Stem cells in adult tissue undergo lifelong continuous self-renewal and generate differentiated cells for maintaining tissue homeostasis by replenishing the lost cells caused by natural turnover, aging, injury, or disease. Adult stem cell self-renewal and proliferation are demonstrated to be controlled by the niche in various tissues and organisms (1, 2). Studies on stem cells from different organisms ranging from Drosophila to mammals have demonstrated that one or multiple signals originated from the niche directly act on stem cells in concert with varieties of different intrinsic factors to control stem cell self-renewal by repressing differentiation pathways (1, 2). Our recent study on germline stem cells (GSCs) in the Drosophila ovary has also demonstrated that stem cell progeny differentiation is also controlled extrinsically by the niche formed by adjacent stromal cells, which is named as the differentiation niche (3). Resident macrophage cells on the surface seminiferous tubule in the adult mouse testis also contribute to the niche for regulating germ cell differentiation, suggesting that the niche dedicated to differentiation might also exist in mammalian tissues (4). Multiple signaling pathways are usually used by niches in various stem cell systems to control either self-renewal or differentiation, but how they cooperate with one another in niches to control stem cell behavior remains poorly understood. In this study, we show that Hh and Wnt signaling pathways use novel cooperative mechanisms to provide the favorable environment for GSC progeny differentiation in the Drosophila ovary by maintaining each other’s signaling activities.
The Drosophila ovary provides an effective system for studying stem cell self-renewal and differentiation due to well-defined GSCs and niches. Two or three GSCs physically interact with the niche consisting of primarily cap cells, whereas early GSC progeny physically interact with their own niche composed of inner germarial sheath (IGS) cells (also known as escort cells) (fig. S1A) (5, 6). The GSCs located at the tip of the germarium continuously generate cystoblasts (CBs), which can further divide four times synchronously with incomplete cytokinesis to form mitotic cysts (2-cell, 4-cell, and 8-cell) and 16-cell cysts. Cap cells and anterior row of IGS cells directly contact GSCs and form the niche for promoting self-renewal (7–10). The niche uses bone morphogenetic protein (BMP) signaling and E-cadherin–mediated cell adhesion to control GSC self-renewal and proliferation (5). In 2011, IGS cells were first proposed to form a niche for promoting GSC progeny differentiation (3). Thus, two distinct niche compartments control stem cell self-renewal and differentiation separately.
IGS cells function as the differentiation niche for GSC progeny through physical interactions and signaling. IGS cells extend long cellular processes to encase early differentiating GSC progeny, including CBs, mitotic cysts, and 16-cell cysts (3, 11). Further genetic studies have demonstrated that IGS cellular process–mediated physical interactions are essential for promoting GSC progeny differentiation (3, 9, 12, 13). In addition, different signaling pathways and gene networks have been identified for their requirement in IGS cells for promoting GSC progeny differentiation by preventing BMP signaling through distinct mechanisms (5). Notably, epidermal growth factor receptor (EGFR), Wnt, Hh, and Jak-Stat signaling pathways are required in IGS cells to promote GSC progeny differentiation by preventing BMP signaling via regulation of dally, dpp, or both (12, 14–21). dally encodes a proteoglycan protein promoting the diffusion of Dpp/BMP protein in Drosophila (22). In addition to repressing BMP signaling, IGS cells might also send direct signals to GSC progeny to promote their differentiation, but these signals remain to be defined. In contrast, EGFR signaling has so far been reported to be required in adult somatic cysts to promote GSC progeny differentiation in the Drosophila testis (23).
Since both Hh and Wnt signaling pathways are required in IGS cells for promoting GSC progeny differentiation by repressing BMP signaling, their functional relationship in the niche remains unclear. This study has revealed that they are dependent on each other for their activities in IGS cells through repressing dally-like protein (dlp). dlp encodes a Dally-related glypican (GPC) protein, which is known to promote BMP, Hh, and Wnt signaling in Drosophila (24). However, Dlp-related GPCs can both promote and inhibit Shh and Wnt signaling in mammals (25–27). Here, we show that Dlp up-regulation can sufficiently inhibit both Hh and Wnt signaling and elevate BMP signaling. dlp knockdown in IGS cells can significantly rescue the GSC progeny differentiation defects caused by defective Hh or Wnt signaling and can also uncouple the interdependence of Hh and Wnt signaling. Hh and Wnt signaling directly repress dlp expression through recruiting Croc and H3K9 trimethylase Eggless into the regulatory region. Therefore, this study has revealed a novel cooperative mechanism of Hh and Wnt signaling and a novel Hh/Wnt-mediated mechanism for dlp repression in the niche for preventing BMP signaling and promoting GSC progeny differentiation.
Hh and Wnt signaling activities are mutually dependent in the niche
Hh and Wnt signaling are both required in IGS cells for proper GSC progeny differentiation. To investigate the relationship between Hh and Wnt signaling in IGS cells, we examined the expression of ptc–green fluorescent protein (GFP) and fz3–red fluorescent protein (RFP), which are Hh and Wnt signaling activity reporters, respectively (28–30), in adult smo and dsh knockdown (smoKD and dshKD) IGS cells. IGS-expressed gal4 line, c587, is combined with a temperature-sensitive mutant gal80 (gal80ts) to achieve RNA interference (RNAi)–mediated gene knockdown in adult IGS cells after shifting adult flies from room temperature to 29°C (fig. S1B) (12, 15). Two independent RNAi lines for smo and dsh, which had been characterized previously (12, 15), were used to inactivate Hh or Wnt signaling in this study, respectively. The enhancer trap line PZ1444 expressing nuclear LacZ is used to label IGS cells and cap cells, which can further be distinguished on the basis of their distinct nucleus size and location (15). Consistent with our previous finding that Hh and Wnt signaling are required for IGS maintenance, most of IGS cells are lost 5 and 7 days after dsh or smo knockdown (fig. S1, C and D). However, IGS numbers remain close to normal 2 days after their knockdown, which is the time when we examined fz3-RFP and ptc-GFP expression throughout this study (fig. S1, C and D, and Fig. 1, A and B).
