Loss of the repressor REST affects progesterone receptor function and promotes uterine leiomyoma pathogenesis.

All human studies were approved by and performed in accordance with the University of Kansas Medical Center Policies and Procedures Relating to Human Subjects (IRB#: 5929). All patients gave informed consent when donating their tissue samples to the University of Kansas Cancer Center’s Biospecimen Repository Core Facility. All mouse experiments were approved by the University of Kansas Medical Center Institutional Animal Care and Use Committee protocols (2019-2508) and adhere to NIH guidelines for care and use of laboratory animals. Matched myometrium and leiomyoma tissue samples were obtained from premenopausal women undergoing hysterectomies at the University of Kansas Hospital (Kansas City, KS). Criteria for patients in this study exclude women undergoing hysterectomy for a primary condition other than UL and those taking hormone therapy in the 3 mo preceding surgery. Myometrial primary cells were prepared from samples as described previously ( 8 ). Details of the generation of mouse models and gene and protein expression analysis are described in SI Appendix , Supplemental Methods .

Previously, we had shown that the loss of REST protein in leiomyoma patient samples leads to derepression of its target genes, including GPR10 ( 8 ). To investigate the loss of REST in vivo, we developed a cKO of Rest in the mouse uterus since the conventional Rest KO is embryonically lethal ( 19 ). Chimeric founder mice were generated using embryonic stem cell clones in which exon 3 of Rest was floxed ( Fig. 1 A and B ) and homozygous reproductive tract-specific Rest f/f Amhr2 Cre/+ cKO mice were generated, as described in SI Appendix , Supplemental Methods . The Rest f/f Amhr2 Cre/+ cKO mice had an increased uterine size compared with the control ( Fig. 1 C and SI Appendix , Fig. S1 ). Uteri of Rest f/f Amhr2 Cre/+ mice were consistently hypertrophic, contained cystic glands, and had abnormal uterine morphology compared with control mice ( SI Appendix , Fig. S1 ). In addition, the cKO mice developed distinct fibroid tumors within the uterus that were not seen in control mice ( Fig. 1 D and E ). Immunostaining of mice uteri showed an increase in myometrial thickness and excessive deposition of ECM components, including collagen 3A1 ( SI Appendix , Fig. S2 ). Loss of Rest under the Amhr2 promoter displays a UL phenotype. ( A ) Rest f/f targeting construct. ( B ) PCR genotyping of Rest fl/fl mice (lanes 4 and 5) Rest heterozygous mice (lanes 2, 3, and 7), and control mice (lane 6). ( C ) Representative image showing increased uterine size in 6-mo-old Rest f/f Amhr2 Cre/+ cKO mouse compared with control mouse during diestrus. ( D ) H&E stain of control mouse uterus. (Scale bar, 500 μm.) ( E ) H&E stain of Rest f/f Amhr2 Cre/+ cKO with a tumor indicated by the red arrow. (Scale bar, 500 μm.) Images are at 4× magnification. ( F ) Western blot showing increased GPR10 expression and activation of downstream tumorigenic signaling proteins in total uterine tissue of Rest f/f Amhr2 Cre/+ cKO mice compared with control and Rest fl/+ Amhr2 Cre/+ mice. GAPDH was used as a protein loading control. ( G ) Gene-expression analysis of Rest and Rest targets, Gria2 , Stmn2 , and Stmn3 in Rest f/f Amhr2 Cre/+ compared with control ( n = 4). Error bars represent ± SEM. Student’s t test was performed, * P < 0.05, ** P < 0.01, *** P = 0.0001. Similar to the aberrant expression of GPR10 in a UL specimen in which REST was lost ( 8 ), Western blot analysis showed the ablation of Rest in the cKO mouse uterus resulted in an increase in GPR10 expression ( Fig. 1 F ). In addition, the cKO mouse uteri contained higher levels of phosphorylated forms of AKT, 4EBP1, and p70S6K, which are known