Traffic jam regulates the function of the ovarian germline stem cell progeny differentiation niche during pre-adult stage in Drosophila

Stem cell self-renewal and the daughter cell differentiation are tightly regulated by the respective niches, which produce extrinsic cues to support the proper development. In Drosophila ovary, Dpp is secreted from germline stem cell (GSC) niche and activates the BMP signaling in GSCs for their self-renewal. Escort cells (ECs) in differentiation niche restrict Dpp outside the GSC niche and extend protrusions to help with proper differentiation of the GSC daughter cells. Here we provide evidence that loss of large Maf transcriptional factor Traffic jam (Tj) blocks GSC progeny differentiation. Spatio-temporal specific knockdown experiments indicate that Tj is required in pre-adult EC lineage for germline differentiation control. Further molecular and genetic analyses suggest that the defective germline differentiation caused by tj-depletion is partly attributed to the elevated dpp in the differentiation niche. Moreover, our study reveals that tj-depletion induces ectopic En expression outside the GSC niche, which contributes to the upregulated dpp expression in ECs as well as GSC progeny differentiation defect. Alternatively, loss of EC protrusions and decreased EC number elicited by tj-depletion may also partially contribute to the germline differentiation defect. Collectively, our findings suggest that Tj in ECs regulates germline differentiation by controlling the differentiation niche characteristics.

The EC-mediated molecular mechanisms to extrinsically modulate GSC progeny differentiation are under extensive study. Based on previous studies, suppression of Dally 16 and expression of Tkv 17 in ECs are necessary to restrict diffusion of GSC niche-derived Dpp within the range of one-cell-diameter. It has been shown that Dally suppression in ECs is modulated by MAPK signaling 18 and Eggless 19 . Interestingly, Wnt signaling in ECs regulates the Tkv expression without repressing Dally and Dpp 17 . On the other hand, Lsd1 20,21 , Piwi 22,23 , COP9/ Hh 24,25 and H1 26 have been reported to control GSC progeny differentiation through repressing Dpp expression in ECs. Moreover, Rho1 10 and Bre1/Set1 27 function in ECs to inhibit both Dpp and Dally production. Hh signaling antagonizes JAK/STAT activity to maintain an optimum Dpp expression 3 and PRC1 represses the transcription of dpp-RB isoform in ECs 28 . In addition, Engrailed (En) has been shown to activate transcription of dpp in CpCs 29 . Besides all the above, the detailed molecular mechanisms for suppressing BMP ligands production in the differentiation niche still remain to be explored.
Traffic jam (Tj), also known as dMaf, is the only large Maf transcription factor in Drosophila melanogaster, whose homologs play pivotal roles in vertebrate development [30][31][32][33][34] . tj mRNA becomes detectable from late embryonic stage 12 onward in somatic gonadal precursor cells (SGPs) 33,35 . Tj protein has already been distributed in the nuclei of SGPs of embryonic gonads at stage 15-16 33 . And the expression is persisted in germline-associated somatic cells during ovarian development and oogenesis, including intermingled cells (ICs), CpCs, ECs and follicle cells (FCs) 33,36,37 . tj mutant ovaries are malformed and lack of ovariole subdivision 37,38 . Tj-mediated suppression of E-cadherin expression is required for SGPs to become ICs 39 . Moreover, Tj in CpCs inhibits the TF cell fate 37 . Considering Tj expression in ECs, we wonder if Tj plays essential extrinsic roles in controlling female germline differentiation.
Here we report evidence that Tj controls CB differentiation in a non-cell-autonomous manner during the pre-adult stage. tj-deficient pre-adult ECs produce ectopic En, which promotes dpp transcription and subsequently up-regulates BMP signaling activity in CBs, thereby causing differentiation defect. We further show that Tj is required in pre-adult ECs to maintain EC protrusions and establish a proper EC number. This study suggests new mechanisms by which Tj preserves the properties of differentiation niche to promote proper germline differentiation.

