Maintenance of niche architecture requires actomyosin and enables proper stem cell signaling and oriented division in the Drosophila testis

Stem cells are essential in supporting and repairing the tissues in which they reside, and are often maintained by a niche, a specialized microenvironment. The niche is important for sending cues that allow stem cells to maintain their identity and regulate the balance between self-renewing and differentiating outcomes. Imbalance of these cues can lead to tumor formation or tissue degeneration (Ohlstein et al., 2004; Morrison and Kimble, 2006; Morrison and Spradling, 2008). Therefore, the proper function of the niche is required for tissue homeostasis.

Stem cell niches across organisms and tissues often adopt a reproducible shape, structure, or organization. For example, the mammalian intestinal epithelium organizes its niche into the base of crypts. This affords a protected environment within which stem cells are maintained, while newly differentiating cells travel out of the crypt and replace cells during intestinal lining turnover (Barker et al., 2007; Sumigray et al., 2018). However, what remains less understood is whether that precise crypt organization is essential for stem cell function. Recent work in intestinal organoids revealed a requirement for crypt invagination and its precise diameter to promote stem cell function (Pentinmikko et al., 2022). While it is speculated that specific niche architectures are crucial to regulate signaling and stem cell behaviors, the mechanisms controlling niche structure and the effects of disruption on stem cell dynamics remain unknown.

The Drosophila testis is an ideal system to study the potential role of niche structure on stem cell function. The identity of the stem cells and niche cells are known, and their respective behaviors can be directly visualized within the tissue through live imaging. The testis niche is composed of a compact group of quiescent somatic cells that reside at the apex of the testis tube (Hardy et al., 1979). The niche signals to two stem cell populations, the germline stem cells (GSCs) and the cyst stem cells (CySCs) (Kiger et al., 2001; Tulina and Matunis, 2001; Kawase et al., 2004; Leatherman and DiNardo, 2008; Leatherman and DiNardo, 2010; De Cuevas and Matunis, 2011; Amoyel et al., 2013; Inaba et al., 2015b). Signals from the niche exhibit spatially restricted effects, with pathway activation reaching only the first tier of cells adjacent to the niche and thus maintaining a restricted stem cell pool. For example, niche secretion of the ligand Unpaired (Upd) activates the Jak/STAT pathway only in nearby cells (Kiger et al., 2001; Tulina and Matunis, 2001). This contained signaling is essential for stem cell maintenance and attachment at the niche and thus allows differentiation of germ cells once displaced from the niche.

In many tissues, the placement of the niche imparts an overall directionality to differentiating cells as they move away from the niche. This is the case in the Drosophila testis because the stem cell divisions are invariably oriented such that differentiating daughters are displaced away from the niche (Yamashita et al., 2003, 2007). The mechanism controlling this is most understood for GSCs, where an intrinsic regulatory network controlling division plane has been identified. Multiple factors, including E-Cadherin, Bazooka (Par-3) and APC, are enriched in the GSC cortex adjacent to niche cells, where they anchor one of the centrosomes (Yamashita et al., 2003; Inaba et al., 2010, 2015a). Centrosome anchoring orients the eventual mitotic spindle so that division occurs orthogonally, generating one daughter cell remaining attached to the niche and one daughter pushed away. Such orthogonal divisions appear to be crucial in GSCs, as a centrosome orientation checkpoint (COC) exists, and is employed specifically in the GSCs to monitor centrosome anchoring (Venkei and Yamashita, 2015). If centrosomes are misoriented, the COC arrests the GSC cell cycle in G2 until centrosomes are properly positioned (Cheng et al., 2008; Venkei and Yamashita, 2015). Currently, only GSC-intrinsic factors regulating the COC have been identified, including Centrosomin and Par-1 (Inaba et al., 2010; Yuan et al., 2012). However, the precise mechanisms controlling centrosome anchoring and how the GSC cortex adjacent to the niche cell must be specialized to sense proper anchoring are not well understood.

Although little is known about how niche architecture is maintained in the adult testis, substantial work has been done to understand how the niche adopts this architecture during its initial formation. The structure of the steady-state niche is initially formed in the embryo where a subset of somatic gonadal cells are specified to become niche cells, migrate and assemble loosely at the anterior of the gonad, and then undergo compaction to achieve a spherical shape (Le Bras and Van Doren, 2006; Anllo et al., 2019; Anllo and DiNardo, 2022; Warder et al., 2024). Work from our lab has shown that actomyosin contractility (AMC) is required during the compaction process to shape the niche. Inhibition of AMC or the AMC regulator Rho Kinase (ROK) in the gonad disrupts compaction and ultimately generates defective niche architecture. In parallel work analyzing snRNA-seq data from the steady-state testis, we found that ROK was predicted to be more enriched in the niche compared to other cells in the testis (Li et al., 2022; Raz et al., 2023). This finding, in conjunction with our lab's work on gonadal niche assembly, suggested a role for AMC beyond formation to include maintenance of niche structure.

Previous analyses disrupting niche structure also altered niche cell and/or fate. For example, knockdown of retinoblastoma (Rb; also known as Rbf) in the niche allowed for huge changes in niche size and structure, but also resulted in niche cells exiting quiescence and growing in number (Greenspan and Matunis, 2018). Additionally, ablation of CySCs can induce niche cells to exit quiescence and can result in ectopic niche formation after recovery from CySC ablation (Hétié et al., 2014). Exiting the normally quiescent state, changes to niche cell number, and/or alterations in fate, all make it difficult to assess how niche architecture might contribute to niche function in governing stem cell behavior.

