Tumors mimic the niche to inhibit neighboring stem cell differentiation

Reviewer #1 (Public review):

Summary:

This preprint from Shaowei Zhao and colleagues presents results that suggest tumorous germline stem cells (GSCs) in the Drosophila ovary mimic the ovarian stem cell niche and inhibit the differentiation of neighboring non-mutant GSC-like cells. The authors use FRT-mediated clonal analysis driven by a germline-specific gene (nos-Gal4, UASp-flp) to induce GSC-like cells mutant for bam or bam's co-factor bgcn. Bam-mutant or bgcn-mutant germ cells produce tumors in the stem cell compartment (the germarium) of the ovary (Figure 1). These tumors contain non-mutant cells - termed SGC for single-germ cells. 75% of SGCs do not exhibit signs of differentiation (as assessed by bamP-GFP) (Figure 2). The authors demonstrate that block in differentiation in SGC is a result of suppression of bam expression (Figure 2). They present data suggesting that in 73% of SGCs, BMP signaling is low (assessed by dad-lacZ) (Figure 3) and proliferation is less in SGCs vs GSCs. They present genetic evidence that mutations in BMP pathway receptors and transcription factors suppress some of the non-autonomous effects exhibited by SGCs within bam-mutant tumors (Figure 4). They show data that bam-mutant cells secrete Dpp, but this data is not compelling (see below) (Figure 5). They provide genetic data that loss of BMP ligands (dpp and gbb) suppresses the appearance of SGCs in bam-mutant tumors (Figure 6). Taken together, their data support a model in which bam-mutant GSC-like cells produce BMPs that act on non-mutant cells (i.e., SGCs) to prevent their differentiation, similar to what is seen in the ovarian stem cell niche. This preprint from Shaowei Zhao and colleagues presents results that suggest tumorous germline stem cells (GSCs) in the Drosophila ovary mimic the ovarian stem cell niche and inhibit the differentiation of neighboring non-mutant GSC-like cells. The authors use FRT-mediated clonal analysis driven by a germline-specific gene (nos-Gal4, UASp-flp) to induce GSC-like cells mutant for bam or bam's co-factor bgcn. Bam-mutant or bgcn-mutant germ cells produce tumors in the stem cell compartment (the germarium) of the ovary (Figure 1). These tumors contain non-mutant cells - termed SGC for single-germ cells. 75% of SGCs do not exhibit signs of differentiation (as assessed by bamP-GFP) (Figure 2). The authors demonstrate that block in differentiation in SGC is a result of suppression of bam expression (Figure 2). They present data suggesting that in 73% of SGCs, BMP signaling is low (assessed by dad-lacZ) (Figure 3) and proliferation is less in SGCs vs GSCs. They present genetic evidence that mutations in BMP pathway receptors and transcription factors suppress some of the non-autonomous effects exhibited by SGCs within bam-mutant tumors (Figure 4). They show data that bam-mutant cells secrete Dpp, but this data is not compelling (see below) (Figure 5). They provide genetic data that loss of BMP ligands (dpp and gbb) suppresses the appearance of SGCs in bam-mutant tumors (Figure 6). Taken together, their data support a model in which bam-mutant GSC-like cells produce BMPs that act on non-mutant cells (i.e., SGCs) to prevent their differentiation, similar to what in seen in the ovarian stem cell niche.

Strengths:

(1) Use of an excellent and established model for tumorous cells in a stem cell microenvironment.

(2) Powerful genetics allow them to test various factors in the tumorous vs non-tumorous cells.

(3) Appropriate use of quantification and statistics.

Weaknesses:

(1) What is the frequency of SGCs in nos>flp; bam-mutant tumors? For example, are they seen in every germarium, or in some germaria, etc, or in a few germaria?

(2) Does the breakdown in clonality vary when they induce hs-flp clones in adults as opposed to in larvae/pupae?

(3) Approximately 20-25% of SGCs are bam+, dad-LacZ+. Firstly, how do the authors explain this? Secondly, of the 70-75% of SGCs that have no/low BMP signaling, the authors should perform additional characterization using markers that are expressed in GSCs (i.e., Sex lethal and nanos).

(4) All experiments except Figure 1I (where a single germarium with no quantification) were performed with nos-Gal4, UASp-flp. Have the authors performed any of the phenotypic characterizations (i.e., figures other than Figure 1) with hs-flp?

(5) Does the number of SGCs change with the age of the female? The experiments were all performed in 14-day-old adult females. What happens when they look at a young female (like 2-day-old). I assume that the nos>flp is working in larval and pupal stages, and so the phenotype should be present in young females. Why did the authors choose this later age? For example, is the phenotype more robust in older females? Or do you see more SGCs at later time points?

(6) Can the authors distinguish one copy of GFP versus 2 copies of GFP in germ cells of the ovary? This is not possible in the Drosophila testis. I ask because this could impact the clonal analyses diagrammed in Figure 4A and 4G and in 6A and B. Additionally, in most of the figures, the GFP is saturated, so it is not possible to discern one vs two copies of GFP.

