Triple Foxo knockout more clearly exposes Foxo roles in SSC self-renewal and the initiation of spermatogenesis.

These genetic analyses demonstrated that Foxo1 is necessary for male fertility and is the most important Foxo in spermatogenesis (since Foxo3 and Foxo4 were previously shown to be dispensable for male fertility) (16–18). However, it remained possible that Foxo3 and Foxo4 partially compensated for Foxo1 deficiency (13). Vasa-cre Foxo1^L/LFoxo3^L/LFoxo4^L/L (hereafter referred to as Foxo1/3/4) males had normal gonocyte numbers at birth but exhibited significant testicular hypotrophy by P14 — earlier than that for Foxo1 males, indicating a more severe phenotype (Figure ​(Figure2,2, A and B). Compared with those of Foxo1 testes, Foxo1/3/4 testes contained fewer germ cells and more tubules depleted of germ cells at all time points (Figure ​(Figure2,2, C–E). It is unclear why some tubules become devoid of germ cells while others do not, but such variation has been documented in other mutants that affect the long-term renewal of SSCs, such as Plzf and Taf4b (21–23). Strikingly, there was also a complete spermatogonial arrest, with failure of spermatogenic differentiation and meiotic initiation evidenced by the absence of multilayer spermatogenesis, including spermatocytes and spermatids in most tubules, even those with abundant spermatogonia (Figure ​(Figure2E,2E, Supplemental Figure 2, and Supplemental Figure 3A). Thus, Foxo3 and Foxo4 partially compensate for Foxo1 in spermatogenesis, and simultaneous inactivation of all 3 Foxos more clearly revealed their collective, essential roles in SSC long-term self-renewal and the initiation of spermatogenesis in the first and subsequent waves of spermatogenesis. Notably, Foxo1/3/4 deficiency phenocopies the failure of spermatogenic initiation and meiotic progression observed in Kit mutants, such as Kit^Y719F, which abrogates signaling through PI3K (24).

Open in a separate window

Figure 2

More severe SSC maintenance and meiotic initiation phenotypes in Foxo1/3/4 male mice.

(A) Intact testes. Scale bar: 3 mm. (B) Testis weights. *P < 0.05. (C) Germ cell numbers by counting GCNA⁺ cells per tubule. *P < 0.0001. (D) P7 testes, H&E, and GCNA immunostain demonstrating empty tubules harboring no germ cells. Control testes contained no empty tubules at these time points. (E) Testis histology for mutant or wild-type controls, H&E, or immunostains as indicated. Controls have multilayer spermatogenesis, whereas Foxo1/3/4 tubules have many spermatogonia (confirmed by GCNA) but no multilayer spermatogenesis. Asterisks indicate empty tubules at P7 and P21. Insets in P21 images highlight basal layer of spermatogonia and complete failure of spermatogenic initiation. Larger boxes contain high-magnification views of the smaller boxes. Scale bar: 20 μm.

Foxo1 is regulated during the gonocyte-SSC transition and is required for the induction of Kit.

To begin to understand how Foxo1 controls SSC self-renewal and differentiation, immunolocalization was performed with a Foxo1-specific antibody. Foxo1 was expressed in all gonocytes at P1–P3, while in P21 animals to adult animals, Foxo1 was confined to basal intratubular cells consistent with spermatogonia. Foxo1 was undetectable in the Sertoli cells or any other somatic cell of the testis (P1 to adult) (Figure ​(Figure3A;3A; see also next section). Strikingly, Foxo1 underwent cytoplasmic-to-nuclear translocation during normal development. At P1, the protein was exclusively cytoplasmic, but nuclear translocation began at P3 and increased by P7. At P3 to P7, the translocation of Foxo1 protein was not perfectly synchronized, with Foxo1 being distinctly nuclear in some gonocytes but cytoplasmic in others (Figure ​(Figure3,3, A and B). The biological basis of this variation is unclear but likely reflects differences in rates of gonocyte maturation, with the gonocytes exhibiting nuclear Foxo1 being more mature. By P21, Foxo1 was nuclear and remained so in adults (Figure ​(Figure3,3, A and B, and Supplemental Figure 3B). Thus, we believe Foxo1 is a novel marker of gonocytes, and its cytoplasmic-to-nuclear translocation demarcates the gonocyte-to-SSC transition. Foxo1 testes served as negative controls in these studies, demonstrating the specificity of immunodetection. Note that the complete absence of Foxo1⁺ cells in Foxo1 testes by P7 proved there was no mosaicism with regards to Vasa-cre–mediated recombination (Figure ​(Figure3A). 3A).

