Gonadal Expression of Foxo1, but Not Foxo3, Is Conserved in Diverse Mammalian Species

ABSTRACT

The Foxos are key effectors of the PI3K/Akt signaling pathway and regulate diverse physiologic processes. Two of these factors, Foxo1 and Foxo3, serve specific roles in reproduction in the mouse. Foxo3 is required for suppression of primordial follicle activation in females, while Foxo1 regulates spermatogonial stem cell maintenance in males. In the mouse ovary, Foxo1 is highly expressed in somatic cells (but not in oocytes), suggesting an important functional role for Foxo1 in these cells. Given that invertebrate model species such as Caenorhabditis elegans and Drosophila melanogaster harbor a single ancestral Foxo homolog, these observations suggest that gene duplication conferred a selective advantage by permitting the Foxos to adopt distinct roles in oogenesis and spermatogenesis. Our objective was to determine if the remarkably specific expression patterns of Foxo1 and Foxo3 in mouse gonads (and, by inference, Foxo function) are conserved in diverse mammalian species. Western blotting was used to validate isoform-specific antibodies in rodents, companion animals, farm animals, nonhuman primates, and humans. Following validation of each antibody, immunohistochemistry was performed to ascertain Foxo1 and Foxo3 gonadal expression patterns. While Foxo1 expression in spermatogonia and granulosa cells was conserved in each species evaluated, Foxo3 expression in oocytes was not. Our findings suggest that Foxo3 is not uniquely required for primordial follicle maintenance in nonrodent species and that other Foxos, particularly Foxo1, may contribute to oocyte maintenance in a functionally redundant manner.

Gonadal expression of Foxo1 is conserved in mammals, but Foxo3 expression in primordial oocytes is limited to rodents.

INTRODUCTION

The Foxo forkhead transcription factors control multiple aspects of development, metabolism, reproduction, and tissue stem cell function [1, 2]. The three principal Foxos (Foxo1, Foxo3, and Foxo4) are broadly expressed and subject to regulation by several posttranslational modifications, most notably phosphorylation by the kinase Akt [3]. A fourth, more distantly related Foxo, Foxo6, is subject to distinct regulatory mechanisms and has a more limited tissue distribution—discrete anatomical locations in the brain, such as the hypothalamus, where Foxo6 regulates memory consolidation and synaptic function [4, 5]. The Foxos are highly conserved in mammals, and each Foxo protein is more similar to its cross-species orthologs (e.g., mouse vs. human Foxo1) than to the other Foxos within the same species (e.g., mouse Foxo1 vs. mouse Foxo3 or Foxo4), suggesting that the Foxos evolved to serve distinct and conserved physiologic functions. Nonetheless, in some contexts (such as tumor suppression), the Foxos are functionally and genetically redundant, demonstrating that the Foxos also share overlapping functions [6, 7].

The Foxos serve remarkably discrete gametogenic functions in mice. Foxo3^−/− female mice are born with a normal complement of oocytes. However, global premature primordial follicle activation (PFA) occurs within a few days, leading to a syndrome of ovarian hypertrophy, accelerated follicular atresia, and hypergonadotropic ovarian failure, with consequent infertility [8, 9]. Detailed phenotypic analyses demonstrated that Foxo3 is specifically required for PFA but not subsequent stages of follicle maturation (e.g., Foxo3^−/− females are initially fertile despite global PFA, becoming sterile at the time of follicle depletion) [10]. The Foxo3 protein is highly expressed in the oocytes of primordial and primary follicles, and, by immunohistochemistry (IHC), Foxo3 is detectable only within oocytes. Oocyte Foxo3 is cytoplasmic at birth but is gradually imported into the nucleus beginning at Postnatal Day (PD) 3, when primordial follicle assembly is completed. Nuclear import concludes by PD14, and the Foxo3 protein remains nuclear in primordial oocytes throughout life. The protein translocates back into the cytoplasm following primordial follicle activation and is degraded by the secondary follicle stage. These observations established that Foxo3 serves as a molecular switch functioning within the oocyte to regulate PFA. Consistent with this idea, oocyte-specific conditional inactivation of Foxo3 also results in a global PFA phenotype. In contrast, germline inactivation of Foxo1 or Foxo4 does not have a discernible impact on female fertility or ovarian function, and triple germline Foxo1/3/4 knockout results in the same phenotype as Foxo3 inactivation alone [6, 11].

