Regulation of Cell Fate Decision of Undifferentiated Spermatogonia by GDNF

See allHide authors and affiliations

Science  25 Feb 2000:
Vol. 287, Issue 5457, pp. 1489-1493
DOI: 10.1126/science.287.5457.1489


The molecular control of self-renewal and differentiation of stem cells has remained enigmatic. Transgenic loss-of-function and overexpression models now show that the dosage of glial cell line–derived neurotrophic factor (GDNF), produced by Sertoli cells, regulates cell fate decisions of undifferentiated spermatogonial cells that include the stem cells for spermatogenesis. Gene-targeted mice with one GDNF-null allele show depletion of stem cell reserves, whereas mice overexpressing GDNF show accumulation of undifferentiated spermatogonia. They are unable to respond properly to differentiation signals and undergo apoptosis upon retinoic acid treatment. Nonmetastatic testicular tumors are regularly formed in older GDNF-overexpressing mice. Thus, GDNF contributes to paracrine regulation of spermatogonial self-renewal and differentiation.

The stem cells for spermatogenesis are single cells in the periphery of seminiferous tubules. The stem cells either self-renew by forming single stem cells or they become interconnected pairs of cells destined to differentiate. Such cells divide further into syncytial chains of usually not more than 16 cells that enter mitosis and apoptosis synchronously (1, 2). Stem cells, pairs, and chains are collectively called undifferentiated spermatogonia, which subsequently become differentiating spermatogonia, spermatocytes, spermatids, and sperm cells. All types of undifferentiated spermatogonia are morphologically and molecularly alike, but they can be distinguished by the absence or presence of synchronized mitotic and apoptotic figures (2) and by their spatial relation to differentiating sperm cells. Sertoli cells, the somatic cells of the seminiferous tubules, are paracrine regulators of spermatogenesis (3). These cells probably regulate the maintenance of the stem cell pool and sperm differentiation, but the molecular mechanisms by which this occurs have remained unresolved. Here we show that the dosage of GDNF, normally secreted by Sertoli cells (4), regulates the fate and lineage determination of undifferentiated spermatogonia.

GDNF, a distant member of the transforming growth factor–β family, promotes survival and differentiation of several types of neurons in the nervous system (5, 6) and regulates ureteric branching in the embryonic kidney (7,8). The signaling receptor complex of GDNF includes Ret receptor tyrosine kinase and GDNF family receptor–α1 (GFR-α1) (9). GDNF-, GFR-α1– and Ret-null mice exhibit similar phenotypes and die during the first postnatal day (7, 10). Their testicular morphology at this stage is normal. GDNF promotes Sertoli cell proliferation in vitro (11). To approach the in vivo function of GDNF in spermatogenesis, we analyzed gene-targeted mice with either decreased or increased GDNF expression.

Although most GDNF +/– mice survived to adulthood and were fertile, histological analysis of their testes showed that spermatogenesis was disturbed (12) (Fig. 1, A and B). However, sperm cells were observed in the epididymal ducts and in well-preserved segments (13). In the degenerating tubules, spermatids were in an abnormal position and Sertoli cells engulfed some of them (Fig. 1C). In older GDNF +/− mice, the depletion of the germ cells often resulted in Sertoli cell–only seminiferous tubules without spermatogonia (13) and the cell proliferation rate was reduced (Fig. 1, D and E), reflecting the depletion of spermatogonia.

Figure 1

Segmental testicular degeneration inGDNF +/− mice at 5 (A andB) and 8 (C through E) weeks of age. (A) About every second tubule is atrophic. (B) High magnification from a degenerating segment. Spermatocytes and spermatids are visible, but spermatogonia have disappeared. The peripheral rim consists mostly of Sertoli cells. (C) Mislocation of spermatids (arrows) in an atrophic tubule. The reduction of spermatid number is not statistically significant (18%) as compared to that in wild-type testes. (D) BrdU incorporation in a wild-type control and (E) reduced incorporation in a GDNF+/− testis. Scale bar in (E), 40 μm; bar indicates 200 μm in (A), 100 μm in (B), 60 μm in (C), and 40 μm in (D).

We then overexpressed GDNF in transgenic mice under the testis-specific (13) human translation elongation factor–1α (EF-1α) promoter (14–16). Four independent transgenic founders were analyzed: two males, C10 and C12, and two females, S6 and E19, with transgene copy numbers of 10, 3, 20, and 4, respectively. The male founders were infertile, and further analysis of the transgenic phenotype was done with offspring of the female founders. Their litters developed normally to adulthood and did not show defects in other organs than the testis. The testicular weights were 41 and 66% less than in controls at 4 and 8 weeks of age, respectively (n = 40 animals in both groups). All transgenic males were infertile. In 5 months of continuous breeding (3 males) and a short-term breeding test of 2 weeks (20 males) with FVB females, the transgenic males produced more than 200 vaginal plugs, but sired no pups. Strain dependence of the infertile phenotype was excluded by crossbreeding the transgenic FVB females to NMRI mouse strain.