Open in a separate window

Figure 3

Foxo1 is specifically expressed in gonocytes and undifferentiated spermatogonia and regulates Kit.

(A) Kit and Foxo1 immunostains at time points and genotypes as shown. P7 Foxo1 testis serves as negative control for Foxo1 immunostaining. Larger boxes contain high-magnification views of the smaller boxes. The Kit inset shows Kit expression in Leydig cells as internal positive control. Other insets illustrate subcellular distributions of Foxo1 and Kit. Scale bar: 20 μm. (B) Subcellular distribution of Foxo1 in spermatogonia. C, cytoplasmic; N, nuclear. (C) Developmental profile of Kit mRNA levels in mouse testis by microarray (arbitrary units on y axis).

Intriguingly, this translocation of Foxo1 into the nucleus preceded the normal induction of Kit. Consistent with previous studies (25), Kit was first detectable in wild-type testes at P3, when Foxo1 first entered the nucleus. Kit increased with continued cytoplasmic-to-nuclear translocation of Foxo1 (Figure ​(Figure3,3, A and B). This induction of Kit protein matched closely with Kit mRNA levels, suggesting that Kit is regulated transcriptionally (Figure ​(Figure3C).3C). Kit was greatly diminished in Foxo1 and Foxo1/3/4 differentiated spermatogonia at P7 and adulthood (Figure ​(Figure3A3A and Supplemental Figure 3, D and E). The presence of diverse steps of spermatogenesis in Foxo1 testes argues that Kit underexpression in these testes is not simply explained by the absence of differentiated spermatogonia that normally express Kit. Plzf⁺ (undifferentiated) and Plzf^– (differentiated) spermatogonia were present in Foxo1/3/4 testes, further arguing against a maturational block as the basis for Kit underexpression (Supplemental Figure 3C). However, it remains possible that some skipping of developmental steps occurs in mice lacking Foxo gene expression, as has recently been described in another spermatogenesis mutant (Dmrt1) (26). The attenuated induction of Kit in Foxo mutants rationalizes the observed defects in meiosis, given that Kit signaling is vital for this process (24).

The converse possibility that Kit regulates Foxo1 (creating a feedback loop) also appeared plausible, given that Kit can activate PI3K-Akt, which in turn regulates (inactivates) the Foxos. We analyzed mice homozygous for Kit^Y719F, a knockin mutation that abrogates Kit signaling through PI3K. As previously reported, Kit^Y719F testes were populated by numerous SSCs that failed to differentiate and initiate meiosis (24). However, there was no alteration in the expression or subcellular distribution of Foxo1 in Kit^Y719F males at P1 to P21 (Supplemental Figure 4). Thus, Foxo1 is required for the induction of Kit expression in differentiating spermatogonia, but the regulation appears unidirectional without evidence of a feedback loop (i.e., Kit is downstream of Foxo1 and acts through signaling intermediaries other than Foxo1 itself).

Foxo1 is a specific marker of undifferentiated spermatogonia in steady-state adult spermatogenesis.

Colabeling studies were conducted to further define the spermatogenic cells expressing Foxo1 in the adult testis. In wild-type testes, Foxo1 and Plzf were always coexpressed (Figure ​(Figure4A),4A), demonstrating that, in steady-state spermatogenesis, Foxo1 expression is restricted to undifferentiated spermatogonia. Next, we studied Gfrα1, which is expressed in single and paired (A_s and A_pr) spermatogonia. Only a subset of Foxo1⁺ cells were also Gfrα1⁺, but all Gfrα1⁺ cells were Foxo1⁺, and furthermore, Foxo1 protein in these cells was predominantly nuclear (Figure ​(Figure4B).4B). These results are consistent with strong Foxo1 activation in SSCs. The abundance of Foxo1 protein in undifferentiated spermatogonia rationalized the unique genetic requirement for Foxo1 in spermatogenesis, as Foxo3 and Foxo4 were undetectable immunohistochemically in spermatogonia or any other cell type in testis sections (our unpublished observations). Finally, Foxo1 and Kit were expressed in distinct cells, as expected given that they mark undifferentiated and differentiated spermatogonia, respectively (Figure ​(Figure4C).4C). Thus, it appears unlikely that Kit is a direct transcriptional target of Foxo1; instead, Foxo1 must act through some less direct mechanism to regulate Kit expression in progenitor cells. These results are summarized in Figure ​Figure4D. 4D.