mediators ( 8 , 20 ) of the dysregulated PI3K/AKT-mTOR pathway in UL ( Fig. 1 F ). qRT-PCR confirmed a significant decrease in Rest mRNA expression in the Rest f/f Amhr2 Cre/+ cKO mice ( Fig. 1 G ). Additionally, REST target genes Gria2 , Stmn3 , and Stmn2 were all significantly overexpressed in the cKO mouse ( Fig. 1 G ), confirming our earlier findings that loss of REST leads to overexpression of its target genes, including GRIA2 , STMN2 , and STMN3 ( 8 ). Many putative REST target genes that have known roles in the uterus and significantly dysregulated in human UL, were also dysregulated in our Rest f/f Amhr2 Cre/+ cKO mice, although species-specific differences in gene regulation were present ( SI Appendix , Table S1 ). To further characterize the Rest f/f Amhr2 Cre/+ cKO mice, gene-expression profiles of the cKO and control mice in similar estrous cycle stage (diestrus) were analyzed using RNA sequencing ( GSE178141 , token qrcdqusibfaxjub). Ingenuity Pathway Analysis (IPA, Qiagen) revealed significant similarities in dysregulated gene-expression profiles in human UL and the Rest f/f Amhr2 Cre/+ cKO mouse model, with most significant disease pathways in the KO mouse uteri being analogous to leiomyomatosis, smooth muscle tumor, benign neoplasia, and leiomyoma ( Table 1 ). Furthermore, the gene-expression profile of the cKO mouse showed similarities to dermatological disorders ( Table 1 ), hepatic fibrosis, and retinoic acid signaling ( SI Appendix , Figs. S3 A–C ), diseases of the reproductive and nervous systems, all bearing overlapping cellular and molecular characteristics with UL ( 21 – 23 ). Finally, the IPA confirmed REST to be down-regulated and its network of targets to be significantly dysregulated, as anticipated in the cKO model ( SI Appendix , Fig. S3 B and D ). Pathway analysis of functions or predicted diseases associated with dysregulated genes in UL and cKO mice Disease predications made by IPA software based off genes which were found to be dysregulated in both human UL (GEO dataset GSE13319 ; n = 23) and RNA-sequencing results (GEO dataset GSE178141 ( 24 ); n = 3) of 4-mo-old Rest f/f Amhr2 +/Cre cKO in diestrus. A highly relevant prediction of the IPA was how the loss of REST affected its upstream regulators, such as estrogen signaling ( SI Appendix , Fig. S3 B ). IPA identified β-estradiol as a highly significant upstream regulator of genes expressed in the Rest f/f Amhr2 Cre/+ cKO, showing an activation of estrogen receptor signaling in the absence of REST. Indeed, the RNA sequencing on the cKO mouse uteri identified ESR1-associated genes as significantly ( P < 0.05) dysregulated in these mice ( SI Appendix , Fig. S4 ). Moreover, we observed an increase in uterine size and hypertrophy in the cKO mice during diestrus when estrogen and progesterone were low and high, respectively ( Fig. 1 C and SI Appendix , Table S3 ). To study the impact of loss of REST in the uterus, we performed single-cell RNA-sequencing using Rest f/f PR Cre/+ cKO mice, instead of the Rest f/f Amhr2 Cre/+ cKO mice, mainly due to the frequent embryonic lethality of Rest cKO due to leaky Cre expression under the Amhr2 promoter, as reported by others ( 25 , 26 ). While the complete phenotypic characterization of the Rest f/f PR Cre/+ is still ongoing, preliminary studies showed an increase in the overall uterine size in 6-mo-old mice compared with age-matched control uteri ( SI Appendix , Fig. S5 ). Additionally, the uteri of the Rest f/f PR Cre/+ cKO mice were hypertrophic with cystic glands and abnormal lumen structure ( SI Appendix , Fig. S5 ), similar to Rest f/f Amhr Cre/+ cKO mice. Results from Seurat single-cell RNA-sequencing data analysis ( 27 ) on uteri from 5-mo-old Rest f/f PR Cre/+ cKO mice and littermate Rest f /f controls ( GSE178141 ), which