This study demonstrates that the Foxos, particularly Foxo1, served critical and nonredundant roles in SSC homeostasis both in vitro and in vivo. In their absence, SSCs displayed a decreased ability to self-renew and differentiate. Foxo function in the male and female germ line shares some similarities. First, the Foxos are not required for the differentiation or early development of the germ lineage in males or females. Rather, the Foxos evolved to control gametogenesis within the gonad itself. Whereas Foxo1 is highly expressed in undifferentiated spermatogonia, Foxo3 is highly expressed in primordial oocytes, in which it serves to restrain their growth (18
). The unique genetic requirement for Foxo1
in males and Foxo3
in females mirrors their high expression at discrete cellular stages in spermatogenesis or oogenesis. Another intriguing similarity is that both Foxo1 and Foxo3 undergo relocalization from the cytoplasm to the nucleus postnatally, coincident with their initial functional requirement, suggesting that the timing of their activation is critical for normal germ line development and function. In mammals, there are 3 principal Foxo
genes (and 1 more distantly related member, Foxo6
), but there is a single ancestral homolog in Caenorhabditis elegans
, which controls life span and stress resistance (12
). Taken together, studies from our laboratory suggest that this gene duplication conferred an evolutionary advantage by permitting the Foxos to adopt distinct but controlling roles in oogenesis and spermatogenesis, 2 processes with many similarities but also important differences.
One of the more interesting and potentially unexpected findings in this study is that Foxo1 controlled various aspects of spermatogenesis, from long-term SSC self-renewal to the initiation of spermatogenesis and meiosis. The Foxo1
phenotype also indicated a requirement in spermiogenesis. The control of diverse steps of spermatogenesis is surprising, because Foxo1 protein is restricted to very early progenitors (undifferentiated spermatogonia). We conclude that in addition to regulating the expression of a network of factors required for spermatogenesis, Foxo1 initiates a cascade of events that influence later steps of spermatogenesis. It is also possible that Foxos are biologically active and regulate gene expression in subsequent steps of spermatogenesis, even if individual Foxos are much less abundant than Foxo1 in undifferentiated spermatogonia. The underexpression of Kit in Foxo
mutants explains the failure of meiotic initiation and represents a molecular foothold in understanding this cascade. However, further investigations will be needed to understand how the individual Foxo targets identified in this study contribute to the diverse steps of spermatogenesis. The Foxos join a growing network of transcription factors — Plzf
— that control early steps of spermatogenesis, including SSC self-renewal (21
). Further dissection of functional interrelationships among these factors and their order of interactions is warranted. In particular, additional studies such as ChIP-seq will be needed to formally distinguish between indirect and direct Foxo targets, ascertain whether the Foxos act directly upon the Ret promoter or via other intermediaries, and also determine whether the 3 Foxos act upon distinct direct targets.
The developmental relocalization of Foxo1 during the gonocyte-SSC transition implies that the subcellular localization of the Foxos is an important determinant of their biological activity, consistent with prior observations (11
). Our studies of Pdk1
mutants showed that the PI3K-Akt pathway is the principal pathway regulating Foxo1 subcellular localization and activity in the context of spermatogenesis. Foxo1 is highly expressed and in an active (nuclear) state in SSCs and is intimately linked with the “stem state” in spermatogonia. For example, in Pten
testes, Foxo1 was inactivated, and SSCs became depleted. In contrast, in Pdk1
testes, Foxo1 remained activated (nuclear), and SSCs were not depleted. These findings also imply that the inactivation of Foxo1 that occurs physiologically triggers the initiation of spermatogenesis. What, in turn, are the extracellular signals influencing this delicate balance of PI3K and hence Foxo1 activity? Candidates are a variety of growth factor ligands that bind cognate receptor tyrosine kinases or G protein–coupled receptors that act via PI3K. Our studies of the KitY719F
mutant argue that Kit ligand is not critical, even though Kit reexpression coincided with Foxo1 relocalization. In addition, the nuclear and active state of Foxo1 in Gfrα1+
spermatogonia further argues that the Gdnf-Gfrα1/Ret signaling axis is not the principal regulator of Foxo activity in vivo (if that were the case, pathway activity would drive Foxo1 out of the nucleus), although this question merits further exploration.
We demonstrated essential roles for Pdk1 and Pten in the male germ line. Interestingly, both factors are required for spermatogenic differentiation, as evidenced by the absence of multilayer spermatogenesis. Pten inactivation resulted in the extinction of SSCs even before the onset of sexual maturity, whereas in Pdk1 testes, SSCs continued to proliferate in aged males. These results show that SSC maintenance and differentiation depend critically on the proper balance of PI3K-Akt activity and suggest a model in which excessive PI3K activity promotes excessive stem cell loss (see below). The Pdk1 phenotype also suggests that PI3K hypoactivity leads to a failure of spermatogenic progression and meiotic initiation. Although interpretation of these phenotypes may be complicated by the diverse functions of this pathway at multiple points of spermatogenesis, our results strongly argue that the Foxos and their targets are pivotal mediators of this balance.
