Gonadotrope-specific Deletion of Dicer Results in Severely Suppressed Gonadotropins and Fertility Defects*

Background: DICER mediates microRNA production, and its functional role in gonadotropes is not known. Results: Deletion of Dicer in gonadotropes leads to suppression of gonadotropin synthesis and secretion and results in fertility defects. Conclusion: DICER-dependent microRNAs are critical for gonadotropin homeostasis and fertility. Significance: Understanding how microRNAs regulate gonadotropin homeostasis provides a new approach to enhance or block fertility in mammals. Pituitary gonadotropins follicle-stimulating hormone and luteinizing hormone are heterodimeric glycoproteins expressed in gonadotropes. They act on gonads and promote their development and functions including steroidogenesis and gametogenesis. Although transcriptional regulation of gonadotropin subunits has been well studied, the post-transcriptional regulation of gonadotropin subunits is not well understood. To test if microRNAs regulate the hormone-specific gonadotropin β subunits in vivo, we deleted Dicer in gonadotropes by a Cre-lox genetic approach. We found that many of the DICER-dependent microRNAs, predicted in silico to bind gonadotropin β subunit mRNAs, were suppressed in purified gonadotropes of mutant mice. Loss of DICER-dependent microRNAs in gonadotropes resulted in profound suppression of gonadotropin-β subunit proteins and, consequently, the heterodimeric hormone secretion. In addition to suppression of basal levels, interestingly, the post-gonadectomy-induced rise in pituitary gonadotropin synthesis and secretion were both abolished in mutants, indicating a defective gonadal negative feedback control. Furthermore, mutants lacking Dicer in gonadotropes displayed severely reduced fertility and were rescued with exogenous hormones confirming that the fertility defects were secondary to suppressed gonadotropins. Our studies reveal that DICER-dependent microRNAs are essential for gonadotropin homeostasis and fertility in mice. Our studies also implicate microRNAs in gonadal feedback control of gonadotropin synthesis and secretion. Thus, DICER-dependent microRNAs confer a new layer of transcriptional and post-transcriptional regulation in gonadotropes to orchestrate the hypothalamus-pituitary-gonadal axis physiology.

Pituitary gonadotropins follicle-stimulating hormone and luteinizing hormone are heterodimeric glycoproteins expressed in gonadotropes. They act on gonads and promote their development and functions including steroidogenesis and gametogenesis. Although transcriptional regulation of gonadotropin subunits has been well studied, the post-transcriptional regulation of gonadotropin subunits is not well understood. To test if microRNAs regulate the hormone-specific gonadotropin ␤ subunits in vivo, we deleted Dicer in gonadotropes by a Cre-lox genetic approach. We found that many of the DICER-dependent microRNAs, predicted in silico to bind gonadotropin ␤ subunit mRNAs, were suppressed in purified gonadotropes of mutant mice. Loss of DICER-dependent microRNAs in gonadotropes resulted in profound suppression of gonadotropin-␤ subunit proteins and, consequently, the heterodimeric hormone secretion. In addition to suppression of basal levels, interestingly, the post-gonadectomy-induced rise in pituitary gonadotropin synthesis and secretion were both abolished in mutants, indicating a defective gonadal negative feedback control. Furthermore, mutants lacking Dicer in gonadotropes displayed severely reduced fertility and were rescued with exogenous hormones confirming that the fertility defects were secondary to suppressed gonadotropins. Our studies reveal that DICER-dependent microRNAs are essential for gonadotropin homeostasis and fertility in mice. Our studies also implicate microRNAs in gonadal feedback control of gonadotropin synthesis and secretion. Thus, DICER-dependent microRNAs confer a new layer of transcriptional and post-transcriptional regulation in gonadotropes to orchestrate the hypothalamus-pituitary-gonadal axis physiology.
Gonadotropes are the least abundant cell type, representing only 5-6% of all cells in the anterior pituitary (1,2). They synthesize and secrete two heterodimeric glycoprotein hormones luteinizing hormone (LH) 2 and follicle-stimulating hormone (FSH) that are essential for gonadal growth, gametogenesis, and steroidogenesis (1,2). Both these gonadotropic hormones consist of a common ␣and a hormone-specific ␤-subunit that are encoded by distinct genes (1,2). The development of the gonadotrope occurs via a combinatorial action of several transcription factors (3)(4)(5)(6)(7)(8). In the embryonic mouse pituitary, the gene encoding the common ␣ subunit (Cga) is expressed around E 11.5, and subsequently the Lhb-and Fshb-encoding genes are first detected between E 15.5 and E 17.5 (9).
Gonadotropin releasing hormone (GnRH) and gonadal steroids regulate the coordinated expression of the glycoprotein hormone subunits in the mature gonadotrope (1,10,11). GnRH activates the GnRH receptor in gonadotropes, a seven membrane-spanning G-protein-coupled receptor, to maintain the basal expression of gonadotropin subunit gene transcription (12). Activins, members of transforming growth factor-␤ superfamily, regulate Fshb transcription via SMAD-and FOXL2-dependent pathways (13). Another member of the same family, follistatin, is a locally produced factor that negatively regulates Fshb transcription by preventing activin action on gonadotropes (14).
A combinatorial action of several activators and repressors is required for transcription of gonadotrope subunit-encoding genes (10,12,15). Steroidogenic factor-1 (SF-1) is an orphan nuclear receptor protein activator that is required for transcription of gonadotropin subunits in vivo (16). Both in vitro and in vivo studies have identified that Msx1, a homeodomain protein, is a repressor for Cga and Gnrhr genes (17). Other studies iden-tified that FOXO1, a homeodomain protein, acts as a repressor for Lhb in immortalized gonadotrope tumor cells (18).
Although transcriptional regulation of gonadotropin subunits has been well studied (10, 12, 19 -22) as described above, the post-transcriptional mechanisms that regulate gonadotropin subunits are unknown (23,24). Of the three pituitary glycoprotein hormone subunit-encoding genes (Cga, Lhb, and Fshb), Fshb gene is unique in its organization. It encompasses a long 3Ј-untranslated region (UTR) whose functional significance is unknown (1). In vitro studies using primary pituitary cells indicate that activins, the gonadal-and pituitary-derived peptides, post-transcriptionally regulate Fshb mRNA stability (19,25). Although regulation of Lhb and Fshb promoter activity in transformed gonadotrope cell lines has been extensively studied (26 -35), post-transcriptional regulation and secretion of LH and FSH in these in vitro models is not clear.
DICER is an evolutionarily conserved endoribonuclease in the RNase III family that synthesizes microRNAs (miRNAs) and small interfering RNAs from double-stranded RNA (36 -41). The miRNAs have recently emerged as critical regulators of tissue development and gene expression at multiple levels including post-transcription events (36 -41). These include mRNA expression, turnover, stability, and translational control (36 -41). Expression profiling studies have identified several miRNAs differentially regulated in whole pituitary (23,24), various pituitary cell lineages (42), human pituitary tumors (43)(44)(45), and in an immortalized gonadotrope cell line in response to GnRH treatment (46 -48). In addition, specific miRNAs regulating a repressor that selectively regulates LH levels in the female have also been identified (49). Although in silico analysis reveals clusters of miRNAs that are predicted to bind to 3Ј-UTRs of mouse Lhb and Fshb mRNAs, their functional significance in vivo is unknown.
