The mammalian ovary is unique in that its reproductive life span is limited by oocyte quantity and quality. Oocytes are recruited from a finite pool of primordial follicles that are usually exhausted from the ovary during midadult life. If regulation of this pool is perturbed, the reproductive capacity of the ovary is compromised. TAF4B is a gonad-enriched subunit of the TFIID complex required for female fertility in mice. Previous characterization of TAF4B-deficient ovaries revealed several reproductive deficits that collectively result in infertility. However, the etiology of such fertility defects remains unknown. By assaying estrous cycle, ovarian pathology, and gene expression changes in young Taf4b-null female mice, we show that TAF4B-deficient female mice exhibit premature reproductive senescence. The rapid decline of ovarian function in Taf4b-null mice begins in early postnatal life, and follicle depletion is completed by 16 wk of age. To uncover differences in gene expression that may underlie accelerated ovarian aging, we compared genome-wide expression profiles of 3-wk-old, prepubescent Taf4b-null and wild-type ovaries. At 3 wk of age, decreased gene expression in Taf4b-null ovaries is similar to that seen in aged ovaries, revealing several molecular signatures of premature reproductive senescence, including reduced Smc1b. One significantly reduced transcript in the young TAF4B-null ovary codes for MOV10L1, a putative germline-specific RNA helicase that is related to the Drosophila RNA interference protein, armitage. We show here that Mov10l1 is expressed in mouse oocytes and that its expression is sensitive to TAF4B level, linking TAF4B to the posttranscriptional control of ovarian gene expression.
INTRODUCTION
The regulation of aging of the mammalian ovary is a complex process and unique among diverse organ systems. A key factor in determining the length of female mammalian reproductive life span is the number of primordial follicles present in the ovary at birth and the maintenance of this finite pool during early postnatal life. In women, ovarian life span is not always directed by chronological age [1]. Often, young women seeking treatment at fertility clinics display characteristics of advanced ovarian age, such as reduced primordial follicle pool and elevated follicle-stimulating hormone (FSH) level. When the follicle reserve is low in women, oocyte quality is low regardless of chronological age, resulting in an increased number of chromosomal abnormalities and miscarriages [2]. Diminished oocyte quality in older women seems to be the major hurdle in conception. A previous study demonstrated that older women could carry and bear children just as well as younger women if oocyte quality were controlled using oocytes donated by younger women [3]. Results of studies in mice are in agreement with observations in humans of diminished oocyte quality with advanced age and reveal that age-related increases in aneuploidy may be attributed to reduced transcript number of spindle assembly checkpoint genes [4]. Taken together, these data suggest that declining fertility with age reflects the declining quality of the oocyte and a depleted follicle pool.
Recent gene expression studies in mice have elucidated age-related gene expression changes in whole ovaries as well as oocyte-specific differences [5, 6]. Consistent with aging in other organs, aged ovaries showed increased levels of genes involved in immune functions and decreased levels of those associated with mitochondrial electron transport, but the majority of transcript changes were related to ovary-specific processes. These results highlight the unique modes of aging found within the ovary. A second important result from molecular profiling of aged oocytes was the alterations in expression patterns of genes encoding proteins involved in chromatin structure, genome stability, and RNA helicases. Thus, the regulation of gene expression might become increasingly impaired as oocytes age [5].
Many of the signaling factors controlling ovary function and reproductive aging are well known, but less is known about the mechanisms of transcriptional integrators and effectors of specific oogenic transcriptional networks. Recent characterization of the general transcription machinery has revealed gonad-specific or -enriched variants of general transcription factors required for gametogenesis in diverse organisms [7–9]. TFIID is a large multiprotein complex that includes the TATA-box binding protein (TBP) and numerous TBP-associated factors (TAFs), which function collectively as core promoter recognition factors and coactivators [10–12]. One variant general transcription factor that is essential for multiple aspects of mammalian gametogenesis is the gonad-enriched TAF4B subunit of TFIID. In the mouse, TAF4B is required for proper female and male fertility. Although dispensable for the first round of spermatogenesis in the mouse testis, TAF4B is required for subsequent rounds and maintenance of spermatogenesis in the early adult [13]. In contrast, female mice that do not express Taf4b are completely infertile because of multiple defects of oogenesis, such as a reduced primordial follicle pool, increased granulosa cell apoptosis, decreased granulosa cell proliferation, and increased FSH levels recorded as early as 23 days of age [14–16]. Taf4b heterozygotes do not show an intermediate phenotype but are comparable to wild type with respect to fertility. Pseudopregnant Taf4b-null mice are able to successfully maintain pregnancy of implanted wild-type embryos, pointing to impairment before implantation. There also have been documented developmental defects in the oocytes derived from Taf4b knockout mice, which exhibit decreased ability to complete meiosis I, extrude a polar body, and cleave properly following fertilization [14]. Ovulation in the Taf4b-null mouse has been assessed under multiple conditions of hormonal stimulation, age, and copulation. The data are not completely consistent. However, there appears to be a reduced level of ovulation in Taf4b-null mice, although the underlying cause is not clear [14]. Conflicting reports of Taf4b-target gene expression and localization of TAF4B in ovarian cell types exist, but it is becoming apparent that the age at which these are assayed is of great importance. Taken together, these studies suggest that rather than play a very specific role in female fertility, TAF4B executes several related functions that coordinate multiple aspects of female fertility, including those regulated by the somatic and oocytic compartments of the ovary.
