Regeneration of Male Germline Stem Cells by Spermatogonial Dedifferentiation in Vivo

See allHide authors and affiliations

Science  28 May 2004:
Vol. 304, Issue 5675, pp. 1331-1334
DOI: 10.1126/science.1097676


Although the ability of engrafted stem cells to regenerate tissue has received much attention, the molecular mechanisms controlling regeneration are poorly understood. In the Drosophila male germline, local activation of the Janus kinase–signal transducer and activator of transcription (Jak-STAT) pathway maintains stem cells; germline stem cells lacking Jak-STAT signaling differentiate into spermatogonia without self-renewal. By conditionally manipulating Jak-STAT signaling, we find that spermatogonia that have initiated differentiation and are undergoing limited mitotic (transit-amplifying) divisions can repopulate the niche and revert to stem cell identity. Thus, in the appropriate microenvironment, transit-amplifying cells dedifferentiate, becoming functional stem cells during tissue regeneration.

In the Drosophila testis, germline stem cells (GSCs) attach to a cluster of quiescent somatic cells called the hub (Fig. 1A), which creates a special localized microenvironment, or niche, by producing the ligand Unpaired (Upd). Upd locally activates the Jak-STAT pathway within GSCs to maintain stem cell fate (1, 2). GSC divisions are stereotypically oriented (3), which ensures that one daughter remains at the hub, while the other (displaced from the niche) differentiates (Fig. 1A).

Fig. 1.

Conditional loss of stat92E causes GSC differentiation. (A) The Drosophila testis apex. Ten GSCs (one shown) attach to the hub. Daughters (gonialblasts, GBs) are displaced from the hub, then undergo four mitotic (transit-amplifying) divisions to form 16-cell clusters of interconnected spermatogonia that exit mitosis and become spermatocytes. Fusomes (red) are spherical in GSCs and GBs; branched in spermatogonia and spermatocytes. Cyst cells (gray), produced by somatic stem cells (gray, at the hub) envelop GBs and form cysts. Color images are confocal sections through the testis apex. (B to H) Testes immunostained for germ cells (with α-Vasa, red), fusomes [with monoclonal antibody (mAb) 1B1, green], the hub (with α-FasIII, green, asterisk), and DNA [with 4′,6′-diamidino-2-phenylindole (DAPI), blue]. (B) Wild-type control shifted to 29°C for 2 days. GSCs contain spherical fusomes (small arrowhead). Spermatogonia and spermatocytes contain branching fusomes (large arrowhead). (C to H) stat92EF/stat92E06346 testes shifted to 29°C. (C) Day 0; a GSC-GB pair (outlined) has a round fusome (arrowhead). (D) Day 1; a four-cell spermatogonial cyst (three cells visible, outlined, arrow), two GSC-GB pairs (outlined, small arrowhead) contact the hub. (E) Day 2; spermatogonia enveloped by cyst cells (outlined, arrow marks one) surround the hub. Most contact it in at least one focal plane; the spermatocyte cyst (large arrowhead) does not. (F) Day 4; spermatogonial (arrow) and spermatocyte cysts (large arrowheads) surround the hub. (G) Day 6; a spermatocyte cyst (outlined, large arrowhead) and a few Vasa-negative cells (somatic stem cells or cyst cells, small arrowhead) remain. (H) Day 13; only the hub remains. (I and J) Testes immunostained to reveal the hub (α-Armadillo, red, asterisk), cell death (ApopTag, green), and DNA (DAPI, blue). (I) Wild-type and (J) stat92EF/stat92E06346 flies shifted to 29°C for 1 day contain dying spermatogonial cysts (outlined); cell death is not detected in GSCs. Scale bars, 10 μm.

Because GSCs null for the Drosophila STAT homolog stat92E differentiate (1, 2), we hypothesized that a temperature-sensitive allele of stat92E (stat92EF) (4) would allow us to reversibly control GSC differentiation. stat92EF/stat92E06346 fruit flies die during embryogenesis at 29°C but are viable and fertile at 18°C (4), with testes indistinguishable from wild type (Fig. 1C). Therefore, to follow the effects of removing stat92E on adult GSCs, we used confocal microscopy to analyze testes from stat92EF/stat92E06346 flies raised at 18°C then shifted to 29°C (5).