(7) More evidence is needed to support the claim of elevated Dpp levels in bam or bgcn mutant tumors. The current results with the dpp-lacZ enhancer trap in Figure 5A, B are not convincing. First, why is the dpp-lacZ so much brighter in the mosaic analysis (A) than in the no-clone analysis (B)? It is expected that the level of dpp-lacZ in cap cells should be invariant between ovaries, and yet LacZ is very faint in Figure 5B. I think that if the settings in A matched those in B, the apparent expression of dpp-lacZ in the tumor would be much lower and likely not statistically significant. Second, they should use RNA in situ hybridization with a sensitive technique like hybridization chain reactions (HCR) - an approach that has worked well in numerous Drosophila tissues, including the ovary.

(8) In Figure 6, the authors report results obtained with the bamBG allele. Do they obtain similar data with another bam allele (i.e., bamdelta86)?

Reviewer #2 (Public review):

While the study by Zhang et al. provides valuable insights into how germline tumors can non-autonomously suppress the differentiation of neighboring wild-type germline stem cells (GSCs), several conceptual and technical issues limit the strength of the conclusions.

Major points:

(1) Naming of SGCs is confusing. In line 68, the authors state that "many wild-type germ cells located outside the niche retained a GSC-like single-germ-cell (SGC) morphology." However, bam or bgcn mutant GSCs are also referred to as "SGCs," which creates confusion when reading the text and interpreting the figures. The authors should clarify the terminology used to distinguish between wild-type SGCs and tumor (bam/bgcn mutant) SGCs, and apply consistent naming throughout the manuscript and figure legends.

a) The same confusion appears in Figure 2. It is unclear whether the analyzed SGCs are wild-type or bam mutant cells. If the SGCs analyzed are Bam mutants, then the lack of Bam expression and failure to differentiate would be expected and not informative. However, if the SGCs are wild-type GSCs located outside the niche, then the observation would suggest that Bam expression is silenced in these wild-type cells, which is a significant finding. The authors should clarify the genotype of the SGCs analyzed in Figure 2C, as this information is not currently provided.

b) In Figures 4B and 4E, the analysis of SGC composition is confusing. In the control germaria (bam mutant mosaic), the authors label GFP⁺ SGCs as "wild-type," which makes interpretation unclear. Note, this is completely different from their earlier definition shown in line 68.

c) Additionally, bam⁺/⁻ GSCs (the first bar in Figure 4E) should appear GFP⁺ and Red⁺ (i.e., yellow). It would be helpful if the authors could indicate these bam⁺/⁻ germ cells directly in the image and clarify the corresponding color representation in the main text. In Figure 2A, although a color code is shown, the legend does not explain it clearly, nor does it specify the identity of bam⁺/⁻ cells alone. Figure 4F has the same issue, and in this graph, the color does not match Figure 4A.

(2) The frequencies of bam or bgcn mutant mosaic germaria carrying [wild-type] SGCs or wild-type germ cell cysts with branched fusomes, as well as the average number of wild-type SGCs per germarium and the number of days after heat shock for the representative images, are not provided when Figure 1 is first introduced. Since this is the first time the authors describe these phenotypes, including these details is essential. Without this information, it is difficult for readers to follow and evaluate the presented observations.

(3) Without the information mentioned in point 2, it causes problems when reading through the section regarding [wild-type] SGCs induced by impairment of differentiation or dedifferentiation. In lines 90-97, the authors use the presence of midbodies between cystocytes as a criterion to determine whether the wild-type GSCs surrounded by tumor GSCs arise through dedifferentiation. However, the cited study (Mathieu et al., 2022) reports that midbodies can be detected between two germ cells within a cyst carrying a branched fusome upon USP8 loss.

a) Are wild-type germ cell cysts with branched fusomes present in the bam mutant mosaic germaria? What is the proportion of germaria containing wild-type SGCs versus those containing wild-type germ cell cysts with branched fusomes?

b) If all bam mutant mosaic germaria carry only wild-type GSCs outside the niche and no germaria contain wild-type germ cell cysts with branched fusomes, then examining midbodies as an indicator of dedifferentiation may not be appropriate.

c) If, however, some germaria do contain wild-type germ cell cysts with branched fusomes, the authors should provide representative images and quantify their proportion.

d) In line 95, although the authors state that 50 germ cell cysts were analyzed for the presence of midbodies, it would be more informative to specify how many germaria these cysts were derived from and how many biological replicates were examined.

(4) Note that both bam mutant GSCs and wild-type SGCs can undergo division to generate midbodies (double cells), as shown in Figure 4H. Therefore, the current description of the midbody analysis is confusing. The authors should clarify which cell types were examined and explain how midbodies were interpreted in distinguishing between cell division and differentiation.