In contrast to this specific requirement for Foxo3 in the female germline, Foxo1 was more recently shown to serve vital functions in the male germline [11]. Within the adult testis, Foxo1 protein is specifically expressed in undifferentiated spermatogonia—cells that reside on the basement membrane and serve as a stem cell population driving spermatogenesis. Conditional inactivation of Foxo1 in the male germline revealed its essential role in male fertility. Foxo1 inactivation leads to severe defects in spermatogonial stem cell (SSC) maintenance and differentiation. Foxo3^−/− and Foxo4^−/− males are fertile with normal spermatogenesis and testis weights, but Foxo1/3/4 triple mutant males exhibit a more severe phenotype with a complete failure of spermatogenic differentiation. These results demonstrated that Foxo1 is, by far, the most important Foxo with respect to spermatogenesis, with Foxo3 and Foxo4 serving relatively subservient roles. In mouse ovaries, Foxo1 is specifically expressed in the granulosa cells of growing follicles, where it serves as the principal Foxo regulating several aspects of granulosa cell function and follicle maturation [12, 13].

Taken together, the above observations demonstrate that, at least in mice, Foxo1 and Foxo3 serve discrete and highly specific gonadal functions (Foxo1-spermatogenesis and granulosa cell maturation, Foxo3-primordial follicle activation). These genetic and functional requirements in mice correlate with the specific expression and localization patterns of the Foxo1 and Foxo3 proteins within discrete gonadal cell types (Foxo1-undifferentiated spermatogonia and granulosa cells; Foxo3-primordial oocytes). Here, we investigated whether these expression patterns (and, by inference, functions) are conserved phylogenetically. We systematically analyzed the expression and distribution of Foxo1 and Foxo3 in the gonads of diverse mammalian species, including rodents, companion animals, farm animals, nonhuman primates, and humans. There is considerable interest in the roles of the Foxos in gametogenesis and reproduction in diverse mammalian species [14–17], including their use as targets and/or biomarkers in contraception and advanced reproductive technologies [18–21]. Therefore, these analyses are an important step toward understanding the potential conservation (and practical applications) of Foxo gonadal functions in mammals.

MATERIALS AND METHODS

Tissue Specimens and Processing

Tissues were obtained from the following genera: Danio, Mus, Rattus, Peromyscus, Felis, Canis, Bos, Sus, Macaca, Papio, Pan, and Homo (Fig. 1A). Wild-type zebrafish (Dr. James Amatruda, UT Southwestern Medical Center, [UTSW]), FVB mice, and Sprague-Dawley rats (Dr. Kent Hamra, UTSW) were obtained from colonies maintained at the UTSW Animal Resource Center under approved Institutional Animal Care and Use Committee protocols. Tissues were harvested from 3- to 4-wk-old mice (ovaries and testes), 6-mo-old rats (testes), and 4-wk-old rats (ovaries). Three-month-old deer mouse tissues (ovaries, testes, and liver) were obtained from the University of South Carolina Peromyscus Genetic Stock Center (Dr. Gabor Szalai, Columbia, SC). Cat and dog tissues included testes from a 4-mo-old cat, ovaries from a 2-yr-old cat, testes from a 1-yr-old dog, and ovaries from an 11-mo-old dog. These specimens were processed immediately following scheduled castrations in a veterinary practice and would have otherwise been discarded (Dr. Bart Owens, West Loop Animal Hospital, Longview, TX). Adult bovine ovaries were obtained from Sierra for Medical Science (Whittier, CA), and an adult bovine testis was obtained from a local meatpacking company. Porcine ovaries and testes were obtained from Sierra for Medical Science and a local rancher, respectively. Nonhuman primate tissues that would have otherwise been discarded following necropsy for other indications were obtained from the Southwest National Primate Research Center (Dr. Jerilyn Pecotte, San Antonio, TX) and included an ovary from a 19-yr-old rhesus, a testis from a 13-yr-old rhesus, an ovary from an 18-yr-old baboon, a testis from an 11-yr-old baboon, a testis from a 44-yr-old chimpanzee, and an ovary from a 17-yr-old chimpanzee. Neonatal rhesus (preterm neonatal death), baboon (neonatal death at term), and chimpanzee (neonatal death at term) ovaries were also obtained in the form of freshly cut 5-μm tissue sections from paraffin blocks previously processed and archived by the center. An adult human ovary (38 yr old) was collected following oophorectomy for an unrelated indication as part of a UTSW Medical Center Institutional Review Board-approved gynecologic research tissue collection/distribution protocol. Paraffin-embedded human testis (normal tissue adjacent to testicular tumor in an orchiectomy specimen from a 15-yr-old male) and neonatal ovary (2 days of life) specimens were obtained from the UTSW Medical Center Tissue Resource Center. All tissues were processed immediately after procurement. Portions were snap frozen in either liquid nitrogen or dry ice and stored at −80° until homogenized for Western blotting. The remaining tissues were fixed in 10% neutral buffered formalin for 24 hours, embedded in paraffin, and cut into 5-μm sections for IHC.