Here, we describe a requirement for AMC in maintaining niche structure in the adult testis niche. We show that reduction in AMC causes changes to the polarized localization of adhesion proteins between niche cells, concomitant with progressive disassembly of the niche. These changes in niche structure occur without increased niche cell numbers, loss of niche quiescence or altered niche cell identity. Thus, we could now ask whether maintenance of niche structure is necessary for proper stem cell regulation in the absence of confounding factors. Importantly, loss of proper niche structure disrupts restriction of Jak/STAT signaling, resulting in increased numbers of STAT⁺ germ cells, most with decreased activity of the pathway. Lastly, we have identified the first GSC-extrinsic control of the COC, as we find that disrupted niche structure prevents proper execution of oriented divisions in GSCs. Taken together, we elucidate mechanisms required for long-term maintenance of niche structure and identify the crucial importance of niche architecture in the proper regulation of stem cells.

Although significant work has been done studying niche formation in the embryo, much less is known about maintenance of niche structure in the adult tissue. To address this, we first examined the niche in three dimensions (3D) to determine if niche architecture remains similar to the spheroid structure established during embryogenesis (Warder et al., 2024). Outlining niche cell membranes using either Fasciclin III (FasIII) or E-Cadherin (Fig. 1A-B′; key reagents, Table S1) revealed niche structure to indeed approximate a spheroid (Fig. 1C, see Materials and Methods), similar to that observed upon niche formation and consistent with the classical description of this niche based on electron microscopy (Hardy et al., 1979).

Our lab recently showed that AMC is crucial in achieving this spheroidal shape during niche formation (Warder et al., 2024). To test whether contractility remained important in preserving niche structure, we examined the distribution of a regulator of contractility, ROK, that was predicted to be upregulated in niche cells by single nuclear (sn) RNA-seq data (Raz et al., 2023). A ROK::GFP transgenic reporter exhibited enriched signal at the niche compared to other cells in the testis (Fig. 1D-D″). In addition, by quantifying ROK::GFP fluorescence intensity we found that ROK was polarized to the niche–GSC interface as opposed to the niche–niche interface (Fig. 1G). As ROK governs contractility through its action on Myosin II (MyoII), we next asked whether MyoII was similarly enriched in niche cells. We used fluorescently tagged transgenic lines for either MyoII heavy chain (HC; Zipper; Fig. 1E-E″) or MyoII regulatory light chain (RLC; Spaghetti Squash; Fig. 1F-F″) and found that both were similarly polarized to the niche–GSC interface (Fig. 1H,I). Therefore, MyoII and a key regulator of contractility are polarized at the niche–GSC interface in the adult testis, precisely located to play a functional role in maintenance of niche structure.

The enrichment of Sqh-GFP to the niche interface suggested that Myosin contractility was present, given that Sqh represents the Myosin RLC (Aranjuez et al., 2016; Bertet et al., 2004). In fact, live-imaging of sqh-GFP revealed the presence of dynamic puncta, that appeared and disappeared over short time intervals, suggesting pulses of activation along the niche–stem cell interfaces (Fig. 2A-B‴). To corroborate that active actomyosin contractility was important for niche structure, we compromised ROK by temporally inducing expression of an RNA interference (RNAi) transgene directed against ROK using a niche-specific driver, Upd-GAL4, regulated by Gal80ts. We confirmed specificity and temporal induction of the system by expressing a UAS-GFP under GAL4-restrictive and GAL4-permissive temperatures and observed GFP expression localized to the niche only upon upshift to the restrictive temperature for Gal80ts (Fig. S1). For all RNAi treatments, flies were initially grown without knockdown until late pupal stages, allowing normal assembly and function of the niche. Flies were then upshifted to activate GAL4 and initiate knockdown by expression of the target RNAi, along with the UAS-GFP, and assayed after 10 days. After ROK knockdown, niche structure was misshapen in about 45% of niches (Fig. 2C-F). A misshapen niche qualifies as a niche that has lost its spheroid structure to adopt more elongate shapes. To quantify the degree of misshaping, we measured niche sphericity, area, and volume in 3D. Niches compromised for ROK exhibited decreased sphericity and increased area and volume in comparison to control niches (Fig. 2G,J,K). Importantly, using a nuclear marker, we quantified niche cell number and found no difference between wild-type (WT) and ROK knockdown niches (Fig. 2H). By contrast, the average internuclear distance among niche cells was significantly increased upon ROK knockdown, suggesting that increased niche area and volume was due to dispersion of niche cells (Fig. 2I; see Materials and Methods). As these niche cells still retained niche-specific FasIII and continued to express Upd-GAL4, these data strongly suggest that, while ROK activity is not necessary for maintenance of niche cell identity, it is essential to maintain proper niche shape.

As a main role for ROK is control of MyoII contractility, and given the enrichment of both at niche–GSC interfaces, we next asked whether MyoII is similarly important for preservation of niche shape. We expressed an RNAi against MyoII HC and quantification of MyoII fluorescence intensity (Zipper) confirmed that protein levels were significantly reduced by RNAi expression in niche cells (Fig. 3J-L). We found that niches depleted of MyoII were severely misshapen compared to controls (Fig. 3A-C′, Fig. S3). Notably, about one-third of these niches were split into separate aggregates (Fig. 3D). Loss of MyoII consistently led to decreased niche sphericity and increased area and volume in comparison to WT niches (Fig. 3E,H,I). Consistent with our results from ROK knockdowns, loss of MyoII in the niche did not alter niche cell number but resulted in greater dispersal of niche cells compared to controls (Fig. 3F,G). Importantly, cells comprising the altered niches remained quiescent, as none were positive after a pulse with the S-phase substrate EdU (Fig. S2). These data together reveal a requirement for MyoII in maintaining the spherical organization of the niche. Furthermore, the similarity in phenotype between ROK and MyoII inhibition suggests an essential role for AMC in niche shape.

