Fibroblast-specific ablation of ESR1 prevents herniation in Aromhum mice. All Aromhum mice developed a fibrotic process characterized by the proliferation of fibroblasts depositing excess ECM in the LAM, weakened muscle tissue, and formation of scrotal hernias — with hernia sacs containing abdominal viscera, gonads, gonadal fat, and urinary bladder (Figure 1A). Primary fibroblasts isolated from Aromhum LAM confirmed the coexpression of both ESR1 and PDGFRA proteins, validating their identity as HAFs (9). This expression pattern aligns with our previous single-cell RNA study of Aromhum LAM (Supplemental Figure 1, A and B; supplemental material available online with this article; https://doi.org/10.1172/JCI179137DS1) (9). To discern the impact of ESR1 signaling on LAM HAFs and hernia development, we engineered a fibroblast-specific ESR1 knockout in Aromhum mice (fEsr1–/– Aromhum) by cross-breeding floxed ESR1, PDGFRA-cre, and Aromhum mice. Notably, the fEsr1–/– Aromhum mice did not show hernia formation during a 20-week observation period (Figure 1B). Similarly, the WT control mice (fEsr1+/+ WT) displayed no herniation (Figure 1B). As expected, all fEsr1+/+ Aromhum (i.e., Aromhum) littermate controls displayed hernia onset at approximately 5 weeks of age, with hernia sizes increasing over time (Figure 1, B and D). The percentage of HAFs marked by expression of PDGFRA and ESR1 was significantly lower in the hernia-free fEsr1–/– Aromhum mice compared with the positive control herniated fEsr1+/+ Aromhum mice (1.8% vs. 32.3%; Figure 1C). This marked reduction in ESR1 expression in PDGFRA-cre–driven Esr1fl/fl mice serves as a crucial validation of our model, confirming the specificity of the ESR1 knockout in targeting HAFs. Furthermore, ESR1 depletion in LAM HAFs effectively prevented LAM fibrosis and muscle atrophy, as corroborated by Masson’s trichrome staining from fEsr1–/– Aromhum mice (Figure 1, D and E). In contrast, fEsr1+/+ Aromhum mice exhibited significantly higher LAM fibrosis than both ESR1-depleted fEsr1–/– Aromhum mice and the fEsr1+/+ WT controls (Figure 1, D and E). These findings underscore the central role of ESR1 signaling in the stimulation and expansion of LAM HAFs to drive scrotal herniation in Aromhum mice.

Fibroblast-specific ablation of ESR1 in Aromhum mice prevents herniation. (A) Representative images of WT and Aromhum mice, and an illustration depicting scrotal hernia and LAMs. Created with BioRender (biorender.com). (B) Measurement of scrotal hernia size with age in fibroblast-specific Esr1-knockout mice (fEsr1–/– Aromhum) and fEsr1+/+ Aromhum and fEsr1+/+ WT littermate controls (n = 3–4 per group, mean ± SEM, repeated-measures ANOVA with Bonferroni multiple comparisons). (C) Flow cytometry dot plots showing the percentage of PDGFRA+ estrogen receptor-α–positive HAFs in LAMs from fEsr1–/– Aromhum and control fEsr1+/+ Aromhum mice (n = 3). (D) Representative images of scrotal hernias (top) and Masson’s trichrome–stained LAMs (bottom). Red arrows point to scrotal hernia, while yellow arrows point to atrophied myofibers. Scale bars: 100 μm. (E) Quantification of the fibrotic area in fEsr1–/– Aromhum, fEsr1+/+ Aromhum, and fEsr1+/+ WT mice (n = 3–4 per group, median ± interquartile range, 1-way ANOVA with Bonferroni multiple comparisons).