Roles of the DICER-dependent miRNA biogenesis pathway in several mouse reproductive tissues have recently been identified (50). Studies in the female germ line confirm that although miRNA activity is suppressed in mouse oocytes, miRNAs are essential for zygotic development (51)(52)(53)(54). Recent studies confirm the oocytes, unlike somatic cells, express high levels of endogenous-small interfering RNAs that directly target many maternal RNAs and regulate early embryo development (55). In contrast to data on oocytes, loss of miRNAs in granulosa cells, the somatic cells of ovary and uterus, causes variable phenotypes ranging from reduced number of ovulations, defects in oviduct and uterine morphogenesis, and embryo implantation (56 -61). In the male germ line, miRNAs are important for the proliferation of primordial germ cells and spermatogonia but are dispensable for the repression of retrotransposons in developing germ cells (62,63). Deletion of Dicer in Sertoli cells causes proliferation and maturation defects and eventually results in failure of germ cell maintenance (62,64,65). All of the above genetic models have illustrated the important roles of DICER-mediated miRNAs in various reproductive tissues. Here, we have deleted Dicer selectively in gonadotropes, the critical cell lineage in the hypothalamus-pituitarygonadal (HPG) axis, and provide functional evidence that DICER-dependent miRNAs play key roles in gonadotropin homeostasis and fertility.

EXPERIMENTAL PROCEDURES
Generation of Gonadotrope-specific Dicer Knock-out Mice-To generate mice with a gonadotrope-specific deletion of Dicer, we set up a two-step breeding scheme (Fig. 1A). Dicer flox/flox males, a generous gift from Dr. Brian Harfe (66) were first crossed with bLhb-Cre ϩ female mice (67) to obtain Dicer f/ϩ Cre ϩ mice. These mice were subsequently intercrossed to finally generate Dicer f/f Cre ϩ , hereafter referred to as Dicer cKO mice. In the subsequent rounds of breeding, we also set up Dicer f/f Cre Ϫ mice with Dicer f/ϩ Cre ϩ mice to increase the frequency of generating the desired Dicer cKO mice. To genetically label gonadotropes in vivo, we set up Dicer f/ϩ Cre ϩ males with ROSATOMATO (Gt(ROSA)26SorϽtm4-(ACTB-tdTomato,-EGFP) LuoϾ/J) mice (The Jackson Laboratory), referred to as RT mice, and eventually generated Dicer f/f Cre ϩ RT KIϩ or Dicer f/f Cre ϩ RT KI/KI mice. Mice lacking either Fshb (68) or Lhb (69) or both were used as controls. Esr1 ϩ/Ϫ mice were purchased from The Jackson Laboratory, Bar Harbor, ME and intercrossed to generate Esr1 Ϫ/Ϫ mice. All experiments with mice were done per National Institutes of Health guidelines and approved by the University of Kansas Medical Center Institutional Care and Use Committee Animal Protocol. Mice were maintained on standard dark:light (12:12) cycles and fed standard rodent chow, and water was provided ad libitum. All mouse strains were identified by genomic PCR reactions performed on tail DNA samples using primers that distinguish wild type and mutant alleles and detect the Cre transgene as described (67)(68)(69). The amplified products were separated on agarose gels and visualized by ethidium bromide staining.
Isolation of Gonadotropes by Fluorescence-activated Cell Sorting-Pools of pituitaries from 3-4 adult mice of identical genotypes were surgically resected and enzymatically dispersed using collagenase treatment at 37°C as described (70). The final cell suspension was filtered through 100-m nylon cap-containing tubes (BD Biosciences) to remove cell clumps. A 10-l aliquot was checked for red/green fluorescence on an epifluorescence microscope (Zeiss), and the remaining cell fraction was subjected to FACS analysis using an AriaIIU flow sorter (BD Biosciences). A UV laser at 488 nm was used to excite GFP protein. The GFP ϩ fraction representing the gonadotropes and GFP Ϫ non-gonadotrope cells was collected on ice, an aliquot was immediately checked for green fluorescence to confirm only GFP ϩ cells were present and was centrifuged at 3000 ϫ g at 4°C for 5 min, and the cell pellet was used for RNA isolation.
Isolation of mRNA, miRNA, and Real Time qPCR Assays-RNA and miRNA from individual mouse pituitaries or flowsorted gonadotropes were isolated using either the RNAEasy mini kit or miRNA isolation kits (Qiagen), respectively, and DNase-treated on columns. After spectrophotometric quantitation, 200 ng to 1 g of RNA was reverse-transcribed using SuperScript III reverse transcriptase (Invitrogen) to measure gene expression by established protocols (71). Real time qPCR assays were done using Taqman primer/probe mixes (Applied Biosystems) using the absolute quantification method (71). For miRNA quantitation, specific forward primers were used to amplify the first strand of cDNA from single strand miRNAs, and the second strand was synthesized by the universal reverse primer using the microScript II-RT kit. Quantitative RT-PCR was carried out using SYBR Green PCR method (Qiagen) according to the manufacturer's instructions on an Applied Biosystems HT7900 Sequence Detector. A standard curve was run in each assay, with an arbitrary value assigned to the highest standard and corresponding values to the subsequent dilutions. Each cDNA sample was run in triplicate, and the abundance of each mRNA was calculated by dividing that of internal control Ppil1 to normalize for the starting quantity of cDNA. For miRNA quantification, 5 S rRNA and small nuclear RNA U19 were used as internal controls (72). For qPCR assay, RNA samples from at least three mice or three independent flow-sorting experiments were used.
Western Blot Analyses-Pituitary glands from adult mice were collected into chilled radioimmuno precipitation assay buffer (Santa Cruz Biotechnology) containing the protease and phosphatase inhibitors (Santa Cruz Biotechnology) and homogenized, and the supernatants were extracted by centrifugation as described (71). ϳ1 g of prestained ready-to-load protein standards (Fermentas) or 20 g of pituitary proteins were heat-denatured in SDS-PAGE sample buffer (final concentration ϭ 32 mM Tris-HCl, pH 6.8, 12.5% glycerol (v/v), 1% SDS, and 31 M ␤-mercaptoethanol) and separated on 12% polyacrylamide gels. The separated proteins were transferred onto PVDF membranes (Bio-Rad) and blocked in 5% nonfat dry milk for 2 h, and the membranes were incubated with primary antibodies for 2 h at room temperature, washed in TBST buffer, pH 7.6 (137 mM sodium chloride, 20 mM Tris-HCl, 0.1% (v/v) Tween 20), and incubated with the corresponding horseradish peroxidase-conjugated secondary antibodies (Santa Cruz Biotechnology and Bio-Rad) at room temperature for 1 h. The immune complexes were visualized with an ECL kit (GE Healthcare) after exposure to a Biomax film (ISC Biosciences). The following antibodies (and dilutions) were used for Western blot analysis: goat anti-human LH␤ (Santa Cruz Biotechnology, SC7824, 1:5000), goat anti-human FSH␤ (Santa Cruz Biotechnology, SC7797, 1:5000), rabbit polyclonal anti-human ␤-tubulin (Santa Cruz Biotechnology, SC9104, 1:5000), and rabbit anti-FOXO1 (Cell Signaling Technologies, #2880, 1: 2000). Recombinant human FSH was a gift from Dr. Irving Boime. Highly purified bovine LH standard was obtained from Dr. A.F. Parlow, UCLA Medical Center.
Acute GnRH Response Assays-Adult mice were injected intraperitoneally with either 100 l of PBS, pH 7.4, or the GnRH analog Buserelin (20 g/ml). Blood was collected within 2 h and processed for serum LH and FSH measurements as described below.