The strikingly separated niche fragments upon MyoII challenge suggested the possibility that, over time, constituent niche cells were drifting apart. To test this, we examined niches in which MyoII was compromised for a shorter period. These niches were still misshapen, though not as severely as in 10 day knockdowns including no disassembled niche fragments (Fig. 4A-C′). This was confirmed by sphericity measurements, where the 3 day knockdowns were significantly less spherical than WT, but not as misshapen as with 10 days of MyoII inhibition (Fig. 4D). This suggests a progressive loss of niche structure the longer MyoII is compromised.

To determine the mechanism by which MyoII inhibition disrupts niche structure, we performed live imaging to directly visualize niche cell behavior. Testes expressing MyoII RNAi in niche cells for 10 days were imaged ex vivo for a 24 h period. We found that control niches generally maintained their compact shape throughout the imaging period, with the niche cells in the z-plane shown remaining in a similar plane throughout imaging. Although membranes of individual niche cells were not always differentiated, there were no obvious changes in cell contacts to other niche cells or GSCs throughout imaging (Fig. 4E-F‴). In contrast, niches compromised for MyoII progressively changed shape throughout imaging beyond their initial misshaped appearance. For example, niche cells in the MyoII knockdown appear to extend outward, potentially exchanging a previous contact with a neighbor niche cell, as well as making a new contact by surrounding a germline cell (Fig. 4G-H‴). Additionally, niche cells can dynamically extend and retract protrusions in between germ cells (Fig. S4, yellow arrows, white arrowheads). Together, this shows a progressive phenotype where niche cells compromised for MyoII are more loosely organized, and even explore surfaces of other cells.

In niches compromised for MyoII activity, the decrease in sphericity, increase in cell dispersion, and the behaviors visible by live-imaging all suggest that individual niche cells in the collective are not as tightly adherent, and can separate from one another. Elegant serial electron microscopy (EM) reconstruction has shown that cells of the niche are extensively interdigitated with each other (Hardy et al., 1979). Subsequent analyses have revealed enrichment of proteins at niche–niche interfaces falling into several different adhesion classes, including E-Cadherin and N-Cadherin (Brower et al., 1981; Le Bras and Van Doren, 2006; Michel et al., 2011). We therefore asked whether loss of MyoII might be compromising the enrichment or polarity of these factors. In control testes, both E-Cadherin and N-Cadherin exhibited enrichment at the niche as expected (Fig. 5A-A″,C-C″). Both proteins significantly polarized to the niche–niche compared to niche–GSC interfaces (Fig. 5E,G). In contrast, when MyoII was compromised, the polarity of E-Cadherin and N-Cadherin to niche–niche interfaces was significantly reduced, thought to a lesser degree for N-Cadherin (Fig. 5B-B″,D-D″,F,H).

To test whether compromising E-Cadherin was sufficient to disrupt niche structure, we expressed an RNAi against E-Cadherin in the niche. Consistent with previous experiments, we observed no disruption in niche structure (Fig. S5; Michel et al., 2011). As a more stringent test for the role of adherens junctions, we expressed an RNAi against β-Catenin (Arm). β-Catenin is likely integral to both E-Cadherin and N-Cadherin linkages and, thus, upon knockdown would compromise all adherens junctions (Oda et al., 1994). Similar to compromising E-Cadherin, we observed no disruption of the niche structure (Fig. S5), even though β-Catenin was significantly reduced. The apparent stability of the niche when adherens junctions are compromised suggests that compromising MyoII affects not only cadherin complexes, but other adhesive complexes that must be acting redundantly among niche cells.

The generation of misshapen and disassembled niches that still retain normal niche cell number allows us to ask how niche architecture might affect stem cell function. One of the key signals emanating from the testis niche, Upd activates the Jak-STAT pathway only locally, in nearby somatic and germline cells. These first-tier cells represent the two stem cell lineages, CySC and GSC, and the restriction of signaling allows the progeny of these stem cells to differentiate properly. To determine whether niche signaling is compromised in misshapen niches, we immunostained for STAT in GSCs, as a readout for efficient Jak-STAT signaling. We observed that GSCs adherent to niches compromised for MyoII activate the Jak/STAT pathway; however, they do so to a lesser degree than controls, suggesting inefficiency in signaling (Fig. 6A-B‴,E). It is known that decreased STAT activation can cause GSC loss from the niche (Brawley and Matunis, 2004; Kiger et al., 2001; Leatherman and DiNardo, 2010; Tulina and Matunis, 2001). We therefore used live imaging to directly track GSC behavior in control and MyoII-compromised niches. We scored loss events as germline cells that were initially adherent to the niche but that subsequently lost contact over the imaging period (see Materials and Methods). We found that there was an increase in GSC loss from niches compromised for MyoII compared to controls (Fig. 6C-D‴,G). Together, these data show that misshapen niches have reduced Jak/STAT signaling and increased GSC loss.

The increase in area for niches compromised for MyoII (Fig. 3H) raised the possibility that these disrupted niches might present more vacancies to attract cells to a stem cell fate. Therefore, we also scored the number of STAT⁺ germ cells that contacted niche cells. We first chose a threshold signal intensity above which a cell would be scored as STAT⁺ (see Materials and Methods). We found that there is a significant increase in the number of STAT⁺ germ cells surrounding those niches compromised for MyoII than control niches (Fig. 6F). These data together show that proper niche shape is important for restricting niche signals to a select population of cells, with disrupted niche structure increasing stem cell number but diminishing their ability to retain contact with the niche.

