Constitutive Activation of Transforming Growth Factor Beta Receptor 1 in the Mouse Uterus Impairs Uterine Morphology and Function

ABSTRACT

Despite increasing evidence pointing to the essential involvement of the transforming growth factor beta (TGFB) superfamily in reproduction, a definitive role of TGFB signaling in the uterus remains to be unveiled. In this study, we generated a gain-of-function mouse model harboring a constitutively active (CA) TGFB receptor 1 (TGFBR1), the expression of which was conditionally induced by the progesterone receptor (Pgr)-Cre recombinase. Overactivation of TGFB signaling was verified by enhanced phosphorylation of SMAD2 and increased expression of TGFB target genes in the uterus. TGFBR1 Pgr-Cre CA mice were sterile. Histological, cellular, and molecular analyses demonstrated that constitutive activation of TGFBR1 in the mouse uterus promoted formation of hypermuscled uteri. Accompanying this phenotype was the upregulation of a battery of smooth muscle genes in the uterus. Furthermore, TGFB ligands activated SMAD2/3 and stimulated the expression of a smooth muscle maker gene, alpha smooth muscle actin (ACTA2), in human uterine smooth muscle cells. Immunofluorescence microscopy identified a marked reduction of uterine glands in TGFBR1 Pgr-Cre CA mice within the endometrial compartment that contained myofibroblast-like cells. Thus, constitutive activation of TGFBR1 in the mouse uterus caused defects in uterine morphology and function, as evidenced by abnormal myometrial structure, dramatically reduced uterine glands, and impaired uterine decidualization. These results underscore the importance of a precisely controlled TGFB signaling system in establishing a uterine microenvironment conducive to normal development and function.

INTRODUCTION

Transforming growth factor beta (TGFB) superfamily signaling plays a pleiotropic role in fundamental cellular and developmental processes. TGFB superfamily ligands (e.g., TGFBs, activins, and bone morphogenetic proteins [BMPs]) interact with their membrane-bound type 2 and type 1 receptors to form a heteromeric complex. Subsequent phosphorylation of the type 1 receptor at the glycine and serine (GS) domain by the constitutively active type 2 receptor activates receptor-regulated intracellular SMAD proteins, which modulate gene transcription in concert with the common SMAD (i.e., SMAD4), coactivators, and corepressors [1, 2].

Signaling activity of the TGFB superfamily is precisely controlled under normal physiological conditions. Multiple regulatory factors, including ligand traps (e.g., follistatin), ligand activators (e.g., tenascin-X), inhibitory SMADs (i.e., SMAD6 and SMAD7), and agonistic/antagonistic pathways may cooperate to govern the normal activity and function of this pathway [2–8]. Accumulating evidence indicates that TGFB superfamily members are key regulators of female reproduction, including, but not limited to, follicular development, ovulation, oocyte-cumulus cell communications, uterine decidualization, and embryo development [9–21].

TGFB ligands (i.e., TGFBs 1–3) are founding members of the TGFB superfamily. TGFBs signal via TGFB receptor 1 (TGFBR1/ALK5) and receptor 2 (TGFBR2) and downstream SMAD2/3 proteins. Identification of the in vivo function of TGFB signaling in the uterus remains a challenging puzzle, partially because of the potential redundancy of the ligands [22, 23] and the lack of appropriate animal models. TGFB signaling components, including TGFB ligands, receptors, and SMADs, are expressed in the mouse and human myometrium and regulate DNA synthesis of human myometrial cells [24–26]. In the rat uterus, myometrial expression of TGFB1 and TGFB3 is increased from midgestation, with TGFB3 strongly localized to the circular myometrial layer at late pregnancy [27]. Interestingly, TGFB3 levels are higher in human leiomyoma cells versus myometrial cells, and TGFB3 is expected to promote leiomyoma development via stimulating cell growth and fibrogenic process [28]. By taking advantage of a conditional knockout approach, we have shown that ablation of Tgfbr1 in the female reproductive tract using anti-Müllerian hormone receptor type 2 (Amhr2)-Cre recombinase leads to smooth muscle defects and reproductive failure [18, 29], suggesting an essential role of TGFB signaling in myometrial development.

That TGFB signaling is finely tuned argues for the need to use both loss-of-function and gain-of-function approaches to fully understand its physiologic and pathologic roles. Constitutively active receptors can be used to investigate the impact of sustained elevation of a signaling pathway on the pathogenesis of diseases. To our knowledge, mouse models with enhanced TGFB signaling in the female reproductive tract are lacking. Notably, overactivation of TGFB signaling is linked to the development of diseases, including cancer [30–32]. Therefore, in the present study, we created a mouse model harboring a constitutively active TGFBR1 in the uterus for which the expression is conditionally induced by the progesterone receptor (Pgr)-Cre. Overactivation of TGFB signaling causes infertility and striking phenotypic alterations in the uteri of these mice. Our results highlight the importance of a precisely controlled TGFB signaling system in establishing a uterine microenvironment conducive to normal development and function.

