Early growth response 1 transcription factor is essential for the pathogenic properties of human endometriotic epithelial cells

Abstract

Although a non-malignant gynecological disorder, endometriosis displays some pathogenic features of malignancy, such as cell proliferation, migration, invasion and adaptation to hypoxia. Current treatments of endometriosis include pharmacotherapy and/or surgery, which are of limited efficacy and often associated with adverse side-effects. Therefore, to develop more effective therapies to treat this disease, a broader understanding of the underlying molecular mechanisms that underpin endometriosis needs to be attained. Using immortalized human endometriotic epithelial and stromal cell lines, we demonstrate that the early growth response 1 (EGR1) transcription factor is essential for cell proliferation, migration and invasion, which represent some of the pathogenic properties of endometriotic cells. Genome-wide transcriptomics identified an EGR1-dependent transcriptome in human endometriotic epithelial cells that potentially encodes a diverse spectrum of proteins that are known to be involved in tissue pathologies. To underscore the utility of this transcriptomic dataset, we demonstrate that carbonic anhydrase IX (CAIX), a homeostatic regulator of intracellular pH, is not only a molecular target of EGR1 but is important for maintaining many of the cellular properties of human endometriotic epithelial cells that are also ascribed to EGR1. Considering therapeutic intervention strategies are actively being developed for EGR1 and CAIX in the treatment of other pathologies, we believe EGR1 and its transcriptome (which includes CAIX) will offer not only a new conceptual framework to advance our understanding of endometriosis but will furnish new molecular vulnerabilities to be leveraged as potential therapeutic options in the future treatment of endometriosis.

Introduction

Early growth response 1 (EGR1; also known as NGFI-A, Zif 268 or Krox 24) is a member of the EGR family of Cys₂-His₂-type zinc finger transcription factors that also includes EGR2, EGR3 and EGR4 (Gashler and Sukhatme, 1995, O'Donovan et al., 1999, Sukhatme, 1990). In response to a broad spectrum of extracellular stimuli, EGR family members mediate transcriptional responses through direct interaction of their three tandem DNA binding motifs with a GC-rich consensus sequence (GCG(T/G)GGGCG) within regulatory regions of target genes (Benos et al., 2002, Swirnoff and Milbrandt, 1995). Through these target genes, EGR1 controls a myriad of cellular properties from proliferation, differentiation, migration, invasion to programmed cell death and stemness (Gururajan et al., 2008, Ma et al., 2021, Madden and Rauscher, 1993, Yan et al., 2021, Yan et al., 2000, Zhang et al., 2021, Zhao et al., 2021). Such cellular responses drive EGR1’s Early investigations in the mouse demonstrated that EGR1 ablation results in a block in the expression of the luteinizing hormones.

Early investigations in the mouse demonstrated that EGR1 ablation results in a block in the expression of the luteinizing hormone β-subunit in the pituitary gonadotrope, resulting in impaired ovulation and luteinization (Lee et al., 1996, Topilko et al., 1998, Wolfe and Call, 1999). In the uterus, studies revealed that Egr1 transcript levels are rapidly induced by estrogen in the epithelial and stromal cellular compartments of the murine endometrium during early pregnancy (Guo et al., 2014, Kim et al., 2014, Kim et al., 2018, Liang et al., 2014). These studies also indicated that endometrial EGR1 expression in the luminal epithelium and pre-decidual stromal cells is required for embryo implantation and subsequent decidualization (Guo et al., 2014, Kim et al., 2014, Kim et al., 2018, Liang et al., 2014). Recent investigations on human endometrial stromal cells in culture support the findings in the mouse as well as a role for EGR1 in priming the pre-decidual stromal cell for decidualization when exposed to a deciduogenic hormone stimulus (Kommagani et al., 2016, Szwarc et al., 2019).

Apart from its function in normal physiological processes, EGR1 plays important roles in the pathogenesis of numerous target tissues (Wang et al., 2021a, Hao et al., 2021). Therefore, we asked whether EGR1 is involved in the pathology of the uterus in addition to its established role in normal uterine functions. Here, we demonstrate that EGR1 is critical for the pathogenic properties of human endometriotic epithelial and stromal cells, which include cellular proliferation, migration and invasion. Moreover, genome-wide transcriptome analysis highlights distinct gene expression programs that mediate EGR1’s contribution to the pathogenic properties of human endometriotic epithelial cells, which encompass cytokine signaling, adaptation to hypoxia, cellular inflammatory responses, epithelial-mesenchymal transition, and cell-cell communication. To underscore the utility of the EGR1 transcriptome dataset, we observed carbonic anhydrase IX (CAIX) is a critical EGR1 responsive molecular target that drives many of the pathogenic properties of the human endometriotic epithelial cell.

Materials and methods

Immunohistochemical analysis of baboon (Papio anubis) endometriotic tissue

Eutopic uterine tissue and matched ectopic endometriotic lesions (at pelvic and peritoneal locations) were obtained from a baboon model for endometriosis (n=4). As previously described (D'Hooghe et al., 1994, Fazleabas, 2006b, Fazleabas et al., 2002), the baboon model for endometriosis entails autologous inoculation of menstrual endometrium into the peritoneal cavity, modeling retrograde menstruation. Eutopic and ectopic tissues were collected during the mid-secretory phase of the cycle (days: 9-12 postovulation), 15 months following disease induction. Importantly, endometriotic lesions derived from the baboon endometriosis model share morphological and histological characteristics similar to those observed in human lesions (Fazleabas et al., 2002). At the time of harvesting these tissues, all animal procedures used for the experimental induction of endometriosis in the baboon were prospectively approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Illinois, Chicago Michigan State University.

For immunohistochemical analyses, tissues were fixed overnight in 4% paraformaldehyde in phosphate-buffered saline (PBS) before paraffin embedding and sectioning onto slides. Immunohistochemical detection of EGR1 and CAIX was achieved using primary rabbit monoclonal antibodies (anti-EGR1 (#4153) and anti-CAIX (#5649); Cell Signaling Technology Inc., Danvers, MA; each diluted 1:100) followed by incubation with a horseradish peroxidase (HRP)-conjugated goat anti-rabbit secondary antibody (Vector Laboratories, Burlingame CA (P-1000); diluted 1:200). Peroxidase activity was detected with the Vectastain Elite ABC-HRP kit (Vector Laboratories Inc., Burlingame, CA). Following immunostaining, tissue sections were counterstained with hematoxylin before applying Permount mounting medium to affix coverslips.

