Epigenetic regulator Lid maintains germline stem cells through regulating JAK-STAT signaling pathway activity

ABSTRACT Signaling pathways and epigenetic mechanisms have both been shown to play essential roles in regulating stem cell activity. While the role of either mechanism in this regulation is well established in multiple stem cell lineages, how the two mechanisms interact to regulate stem cell activity is not as well understood. Here we report that in the Drosophila testis, an H3K4me3-specific histone demethylase encoded by little imaginal discs (lid) maintains germline stem cell (GSC) mitotic index and prevents GSC premature differentiation. Lid is required in germ cells for proper expression of the Stat92E transcription factor, the downstream effector of the Janus kinase signal transducer and activator of transcription (JAK-STAT) signaling pathway. Our findings support a germ cell autonomous role for the JAK-STAT pathway in maintaining GSCs and place Lid as an upstream regulator of this pathway. Our study provides new insights into the biological functions of a histone demethylase in vivo and sheds light on the interaction between epigenetic mechanisms and signaling pathways in regulating stem cell activities.


INTRODUCTION
Extrinsic signals from cells comprising the stem cell niche are essential in maintaining stem cell activity (Morrison and Spradling, 2008). In addition, epigenetic regulation that changes stem cell chromatin structure without altering DNA sequences acts as an important intrinsic mechanism to maintain stem cells (Eun et al., 2010). Although both mechanisms are important to regulate stem cell activities, our understanding of the crosstalk between the two events is limited to a few examples (Cherry and Matunis, 2010;Eliazer et al., 2011;Tarayrah et al., 2013).
Drosophila spermatogenesis is a paradigmatic system to investigate the molecular mechanisms responsible for the maintenance of adult stem cell activities in their physiological environment (Kiger et al., 2001;Tulina and Matunis, 2001;Yamashita et al., 2003Yamashita et al., , 2007. The Drosophila male germline stem cell (GSC) niche is one of the best characterized niches in which GSCs associate with two types of somatic cells: hub cells located at the tip of the testis, and cyst stem cells (CySCs) two of which surround each GSC (Fig. 1A). Hub cells and CySCs contribute to a niche that provides the critical signaling necessary to preserve GSC identity and activity (Kiger et al., 2001;DiNardo, 2008, 2010;Tulina and Matunis, 2001). Janus kinase signal transducer and activator of transcription (JAK-STAT) and bone morphogenetic protein (BMP) signaling pathways are the two major pathways involved in the maintenance of the male GSC niche. Activation of the JAK-STAT pathway is initiated by the secretion of the cytokine Unpaired (Upd) from the hub cells. Upd binds the receptor Domeless activating Hopscotch (Hop), the JAK kinase homolog in Drosophila, and Stat92E, the STAT homolog, in both GSCs and CySCs (reviewed in Hou et al., 2002). The intrinsic activation of Stat92E in CySCs activates BMP signaling and is thought to be sufficient to cause continuous GSC self-renewal while Stat92E in the GSCs has been reported to only regulate their adhesion to the niche DiNardo, 2008, 2010).
In addition to signaling pathways, the chromatin structure of GSCs can profoundly influence critical decisions of stem cell maintenance versus differentiation. In Drosophila, histone modifying enzymes and chromatin remodeling factors are important chromatin regulators (Becker and Horz, 2002). Both have been shown to act cell autonomously to maintain GSCs in the testis niche (Buszczak et al., 2009;Cherry and Matunis, 2010;Eliazer et al., 2014). Among the histone modifying enzymes, histone demethylases have been identified as 'epigenetic erasers' that remove methyl-groups from methylated Lysine residues of histones (Klose et al., 2006). Specifically, the repressive trimethylation on lysine 27 of histone H3 (H3K27me3) and the active H3K4me3 histone modifications have been the focus of many studies. While H3K27me3 is laid down by a member of the Polycomb Group (PcG) proteins and associates with silent gene regions (Cao et al., 2002;Czermin et al., 2002;Kuzmichev et al., 2002;Muller et al., 2002), H3K4me3 is generated by the Trithorax Group (TrxG) family of proteins and has been shown to associate with active regions of chromatin (Byrd and Shearn, 2003). A recent study from our lab on the function of dUTX, a histone demethylase that targets the repressive H3K27me3 modification, revealed that dUTX regulates JAK-STAT pathway activity in the CySCs to maintain proper niche structure and gene expression (Tarayrah et al., 2013).
