Compressive force induces differential gene and protein expression in uterine fibroids.

Once embedded in agarose the spheroids contracted, Figures 1C and D . The change in area was analyzed from the five biologic replicates; the five sets of patient-matched fibroid and myometrial cells. Technical replicates, typically 4–8 per biological replicate, were imaged and analyzed. A total of 35–41 spheroids were imaged for each of the four different cells (fibroid or myometrial) and compression (0 or 6.4 mm Hg) conditions. All spheroids contracted less when exposed to compressive force ( P <.0001); myometrial spheroids contracted more than fibroid spheroids ( P = .022). TUNEL staining was performed to ensure that the cells within the spheroids were viable ( Fig. 1B ). Only one of the myometrial samples evaluated showed evidence of a necrotic core. There was significant variation between samples, as seen in the PCA plots, Figure 2 . Despite these differences, the general shift due to compression was in the same direction for all samples. A summary of the DEGs is provided in Table 1 . It also includes significantly enriched pathways in GO and KEGG databases that were identified using R. Both cell types responded to compression by altering receptor ligand activity ( q = 2.54 × 10 −17 myometrial and q = 7.30 × 10 −11 fibroid). Different pathways between the cell types under compression included muscle tissue morphogenesis ( q = 1.00 × 10 −4 ) and cell-cell adhesion ( q = 1.07 × 10 −5 ). The networks of the top 20 enriched processes and pathways as identified by Metascape analysis of DEGs are shown in Figure 3A – C . To ensure that DEGs had altered expression due to compression and were clinically relevant, additional analysis was performed, starting with the DEGs between cell types under compression. First, we removed all DEGs identified between the cells in the control conditions unless the differences were enhanced by at least 1.5-fold. We then removed all DEGs that were not also DEGs in at least one of the four tissue datasets. Principal component analysis and volcano plots from our analysis of the datasets are shown in Supplemental Figure 1 (available online). The list of remaining clinically relevant DEGs can be found in Supplemental Table 1 (available online). A total of 76 clinically relevant genes were identified as being affected by compression; of these, 61 were only DEGs when the spheroids were under compression. A total of 96 clinically relevant genes were differentially expressed between the cells but were not affected by compression, Figures 3D and E . The compression-related DEGs were related to the ECM (GO, q = 6.99 × 10 −6 ), with a significant enrichment of glycoproteins ( q = 4.02 × 10 −10 ). Of the genes that were only affected by compression, KLK10 was the most overexpressed DEG in fibroids with a log2 fold-change of 5.66 ( q = 1.06 × 10 −15 ). KLK10 was upregulated in fibroids in two of the four tissue datasets. EPHB1 and one of its ligands, EFNB2 were upregulated in fibroid cells compared with myometrial cells only when under compression. EPHB1 was upregulated by 2.81 log2 fold-change and found in three of the four datasets ( q = 5.35 × 10 −3 ); it was upregulated by fibroid cells under compression ( q = 1.06 × 10 −4 ). EFNB2 up upregulated 0.97 log2 fold-change and was found in two of the four datasets ( q = 4.29 × 10 −2 ); it was significantly downregulated in myometrial cells under compression (−1.15 log2 fold-change; q = 4.45 × 10 −21 ). A total of 96 clinically relevant DEGs between the cell type were not affected by compression, Figure 3F . These constant DEGs were enriched for tissue morphogenesis ( q = .0077). Two DEGs of particular interest were COL23A1 and MMP11. COL23A1 was the DEG with the largest difference; fibroid cells expressed this gene l0.03 log2 fold-change more than myometrial cells ( q = 5.35 × 10 −91 ). MMP11 was upregulated in fibroid cells by 2.87 log2 fold-change ( q = 3.08 × 10 −3 ). Collagen XXIII overexpression was confirmed at the protein level in control spheroids ( Supplemental Fig. 2 ). A total of 2,912 proteins were identified in the spheroids; the PCA analysis and heatmap are shown in Figure. 