Immortalized human endometriotic cell lines

The immortalized human endometriotic epithelial cell (iHEEC/Luc (referred to as iHEEC hereon) line has been described (Bono et al., 2012, Han et al., 2012). Briefly, the iHEEC line (formerly known as EMOSIS-CC/TERT1 (Bono et al., 2012)) was derived from a human ovarian endometrioma and immortalized by transfection with the human telomerase reverse transcriptase (hTERT) gene and subsequently modified with a luciferase reporter using lentivirus (Bono et al., 2012, Han et al., 2012). The iHEEC line was cultured in DMEM/F12 medium supplemented with 10% fetal bovine serum (FBS; Sigma-Aldrich, St. Louis, MO) and a 1% penicillin-streptomycin antibiotic solution (ThermoFisher Scientific Inc. Waltham, MA); medium was changed every other day. Using the American Type Culture Collection (ATCC) cell line authentication service, the iHEEC line was authenticated by short tandem repeat (STR) profiling analysis. The immortalized human endometriotic stromal cell line (iEc-ESC) and its culture have been described previously (Song et al., 2020a).

Transfection of small interfering RNAs

Human endometriotic cells were cultured in six-well plates in triplicate before transfection with sixty picomoles of the non-targeting (NT) siRNA ((D-001810-10-05) Dharmacon Inc., Lafayette, CO), or siRNAs targeting either EGR1 ((L-006526-00-0005) Dharmacon Inc.) or CAIX ((L-005244-00-00005) Dharmacon Inc.) using the Lipofectamine RNAiMAX transfection reagent (Invitrogen Corporation, Carlsbad, USA) (Szwarc et al., 2019). Forty-eight hours post-transfection, cells were harvested for quantitative real-time (qRT) PCR, RNA-seq, or western immunoblot analysis. Alternatively, cells were trypsinized and re-plated to assay for cell proliferation/viability, clonogenic survival or invasion capabilities (Cagle et al., 2019, Szwarc et al., 2018a, Zhang et al., 2018).

Quantitative real-time PCR

Cells were lysed in RNA lysis buffer before total RNA was isolated with the Purelink RNA Mini Kit ((#12183020) ThermoFisher Scientific Inc.). The Nano-Drop 2000 UV/Visual spectrophometer (ThermoFisher Scientific Inc.) was used for RNA quantification; RNA (1 μg) was reverse transcribed using the High-Capacity cDNA Reverse Transcription Kit ((#4368814) ThermoFisher Scientific Inc.). Amplified cDNA was diluted to 10 ng/μl before qRT-PCR was performed using the Fast TaqMan 2X Mastermix (Applied Biosystems/Life Technologies, Grand Island, NY). The TaqMan assays used in this study are listed in Table 1. All qRT-PCR experiments were performed using the 7500 Fast Real-time PCR system (Applied Biosystems/Life Technologies, Grand Island, NY); the delta-delta cycle threshold was used to normalize expression to the 18S reference.

Table 1.

Human TaqMan expression assays

Gene	ID	Catalog number
CAIX	768	Hs00154208_m1
DRICH1	51233	Hs01589059_m1
DPCR1	135656	Hs00369879_m1
EBF2	64641	Hs00970588_m1
EGR1	1958	Hs00152928_m1
EGR2	1959	Hs00166165_m1
EGR3	1960	Hs04935588_m1
ERVH48-1	90625	Hs05577546_g1
HCAR3	8843	Hs02341102_s1
HTR3E	285242	Hs00704511_s1
IL6	3569	Hs00174131_m1
MT1G	4495	Hs04401199_s1
PLPPR1	54886	Hs00214827_m1
SERPINA9	327657	Hs00900935_m1
SERPINB4	6318	Hs01691258_g1
SYN2	6854	Hs00923900_m1
18S rRNA		4319413E

Global RNA expression profiling

Genome-wide RNA-sequencing (RNA-seq) and analysis were performed as previously described (Szwarc et al., 2019, Szwarc et al., 2018b). Briefly, total RNA purity and integrity were assessed using the NanoDrop spectrophotometer (ThermoFisher Scientific Inc.), and the 2100 Bioanalyzer with RNA chips (Agilent Technologies, Santa Clara, CA) respectively. Only RNA samples scoring a RNA integrity number (RIN) of 8 or greater were included. For each experimental group, RNA samples from three replicates were used. Sequencing libraries were prepared using the TruSeq Stranded mRNA kit (Illumina Inc., San Diego, CA) from 250 ng of RNA and PCR amplified. Quality analysis of resultant libraries was performed on the 4200 TapeStation with D1000 ScreenTape assays (Illumina Inc.). Adapter-ligated fragment concentration was estimated by qRT-PCR assay with a KAPA Library Quantification Kit (KAPA Biosystems, Wilmington, MA). After equimolar pooling, libraries were quantified on the 2100 Bioanalyzer (using the High Sensitivity DNA Kit and DNA chips) and the KAPA Library Quantification Kit. Sequencing of libraries was performed on the NovaSeq platform (Ilumina Inc.). Paired-end 100 base pair (bp) sequencing reads were generated at mid-output and mapped to the human genome. Raw sequenced reads in Ilumina fastq file format were aligned to the human genome (Genome Reference Consortium Human Build hg38 (National Center for Biotechnology Information (NCBI))) through use of the ultrafast universal RNA-seq aligner: spliced transcripts alignment to a reference (STAR) (Dobin et al., 2013, Anders et al., 2015). The number of reads aligned to known genes was determined by the Python-based software package HTSeq (Anders et al., 2015) (http://www-huber.embl,de/users/anders/HTSeq). To reduce possible PCR bias, read duplicates were removed with Picard Tools (http://broad.institute.github.10/picard).