Poor oocyte quality, reduced primordial follicle pool, and elevated FSH in young Taf4b-null mice lead us to hypothesize that these mice may be experiencing aspects of premature reproductive senescence. Many of the initial defects observed in the adult TAF4B-null ovary, such as decreased expression of several genes important in folliculogenesis, are not observed in the prepubescent ovary, and cursory examination of the ovary suggests that the Taf4b-null phenotype worsens with age. Initial characterization of Taf4b-null mice included a microarray that compared wild-type and Taf4b-null ovarian gene expression in adult mice, at which point TAF4B-null ovaries are largely depleted of follicles and granulosa cell proliferation is impaired [15]. To examine further the potential role of TAF4B in the transcriptional control of ovarian life span regulation, we compared the estrous cycle, ovarian development, and gene expression profiles between prepubertal and adult TAF4B-null ovaries. Here, we show that in the absence of TAF4B, female mice prematurely enter and exit the estrous cycle compared to their corresponding wild-type counterparts and develop hemorrhagic cysts. Candidate gene expression assays showed that disrupted expression of a negative regulator of activin, follistatin (Fst), occurred between 2 and 3 wk of age in the TAF4B-null ovary and pituitary. Components of the AKT/FOXO/PTEN ovarian survival pathway were examined between juvenile and adult mice to determine if their misexpression was involved in TAF4B-null ovarian cell death. Genome-wide whole-ovary expression profiling at 3 wk of age revealed the reduced expression of NACHT, leucine-rich repeat and PYD-containing (NLRP or NALP) gene family members, and oocyte-specific genes that have been shown to change their expression levels in aged ovaries [5] as well as Obox gene family members, which are exclusively expressed in germ cells [17]. Smc1b and Mad2l1, important meiotic genes that show decreased expression with human maternal aging, also are decreased, as is the putative germline-specific RNA helicase, Mov10l1. We do not observe a global down-regulation of oocyte-specific genes, but the gene expression changes we see indicate that young TAF4B-null oocytes may be prematurely aged. In addition, these data suggest that deregulated TAF4B-dependent events may underlie premature reproductive senescence and infertility in women.
MATERIALS AND METHODS
Animals
Wild-type and Taf4b-null or heterozygous mice were generated by mating heterozygous Taf4b+/− male and female mice as described previously [15]. Offspring were genotyped by PCR analysis of tail-snip genomic DNA amplifying the region targeted by homologous recombination. Daily vaginal smear cytology was performed to determine estrous cycle stage. Mice were classified as in diestrus, proestrus, or estrus based on the majority cell type present in the vaginal sample. Estrous cycle entry was deemed to be opening of the vagina and presence of primary leukocytes in the sample. Samples in prepubescent mice were not taken if the vagina had not opened. No mouse was cycled for more than 27 consecutive days. All animal protocols were reviewed and approved by Brown University Institutional Animal Care and Use Committee and were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals.
Ovarian Histology
Ovaries were removed, cleaned of excess fat and bursal sac, and fixed in Bouin solution overnight. Standard paraffin embedding, sectioning, and hematoxylin-and-eosin staining were performed. Images were taken using the Scanscope CS (Aperio).
Ovarian Cell Isolation
Oocyte isolation.
Oocytes were collected from 8-wk-old wild-type mice as described by Ebert et al. [18]. Ovaries were removed, cleaned of excess fat and bursal sac, washed in PBS, and agitated in an enzyme solution (0.1% collagenase type 3 [Worthington], 0.02% DNase type I [Worthington], and 0.1% trypsin [Invitrogen] in Hanks balanced salt solution [Invitrogen]), for 30 min at 37°C. Ovaries were dispersed using a 20-G needle, the solution was diluted with Dulbecco modified Eagle medium/Ham F-12 (DMEM-F12), and oocytes free of cumulus cells were collected. Oocytes were washed once with PBS and resuspended in Trizol reagent (Bio-Rad) for subsequent RNA isolation. Ovaries of three mice (∼120 oocytes) were combined per sample.
Granulosa cell isolation.
Ovaries from 8-wk-old wild-type mice were removed, cleaned of excess fat and bursal sac, and placed in DMEM-F12. Granulosa cells were liberated by pricking with a 26-G needle. Resulting granulosa cell suspensions were passed through a 40-μm nylon filter to remove large tissue clumps and oocytes. Purified granulosa cells were washed once in PBS and resuspended in Trizol reagent for RNA isolation or in extraction buffer (0.2 M KCl, 0.1 M Tris [pH 8], 0. 2 mM ethylenediaminetetra-acetic acid [EDTA], 0.1% NP-40, 10% glycerol, and 1 mM PMSF) for protein isolation. Granulosa cells of two mice were combined per sample.
Western Blot Analysis
Whole ovaries were Dounce homogenized and soluble protein extracted in low-salt protein extraction buffer (0.2 M KCl, 0.1 M Tris [pH 8], 0.2 mM EDTA, 0.1% NP-40, 10% glycerol, and 1 mM PMSF). Protein concentration was determined by Bradford assay, and equivalent amounts of protein were loaded per lane. The following antibodies were used: from Cell Signaling Technologies, FOXO1 2880, Phospho-FOXO1 (Thr24)/FOXO3a (Thr32) 9464, FOXO3 9467, Phospho-FOXO3 (Ser253) 9466, Phospho-FOXO3 (Ser318/321) 9465, FOXO4 9472, AKT 9272, Phospho-AKT (Ser473) 9271, Phospho-AKT (Thr308) 9275, PTEN 9552, and Phospho-PTEN (Ser380/Thr382/383) 9554; from BD Transduction Laboratories, TAF4A (TAF135) 612054; from Novus Biologicals, β-Tubulin RB-9249-P; and from Santa Cruz Biotechnology, CDKN1B (C-19), sc-528.
Quantitative RT-PCR
Total RNA was isolated from indicated tissues using Qiagen RNeasy micro kits, and concentration was determined using a Nanodrop (Thermo Scientific). Twenty microliters of cDNA were prepared from approximately 500 ng of total RNA using iScript cDNA Synthesis Kit (Bio-Rad), according to the manufacturer's directions, using a mix of oligo(dT) and random primers. Real-time PCR reactions were performed in triplicate using 2 μl of cDNA template, ABI Power SYBR green PCR master-mix, and gene-specific primers in the ABI 7500 Real Time PCR System. Dissociation curves were generated, and PCR products were run on an agarose gel to verify amplification of a single product. Quantitative PCR data were analyzed by the ΔΔCt method. Relative expression levels were normalized to 18S ribosomal RNA levels to correct for equivalent total RNA levels. Primer sequences can be found in Supplemental Table S1 (all Supplemental Data are available online at www.biolreprod.org).