Among the 14 demethylases in Drosophila (Klose et al., 2006), little imaginal discs (lid) encodes a demethylase that specifically removes the active H3K4me3 modification (Eissenberg et al., 2007;Lee et al., 2007). Lid belongs to the JARID1 family of H3K4me3 demethylases . While human cells encode four JARID1 family members (JARID1a, JARID1b, JARID1c, and JARID1d), Lid is the sole Drosophila homolog (Eissenberg et al., 2007). Previous studies have shown that lid mutant adult flies have increased global H3K4me3 levels (Eissenberg et al., 2007;Lee et al., 2007) and demonstrated a role for Lid in dMyc-induced cell growth (Li et al., 2010). However, the function of Lid in an endogenous stem cell system has not yet been elucidated. The mammalian Lid homologue JARID1a was reported to play a critical role in breast cancer metastatic progression, suggesting a role for H3K4me3 demethylases in inhibition of tumor progression and metastasis (Cao et al., 2014). Therefore, understanding the functions of Lid in an adult stem cell system might facilitate the targeting of histone demethylases for cancer treatment. Here we report that Lid is required cell autonomously to prevent premature differentiation of male GSCs by maintaining Stat92E levels. Our findings support a cell autonomous role for the JAK-STAT pathway in maintaining GSCs and provide insight into the in vivo functions of a histone demethylase.
To explore the expression pattern of Lid in testis, we performed immunostaining using anti-Lid antibody  in wt testes. The Lid protein was detected in the nuclei of cells throughout the testis with strong signals in the germline ( Fig. S1C-C″). To determine the role of lid in the male GSC niche, we analyzed testes isolated from third instar larvae of lid males. We detected a marked decrease of GSC number in lid testes compared to wt testes. The decrease in GSC number was visible in immunostained images when comparing wt testes ( Fig. 1B) with lid testes (Fig. 1C, dots). When quantified, we found that wt testes had an average of 9.7±1.6 GSCs compared to an average of 6.03±1.4 GSCs in lid testes (P<0.01; Fig. 1D). Furthermore, GSCs appeared to be detached from the hub cells in the majority of lid testes (Fig. 1C, arrow). These results suggest that Lid is required to maintain GSCs at the Drosophila testis niche.
Due to adult lethality we next analyzed the lid mutant phenotype in adult testes using the FLP-mediated FRT recombination system (Xu and Rubin, 1993). We generated mutant clones for lid using the strong loss-of-function allele lid k6801 (Gildea et al., 2000;. The percentage of testes with at least one lid GSC clone declined over time compared to the wt control ( Fig. 1E), suggesting that Lid is required in the germline to maintain GSCs at the Drosophila testis niche.
To confirm this conclusion, we used different cell-type specific Gal4 drivers in combination with a UAS-lid small hairpin microRNA (shmiRNA) (Ni et al., 2011) to knockdown lid in a cell type-specific manner. Knockdown of lid exclusively in early stage germ cells using nanos (nos)-Gal4 (Van Doren et al., 1998) led to a significant decrease in GSC number ( Fig. 1F-G, dots; 1H). By contrast, knockdown of lid using the early cyst cell driver c587-Gal4 (Manseau et al., 1997) (Fig. 1H and Fig. S2A-B) or the hub driver upd-Gal4 (Boyle et al., 2007) (Fig. 1H and Fig. S2C-D) did not lead to a change in GSC number. These results demonstrate that normal function of Lid is required in early stage germ cells including GSCs, but not in somatic gonadal cells, to prevent GSC loss at the testis niche.
