GATA4 and GATA6 Silencing in Ovarian Granulosa Cells Affects Levels of mRNAs Involved in Steroidogenesis, Extracellular Structure Organization, IGF-I Activity, and Apoptosis

Abstract

Knockdown of the transcription factors GATA4 and GATA6 in granulosa cells (GCs) impairs folliculogenesis and induces infertility. To investigate the pathways and genes regulated by these factors, we performed microarray analyses on wild-type GCs or GCs lacking GATA4, GATA6, or GATA4/6 (G4^gcko, G6^gcko, and G4/6^gcko) after in vivo treatment with equine chorionic gonadotropin. GATA4 deletion affected a greater number of genes than GATA6, which correlates with the subfertility observed in G4^gcko mice and the normal reproductive function found in G6^gcko animals. An even greater number of genes were affected by the deletion of both factors. Moreover, the expression of FSH receptor, LH receptor, inhibin α and β, versican, pregnancy-associated plasma protein A, and the regulatory unit 2b of protein kinase A, which are known to be crucial for ovarian function, was greatly affected in double GATA4 and GATA6 knockouts when compared with single GATA-deficient animals. This suggests that GATA4 and GATA6 functionally compensate for each other in the regulation of key ovarian genes. Functional enrichment revealed that ovulation, growth, intracellular signaling, extracellular structure organization, gonadotropin and growth factor actions, and steroidogenesis were significantly regulated in G4/6^gcko mice. The results of this analysis were confirmed using quantitative polymerase chain reaction, immunohistochemical, and biological assays. Treatment of GCs with cAMP/IGF-I, to bypass FSH and IGF-I signaling defects, revealed that most of the affected genes are direct targets of GATA4/6. The diversity of pathways affected by the knockdown of GATA underscores the important role of these factors in the regulation of GC function.

The transcription factors GATA4 and GATA6 are crucial for normal granulosa cell (GC) function (1). These factors are coexpressed in the GCs of numerous species, including human (2), pig (3), rat, and mouse (4). In the mouse, the GATA4 and GATA6 genes encode proteins of 48 and 45 kDa, respectively, which are 85% identical at the amino acid level within the DNA-binding region (5). Consequently, both GATA factors recognize a conserved binding motif characterized by the core A/T-GATA-A/G (6). This property of GATA4 and GATA6 impedes determination of the genes and functional pathways targeted by each factor in the ovary. On the other hand, this particularity of GATA4 and GATA6 could account for the functional compensations observed when one or the other is silenced in GCs. For instance, GATA4 GC conditional knockout mice are subfertile (1, 7), whereas GATA6 conditional knockout mice have no reproductive defects (1); in marked contrast, animals lacking both GATA4 and GATA6 in GCs are infertile (1). The mechanisms responsible for the redundant or compensatory roles of GATA4 and GATA6 in the ovary are not fully understood.

These findings also indicate that GATA4 and GATA6 do not contribute equally to regulate ovarian function and that GATA4 plays a major role in the regulation of follicle growth and maturation. Thus, mice lacking GATA4 in GCs release significantly fewer oocytes at ovulation than wild-type (WT) animals (1, 7). However, mice lacking GATA6 ovulate normally. In addition, GATA4, but not GATA6, binds to the promoters of the aromatase (CYP19a1) and the FSH receptor (FSHR) genes (1, 8), which are essential for normal follicle growth. Moreover, we have previously shown a significant decrease in the expression of CYP19a1, LH/chorionic gonadotropin receptor (Lhcgr), cholesterol side chain cleavage (CYP11a1), and FSHR only in the absence of GATA4. Interestingly, although genes targeted specifically by GATA6 have not been described in GCs, GATA6 compensates for the absence of GATA4 and partially sustains GC function (1), suggesting that GATA6 is able to replace GATA4 in the stimulation of key genes involved in folliculogenesis. The identity of these genes remains unknown.

Here, we examined the impact that the lack of GATA4, GATA6, or both has on the steady-state levels of mRNAs during the process of GC differentiation using microarrays. Elucidation of the genes regulated by GATA4 and/or GATA6 is essential to provide novel insights into the transcriptional regulatory programs controlled by each factor. The results of this analysis revealed a role for GATA factors in the regulation mRNA levels of genes involved in ovulation, steroid metabolic processes, extracellular structure organization, IGF metabolism, and intracellular signaling.

Materials and Methods

Reagents

DMEM/F12 medium, equine chorionic gonadotropin (eCG), estradiol, and all other media components were purchased from Sigma unless otherwise specified. Antibodies used were cleaved caspase 3 (no. 9661; Cell Signaling Technology), proliferating cell nuclear antigen (PCNA) (no. 2714–1; Epitomics), versican (Vcan) (no. AB1033; Millipore), and secondary antibody (no. ab6721; Abcam).

Animals

The generation of GATA4 (G4^gcko), GATA6 (G6^gcko), and GATA4/6 (G4/6^gcko) GC conditional knockout animals was described previously (1). Animals were treated in accordance with the National Institutes of Health Guide for Care and Use of Laboratory Animals, and all protocols were approved by the University of Illinois at Chicago Animal Care Committee.