Microarray
Total RNA from nine 3-wk-old mice (n = 3 wild-type, 3 Taf4b-heterozygous, and 3 Taf4b-null mice) was obtained as described above. RNA quality was checked using a Bioanalyzer (Agilent), and concentration was determined using a Nanodrop. Three-hundred nanograms of each RNA sample were used in the Affymetrix Whole-Transcript Sense Target Labeling Assay (Rev 3), followed by hybridization to a GeneChip Mouse Gene 1.0 ST Array. Nine GeneChips were used to provide biological triplicates of each genotype. The Affymetrix Expression Console (v 1.1) was used to normalize data and determine signal intensity (RMA-Sketch). Analysis was performed in Microsoft Excel, and DAVID Bioinformatic software was used to identify gene expression pathways [19].
Small Interfering RNA-Mediated Gene Knockdown
The TAF4A and TAF4B SMARTpool siRNAs and ON-TARGETplus Non-Targeting Pool (mouse) were purchased from Dharmacon and transfected into the mouse GC1 cell line using Dharmafect reagent 1. Optimization was performed according to the manufacturer's instructions. Cells were harvested at 60 h posttransfection.
RESULTS
Estrous Cycle Disruption in Taf4b-Null Mice
High FSH levels are often associated with irregular cycling in mouse models of infertility and in both premature ovarian failure and menopause in women. Taf4b-null mice exhibit high levels of FSH as early as 3 wk of age; therefore, we set out to characterize the regularity of their estrous cycle. Daily examination of cell type by vaginal smear cytology revealed that Taf4b-null females have irregular cycling; a representative sample is shown in Figure 1A. Initial cycling trials, lasting from 11 to 27 days, were completed with sexually mature mice from 6 wk to 7 mo of age (Taf4b−/−, n = 7; Taf4b+/−, n = 9; Taf4b+/+, n = 5). While all heterozygous and wild-type mice cycled appropriately, all Taf4b-null littermates were characterized as being in persistent estrus, exhibiting at least one period of more than 4 days stalled in estrus. Persistent estrus usually accompanies the natural reproductive senescence occurring around 15 mo (465 days) in wild-type C57BL/6 mice [20]. Surprisingly, 6-wk-old Taf4b-null mice displayed relatively normal cyclicity during the first week of sample collection, whereas older Taf4b-null mice spent the vast majority of time in estrus with sporadic and brief reentry into the estrous cycle.
FIG. 1.
Estrous cycle disruption in Taf4b-null mice. Vaginal smear cytology was performed daily to assess the estrous stage of Taf4b-null (−/−) and their wild-type (+/+) and heterozygous (+/−) littermates. A) Representative samples of wild-type and Taf4b-null mice, assessed daily, starting at 6 and 18 wk of age. Each block indicates the stage of the mouse for that day. Days indicating persistent estrus are underlined. B) Summary of wild-type (Taf4b+/+) and Taf4b-null (Taf4b−/−) estrous cycle with age at onset of puberty at 24.5 ± 1.3 days for Taf4b-null mice (n = 4) and 29.8 ± 0.5 days for wild-type mice (n = 4). Taf4b-null mice (n = 3) enter persistent estrus at 50 ± 1.5, whereas wild-type mice do not enter until approximately 465 days (15 mo) [20].
Upon discovering that the younger Taf4b-null mice could temporarily cycle normally, we set out to better characterize estrous cycling at the onset of estrus. We examined litters beginning at 3 wk (TAF4B−/−, n = 4; TAF4B+/−, n = 4; TAF4B+/+, n = 4), 4 wk (TAF4B−/−, n = 1; TAF4B +/−, n = 1), and 5 wk (TAF4B−/−, n = 2; TAF4B−/−, n = 1; TAF4B+/+, n = 1) of age and found that females did not remain in estrus for periods of longer than 3 days until 6–7 wk of age, with persistent estrus occurring at Postnatal Day (pnd) 50 ± 1.5 (Fig. 1B). The timing of entry into estrus also is significantly earlier for Taf4b-null mice, with onset at pnd 24.5 ± 1.3, compared to onset at 28.2 ± 1.1 for Taf4b-heterozygotes and at pnd 29.8 ± 0.5 for wild types. Thus, defects in prepubertal Taf4b-null mice most likely lead to both accelerated activation and senescence of the TAF4B-deficient ovary. These data reveal a small window of normal cycling for Taf4b-null mice from an early onset at pnd 24.5 ± 1.3 to a state of persistent estrus beginning at pnd 50 ± 1.5 (Fig. 1B).
Histopathological Examination of TAF4B-Null Ovaries
Disruption of the estrous cycle in Taf4b-null mice suggested that follicular development may be similarly deregulated. Our previous study included a comprehensive count of follicle stages at 3, 6, and 12 wk of age and revealed that TAF4B-null ovaries contain reduced numbers of primordial follicles and increased atretic follicle number at each time point, whereas recruitment to the primary follicle stage is comparable to that in wild-type ovaries. We did not set out to uncover numerical follicle-stage data in the present study; instead, we turned to histopathological examination of the TAF4B-null ovary to capture its rapid degeneration. To better understand the effects on the ovary during the progression from normal to irregular cycling, we harvested TAF4B-null and matched wild-type ovaries between 3 and 17 wk of age for histological staining. From 3 wk of age (before puberty and hormonal signaling has commenced in wild-type mice) until sexual maturity at 8 wk, ovaries from Taf4b-null and wild-type mice were indistinguishable (Fig. 2, A–C). At 8 wk, TAF4B-null ovaries contained healthy antral follicles, which is interesting given that these mice were beginning to cycle irregularly (Fig. 2C). Strikingly, only 1 wk later, at 9 wk of age, the TAF4B-null ovaries displayed obvious signs of premature depletion of follicles (Fig. 2D). By 10 wk, the TAF4B-null ovary contained very few oocytes (Fig. 2E). By 12 wk, hemorrhagic cysts became evident (Fig. 2F), and many fragmented oocytes were present in degenerating follicles (Supplemental Fig. S1). The 17-wk-old TAF4B-null ovary appeared to be devoid of all developing follicles and oocytes (Fig. 2G). In contrast, the wild-type ovary still contained some healthy follicles at 1 yr of age and beyond (Fig. 2H). We also observed an increased occurrence of hemorrhagic cysts in the Taf4b-null mice. TAF4B-null ovaries were observed with one, two, or multiple cysts, with the earliest identified at 16 wk of age. The ovaries collected from an aged wild-type mouse contained multiple cysts resembling those of the younger Taf4b-null animals (Supplemental Fig. S2). Taken together, these data indicate that follicular development is relatively intact, whereas the Taf4b-null mice are cycling normally and premature depletion of all follicles occurs rapidly by 17 wk. Additionally, early cyst formation closely follows the onset of persistent estrus.