Serum Hormone Assays-Mice were anesthetized under isoflurane anesthesia, and blood was collected by cardiac puncture and allowed to clot at room temperature. Serum was separated in MICROTAINER (BD Biosciences) tubes by centrifugation and stored frozen at Ϫ20°C until assayed. Serum FSH, LH, and steroid hormones were measured at the University of Virginia Ligand Assay and Analysis Core Laboratory, Charlottesville, VA. Serum FSH was measured by a mouse FSH RIA (sensitivity ϭ 2 ng/ml; intra assay CV ϭ 6.1%; inter assay CV ϭ 8.2%), LH was measured by a sandwich immunoradiometric assay (sensitivity ϭ 0.07 ng/ml; intra assay CV ϭ 3.9%; inter assay CV ϭ 8.5%), testosterone was measured by RIA (sensitivity ϭ 10 ng/dl; intra assay CV ϭ 3.9%; inter assay CV ϭ 7.8%, and estradiol was measured by an ELISA method (Calbiotech, Inc). The sensitivity of the estradiol assay was 3.94 pg/ml, the intra assay CV was 8.1%, and inter assay CV ϭ 5.8% progesterone was measured by a RIA (sensitivity ϭ 0.1 ng/ml; intra assay CV ϭ 4.9%; inter assay CV ϭ 6.9).
Gross and Histological Analyses-Gonads were collected from adult mice at different ages, gross images were photographed, and wet weights were recorded. The tissues were fixed in Bouin's reagent (Sigma) overnight at 4°C, processed, and paraffin-embedded. Approximately 6-m sections were cut and stained with periodic acid-Schiff's reagent/hematoxylin for routine histological analysis as described (71).
Immunofluorescence Assays-Adult mice were transcardially perfused under isoflurane (Piramal Critical Care) anesthesia first with 10 ml of cold 0.9% saline and followed by 60 ml of cold 4% paraformaldehyde, pH 7.2. The pituitaries were collected and post-fixed overnight in 10% sucrose at 4°C. They were later embedded in Tissue-Tek OCT compound (Sakura FineTek) in cryomolds, and ϳ10-m frozen sections were cut onto glass slides using a Leica CM 3050S model cryostat. The sections were rinsed in phosphate-buffered saline, pH 7.4 (PBS), blocked in ready-to-use normal goat or rabbit serum (Invitrogen), and incubated overnight with pairs of primary antibodies at 4°C. The sections were rinsed in PBS incubated with the appropriate fluorochrome-conjugated secondary antibodies in a dark chamber for 1 h at room temperature and mounted in antifade (Invitrogen). The following primary antibodies were used at 1:500 dilution: mouse monoclonal anti-human FSH 4B (a gift from Dr. Irving Boime), rabbit polyclonal anti-human chorionic gonadotropin serum that recognizes mouse LH (generated in the Boime laboratory), rabbit polyclonal anti-SF-1 (a gift from Dr. Ken Morohashi), and all other pituitary hormonespecific antisera obtained from Dr. A. F. Parlow. All the conjugated secondary antibodies were used at a final dilution of 1:200.
Epididymal Sperm Quantification-Freshly isolated cauda epididymis from adult male mice were minced into small fragments in 1 ml of sterile M2 medium (Sigma) and incubated at 37°C for 15 min to release the sperm. An aliquot containing the released sperm was diluted at 1:10 in PBS and counted with a hemocytometer as described (68,71).
Fertility Assays-Adult control and cKO mice at 42 days of age were caged with proven fertile control mice (1 male with one female; a total of 5 pairs) over a period of 6 months. The number of litters and litter sizes was recorded and used to evaluate the breeding performance (68,71).
Gonadectomy and Superovulation Assays-Adult mice were surgically gonadectomized under isoflurane anesthesia and maintained for 1 week to totally deplete the endogenous steroids as described (70,73,74). At this point, blood was collected for hormone assays; one-half pituitary was used for RNA and the other half for Western blot analyses as described above. Immature female mice at 21-23 days of age were superovulated using the standard protocol of equine and human chorionic gonadotropin hormone injection regimen and mated with proven fertile males as described (71). One-cell fertilized embryos were recovered in M-2 medium from oviducts, hyaluronidase-treated, counted, and photographed as described (71).
Statistical Analysis-Each experiment was performed at least three times and consisted of 5-6 mice per group. Statistical analysis was done by the PRISM program using Student's t test. When appropriate, one-way analysis of variance followed by Turkey's post hoc test was used. A p value Ͻ0.05 was considered statistically significant. Data are represented as the mean Ϯ S.E.

Gonadotrope-specific Deletion of Dicer-Anterior pituitary
consists of a heterogeneous population of hormone-producing cells (9). To delete Dicer selectively in gonadotropes, we used a well characterized bLhb-Cre ϩ driver line in which ϳ700 bp of bovine Lhb promoter sequences direct Cre transgene expression in gonadotropes (67). CRE-mediated recombination should occur between the loxP sequences and remove the coding sequences of Dicer, thus resulting in loss of DICER enzyme (Fig. 1A). To test this, we generated Dicer cKO mice on an RT genetic background that allowed visualization of CRE-expressing gonadotrope cells in green (GFP ϩ ) and their enrichment by flow sorting, whereas non-gonadotrope cells that did not express CRE remained red (Fig. 1B). Real time quantitative PCR assay showed that Dicer mRNA is totally suppressed in gonadotropes obtained from pituitaries of Dicer cKO mice when compared with that in gonadotropes obtained from pituitaries of Ctrl mice (Fig. 1C). Immunolabeling showed that flow-sorted GFP ϩ gonadotropes expressed CRE but not DICER and further confirmed that Dicer was deleted specifically in the gonadotrope population (Fig. 1D). Furthermore, quantitative real time PCR indicated many of the miRNAs, computationally predicted to bind Fshb, Lhb, and Cga mRNAs, were significantly suppressed in gonadotropes of Dicer cKO mice (Fig. 2). Together, these data confirm that gonadotrope-specific deletion of Dicer results in loss of miRNAs that target gonadotropin ␤ subunit-encoding mRNAs.

Loss of Dicer in Gonadotropes Results in Suppressed Gonadotropin Synthesis and Secretion in cKO Male
Mice-Loss of microRNAs could result in up-or down-regulation of mRNAs and the corresponding proteins depending on cell type and developmental status and for other reasons (75)(76)(77)(78). To test, how loss of DICER-dependent miRNAs affects gonadotropin synthesis in pituitaries of male mice, we first quantified gonadotropin subunit mRNAs by real time qPCR assays. All the three subunit-encoding mRNAs, namely, Fshb, Lhb, and Cga, were significantly suppressed in purified gonadotropes of Dicer cKO mice compared with those in control mice (Fig. 3A). As expected, the expression of gonadotropin subunit mRNAs was nearly undetectable in the GFP Ϫ negative non-gonadotrope population of cells (Fig. 3A) and thus further validated the flow sorting strategy to purify gonadotropes. Similarly, direct visualization of immunolabeled gonadotropes in the pituitaries of control and Dicer cKO mice using antibodies against FSH␤ and LH␤ ( Fig. 3B) revealed that expression of only gonadotropin subunits, i.e. FSH␤ and LH␤, but not other non-gonadotrope markers, i.e. TSH␤ (thyrotropes), adrenocorticotropic hor- A, Dicer f/f male mice were first intercrossed with bLhb-Cre ϩ female mice to generate Dicer f/ϩ bLhb-Cre ϩ mice. These mice were intercrossed in a second step to generate the desired Dicer f/f bLHb-Cre ϩ (Dicer cKO) mice. In addition, we also intercrossed Dicer f/ϩ bLhb-Cre ϩ male mice with Dicer f/f bLHb-Cre Ϫ females and vice versa to increase the frequency of generating Dicer cKO mice. Pituitaries from adult Dicer cKO mice on a Rosa tomato genetic background were harvested, and an aliquot of enzyme-dispersed cell fraction was visualized before sorting. In B, the merged image shows that Cre is expressed selectively in gonadotropes (GFP ϩ green fluorescent cells); the non-gonadotropes exhibit red fluorescence (GFP Ϫ Cre Ϫ ). C shows Dicer mRNA expression by Taqman real time qPCR assay using FACS-sorted cells. Note that Dicer mRNA was significantly suppressed in GFP-expressing gonadotropes obtained from pituitaries of Dicer cKO mice compared with that in GFP Ϫ non-gonadotrope cells and GFPexpressing gonadotropes obtained from age-matched control (Ctrl) mice. *, p Ͻ 0.01 versus Ctrl (5 mice per genotype; each sample was run in triplicate). Immunofluorescence visualization post-sort (D) GFP ϩ gonadotrope cells were immunoreactive with an antibody against CRE (yellow cells in the merge image) but not DICER (presence of only green fluorescence but no yellow in merge image). The white bar represents 50 m. mone (corticotropes), prolactin (lactotropes), and growth hormone (somatotropes) was significantly reduced in Dicer cKO mice. The immunolocalization data were further confirmed by Western blot analysis. These data indicated that expression of both FSH␤ and LH␤ proteins was significantly suppressed in pituitaries of Dicer cKO males compared with that in controls (Fig. 3C). Loss of one allele of Dicer (Dicer f/ϩ Cre ϩ or Dicer ⌬/ϩ ) did not significantly affect the expression of gonadotropin subunits compared with that in controls. Recombinant human FSH (rhFSH) and bovine LH (bLH) standards served as positive controls, whereas pituitary extracts prepared from Fshb Lhb double null mice served as negative controls and confirmed the specificity of the antibodies used for immunoblots (Fig. 3C). Collectively, these results indicate that loss of Dicer in gonadotropes leads to suppression of gonadotropin synthesis in male mice.