MATERIALS AND METHODS

Animals and Treatment

All protocols using laboratory mice were approved by the Institutional Animal Care and Use Committee at Texas A&M University. Mice were maintained on a mixed C57BL/6/129SvEv genetic background. Mice were exposed to a 12L:12D photoperiod with access to food and water ad libitum during the entire experimental period. The Pgr-Cre mice were created as described previously [33]. Mice harboring a constitutively active TGFBR1 were generated earlier according to strategies, including genetic modifications, described elsewhere [34]. Briefly, constitutive activation of the receptor in the absence of ligand results from three missense mutations: T204D that constitutively activates the TGFBR1 kinase [35] and L193A/P194A that prevent binding of the TGFBR1 inhibitor, FKBP12 [36]. Mice harboring the Rosa26-LacZ allele (Gt [ROSA] 26Sortm1Sor/J) [37] were purchased from the Jackson Laboratory and used as a reporter for Pgr-Cre expression. Fertility testing was performed by breeding the control and experimental female mice with proven fertile males for a period of 3 mo. Superovulation of immature females was performed as described [16]. Analysis of uterine decidualization was conducted as reported previously [18, 38]. Briefly, ovariectomized mice received subcutaneous injections of estradiol (E2; 100 ng/mouse for 3 days; Sigma). The mice were rested for 2 days and then treated daily with E2 (6.7 ng/mouse) and progesterone (P4; 1.0 mg/mouse; Sigma). After 2 days of daily injection of E2 and P4, one uterine horn was traumatized using a burred needle, whereas the other served as an unstimulated control. Two days after the decidual stimulus, mice were euthanized and uterine samples collected. The uterine horns were weighed, and total RNA was prepared for real-time PCR analysis as described below. Alkaline phosphatase staining was performed to examine uterine stromal cell differentiation as described previously [39].

Mouse Breeding, Genotyping, and DNA Recombination Analysis

To generate mice with conditionally activated TGFBR1, mice containing a latent constitutively active TGFBR1 were bred with Pgr-Cre mice. Genomic DNA was isolated from mouse tails using NaOH buffer. Genotyping and DNA recombination analyses of the conditional allele were carried out using PCR [34, 40, 41]. Primers used for genotyping of Pgr-Cre have been described elsewhere [19]. PCR products were separated on 1% agarose gels containing ethidium bromide and digital images captured using a VWR Gel Imager.

Histological Analysis

Uterine and ovarian samples were collected from control and experimental groups at defined ages (1 wk or 1–3 mo) and fixed in neutral buffered formalin. The histology core facility of the Department of Veterinary Integrative Biosciences at Texas A&M University was used for sample processing and embedding. Paraffin sections (thickness, 5 μm) were serially generated, and hematoxylin-and-eosin staining was conducted using a standard protocol. After staining, slides were mounted and examined with a microscope, and images were captured using a digital camera (DP25; Olympus) with cellSens Digital Imaging Software (Olympus).

X-Gal Staining

Uteri from mice harboring Rosa26-LacZ and Pgr-Cre were collected and fixed as described previously [18]. The samples were washed and stained using staining buffer containing 1 mg/ml of X-gal, 5 mM potassium ferricyanide, and 5 mM potassium ferrocyanide (Sigma). Samples were then sequentially processed for postfixation, paraffin embedding, sectioning, and fast red counterstaining [18].

Human Uterine Smooth Muscle Cell Line Culture and Treatment

A well-characterized human myometrial cell line, PHM1-41 [42–44], was used to determine the effect of TGFB ligands on alpha smooth muscle actin (ACTA2) expression in uterine smooth muscle cells. Briefly, PHM1-41 cells were cultured in Dulbecco modified Eagle medium supplemented with 10% fetal bovine serum, 100 U/ml penicillin, 0.1 mg/ml streptomycin, and 0.1 mg/ml of G418 sulfate. Before the experiment, G418 was omitted. The cells were serum-starved overnight before treatment. For the SMAD2/3 phosphorylation study, the cells were treated with TGFB1 (2.5 ng/ml; R&D Systems) and collected after 40 min of treatment. The cells were then processed for immunofluorescence as described previously [29]. Briefly, the cells were fixed with 4% paraformaldehyde in cold PBS for 15 min and permeabilized with 0.1% Triton X-100 for another 15 min. The cells were then blocked with 10% normal goat serum and incubated with rabbit anti-SMAD2/3 antibody (1:100; Cell Signaling) overnight at 4°C. After washing, the cells were incubated with Alexa Fluor 594 conjugated anti-rabbit immunoglobulin (Ig) G (Invitrogen) for 1 h. The nuclei were labeled with 4′,6-diamidino-2-phenylindole (DAPI) using ProLong Gold Antifade Reagent (Invitrogen). For gene expression analysis, the cells were treated with 2.5 ng/ml of TGFB1, TGFB2 (HumanZyme), and TGFB3 (HumanZyme). Selection of the dosage for TGFBs was based on a pilot experiment. Cells were collected after 20 h of treatment and processed for RNA isolation and real-time PCR analysis.