In addition to its role in GSC maintenance, we also found that Lid is required for proper hub architecture (compare hub region in Fig. 1B and C). Loss of lid led to a dramatic increase in hub size compared to wt testes (Fig. S3A). The increase in hub size is a secondary effect due to GSC loss, as reported previously (DiNardo et al., 2011;Gonczy and DiNardo, 1996;Monk et al., 2010;Tazuke et al., 2002). Consistently, this phenotype was recapitulated by knocking down lid in germ cells but not in somatic gonadal cells (Fig. S3B).

Lid is required to maintain GSC proliferation and prevent premature differentiation
To determine the mechanism leading to GSC loss in the lid testes, we used phospho-Histone H3 (PH3) immunostaining to assess the mitotic index of GSCs. We observed a significant decrease in the mitotic activity of GSCs in lid testes compared to wt control ( Fig. 2A, first two columns), suggesting a role for Lid in maintaining GSC proliferation. While the GSC mitotic index in wt third instar larvae is comparable to what was reported previously (Parrott et al., 2012), it is worth noting that it is higher than what was reported for GSCs from wt adult flies (Sheng and Matunis, 2011;Yadlapalli et al., 2011;Yadlapalli and Yamashita, 2013). This is likely because GSCs undergo an expansion at the third instar larval developmental stage. Indeed when we counted H3T3P-positive GSCs in adult testes from nos-Gal4 control, we detected a lower mitotic index at 2.5% (15/607 total GSCs). In comparison, GSCs in nos>lid shmiRNA testes showed an approximate ∼2-fold reduction at 1.4% (8/579 total GSCs). We further confirmed this reduction by using another mitosis-enriched H3S10P antibody, which revealed ∼3-fold reduction of H3S10P-positive GSC ratio in nos>lid shmiRNA testes (3.2%, 9/283 total GSCs) compared to the nos-Gal4 control (9.3%, 30/324 total GSCs). Lid is required to maintain GSC proliferation and prevent premature differentiation. (A) Percentage of PH3 positive GSCs in testes from males of the following genotypes: wt (7.4%); lid (2.7%); lid, UAS-DE-Cadherin; nos-Gal4 (1.7%); and lid, UAS-Stat92E; nos-Gal4 (6.8%). P-value calculated using Fisher's test. (B,C) Immunostaining using antibodies against Vasa (red), Arm (green), and α-spectrin (green) in wt (B) and lid (C) testes. Arrow points to round spectrosome in (B). 4-cell spermatogonia cyst at the hub is outlined (yellow dotted line) in (C). Hub area is outlined (white dotted line). (D-E′) Immunostaining with antibodies against Vasa (red), Arm (magenta), and Bam-GFP (green) in Bam-GFP (D,D′) and lid; Bam-GFP (E,E′) testes. Arrows point to Bam-expressing GSCs in (E′). Hub area is outlined (white dotted line). Scale bars: 10 µm. We next asked whether the loss of GSCs is accompanied by premature differentiation. Fusomes are branched cytoplasmic structures, comprised of membrane skeletal proteins such as αand β-spectrin, that connect germ cells in a cyst (Deng and Lin, 1997). When we used antibody against α-spectrin, the fusome appears spherical (called spectrosome) in all GSCs (Fig. 2B, arrow) and becomes branched in further differentiated spermatogonia of wt testes (n=44). By contrast, we observed spermatogonial cysts containing up to four interconnected germ cells in direct association with the hub region (Fig. 2C, yellow dotted outline) in 33% of the lid testes (n=52). This phenomenon could be due to premature differentiation of GSCs or dedifferentiation of spermatogonial cells. It has been reported that dedifferentiated spermatogonial cells tend to have a higher percentage of misoriented centrosomes when homing back to the niche (Cheng et al., 2008). We labeled centrosomes using antibody against γ-tubulin, a major component of the centrosome (Yamashita et al., 2003) and found that the percentage of GSCs with misoriented centrosomes was not significantly different in lid (6.3%, n=59) compared to wt testes (7.1%, n=104) (P>0.05). In addition to misoriented centrosomes, it has been reported that transiently disintegrating fusome remnants are detectable in dedifferentiated spermatogonial cells (Brawley and Matunis, 2004;Cheng et al., 2008). Using antibody against α-Spectrin, we did not observe any disintegrating fusome remnants in GSCs from lid testes (n=52). Taken together, these data suggest that GSCs in lid testes are lost due to decreased proliferation and premature differentiation.
