Abstract
Stem cells divide both to produce new stem cells and to generate daughter cells that can differentiate1. The underlying mechanisms are not well understood, but conceptually are of two kinds2. Intrinsic mechanisms may control the unequal partitioning of determinants leading to asymmetric cell divisions that yield one stem cell and one differentiated daughter cell. Alternatively, extrinsic mechanisms, involving stromal cell signals, could cause daughter cells that remain in their proper niche to stay stem cells, whereas daughter cells that leave this niche differentiate. Here we use Drosophila spermatogenesis as a model stem cell system3 to show that there are excess stem cells and gonialblasts in testes that are deficient for Raf activity. In addition, the germline stem cell population remains active for a longer fraction of lifespan than in wild type. Finally, raf is required in somatic cells that surround germ cells. We conclude that a cell-extrinsic mechanism regulates germline stem cell behaviour.
The testis proliferation centre in Drosophila melanogaster includes the germline and somatic stem cells that maintain spermatogenesis ( Fig. 1a)3, 4. As a germline stem cell divides, one daughter becomes a gonialblast, while the other remains a stem cell. To amplify the germline population, each gonialblast executes four divisions as 2° spermatogonia, which exit the mitotic cycle and enter a meiotic and differentiation programme as a clone of 16 spermatocytes. The gonialblast and its progeny are encysted by somatic cells derived from cyst progenitor stem cells3. We previously characterized mutants affecting 2° spermatogonia5, 6, and have begun to test signal transduction pathways for possible involvement at earlier decision points in this germline stem cell lineage. Null mutations in raf, encoding a serine/threonine kinase involved in receptor tyrosine kinase pathways7, 8, are lethal in larvae, but such flies carrying a heat shock (HS)-Raf transgene can survive to fertile adults by using daily heatshocks9. By withdrawing heat shock when adults eclose, animals become progressively raf deficient as the protein decays. Testes from such hypomorphic raf-deficient males 5 days after eclosion were indistinguishable from controls. By day 7, however, 43 of 44 raf mutant testes showed great proliferation centre expansion. The increase in cell number is due to excess, early stage germ cells, and continues such that on day 15, raf -deficient testes are filled with these cells at the expense of post-mitotic cells (compare Fig. 1b and c).
Figure 1: raf function restricts early germ cell number.
a, Testis apex, where 5–9 germline stem cells (one shown for
clarity) and approximately twice as many somatic cyst progenitor cells are
anchored around somatic hub cells. The testis proliferation centre, comprising
hub, germline and somatic stem cells, gonialblasts and 2° spermatogonia
(which undergo incomplete cytokineses; white bars), is restricted to the tip
as shown by bright DNA stain (b, arrowhead). Spermatocyte and postmeiotic
spermatids, which fluoresce more weakly owing to chromatin reorganization,
fill the remainder of the testis. c, raf-deficient testis
on day 15. Scale bar, 100 
m.
To identify the earliest defect in germ-cell progression, we examined the fusome, a membrane and cytoskeletal organelle specific to the germline that is spheroid throughout stem cells and gonialblasts, but branches extensively throughout interconnected 2° spermatogonia (Fig. 2a )10, 11, 12. In raf-deficient testes, fusome structure is not normal. Unbranched fusomes are found in many cells ( Fig. 2b), even those located some distance from the hub, where, in wild-type testes, only branched fusomes interconnecting 2° spermatogonia are found. Although branched fusomes do appear in raf-deficient testes, unlike wild-type testes, these fusomes usually appear only after an intervening region containing many excess germ cells that have only spheroid fusomes. These excess germ cells do not result from an increased frequency of germline stem cell divisions, as the M-phase index for the tier of cells adjacent to the hub in raf-deficient testes was almost identical to that in the wild type (about 3 M phases in 14 testes on day 13).
Figure 2: Early stage germ cells accumulate in raf-deficient testes.
Confocal projection of half testis depth; DNA, purple (a, b,
e, f). a, Wild-type (WT) testis fusome (mAb1b1) is spherical
in stem cells (small arrow) and gonialblasts, which are located only near
the hub (arrowhead). Fusome branches in 2° spermatogonia (larger arrows).