4 . Of these proteins, 1,782 are associated with the membrane (61.2%), 1,157 are associated with the extracellular region (39.7%), and 453 are associated with the cytoskeleton (15.6%). Differentially expressed proteins (DEPs) were identified; an overview is given in Table 2 . Metascape analysis of myometrial spheroid proteins that were affected by compression identified the enrichment of genes associated with uterine fibroids ( q = 6.31 × 10−8) and fibroid tumors ( q = 1.58 × 10−7); this analysis used DisGeNET datasets ( 37 ). In fibroid cells, proteins that changed with compression were enriched for collagen-containing ECM ( q = 7.50 × 10−4). To ensure that differences in the weighted fibroid and myometrial cells were due to compression, DEPs between cell types under compression and control conditions were compared. Those that were DEPs in the control condition were removed from the weighted DEPs unless the abundances were enhanced by at least 1.5-fold. The STRING networks are depicted in Figures 4B and D . The list of clinically relevant DEPs is given in Supplemental Table 2 . A total of 150 proteins were only different in fibroid vs. myometrial spheroids when under compression. A total of 99 additional DEPs had differences that were enhanced by compression, whereas 23 remained consistently altered between the cell types. No differences in collagen I were found between cell types under compression; however, the collagenase MMP-1 was downregulated in these fibroid spheroids −1.85 log2 fold-change ( q = 4.66 × 10 −12 ). Of the 813 DEGs in the myometrial cells due to compression, 35 were also found as DEPs (4.3%). For fibroid cells, 33 of the 853 DEGs were also DEGs (3.9%), in control spheroids 18 of the 600 DEGs were DEGs (3.0%), and in compressed spheroids, 11 of the 317 DEGs were also DEGs (3.5%). Differentially expressed genes of interest were confirmed at the protein level, either confirmed with proteomics or by immunohistochemistry or western blot, Figure 5 . Additionally, we investigated collagen XXIII and EphB1 at the protein level. These did not show up in proteomics data due to the relatively low abundance compared with ECM proteins. Collagen XXIII was a DEG with extremely high fold-change despite not being affected by compression. It was almost undetectable in fibroid cultures. EphB1 was identified in all samples; however, due to significant sample variation, no significant differences were found at the protein level.

From June 2021 through August 2023, tissues were obtained from patients undergoing surgical treatment of uterine fibroids who provided informed consent per our protocol that was approved by the University of Cincinnati (UC) institutional review board. The five patients whose tissue was used for this study were pre-menopausal, with intramural fibroids, who had not taken hormonal modulation within 3 months of surgery. Surgeries were performed at the University of Cincinnati Medical Center. The patients had a mean age of 40 ± 4 years. Three underwent myomectomy whereas 2 had a hysterectomy. The biopsies and cells were isolated as described in Warwar et al. ( 25 ). Briefly, after tissue resection by pathology, samples were transported in Hanks’ Balanced Salt Solution (HBSS; Gibco, Grand Island, NY, USA) with 3% antibiotic-antimycotic (#15240096, Gibco). The psuedocapsule was not included. Tissues were rinsed with Dulbecco’s Phosphate-Buffered Saline (DPBS; Cytiva, Westborough, MA, USA). Tissues were dissected into small pieces, excluding the center of the fibroid (>0.5 cm from the edge). Cells were isolated from the fragments using a combination of enzymatic digestion and explant migration techniques as previously described ( 25 ). Briefly, tissue fragments were incubated in 600 U/mL collagenase II (Worthington Biochemical, Lakewood, NJ, USA) in Hanks’ Balanced Salt Solution (HBSS; Gibco, Grand Island, NY, USA). The tissues were resuspended in human smooth muscle cell (hSMC) culture medium consisting of Dulbecco’s Modified Eagle Medium (DMEM; Corning, Tewksbury, MA, USA) with 5% fetal bovine serum (FBS; Corning), 1% antibiotic-antimycotic (Gibco), 5 ng/mL basic fibroblast growth factor (FGFß; Gibco), 5 μ g/mL insulin (Sigma-Aldrich, St. Louis, MO, USA), 5 ng/mL epidermal growth factor (Peprotech, Rocky Hill, NJ, USA), 6 mM L-glutamine (Cytiva, Westborough, MA, USA), and 50 μ g/mL L-ascorbic acid 2-phosphate (AA2P; Sigma-Aldrich). Cells were used at passages 2–4. An overview of the experiment is depicted in Figure 1A . Fibroid and myometrial spheroids were fabricated by pelleting 5 × 10 5 cells in 1.5 mL conical tubes according to a published protocol by Zanoni et al. ( 26 ). Spheroids were cultured for 7 days in the conical