Inhibition of E2/ESR1 signaling prevents and reverses hernias in Aromhum mice. To explore pharmacological interventions for hernias, we used fulvestrant, an E2/ESR antagonist that competitively blocks E2 binding to ESRs, leading to subsequent ESR degradation. Analogously to fEsr1–/– Aromhum mice in Figure 1, Aromhum mice given slow-release fulvestrant pellets at 3–4 weeks of age (prior to hernia formation) did not develop hernias, whereas all Aromhum mice given placebo pellets exhibited progressive hernia growth over time (Figure 2A). To explore the potential of fulvestrant to reverse established hernias, Aromhum mice with large hernias (>200 mm2) at approximately 6–10 weeks of age were treated with fulvestrant slow-release pellets. Remarkably, within 2 weeks, fulvestrant-treated mice displayed a significant reduction in hernia size (Figure 2B). Subsequently, all fulvestrant-treated mice exhibited complete hernia regression with scrotal sizes comparable to those of WT mice after 4 weeks of treatment (Figure 2C). Histological examination after 12 weeks of treatment demonstrated extensive fibrotic degeneration in the LAM of placebo-treated Aromhum mice, whereas those receiving fulvestrant displayed normal muscle tissue and no fibrosis (Figure 2, C and D, and Supplemental Figure 3A). Remarkably, within just 1 week, fulvestrant-treated mice exhibited a stalling of herniation (Supplemental Figure 2A). Although hernia regression was not achieved during this short period, subsequent histological and immunohistochemical analyses show reduction in fibrosis and increase in muscle regeneration (Supplemental Figure 3, B–D) (20, 21). Furthermore, collagen content in the LAM tissue of Aromhum mice after the longer 90-day fulvestrant treatment was comparable to the levels in WT mice (Figure 2E). These findings underscore the translational potential of fulvestrant as a pharmacological approach for both preventing and reversing hernias.

Fulvestrant treatment prevents and reverses well-established large hernias in mice. (A) Schematic of hernia prevention study design (top) and measurement of scrotal hernias (bottom); fulvestrant was administered before hernia formation. Arrow indicates the week of pellet implantation (n = 10–15 per group, mean ± SEM, repeated-measures ANOVA). (B) Schematic of hernia treatment study design (top) and measurement of scrotal hernias (bottom); fulvestrant was administered after large hernias were formed. Arrow indicates the week of pellet implantation (n = 10–15 per group, mean ± SEM, repeated-measures ANOVA with Bonferroni multiple comparisons). In both A and B, the dotted line at 140 mm2 represents normal scrotum size before hernia development, and the orange shaded region represents large scrotal hernia size (>200 mm2). Created with BioRender (biorender.com). (C) Representative images of LAM morphology and Masson’s trichrome staining of LAMs from mice in the treatment study (B). Red arrows point to bilateral scrotal hernias in placebo-treated mice, while yellow arrows point to atrophied myofibers. (D and E) Quantification of the fibrotic area (D) and collagen content by hydroxyproline assay (E) in the mouse LAM treatment study (B and C) (n = 5–6 per group, median ± interquartile range, 2-way ANOVA with Bonferroni multiple comparisons; scale bars: 100 μm).

Subsequently, we administered raloxifene HCl — a partial antagonist of E2/ESR — to Aromhum mice harboring large scrotal hernias (>200 mm2). Raloxifene administration effectively reduced hernia sizes, with results similar to those obtained with fulvestrant (Figure 2B and Supplemental Figure 2B). Hernia size in placebo-treated Aromhum mice continued to increase to about 300 mm2 during treatment. However, a 10-week raloxifene treatment reduced hernia from large to small/medium sizes, suggesting stoichiometric effects of the partial E2/ESR antagonist raloxifene on hernia regression compared with the E2/ESR antagonist fulvestrant (Figure 2B and Supplemental Figure 2B). To ascertain the specificity of hernia regression to ESR1, we used methyl-piperidino-pyrazole (MPP), an ESR1-selective antagonist, in Aromhum mice with large scrotal hernias. A significant reduction in hernia sizes was evident following a 21-day treatment regimen (Supplemental Figure 2C). Despite a significant reduction in hernia sizes, the effects of MPP were relatively partial compared with those of fulvestrant treatment. This disparity may be attributed to MPP’s short half-life and its pharmacokinetic properties, which might have limited its therapeutic efficacy. In contrast, ESR1 deletion and fulvestrant treatment provide more comprehensive and stronger inhibition of ESR1 action, leading to more pronounced effects on hernia regression. The ESR2- and GPER1-selective antagonists PHTPP and G-15, respectively, exhibited no discernible effects on hernia size compared with placebo treatment (Supplemental Figure 2, D and E). These findings underscore the predominant role of ESR1 as the primary ESR driving and regressing herniation in Aromhum mice.