The GSCs, spermatogonia, and spermatocytes were distinguished by fusome morphology (6) (Fig. 1, A and B); GSCs contacted the hub; spermatogonia and spermatocytes were displaced away from it in wild-type (Fig. 1B) and stat92EF/stat92E06346 flies at 18°C. Normally, only GSCs (with round fusomes, Fig. 1C) contact the hub. However, after a 1-day shift to 29°C (day 1), one to three clusters of germ cells with branching fusomes contacted the hub in nearly all stat92EF/stat92E06346 testes (95%, n = 19) (Fig. 1D), and a burst of two- and four-cell cysts occurred (Table 1), indicating that multiple GSCs differentiated. Concomitantly, GSCs, then spermatogonia, and finally spermatocytes were lost over time at 29°C (Table 1; representative examples, Fig. 1, C to H). Dying GSCs were not detected in wild-type or stat92EF/stat92E06346 testes at 29°C, and similar numbers of dying spermatogonial cysts were detected in both genotypes (four to five dying cysts per testis, n = 50 wild-type and 104 mutant testes) (Fig. 1, I and J). Thus, consistent with previous findings (1, 2), differentiation accounts for the progressive loss of GSCs and their progeny within stat92EF/stat92E06346 testes at 29°C.

Table 1.

Progression of GSC differentiation in stat92EF/stat92E06346 testes.

Days at 29°C GSCs per testis Two- and four-cell cysts per testis Percentage testes with Testes scored
GSCs Spermatogonia Spermatocytes
0 10.00 ± 2.00 5.79 ± 2.30 100 100 100 25
1 3.37 ± 0.93 11.26 ± 2.23 100 100 100 19
2 0.25 ± 0.60 1.40 ± 1.48 22.50 100 100 40
4 0.03 ± 0.24 0 1.40 50 50 73
6 0 0 0 0 67 18
13 0 0 0 0 0 17

To further characterize GSC differentiation, we used the spermatogonial marker Bag-of-marbles (Bam) to mark four- and eight-cell cysts (table S1) (7); these do not contact the hub (Fig. 2A). In stat92EF/stat92E06346 testes shifted to 29°C for 1 day, Bam-positive cells started to contact the hub (Fig. 2B) and surrounded it by day 2 (Fig. 2C).

Fig. 2.

Differentiating spermatogonial cysts revert to stem cell identity when stat92E function is restored. (A to E) Confocal sections through the apex of testes stained to reveal the hub (α-Armadillo, red, asterisk), fusomes (mAb 1B1, red), four- or eight-cell spermatogonial cysts (α-BAM-C, green, outlined) and DNA (DAPI, blue). (A) Wild-type control. GSCs, GBs and two-cell cysts separate Bam-positive cysts from the hub. (B to E) stat92EF/stat92E06346 testes shifted to 29°C. (B) Day 1; Bam-negative and Bam-positive cells contact the hub. (C) Day 2, Bam-positive cysts surround hub. (D) Day 2, followed by 2 days at 18°C; Bam-negative cells reappear around the hub. (E) Day 2, followed by 10 days at 18°C; the testis apex appears normal. (F) The number of somatic cells (excluding hub cells) in serially reconstructed apical ends of stat92EF/stat92E06346 testes declines progressively at 29°C. (G) In stat92EF/stat92E06346 flies shifted to 29°C for 4 days, a normal number of GSCs returns after recovery at 18°C. (H to J) Confocal sections through the apex of testes as in Fig. 1. The rosette of GSCs (outlined), somatic stem cells (small arrowhead), spermatocytes (large arrowhead), and zone of spermatogonia (line) are indicated. (H) Wild type; GSCs contact the hub; followed by spermatogonia then spermatocytes. (I) stat92EF/stat92E06346 shifted to 29°C for 2 days, then 18°C for 2 days. Newly regenerated GSCs surround the hub; spermatogonia are absent. Spermatocytes (large arrowheads) fill the remainder of the testis apex. (J) After a 4-day recovery at 18°C, GSCs surround the hub; a zone of spermatogonia now separates them from spermatocytes (large arrowhead). Somatic stem cells are present between GSCs. Scale bars, 10 μm.

We next asked whether restoring stat92E function would restore lost GSCs. When stat92EF/stat92E06346 flies shifted to 29°C for 2 days were returned to 18°C for 2 days, Bam-negative cells (including germ cells, indicated by fusomes) returned around the hub in most testes (76%, n = 21) (compare Fig. 2C and 2D). With additional time at 18°C, the Bam-negative zone expanded in 79% of testes (n = 14), which suggested that functional GSCs returned to the niche (compare Fig. 2D and 2E). Therefore, we quantified GSCs in recovering testes (5). When stat92EF/stat92E06346 flies shifted to 29°C for 1 day were returned to 18°C for 4 days, the number of GSCs increased significantly (from 3.37 ± 0.93 to 6.71 ± 1.62 GSCs per testis, P = 0.0154). Thus, restoration of Jak-STAT signaling restores lost GSCs.