The Bioconductor package EdgeR was applied to the gene expression data to detect differentially expressed genes between the two groups (Robinson et al., 2010). The false discovery rate (FDR) of differentially expressed genes was estimated using the Benjamini and Hochberg method (Benjamini, 1995). Gene expression comparisons with an FDR ≤ 0.05 and an absolute fold change (IFCI) ≥ 1.5 were considered to be significantly differentially expressed between the two groups. Genes with significantly altered expression (IFCI ≥ 1.5; FDR ≤ 0.05) between the two groups were used further to identify affected pathways (Huang da et al., 2009). Fragments per kilobase of transcript per million (FPKM) values of transcripts were used for hierarchical clustering; the pheatmap package in R was used to draw the clustered heatmap. Using raw gene count data, principal component analysis (PCA) was performed with the R function prcomp package (https://cran.r-project.org). All raw data files were deposited in Gene Expression Omnibus repository at the NCBI ((GSE199526) www.ncbi.nlm.gov/geo). Gene ontology enrichment analysis was performed using the DAVID (Database for Annotation, Visualization, and Integrated Discovery) functional annotation clustering tool (http://david.abcc.ncifcrff.gov/) (Sherman et al., 2007). Established gene sets overrepresented in our RNA-seq datasets were identified by Gene Set Enrichment Analysis (GSEA; http://software.broadinstitute.org/gsea/) (Mootha et al., 2003, Subramanian et al., 2005). Hallmark gene sets from the Molecular Signatures Database (MSigDB) were used in these GSEA studies (Liberzon et al., 2011).

Immunoblotting

Protein (20 μg) from cell lysates was resolved on 4-15% sodium dodecyl sulfate-polyacrylamide (SDS-PAGE) gels before transfer to polyvinylidene difluoride (PVDF) membranes. Following protein transfer, PVDF membranes were blocked for 1 hour with 5% non-fat dry milk ((sc-2324 (Blotto)) Santa Cruz Biotechnology Inc., Dallas, Texas) in Tris-buffered saline with Tween 20 (TBS-T) and incubated overnight at 4°C with the following primary antibodies: anti-EGR1 (#4154, Cell Signaling Technology, Inc., Danvers, MA) diluted 1:1000 and anti-β-actin (#A00702, GenScript Biotech, Piscataway, NJ) diluted 1:100000 in 5% non-fat milk in TBS-T. Blots were then probed with anti-rabbit ((A27036 (1:5000 dilution)) ThermoFisher Scientific Inc.) and anti-mouse IgG secondary antibodies conjugated with HRP ((#7076 (1:10000 dilution)) Cell Signaling Technology, Inc.) respectively in 5% non-fat milk in TBS-T for 1 hour at room temperature. Chemiluminescence was detected with the SuperSignal West Pico PLUS Chemiluminescent Substrate ((#1863097) ThermoFisher Scientific, Inc.). Immunoreactive bands were digitally imaged using the Bio-Rad ChemiDoc imaging system (Bio-Rad Laboratories, Hercules, CA).

Cell proliferation/viability assay

Cells were seeded in 96-well culture plates in triplicate, at a density of 5x10³ cells per well. Cells transfected with siRNAs for 48 hours were further cultured for 0, 24, 48, 72 or 96 hours before cell proliferation was measured using the CellTiter 96^® Non-Radioactive Cell Proliferation Assay kit ((#G4000) Promega Inc. Madison, WI). After a specific time period in culture, 15μl of 3-(4, 5-dimethylthiazol-2-yl)-2, 5-diphenyltetrazolium bromide (MTT); Promega, Madison, WI) was added to each well to a final concentration of 0.5mg/ml. Cells were then incubated at 37°C for an additional three hours in the dark. Following the three-hour incubation period, the supernatant was removed. Formazan crystals within wells were dissolved by the addition of the stop/solubilizing solution (100 μl (dimethyl sulfoxide (DMSO)/well)) before a further incubation period of 15 minutes at 37°C with gentle agitation. The absorbance of the final mixture was recorded at 570 nm (formazan absorbance maximum (Campling et al., 1988, Mosmann, 1983)) using a 96 well microplate ELISA reader. Relative cell proliferation was calculated as: mean absorbance at “N” time point/mean absorbance at 0 hour (h); N= 0, 24, 48, 72, and 96h. Each experiment was repeated three times with three to five technical replicates for each treatment group.

Clonogenic survival assay

Forty-eight hours post siRNA transfection, cells were cultured in six-well culture plates (3x10³ cells per well in triplicate). Cells were incubated for ten days; medium was replaced every other day. After the ten-day culture period, cells were fixed with 4% paraformaldehyde for 15 minutes, washed with PBS for 10 minutes before colonies were stained with crystal violet solution (0.5%) for 15 minutes (Kommagani et al., 2013). After the sequential steps of de-staining in tap water and air-drying, stained colonies were photographed and counted. Each experiment was repeated three times with triplicates for each treatment group.

Cell migration assay

Cell migration was assessed using the in vitro wound-healing assay (Grada et al., 2017, Todaro et al., 1965). Briefly, cells were seeded in six-well culture plates and cultured until reaching 70–80% confluency before siRNA transfection. Using a 200-μl sterile pipette tip, a linear scratch (wound) was created in the middle of the cell monolayer within each well. Wells were gently washed to remove detached cells before image capture with an inverted phase-contrast microscope (EVOS™ XL Core Imaging System, #AMEX1000 (ThermoFisher Scientific Inc.)). Cells were incubated for forty-eight hours before the degree of wound closure was recorded by digital image capture. The wound area was calculated by manual tracing the cell-free area within captured images using ImageJ software (https://imagej.nih.gov/ij/). Results were expressed as a percent of wound closure in comparison to control after a forty-eight hours culture period (percent cell migration area = wound width at 0 h – wound width at 48 h/wound width at 0 h). Each experiment was repeated three times with triplicates for each treatment group.