Testicular morphology of GDNF-overexpressing mice was normal at birth. After 2 to 3 weeks, a chimeric histological pattern was observed. The tubular cross sections not only showed normal spermatogenesis but also displayed large cell clusters (Fig. 2, A and B). Because these cells in the clusters did not show much nuclear heterochromatin and did express a spermatogonial marker, EE2 (17), they could be morphologically classified as type A spermatogonia (13, 18). The clusters gradually degenerated after puberty, resulting in tubular atrophy, and Sertoli cells phagocytosed the dead cells (Fig. 2C). At 10 weeks of age, only remnants of clusters were seen, but a rim of spermatogonia at the periphery of atrophic seminiferous tubules remained (Fig. 2, D and E). No sperm was observed in seminiferous tubules or the epididymis (13).

Figure 2

Features of the spermatogonia in GDNF-overexpressing testes. Insets in (A), (B), (D), and (E) show EE2 labeling of spermatogonia. (A) Wild-type testis and (B) transgenic testis with clusters of spermatogonia (arrow) at 3 weeks of age. (C) High magnification of a live cell preparation (18) from a transgenic seminiferous tubule with dying cells that resemble type A spermatogonia (arrow) and are engulfed by Sertoli cells. n, Sertoli cell nucleus. Histology of (D) normal 10-week-old testis and (E) advanced atrophy with Leydig cell hyperplasia (asterisk) in transgenic testis. Scale bar in (C), 10 μm; bar indicates 100 μm in (A), (B), (D), and (E).

The levels of GDNF and its receptors were determined by Northern blotting in normal and GDNF-overexpressing testes (Fig. 3, A through F) (16). Although GDNF, Ret, andGFR1 were down-regulated in wild-type testes after the first postnatal weeks, their expression stayed high into adulthood in transgenic testes. Also, the endogenousGDNF mRNA levels (16) were elevated (Fig. 3G), reflecting altered cell type ratios. Immunoprecipitation with Western blotting (Fig. 3H) confirmed the high transgene expression. In situ hybridization for GDNF and its receptors is shown in Fig. 4 (19). Ret and GFR1 expression by a spermatogonial subset indicates that this group represents the GDNF-responsive cells. To further specify the target cell type of GDNF, GDNF-overexpressing testes were analyzed with markers for differentiating spermatogonia and Sertoli cells. The clusters did not express c-Kit, a marker for differentiating spermatogonia (12, 20). The distribution of GATA-1–positive nuclei (a DNA binding protein in Sertoli cells) (21) was unaltered, suggesting that Sertoli cell proliferation is not regulated in vivo by GDNF (13).

Figure 3

Northern blotting of GDNF and its receptors in wild-type and transgenic testes. GDNF(A), Ret (B), andGFR1 (C) mRNA levels had been strongly down-regulated since the second postnatal week. In the transgenic testis, the levels of GDNF (D),Ret (E), and GFR1(F) remain high to adulthood. (G) Also, the endogenous GDNF transcripts in transgenic (TG) versus wild-type (WT) testes are elevated (1- and 4.5-kb transcripts, respectively). (H) Western blotting of immunoprecipitated GDNF at 3 and 6 weeks of age. In WT testes, the GDNF protein is not detectable at these stages. Twenty-five nanograms of human GDNF protein serve as a control (right lane).

Figure 4

In situ hybridization for GDNF and its receptors in wild-type (left) and transgenic (right) testes. (A) At 1 week of age, GDNF is highly expressed (red). The inset in (A) depicts GDNF protein expression by Sertoli cells in a newborn mouse. At 2 weeks of age, Ret(B) and GFR1 (C) are expressed in a subset of spermatogonia. They have a “starry sky” distribution. The spermatogonial clusters in an adult transgenic mouse continuously express the GDNF transgene (D),Ret (E), and GFR1(F). There are peripheral rows of Ret- andGFR1–positive cells outside the clusters. The inset in (E) shows Ret protein. Scale bar, 33 μm.

The cell cycle kinetics were analyzed by bromodeoxyuridine (BrdU) incorporation and apoptosis labeling (22).After the first postnatal week, the normal segmental nature of cell proliferation was disrupted in transgenic mice, but the overall proliferation rate of spermatogonia was not enhanced (Fig. 5, A through C). Thus, a differentiation block rather than hyperproliferation probably causes the accumulation of spermatogonia. Apoptosis was increased from the second postnatal week with a peak at 4 weeks of age (Fig. 5, D and E).