Open in a separate window

Figure 4

Foxo1 is specifically expressed in undifferentiated spermatogonia in adult testis.

Panels show confocal images of intact tubules. (A) Foxo1 and Plzf coexpression. Shown are two 16-cell clusters. Scale bar: 20 μm. (B) Gfrα1 and Foxo1 expression. Scale bar: 10 μm. (C) Kit and Foxo1 nonoverlapping expression. Scale bar: 20 μm. (D) Schematic illustrating a subset of spermatogonia expressing Foxo1. A_s-B spermatogonia to intermediate (Int) and preleptotene (Pl) spermatocyte are shown.

PI3K-Akt signaling operates through the Foxos in the control of SSC homeostasis.

Previous studies implicated PI3K-Akt signaling in SSC function (27, 28), prompting us to examine whether Foxo1 was an effector of this pathway in spermatogenesis. If so, then phosphatase and tensin homolog (Pten) inactivation should at least partially phenocopy Foxo1/3/4, since Pten inhibits Akt, which in turn inhibits the Foxos (11). Vasa-cre Pten^L/L (hereafter referred to as Pten) testes contained normal numbers of gonocytes at P1. There was an initial wave of SSC expansion, as in Foxo1 and Foxo1/3/4 testes; however, severe defects in SSC self-renewal and differentiation became evident later, resulting in testicular hypo­trophy and sterility (Figure ​(Figure5A).5A). Germ cell numbers were normal up to P7 but decreased by P21, and there was a significant age-dependent increase in the number of empty tubules (Figure ​(Figure5,5, B–D). There was a complete absence of postmeiotic round spermatids (Figure ​(Figure5D).5D). By P21, no germ cells remained attached to the basement membrane, demonstrating exhaustion of the SSC pool. Testes were almost entirely depleted of germ cells by 4 weeks (Figure ​(Figure5D).5D). Thus, Pten shares essential roles in SSC self-renewal and spermatogenesis with the Foxos.

Open in a separate window

Figure 5

Regulation of Foxo1 by the PI3K-Akt pathway, Pten, and control testes.

(A) Intact testes. Scale bar: 3 mm. (B) GCNA⁺ cells per tubule. *P < 0.0001. (C) Empty tubules harboring no germ cells per GCNA. Control testes contained no empty tubules at these time points. (D) Testis histology, H&E stains or immunostains as indicated. Asterisks at P1 highlight gonocytes, which were morphologically normal. Asterisk at 4 weeks shows rare residual germ cell. (E) Foxo1 and Kit immunostains of Pten testes at P7. Scale bar: 20 μm (D and E). Larger boxes contain high-magnification views of the smaller boxes. See Figure ​Figure3A3A for predominantly nuclear localization of Foxo1 and expression of Kit in controls.

Pten loss led to Akt hyperphosphorylation and cytoplasmic localization of Foxo1 by P7, when Foxo1 is normally predominantly nuclear. Based on its weak staining intensity, cytoplasmic Foxo1 also appeared to be rapidly degraded. This Pten-mediated functional inactivation of Foxo1 was associated with defective Kit induction at P7 (Figure ​(Figure5E).5E). Thus, Pten acts at least in part through PI3K-Akt and the Foxos to regulate SSC maintenance and differentiation. The more severe phenotype of Pten versus Foxo1/3/4 suggests that additional effectors mediate other actions of PI3K-Akt in the male germ line.