passed the quality-control markers before being analyzed ( SI Appendix , Fig. S6 ), identified 19 different clusters in the uterus based on their expression profiles ( Fig. 2 A ). Cell types were determined by gene-expression profiles in each cluster using the SingleR software ( 28 ) and expert curation. A χ 2 test of homogeneity comparing control and Rest f/f PR Cre/+ cKO single-cell RNA-sequencing cell counts revealed significant differences in population of neutrophils, epithelial, stromal, and myometrial cells ( Fig. 2 B and SI Appendix , Fig. S7 ). The results showing an increased presence of neutrophils corroborates with reported up-regulation of these cells in human UL ( 29 ). Based on conserved gene expression, a cluster of cells representing SMCs ( Acta2 , Cnn1 , Myh11 , Actg2 , Myl9 , Tpm2 , Pcp4 , Mylk , and Tagln ) ( SI Appendix , Fig. S8 ), as well as uterine stromal, myometrial fibroblast lineage ( Slco5a1 , Col5a2 , Col6a3 , Dpt , Ddr2 , Adamts4 , Ifi205 , Tgfb2 , and Cd34 ) ( SI Appendix , Fig. S9 ) were identified. Analysis of top TCA features (average cluster expression) of genes showed up-regulation of ECM components in myometrial and stromal fibroblasts, as well as SMCs (clusters 0 to 4, 8, and 17) in Rest f/f PR Cre/+ mice ( SI Appendix , Fig. S10 ). Single-cell RNA-sequencing analysis. ( A ) A t-distributed stochastic neighbor embedding (t-SNE) plot of the hierarchical clustering from uteri of 5-mo-old control and Rest f/f PR Cre/+ cKO mice showed 19 distinct clusters. Populations of cells included epithelial, neutrophils, mast, immune, myometrial, endothelial, natural killer (NK), and stromal cells. ( B ) Individual t-SNE plots of control and Rest f/f PR Cre/+ cKO showed differences in subpopulations. Uteri of Rest f/f PR Cre/+ cKO mice showed increases in REST target gene expression, including overexpression of Gria2 ( SI Appendix , Fig. S11 A ), Stmn3 ( SI Appendix , Fig. S11 B ), and Stmn2 ( SI Appendix , Fig. S11 C ). Additionally, up-regulation of REST target genes was found in the stromal, epithelial, and myometrial cell clusters as represented in a violin plot ( SI Appendix , Fig. S12 ). Moreover, genes downstream of estrogen receptor α (ERα) signaling, including Mmp24 (matrix metallopeptidase 24) and Snap25 (synaptosome associated protein 25), were activated in the Rest f/f PR Cre/+ cKO mice ( SI Appendix , Fig. S11 D and E ), indicating enhanced estrogen signaling upon the loss of REST. IPA analysis of stromal and myometrial cells (Cluster 0, fibroblast lineage) in cKO uteri showed gene-expression profiles similar to those found in benign solid tumors ( SI Appendix , Fig. S13 A ). In addition, estrogen signaling genes were altered in this cluster ( SI Appendix , Fig. S13 B ). IPA also identified an adenomyosis phenotype in this cluster, which needs to be further explored ( SI Appendix , Fig. S13 C ). Finally, the Rest f/f PR Cre/+ cKO mice had higher levels of collagens and Acta2 compared with control uteri ( SI Appendix , Figs. S14 and S15 ). This phenotype is consistent with data from Rest f/f Amhr Cre/+ cKO mice. To faithfully recapitulate the loss of REST in human UL, we next developed a cKO mouse model in which REST was specifically deleted in the myometrial layer of the uterus. Using a proximal promoter sequence from the rat calbindin-D9K ( CaBP9K ) promoter, which drives transgene expression specifically in myometrium ( 8 , 30 ), we generated myometrial-specific Cre recombinase transgenic mice (CaBP9K-iCre, Myometrial-specific M-iCre or MiC) ( Fig. 3 ). Myometrial specificity of iCre expression was confirmed by crossing MiC mice with Rosa mt/mG reporter mice ( Fig. 3 B and SI Appendix , Fig. S16 ). The Rest cKO mouse, Rest f/f M-iCre cKO, was generated by breeding the Rest f/+ MiCre mice with the Rest f/f mice. Generation