E2 or fulvestrant modifies chromatin accessibility, its occupancy by ESR1, and the transcriptome in HAFs. We investigated the genome-wide, epigenomic, and transcriptomic effects of E2 and fulvestrant in HAFs to reveal underlying mechanisms responsible for LAM fibrosis and herniation. Based on our earlier scRNA-Seq and flow cytometry results, HAFs make up 50%–80% of the fibroblasts present in LAM (9). To ensure a higher purity of HAF populations, we used a preplating procedure, allowing us to selectively obtain adherent, pathogenic fibroblasts while minimizing the presence of other cell types. By the second passage, our cultures consistently contained only HAFs, as confirmed by immunostaining for ESR1 and PDGFRA (Figure 1B). This preplating method effectively excluded myogenic cells and other nonfibroblast populations, ensuring the purity of our HAF cultures. Notably, we observed that HAFs cultured in serum-rich or E2-supplemented conditions exhibited rapid proliferation, higher viability, and increased secretion of ECM, further underscoring their pathogenic role in fibrosis. HAFs treated with E2 alone or E2 with fulvestrant were subjected to multiomics analyses (Figure 3A). ChIP-Seq using an antibody against ESR1 revealed higher ESR1 binding in distal intergenic regions, suggesting a critical influence of enhancer regions on transcriptional regulation (Figure 3B, Supplemental Figure 5, A and B, and Supplemental Figure 6A). In contrast, the E2/ESR antagonist fulvestrant decreased total ESR1 binding and accessible chromatin regions (Figure 3B, Supplemental Figure 5, A and B, and Supplemental Figure 6A). RNA-Seq revealed distinct transcriptomic changes induced by E2 treatment, including the upregulation of pathways related to mesenchymal cell proliferation, ECM organization, and TGF-β/WNT signaling pathways (Supplemental Figure 7, A, B, and D). In contrast, fulvestrant binding to ESR1 downregulated these E2-driven effects with marked downstream transcriptional effects on angiogenesis and regulatory pathways (Supplemental Figure 7, A, C, and E).

Multiomics analysis reveals E2/ESR1 signaling changes in HAFs. (A) Illustration of experimental design for multiomics studies. Created with BioRender (biorender.com). (B) Genomic distribution of ESR1 binding events in ChIP-Seq and open chromatin peaks in ATAC-seq in HAFs after E2 or E2 plus fulvestrant treatment (n = 3 per group). (C) Venn diagram showing overlap of genes upregulated with E2 treatment compared with E2 plus fulvestrant treatment in multiomics assays: RNA-Seq, ChIP-Seq, and ATAC-seq (fold change > 1.2, P < 0.05). (D) Significantly upregulated pathways in HAFs after E2 treatment. (E) Unique motifs enriched at the promoter and distal regions from both ChIP-Seq and ATAC-seq after E2 treatment.

