Approximately ten somatic cells, called the hub, are found at the apical tip of the Drosophila testis (Fig. 1a)2. Germline stem cells (GSCs) and somatic stem cells (SSCs) surround and are in contact with hub cells. Whereas GSCs sustain spermatogenesis, SSCs produce cyst cells that encapsulate the maturing germ cells and ensure differentiation6,7. Hub cells secrete the growth factor Unpaired (Upd)3,4, which activates the JAK-STAT signal transduction pathway in adjacent stem cells. JAK-STAT signalling is necessary for stem cell maintenance and is sufficient to specify self-renewal of both GSCs and SSCs in the testis3,4,8.

The apical hub is typically described as a post-mitotic, static structure. However, in agametic flies, SSCs proliferate and express hub markers, leading to an apparent expansion of the apical hub9. Furthermore, recent studies have noted that hub cell number and function decrease with age10,11, indicating that the stem cell niche in the testis is dynamic. Hub cells and SSCs share numerous features, including similar gene expression patterns and close association with GSCs and each other2,12; however, the precise relationship between SSCs and hub cells has not been explored.

We proposed that SSCs may serve as a source of cells that contribute to the apical hub and, consequently, the stem cell niche. To address whether SSCs give rise to hub cells, positively marked β-galactosidase-expressing (β-gal⁺) SSCs were generated using mitotic recombination, a technique typically used for lineage tracing analyses. Labelled SSCs were generated by heat-shocking flies of the appropriate genotype to initiate FLP-mediated recombination, resulting in reconstitution of the α-tubulin promoter upstream of the lacZ gene13. Hub cells were identified by immunolabelling with antibodies to Fasciclin III (FasIII, Fig. 1b, third panel)14 or DE-cadherin (Fig. 1c)15, cell-surface proteins concentrated at hub cell junctions. SSCs and early cyst cells were identified by immunolabelling with antibodies to Traffic Jam (TJ), a transcription factor that is strongly expressed in early cyst cell nuclei16 (Fig. 1b, f, first panels) and weakly expressed in hub cells (Fig. 1f, fourth panel).

Wild-type SSC clones were identified as β-gal⁺ cells adjacent to the apical hub and surrounding germ cells (Fig. 1c and Supplementary Table 1). A series of heat shocks after eclosion (hatching) led to at least one β-gal⁺ SSC in 57% (n = 47) of testes from 3-day-old males analysed 1 day after heat shock. At 5 days after heat shock, 28% (n = 74) of testes contained at least one β-gal⁺ SSC, and this frequency decreased to 10% (n = 183), 10% (n = 225) and 4% (n = 142) at 10, 15 and 30 days after heat shock, respectively (Supplementary Table 1).

In addition to β-gal⁺ SSCs, β-gal⁺ hub cells were also observed that co-labelled with DE-cadherin and FasIII. In fact, β-gal⁺ cells were found within the hub in 34% (n = 47), 47% (n = 74) and 60% (n = 225) of testes from males at 1, 5 and 15 days after heat shock, respectively (Fig. 1e, f and Supplementary Table 2). No β-gal⁺ cells were observed in flies not exposed to the heat-shock protocol (Supplementary Table 2).

Previous reports concluded that hub cells are post-mitotic2,9; however, it is possible that hub cells undergo rare divisions to become marked during recombination. To test whether hub cells are mitotic, dividing cells in the testis were labelled with 5′-bromo-2-deoxyuridine (BrdU), which is incorporated into newly synthesized DNA during S phase. Flies were fed (‘pulsed’) BrdU for 30 min and aged (‘chased’) for up to 15 days. Subsequently, labelled testes were costained with antibodies to BrdU, as well as to the hub marker FasIII. No BrdU-positive (BrdU⁺) hub cells were detected after a 1-day chase (n = 82), although cells adjacent to the hub were clearly labelled (Fig. 2a).