tubes under standard culture conditions (37°C, 5% CO 2 ) with media changes every 2 days. Human smooth muscle cell culture medium was used with 50 μ g/mL AA2P added right before use to promote cell-cell attachment and ECM production. The spheroids were embedded in 1% agarose mixed with culture medium on a 0.4 μ m pore size Transwell permeable insert in a 6-well plate (VWR International, Radnor, PA, USA). Four spheroids were added per well; they were placed approximately halfway between the center and outer edge of the well, equidistant from each other. After 24 hours, weights were added to obtain 6.4 mm Hg compression; no weight, 0 mm Hg controls were used. An additional seven spheroids were preserved in RNAlater (Thermo Fisher Scientific, Waltham, MA, USA) and stored at −20°C before RNA-sequencing (RNA-seq). Additional samples were fixed in 10% buffered formalin for immunostaining or frozen and stored at −20°C for proteomic analysis. The University of Cincinnati Genomics, Epigenomics and Sequencing Core performed RNA isolation and RNA-seq. Sequencing reads were performed using the Illumina NextSeq2000 and the BaseSpace Sequence Hub app was used for initial bioinformatic analysis. Specifically, the RNA-Seq Alignment app v2.0.2 was used to align RNA-seq reads. After read mapping, Salmon was used to quantify transcript expression ( 27 ). Differentially expressed genes (DEGs) for protein-coding genes were identified using edgeR. The data files were deposited in the NCBI Gene Expression Omnibus (GEO) data repository, accession number GSE282495 . Enriched pathways and associated genes were identified using Metascape, STRING, and multiple packages in R ( 28 – 30 ). Metascape analysis was performed using the whole human genome as background, whereas our list of detected genes was used as background for other analyses. Pathway enrichment analysis was limited to DEGs with false discovery rate corrected P -value ( q ) 0.585 (minimal fold-change = 1.5). All genes included for analysis were expressed in more than one patient, otherwise, they were discarded as physiologic outliers. The Gene Ontology (GO) knowledgebase on the function of genes (molecular function, cellular component, and biological process) ( 31 ), in addition to the Kyoto Encyclopedia of Genes and Genomes (KEGG) ( 32 ), was applied to enrichment analyses to guide the investigation. To compare our fibroid model to fibroid tissue, we analyzed 4s RNA-seq datasets accessed through the Gene Expression Omnibus (GEO) from studies that compared fibroid and myometrial tissue: GSE100338 , GSE128229 , GSE169225 , and GSE199849 ( 33 – 36 ). RNA-seq count data were downloaded, and differential expression analysis was completed as described above. These data were used to ensure our results were clinically relevant. The fixed tissue was paraffin-embedded and sectioned. TUNEL staining (TUNEL Assay Kit – BrdU-Red; Abcam, Cambridge, UK), with Hoechst dye counterstain, was performed to assess cell viability. Collagen XXIII primary antibody (COL23A1; RayBiotech 102–25845, 1:100) was used with a fluorescent secondary antibody (TRITC-conjugated AffiniPure goat anti-rabbit; Jackson ImmunoResearch, West Grove, PA); tissues were counter-stained with Hoechst dye (Invitrogen) for nuclei visualization. Sections were imaged using a Nikon Ni-E upright motorized fluorescence microscope. For collagen XXIII images, blur was removed from images using the Nikon Clarify.ai module, and shot noise was removed using the Nikon Denoise.ai module. A cell was considered positive when there was an area of positive staining within 4 μ m of a nucleus. EphB1 content was determined by ELISA (MBS1606163, MyBioSource, San Diego, CA, USA). The results were normalized to deoxyribonucleic acid (DNA) content which was quantified using the AccuBlue NextGen dsDNA Quantitation Kit (biotium, Fremont, CA, USA) Three biological samples were used for the proteomics study. Proteins in the spheroids were identified using liquid chromatography-mass spectrometry with a label-free quantification (LFQ) method. A 660 nm protein assay with the ionic detergent compatibility reagent was performed on the samples to determine protein concentration. Equal protein contents from the samples were reduced, alkylated, and digested with LysC/trypsin according to the Easypep MS sample prep kit (A40006) instructions. The peptides were desalted and then dried in a speed vac. Each sample was analyzed by nanoLC-MS/MS (Orbitrap Eclipse) and was searched against a combined contaminant database plus the SwissProt homo sapiens database using the Sequest HT search algorithm and the LFQ workflow in Proteome Discoverer ver 3.0 (Thermo Scientific). Proteins were filtered for high protein false discovery rate confidence (99%) and having two peptides per protein. Samples were normalized to total peptides. Protein abundances were normalized to total peptides, and ratios were calculated using the pairwise method. Data are presented as mean ± standard deviation. Principal component analysis (PCA) was used to show sample separation using ggplot2 (v3.5.1). For RNA-seq data, bioinformatic statistics were performed using packages in R (version 4.3.0). Significant DEGs were identified as those with a q -value <.05. The UC Proteomics Core performed analysis of the proteomics data. P -values were calculated using the t-test background-based method. For spheroid size and EphB1 content, a two-way analysis of variance (ANOVA) with post hoc Tukey’s test was performed; for collagen XXIII content, a one-way ANOVA with post hoc Tukey’s test was performed using GraphPad Prism (version 10.2.0). A 95% confidence interval was used to determine significance.

Compressive forces are responsible for some changes in fibroid cells and may be a factor in the initiation of fibroid tumors. More investigation is necessary to understand the forces in vivo and how these compression-induced changes affect fibroid progression.

Compression-induced significant changes both in fibroid and myometrial spheroids that resulted in clinically relevant differences, with enrichment for signaling receptors binding and ECM components. Increases in glycoproteins were found, which is consistent with tissues under compression. EPHB1 and one of its ligands, EFNB2, were identified as clinically relevant DEGs between fibroid and myometrial cells under compression. They encode for EphB1 and ephrin B2, respectively. Although found in the RNA-seq datasets, the receptor EphB1 has also been identified as being overabundant in fibroid tissue ( 38 ). Ephrin signaling regulates cell adhesion and actin organization and EphB1 is upregulated in several cancers; it can be tumor-promoting or suppressing ( 39 , 40 ). It is possible that the differences in ligand binding, such as with ephrin B2, may be involved in the differences in tumor progression. The upregulation of these genes may be responsible for the differences in actin organization that we found previously ( 25 ), as well as the differences in integrin abundance that have been found in this study and others ( 41 – 44 ). Unfortunately, these changes were not found at the protein level as expected. This could be due to the model, variation in sample size, or differences in expression due to the size of the spheroid compared with clinical fibroid samples. The data also indicated changes in the myometrium due to compression-enhanced genes related to uterine fibroids. This suggests that compressive forces may promote an environment that is conducive to fibroid initiation. This is consistent with data from Bariani et al. ( 45 ), where the myometrium in at-risk patients, had altered ECM and mechanotransduction pathways compared with that of women without fibroids. Increased compressive forces due to aberrant uterine contractions or altered tissue integrity in diseases like adenomyosis or endometriosis may initiate these responses and predispose individuals to developing uterine fibroids ( 46 , 47 ). Although we believe that compressive forces are the reason for many of these changes, we cannot rule out the role of lower oxygen tension in the myometrial cell responses. The myometrial cells were more responsive to lower oxygen levels in our model, so this cannot be ruled out as an alternate mechanism ( Supplemental Fig. 2 ). This study is also too preliminary and short in duration to determine if the changes could induce MED12 mutations or if the myometrial cells may just perform a supporting role as cancer-associated fibroblasts or a similar phenotype. Close to 4% of the DEGs were also found as DEPs in the spheroid cultures. This helps validate our results; however, it is a low estimate of actual agreements due to the strengths and limitations of the two assays. Proteomics data are skewed toward the ECM and other high-abundance proteins that can build up over time. Less abundant proteins may not be detected without doing additional