FIG. 2.
Progressive follicle loss in TAF4B-null ovaries. A–G) Gross ovarian histology using hematoxylin-and-eosin staining of Bouin-fixed ovaries at 3 (A), 6 (B), 8 (C), 9 (D), 10 (E), 12 (F), and 17 (G) wk of age from Taf4b−/− mice and their wild-type (Taf4b+/+) littermates shows the progressive and rapid loss of maturing oocytes. H) One-year-old wild-type ovary serves as a reference for normal aging in the C57BL/6 mouse line and shows the presence of both developing follicles and a hemorrhagic cyst. Asterisk indicates hemorrhagic cyst. Bar = 500 μm.
Progressive Loss of the AKT Survival Pathway in TAF4B-Deficient Ovaries
The rapid depletion of follicles led us to investigate the possible role of a disrupted AKT survival pathway in the TAF4B-deficient ovary. At 4 wk, all of the total and phosphorylated FOXO proteins tested were present at near-equivalent levels between the wild-type and TAF4B-null ovary (Fig. 3A). In 16-wk-old TAF4B-null ovaries, FOXO1 and phosphorylated FOXO3 were almost undetectable, whereas total FOXO3 and FOXO4 were present but with reduced levels. Although almost complete loss of FOXO3 phosphorylation was found in the 16-wk-old TAF4B-null ovary, only an approximately 50% loss of total FOXO3 protein was found, suggesting that at 16 wk, phosphorylation of FOXO3 is reduced in TAF4B-null ovaries. Of the factors tested that are upstream of FOXO proteins in this survival pathway, we found a slight decrease in total AKT at 16 wk in the TAF4B-null ovary, accompanied by a very slight decreases in phosphorylated AKT (Fig. 3B). PTEN and p-PTEN were equivalent between genotypes at both 4 and 16 wk. Interestingly, CDKN1B (also known as p27) levels were equivalent at 4 wk and elevated in the 16-wk-old knockout. At the earliest time point examined, defects in the AKT survival pathway were not prevalent, and these later defects may be the consequence, rather than the cause, of premature senescence of the TAF4B-null ovary.
FIG. 3.
Age-dependent reduction of key ovarian regulatory proteins in TAF4B-null ovaries. Western blot analysis was used to examine protein levels in the FOXO (A), PTEN/AKT pathway (B), and TFIID (C) using whole-ovary extracts collected from 4- and 16-week-old TAF4B−/− and TAF4B+/+ ovaries. An equivalent amount of whole-ovary protein extract was loaded per lane, confirmed by tubulin level (A).
To examine the relative levels of TFIID components between 4 and 16 wk, we used antibodies against TAF4, TAF4B, and TBP. Loss of TAF4B did not change levels of the paralogous TAF4 subunit in the ovary, which remained constant between genotypes at both time points. However, TBP was slightly reduced at 16 wk in the TAF4B-null ovary (Fig. 3C). Interestingly, the level of TAF4B in the wild-type ovary was elevated at 4 wk compared to 16 wk. Whereas 4 wk of age is just at, or before, the onset of estrus in a wild-type mouse, these mice were cycling regularly at 16 wk, so if TAF4B protein level were dynamic through the estrous cycle, this lower level may have been a result of examining the mouse in a phase during which TAF4B normally is low. To this end, we collected ovaries at different points during the estrous cycle and found that Taf4b mRNA and TAF4B protein expression do not change between estrus, diestrus, and proestrus (Supplemental Fig. S3). The elevated levels of TAF4B at 4 wk in wild-type ovaries may suggest a greater need for TAF4B action at early stages of ovarian function, a lack of which leads to premature senescence.
Follistatin mRNA Down-Regulation in TAF4B-Null Ovary and Pituitary
Similar to FOXO and AKT expression, many TAF4B-target genes were reduced in older, but not in younger, TAF4B-null ovaries. Although young TAF4B-null ovaries appeared to be healthy, without a complete TFIID complex there may be underlying transcriptional deregulation from early time points, which ultimately may lead to premature reproductive senescence. Previous data implicating TAF4B in control of Fst expression, combined with high FSH levels, led us to examine Fst expression between Taf4b-null mice and wild-type littermates. To determine if TAF4B is necessary for proper Fst expression, we performed quantitative RT-PCR on ovary-derived RNA. At 7 mo of age, when the TAF4B-null ovary contains no growing follicles, Fst mRNA levels were decreased between three- and sixfold compared to wild-type and heterozygous littermates (Fig. 4A). In addition to granulosa cells of the ovary, the anterior pituitary also produces Fst, which acts there as a paracrine signaling factor [21]. At 3 wk of age, Fst mRNA levels were significantly reduced by threefold in the TAF4B-null ovary (Fig. 4B) and by twofold in the pituitary (Fig. 4C). However, Fst misregulation is probably not the sole cause of the TAF4B-null female phenotype, because at 2 wk of age, Fst levels in the TAF4B-null ovary (Fig. 4D) and pituitary (data not shown) were equivalent to those of its heterozygous littermates. Regulation of other key pituitary factors, Fshb, Inhbb, and Foxl2, seemed to be largely intact (Fig. 4C). Follistatin, a key component in the ovary-pituitary axis, appeared to be deregulated in the Taf4b-null mouse, both in the ovary and in the pituitary, and may contribute to elevated levels of FSH in Taf4b-null mice, but its normal expression level at 2 wk of age suggests that TAF4B may regulate follistatin in a temporal fashion.
FIG. 4.