Our results seem at odds with a report that Pten
inactivation led to enhanced germ cell production and testicular teratomas (42
). In the prior study, conditional Pten
deletion was performed with TNAP-cre
, which is induced in primordial germ cells by E13.5, earlier than that with Vasa-cre
. All TNAP-cre Pten
males developed multiple bilateral teratomas by birth, whereas we did not observe teratomas or any other type of abnormal cellular proliferation. It is possible that the somewhat earlier timing of Pten
deletion promoted teratoma formation. Our results, which seem paradoxical given the role of Pten
as a cell growth and tumor suppressor, are more in line with those from studies of Pten
in HSCs, in which conditional inactivation resulted in abnormalities in differentiation and long-term decline of HSCs due to decreased self-renewal potential (43
). These HSC phenotypes can be partially rescued by rapamycin, an mTOR inhibitor. Plzf, a transcription factor needed for SSC maintenance, indirectly regulates mTOR activity (45
). Thus, it will be interesting to study the relative contributions and interactions between PI3K-Akt-Foxo and the mTOR pathways in SSCs.
Recent studies have implicated the Foxos in stem cell maintenance. The Foxos, particularly Foxo3, coordinately regulate neural stem cell homeostasis through genes influencing stress responses and oxygen metabolism (15
). The Foxos also regulate HSC differentiation and assist long-term maintenance by protecting against oxidative stress. In contrast to these studies, our transcriptomic analyses did not yield targets relevant to oxygen metabolism or cellular stress responses. Instead, the Foxos appeared to control a network of genes unique to spermatogenesis, consistent with diverse studies showing that Foxo functions are biologically numerous and highly context-dependent (10
The regulation of Ret rationalizes to a large extent the observed Foxo defects in long-term SSC maintenance. By in situ methods, Ret was virtually undetectable in Foxo1
undifferentiated spermatogonia in vivo. In enteric neuron progenitors, Ret acts through diverse pathways, including Ras-MAPK, PI3K-Akt, PLC-γ, and Src pathways, to regulate proliferation, differentiation, and long-term survival (46
). Ret also acts through these pathways to promote SSC survival and proliferation (2
). Ret inactivation results in severe defects in SSC proliferation and differentiation by P7, eventually leading to SSC depletion (8
). The similarity of the Ret and Foxo phenotypes and these previous studies together strongly argue that Ret downregulation accounts for the observed Foxo phenotypes in SSC maintenance. Finally, the possibility that Foxos control Ret expression in diverse neuronal populations also merits further investigation, particularly as the Foxos promote neuronal survival (47
The studies presented here, which were based on in vivo analyses of diverse genetic models, including conditional germ line mutants, imply an intimate relationship between Foxo1 expression and stem cell potential. They also argue that the Foxos function in a cell-autonomous manner to maintain SSC potential. Future dissection of the role of the Foxos and the PI3K pathway in the regulation of SSC long-term maintenance would likely benefit from the development and exploitation of in vitro models. One such approach would entail the derivation and culture of SSCs that are null or conditionally null for the Foxos or other PI3K pathway components. Such in vitro models, perhaps coupled with transplantation experiments, would enable a diversity of other studies, such as cell cycle analyses, that would likely lead to further insights into the diverse molecular mechanisms by which the PI3K-Foxo pathway regulates the long-term maintenance and differentiation of SSCs.
In conclusion, the Foxos are pivotal regulators of SSC self-renewal and differentiation (summarized in Figure ). Like other adult stem cells, SSCs represent a finite reserve that needs to be maintained throughout life. The importance of the Foxos in organismal aging (12
), combined with our discovery of their function in SSC self-renewal suggests that the Foxos function throughout life to protect this finite resource. It seems likely that abnormalities in SSC maintenance account for some cases of male infertility, one of the most common conditions for which medical attention is sought. Conversely, the overexpansion of cells with stem cell–like properties is thought to be a requisite for cancer. Testicular cancer is the most common cancer in young men, and most of these tumors are composed of cells with SSC-like properties (48
). Thus, these findings and future investigations of the PI3K-Akt-Foxo pathway in SSCs using the models described herein should lead to important insights into the etiology of these clinically important but poorly understood conditions.
Model for essential role of Foxos in control of SSC self-maintenance and initiation of spermatogenesis.