RNA isolation and quantitative real-time PCR analysis

WT, G4^gcko, G6^gcko, and G4/6^gcko immature 23-day-old (D23) animals were treated with 7.5 IU of eCG for 48 hours. GCs were isolated from preovulatory follicles in WT, G4^gcko, G6^gcko, and G4/6^gcko animals or from large preantral follicles in G4/6^gcko mice, which do not form preovulatory follicles. Cells were filtered through a 40-μm mesh to partially eliminate cumulus oocyte complexes. Total RNA from GCs was isolated using TRIzol (Invitrogen) following the manufacturer's instructions. mRNA quantification was performed as previously described (1). The sequences of the primers used are available upon request. The identity and size of all PCR products were confirmed by sequencing and gel analysis. The results are expressed as the ratio between the copies per nanograms of total RNA of the gene of interest and ribosomal L19.

Microarray analysis

Total RNA was labeled and hybridized to the Affymetrix Mouse Gene 1.0 ST Array according to the protocol recommended by Affymetrix. Scanned images of each chip were analyzed for the following quality metrics: total background, raw noise, average signal present, signal intensity of housekeeping genes, 3′/5′ signal ratio of housekeeping genes, relative signal intensities of labeling controls, and absolute signal intensities of hybridization controls. All hybridizations passed according to indicated quality criteria. Microarray data analyses were performed using the software package BRB Array Tools. Filter threshold values were set to a minimum value of 25. Array normalization was performed by using the median of a set of housekeeping genes as reference. ANOVA tests were used to calculate significance of the differential expression. For all genes, the fold change was calculated by dividing the mutant value by the WT value. If this number was less than 1, the reciprocal is listed. The reported fold changes are the average of 3 animals for each genotype. A change was deemed significant only if the P value was less than 0.01 and the fold change was more than 2. Differentially regulated genes were then analyzed in Database for Annotation, Visualization, and Integrated Discovery (DAVID), Gene Set Enrichment Analysis (GSEA), and Significance Analysis of Microarrays (SAM) for functional pathway analysis.

Immunohistochemistry (IHC)

IHC was performed as previously described (1). The primary antibodies used were: cleaved caspase 3 (1:200), PCNA (1:300), and Vcan (1:300). Slides were developed using Vectastain elite ABC kit and 3,3′-diaminobenzidine peroxidase (Vector Laboratories) following manufacturer's recommendations and counterstained with Gill's hematoxylin.

GC culture

Undifferentiated GCs were isolated from preantral follicles from immature D23 mice treated for 3 days with 1-mg/mL estradiol. Cells were cultured as previously described (1). Cells were transfected with adenoviral Cre-recombinase (adCre) (University of Iowa Gene Vector Transfer Core) at a multiplicity of infection of 10. For 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay, cells were treated with or without 100-ng/mL FSH for 48 hours. Then, 50 μL of 50-mg/mL MTT dissolved in PBS was added to each culture well. Plates were incubated for 2 hours at 37°C. After incubation, medium was removed, and 500 μL of dimethyl sulfoxide were added. Plates were shaken for 5 minutes at room temperature, and the absorbance was read at 560 nm. For RNA expression, cells were treated with adCre for 24 hours before treatments with 1mM dibutyryl cyclic-AMP (dbcAMP) (Sigma), 50-ng/mL IGF-I (Sigma), or both. Cells were cultured for 48 hours after treatments and then harvested for RNA isolation.

Statistics

Data are expressed as the mean ± SEM. Multiple group statistical analyses were performed by one-way ANOVA followed by the Tukey test for multiple comparisons using GraphPad Prism 5 (GraphPad).

Results and Discussion

The knockdown of GATA4 and GATA6 in GCs impairs folliculogenesis and causes female infertility (1). In an effort to uncover the transcriptional defects that lead to these phenotypes, we performed mRNA microarray analyses on GCs of WT, G4^gcko, G6^gcko, or G4/6^gcko animals treated with eCG for 48 hours. Analysis of differentially expressed genes between the GATA conditional knockouts and WT demonstrated that more genes were affected by the absence of both GATA4 and GATA6 than in the absence of either factor alone (Figure 1, A–C). Microarray data were analyzed using SAM (false discovery rate: 0.01, 1000 permutations; confidence level: 90%) (9). One-way ANOVA of 3 independent samples for each genotype was used to identify genes in which expression changed by 2-fold or more between knockouts and WT. This analysis revealed that 493 genes in G4/6^gcko, 224 genes in G4^gcko, and 34 genes in G6^gcko were significantly regulated by 2-fold or more (P < .01). A list of all differentially expressed genes can be found in Supplemental Tables 1 (G4^gcko), 2 (G6^gcko), and 3 (G4/6^gcko), published on The Endocrine Society's Journals Online web site at http://endo.endojournals.org.

Figure 1.