Immunofluorescence Microscopy

Immunofluorescence was performed using serial paraffin sections [18]. Antigen retrieval was performed to expose the hidden antigenic sites by boiling the sections for 20 min in 10 mM citrate buffer (pH 6.0) using a microwave oven. Following antigen retrieval, the sections were blocked with 5% bovine serum albumin (Sigma) and incubated with the following primary antibodies overnight at 4°C in a humid box: mouse anti-ACTA2 (1:2000; Abcam), rat anti-cytokeratin 8 (KRT8; 1:100; Developmental Studies Hybridoma Bank), rabbit anti-vimentin (1:200; Cell Signaling), and goat anti-3-beta-hydroxysteroid dehydrogenase (HSD3B; 1:1000; Santa Cruz). The slides were next incubated with secondary antibodies conjugated with Alexa Fluor 488 or 594 (Invitrogen) at room temperature for 1 h, and the sections were mounted using ProLong Gold Slowfade media with DAPI (Invitrogen). The slides were examined using an IX73 microscope (Olympus) interfaced with an XM10 CCD camera (Olympus) and cellSens Digital Imaging Software. To determine the background levels of the above assays, controls in which primary antibodies were replaced by isotype-matched IgGs purified from the same species as the primary antibodies were included.

Western Blot Analysis

Protein lysates were prepared using radioimmunoprecipitation assay buffer [16] containing proteinase and phosphatase inhibitors (Roche) and quantified using a BCA Protein Assay Kit (Thermo Scientific). Protein samples (∼30 μg) were separated on 10% Tris gel, transferred to polyvinylidene difluoride membranes (Bio-Rad), and incubated overnight at 4°C with the following primary antibodies: mouse anti-hemagglutinin (HA; 1:200; Santa Cruz), rabbit anti-phospho-SMAD2 (1:500; Millipore); rabbit anti-SMAD2 (1:1000; Cell Signaling), and mouse anti-beta actin (ACTB; 1:50 000 Sigma). Then, the membranes were washed and further incubated with horseradish peroxidase (HRP)-conjugated donkey anti-rabbit antibody (1:20 000; Jackson ImmunoResearch) at room temperature for 1 h. The signals were developed with Immobilon Western Chemiluminescent HRP Substrate (Millipore) and scanned using a Kodak Image Station 4000 mm PRO. ACTB was used as an internal control. ImageJ software (version 1.47; National Institutes of Health) was used for Western blot quantification, in which the intensity of target bands was normalized to that of ACTB to correct for variations in sample loading.

RNA Isolation and Real-Time PCR

Mouse uterine tissues were homogenized and total RNA isolated using an RNeasy Mini Kit (Qiagen) according to the manufacturer's instruction. On column DNase digestion was performed during the procedure to eliminate potential DNA contamination. Total RNA was quantified using a NanoDrop Spectrophotometer ND 1000 (NanoDrop Technologies). Two hundred nanograms of total RNA were reverse transcribed to generate cDNA [29]. Then, real-time PCR was performed using CFX Connect Real-time PCR Detection System (Bio-Rad) and iTaq Universal SYBR Green Supermix (Bio-Rad) [29].

Quantification of matrix metalloproteinase 9 (Mmp9), Acta2, calponin 1 (Cnn1), desmin (Des), smoothelin (Smtn), transgelin (Tagln), smooth muscle actin gamma (Actg2), myosin heavy chain 11 (Myh11), Bmp2, follistatin (Fst), and ACTA2 was conducted as described previously [29, 45, 46]. Primers used were as follows: serine (or cysteine) peptidase inhibitor clade E member 1 (Serpine1; ID 6679373a1): forward, 5′-TTCAGCCCTTGCTTGCCTC-3′; reverse, 5′-ACACTTTTACTCCGAAGTCGGT-3′; connective tissue growth factor (Ctgf; ID 6753878a1): forward, 5′-GGGCCTCTTCTGCGATTTC-3′; reverse, 5′-ATCCAGGCAAGTGCATTGGTA-3′; Bmp2 (ID 6680794a1): forward, 5′-GGGACCCGCTGTCTTCTAGT-3′; reverse, 5′-TCAACTCAAATTCGCTGAGGAC-3; Fst (ID 6679867a1): forward, 5′-TGCTGCTACTCTGCCAGTTC-3′; reverse: 5′-GTGCTGCAACACTCTTCCTTG-3′; and TGFB-induced factor homeobox 1 (TGIF1): forward, 5′-GGGATCAGTTTTGGCTCGTCC-3′; reverse, 5′-GCAGTCACAGTGGTATGGCAG-3′. Serpine1, Ctgf, Bmp2, and Fst primers were obtained from PrimerBank [47]. Rpl19 [29] and RPLP0/36B4 [48] were used as internal controls for the respective mouse and human gene expression analyses. The assays were performed in duplicate for each biological replicate. Relative levels of gene expression were calculated as described [49].

Statistical Analysis

Statistical analysis was performed using Statistical Package for the Social Sciences (version 21; IBM). A one-way ANOVA was applied to determine the difference of means among treatment groups, followed by a Bonferroni post-hoc test. Comparisons of mean values from two groups were made using Student t-test. Data are presented as the mean ± SEM. P values less than 0.05 were defined as statistically significant.

