Endocrine disruptor exposure during development increases incidence of uterine fibroids by altering DNA repair in myometrial stem cells

Abstract

Despite the major negative impact uterine fibroids (UFs) have on female reproductive health, little is known about early events that initiate development of these tumors. Somatic fibroid-causing mutations in mediator complex subunit 12 (MED12), the most frequent genetic alterations in UFs (up to 85% of tumors), are implicated in transforming normal myometrial stem cells (MSCs) into tumor-forming cells, though the underlying mechanism(s) leading to these mutations remains unknown. It is well accepted that defective DNA repair increases the risk of acquiring tumor-driving mutations, though defects in DNA repair have not been explored in UF tumorigenesis. In the Eker rat UF model, a germline mutation in the Tsc2 tumor suppressor gene predisposes to UFs, which arise due to “second hits” in the normal allele of this gene. Risk for developing these tumors is significantly increased by early-life exposure to endocrine-disrupting chemicals (EDCs), suggesting increased UF penetrance is modulated by early drivers for these tumors. We analyzed DNA repair capacity using analyses of related gene and protein expression and DNA repair function in MSCs from adult rats exposed during uterine development to the model EDC diethylstilbestrol. Adult MSCs isolated from developmentally exposed rats demonstrated decreased DNA end-joining ability, higher levels of DNA damage, and impaired ability to repair DNA double-strand breaks relative to MSCs from age-matched, vehicle-exposed rats. These data suggest that early-life developmental EDC exposure alters these MSCs’ ability to repair and reverse DNA damage, providing a driver for acquisition of mutations that may promote the development of these tumors in adult life.

Effects of perinatal endocrine disruptor-exposure on DNA repair of adult rat myometrial stem cells(MSCs) were examined. MSCs from exposed animals showed increased DNA damage and diminished DNA repair.

Introduction

Uterine fibroids (UFs), benign myometrial tumors, have a profound negative impact on female reproductive health. Each year, roughly $34.4 billion healthcare dollars spent in the United States can be attributed to UF-related costs [1]. Although benign, these tumors can cause severe obstetric- and gynecologic-related complications, including excess uterine bleeding, often causing anemia, infertility, recurrent pregnancy loss, preterm labor, and postpartum hemorrhage [2–4].

The currently accepted model for UF origin is based on the conversion of myometrial stem cells (MSCs) into UF-forming MSCs (sometimes referred to as tumor-initiating cells) [5–7]. Increased tissue-specific progenitor cell number has been linked with an increased risk of genomic instability and neoplasia, given these cells’ high number of cell division events to maintain homeostasis [8–10]. Given that smooth muscle cells are among the very few cell types which do not terminally differentiate [11–14], and the uterus requires continuous remodeling during reproductive cycles and pregnancy [15–18], it is reasonable to hypothesize that an increase in the MSC progenitor population contributes to UFs risk [18–21]. Genome instability, defined as higher than normal rates of mutation, may originate from such normal cellular processes, such as transcription and replication, in such highly replicative cells [22–24].

Conversion from normal MSC to UF-forming MSCs is thought to be due to acquisition in one or more mutations in genes, such as somatic fibroid-causing mutations in gene MED12, detected in ∼85% of all sporadic human UFs [5, 6, 25, 26]; MED12 defects have been confirmed in animal models as causal for UF development [27]. Despite the high frequency of UFs (estimates of the prevalence of these tumors lie around 80%), the underlying origin(s) for these MED12 driver and other mutations remains unknown. UFs, by definition, have a very low mitotic index [28–31], adding to the paradoxical nature of the observed high frequency of MED12 driver mutations in these tumors.

For this study we have used an animal model of UFs, the Eker (Tsc2^Ek/+) rat, which possesses a germline heterozygous tumor suppressor gene, Tsc2, mutation and spontaneously develops UFs after ∼16 months of age following loss of the second normal Tsc2 allele [32–34]. In previous studies, we have shown that Eker rats developmentally exposed to diethylstilbestrol (DES, a tool compound of environmental endocrine-disrupting chemicals [EDCs]) during early life form fibroids later in adult life at higher frequency (100% tumor penetrance) and increased size, number, and severity versus unexposed counterparts (∼65% penetrance) [2, 35]. These DES-exposed rats have also been shown to develop the UF tumors earlier in adult life (∼12 months of age), which may be attributed to more rapid accumulation of DNA damage, potentially leading to an earlier loss of the second normal Tsc2 allele. These findings strongly suggest increased mutagenesis and decreased ability to appropriately repair DNA damage/breaks in these developmentally early-life EDC-exposed animals [2, 35].

We have recently shown that the Stro1⁺/CD44⁺ myometrial stem/progenitor-like cell (MSC) population isolated from these DES-exposed rats’ uteri are expanded in number and proliferate significantly faster than normal MSCs, suggesting their integral role in increased penetrance of UF tumors in DES-exposed rats [21]. A normal eukaryotic cell generates ∼70,000 DNA lesions each day that may generate mutations; most frequently, single-nucleotide substitutions incorporated by DNA polymerases, through normal replication processes, accumulate at a low, but constant rate [8, 22]. Even small errors in DNA synthesis can cause many mutations, thus underscoring replication machinery as a source of mutagenesis; post-replication DNA repair processes involve homologous recombination (HR) double-strand break (DSB) repair, thus suggesting that DNA DSBs persisting post-replication may remain irreparable if HR components are compromised [22, 36]. Thus, this expanded, highly proliferative MSC population is at increased risk of DNA mutations during replication; this increased chance of mutation, combined with putatively impaired DNA repair systems, implicates the DES-exposed rat MSCs in the more penetrant development of UF tumors in the exposed Eker rats. In this work, we aimed to evaluate the DNA repair system in the Stro1⁺/CD44⁺ MSC population of an early-life EDC-exposed versus unexposed (normal) rat fibroid model to explore whether the changes induced in these rats exposed to DES during the sensitive window of early uterine development include dysregulation and/or impairment of DNA repair systems as well.

Materials and methods

Environmental exposure animal model

Roughly 65% of Eker rats, which carry a germline mutation in the tuberous sclerosis 2 (Tsc2) tumor suppressor gene, spontaneously develop uterine leiomyoma between ages 12 and 16 months [34]. Recently, we used this model to investigate the role of MSCs in the pathogenesis of these benign tumors [21]. Female newborn Eker (Tsc2 ^Ek/+) rats (n = 5 per group) were treated S.C. with vehicle (VEH) or 1 μg/kg of DES (a tool compound of environmental EDCs) on postnatal days (PND) 10–12, a key period of uterine development. They were then maintained until sacrifice after 5 months of age. It has been previously demonstrated that exposure of Eker rats to DES during PND 10–12 (key period of uterine development) increases incidence of UF formation to 100% between ages 12 and 16 months [2, 35]. Upon sacrifice, reproductive tissues were collected for immunohistochemistry (IHC) and for quantification and isolation of Stro1⁺/CD44⁺ MSCs from DES- and VEH-exposed rats (DES and VEH, respectively). Protocols involving the use of these animals were approved by the Committee on the Ethics of Animal Experiments, Augusta University (Augusta, GA).