On the basis of fz3-RFP and ptc-GFP expression, knocking down smo or dsh for 2 days can effectively inactivate Hh and Wnt signaling in adult IGS cells, respectively (Fig. 1, A to D). Adult dshKD IGS cells significantly decrease ptc-GFP expression, while smoKD IGS cells significantly reduce fz3-RFP expression, indicating that Hh and Wnt signaling regulate each other (Fig. 1, A to D). Our previous RNA sequencing (RNA-seq) results on purified dshKD IGS cells did not show significant changes in fz3 and ptc mRNAs compared with control IGS cells (table S1) (15). One of the concerns is that enzymatic dissociation of IGS cells destroys Hh and Wnt proteins in the extracellular space, which result in the loss of Hh and Wnt signaling in control IGS cells. As the whole IGS purification process lasts about 4 hours, which might be long enough for fz3 and ptc mRNA decay, fluorescence-activated cell sorting (FACS)–purified control and dshKD IGS cells behave similarly on fz3 and ptc mRNA levels. In the future, it should be extremely cautious to use FACS-purified cells for examining gene expression changes caused by secreted factors. Then, we performed fluorescent mRNA in situ hybridization (FISH) using quantitative hybridization chain reaction technology to further examine fz3 and ptc mRNA expression changes in dshKD or smoKD IGS cells (31). dshKD or smoKD significantly decreases the expression of both fz3 and ptc mRNAs in IGS cells (Fig. 1, E to H). To exclude the possibility that germ cell defects cause the loss of fz3-RFP and ptc-GFP expression in IGS cells, we examine fz3-RFP and ptc-GFP expression in IGS-specific tkv knockdown germaria, which exhibit the severe germ cell differentiation defect as reported previously (fig. S1, E and F) (18, 32). fz3-RFP and ptc-GFP expression remain normal in tkvKD IGS cells despite the presence of the severe differentiation defect (fig. S1, E and F). Together, these results indicate that Wnt and Hh signaling are mutually dependent in IGS cells.
Hh and Wnt signaling in the niche repress the expression of dlp, whose up-regulation sufficiently represses Hh and Wnt signaling
To investigate the mechanism underlying the Hh and Wnt signaling interdependence, we examined the previous RNA-seq results on purified control and dshKD IGS cells (15). dlp mRNA is up-regulated by about fourfold, but dally expression remains unchanged, in dshKD IGS cells compared with control ones (fig. S2A and table S1). dally and dlp encode highly related glypican proteins known to modulate Dpp/BMP, Hh, and Wnt signaling in Drosophila (24). Although its mRNA and protein levels are very low in control IGS cells, dlp mRNA and protein levels are drastically up-regulated in dshKD and smoKD IGS cells based on FISH and immunostaining results, respectively (fig. S2, B and C, and Fig. 2, A and B). These results reveal that Hh and Wnt signaling are required in IGS cells to repress Dlp expression.
Then, we determined whether dlp up-regulation in IGS cells can affect GSC progeny differentiation, Hh and Wnt signaling. In contrast with the control germaria containing about one CB, the 5- and 10-day Dlp-overexpressing (dlpOE) germaria accumulate approximately 10 and 20 spectrosome-containing single germ cells (SGCs), respectively, indicating that Dlp overexpression blocks CB differentiation (Fig. 2, C and D). dlp overexpression also diminishes the expression of fz3-RFP and ptc-GFP in IGS cells, indicating that Dlp up-regulation can sufficiently repress both Hh and Wnt signaling activities in the niche (Fig. 2, E to H). Dlp is known to be directly associated with Dpp, Hh, and Wg proteins to modulate their signaling activity (22, 26, 33, 34), but it remains undermined whether Dlp can also directly bind to Wnt2 and Wnt4, two highly expressed Wnt proteins in IGS cells (15). We used coimmunoprecipitation (co-IP) experiments in S2 cells to show that Dlp can also be associated with Wnt2 and Wnt4 (fig. S2, D to G). Together, our findings indicate that up-regulated Dlp expression sufficiently and directly represses Hh and Wnt signaling in IGS cells.
Niche-specific Dlp overexpression blocks GSC progeny differentiation primarily by elevating BMP signaling
BMP signaling elevation is known to be linked to the CB differentiation defects caused by defective Hh and Wnt signaling in IGS cells (12, 15–18). In control germaria, BMP signaling leads to production of phosphorylated Mad (pMad), activation of Dad-lacZ reporter expression in GSCs, and represses bam-GFP (Fig. 2, I to K). In GSC progeny, including CBs, pMad and Dad-lacZ are turned off and bam-GFP is activated due to the absence of BMP signaling (Fig. 2, I to K). However, in the niche-specific Dlp-overexpressing germaria, most of the accumulated SGCs are positive for pMad and Dad-lacZ but negative for bam-GFP, indicating that Dlp overexpression in IGS cells elevates BMP signaling in GSC progeny, thus blocking their differentiation (Fig. 2, I to K). dpphr56 and dpphr27 heterozygous mutations were used previously to effectively decrease BMP signaling in the Drosophila ovary because dpp encodes a major BMP ligand in the ovary (12, 15, 35). Compared with dlpOE, the removal of one copy of dpp significantly reduces pMad level in germarium (Fig. 2L). dpphr56 and dpphr27 heterozygous mutations can significantly rescue the CB differentiation defects caused by dlp overexpression, but do not affect GSC numbers significantly (Fig. 2, M and N). Therefore, our experimental results demonstrate that Dlp up-regulation in IGS cells increases BMP signaling, thereby disrupting GSC progeny differentiation.
dlp repression is required for the interdependence of Hh and Wnt signaling in the niche
Since Dlp is known to regulate Hh and Wnt signaling in Drosophila (36, 37), we then determined if Dlp is required for modulating Hh and Wnt signaling in IGS cells. Although two dlp short hairpin RNA (shRNA) lines can efficiently knock down Dlp protein expression (fig. S3, A to D), dlp knockdown does not affect fz3-RFP and ptc-GFP expression in IGS cells, indicating that endogenous Dlp is dispensable for Hh and Wnt signaling activities in the differentiation niche (fig. S3, E and F). dlp knockdown in IGS cells can significantly rescue fz3-GFP expression in the smoKD IGS cells as well as ptc-GFP expression in the smoKD IGS cells (Fig. 3, A to C). These results indicate that Dlp repression is required for maintaining Hh and Wnt signaling independence in IGS cells.