Transwell cell invasion assay

Cell invasion was analyzed using the Corning BioCoat Matrigel Invasion Chamber kit ((#354480) ThermoFisher Scientific Inc.). Following siRNA transfection, cells were first suspended in Opti-MEM medium. A culture medium with 20% FBS was added (0.6 ml) to the bottom of each transwell of the invasion chamber plate. Suspended cells (1x10⁵ cells/250 μl) were then added to each transwell insert and allowed to migrate. After forty-eight hours, cells on the upper surface of the transwell were removed using a cotton swab. Migrated cells were fixed with 4% paraformaldehyde in PBS for fifteen minutes and stained with crystal violet solution for ten minutes (Justus et al., 2014, Zhang et al., 2020). After washing with distilled water, the inserts were digitally imaged using a Zeiss stereo-microscope with an attached AxioCam MRC-5 digital camera (Zeiss, Jena Germany). Migrated cells were counted within four separate areas of the insert; an average number of migrated cells was calculated (Pijuan et al., 2019). Each experiment was repeated three times with triplicates for each treatment group.

Flow cytometry and cell cycle analysis

Seeded at a density of 2x10⁵ cells per well in triplicate in six-well plates, iHEECs were transfected with NT or EGR1 targeted siRNAs. Forty-eight hours post-transfection, cells were harvested, washed with PBS, fixed in 70% chilled ethanol, before staining with propidium iodide (PI)/RNase staining solution (#4087, Cell Signaling Technology Inc.). Analysis of cell cycle stage was conducted using a flow cytometer (Attune NXT Acoustic Focusing Flow Cytometer, Invitrogen) with installed FlowJo software (version 10.7.1). Cell cycle analysis experiments were performed in triplicate and repeated three times.

Statistical analysis

Two-tailed unpaired Student t-tests were used to estimate the statistical significance of differences between the two groups. Unless otherwise stated, data were graphically presented as mean ± standard error of the mean (SEM). Differences with p-values <0.05 were considered statistically significant; asterisks represent the level of significance: *p<0.05, **p<0.01, ***p<0.001, and ****p<0.0001. Prism software version 9 (GraphPad Software Inc., San Diego CA) was used for the majority of the reported statistical analyses.

Results

Expression of EGR1 is significantly increased in Baboon endometriotic lesions

Analyses of published microarray datasets from human endometriotic tissues compared with matched eutopic endometria revealed that EGR1 transcript levels are significantly elevated in ovarian endometrioma when compared to matched eutopic endometrium (Hever et al., 2007, Crispi et al., 2013) (Supplementary Fig. 1), suggesting a possible link between aberrant EGR1 expression and endometriosis pathology progression. This supposition is supported by the observed increase in EGR1 protein expression in the glandular epithelial and stromal compartments within endometriotic lesions in a baboon model for experimental induced endometriosis (Fazleabas, 2006b, Fazleabas, 2006a) (Fig. 1). Noteworthy, the surrounding host tissue also scores positive for EGR1 immunopositivity; however, the staining is less intense as compared with the endometriotic lesion. Because the host microenvironment (i.e. the reactive stromal environment) is expected to display an inflammatory response (Wilson, 2018), and because EGR1 is induced by inflammatory stimuli (Bhattacharyya et al., 2011), it’s not unexpected that EGR1 expression levels would increase within the surrounding host microenvironment in response to local ectopic lesion development.

Figure 1.

Significant EGR1 protein expression in epithelial and stromal cells of baboon ectopic endometriotic lesions. (A) Low power magnification image of an endometriotic lesion biopsied from the pelvic region of the baboon. Note the strong expression of EGR1 in the glandular epithelium and underlying stroma. The inset shows negative control staining, which does not include the primary antibody against EGR1. (B) Higher magnification image of a region demarcated with a blue box in (A); the epithelium and stroma are denoted by “E” and “S” respectively. (C) Higher magnification image of an elongated epithelial gland within the same endometriotic lesion (outside the field shown in (A)). Again, note the marked immunoreactivity for EGR1 in all epithelial cells that comprise the gland. (D) Low power magnification image of an endometriotic lesion in the peritoneal region. Again, note the significant EGR1 immunopositivity within the endometriotic lesion. (E) Higher power magnification image of a region delineated by the blue box in (D). Again, note the strong expression for EGR1 in the epithelial and stromal compartments that comprise the endometriotic lesion. (F) Matched eutopic endometrium exhibits significantly lower EGR1 expression in the epithelial and stromal compartments. (G) Higher magnification image of the region outlined by the blue box in (F). Note that the eutopic uterus and both ectopic endometriotic lesions shown are derived from the same animal during the secretory phase of the cycle (see: Materials and methods sub-section). The data are representative of a group size of n=4.

Pathogenic properties of human epithelial endometriotic cells require EGR1

Based on our histological findings (Fig. 1), we investigated whether EGR1 is required for the proliferative, migratory, and invasive properties of a cultured human epithelial endometriotic cell line (iHEEC (Bono et al., 2012)). A 50-60% reduction in EGR1 transcript levels was shown to result in a marked attenuation in iHEEC proliferation (Fig. 2A), a significantly reduced number of colonies were also observed with iHEECs following EGR1 knockdown (Fig. 2B). Linked to the aforementioned, cell cycle analyses of iHEECs with reduced EGR1 levels demonstrate that a significant number of these cells are arrested during the S-phase of the cell cycle (Supplementary Fig. 2). Note that these changes in cellular properties of the iHEEC line are solely due to changes in EGR1 levels and not due to alteration in levels of other members of the EGR family (Supplementary Fig. 3). Endogenous EGR1 expression levels also are required to maintain the full migratory and invasive properties of the iHEEC line (Figs. 3 and 4). Because EGR1 is highly expressed in stromal cells of ectopic endometriotic lesions, we also demonstrated that maintenance of EGR1 expression levels in a recently generated hTERT-immortalized human stromal endometriotic cell line (iEc-ESC) (Song et al., 2020a) is required for its full pathogenic cellular properties (Supplementary Fig. 4). Collectively, the above findings provide strong support for critical roles for both epithelial- and stromal-derived EGR1 in the pathogenic properties of endometriotic lesions.

Figure 2.