By integrating RNA-Seq, ESR1 ChIP-Seq, and assay for transposase-accessible chromatin using sequencing (ATAC-seq) data, we identified a core set of genes and pathways regulated by activated E2/ESR1 signaling in HAFs in all 3 datasets from Aromhum LAM (Figure 3C). This contained 58 E2/ESR1-upregulated genes, including well-known E2/ESR1-responsive genes such as Pgr, Pbx1, and several E2/ESR1-related profibrotic genes (e.g., Adamts6, Fbln7; Supplemental Table 1). The increase in expression of Pgr, a hallmark E2-responsive gene, further demonstrates the successful activation of E2 pathways in HAFs. This elevation was also observed in our prior scRNA-Seq study of Aromhum LAM (Supplemental Figure 4, A and B) (9). Additionally, Ltbp1, an ECM protein involved in TGF-β signaling, was consistently upregulated. The mechanotransduction modulator Piezo2, the cell adhesion molecule Ncam1, and the semaphorin receptor Nrp2 were other notable genes upregulated with E2/ESR1 signaling in HAFs (Supplemental Table 1). Functional enrichment analysis of integrated ESR1 ChIP-Seq and RNA-Seq data highlighted the activation of key profibrotic pathways associated with fibroblast proliferation and ECM formation, including WNT, ubiquitin-mediated proteolysis, N-glycan biosynthesis, TGF-β, Hedgehog, and chemokine signaling, in response to E2 treatment (Figure 3D). In contrast, inhibition of E2/ESR1 signaling by fulvestrant uncovered a common set of 34 genes, including cell cycle inhibitors (e.g., Wee1, Cdkn1c, Cdc7) and pathways related to post-transcriptional and translational regulation machinery, as well as phagocytosis and endocytosis (Supplemental Figure 8, A and B, and Supplemental Table 2).

Differential motif analysis from ChIP-Seq and ATAC-seq suggested highly significant enrichment of PLAG1 binding sites adjacent to ESR1 binding sites in distal genomic regions in E2-treated HAFs (Figure 3E). Interestingly, previous research showed that the ectopic expression of PLAG1 in skeletal muscle induces fibrosis and atrophy (22). While we did not observe a significant change in Plag1 expression with fulvestrant treatment, several other Plag1-like genes (Plagl1, Plagl2, Plag2l2) may bind to similar motifs and influence downstream responses. We also identified other enriched regulatory elements unique to E2-treated HAFs, such as NR3C2 (distal region) and HIF1A (promoter region), which were also previously implicated in tissue fibrosis (Figure 3E and Supplemental Figure 8C) (23, 24). Combined network analysis revealed perturbations in pathways associated with cellular and tissue morphogenesis, mesenchymal development, matrisome core, and cell-cell signaling in response to E2 treatment, whereas E2 with fulvestrant led to the enrichment of regulatory transcriptional and developmental pathways (Supplemental Figure 8, D and E). Furthermore, fulvestrant treatment led to an upregulation of apoptotic pathways, as well as heat shock and hypoxic response genes, suggesting that the absence of E2 causes cells to stall in the cell cycle and undergo apoptosis (Supplemental Figure 5D, Supplemental Figure 6C, and Supplemental Figure 8, B and E). Overall, these findings support the notion that E2/ESR1 action at distal genomic regions contributes to the fibrotic pathogenicity of HAFs and that fulvestrant treatment reverses fibrosis by reducing HAF activation and inducing tissue repair pathways.