Ex vivo immunohistochemistry for DNA repair protein expression analysis

Ex vivo tissue sample slide preparation, IHC staining, imaging, and imaging analysis were completed by the Research Histology and Tissue Imaging Core at the University of Illinois at Chicago. Briefly, formalin-fixed paraffin-embedded (FFPE) myometrial tissues from DES- and VEH-exposed animals were constructed into single-layer tissue slides for staining and analysis with traditional IHC. Tissue sections were stained on Bond RX autostainer (Leica Biosystems) following a preset protocol. Sections were deparaffinized and subjected to citrate-based (Bond ER1 solution, pH6) antigen retrieval for 20 min at 100°C for Rad51 staining and to EDTA-based (Bond ER2 solution, pH9) antigen retrieval for 40 min at 100°C for BRCA2 staining. Sections were blocked for 5 min with hydrogen peroxide block and washed with Bond Wash Solution, followed by incubation with anti-BRCA2 (bs­1210R, Bioss Antibodies, Woburn, MA; 1:250 dilution) or anti-Rad51 (ab133534, Abcam, Cambridge, MA; 1:1000 dilution) antibodies for 30 min. The detection was performed using Bond Polymer Refine Detection kit (Leica Biosystems, DS9800). All slides were counterstained with Mayer's hematoxylin for 5 min and mounted with Surgipath Micromount Media (Leica Biosystems).

Image acquisition and analysis

Slides were scanned at ×200 magnification using a Leica Aperio AT2 slide scanner (Leica Biosystems). After slide scans were completed, images were loaded into Leica eSlide Manager and regions of interest were manually drawn around the compartment of interest, where myometrial tissue areas were annotated for DES and VEH tissues for further staining analysis and quantification. The Leica Aperio Positive Pixel v9 algorithm was used to measure the percent positive area, average brown stain intensity in the entire tissue of interest, and the average brown stain intensity in the positive tissue of interest. All cells (both positively and negatively stained) throughout the myometrium annotated in each image were identified using this algorithm with thresholds for negative, weak, moderate, and strong positive staining. Staining intensity for all positively stained cells in the field was also quantified. Once scored, each category of staining was then automatically counted. From this, we calculated the percentage of negative, weak, moderate, and strong positively stained cells in each respective cell type, DES and VEH, and presented these data in stacked columns to show the staining percentages.

DES- and VEH-exposed Eker rat myometrial Stro1⁺/CD44⁺ stem cell isolation and culture

Tissue digestion and stem cells isolation was performed as previously described [21]. Briefly, uterine tissues from DES- (N = 5) and VEH- (N = 5) exposed Eker rats were collected and rinsed well in freshly prepared buffer solution three times to remove residual blood, and endometrial and serosal tissues were removed from the myometrial tissues by scraping with a sterile scalpel to ensure only myometrial tissue was present for digestion (Life Technologies). Myometrial tissue was digested into single-cell suspensions, which were first subjected to immunophenotypic analysis to exclude hematopoietic (CD34 and CD45) and endothelial stem cells (CD31), by which it was determined that the myometrial cell suspensions were negative for these markers [18, 21]. Then, these fresh myometrial cell suspensions were subjected to selection for Stro1/CD44 double-positivity by fluorescence-assisted cell sorting (FACS) to isolate Stro1⁺/CD44⁺ MSCs, which contained two distinct cell populations: a small population of Stro-1⁺/CD44⁺ MSC population (4.58% of total myometrial cells) and primary/differentiated Stro-1⁻/CD44⁻ myometrial cells (82.77% of total myometrial cells), as previously determined [18, 21]. Then, isolated DES and VEH MSCs were cultured in normal growth media (SmGM-2, Lonza) separately in collagen-coated dishes, and maintained under hypoxic conditions (37°C, 5% CO₂, 2% O₂). Using IHC, it was also demonstrated that these Stro-1⁺/CD44⁺ MSCs co-localize with markers of stemness (e.g. NANOG, c-KIT, and OCT-4) and are localized within murine myometrium in the cervix and horns of the murine uterus [21]. Studies have been completed previously in other systems using similar methods to identify, isolate, and characterize tissue-specific, tissue-resident stem/progenitor cells [37–40]. For immunostaining, DES and VEH cells were grown in 6-well plates containing 25 mm sterile circular microscope cover glasses.

RNA library preparation and whole genome RNA sequencing and assessment

RNA isolation and library preparation of DES and VEH MSCs was completed at the UNC Lineberger Comprehensive Cancer Center Genomics Core (Chapel Hill, NC) for Illumina sequencing. RNeasy RNA isolation kit (Qiagen, Valencia, CA) was used to isolate RNA for RNA-seq analysis. RNA samples were treated with DNase I, and the RNA purity and quality was checked by Bioanalyzer prior to cDNA synthesis. Complementary DNA libraries were constructed using SPRI-works Fragment Library System I (Beckman Coulter, Brea, CA), which were then PCR-enriched and purified. Ten pM DNA were loaded into the paired end flow cell on cBOT for cluster generation and thereafter loaded on the HiSeq2500 platform (Illumina Inc., San Diego, CA) to generate single 36 bp sequence reads. Sequence reads were aligned to the rat reference genome (rn4; version 3.4 by BCM HGSC) using TopHat aligner and aligned read counts were summarized using ShortRead and associated Bioconductor packages.

For initial quality control (QC) assessment, FastQC was used to ensure overall good quality of data before trimming for further analysis. Read trimming was then performed using BBduk.sh in the BBMap suite. Any reads <20 bp after trimming were discarded. Only Illumina adapters were scanned and trimmed. No Q-scores-based trimming was done as it was not necessary. Using BBmap.sh for read alignments, rn5 genome build (University of California, Santa Cruz), and annotation files (iGenomes), BBMap aligner indexes were created using the genome.fa file (iGenomes). Parameters for alignments: maximum indels allowed 100 kb, any multimapping reads were randomly assigned, and SAM v.1.3 format tags were used in SAM/BAM files. BAM files were then sorted and indexed using Samtools [41]. To estimate expression of genes of interest, featureCounts was used to count reads that map to genomic features, and DESeq2 analysis was completed using the summarized count matrix/gene level [42, 43] (files have been submitted to Gene Expression Omnibus (GEO) data repository, pending accession number).

RNA Seq data (Supplemental Table S4) was used to screen for altered DNA repair-related gene expression in DES vs. VEH cells; a review of the literature, DNA repair PrimePCR templates, and KEGG Pathway Maps was completed, and 202 genes from the RNA Seq analysis were categorized as DNA repair-related genes (Supplemental Table S1). Of these, 76 of 202 genes were upregulated (Log2 Fold Change (F.C. ≥ 0.0), 8 markedly (F.C. ≥ 1.0) upregulated, while 122 of 202 genes were downregulated (F.C. ≤ 0.0), 20 of these 122 appreciably (F.C. ≤ –1.0). These 28 genes were selected for further analysis and validation by quantitative PCR (Supplemental Table S1). In addition, Gene Ontology (GO) enrichment analysis (PANTHER13.1 Classification System) was performed (Reference List: Rattus norvegicus) on the respective up- and downregulated gene sets to determine the significant shared GO terms used to describe the sets of genes [44].