Genomic-wide mRNA profile in WT, GATA4, GATA6, and GATA4/6 conditional knockout animals. (A–C) Scatter plot of mRNA profiles of GATA knockouts vs WT ovarian GCs. Each point represents a unique probe set. Y- and x-axis values are expressed as the logarithm of expression intensity for each probe set. The middle diagonal line represents equal expression. Probe sets that yielded a 2-fold difference (as determined by SAM analysis) are located outside (up or down) of the outlier lines that indicate ±2-fold between the mean of the ratios. (D–F) Venn diagrams of mRNA profiles for each genotype. Numbers indicate total mRNAs significantly regulated in common between GATA4, GATA6, and GATA4/6 as well as those individually regulated by each genotype. Lists of the genes included in each one of these categories can be found in Supplemental Tables 1–4.

The overlap of all differentially regulated genes between the 3 phenotypes was represented using a Venn diagram (Figure 1D). This diagram revealed a greater degree of overlap between G4^gcko and G4/6^gcko than between G6^gcko and G4/6^gcko or between G6^gcko and G4^gcko (Figure 1D). Venn diagrams of down- or up-regulated genes yielded similar findings (Figure 1, E and F). Genes regulated by each phenotype as well as the elements that are common between genotypes are listed in Supplemental Table 4, A–D. These findings confirm the predominant role of GATA4 in the regulation of GC function and provided for the first time a short list of genes that seem to be exclusive targets of GATA6. Based on these findings, it is also possible to conclude that GATA4 and GATA6 compensate for one another in the regulation of GC function.

Functional classification of GATA-regulated genes

In view of the compensatory actions observed between GATA4 and GATA6, we next performed functional analyses of genes significantly (P < .01) regulated in the absence of both GATA4 and GATA6 using the DAVID (http://david.abcc.ncifcrf.gov/), SAM, and GSEA (10, 11). Multiple pathways, including ovulation-related genes, ovarian/infertility genes, steroid metabolic process, extracellular organization, regulation of growth, and intracellular signaling, were found to be affected (Supplemental Table 5).

Ovulation-related and ovarian/fertility genes

GSEAs identified gene pathways associated with ovarian defects and fertility to be significantly represented within the differentially regulated genes in G4/6^gcko GCs (Figure 2A). A partial list of these genes is shown in Figure 2B. Of these genes, inhibin βa- and βb-subunits (Inhβa and Inhβb), inhibin α (Inhα), FSHR, Lhcgr, gremlin (Grem)1 and Grem2, peroxisome proliferator activated receptor γ, CCAAT/enhancer binding protein α, and prolactin receptor were down-regulated. However, genes such as endothelin receptor type A; follistatin-like 3; cytochrome P450, family 1, subfamily b, polypeptide 1; and anti-Mullerian hormone (Amh) were up-regulated.

Figure 2.

Ovarian- and fertility-related genes. (A) GSEA enrichment plot for ovulatory and infertility genes. (B) List of selected ovulatory/fertility genes from the GSEA and DAVID analysis. (C) qPCR determination of selected differentially regulated genes in untreated D23 WT animals and in eCG-treated WT, GATA4, GATA6, and GATA4/6 conditional knockout animals. Three or more animals were included for each genotype. Columns represent the mean ± SEM (*, P < .05; **, P < .01; ***, P < .001 vs WT one-way ANOVA, Tukey test). (D) FSHR protein levels in GCs from D23 eCG-treated WT and G4/6^gcko animals. β-Actin was used as a loading control.

To confirm the microarray results, we performed new experiments, in which immature WT or GATA conditional knockout mice were treated with eCG to stimulate follicle maturation and GC differentiation. As a control, D23 females were also included. The expression of selected genes was determined using quantitative polymerase chain reaction (qPCR) (Figure 2C). Confirming our previous report (1) and microarray results (Figure 2B), qPCR assays demonstrated the essential role of GATA4 in the regulation of FSHR and Lhcgr expression (Figure 2C). In addition, Western blot analyses demonstrated that FSHR protein expression is undetectable in GCs of G4/6^gcko animals (Figure 2D).

Inhibins are heterodimers of the common α-subunit with either βa- or βb-subunits (inhibins A and B, respectively). The expression of the α-subunit is regulated by GATA factors in vitro (12, 13). Our findings demonstrated that GATA factors are also required for the mRNA expression of the α-subunit in vivo. Thus, microarray results revealed a 2-fold decrease in the expression of the α-subunit in G4/6^gcko when compared with WT animals. This finding was confirmed by qPCR showing that the lack of GATA4 or GATA4 and GATA6 decreased α-subunit expression to levels observed in untreated D23 animals (Figure 2C). We also provided evidence that GATA4 and GATA6 participate in the regulation of the 2 β-subunits. Thus, in eCG-treated animals, microarray and qPCR data demonstrated a decrease in the mRNA expression of the βa (6-fold)- and βb (3-fold)-subunits in the absence of GATA4 and GATA6 when compared with WT animals treated with eCG (Figure 2, B and C). Moreover, qPCR data demonstrated that the expression of the βa- and α-subunits, but not that of βb-subunit, is stimulated by eCG (Figure 2C). The ratio of inhibin to activin changes as follicles grow (14). Thus, preantral follicles produce mainly activins (ββ dimer), whereas preovulatory follicles produce mostly inhibin A (βa/α dimer) (15). Our in vivo results suggest that GATA factors are required for the increase in inhibin A observed in preovulatory follicles.