(5) The data in Figure 5 showing Dpp expression in bam mutant tumorous GSCs are not convincing. The Dpp-lacZ signal appears broadly distributed throughout the germarium, including in escort cells. To support the claim more clearly, the authors should present corresponding images for Figures 5D and 5E, in which dpp expression was knocked down in the germ cells of bam or bgcn mutant mosaic germaria. Showing these images would help clarify the localization and specificity of Dpp-lacZ expression relative to the tumorous GSCs.

(6) While Figure 6 provides genetic evidence that bam mutant tumorous GSCs produce Dpp to inhibit the differentiation of wild-type SGCs, it should be noted that these analyses were performed in a dpp⁺/⁻ background. To strengthen the conclusion, the authors should include appropriate controls showing [dpp⁺/⁻; bam⁺/⁻] SGCs and [dpp⁺/⁻; bam⁺/⁻] germ cell cysts without heat shock (as referenced in Figures 6F and 6I).

(7) Previous studies have reported that bam mutant germ cells cause blunted escort cell protrusions (e.g., Kirilly et al., Development, 2011), which are known to contribute to germ cell differentiation (e.g., Chen et al., Frontiers in Cell and Developmental Biology, 2022). The authors should include these findings in the Discussion to provide a broader context and to acknowledge how alterations in escort cell morphology may further influence differentiation defects in their model.

(8) Since fusome morphology is an important readout of SGCs vs differentiation. All the clonal analysis should have fusome staining.

(9) Figure arrangement. It is somewhat difficult to identify the figure panels cited in the text due to the current panel arrangement.

(10) The number of biological replicates and germaria analyzed should be clearly stated somewhere in the manuscript-ideally in the Methods section or figure legends. Providing this information is essential for assessing data reliability and reproducibility.

Reviewer #3 (Public review):

Summary:

Zhang et al. investigated how germline tumors influence the development of neighboring wild-type (WT) germline stem cells (GSC) in the Drosophila ovary. They report that germline tumors inhibit the differentiation of neighboring WT GSCs by arresting them in an undifferentiated state, resulting from reduced expression of the differentiation-promoting factor Bam. They find that these tumor cells produce low levels of the niche-associated signaling molecules Dpp and Gbb, which suppress bam expression and consequently inhibit the differentiation of neighboring WT GSCs non-cell-autonomously. Based on these findings, the authors propose that germline tumors mimic the niche to suppress the differentiation of the neighboring stem cells.

Strengths:

This study addresses an important biological question concerning the interaction between germline tumor cells and WT germline stem cells in the Drosophila ovary. If the findings are substantiated, they could provide valuable insights applicable to other stem cell systems.

Weaknesses:

Previous work from Xie's lab demonstrated that bam and bgcn mutant GSCs can outcompete WT GSCs for niche occupancy. Furthermore, a large body of literature has established that the interactions between escort cells (ECs) and GSC daughters are essential for proper and timely germline differentiation (the differentiation niche). Disruption of these interactions leads to arrest of germline cell differentiation in a status with weak BMP signaling activation and low bam expression, a phenotype virtually identical to what is reported here.

Thus, it remains unclear whether the observed phenotype reflects "direct inhibition by tumor cells" or "arrested differentiation due to the loss of the differentiation niche". Because most data were collected at a very late stage (more than 10 days after clonal induction), when tumor cells already dominate the germarium, this question cannot be solved. To distinguish between these two possibilities, the authors could conduct a time-course analysis to examine the onset of the WT GSC-like single-germ-cell (SGC) phenotype and determine whether early-stage tumor clones with a few tumor cells can suppress the differentiation of neighboring WT GSCs with only a few tumor cells present. If tumor cells indeed produce Dpp and Gbb (as proposed here) to inhibit the differentiation of neighboring germline cells, a small cluster or probably even a single tumor cell generated at an early stage might prevent the differentiation of their neighboring germ cells.

The key evidence supporting the claim that tumor cells produce Gpp and Gbb comes from Figures 5 and 6, which suggest that tumor-derived dpp and gbb are required for this inhibition. However, interpretation of these data requires caution.

In Figure 5, the authors use dpp-lacZ to support the claim that dpp is upregulated in tumor cells (Figure 5A and 5B). However, the background expression in somatic cells (ECs and pre-follicular cells) differs noticeably between these panels. In Figure 5A, dpp-lacZ expression in somatic cells in 5A is clearly higher than in 5B, and the expression level in tumor cells appears comparable to that in somatic cells (dpp-lacZ single channel). Similarly, in Figure 5B, dpp-lacZ expression in germline cells is also comparable to that in somatic cells. Providing clear evidence of upregulated dpp and gbb expression in tumor cells (for example, through single-molecular RNA in situ) would be essential.

Most tumor data present in this study were collected from the bam[86] null allele, whereas the data in Figure 6 were derived from a weaker bam[BG] allele. This bam[BG] allele is not molecularly defined and shows some genetic interaction with dpp mutants. As shown in Figure 6E, removal of dpp from homozygous bam[BG] mutant leads to germline differentiation (evidenced by a branched fusome connecting several cystocytes, located at the right side of the white arrowhead). In Figure 6D, fusome is likely present in some GFP-negative bam[BG]/bam[BG] cells. To strengthen their claim that the tumor produces Dpp and Gbb to inhibit WT germline cell differentiation, the authors should repeat these experiments using the bam[86] null allele.