To determine whether dlp up-regulation is responsible for the germ cell differentiation defects caused by defective Hh and Wnt signaling, we examined the SGC accumulation in dlpKD dshKD and dlpKD smoKD germaria. The dlpKD lucKD germaria contain similar GSC and CB numbers to those lucKD germaria, indicating that Dlp is also dispensable in IGS cells for promoting GSC progeny differentiation (Fig. 3, D and E). As expected, dlp knockdown in IGS cells can significantly decrease the Dlp up-regulation in IGS cells and reduce the SGC accumulation caused by smoKD or dshKD but cannot rescue the germ cell differentiation defects completely, indicating that Hh and Wnt signaling promote GSC progeny differentiation partly by repressing dlp expression (Fig. 3, D and E, and fig. S3, A to D). Reducing the dlp dosage by three independent heterozygous mutations can significantly decrease Dlp protein expression in IGS cells and also rescue the germ cell differentiation defects caused by defective Hh and Wnt signaling in IGS cells (fig. S4, A to E). This partial rescue by dlpKD and heterozygous dlp mutations can be explained by the previous findings that Hh and Wnt signaling have additional important downstream targets in addition to dlp (12, 15, 20). dlpKD can only decrease the IGS cell loss caused by dshKD or smoKD 3 days, but not 5 days, after knockdown, suggesting that Hh and Wnt signaling maintain IGS cells largely independent of Dlp repression (fig. S4, F to G). Therefore, these results show that Hh and Wnt signaling–mediated dlp repression in IGS cells is required for their signaling interdependence and normal GSC progeny differentiation.
Hh/Wnt signaling–mediated dlp repression in the niche is controlled by a regulatory region in the second intron
To further investigate how Wnt and Hh signaling repress dlp expression in IGS cells, we generated a series of transgenes carrying different dlp genomic fragments and followed with the GFP complementary DNA (cDNA) by using the pGreenRabbit (pGR) vector (Fig. 4A)(38). Through two rounds of genomic fragment screens, a 900–base pair (bp) genomic region (dlp2.1.5) in the second intron is identified to sufficiently recapitulate Dlp expression patterns in the germarium (Fig. 4B and fig. S5A). Both dlp2.1.5-GFP and endogenous Dlp protein show low expression in IGS cells and high expression in follicle cells (Fig. 4B). Consistent with the idea that the dlp2.1.5 genomic region carries most, if not all, of regulatory elements for Hh/Wnt signaling–mediated repression, dlp2.1.5-GFP is up-regulated in both smoKD and dshKD IGS cells (Fig. 4C and fig. S5B). Transcription factors Ci and Pan function downstream of Hh and Wnt signaling, respectively, to regulate target gene expression (39–41). dlp2.1.5-GFP is also up-regulated in ciKD and panKD IGS cells, indicating that canonical Hh and Wnt signaling likely repress dlp in IGS cells via the 900-bp region (Fig. 4C and fig. S5B). dlp overexpression can also up-regulate dlp2.1.5-GFP expression, suggesting that there is a feedforward loop via Hh/Wnt signaling for the dlp control. These results indicate that Hh and Wnt signaling repress dlp transcription via a small regulatory region.
To further define individual elements in the dlp2.1.5 region for Hh/Wnt signaling–mediated repression in IGS cells, we generated GFP reporter transgenic flies carrying nested deletions from both the ends of the 900-bp dlp2.1.5 region with each deleting 100 bp, dlp2.1.5 Δ1-GFP to dlp2.1.5 Δ10-GFP (fig. S6A). Our nested deletion analyses in wild-type, dshKD, and smoKD IGS cells have yielded three pieces of important information. First, only an 800-bp continuous genomic region is sufficient for recapitulating dlp expression in the germarium since dlp2.1.5 Δ6-GFP has the same expression patterns and levels as dlp2.1.5-GFP (fig. S6H). Second, multiple repressive elements likely scatter along the 800-bp genomic region to work synergistically for Hh/Wnt signaling–mediated dlp repression in IGS cells since no single 100-bp deletion in the 800-bp region alone sufficiently causes the up-regulation of the GFP reporter in IGS cells (fig. S6, B to G and I to L). Third, multiple activators in the 800-bp region are required for Dlp gene expression in follicle cells and also for defective Hh/Wnt signaling–caused Dlp up-regulation in IGS cells. On the basis of GFP expression in follicle cells, the deleted regions in dlp2.1.5 Δ1, dlp2.1.5 Δ5, dlp2.1.5 Δ7, and dlp2.1.5 Δ8 are important for dlp expression in follicle cells and equally important for dshKD/smoKD-mediated dlp up-regulation. Among them, the deleted regions by dlp2.1.5 Δ1 and dlp2.1.5 Δ8 have stronger effects on dlp gene activation in follicle cells and also have stronger suppression effects on defective Hh/Wnt signaling–mediated dlp up-regulation than those deleted ones in dlp2.1.5 Δ5 and dlp2.1.5 Δ7, suggesting that scattered repressive elements in the 800-bp region suppress dlp expression in IGS cells likely by antagonizing the activators. To determine whether the 800-bp-long regulatory region is required for endogenous Dlp protein expression in follicle cells, we used CRISPR-Cas9 to generate a dlpΔ215 mutant deleting the region, which homozygotes are lethal likely due to its requirement for Dlp expression during early development (fig. S6N). In the dlpΔ215 heterozygous mutant germarium, Dlp protein expression is deceased in follicle cells as predicted (fig. S6, O and P). Together, these results suggest that Hh/Wnt signaling–mediated dlp repression in the niche is accomplished through multiple cooperative repressive elements in the dlp regulatory region.