Gene validation by real-time quantitative reverse transcription polymerase chain reaction

To confirm the results obtained by whole genome RNA sequencing, real-time quantitative reverse transcription polymerase chain reaction (qRT-PCR, Bio-Rad CFX96) was performed for the selected 8 and 20 most up- and downregulated genes, respectively. The mRNA levels of 28 genes were quantified using gene-specific primers (Supplemental Table S2). The mRNA values for each gene were normalized to internal control 18S rRNA, and gene expression of DES vs. VEH control cells was calculated from 2^−ΔΔCt and expressed as Log2 Fold Change. All experiments were performed in triplicates.

Drug preparation

Bleomycin (BLM, Sigma, St. Louis, MO) stock solution was prepared and then diluted in culturing medium to 15 μg/mL for each experimental treatment.

γ-H2AX foci induction by bleomycin treatment

DES vs. VEH cells were plated in 6-well plates (50,000 cells/well) and cultured to confluency. Cells were treated with 0 and 15 μg/mL BLM (Sigma) for 1 h as a radiomimetic used for inducing DNA DSBs [45, 46]. Following 1 h of treatment, BLM was removed and cells were either fixed immediately with 4% paraformaldehyde (PFA) for 15 min at room temperature (RT) or maintained in SmGM-2 to recover until the following time points: 30 min, 6 h, and 24 h, at which points media was removed and cells were fixed in the same manner (4% PFA, 15 min, 25°C), washed, and stored at 4°C until immunofluorescence staining.

γ-H2AX foci staining for immunofluorescence

For staining, cells were permeabilized for 30 min with 0.1% Triton X-100 in PBS (0.1% PBS-T). After washing, cells were then blocked for 30 min with 1% bovine serum albumin (BSA) in 0.1% PBS-T blocking buffer. Staining with the phospho-histone H2A.X (Ser139) rabbit antibody (2577, Cell Signaling, Danvers, MA) was performed for 1 h at a 1:250 dilution in blocking buffer (Supplemental Table S3). Samples were washed three times with PBS and then stained with goat anti-rabbit IgG (H + L) secondary antibody, Alexa Fluor 555 (A-21429, Thermo Fisher, Waltham, MA) for 1 h in the dark at a 1:1000 dilution in blocking buffer. Samples were again washed three times with PBS and stained with 4′,6-diamidino-2-phenylindole (DAPI) (Sigma) for nuclear counterstaining. Cells were washed three times again, and coverslips were fixed on slides for imaging using Fluoroshield mounting medium (Sigma) and dried overnight in the dark.

Confocal fluorescence imaging of γ-H2AX foci

For cell quantification and the study of γ-H2AX foci formation in Stro-1⁺/CD44⁺ DES vs. VEH rat MSCs (DES vs. VEH), confocal images were captured using a Plan apochromat ×40/1.4 oil DIC M27 objective lens (Zeiss 780 Upright Confocal, 1024  ×  1024 pixels) in a blinded manner of 10 representative fields for each cell, treatment, and recovery time point. For each field, two images were taken: (1) a DAPI image with a band filter of 410–559 nm (405 nm laser, 646 master gain, 30 μm pinhole), and (2) a γ-H2AX image with a 548–697 nm band filter (543 nm laser, 621 master gain, 37 μm pinhole). The pictures were saved as 16-bit Zeiss Vision Image ZVI files with no further editing.

Quantification of γ-H2AX foci-positive cells and foci fluorescence density

To quantify γ-H2AX foci-positive cells, using inForm(R) Cell Analysis(TM) v.2.3 software (PerkinElmer), individual cell nuclei were phenotyped and deemed γ-H2AX foci-positive based on the presence of ≥5 γ-H2AX foci or γ-H2AX foci-negative (normal). Standardized threshold for nuclei detection and quantification was used in a “batch” application for all already-acquired images. Percent (%) γ-H2AX foci-positive nuclei were quantified automatically based on the two cell phenotypes (γ-H2AX foci-positive or normal) by the inForm(R) Cell Analysis(TM) software. In addition, the integrated fluorescence intensity of individual cells was also quantified for each nucleus, as it accurately measures γ-H2AX expression when considering the inability to accurately discriminate between foci in a highly damaged cell [47]. Using ImageJ v.1.48k software (NIH, Bethesda, MD), the integrated density (IntDen) was calculated in all images for each nucleus to determine the relative fluorescence density and intensity, corresponding to γ-H2AX expression, after subtracting the background signal of each nucleus. The IntDen is a calculus of the mean stained area times the intensity of stain in each pixel in the area and indicates the total amount of stained material in that area.

DNA double-strand breaks induction by bleomycin

DES vs. VEH cells were cultured to confluency. Cells were treated with 0 and 15 μg/mL BLM (Sigma) for 1 h; then, BLM was removed and cells were maintained in SmGM-2 to recover until the following time points: 30 min, 6 h, and 24 h, at which points media was removed and cells were fixed immediately with 4% PFA for 15 min at RT (25°C), washed, and collected for protein isolation.

Protein expression analysis by western blot

DES and VEH cell pellets were lysed in RIPA buffer (10 mm tris-HCl, pH 7.4, 150 mm NaCl, 1% sodium deoxycholate, 1% Triton X-100) containing protease inhibitors (2 μg/ml aprotinin, 2 μg/ml leupeptin, 5 μg/ml pepstatin, 1 mm phenylmethane sulphonyl fluoride) for 30 min on ice with thorough agitation every 5 min, and centrifuged for 10 min at 12,000 RPM at 4°C. Per cell (DES vs. VEH) and treatment/recovery time point, three experimental replicates were run (e.g. Lane #1: VEH, untreated (UT)-1; Lane #2: DES, UT-1; Lane #3: VEH, UT-2, etc.). Samples equivalent to 30 μg of protein were separated using 4–20% MiniPROTEAN TGX Precast Protein Gels (Bio-Rad, Hercules, CA) and transferred to PVDF membranes according to standard procedures. Equal loading and transfer were assessed by re-probing the blots with anti-β-actin antibody (A5441, Sigma). Membranes were blocked for 1 h at RT in either 5% w/v nonfat dry milk or 5% BSA in 0.1% Tween-supplemented PBS (0.1% PBS-T) per antibody specification. Membranes were then incubated with primary antibodies (Supplemental Table S3): mouse anti-phospho-ATM (LS-C2763, LifeSpan Biosciences, Seattle, WA; 1:1000 dilution), rabbit anti-ATM (ab199726, Abcam; 1:1000 dilution), rabbit anti-phospho-histone H2A.X (Ser139) (2577, Cell Signaling, Danvers, MA; 1:1000 dilution), rabbit anti-phospho-p95/NBS1 (Ser343) (SAB4504479, Sigma; 1:1000 dilution), rabbit anti-p95/NBS1 (3002, Cell Signaling; 1:1000 dilution), rabbit anti-phospho-CHK1 (Ser345) (2348, Cell Signaling; 1:1000 dilution), mouse anti-CHK1 (2360, Cell Signaling; 1:1000 dilution) overnight at 4°C in either 1% w/v nonfat dry milk or 1% BSA in 0.1% PBS-T per antibody specification. Membranes were washed in 0.1% PBS-T and then incubated with anti-rabbit (7074, Cell Signaling) or anti-mouse (7076, Cell Signaling) horseradish peroxidase-labeled secondary antibody (Cell Signaling; 1:5000 dilution) in 1% w/v nonfat dry milk or 1% BSA in 0.1% PBS-T. β-Actin was used as an internal standard to account for variations in the amount of protein loaded in each lane. Protein bands were detected using an enhanced chemiluminescence system, ChemiDoc XRS + System (Bio-Rad). Protein expression for each lane was quantified using the Bio-Rad Image Lab Software (Bio-Rad) and normalized β-actin. Normalized protein expression for each DES MSC sample and treatment/recovery time point (UT, BLM 15 + 0.5, 15 + 6, 15 + 24 h) was then normalized to each respective VEH MSC sample and treatment/recovery time point to determine the DES: VEH expression ratios; then, the mean DES: VEH expression ratio was determined for each treatment/recovery time point from the respective experimental replicates.