Validation of protein and mRNA expression of E2/ESR1 profibrotic genes identified by multiomics analyses of HAFs. First, we verified protein expression of key E2/ESR1-responsive genes (Pgr, Pbx1) and E2/ESR1-related profibrotic genes (Adamts6, Piezo2, Ncam1) identified in multiomics analyses. In E2-treated primary HAFs from Aromhum mice, we showed expression of PGR, PBX1, ADAMTS6, PIEZO2, and NCAM1 (Figure 4, A and B). Immunofluorescence staining demonstrated coexpression of these proteins with PDGFRA, which can localize in various cellular compartments, including the cell membrane, nucleoplasm, and gap junctions (Figure 4B) (25–28). We also used NIH 3T3 fibroblasts with low ESR1 expression as surrogate controls for fibroblasts from WT mice. As expected, these proteins were either absent or minimally expressed in NIH 3T3 control fibroblasts (Figure 4, A and B). Additionally, we quantified expression of PGR, PIEZO2, CCN3, and PBX1 through flow cytometry following E2 and fulvestrant treatments (Supplemental Figure 4C). Next, we demonstrated significantly increased in vivo mRNA expression of the key E2/ESR1-target genes using the tissues of Aromhum mice (see Figure 2B), which developed large hernias for 12 weeks (Figure 4C). We verified the upregulation of the core E2/ESR1-related profibrotic genes, including Fbln7, Piezo2, Ltbp1, Ncam1, and Nrp2, in the LAM of placebo-treated Aromhum mice. Further, treatment of Aromhum mice with the E2/ESR1 antagonist fulvestrant significantly decreased LAM mRNA levels of Fbln7, Piezo2, and Ltbp1 (Figure 4C). Fulvestrant treatment also decreased expression of Ncam1 and Nrp2, though this did not reach significance (Figure 4C). These in vitro and in vivo data suggest that a critical signature of E2/ESR1-responsive profibrotic genes identified from multiomic genome-wide analyses may be responsible for increased fibroblast proliferation and ECM production, leading to LAM fibrosis and herniation.

Validation of E2/ESR1-modulated genes in vitro and in vivo. (A) In vitro staining of primary cultured HAFs and NIH 3T3 control cells for ESR1. (B) PDGFRA- and E2/ESR1-regulated genes identified from multiomics analyses in HAFs and NIH 3T3 cells (n = 3–5 mice for HAFs, 3–6 technical replicates; scale bars: 200 μm). (C) mRNA expression of the E2/ESR1-targeted genes identified via multiomics analyses of LAMs from Aromhum mice in the fulvestrant treatment study shown in Figure 2B (n = 4–5 per group, mean ± SEM, 2-way ANOVA with Bonferroni multiple comparisons). Plc, placebo; Fulv, fulvestrant.

Given the established role of Pbx1 as a pioneer factor for ESR1 in breast cancer cells and its necessity for E2 signaling, we conducted siRNA knockdown of Pbx1 to assess its involvement in E2-induced hernia pathogenesis (Figure 5, A and B) (3, 29). HAFs with successful Pbx1 knockdown exhibited reduced DNA content following E2 treatment as compared with E2-treated control siRNA knockdown HAFs, suggesting an impairment in cell cycle progression (Figure 5B). We further performed flow cytometry to assess the effects of Pbx1 on the cell cycle. In control siRNA knockdown cells, E2 treatment significantly decreased the percentage of cells in G0/G1 phase with a concomitant increase in the percentage of cells in S and G2 phases (Figure 5C). Pbx1 knockdown eliminated the effect of E2 in all phases of the cell cycle (Figure 5, C and D). These findings indicate that E2’s proliferative effects are, in part, mediated through Pbx1. Similar results were observed with the knockdown of Ccn3 (Nov), which led to reduced cell proliferation and impacted the downstream production of β-catenin, a known intermediary in Ccn3 signaling (Supplemental Figure 9) (30).

Pbx1 plays a key role in mediating E2-driven proliferation of HAFs. (A) Pbx1 RNA expression at various siRNA concentrations (left) and following vehicle or E2 treatment (right; 25 nM si-Pbx1) (n = 3, mean ± SEM, 2-way ANOVA with Bonferroni multiple comparisons). (B) DNA content in HAFs treated with vehicle or E2, with and without Pbx1 knockdown (n = 3, mean ± SEM, 2-way ANOVA with Bonferroni comparisons). (C and D) Flow cytometry scatterplots of HAFs (C) and their distribution across cell cycle stages (G0/G1, S, and G2 phases) (D) following Pbx1 knockdown and E2 treatment (n = 3 per group, χ2 test for proportions).