Both Ci and Pan directly bind to multiple sites in the regulatory region to repress dlp expression in the niche
Then, we performed the electrophoretic mobility shift assay (EMSA) to determine whether Ci and Pan directly bind to multiple sites in the 800-bp region using overlapped biotin-labeled 24-bp DNA fragments and purified glutathione S-transferase (GST) fusion proteins with Ci and Pan DNA binding domains (Fig. 5Aand fig. S7A). Our EMSA assay has identified four strong Ci binding regions and three strong Pan binding regions in the 800-bp genomic region in addition to some weak binding sites (Fig. 5A). Two of the four Ci binding sites, CGTGGCTGGC and GACAAGGGACT, are consistent with bioinformatic prediction, whereas only one of the three Pan binding sites, GGATACCAAAAATAGG, is predicted, suggesting that Ci and Pan are capable of binding to the previously uncharacterized new sites. Chromatin immunoprecipitation (ChIP) results have further confirmed that Ci and Pan also associate with dlp2.1.5 region in vivo and show stronger enrichment near in vitro–identified binding regions (Fig. 5, B and C, and fig. S7B).
These Ci and Pan binding regions overlap with the activating regions, suggesting that Ci and Pan binding to the regulatory region might preclude the binding of currently uncharacterized transcription activators (Fig. 5C). Diminishing the binding activities of Ci, Pan, or both in dlp2.1.5 by mutating the experimentally defined sites results in up-regulated GFP reporter expression in IGS cells, showing that Wnt and Hh signaling directly repress dlp expression in IGS cells (Fig. 5D and fig. S7, C and D). It is worth noting that dlp2.1.5-GFP reporter up-regulation caused by the mutated Ci or Pan binding sites is relative moderate compared with that caused by defective Hh and Wnt signaling, suggesting that Ci and Pan binding sites might overlap with the activators’ binding sites in the identified dlp regulatory region. Together, Hh and Wnt pathway downstream transcription factors Ci and Pan can directly bind to multiple sites in the dlp regulatory region to repress dlp expression in the niche.
Ci and Pan cooperatively recruit transcription factor Croc and H3K9 trimethylase Eggless to repress dlp transcription in the niche
To further investigate how Hh and Wnt signaling repress dlp transcription via the 800-bp genomic region in IGS cells, we then investigated how it works with Wnt signaling in IGS cells to directly repress dlp expression by carrying out the shRNA knockdown of transcription factors expressed in IGS cells. In the screen, crocodile (croc) and eggless (egg) were identified for their requirement in repressing dlp2.1.5-GFP expression in IGS cells. Knockdown of croc or egg results in the up-regulated expression of dlp2.1.5-GFP compared with the control (Fig. 6A). Consistent with this, crocKD or eggKD IGS cells also show increased Dlp protein expression (fig. S8, A and B). egg encodes an H3K9 trimethylase, which has been shown previously to be required in IGS cells for promoting GSC progeny differentiation (9, 42). croc encodes a fork head domain–containing transcriptional factor (43). Consistent with the idea that Egg and Croc are involved in Dlp repression in IGS cells, knocking down egg or croc in IGS cells also leads to a significant down-regulation of Wnt and Hh signaling (fig. S8, C to F). These results suggest that Egg and Croc might be involved in Hh/Wnt signaling–mediated dlp repression in IGS cells.
Then, we determined whether Croc is also required in IGS cells to promote GSC progeny differentiation by directly repressing dlp expression. Knocking down croc in IGS cells by two independent shRNAs results in the accumulation of SGCs but does not affect GSC maintenance, indicating that Croc is required in the niche to promote GSC progeny differentiation (Fig. 6, B and C). In addition, our EMSA results indicate that Croc protein can also bind to two sites in the 800-bp dlp regulatory region (Fig. 6D and fig. S8, G and H). ChIP–quantitative polymerase chain reaction (qPCR) results have further shown that IGS–specifically expressed Croc-HA binds strongly in vivo to the in vitro–identified binding sites in the dlp2.1.5 region (fig. S8, I and J). Notably, the Croc in vivo binding ability to the dlp regulatory region is significantly decreased in the smoKD and dshKD IGS cells compared with the control (Fig. 6E). Together, these results reveal that Croc binds to the dlp regulatory region to repress dlp expression in IGS cells, thereby promoting GSC progeny differentiation.
To further investigate how they work together to repress dlp in IGS cells, we tested whether Croc, Ci, and Pan can associate with each other in S2 cells. Myc-tagged stable full-length CiPKA-Myc, which is PKA phosphorylation resistant for preventing its cleavage (44), can pull down Flag-tagged Croc (Flag-Croc) (Fig. 6F). Similarly, hemagglutinin (HA)–tagged Pan (Pan-HA) can also bring down Croc-Flag in the presence of Wnt signaling activated ArmS10, which forms a protein complex with Pan for its nuclear import (Fig. 6G)(45). HA-tagged Croc (Croc-HA) can also precipitate Croc-Flag, indicating that Croc proteins can dimerize or oligomerize (Fig. 6H). Flag-tagged Egg (Egg-Flag) could also be coimmunoprecipitated by Croc-HA (Fig. 6I). Croc can also significantly stabilize Egg-Flag in S2 cells in a dosage-dependent manner (Fig. 6, J and K). In summary, these results indicate that signaling-activated nuclear-localized Pan and Ci recruit Croc and, subsequently, Egg for promoting H3K9 trimethylation and, thus, blocking the access of the transcriptional activators to the dlp2.1.5 region and preventing dlp transcription in IGS cells.