FIG. 1.

Comparative analysis of Foxo1 and Foxo3 proteins in diverse vertebrate species. A) Scientific and common names of the species included in this study. Pairwise alignment scores (HomoloGene, NCBI database) were used to calculate the identity (%) shared with the human FOXO1 and FOXO3 homologs. Sequence data were not available for deer mouse, cat, pig, and baboon. B) Phylogrammatic representation of the data (ClustalW2) reveals that each Foxo protein is more similar to its cross-species orthologs (e.g., mouse vs. human Foxo1) than to the other Foxos within the same species (e.g., mouse Foxo1 vs. mouse Foxo3).

Antibodies for Immunohistochemistry and Western Blotting

Antibodies and titers used were Foxo1 (rabbit monoclonal; Cell Signaling Technology, no. 2880; 1:500 IHC and immunofluorescence, 1:1000 Western blotting), Foxo3 (rabbit polyclonal; Santa Cruz Biotechnology, no. sc-11351; 1:500 IHC, 1:1000 Western blotting), Foxo3 “ab2” (rabbit monoclonal; Cell Signaling Technology, no. 2497; 1:1000 Western blotting), glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (rabbit monoclonal; Cell Signaling Technology, no. 2118; 1:5000 Western blotting), actin (mouse monoclonal; Sigma-Aldrich, no. A-2547; 1:5000 Western blotting), and Plzf (mouse monoclonal; Calbiochem; no. OP128; 1:200 immunofluorescence).

Immunohistochemistry and Immunofluorescence

Slides were deparaffinized in xylene, hydrated in an ethanol series, subjected to antigen retrieval by boiling in 10 mM sodium citrate (pH 6.0) for 15 min, and cooled at room temperature (RT) for 20 min. For IHC, after peroxidase blocking (3% hydrogen peroxide in water) for 30 min, slides were washed in water and then blocked in bovine serum albumin (BSA, 1% in 1× PBS) for 15 min. Slides were then incubated with primary antibody for 1 h at RT, subjected to a TBST (1 M Tris-HCL, 5 M NaCl, 1× Tween-20) wash series, incubated with secondary antibody for 30 min at RT (Immpress; Vector Laboratories), and subjected to a second TBST wash series. Signal was detected using a DAB liquid chromogen substrate kit (DAKO). Slides were then counterstained with hematoxylin, rinsed in water, and air-dried. For immunofluorescence, slides were blocked in BSA (3% in 1× PBS) for 1 h and then incubated with both primary antibodies (diluted in 3% BSA) for 1 h at RT. The slides were then subjected to a PBS wash series (10 min × 3) followed by a 1-h incubation at RT with secondary antibodies (Alexa Fluor 488 anti-mouse, no. A-11029, and Alexa Fluor 555 anti-rabbit, no. A-21429; Invitrogen) diluted to 1:1000 in 3% BSA. Following a second PBS wash series, slides were incubated with DAPI (no. 46290, 1:10 000 in PBS; Pierce Protein Research Products,) for 10 min at room RT. Slides were then briefly rinsed in PBS and mounted in Vectashield (Vector Laboratories). Microscopy was performed with a Leica TCS SP5 confocal microscope with 40× and 63× oil immersion lenses using 543-nm HeNe and 488-nm Ar lasers.

Tissue Homogenates and Cell Lysates

Previously frozen tissues were weighed and then mechanically homogenized on ice. The lysis buffer consisted of RIPA buffer (Thermo Scientific), cOmplete Mini EDTA-free protease inhibitor cocktail (Roche Diagnostics), and Phosphatase Inhibitor Cocktail 2 (Sigma-Aldrich). Following homogenization, suspensions were centrifuged at 13 000 rpm for 10 min in a microcentrifuge. The resulting supernatants were then divided into aliquots for protein quantification using a BCA protein assay kit (Pierce Biotechnology) and frozen storage at −20°C. Previously processed cell lysates from a mouse spermatogonial stem cell line harboring floxed alleles for the Foxos and a tamoxifen-inducible Cre^ERT2 transgene (Dr. Zhuoru Wu, UTSW) and both Foxo3^−/− and Foxo3^+/+ mouse embryonic fibroblasts (Dr. Diego Castrillon, UTSW) were obtained for use as positive and negative controls for immunoblotting experiments.

