Introduction

The homeostasis of many tissues in our bodies is maintained by adult stem cells, but this balance can be disrupted by tumor cells. What occurs when tumorigenesis intersects with stem cell development? To address this question, a mosaic-analysis model system is essential, where wild-type stem cells develop alongside tumor cells. Drosophila offers an exceptional model for such studies, as it allows for the efficient generation of mosaic clones through various established methods (Germani et al., 2018; Pastor-Pareja and Xu, 2013).

In Drosophila ovaries, germline stem cells (GSCs) play a crucial role in sustaining normal oogenesis and maintaining fertility (Fuller and Spradling, 2007; Lin, 1997). These GSCs reside in a specialized microenvironment known as the stem cell niche (hereafter referred to as niche) (Xie and Spradling, 2000). Typically, a GSC undergoes asymmetric division, generating two distinct daughter cells: one remains in the niche to self-renew as a GSC, while the other, called a cystoblast, exits the niche and initiates differentiation. During the differentiation process, each cystoblast performs exactly four rounds of mitotic division with incomplete cytokinesis to produce16 interconnected cystocytes, forming a germline cyst. In each germline cyst, only one germ cell is destined to become the oocyte, while the remaining 15 differentiate into nurse cells that support the development of the oocyte (Figure 1A) (Fuller and Spradling, 2007; Lin, 1997). The principal niche signals are bone morphogenetic protein (BMP) ligands, including Decapentaplegic (Dpp) and Glass bottom boat (Gbb), which are secreted predominantly by cap cells, the major somatic niche cells (Chen and McKearin, 2003a; Li et al., 2016; Song et al., 2004; Xie and Spradling, 1998, 2000). These ligands activate BMP signaling in GSCs, leading to the transcriptional repression of bag of marbles (bam), a key gene that promotes differentiation. In contrast, BMP signaling is inactive in cystoblasts, allowing Bam to be expressed and drive their differentiation (Chen and McKearin, 2003a; Song et al., 2004). Bam carries out this function in collaboration with its partner, Benign gonial cell neoplasm (Bgcn) (Li et al., 2009; Ohlstein et al., 2000).

Figure 1.

GSCs mutant for bam or bgcn fail to differentiate and instead hyper-proliferate, forming a well-established Drosophila germline tumor model (Lavoie et al., 1999; McKearin and Ohlstein, 1995; McKearin and Spradling, 1990; Niki and Mahowald, 2003). Notably, these germline tumor cells competitively displace wild-type GSCs from the niche (Jin et al., 2008). The resulting displacement creates a microenvironment where wild-type GSCs are surrounded by tumor cells, providing an excellent model system to study stem cell behavior in tumor neighborhoods.

Here, we demonstrate that bam or bgcn mutant germline tumors inhibit the differentiation of neighboring wild-type GSCs by functionally mimicking the stem cell niche. This mechanism may be conserved in mammals, including humans, during tumorigenesis, where malignant cells could similarly disrupt normal stem cell development.

Results

Germline tumors inhibit the differentiation of neighboring wild-type GSCs

Using the nos>FLP/FRT or hs-FLP/FRT methods we previously established (Zhang et al., 2023; Zhao et al., 2018), we generated bam or bgcn mutant germline clones in Drosophila ovaries (Figure 1B). Surprisingly, we found that many wild-type germ cells located outside the niche retained a GSC-like single-germ-cell (SGC) morphology, even when encapsulated within egg chambers (Figure 1C, D, Figure 1—figure supplement 1). Under normal conditions, GSCs that exit the niche differentiate into interconnected germline cysts, where germ cells are linked rather than remaining as individual, isolated cells (Fuller and Spradling, 2007; Xie and Spradling, 2000). To rule out the possibility that the SGC phenotype is an artifact caused by GFP expression, we repeated the experiments using RFP and arm-lacZ as alternative mosaic-analysis markers. Consistent results were observed (Figure 1E, F), confirming that the phenotype is not attributable to GFP.