Plasmid DNA preparation

Briefly, SV40-luc pGL2-control vector plasmid DNA (Supplemental Figure S1A, Promega, Madison, WI) was used to transform DH5α Escherichia coli cells for plasmid amplification and was then isolated using PureYield Plasmid Midiprep System (Promega). HindIII restriction enzyme digestion (New England Biolabs) in CutSmart Buffer (New England Biolabs) was completed for SV40-luc pGL2-control plasmid to linearize the plasmid and disrupt promoter-driven luciferase activity. HindIII-digested plasmid (a.k.a. HindIII/pGL2-control) was purified on agarose gels to confirm and isolate linearized plasmid.

Luciferase gene reporter assay to detect DES vs. VEH cells’ end-joining capacity of DNA double-strand breaks

The luciferase reporter in the pGL2-control plasmid, driven by the SV40 gene promoter and enhancer (6 kb) (SV40-Luc), was used to determine end-joining activity in DES and VEH MSCs following co-transfection with HindIII/pGL2-control + CMV-β-galactosidase (CMV-β-gal) plasmids. Experimental conditions and controls were based on previously established DNA repair analyses assays that utilize transfection methods and exogenously damaged gene reporter plasmids to measure DNA repair/end-joining efficiencies [48–50]. X-tremeGENE HP (Roche Applied Science, Indianapolis, IN, USA) was used to transfect cells with plasmid DNA according to the vendor's protocol. Briefly, fresh cultures of VEH and DES cells were each co-transfected with 2 μg of HindIII/pGL2-control plasmid and 0.2 μg (1/10 w/w) of CMV-β-gal plasmid (to normalize for respective cell transfection efficiency between DES and VEH cells) in 6 cm dishes using XtremeGENE HP transfection reagent (Roche). In parallel, fresh cultures of VEH and DES cells were each co-transfected [in a ratio of 1:9 (2 μg total)] with supercoiled pGL2-control: pGL2-basic plasmids and 0.2 μg (1/10 w/w) of CMV-β-gal plasmid. These co-transfected cells controlled for any differences in endogenous SV40 promoter/luciferase activity and pGL2 plasmid transfection efficiency between DES and VEH MSCs. Cells were incubated for 48 h and then harvested for luciferase activity using a luminometer. Cell lysates from respective samples were also analyzed using Tropix Galacto-Light chemiluminescent reporter gene assay system (T1006, Applied Biosystems, Bedford, MA) to detect β-galactosidase reporter enzyme in order to quantify transfection efficiency of DES and VEH MSCs. Luciferase readings in HindIII/pGL2-control/CMV-β-gal co-transfected cells were first normalized to β-gal readings to account for DES and VEH cell transfection efficiency. Then, these β-gal-normalized luciferase readings were normalized to the DES: VEH luciferase activity ratio obtained from pGL2-control: pGL2-basic plasmid/CMV-β-gal co-transfected cell luciferase readings; this controlled for any differences in endogenous SV40 promoter/luciferase activity between DES and VEH cells. Three independent experiments were performed, each in triplicate for each transfection condition, for each cell type. Results were expressed as normalized (as described above) luciferase activity of HindIII/pGL2-control/CMV-β-gal co-transfected DES vs. VEH MSCs. Because DES and VEH MSCs undergo identical experimental conditions (aside from initial DES exposure in the animal model, e.g. isolation, characterization, culture, maintenance, treatment, etc.), any difference in luciferase activity is likely due to the initial DES exposure event in the respective animals, which we have hypothesized to implicate reprogramming of DNA repair-related genes and mechanisms.

Statistical analyses

Quantitative RT-PCR data were analyzed using unpaired Student t-test for comparative parametric analysis with a significance level of P-value < 0.05 considered statistically significant. For γ-H2AX quantification by ICC/IF, the Kruskal-Wallis test for one-way analysis of variance (ANOVA) was utilized for nonparametric analysis of all treatment and recovery time points of DES vs VEH groups. ANOVA justified post hoc comparisons between group medians, which were conducted using the Dunn test for multiple comparisons. Differences were considered significant at P < 0.05. For western blot analyses, the mean relative protein expression ratio (DES: VEH) was compared to 1 at each time point (for each experimental replicate) using a two-tailed one-sample t test; a mean significantly different from 1 indicated DES protein expression differed from that in VEH cells. Differences were considered significant at P < 0.05. For Luciferase gene reporter assay analyses, luciferase activity in DES vs. VEH cells was compared using a two-tailed unpaired Student t-test for comparative parametric analysis with a significance level of P-value < 0.05 considered statistically significant. All of these analyses were completed using GraphPad Prism 7.03.

Results

Expression of DNA repair-related genes/proteins is reprogrammed by endocrine-disrupting chemical exposure

Ex vivo IHC analyses of myometrial tissues extracted from DES-exposed animals demonstrated decreased overall staining of DNA repair proteins, BRCA2 and RAD51 (Supplemental Figure S2A and C). Although both tissues demonstrate large percentages of negatively stained cells, myometrial tissue from DES-exposed animals demonstrated lower percentages of cells in weakly positive, positive, and strongly positive staining categories than was found in that of VEH-exposed animals (Supplemental Figure S2B and D). Transcriptional profiling of MSCs (the cells of origin for UFs) from uteri of adult rats exposed neonatally to DES or VEH revealed that developmental DES exposure had reprogrammed the expression of DNA repair-related genes (Figure 1). These genes were categorized as they either directly or indirectly (e.g. via transcription regulation, cell cycle processes, etc.) participate in DNA repair; interestingly, of those genes found by RNA Seq to be up- or downregulated, there were more downregulated genes in each category relating to DNA repair-related processes (as characterized using GO: Biological Processes) versus those upregulated genes (Figure 1B). RNA seq data were validated by quantitative PCR, which confirmed 7 of 8 upregulated and 17 of 20 downregulated genes exhibited significantly altered expression in DES- relative to VEH-exposed MSCs (Figures 1 and 2; Supplemental Table S1). Remarkably, over half of the significantly downregulated genes were involved in HR (log2 fold change, P-value): Bard1 (–1.99, P = 0.004), Brca2 (–1.16, P = 0.004), Cdk1 (–1.05, P = 0.0002), Fancd2 (–3.10, P < 0.0001), Palb2 (–1.27, P = 0.023), Rad51ap1 (–1.57, P < 0.0001), Rad52 (–1.30, P = 0.002), and Xrcc2 (–1.07, P = 0.019). Several genes were related to nonhomologous end-joining (NHEJ): Polq (–1.93, P < 0.0001), Nhej1 (–1.59, P = 0.005), Gstp1 (–0.77, P = 0.002), and cell cycle regulation or other DNA repair-related processes: Ccnb1 (–1.56, P = 0.001), Gadd45a (–0.78, P = 0.0008), Gadd45b (–1.15, P = 0.001), Hap1 (–1.95, P = 0.0005), Recql4 (–1.67, P = 0.002), and Xpa (–0.85, P = 0.014) (Figure 2).