It is well established that the stem niche provides multiple functional supports for maintaining resident stem cells, including physical anchorage and signaling regulation. In Drosophila, several signaling molecules produced by the niche have been identified, each with a distinct function - some promoting stemness, while others regulate differentiation. Expression of Dpp and Gbb alone does not substantiate the claim that these tumor cells have acquired the niche-like property. To support their assertion that these tumors mimic the niche, the authors should provide additional evidence showing that these tumor cells also express other niche-associated markers. Alternatively, they could revise the manuscript title to more accurately reflect their findings.

In the Method section, the authors need to provide details on how dpp-lacZ expression levels were quantified and normalized.

Author response:

Reviewer #1 (Public review):

Summary:

This preprint from Shaowei Zhao and colleagues presents results that suggest tumorous germline stem cells (GSCs) in the Drosophila ovary mimic the ovarian stem cell niche and inhibit the differentiation of neighboring non-mutant GSC-like cells. The authors use FRT-mediated clonal analysis driven by a germline-specific gene (nos-Gal4, UASp-flp) to induce GSC-like cells mutant for bam or bam's cofactor bgcn. Bam-mutant or bgcn-mutant germ cells produce tumors in the stem cell compartment (the germarium) of the ovary (Figure 1). These tumors contain non-mutant cells - termed SGC for single-germ cells. 75% of SGCs do not exhibit signs of differentiation (as assessed by bamP-GFP) (Figure 2). The authors demonstrate that block in differentiation in SGC is a result of suppression of bam expression (Figure 2). They present data suggesting that in 73% of SGCs, BMP signaling is low (assessed by dad-lacZ) (Figure 3) and proliferation is less in SGCs vs GSCs. They present genetic evidence that mutations in BMP pathway receptors and transcription factors suppress some of the non-autonomous effects exhibited by SGCs within bam-mutant tumors (Figure 4). They show data that bam-mutant cells secrete Dpp, but this data is not compelling (see below) (Figure 5). They provide genetic data that loss of BMP ligands (dpp and gbb) suppresses the appearance of SGCs in bam-mutant tumors (Figure 6). Taken together, their data support a model in which bam-mutant GSC-like cells produce BMPs that act on nonmutant cells (i.e., SGCs) to prevent their differentiation, similar to what is seen in the ovarian stem cell niche.

Strengths:

(1) Use of an excellent and established model for tumorous cells in a stem cell microenvironment.

(2) Powerful genetics allow them to test various factors in the tumorous vs nontumorous cells.

(3) Appropriate use of quantification and statistics.

We greatly appreciate these comments.

Weaknesses:

(1) What is the frequency of SGCs in nos>flp; bam-mutant tumors? For example, are they seen in every germarium, or in some germaria, etc, or in a few germaria?

This is a great question. Because the SGC phenotype depends on the presence of germline tumor clones, our quantification was restricted to germaria that contained them.These quantification data ("SGCs and/or germline cysts per germarium with germline clones") will be presented in the revised Figure 1.

(2) Does the breakdown in clonality vary when they induce hs-flp clones in adults as opposed to in larvae/pupae?

Our initial attempts to induce ovarian hs-flp germline clones by heat-shocking adult flies were unsuccessful, with very few clones being observed. Therefore, we shifted our approach to an earlier developmental stage. Successful induction was achieved by subjecting late-L3/early-pupal animals to a twice-daily heatshock at 37°C for 6 consecutive days (2 hours per session with a 6-hour interval, see Lines 325-329) (Zhao et al., 2018).

(3) Approximately 20-25% of SGCs are bam+, dad-LacZ+. Firstly, how do the authors explain this? Secondly, of the 70-75% of SGCs that have no/low BMP signaling, the authors should perform additional characterization using markers that are expressed in GSCs (i.e., Sex lethal and nanos).

These 20-25% of SGCs are bamP-GFP⁺ dad-lacZ-, not bam⁺ dad-lacZ⁺ (see Figure 2C and 3D). They would be cystoblast-like cells that may have initiated a differentiation program toward forming germline cysts (see Lines 109-117). The 70-75% of SGCs that have low BMP signaling exhibit GSC-like properties, including: 1) dot-like spectrosomes; 2) dad-lacZ positivity; 3) absence of bamP-GFP expression. While additional markers would be beneficial, we think that this combination of properties is sufficient to classify these cells as GSC-like.

(4) All experiments except Figure 1I (where a single germarium with no quantification) were performed with nos-Gal4, UASp-flp. Have the authors performed any of the phenotypic characterizations (i.e., figures other than Figure 1) with hs-flp?