Proliferative and colony-forming properties of human endometriotic epithelial cells require EGR1. (A) Following NT or EGR1 siRNA transfection, iHEEC viability was assessed by the MTT assay. (B) A representative result from a colony-formation assay to assess the colony-forming abilities of human endometriotic epithelial cells forty-eight hours following NT or EGR1 siRNA transfection and subsequent culture for 10 days. (C) The histogram displays quantification of stained colony numbers. (D-E) Representative results of qPCR and Western immunoblot analyses confirm EGR1 depletion in the iHEEC line. For Western analyses, β-actin was used as a control for protein loading. Results are indicated as mean ± SE and are representative of three independent experiments; **p-value<0.01; ***p-value<0.001; and ****p-value<0.0001.

Figure 3.

EGR1 is required for iHEEC migration in vitro. (A) The migration ability of iHEEC was assessed by the wound healing assay. Representative bright-field images of the migrated area forty-eight hours following the scratch; scale bar applies to both images. (B) The histogram displays the reduced migration ability of iHEECs following EGR1 knockdown, specifically showing 60% reduced migration ability of iHEECs compared to control. (C-D) Both qPCR and Western immunoblot results confirm that EGR1 expression levels are significantly attenuated at the RNA and protein level respectively forty-eight hours following transfection with NT or EGR1 siRNAs. For Western immunoblot analyses, β-actin served as a control for equal protein loading per lane. Results represent the mean ± SE and representative of three independent experiments; **p-value<0.01; and ***p-value<0.001.

Figure 4.

Attenuation of EGR1 levels reduces the invasive capability of the iHEEC line in vitro. (A) Forty-eight hours following transfection with siRNAs targeting NT or EGR1, cell invasion analysis was initiated. Representative cell images are shown following the invasion experiment using either NT siRNA or EGR1 siRNA transfected cells; the scale bar applies to all images. (B) Representative histogram quantitatively displays the number of EGR1 siRNA transfected endometriotic epithelial cells that invaded the lower chamber compared with endometriotic epithelial cells transfected with NT siRNAs. (C-D) Both qPCR and Western immunoblot analyses confirm a significant reduction in EGR1 expression at the RNA and protein level respectively. Note: β-actin was used as a protein loading control. Results are represented as mean ± SE and representative of three independent experiments; **p-value<0.01 and ***p-value<0.001.

Transcriptomic changes in iHEECs in response to decreased EGR1 levels

Because EGR1 is a transcription factor, genome-wide RNA profiling was conducted to identify the downstream genes, pathways and networks that may mediate EGR1’s role in the above pathogenic properties of iHEECs. The experimental design for the RNA-seq study is schematically shown (Fig. 5A). Briefly, cultured iHEECs in six-well plates were transfected with NT or EGR1 targeted siRNAs forty-eight hours before RNA isolation and sequencing; triplicate samples were used per treatment group. All genes differentially expressed between NT and EGR1 siRNA treated iHEEC groups are tabulated in an Excel spreadsheet in the supplementary section (Supplementary Folder 1), also included in this folder is the gene ontology (GO) analysis by DAVID for sets of differentially expressed genes. As expected, the expression of the EGR1 gene is significantly downregulated (Log₂ FC: −2.8) in the differentially expressed gene list (yellow highlighted row).

Figure 5.

Changes in the iHEEC transcriptome following EGR1 knockdown. (A) Experimental design of the RNA-seq experiment; triplicate samples were used for NT siRNA and EGR1 siRNA groups. (B) Heatmap of clustering of genes with the same expression level differentially expressed (up or down) between the NT siRNA and EGR1 siRNA groups. With a FDR<0.05 and a IFCI >1.5, 76 genes differentially expressed between NT siRNA and EGR1 siRNA groups were clustered and presented as a heat map; each horizontal row represents a single gene. Warmer (i.e. reds) and cooler colors (i.e. blues) represent higher and lower expression respectively; the vertical color key on the right indicates the intensity with normalized expression values.

In total, 22632 (11945 upregulated and 10687 downregulated) expressed genes were detected by RNA-seq (Supplementary Folder 1); 2231 and 1417 genes were significantly upregulated and downregulated respectively. Forty upregulated and thirty-six downregulated genes met the predetermined FDR (≤0.05) and FC (≥1.5) cutoffs. FKPM values for all 22632 genes were analyzed by principal component analysis (PCA (Supplementary Folder 1)). The PCA showed that the NT siRNA and EGR1 siRNA treated groups were significantly separated in terms of their respective triplicates. Tables 2 and 3 list the top 35 genes down and upregulated respecitively that meet the FDR (≤0.05) and FC (≥1.5) cutoffs whereas the expression heatmap (Fig. 5B) shows the top 76 genes (40 upregulated and 36 downregulated) between the NT siRNA and EGR1 siRNA groups with an FDR ≤0.05 and IFCI ≥1.5 cut-off. With EGR1 knockdown, GSEA showed that pathways involved in inflammation, estrogen early and late response, adaptation to hypoxia, epithelial-mesenchymal transition, and cell-cell junctions were significantly enriched (Fig. 6A). Furthermore, DAVID analysis revealed enrichment of biological processes related to cytokine activity, inflammatory response, and cell proliferation and cell death within the differential gene expression set (Fig. 6B). Together, these biological responses are in line with known cellular phenotypes that drive endometriotic lesion progression (Bulun et al., 2019, Taylor et al., 2021).

Table 2.