Results
Tj regulates germ cell differentiation in Drosophila ovaries. It has been reported that loss of Tj blocks germ cell differentiation in Drosophila testis 33 . This prompted us to investigate a potential role of Tj in regulating female germline differentiation. We first assayed aberrant germ cell differentiation in tj mutant ovaries by immunofluorescence. Beginning at late third larval instar (LL3), primordial germ cells (PGCs) -derived germ cell differentiation initiates and proceeds in Drosophila female gonads 40,41 , producing differentiating germ cells including CBs and actively dividing cysts. Those germ cells at different developmental stages can be morphologically distinguished by the fusome, a germline specific cytoplasmic organelle which appears spherical in PGCs or GSCs and grows new branches after each cyst division 42 . At white pre-pupal stage (120 h AEL), germ cell differentiation in control gonads is evident by the presence of germline cysts containing branched fusomes, as visualized by staining with anti-α-spectrin antibody (4 out of 20 gonads examined, Fig. 1a,a' , white arrow). Compared with the control, differentiating germline cysts were detected at a smaller percentage in the gonads transheterozygous for tj Δ1 and tj PL3 (1 out of 20 gonads examined). The vast majority of germ cells in tj Δ1/PL3 gonads also harbored a spherical fusome (Fig. 1b,b'). Notably, PGCs formed a cluster and were surrounded by a layer of somatic cells labeled by anti-Zfh-1, as previously reported 33 . At early pupal stage (EP, 144 h AEL), the wild type gonads have developed into 18-20 primordial ovarioles. We checked each ovariole in the EP gonads and scored the frequency of ovarioles lacking branched fusomes. We found that all the ovarioles contained differentiating germline cysts in control gonads (180 primordial ovarioles, Fig. 1c,c' , white arrows). In contrast, about 80% primordial ovarioles of tj mutant EP gonads lacked germline cysts with branched fusomes, but exhibited spectrosome-containing germ cells accumulation (152 primordial ovarioles, Fig. 1d,d' , white arrowheads). CpCs were disc-shaped and arranged in a single row similar to TF cells ( Fig. 1d' , white open ovals and yellow dashed lines). These results presented a similar phenotype to the observation of Panchal, et al. 37 . In addition, the ovaries from adult tj Δ1/PL3 were small and lacked mature germ cells, as previously reported 33,37 .
As Tj is expressed in IC/EC lineage in developing female gonads 33,36 , we sought to determine its extrinsic effects on germ cell differentiation in the ovaries. For this purpose, we knocked down the expression of tj specifically in somatic cells using either UAS-tj RNAi-1 or UAS-tj RNAi-2 driven by somatic specific tj-GAL4. Prior to the test, the RNAi knockdown (KD for short) efficiency was evaluated at the protein level using antibody against Tj in LL3 gonads. As previously reported, Tj protein is concentrated in ICs since L3 33,36 (Fig. S1a,a'). Our results revealed that these two independent RNAi lines driven by tj-GAL4 reduced the Tj protein levels substantially in somatic lineage (Fig. S1b-c') and led to a drastic phenotype similar to tj transhterozygotes ( Fig. S1d-f '). However, we could only get severely degenerated ovaries after eclosion probably due to a lack of mature germline or a dramatic loss of ECs. To restrict the expression time window for tj RNAi, tj-GAL4 combined with tub-GAL80 ts was used. The flies were initially raised at 18 °C and then switched to 29 °C at different time points during pupal stage or adulthood. However, once the temperature shift carried out, the ovaries from any experimental groups degenerated rapidly, even though the flies were switched to 25 °C. This technical setback prompted us to express tj RNAi with C587-GAL4, another somatic specific driver active in developing gonad and adult germarium 6,43 . This led to a minor impairment of Tj expression since larval stage (Fig. S2a-i"), thus bypassing the strong phenotype of tj loss-of-function in earlier stage. In control germaria (Fig. 1e), 0 to 4 CBs were identified by spherical fusomes located not immediately adjacent to CpCs. As shown in Fig. 1f, reduced Tj expression in the somatic lineage led to a remarkable increase of spherical fusome-containing single germ cells classified as undifferentiated germ cells (UGCs). Note that the interaction between PGCs and ICs was not affected by tj KD in the LL3 gonads (Fig. S2a-c"), excluding the possibility that PGC cluster contributed to the adult phenotype. Next, we generated MARCM 44 (the Mosaic Analysis with a Repressible Cell Marker) clones homozygous for tj PL3 to further analyze the phenotype in adulthood. The control (FRT40A) and the tj mutant (FRT40A, tj PL3 ) clones were induced in parallel (see Methods). The ovaries were stained for GFP as a positive clone marker and for α-spectrin to label the fusome. In control germaria (Fig. 1g), most of ECs were labeled with GFP. Control EC clones almost never exhibited UGC accumulation, with only one exception that contained 5 UGCs (1 out of 233 germaria). Although the percentage of germaria with clonal EC was similar to the control group, only small clones of ECs mutant for tj were recovered in this condition (Fig. 1h). There was 6.23% of the germaria (17 out of 273 germaria with clonal EC and without clonal TF and CpCs) exhibiting excess UGCs (Fig. 1i). Lower penetrance is potentially owing to the smaller clone size. Taken together, these data indicate that Tj is required to regulate GSC progeny differentiation in Drosophila ovarian development.