Western Blotting

Tissue homogenates (liver, ovary, or testis, depending on availability) and cell lysates (with previously determined protein concentrations) were combined with an equal volume of Laemmli buffer (Bio-Rad Laboratories), boiled for 10 min, subjected to SDS-polyacrylamide gel electrophoresis (Mini-PRO-TEAN TGX 12% Precast Gels; Bio-Rad Laboratories; 130 V until migration complete), and transferred to polyvinylidene difluoride membranes (Immobilon-P; Millipore; 100 V for 1 h). Nonspecific antibody binding was blocked by incubation with TBST plus nonfat milk (5% in TBST) for 1 h at RT. Membranes were then incubated with primary antibody (see above) overnight at 4°C. After a TBST wash series, membranes were incubated with HRP-linked secondary antibody raised in donkey (ECL; GE Healthcare) for 1 h at RT and then subjected to a second TBST wash series. Signal was detected using the SuperSignal West Dura chemiluminescent substrate detection kit (Thermo Scientific). Where indicated, membranes were stripped with Restore Western Blot Stripping Buffer (Thermo Scientific) and reblocked prior to incubation with primary antibody specific for loading control proteins. In these experiments, the total amount of protein loaded in each lane was normalized to the levels of Foxo1 or Foxo3 in a preliminary Western analysis. The total amounts loaded ranged from 0.25 to 10 μg of protein for the Foxo1 blots and 0.6 to 14 μg for Foxo3.

RESULTS

Phylogenetic Analysis of Foxo1 and Foxo3 Proteins in Diverse Mammalian Species

ClustalW2 was used to investigate the phylogeny of the mammalian Foxo1 and Foxo3 proteins for which sequence data were available [22]. For both, amino acid sequences were highly similar (at least 93% identity) to the human protein in each species (Fig. 1A). The phylogramatic representation of the data emphasizes not only the striking degree of conservation for each Foxo but also that each Foxo protein is more similar to its cross-species orthologs than to the other Foxos within the same species. This was true even when a more distant vertebrate, zebrafish, was included in the analysis (Fig. 1B). These data are consistent with the idea that Foxo gene duplication was important in the evolution of vertebrates and that the each Foxo diverged to serve distinct functions.

Validation of Antibodies for Analyses of Foxo1 and Foxo3 Expression in Tissue Sections

Previous studies have shown that Foxo mRNA expression does not correlate well with protein levels in tissue sections. For example, in the ovary, the Foxo3 protein is detectable only in primordial to primary oocyte nuclei, whereas at the mRNA level, Foxo3 is ubiquitously expressed [8, 12, 23]. Foxo1 protein is detectable only in undifferentiated spermatogonia [11], while mRNA in situ hybridization shows the presence of Foxo1 transcripts throughout the testis (unpublished data). These findings, which point to the existence of important posttranslational mechanisms regulating Foxo protein stability [24], argue that analyses of Foxo expression within gonads should be conducted at the protein level. To this end, we sought to identify antibodies suitable for the detection of Foxo1 and Foxo3 proteins in tissue sections in a wide range of mammalian species.

Given that most antibodies detect epitopes that can differ in amino acid sequence even in closely related species, it was essential to first formally validate the antibodies and demonstrate their ability to bind to the protein in each species to be investigated. As shown in Figure 2A, the Foxo1 antibody used for subsequent IHC experiments (a rabbit monoclonal to a C-terminal human epitope) identified principal bands of the expected molecular weights of 72–75 kDa in each of the eleven mammalian species studied. Immunoblotting for actin using an antibody commonly used as a loading control demonstrated cross-reactivity with actin in all species except one. A second antibody also used for loading controls (GAPDH) detected the protein in 9 of the 11 species (Fig. 2A). Strikingly, the Foxo1 antibody detected Foxo1 in all mammalian species, while the highly conserved actin and GAPDH proteins were not similarly detected in all species. These Western analyses thus demonstrated that this Foxo1 antibody is extremely well suited for phylogenetic analyses. To further confirm the specificity of this antibody, we analyzed a mouse spermatogonial stem cell line harboring floxed alleles for the Foxos and a tamoxifen-inducible Cre^ERT2 transgene. The principal band disappeared following tamoxifen administration, confirming that this band corresponds to Foxo1 (Fig. 2, A and B).

FIG. 2.