Figure 1.

DES differentially express pivotal DNA repair-related genes vs. unexposed VEH MSCs. (A) A summary of RNA Seq results of 202 DNA repair-related genes, in DES cells, 8 genes were strongly upregulated (cut-off: Log2 fold change (FC) ≥ 1.0), and 20 notably downregulated (cut-off: FC ≤ –1.0). (B) DNA repair-related genes found by RNA Seq to be up- or downregulated in DES cells and categorized by gene ontology (GO) biological process; overall, for each GO category shown, more genes were downregulated in DES cells than were upregulated.

Figure 2.

Confirmation of DNA repair-related gene expression shown most differentially expressed in DES vs. VEH MSCs. Quantitative RT-PCR validation of 8 upregulated and 20 downregulated genes confirmed 7/8 genes were significantly upregulated, and 17/20 genes were significantly downregulated in DES as compared to VEH cells (FC, P-value): Myh1 (7.02, P < 0.0001), Myh4 (4.37, P < 0.0001), Myh8 (1.29, P = 0.058), Cdkn1a (2.32, P < 0.0001), Mgmt (2.76, P < 0.0001), Plk3 (3.13, P < 0.0001), Xrcc5 (1.69, P = 0.0003), Tyms (1.94, P = 0.0002), Bard1 –1.99, P = 0.004), Brca2 (–1.16, P = 0.004), Cdk1 (–1.05, P = 0.0002), Fancd2 (–3.10, P < 0.0001), Palb2 (–1.27, P = 0.023), Rad51ap1 (–1.57, P < 0.0001), Rad52 (–1.30, P = 0.002), Xrcc2 (–1.07, P = 0.019), Polq –1.93, P < 0.0001), Nhej1 (–1.59, P = 0.005), Gstp1 (–0.77, P = 0.002), Ccnb1 (–1.56, P = 0.001), Gadd45a (–0.78, P = 0.0008), Gadd45b (–1.15, P = 0.001), Hap1 (–1.95, P = 0.0005), Recql4 (–1.67, P = 0.002), and Xpa (–0.85, P = 0.014). Bars represent Log2 Fold Change expression from RNA Seq data (black bars) or qRT-PCR validation (gray bars) in DES (vs. VEH control) ± SD. *P < 0.05, **P < 0.0001.

Reprogrammed myometrial stem cells demonstrate decreased ability to repair DNA double-strand breaks

Decreased expression of DNA damage response (DDR) genes was accompanied by attenuated DNA repair as determined by persistent activation (phosphorylation, “p-”) of ataxia telangiectasia mutated (ATM), checkpoint kinase 1 (CHK1), Nibrin [also known as cell cycle regulatory protein p95 (p95) or Nijmegen breakage syndrome protein 1 (NBS1)], and histone 2A, variant X (H2AX). DES cells demonstrated an insignificant increase in mean phosphorylated H2AX expression in DES vs. VEH MSCs (mean DES: VEH expression ± SD, P-value): γ-H2AX (1.21 ± 0.22, P = 0.24), suggesting increased DNA damage and/or an exaggerated DDR in DES-exposed MSCs (from adult rats), even prior to BLM treatment vs. unexposed VEH cells (Figure 3C and D). After DNA DSB induction with BLM, DES MSCs demonstrated significantly higher mean γ-H2AX, p-CHK1, and p-ATM expression after 30 min of recovery: γ-H2AX (1.41 ± 0.13, P = 0.03) (Figure 3D), p-ATM (1.21 ± 0.06, P = 0.03) (Figure 4B), and p-CHK1 (1.55 ± 0.21, P =0 .04) (Figure 4D). DES: VEH expression ratio of p-Nibrin ratio was increased as well, though not significantly (1.80 ± 0.33, P = 0.052) (Figure 3B). Even after 6 h of recovery, both γ-H2AX and p-CHK1 mean DES: VEH expression ratios increased, though neither reached statistical significance [(1.49 ± 0.20, P = 0.052) (Figure 3D) and (2.09 ± 0.55, P = 0.07) (Figure 4D)]. DES: VEH expression ratios, however, remained significantly elevated in p-Nibrin (1.50 ± 0.20, P = 0.04) (Figure 3B) and p-ATM (1.88 ± 0.29, P = 0.04) (Figure 4B). Moreover, even after 24 h of recovery to allow time to repair DSBs, the DES: VEH expression ratio for all four phosphorylated proteins remained significantly elevated: p-Nibrin (1.50 ± 0.15, P = 0.03) (Figure 3B), γ-H2AX (2.18 ± 0.38, P = 0.03) (Figure 3D), p-ATM (2.05 ± 0.41, P = 0.047) (Figure 4B), and p-CHK1 (1.26 ± 0.07, P = 0.03) (Figure 4D). Altogether, these data suggest that DDR initiation may be intact, but early-life exposed DES cells are unable to repair DNA DSBs as efficiently as VEH control cells and may harbor residual long-term DNA damage.

Figure 3.

Increased Nibrin activation and DNA double-strand breaks (DSBs) in DES vs. VEH MSCs measured by increased phospho-Nibrin and γ-H2AX expression. (A) Western blot of phospho-Nibrin expression in DES vs. VEH MSCs: untreated (UT), BLM-treated + 0.5, 6, and 24 h of recovery. (B) By western blot analysis, DES cells demonstrated significantly increased activation of Nibrin relative to VEH cells following treatment with BLM, even after 24 h of recovery. (C) Western blot of γ-H2AX expression in DES vs. VEH MSCs: untreated (UT), BLM-treated + 0.5, 6, and 24 h of recovery. (D) By western blot analysis, DES cells demonstrated significantly increased DNA DSBs (reflected by increased expression of γ-H2AX) relative to VEH cells following treatment with BLM, even after 24 h of recovery. Note: Each treatment/recovery time point for each VEH or DES MSC sample displays only one representative image from all of the experimental replicates completed for each treatment/recovery time point. For each time point, the mean expression ratio of p-Nibrin: Nibrin or γ-H2AX in DES: VEH was compared to 1 using a two-tailed one-sample t-test. Lines represent means ± SD. *P < 0.05

Figure 4.