Testosterone and gonadal peptides exert a negative feedback control to regulate gonadotropin synthesis and secretion in the male (1,10,11). We next tested whether castration, which removes gonadal negative feedback, affects gonadotropin synthesis and secretion in the absence of DICER-dependent miRNAs in gonadotropes. Compared with the levels in the intact group of control mice, castration resulted in a significant increase in expression of gonadotropin mRNAs encoding ␣and FSH␤ but not LH␤ subunits when assayed by real time qPCR (Fig. 3D). In contrast, castration did not significantly affect expression of mRNAs encoding gonadotropin ␣ and ␤ subunits compared with those in the corresponding intact group of Dicer cKO male mice (Fig. 3D). Similarly, Western blot analysis showed castration resulted in a significant decrease in FSH␤ and LH␤ protein expression compared with that in intact group of control mice (Fig. 3E). However, no significant change in expression of FSH␤ and LH␤ proteins was detected when intact and castrated groups of Dicer cKO mice were compared (Fig. 3E). Quantification by specific RIAs further confirmed that gonadectomy-induced rise in serum FSH and LH was observed only in control but not Dicer cKO mice (Fig. 3F). Although Dicer cKO mice demonstrated suppressed basal levels of serum LH, acute GnRH response to release LH after a single bolus of buserelin injection was maintained similar to that seen in control male mice (Fig. 3G). Thus, these data reveal that DICER-dependent miRNAs in gonadotropes are critical for post-castration rise in gonadotropin synthesis and secretion in male mice.

Altered Expression of Transcriptional Activators and Repressors in Pituitaries of Dicer cKO Mutant Male
Mice-If DICERdependent miRNAs directly bind to and post-transcriptionally regulate gonadotropin subunit mRNAs, loss of DICER should result in up-regulation of the corresponding gonadotropin subunit proteins. In contrast, we observed suppression of gonadotropin proteins in the pituitaries of Dicer cKO male mice (Fig.  3). We, therefore, next investigated two possibilities. First, we computationally identified miRNAs that were predicted to bind some of the known mRNAs encoding transcriptional repressors. In the absence or severe suppression of a set of these miRNAs in purified gonadotropes (Fig. 4A), we found that some of the repressor-encoding mRNAs including Foxo1, Sox2, and Fst and at least one repressor protein, FOXO1, were upregulated ( Fig. 4C) in the pituitaries of Dicer cKO mice compared with those in control mice. Other known repressor-encoding mRNAs such as Msx1, Esr1, and Zeb1 did not show any significant differences in Dicer cKO mutant pituitaries (data not shown). Immunolocalization studies also failed to detect any significant baseline expression of other repressor proteins including RGS3, ATF3, and ZEB1 in the pituitaries of control mice (data not shown). Second, we found that Gnrhr and Acvr2 but not Foxl2 mRNAs ( Fig. 4D and data not shown) and SF-1 (Fig. 4E) were also significantly suppressed in the pituitaries of cKO mutants. Together, these data indicate that loss of DICERdependent miRNAs results in an increase in gonadotropin subunit transcriptional suppressors and suppression of gonadotropin subunit transcriptional activators with a net effect of severely suppressed gonadotropin subunit proteins in the pituitaries of Dicer cKO male mice.
Dicer cKO Male Mice Display Fertility Defects-Gonadotropins serve as key trophic hormones for male fertility. FSH regulates Sertoli cell number and consequently the germ cell-carrying capacity and testis size (68,79,80). Fshb null male mice, hence lacking FSH dimer, are fertile despite reduced testis size, Sertoli and germ cell number, and sperm (68). LH is required Specific primers for each of the computationally predicted miRNAs that bind gonadotropin subunits were designed and tested in real time qPCR assays using gonadotrope cells purified from control (Ctrl) and Dicer cKO mice. Loss of DICER resulted in suppression of all these predicted miRNAs in the pituitaries of Dicer cKO mice compared with those in controls (*, p Ͻ 0.01). FACS-sorted GFP ϩ gonadotropes were obtained from age-matched adult males (three mice per genotype), and cDNA samples were run in triplicate. Expression of 5 S rRNA was used as an internal control.