There are a few ways GSC number might increase at the niche. The first possibility is an increased GSC cell cycling rate. However, an EdU pulse ruled out any change in S-phase frequency in GSCs between MyoII knockdowns and controls (Fig. S2). The other possibilities each involved alterations in GSC divisions. One possibility was an increase in executing a ‘symmetric renewal’ event. Normally, GSCs divide perpendicular to the niche producing one daughter that renews as a stem cell and one gonialblast (Gb) daughter moved farther from the niche that will eventually abscise and differentiate (Yamashita et al., 2003, 2007; Lenhart and DiNardo, 2015; Lenhart et al., 2019). A symmetric renewal can occur when, after a properly oriented division, the Gb daughter rotates inward and contacts the niche (Sheng and Matunis, 2011). The second way GSC number might increase is through a ‘symmetric division’, where the orientation is incorrect and the GSC divides parallel to the niche such that the two daughter cells remain in contact with the niche through and after division. We hypothesized that GSCs adherent to a misshapen niche might execute symmetric renewals and/or symmetric divisions at a higher frequency than in controls.

To test for symmetric renewal frequency, we live imaged GSCs in control testes and testes with depletion of MyoII from the niche and quantified the number of correctly oriented GSC divisions that were followed by contact of the Gb daughter with the niche. We observed an increase in symmetric renewal events at niches compromised for MyoII (Fig. 7A-C‴,F). In these cases, the prospective Gb would appear to rotate in to contact the niche, although in some cases a niche cell appeared to extend outward to contact the daughter cell (Fig. 7C‴, white arrowhead). Therefore, niches compromised for MyoII allow for more cells to respond to niche signaling by an increase in symmetric renewal events. To test for symmetric divisions, we live imaged MyoII knockdown testes compared to controls and tracked each GSC division. While all GSCs in controls exhibited oriented divisions, about a quarter of GSC divisions in MyoII knockdowns were misoriented. These resulted in both daughter cells adhering to the niche (Fig. 7D,F). Taken together, niches compromised for MyoII have less efficient STAT activation and significant deviation from proper stem cell division control.

GSCs orient their divisions, anchoring one centrosome at the niche–GSC interface, resulting in a spindle orientation that pushes the distal daughter cell away from the niche. A GSC-specific checkpoint prevents M-phase entry if centrosomes are misoriented and COC is only released when a centrosome is re-captured properly (Cheng et al., 2008; Venkei and Yamashita, 2015). Because we observed symmetric divisions in MyoII knockdown niches, we hypothesized that this checkpoint was being bypassed in GSCs. We first tested whether there was centrosome misorientation using γ-tubulin as a centrosome marker. Indeed, we observed an increase in misorientation in GSCs with MyoII knockdown niches compared to aged-matched controls (Fig. 8). In some cases, a GSC was observed nestled between separated niche cells. In these GSCs, the two centrosomes abutted a niche interface. It is likely that division of such a GSC would result in both daughter cells adherent to a niche cell, producing two stem cell daughters rather than one stem cell and one differentiating daughter (Fig. 8A-D).

The centrosome misorientation suggested that the COC might be defective in GSCs adherent to MyoII-compromised niches. The established assay for COC activation takes advantage of differences between typical cells and GSCs in terms of when the cell cycle stalls following forced microtubule disassembly. Treatment with colcemid induces microtubule depolymerization, which eliminates cytoplasmic asters necessary for centrosome anchoring, as well as mitotic spindles (Venkei and Yamashita, 2015). In most cycling cells, the absence of a bipolar spindle will cause arrest in M phase due to the spindle assembly checkpoint present in all cells (Musacchio and Salmon, 2007). This M-phase arrest is detected as an increased proportion of cells containing phosphorylated histone H3 (PH3). Under the same conditions, testis GSCs uniquely arrest before M phase due to loss of centrosome anchorage and activation of the COC, which occurs in G2 before the spindle assembly checkpoint (Venkei and Yamashita, 2015). Thus, loss of COC function can be identified as a failure of GSCs to arrest in G2 and instead now arresting in M, indicated by an increased proportion of GSCs positive for PH3.

To assess the status of the COC, we treated control testes and testes with MyoII-compromised niches with colcemid and quantified the GSC M-phase index. In controls, we found no significant difference in M-phase index between vehicle and colcemid treatments, as expected for GSCs with an active COC (Fig. 8E-F″,I). However, with MyoII-compromised niches there was a significant increase in M-phase-positive GSCs after colcemid treatment (Fig. 8G-I). This demonstrates that the COC is not functional in these GSCs. Taken together, these data suggest that MyoII-compromised disruptions to niche structure interfere with the establishment or activation of the COC in the adjacent germ cells.

We have shown that proper structure of the testis niche is required for its proper function. Disrupting the normally organized niche cell cluster leads to changes in signaling to stem cells, decreased retention of GSCs at the niche, and key changes in their tightly orchestrated division orientation program. Each of these alterations would be detrimental to robust spermatogenesis.

AMC has been shown to induce cellular movements and migrations that can ultimately lead to specific organizations of cells. This is the case for the organization and/or placement of niches within tissues. In the formation of the intestinal crypt, contractility is required for its invagination (Sumigray et al., 2018). Our lab has recently shown that the early shaping of the male gonadal niche is dependent on AMC (Warder et al., 2024). Intriguingly, the continued enrichment of MyoII at the niche–stem cell interface in the adult testis coupled to the enrichment of ROK predicted by adult testis snRNA-seq data suggested that contractility would play a key role in also maintaining niche architecture (Li et al., 2022; Raz et al., 2023).