Tj functions mainly in pre-adult ECs to control proper germ cell differentiation. To characterize the developmental stage-specific requirements for Tj in GSC progeny differentiation control, C587-GAL4 combined with tub-GAL80 ts (C587 ts for short) was employed to control tj RNAi expression at a particular stage. Available UAS-tj RNAi transgenic lines and control UAS-mcherry RNAi were crossed to C587 ts and the progeny were subjected to time-course analysis. Flies were maintained at the permissive temperature (18 °C) until a given developmental stage and then shifted to the restrictive temperature (29 °C) for several days in order to knock down tj specifically during developmental or adult stage. First, the newly eclosed females (<1-day old) with developmental knockdown of tj were examined. Shifting the animals of tj KD to 29 °C at either LL3 (Fig. 2a) or EP (Fig. 2b) stage resulted in a striking expansion of UGC number. We subsequently examined the 10-day old  www.nature.com/scientificreports www.nature.com/scientificreports/ females shifted to 29 °C immediately after eclosion (Fig. 2c). However, we did not find UGC phenotype with statistical significance upon adult-specific depletion of tj, in which only about 1% germaria exhibited 5 UGCs. As a control group, animals were raised at 18 °C until eclosion, and then kept at 18 °C for 10 days. Ovaries from these females exhibited a normal number of CB (Fig. 2d). These data demonstrate that Tj is critically required during developmental stage, rather than adulthood, for proper germline differentiation (Figs 2e and S3a).
It is known that C587-GAL4 is active in most of the somatic lineage in developing female gonads 6,43 , which will differentiate into more specialized cell types in germaria during subsequent development, such as TF, CpC, IC/EC and early FCs 45 . To determine Tj functions in which cell types, hh-GAL4 combined with tub-GAL80 ts (hh ts for short) was used to express tj RNAi exclusively in TF and CpCs since their specification 39 . Flies were initially kept at 18 °C and shifted to 29 °C since LL3. Interestingly, the tj KD germaria from newly eclosed females (<1-day old) displayed greatly diminished Tj antibody staining in CpCs and normal germ cell differentiation (Figs 2f and S3b-g). However, about 50% of the germaria still exhibited detectable expression of Tj in 1-2 CpCs. Thus, we could not exclude the possibility that the remaining amount of Tj is enough to maintain normal germline differentiation. Taken together with the observation that germaria bearing tj mutant EC clones presented UGCs accumulation, we argue that Tj is predominantly required in pre-adult ECs for regulating Drosophila ovarian germ cell differentiation.

Tj is required in pre-adult ECs to promote CB differentiation by restricting BMP signaling.
Previous findings suggest that ectopic BMP signaling activated in germline differentiation zone results in an expansion of GSC/CB-like cells 10,[19][20][21][22][23][24][25][26][27]46 . To determine if Tj is required extrinsically for restricting the BMP signaling activity outside the GSC niche, we examined the BMP signaling activity by various reporters in flies expressing tj RNAi by C587-GAL4. In these experiments, crosses were performed and the resulting progeny were raised at 25 °C, and newly eclosed females (<1-day old) were dissected for detection. We first evaluated BMP signaling activity with antibody against phosphorylated Mad (pMad). In wild type germaria, pMad expression is restricted to the GSCs (Fig. 3a). Unexpectedly, in tj KD germaria, the pMad was barely detectable in UGCs (Fig. 3b,c). GFP reporter labeling the activity of the bam transcription, bamP-GFP 4,5 , was highly visible in all UGCs in tj KD germaria ( Fig. 3e,f). Whereas it is normally absent in GSCs and first become detectable in CBs and upregulated in mitotic cysts (Fig. 3d). Next, we checked the enhancer trap dad-lacZ 47 which had been reported as a low threshold reporter of BMP signaling 3 . dad (daughters against dpp) is a target gene for BMP signaling and the expression of dad-lacZ in germline is confined to GSCs and intermediate CBs in control germaria (Fig. 3g,g'). In contrast, the number of dad-lacZ-positive germ cell was significantly expanded in germaria of C587 > tj RNAi females ( Fig. 3h-j). These results indicate that tj depletion in pre-adult ECs causes an up-regulation of BMP signaling outside GSC niche. Moreover, the UGCs are blocked at a late CB stage.