Top 35 downregulated genes with ≥ 1.5 log2FC and ≤ 0.05 FDR

GENE SYMBOL	GENE ID	GENE NAME	log2FC
DPCR1	135656	diffuse panbronchiolitis critical region 1	−5.13541
HPCA	3208	hippocalcin	−5.02329
LINC02043	102724699	long intergenic non-protein coding RNA 2043	−5.00908
SFTPA2	729238	surfactant protein A2	−4.98471
ERVH48-1	90625	endogenous retrovirus group 48 member 1	−4.20807
HCAR3	8843	hydroxycarboxylic acid receptor 3	−4.18165
LINC00302	388699	long intergenic non-protein coding RNA 302	−3.96066
MIR4531	100616355	microRNA 4531	−3.66491
PPIAL4G	644591	peptidylprolyl isomerase A like 4G	−3.52393
MIR2116	100313886	microRNA 2116	−3.44583
SERPINA9	327657	serpin family A member 9	−2.96628
EGR1	1958	early growth response 1	−2.80233
MIR6071	102466516	microRNA 6071	−2.76669
CA9	768	carbonic anhydrase 9	−2.5501
IGBP1P1	280655	immunoglobulin (CD79A) binding protein 1 pseudogene 1	−2.45972
IL6	3569	interleukin 6	−2.44627
WARS2-IT1	104472716	WARS2 intronic transcript 1	−2.32034
EBF2	64641	early B-cell factor 2	−2.18673
SNORA69	26779	small nucleolar RNA, H/ACA box 69	−2.14905
MT1G	4495	metallothionein 1G	−2.13078
TOMM20L	387990	translocase of outer mitochondrial membrane 20 like	−2.06042
GOLGA2P7	388152	golgin A2 pseudogene 7	−2.03929
LINC02056	102477328	long intergenic non-protein coding RNA 2056	−1.88876
LINC01364	100505768	long intergenic non-protein coding RNA 1364	−1.81405
DRICH1	51233	aspartate rich 1	−1.77817
SNORD55	26811	small nucleolar RNA, C/D box 55	−1.72796
CHRNA3	1136	cholinergic receptor nicotinic alpha 3 subunit	−1.66121
ZNF804A	91752	zinc finger protein 804A	−1.63954
SERPINB4	6318	serpin family B member 4	−1.63168
SCOC-AS1	100129858	SCOC antisense RNA 1	−1.61592
CERCAM	51148	cerebral endothelial cell adhesion molecule	−1.57345
ARTN	9048	artemin	−1.56061
TAGLN3	29114	transgelin 3	−1.55785
JPH1	56704	junctophilin 1	−1.54813
MYO16	23026	myosin XVI	−1.52762

Table 3.

Top 35 upregulated genes with ≥ 1.5 log2FC and ≤ 0.05 FDR

GENE SYMBOL	GENE ID	GENE NAME	log2FC
SYN2	6854	synapsin II	5.082034
MIR579	693164	microRNA 579	4.987176
CFAP99	402160	cilia and flagella associated protein 99	4.84547
AGBL4	84871	ATP/GTP binding protein like 4	4.717173
ALLC	55821	allantoicase	4.706209
LINC02324	100128233	long intergenic non-protein coding RNA 2324	4.702674
MIR7112	102465906	microRNA 7112	4.700377
HTR3E	285242	5-hydroxytryptamine receptor 3E	4.693637
TCERG1L	256536	transcription elongation regulator 1 like	3.674137
ADAMTSL2	9719	ADAMTS like 2	3.656694
TJP3	27134	tight junction protein 3	3.483605
MAGI1-AS1	100873983	MAGI1 antisense RNA 1	3.321479
TEC	7006	tec protein tyrosine kinase	3.218295
SPN	6693	sialophorin	3.124289
FAM213A	84293	family with sequence similarity 213 member A	3.017358
TDGF1	6997	teratocarcinoma-derived growth factor 1	2.81775
CCDC102B	79839	coiled-coil domain containing 102B	2.622383
HMMR-AS1	101927813	HMMR antisense RNA 1	2.587966
ZFX-AS1	100873922	ZFX antisense RNA 1	2.486735
LRMP	4033	lymphoid restricted membrane protein	2.459547
TSPAN2	10100	tetraspanin 2	2.38303
HOXA6	3203	homeobox A6	2.30578
HR	55806	hair growth associated	2.263117
CXCL10	3627	C-X-C motif chemokine ligand 10	2.237354
SCG2	7857	secretogranin II	2.187924
TEKT2	27285	tektin 2	2.183893
SYTL5	94122	synaptotagmin like 5	2.120299
TIGIT	201633	T-cell immunoreceptor with Ig and ITIM domains	2.11793
RCAN2	10231	regulator of calcineurin 2	1.884025
PLPPR1	54886	phospholipid phosphatase related 1	1.781885
TEX14	56155	testis expressed 14, intercellular bridge forming factor	1.761584
FAM84B	157638	LRAT domain containing 2	1.735547
TNFSF10	8743	TNF superfamily member 10	1.653359
CPA4	51200	carboxypeptidase A4	1.604515
MAP3K14-AS1	100133991	MAP3K14 antisense RNA 1	1.543323

Figure 6.

Pathway analyses of differential expressed genes in iHEECs following EGR1 knockdown. (A) GSEA of differential expressed genes between the NT siRNA and EGR1 siRNA groups showing normalized enrichment scores for listed Hallmark pathways. On the x-axis, normalized enrichment scores for gene expression changes (up or down represented by red and blue bars respectively) following a reduction in EGR1 levels in iHEECs; the y-axis displays hallmark gene-sets representing well-defined biological states or processes (Liberzon et al., 2011). (B) DAVID gene functional clustering analysis of genes differentially expressed between the NT siRNA and EGR1 siRNA treated iHEEC groups.

The volcano plot furnished a global perspective of the transcriptional changes that occur due to EGR1 knockdown (Fig. 7A). To illustrate the diverse functionality of the genes differentially expressed between the NT siRNA and EGR1 siRNA treated groups, 12 genes (10 downregulated genes: Carbonic anhydrase IX (CAIX); diffuse panbronchiolitis critical region 1 (DPCR1); aspartate-rich protein 1 (DRICH1); early B-cell factor 2 (EBF2); endogenous retrovirus group 48 member 1 (ERVH48-1); hydroxycarboxylic acid receptor 3 (HCAR3); interleukin-6 (IL6); metallothionein 1G (MT1G); serpin A9 (SEPINA9); serpin B4 (SERPINB4); and 2 upregulated: phospholipid phosphatase related 1 (PLPPR1) and synapsin 2 (SYN2)) are highlighted in the volcano plot that are significantly differentially expressed in iHEECs following EGR1 knockdown (Fig. 7A). In addition, the differential expression of these genes in iHEECs following EGR1 knockdown was validated at the RNA level by qRT-PCR (Fig. 7B).

Figure 7.