In contrast to the inhibins, AMH is highly expressed in GCs of preantral and early antral follicles (16) and progressively decreases toward the preovulatory stage (17–19). In agreement with the lack of follicle maturation observed in the absence of GATA factors (1), AMH mRNA expression was higher in the ovary of G4/6^gcko animals when compared with WT (Figure 2, B and C). These are intriguing findings, because the expression of AMH has been shown to be stimulated by GATA4 (20). The mechanism that leads to the sustained mRNA expression of AMH in GATA4-deficient GCs needs further analysis.

Microarray analysis also revealed that the lack of GATA4 and GATA6 affects the mRNA expression of Grem1 and Grem2 in GCs. Grem1 was significantly lower in G4^gcko and G4/6^gcko animals. In contrast, Grem2 expression decreased by 2-fold in the G6^gcko and in G4/6^gcko animals (Figure 2B), suggesting that GATA4 and GATA6 may specifically target Grem1 and Grem2, respectively. Grem1 expression in GCs is stimulated by eCG (21, 22) and by growth differentiation factor 9 (GDF9) and bone morphogenetic protein 4 (22). Both Grem1 and Grem2 prevent the inhibitory effect of bone morphogenetic protein 4 on GC steroidogenesis (21). Our findings demonstrate that the absence of GATA factors abolished the increase of Grem1 and Grem2 expression induced by eCG. Therefore, GATA4 and GATA6 may contribute to the normal development of folliculogenesis by mediating the effects of GDF9 on Grem1/2 expression. Noteworthy, Grem2 is one of the few genes regulated in the absence of GATA6. However, qPCR analysis indicated that the expression Grem2 was low in the absence of either GATA factor. However, Grem2 expression tended to be lower in the absence of GATA6 than in the absence of GATA4. Whether the Grem2 gene can be specifically regulated by GATA6 remains to be determined.

Steroid synthesis

Several genes involved in steroid metabolism were differentially regulated in GATA conditional knockout animals (Figure 3, A and B). Steroidogenic genes, including aldo-keto reductase family 1, member C18 (progesterone metabolism), CYP11a1 (progesterone synthesis), and CYP19a1 (estrogen synthesis), were decreased in eCG-treated G4/6^gcko mice by 9-, 2-, or 11-fold, respectively, when compared with eCG-treated WT animals. A down-regulation of CYP19a1, aldo-keto reductase family 1, member C18, and CYP11a1 expression was also seen in the G4^gcko animals. Ferredoxin 1, which shuttles electrons from ferredoxin reductase to CYP11a1, was significantly decreased in G4/6^gcko when compared with WT and single knockout animals treated with eCG. Additionally, eCG stimulation of low-density lipoprotein receptor, which is required for the uptake of cholesterol needed for steroid synthesis, was significantly decreased in G4/6^gcko animals (Figure 3, A and B). Cytochrome P4501b1, that inactivates estradiol (23), increased by 5-fold in G4^gcko and G4/6^gcko animals. These findings suggest that, in the absence of GATA4 and GATA6 expression, the synthesis of estradiol and progesterone is significantly impaired. This conclusion is supported by the decrease in estradiol and progesterone serum levels observed in G4/6^gcko when compared with WT animals (Figure 3C).

Figure 3.

Steroidogenic genes. (A) List of selected steroidogenic genes from DAVID and GSEA of differentially regulated genes. (B) Relative expression of steroidogenic genes in untreated D23 WT animals and in eCG-treated WT, GATA4, GATA6, and GATA4/6 conditional knockout animals. (C) Progesterone and estradiol levels after a 48-hour treatment with eCG of WT or G4/6^gcko animals. Three or more animals were included for each genotype. Columns represent the mean ± SEM (*, P < .05; **, P < .01; ***, P < .001 vs WT one-way ANOVA, Tukey test).

A 6-fold increase in apolipoprotein-E (Apoe) was observed in G4/6^gcko animals. In the ovary, Apoe may limit androgen production, thereby limiting follicular estrogen synthesis. In fact, in humans, the levels of Apoe in follicular fluid decrease as serum estrogen levels increase during the menstrual cycle (24). In addition, Apoe has been shown to increase in atretic follicles (25), which corresponds with the increase in apoptosis observed in follicles lacking GATA factors (see below).

Taken together, these findings suggest that the increased mRNA expression of cytochrome P4501b1 and Apoe along with a decrease in aromatase expression may contribute to the decrease in estradiol levels observed in and G4/6^gcko animals (Figure 3C).