Since 33% of lid testes contain spermatogonial cysts with branched fusomes next to the hub, the counting of GSCs as Vasapositive cells that are directly associated with the hub region in lid testes might be an overestimation. Therefore we re-quantified the number of GSCs in lid testes: we defined GSCs as Vasa-positive, spectrosome-containing cells in direct association with the hub cells. Indeed, we find that using this method, the average number of GSCs is 5.86±1.69 per lid testis (n=52), which is slightly, but not significantly, lower than the 6.03±1.4 GSCs reported in Fig. 1D. These results suggest that the germ cells that are part of differentiating spermatogonial cysts do not contribute significantly to the overall GSC scoring in lid testes.
To further understand the GSC premature differentiation phenotype, we performed immunostaining using a germ cell differentiation marker, Bag of marbles (Bam) (Gonczy et al., 1997), in both wt and lid testes. Bam is normally expressed in fourto-16-cell spermatogonial cells (Gonczy et al., 1997). Consistently, using a Bam-GFP reporter transgene (Chen and McKearin, 2003), we found that the GFP signal is only detectable in differentiating spermatogonial cysts away from the hub in wt testes (n=20; Fig. 2D-D′). By contrast, Bam-GFP-positive cells were detected in direct association with the hub cells in 15% of the lid testes (n=20; Fig. 2E-E′, arrows in 2E′), further indicating that Lid is required to prevent GSCs from undergoing premature differentiation.

Lid acts in germ cells to maintain the proper level of the Stat92E transcription factor
The JAK-STAT pathway is one of the major pathways that maintain stem cell activity and identity in the testis niche. Testes depleted of Stat92E display severe loss of both CySC and GSC populations (Kiger et al., 2001;DiNardo, 2008, 2010;Tulina and Matunis, 2001). In wt testes, Stat92E is highly enriched in GSCs and some of their immediate daughter cells, but rapidly declines in further differentiated cells (Fig. 3A-A′). By contrast, lid testes had no detectable Stat92E signal (Fig. 3B-B′), even though testes from both genotypes were immunostained together, mounted on the same slide and imaged using the same microscope settings.
To confirm the decrease of Stat92E in lid GSCs, we stained testes containing lid GSC clones with antibody against Stat92E. We observed lower Stat92E immunostaining signal in lid GSCs compared to the neighboring control GSCs (Fig. 3C-C″). In addition, we used quantitative reverse transcription PCR (qRT-PCR) to measure the Stat92E mRNA levels. The Stat92E transcript level in lid testes decreased to ∼50% of the level in the wt control when a constitutively expressed RpL32 gene was used as an internal control (Fig. 3D). It is worth noting that halving the level of Stat92E in Stat92E/+ testes did not cause any obvious phenotypes. Therefore it is likely that the decrease of Stat92E transcript is higher than 50% in lid GSCs, but using the whole tissue with mixed cell types and different stages of germ cells to measure Stat92E mRNA levels compromised this effect.
Finally, we compared nos>lid shmiRNA testes with the nos-Gal4 control using the similar experimental strategy as described above. We also found a marked decrease of the Stat92E immunostaining signal at the niche in the nos>lid shmiRNA testes (Fig. 3F-F′) compared to the nos-Gal4 control (Fig. 3E-E′). In summary, these results demonstrate that Lid acts in germ cells to maintain proper expression of the Stat92E transcription factor. Interestingly, the decrease of Stat92E in nos>lid shmiRNA testes is also accompanied by a decrease of Stat92E signals in the hub cells ( Fig. 3F′ vs 3E′), suggesting a potential non-cell autonomous role for Lid in maintaining JAK-STAT signaling in the hub cells.