enrichment steps. Differentially expressed proteins may not show up in the RNA-seq data due to feedback loops. Therefore, additional differences may be clinically significant and could be considered further; however, additional validation must be considered. Of the 317 DEGs identified between fibroid and myometrial cells under compression, 175 were also identified in at least one dataset from tissue. This means that almost 45% of the genes were not altered in any of the studies of fibroid and myometrial tissues. Many of these genes are likely artifacts due to our model. It is also possible that some of these genes are altered only in small fibroids or that the location within the fibroid is important for expression. The use of model systems such as this can provide important, clinically relevant data; however, the results must be interpreted carefully. The proteomics data indicated a significant alteration in the ECM and integrins due to compression between cell types. We expected collagen I to be altered with compression, but it was not in either cell type. Instead, we found that MMP-1 content was downregulated in fibroid cells. If this remains, the amount of collagen being degraded in fibroid spheroids will be lower and could lead to later differences. We anticipated the growth of the spheroids during culture; however, this did not happen, even when the spheroids were cultured for up to 4 weeks (data not shown). The choice of agarose ensures that the cells will not invade the surrounding tissue and weights can be added for compression, but the model needs to be refined to study fibroid growth. The strengths of this model are that it is a spheroid culture allowing for cell-cell interactions in a tumor-mimetic manner and it allows for compressive forces to be imparted. Patient-matched myometrial and fibroid cells provide translatable data and paired sample analysis with more power. Although providing a more physiologically relevant culture system for fibroid cells, the model includes limitations as it does not completely recapitulate the in vivo condition. Model artifacts led to differences between cell types that were not found in datasets from tissues. The location of the fibroids within the uterus, the size of the fibroids, and the location of the cells within the fibroid tissue will affect how the forces are distributed and may affect how cells respond to mechanical cues. We were able to identify significant differences due to compression that were relevant. Other differences or location-specific differences may have been missed. More information on cell and tissue location is needed, as are computational models of the force and strain distributions within the fibroids and the uterus for more in-depth future studies. The myometrial cells were more susceptible to changes in oxygen tension when embedded, as myometrial tissue is more vascularized than fibroids; this is a likely source of error in our data. Our sample size is small with significant variation between the biological samples; therefore, small, subtle changes may not be detected. The actual compressive stresses that the fibroids and myometrial tissues experience are not known and will vary based on the size of the fibroid and its location in the uterus. We believe that this strain is within the physiologic range of stresses due to other studies, but this cannot be confirmed at this time. Only one pressure was used for this study; it is expected that different forces will lead to different outcomes. We chose the pressure as it is in a likely physiologic range, and we found differences in cellular stiffening at this force ( 14 ); the actual pressure was determined from the weights after they were cut. Future studies of the effects of a large range of pressures would be beneficial to understanding the changes due to fibroid growth, the location of the fibroid in the uterus, and the effects of cells in different locations within the fibroid. The compressive forces that were used in this study were constant. Fibroids will experience consistent compression due to tumor growth; however, they will also experience fluctuations that can be significantly higher due to uterine contractions that occur throughout the menstrual cycle. This cyclic strain will likely elicit additional differences through other mechanotransduction pathways.

Supplemental data for this article can be found online at https://doi.org/10.1016/j.xfss.2025.07.004 .