Increased DNA damage response in DES vs. VEH MSCs measured by increased phosphorylation of ATM and Nibrin following bleomycin treatment. (A) Western blot of p-ATM expression normalized to total ATM expression in DES vs. VEH cells: untreated (UT), BLM-treated + 0.5, 6, and 24 h of recovery. By western blot analysis, DES cells demonstrated (B) significantly elevated p-ATM expression relative to VEH following BLM treatment, even after 24 h of recovery. (C) Western blot of p-CHK1 expression normalized to total CHK1 expression in DES vs. VEH cells: untreated (UT), BLM-treated + 0.5, 6, and 24 h of recovery. By western blot analysis, DES cells demonstrated (D) significantly elevated p-CHK1 expression relative to VEH following BLM treatment, after 0.5 and even 24 h of recovery. Note: Each treatment/recovery time point for each VEH or DES MSC sample displays only one representative image from all of the experimental replicates completed for each treatment/recovery time point. For each time point, the mean expression ratio in DES: VEH was compared to 1 using a two-tailed one-sample t-test. Lines represent means ± SD. *P < 0.05

Quantifying γ-H2AX foci immunofluorescence, significantly more UT DES MSCs were γ-H2AX foci-positive (49 ± 8%) vs. VEH control MSCs (18 ± 4%), P < 0.0001 (mean % γ-H2AX foci-positive ± SEM) (Figure 5B). In addition, DES MSCs showed a significantly higher median fluorescence intensity of γ-H2AX foci per nucleus: 2.91 (interquartile range (IQR): 2.05–4.20) as compared to VEH MSCs, 1.27 (IQR: 0.26–2.51), P < .0001 (Figure 5C). Moreover, both DES and VEH MSCs demonstrated increased mean % γ-H2AX foci-positive nuclei: (90 ± 6%) and (88 ± 5%) (Figure 5B) and median γ-H2AX foci fluorescence following induction of DSBs with BLM and 30 min of recovery: 10.08 (IQR: 5.00–22.76) and 7.97 (IQR: 5.20–13.07), respectively (P < 0.0001) (Figure 5C). This indicated that BLM had induced DSBs and showed a highly significant increase in γ-H2AX foci formation in BLM-treated DES and VEH MSCs vs. UT MSCs, as expected (Figure 5B and C). Interestingly, however, even after 6 h of recovery time, mean % γ-H2AX foci-positive nuclei remained elevated in both DES (86 ± 4%) and VEH (83 ± 5%) MSCs (Figure 5B). In addition, median integrated γ-H2AX foci fluorescence in DES MSC nuclei remained significantly elevated, 7.836 (IQR: 5.97–10.68) as compared both to UT MSCs, DES: 2.91 (IQR: 2.05–4.20) P < 0.0001 and VEH: 1.27 (IQR: 0.26–2.51) P < 0.0001; and to DNA DSB-induced VEH cells, 4.02 (IQR: 2.57–7.54) P < 0.0001 (Figure 5C). Importantly, no significant difference in median integrated γ-H2AX foci fluorescence existed between DES cells allowed to recover 30 min vs. 6 h post-DNA DSB induction: 10.08 (IQR: 5.00–22.76) and 7.836 (IQR: 5.97–10.68), respectively, P > 0.9999 (Figure 5C). Finally, after 24 h of recovery, though both DES and VEH MSCs demonstrated decreased % γ-H2AX foci-positive nuclei, DES MSCs maintained a greater % of γ-H2AX foci-positive nuclei (80 ± 3%) vs. VEH MSCs (57 ± 4%), P = 0.03 (Figure 5B) and a significantly higher median integrated γ-H2AX foci fluorescence: 6.37 (IQR: 4.71–8.34) vs. VEH MSCs: 3.66 (IQR: 2.33–5.61), P < 0.0001 (Figure 5C). These data suggest DES MSCs have a slowed rate of DNA DSB repair vs. healthy control MSCs, which may ultimately contribute to decreased ability to repair probable mutations in DNA or DNA DSBs, such as the loss of the remaining normal Tsc2 allele, leading to increased risk of fibroid tumorigenesis.

Figure 5.

Increased DNA double-strand breaks (DSBs) in DES vs. VEH MSCs measured by increased percent (%) γ-H2AX-positive cells and increased density of γ-H2AX foci fluorescence. (A) Immunocytochemistry/immunofluorescence demonstrates increases in DNA DSBs (γ-H2AX foci reflect DNA DSB breaks) in DES vs. VEH cells prior to and following treatment with bleomycin (BLM). (B) Percent (%) γ-H2AX-positive (≥5 foci/nucleus) nuclei per high-power field (hpf) were quantified for each cell type, treatment, and time point. Mean % were compared using one-way analysis of variance (ANOVA) followed by multiple comparisons. Lines represent means ± SEM. *P < 0.05, n.s. = not significant. (C) Fluorescence density quantified by ImageJ “RawIntDen” measurements. Scale bars = 20 μm. Medians were compared using non-parametric Kruskal-Wallis followed by multiple comparisons. Lines represent medians, interquartile range (IQR). *P < 0.0001, n.s. = not significant

Developmentally DES-exposed adult rat myometrial stem cells demonstrate decreased ability to complete end-joining of DNA double-strand breaks in HindIII-linearized SV40-Luc pGL2-control vector

While both DES and VEH cells demonstrated appropriate transfection and SV40 promoter activity, mean (±SD) DES stem cell DNA end-joining efficiency (AU), 415.8 ± 49.8, was significantly less than that of VEH stem cells: 3095 ± 695, P < 0.0001 (relative to their respective pGL2-control luciferase activities, normalized for transfection efficiency) (Figure 6). The pGL2-control luciferase (luc) reporter vector contains both SV40 promoter (bp 42–244) and SV40 enhancer sequences upstream of the luciferase (luc) gene (bp 268–1920) to enable strong expression of the luciferase coding region in order to confirm mammalian cell-contextual transcriptional activity (Supplemental Figure S1A). The pGL2-control vector contains a HindIII restriction site (bp 239) that can interrupt the SV40 promoter activity of Luc transcription (Supplemental Figure S1A). As such, loss of luciferase activity is indicative of complete linearization of the pGL2-control plasmid using HindIII restriction enzyme (HindIII/pGL2), and resurgence of luciferase activity indicates efficient DNA double-strand re-end joining, as observed with other plasmid transfection-DSB repair systems [51, 52]. When HindIII/pGL2-control/CMV-β-gal is co-transfected into DES and VEH cells, therefore, the relative, normalized luciferase activity demonstrated in each cell type corresponds to DES vs. VEH cells’ ability to complete DNA double-strand re-end joining. Thus, the decreased DNA DSB end-joining activity may be ultimately indicative of DES stem cells’ decreased functional capacity for repairing DNA DSBs. As described in the Methods section, to compare transfection efficiency, DES and VEH cells were co-transfected with the pGL2-control vector, in which luciferase is expressed from the same regulatory elements as HindIII/pGL2-control, and with CMV-β-gal plasmid vector. Similar transfection efficiencies were found for both DES and VEH cells, indicating that the increased luciferase activity in the VEH cells was not attributable to a higher transfection efficiency (Supplemental Figure S3) [48–50]. Therefore, this robust, significantly decreased DNA DSB end-joining in DES cells vs. VEH control cells (Figure 6) suggests that MSCs from adult Eker rats briefly exposed to DES in early life will permanently exhibit impaired DNA DSB repair capacity.

Figure 6.