To gain more evidence to support our observation, we further overexpressed dpp with C587 ts , which has been demonstrated to result in the formation of the accumulated GSC-like cells 10,48 . First the crosses were set up and the resulting progeny were raised at 18 °C until eclosion and then kept at 29 °C for 2-3 days. As expected, the accumulated UGCs in germaria exhibited robust pMad staining (Fig. 3k) and were devoid of bamP-GFP expression (Fig. 3l). As a second strategy for overexpression, the crosses were set up and the resulting progeny were raised at 18 °C until EP stage. And then the flies were shifted to 25 °C for a moderate overexpression of dpp in pre-adult ECs, as tub-GAL80 ts is not fully functional at this temperature 49 . The newly born ovaries of those females exhibited an expansion of UGCs devoid of pMad staining (Fig. 3m) and positive for bamP-GFP (Fig. 3n). These results indicate that a moderate elevation of dpp in pre-adult ECs results in an accumulation of CB-like germ cells. Taken together, we argue that depletion of tj in pre-adult ECs may induce an elevation of dpp outside the GSC niche.
Tj is required extrinsically to control germline differentiation by repressing dpp expression. To further elucidate mechanisms underlying expanded BMP signaling elicited by tj KD, quantitative RT-PCR was conducted to determine the dpp expression at the transcriptional level. Due to the limitation of the technique, we were not able to purify the ECs from ovaries. We isolated ovary RNAs of virgins carrying tj-GAL4 > tj RNAi or tj-GAL4 > mcherry RNAi, as the ovaries of each genotype were morphologically similar and contain few egg chambers, which made the comparison more valid. Notably, dpp mRNA was elevated in tj-GAL4-mediated tj RNAi ovaries compared with the control group (Fig. 4a), suggesting that Tj mediates the repression of dpp transcription. To visualize the changes of dpp expression in germaria, two recently published enhancer trap lines, dpp2.0-lacZ 29 and P4-lacZ 28 , were employed to trace the dpp transcription in vivo. Both reporters were evaluated when expressing tj RNAi using C587-GAL4 at 25 °C. In the wild type germaria, expression of dpp2.0-lacZ, which faithfully recapitulates dpp expression in germarium 29 , is strongly visible in CpCs (Fig. 4b,b' , open ovals) and occasionally found in 1-3 ECs (about 12% of total control germaria, n = 640, Fig. S4a,a' , white arrowhead) as reported. Remarkably, C587-GAL4 driven expression of tj RNAi caused a more frequent dpp2.0-lacZ expression in ECs (Fig. 4c-d' , white arrows). On the other hand, tj-depleted ECs also showed an ectopic P4-lacZ activity, which is typically expressed in TF and CpCs in wild type germaria (Fig. S4b-e). Here, ECs could be distinguishable from other somatic cells by their position and nuclear morphology. CpCs were identified with round-or disc-shaped nuclei upon DAPI staining. ECs were identified with triangle-shaped or flattened and curved spindle-shaped nuclei upon DAPI staining. It is noteworthy that CpCs adopted a stalk-like or branched organization (Figs 4b-d and S4b-d, white broken ovals and yellow dashed lines) and GSC number decreased (Fig. S4f,g) in gemaria expressing tj RNAi by C587-GAL4, which are similar to previous findings 37 . Thus, the ectopic dpp-lacZ-expressing cells might not be a mislocalization of CpCs. Meanwhile, TFC and CpC number were not affected by C587-mediated tj RNAi, as manifested by Dl-lacZ ( Fig. S5a-b') and hh-lacZ respectively ( Fig. S5c-d'). Furthermore, C587 > Flp-mediated generation of CpC clones during niche formation resulted in similar clone size of CpCs between control and tj PL3 mutant mosaic germaria (Fig. S5e-j"'). These results implicate that tj-depletion does not induce expansion of TFCs or CpCs, and the ectopic dpp-lacZ-expressing cells are ECs that display a characteristic feature of CpCs. ( www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ Then we tested whether excess Dpp signaling outside the GSC niche is responsible for accumulated UGC in tj KD germaria. To this aim, the UGC number was assessed in germaria of reducing dpp expression levels in flies expressing tj RNAi by C587-GAL4. Heterozygotes of either dpp alleles (dpp hr4 or dpp e90 ) in the presence of tj KD exhibited a partial suppression of UGC accumulation. Likewise, C587-mediated double knockdown of dpp and tj led to a significant reduction of the UGC number (Fig. 4e-p). Furthermore, similar results were obtained with a pupal-specific tj dpp double knockdown with C587 ts (Fig. S6a,b). Taken together, our results support the idea that Tj in pre-adult ECs regulates CB differentiation at least partly by restricting dpp expression. The incomplete rescue could be attributed to the incomplete knockdown of dpp or an unidentified mechanism independent of dpp repression. www.nature.com/scientificreports www.nature.com/scientificreports/ En dysregulation in ECs with tj depletion contributes to the dpp expansion in germaria. One recent study confirmed that Engrailed (En), a homeodomain-containing transcription factor, regulates dpp expression in CpCs by binding to