Defects in TAF4B-null follistatin expression. RNA samples were obtained from 7-mo-old ovaries (A), 3-wk-old ovaries (B), 3-wk-old pituitaries (C), and 2-wk-old ovaries (D). Quantitative RT-PCR was used to compare gene expression between Taf4b-null cDNA to that derived from heterozygous and/or wild-type littermates. Data were normalized to 18S rRNA present in the cDNA sample. Relative mRNA fold-difference is displayed after setting the wild-type or heterozygote value to one. Error bar represents the SD (A–C) or SEM (D; n = 2 TAF4B+/− and 5 TAF4B−/−). Fshb, follicle-stimulating hormone subunit beta; Fst, follistatin; Foxl2, forkhead box protein L2; Inhbb, inhibin beta B.
Genome-Wide Expression Profiles of Immature TAF4B-Null Ovaries
A candidate gene approach revealed surprisingly few gene expression changes at 3 wk; thus, we likely missed some novel aspect of transcriptional deregulation within the prepubertal TAF4B-null ovary. To this end, we performed genome-wide expression profiling using microarrays with total RNA derived from the ovaries of 3-wk-old wild-type, heterozygous, and Taf4b-null mice in triplicate. By examining this time point, when the ovaries appeared to be healthy, we hoped to obtain an unbiased group of gene expression differences that result more directly from the absence of TAF4B and less from the complete loss of follicles. Of the 28 853 genes represented on the Affymetrix Mouse GeneST GeneChip, 1488 nonredundant genes were present with significant (P < 0.05) changes ranging from to 2.7- to 0.27-fold. Complete microarray results can be found at the NCBI GEO under accession number GSE15228. The majority of the gene changes were small. Only 25 genes were expressed in the TAF4B-null ovary at levels 1.5-fold greater than in the wild-type ovary, and 60 genes were down-regulated by 1.5-fold or more. Genes with less than a 1.5-fold difference from wild type were not further analyzed.
Previous expression studies of aging in the ovary have found that the majority of gene changes are reductions [6]. Among the classes down-regulated are mitochondrial and those involved in oxidative stress, DNA metabolism, as well as oocyte-specific genes. Up-regulated genes tended to be involved in receptor binding and immune response/complement activation. The expression changes in TAF4B-null ovaries overlap with the genes disrupted in ovarian aging. Functional annotation analysis of TAF4B-null versus wild-type 3-wk-old ovaries revealed that the main group of down-regulated genes in the TAF4B-null ovary belongs to the Homeobox family and related oocyte transcription factors called the Obox family (Table 1). A second down-regulated group includes gene products with nucleotide-binding activity. Within this group are Moloney leukemia virus 10-like 1 (Mov10l1) and multiple Nlrp genes. The functions of the Mov10l1 and Nlrp genes have not been determined in the ovary. However, they were identified in screens for down-regulated factors in ovary aging [5], and preliminary data hint at important roles in oocyte development and aging [22].
TABLE 1.
Functional characterization of decreased transcripts in TAF4B-null ovaries at 3 wk.
To confirm expression changes identified by the microarray, we used quantitative RT-PCR. Complementary DNA was made from the 3-wk-old ovarian RNA used in the microarray representing biological triplicates for each genotype. Complementary DNA for each genotype was pooled (n = 3 samples/genotype), and quantitative RT-PCR was performed on several interesting genes with roles in folliculogenesis or premature aging (Table 2). We found the largest reduction was of Mov10l1 (Fig. 5A), with nearly a sixfold reduction from wild type and fourfold reduction from heterozygotes, followed by follistatin (Fig. 4B) and the Nlrp genes (Fig. 5A). The reduced level of Mov10l1 also was present in the ovaries of 2-wk-old TAF4B-null ovaries (Fig. 5B). The majority of fold-changes in the present study were subtle. By examining functional gene groups, we begin to build a picture of how numerous, small gene changes compound to exhibit the defects in TAF4B-null and the aging ovary. Taken together, these data indicate that TAF4B-deficient oocytes display a molecular signature of premature aging and are consistent with the complete infertility of the Taf4b-deficient female mice.
TABLE 2.
Fold changes of selected ovary transcripts between 3-wk TAF4B-null and wild-type (WT) mice by microarray and qRT-PCR.
FIG. 5.
Confirmation of ovarian gene expression changes from microarray in 3-wk-old TAF4B-null ovaries versus wild-type and heterozygous littermates. A) Complementary DNA from the biological triplicate samples used in the microarray were pooled by genotype (Taf4b−/−, Taf4b+/−, and Taf4b+/+) and used in quantitative RT-PCR using gene specific primers. Data was normalized to 18S rRNA present in the cDNA sample. Relative mRNA fold-difference is displayed after setting the TAF4B+/+ sample to one. Error bar represents the SD. B) Mov10l1 transcript levels were measured using quantitative RT-PCR from cDNA of 2-wk-old ovaries: TAF4B−/−, n = 5; TAF4B+/−, n = 2. Data were normalized to 18S rRNA present in the cDNA sample. Relative mRNA fold-difference is displayed after setting the TAF4B+/− sample to one. Error bar represents the SEM. **P < 0.001.
Ovarian Mov10l1 expression
Mov10l1 is a very interesting gene that may play a significant role in the TAF4B-null phenotype. Little is known about this putative RNA helicase, but Mov10l1 is the mammalian homolog of the Drosophila protein, armitage, which has been identified as a germline factor required for RNA interference in the ovary [23, 24]. In mammals, there is tissue-specific expression of Mov10l1 and its splice variants in the heart, Champ and Csm. In the mouse testis, Mov10l1 is expressed at its full length of 25 exons [25, 26]. The heart-specific splice variants, Champ and Csm, lack the N-terminal coding sequence, and the majority of these proteins are composed only of the helicase domain [26, 27].