for Leydig cell development and testosterone production, and Lhb null male mice have hypogonadism and defects in steroid biosynthesis and display infertility (69). To determine how suppressed gonadotropins affect fertility in Dicer cKO males, we first evaluated both morphology and histology of testes collected from reproductively mature adult males at 42 days of age. Testis size was significantly reduced in Dicer cKO mice compared with age-matched controls (Fig. 5, A and B). Histological analyses of periodic acid/Schiff 's reagent-hematoxylin-stained testis sections from Dicer cKO mice indicated decreased tubule size with apparently heterogeneous tubule architecture including some tubules with grossly normal spermatogenesis but sev-eral abnormal tubules containing very few or no germ cells (Fig.  5F). The majority (23/26 ϭ 88%) of the Dicer cKO male mice displayed this abnormal testis tubule histology. Consistent with testis histology, epididymal sperm number (Fig. 5C) and histology of cauda epididymis showed a significantly lower number of sperm in mutants (Fig. 5F). These testicular defects did not affect mating performance per se of mutant males because they sired similar number of litters compared with controls over a period of 6 months (Fig. 5D). However, the litter size, i.e. the number of pups produced per litter during this time period was significantly reduced when Dicer cKO mutants were compared with control males (Fig. 5E). In A, expression of the gonadotropin subunit encoding mRNAs (Fshb, Lhb, and Cga) was measured by Taqman real time qPCR assays. Purified gonadotrope (GFP ϩ ) and non-gonadotrope (GFP Ϫ ) cells were obtained from the pituitaries of adult male control and Dicer cKO mice. Loss of DICER in gonadotropes led to severe suppression of all three gonadotropin subunit mRNAs compared with that in corresponding controls (p Ͻ 0.01). Note that GFP Ϫ non-gonadotrope cells do not express gonadotropin subunits, confirming the validity of the flow sorting to enrich GFP ϩ gonadotropes. In B, anterior pituitary markers were evaluated in frozen tissue sections from control and Dicer cKO adult male mice. Immunofluorescence was performed using specific antisera against each of the hormones or hormone subunits as indicated and visualized by appropriate fluorochromeconjugated secondary antibodies. Loss of DICER in gonadotropes resulted in marked suppression of only FSH␤ and LH␤ but not non-gonadotrope markers including thyroid stimulating hormone ␤-subunit (TSH␤), corticotropin (adrenocorticotropic hormone (ACTH)), prolactin (PRL), and growth hormone (GH). Hormone expression was indicated by green fluorescence, whereas the nuclei were stained red. The white bar represents 20 m. Immunofluorescence data were further confirmed by Western blot analysis (C) using total pituitary extracts from control (Ctrl), Dicer heterozygous (Dicer ϩ/⌬ ), and homozygous (cKO) mice. Pituitary extracts from double null mice lacking Fshb and Lhb, recombinant human FSH (rhFSH), and purified pituitary bovine LH (bLH) were used, respectively, as negative and positive controls for specificity of the antibodies used. Densitometry quantification showed that both gonadotropins were suppressed in the pituitaries of male mice lacking Dicer in gonadotropes compared with those in controls or heterozygous Dicer cKO mice (p Ͻ 0.01). Panels D and E show RNA (D) and protein (E) expression, respectively, by Taqman real time qPCR assays and Western blot analysis of gonadotropin subunits in the pituitaries of control (Ctrl) and Dicer cKO mice after castration (CASTN). Castration resulted in significant up-regulation (panel D; *, p Ͻ 0.01) of Cga and Fshb gonadotropin subunit mRNAs in the pituitaries of control (Ctrl) but not Dicer cKO mice (panel D; **, p Ͼ 0.05). Densitometry data of Western blots show that castration caused a significant reduction in gonadotropin ␤ subunits in control (panel E; *, p Ͻ 0.01) but not Dicer cKO (panel E; **, p Ͼ 0.05) mouse pituitaries. In F, serum gonadotropins in adult intact and castrated Ctrl and Dicer cKO male mice were measured by RIAs. Castration resulted in significantly elevated serum gonadotropin levels in Ctrl but not Dicer cKO male mice, indicating that DICER-dependent miRNAs are critical regulators of post-gonadectomy rise in gonadotropins. Adult mice were injected with PBS or 2 g of buserelin, a GnRH agonist. After 2 h, serum gonadotropins were measured by RIAs to quantify the acute GnRH response. The -fold changes of GnRH response with respect to PBS-injected values were presented. Although FSH was not released in control or Dicer cKO mice (G, top panel), LH was robustly released (G, bottom panel) in Ctrl mice in response to acute GnRH. Although LH release was not of the same magnitude as in Ctrl mice (5 versus 40-fold), GnRH did cause a significant LH release in Dicer cKO mice (*, p Ͻ 0.05 versus corresponding PBS group). For Taqman real time qPCR assays, expression of Ppil1 was used as an internal control. For Western blot experiments, antibodies specific to gonadotropin ␤ subunits were used at a 1:5000 dilution, and expression of ␤-tubulin was used as an internal control. A representative Western blot was shown, and the densitometry data represent the mean Ϯ S.E. of three independent experiments.
To further evaluate the male fertility defects in Dicer cKO mutants, we assessed by immunofluorescence cell-specific expression of functional markers in the testes of reproductively mature adult mice (Fig. 5G). The transcription factor SOX9 is expressed downstream of SRY, the testis determining factor, and defines the Sertoli cell lineage in the testis (81)(82)(83). The expression of SOX9 did not change irrespective of whether the male fertility was moderately or severely affected in Dicer cKO mutant mice (Fig. 5G). Whereas the expression of other germ cell markers including GCNA1 (spermatogonia), PLZF (stem cells), phospho-histone H3 (mitotically diving germ cells), and SP10 (round spermatids) was similar in moderately affected mutants (Fig. 5G), these markers were undetectable in testes of severely affected mutants. Thus, there were mostly Sertoli cellonly tubules in the testes of severely affected mutants (Fig. 5G). We next tested whether the defects in tubules manifest as a result of defects in Leydig cells that produce the androgen testosterone. Immunofluorescence analysis of three testosterone biosynthetic pathway enzymes, namely, 3␤-HSD1, SCC, and CYP17, that are adult mature Leydig cell markers showed identical expression pattern in the testes of mutants and control mice (Fig. 5H). Consistent with the Leydig cell marker expres-sion, serum testosterone levels were not significantly different in Dicer cKO mutants compared with those in controls (Fig. 5J). However, 50% of mutant males in which testes size was moderately affected showed elevated serum testosterone (Fig. 5J). Similarly, real time qPCR assays of Sertoli, Leydig, and germ cell markers did not reveal any significant differences between Dicer cKO mutants and controls (Fig. 5I). Interestingly, immunofluorescence of gross testis from mutants on an RT genetic background did confirm that Cre was active locally within the testis (Fig. 6A), and expression of Dicer was significantly reduced in the testes of Dicer cKO male mutants (Fig. 6B).

Loss of Dicer in Gonadotropes Results in Suppressed Gonadotropin Synthesis and Secretion in Female
Mice-Gender-specific regulation of gonadotropins has been well studied (68,70,74). To determine if loss of Dicer in gonadotropes affects gonadotropin synthesis and secretion in females similar to males, we first analyzed gonadotropin subunit mRNAs by real time qPCR assays using purified gonadotropes (GFP ϩ ) prepared from the pituitaries of adult female mice. Unlike Lhb and Cga subunit mRNAs, which were both suppressed (Fig. 7A, middle and right panels), Fshb mRNA was unaffected in pituitary in the absence of Dicer (Fig. 7A, left panel) compared with controls. As expected, gonadotropin subunit-encoding mRNAs were undetectable in GFP Ϫ non-gonadotropes (Fig. 7A). Immunofluorescence was used to further evaluate and visualize how gonadotropin protein expression was affected in the absence of DICER. Consistent with RNA expression data, we observed that LH immunostaining was mostly undetectable and, even when present, less intense in only few gonadotropes in pituitary sections obtained from Dicer cKO female mice compared with that in control sections (Fig. 7B). In contrast to LH expression, although Fshb RNA levels remain unchanged, the corresponding protein was undetectable in pituitary sections obtained from Dicer cKO mice compared with that in control sections (Fig. 7B). Loss of DICER in gonadotropes did not affect the expression of non-gonadotrope markers such as adrenocorticotropic hormone, prolactin, and growth hormone whose staining in Dicer cKO pituitary sections was similar to that in controls. Western blot analysis followed by densitometry was used to further quantify gonadotropin subunit expression normalized to that of an internal control ␤-tubulin (Fig. 7C). These values closely paralleled the qualitative data obtained with immunofluorescence, i.e. whereas expression of RNA and protein correlated to each other in the case of LH, there was a discordant correlation between the two in case of FSH in pituitaries of Dicer cKO mice. Pituitary extracts from double null mice lacking both Fshb and Lhb and (hence FSH and LH) and FSH (recombinant human FSH) and LH (bovine LH) standards served as negative and positive controls, respectively (Fig. 7C), and validated the specificity of the antibodies used in Western blot analysis. The acute GnRH response to release a bolus of LH was unaffected in Dicer cKO females similar to that seen in males (Fig. 7G). Together, these data indicate that gonadotrope-specific loss of DICER affects gonadotropins in a sexually dimorphic pattern and selectively regulates the expression of Fshb mRNA at the post-transcriptional level within the female pituitary. In A, real time qPCR analyses indicate that several miRNAs computationally predicted to bind some of the repressors (that normally inhibit gonadotropin subunit gene transcription) were significantly suppressed in purified gonadotropes of Dicer cKO male mice. Consequently, mRNAs encoding the repressors were significantly up-regulated in pituitaries of cKO males (B). In C, Western blot analysis (left) followed by densitometry quantification (right) shows that FOXO1, a repressor for Lhb expression, was significantly up-regulated in the pituitaries of Dicer cKO compared with that in control (Ctrl) male mice. In D, real time qPCR assays indicate that Gnrhr and Acvr2 mRNAs that encode two major signaling pathway receptors required for activation of gonadotrope subunit gene transcription were significantly suppressed in gonadotropes of Dicer cKO male mice. In E, dual label immunofluorescence localization confirms that SF-1, a gonadotrope-specific transcriptional activator, was severely suppressed in a pituitary section obtained from an adult Dicer cKO male mouse pituitary. *, p Ͻ 0.05 versus control (Ctrl) group; n ϭ 4 mice; for real time qPCR assays each sample was run in triplicate. For immunofluorescence, formalin-fixed pituitaries were obtained from adult mice (n ϭ 3), incubated with a rabbit antiserum raised against SF-1 and an FSH␤-specific mouse monoclonal antibody, and visualized by appropriate secondary antibodies conjugated with fluorochromes. The white arrow in E indicates SF-1 nuclear localization in a FSH␤ϩ gonadotrope that was suppressed in a section obtained from Dicer cKO male mouse pituitary. Multiple sections were observed per pituitary and per genotype. The white bar in E represents 100 m.