and characterization of Rest cKO mouse under a myometrial-specific iCre (MiC) ablation. ( A ) Transgenic construct for the myometrial-specific iCre expression using the proximal region of the rat Cabp9k promoter. ( B , Left Upper ) ROSA mT/mG mouse uterine horn expressing tdTomato. ( Left Lower ) MIC ROSA mT/mG mouse uterine horn expressing EGFP in the myometrium after tdTomato was removed using the myometrial-specific iCre, magnification at 4× (Scale bar, 2 mm). ( Right ) EGFP expression is specific to the mouse myometrium, magnification at 10×, (Scale bar, 100 μm). ( C ) H&E stain of uterine horn section in a control mouse. (Scale bar, 500 μm.) ( D ) H&E stain of uterine horn showing changes in morphology of Rest f/f MiC mouse. Tumor formation in the Rest f/f MiC mouse indicated by arrow. Magnification at 4×. (Scale bar, 500 μm.) ( E and F ) Representative image of increased uterus size of Rest f/f MiC mouse compared with control uterus during diestrus. ( G ) Gene expression of known Rest targets Stmn3 , Stmn2 , and Gria2 in 6-mo-old Rest f/f MiC mice compared with control ( n = 5). Error bars represent ± SEM. Student’s t test was performed, * P < 0.05, ** P < 0.01. Similar to the Rest f/f Amhr2 +/Cre mice, the Rest f/f M-iCre cKO mice had increased uterine size, cystic glands, and abnormal uterine morphology compared with control uteri in diestrus ( Fig. 3 E and F and SI Appendix , Fig. S17 ). Additionally, we observed SMC tumors expressing αSMA, originating in the myometrial layer of the uterus of the Rest f/f M-iCre cKO mice ( Fig. 3 C and D and SI Appendix , Fig. S18 ). Furthermore, REST target genes Stmn3 , Stmn2 , and Gria2 were all significantly overexpressed in the Rest f/f M-iCre cKO mouse, confirming the loss of REST function ( Fig. 3 G ). ULs are known to be associated with altered expression of genes involved in transforming growth factor (TGF)-β signaling, SMCs, and ECM components ( 31 ). Using uteri from 6-mo-old mice, we tested the expression of Tgfb3 , dermatopontin ( Dpt ), α-smooth muscle actin ( Acta2 ), and collagens Col1A1 and Col3A1 to determine if they contributed to the increased uterus size. Results indicated that the Rest f/f M-iCre cKO mice expressed higher levels of Col3A1 , and statistically significantly higher levels of Dpt , Col1A1 , Tgfb3 , and Acta2 compared with control mice ( SI Appendix , Fig. S19 ). Dysregulation of these genes in the Rest f/f M-iCre cKO mice—except for Dpt , which showed an increase in expression unlike the decrease reported in UL—represents a phenotype similar to human UL. Our three Rest cKO models showed altered signaling, including activation of the ER pathway as well as uterine hypertrophy during diestrus ( Fig. 1 C ), even when progesterone was present. To understand how REST regulated this phenotype, we investigated whether REST had a direct relationship with the steroid hormone receptors, ER and progesterone receptor (PGR). Using available chromatin immunoprecipitation-sequencing (ChIP-seq) datasets for REST, PGR-A, PGR-B, and ERα ( SI Appendix , Table S2 ) [ GSE62475 , GSE36455 ( 32 , 33 )], we identified a high frequency of conserved RE1 sites associated with REST within 100 bp of PGR-A binding sites ( Fig. 4 A ). This conserved relationship was found on roughly 200 REST target genes. Additionally, analysis of binding sites of PGR-B and REST in available ChIP-seq data identified conserved binding sites within 300 bp of each other ( SI Appendix , Fig. S20 ). Interestingly, there was no overlap or proximal relationship between ERα and RE1 binding sites ( Fig. 4 B ). These results indicate that the interaction between REST and PGR is unique. REST’s interactions with steroid hormone receptors. ( A ) A conserved frequency of occurrence of REST binding sites (0) within 100 bp of PGR-A