Although Shh and Wnt signaling work synergistically to promote neural stem cell proliferation/differentiation in the mouse developing brain as well as cell proliferation in human medulloblastoma, the underlying mechanisms remain missing (46, 47). In addition, both Hh and Wnt signaling have also been shown to be required in the niche to promote GSC progeny differentiation in the Drosophila ovary by repressing BMP signaling, but the cooperative mechanisms are also not understood as well (15, 17–19). In this study, we show that Hh and Wnt signaling sustain each other in the niche by directly repressing Dlp expression through Ci/Pan-recruited transcription factor Croc and H3K9 trimethylase Egg, and such repression is critical for preventing BMP signaling and, thus, promoting GSC progeny differentiations (Fig. 6L).
Dlp repression–mediated interdependence in the niche is a novel mechanism for Hh and Wnt signaling cooperation
The cooperative and antagonistic relationships between Hh and Wnt signaling have been well established in normal developmental contexts and various human tumors. The antagonistic relationship between Hh and Wnt signaling is often accomplished through intrinsic signal transducers, target genes, and secreted inhibitors (48, 49). In Drosophila, Wg and Hh often regulate the same developmental processes synergistically through regulating each other’s expression (50, 51). However, the molecular mechanisms underlying Hh and Wnt signaling synergistic relationships remain largely unknown. Our findings have revealed a new Hh and Wnt signaling interdependent relationship maintained by a novel Dlp-mediated mechanism.
This study has provided the convincing experimental evidence supporting the Dlp repression–mediated Hh and Wnt signaling interdependence. First, Wnt and Hh signaling are required for each other to sustain their signaling activities in the niche. Second, Hh and Wnt signaling are required in the niche to directly repress Dlp expression since dlp mRNA and protein are significantly up-regulated in the Hh/Wnt signaling–defective niche. Third, niche-specific Dlp overexpression eliminates Hh/Wnt signaling in the niche. Fourth, Dlp overexpression can also further induce its own transcription, likely through inactivating Hh and Wnt signaling, suggesting that there is a feedforward loop for dlp regulation in the niche. Last, Dlp repression in the niche is critical for the interdependence of Hh and Wnt signaling. Together, our findings demonstrate that Hh/Wnt signaling–mediated dlp repression is essential for maintaining the Hh/Wnt signaling interdependence in the niche (Fig. 6L). Although this study has revealed a novel mode for Hh/Wnt signaling cooperation as well as a novel mechanism mediating such cooperation, many important questions remain to be answered, such as how Dlp up-regulation inhibits Hh and Wnt signaling mechanistically in the niche and whether such regulatory mechanism operates in other developmental contexts and some diseased conditions.
Hh/Wnt signaling–mediated Dlp repression represents a new mechanism for the differentiation niche to prevent BMP signaling
BMP signaling activated by GSC niche-secreted Dpp is necessary and sufficient for maintaining GSC self-renewal by repressing differentiation (5). IGS cells function as a niche for promoting GSC progeny differentiation partly by preventing BMP signaling (3). Recent several studies have shown that Hh and Wnt signaling are required in IGS cells to promote GSC progeny differentiation partly by preventing BMP signaling via multiple mechanisms. Hh signaling functions in IGS cells to repress dpp expression and antagonize Hippo signaling, thereby preventing BMP signaling (12, 20). Wnt signaling is required in IGS cells to prevent BMP signaling activities in GSC by maintaining Tkv expression, IGS cellular processes, and the redox state, as well as by repressing dpp expression (15–18). On the basis of our data, we propose a model that Hh and Wnt signaling function in IGS cells to prevent BMP signaling in GSC progeny by repressing dlp expression (Fig. 6L).
This study has provided several pieces of experimental evidence demonstrating that Hh/Wnt signaling–mediated Dlp repression in IGS cells is essential for preventing BMP signaling and promoting GSC progeny differentiation. Dlp up-regulation in IGS cells causes BMP signaling elevation in GSC progeny as well as severe differentiation defects. In addition, decreasing BMP signaling by dpp mutations can significantly and drastically rescue the differentiation defects caused by Dlp overexpression in the niche. Furthermore, dlp knockdown in the niche can significantly rescue the GSC differentiation defects caused by defective Hh or Wnt signaling. Consistent with our findings, Dlp has been suggested to promote BMP signaling by increasing BMP concentration at the cell surface or functioning as a BMP coreceptor in Drosophila (24). It will be of great interest to investigate how Dlp mechanistically promotes BMP signaling in the differentiation niche.
Hh and Wnt signaling use a novel mechanism for repressing dlp expression in the niche
Wnt signaling has been shown to directly repress the transcription of dpp in the leg imaginal disc by recruiting the Pan/Arm/Brinker complex to canonical T cell factor (TCF) binding sites and the transcription of Ugt36Bc in the hemocyte by recruiting the TCF/Pan complex to uncanonical TCF binding sites (52, 53). Hh signaling has only been reported to directly repress tkv expression in Drosophila wing imaginal disc via the full-length Ci, but the underlying mechanism remains unclear (54). This study shows that Hh and Wnt signaling can directly repress dlp expression by recruiting the Croc-Egg protein complex to TCF and Ci binding sites in the dlp regulatory region (Fig. 6L).