We found that both ROK and MyoII are polarized at the niche–GSC interface. Reducing MyoII in niche cells led to decrease of MyoII signal, suggesting that most of the MyoII contribution was by niche cells and not the germ cell cortex. The polarization of AMC components is initially established in the early embryo during compaction, where it generates tensile forces at the niche–GSC boundary (Warder et al., 2024). The ROK and MyoII enrichment at this boundary in the adult suggests that this tension remains at the niche–GSC interface to maintain the niche in a spherical organization. Indeed, we showed that knockdown of MyoII within niche cells severely disrupts niche structure. Importantly, loss of germ cells does not significantly alter niche architecture (Brawley and Matunis, 2004). Together, this suggests that preservation of AMC at the niche–GSC boundary from embryogenesis through adulthood is essential to maintain niche structure and that this polarization is primarily driven by the niche cells.

We show that absence of ROK and MyoII can lead to severe consequences to niche structure that ultimately cause defects in the neighboring GSCs. Since depletion of ROK and the MyoII HC generated similar defects, we conclude that AMC is regulated by ROK in the cortex of cells, specifically along the niche–GSC interface. However, the lower number of niches with misshaping defects and the lesser degree to which niches are misshapen suggests other regulators also act in maintaining AMC at the niche. It is also possible that incomplete knockdown of ROK by RNAi might account for the difference in severity of the effect. In any case, our observations raise questions about how contractility factors are enriched in niche cells, how they are polarized to the correct interface with stem cells, and what the specific effects of cortical contractility might be that, when compromised, disrupt niche cell function, and generate substantial defects in stem cell behavior.

A potential source for candidate factors acting upstream to regulate AMC is the snRNA-seq dataset of genes, the expression of which is enriched in niche cells. It is also possible that extrinsic signaling could help maintain niche shape. Interestingly, the initial architecture of this niche is shaped by feedback from recruited GSCs (Warder et al., 2024). Some of that feedback appears to be in the form of mechanical forces generated by GSC divisions (Warder et al., 2024). In the adult testis, the ablation of germ cells does not appear to alter niche shape (Brawley and Matunis, 2004), suggesting that, if extrinsic factors or forces exist, these cannot emanate from GSCs. Manipulations in the CySCs, as well as manipulations in niche cells themselves, have been shown to affect niche architecture. However, the manipulations involved in those studies also caused changes in fate and proliferation of niche cells (Hétié et al., 2014; Greenspan and Matunis, 2018). Thus, it is not clear that CySCs specifically affect the cellular organization of the niche.

We found that the niche progressively loses its shape when compromised for MyoII over time. Live imaging first hinted at this, and afforded a unique perspective to understand niche shape changes, something typically challenging in other systems. Examining the effects of knockdown after different lengths of time revealed that the niche cells spread apart and suggested the possibility that adhesion was progressively compromised. Indeed, we observed changes to cadherin polarization when MyoII was compromised. Normally, the niche polarizes both E- and N-Cadherin to the niche–niche interface as opposed to the niche–GSC interface. Interestingly, this polarization is opposite to that observed for AMC factors that are enriched at niche–GSC interfaces. The mechanism underlying this polarity opposition for these niche cells is not yet clear. However, this is a common phenomenon in other polarized tissues, such as epithelia (Simões et al., 2010; Bulgakova et al., 2013). There, a cascade involving microtubules, Par3 and RHO-GEF help maintain the enrichment of E-Cadherin to interfaces opposite to those with high contractility (Bulgakova et al., 2013). When these factors are compromised, epithelial organization is also compromised. Given that we observe niche cells drifting apart it will be of interest to test whether a similar cascade affects niche cell polarization.

It is likely that MyoII and contractility in niche cells affects more than cadherin polarization. Knocking down E-Cadherin, or even β-Catenin, which likely challenges both E- and N-Cadherin complexes, did not cause defects in niche shape (Michel et al., 2011). Indeed, a survey of the literature and the snRNA-seq data reveals a large number of junctional, adhesive and transmembrane factors that are enriched on these niche cells, such as Echinoid, FasIII, Discs Large, and Polychaetoid (Raz et al., 2023). This suggests significant redundancy to the maintenance of niche organization. It will be of interest to pursue precisely which complexes are affected, and how, when AMC is compromised.

We observed that changes to niche structure had two effects on Jak/STAT signaling. First, we found that GSCs respond less robustly to Jak/STAT signaling when MyoII is compromised in the niche. It is known that GSCs unable to activate STAT properly are lost from the niche and we confirmed an increase in GSC loss when niche structure was disrupted (Kiger et al., 2001; Tulina and Matunis, 2001; Brawley and Matunis, 2004; Leatherman and DiNardo, 2010). Thus, the decreased GSC responsiveness to Jak/STAT signaling upon disruption of niche structure directly impacts a crucial feature of GSC function (persistent adhesion to the niche).

Second, we found that there was an increase in the number of STAT-responding germ cells at MyoII-compromised niches. One hypothesis is that the area increase of the niche allows the signal to disperse more, which could afford more cells access to the signal, albeit each to a lesser degree. However, this cannot be confirmed without directly observing secreted ligand levels upon disruption of niche structure. Alternatively, there is a possibility that the activation of STAT itself is being disrupted upon loss of niche structure. There is evidence that STAT can be polarized to specific interfaces within the cell, and this polarization is required for proper STAT activation (Sotillos et al., 2008). Further studies could address whether STAT activation occurs at the interface in this system, and whether that is disrupted upon the misshaping or disassembly of the niche.