Extracellular structure and organization

DAVID analysis of differentially expressed genes revealed an enrichment of genes involved in the reorganization of the extracellular matrix (ECM) and cell adhesion (Supplemental Table 5). Similarly, GSEA identified a set of genes associated with ECM proteins (data not shown). A list of selected genes identified by these 2 analyses is shown in Figure 4A. Within these genes, A disintegrin and metalloproteinase with thrombo spondin motifs (ADAMTS)1 and Vcan are known to be crucial for normal ovulation (26, 27). ADAMTS are proteinases that cleave proteoglycans present in the ECM that surrounds all cells and tissues. Of the proteoglycans present in the ovarian ECM, Vcan is produced by GCs and can be found in the granulosa layer of small growing follicles and in antral follicles (28, 29). These findings suggest that Vcan is a matrix component of the follicle expressed throughout folliculogenesis. Data from the microarray and qPCR experiments suggest that the expression of ADAMTS1 and Vcan increases after treatment of WT animals with eCG. However, this increase in the mRNA expression of ADAMTS1 and Vcan was abolished by the deletion of both GATA4 and GATA6 (Figure 4B). Thus, after treatment with eCG, the expression of ADAMTS1 and Vcan was 5- and 10-fold lower, respectively, in G4/6^gcko when compared with WT (Figure 4B). This decrease in the expression of Vcan was confirmed by IHC staining with an antibody that detects the V0 and V1 isoforms of Vcan. There was diffuse Vcan staining in the granulosa layer of antral follicles but very intense staining within the follicular antrum and cumulus cells (Figure 4C). Similarly, Vcan was detected in antral and secondary follicles of WT, G4^gcko, and G6^gcko animals. In contrast, in G4/6^gcko animals, Vcan was detectable in early secondary follicles (Figure 4C, arrows) but not in any of the few early antral follicles that these animals develop (Figure 4C, arrowhead). This pattern of Vcan expression was also observed at the RNA level (Figure 4B).

Figure 4.

ECM and tissue remodeling genes. (A) List of selected differentially regulated genes involved in ECM/structural remodeling as determined by DAVID analysis. (B) qPCR quantification of key extracellular/remodeling genes in the different genetic backgrounds. Three or more animals were included for each genotype. Columns represent the mean ± SEM (*, P < .05; **, P < .01; ***, P < .001 vs WT one-way ANOVA, Tukey test). (C) IHC analysis of Vcan in WT, G4^gcko, G6^gcko, and G4/6^gcko ovaries from D23 eCG-stimulated mice (n = 3 for each genotype; representative pictures are shown). #, secondary follicles; *, antral follicles; arrow, secondary follicles; arrowhead, early antral follicles. Vcan staining is depicted in brown, counterstaining by hematoxylin is depicted in light blue.

Cartilage oligomeric matrix protein (Comp) mRNA levels were significantly decreased (22-fold) in G4^gcko and G4/6^gcko mice when compared with WT mice. In the absence of GATA6, a 2-fold decrease in Comp expression was also observed (Figure 4A). qPCR analysis confirmed the down-regulation of Comp in G4^gcko, G6^gcko, and G4/6^gcko animals (Figure 4B) and demonstrated that eCG treatment induces a 20-fold stimulation of Comp expression. The role that Comp may play in the regulation of follicle development remains to be determined.

In contrast to the dramatic decrease in the expression of Comp, ADAMTS1, and Vcan observed in conditional knockout animals, we observed an increase in the mRNA expression of some ECM-related proteins and enzymes. For instance, the expression of collagen 12 and 6, laminin a1/2, b1, and c3, and several integrins increased in the absence of GATA factors. Similarly, the expression of ADAMTS2 increased in GATA4^gcko and GATA4/6^gcko animals.

Taken together, these findings suggest that in the absence of GATA factors, the remodeling of the follicular ECM is greatly compromised. One important step in the process of follicle maturation is the formation of the antrum. Antrum formation does not occur in G4/6^gcko animals, suggesting that GATA4 and GATA6 are needed for the build-up of the extracellular components involved in this process.

IGF-I and IGF binding protein (IGFBP) system

DAVID analysis indicated that the lack of GATA4 and GATA6 expression in GCs negatively affects the biological activity of IGF-I (P < 6.8 × 10⁻⁵). IGF-I is required for the differentiation of GCs to the preovulatory stage (30). Although a small 1.7-fold reduction of IGF-I expression was observed, microarray results demonstrated that the mRNA of IGFBP2, IGFBP4, and IGFBP5 remained highly expressed in GCs lacking GATA factors (Figure 5A). In particular, IGFBP4 was significantly up-regulated in G4^gcko (4.4-fold), G6^gcko (1.93-fold), and G4/6^gcko (21.92-fold) and was within the most up-regulated genes in all 3 knockouts. IGFBPs inhibit the interaction of IGF-I with the IGF-I receptor and are known to prevent follicular growth and maturation (31, 32). IGFBP expression has been shown to be down-regulated during the differentiation of GCs to the preovulatory stage (33–35), more specifically by action of FSH (36–38). qPCR assays confirmed the up-regulation of IGFBP2 (3-fold), IGFBP4 (4-fold), and IGFBP5 (2-fold) in the absence of GATA4. However, in the absence of both GATA factors, only IGFBP4 (11-fold) and IGFBP5 (4-fold) were up-regulated (Figure 5B).

Figure 5.

IFG-related genes. (A) Differentially regulated genes related to the IGF-I signaling pathway in GATA conditional knockout and WT animals. (B) qPCR results for IGF-I signaling pathway genes in the different genetic backgrounds. Three or more animals were included for each genotype. Columns represent the mean ± SEM (*, P < .05; **, P < .01; ***, P < .001 vs WT one-way ANOVA, Tukey test).