Yes, we initially identified the SGC phenotype through hs-flp-mediated mosaic analysis of bam or bgcn mutant in ovaries. However, as noted in our response to Weakness (2), this approach was very labor-intensive. Therefore, we switched to using the more convenient nos::flp system for subsequent experiments. To our observation, there was no difference in the SGC phenotype between these two approaches, confirming that the nos::flp system is a valid and more practical alternative for its study.

(5) Does the number of SGCs change with the age of the female? The experiments were all performed in 14-day-old adult females. What happens when they look at a young female (like 2-day-old). I assume that the nos>flp is working in larval and pupal stages, and so the phenotype should be present in young females. Why did the authors choose this later age? For example, is the phenotype more robust in older females? Or do you see more SGCs at later time points?

These are very good questions. Such time-course analysis data will be provided in revised Figure 1. The SGC phenotype depends on the presence of bam or bgcn mutant germline clones. Germaria from 14-day-old flies contained bigger and more such clones than those from younger flies. This age-dependent increase in clone size and frequency significantly enhanced the efficiency of our quantification (see Lines 129-131).

(6) Can the authors distinguish one copy of GFP versus 2 copies of GFP in germ cells of the ovary? This is not possible in the Drosophila testis. I ask because this could impact the clonal analyses diagrammed in Figure 4A and 4G and in 6A and B. Additionally, in most of the figures, the GFP is saturated, so it is not possible to discern one vs two copies of GFP.

We greatly appreciate this comment. It was also difficult for us to distinguish 1 and 2 copies of GFP in the Drosophila ovary. In Figure 4A-F, to resolve this problem, we used a triplecolor system, in which red germ cells (RFP^+/+ GFP^-/-) are bam mutant, yellow germ cells (RFP^+/- GFP^+/-) are wild-type, and green germ cells (RFP^-/- GFP^+/+) are punt or med mutant. In Figure 4G-J, we quantified the SGC phenotype only in black germ cells (GFP^-/-), which are wild-type (control) or mad mutant. In Figure 6, we quantified the SGC phenotype only in green germ cells (both GFP^+/+ and GFP^+/-), all of which are wild-type.

(7) More evidence is needed to support the claim of elevated Dpp levels in bam or bgcn mutant tumors. The current results with the dpp-lacZ enhancer trap in Figure 5A, B are not convincing. First, why is the dpp-lacZ so much brighter in the mosaic analysis (A) than in the no-clone analysis (B)? It is expected that the level of dpplacZ in cap cells should be invariant between ovaries, and yet LacZ is very faint in Figure 5B. I think that if the settings in A matched those in B, the apparent expression of dpp-lacZ in the tumor would be much lower and likely not statistically significant. Second, they should use RNA in situ hybridization with a sensitive technique like hybridization chain reactions (HCR) - an approach that has worked well in numerous Drosophila tissues, including the ovary.

We appreciate this critical comment. The settings of immunofluorescent staining and confocal parameters in Figure 5A were the same as those in 5B. To our observation, the level of dpp-lacZ in cap cells was variable across germaria, even within the same ovary, as quantified in Figure 5C. We will provide RNA in situ hybridization data to further strengthen the conclusion that bam or bgcn mutant germline tumors secret BMP ligands.

(8) In Figure 6, the authors report results obtained with the bamBG allele. Do they obtain similar data with another bam allele (i.e., bamdelta86)?

No. Given that bam^BG was functionally indistinguishable from bam^Δ86 in inducing the SGC phenotype (compare Figure 6F, I with Figure 6-figure supplement 3C), we believe that repeating these experiments with bam^Δ86 would be redundant and would not alter the key conclusion of our study. Thanks for the understanding!

Reviewer #2 (Public review):

While the study by Zhang et al. provides valuable insights into how germline tumors can non-autonomously suppress the differentiation of neighboring wild-type germline stem cells (GSCs), several conceptual and technical issues limit the strength of the conclusions.

Major points:

(1) Naming of SGCs is confusing. In line 68, the authors state that "many wild-type germ cells located outside the niche retained a GSC-like single-germ-cell (SGC) morphology." However, bam or bgcn mutant GSCs are also referred to as "SGCs," which creates confusion when reading the text and interpreting the figures. The authors should clarify the terminology used to distinguish between wild-type SGCs and tumor (bam/bgcn mutant) SGCs, and apply consistent naming throughout the manuscript and figure legends.

We apologize for any confusion. In our manuscript, the term "SGC" is reserved specifically for wild-type germ cells that maintain a GSC-like morphology outside the niche. bam or bgcn mutant germ cells are referred to as GSC-like tumor cells (Lines 87-88), not SGCs.

(a) The same confusion appears in Figure 2. It is unclear whether the analyzed SGCs are wild-type or bam mutant cells. If the SGCs analyzed are Bam mutants, then the lack of Bam expression and failure to differentiate would be expected and not informative. However, if the SGCs are wild-type GSCs located outside the niche, then the observation would suggest that Bam expression is silenced in these wildtype cells, which is a significant finding. The authors should clarify the genotype of the SGCs analyzed in Figure 2C, as this information is not currently provided.