In this study, we have revealed that Hh and Wnt signaling downstream transcription factors Ci and Pan bind to multiple sites of a dlp regulatory region to antagonize activators, thereby repressing dlp expression. The 800-bp-long regulatory region in the second intron of dlp (dlp2.1.5) sufficiently recapitulates the expression pattern of Dlp protein in the Drosophila ovary based on the GFP reporters containing different dlp genomic fragments. This region carries all the necessary elements capable of responding to Hh/Wnt signaling properly. Further analysis on 100-bp-long nested deletions has shown that four deletions decease dlp2.1.5-GFP up-regulation in Hh/Wnt signaling–defective IGS cells, but no single deletion up-regulates dlp2.1.5-GFP expression, suggesting that multiple repressive elements in the regulatory region are required for repressing dlp expression in the niche, likely by antagonizing the function of the activating elements. Consistently, both Pan and Ci can bind to multiple sites of the identified regulatory region in vitro and in vivo. These Pan and Ci binding sites are also overlapped with the regions containing the activating elements. Last, mutating either Pan binding sites or Ci binding sites in the dlp regulatory region causes the moderate up-regulation of dlp2.1.5-GFP in the niche compared with its strong up-regulation in the Hh/Wnt signaling–defective niche. Curiously, these mutations also decrease the expression of dlp2.1.5-GFP in follicle cell progenitors. Together, these findings lead us to propose a model that on Hh and Wnt signaling activation, Ci and Pan bind the regulatory region of dlp and repress its expression in the niche partly by preventing the recruitment of unknown transcriptional activators.
This study has further suggested that Ci and Pan sequentially recruit Croc and Egg/H3K9 to the dlp regulatory region to maintain transcriptional repressive mark H3K9me3 and, thus, prevent dlp transcription. Both Croc and Egg are required in the niche for repressing dlp expression and for promoting GSC progeny differentiation. egg is an H3K9 trimethylase required in the niche for promoting GSC progeny differentiation (9), whereas Croc is a known fork head domain–containing transcriptional factor (43). Croc can also directly bind to two independent sites in the same dlp regulatory region in an Hh/Wnt signaling–dependent manner in vivo, and one of them is also in close proximity to Pan and Ci binding sites. In S2 cells, both Pan and Ci are associated with Croc, which is also associated with and stabilizes Egg. On the basis of these results, we propose that on Hh and Wnt signaling activation, Ci and Pan recruit the Croc-Egg protein complex to the dlp regulatory region to directly repress dlp expression, likely through maintaining H3K9me3 (Fig. 6L).
Among six Dlp-related mammalian GPC proteins, GPC4 and GPC6 can functionally replace Dlp to promote Hh signaling in Drosophila, whereas GPC2, GPC3, and GPC5 are inhibitory on Hh signaling when overexpressed (26). In mammals, Dlp homologs GPC3 and GPC5 can inhibit and activate Hh signaling, respectively (25, 55), whereas GPC3 and GPC4 can promote and repress canonical Wnt signaling (27, 56), indicating that the ability of Dlp to repress and activate Hh and Wnt signaling is conserved from Drosophila to mammals. These findings raise the interesting possibility that the Dlp-mediated feedback control of Hh and Wnt signaling interdependence might also help elucidate their cooperative mechanisms in mammalian development, stem cell regulation, and cancer.
MATERIALS AND METHODS
Drosophila melanogaster stocks
The following Drosophila stocks used in this study are described in FlyBase, unless specified: c587, tubulin-gal80ts, smoRNAi (BL27037 and BL62987) (12), dshRNAi (BL31306 and BL31307), ciRNAi (BL64928), PanRNAi (BL40848), dlpRNAi (BL34089 and BL34091), crocRNAi (BL27071 and BL34647), eggRNAi (BL32445 and BL34803), tkvRNAi (BL40937 and BL57303), lucRNAi (BL31603), bamGFP, Dad-lacZ, ptc-GFP (29, 30), fz3-RFP (28, 30), UAS-CD8::GFP, UAS-dlp, UAS-Croc-3×HA (FlyORF: F000139), dlpMI04217, dlpMI09937, and dlpMI10064. Drosophila strains were maintained and crossed at room temperature on standard cornmeal/molasses/agar media unless specified. To maximize the RNAi-mediated knockdown effect, newly eclosed flies at room temperature were cultured at 29°C for the specified days before phenotypic analysis.
The Invitrogen Gateway Technology was used to make the constructs for expressing Flag-tagged Ci, Flag-tagged Pan, Flag-tagged Dlp, HA-tagged Wnt2, HA-tagged Wnt4, Flag-tagged Egg, Flag-tagged Croc, and HA-tagged Croc in S2 cells for co-IP experiments or for making transgenic flies. The coding sequences for ci, pan, dlp, wnt2, wnt4, egg, and croc were amplified from Drosophila ovarian cDNAs using PCR. The armS10 sequence was amplified from the genomic DNA from the UAS-armS10 transgenic strain (BL4782). All the PCR products were cloned into the pENTR-TOPO cloning vector and were completely sequenced. These pENTR vectors were subsequently recombined into Flag-, Myc-, or HA-tagged destination vectors (pAWF, pAWH, pAWM, and pTWF) by using LR Clonase (Invitrogen). UAS-Myc-CiPKA was a gift from J. Jiang (57). Since Dlp protein undergoes internal proteolytic cleavage, the 3× Flag tag was inserted after the 18th amino acid residue, and the termination codon was added to the reverse primer to skip the Flag tag in the pAWF destination vector. The GST fusion proteins with Ci (GST-Ci), Pan (GST-Pan), or Croc (GST-Croc) were constructed by cloning the DNA fragments encoding five Ci ZnF_C2H2 domains, the Pan HMG domain, or the Croc FH domain into the Eco RI and Xho I sites of pGEX4T1, respectively.
Ovaries were dissected, fixed, and stained according to the method described previously (58, 59). The following antibodies were used in this study: mouse monoclonal anti-Dlp antibody [1:10; Developmental Studies Hybridoma Bank (DSHB)], mouse monoclonal anti-Hts antibody (1:50; 1B1, DSHB), rabbit polyclonal anti-β-galactosidase (LacZ) antibody (1:500; MP Biomedical, no. 08559761), mouse monoclonal anti-β-galactosidase (LacZ) antibody (1:50; JIE7, DSHB), rabbit monoclonal anti-Smad3 antibody (pS423/pS425) (1:200; Epitomics, ab52903), rabbit polyclonal anti-RFP (1:1000; Rockland, no. 600-401-379), and chicken polyclonal anti-GFP antibody (1:500; Invitrogen, no. A10262).