Validation of Foxo1 antibody in diverse mammalian species. Tissue homogenates and cell lysates were subjected to SDS-PAGE and Western blot analysis. A) The Foxo1 (rabbit monoclonal) antibody employed in this study detected 72- to 75-kDa proteins consistent with Foxo1. Note that the image is uncropped to show all detected molecular species; the minor secondary bands appear to be related to Foxo1. The membrane was then stripped and reprobed with actin and GAPDH control antibodies. Although actin and GAPDH are highly conserved phylogenetically, the respective antibodies did not detect the proteins in all species. B) The same Foxo1 antibody failed to detect a protein in Foxo1 null cells. The principal band remained detectable in untreated cells as well as control homogenate (liver).

The Foxo3 antibody selected for IHC was a rabbit polyclonal raised against a polypeptide corresponding to amino acids 329–472 near the C-terminus of human Foxo3. This antibody detected principal bands of 80–86 kDa, consistent with the predicted molecular weights for Foxo3 in these species (Fig. 3A). Other minor protein bands seen in this blot appear to correspond to alternative Foxo3 splice forms or degradation products (see below). To further confirm the specificity of this antibody, we analyzed mouse embryonic fibroblasts obtained from Foxo3^+/+ and Foxo3^−/− mice. As expected, the principal band was absent in the Foxo3 null cells. Furthermore, most of the lower-molecular-weight bands also disappeared (Fig. 3B and data not shown), arguing that these also correspond to Foxo3 (alternative splice forms, degradation products, and so on).

FIG. 3.

Validation of Foxo3 antibody in diverse mammalian species. Tissue homogenates and cell lysates were subjected to SDS-PAGE and Western blot analysis. A) The Foxo3 (rabbit polyclonal) antibody employed in this study for IHC detected 80- to 86-kDa proteins consistent with Foxo3. In one species (cat), the Foxo3 protein was of lower apparent molecular weight (65 kDa, asterisk). The secondary bands in this uncropped image appear to be degradation products or variants of Foxo3. The membrane was stripped and reprobed with the control antibodies actin and GAPDH. B) The same Foxo3 antibody failed to detect a protein in Foxo3 null cells. C) Western blotting with an unrelated Foxo3 antibody (ab2, rabbit monoclonal) confirmed that the cat 65-kDa protein detected in A is indeed Foxo3 (asterisk). The principal bands detected for mouse, dog, and human were of the expected molecular weight range.

The principal band in one species (cat) was of significantly lower apparent molecular weight (∼65 kDa) than in all other species, raising the question as to whether this indeed corresponded to Foxo3. Western blotting with a second, unrelated Foxo3 antibody (“ab2,” a rabbit monoclonal) detected identical bands, confirming that the cat Foxo3 is of lower molecular weight than all other mammalian Foxo3 proteins (Fig. 3C). This second Foxo3 antibody did not prove suitable for IHC (data not shown). The current Felis catus genome assembly (Ensembl Build 70) is of relatively low coverage (1.87×) and highly fragmentary [25]. Subsequent builds with complete Foxo3 gene sequences may provide an explanation for the unusually low molecular weight of Foxo3 in house cats.

The above results confirmed the utility of this antibody (rabbit polyclonal) in detecting Foxo3 in all the mammalian species in this study. Incidentally, these species were selected on the basis of their interest as companion animals, domesticated species, or closely related primates, but no species was excluded from this study because of a failure of the antibodies to detect their respective Foxos. Thus, these Foxo1 and Foxo3 antibodies are likely suitable for analyses of the vast majority of mammalian species.

Foxo1 Is Specifically Expressed in Undifferentiated Spermatogonia in Diverse Mammalian Species

Foxo1 is required for SSC self-renewal and differentiation in mice. Concordantly, the Foxo1 protein has been immunolocalized in mice to undifferentiated spermatogonia, a relatively rare subpopulation of spermatogonia on the seminiferous tubule basement membrane [11]. This pattern is distinct (Fig. 4A) and easily distinguishable from that of differentiated spermatogonia, which also reside on the basement membrane but are much more numerous [11, 26, 27]. To determine if Foxo1 was also specifically expressed in undifferentiated spermatogonia in other mammals, IHC against Foxo1 was performed under identical experimental conditions (fixation time, antigen retrieval, antibody titer, chromogen incubation time, and so on). The characteristic staining pattern of undifferentiated spermatogonia was observed in all 11 species, including human (Fig. 4A). Immunofluoresent colabeling of Foxo1 and Plzf, an established marker of undifferentiated spermatogonia, confirmed in all species analyzed that the Foxo1-positive cells indeed corresponded to undifferentiated spermatogonia (Fig. 4B). We infer that Foxo1's importance in controlling the self-renewal and differentiation of undifferentiated spermatogonia is likely conserved in most mammals, including humans.

FIG. 4.

