Control of Stem Cell Self-Renewal in Drosophila Spermatogenesis by JAK-STAT Signaling

See allHide authors and affiliations

Science  21 Dec 2001:
Vol. 294, Issue 5551, pp. 2546-2549
DOI: 10.1126/science.1066700


Stem cells, which regenerate tissue by producing differentiating cells, also produce cells that renew the stem cell population. Signals from regulatory microenvironments (niches) are thought to cause stem cells to retain self-renewing potential. However, the molecular characterization of niches remains an important goal. InDrosophila testes, germ line and somatic stem cells attach to a cluster of support cells called the hub. The hub specifically expresses Unpaired, a ligand activating the JAK-STAT (Janus kinase–signal transducer and activator of transcription) signaling cascade. Without JAK-STAT signaling, germ line stem cells differentiate but do not self-renew. Conversely, ectopic JAK-STAT signaling greatly expands both stem cell populations. We conclude that the support cells of the hub signal to adjacent stem cells by activation of the JAK-STAT pathway, thereby defining a niche for stem cell self-renewal.

A small population of germ line stem cells (GSCs) in a morphologically well-defined environment sustains Drosophila spermatogenesis (1). GSCs attach to a cluster of nonmitotic somatic support cells at the testis apex, called the hub (Fig. 1A). GSC divisions are asymmetric, producing GSCs remaining at the hub as well as differentiating daughters (gonialblasts) displaced away from the hub. Recently, a somatic signal regulating the GSC lineage was found. Although the identity of this signal is unknown, it promotes the GSC-to-gonialblast transition (2, 3). We hypothesized that an opposing signal, likely produced by hub cells, should promote GSC self-renewal.

Figure 1

The ligand Upd is specifically expressed in the hub. (A) The Drosophila testis apex. Germ line stem cells (five to nine cells; yellow) attach to a cluster of about 12 somatic hub cells (red). GSC daughters adjacent to the hub remain stem cells (yellow); daughters displaced from the hub differentiate into gonialblasts (blue). Gonialblasts undergo four synchronous, incomplete mitotic divisions, forming 16 interconnected spermatogonia (light blue; cysts representing the first, second, and third divisions are shown). Later, spermatogonia exit mitosis and enter meiosis, becoming spermatocytes. Cells move progressively further from the apex as differentiation proceeds. Each gonialblast and its progeny are enveloped by two somatic support cells throughout spermatogenesis (cyst cells, colorless) arising from asymmetric divisions of cyst progenitor cells (herein termed SSCs, gray). (B) In situ hybridization to whole adult testes (3); Upd mRNA (4) is localized to the cells of the hub. (C) X-gal activity stain (26); an enhancer detector inserted inUpd (27) is expressed in the hub. Apical is to the left in all panels. Scale bars, 10 μm.

We found that Unpaired (Upd), a secreted ligand activating the highly conserved JAK-STAT signaling pathway (4), is expressed specifically in hub cells (Fig. 1, B and C). Upd signals through the recently identified receptor Domeless (5), activating the JAK kinase homolog Hopscotch (Hop) and the STAT homolog Stat92E, which enters the nucleus to activate target genes (6). Localized expression of Upd by hub cells suggests that Upd activates the JAK-STAT pathway in adjacent stem cells.

To test this possibility, we analyzed the effects of removing Hop and Stat92E on stem cells. In contrast to the wild type, testes from the male-sterile, partial loss-of-function mutanthop[25] (7) contain hub cells but lack stem cells and spermatogonia (Fig. 2, A and B). Analysis of markers for the hub and either stem cell lineage (8) in hop[25] testes (9), and ofhop[2] (null) testes (Fig. 2, C and D) (10), confirms that Hop maintains GSCs and somatic stem cells (SSCs). Because Stat92E is similarly required (see below), we suggest that Upd, signaling through JAK-STAT, maintains either the self-renewing capacity or viability of GSCs and SSCs.