To further confirm that these SGCs exhibit GSC-like characteristics, we conducted anti-α-Spectrin immunofluorescent staining, a method that labels a germline-specific organelle known as the spectrosome in GSCs and cystoblasts, and the fusome in cystocytes. GSCs perform complete cell division, whereas cystocytes undergo incomplete cytokinesis, remaining interconnected through fusomes and ring canals. Consequently, spectrosomes appear as dot-like structures, while fusomes exhibit branched morphologies (Figure 1G) (Lin et al., 1994). To accurately capture the three-dimensional (3D) architecture of spectrosomes and fusomes, we acquired z-stack images using confocal microscopy. Strikingly, these SGCs displayed dot-like spectrosomes, closely resembling those observed in wild-type GSCs and bam or bgcn mutant GSC-like tumor cells (Figure 1H, I). We also considered the possibility that SGCs might arise through the dedifferentiation of the cystocytes in germline cysts surrounded by germline tumors. If this were the case, such cystocytes would initially undergo complete cell division, leaving behind midbodies as markers of the late cytokinesis stage. When visualized by anti-α-Spectrin immunofluorescence, midbody appears as a central sphere that is slightly connected to two larger flanking structures, resembling a variant of nunchucks (Mathieu et al., 2022). Notably, in our analyses of over 50 germline cysts surrounded by bam mutant germline tumors, none contained midbodies, suggesting that dedifferentiation is unlikely to be the primary mechanism responsible for the SGC phenotype. Together, these findings indicate that bam or bgcn mutant germline tumors inhibit the differentiation of neighboring wild-type GSCs.

The inhibition of differentiation in SGCs relies on the lack of Bam expression

Given that Bam is the key factor promoting GSC differentiation (McKearin and Ohlstein, 1995; Ohlstein and McKearin, 1997), we were very curious about the expression of Bam in SGCs. At first, we assessed Bam protein levels using immunofluorescent staining with an anti-BamC antibody (McKearin and Ohlstein, 1995). Strikingly, none of the SGCs examined (n > 100) were BamC-positive (Figure 2A). Then, we analyzed bam transcription levels using a bamP-GFP reporter (Chen and McKearin, 2003b). Under normal conditions, 100% of GSCs within the niche (n = 153) were GFP-negative, while 98% of cystoblasts (n = 106) were GFP-positive (Figure 2B, C), confirming that bam transcription is associated with the initiation of GSC differentiation (McKearin and Ohlstein, 1995). Notably, 74% of SGCs (n = 132) were GFP-negative, while the remaining 26% were GFP-positive (Figure 2B, C). This suggests that SGCs can be categorized into two distinct groups: those resembling GSCs (GSC-like) and those resembling cystoblasts (cystoblast-like). The cystoblast-like SGCs may have already initiated their differentiation program toward becoming cystocytes. Since bam transcription initiates in cystoblasts (McKearin and Spradling, 1990) but Bam proteins accumulate predominantly in cystocytes (McKearin and Ohlstein, 1995), the Bam protein levels in these cystoblast-like SGCs are likely below the detection threshold at this early stage.

Figure 2.

Next, we asked whether ectopic expression of Bam can drive SGCs to differentiate. To address this, we established two experimental scenarios: one with the hs-bam element and one without as the control. In the hs-bam scenario (with hs-bam), GFP-positive germ cells are wild-type (carrying hs-bam), while GFP-negative cells are bam^−/− (lacking hs-bam). In the control scenario (without hs-bam), GFP-positive cells are wild-type, and GFP-negative cells are bam^−/− (Figure 2D, see genotypes in Source data 1). To quantify the SGC phenotype, we calculated the percentage of SGCs relative to the total number of SGCs and germline cysts, considering the out-of-niche germ cells that are either fully enclosed by germline tumors (e.g., the right SGC in Figure 1I and the marked germline cyst in Figure 2G) or in contact with wild-type germ cells or somatic cells on just one side (e.g., the left SGC in Figure 1I and the outlined germline cyst in Figure 4C). Although the SGC phenotype was detectable after fly eclosion, we quantified it in 14-day-old flies to ensure robust germline tumor growth, improving quantification efficiency. To induce ectopic Bam expression, 12-day-old female flies were subjected to heatshock treatment, which involved heating at 37°C for two hours, twice daily with a 6-hour interval, and over two consecutive days. In the absence of heatshock treatment, the percentage of SGCs in ovaries of both genotypes showed no significant difference at either 12 or 14 days (Figure 2E-H), indicating that the hs-bam element alone, without heatshock, does not affect the phenotype. However, following heatshock treatment, the percentage of SGCs in ovaries with hs-bam was markedly reduced compared to those without hs-bam (Figure 2E-H), suggesting that ectopic Bam expression can drive SGCs to differentiate. Collectively, these results support that the differentiation defects of SGCs are due to the lack of Bam expression.