IGFBP5-overexpressing female mice are subfertile (39), whereas IGFBP4 inhibits LH-induced progesterone and FSH-induced estradiol production in human GCs (40). IGFBP levels are mainly regulated by proteolytic degradation by specific proteinases, such as pregnancy-associated plasma protein-A (Papp-a), which is highly expressed in healthy antral follicles and positively correlates with dominant follicle development (41, 42). Accordingly, Papp-a knockout animals have a reduced number of pups per litter, a reduced number of oocytes ovulated, and low estradiol levels after eCG stimulation (43, 44). Papp-a expression in GCs was significantly decreased in GATA4 (4.24-fold) and GATA4/6 (6.47-fold), suggesting that GATA factors indirectly contribute to regulate IGF-I activity by regulating Papp-a expression.

In addition, a 2-fold increase of phosphatidylinositol-3-kinase (PI3K)-interacting protein 1 (PIK3IP1), which binds to the p110 catalytic subunit of PI3K and reduces its activity (45), was found in animals lacking GATA4 and GATA6. Because activation of PI3K is part of the canonical pathway activated by IGF-I, the increase in PIK3IP1 may further contribute to a decrease in IGF-I signaling.

These findings suggest that in the absence of GATA factors, IGF-I biological activity declines due to a reduction of IGF-I, an increase in IGFBPs and PIK3IP1 mRNA expression, and a decrease in Papp-a expression. These effects could significantly contribute to the lack of follicle growth and the infertility observed in G4/6^gcko animals.

Apoptosis/cell division

Genes involved in apoptosis and cell proliferation were affected by the deletion of GATA factors in GCs (Figure 6A). Within these genes, defender against apoptotic cell death, a negative regulator of programmed cell death (46), was expressed at significantly lower levels in cells lacking GATA than in WT cells. In agreement with this finding, staining for cleaved caspase 3, a marker of apoptosis, increased in the ovaries of G4^gcko and G4/6^gcko animals treated with eCG in comparison to WT treated animals (Figure 6B).

Figure 6.

Cell growth- and apoptosis-related genes. (A) List of selected genes involved in apoptosis and cell growth found to be significantly affected by the lack of GATA factor expression in GCs. (B) IHC for cleaved caspase 3 protein in WT, G4^gcko, G6^gcko, and G4/6^gcko ovaries from D23 eCG-treated mice (n = 3 for each genotype; representative pictures are shown). Cleaved caspase 3 staining is depicted in brown, counterstaining by hematoxylin is depicted in light blue. (C) Proliferation, determined using MTT assays, of WT or GATA4/6-deficient GCs. Proliferation was stimulated with 50 ng/mL of FSH. The experiment was repeated at least 3 times. Columns represent the mean ± SEM, columns with different letters differ significantly. (D) IHC for PCNA in D23 eCG-treated WT or conditional knockout animals (n = 3 for each genotype; representative pictures are shown). PCNA staining is depicted in brown, counterstaining by hematoxylin is depicted in light blue.

In addition, an increase in the mRNA expression of growth arrest-specific (Gas)1 and Gas6 genes, which are both known to inhibit cell proliferation (47, 48), was observed in G4/6^gcko mice. We also found an increase in the expression of growth arrest and DNA damage-inducible (GADD45) proteins: GADD45A, GADD45B, and GADD45G. Supporting a decrease in cell proliferation in the absence of GATA factors, FSH-induced stimulation of cell proliferation was significantly reduced in G4^gcko and G4/6^gcko cells (Figure 6C). This finding was confirmed in vivo by a decrease in PCNA, a marker of proliferation, staining in G4/6^gcko when compared with WT animals (Figure 6D).

Dishevelled, Egl-10 and Pleckstrin mammalian target of rapamycin (mTOR)-interacting protein (Depdc6) is a component of both mTOR1 and mTOR2 complexes and negatively regulates mTOR function (49). mTOR activity is a reliable indicator of cell growth (50, 51). Microarray results showed that Depdc6 remains highly expressed in the absence of GATA4 and GATA6 when compared to WT cells. In addition, the homeodomain-interacting protein kinase 2 (HIPK2) was significantly reduced in G4^gcko and G4/6^gcko. HIPK2 is involved in the regulation of cell survival and proliferation (52). Accordingly, Hipk2 null mice show reduced cell proliferation and accumulation of cells in the G₀/G₁ phase of the cell cycle (52). Thus, increased Depdc6 expression and decreased Hipk2 may contribute to the diminished amount of proliferation observed in conditional knockout cells.

These findings suggest that the lack of GATA factors in GCs not only increases apoptosis but also halts GC proliferation. These effects are accomplished by an augmented expression of genes involved in growth arrest, such as GAS1/6, GADD45a/b/g, and Depdc6, as well as decreased expression of proliferative factors such as HIPK2. To our knowledge, this is the first report, suggesting a role for these genes in the proliferation of GCs.