b, raf-deficient testis, day 8. Spherical fusomes (small arrows)
are found both next to and far away from the hub (arrowhead; anti-Fasciclin
(Fasc) III). Fusome branching is detected (larger arrow), more so farther
from the hub (not shown). c, WT testis, day 6; Bam-C (red) is expressed
only in 2° spermatogonial cells, four groups of which are shown, all having
branched fusomes (large arrow; white). Cells to the right of this are maturing
spermatocytes, in which Bam-C protein has decayed, and all of which possess
fusomes branching into and out of the focal plane. The hub is just below the
focal plane (arrowhead). d, raf-deficient testis, day 6; two
of nine cells with unbranched fusomes are indicated (small arrows), none of
which accumulates Bam-C. The fusome is branched in the spermatogonial cluster
expressing Bam-C (large arrow). The asterisk shows one of two 2° spermatogonial
cysts with interconnected fusomes that either did not accumulate Bam-C, or
lost it precociously. The hub is just below focal plane (arrowhead). e
, WT testis, day 8; only a thin band of cells (line) separate Bam-C-expressing
2° spermatogonial cells (white) from the hub (arrowhead; white). f
, raf-deficient testis, day 8; many (line) germline stem cells
and gonialblasts separate Bam-C-expressing 2° spermatogonial cells (white)
from the hub (arrowhead; white). Further along the testis, Bam-C-expressing
2° spermatogonia accumulate in cysts of up to 35 cells, as shown by S-
or M-phase synchrony (not shown). Interspersed with these clusters are individual
cells that cycle in isolation and do not express Bam-C. Scale bar, 20 m
in a, b, e, f; 10 m in c,
d.
We next examined cytoplasmic Bag-of-marbles protein (Bam-C), which first accumulates in 2° spermatogonial cells (Fig. 2c) and is required for their progression into spermatocytes6, 13. In the wild type, the total population of germline stem cells and gonialblasts comprises the few non-staining cells between the hub and the first rows of 2° spermatogonia (Fig. 2e). In raf-deficient testes, the first Bam-C-expressing germ cells are located much further from the hub (Fig. 2f). Also, most of the intervening non-staining cells contain unbranched fusomes (Fig. 2d), consistent with these cells representing excess gonialblasts and/or germline stem cells.
To verify this, we examined M5-4 and S1-33, markers expressed in hub cells, germline stem cells and gonialblasts, but not in 2° spermatogonia (Fig. 3a)6. In raf-deficient testes, while hub cells appeared normal (Fig. 2d, 3b), the number of germ cells expressing these markers was greatly increased (Fig. 3b). Furthermore, marker expression persisted for a significantly longer fraction of adult lifespan. For instance, by day 19, all control testes (n = 14) had M5-4 marker expression in fewer than three germ cells (Fig. 3c), rather than the average 5 to 9 germline stem cells plus associated gonialblasts contained in testes from young adults14. In contrast, only 17% of raf-deficient testes (n = 23) on day 19 showed loss of germ cell expression, whereas 48% maintained expression at least equivalent to that of day 1, and a further 35% still showed increased expression ( Fig. 3d). These data suggest that germline stem cells and gonialblasts remain active for a significantly longer period than in the wild type. We tested this further by directly counting the number of cycling germline stem cells, which we judged to be cells that were located adjacent to the hub and contained a spherical fusome and nuclear anillin, a late interphase marker. Whereas wild-type testes averaged 3.1 late interphase germline stem cells per testis (n = 10 testes; standard deviation, s.d. = 1.0) on day 1, on day 19 aged cohorts had only 0.8 (n = 10; s.d. = 1.0; P < 10-4; Fig. 3e). This suggests an age-induced quiescence of the germline stem cell population. In contrast to this, in raf-deficient testes late interphase germline stem cell numbers were maintained over 19 days (2.5 per testis on day 1, n = 10, s.d. 1.0; 2.8 per testis on day 19, n = 10, s.d. = 1.7; Fig. 3f ) suggesting that germline stem cells as a population remain active for a longer period than in wild-type.
Figure 3: Prolonged activity for the stem cell population in raf-deficient testes.
a, WT testis, X-gal activity; M5-4 expressed in the hub (arrowhead),
germline stem cells (large arrow), and gonialblasts (smaller arrow). b
, raf-deficient testis; M5-4 is expressed in many cells near hub
(arrowhead). c, WT testis, day 19; expression solely in hub. d,
raf-deficient testis, day 19. e, WT testis, day 19; branched fusomes
(green) are visible further along the testis (arrow), but there are no cells
with spherical fusomes adjacent to hub (arrowhead). f, raf-deficient
testis, day 19; an example of late-interphase (nuclear Anillin, red) germline
stem cell (arrow), as indicated by the presence of a spherical fusome (green)
in the cell adjacent to the hub (arrowhead). Scale bar, 40 m.