SGCs retain moderate BMP signaling activation

Within the niche, BMP signaling functions to repress bam transcription to inhibit GSC differentiation (Chen and McKearin, 2003a; Song et al., 2004). To investigate BMP signaling activation in SGCs, we employed immunofluorescent staining for pMad, a well-characterized marker of BMP signaling activity (Kai and Spradling, 2003). Surprisingly, we observed undetectable pMad levels in all SGCs examined (n > 100) (Figure 3A, B). To investigate this further, we examined the activity of Dad-lacZ, a highly sensitive BMP signaling reporter known to be activated not only in GSCs but also in cystoblasts (Kai and Spradling, 2003; Song et al., 2004). Notably, 73% of SGCs were lacZ-positive (n = 107), a proportion lower than that of GSCs within the niche, which showed 100% lacZ positivity (n = 122) (Figure 3C, D). Furthermore, when comparing Dad-lacZ expression levels exclusively in lacZ-positive cells, we found that SGCs exhibited significantly lower expression levels than GSCs within the niche (Figure 3C, E). These findings indicate that BMP signaling is activated in SGCs but at lower levels than those in GSCs within the niche.

Figure 3.

Beyond maintaining Drosophila female GSCs in the niche, BMP signaling also promotes their division (Xie and Spradling, 1998). Since the activation levels of BMP signaling in SGCs were lower than those in GSCs within the niche, we hypothesized that SGCs would exhibit slower proliferation rates than GSCs. To test this hypothesis, we performed BrdU incorporation assays. The results revealed that only 4.5% of SGCs were BrdU-positive (n = 1034), a significantly lower proportion than the 7.8% observed in GSCs within the niche (n = 1337) (Figure 3F-H). These findings further corroborate the reduced activation of BMP signaling in SGCs relative to GSCs.

BMP signaling inhibits SGC differentiation

Then we investigated whether BMP signaling functions to inhibit SGC differentiation. The BMP type II receptor Punt and the co-Smad Medea (Med) are essential for maintaining GSC stemness within the niche (Xie and Spradling, 1998). To investigate whether they are also required to inhibit SGC differentiation, we established a genetic scenario, in which GFP^+/+ RFP^-/- germ cells are punt^-/-or med^-/-; GFP^+/- RFP^+/- germ cells are punt^+/- bam^+/- or med^+/- bam^+/-(similar to wild-type); and GFP^-/- RFP^+/+ germ cells are bam^-/-. In control experiments (with no punt or med mutation), GFP^+/+ RFP^-/- germ cells are wild-type; GFP^+/- RFP^+/- germ cells are bam^+/- (similar to wild-type); and GFP^-/- RFP^+/+ germ cells are bam^-/-(Figure 4A, see genotypes in Source data 1). Strikingly, the proportion of punt^-/- or med^-/-SGCs relative to total SGCs was significantly lower than in controls (Figure 4B-E). Conversely, among punt^-/- or med^-/-germ cells meeting our established criteria for SGC phenotype quantification, germline cysts constituted a higher percentage compared to controls (Figure 4F). These results indicate that Punt and Med function to inhibit SGC differentiation.

Figure 4.

Mothers against dpp (Mad) is the primary transcription factor of BMP signaling, and it is also essential for GSC maintenance in Drosophila ovaries (Xie and Spradling, 1998). Unlike punt and med, which reside on the same chromosome arm (3R) as bam, mad is located on a separate chromosome arm (2L). To investigate whether Mad is required to inhibit SGC differentiation, we established a genetic scenario, in which GFP^-/- germ cells are mad^-/- and GFP^+/+ germ cells are bam^-/-. In control experiments (with no mad mutation), GFP^-/- germ cells are wild-type and GFP^+/+ germ cells are bam^-/- (Figure 4G, see genotypes in Source data 1). Notably, mad mutation significantly decreased the SGC proportion relative to controls (Figure 4H-J). These results suggest that, like Punt and Med, Mad also plays a crucial role in suppressing SGC differentiation. Together, these findings demonstrate that BMP signaling contributes to inhibiting SGC differentiation, despite at reduced activation levels.

Germline tumors secret Dpp and Gbb

The next question we sought to address was which cells surrounding SGCs secrete BMP ligands, such as Dpp and Gbb. Guided by the Occam’s Razor principle, we focused on bam or bgcn mutant germline tumor cells, as the SGC phenotype is dependent on their presence. Initial attempts to detect expression of these two ligand genes using in situ hybridization proved unsuccessful. We therefore employed an established dpp-lacZ (P4-lacZ) reporter (Li et al., 2016) to monitor Dpp expression. Consistent with previous findings, we observed robust lacZ signals in cap cells (Figure 5A), the primary niche cells that are responsible for BMP ligand secretion (Chen and McKearin, 2003a; Song et al., 2004; Xie and Spradling, 2000). Notably, bam mutant germline tumor cells displayed lower lacZ expression levels than cap cells but significantly higher levels than differentiating cystocytes (Figure 5A-C).