RESULTS

Generation of Mice Harboring a Constitutively Active TGFBR1 in the Uterus

To create a gain-of-function model of TGFBR1 in the mouse uterus, we took advantage of a latent constitutively active TGFBR1 allele (TGFBR1^CA), in which a constitutively active TGFBR1 was knocked into the hypoxanthine guanine phosphoribosyl transferase (Hprt) locus (Fig. 1A) [34]. We created mice harboring the TGFBR1^CA for which the expression was conditionally induced by Pgr-Cre recombinase (TGFBR1 Pgr-Cre CA) (Fig. 1, B and C). Pgr-Cre is expressed in the uterus [50] and has been extensively utilized as a Cre deletor for genes expressed in the uterus [19, 51–53]. By crossing the Pgr-Cre mice with Rosa26 LacZ reporter mice [37], we verified the expression of Pgr-Cre in the uterus (data not shown). Pgr-Cre recombinase can remove the “stop” sequence of the constitutively active TGFBR1 allele in the uterus, leading to the expression of the constitutively active TGFBR1.

FIG. 1.

Generation of mice containing a constitutively active TGFBR1 in the uterus. A) Schematic representation of the latent constitutively active TGFBR1 allele. Pt, Hprt promoter; HA, hemagglutinin tag; CAG, composite constitutive CAG (human cytomegalovirus enhancer and chicken beta actin); Hprt, hypoxanthine-guanine phosphoribosyl transferase. B) The TGFBR1^{CA Lox/Lox} mice were bred with Pgr-Cre mice to produce control mice (TGFBR1^{CA Lox/Lox}; Ctrl) and experimental mice (TGFBR1^{CA Lox/Lox}; Pgr-Cre; TGFBR1 Pgr-Cre CA). C) Illustration of genotyping PCR of female pups. Lanes 1–3 represent WT, TGFBR1^{CA Lox/Lox}, and TGFBR1^{CA Lox/Lox}; Pgr-Cre, respectively.

To validate this model, we first determined whether the TGFBR1^CA allele could be recombined at the genomic DNA level. Using specific recombination primers for TGFBR1^CA [40] and uterine DNA isolated from control and TGFBR1 Pgr-Cre CA mice, a band with the expected size was detected in the uterus of TGFBR1 Pgr-Cre CA mice but not controls (Fig. 2A). Next, we performed quantitative real-time PCR and demonstrated that TGFBR1^CA mRNA level was significantly higher in the uterus of TGFBR1 Pgr-Cre CA mice compared with controls (Fig. 2B). Using an antibody directed against HA, we further showed that TGFBR1^CA-HA fusion protein expression was induced in the TGFBR1 Pgr-Cre CA uterus (Fig. 2C) and correlated with increased phospho-SMAD2, an indicator of TGFB signaling activity (Fig. 2D). Consistent with enhanced TGFB signaling activity in the uterus, mRNA levels of TGFB targets, including Serpine1 (Fig. 2E), Ctgf (Fig. 2F), and Mmp9 (Fig. 2G) were upregulated in the uterus of TGFBR1 Pgr-Cre CA mice at the age of 1 wk, a critical time point for early postnatal uterine development. Therefore, we created a mouse model with enhanced TGFB signaling in the uterus.

FIG. 2.

Constitutively active TGFBR1 enhances TGFB signaling in the mouse uterus. A) DNA recombination. Rec, recombination. B) TGFBR1^CA mRNA levels were significantly higher in mice containing a constitutively active TGFBR1. Real-time PCR was performed using uteri samples from control (Ctrl) and TGFBR1^{CA Lox/Lox}; Pgr-Cre (CA) mice at the age of 1 wk and 1 mo (n = 4). Note that the primers amplify the transgene. C and D) Western blot analysis of TGFBR1-HA fusion protein, phospho-SMAD2, and SMAD2 in the uteri of control and TGFBR1^{CA Lox/Lox}; Pgr-Cre mice. Upper panels show the representative Western blot images, whereas lower panels depict the results of quantification for TGFBR1CA (C) and phospho-SMAD2 (D). Each lane represents an independent sample. Uterine samples were collected from mice at the age of 6 wk (n = 3–5). E–G) Increased mRNA expression of Serpine1, Ctgf, and Mmp9 in the uteri of TGFBR1^{CA Lox/Lox}; Pgr-Cre mice compared with controls. Uterine samples were collected from mice at 1 wk of age (n = 4). Data are presented as the mean ± SEM. *P < 0.05 versus controls.

Mice Harboring a Constitutively Active TGFBR1 in the Uterus are Infertile and Develop Myometrial and Glandular Defects

During a 3-mo fertility test period, the TGFBR1 Pgr-Cre CA mice were sterile compared with controls (Table 1). To determine the potential causes of the infertility and investigate the effect of TGFB signaling activation in the uterus, we first performed immunofluorescence microscopy using antibodies directed to ACTA2, a smooth muscle marker. Hypermuscled uteri were identified in sexually immature TGFBR1 Pgr-Cre CA mice (Fig. 3, D and F) compared with controls (Fig. 3, A and C). Immunofluorescence examination of KRT8, an epithelial marker, was performed to evaluate the integrity of uterine epithelia in the TGFBR1 overactivation mice. Interestingly, a marked reduction of uterine glands in TGFBR1^{CA Lox/Lox}; Pgr-Cre mice was detected (Fig. 3, B, C, E, and F).