Foxo1 expression in spermatogonial stem cells is conserved across all mammalian species studied. A) Immunohistochemical detection and localization of Foxo1 protein in tissue sections from paraffin-embedded testes. Foxo1 is clearly localized to a rare population of cells that rest on the seminiferous tubule basement membrane, consistent with undifferentiated spermatogonia. Bar = 25 μm. B) Foxo1 and Plzf immunofluorescence in tissue sections from paraffin-embedded testes. The two markers were coexpressed in all species studied. Representative fields from four species are shown. All images are at ×63 magnification except mouse (×40).

Foxo1 Is Specifically Expressed in Granulosa Cells of Developing Follicles in Diverse Mammalian Species

Foxo1 is highly expressed in the granulosa cells of developing follicles in the mouse ovary [12, 28, 29]. Contrary to its specific germline expression pattern in the testis, Foxo1 is not detectable by IHC in female murine germ cells postnatally [12]. To investigate if Foxo1 expression in granulosa cells is conserved, Foxo1 IHC was performed on each of the species described above. Foxo1 expression was readily and specifically detected in granulosa cells of developing follicles (Fig. 5, A–K) in similar patterns. Foxo1 expression was undetectable in the pregranulosa cells of primordial follicles (Fig. 5, A–K, insets) but was induced in the cuboidal granulosa cells of primary to secondary follicles and remained highly expressed in the granulosa cells of more advanced follicles (Fig. 5, A–K). In some species (rat, deer mouse, cat, baboon, and human), expression was apparent by the primary follicular stage, whereas it did not appear until the secondary follicular stage in other species (mouse, dog, cow, pig, rhesus, and chimpanzee). In addition, all species except rhesus exhibited prominent Foxo1 expression in theca cells (data not shown). Foxo1 expression in corpora lutea was more variable but prominent in cat, pig, baboon, chimpanzee, and human (data not shown). We conclude that, in mammals, Foxo1 evolved to serve divergent functions in the male and female gonad, that is, germline in testis and somatic compartment(s) in ovary.

FIG. 5.

Foxo1 expression in ovarian granulosa cells is conserved across all mammalian species studied. Foxo1 has been previously shown to be the principal Foxo in ovarian granulosa cells in mice. At low magnification (upper panels), immunohistochemistry against Foxo1 showed expression in developing follicles in all species. Granulosa cell expression was induced at the primary to secondary follicle stage and remained prominent in antral/preovulatory follicles in all species. Higher magnification (lower panels) contrasts the absence of expression in pregranulosa cells in primordial follicles (insets) with prominent granulosa cell expression in adjacent, more advanced follicles. Bars = 200 μm and 50 μm for upper and lower panels, respectively.

Oocyte Foxo3 Expression Is Conserved Only in Rodents

In mice, the Foxo3 protein functions within the oocyte nucleus to suppress PFA during postnatal life. Inactivation of Foxo3 either genetically [8] or by artificial hyperactivation of PI3K (which leads to Foxo3 hyperphosphorylation and nuclear-to-cytoplasmic export) [23] results in global activation of the primordial follicle pool. Following physiologic PFA, the Foxo3 protein is rendered inactive by nuclear export and subsequent degradation. Figure 6A shows this natural sequence in the adult mouse ovary. In normal primordial follicles (two left panels), Foxo3 localizes to the oocyte nucleus and is not detected in other ovarian compartments (apart from the smooth muscle layer of blood vessels). Foxo3 regulates the proliferation of vascular smooth muscle cells [30–33], and its expression in this compartment served as a reliable and convenient internal positive control in all species (two examples shown in Fig. 6D). However, by the late primary follicle stage, the Foxo3 protein was exported from the nucleus to the cytoplasm (Fig. 6A, right panel). An identical pattern of Foxo3 expression and relocalization during early follicle growth was found in all rodent species examined (mouse, rat, and deer mouse).

FIG. 6.

Foxo3 expression in primordial oocytes and Foxo3 nuclear-to-cytoplasmic translocation following primordial follicle activation is conserved only in rodents and not other mammalian species. Foxo3 is the principal Foxo in the mouse oocyte, and its nuclear-to-cytoplasmic translocation is believed to be an important regulator of primordial follicle activation. A) Immunohistochemistry against Foxo3 shows exclusive nuclear expression in primordial follicle oocytes in mouse, rat, and deer mouse. Nuclear-to-cytoplasmic translocation was evident in larger primary follicles in all three species, followed by degradation during the secondary follicle stage, as previously demonstrated in mouse (not shown). In companion and farm animals (B) and primates, including human (C), no specific Foxo3 localization was evident. Only a faint and likely nonspecific signal was apparent in some nonrodent species. D) Staining in vascular smooth muscle in all species served as an internal positive control. Examples for dog and cow are shown. Bars = 10 μm.