Lid regulates Stat92E in GSCs to maintain proliferation and prevent premature differentiation
Because Lid is required for proper expression of Stat92E, we tested whether further reducing Stat92E levels could enhance lid mutant phenotype. By removing one copy of Stat92E using a strong loss-offunction allele (Stat92E 06346 ) (Hou et al., 1996) at lid mutant background, we found a significant enhancement of the GSC loss and premature differentiation phenotypes. In wt testes, after GSC asymmetric cell division, the differentiating daughter cell called gonialblast (GB) undergoes four rounds of mitosis and then enters meiosis. During the elongated G2 phase of meiosis, male germ cells grow 25 times in volume as spermatocytes. Spermatocytes are distinguishable from spermatogonia based on their large size and distinct nuclear morphology (White-Cooper et al., 1998). In wt testes, spermatocytes are distant from the niche and never observed in close proximity with the hub cells (Fig. 4A, arrow). In lid mutant testes, although we found differentiated spermatogonial cysts in direct contact with hub cells, we have never observed spermatocytes in direct contact with the hub cells (Fig. 4B, arrow). However, in drastic contrast, we found that spermatocytes are in direct association with the hub cells in 100% of lid; Stat92E/+ testes (Fig. 4C, arrows). In these testes, the earliest stage of germ cells we could detect is the eight-cell spermatogonia (Fig. 4C, yellow dotted outline). These results demonstrate that both the severity and the penetrance of the lid mutant phenotype are enhanced by loss of one copy of Stat92E, suggesting that Lid acts in synergy with Stat92E to prevent GSC premature differentiation. On the other hand, the enhanceable phenotype of lid mutant is consistent with the measurement of Stat92E transcript level in lid mutant testes (Fig. 3D), suggesting a partial loss of Stat92E in lid mutant testes. These results also explain the reason why GSCs mutant for Stat92E using a strong loss-of-function allele (Stat92E 06346 ) are lost at a more rapid rate (Tulina and Matunis, 2001) compared to lid mutant GSCs (Fig. 1E).
To explore whether Stat92E acts downstream of Lid, we next examined whether overexpression of a Stat92E cDNA is sufficient to rescue the lid mutant defects. We found that driving a UAS-Stat92E transgene (Bach et al., 2003) using nos-Gal4 rescued both GSC loss (Fig. 4D,F 3rd vs 2nd column) and premature differentiation phenotype of lid mutant testes: the percentage of testes with spermatogonial cysts next to the hub decreased from 33% (n=52) in lid mutant to 5% (n=40) in lid; nos>UAS-Stat testes. In addition to lid mutant testes, we tested whether nos>UAS-Stat can rescue nos>lid shmiRNA phenotype in which the function of Lid is only compromised in germ cells. We found that nos>UAS-Stat rescued GSC loss in lid knockdown testes (Fig. 4E,F, 6th vs 5th column). As a control for both experiments, we drove UAS-Stat92E expression using the same nos-Gal4 at a wt background and did not observe an increase in GSC number. This result is consistent with the previous report that increasing Stat92E level in the germline using an activated allele of jak (Hop Tuml ) does not lead to any ectopic phenotype (Fig. 4F, last column) (Leatherman and DiNardo, 2008). In summary, these results demonstrate that Stat92E acts downstream of Lid in maintaining GSC proliferation and preventing their premature differentiation.