We found that more germ cells responded to STAT signaling when the niches were compromised for MyoII. It has been shown that there is competition for space at the niche, so it is not surprising that cells quickly fill a space that might be generated by increases to niche area and volume (Issigonis et al., 2009; Sheng et al., 2009). In principle, while dedifferentiation of gonial cells could account for such an increase, this is a relatively rare event and was not obvious in our live imaging. Instead, our imaging illuminated two behaviors in GSCs attached to compromised niches that could account for their increase in number at the niche. First, we observed GSCs that underwent a properly oriented division to push the Gb away from the niche. However, the still-attached Gb instead contacted the niche as opposed to remaining pushed away from the niche. This type of ‘symmetric renewal’ is normally a rare event but was significantly increased in MyoII-compromised niches (Fig. 7E). In some cases, we observed niche cells that appeared to extend outward to contact the Gb cell (Fig. 7C‴). Whether this reflects some attraction or is purely a stochastic consequence of the aberrant niche shape remains to be investigated. Second, we observed a higher frequency of misoriented GSC divisions (discussed below). Such ‘symmetric divisions’ do not occur in GSCs normally. Thus, when niche organization is disrupted, both symmetric renewal and symmetric division occurs, leading to Gb daughter cells that are meant for differentiation potentially maintaining a stem cell fate instead.

In various tissues, stem cells divide in an oriented manner by precisely positioning one or both centrosomes so that the proper orientation of the eventual spindle is established. For example, in Drosophila neuroblasts, centrosome asters are regulated such that one stabilizes at the apical cortex and the other travels to the opposite pole, due to its loss of microtubule stability (Rusan and Peifer, 2007). In the case of the Drosophila testis, one centrosome in the GSC anchors at the niche–GSC interface to ensure a properly oriented division. Several factors have been identified that all act intrinsically in the GSC to anchor the centrosome. For example, it has been shown that there are polarity proteins like Bazooka and pericentriolar proteins like centrosomin that localize to the niche–GSC interface within the GSC and are required for proper centrosome orientation (Yamashita et al., 2003; Inaba et al., 2010, 2015a). The oriented division is an important feature in this tissue, and as such GSCs uniquely engage a centrosome orientation checkpoint (COC), which will cause arrest in G2 if the centrosomes are not properly oriented for division (Cheng et al., 2008; Yuan et al., 2012; Venkei and Yamashita, 2015). Significantly, in disrupted niches we observed misoriented centrosomes, and a failure of the checkpoint resulting in misoriented GSC divisions. This is the first description of a niche-dependent role in centrosome orientation and the subsequent execution of the COC in GSCs. We hypothesize that the cytoskeletal connections among MyoII, actin and adherens (or other junctional proteins) in the niche cell cortex must be what link to junctional factors within the GSC cortex to ensure proper centrosome orientation. Taken together, this work emphasizes how AMC-dependent niche structure is crucial for proper stem cell behavior and ultimately necessary for maintaining the proper balance of stem cells within a tissue.

Although our work shows preservation of niche structure through AMC, we were unable to distinguish the precise mechanism by which niche AMC maintains this structure. Additionally, our work shows a decrease in STAT response in GSCs with misshapen niches, however we could not directly assay ligand levels or distribution. Lastly, we determined that the COC is bypassed occasionally in GSCs with misshapen niches, however we could not precisely determine the mechanism by which the COC is inactivated.

Virgin females with Upd-GAL4 tubGal80ts and a UAS-GFP construct were crossed with males of either the MyoII HC, ROK, E-Cadherin, or Armadillo RNAi, depending on the experiment. These crosses were kept at 18°C until late pupation and were then upshifted to 29°C to induce expression of the RNAi and GFP. Males were then dissected after 10 days at 29°C unless otherwise indicated. For MyoII HC RNAi, E-Cadherin RNAi, and Armadillo RNAi experiments, controls were aged matched and dissected at the same time as knockdown, and did not include the Gal4 driver unless otherwise indicated. For ROK RNAi experiments, controls were kept at 18°C for 10 days after late pupation and had the same genotype as ROK knockdowns.

Testes were dissected in ringers and fixed in 4% formaldehyde in Buffer B (16.7 mM KPO₄, pH 6.8; 75 mM KCl; 25 mM NaCl; 3.3 mM MgCl₂) and 0.1% Triton for 20 min. They were then washed in PBS plus Triton for 20 min and then blocked in 4% normal donkey serum for 1 h. Primary antibodies were diluted in block and added overnight at 4°C. Secondary antibodies were diluted in block and used at 3-4 µg/ml for 1 h at room temperature (Alexa488, Cy3, orAlexa647; Molecular Probes and Jackson ImmunoResearch). Hoechst at 0.2 µg/ml was used to stain DNA for 3 min. Testes were then equilibrated in 50/50 Ringers solution (5 mM HEPES, pH 7.3; 130 mM NaCl; 5 mM KCl; 2 mM MgCl₂; 2 mM CaCl₂) and glycerol and mounted in 2% n-propyl-gallate, 80% glycerol.

We used chick antibody against GFP (Aves labs, GFP-1020, 1:3000), rabbit antibody against RFP (Abcam, ab62341, 1:500), Vasa (gift from R. Lehmann, MIT, USA, 1:5000), STAT92E (gift from E. Bach, NYU, USA, 1:375), MyoII HC (zipper) (gift from D. Kiehart, Duke University, 1:750), Phosphohistone H3 (PH3) (Upstate Biotech, 06-570, 1:1000); mouse antibody against FasIII (DSHB, 7G10, 1:50), Gamma Tubulin (Sigma-Aldrich, GTU-88, 1:200), β-Catenin (armadillo) (DSHB, N2 7A1, 1:100), rat antibody against DE-Cadherin (DSHB, DCAD2, 1:20), DN-Cadherin (DSHB, Ex#8, 1:50). Images of fixed testes were acquired using a Zeiss Imager with an apotome using a 40×1.2 N.A. water immersion lens.