However, BrdU⁺ hub cells were observed 3-10 days after labelling (Fig. 2b, d). BrdU⁺ hub cells were present in 4% (n = 143), 8% (n = 103) and 3% (n = 96) of testes assayed at 5, 7-8 and 10 days after labelling (Supplementary Table 3). Moreover, these BrdU⁺ cells co-expressed FasIII (Fig. 2b) and an upd reporter, indicating that these cells function as hub cells (Fig. 2d). These data are consistent with previous results indicating that hub cells are post-mitotic2,9 and support the hypothesis that mitotically active SSCs act as a source of cells that can contribute to the apical hub.

SSCs are reported to be the only dividing somatic cells in the testis; however, we observed two distinct populations of somatic cells dividing near the testis tip. One group, which constituted 69% (20 out of 29) of phospho-histone-H3-positive somatic cells (n = 453 testes), appeared to be immediately adjacent to the hub, similar to GSCs (Fig. 3a, b). We also observed somatic cells that were dividing 1-2 cell distances away from the hub (9 out of 29) (Fig. 3c, d). Several scenarios could explain these observations: there are two SSC populations, one which gives rise to hub cells and another that sustains the cyst cell lineage, or there are SSCs that produce both hub cells and a transient amplifying somatic cell population.

To distinguish between these possibilities, we adjusted the heat-shock regime to label, on average, only one β-gal⁺ somatic cell and analysed the clones derived from these marked cells (Supplementary Fig. 1). Thirteen per cent of testes examined contained marked somatic cells adjacent to the hub at 1 day after heat shock (n = 185), which decreased to 9.5% (n = 190), 3.3% (n = 212) and 2.8% (n = 72) of testes at 5, 10 and 15 days after heat shock, indicating a half-life for SSCs between 5-10 days. Single, marked somatic cells displaced away from the hub were found initially in 11.3% of testes examined at 1 day after heat shock, but this number decreased to 0.5% by 5 days after heat shock, which is consistent with these cells being transient non-stem-cell clones (Fig. 3e, f and Supplementary Fig. 1). In contrast, the number of testes that contained marked hub cells increased from 11.9% at 1 day after heat shock to 25.8%, 24.5% and 18.1% at 5, 10 and 15 days after heat shock, respectively. Clones containing all three cell types were observed in 22% (n = 18) and 14% (n = 7) of testes that contained marked SSCs at 5 and 10 days after heat shock, respectively (Fig. 3g-i). From these data we conclude that multipotent SSCs self-renew and contribute to both hub and cyst cell lineages, whereas dividing cyst cells, which we call cyst progenitor cells (CPCs), expand the pool of cells capable of encapsulating newly divided gonialblasts and maturing spermatogonia to ensure terminal differentiation6,7,17 (Supplementary Fig. 1).

To identify factors required for incorporation of cells into the apical hub, SSCs were generated that were mutant for genes expressed in both cell types: DE-cadherin, which is encoded by the shotgun (shg) gene, and the transcriptional repressor Escargot (Esg). DE-cadherin is expressed in cyst cells and is strongly enriched in the hub (Fig. 1c, d). SSC clones were generated that were homozygous mutant for either the loss-of-function shg^IG29 or amorphic shg^IH allele. SSC maintenance and frequency of marked hub cells were assayed at various time points. In this experiment, marked cells and their progeny subsequently become permanently labelled by ubiquitous green fluorescent protein (GFP) expression (Fig. 4).

Heat-shocked wild-type testes possessed GFP⁺ GSCs and SSCs, as well as GFP⁺ hub cells that co-stained with FasIII and DE-cadherin (Fig. 4 a-c, e). In contrast to wild type, shg mutant GSC and SSC clones were not maintained (Fig. 4d, Supplementary Fig. 2 and Supplementary Table 1), indicating that DE-cadherin has a role in stem cell maintenance in the testis, similar to its role in the ovary18,19, presumably by holding stem cells within the niche and close to self-renewal signals.

Marked hub cells were observed in 14% (n = 144), 35% (n = 114) and 65% (n = 110) of wild-type testes examined at 5, 10 and 15 days after heat shock, respectively (Fig. 4e and Supplementary Table 2). Notably, progeny of DE-cadherin mutant SSCs contributed to the apical hub at a frequency similar to progeny from wild-type SSCs (Fig. 4f and Supplementary Table 2). These data indicate that although DE-cadherin is required for SSC maintenance, it is not absolutely required for mediating the contribution of SSC progeny to the hub.