Expression validation of a select number of genes for which expression levels change in response to EGR1 knockdown in iHEECs. (A) Global gene expression changes displayed as a volcano plot represent the statistical significance (plotted as the log-transformed p-value) versus the fold change across all genes. To aid visualization, the insert on the right represents a magnification of the volcano plot of genes with a – log10 (p-value) up to 5. Individual genes are presented by open colored circles. Orange circles represent genes with an absolute fold change ≥ 1.5 and a p-value ≤ 0.05; red circles denote genes that also have an FDR ≤ 0.05. (B) Genes (10 downregulated and 2 upregulated following EGR1 knockdown) annotated in (A) were validated by qPCR.

The pathogenic cellular properties of iHEECs require CAIX

The CAIX gene was further validated to showcase the usefulness of this EGR1 transcriptome dataset in identifying potential new pathogenic molecular mediators of endometriosis (Fig. 8). Analysis of published human endometriotic microarray datasets (Gene Expression Omnibus dataset GSE25628 (Crispi et al., 2013)) reveals that CAIX transcript levels are significantly increased in ectopic endometriotic lesions when compared to eutopic endometrial tissue (Fig. 8A). Immunohistochemical studies also show that CAIX protein is significantly expressed in epithelial cells of baboon ectopic endometriotic lesions as compared to matched eutopic endometrium (Fig. 8B). Using the iHEEC culture model, we show that CAIX is essential for the pathogenic properties of human endometriotic epithelial cells, which include cellular proliferation, colony formation capability and cell invasion (Fig 8C-H). Together, these results underscore the general utility of the above EGR1 RNA-seq dataset to uncover new potential drivers of endometriosis progression and highlight CAIX in particular as a molecular EGR1 target that may provide a new perspective in examining the pathogenic properties of endometriotic epithelial cells.

Figure 8.

Carbonic anhydrase IX is required to maintain the pathogenic properties of iHEECs. (A) The relative raw abundance of CAIX transcripts in human ectopic and matched control endometrium (from: Gene Expression Omnibus dataset GSE25628 (Crispi et al., 2013)). Note: CAIX transcripts are significantly elevated in human ectopic endometriosis compared with control endometrium (AU denotes arbitrary units). Human control and ectopic endometrial tissues were obtained during the proliferative phase. Data are presented as mean SE (control endometrium: n=6; ectopic endometrium: n=7); *p-value ≤0.05. (B) Immunohistochemical analysis shows that CAIX is undetectable in the baboon eutopic endometrium (top and bottom left panels represent two separate baboon eutopic endometrial tissues). Ectopic endometriotic lesions (right panels (top and bottom)) express CAIX that is restricted to epithelial cells (white arrowhead); “S” indicates stromal compartment. The scale bar shown in left top panel applies to all four panels in (B). (C-H) Cell viability, clonogenic survival, and wound healing assays respectively show that CAIX depletion in iHEECs results in a compromised ability to proliferate, form colonies, and migrate—all pathogenic properties of iHEECs. Results are represented as mean ± SE and representative of three independent experiments; *p-value<0.05, ***p-value<0.001 and ****p-value<0.0001.

Discussion

As an inflammatory gynecological disorder of reproductive-aged women, endometriosis is diagnosed based on the presence of endometrial-like tissue (epithelial glands and stroma) outside the uterine cavity, usually the abdominal organs and cavities such as the peritoneal mesothelium, ovaries and fallopian tubes (Bulun et al., 2019, Giudice, 2010, Taylor et al., 2021). Superficial peritoneal lesions, deep-infiltrating endometriosis, and ovarian endometriotic cysts (endometriomas) represent the most common anatomic types of pelvic endometriosis. Endometriosis is a debilitating disease, which shares some characteristics with malignancy, such as cell migration, invasion and adaptation to hypoxia. Severe symptoms of this systemic disorder often include dysmenorrhea, chronic pelvic pain, dyspareunia, infertility, and an elevated risk for ovarian cancer. Although the exact incidence of endometriosis remains uncertain, estimates suggest that the disease affects one in ten women of reproductive age (Simoens et al., 2007), 50-60% of women and teenage girls with pelvic pain (Eskenazi and Warner, 1997), and up to 50% of women with infertility (Ozkan et al., 2008). Despite the adverse impact on a patient’s quality of life that can extend well into menopause (Moradi et al., 2014), the etiopathogenesis of endometriosis remains unclear.

For the first time, we provide support for a functional role for the EGR1 transcription factor in the pathogenic properties of human endometriotic cells (epithelial and stromal), which include cell proliferation, migration, and invasion. To our knowledge, EGR1 has not been previously associated with endometriosis. In the case of endometriotic epithelial cells, our RNA-seq analysis identifies a number of biological processes and signaling pathways that may be critical for promoting EGR1-dependent pathogenic cellular processes. Such cellular processes—cell survival, proliferation, migration, and invasion—are essential for endometriotic cells migrating from the menstrual efflux to attach, colonize and invade distant anatomic sites with harsh microenvironments.

Focusing on a subset of molecular targets for which expression levels are downregulated in iHEECs following EGR1 knockdown, we find that many have been implicated in promoting pathologies in other physiological systems. For example, DPCR1—a family member of the major histocompatibility complex class I molecules—has been reported to promote cancer cell proliferation, migration and invasion (Yan et al., 2018). Exclusive to human and higher-order primates, HCAR3 has been correlated with poor long-term survival for patients diagnosed with cancers of the colon (Yang et al., 2021) and cervix (Ding et al., 2020). A member of the serpin superfamily of protease inhibitors, SERPIN A9 has been associated with the cell migration properties of endometriotic cells (Li et al., 2018) whereas SERPIN B4 has been linked to promoting cell proliferation and migration in various cancer cell lines (Heit et al., 2013). As an inflammatory cytokine, IL-6 is involved in a broad spectrum of pathophysiologies (Carmona et al., 2012, Bergqvist et al., 2001, Hirano, 2021), including endometriosis (Song et al., 2020b). A transcription factor containing a non-basic HLH dimerization domain and an atypical zinc-finger DNA–binding domain (Wang et al., 2021b), EBF2 has been linked to a set of pathologies (Li et al., 2019, Mallm et al., 2019, Ng et al., 2020), including endometriosis (Sohler et al., 2013). A member of the metallothionein family (West et al., 1990), MTIG has been associated with the promotion of cell proliferation in many cancer types (Si and Lang, 2018). In aggregate, these findings support EGR1 as a potent transcriptional regulator of diverse signals that individually or together promote the pathogenic properties of endometriotic epithelial cells.