Figure 5

Cell cycle kinetics of the spermatogonia in GDNF-overexpressing mice, their contacts with Sertoli cells, and their response to all-trans retinol. (A) In a 3-week-old wild-type testis, S-phase cells are only seen in the periphery of seminiferous tubules. Absent labeling of some tubules (stars) shows the segmental distribution of cell proliferation during normal spermatogenesis. (B) In a 3-week-old transgenic testis, numerous BrdU-labeled cells (arrow) are abnormally distributed in the luminal area of the tubules. (C) The average proliferation indices of spermatogonia in 100 trans-sections of wild-type (squares) and transgenic (triangles) tubules (BrdU-labeled nuclei per 100 spermatogonia) are not significantly different (0.28 ± 0.12 versus 0.25 ± 0.31, respectively;n = 19645 cells), but the peak proliferation index is higher in wild-type than in transgenic mice (Student'st test, P < 0.001). (Dand E) In situ labeling for apoptosis in a 4-week-old wild-type (D) and transgenic (E) testis. Apoptotic cells inside (arrow) and outside the clusters are nine times more frequent in the transgenic than in the wild-type testes (the average ± SD of apoptotic cells in tubular transsection is 0.47 ± 0.16 in wild-type and 4.22 ± 0.23 in transgenic testes). (F) Transgenic mouse testis without retinol treatment at the age of 4 weeks. (G) In its retinol-treated littermate, the clusters have almost disappeared after 8 days of all-trans retinol injections, and spermatogenesis has been partially recovered. The stem cells would have needed at least 22 days to reach the most advanced stages seen here (31). Thus, the differentiating cells are obviously not derived from the spermatogonial clusters but from the disturbed spermatogenesis outside the clusters (see Fig. 1B). Inset, in situ labeled apoptotic spermatogonia in a transgenic mouse injected for 3 days with retinol. (H) Electron micrograph depicts cell contacts (arrows) between a spermatogonium (asterisk) and a Sertoli cell within a cluster of spermatogonia in a 3-week-old transgenic mouse. Scale bar in (H), 0.4 μm; bar indicates 100 μm in (A) through (E), and 60 μm in (F) and (G).

The interconnected spermatogonia enter apoptosis and undergo mitoses synchronously (2). The survey of 366 mitotic figures in spermatogonial clusters of 3-week-old transgenic mice (n = 3 mice) revealed predominantly single figures: 252 singles, 41 pairs, six groups of four, and one group of eight. Accordingly, only single or small groups of apoptotic cells were seen. The predominance of single mitotic figures and the morphological and molecular characteristics of the cells suggest that the clusters were mainly composed of stem cells. However, the changed microenvironment could have caused a premature disruption of the syncytial spermatogonial colonies, and the cluster cells could therefore also represent abnormal undifferentiated spermatogonia at other stages.

A lack of proper cell contacts between the clusters of spermatogonia and Sertoli cells could explain the disturbed spermatogonial differentiation. However, cytoplasmic extensions of Sertoli cells (Fig. 5H) often protruded into the clusters (23). The expression of Ret and EE2 by the clusters further suggests that the differentiation defect is not caused by unspecific distortion of the translational machinery by the transgene expression. This was also excluded by expressing neurturin, another GDNF family member expressed by Sertoli cells (24) in the presence of the EF-1α promoter. These mice showed a different phenotype. They formed no clusters, had normal fertility, and had only a transient delay of spermatogenesis (25).

When rats or mice are fed a vitamin A– deficient diet or when the retinoic acid receptor–α gene is disrupted, differentiation of spermatogonia is blocked (26). We challenged the transgenic spermatogonia with daily injections of all-trans retinol (27).Instead of undergoing differentiation, the clusters underwent apoptosis (Fig. 5, F and G). Thus, the overactivation of Ret or the abnormal localization of transgenic spermatogonia renders them incapable of proper response to a differentiation signal.

Testicular tumors developed regularly in older GDNF-overexpressing mice. The transgenic spermatogonia remained dormant after the clustering period, but invariably started to spread into the interstitium and formed nonmetastatic tumors after a year of age. Of 12 old mice, 10 had bilateral and 2 had unilateral tumors.

The atrophy in adult GDNF-overexpressing testes might be caused by several mechanisms. First, the clusters block fluid flow in seminiferous tubules, which triggers testicular degeneration (28). Thus, obstruction might be the most important reason for the atrophy because spermatogenesis was partially restored after dissolution of the clusters by retinol treatment. Second, the blood-testis barrier, developing around three weeks of age (29), might prevent the protrusion of undifferentiated spermatogonia into the lumen. This would force them to spread horizontally under Sertoli cells. Indeed, sheets of Ret-positive spermatogonia were found in atrophic testes, suggesting that this subpopulation had overgrown and replaced Ret-negative spermatogonia.

Thus, GDNF dosage regulates the differentiation of undifferentiated spermatogonia. At a low GDNF level, spermatogonia favor differentiation, and at a high level, they favor self-renewal (13). The disturbed spermatogenesis in the GDNF+/− mice closely mimics the morphology of many human cases of impaired spermatogenesis and can be used as a new model for male infertility (30). Molecules activating the Ret signaling cascade may provide us with means to restore reduced spermatogonia pools in infertile men. Although the tumors in older GDNF-overexpressing mice remain uncharacterized, their regular development in GDNF-overexpressing mice suggests that GDNF is involved in the pathogenesis of germ line tumors. Furthermore, the effects on spermatogenesis should be taken into consideration when lead molecules activating the GDNF signaling cascade are designed for use in therapy for neurodegenerative disorders.

  • * These authors contributed equally to this work.

  • To whom correspondence should be addressed. E-mail: hannu.sariola{at}


View Abstract

Navigate This Article