinogen activated with 1 U/ml thrombin, Sigma Aldrich). This hydrogel composition was selected to combine the vascularization promoting properties of fibrin with the suprastructural stability provided by collagen gels. Hydrogels were allowed to solidify at 37 °C for 30 min. Stromal vascular tissues were cultured in vascular cell growth media supplemented with VEGF (ATCC), 25 μg/mL aprotinin (Sigma Aldrich), and 2% FBS for 96 h. Media was changed to phenol red free media containing 2% charcoal stripped serum (Gibco) for complete hormone starvation for 24 h before exposing the gels to 0 nM, 1 nM, or 10 nM E2 in ethanol or 0 nM, 1 nM, or 10 nM DHT for 4 days. For each sex, 3 donor combinations were utilized for 3 gels per condition and 3 images per gel. See Supplemental Fig. 7H for workflow.
3D tissues from the PDMS tissue arrays were processed for whole mount staining. Tissues were fixed in 4% paraformaldehyde (Sigma Aldrich) for 1 h at room temperature and then overnight at 4 °C. Tissues were then washed with several changes of PBS and stored in PBS at 4 °C until stained. Gels were stained using 4 μL/mL (DAPI) to label nuclei, 4 μL/mL Alexa Fluor 488-conjugated Phalloidin (Thermo Fisher, A12379) to label F-actin, and 20 μL/mL Ulex Europeas agglutinin I (UEA-1, Vector laboratories, RL-1062-2) to specifically label endothelial cell lectins. Stains were prepared in PBS with 0.2% Triton-X and 1% BSA (Sigma Aldrich). Devices were loaded with staining cocktail, ensuring complete coverage of the gels, and rocked gently for 1 h at room temperature. Gels were then refrigerated overnight, rocked for an additional hour the next day at room temperature, washed in PBS, and stored covered with PBS at 4 °C until imaging.
All 2D and 3D samples were imaged on an inverted Nikon C2 laser scanning confocal microscope (LCSM) equipped with a Nikon DS-FI3 camera. Samples from a given staining cohort were imaged at a fixed laser intensity and exposure time. For 2D cultures, 3 images of each sample were taken at randomized points in the central regions of the well to avoid edge effects when imaging. Analysis of 2D stains was performed in FiJi. TMRM images were converted to grayscale and segmented to the area of the cells utilizing an overlayed DIC image of the sample. Mean fluorescence intensity of the region of the cells was measured, and the average intensity across the image was recorded. This process was completed for each of the 3 images taken from each well. The average of these 3 measurements was then used as the average intensity across the entire well. Ki67 staining was analyzed in FiJi, using the DAPI co-stain as the mask for regions of interest (cell nuclei). The threshold intensity of DAPI images was adjusted to the sharply capture cell nuclei and turned into a binary mask which was overlaid on the Ki67 channel. Ki67 positive cells were counted via particle analysis.
Each tissue from the vasculogenesis assay was imaged with 3 full thickness Z-stacks, excluding the excluding a rim of tissue at the interface with PDMS surfaces. 3D image datasets were analyzed via MATLAB (Mathworks, R2021b). Max intensity projections of the Z-stacks were saved at TIFF files and exported to MATLAB for filtering and quantification. Vascular images were smoothed with a gaussian filter, low intensity noise was filtered out, and images were then denoised via a pretrained neural network. The open-source segmentation tool REAVER was utilized to segment vascular networks and quantify morphometric parameters. For non-participating cells, TRITC channels were isolated and analyzed for particles with high roundness. Image analysis results were exported to GraphPad Prism V 9.2 for statistical analysis.
All statistics were completed in GraphPad Prism V 9.2. For analysis of TMRM and Ki67 indexes, statistical analysis was completed via two-way ANOVA between sex and hormonal conditions with a 95% confidence interval (* = p < 0.05; ** = p < 0.01; *** = p < 0.001). Vasculogenesis data was compared via two-way ANOVA comparing between sex and hormone conditions. Data was assessed and found to be normally distributed per a Shapiro-Wilk test. Variances were homogeneous per Levene's test for homogeneity of variances. All samples used to collect datasets were independent, thereby meeting the necessary assumptions for two-way ANOVA. The number of donors per cell type per sex and the number of technical replicates per experiment are included in previous Methods sections for each assay and in the corresponding Figure captions.
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.