S2 cell transfection and co-IP
S2 cells were grown at 25°C in the HyClone SFX-Insect Cell Culture Media (Thermo Fisher Scientific). Transfections were performed using the X-treme GENE HP (6366546001, Roche) transfection reagent according to the manufacturer’s instructions. For co-IP experiments in S2 cells, 12 ml of S2 cells was transfected by indicated plasmids. The transfected S2 cells were then lysed with 800 μl of ice-cold lysis buffer [50 mM tris-HCl (pH 7.5), 150 mM NaCl, 0.5% Triton X-100, 1 mM EDTA, and a mixture of protease inhibitors]. The supernatant of the lysate was incubated with 2 μg of mouse anti-HA, mouse anti-Flag, or mouse anti-Myc. Protein A/G agarose (40 μl; sc-2003; Santa Cruz Biotechnology), which was prewashed in 5% bovine serum albumin at 4°C for 1 hour, was added to the supernatant. The supernatant-antibody-agarose mix was incubated overnight at 4°C. After six washes with the lysis buffer, the bound complexes were eluted with 2× SDS sample buffer and subjected to SDS–polyacrylamide gel electrophoresis and immunoblotting. Mouse anti–α-tubulin (T9026, Sigma-Aldrich; 1:10000), mouse anti-Flag (F1804, Sigma-Aldrich; 1:2000), mouse anti-Myc (M4439, Sigma-Aldrich; 1:2000), or mouse anti-HA (H3663, Sigma-Aldrich; 1:2000) antibodies were used for immunoblotting. To avoid the interference of immunoglobulin G (IgG) heavy chain (~55 kDa), horseradish peroxidase–goat anti-mouse IgG light chain secondary antibodies were used. Inputs were extracted before IP.
Construction of the GFP reporters for dlp
To construct the dlp reporter plasmids, we then used the following primer pairs carrying either Xba I or Kpn I at the 5′ end to amplify the DNA fragments from the Drosophila genomic DNA, which were confirmed by complete sequencing and then cloned into the pGR vector:
dlp promotor: agtctctagactttcgatagtgtggaccttcctt; aagtggtaccgtatgtacagtgtcactaggctat.
dlp2.1: cgactctagagtatgtccgatattatataccaat; cgacggtaccgcatttataactttgttgtagttg.
dlp2.2: cgactctagataataatagtaggca; cgacggtaccttgccacattccaccttagctatt.
dlp2.3: aaggtctagaaatggggctagctta; cgacggtaccaagggagaacggagccaaactcca.
dlp2.4: ttggtctagaacaagttttcgaatga; cgacggtaccatgtggacataatcgagcataa.
dlp2.5: attttctagaatgtatttctggagt; cgacggtaccagactctgatacgcatacaggata.
dlp2.6: agtctctagatggtgccacactcca; aagtggtaccattttgttaatctct.
dlp4: agtctctagagtgagtagtagtctgcgaaatcca; aagtggtacctggaaaataagattaaatcggtg.
dlp6: agtctctagagtgagatctacagcggaataatt; aagtggtacctgcaatgaattaatttgagagtt.
dlp2.1.1: cgactctagagtatgtccgatattatataccaat; cgacggtacctcacgcagttcacgccaacgatgct.
dlp2.1.2: cgactctagaagcatcgttggcgtgaactgcgtga; cgacggtaccaatctgttattaaaatttgtccta.
dlp2.1.3: cgactctagataggacaaattttaataacagatt; cgacggtaccagttgcgatctacaaagccaatct.
dlp2.1.4: cgactctagaagattggctttgtagatcgcaact; cgacggtaccacaatggtcaacaattgcagaagt.
dlp2.1.5: cgactctagaacttctgcaattgttgaccattgt; cgacggtacctggccacgtttgacctgctcgaga.
dlp2.1.6: cgactctagatctcgagcaggtcaaacgtggcca; cgacggtaccgcatttataactttgttgtagttg.
dlp2.1.5Δ1: gctctagactgtctggtgtttgtttatgagg; cgacggtaccgcatttataactttgttgtagttg.
dlp2.1.5Δ2: gctctagaaaaacttatgaagcttttttaatatgattagcaaac; cgacggtaccgcatttataactttgttgtagttg.
dlp2.1.5Δ3: gctctagacatctggtaaaccgaaagctt; cgacggtaccgcatttataactttgttgtagttg.
dlp2.1.5Δ4: gctctagatacaattactcagttcctagggg; cgacggtaccgcatttataactttgttgtagttg.
dlp2.1.5Δ5: gctctagacggtgctgggattccaga; cgacggtaccgcatttataactttgttgtagttg.
dlp2.1.5Δ6: cgactctagaacttctgcaattgttgaccattgt; agacggtacctgccggcaattaagtcgt.
dlp2.1.5Δ7: cgactctagaacttctgcaattgttgaccattgt; agacggtacctccacaggattcattcttagaaaatttgc.
dlp2.1.5Δ8: cgactctagaacttctgcaattgttgaccattgt; agacggtacctcagctaattacgcgaaattgc.
dlp2.1.5Δ9: cgactctagaacttctgcaattgttgaccattgt; agacggtaccatcactggatcagatagcacc.
dlp2.1.5Δ10: cgactctagaacttctgcaattgttgaccattgt; agacggtaccatggcatattagggggcg.
To make dlp2.1.5(3Pan*)-GFP transgenes, the three identified Pan binding sites in gcccacaaagtcaacacttgctga, ctgacgatgctgacagaaatggga, and tcagcaaattttctaagaatgaat were mutated to gcccacaaagtttacacttgctga, ctgacgatgcaaacagaaatggga, and tttgcaaattttctaagaatgaat, respectively. To make dlp2.1.5(4 Ci*)-GFP transgenes, the four identified Ci binding sites in cgtttatcacgggggcttttcgca, actgacaacccactaaactagatc, agcaaactctttcacgcgatctcg, and atgggatctcccagccggcagcca were mutated to cgtttatcacggggtttttttgca, actaacaaaacactaaactagatc, agcaaactctttcacgttatcttg, and atggaatctaacagccggcagcca, respectively. For the dlp2.1.5(3Pan*+4Ci*)-GFP reporter, all of the seven binding sites were mutated. All the constructs were inserted into the attp40 site on the second chromosome using PhiC31 integrase–mediated transgenesis by Rainbow Transgenic Flies Inc.