binding sites. The x axis represents distance in base pairs and y axis represents density. ( B ) No overlap or proximal relationship between REST binding sites (0) and ERα binding sites. The x axis represents distance in base pairs and y axis represents density. ( C and D ) Representative image of co-IP of REST and PGR in human myometrial tissue and decreased interaction in leiomyoma tissue. ( E ) Working model depicting a REST–PGR interaction in the normal myometrium and loss of REST’s impact in leiomyoma. Further analysis of available ChIP-seq data revealed that there were 1,598 PGR-A ChIP binding sites with at least one REST ChIP binding site in its vicinity. There were 2,215 PGR-B ChIP binding sites with at least one REST ChIP binding site in its vicinity. Of these sites, 1,135 sites were common to both PGR-A and PGR-B ( Datasets S1 and S2 ). We analyzed the putative roles of REST/PGR-A and REST/PGR-B target genes using gene ontology (GO) enrichment analysis. First, top enriched GO-terms for genes with PGR-A/PGR-B and REST binding sites within 10k from their transcription start sites (TSSs) were plotted ( SI Appendix , Figs. S21 and S22 ). Then, we analyzed the top-level GO-biological processes associated with genes having a PGR-A/PGR-B and REST binding sites within 10k from their TSSs ( SI Appendix , Figs. S23 and S24 ). As expected, GO terms related to distinct pathways were enriched (significantly associated) with REST/PGR-A and REST/PGR-B regulated genes. We next hypothesized that the absence of interaction between REST and PGR due to the loss of REST could account for altered sex steroid hormone signaling seen in UL and in our mouse models. Results from coimmunoprecipitation (co-IP) studies with PGR antibodies on healthy human myometrial tissue and paired UL specimen showed association of REST with PGR in myometrial samples ( Fig. 4 C ). This interaction was reduced in the leiomyoma sample where REST expression was low ( Fig. 4 C ). Reciprocal IP with REST antibodies confirmed IP results with PGR antibodies ( Fig. 4 D ). These results indicate that REST and PGR interact in the healthy myometrium. To further study the interaction between REST and PGR, expression of REST–PGR target genes that contained conserved binding sites were further investigated. Many of these targets were found to be dysregulated in the Rest f/f Amhr2 +/Cre cKO mice ( SI Appendix , Table S4 ). Among the top 24 dysregulated REST–PGR targets, which were also dysregulated in UL ( SI Appendix , Table S5 ), the cell migration inducing hyaluronidase 1 (CEMIP, KIAA1199) contained both RE1 and PGR binding sites within 1,000 bp of each other ( Fig. 5 A ). We also found CEMIP to be significantly up-regulated in human UL specimens compared with a healthy myometrial specimen ( Fig. 5 B ) and correlated with low levels of REST ( Fig. 5 B ). Similarly, Cemip was found to be overexpressed in the Rest f/f Amhr2 +/Cre cKO, and significantly overexpressed in the Rest f/f M-iCre cKO mice uteri, at both the mRNA and protein levels ( Fig. 5 C and D ). Finally, in order to investigate the effect of the loss of REST on PGR binding to its target sequences, we performed ChIP-PCR experiments in uterine tissue from Rest f/f and Rest f/f PR Cre/+ mice (three pairs each, 6-mo old). Interestingly, ChIP-PCR results indicated that in the absence of REST there was significant increase in PGR binding to a PGRE/RE1 locus upstream of the TSS and a potential trend in the same direction in a locus within the first intron of CEMIP ( SI Appendix , Fig. S25 ). A detailed picture of REST-dependent PGR binding and regulation of its target genes may emerge in the future from sequencing of ChIP samples generated in the above study. Our results indicate that, upon the loss of REST, PGR binding to its target genes could