Based on our data on Lid's function, we hypothesize that Lid is required for normal expression of Stat92E, which is required for GSC maintenance. Then loss of Stat92E in the germline should phenocopy the lid mutant phenotype. To test this hypothesis more directly, we compromised Stat92E expression by driving a UAS-Stat92E shmiRNA exclusively in early germ cells using the same nos-Gal4 driver. A previous study reported that knockdown of germline Stat92E using RNA interference (RNAi) resulted in testes with displaced GSCs that clustered next to CySCs instead of associating with the hub (Leatherman and DiNardo, 2010). While the detachment of GSCs was not observed in the nos-Gal4 control (Fig. S4A-A′), we did observe a similar phenotype in 30% of nos>UAS-Stat92E shmiRNA testes (n=33; Fig. S4B-B′). In addition, we found that knockdown of Stat92E specifically in the germline led to phenotypes similar to lid mutant testes in which differentiated spermatogonial cysts directly contact the hub cells in 48% of testes (n=33; Fig. 4H, arrow). By contrast, only spectrosome-containing GSCs were located next to the hub in all nos-Gal4 control testes (n=30; Fig. 4G, arrow). The premature differentiation phenotype is more penetrant in UAS-Stat92E shmiRNA testes (48%) compared to that in lid mutant testes (33%), consistent with the partial loss of Stat92E phenotype in lid mutant testes as discussed previously. Taken together, these results suggest that both Lid and Stat92E act in the germline to prevent GSCs from undergoing premature differentiation.

Ptp61F and DE-Cadherin interact with Lid to maintain GSC number
A previous study reported that Stat92E regulates adhesion of GSCs to the hub by maintaining DE-Cadherin levels (Leatherman and DiNardo, 2010). The increased detachment of GSCs in lid testes (Fig. 1C, arrow) could therefore be due to decreased DE-Cadherin because of compromised Stat92E (Fig. 4). Indeed, we find that driving wild-type DE-Cadherin (UAS-DE-Cad DEFL ) (Inaba et al., 2010) using nos-Gal4 rescued GSC loss in lid testes (Fig. 5A-B). However, the rescue of GSC number was not accompanied by restoring GSC mitotic activity measured by anti-PH3 immunostaining ( Fig. 2A, 3rd column). While Lid could control DE-Cadherin levels directly or indirectly by regulating Stat92E levels, our data suggest that lid mutant GSCs have more defects than decreased DE-Cadherin levels. By contrast, both the number and the mitotic index of GSCs in lid; nos>UAS-Stat testes were restored to the level in wt testes ( Fig. 2A, 4th column). Therefore our results demonstrate that Lid acts in synergy with Stat92E to maintain GSCs by regulating their proper proliferation.

DISCUSSION
In this study, we identify a new epigenetic regulator of the JAK-STAT signaling pathway in the Drosophila testis niche (Fig. 5G). Lid acts in the germ cells to regulate Stat92E levels, maintain GSC proliferation and prevent premature differentiation of GSCs. Our results demonstrate an essential role for a histone demethylase in regulating a major signaling pathway to maintain stem cell activity in vivo.
Lid is a new epigenetic regulator of the JAK-STAT signaling pathway Mechanisms responsible for the maintenance of stem cell activities are vital as disruption in these mechanisms is implicated in many types of disease including cancer. The JAK-STAT pathway is one of these mechanisms that play a critical role in stem cell maintenance across a range of species. Abnormalities in this pathway have been reported to lead to tumorigenesis in mammals (Bowman et al., 2000;Leeman et al., 2006). Here we report that Lid maintains GSCs at the Drosophila testis niche by regulating JAK-STAT pathway activity. Loss of lid leads to decreased Stat92E levels and premature GSC differentiation. Our data provide a novel epigenetic mechanism that regulates the JAK-STAT signaling pathway in an endogenous stem cell niche. It will be imperative to determine whether these links between epigenetic and transcriptional regulation occur in other stem cell systems.
The JAK-STAT pathway is required cell autonomously for GSC maintenance The JAK-STAT pathway in the Drosophila testis niche is a major pathway required for both GSCs and CySCs activities. Several lines of evidence in our study support a direct role of Stat92E in maintaining GSC identity and activity. First, germline-specific knockdown of Stat92E leads to premature differentiation of GSCs. Second, expressing a Stat92E cDNA in germ cells rescues GSC loss, decreased mitotic activity, and premature differentiation phenotypes in lid testes. Finally, expression of DE-Cadherin in lid GSCs was insufficient to restore GSC mitotic index. Therefore we hypothesize that STAT activity is required in GSCs to maintain GSC number and proliferation and prevent premature differentiation.