DNA end-joining efficiency is decreased in DES vs. VEH MSCs. Luciferase gene reporter assay detected lesser DNA end-joining activity in DES vs. VEH MSCs following co-transfection with linearized HindIIIpGL2-control luc and CMV-β-gal plasmids. Means of combined experimental and biological replicates were compared using a two-tailed Student unpaired t-test. Lines represent the mean normalized end-joining activity ± SD. *P < 0.0001

Discussion

Although a variety of risk factors have been linked with the development of UF in humans, specifically low parity, obesity, vitamin D deficiency, and family history, ethnicity plays a strong role in UF prevalence in the United States, as UFs occur four times more commonly in African-American women as compared to Caucasian women [53–56]. African-American women typically experience not only greater tumor burden (number and size of tumors), but also more severe symptoms [54]. Additional epidemiological studies have demonstrated that early-life environmental exposure to EDCs is strongly associated with UF development later on in adult life [57], such as in utero exposure of female fetuses to DES from 1940 to 1971 that led to severe reproductive system abnormalities later in life, including structural reproductive tract aberrations, increased rates of infertility, poor pregnancy outcomes, increased risk of vaginal clear cell adenocarcinoma, and increased prevalence of UF [58–62]. Additionally, it has been shown through the Study of Environment, Lifestyle & Fibroids (SELF), in which women underwent baseline ultrasound screening for fibroids, that those who were fed soy formula in infancy had larger fibroids and increased total tumor volume than soy formula-unexposed women [63]. Importantly, social epidemiology literature supports the higher tendency of minority populations to live in and be exposed to disproportionate levels of environmental pollution including EDCs [64–66]. Thus, the compelling question arises: Does the potential exposure of African-American women to various EDCs found in the environment, e.g. bisphenol A, genistein (a phytoestrogen in soy products), and phthalates, in early fetal life (during sensitive periods of uterine development) contribute to the higher prevalence of UF in women of color later in adult life?

Not only human epidemiology correlates these early-life exposure incidents with adult-onset UF disease, but experimental animal studies have also provided direct evidence that early-life exposure to EDCs, including xenoestrogens (e.g. DES, genistein), induces permanent reprogramming of the epigenome and, more specifically in the reproductive tract, anatomic abnormalities, and neoplastic diseases, including UFs [2, 35, 62, 67]. The Eker rat model has been invaluable for studying gene–environment interaction in tumorigenesis [2, 21, 34, 35]. We have recently demonstrated that early-life exposure of Eker pups to DES leads to significant, long-lasting changes, such as expanding the number and altering the characteristics (e.g. proliferation rate) of the MSC population [21]. These alterations in MSCs have been implicated in the increased prevalence (100%) of UF formation in these early-life exposed Eker rats, though the mechanism of their initial transformation into UF tumor-forming cells remains to be elicited [21]. Recent studies of stem cell distribution demonstrated that with increased numbers of tissue-specific stem cells in highly proliferative tissues, which undergo equally high numbers of cell divisions, the risk of cancer development in these tissues also increases [8, 23, 24, 36, 68]. An increase in overall stem cell proliferation suggests increased numbers of DNA DSBs, events that, while normal and easily fixed in a healthy, replicating cell, can wreak havoc in developmentally reprogrammed stem cells, such as these adult MSCs exposed in early life to EDCs. Other studies have shown a connection between environmental conditions and altered estrogen receptor signaling and accumulation of DNA damage, decreased expression of genes and proteins critical to DNA repair, and a shift toward pro-proliferative signaling in both human patients and animal models [69–78].

As we have shown, myometrial tissues from adult Eker rats exposed during early-life uterine development (to DES) demonstrate decreased expression of BRCA2 and RAD51, critical DNA DSB HR proteins, and the MSCs isolated from these tissues showed an increased level of DNA damage, an altered baseline expression of genes and proteins critical for detection and repair of DNA DSBs, and a decreased functional capacity for efficiently repairing DNA DSBs once they occur, as compared to their unexposed counterparts (VEH). As shown by whole genome RNA sequencing and qRT-PCR confirmation, a strongly significant difference exists in DES cells in the expression levels of upregulated genes relating to motor activity (Myh1, Myh4, Myh8), cell cycle regulation, specifically the G1/S transition, and DNA DSB repair (Plk3, Cdkn1a), methyltransferase activity (Mgmt), and translation regulation (Tyms) (Figure 1). Mechanistically, Cdkn1a, whose expression is tightly regulated by DNA damage-response gene, p53, mediates the cell cycle G1 phase arrest in response to stressful stimuli and is shown to be upregulated following DNA damage [79, 80]. In addition, more genes were downregulated in DES cells, including those related to cell cycle regulation and stress response and, more importantly, DNA DSB-repair pathways, specifically those relating to NHEJ and HR, Polq, Nhej1, and Gstp1; and Bard1, Brca2, Cdk1, Fancd2, Palb2, Rad51ap1, Rad52, and Xrcc2, whose protein products are essential in the DDR and DSB repair. The HR pathway in particular is thought to be central in mediating genomic instability; thus, null mutations in many of these genes have been linked with increased risk of genomic instability and earlier onset of cancers [81–84]. In addition, even following 24 h of recovery, DES stem cells demonstrated increased phosphorylation of ATM, which is rapidly autophosphorylated in response to DNA DSBs to phosphorylate and activate downstream signaling and effector molecules; this includes histone 2A, variant X (γ-H2AX) which serves as a marker of DNA DSBs, aggregating to form discrete foci that recruit additional repair proteins, and CHK1, a kinase involved of further downstream effector activation/regulation, including mediators of cell cycle arrest and apoptosis (e.g. p53) [85–88]. In addition, either ATM activation can lead to phosphorylation and activation of Nibrin, a member of the MRN (MRE11, RAD50, Nibrin/p95/NBS1) complex that initiates downstream signaling, or MRN complex activation can result independently of ATM phosphorylation [83, 84, 87, 89, 90]. Although ATM and Nibrin activation may indicate intact DDR initiation, the prolonged presence of γ-H2AX and CHK1 activation and the downregulation of several downstream DNA DSB repair components suggest delayed DNA repair in the exposed DES stem cells. Our data collectively suggest that these genes’ dysregulation in DES-exposed adult MSCs may be implicated in the genomic instability that may lead to earlier and more frequent loss of the second, normal Tsc2 allele, leading to earlier, more severe, and more penetrant occurrence of UF in the developmentally reprogrammed myometrial tissue of exposed rats [91, 92].