E2/ESR1-modulated mRNA or protein expression in Aromhum LAM is comparable to that observed in men with inguinal hernias. We analyzed LAM from men undergoing hernia surgery to examine E2/ESR1-mediated mRNA and protein expression and associated histological changes in human inguinal hernias. We collected matched biopsies from the herniated LAM and adjacent healthy-appearing LAM from 25 men undergoing hernia repair surgery (21–76 years of age). The adjacent healthy-appearing tissue exhibited lower levels of fibrosis (<15%), consistent with our previous findings in LAM tissues from nonhernia patients (Supplemental Figure 10A) (6). Moreover, this 15% threshold aligns with the collagen levels observed in WT mice without herniation, providing a meaningful baseline (Figure 1E). We observed extensive muscle fibrosis containing atrophic myofibers in human herniated LAM by Masson’s trichrome staining, with fibrosis ranging from 5% to 70% (Figure 6, A and B). Immunoreactive PDGFRA was found only in stromal fibroblasts but not myofibers (Figure 6C). Expression of ESR1 and the cell proliferation marker Ki67 were observed in a strikingly higher number of cells in hernia site LAM compared with adjacent healthy muscle (Figure 6, E and F). Expression of PDGFRA, ESR1, and Ki67 was significantly higher in LAM from herniated samples (16%–70%) compared with healthier tissues (<15%; Figure 6, D, G, and I). Moreover, we observed a significant correlation between the expression of ESR1 and Ki67 and the degree of fibrosis in herniated LAM (Figure 6H).

Histopathology of LAM in men with inguinal hernias. (A–C, E, and F) Representative images of H&E (A), Masson’s trichrome (B), and immunohistochemistry staining for PDGFRA (n = 25 patients) (C), ESR1 (n = 34 patients) (E), and Ki67 (n = 25 patients) (F) in human LAM from inguinal hernia sites and adjacent healthy muscle tissues (t test; scale bars: 100 μm). Original magnification, ×20 (C and E, insets). Yellow arrows point to atrophied myofibers, while black arrows point to positive staining. (D, G, and I) Quantification of PDGFRA+ (D), ESR1+ (G), and Ki67+ (I) nuclei from C, E, and F, respectively, stratified by the percentage of fibrosis observed. (H) Spearman’s ρ (rs) correlation between percentage fibrosis, ESR1, and Ki67 scores (44 samples from 22 patients, median ± interquartile range).

To gain deeper molecular insights, we used RNA in situ hybridization and verified the expression of E2/ESR1-modulated genes identified in Aromhum LAM (e.g., NCAM1, LTBP1, ADAMTS6, NRP2, PBX1, and PIEZO2) in the fibrotic regions of herniated muscle tissue from men (Figure 7, A–D, and Supplemental Figure 10, B–D). Moreover, both NCAM1 and LTBP1 were consistently expressed in all patient samples, providing evidence of their involvement in the hernia development (Figure 7, A and B). Additionally, we detected PGR protein expression via immunohistochemistry (Supplemental Figure 10D) and mRNA expression of ADAMTS6, NRP2, PBX1, and PIEZO2 in more than 30% of the herniated and fibrotic LAM samples (Figure 7, C and D, and Supplemental Figure 10, B and C). Overall, our findings demonstrate that the activation of E2/ESR1 signaling in LAM fibroblasts from a large subset of men with inguinal hernias is similar to that observed in HAFs from Aromhum mice, emphasizing the clinical relevance of E2/ESR1 signaling in inguinal hernias in men.

E2/ESR1-modulated genes in men with inguinal hernias. Representative RNAscope images of the genes NCAM1 (A) and LTBP1 (B) identified from multiomics studies that were observed in all patient samples and their quantification, stratified by the size of the fibrotic area (n = 12 tissues from 6 patients denoted by different shapes, mean ± SEM; scale bars: 200 μm). RNAscope images of PBX1 (C) and PIEZO2 (D) identified from multiomics studies that were observed in some patient samples. Black arrows point to positive staining (n = 8 tissues from 4–5 patients denoted by different shapes, mean ± SEM, nested t test; scale bars: 200 μm). Original magnification, ×40 (A–D, insets).