To quantify niche shape defects, we stained with FasIII and evaluated in 3D whether each niche was normal, misshapen or disassembled using Imaris software. Normal niches were scored as such if the FasIII staining resembled a compact, relatively spherical shape. Misshapen niches were scored as such if they were intact niches that had lost shape or were less compact, typically adopting elongate shapes. Disassembled niches were scored as such if they were fragmented. This type of categorization was used for MyoII knockdown and ROK knockdown experiments, with Chi-squared tests used to evaluate comparisons.

Imaris software was also used to extract sphericity, area, and volume of the niche. To extract these values we used FasIII labeling of the niche to draw a surface that encompasses the whole niche. The surface was then generated into a mask and further refined by drawing a more accurate surface on the new mask that encompasses only the niche. This second surface was used to extract sphericity, area, and volume calculations. For all sphericity measurements, FasIII was used except for when E-Cadherin was used to draw surfaces for WT in Fig. 1. Data were imported into Prism v10.0 to generate scatterplots where each dot represents a testis. To quantify niche cell number and average nuclear distance from one niche cell to its three nearest neighbors, we used Hoechst to label niche cell nuclei that were only within the confines of the finalized niche surface. This allowed us to accurately quantify niche cell number and the average distance between them. Data were imported into Prism v10.0 to generate scatterplots, where each dot represents a niche cell nuclei. Mann–Whitney tests were used to evaluate comparisons.

Live Imaging was carried out as in Lenhart and Dinardo (2015). Control and MyoII RNAi samples were dissected after 10 days at 29°C and mounted on a MatTek dish. Imaging media was carefully added, which consisted of 15% fetal bovine serum (Gibco, 10082) and 0.5× penicillin/streptomycin (Corning, 30-002-Cl) in Schneider's Insect Media (Gibco, 21720-024) (Prasad et al., 2007; Sheng and Matunis, 2011; Lenhart and DiNardo, 2015). The stock imaging media was prepared ahead of time, kept at 4°C, and used within 2 weeks. Bovine insulin (Sigma-Aldrich, l0516) was added to an aliquot of imaging media, just before use, to a final concentration of 0.2 mg/ml. Sqh-GFP-labeled MyoII RLC to assess Myosin II activity. Live imaging was carried out for 1 h using 60 s intervals and a z-step size of 0.5 µM; stack depth was 40 µM to include the whole niche. Upd>GFP labeled the niche and nanos-lifeact::tdTomato (Lin et al., 2020) labeled germ cells to assess niche changes and germ cell behaviors, respectively, throughout imaging. Live imaging was carried out overnight using 30 min intervals, and z-step sizes of 0.5 µM; stack depth was 40 µM to include the whole niche. Testes were imaged on an Olympus iX83 microscope with a Yokagawa CSU-10 spinning disk scan head, 60×1.4 N.A. silicon oil immersion objective, and Hamamatsu EM-CCD camera.

To quantify polarity of either ROK and MyoII between niche–niche interfaces and niche–GSC interfaces we examined transgenic lines for ROK (ROK-GFP), MyoII HC (zip::GFP) and MyoII RLC (sqh3xGFP). We identified niche–niche interfaces by FasIII staining and identified niche–GSC interfaces by comparing FasIII staining to germ cell markers, which consisted of either a vasa stain or using the nanos-lifeact::tdTomato transgenic line. To characterize the polarity of either ROK or MyoII at the niche–GSC interface we used ImageJ to draw a 5 pixel-wide line along three niche–GSC interfaces and extracted the average fluorescence intensity. These values were normalized to the average fluorescence intensity along three niche–niche interfaces, after background subtraction.

To quantify polarity of either E-Cadherin and N-Cadherin between niche–niche interfaces and niche–GSC interfaces we analyzed testes stained for either E-Cadherin or N-Cadherin, and FasIII. We identified niche–niche interfaces by FasIII staining and identified niche–GSC interfaces by comparing FasIII staining to the germ cell marker, nanos-lifeact::tdTomato transgenic line. Using ImageJ, we measured the average fluorescence intensity along three niche–niche interfaces and normalized them to the average fluorescence intensity along three niche–GSC interfaces, after background subtraction, to quantify the polarity of E-Cadherin and N-Cadherin at the niche–niche interface. We repeated these measurements in MyoII 10 day knockdown niches to determine whether there was a change in polarity of E-Cadherin or N-Cadherin in niche cells. Data were imported into Prism v10.0 to generate scatter plots, with each dot representing an interface. Mann–Whitney tests were used to evaluate all comparisons.

To quantify the level of MyoII knockdown, Myosin II knockdown and control testes were stained with an antibody against MyoII HC. Using ImageJ, we chose representative z-slices and drew a region of interest (ROI) that encompassed the niche, identified by FasIII staining. This same ROI was moved further along the testis sample, to a position among the germ cells, for normalization. After background subtracting, the sum of fluorescence within the niche ROI was normalized to the sum of fluorescence of the germ cell ROI. This was done for each testis, and those values were imported into Prism v10.0 and scatterplotted for MyoII knockdowns and controls, where each dot represents a testis. A Mann–Whitney test was used to evaluate the comparison.