The SGCs analyzed in Figure 2A-C are wild-type, GSC-like cells located outside the niche. They were generated using the same genetic strategy depicted in Figures 1C and 1E (with the schematic in Figure 1B). The complete genotypes for all experiments are available in

Source data 1.

(b) In Figures 4B and 4E, the analysis of SGC composition is confusing. In the control germaria (bam mutant mosaic), the authors label GFP⁺ SGCs as "wild-type," which makes interpretation unclear. Note, this is completely different from their earlier definition shown in line 68.

The strategy to generate SGCs in Figure 4B-F (with the schematic in Figure 4A) is completely different from that in Figure 1C-F, H, and I (with the schematic in Figure 1B). In Figure 4B-F, we needed to distinguish punt^-/- (or med^-/-) with punt^+/- (or med^+/-) germ cells. As noted in our response to Reviewer #1’s Weakness (6), it was difficult for us to distinguish 1 and 2 copies of GFP in the Drosophila ovary. Therefore, we chose to use the triple-color system to distinguish these germ cells in Figure 4B-F (see genotypes in Source data 1).

(c) Additionally, bam⁺/⁻ GSCs (the first bar in Figure 4E) should appear GFP⁺ and Red⁺ (i.e., yellow). It would be helpful if the authors could indicate these bam⁺/⁻ germ cells directly in the image and clarify the corresponding color representation in the main text. In Figure 2A, although a color code is shown, the legend does not explain it clearly, nor does it specify the identity of bam⁺/⁻ cells alone. Figure 4F has the same issue, and in this graph, the color does not match Figure 4A.

The color-to-genotype relationships for the schematics in Figures 2A and 4E are provided in Figures 1B and 4A, respectively. Due to the high density of germ cells, it is impractical to label each genotype directly in the images. In contrast to Figure 4E, the colors in Figure 4F do not represent genotypes; instead, blue denotes the percentage of SGCs, and red denotes the percentage of germline cysts, as indicated below the bar chart.

(2) The frequencies of bam or bgcn mutant mosaic germaria carrying [wild-type] SGCs or wild-type germ cell cysts with branched fusomes, as well as the average number of wild-type SGCs per germarium and the number of days after heat shock for the representative images, are not provided when Figure 1 is first introduced. Since this is the first time the authors describe these phenotypes, including these details is essential. Without this information, it is difficult for readers to follow and evaluate the presented observations.

Thanks for this constructive suggestion. We will include such quantification data in the revised manuscript.

(3) Without the information mentioned in point 2, it causes problems when reading through the section regarding [wild-type] SGCs induced by impairment of differentiation or dedifferentiation. In lines 90-97, the authors use the presence of midbodies between cystocytes as a criterion to determine whether the wild-type GSCs surrounded by tumor GSCs arise through dedifferentiation. However, the cited study (Mathieu et al., 2022) reports that midbodies can be detected between two germ cells within a cyst carrying a branched fusome upon USP8 loss.

Unlike wild-type cystocytes, which undergo incomplete cytokinesis and lack midbodies, those with USP8 loss undergo complete cell division, with the presence of midbodies (white arrow, Figure 1F’ from Mathieu et al., 2022) as a marker of the late cytokinesis stage (Mathieu et al., 2022).

(a) Are wild-type germ cell cysts with branched fusomes present in the bam mutant mosaic germaria? What is the proportion of germaria containing wild-type SGCs versus those containing wild-type germ cell cysts with branched fusomes?

(b) If all bam mutant mosaic germaria carry only wild-type GSCs outside the niche and no germaria contain wild-type germ cell cysts with branched fusomes, then examining midbodies as an indicator of dedifferentiation may not be appropriate.

We greatly appreciate this critical comment. bam mutant mosaic germaria indeed contained wild-type germline cysts, as evidenced by an SGC frequency of ~70%, rather than 100% (see Figures 2H, 4F, 4J, 6F, 6I, and Figure 6-figure supplement 3C). Since the SGC phenotype depends on the presence of bam or bgcn mutant germline tumors, we quantified it as “the percentage of SGCs relative to the total number of SGCs and germline cysts that are surrounded by germline tumors” (see Lines 124-129). Quantifying the SGC phenotype as "the percentage of germaria with SGCs" would be imprecise. This is because the presence and number of SGCs were highly variable among germaria with bam mutant germline clones, and a small number of germaria entirely lacked these clones. We will provide the data of "SGCs and/or germline cysts per germarium with germline clones" in revised Figure 1.

(c) If, however, some germaria do contain wild-type germ cell cysts with branched fusomes, the authors should provide representative images and quantify their proportion.

Such representative germaria are shown in Figure 2G, 3B, 3C, 6D, 6E, and 6H. The percentage of germline cysts can be calculated by “100% - SGC%”.

(d) In line 95, although the authors state that 50 germ cell cysts were analyzed for the presence of midbodies, it would be more informative to specify how many germaria these cysts were derived from and how many biological replicates were examined.