Having established that these differences in DNA repair-related molecules exists, it was important to understand if the DNA DSB-related repair systems may also be altered in their functional capacity in DES stem cells. We demonstrated (Figure 6) that the functional capacity for DES cells to repair DNA DSBs was dramatically diminished relative to that of VEH cells. When developmentally reprogrammed adult DES stem cells were transfected with equal amounts of linearized HindIII/pGL2-control luc plasmid vector, they exhibited only about one-tenth the capacity for DNA double-strand end-joining as normal, unexposed VEH stem cells did (Figure 6). Moreover, when DES and VEH stem cells were briefly treated with BLM, a tool compound used as a radiomimetic for inducing DNA damage via DNA DSBs, they both exhibited the expected increase in DNA DSBs, as shown by increased integration and intensity of γ-H2AX; following rapid phosphorylation of histone H2AX by kinases, like ATM, or by the MRN complex, γ-H2AX foci both immediately label DNA damage/DNA DSBs and induce cellular repair machinery [93, 94]. Importantly, however, while both cell types demonstrated decreasing % γ-H2AX foci-positive nuclei and integration of γ-H2AX foci per nuclear area given time to recover and repair DNA DSBs, after 6 h, DES stem cells still showed no significant decrease in γ-H2AX foci per nuclear area, while that in VEH stem cells decreased significantly from that after 30 min of recovery (Figure 5). In addition, though both cell types continued to recover after 24 h, there remained a significantly increased % γ-H2AX foci-positive nuclei in DES MSCs, while that of VEH MSCs had decreased significantly as compared to the 6 h time point P = 0.02 (Figure 5). This suggests that DES cells may repair DNA DSBs but require a longer period of time to do so. It has been demonstrated using γ-H2AX foci as a marker of DNA DSBs (DNA damage) that the rate of disappearance of γ-H2AX foci correlates directly with the rate of DNA repair, and increased residual γ-H2AX foci following induction of DNA DSBs have been correlated with increased numbers of chromosome aberrations [93–95].

A proposed limitation of this study is the translational applicability between this animal model, which harbors a heterozygous mutation in Tsc2, and the human female, who likely does not house a similar mutation. Most importantly, it should be noted that any causal effects between the Tsc2 germline mutation and alterations in the DNA repair system may be disregarded in this particular work, as this study focuses on MSCs isolated from animals harboring the heterozygous germline mutation in Tsc2 (Tsc2^Ek/+), both in animals developmentally exposed to DES and in those developmentally unexposed (VEH), prior to eventual loss of heterozygosity (LOH). It has been demonstrated in both human and animal studies that LOH is required for cellular consequences, i.e. tumorigenesis to occur [96–98]. Moreover, only one study has shown any relationship between a specific variant of DNA repair enzyme, 8-oxoguanine glycosylase 1 (OGG1), and increased risk of angiomyolipoma tumor development in tuberous sclerosis (TSC) human patients [99]. In addition, most second-hit mutations leading to LOH observed in the Tsc2 gene of the Eker rat model have been shown primarily to be related to aberrant expression of genes relating to cell proliferation and adhesion pathways, including those regulated by the PI3K/Akt and mTOR, suggesting changes in these pathways may explain Tsc2 involvement in tumorigenicity via deregulation of protein synthesis and cell growth [100, 101]. Thus, the effects of early-life DES exposure on DNA repair systems of the MSC populations may be studied in exposed vs. unexposed MSCs from animals harboring the germline heterozygous Tsc2 mutation.

Uterine fibroid-causing MED12 mutations are emerging as the most common mutations associated with sporadic UFs as they are detected in up to 85% of such lesions [5, 6]. Remarkably, in the same uterus, separate UF lesions may harbor different MED12 mutations [5, 6, 25, 26]. This strongly suggests that each UF lesion arises separately after an independent mutagenic event occurs in a perturbed MSC, changing it to a tumor-forming stem cell. As it is very common to have multiple UF lesions in same uterus, sometimes up to hundreds of small UF lesions (miliary leiomyomatosis), this suggests that myometrium from individuals at risk for UF development is in a state of heightened mutagenesis. Indeed, we have in our lab compared DNA repair in such myometrium (MyoF) with age-matched healthy fibroid-free myometrium (MyoN). Interestingly, MyoF demonstrated downregulation of DNA repair-related gene, p21, and upregulation of Enhancer of zeste homolog 2 (EZH2) [21], a related transcription repressor, suggesting that the gene–environment interactions observed in the Eker rat may indeed assist in understanding the pathogenesis of human UFs (Supplemental Figure S4). Altogether, these data imply that such a brief, temporary exposure to an EDC like DES, during the sensitive window of uterine development, not only alters expression of genes and phosphorylation of protein products important in DNA repair, but may also permanently alter the normal functional DNA repair capacity of the MSC population, by hindering their ability to repair DNA DSBs leading to increased predisposition to neoplastic development in adulthood (Supplemental Figure S4). We have preliminary data showing that DES MSCs demonstrate increased responsiveness to estrogen, as shown by increased expression of estrogen-responsive genes in DES MSCs as compared to VEH-exposed MSCs [102]. Because estrogen signaling has been implicated in impairing the DNA DSB DDR, specifically by impairing repair and apoptosis in favor of proliferation, additional studies are needed to investigate the mechanisms through which estrogen signaling in these MSCs is putatively altered, ultimately leading to defects in DNA DSB repair [103].

Supplementary data

Supplemental Figure S1. pGL2-control and pGL2-basic plasmid maps for plasmid vector transfection. (A) pGL2-control vector contains both SV40 promoter and SV40 enhancer of luc reporter gene and contains HindIII restriction site at bp 239. (B) pGL2-basic vector contains neither SV40 promoter nor SV40 enhancer.

Supplemental Figure S2. Immunohistochemistry (IHC) of DES- and VEH-exposed ex vivo myometrial tissues demonstrates differential RAD51 and BRCA2 expression. (A) IHC staining of BRCA2 in DES- and VEH-exposed myometrial tissues. (B) Quantification of negative, weakly positive, positive, and strongly positive staining of BRCA2 in DES- and VEH-exposed myometrial tissues. (C) IHC staining of RAD51 in DES- and VEH-exposed myometrial tissues. (D) Quantification of negative, weakly positive, positive, and strongly positive staining of RAD51 in DES- and VEH-exposed myometrial tissues.

Supplemental Figure S3. Transfection efficiency does not differ between DES and VEH MSCs. No significant difference in transfection efficiency exists between DES vs. VEH as demonstrated by β-galactosidase reporter plasmid enzyme activity.

Supplemental Figure S4. Model of hypothesis of pathogenesis of uterine fibroid development in the murine (Eker rat) and human reproductive tract as a result of early-life exposure to EDCs. Early-life exposure to EDCs (e.g. DES) reprograms the epigenome of MSCs, altering the expression and function of DNA repair-related genes/proteins, ultimately transforming normal MSCs into tumor-forming MSCs and increasing UF tumor penetrance in DES-exposed rats (and respective “exposed” human myometrium) later in adult life.

Supplemental Table S1. List of genes shown by RNA Seq to be most up- and downregulated—used for gene expression validation in rat.

Supplemental Table S2. List of primer sequences used for gene expression validation in rat.

Supplemental Table S3. Antibody table.

Supplemental Table S4. Relative gene expression data from RNA-Seq of DES relative to VEH control myometrial stem cell expression. Column 1: Gene identifier. Column 2: base mean. Column 3: log2foldchange DES normalized to VEH. Column 4: logfoldchange standard error (lfcSE). Column 5: stat. Column 6: P-value Column 7: Adjusted p-value (padj). Column 8: Fold change (calculated from Column 3). Column 9: Gene category.

Notes

Gene expression data was deposited into the Gene Expression Omnibus (GEO): GSE106655.

Edited by Dr. Haibin Wang, PhD, Xiamen University

FUNDING

This work was supported by the National Institutes of Health: National Institute of Environmental Health Sciences grant R01 ES028615 (AA) and the National Institute of Child Health and Human Development grant F30 HD089585 (LPF); the Augusta University (AU) Start-up Package; and the AU Intramural Grants Program (QY).