To quantify the level of STAT accumulation, MyoII knockdown and control testes were stained with an antibody against STAT. Controls and knockdowns are of the same genotype, with controls remaining at 18°C (inactive GAL4) for 10 days. Dissections and immunostaining of controls and knockdowns were carried out simultaneously. Using ImageJ, ROIs were taken from all GSCs to extract mean fluorescence intensity from each cell. A cell was scored as a GSC if it was contacting the niche, as determined by using nanos-lifeact::tdTomato for the germ cells and FasIII for the niche. Within each testis, three additional germ cells (non-GSCs) were measured and averaged together after background subtraction to yield a ‘germ cell mean intensity’ value. Each GSC measurement within that testis was normalized to that germ cell mean fluorescence intensity. Data were imported into Prism v10.0 to generate scatterplots, where each dot represents a GSC. A Mann–Whitney test was used to evaluate the comparison.

To determine whether a germ cell adherent to the niche was STAT⁺, we quantified the mean fluorescence intensity from each cell that was adherent to the niche as described above. We determined the mean STAT enrichment for all cells measured for both the controls and MyoII knockdowns. We established individual thresholds for the control group as well as the MyoII knockdown group. We determined the threshold to be one standard deviation below the relative mean fluorescence intensity of controls. For controls, the relative mean fluorescence intensity was 2.61, with a standard deviation of 0.83, therefore any cells with a mean fluorescence intensity above 1.78 was counted as STAT⁺. For MyoII knockdowns, the relative mean fluorescence intensity was 1.99, with a standard deviation of 1.00, therefore any cell with a mean fluorescence intensity above 0.99 was counted as STAT⁺. Data were imported into Prism v10.0 to generate scatterplots where each dot represents a testis. A Mann–Whitney test was used to evaluate the comparison.

To analyze GSC behavior, each GSC was tracked from the start of imaging (t=0) until the end of the 24 h imaging session. GSCs were scored for division orientation, symmetric renewal, and loss. GSCs that divided and retained one daughter cell at the niche and the other pushed away from the niche were scored as properly oriented. GSCs that divided such that both daughter cells remained in contact with the niche for multiple time points after division were scored as misoriented. The frequency of misoriented divisions relative to all divisions was plotted comparing MyoII knockdown niches and controls. A Fisher's exact test was used to evaluate the comparison.

GSCs that had oriented divisions were simultaneously tracked for whether the GSC-Gb pair exhibited a symmetric renewal event. The criteria for such would mean that a newly produced Gb initially not touching a niche cell would ultimately rotate in and make contact with the niche and remain in contact for multiple time points. The frequency of symmetric renewal events out of every oriented division scored was plotted between MyoII knockdown niches and controls. A Fisher's exact test was used to evaluate the comparison.

GSCs that were lost from the niche were scored as those cells initially touching a niche cell but ultimately losing contact and drifting away after multiple time points. The frequency of GSC loss out of all GSCs scored was plotted between MyoII knockdown niches and controls. A Fisher's exact test was used to evaluate the comparison.

To determine whether centrosomes were properly oriented in MyoII knockdowns compared to WT testes, we stained for centrosomes using an antibody against γ-tubulin. We used FasIII staining to visualize the niche and used nanos-lifeact::tdTomato to visualize germ cells. GSCs were scored as having oriented centrosomes if they had two visible centrosomes with one touching the niche–GSC interface within the GSC. GSCs were scored as having misoriented centrosomes if they had two visible centrosomes, neither of which were touching the niche–GSC interface. GSCs were scored as having doubly oriented centrosomes if they had two visible centrosomes where both were touching a niche–GSC interface. A Chi-squared test was used to evaluate the comparison.

To quantify frequency of S-phase entry in GSCs and in niche cells, EdU pulses were carried out using the Click-iT EdU plus kit (Thermo Fisher Scientific, C10419) (Salic and Mitchison, 2008). Immediately after dissection in Ringers, testes were incubated in 10 µM of EdU in Ringers for 30 min at room temperature before fixing. The azide reaction to couple EdU to Alexa 647 was carried out after fixing, but before the remainder of the staining protocol. GSCs positive for EdU were scored and the frequency of EdU⁺ GSCs out of all GSCs was plotted per testis between MyoII knockdown niches and controls. Data were imported into Prism v10.0 to generate scatterplots, where each dot represents a testis. A Mann–Whitney test was used to evaluate the comparison. Niches were also screened for any EdU⁺ niche cells, but they were not observed.

To determine whether the COC was active, we assayed for its activity using the protocol of Venkei and Yamashita (2015). Immediately after dissection in Ringers, testes were transferred into imaging media plus insulin, containing either DMSO-only as a control, or DMSO+10 mM colcemid, for a final concentration of 100 µM. The samples were incubated for 4.5 h at room temperature. Samples were fixed as usual and stained with a rabbit antibody against PH3 to score for cells in mitosis. GSCs positive for PH3 were scored and the frequency of PH3⁺ GSCs out of all GSCs was plotted for MyoII knockdown niches and controls between colcemid-treated and DMSO-only conditions. Data were imported into Prism v10.0 to generate scatterplots, where each dot represents a testis. A Kruskal–Wallis test with Dunn comparisons was used to evaluate all comparisons.

We thank the Bloomington Drosophila Stock Center (NIH P40OD018537), Vienna Drosophila Research Center, R. Lehmann, E. Bach, Y. Bellaïche, D. Kiehart, and Zallen for antibodies and stocks. We thank the director of our CDB core for use of Imaris workstations. Thank you to V. Desikan for initial experimental set up. Thanks to B. Warder, K. Nelson and the Lenhart Lab for input throughout this project. Special thanks to Kari Lenhart for her manuscript comments and her microscope, without which our live imaging experiments would not be possible.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.204498.reviewer-comments.pdf