TABLE 1.

Fertility test of TGFBR1 Pgr-Cre CA and control mice.*

^*The fertility test was conducted during a 3-mo period. Data are presented as the mean ± SEM.

FIG. 3.

Constitutively active TGFBR1 in the mouse uterus causes myometrial and glandular defects. A–F) Immunofluorescence of ACTA2 (red) and KRT8 (green) using uterine samples from control (Ctrl) and TGFBR1^{CA Lox/Lox}; Pgr-Cre (CA) mice. Note the increased thickness of myometrium (D and F), reduced uterine glands (E and F), and myofibroblast-like cells within stroma (D and F) in TGFBR1 Pgr-Cre CA mice compared with age-matched controls (A–C). Uteri samples from 1-mo-old control and TGFBR1^{CA Lox/Lox}; Pgr-Cre mice (n = 3) were analyzed, and representative images are shown. G–R) Immunofluorescence of ACTA2 in the uteri of control and TGFBR1^{CA Lox/Lox}; Pgr-Cre mice. Note the disorganized smooth muscle cell orientation (green; H and K) and the presence of muscle structures with longitudinal orientation within the circular myometrial layer region (I and L) compared with controls (G and J). Less pronounced alteration of longitudinal muscle structure was found in TGFBR1^{CA Lox/Lox}; Pgr-Cre mice (N and Q) versus controls (M and P). O and R) Representative negative controls in which the primary antibody for ACTA2 was replaced with mouse IgG. LE, luminal epithelium; GE, glandular epithelium; Myo, myometrium; Neg, negative controls. Bar = 100 μm (A–F) and 25 μm (G–R).

To determine whether these morphological alterations were preserved in older mice, circular and longitudinal layers were examined in adult mice at the age of 2–3 mo (Fig. 3, G–R, and Supplemental Fig. S1 [available online at www.biolreprod.org]). Whereas organized myometrial layers were evident in control mice (Fig. 3, G and J, and Supplemental Fig. S1, A and B), an increased thickness of the myometrium and a disorientation of inner circular layer of myometrium with gaps between smooth muscle bundles were observed in the TGFBR1 Pgr-Cre CA mice (Fig. 3, H and K, and Supplemental Fig. S1, C and D). Longitudinal orientation of muscle structures could also be found within the circular layer region in TGFBR1 Pgr-Cre CA mice (Fig. 3, I and L). In contrast to the circular myometrium, defects in the longitudinal layer were less pronounced at the examined time stages (Fig. 3, M, N, P, and Q). Altogether, these data showed that constitutive activation of TGFBR1 in the mouse uterus promoted the development of hypermuscled uteri with a disorganized myometrial structure.

We next examined whether development of the endometrium was altered in TGFBR1 Pgr-Cre CA mice. In contrast to controls (Fig. 4, A and E), the endometrium of the TGFBR1^{CA Lox/Lox}; Pgr-Cre mice contained abundant ACTA2-positive cells (Fig. 4, B and F), suggesting a potential increase in the myofibroblast population. These ACTA2-positive cells were negative for vimentin, as demonstrated by double immunofluorescence of ACTA2 and vimentin (Fig. 4, F and H), the latter of which was predominantly expressed in normal uterine fibroblasts (Fig. 4, C, E, and G). Therefore, overactivation of TGFB signaling appears to promote myofibroblast differentiation in the endometrium of the uterus.

FIG. 4.

Colocalization of ACTA2 and vimentin in the endometrium. A–H) Representative images of immunofluorescence for ACTA2 and vimentin using uterine samples from control and TGFBR1^{CA Lox/Lox}; Pgr-Cre mice. Note the presence of abundant ACTA2-positive cells (green) in the stroma (B), and these cells did not express vimentin (red; F and H). G and H are higher-power images for E and F, respectively. DAPI (blue) was used to counterstain the nucleus. Arrowheads in B and H show the presence of myofibroblast-like cells in the endometrium. VIM, vimentin. Bar = 20 μm (A–F) and 10 μm (G and H).

Increased Smooth Muscle Gene Expression in the TGFBR1 Constitutively Active Uterus

Along with the hypermuscled uterine phenotype, the expression of a battery of smooth muscle genes, including Acta2, Cnn1, Des, Smtn, Tagln, Actg2, and Myh11, was upregulated in the uteri of TGFBR1 Pgr-Cre CA mice at 1 wk of age compared with age-matched controls (Fig. 5A). To independently test whether TGFB signaling enhanced smooth muscle gene expression in myometrial cells in vitro, we utilized of a well-characterized uterine smooth muscle cell line, PHM1. TGFB1 activated SMAD2/3 in PHM1 cells, as evidenced by the nuclear accumulation of SMAD2/3 in TGFB1-treated cells versus vehicle controls (Fig. 5B). The addition of multiple TGFB isoforms stimulated the expression of ACTA2 (Fig. 5C). TGIF1, a known TGFB-induced gene in human myometrial cells [54], was included as a positive control (Fig. 5D). These data suggest that TGFB signaling promotes uterine smooth muscle gene expression in myometrial cells.