Figure 5.

To more sensitively assess Dpp and Gbb expression, we performed real-time quantitative PCR (RT-qPCR) analyses in bam or bgcn mutant ovaries, comparing samples with and without germline-specific knockdown of dpp or gbb. Detection of reduced transcript levels in knockdown conditions would confirm active expression of these genes in the respective genetic backgrounds. Consistent with the essential roles of these two genes in fly viability, ubiquitous knockdown using act-GAL4 with either dpp-RNAi or gbb-RNAi caused lethality, which results also validated the efficacy of these RNAi lines. Notably, germline-specific knockdown of dpp or gbb significantly reduced their transcript levels compared to yellow (y) or white (w) knockdown controls (Figure 5D, E). These findings demonstrate that bam or bgcn mutant germline tumors secrete the BMP ligands, albeit at lower levels than cap cells.

Dpp and Gbb secreted by germline tumors are required to inhibit SGC differentiation

Finally, we investigated whether Dpp and Gbb secreted by germline tumors are required to inhibit SGC differentiation. Using a previously established double-mutant mosaic-analysis strategy for two genes on different chromosomes (Zhang et al., 2024; Zhang et al., 2023), we generated dpp bam or gbb bam double-mutant germline clones using two dpp mutant alleles, dpp^d6, dpp^d12, and one gbb allele, gbb¹ (Figure 6A, B, see genotypes in Source data 1). Heterozygotes in any of these alleles did not affect GSC maintenance, germ cell differentiation, and female fly fertility (Figure 6—figure supplement 1). However, both dpp bam and gbb bam double-mutant germline tumor cells exhibited reduced proliferation rates compared to bam single-mutant controls (Figure 6—figure supplement 2), indicating that autocrine BMP signaling promotes bam mutant tumor growth. As mentioned earlier, our evaluation focused on germ cells that have exited the niche and are surrounded by germline tumors to quantify the SGC phenotype. However, this raises the question of whether the extent of tumor encirclement (i.e., being surrounded by more or fewer tumor cells) influences the phenotype. To investigate this, we compared the SGC phenotype in bigger and smaller bam mutant germline tumors. Using confocal microscopy, we analyzed 70 germaria containing bam mutant germline clones that met our criteria for SGC phenotype quantification. For each bam mutant germline clone, we scanned and measured its maximal 2D cross-sectional area as a proxy for its 3D size. The 35 bigger and 35 smaller clones were categorized as ‘bigger’ and ‘smaller’ tumors, respectively. Strikingly, we observed no significant difference in the SGC phenotype between these two groups of tumors (Figure 6—figure supplement 3), suggesting that direct contact between tumorous and wild-type germ cells, rather than tumor size, is the primary determinant of this phenotype.

Figure 6.

The results above demonstrate that comparing the severity of the SGC phenotype is feasible between germ cells surrounded by smaller dpp bam or gbb bam double-mutant germline tumors and those surrounded by larger bam single-mutant germline tumors. Remarkably, both dpp bam and gbb bam double-mutant germline tumors enclosed fewer SGCs but more germline cysts than their bam single-mutant counterparts (Figure 6C-I). Thus, we concluded that the BMP ligands from directly-contacting germline tumor cells mediate the dominant inhibition of SGC differentiation. This amazingly parallels the mechanism observed in normal stem cell niche, where only germ cells in direct contact with cap cells are maintained as GSCs (Chen and McKearin, 2003a; Song et al., 2004; Xie and Spradling, 2000).

Discussion

Our study reveals that bam or bgcn mutant germline tumors in Drosophila ovaries secrete lower levels of BMP ligands Dpp and Gbb than cap cells, resulting in moderate BMP signaling activation in adjacent wild-type GSCs (called SGCs in this study). Such BMP signaling activation is sufficient to repress bam transcription, thereby blocking SGC differentiation (see our working model in Figure 7). Strikingly, this mechanism closely recapitulates the normal niche signaling program mediated by cap cells (Chen and McKearin, 2003a; Song et al., 2004; Xie and Spradling, 1998, 2000). To our knowledge, this represents the first evidence that tumor cells can functionally mimic a stem cell niche to arrest neighboring wild-type stem cells in an undifferentiated state.