To further investigate the control of the Foxos by the PI3K pathway, we conditionally inactivated 3-phosphoinositide–dependent protein kinase 1 (Pdk1), which is required for Akt activation and thus has a biochemical role opposite to that of Pten. Whereas Pten promotes Akt hyperactivation and Foxo inactivation, Pdk1 loss conversely results in Akt hypoactivation and constitutive Foxo activation (Figure ​(Figure6A).6A). Vasa-cre Pdk1^L/L (hereafter referred to as Pdk1) males were sterile and their testes were small due to decreased germ cell numbers (Figure ​(Figure6,6, B–D). Histologic analyses revealed a complete failure of multilayer spermatogenic differentiation. Notably, the subcellular distribution of Foxo1 was dramatically altered in Pdk1 spermatogonia. Foxo1 was nuclear in virtually all Pdk1 spermatogonia at time points in which it is normally cytoplasmic (Figure ​(Figure6E;6E; P < 0.0001 at P3 and P14), an effect opposite to that of Pten. Also in sharp contrast to Pten and Foxo, Pdk1 testes did not exhibit an age-dependent SSC self-renewal phenotype. Whereas Pten males were depleted of germ cells by 4 weeks, resulting in empty tubules, Pdk1 males maintained large numbers of SSCs, which continued to proliferate, as evidenced by abundant Ki67⁺ positivity. Spermatogonia were numerous, even at 5 months of age, and were Plzf⁺ or Plzf^–, consistent with undifferentiated and differentiated spermatogonia (Figure ​(Figure6F).6F). In conclusion, although the signals that regulate Foxo in spermatogonia are not fully known, these findings strongly suggest that they act through PI3K-Akt. Furthermore, our results demonstrate that PI3K-Akt signaling must be carefully titrated in vivo to maintain SSC self-renewal and differentiation and argue that the Foxos are pivotal intermediaries of this balance.

Open in a separate window

Figure 6

Regulation of Foxo1 by the PI3K-Akt pathway, Pdk1, and control testes.

(A) Schematic of the PI3K-Akt-Foxo pathway. A receptor (typically a receptor tyrosine kinase [RTK]) results in PI3K activation and synthesis of 3-phosphoinositides (PIP3). Pten metabolizes PIP3 and acts as a pathway inhibitor. (B) Pdk1 testes were small by P14. Testes from 4-month-old males are shown. Scale bar: 3 mm. (C) GCNA⁺ cells per tubule and tubules devoid of germ cells per GCNA. *P < 0.0001. (D) Testis histology (H&E) and immunohistochemistry with antibodies as shown. Inset shows alternating Sertoli cells and spermatogonia (asterisks) on basement membrane with complete absence of differentiation and multilayer spermatogenesis. Ki67 demonstrates that spermatogonia are actively proliferating; the Ki67 index in spermatogonia was more than 10% (data not shown). (E) Foxo1 localization in Pdk1 testes at various ages as shown. Pdk1 inactivation promotes the shift of Foxo1 protein from cytoplasm to nucleus at P3. *P < 0.0001. (F) Persistence of SSCs in aged (5-month-old) males. GCNA shows abundant spermatogonia and absence of empty tubules. Plzf and Vasa staining of intact tubules (confocal microscopy) shows the presence of both undifferentiated (Plzf⁺) and differentiated (Plzf^–) Vasa⁺ spermatogonia in aged Pdk1 testes. Scale bar: 20 μm (D–F).

Tissue processing, immunohistochemistry, and immunofluorescence.

For immuno­histochemistry, tissues were fixed in 10% buffered formalin overnight, embedded in paraffin, and cut into 5-μm sections. Slides were deparaffinized in xylene, hydrated in an ethanol series, subjected to antigen retrieval by boiling in 10 mM sodium citrate, and cooled at room temperature (RT) for 20 minutes. The detection system was Immpress (Vector Laboratories). Tissue sections from experimental and control samples were placed on the same slide to ensure identical processing. For whole-mount immunofluorescence, seminiferous tubules were mechanically dissociated in PBS on ice and fixed overnight in 4% paraformaldehyde. Tubules were dehydrated in a series of methanol washes and stored at –20°C. To rehydrate and permeabilize, tubules were put through a series of washes with methanol/PBS plus 0.1% Tween-20. Tubules were blocked with 1% BSA/PBS at RT (Thermo Scientific Blocker no. 37525) for 2 hours, and primary antibody was added in 0.5% BSA/PBS with 0.02% sodium azide and incubated at 4°C overnight. Tubules were washed 3 times for 10 minutes each time in PBS at RT. Secondary antibody (Invitrogen; Alexa Fluor 555 α-mouse, catalog no. A-21422, or Alexa Fluor 488 α-rabbit, catalog no. A-11008) was added at 1:1,000 in 0.5% BSA for 2 hours at RT followed by DAPI (Pierce Protein Research Products, catalog no. 46290; 1:10,000 in PBS). Tubules were placed on glass slides and mounted in Vectashield (Vector Laboratories). Microscopy was performed with a Leica TCS SP5 confocal microscope with a ×40 oil immersion lens using 543-nm HeNe laser (26% full power) and 488-nm Ar laser (15% full power) excitation.