Strikingly, however, Foxo3 was not specifically detected in the primordial oocytes of any nonrodent species (three of three rodents vs. zero of eight nonrodents; P = 0.0061 by Fisher exact test; Fig. 6, B and C). Faint cytoplasmic signal observed in primordial/primary oocytes in some species could not be distinguished from background (i.e., nonspecific) staining of the ooplasm. In nonrodent species, neither nuclear localization of Foxo3 in primordial oocytes nor evidence of nucleocytoplasmic translocation in primary follicles was observed (Fig. 6, B and C). Internal positive controls (staining of vascular smooth muscle) demonstrated successful immunodetection of Foxo3 in all sections.

Foxo1 Is Expressed in Bovine Oocytes

The cow was unique among the species studied in that Foxo1 was detected in both granulosa cells and oocytes. With respect to granulosa cells, the expression pattern mirrored that seen in all other species. Unlike other species, though, Foxo1 was readily detected in most primordial oocytes (Fig. 7). Expression persisted in oocyte nuclei within some primary follicles, whereas it was absent in others. Some primary follicles suggested nucleocytoplasmic translocation of Foxo1, while others did not. Oocyte nuclear Foxo1 expression was largely absent by the secondary follicular stage but was occasionally observed (Fig. 7). Thus, while Foxo1 expression and localization in bovine oocytes was somewhat variable, it is consistent with the idea that other Foxos could contribute to Foxo function (and thus substitute for Foxo3) in oocytes in nonrodent species.

FIG. 7.

Foxo1 is expressed in bovine primordial oocytes. Immunohistochemistry against Foxo1 shows nuclear expression within primordial follicle oocytes. Expression persisted in some primary follicles, whereas it was absent in others. Evidence of nuclear-to-cytoplasmic translocation was detected in some primary follicles. Nuclear Foxo1 expression was variable in the oocytes of secondary follicles. Bars = 25 μm.

DISCUSSION

Studies in mice have firmly established the importance of the Foxos for gametogenesis and long-term germline maintenance. Foxo1 is required for SSC self-renewal and differentiation in males, while in females Foxo3 is an essential regulator of primordial follicle activation and maintenance. Although most mammals have three closely related Foxo genes (Foxo1, Foxo3, and Foxo4, plus the more distantly related Foxo6), invertebrates such as Caenorhabditis elegans and Drosophila melanogaster have single ancestral Foxo homologs (DAF-16 and dFOXO, respectively). Notably, these ancestral homologs are also required for maintenance of the germline in a germline-autonomous manner, demonstrating that these Foxo functions are quite ancient [34, 35]. Our goal here was to explore the extent to which the well-established dichotomy of Foxo1/Foxo3 function in the murine male versus female germline and somatic gonad was conserved in other mammals, including humans.

A number of mechanisms and factors directing oogenesis are remarkably conserved not only among mammals but also between more evolutionarily distant species [36]. These include primordial germ cell differentiation and migration to the developing ovary, specialization of somatic cells within the gonad, and follicular assembly. While menopause may be unique to primates, the phenomenon of gradually declining reproductive performance and ultimate senescence is also seen in nonprimate mammalian species as well as nonmammalian vertebrates and invertebrates [37, 38]. For example, the dynamics of primordial follicle decline throughout life are quite similar between rodents and humans, and mechanisms regulating programmed cell death in the germline exhibit a high degree of interspecies conservation [36]. While similarities in essential features of germline establishment, maintenance, and eventual decline are clearly evident among species, very striking differences also exist. The higher incidence of polyovular follicles (i.e., the presence of multiple oocytes in an individual follicle) in canine ovaries is one notable example of variation in follicular architecture [39], whereas the phenomena of seasonal (long-day or short-day) [40] versus opportunistic (e.g., golden spiny mouse) [41] versus continuous breeding (humans) and copulation-induced ovulation (in felids) [42] reflect profound differences in the hypothalamic-pituitary-ovarian endocrine axis.

The most striking and unexpected finding in our study was that Foxo3 was specifically expressed in primordial oocytes only in rodents but not in any of the other species studied, including humans. Patterns of Foxo3 ovarian expression (including nuclear-to-cytoplasmic shuttling followed by degradation) were remarkably consistent among rodents, including Peromyscus (deer mice), further supporting conservation among rodents since Peromyscus is only distantly related to the Mus and Rattus genera. This conservation of Foxo3 expression patterns in rodents is in stark contrast to the fact that Foxo3 was undetectable within oocytes (or indistinguishable from background) in all nonrodent species. We also note that the antibodies employed for both Foxo1 and Foxo3 were raised to the human protein. This, in addition to our Western analyses and other results, strongly argues against the possibility that these antibodies are suitable only for immunolocalization of the Foxos in rodents.

