Coupling of Hedgehog and Hippo pathways promotes stem cell maintenance by stimulating proliferation

Abstract

It is essential to define the mechanisms by which external signals regulate adult stem cell numbers, stem cell maintenance, and stem cell proliferation to guide regenerative stem cell therapies and to understand better how cancers originate in stem cells. In this paper, we show that Hedgehog (Hh) signaling in Drosophila melanogaster ovarian follicle stem cells (FSCs) induces the activity of Yorkie (Yki), the transcriptional coactivator of the Hippo pathway, by inducing yki transcription. Moreover, both Hh signaling and Yki positively regulate the rate of FSC proliferation, both are essential for FSC maintenance, and both promote increased FSC longevity and FSC duplication when in excess. We also found that responses to activated Yki depend on Cyclin E induction while responses to excess Hh signaling depend on Yki induction, and excess Yki can compensate for defective Hh signaling. These causal connections provide the most rigorous evidence to date that a niche signal can promote stem cell maintenance principally by stimulating stem cell proliferation.

Figure 1.

Yki is required for FSC maintenance. (A) Drosophila germarium: germline stem cells (GSC) produce cystoblast (CB) daughters, which proliferate to form 16-cell cysts (black) while surrounded by somatic escort cells (ECs). Follicle stem cells (FSC) just anterior (left) to region 2b cysts, which span the germarium, produce follicle cells (FC), which express Fas3 (red) at their surface and envelop germline cysts. A typical single FSC clone lineage is indicated by green nuclei. Hh is expressed strongly in anterior terminal filament (TF) and cap cells (CC), whereas the JAK-STAT pathway ligand is expressed strongly in polar FCs at the posterior of the germarium. (B) Percentage of ovarioles that contain wild-type (WT) or yki^B5 FSC or GSC clones 9–18 d after clone induction in 2-d-old adults. (C) The percentage of ovarioles that maintain yki^B5 FSC clones 12 d after induction in larvae was greatly increased by coexpression of both UAS-diap1 and UAS-cycE transgenes. Error bars in B and C show SDs; n = 3, with ≥100 ovarioles scored in each measurement. (D–G′) D and E show clones marked by the loss of GFP, whereas F and G show MARCM clones marked by expression of GFP. In all cases, clones were induced 12 d earlier, ovarioles were stained with Fas3, and close-ups of the FSC region are shown (D′, E′, F′, and G′). Red bars, 10 µm. (D) The wild-type FSC clone includes an FSC (arrowhead) just anterior (left) to Fas3 staining and many FC derivatives (white lines). (E) In most yki^B5 ovarioles, all FSCs (arrowhead) and FCs express GFP, indicating the absence of yki mutant FSCs, but GFP-negative germline cysts (asterisks) and GSCs were frequently present, indicating continued activity of yki mutant GSCs. (F) yki mutant FSCs expressing transgenic DIAP1 were generally not maintained for 12 d, but some ovarioles include a patch of GFP-marked FCs (white line), indicating that the marked FSC (arrowhead) was lost within the last 5 d. (G) Excess DIAP1 and CycE together rescued the maintenance of yki mutant FSCs in many ovarioles to produce clones with a marked FSC (arrowhead) and FC derivatives (white lines).

Figure 2.

Yki is important for net FC amplification principally by preventing apoptosis. (A and B) Stage 10 egg chambers with control (wild type [WT]) or yki mutant clones lacking GFP (green, outlined in yellow) and adjacent wild-type twin spot clones with two copies of the GFP transgene (brighter green, outlined in white). Control clones were of similar size to simultaneously generated twin spots (A), but yki mutant clones were much smaller (B). (C) Mean number of cells in a GFP-negative clone divided by the number of cells in a twin spot, calculated for 20 (wild type) or 25 (yki) clones. Error bars show SDs. (D–G) Clones derived from single FCs formed patches (green nuclei, outlined in yellow) in stage 10 egg chambers that were much smaller than wild type (D) for yki mutant FCs (E); normal clone size was restored by expressing excess DIAP1 (G), but not CycE (F), in the clone. (H) Mean number of cells in a clone for wild-type and yki FCs expressing the indicated transgenes in the clone. Error bars show SDs, and number of clones measured is indicated in parentheses. Red bars, 15 µm.

Figure 3.