If missing GSCs are replaced by symmetric divisions of remaining GSCs, as is thought to occur in the Drosophila ovary (8), testes without GSCs should not regain them after restoration of Jak-STAT signaling. By day 2, only one or two GSCs remained in 22.5% of testes; the rest completely lacked GSCs (Table 1). To our surprise, after a brief recovery at 18°C (2 days), the number of testes containing GSCs increased from 22.5 to 75.8% (n = 33 testes; P = 2.29 × 10–9). Therefore, testes completely lacking GSCs regained GSCs. These testes appeared strikingly different from wild type; although GSCs surrounded the hub, they contacted spermatocytes, because intermediate cells (spermatogonia) were absent (compare Fig. 2 H and I). After a longer recovery period, 76.9% of testes (n = 13) regained a normal zone of spermatogonial cysts (Fig. 2J) expressing Bam (Fig. 2E). Thus, regenerated GSCs are functional. Similarly, when stat92EF/stat92E06346 flies shifted to 29°C for 4 days recovered at 18°C for 4 to 22 days, the number of testes containing GSCs rose from 1.4% (n = 73 testes) to 13.9% (n = 72 testes; P = 0.0017). Again, a normal spermatogonial zone re-formed (9), and the number of GSCs per testis returned to that of wild type (Fig. 2G), which confirmed that functional GSCs are regenerated in testes completely lacking GSCs.

We suspected that spermatogonia reverted to stem cell fate, because all or part of the testes from stat92EF/stat92E06346 flies shifted to 29°C for 2 and 4 days contained spermatogonia (Table 1). Furthermore, males shifted to 29°C for 6 days (which lacked spermatogonia but contained spermatocytes, Fig. 1I, Table 1) never regained lost GSCs when returned to 18°C (n = 42 testes). Thus spermatogonia, not spermatocytes, revert to GSCs. Interestingly, somatic stem cells always returned in testes undergoing repopulation (n = 81) (compare small arrowheads, Fig. 2, H to J) but were absent in 14% (n = 4 out of 28) of day 2 testes (Fig. 2F). This may explain why, although 100% of day 2 testes contain spermatogonia, only 75.8% undergo repopulation; perhaps somatic stem cells and spermatogonial cysts cooperate during the process of cyst breakdown.

Because spermatogonia that produce GSCs were in clusters of 8 or 16 interconnected cells (5) and enveloped by somatic cyst cells separating germ cells from the hub (9), these multicellular cysts must break apart to produce GSCs when signaling is restored. Therefore, we examined the distribution of Anillin, which marks stable ring canals between spermatogonia (Fig. 3A, arrow), as well as the transient ring canal remnant specifically marking the completion of divisions between GSCs and gonialblasts (GBs) (Fig. 3A, arrowhead) (10, 11). In stat92EF/stat92E06346 flies shifted to 29°C for 2 days, large spermatogonial cysts containing branching fusomes decorated with multiple ring canals surrounded the hub (Fig. 3B). After recovery at 18°C, multiple GSCs each containing a transient ring canal remnant appeared in 3 out of 31 testes (Fig. 3C, arrowhead). However, GBs and two-cell cysts were not found in these serially reconstructed testes. Therefore, the remnants represent spermatogonial cyst breakdown, not GSC-GB divisions. Remnants were only detected in cysts at the hub. Thus, when Jak-STAT signaling is restored, ring canals between interconnected spermatogonia adjacent to the hub close, which allows spermatogonia to pinch off from the cyst and to become stem cells.

Fig. 3.

Spermatogonial cysts break apart to form GSCs. (A to C) Confocal projection through testis apex, ring canals (α-Anillin, green) and fusomes (mAb 1B1, red) marked. (A) Testis from wild-type fly shifted to 29°C for 2 days. A GSC-GB pair (outline) completing cytokinesis shares a ring canal remnant (arrowhead) adjacent to the spectrosome. Four-, 8-, and 16-cell spermatogonial cysts (top to bottom) containing branching fusomes and multiple ring canals are outlined. (B) stat92EF/stat92E06346 shifted to 29°C for 2 days. Spermatogonial cysts (outlined) of 16 cells (top and middle) and 8 cells (bottom) contact the hub. (C) stat92EF/stat92E06346 shifted to 29°C for 2 days, then 18°C for 1 day. Ring canal remnants were found in four GSCs in this testis (one in this plane, small arrowhead, and inset). A mitotic 4-cell spermatogonial cyst (outlined, arrow) and a 16-cell cyst (outlined, large arrowhead) are visible. (D to F) Confocal projections through serially reconstructed testes; BrdU incorporation (green), fusomes (1B1, red), hub (FasIII, red, asterisk) and DNA (DAPI, blue) are shown. (D) Testis from stat92EF/stat92E06346 shifted to 29°C for 4 days, then labeled by BrdU incorporation in vivo. A 16-cell spermatogonial cyst with a branching fusome contacts the hub (outlined). (E) stat92EF/stat92E06346 shifted to 29°C for 4 days, then to 18°C for 2 days. Four GSCs with dot fusomes (small arrowhead) contact the hub (three are visible in this plane); these GSCs originated from an eight-cell cyst, producing a four-cell cyst in a lower focal plane (not visible). (F) Four labeled GSCs contact the hub (each containing a round fusome, arrowhead). An accompanying partial cyst (4 cells, outlined, arrow) and a 16-cell cyst (large arrowhead) are similarly labeled. Scale bars, 10 μm.