Figure 7.

bam or bgcn mutant germline tumors consist of GSC-like cells (Lavoie et al., 1999; McKearin and Ohlstein, 1995), which are expected to resemble SGCs. However, out-of-niche bgcn mutant germline tumor cells did not activate BMP signaling to the same extent as SGCs, as evidenced by much lower Dad-lacZ expression in GFP-negative bgcn mutant tumor cells than in neighboring GFP-positive SGCs (Figure 3C). Consistent with this, while only 26% of SGCs expressed bamP-GFP, most of out-of-niche bam mutant germline tumor cells were bamP-GFP-positive (Figure 2B, C), further supporting reduced BMP signaling in these tumor cells relative to SGCs. These findings suggest that SGCs are more responsive to BMP signals secreted by germline tumors than the tumors themselves. Future studies are needed to elucidate the underlying mechanisms.

One interesting finding is that bam or bgcn mutant germline tumors secrete lower levels of BMP ligands than cap cells (Figure 5A-C). This aligns with earlier microarray data showing that purified Drosophila female GSCs express minimal Dpp and Gbb (Kai et al., 2005). However, our work reveals that such BMP levels in germline tumors are functionally critical to suppress SGC differentiation (Figure 6). Unlike normal GSCs, which receive unidirectional BMP ligands from cap cells (Chen and McKearin, 2003a; Li et al., 2016; Song et al., 2004; Xie and Spradling, 2000), SGCs are often fully surrounded by bam or bgcn mutant germline tumors. This spatial advantage likely enables tumors to inhibit SGC differentiation efficiently without matching the high BMP output of cap cells. Moreover, since BMP signaling is known to both inhibit normal GSC differentiation and promote their proliferation (Xie and Spradling, 1998), it should similarly stimulate SGC expansion, which is detrimental for bam or bgcn mutant germline tumors. We propose that these tumor cells finely regulate BMP secretion to balance these opposing demands: maintaining differentiation blockade of SGCs while avoiding stimulation of their excessive proliferation.

A well-established principle in oncology is that tumor aggressiveness correlates with poor differentiation, with less-differentiated tumors exhibiting enhanced transformative capacity and metastatic potential (Jogi et al., 2012; Lytle et al., 2018). In Drosophila ovaries, bam or bgcn mutant germline tumors consist of GSC-like cells that may resemble these poorly differentiated human tumors (Lavoie et al., 1999; McKearin and Ohlstein, 1995). This similarity raises the possibility that stem cell-like human tumors may similarly inhibit the differentiation of adjacent wild-type stem cells. By blocking differentiation, such tumors could deplete terminally differentiated cell populations, potentially exacerbating patient mortality. This mechanism may contribute to the heightened lethality of poorly differentiated tumors. Further investigation is needed to test this hypothesis.

The differentiation of a single GSC into a 16-cell germline cyst, comprising 15 polyploid nurse cells and one developing oocyte, represents a substantial metabolic investment (Fuller and Spradling, 2007; Lin, 1997). We propose that bam or bgcn mutant germline tumors block this process to divert nutrients toward their own uncontrolled growth. This phenomenon could have broad implications, as many human tissues and organs (intestine, muscle, skin, blood system, male germline, etc.) similarly depend on adult stem cells for homeostasis (Blanpain and Fuchs, 2006; Gehart and Clevers, 2019; Sousa-Victor et al., 2022; Spradling et al., 2011; Wilkinson et al., 2020). Notably, these stem cell-dependent tissues and organs are frequent sites of tumorigenesis, raising the possibility that human cancers may similarly impair neighboring stem cell differentiation to optimize nutrient allocation for malignant growth. A key limitation of our study is that the evidence is derived solely from Drosophila germline. Future work should explore whether similar regulatory paradigms operate in mammalian tissues during tumorigenesis.

Materials and Methods

Key resources table

Data availability

All genotypes are described in Source data 1, and the raw quantification data are included in Source data 2.

Fly husbandry

Flies were raised at 25°C on standard cornmeal/molasses/agar media.

Transgenic flies

hs-bam on chromosome 3R

The coding sequence of the bam gene, amplified from the cDNA clone, was cloned into the BglII-XbaI sites of the pCaSpeR-hs vector (a gift from Eric Baehrecke), while the attB sequence was inserted into the XhoI site. The resulting attB-pCaSpeR-hs-bam plasmid was then microinjected into the attP154 (Chromosome 3R, 97D2) fly strain to generate site-specific transgenic flies.