FSCs with increased Yki activity displace wild-type FSCs and produce FC hypertrophy. (A and B) Percentage of ovarioles containing marked FSC clones 12 d after induction in larvae or adults. Dark red and blue columns show the percentage of all-marked (A.M.) ovarioles, containing only marked FSCs and FCs. To the left of each group of mutant genotypes is the control (wild type [WT]) for the appropriate chromosome arm (2R, 3R, X, or 2L) that becomes homozygous in clones. Error bars show SDs; n = 3, with ≥100 ovarioles scored in each measurement. Values for adult FSC clones are from one experiment. Significant differences from WT all-marked values for all-marked FSC clones using Fisher’s exact two-tailed test with P < 0.05 are indicated by asterisks. (C–D′) All-marked ovarioles (C and D) and enlarged germarial regions (C′ and D′), stained for Fas3, contained only GFP-negative FSCs (arrowheads) and FCs, which are homozygous for the indicated mutations, and included multilayering of FCs (yellow lines) and accumulation of FCs between egg chambers (white lines). Red bars, 10 µm.

Figure 4.

Regulation of FSC proliferation by the Hpo pathway is critical for FSC maintenance. (A) Percentage of marked FSCs and FCs of indicated genotypes labeled by EdU during a 1-h incubation 6 d after clone induction. Significant differences from control (wild type [WT]) values by Fisher’s exact two-tailed test are indicated for P < 0.05 by the asterisks. For yki and smo FSCs (marked by #), P < 0.05 after adding the data shown in Fig. S5 B. Key comparisons are highlighted by red brackets. (B and C) Examples of EdU labeling of germaria containing positively marked FSC clones (green) and stained for Fas3, showing positions of FSCs, FCs, region 1 and 2a ECs (1 EC and 2a EC), and cells in the same dorsoventral plane as FSCs (2a/2b cells), all of which were scored separately (Table S1). (D) Percentage of ovarioles with marked FSC clones of indicated genotypes 12 d after induction in larvae (dark red indicates all-marked clones). For all genotypes with cycE mutations, both hypomorphic (cycE^WX, left, solid shading) and null (cycE^AR95, right, angled stripes) alleles were tested. Error bars show SDs; n = 3, with ≥100 ovarioles scored in each measurement. Significant differences from wild type for all-marked FSC clones using Fisher’s exact two-tailed test with P < 0.05 are indicated by asterisks. Key comparisons are highlighted by blue brackets. (E and F) Ovarioles with clones induced 12 d earlier and marked by the presence of GFP, shown together with Fas3 to reveal FSC location (arrowheads). The all-marked FSC clone phenotype and FC multilayering (yellow line) produced by loss of ex (E) was not observed when UAS-cycE complements a cycE mutation (F). Red bars, 10 µm.

Figure 5.

Hh signaling increases Yki activity at the transcriptional level in the FSC lineage. (A–J′) FSC clones of the indicated genotypes, marked by loss of GFP in germaria stained for Fas3 and β-galactosidase (red) encoded by a diap1-lacZ reporter. diap1-lacZ expression was lower in wild-type (WT) FSCs (arrowhead; A) than FCs (white lines) and was increased in both FSCs (arrowheads) and FCs (arrows in F–J′) for hpo and ptc genotypes (B, C, G, and H) but not in ptc yki; tub-yki (D and I) or ptc mam (E and J) genotypes. Red bars, 10 µm. (K) Quantification of diap1-lacZ expression levels in FSCs and FCs of the indicated genotypes. Error bars show SDs from three measurements of the same samples, as described in the Materials and methods (n > 18 for FSCs and n > 300 for FCs). Dotted lines indicate values for wild-type FSCs and FCs. (L) Relative mRNA levels of diap1, cycE, ex, yki, and ptc (indicated on top) in ovarioles containing control (wild type, blue bars), hpo (red bars), or ptc (green bars) FSC clones induced 12 d earlier in larvae. rp49 mRNA was used to normalize total mRNA concentration for each genotype. Error bars show SDs; n = 3. Values significantly different from wild type at P < 0.05 by Student’s t test in K and L are indicated by asterisks.

Figure 6.

Regulation of FSCs by Hh is mediated principally by induction of Yki. (A–C) Percentage of ovarioles containing positively marked FSC clones 12 d after larval heat shock induction; dark red indicates all-marked clones. Each set of genotypes has the appropriate control (wild type [WT]) to the left. Horizontal black lines clarify correspondence between bars and genotypes. Blue brackets indicate critical comparisons. (B and C) All tests used a hypomorphic smo allele (smo^7.6.6, right, angled stripes) and some (C) also used a strong, effectively null allele (smo², left, solid shading). Error bars show SDs; n = 3, with ≥100 ovarioles scored in each measurement. Significant differences from wild type for all-marked FSC clones using Fisher’s exact two-tailed test with P < 0.05 are indicated by asterisks. (D–G) Ovarioles with clones induced 12 d earlier and marked by the presence of GFP, shown together with Fas3 and with FSCs indicated (arrowheads). All-marked FSC clones were induced by excess Yki (D) but not by ptc when yki is replaced by a tub-yki transgene (E). mam FSC clones were rescued by activated Yki (F), and smo FSC clones were rescued by excess CycE together with DIAP1 (G). Red bars, 10 µm.