We also provide support for CAIX as a new and important downstream EGR1 target that is critical for maintaining many of the pathogenic properties of endometriotic epithelial cells (Fig. 8). As a membrane-associated zinc metalloenzyme, CAIX was first shown to be upregulated in hypoxic tumors where it plays a central role as an intra- and extra-cellular pH regulator (Pastorekova and Gillies, 2019, Becker, 2020, Benej et al., 2020, Aldera and Govender, 2021, Queen et al., 2022). In hypoxic tumors, CAIX maintains an intracellular pH (pHi) that is favorable for continued tumor cell growth and survival, while at the same time is involved in generating an acidic extracellular microenvironment that enables tumor cell invasiveness (Shin et al., 2011, Daunys and Petrikaite, 2020). Apart from providing a survival advantage to cancer cells through intracellular neutralization while promoting tumor invasion by extracellular acidification, CAIX has been implicated as essential for modulating cell proliferation, loss of cell adhesion, increased tumor cell migration, invasion, and metastasis (Daunys and Petrikaite, 2020, Shin et al., 2011). Because CAIX is now considered an important molecular marker of poor prognosis in many cancers and diseases (Zamanova et al., 2019), CAIX has attracted increasing attention as a possible drug target to treat various forms of cancers as well as other pathologies (Pastorek and Pastorekova, 2015, Ciccone et al., 2020, Angeli et al., 2020, Supuran, 2020). Considering a growing body of evidence shows that hypoxia regulates the disease phenotype of endometriosis (Kobayashi et al., 2021, Li et al., 2021) and that therapeutic use of CAIX inhibitors has reached phase I clinical trials (McDonald et al., 2020), we believe further investigations on CAIX’s involvement in the etiopathogenesis of ectopic endometriotic lesions, and the role of CAIX inhibitors as a treatment option for this disease, is warranted.

In summary, our studies support a critical role for EGR1 in maintaining the pathogenic properties of endometriotic cells through a transcriptome that is derived from a myriad of genes that are known to mediate pathogenic responses in other physiological systems. Future investigations will focus on whether EGR1 inhibitors can be considered a feasible treatment option for endometriosis as shown in other pathological systems (Bhattacharyya et al., 2011). The fact that the use of an EGR1 inhibitor (mithramycin) has reached a phase I/II clinical trial for certain cancer types (Grohar et al., 2017), and that we have shown that mithramycin significantly suppresses the pathogenic cellular properties of the iHEEC line (data not shown), provide additional motivation to further study EGR1 inhibitors in the context of endometriosis treatment in the future. An important focus will also be the identification of the transcriptome that mediates EGR1’s pathogenic responses in endometriotic stromal cells and identify the commonalities and differences between the EGR1 transcriptomes derived from human endometriotic epithelial and stromal cells. Finally, these in vitro findings will need to be functionally validated in vivo using established animal models for experimental endometriosis in the future.

Supplementary Material

01

Supplementary Figure 1 The level of EGR1 transcript is significantly increased in human ovarian endometriomas. Expression data were obtained from the Gene Expression Omnibus datasets: GSE7305 and GSE25628 (Crispi et al., 2013, Hever et al., 2007). (A) The relative raw abundance of EGR1 transcripts in human ovarian endometrioma and matched control endometrium (A.U. denotes arbitrary units). Human ovarian endometrioma and matched control endometrium were biopsied during the proliferative (n=2) and secretory (n=8) phases of the menstrual cycle for the GSE7305 dataset. (B) Relative raw abundance of EGR1 transcripts in human ovarian endometrioma and matched control endometrium. Human eutopic, ectopic and control endometrium was obtained during the proliferative phase for the dataset obtained from GSE25628. Data are presented as mean ± SE (n= 6-10); *p-value<0.05, **p-value<0.01, ****p-value<0.0001.

02

Supplementary Figure 2 Reduction in EGR1 levels causes S phase cell cycle arrest in iHEECs. Forty-eight hours following transfection with NT or EGR1 siRNAs, iHEECs were sorted by flow cytometry to assess their stage in the cell cycle. (A-B) Histograms show the percentage of cells in each phase of the cell cycle. Data are presented as mean ± SE (*p-value<0.05) and are representative of three independent experiments.

03

Supplementary Figure 3 Knockdown of EGR1 expression does not alter EGR2 or EGR3 expression levels in the iHEEC line. (A-B) Transcript levels as measured by qPCR of EGR2 and EGR3 in iHEECs forty-eight hours following transfection with NT siRNAs or siRNAs targeting EGR1.

04

Supplementary Figure 4 Human endometriotic stromal cells require EGR1 to maintain their pathogenic cellular properties. (A) Knockdown of EGR1 in iEc-ESCs results in a significantly decreased capacity to proliferate. (B) Colony formation ability is markedly attenuated following EGR1 knockdown in iEc-ESCs; histogram in (C) shows the quantitation of crystal violet staining as a measure of colony number and size. (D) The wound-healing assay demonstrates that the migration ability of iEc-ESCs is significantly decreased with EGR1 knockdown; histogram displays the percentage area migrated for iEc-ESCs transfected with either NT or EGR1 siRNAs. The scale bar shown in bottom left panel applies to all four panels in (G). Results are represented as mean ± SE and representative of three independent experiments; *p-value<0.05, and **p-value<0.01.

05

Supplementary Folder 1 List of differential expressed genes (EGR1 siRNA versus NT siRNA groups). The first Excel sheet contains the log₂ fold change (log₂FC) between the two groups with the p-value, the false discovery rate corrected p-value (FDR) and the individual read counts by gene for each sample. Note the yellow highlighted row for EGR1 (log₂FC: −2.8), confirming significant EGR1 knockdown in this group. The second Excel sheet lists the respective GO terms by DAVID analyses. The third Excel sheet displays the PCA result for the two treatment groups in this study.