Both ovarian steroids (progesterone and estrogen) and peptides (inhibins and follistatin) act at the level of hypothalamus and pituitary (gonadotropes) and regulate gonadotropins in the female (1,10,11). We next tested if DICER is required for ovariectomy-induced changes in gonadotropins at the level of pituitary. Ovariectomy caused a significant up-regulation of three gonadotropin subunit mRNAs in pituitaries of control mice as assessed by real time qPCR (Fig. 7D), an increase in the corresponding gonadotropin ␤-subunit proteins in pituitaries determined by Western blot analysis (Fig. 7E), and resulted in elevated serum gonadotropin levels controls (Fig. 7F). In contrast, ovariectomy failed to up-regulate gonadotropin expression in Dicer cKO mice at all levels analyzed including RNA and protein in the pituitary and circulating hormones in serum (Fig.  7F). Thus, DICER-dependent miRNAs in gonadotropes are essential for ovariectomy-induced up-regulation of gonadotropins in female mice.

Altered Expression of Transcriptional Activators and Repressors in Pituitaries of Dicer cKO Female
Mice-Loss of DICERdependent miRNAs in the pituitaries of Dicer cKO female mice also caused suppression of gonadotropin subunit proteins similar to that in males. To check if transcriptional repressors and activators were, respectively, up and down-regulated similarly in pituitaries of female cKO mutants, we first assayed computationally predicted miRNAs that target repressors by Taqman real time qPCR. We found that these miRNAs were significantly suppressed in purified gonadotropes of Dicer cKO females compared with those in controls (Fig. 8A). Unlike in Dicer cKO males, only Esr1 mRNA, but not other repressors, was significantly up-regulated in the pituitaries of female Dicer cKO mice (Fig. 8B). Moreover, in pituitaries of female Esr1 Ϫ/Ϫ mice, Cga and Lhb, but not Fshb mRNAs, were significantly up-regulated by real time qPCR (Fig. 8C) as has been shown by others (84). In addition to up-regulation of repressors, either suppression or loss of activators could also result in suppression of gonadotropin subunit proteins. To address this, we tested the expression of activators for gonadotropins and found that Gnrhr and Nr5a1 (encodes SF-1) mRNAs were both significantly suppressed (Fig. 8D). Immunolocalization data further confirmed that SF-1 protein levels in FSH␤-stained gonadotropes were significantly suppressed in the pituitaries of Dicer cKO female mice (Fig. 8E). Collectively these studies suggest an indirect mechanism likely involving regulation of transcriptional activators and ESR1-mediated repression contributed to the observed defects in gonadotropin subunit protein expression in pituitaries of Dicer cKO female mice.
Dicer cKO Female Mice Display Fertility Defects-Loss of gonadotropins and their cognate receptors results in impairment of ovarian folliculogenesis at distinct phases of ovarian development (85,86). Fshb/Fshr null females mostly phenocopy each other and demonstrate a pre-antral stage block in ovarian folliculogenesis (85,86). Lhb/Lhr null females also phenocopy each other and display a pre-ovulatory stage block in ovarian folliculogenesis (69,87,88). To determine how severely suppressed gonadotropins affect female fertility of Dicer cKO mice, first, we analyzed grossly the female reproductive tracts. At 9 weeks of age, both uteri and ovaries were grossly hypoplastic in Dicer cKO mice compared with controls, and the phenotype was more severely affected by 18 weeks of age (Fig. 9A). Mating trials between Dicer cKO adult females with proven fertile males over a period of 6 months resulted in severely reduced number of litters (Fig. 9B) and litter size, i.e. number of pups per litter (Fig. 9C). During the same time period, one of the three mutant "test" females did not produce a single litter. The remaining two females delivered only 1-2 litters and then FIGURE 5. Male reproductive phenotypes in Dicer cKO mice. Gross testis morphology (A) in adult male mice at 42 days of age is shown. Testis size (B) and cauda epididymal sperm (C) were significantly reduced in Dicer cKO mice compared with control (Ctrl) or heterozygous mice lacking one Dicer allele (Dicer ϩ/⌬ ) in gonadotropes. Although both control and Dicer cKO male mice sired the same number of litters (D) in a 6-month period, the litter size (pups per litter) was significantly reduced in mutants compared with controls (E). Periodic acid/Schiff's reagent-hematoxylin-stained testis histology (F) appears grossly normal in some tubules in Dicer cKO mice, similar to that compared in Ctrl and Dicer ϩ/⌬ mice. Abnormal histology with mostly Sertoli cell-only like phenotype (black star) was also apparent in some tubules within the same testis sections. Consistent with the testis phenotype, very few sperm were present in the epididymis of mutant males compared with control groups of mice (F, lower panels). G, expression of testis markers was shown. Immunolocalization was done on formalinfixed testis sections using antibodies against Sertoli (SOX9)-, PAN germ cell (GCNA1)-, undifferentiated spermatogonia (PLZF; white arrows)-, mitotic germ cell (pHistone H3)-, and round spermatid (SP10)-specific markers. Note that histologically normal tubules in testis of Dicer cKO mouse express all the testis cell markers similar to those in control sections. However, abnormal tubules in Dicer cKO mouse testis expressed only SOX9 but none of the germ cell markers confirming that this is truly a Sertoli cell-only tubule phenotype. Specific marker expression was shown in green, and nuclei are stained red. Although defects in tubules were apparent, Leydig cell markers were all normally expressed in Dicer cKO mutant testes (H). Taqman real time qPCR assays (I) indicate that Sertoli and Leydig cell mRNAs were aberrantly regulated in the testes of Dicer cKO mice compared with those in control mice (indicated by the *). Serum testosterone levels were elevated in moderately affected but not severely affected Dicer cKO (open bars in J) male mice compared with controls (black bar in J). *, p Ͻ 0.05 versus corresponding Ctrl group. Groups of 5-6 adult male mice per genotype were used. White and black bars in F-H panels represent 20 m. stopped breeding. We next analyzed ovarian histology to identify if folliculogenesis was defective and could explain the observed fertility defects in Dicer cKO mutant females. The presence of corporal lutea (CL) indicating ovulations and normal estrus cycles was readily apparent in PAS/hematoxylinstained ovarian sections from control mice at 9 weeks of age FIGURE 7. Regulation of gonadotropins in Dicer cKO female mice. Taqman real time qPCR assays (A) confirm that although Lhb and Cga mRNAs were significantly suppressed, Fshb mRNA was not affected in the absence of DICER in gonadotropes in pituitaries of female mice. Immunofluorescence localization (B) and Western blot followed by densitometry analyses (C) indicate that both FSH␤ and LH␤ were significantly suppressed in the pituitaries of Dicer cKO mice. In B, expression of non-gonadotrope cell markers was not affected and similar to that in control mouse pituitaries. Ovariectomy (OVX) caused a marked up-regulation of gonadotropin subunit mRNAs (D) and FSH␤ and LH␤ proteins (E) in the pituitaries of intact control (Ctrl) but not Dicer cKO mice. The pituitary expression data closely matched that