be altered in a gene environment-dependent manner and may lead to aberrant expression of PGR–REST target genes. ( Fig. 4 E ). Overexpression of CEMIP in uterine leiomyoma and Rest cKO mice. ( A ) Human CEMIP gene locus showing REST binding sites (RE1 sites) and PGR binding sites (PGRE), and their locations relative to the TSS (+1). ( B , Left ) Western blot showing CEMIP overexpression and loss of REST in leiomyoma patient samples compared with control myometrium. ( Right ) TaqMan qRT-PCR of CEMIP in leiomyoma patient samples compared with control myometrium ( n = 11). ( C , Left ) Western blot showing CEMIP overexpression and loss of Rest in Rest f/f Amhr2 Cre/+ cKO compared with control. ( Right ) TaqMan qRT-PCR of Cemip in Rest f/f Amhr2 Cre/+ cKO compared with control ( n = 4). ( D , Left ) Cemip overexpression shown by Western blot in Rest fl/fl M-iCre cKO compared with control. ( Right ) TaqMan qRT-PCR of Cemip in Rest f/f M-iCre cKO compared with control ( n = 4). β-Actin was used as a loading control for Western blots. Error bars represent ± SEM. Student’s nonparametric t test was performed, * P < 0.05.

Despite the long-recognized roles of epigenetic and environmental factors that predispose women to UL, molecular events that lead to the initiation of UL growth and pathogenesis are still not well understood ( 34 – 36 ). Here we provide evidence that the loss of REST, a known tumor suppressor and a major epigenetic regulator of gene expression, plays a significant role in UL pathogenesis. We provide evidence that the loss of REST function leads to aberrant expression of REST target genes that contribute to cell proliferation and ECM accumulation in the uterus. We further provide evidence that loss of Rest in tissue-specific cKO mouse models leads to UL formation, and changes in uterine morphology and gene-expression profiles with extensive similarities to human UL. Crucially, we show Rest cKO mice have altered responses to endogenous hormones, estrogen, and progesterone, due to a novel interaction between REST and PGR. REST is a master epigenetic silencer that represses neuronal genes in nonneuronal cells ( 37 ) through binding to a 21- to 23-bp repressor element sequence (RE1 site) located on an estimated 2,000 target genes within the human genome ( 38 ). Previously, we demonstrated that loss of REST leads to derepression of GPR10 ( PRLHR ), which plays a role in UL pathogenesis by activating the PI3K/AKT-mTOR pathway ( 8 , 39 , 40 ). Additionally, PRLHR was shown to be overexpressed ubiquitously in a UL specimen carrying wide range of known mutations or translocations ( 4 ), indicating that loss of REST may be a crucial upstream event in uterine fibroids ( 8 ). Our data demonstrate that loss of Rest in vivo leads to tumorigenesis in uterine SMCs and accumulation of ECM. Preclinical models that accurately recapitulate altered steroid hormone pathways in UL, which are necessary for the development of safe and efficacious treatments for UL, are nonexistent in the field. Although Eker rats, which harbor a mutation in the Tsc2 gene ( 41 , 42 ), develop spontaneous, estrogen-sensitive UL tumors, they also develop fatal renal and liver cancers ( 43 ). Additionally, UL growth in this model can only be induced by estrogen and not progesterone ( 44 ). In addition, formation of UL in another model in guinea pig is also dependent on estrogen ( 45 ). These tumors do not have similar histological features to human UL and are inhibited by progesterone treatment ( 3 ). Although a recently developed xenograft model of UL shows tumor growth in the presence of estrogen and progesterone ( 46 ), this model is not useful to study initiation and development of UL. There is, therefore, an urgent need for UL-relevant animal models which are sensitive to estrogen and