a 2.0 kb fragment 5′ enhancer region of dpp 29 . And forced expression of En in ECs causes UGC accumulation as well as ectopic expression of dpp2.0-lacZ 21,29 . These results raise a possibility that the ectopic En could induce UGCs accumulation in tj KD germaria through regulating dpp transcription. To test this hypothesis, we investigated whether reduced expression of tj causes an aberrant change in En expression in germaria. The crosses were performed and the resulting progeny were raised at 25 °C, and newly eclosed females (<1-day old) were dissected for detection. Immunostaining revealed that En is specifically expressed in the nuclei of TF and CpCs in control germaria 21,50,51 (Fig. 5a,a'). In contrast, some ECs in the anterior portion of the germaria exhibited ectopic En expression when tj was knocked down (Fig. 5b-c' , white arrowheads). As expected, we also found the deformed TF and CpC clusters similar to that have been previously reported 37 (Fig. 5b-c' , white broken ovals and yellow dashed lines). And ECs were distinguishable from other somatic cells by their position and nuclear morphology, as stated above. Similarly, ectopic En expression in ECs was caused by pupal-specific, but not adult-specific tj KD (Fig. S7a-c' , white arrows and data not shown). To link the ectopic En expression with the UGC accumulation elicited by tj KD, the effect of heterozygous en allele or UAS-en RNAi lines on the tj RNAi differentiation phenotype was examined. Phonotypical analysis showed that compromised en expression significantly suppressed germline differentiation defects induced by tj RNAi (Fig. 5d-m). Similarly, pupal-specific tj en double knockdown with C587 ts significantly repressed the tj KD UGC phenotype (Fig. S6a,b). These results indicate that Tj acts in pre-adult ECs for germ cell differentiation control partially through negative regulation of En.
Given that En regulates dpp transcription in the GSC niche, we reasoned that ectopic En expression induced by tj KD might upregulate dpp transcription. To verify this assumption, we tested whether attenuation of en expression in tj KD germaria reduces the ectopic dpp expression. As expected, heterozygous en allele or en RNAi partially suppressed elevated dpp expression in tj KD germaria, as indicated by a significant decline in the number of EC positive for dpp2.0-lacZ (Fig. 6a-h). On the contrary, the ectopic En staining in tj KD ECs remained statistically unchanged when heterozygous dpp allele or UAS-dpp RNAi lines were present (Fig. 7a-h, white arrow heads). These results support the idea that Tj functions in ECs to control GSC progeny differentiation partly by En-regulated restriction of dpp expression. www.nature.com/scientificreports www.nature.com/scientificreports/

Tj regulates EC protrusions cell-autonomously in an En-independent context. Abundant studies
have demonstrated that EC protrusions-mediated interactions with germ cells are important for proper germline differentiation 10,12,21,[23][24][25]46 . Given that Tj participates in the interaction between ICs and PGCs 33 , we further determined if C587-mediated tj depletion impairs EC protrusions. To test this, UAS-mCD8-GFP combined with C587-GAL4 was used to visualize the EC protrusions. The crosses were set up and the progeny were maintained at 25 °C and newly eclosed females (<1-day old) were dissected for detection. In control germaria, EC protrusions were detected to encapsulate the germ cells outside the GSC niche (Fig. 8a, white arrows). Interestingly, some of the ECs lost protrusions and left germ cells unwrapped in tj KD flies (Fig. 8b,c, white arrowheads). Similar results were obtained for pupal-specific tj KD with C587 ts . As visualized by anti-Coracle staining, ECs in germaria from newly  (a,a'), C587 > tj RNAi-1 + en RNAi-1 (b,b') and C587 > tj RNAi-1 + en RNAi-2 (c,c') groups exhibit relieved expression of dpp2.0-lacZ in ECs, compared with Fig. 4(c,c'). The C587 > tj RNAi-2; en 4 /+ (d,d'), C587 > tj RNAi-2 + en RNAi-1 (e,e') and C587 > tj RNAi-2 + en RNAi-2 (f,f ') groups exhibit relieved expression of dpp2.0-lacZ in ECs, compared with Fig. 4(d,d'). (g,h) Graphs show the percentage of germaria that contain a given number of EC with ectopic dpp2.0-lacZ expression of indicated genotypes. ( www.nature.com/scientificreports www.nature.com/scientificreports/ eclosed females failed to penetrate into the germ cells (Fig. S8a-c'). However, no evidence of significant impairment in EC protrusions was found in C587 ts -mediated adult-specific tj KD germaria. As the vast majority (about 99%) of the germaria exhibited normal EC protrusions (Fig. S8d-f '). These results are consistent with Tj mainly functioning in the pre-adult stage to prevent the formation of supernumerary UGCs. In addition, the MARCM analysis displayed small protrusions of tj clonal ECs (Fig. S8g-i, white arrows), which is consistent with the results from previous studies that wild type neighbors stabilized mutant EC extensions to encapsulate germ cells 15 . Thus, we conclude that Tj is required for proper EC protrusions, helping to facilitate germline differentiation.