To determine whether raf function is required in the germ line, which exhibits the phenotypes described above, or in the surrounding somatic cells, we generated raf null mutant clones. Persistent germline clones indicate the existence of a raf null germline stem cell3, 6. In all cases (34% of testes; n = 50), progression through spermatogenesis is normal as judged by groups of 16 mutant germ cells that are morphologically indistinguishable from surrounding wild-type spermatocytes ( Fig. 4a). Thus, raf function is dispensable in germ cells. In contrast, we recovered no cyst cell clones, suggesting that raf is necessary for viability or proliferation of cyst progenitor cells. As our previous analysis was under hypomorphic conditions for raf, we repeated the mosaic analysis, introducing a HS–Raf transgene to provide a basal level of raf function so that raf-deficient cyst cells might survive. Persistent raf-deficient cyst cell clones (5% of testes; n = 204), indicate the existence of a raf-deficient cyst progenitor cell. Such testes had excess early stage, raf+ germ cells ( Fig. 4b, and inset). Thus, raf is required in the cyst cell lineage. In addition, in raf-deficient testes two cyst cell markers normally expressed in later-stage cyst cells, LacZ600 (Fig. 4c)15 and Eyes absent (not shown), were now expressed prematurely in somatic cells adjacent to the hub, probably the cyst progenitor cells (Fig. 4d). This shows that raf function is required in somatic cells surrounding germline stem cells.
Figure 4: raf is required in cyst progenitor cells.
a, Normal spermatocytes (
-galactosidase-positive, arrow) produced
10 days after induction of a raf germline stem cell clone. Similar
data were obtained for null mutations in egfr and ras1.
b, Deficiency of raf in somatic cyst cells (arrowheads) leads to
excess early stage germ cells (inset; DNA stain). Glutaraldehyde fixation
for LacZ activity meant that we could not assay germ cells for stem or gonial
cell markers. c, WT testis, day 18. LacZ600 (white; anti--galactosidase)
is expressed in the hub (arrowhead; green) and in later cyst cells (arrow).
d, raf-deficient testis, day 18. LacZ600, expressed in cyst
progenitor cells (arrows) adjacent to the hub (arrowhead); this is visible
before day 9. Scale bar, 100 m in a, b (125 m
in inset); 20 m in c, d.
Germline stem cell divisions lead to a distinction between a self-renewing daughter cell and a sister cell committed to differentiation as a gonialblast. Our data suggest that a somatic signal influences this decision by limiting the self-renewing potential. We cannot yet say how this potential is encoded. Our hypothesis is that in raf-deficient testes, where cyst progenitor cell identity is disturbed, the signal is lost, and excess stem cell potential is produced. Upon division, both daughters of the germline stem cell inherit some stem cell character. At steady state, this increases the number of cells that become gonialblasts, and somehow prolongs the active state of the stem cell population. Thus, a somatic cell defect leads to a tumour in the germline stem cell lineage, which suggests that some tumours of progenitor cell populations could be initiated by genetic lesions in support cells, rather than in the tumorous cells themselves. In the testis, the Raf-dependent signal may be delivered by cyst progenitor cells, or their cyst-cell daughters. Our mosaic analysis shows that depletion of Raf from just one of the two cyst progenitor cells surrounding a germline stem cell causes a defect. This may indicate a dose effect, where one heterozygous somatic cell is not sufficient to allow normal signalling. It is likely that the signal transducer Raf is engaged, owing to activation of the Epidermal growth factor receptor pathway in somatic cells16. We note that in raf-deficient testes, differentiation of 2° spermatogonia is blocked as they do not transit to the spermatocyte stage. We believe this is a secondary effect owing to the defective cyst lineage, as we previously showed that a cyst cell signal governs this later transition from 2° spermatogonia to spermatocytes5.
Somatic signals have been postulated to affect germline stem cell behaviour in the Drosophila ovary17, 18. However, the characterized signals are necessary to maintain germline stem cells, rather than restrict their self-renewing potential, as we find here. Additionally, raf-deficient ovaries exhibit no increases in early stage germ cells19. Thus, despite the superficial similarities of early germ cell development in ovary and testis, oogenesis and spermatogenesis are emerging as complementary systems from which different principles of stem cell regulation will emerge.
Methods
The raf 11-29 or raf 400B6; HS-Raf/+ larvae were heat shocked for 1 hour at 38 °C, daily. At eclosion (day 1), heat shocking was halted, and flies were aged with wild-type FM6/Y siblings. Stains were carried out as in refs 3,11 and 15. Figure 3e, f shows wild-type and mutant cohorts that were aged and processed using Anti-Anillin for cell-cycle stage20, Anti-Fasc III and Anti-1B1. Cells adjacent to the hub containing dot fusomes and exhibiting nuclear anillin accumulation were scored as late interphase germline stem cells. Wild-type and raf-deficient testes differed significantly at day 19 (P < 0.003). Figure 4a, b shows clones induced by a 1-hour heat shock at 38 °C in day 1 raf400B6Act> DRaf+>nuc-lacZ / Y; HS-Flp/+ or raf400B6Act5C>DRaf+>nuc- lacZ / Y; HS-Raf /+; HS-Flp/+ males that were fixed and stained on day 10. The DRaf+ flip-out is described in ref. 21. The continued production of persistent marked clones demonstrates the marking of a stem cell6.