Figure 2

JAK-STAT signaling is required for germ line stem cell self-renewal. (A and B) Confocal section through the apical region of adult testes stained for DNA. In a wild-type testis (A), nuclei of hub cells (outlined), GSCs (arrow), SSCs (small arrowhead), and spermatogonia (large arrowhead) are visible. In a hop[25] testis (B), only hub cells remain (narrow outline); heavy outline denotes testis edge. (C and D) DNA-stained whole third- instar larval testes. A wild-type testis (C) shows brightly staining GSCs and spermatogonia at the apex (arrowhead), diffusely staining spermatocytes distal from the apex (arrow), and brightly staining terminal epithelial cells at the basal end (asterisk). A hop[2] testis (D) is much smaller than in the wild type; notably, GSCs and spermatogonia are absent. A few spermatocytes (arrow) and terminal epithelial cells (asterisk) remain. (E and F) Confocal sections through the apex of testes containing stat92E clones at day 5. Wild-type cells are marked by a nuclear and cytoplasmic signal (red); unmarked cells are null for stat92E but are visible by DNA counterstain (green). In (E), a stat92E null GSC-gonialblast pair (arrow) is at the hub (outlined). In (F), stat92E null spermatogonia (arrow) are distal from the hub (outlined). Scale bars, 100 μm [(C) and (D)], 10 μm (other panels).

We next determined whether Upd signals directly to GSCs, and whether it instructs GSC self-renewal or maintains GSC viability. For example, if Upd maintains GSC viability, GSCs lacking Hop or Stat92E (and thus unable to transduce the signal) would not survive or produce progeny. Conversely, if Upd instructs GSC self-renewal, GSCs unable to transduce the signal would survive but would produce spermatogonia rather than replenish the GSC population. To create GSCs unable to respond to Upd, we induced stat92E null clones (11). Two days later (day 2), stat92E GSCs were frequently detected (Fig. 2E). However, stat92E GSCs were rare by day 5 and were absent by day 9, whereas control GSC clones remained. A different strong loss-of-function allele ofstat92E yielded similar results (Table 1). The finding that stat92EGSCs are initially seen but do not persist shows that Stat92E is required directly in the germ line for GSC maintenance.

Table 1

stat92E null GSCs are not maintained.

View this table:

To distinguish whether Stat92E maintains GSC self-renewal or viability, we tested whether stat92E GSCs were chased into spermatogonia. Because gonialblasts become spermatocytes in about 2 days, marked spermatogonia present at day 5 originate from marked GSCs. stat92E spermatogonia are found in 48% (14/29) of testes at day 5 (Fig. 2F). A similar number of spermatogonial clones (41%, 9/22) is detected in day 5 control testes. Because progeny of stat92E GSCs differentiate rather than die, JAK-STAT signaling is required in the germ line for GSC self-renewal rather than for GSC viability.

If JAK-STAT signaling alone promotes stem cell self-renewal, then ectopic signaling could increase the GSC population. To test this possibility, we used the binary GAL4/UAS system to overexpress Upd (12). The proliferation center in wild-type testes is restricted to the apex. About half (43/91) of the testes ectopically expressing Upd have a striking accumulation of early-stage cells distant from the hub (compare bracketed areas, Fig. 3, A and B). By their morphology and DNA staining characteristics, these cells are a mixture of undifferentiated germ line and somatic cells. This interpretation was confirmed by examining the fusome, a germ cell–specific organelle that is spherical in GSCs and gonialblasts (13). Typically, GSCs are connected to gonialblasts by a cytoplasmic bridge (14); the fusome occupies the bridge. In contrast, spermatogonia have branched fusomes (compare arrow and arrowhead, Fig. 3C). In testes expressing Upd ectopically, the number of cells with spherical fusomes increases markedly over the wild-type number of about 12. Typically, in a field of several hundred ectopic early germ cells, two-thirds of the cells have spherical fusomes. The fusomes are often located in the cytoplasmic bridge between the GSC and gonialblast, as in the wild type (compare arrows, Fig. 3, C and D), indicating that the GSC and/or gonialblast population is greatly expanded. This was verified by analyzing Bag-of-marbles (Bam) expression, which marks two-, four-, and eight-cell spermatogonial cysts (Fig. 3E) (15). Testes expressing ectopic Upd contain a greatly increased number of germ cells lacking Bam (Fig. 3F), consistent with a GSC and/or gonialblast identity. This finding was confirmed by simultaneously analyzing fusome morphology and Bam expression. Bam-negative germ cells have spherical fusomes (9). The population of Bam-positive spermatogonia also expands (Fig. 3F, arrow). These cells likely derive from excess gonialblasts as they differentiate, whereas the gonialblasts derive from excess GSCs. We conclude that ectopic Upd is sufficient to expand the GSC population.