As noted in our response to points a) and b) above, the germ cells surrounded by germline tumors, rather than germarial numbers, are more precise for analyzing the phenotype. For this experiment, we examined >50 such germline cysts via confocal microscopy. As the analysis was performed on a defined cellular population, this sample size should be sufficient to support our conclusion.

(4) Note that both bam mutant GSCs and wild-type SGCs can undergo division to generate midbodies (double cells), as shown in Figure 4H. Therefore, the current description of the midbody analysis is confusing. The authors should clarify which cell types were examined and explain how midbodies were interpreted in distinguishing between cell division and differentiation.

We assayed for the presence of midbodies or not specifically within the germline cysts surrounded by bam mutant tumors, not within the tumors themselves (Lines 94-95). As detailed in Lines 88-97, the absence of midbodies was used as a key criterion to exclude the possibility of dedifferentiation.

(5) The data in Figure 5 showing Dpp expression in bam mutant tumorous GSCs are not convincing. The Dpp-lacZ signal appears broadly distributed throughout the germarium, including in escort cells. To support the claim more clearly, the authors should present corresponding images for Figures 5D and 5E, in which dpp expression was knocked down in the germ cells of bam or bgcn mutant mosaic germaria. Showing these images would help clarify the localization and specificity of Dpp-lacZ expression relative to the tumorous GSCs.

We greatly appreciate this comment. RNA in situ hybridization data will be provided to further strengthen the conclusion that bam or bgcn mutant germline tumors secret BMP ligands.

(6) While Figure 6 provides genetic evidence that bam mutant tumorous GSCs produce Dpp to inhibit the differentiation of wild-type SGCs, it should be noted that these analyses were performed in a dpp⁺/⁻ background. To strengthen the conclusion, the authors should include appropriate controls showing [dpp⁺/⁻; bam⁺/⁻] SGCs and [dpp⁺/⁻; bam⁺/⁻] germ cell cysts without heat shock (as referenced in Figures 6F and 6I).

Schematic cartoons in Figure 6A and 6B demonstrate that these analyses were performed in a dpp^+/- background. Figure 6-figure supplement 1 indicates that dpp^+/- or gbb^+/- does not affect GSC maintenance, germ cell differentiation, and female fly fertility. Figure 6C is the control for 6D and 6E, and 6G is the control for 6H, with quantification in 6F and 6I. We used nos::flp, not the heat shock method, to induce germline clones in these experiments (see genotypes in Source data 1).

(7) Previous studies have reported that bam mutant germ cells cause blunted escort cell protrusions (e.g., Kirilly et al., Development, 2011), which are known to contribute to germ cell differentiation (e.g., Chen et al., Frontiers in Cell and Developmental Biology, 2022). The authors should include these findings in the Discussion to provide a broader context and to acknowledge how alterations in escort cell morphology may further influence differentiation defects in their model.

Thanks for teaching us! Such discussion will be included in the revised manuscript.

(8) Since fusome morphology is an important readout of SGCs vs differentiation. All the clonal analysis should have fusome staining.

SGC is readily distinguishable from multi-cellular germline cyst based on morphology. In some clonal analysis experiments, fusome staining was not feasible due to technical limitations such as channel saturation or antibody incompatibility. Thanks for the understanding!

(9) Figure arrangement. It is somewhat difficult to identify the figure panels cited in the text due to the current panel arrangement.

The figure panels were arranged to optimize space while ensuring that related panels are grouped in close proximity for logical comparison. We would be happy to consider any specific suggestions for an alternative layout that could improve clarity. Thanks!

(10) The number of biological replicates and germaria analyzed should be clearly stated somewhere in the manuscript-ideally in the Methods section or figure legends. Providing this information is essential for assessing data reliability and reproducibility.

Thanks for this constructive suggestion. Such information will be included in figure legends in the revised manuscript.

Reviewer #3 (Public review):

Summary:

Zhang et al. investigated how germline tumors influence the development of neighboring wild-type (WT) germline stem cells (GSC) in the Drosophila ovary. They report that germline tumors inhibit the differentiation of neighboring WT GSCs by arresting them in an undifferentiated state, resulting from reduced expression of the differentiation-promoting factor Bam. They find that these tumor cells produce low levels of the niche-associated signaling molecules Dpp and Gbb, which suppress bam expression and consequently inhibit the differentiation of neighboring WT GSCs non-cell-autonomously. Based on these findings, the authors propose that germline tumors mimic the niche to suppress the differentiation of the neighboring stem cells.

Strengths:

This study addresses an important biological question concerning the interaction between germline tumor cells and WT germline stem cells in the Drosophila ovary. If the findings are substantiated, they could provide valuable insights applicable to other stem cell systems.

We greatly appreciate these comments.