of serum gonadotropin levels shown in F. In G, in an acute GnRH response assay, female mice responded (G) similarly to male mice (Fig. 3G). LH release response to buserelin, although not at the same magnitude as in control (Ctrl) group (5 versus 15-fold), there was a significant increase in LH release response in Dicer cKO female mice compared with the PBS-injected group. Other experimental details were identical to those described in Fig. 3 legend. The white bar in B represents 20 m. In A, real time qPCR assay data confirm that computationally predicted miRNAs that bind repressor-encoding mRNAs were significantly suppressed in gonadotropes of Dicer cKO female mice. However, only Esr1 expression, but not other repressor encoding mRNAs (Foxo1, Foxo3, and Sox2 in B), was found significantly up-regulated in cKO female mutants. Loss of Esr1 resulted in significant up-regulation of Cga and Lhb mRNAs but not Fshb mRNA in pituitaries of cKO mutants compared with those in control (C) as measured by real time qPCR assays. In D, expression of Gnrhr and Nr5a1 mRNAs encoding two critical activators were reduced in gonadotropes of cKO mutants. *, p Ͻ 0.05; n ϭ 3-4 mice; each sample was run in triplicate for real time qPCR assays. Dual immunolabeling (E) confirmed that SF-1 protein expression was severely reduced in a pituitary section obtained from an adult female cKO mouse (E, right) compared with that in control mouse (E, left). Gonadotropes were labeled red, and SF-1 was localized to nuclei and visualized green (E, white arrows). Multiple pituitary sections from three mice were analyzed. The white bar represents 100 m. (Fig. 9D). However, we rarely found CL in ovarian sections obtained from 9 of 10 (90%) age-matched Dicer cKO females (Fig. 9D). Only 1-2 CL (compared with 6 -7 CL in control ovary) were present in ovarian sections obtained from 1 of 10 (10%) mutant mice analyzed at ϳ6 months of age. Although CL were rarely present, both antral and pre-ovulatory follicles were apparent in the ovaries of Dicer cKO mutants (Fig. 9D). Consistent with ovarian histology, some (Cyp19a1 and Inhbb) but not other (Fshr and Lhr) ovarian marker genes regulated by FSH and LH (Ptgs2 and Cyp11a1) were suppressed in adult mutant ovaries (Fig. 9E). Serum estradiol levels were not significantly different (Fig. 9F), but progesterone levels were suppressed in Dicer cKO mutants compared with those in control mice (Fig.  9G). Collectively these data indicate that Dicer cKO female mice display a hypoplastic reproductive tract and severely suppressed fertility.
Superovulation Rescues the Ovulation Defect in Dicer cKO Female Mice-Female mice null for Fshb or Lhb lack either FSH or LH, respectively, but retain the responsiveness to exogenous gonadotropins (68,69). Thus, these null mice can be pharma-cologically rescued in a superovulation assay. To test if reduced number of ovulations in cKO mice is secondary to suppressed gonadotropins, we injected immature female mice with equine chorionic gonadotropin (analog containing both FSH-and LHlike activities)/human chorionic gonadotropin (analog containing LH-like activity) and counted the number of eggs released into oviducts. Dicer cKO mice responded better than double null mice lacking both LH and FSH but less efficiently than control mice in this superovulation assay (Fig. 10, A and B). Histological analyses revealed CL were readily detectable in ovarian sections obtained from all three genotypes of hormonetreated mice (Fig. 10, C-E). The ovarian histology correlated well with induction of most of the gonadotropin-induced marker genes that showed a response in Dicer cKO mice nearly identical to those in PBS-injected control mice and better than that in gonadotropin-deficient double null mice (Fig. 10F). Together, these data confirm that Dicer cKO female mice retain the gonadotropin responsiveness and some of the ovarian defects are secondary to suppressed gonadotropins in these mutant mice.

DISCUSSION
Transcriptional regulation of pituitary gonadotropin subunits has been extensively studied using both in vitro and in vivo models. However, post-transcriptional regulation of these subunit genes is not well understood. DICER is an essential enzyme for biogenesis of mature miRNAs that are key players in posttranscriptional regulation and various other events in gene expression (75). Although several cell-specific Dicer loss-offunction mouse models with defective reproductive function have been generated (75), functional analysis of DICER specifically within pituitary gonadotrope lineage has not been tested. Here, we used a Cre-lox genetic inactivation strategy using the bLhb-Cre driver line and successfully deleted Dicer specifically within the gonadotrope lineage. Loss of DICER beginning around embryonic day (E) 16.5, when bLhb-Cre is first activated (67) in the mouse pituitary, did not affect gonadotropes per se, because gonadotropes could be routinely purified from Dicer cKO mouse pituitaries by flow sorting. Whether pituitary lineage specification would be affected by deleting Dicer using other gonadotrope-specific Cre lines in which Cre recombinase is expressed earlier than E16.5 (89,90) remains to be tested.
Loss of DICER selectively in gonadotropes resulted in suppression of all of the miRNAs computationally predicted to target gonadotropin subunit-encoding mRNAs. It is interesting to note that many of the predicted miRNAs target the Fshb mRNA 3Ј-UTR region that is strikingly longer compared with that of Cga and Lhb mRNAs. The observation that this long 3Ј-UTR of Fshb is the target for binding of many miRNAs and consequently post-transcriptional regulation is consistent with our previous in vivo studies in which we expressed human FSHB transgenes carrying different deletions within the 3Ј region and analyzed their regulation within mouse gonadotropes (91). It is likely that miRNAs respond to multiple signaling networks (GnRH, activins, bone morphogenetic proteins) and orchestrate the coordinated regulation of gonadotropin subunits within gonadotropes. Delineating which sets of miRNAs respond to each of these signaling pathways will be FIGURE 9. Reproductive phenotypes in Dicer cKO female mice. Gross morphology of the female reproductive tract (A) shows hypoplastic ovaries and uteri at two different ages in Dicer cKO mice. Fertility analysis over a 6-month period indicated severely reduced litter number (B) and litter size (C) in female Dicer cKO mice compared with those in age-matched controls. Ovarian histology (D) in control mice shows CL; they were rare and apparent only in 1 of the 20 Dicer cKO female mice analyzed. Black stars indicate antral follicles, and the black bar represents 20 m in D. Ovarian gene expression by Taqman real time qPCR assays (E) indicated that FSH-responsive Cyp19a, Inhbb, and LH-responsive Ptgs2 were significantly suppressed in Dicer cKO mutants (p Ͻ 0.05 versus controls) consistent with suppressed serum FSH and LH in these cKO mice. Serum estradiol showed a trend toward reduction (F), whereas progesterone was suppressed (G) in Dicer cKO female mice compared with controls. Adult mice at 6 -9 weeks of age were used. *, p Ͻ 0.05 versus corresponding Ctrl group. Groups of 5-6 adult mice per genotype were used. critical for future understanding of gonadotropin secretion dynamics that plays a fundamental role in gonad development and gametogenesis.