To further support our conclusion, we took an advantage of hs-bam transgene, in which bam expression is under the control of the heat-shock inducible promoter, to induce germline differentiation in tj KD germaria. The heat shock was conducted at the pupal stage (see Methods) and the newly eclosed females (<1-day old) were dissected for detection. Heat-shock induced GSC differentiation resulted in differentiating cysts, instead of GSCs, occupied the tip of germaria in control groups, whereas the EC protrusions remained extended and wrapped the germ cells underneath (Fig. 8d, white arrows). Similarly, C587 > tj RNAi germaria were occupied with multiple cell-cysts under  + (a,a'), C587 > tj RNAi-1; dpp hr4 /+ (b,b'), C587 > tj RNAi-1 + dpp RNAi-1 (c,c'), C587 > tj RNAi-2; dpp e90 /+ (d,d'), C587 > tj RNAi-2; dpp hr4 /+ (e,e') and C587 > tj RNAi-2 + dpp RNAi-1 (f,f ') groups exhibit ectopic En expression in ECs, as indicated by white arrowheads. www.nature.com/scientificreports www.nature.com/scientificreports/ heat-shock induced bam expression. However, no restoration of EC protrusions was observed (Fig. 8e,f, white  arrowheads). In short, forced differentiation of UGC was not able to suppress the EC morphology defect induced by tj-depletion (Fig. 8m). Altogether, we conclude a cell-autonomous contribution of Tj in regulating EC protrusions.
To test whether Tj regulates EC protrusions through En, en RNAi and tj RNAi were coexpressed using C587-GAL4 at 25 °C. No restoration of EC morphology defects was observed following co-expression of en RNAi in newly eclosed (<1-day old) females, suggesting that En is dispensable for tj-mediated regulation of EC protrusions ( Fig. 8g-l,n). www.nature.com/scientificreports www.nature.com/scientificreports/

Tj regulates the establishment of a proper EC number in an En-independent context. Previous
studies have also shown that UGC accumulation caused by extrinsic EC defects is always coupled with severe loss of ECs 10,[24][25][26]28,46 , indicating that a certain number of EC is required for non-cell autonomous control of germline differentiation. To determine whether Tj controls EC number, the enhancer trap line PZ1444 52,53 , which labels both CpCs and ECs, was used to directly quantify the EC number in germaria. The crosses were set up at 25 °C and newly eclosed females (<1-day old) were dissected for detection. In germaria of control group (Fig. 9a), PZ1444 is expressed higher in CpCs than in ECs. ECs also differ from CpCs by location and the nuclear morphology. In contrast, C587-mediated tj KD led to a significant reduction of β-gal positive ECs (Fig. 9b-d). Furthermore, the lacZ expression was dampened in ECs upon tj-depletion. Pupal-specific tj KD with C587 ts induced consistent phenotype (Fig. S8j). Surprisingly, no significant reduction in EC number was detected in the adult-specific regimens for C587 ts -mediated tj KD (Fig. S8k). Together with observations on the small EC clone size for tj mutant (Figs 1g,h and S8g-i), these pieces of evidence lead to a conclusion that Tj is required in pre-adult stage for the establishment of a proper EC number, which may contribute to proper GSC progeny differentiation.