We also used another strategy to observe products of cyst breakdown. Because spermatogonia labeled before restoration of signaling should become labeled GSCs after signaling is restored, we fed bromodeoxyuridine (BrdU) to flies that had been shifted to 29°C for 4 days (which lacked GSCs but contained spermatogonia, Table 1); labeled spermatogonial cysts were detected near the hub in 6% (n = 2 out of 33) of testes before recovery (Fig. 3D) (5). As expected, after a 2-day recovery, 6% of testes (n = 2 out of 32) regained labeled GSCs. Consistent with our above data (Fig. 3C), multiple marked GSCs were detected, but only near the hub (Fig. 3, E and F; Table 2). The level of labeling was strong in each GSC, indicating that labeled GSCs were breakdown products of a labeled cyst, rather than a single labeled GSC (which would produce diluted levels of label). Furthermore, in each instance where labeled GSCs were found, an adjacent spermatogonial cyst with labeling corresponding to that of the newly formed GSCs was present (Table 2); these most often represented products of eight-cell cysts (Table 2). This confirms that single spermatogonial cysts break apart, to regenerate multiple single GSCs when Jak-STAT signaling is restored.

Table 2.

Labeled GSCs and accompanying partial spermatogonial cysts in stat92EF/stat92E06346 testes after BrdU incorporation and 2-day recovery at 18°C. Cysts were labeled to the same extent as adjacent GSCs and contained fewer than 16 cells.

Days at 29°C Sample number Labeled GSCs Accompanying spermatogonia (n)
4 1View inline 4 4
2 3 13
2View inline 3 7 1
4 5 4
5View inline 4 4
6 4 5, 2
7 4 4
8 4 8, 4
9 3 5
10 2 6
11 1 0
12 1 0
  • View inline* Fig. 3E.

  • View inline 22% of testes contained one or two remaining GSCs but were distinguishable from GSCs derived from cyst breakdown, because the former contained a single labeled GSC and no accompanying labeled spermatogonia after recovery (samples 11, 12).

  • View inline Fig. 3F.

  • These data reveal a mechanism for replacement of lost stem cells in an intact stem cell niche. Removal of Jak-STAT signaling causes GSCs to differentiate into clusters of interconnected spermatogonial cysts that express the differentiation marker Bam but remain within the niche. When Jak-STAT signaling is restored, spermatogonia adjacent to the hub lose expression of Bam, pinch off from 8-cell (or possibly 16-cell) cysts, and form single functional GSCs contacting the hub. Thus, there is considerable plasticity among Drosophila spermatogonial cells, and restoration of Jak-STAT signaling to the testis is sufficient to induce dedifferentiation of spermatogonia into GSCs.

    In mammalian testes, as in most adult mammalian tissues, the precise identity and location of the stem cells, as well as the mechanisms by which repopulation occur, are unknown. However, mammalian testes that have undergone severe germ cell loss commonly contain clusters of intermediate numbers of spermatogonia(e.g., 3, 5, or 11 cells, rather than 2n), which suggests that breakdown of interconnected spermatogonia represents an emergency strategy to regain stem cells (12). Therefore, the ability of spermatogonia to dedifferentiate into GSCs may be highly conserved. In addition, the ability of transit-amplifying cells to become stem cells may be shared by other stem cell systems [reviewed in (13)]. Thus, either stem cells or their differentiating progeny may function as stem cells, if they can respond to appropriate signals from the niche. The ability to study induced dedifferentiation in an intact stem cell niche should greatly aid our understanding of the mechanisms underlying tissue regeneration in vivo.

    Supporting Online Material

    Materials and Methods

    SOM Text

    Table S1


    References and Notes

    View Abstract

    Stay Connected to Science


    Navigate This Article