progesterone, and represent cellular, molecular, and genetic features of human UL; the Rest f/f M-iCre cKO and Rest f/f Amhr2 Cre/+ cKO mouse models fulfill this unmet need in the field. Specifically, the Rest cKO mice showed increased uterine size, SMC proliferation, and ECM deposition, which are key features of human UL tumors. Similar to increased sensitivity of human UL to estrogen and the abnormal expression of estrogen-responsive genes during the luteal phase ( 1 ), the cKO mice showed increases in uterine size during diestrus when estrogen and progesterone were both present. Moreover, results from the single-cell RNA sequencing of the Rest f/f PR Cre/+ cKO mice provided compelling evidence that loss of REST has significant effects on uterine tissue architecture. While Amhr2 Cre/+ and PR Cre/+ drivers are traditionally used to delete floxed genes in the reproductive tract, we developed and utilized a myometrial-specific Cre driver to delete Rest in the myometrium, to further confirm a distinct role of REST in UL pathogenesis. This Cre driver (CaBP9K-iCre, M-iCre) will be invaluable for future studies involving myometrial-specific gene deletion. Importantly, our data show that the loss of REST can lead to alterations in sex steroid hormone signaling and UL development. Our results show: 1) REST and PGR interact in the heathy myometrium, 2) loss of REST may influence PGR interaction with its target sequences, and 3) REST and PGR target genes are dysregulated in UL. One of the novel targets of REST, CEMIP, is significantly overexpressed in both human UL and in our cKO mouse models. CEMIP has been shown to induce fibrosis in arthrofibrosis ( 47 ), and its up-regulation in several cancer types is linked to cell proliferation, migration, and changes in cell signaling, such as PI3K/AKT ( 48 , 49 ), which is overactivated in UL. In addition, CEMIP is an HA binding protein involved in HA depolymerization ( 50 , 51 ). HA is ubiquitously present in the ECM and provides structural integrity to the uterus. It has also been shown that under certain pathological conditions, HA degradation is enhanced, and the lower molecular weight molecules, resulting from HA degradation, can contribute to tumor growth and angiogenesis ( 52 ). We recently showed preferential association of REST with RE1 sites within the CEMIP locus as well as aberrant expression of CEMIP in the absence of REST in breast cancer cells ( 53 ). Based on our results showing loss of REST leads to overexpression of CEMIP, it will be important to test what role CEMIP plays in UL pathogenesis and how PGR and progesterone affect this gene. Current hormone therapies for UL suffer from poor safety profiles, precluding them from long-term use ( 54 , 55 ). Current generation of selective progesterone receptor modulators, which are developed based on PGR antagonism, reduce UL tumor size in randomized control trials ( 16 , 56 ). However, side effects of endometrial hyperplasia have been reported and concerns of their effects on nontargeted tissues (breast, ovary, liver) exist ( 56 ). Our work provides a link between the altered response to progesterone and loss of REST in UL. We have found that REST interacts with PGR and influences the expression of target genes. Future studies are needed to identify isoform selective effects of REST–PGR-A and REST–PGR-B interactions in the uterus since these isoforms are known to have functionally distinct roles in the uterus. We believe this unique relationship between REST and PGR presents a mechanism that can be targeted to develop a new generation of selective progesterone receptor modulators. In addition to the myometrium, the role of REST in other progesterone-responsive tissues, including mammary epithelium and uterine endometrium, needs to be investigated in the future.