To test whether Tj regulates EC cell number through En, en RNAi and tj RNAi were coexpressed using C587-GAL4 at 25 °C. No rescue of EC number was observed following co-expression of en RNAi in newly eclosed (<1-day old) females, suggesting that En is dispensable for tj-mediated regulation of EC number (Fig. 9e-j). www.nature.com/scientificreports www.nature.com/scientificreports/

Discussion
Results from earlier studies 33,37 and ours have shown that Tj absence as induced by a strong loss-of-function allele has deleterious effects on ovariole formation and maintenance. Here we generate a hypomorphic condition of Tj knockdown using C587-GAL4-mediated tj RNAi, and show that Tj is required in pre-adult ECs for CB differentiation control. In our hands, the GSC progeny differentiation process was not disrupted upon hh-mediated tj KD, though it could be due to an incomplete Tj depletion in CpCs. However, our molecular and genetic studies reveal that tj KD drives UGC formation, possibly through an En-dependent mechanism of repressing dpp transcription outside the GSC niche. Tj also intrinsically maintains EC protrusions and EC number in pre-adult stage independent of En. Therefore, we propose that Tj functions to safeguard against the acquisition of GSC niche cell characteristics by immature ECs and promote pre-adult EC to adopt the identities of differentiation niche cells.
Strikingly, UGCs forming upon tj depletion express bamP-GFP, which is similar to some previous studies 10,12,54 . Furthermore, bamP-GFP-positive UGCs are considered as CB tumors, and in a recent study they are believed to have been already escaped niche signal 15 . Remarkably, our data support the idea that the accumulated CB-like cell is probably resulting from an ectopic BMP activity sustained at a certain level outside the GSC niche, as evident by the supernumerary pMad-negative dad-lacZ-positive UGCs in C587-mediated tj RNAi germaria. Meanwhile, the elevated dpp transcription in tj-depleted ECs was also confirmed with two distinct enhancer traps, dpp2.0-lacZ and P4-lacZ. As a consequence of ectopic BMP activity, the amount of Bam protein failed to reach the critical level for 2-cell cyst formation. Taken together with the observation that hs-bam rescued tj differentiation phenotype, we suggest a direct link between CB-like cell accumulation and disreugulated BMP activity.
What motivates the ectopic BMP activity outside the GSC niche? Several correlative studies suggest complex mechanisms, in which a population of ECs is capable of expressing Dpp as well as other niche cell markers under certain conditions 20,21,25,26,28 . Together, they reveal that tightly regulated dpp transcription is required to maintain homeostasis of adult ovaries. Our study highlights the role of Tj required for modulating dpp transcription activity partly through repressing En expression during ovarian development. This suggests that, in wild-type developing ovaries, the enhancer element dpp2.0 is maintained in an inactive state in most of the ECs by Tj-mediated repression of En, an activator of the enhancer element 29 . How disrupted Tj expression leads to upregulation of En remains an open question. Meanwhile, how the enhancer element P4 participate in the Tj-involved regulation of dpp transcription remains to be uncovered. Thus, further studies investigating the regulatory network of dpp transcription would be of great interest.
Our observation that tj-depleted ECs exhibit a dual identity, simultaneously harboring features of CpCs and ECs, particularly recalls the previous observation that Tj functions in CpCs for TF fate repression 37 . Thus, it is reasonable that different expression levels of Tj 37 can be instructive for the cellular differentiation status of TF, CpCs and ECs. Meanwhile, these observations also explain why the immunostaining of En in C587 > tj RNAi CpCs or tj PL3 clonal CpCs (Fig. S5j-j"') appeared to be visibly unchanged. (It is still possible that the upregulated En expression has been below our detection threshold.) Due to adoption of TFC fate, tj-depleted CpCs exhibit comparable En staining to TFCs, in which En staining is normally indistinguishable from CpCs. And here, we consider Tj as a factor that regulates the balance between stem cell and differentiation niches so as to coordinate the behavior or GSC and GSC progeny during development.
On the other hand, slightly extra UGCs without inappropriate En expression were uncovered by MARCM analysis, probably owing to the small clone size of tj mutant. Accordingly, we argue that Tj helps to establish a certain number of ECs. Consistently, PZ1444 reporter analysis with pupal-specific tj KD gave the similar conclusion. Meanwhile, we found altered clonal EC morphology in the MARCM analysis ( Fig. S8g-i, white arrows). Notably, RNAi based pupal-specific depletion of Tj caused a severe loss of EC protrusions. These results are consistent with the previous study that the presence of wild-type ECs is sufficient to partially rescue the phenotype induced by woc or stat mutant ECs 15 . The tight association between germline and ECs is indispensable for proper germline differentiation 10,[23][24][25]46 . Tj has been reported to participate in the interaction between PGCs and ICs 33,37 . Here we further identify that Tj is critical for protrusions of ECs in a cell-autonomous manner. As EC morphology was not regained when germline was forced to differentiate in tj KD germaria. Together with the observation that tj phenotype of EC number and cellular protrusion were not rescued upon tj en double knock down, our data strongly suggest that Tj controls germline differentiation via multiple types of mechanism, including the ones dependent and independent of En.