To follow up on the reduction of Mov10l1 in TAF4B-null ovaries, we set out to determine what splice variant is present in the mouse ovary (Fig. 6A). Using quantitative RT-PCR with primers amplifying the N and C terminus regions of Mov10l1, we analyzed wild-type mouse testis, heart, and ovary (Fig. 6B). The testis exhibited very high levels of expression of both the N and C termini, whereas heart expression was limited to the C-terminal domain, as shown previously [26]. The 3-wk-old ovary showed expression of both N and C termini, similar to the testis, but at much lower levels. Although well within the detection limits of the assay, juvenile ovary only expressed approximately 1% of the testis Mov10l1 level. Given that Mov10l1 is proposed to be restricted to germ cells, lower expression in the ovary is not surprising. Unlike the testis, ovaries contain relatively few germ cells and do not have the ability to repopulate, thereby reducing their oocyte number with age. Indeed, we saw an approximately 10-fold reduction in Mov10l1 from the 3-wk- to 1-yr-old ovary, consistent with germ cell reduction as ovaries age (Fig. 6B).
FIG. 6.
Mov10l1 gonad-specific gene expression and TAF4B dependence. A) Full-length Mov10l1 and its splice variants, Champ and Csm. Of the 25 exons comprising full-length Mov10l1, Champ is missing the first 13, whereas Csm is missing the first 15 and contains only the helicase domain. B) Primers specific to the N or C terminus of Mov10l1, as diagrammed in A, were used in quantitative RT-PCR to determine splice variation between adult testis, adult heart, juvenile ovary (age, 3 wk), and aged ovary (age, 1 yr). With the highest expression of both the N and C termini, adult testis was set to 100% expression. Percentage expression versus testis is indicated with 95% confidence intervals from methodological triplicates. C–F) Taf4b (C), Taf4a (D), Mov10l1 (E), and Tbp (F) mRNA expression following siRNA knockdown of Taf4a and Taf4b and using a nontargeting control (NT). Data were normalized to 18S rRNA present in the cDNA sample. Relative mRNA fold-difference is displayed after setting nontargeting value to one. Graphs represent data from two independent trials. Error bar represents the SEM. *P = 0.03.
To determine if Mov10l1 expression is directly dependent on TAF4B, we used an siRNA-mediated knockdown approach in vitro. Because Mov10l1 is expressed in male germ cells, we used the testis germ cell-derived GC1 cell line to evaluate Mov10l1 expression levels upon TAF4B reduction. GC1 cells were transfected with siRNA targeting either Taf4a, Taf4b, or a nontargeting control, followed by 60 h of incubation, at which point target protein levels were reduced by approximately 75% (data not shown) and mRNA levels, while still reduced by approximately 60% (Fig. 6, C and D), were starting to return to normal (24–48 h post-siRNA transfection show the highest Taf mRNA reduction). It has been our experience that knockdown of Taf4b in cell culture causes an increase in Taf4a transcript, which may be a compensatory mechanism (Fig. 6D). Mov10l1 transcript is elevated 1.3-fold upon knockdown of Taf4a and significantly reduced by twofold upon Taf4b knockdown, which suggests that its expression is modulated by TAF4B levels (Fig. 6E). Mov10l1 reduction upon Taf4b knockdown is not the effect of global transcriptional down-regulation, because Tbp levels remained constant between siRNA targets (Fig. 6F). Perhaps a Mov10l1 reduction in TAF4B-null oocytes reduces oocyte quality, leading to the appearance of advanced age.
Data about the expression pattern of Taf4b within the ovary remain somewhat ambiguous. To suggest that loss of Taf4b leads to a reduction of Mov10l1 in the oocyte, we first need to determine the expression levels of Taf4b, Mov10l1, and their paralogs in purified populations of ovarian cell types. Using quantitative RT-PCR, we compared oocytes, granulosa cells and total ovarian RNA from 8-wk-old wild-type mice for expression of TAF4a, Taf4b, Mov10, and Mov10l1 (Fig. 7). Here, we normalized to Actb (beta-actin) as a more appropriate control when comparing oocytes, which stockpile rRNA [28, 29]. Levels of Gdf9, an oocyte-specific transcript, were measured to control for accurate separation of cell types and displayed a 30-fold enrichment over total ovary, whereas detection in granulosa cells was less than 0.09-fold (data not shown). Similarly, Taf4b was detected in both oocytes and granulosa cells, with a slight enrichment in oocytes. Taf4a showed a significant enrichment in oocytes yet was still expressed in granulosa cells. Mov10l1 had comparatively very little expression in granulosa cells (0.06-fold of total ovary), but its expression in oocytes was enriched eightfold over total ovary. Unlike its paralog, Mov10l1, Mov10 was present in granulosa cells with only a slight bias toward oocytes. Although this assay did not examine other ovarian cell types, such as epithelial, luteal, or thecal, we now have evidence that Taf4b is expressed in both the granulosa cell and oocyte compartments of the adult mouse ovary. We also show, to our knowledge for the first time, that Mov10l1 is enriched in mammalian oocytes.
FIG. 7.
Ovarian mRNA expression levels by cell type. Oocyte, granulosa cell (GC), or total ovary RNA was collected from 8-wk-old wild-type mice. Quantitative RT-PCR was performed using gene-specific primers for Taf4b, Taf4a, Mov10l1, and Mov10. Data were normalized to Actb mRNA present in the cDNA sample. Relative mRNA fold-difference is displayed after setting the ovary sample to one. Error bars represent the SEM (oocyte, n = 2; GC, n = 3; ovary, n = 3).