Figure 3

Ectopic JAK-STAT signaling expands the germ line and somatic stem cell populations in adult testes. (Aand B) DNA stain; arrowhead indicates apex. In (A), a wild-type testis contains a small domain of brightly staining GSCs and spermatogonia (bracket); weakly staining spermatocytes are distal from this domain. In (B), ectopic Upd greatly expands the domain of brightly staining cells (bracket) into the region of the testis normally containing spermatocytes. (C and D) Fusome morphology. In (C), a wild-type GSC-gonialblast pair at the hub (red outline) contains a spherical fusome (arrow); a cyst of spermatogonia contains a branching fusome (arrowhead). In (D), ectopic Upd produces ectopic GSC-gonialblast pairs with spherical fusomes (arrow); somatic cells are also generated (arrowhead). (E and F) Bam expression (red) and DNA (green). In (E), cytoplasmic Bam in the wild type is present in mitotic spermatogonia (arrow), absent from GSCs (small arrowhead) at the hub (outlined), and absent from spermatocytes (large arrowhead). In (F), ectopic Upd expands Bam-positive (arrow) and Bam-negative (arrowhead) germ cell populations, magnified in the inset. (G) A testis expressing ectopic Upd; somatic cells (SC) envelop an ectopic GSC-gonialblast pair (GP). (H andI) Anillin expression (green), fusomes (red), and DNA (blue). In (H), Anillin marks all interphase nuclei in the wild type, particularly SSC nuclei (small arrowheads) near the hub (outlined). Their nonmitotic progeny (cyst cells) lack Anillin (outlined; large arrowhead). At metaphase, Anillin is cortical; an SSC is outlined (arrow). Anillin also marks ring canals (yellow, due to colocalization with fusomes, red). In (I), ectopic Upd expands the Anillin-positive somatic cell population (small arrowheads); one in mitosis is outlined (arrow). Fusomes with one ring canal (large arrowhead) indicate ectopic GSC-gonialblast pairs. Ectopic Upd produced by SSCs [(B) and (D)] or GSCs [(F), (G), and (I)] gives indistinguishable phenotypes. All panels show confocal sections except (G), which is an electron micrograph. Scale bars, 10 μm.

In addition to excess GSCs, gonialblasts, and spermatogonia, testes with ectopic Upd contain excess undifferentiated somatic cells (Fig. 3D, arrowhead). This was confirmed using the late interphase marker Anillin, which strongly stains SSCs but decreases as cyst cells differentiate (compare small and large arrowheads, Fig. 3H) (16). In testes expressing Upd ectopically, ectopic Anillin-positive somatic cells are observed (Fig. 3I), consistent with a SSC identity. Occasionally, Anillin is observed cortically, a distribution that occurs only during M phase; hence, these somatic cells are cycling (Fig. 3I, arrow). Because the only cycling somatic cells in wild-type testes are SSCs (Fig. 3H, arrow), this is also consistent with a SSC identity. Conversely, Eyes-absent (17) and S359 (8), which mark cyst cells rather than SSCs, are not expressed in ectopic somatic cells (9). We conclude that the SSC population expands when Upd is ectopically expressed. Ultrastructural analysis of these cells reveals somatic cells (Fig. 3G, SC) encircling GSC and/or gonialblast pairs (Fig. 3G, GP). This intimate association between GSCs and SSCs is similar to that observed in the wild type (14), which suggests that close contact between the two lineages coordinates the expansion of both stem cell populations. Because activation of the Upd-JAK-STAT pathway is sufficient to generate excess stem cells in both lineages, we conclude that Upd normally instructs GSCs and SSCs to undergo self-renewal.

These data reveal a molecular mechanism governing stem cell self-renewal in a stem cell niche in vivo. We conclude that in theDrosophila testis, hub cells are a localized source of the ligand Upd, which activates the JAK-STAT pathway in adjacent stem cells. This signaling ensures that GSCs and SSCs attached to the hub undergo self-renewal while daughter cells that are displaced away from the niche differentiate. Perhaps cells displaced from the hub do not receive sufficiently high levels of Upd to activate the JAK-STAT pathway, and therefore lose self-renewing capacity. This decision to differentiate is reinforced by a subsequent signal, relayed to the gonialblast from the cyst cell lineage, that promotes the transition from GSC to gonialblast. Although the differentiation signal is currently unidentified, its production requires activation of mitogen-activated protein (MAP) kinase in the soma (2, 3). An intriguing parallel to this emerging pathway in stem cell self-renewal is a requirement for JAK-STAT signaling in the maintenance of mammalian embryonic stem (ES) cells (18). The JAK-STAT signal is counterbalanced by the requirement for MAP kinase activation, which promotes ES cell differentiation (19). Because of these parallels, genetic identification of targets of the JAK-STAT signaling pathway in Drosophila germ line and somatic stem cells holds great potential for revealing the molecular basis of stem cell self-renewal in more complex systems.

  • * To whom correspondence should be addressed. E-mail: matunis{at}


View Abstract

Navigate This Article