Distinct biological functions of histone demethylases
In this study, we report an important cell autonomous function of H3K4me3-specific histone demethylase Lid in male GSC maintenance in Drosophila, which is different from the H3K27me3-specific histone demethylase dUTX reported to have a non-cell autonomous role in the germline (Tarayrah et al., 2013). While both demethylases are required to regulate stem cell activities, Lid mainly regulates GSC while dUTX regulates CySC activities at the niche. Interestingly, both Lid and dUTX maintain the activities of their target stem cell population by regulating the JAK-STAT pathway. However, while dUTX controls the transcription of Socs36E, an inhibitor of JAK-STAT in CySCs, Lid targets another JAK-STAT inhibitor Ptp61F in the germline.
Interestingly, we did observe a comparable role for Lid in CySCs. Using an antibody against zinc finger homeodomain 1 (Zfh1) to label CySCs and early cyst cells, we detected niche architectural defects in lid testes. In wt testes, Zfh1-expressing CySCs surround GSCs and extend thin protrusions toward the hub forming a clear rosette structure (Fig. S5A-A′, arrows). However, in lid testes, Zfh1expressing cells had nuclei that directly contacted the hub surface ( Fig. S5B-B′, white arrows). Furthermore, knockdown of lid using the cyst cell driver c587-Gal4, but not the germ cell driver nos-Gal4 or the hub cell driver upd-Gal4, led to the overpopulation of Zfh1expressing cells around the hub (Fig. S5C-D, compare white arrows). This suggests that Lid is required in CySCs and/or early cyst cells to prevent overpopulation of Zfh1-expressing cells around the hub.
Our results also revealed that 61% (n=76) of lid testes had hub cells that ectopically expressed Zfh1 compared to only 14% (n=41) of wt testes exhibiting this phenotype (Fig. S5B-B′, yellow arrows). Interestingly, knockdown of lid using the cyst cell driver c587-Gal4, but not the germ cell driver nos-Gal4 or the hub cell driver upd-Gal4, led to ectopic Zfh1 expression in the hub (Fig. S5D, yellow  arrow; Fig. S5E). These data indicate that the function of Lid in CySCs or early cyst cells is essential to prevent Zfh1 from ectopically turning on in hub cells. These results reveal dynamic communication among the different cell types in the testis niche where CySCs send feedback to hub cells to maintain proper gene expression. We do not however understand the biological relevance of ectopic Zfh1 expression in hub cells. When we drive the UAS-Zfh1 transgene using the upd-Gal4 driver we do not observe any apparent phenotypes in niche architecture and function, therefore the biological consequence of ectopic Zfh1 expression remains unclear.
Isolation of total RNA and quantitative reverse transcription polymerase chain reaction (qRT-PCR) Total RNA was collected and isolated from wild-type (wt) and lid third instar larval testes using TRIzol reagent (Invitrogen, #15596-018) according to the manufacturer's instructions. The yield and quality of the RNA was determined using a NanoDrop spectrometer (NanoDrop Technology, San Diego, CA, USA). Reverse transcription was performed using the RevertAid First Strand cDNA Synthesis Kit (Fermentas, #K1621). RNA Transcript levels were then measured using SYBR Green PCR Master Mix (Fermentas, #K0221) and normalized by RpL32. Sequences of primers used for qRT-PCR were as follows: AAGGTGAGTGATTTGCTGTGCTGC (Stat92E-forward) CAACAAGCGAGCATGAGAATGCCA (Stat92E-reverse) CATGCTGCCCACCGGATTCAAGAAG (RpL32-forward) CTCGTTCTCTTGAGAACGCAGGCGA (RpL32-reverse)

Statistical analysis
Statistical significance was calculated using two-tailed Student's t-test or Fisher's test. P-values are indicated in figures or in figure legends. Error bars indicate standard deviation (s.d.).