Poor oocyte quality often leads to meiotic defects in older women, resulting in aneuploidy [30]. Several genes with confirmed roles in meiosis have been found to have down-regulated transcripts in oocytes of older women, likely contributing to the higher occurrence of chromosomal segregation abnormalities observed. We compared spindle checkpoint protein, Mad2l1 [31], and kinetochore protein, Bub1 [32], mRNA levels using quantitative RT-PCR in TAF4B-null and heterozygous 2-wk-old ovaries and a 1-yr-old wild-type mouse ovary. Of these genes, only Mad2l1 transcripts were reduced in TAF4B-null ovaries compared to their heterozygous siblings, whereas both Mad2l1 and Bub1 were further reduced in aged, 1-yr-old wild-type ovaries (Fig. 8). Mad2l1 expression decreased 3.3-fold from the 2-wk-old TAF4B heterozygous ovaries to the 1-yr-old ovary, consistent with the trend reported in the published data [33]. Mad2l1 was significantly reduced in TAF4B-null littermate ovaries by approximately 30%. Bub1 showed a similar pattern of expression, with an eightfold reduction from 2 wk to 1 yr of age, but only showed a 10% reduction from heterozygote to TAF4B-null 2-wk-old ovaries. Chromosomal segregation malfunctions can be caused by the premature separation of sister chromatids because of loss of cohesion. Structural Maintenance of Chromosome (SMC) proteins are integral components of the cohesin complex [34]. Whereas SMC1A is involved in mitosis, SMC1B is meiosis-specific and down-regulated with age [35, 36], contributing to aneuploidy. Smc1a expression was equivalent between TAF4B-null and heterozygous 2-wk-old ovaries and the 1-yr-old wild-type ovary (Fig. 8). Expression of Smc1b in 1-yr-old wild-type ovaries was very low, with approximately 45-fold less expression than in the 2-wk-old TAF4B heterozygote. Nearly a 60% reduction occurred in Smc1b from the 2-wk-old Taf4b-heterozygous and Taf4b-null mice. This is one of the larger disparities we have documented at such a young age, and it may contribute to the TAF4B-null phenotype. Taken together, the alterations in gene expression are consistent with the hypothesis that TAF4B promotes the correct temporal expression of oocyte-specific genes required for the regulation of oocyte quality and ovarian life span.
FIG. 8.
Reduced expression of meiotic genes in aged and TAF4B-null ovaries. Ovaries from 2-wk-old Taf4b+/− (n = 2) and Taf4b−/− (n = 5) littermates and a 1-yr-old wild-type mouse were used in quantitative RT-PCR using gene-specific primers for Smc1a, Smc1b, Mad2l1, and Bub1. Data were normalized to 18S rRNA present in the cDNA sample. Relative mRNA fold-difference is displayed after setting the 2-wk-old TAF4B sample to one. Error bar represents the SEM. *P = 0.02, **P < 0.001.
DISCUSSION
Unlike other organ systems, the ovary seems to exhibit two distinct aging pathways: age-related oxidative damage, which can be accounted for by the buildup of reactive oxygen species over time, and the reduction of oocyte-specific transcripts, resulting in diminished oocyte quality. Women well below the age of menopause often present with characteristics of older women and require assisted reproductive technology to conceive. This subset of patients undergoing in vitro fertilization may be experiencing advanced ovarian senescence resulting from acceleration of normal ovarian aging pathways. TAF4B is a major regulator of multiple ovarian processes, and disruption of this transcription factor could be an underlying cause of human infertility. In fact, a recent report has linked the integrity of TFIID and Taf4b expression levels to oocyte quality in women [37].
The reproductive defects observed in Taf4b-null female mice do not mimic the phenotype of any other transgenic mouse model of infertility. No single developmental block exists; instead, TAF4B seems to be playing a critical role in regulating the timing of multiple reproductive processes. The four major disruptions appear to be with 1) the number of primordial follicles in neonates, 2) the endocrine environment of the ovary beginning in prepubescence, 3) the quality of the mature oocyte, and 4) granulosa cell proliferation and survival. Arguably, the most interesting aspect of the phenotype is that whereas TAF4B is essential for normal transcription in the gonad, it does not appear to be absolutely required for the expression of any one gene at all time points.
In an attempt to find underlying factors that may be disrupted in response to elevated FSH levels, we examined the PTEN/AKT/FOXO pathway. Disruption of the components of this pathway in the TAF4B-null ovary could explain the increases in apoptosis and reduced proliferation in response to FSH [16]. If this pathway were misregulated at an early time point in development, it may explain the loss of follicles in the mature TAF4B-null ovaries. At 4 wk, however, PTEN, AKT, and FOXO proteins are present and phosphorylated in TAF4B-null ovaries at levels very similar to those in wild-type mice, indicating that this pathway is largely intact. As expected, FOXO1 is nearly undetectable in the 16-wk-old TAF4B-null ovary given that FOXO1 expression is concentrated in the granulosa cells of growing follicles [38] and, by this time point, the TAF4B-null ovary is nearly devoid of such follicles. There appears to be a reduction in FOXO3 phosphorylation at 16 wk, which supports our observation of increased apoptosis, but the disruptions in this pathway do not precede but, rather, follow ovary deterioration in the TAF4B knockout.
To uncover gene regulation defects underlying the rapid deterioration of the TAF4B-null ovary, we first took a candidate gene approach, evaluating mRNA levels of genes previously found to be affected by TAF4B expression. Our initial microarray [15] compared TAF4B-null and wild-type ovaries from adult mice. Attempts to recapitulate gene expression changes by quantitative RT-PCR in younger mice were largely unsuccessful, because most genes are expressed at similar levels at 3 wk of age. Both chromatin immunoprecipitation (ChIP) and RT-PCR data using cultured granulosa cell lines have identified TAF4B occupancy at target genes such as Ccnd2, Jun, Igfbp3, and Fst [39, 40]. In many cases, genes that have shown differential regulation by TAF4B in cultured cells are expressed normally in the Taf4b-null mouse at younger ages. However, one gene that was found to be affected both in the present and in previous studies is follistatin, which is reduced by threefold in the 3-wk-old TAF4B-deficient ovary [15, 39]. In fact, Fst is up-regulated upon overexpression of TAF4B, and TAF4B, as determined by ChIP, is bound to the Fst promoter in cultured granulosa cells [39]. Follistatin is an important negative regulator of activin, thereby acting to suppress FSH release from the pituitary [41]. FSH is consistently elevated in Taf4b-null mice [14, 16]. The youngest reported measurement of FSH level in the Taf4b-null mouse is 23 days, at which point FSH is elevated [14]. We show here that at 3 wk of age and into adulthood, follistatin is reduced in the Taf4b-null mouse. To see if follistatin levels were reduced at even younger ages, we examined 2-wk-old ovaries and pituitaries and found their levels to be equivalent between those of Taf4b-null and heterozygous mice. We do not currently know the FSH level of Taf4b-null 2-wk-old mice, but it is possible that between 2 and 3 wk of age, follistatin expression is disrupted, leading to the phenotype of elevated FSH. This disruption could stem from a failure of Taf4b-null mice to up-regulate follistatin as the first cohort of follicles approaches the antral stage. Alternatively, follistatin may be down-regulated upon entry into estrus, when the ovary starts to receive FSH signals. Because Taf4b-null mice enter estrus earlier than their wild-type littermates, this decrease may occur prematurely at 3 wk of age. Thus, regulation of gene expression by TAF4B may be more temporal in nature than expected for a general transcription factor.