To confirm that shg is not required in hub cells for maintaining the apical hub, we used RNAi-mediated knockdown of shg expression in hub cells. A FasIII⁺ apical hub was detected in 100% of testes from 1-day-old (n = 12), 10-day-old (n = 15) and 20-day-old males (n = 28), despite a reduction in DE-cadherin expression in hub cells (Supplementary Fig. 3). Testes collected at 20 days also displayed normal expression of a upd reporter (98%, n = 115) and contained TJ⁺ (100%, n = 25) cells near the apical tip (Supplementary Fig. 3). These data support our findings that DE-cadherin is not absolutely required in hub cells to maintain a functional stem cell niche.

However, shg is required in SSCs and early cyst cells for maintaining the apical hub. Knockdown of shg in all SSCs and early cyst cells resulted in a decrease in the number of TJ⁺ cells in 1-day-old males, consistent with a role for shg in SSC maintenance (Fig. 4g, h and Supplementary Fig. 4). Surprisingly, decreased levels of DE-cadherin were also observed in hub cells (Supplementary Fig. 4). In 29% (n = 42) of 15-day-old and 44% (n = 36) of 20-day-old males, the apical hub was severely diminished or lost, as determined by FasIII expression (Fig. 4i, j and Supplementary Fig. 4). These data support our model that SSCs act as a source of cells to maintain the apical hub.

The transcriptional repressor Escargot is expressed in many tissues, including GSCs, early cyst cells and hub cells in the testis20,21. Males carrying a viable, hypomorphic allele of esg, called shutoff, exhibit loss of apical hub cells during development (J.V., manuscript in preparation). Therefore, we hypothesized that esg may be required for regulating the contribution of SSCs to the hub. Mutant labelled SSCs were generated using two amorphic esg alleles, as described above.

Unlike progeny from shg mutant SSCs, progeny from esg mutant SSCs did not contribute to the hub at the same frequency as wild-type controls: esg^L2 mutant GFP⁺ hub cells were observed in 5% (n = 38), 3% (n = 40), 0% (n = 36) and 4% (n = 52) of testes examined at 1, 5, 10 and 15 days after heat shock, respectively (Supplementary Table 2). In instances when esg mutant GFP⁺ hub cells were observed, normal hub morphology was often severely disrupted (Fig. 4k, l). These data suggest that esg regulates either the contribution of SSC progeny to the hub, perhaps by facilitating the cell fate transition between SSC and hub cell, or maintenance of hub cell fate.

To explore Esg function further, we used the agametic oskar (osk) mutant phenotype, in which SSCs proliferate and express hub markers, resulting in an apparent expansion of the apical hub9. If Esg is required for mediating the transition of somatic cyst cells to the apical hub, we predicted that the expansion of FasIII⁺ cells would be blocked in an esg;osk double mutant background. In contrast to the expansion of FasIII⁺ cells in 82% (n = 16) of osk mutant testes, only 22% (n = 63) of testes from esg^shof;osk mutant males showed expansion of FasIII, despite there being clearly more TJ⁺ somatic cyst cells (Fig. 4m, n). These data support our previous results and indicate that esg is required for the ability of somatic cells to assume and/or maintain hub cell fate.

Our findings demonstrating that SSCs can adopt a hub cell fate highlight the dynamic nature of the stem cell-niche relationship and provide a mechanism to regulate the size and function of the stem cell niche in the Drosophila testis. In our model, as somatic cells are displaced from the hub, there is a decline in self-renewal and proliferation potential, which could be reinforced by encapsulation of differentiating germ cells (Supplementary Fig. 1). Interestingly, expansion of the somatic cyst cells as a consequence of germline loss suggests that germ cells exert an anti-proliferative influence that must be overcome in SSCs.

A better understanding of how stem cell niches are established and regulated in mammalian systems could facilitate modulation of the niche to enhance transplantation of stem cells in regenerative medicine22,23. Conversely, if an expanded or modified niche accompanies tumour progression or metastasis, then blocking niche maintenance programmes (niche ablation) could be used as an important anti-cancer therapeutic24,25.

SuppData