Weaknesses:

Previous work from Xie's lab demonstrated that bam and bgcn mutant GSCs can outcompete WT GSCs for niche occupancy. Furthermore, a large body of literature has established that the interactions between escort cells (ECs) and GSC daughters are essential for proper and timely germline differentiation (the differentiation niche). Disruption of these interactions leads to arrest of germline cell differentiation in a status with weak BMP signaling activation and low bam expression, a phenotype virtually identical to what is reported here. Thus, it remains unclear whether the observed phenotype reflects "direct inhibition by tumor cells" or "arrested differentiation due to the loss of the differentiation niche". Because most data were collected at a very late stage (more than 10 days after clonal induction), when tumor cells already dominate the germarium, this question cannot be solved. To distinguish between these two possibilities, the authors could conduct a time-course analysis to examine the onset of the WT GSC-like singlegerm-cell (SGC) phenotype and determine whether early-stage tumor clones with a few tumor cells can suppress the differentiation of neighboring WT GSCs with only a few tumor cells present. If tumor cells indeed produce Dpp and Gbb (as proposed here) to inhibit the differentiation of neighboring germline cells, a small cluster or probably even a single tumor cell generated at an early stage might prevent the differentiation of their neighboring germ cells.

Thanks for this critical comment. Such time-course analysis data will be provided in revised Figure 1.

The key evidence supporting the claim that tumor cells produce Gpp and Gbb comes from Figures 5 and 6, which suggest that tumor-derived dpp and gbb are required for this inhibition. However, interpretation of these data requires caution. In Figure 5, the authors use dpp-lacZ to support the claim that dpp is upregulated in tumor cells (Figure 5A and 5B). However, the background expression in somatic cells (ECs and pre-follicular cells) differs noticeably between these panels. In Figure 5A, dpp-lacZ expression in somatic cells in 5A is clearly higher than in 5B, and the expression level in tumor cells appears comparable to that in somatic cells (dpplacZ single channel). Similarly, in Figure 5B, dpp-lacZ expression in germline cells is also comparable to that in somatic cells. Providing clear evidence of upregulated dpp and gbb expression in tumor cells (for example, through single-molecular RNA in situ) would be essential.

We greatly appreciate this critical comment. In our data, the expression of dpp-lacZ in cap cells was variable across germaria, even within the same ovary, as quantified in Figure 5C. The images in Figures 5A and 5B were selected as representative examples of positive signaling. To directly address the reviewer's point and strengthen our conclusion, we will perform RNA in situ hybridization data in the revised manuscript to visualize the expression of BMP ligands within the bam or bgcn mutant germline tumor cells.

Most tumor data present in this study were collected from the bam[86] null allele, whereas the data in Figure 6 were derived from a weaker bam[BG] allele. This bam[BG] allele is not molecularly defined and shows some genetic interaction with dpp mutants. As shown in Figure 6E, removal of dpp from homozygous bam[BG] mutant leads to germline differentiation (evidenced by a branched fusome connecting several cystocytes, located at the right side of the white arrowhead). In Figure 6D, fusome is likely present in some GFP-negative bam[BG]/bam[BG] cells. To strengthen their claim that the tumor produces Dpp and Gbb to inhibit WT germline cell differentiation, the authors should repeat these experiments using the bam[86] null allele.

Although a structure resembling a "branched fusome" is visible in Figure 6E (right of the white arrowhead), it is an artifact resulting from the cytoplasm of GFP-positive follicle cells, which also stain for α-Spectrin, projecting between germ cells of different clones (see the merged image). In both our previous (Zhang et al., 2023) and current studies, bam^BG was functionally indistinguishable from bam^Δ86 in its ability to block GSC differentiation and induce the SGC phenotype (compare Figure 6F, I with Figure 6-figure supplement 3C). Given this, we believe that repeating the extensive experiments in Figure 6 with the bam^Δ86 allele would be scientifically redundant and would not change the key conclusion of our study. We thank the reviewer for their consideration.

It is well established that the stem niche provides multiple functional supports for maintaining resident stem cells, including physical anchorage and signaling regulation. In Drosophila, several signaling molecules produced by the niche have been identified, each with a distinct function - some promoting stemness, while others regulate differentiation. Expression of Dpp and Gbb alone does not substantiate the claim that these tumor cells have acquired the niche-like property. To support their assertion that these tumors mimic the niche, the authors should provide additional evidence showing that these tumor cells also express other niche-associated markers. Alternatively, they could revise the manuscript title to more accurately reflect their findings.

Dpp and Gbb are the key niche signals from cap cells for maintaining GSC stemness. Our work demonstrates that germline tumors can specifically mimic this signaling function, not the full suite of cap cell properties, to create a non-cell-autonomous differentiation block. The current title “Tumors mimic the niche to inhibit neighboring stem cell differentiation” reflects this precise concept: a partial, functional mimicry of the niche's most relevant activity in this context. We feel it is an appropriate and compelling summary of our main conclusion.

In the Method section, the authors need to provide details on how dpp-lacZ expression levels were quantified and normalized.

Thanks for this suggestion. Such information will be included in the revised manuscript.