Young Taf4b-null mice display characteristics of premature ovarian aging, such as persistent estrus, elevated FSH levels, depleted primordial follicle pool, hemorrhagic cysts, fragmented oocytes, poor oocyte quality, and gene expression profiles consistent with advanced oocyte age. Cysts are not prevalent in the C57BL/6 strain, into which the Taf4b-null mice have been backcrossed [42]. In the wild-type MRL mouse strain, however, cysts are common, and the incidence of cyst formation increases with age [42]. The etiology of cyst formation is still speculative at this point, but disruptions of genes involved in diverse gonadal function increase their occurrence. Similar cysts have been reported in ovaries of granulosa cell-specific knockouts of follistatin [43], mice lacking the tissue-bound isoform of follistatin [44], granulosa cell-specific knockouts of Nr5a1 (also known as SF-1) [45], the Cyp19a1 (also known as aromatase) knockout [46], and mice with chronically high luteinizing hormone levels [47].
To uncover underlying changes in gene expression that may lead to the rapid deterioration of the TAF4B-null ovary, we performed microarray analysis using 3-wk-old ovaries from Taf4b-null females and their wild-type and heterozygous siblings. This genome-wide comparison revealed a number of oocyte-specific gene changes that may be indicative of advanced ovarian aging of the TAF4B-null ovary. The most dramatic transcript down-regulated in the knockout, sixfold below that of wild type, was MOV10L1, which is a putative RNA helicase with two splice variants, Csm and Champ, that are involved in heart development [26]. The full-length Mov10l1, isoform 1, was once identified as a testis germ cell-specific gene [25], but expressed sequence tags (ESTs) have since been identified in oocytes in mice, frogs, and humans. MOV10L1 shares a very high degree of homology with the Drosophila protein, armitage [48]. Disruptions in armitage lead to male and female sterility because of a RNA-induced silencing complex (RISC) maturation failure [24]. Here, we show that not only the mouse testis but also the ovary expresses Mov10l1; that its expression is decreased in aged ovaries, presumably because of lower oocyte number; and that its expression is dependent on TAF4B.
An interesting and unexpected finding was the reduction of a number of Nlrp (NACHT, leucine-rich repeat and PYD-containing), also referred to as NALP, genes. The Nlrp family, identified in a screen for genes involved in oocyte aging [5], is composed of 14 members in humans and 20 members in mice, presumably stemming from several gene duplication events [49]. The Nlrp genes belong to a superfamily of NOD-like receptors, which have roles in inflammation and innate immunity. Many of these genes are specifically expressed in the ovary, and loss of function in Nlrp5 (Mater), Nlrp4, Nlrp7, or Nlrp14 has been shown to cause female reproductive failure [49]. In the TAF4B-null ovary, Nlrp4e, Nlrp9c, and Nlrp4b were confirmed by RT-PCR to be reduced by 3.11-, 2.68-, and 1.79-fold, respectively. Interestingly, many of the Nlrp genes are clustered on chromosome 7 [49], very close to the location of the Obox genes, which also are down-regulated in the TAF4B-null ovary. Factor in the germline alpha (FIGLA), a critical oocyte transcription factor, also regulates the expression of several Nlrp genes [50]. Roles for Nlrp genes in oocyte and preimplantation development and innate immunity have been established, but functional data are scarce. Given that expression of these genes is decreased with oocyte age and is regulated by two critical ovarian transcription factors, TAF4B and FIGLA, the Nlrp family may prove to be very interesting with respect to reproduction and aging.
It is evident that the reduced number of primordial follicles in the TAF4B-null oocytes contributes to the very early depletion observed by 17 wk. TAF4B-null ovaries from pnd 1 mice contain the same number of primordial follicles as their fertile littermates, but by pnd 3, this number is reduced [14]. Interestingly, no increase in apoptosis is observed, leading us to question what has happened to these oocytes. A recent study on germ cell loss in the perinatal mouse has highlighted the role of autophagy in the initial wave of germ cell culling that takes place after birth [51]. Indeed, the microarray in the present study showed a slight increase in autophagy-related genes, which may be responsible for the lower number of germ cells in TAF4B-null ovaries, but this supposition requires much further work. Thus, a complex syndrome of ovarian defects that occurs from shortly after birth and into adulthood culminates in premature reproductive senescence and infertility in the absence of TAF4B.
The emergence of Smc1b as a disrupted gene in TAF4B-null ovaries offers new mechanistic possibilities to explain infertility. Reduction of Smc1b results in cohesion complexes that are unable to maintain attachments between sister chromatids, leading to premature segregation that results in oocyte death. Targeted deletion of Smc1b yields infertile mice, and deletion of Spo11, an important factor required for meiotic chromosome synapsis, results in a very similar phenotype to that of Taf4b-null females [34, 52].
The precise mechanism of how loss of Taf4b in mice leads to advanced ovarian age remains to be determined. The microarray performed in the present study on prepubescent mice provides a new set of possible TAF4B-target genes involved in regulation of ovarian life span. Oocyte aging is not only regulated by factors within the germ cell but also profoundly influenced by the follicular microenvironment [53, 54]. Elucidating how TAF4B acts in granulosa cells, oocytes, or both to maintain proper timing of ovarian events will yield insight regarding normal regulation of ovarian aging and may uncover new therapeutic opportunities for women who experience premature reproductive senescence and infertility.
Acknowledgements
We thank Mary Hixon, Aron Gyuris, and Diana Donovan for helpful suggestions during the course of these studies and comments on the manuscript. We thank Paula Weston and Michele Gardner of the Molecular Pathology Core facility at Brown for ovarian histology.