Besides, Tj appears to function primarily during the developmental stage rather than adulthood. As the En-dependent and independent roles of Tj in CB differentiation control are mainly played during the pupal stage. In this context, a better understanding of hormones, developmental cues and chromatin accessibility is essential for understanding the normal roles of Tj in development.
www.nature.com/scientificreports www.nature.com/scientificreports/ Larval and pupal staging. Staging of larvae and pupae was performed as reported 40,41,56 . In short, synchronized eggs were collected in a fresh vial within 2 hours. After the parents were removed, the offspring were cultured with optimal nutrition and uncrowded conditions. For flies without tub-Gal80 ts , the vials were kept at 25 °C all the time. Under these conditions, the middle of larval third instar (ML3) is at 96 h after egg laying (AEL), the late of larval third instar (LL3) is between 114-116 h AEL, and larval-to-pupal transition (LL3 to white pre-pupa) is at 120 h AEL. The early pupal (EP) is at 144 h AEL. For flies that carry tub-Gal80 ts , vials were kept at 18 °C until temperature shift. Under these conditions, the LL3 is 9 days AEL, and the EP is 12 days AEL.
Heat shock and clone generation. MARCM clone of tj mutant and control were generated by crossing FRT40A, tj PL3 or FRT 40 A with y, w, hs-Flp, UAS-mCD8-GFP; tubP-GAL80, FRT 40A; tubP-GAL4. The resulting larvae were heat shocked at 37 °C with three 60 min pulses at 5 h intervals each day since ML3 to EP. After heat shock, the flies were kept at 29 °C until dissection. Newly born (<1-day old) females were dissected and stained with the appropriate antibodies.
Cap cell clones were generated by crossing FRT40A, tj PL3 or FRT 40A with C587-GAL4, UAS-Flp; FRT40A GFP at 25 °C. The resulting larvae were kept at 29 °C until eclosion. Newly born (<1-day old) females were dissected and stained with the appropriate antibodies.
Flies carrying hs-bam were heated shocked at 37 °C for 60 min at the 8 th day AEL. Newly born (<1-day old) females were dissected and stained with the appropriate antibodies.
Antibodies and immunofluorescence. The staining of ovaries or larvae gonads or pupae gonads was carried out as described previously 57,58 . In brief, females were dissected in PBS and fixed in 4% paraformaldehyde (Sigma) in PBS for 30 min, rinsed with 0.3% PBST [PBS containing 0.3% Triton X-100 (Bio-Rad)] three times, permeated with 1.0% PBST for 1 h, blocked with 10% goat serum (Life Technology) in 0.3% PBST for 2 h, and stained overnight at 4 °C with primary antibodies. Ovaries or gonads were then washed three times in 0.3% PBST and incubated with secondary antibodies for 2 h at room temperature and stained with DAPI for 10 minutes. Finally, ovaries were rinsed four times with 0.3% PBST and mounted.
The RNA extraction and qRT-PCR. RNA was isolated from newly eclosed female ovaries of tj-GAL4 > mcherry RNAi, tj-GAL4 > tj RNAi-1 and tj-GAL4 > tj RNAi-2. And rp49 was served as normalization control. To exclude the interference of the vitellarium region of ovariole, we chose the female eclosed within 2 h for dissection. RNA was isolated using TRIzol (Invitrogen) and then subjected to reverse transcription. ReverTra Ace ® qPCR RT Master Mix with gDNA Remover (TOYOBO) was used for cDNA synthesis according to the manufactures' instructions. Quantitative real-time PCR was performed on Bio-Rad CFX96 PCR system by using SYBR green qPCR Master Mix (TOYOBO).
The primers used for amplifying dpp and rp49 mRNA are as follows: dpp forward: AGCCGATGAAGAAGCTCTACG dpp reverse: ATGTCGTAGACAAGCACCTGGTA rp49 forward: TCCTACCAGCTTCAAGATGAC rp49 reverse: CACGTTGTGCACCAGGAACT Relative concentration was determined using the 2 −ΔΔCt method in Bio-Rad CFX Manager Software Ver 3.0.

Statistical analysis.
All statistical data were recorded in Excel (Microsoft) and graphed in Prism 7.0 (GraphPad Software). The Chi-square (and Fisher's exact) test and the two-tailed Student's t-test were used to calculate the p-values. Data Availability. All data generated or analyzed during this study are included in this published article (and its Supplementary Information files).