FIG. 5.

TGFB signaling promotes uterine smooth muscle gene expression. A) Increased expression of smooth muscle genes in the uteri of TGFBR1 Pgr-Cre CA mice at 1 wk of age compared with age-matched controls (n = 4). The dotted line marks the gene expression level of the controls. Data are presented as the mean ± SEM. *P < 0.05 versus corresponding controls. B) TGFB1 activated SMAD2/3 in PHM1 cells. DAPI was used to counterstain the nuclei of PHM1 cells. Images are representative for immunofluorescence staining from two independent culture experiments. Original magnification ×40. C and D) TGFB ligands stimulated ACTA2 and TGIF1 expression in PHM1 cells. Data are presented as the mean ± SEM from three independent cell-culture experiments. Bars without a common letter are significantly different (P < 0.05).

Impaired Uterine Decidualization in Mice with Constitutively Active TGFBR1

The significantly altered endometrial property of TGFBR1 Pgr-Cre CA mice pointed to potential defects in uterine function. To test this possibility, we assessed the ability of the uterus to undergo decidualization, a process in which endometrial stromal cells differentiate into decidual cells to support implanted embryos during pregnancy. An artificial decidualization experiment was performed using a well-established protocol (Fig. 6A) to determine whether constitutively active TGFBR1 in the mouse uterus compromised uterine decidual response. Two days after the deciduagenic stimulus, the ratio of the stimulated horn to unstimulated horn weight of the TGFBR1 Pgr-Cre CA mice was significantly lower than that of control mice (Fig. 6, B–D), suggesting decidualization defects in mice with enhanced TGFB signaling in the uterus. Further histological analysis and alkaline phosphatase (AP) staining assay (Fig. 6, E–J) demonstrated impaired decidualization and uterine stromal cell differentiation in the TGFBR1 Pgr-Cre CA mice, as evidenced by low levels of AP staining (Fig. 6J) compared with controls (Fig. 6G).

FIG. 6.

Constitutively active TGFBR1 in the uterus impairs uterine decidualization. A) Illustration of the artificial decidualization procedure. OVX, ovariectomy. B and C) Gross morphology of control (B) and TGFBR1 Pgr-Cre CA (C) uteri collected 2 days after decidual stimulation. DH, decidualized uterine horn; CH, control uterine horn that was not stimulated. D) Ratio of wet uterine weight between stimulated and unstimulated horns of control and TGFBR1 Pgr-Cre CA mice (n = 4–5). E–J) Histological analysis and alkaline phosphatase staining of uterine tissues from TGFBR1 Pgr-Cre CA and control mice. Note the strong AP signals in the decidualized horns of control mice (G) versus those of TGFBR1 Pgr-Cre CA mice (J). Bar = 100 μm. K and L) Expression of Bmp2 and Fst in TGFBR1 Pgr-Cre CA and control uteri during artificial decidualization. Note that Bmp2 and Fst transcripts were upregulated by decidual stimulus in control mice but not TGFBR1 Pgr-Cre CA mice. In the decidualized horns, significantly lower Bmp2 mRNA levels were detected in the TGFBR1 Pgr-Cre CA mice compared with controls (n = 3–4). ns, not significant. Data are presented as the mean ± SEM. *P < 0.05.

To further explore the molecular mechanism responsible for the defective decidualization, we assessed the ability of a deciduagenic stimulus to induce the expression of critical regulators of uterine decidualization in the TGFBR1 Pgr-Cre CA uteri by examining whether expression and induction of Bmp2 and Fst were altered. Interestingly, we found that the deciduagenic stimulus failed to induce Bmp2 and Fst expression in the TGFBR1 Pgr-Cre CA mice, whereas upregulation of these genes following the deciduagenic stimulation occurred in the uteri from control groups (Fig. 6, K and L). Moreover, a marked reduction of Bmp2 transcript abundance occurred in the decidualized horns of TGFBR1 Pgr-Cre CA mice compared with controls (Fig. 6K). No significant differences in Fst mRNA levels between decidualized horns of TGFBR1 Pgr-Cre CA and control groups were detected (Fig. 6L). Therefore, enhanced TGFB signaling compromises uterine decidualization.

Ovarian Defects in Mice Harboring a Constitutively Active TGFBR1

Because Pgr-Cre is also expressed in the oviduct and preovulatory follicles in the ovary after gonadotropin surge, the oviductal and ovarian histology was analyzed to determine the effect of overactivation of TGFBR1 in these tissues. The histological structure of the oviduct of TGFBR1 Pgr-Cre CA mice was comparable to that of controls, as demonstrated by immunofluorescence staining of ACTA2 (Fig. 7, A–D). However, histological studies of the ovary showed that three out of eight TGFBR1 Pgr-Cre CA mice developed cystic structures filled with blood (Fig. 7, F–H). In contrast, none of the control mice developed this pathology (Fig. 7E). Further analysis of mice at the age of 2–3 mo revealed the presence of abnormal luteinized structures within the ovary of the TGFBR1 transgenic mice versus controls (Fig. 7, I–L). These structures were positive for HSD3B (Fig. 7, O and P), which was localized to corpora lutea but not ovarian granulosa cells in the control mice (Fig. 7, M and N). To evaluate the ovulatory potential of mice containing a constitutively active TGFBR1, we performed a superovulation experiment and demonstrated that immature TGFBR1 Pgr-Cre CA mice ovulated a similar number of oocytes as controls in response to exogenous gonadotropin injection (39.3 ± 5.9 vs. 40.0 ± 3.9 oocytes/mouse, n = 3–4; P > 0.05). These results collectively suggest that overactivation of TGFB signaling profoundly affects ovarian cell differentiation and/or function, although the ovulation potential in prepubertal mice is not altered.