Antibodies for immunofluorescence and immunohistochemistry.

Antibodies and titers used were Foxo1 (Cell Signaling Technology, no. 2880; 1:200 immuno­histochemistry, 1:50 immuno­fluorescence), Kit (Cell Signaling Technology, no. 3074; 1:200 immuno­histochemistry, 1:50 immuno­fluorescence), Plzf (1:150 immuno­histochemistry, 1:50 immuno­fluorescence), germ cell nuclear antigen (GCNA; 1:150 immuno­histochemistry) (51), Ret (IBL America, no. 18121; 1:20 immuno­histochemistry, 1:50 immunofluorescence), p-AKT (Cell Signaling Technology, no. 9271S; 1:100 immunohistochemistry), and Gfrα1 (R&D Systems, catalog no. AF560; 1:100 immunofluorescence).

Germ cell counts and differentiation index.

After GCNA immuno­histo­chemistry (see above), testis sections were analyzed by counting GCNA⁺ cells per tubule (n = 3 testes). The average number of GCNA⁺ germ cells per tubule was determined per animal in each genotype. A minimum of 50 tubules was included per genotype of P1 to P7 animals or 75 tubules per genotype of P14 to adult animals. Empty tubules were defined by the complete absence of GCNA⁺ cells. The differentiation index was used as a measure of the percentage of tubules showing multilayer spermatogenesis, defined as the presence of any viable germ cells forming an additional layer not associated with the tubular basement membrane. At least 100 tubules were counted per testis, and testes from at least 2 animals were analyzed for each genotype. Student’s 2-tailed t test was used to assess statistical significance (defined as P ≤ 0.05).

Expression profiling and real-time PCR.

cDNA was generated using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, catalog no. 4368814). qPCR was performed as previously described (52) using TaqMan Gene Expression Assays (Applied Biosystems ID numbers, Mm01343388_g1 Dppa4; Mm00490671_m1 Foxo1; Mm00436304_m1 Ret; Mm00456650_m1 Egr2; Mm00842279_g1 Egr4; Mm01240680_m1 Sall4; Mm00521776_m1 Lhx1; Hs99999901_s1 eukaryotic 18s rRNA). Additional information regarding the probe target sequences is available through the Applied Biosystems website ( http://www.appliedbiosystems.com) or through the NCBI TaqMan probe database ( http://www.ncbi.nlm.nih.gov/projects/genome/probe/doc/ProjTaqMan.shtml). PCR reactions were set up using TaqMan Gene Expression Master Mix (Applied Biosystems, catalog no. 4369016) per the manufacturer’s instructions.

Expression profiling.

Testes were dissected from P4 mice and immediately homogenized in 1 ml Tripure (Roche). RNA was prepared per the manufacturer’s instructions and resuspended in RNase-free water (30 μl/testes). Quality and purity of the RNA was checked by spectrophotometry and gel electrophoresis. RNA samples were submitted to the UTSW Microarray Core for labeling and hybridization per standard protocols. Each genotype was submitted in triplicate using independent testis samples and hybridized to the Illumina Mouse-6 V2 BeadChip. Analysis was performed using BeadStudio software with quantile normalization. Microarray data on mouse Kit expression during postnatal development was obtained from the NCBI, Gene Expression Omnibus Profiles, series GSE640 (53). Microarray data from our study was deposited in the Gene Expression Omnibus (series GSE31282; http://www.ncbi.nlm.nih.gov/projects/geo/query/acc.cgi?acc=GSE31282).

We thank Peter Besmer (Sloan-Kettering Institute, New York, New York, USA) for Kit^Y719F mice, Dario Alessi (MRC Protein Phosphorylation Unit, Dundee, Scotland, United Kingdom) for Pdk1^L/L mice, George Enders (University of Kansas Medical Center, Kansas City, Kansas, USA) for the GCNA antibody, and Ned Sharpless, George John, and Wei Guo for technical advice and assistance. This project was supported by award R01HD048690 from the Eunice Kennedy Shriver National Institute Of Child Health and Human Development (NICHD), Award K26RR024196 from the National Center for Research Resources (NCRR), the Lance Armstrong Foundation, and the State of Texas through the Norman Hackerman Advanced Research Program. M.J. Goertz was supported by the UTSW Cell and Molecular Biology training grant 5T32GM008203 from the NIH. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NICHD, NCRR, or the NIH.