Generation of the dlp2.1.5 deletion mutant
The dlpΔ215 (deleting first 800 bp in dlp2.1.5 region) mutant was designed and generated by Rainbow Transgenic Flies Inc. using the CRISPR-Cas9 technology. The following guide RNAs (gRNAs) were used: gRNA1 target, 5-gaattgttgaccattgtatgg; gRNA2 target, 5-ggccaacgacttaattgccgg. Mutants were confirmed by PCR and sequencing. Primers used for PCR identification were 5-gcaaccaccgcatgactatta and 5-gatgggaaagagacagcaact.
Purification of bacterially expressed GST, GST-Ci, GST-Pan fusion proteins and in vitro DNA binding assays
The Escherichia coli bacteria strains were transfected with GST, GST-Ci-ZNF, GST-Pan-HMG, or GST-Croc-FH plasmid, and the culture for each bacteria strain was grown to the density of OD650 (optical density at 650 nm) = 0.1 to 0.25. The expression of the fusion proteins was then induced by the addition of 0.2 mM isopropyl-β-d-thiogalactopyranoside for overnight at 16°C. The cells were then harvested and lysed with B-PER with Enzymes Bacterial Protein Extraction Kit (90079, Thermo Fisher Scientific), and the proteins were purified with glutathione agarose (16100, Thermo Fisher Scientific). The in vitro DNA-protein binding assay was performed according to the LightShift Chemiluminescent EMSA Kit (20148, Thermo Fisher Scientific). Glycerol (4.35%), magnesium chloride (5 mM), poly(dI-dC) (50 ng/ml), and NP-40 (0.05%) were included in the binding reaction. For each 20 μl of binding assay, 0.1 nM biotin-labeled probe (synthesized by Integrated DNA Technologies) and 10 μg of purified GST protein or GST fusion protein were used.
ChIP and PCR or qPCR
ChIP was performed essentially as described by the Pierce Agarose ChIP Kit (26156, Thermo Fisher Scientific). For each genotype, 200 pairs of ovaries were dissected and then digested with type II collagenase (50D11833; Worthington). The late-stage egg chambers and mature eggs were filtered and removed. Primers used for regular PCR are as follows:
1–96: acttctgcaattgttgaccattgt; tcctactcgtttatataccccgcc.
91–204: gtaggagttgctgtctggtgtttg; ttttagattttatatacccaaagc.
199–294: ctaaaacttatgaagcttttttaa; gatctagtttagtgggttgtcagt.
289–402: tagatcacctaacatctggtaaac; taatggcatattagggggcgagat.
397–512: ccattacaattactcagttcctag; atcccagcaccgatcactggatca.
507–602: tgggattccagacattttgcccac; cgtcagctaattacgcgaaattgc.
597–710: gcaatttcgcgtaattagctgacg; tcatccgcgatccacaggattcat.
705–800: ggatgattcaagttggattcgagt; tgccggcaattaagtcgttggccc.
For comparing the binding affinities of Croc to the dlp regulatory region between WT, smoKD, and dshKD IGS cells, qPCR was performed using PerfeCTa SYBR Green FastMix (Quantabio, no. 022048) according to the manufacturer’s recommendations and analyzed using the 2−ΔΔCT method. Sequences of primers are tgggattccagacattttgcccac and gtcagctaattacgcgaaattgc. actin5C was used as internal control (primers: atcgggatggtcttgattctg and actccaaacttccaccactc).
Hybrid chain reaction combined with immunostaining
To assess the expression level of fz3, ptc, and dlp mRNA in control, smoKD, and dshKD IGS cells, we performed FISH on ovaries, which are immunostained for PZ1444-LacZ (labeling IGS cells). Hybridization chain reaction (HCR) was used to achieve high-sensitivity FISH for quantification. Probe sets against fz3-mRNA (lot: PDR091), ptc-mRNA (lot: PDR092), or dlp-mRNA (lot: PDR093) were ordered from Molecular Instruments Inc. Immunostaining ovaries using anti-LacZ antibodies was performed according to the previous publication before in situ hybridization (60). Then, standard steps following HCR v3.0 protocol for whole-mount fruit fly embryos were applied. At the end of HCR in situ hybridization, 4′,6-diamidino-2-phenylindole (DAPI) was added at 1 μg/μl for 10 min in 5× SSCT (1X saline-sodium citrate buffer with 0.1% Triton X-100) buffer to the ovaries and then washed 15 min in 5× SSCT for four times. Last, the ovaries were mounted, and images were captured according to the regular immunostaining protocol.
Imaging, quantification, and statistical analysis
GSCs and CBs were quantified under the fluorescence microscope according to the method described previously (58). The germaria were imaged by the Leica SP5 confocal microscope, and the images with all sections were merged unless specified. For confocal images, fluorescence intensities for the highlighted areas of interest were quantified using the Leica software, and the mean values of fluorescence intensities and internal controls were collected. The ratio of mean values of intensities of interest to internal controls was calculated and subjected for statistical analysis using Student’s t test in Microsoft Excel or GraphPad Prism 7. For fz3-RFP and ptc-GFP reporters, the intensity of single IGS cell nuclear was measured, because these reporters express nuclear located RFP or GFP. For other staining, intensity of IGS cell region in each germarium was measured. All bar graphs are represented as means ± standard error and with individual value (***P ≤ 0.001; **P ≤ 0.01; *P ≤ 0.05; n.s., no significance).