FIG. 7.

Ovarian pathology of TGFBR1^{CA Lox/Lox}; Pgr-Cre mice. A–D) Intact smooth muscle structure of the oviduct in TGFBR1^{CA Lox/Lox}; Pgr-Cre mice. Immunofluorescence of ACTA2 (red) using oviduct collected from 2-mo-old control (A and B) and TGFBR1^{CA Lox/Lox}; Pgr-Cre mice (C and D). E–H) Hemorrhagic cystic structure in the ovary of TGFBR1^{CA Lox/Lox}; Pgr-Cre mice. G and H (dotted yellow line) demonstrate the different cystic structures in the same ovary. I–P) TGFBR1^{CA Lox/Lox}; Pgr-Cre mice contained abnormal luteinized structures (K and L; dotted yellow line and yellow arrowheads) that were positive for HSD3B (red; O and P; yellow arrowheads). HSD3B was mainly localized to the corpus luteum (yellow arrowheads) but not granulosa cells within an adjacent follicle in the control mice (M and N). ACTA2 (green) was used to mark the outer theca layer of follicles. DAPI (blue) was used to counterstain the nucleus. Ct, cystic structure; CL, corpus luteum; AF, antral follicle. Bar = 100 μm (A–D and H), 200 μm (E–G, I, and K), and 50 μm (J, L, and M–P).

DISCUSSION

In mice, müllerian duct formation starts from the specification of coelomic epithelial cells that express LIM homeobox protein 1 (Lim1) at Embryonic Day (E) 11.75 [55]. Subsequent invagination of coelomic epithelium through the mesonephros and elongation of the duct to urogenital sinus are completed by E13.5 [55]. The müllerian duct further develops into the female reproductive tract, including oviduct, uterus, cervix, and the upper part of vagina. The uterus contains simple epithelium and supporting mesenchyme at birth and gradually acquires basic myometrial structures by Postnatal Day 15 [56, 57]. The role of TGFB signaling in myometrial development has been revealed by our early studies using a Tgfbr1 conditional knockout mouse model. We showed that conditional deletion of TGFBR1 in the female reproductive tract disrupts the formation of uterine smooth muscle layers [18, 29]. The present study was aimed at developing a gain-of-function model to achieve further insights into the role of TGFB signaling in uterine development and function.

The role of TGFB superfamily signaling has been identified in many reproductive processes using a functional genomics approach. However, the majority of currently available mouse models were generated to induce loss-of-function of genes encoding the components of the TGFB signaling pathway [16, 18, 58–60]. It is noteworthy that overactivation of TGFB signaling has been associated with diseases, including fibrosis, Marfan syndrome, and late-stage cancers [30–32]. Because TGFB signaling is fine-tuned and controls homeostatic cellular processes in a highly contextually dependent manner [61], the use of loss-of-function and gain-of-function mouse models is complementary and beneficial to a more comprehensive understanding of TGFB signaling in both physiologic and pathologic conditions. Until now, no mouse models with overactivation of TGFB signaling in the female reproductive tract have been available. Several strategies have been utilized to develop mouse models with overactivation of TGFB signaling. It has been shown that targeting inhibitors of TGFB signaling, such as inhibitory SMADs (SMAD6 and SMAD7) [62] and fibrillin [30], can enhance TGFB signaling activity. Transgenic mice expressing TGFB1 or a constitutively active TGFBR1 driven by a fibroblast-specific promoter have been reported [63, 64]. By utilizing a latent conditional constitutively active TGFBR1 allele [34, 40] and Pgr-Cre, we were able to create a mouse model containing a constitutively active TGFBR1 in the uterus. To our knowledge, this is the first mouse model with overactivation of TGFB signaling in the uterus.