Heatshock method to induce germline clones

To ensure developmental synchrony and maintain low-density growth, eggs within 8 hours of laying were collected for heatshock treatment. The animals (late-Larva 3/early-Pupa stage) were subjected to twice-daily heatshocks at 37°C (2 hours per session, with a 6-hour interval between the two sessions) for 6 consecutive days.

Fertility test

For each genotype, three independent crosses were performed. Each cross vial contained two females and four w¹¹¹⁸ (wild-type) males, all aged three days old. The crosses were transferred to fresh vials every two days, with five replicate vials quantified per genotype. After all adult flies eclosed, offspring production was assessed by counting the number of empty pupae on the vial walls.

BrdU labeling

Ovaries were dissected in Schneider’s insect medium (SIM) and incubated in freshly prepared BrdU solution (100 μg/mL in SIM) for five hours at 25°C. After washing with PBS (30 min), samples were fixed in 4% paraformaldehyde (PBS-diluted) for three hours, followed by another PBS wash (30 min). Samples were then treated with RQ1 DNase reaction solution (Promega, Madison, WI, USA) for one hour, washed with PBST (0.3% Triton X-100 in PBS, 30 min), and incubated overnight at 4°C with mouse anti-BrdU antibody. Following a PBST wash (one hour), ovaries were incubated with goat anti-mouse 546 and DAPI (0.1 μg/mL) in PBST for three hours, washed again in PBST (one hour), and mounted in 70% glycerol (autoclaved).

Immuno-florescent staining, image collection, and data Processing

Ovaries were dissected in PBS, fixed in 4% paraformaldehyde (in PBS) for three hours, washed with PBST (0.3% Triton X-100 in PBS) for 30 min, and then incubated overnight at 4°C with primary antibodies diluted in PBST as follows: mouse anti-BamC (1:5), mouse anti-β-Gal (1:200), mouse anti-α-Spectrin (1:100), rabbit anti-pMad (1:500), and rabbit anti-Vasa (1:2000). After washing with PBST (one hour), samples were incubated with Alexa Fluor-conjugated secondary antibodies (1:1000) and 0.1 μg/mL DAPI (in PBST) for three hours, followed by a final PBST wash (one hour). Ovaries were mounted in 70% glycerol and imaged using a Zeiss LSM 710 confocal microscope (Carl Zeiss AG, Baden-Württemberg, Germany). Images were processed with ZEN 3.0 SR imaging software (Carl Zeiss) and Adobe Photoshop 2022. The quantification data were processed by Microsoft Excel or GraphPad Prism.

RT-qPCR

Ovaries from 14-day-old flies were dissected, and total RNA was extracted using the RNeasy Micro Kit. Equal amounts of RNA were reverse-transcribed into cDNA using the HiFiScript cDNA Synthesis Kit. RT-qPCR was performed on a CFX Connect Real-Time PCR System (Bio-Rad) with ChamQ SYBR qPCR Master Mix. The PCR protocol consisted of an initial denaturation at 95°C for 30 min, followed by 40 cycles of 95°C for 10 sec and 60°C for 30 sec. Relative gene expression was calculated using the 2^−ΔΔCT method (Livak and Schmittgen, 2001). The primers described previously were used (Huang et al., 2017): dpp primer-1: 5’-TACCACGCCATCCACTCAAC-3’ dpp primer-2: 5’-GCTCGTTACTCGATACGGCT-3’ gbb primer-1: 5’-CTGGATCATCGCACCAGAGG-3’ gbb primer-2: 5’-GTCTGGACGATCGCATGGTT-3’, rp49 (internal control) primer-1: 5’-CACCGGATTCAAGAAGTTCC-3’ rp49 (internal control) primer-2: 5’-GACAATCTCCTTGCGCTTCT-3’

Acknowledgements

We thank Eric Baehrecke, Michael Buszczak, Zheng Guo, Ed Laufer, Ruth Lehmann, Rongwen Xi, Ting Xie, Zhaohui Wang, BDGP, BDSC, DSHB, and THFC for providing antibodies, plasmids, and fly strains. This study was supported by grants 32270841 and 32070871 from the National Natural Science Foundation of China (NSFC) to Shaowei Zhao.

Additional information

Author contributions

S.Z. conceived and supervised this study. Y.Z., Y.W., J.S., L.Y., Z.W., D.S., Y.D.Z., and S.Z. performed the experiments. S.Z. wrote the manuscript and all authors commented on it.

Additional files

Source data 1. All genotypes.

Source data 2. Raw quantification data.