Intracellular signaling

DAVID analysis and GSEA showed an enrichment of genes involved in intracellular signaling. Of these genes, the expression of the protein kinase A-regulatory subunit R2b (Prkar2b) was significantly down-regulated in the absence of GATA factors. Protein kinase A, which is crucial for normal ovarian function, is formed by 2 catalytic units and 2 regulatory (R) subunits, of which 4 R subunits (R1a, R1b, R2a, and R2b) have been described (53). The mRNA expression of Prkar2b was decreased by 3-fold in G4/6^gcko when compared with WT animals (Figure 7A). Prkar2b is the most abundant R subunit expressed in GCs, where it is stimulated by FSH (54, 55). Our microarray analysis confirmed the abundance of R2b with respect to other units (ratio 2b/1a/2a/1b: 1/0.08/0.02/0.01), whereas qPCR results confirmed the stimulatory effect of FSH on the expression of Prkar2b (Figure 7B). Moreover, these findings demonstrated that GATA4 and GATA6 are required for the stimulation of Prkar2b mRNA expression by FSH (Figure 7B).

Figure 7.

Intracellular signaling-related genes. (A) Partial list of intracellular signaling-related genes differentially regulated in conditional knockout vs WT animals. (B) qPCR results for selected intracellular signaling genes in the different genetic backgrounds. Three or more animals were included for each genotype. Columns represent the mean ± SEM (*, P < .05; **, P < .01; ***, P < .001 vs WT one-way ANOVA, Tukey test).

A 6- to 10-fold decrease in the mRNA expression of the membrane receptor plexin C1 (PlxnC1) was observed in G4^gcko and G4/6^gcko animals (Figure 7, A and B). Plexins are receptors for semaphorins (Semas). Plexin B1 and its ligand Sema4D are expressed in ovarian follicles of mice under the control of FSH (56, 57). We found that PlxnC1 and its ligand Sema7A are also highly expressed in GCs (Figure 7, A and B). No significant changes in the expression of Sema7A were observed in the microarray analysis. However, by qPCR, a 5-fold decrease was observed in G4/6^gcko when compared with eCG-treated WT animals, whereas a significant increase in this mRNA was observed in both single knockouts (Figure 7B). Although the function of these proteins in the ovary is unknown, PlxnC1 and Sema7A are involved in the regulation of cytoskeleton components, including actin and cofilin (58, 59). Cofilin inactivation plays an important role in ovarian steroidogenesis (60), suggesting that PlxnC1 may participate in the regulation of steroid synthesis in the ovary via regulation of cofilin.

The MAPK pathway has a crucial role in the regulation of folliculogenesis. For instance, disruption of Erk1/2 in mouse GCs impairs LH-induced oocyte resumption of meiosis, ovulation, and luteinization (61). The lack of GATA factors affected several members of the MAPK signaling pathways, including Mapkbp1, Map3k5, and Mapk10. In addition, the mRNA expression of Rho and Ras small GTPases, including Rhobtb1, Rnd2, Rassf2, Rasgrp4, and Ralgapa2 was affected in conditional knockout animals. In particular, expression of Ras was inhibited in the G4/6^gcko. Ras, via activation of the MAPK pathway, promotes growth, proliferation, differentiation, and survival of cells (62). These findings suggest that a decrease in the expression of Ras signaling may lead to the reduction in proliferation observed in GCs lacking GATA factors.

Assessment of gene regulation by GATA vs FSH signaling pathway

The results presented above demonstrated that one of the most important targets of GATA factors in GCs is the FSHR. Because of the central role of FSH in follicle maturation and GC differentiation, we further examined whether the changes in the levels of mRNA observed in GCs lacking GATA factors could be attributed entirely or in part to the decrease in FSHR expression. For this purpose, we isolated GCs from WT or GATA4^F/F:GATA6^F/F immature animals. To induce the recombination of the floxed alleles, cells were infected with an adenovirus encoding Cre-recombinase (adCre). As expected, adCre reduced GATA4 and GATA6 expression by 98% and 97%, respectively, in GCs containing floxed genes, but not in WT cells. Similarly, silencing of GATA factors resulted in the down-regulation of the FSHR (data not shown).

Next, we assessed the response of WT or GATA4/6-deficient GCs to cAMP, IGF-I, or their combination. This approach bypassed the decrease of FSH and IGF-I signaling observed in cells lacking GATA factors. The results of these experiments suggested the presence of 3 groups of mRNAs based on whether or not their stimulation by cAMP/IGF-I requires GATA4/6 or whether GATA factors only enhance the effects of FSH on the steady-state levels of these mRNAs. Thus, cAMP/IGF-I stimulated the expression of CYP19a1, inhibin-α, Papp-a, and Prkar2b in WT cells but not in GATA4/6 knockout cells (Figure 8A). In contrast, the increase of CYP11a1 expression induced by cAMP/IGF-I was not affected by the absence of GATA4/6 expression. Finally, cAMP/IGF-I stimulated the expression of Lhcgr and Vcan in both WT and GATA4/6 knockout cells. However, full stimulation of these genes was only attained in WT cells.

Figure 8.