The most striking phenotype of the TGFBR1 Pgr-Cre CA mice is the development of hypermuscled uteri, supporting a clear role of TGFB signaling in uterine smooth muscle biology. The role of TGFB signaling in smooth muscle cell differentiation and function has been unambiguously demonstrated in vascular smooth muscle cells [65–68]. Whereas TGFB signaling promotes smooth muscle gene expression and smooth muscle formation in the TGFBR1 gain-of-function model, our studies using Tgfbr1 conditional knockout mice [29] have suggested that TGFBR1 may not be required for the lineage commitment of uterine smooth muscle cells, because loss of TGFBR1 does not block the expression of smooth muscle genes by myometrial cells. It is conceivable that compensatory mechanisms operate in the absence of TGFB signaling in the TGFBR1 loss-of-function model. As direct in vitro evidence that TGFB signaling regulates smooth muscle gene expression in myometrial cells, we herein demonstrated that TGFB signaling activated SMAD2/3 and induced ACTA2 expression in human uterine smooth muscle cells. This result suggests a common regulatory mechanism operated by TGFB signaling in both mouse and human uterine smooth muscle cells. TGFB signaling induces the differentiation of mesenchymal stem cells into smooth muscle cells [69]. It is tempting to speculate that enhanced TGFB signaling resulting from constitutively active TGFBR1 may cause endometrial stem cell differentiation. Recent evidence suggests the existence of stromal stem/progenitor cells in the mouse endometrium, although the properties of these cells are incompletely defined [70, 71]. Thus, it remains to be determined whether enhanced TGFB signaling affects endometrial stem/progenitor cell differentiation and promotes smooth muscle cell formation.

The TGFB signaling is involved in myofibroblast differentiation [72]. Myofibroblasts are cells that differentiate from quiescent fibroblasts in response to a number of stimuli, including injury. They are characterized by development of contractile machinery induced by changes in the composition and mechanical properties of the extracellular matrix [73]. Myofibroblasts express ACTA2 and can be differentiated/transdifferentiated from several potential precursor cells including local fibroblasts, epithelial cells, and blood-borne cells [74]. Myofibroblasts are generally absent in most tissues, but their differentiation can be induced during tissue repair [75]. It has been shown that human decidual stromal cells resemble myofibroblasts and express ACTA2 [76]. The presence, origin, and role of myofibroblasts in the mouse uterus are poorly understood. TGFB signaling is a key regulator of myofibroblast differentiation in other systems [77]. Supporting a role of TGFB signaling in uterine myofibroblast differentiation, we found that overactivation of TGFB signaling led to the differentiation of myofibroblast-like cells in mouse endometrium, accompanied by an increased expression of smooth muscle genes. Furthermore, our results showed that TGFBR1 Pgr-Cre CA mice had compromised decidualization ability, coincident with uterine morphological aberrations in these mice. Uterine gland development supports decidualization [78]. It is not known whether reduced uterine glands in the TGFBR1 Pgr-Cre CA mice affect decidualization induced by an artificial stimulus. The potential contribution of myofibroblast cells to the observed decidualization defects also awaits elucidation.

Uterine glands are essential for female fertility and pregnancy in multiple species, including rodents [79]. Mouse uterine glands develop postnatally. The mechanisms underlying uterine gland development are not well understood, although recent studies using mouse models have identified a few important genes in this developmental process, such as forkhead box A2 (Foxa2), wingless-type MMTV integration site family member 4 (Wnt4), Wnt5a, Wnt7a, and cadherin 1 (Cdh1) [80–86]. Elegant studies have utilized progestin treatment to disrupt uterine gland development and fertility in both sheep and mouse models [87, 88]. Likewise, the synthetic estrogen, diethylstilbestrol, can also block adenogenesis when administered within an appropriate time frame during postnatal uterine development [89]. The exact mechanisms of how these endocrine disruptors suppress uterine gland development are not clear but may be partially associated with reduced uterine epithelial cell proliferation and/or altered expression of Wnt genes [89]. TGFB signaling regulates multiple developmental events, and several lines of evidence suggest a link between estrogen action and TGFB signaling pathways in various types of cells, including uterine cells [90–93]. In addition, TGFB signaling interacts with Wnts [94]. Because overactivation of TGFB signaling reduces uterine gland formation, it is tantalizing to speculate that TGFB signaling may partially mediate the adverse effect of endocrine disruptors on uterine gland development through regulating adenogenic genes. However, further investigation is warranted to identify a potential link.

Although the primary focus of the present study targeted uterine development and function, we noted histological alterations of the adult ovary. The in vivo role of TGFB signaling in the ovary remains largely unknown. To determine whether potential ovarian defects contribute to the observed infertility phenotype of TGFBR1 Pgr-Cre CA mice, we performed a timed mating study and found that no blastocysts could be recovered from the TGFBR1 Pgr-Cre CA uteri except for a reduced number of morula stage embryos (data not shown). Therefore, the ovarian defects of TGFBR1 Pgr-Cre mice could be a contributing factor to the reproductive failure observed in these mice. Counterintuitively, the ovulatory potential seems to be integral in sexually immature mice based on a superovulation experiment. Of note, Pgr-Cre is expressed in granulosa cells within the preovulatory follicles, and interpretation of the ovarian phenotype of TGFBR1 Pgr-Cre CA mice should take into account the differentiation status of granulosa cells and the timing of Cre expression. Therefore, the observed ovarian defect in adult mice may reflect a cumulative effect of constitutive activation of TGFBR1 as opposed to the superovulation results obtained from prepubertal mice, in which Pgr-Cre activity is expected to be minimal before gonadotropin stimulation. An independent study is currently ongoing to characterize the role of overactivation of TGFBR1 in the mouse ovary.

Collectively, our results show that constitutive activation of TGFBR1 in the uterus leads to morphological abnormalities and functional deficiency of the uterus. These findings further reinforce the importance of a precisely controlled TGFB signaling system in normal uterine development and function.