We found gender-specific differences in transcriptional regulation of gonadotropin ␤ subunits in the absence of DICER in gonadotropes. In the cKO male, basal levels of all gonadotropin subunit mRNAs were suppressed, whereas in the cKO female basal levels of only Cga and Lhb but not Fshb mRNA were suppressed. This suggests a female-specific post-transcriptional mechanism of Fshb gene regulation. This gender-specific regulation could also be due to differences in either the local growth factor signaling pathways within gonadotropes of male versus female or differences in negative feedback control because of the presence of estrous cycles in the female.
Our expression data on transcriptional repressors and activators is consistent with a model wherein a balance between DICER-dependent miRNAs and repressors/activators is critical for gonadotropin subunit protein synthesis (Fig. 11). This model based on our data suggests an indirect mechanism in which simultaneous up-regulation of repressors and loss of activators contributes to suppression of Cga, Lhb, and Fshb mRNAs and the corresponding subunit proteins in pituitaries of Dicer cKO male mice. This could be the net affect, although it cannot be ruled out that DICER-dependent miRNAs could also directly bind to individual gonadotropin subunit miRNAs. A similar mechanism could be invoked to explain the suppression of common ␣ and LH␤ subunit proteins in the pituitaries of female Dicer cKO mice. Interestingly, Fshb mRNA levels unlike Cga and Lhb mRNAs were not suppressed in female cKO mutants nor they were suppressed in Esr1 Ϫ/Ϫ mice. These observations suggest it is likely that yet unknown DICER-independent miRNAs (76) could at least in part contribute to suppression of FSH␤ synthesis in female mice (Fig. 11). This mechanism needs to be tested further in the future.
Irrespective of the gender-specific differences in Fshb gene regulation at the transcriptional or post-transcriptional level, the post-gonadectomy increase in serum gonadotropins was completely abolished in the absence of DICER in gonadotropes of both sexes. Similarly, acute GnRH treatment that normally results in robust LH release in control mice was abolished in Dicer cKO mice. Thus, our studies uncover that DICER-dependent miRNAs play an important role in post-gonadectomy rise in gonadotropins. Whether DICER-dependent miRNAs also play critical roles in temporal regulation of gonadotropin synthesis and secretion throughout the estrous cycle in the female will need to be investigated further. Our future studies will examine how other local signaling pathways regulated by TGF-␤ members, particularly activins and their signaling components, were affected in Dicer cKO mice.
Loss of DICER in gonadotropes resulted in suppressed gonadotropins and altered reproductive function of Dicer cKO male mice. The male reproductive phenotypes of Dicer cKO mice are reminiscent of those in Fshb null males (68) but less severe than those in Lhb null mice (69). The majority of Dicer cKO mutant males demonstrated a moderate reduction in testis size, sperm number, litter size, and thus a sub-fertility phenotype. In addition, testes in the majority of the mutant males (18 of 20; 90%) contained atypical tubules with germ cell aplasia, confirmed by a lack of germ cell marker protein expression. Although Leydig cell marker expression and serum testosterone levels were unaffected in the majority of mutant males, only 50% of mutant males with moderate testis weight reduction showed elevated levels of testosterone. Suppression or the absence of gonadotropins alone may not account for these abnormal testes phenotypes, as male mice lacking either FSH (68) or LH (69) or both (92, 93) do not display this severe germ cell aplasia phenotype. Because bLhb-Cre transgene is residu- FIGURE 11. A model for miRNA-mediated regulation of gonadotropin subunit expression. In pituitaries of control male mice, DICER-dependent miRNAs negatively regulate repressors that normally suppress gonadotropin subunit expression. MicroRNAs can also directly bind to the 3Ј-UTR regions of gonadotropin subunit mRNAs and regulate them directly. GnRH acts via GnRH receptors on gonadotropes and controls expression of gonadotropin subunits. Locally produced activins, primarily activin B, acts via activin receptor type 2 (ACVR2) expressed on gonadotropes and regulates Fshb expression. SF-1 is one of the key transcriptional activators that regulates all gonadotropin subunit genes. A balance between activators and repressors ultimately dictates the coordinated regulation of gonadotropin subunit expression. In Dicer cKO mutants, in the absence of DICER-dependent miRNAs, repressors get up-regulated in pituitary gonadotropes and downregulate expression of gonadotropins in two ways. They can bind Cga, Lhbv and Fshb directly and suppress their transcription. Alternatively, they can down-regulate the expression of activator pathways (Gnrhr, Acvr2, and Sf1), which in turn results in suppression of gonadotropin subunit expression. Thus, in the absence of DICER-dependent miRNAs the balance shifts toward up-regulation of repressors resulting in suppression of gonadotropins. In pituitaries of control female mice, DICER-dependent miRNAs directly bind gonadotropin subunit mRNAs or suppress expression of various repressors including Esr1. ESR1 binds to promoters of Gnrhr, Cga, and Lhb and negatively regulates their expression. Fshb could also be negatively regulated by DICERindependent miRNAs. In the absence of DICER-dependent miRNAs, repressors including ESR1 get up-regulated. Repressors can act at multiple levels including down-regulation of Gnrhr, Sf1, Cga, and Lhb by direct binding or indirectly via suppression of only activators such as Sf1 and Gnrhr, which in turn cause a reduction in gonadotropin subunit expression. Loss of DICER-dependent miRNAs did not cause a reduction in Fshb mRNA in Dicer cKO female mice. Hence, an alternate mechanism to explain suppressed FSH␤ protein could be that there are yet uncharacterized DICER-independent miRNAs (that bind Fshb mRNA) that get up-regulated and suppress FSH␤ protein expression in female mutants. This proposed mechanism remains to be experimentally tested.
ally active in mouse testes (67), we suspect that the above abnormal testis phenotypes could have emerged as a result of recombination events locally within testes. We are currently testing this possibility by generating Dicer conditional mutants using a more tightly regulated gonadotrope-specific Cre expressing mouse line in which Cre is not ectopically expressed in testis.
In contrast to variable reproductive phenotypes in Dicer cKO male mice, Dicer cKO female mice demonstrated hypoplastic uteri and ovaries. Their fertility was more severely suppressed with significantly reduced litter numbers and pups produced per litter. These observations reinforce that female reproductive function, unlike that in the male, is more sensitive to suppression or the absence of gonadotropins. Consistent with suppression of both FSH and LH levels, ovarian cycles were markedly impaired, and ovarian histology showed the presence of a rare corpus luteum in Dicer cKO female mice. These phenotypes were secondary to suppressed gonadotropins because exogenous gonadotropins pharmacologically rescued Dicer cKO female mice. It is interesting to note that Dicer cKO female mice responded to exogenous gonadotropins somewhat intermediately compared with similarly treated control and double mutants lacking both FSH and LH. Whether there are any differences in egg/embryo quality among these genotypes of mice remains to be examined further.
In summary, our studies indicate that gonadotrope-specific deletion of Dicer results in altered gonadotropin homeostasis and as a secondary consequence reproductive dysfunction leading to reduced fertility. Our studies have also unraveled a previously unrecognized role for DICER-dependent miRNA pathway in post-gonadectomy rise in gonadotropin synthesis and secretion. Identification of miRNAs that selectively target individual gonadotropin subunits would provide novel in vivo therapeutic reagents and provide us with a feasible approach to manipulate pituitary gonadotropin secretion as desired.