Role of FSHR silencing on the regulation of the steady-state levels of mRNA by GATA factors. (A) Effect of dibutyryl cAMP (a cAMP analog) and/or IGF-I on the expression of selected genes in WT or GATA4- and GATA6-deficient GCs obtained from immature WT or GATA4F/F;GATA6F/F animals. GCs were cultured for 24 hours with a Cre-recombinase expression adenovirus at a multiplicity of infection of 10. Cells were then treated for 48 hours with vehicle, dbcAMP (1mM), IGF-I (50 ng/mL), or their combination. Columns represent the mean ± SEM of 6 different samples. Columns with different letters differ significantly (one-way ANOVA, Tukey test). (B) General scheme indicating genes regulated directly by GATA factors or indirectly via the inhibition of FSHR expression.

The results also suggest that the basal expression of several genes is affected by GATA4 and GATA6. In untreated cells, the knockdown of these factors decreased the basal expression of CYP19a1, Vcan, inhibin-α, PlxnC1, and Grem2 and increased the basal expression of IGFBP4 and Papp-a. Thus, the up-regulation of IGFBP4 in the absence of GATA factors observed in vivo and in vitro suggests that GATA4 and GATA6 are required to maintain low expression levels of this binding protein.

Noteworthy, we observed that the expression of Comp, PlxnC1, and Grem2 was stimulated by eCG in vivo but inhibited by dbcAMP in the in vitro experiments. These results suggest that FSH may not be the main factor regulating the expression of Comp, PlxnC1, and Grem2. In fact, the expression of Grem1 and Grem2 is stimulated by GDF9 (22). Our findings suggest that GATA4 and GATA6 may mediate this effect of GDF9.

These findings suggest that the knockdown of GATA4 and GATA6 directly and indirectly affects the steady-state levels of several mRNAs in GCs (Figure 8B). Direct effects occur when basal or FSH-induced stimulation of mRNA requires GATA factors, whereas indirect effects are mediated by the reduction of FSHR expression in the absence of GATA factors. Thus far, only the mRNA for CYP11a1 meets the latter criterion.

Conclusions

Our recent report demonstrated differential, but also overlapping, actions of GATA4 and GATA6 in the ovary. Thus, it was shown that G6^gcko mice are fertile, that G4^gcko mice are subfertile, and that the absence of both factors causes infertility. Here, we intended to determine why GATA4 is more dominant than GATA6 and to examine the mechanisms that cause infertility in the absence of both factors. In answering these questions, this report revealed that more genes were regulated by GATA4 than by GATA6 and that even more genes were affected when both factors were absent. These findings also provide a possible answer to our initial question regarding the compensatory role that GATA4 and GATA6 have in the ovary and suggest that genes involved in the final stages of follicle maturation might be controlled by both GATA4 and GATA6, an observation that is supported by the fact that many genes are only affected by the absence of both factors. This may represent an evolutionary adaptation to guarantee the normal development of preovulatory follicles. In addition, our findings demonstrated that the expression of the FSHR decreases only when GATA4 expression was targeted (G4^gcko or G4/6^gcko) but not in the absence of GATA6 alone. Because the FSHR is absolutely essential for the differentiation of GCs, this finding may explain the predominant role GATA4 in ovarian folliculogenesis.

Deletion of GATA4 and GATA6 in GCs demonstrated not only the crucial role that these factors have in ovarian function and female fertility but also offer a unique experimental paradigm to examine genes and pathways involved in the regulation of antral follicle formation. In this regard, this report indicates that the main and probably one of the early defects occurring in GATA-deficient follicles is the inhibition of proliferation and differentiation programs needed for the formation of large preovulatory follicles. A direct consequence of these 2 actions is the lack of follicular antrum formation. A key step in antrum formation is the production of hyaluronan and Vcan, both of which generate an osmotic gradient that draws fluid from the thecal vasculature (63). In the absence of GATA factors, no changes in the expression of hyaluronan synthases were observed. However, the expression of Vcan was abolished. Because Vcan is a large proteoglycan that cross-links with hyaluronan (64, 65), it may contribute greatly to the osmotic potential of the follicular fluid and to the formation of the antrum. Therefore, lack of Vcan expression may explain, at least in part, the lack of antral follicle formation observed in GATA conditional knockout animals.

Limited information is available regarding the transcriptional defects that lead to the halt in folliculogenesis observed in animals lacking FSHβ, FSHR, or IGF-I (66–68). The lack of GATA4 and GATA6 in GCs decreases FSHR expression and increases genes known to reduce IGF-I receptor signaling. Therefore, the genome-wide changes in steady-state levels of mRNAs observed in GATA4/6^gcko animals could shed some light on the mechanisms involved in the deregulation of the folliculogenesis process in the absence of FSH and/or IGF-I signaling. It should also be mentioned that although GATA4 and GATA6 are transcription factors, indirect and nontranscriptional effects may account for the phenotypes observed in the absence of these factors in GCs.

In conclusion, our results suggest that GATA4 regulates, directly or indirectly, a greater number of genes than GATA6. However, because more mRNAs are affected by the absence of both factors, we propose that GATA4 and GATA6 functionally compensate for each other during GC differentiation.