We do not believe that our results relating to oocyte Foxo3 should be construed as evidence that the Foxos do not regulate PFA in nonrodent mammals. For example, it is possible—and we consider it likely—that the Foxos function collectively and redundantly in these other species to regulate PFA. Consistent with this idea, the Foxos are functionally redundant with respect to several physiologic functions, even in mice [6, 7]. In murine granulosa cells, Foxo1 is highly expressed, while Foxo3 is undetectable by IHC, yet genetic studies showed that Foxo1 and Foxo3 are functionally redundant (i.e., Foxo1 or Foxo3 knockout alone led to no overt phenotype, while simultaneous Foxo1/Foxo3 knockout led to female infertility and severe follicle defects). Moreover, we documented Foxo1 expression in bovine primordial oocytes. Although the pattern in bovine oocytes was not highly conserved in relation to rodents, this finding provides some support for the idea that the other Foxos are expressed and hence could function within primordial oocytes.

Unlike Foxo1 and Foxo3, Foxo4 serves no essential reproductive functions in mice. Foxo4^−/− males and females have normal gonadal morphology and are fertile. We were nonetheless interested in studying the gonadal patterns of Foxo4 expression in mice and other mammals under similarly controlled conditions but were unable to do so for technical reasons. We tested commercially available antibodies under optimal conditions (i.e., employing tissues from Foxo4 wild-type vs. null mice as controls) but were unable to demonstrate specific Foxo4 immunostaining in any tissue (data not shown), including skeletal muscle, where Foxo4 is most abundantly expressed [43]. The antibodies used in this study failed to detect zebrafish Foxos by Western blot, indicating that their utility may be limited to mammals (unpublished data). We also note that we cannot exclude the possibility that Foxo gonadal expression in some mammals might occur at developmental time points not analyzed in this study. We failed to detect Foxo3 expression via IHC in neonatal rhesus, baboon, chimpanzee, and human ovaries but again cannot exclude functional significance in these and other species at similar or earlier developmental stages.

Mammalian spermatogenesis, including the role of SSCs, is remarkably conserved among diverse species. Nonetheless, differences in Sertoli cell function and the organization of the seminiferous epithelium are well described, such as the striking variation in the stage-specific patterns of spermatogenic progression in rodents versus humans [44, 45]. Our finding that Foxo1 is specifically expressed in undifferentiated spermatogonia in all mammals studied argues that Foxo1 function in SSCs is likely universal among mammals. Interestingly, specific Foxo1 expression in granulosa cells was also found in all mammals, suggesting that, unlike Foxo3, both Foxo1 somatic and germline functions appear to be conserved.

These findings have diverse clinical implications. For example, there is considerable interest in primary ovarian insufficiency (POI). While many studies have pointed to a genetic basis for this condition, no common causal genetic variants have been identified [19, 46]. Although the striking “early menopause” phenotype observed in Foxo3^−/− mice raised the possibility that FOXO3 mutations or genetic variants might have a causal role in POI, several studies that have carefully addressed this question have failed to identify such mutations. The lack of Foxo3 expression in human oocytes and the possibility of functional redundancy provide a plausible explanation for these findings [18, 47–49]. Functional redundancy of the Foxos in primordial oocytes (suggested by this study) would make it unlikely that mutation of a single FOXO would have a prominent phenotype, that is, due to underlying genetic redundancy.

There is also considerable interest in other clinical applications that would benefit from the control of primordial follicle activation in vitro. For example, pharmacologic manipulation of explanted primordial follicles grown in 3D matrices may someday prove useful in diverse settings, such as fertility preservation in women receiving chemo- or radiotherapy or even the cryopreservation of primordial follicles by young women for use later in life. One promising general approach being considered is cryobanking of ovarian tissue followed by in vitro follicle maturation. However, the ex vivo control of follicle maturation remains problematic [50]. Our findings suggest that Foxo3 itself is not likely to be a suitable target or biomarker for such efforts. However, pharmacologic manipulation of the PI3K pathway components upstream of the Foxos remains a viable and promising target for such efforts, as has recently been shown [21]. On the other hand, our studies argue for conserved roles for Foxo1 in male and female reproductive function and fertility across mammals.