mastermind regulates niche ageing independently of the Notch pathway in the Drosophila ovary

Proper stem cell activity in tissues ensures the correct balance between proliferation and differentiation, thus allowing tissue homeostasis and repair. The Drosophila ovary develops well-defined niches that contain on average 2–4 germline stem cells (GSCs), whose maintenance depends on systemic signals and local factors. A known player in the decline of tissue homeostasis is ageing, which correlates with the waning of resident stem cell populations. In Drosophila, ovaries from old females contain fewer GSCs than those from young flies. We isolated niche cells of aged ovaries, performed a transcriptomic analysis and identified mastermind (mam) as a factor for Drosophila ovarian niche functionality during ageing. We show that mam is upregulated in aged niche cells and that we can induce premature GSC loss by overexpressing mam in otherwise young niche cells. High mam levels in niche cells induce reduced Hedgehog amounts, a decrease in cadherin levels and a likely increase in reactive oxygen species, three scenarios known to provoke GSC loss. Mam is a canonical co-activator of the Notch pathway in many Drosophila tissues. However, we present evidence to support a Notch-independent role for mam in the ovarian germline niche.


Introduction
Stem cells are essential for tissue homeostasis and are involved in the regeneration of damaged organs. Stem cells respond to signals of different types and realms of action so that their activity is regulated by environmental factors and by local and systemic signals, including ageing. Indeed, the decline in regenerative capacity characteristic of aged organs is linked to the impaired activity of their hosted stem cells [1]. The fact that ageing impacts tissue-specific stem cells, and thus organ function, demands the identification of the precise molecular mechanisms behind stem cell obsolescence.
The gradual loss of tissue homeostasis during ageing is a consequence of local modifications in the stem cells themselves and in their supporting microenvironment or niche. Thus, ageing stem cells show changes in the expression of genes involved in the preservation of genomic integrity, and in transcriptional and epigenetic regulation; they can display both declined responsiveness to extrinsic signals and impaired adhesion to support cells or extracellular matrix; and they can increase intracellular concentrations of reactive oxygen species (ROS), accumulate DNA damage and show slower proliferation and deficient selfrenewing divisions [2][3][4][5][6][7][8]. Similarly, the support cells that form part of the ageing niches undergo changes that eventually affect their homed populations of tissue-specific stem cells. For example, experiments involving the parabiosis of heterochronic mice have identified molecular and cellular mechanisms behind the age-related dysfunction of muscles that map to the aged niches [9]. Other studies in Drosophila have shown that niche cell numbers can decrease with time, which in turn affects the pool of stem cells hosted within [10]. In addition, ageing niche cells may reduce the production of niche signals and of adhesion molecules and increase ROS contents, all of which compromise the maintenance of stem cell populations [2,6]. Therefore, as a result of ageing, stem cell populations shrink in numbers or become functionally impaired with time.
The Drosophila ovary is composed of functional eggproducing tubes called ovarioles. Each ovariole contains tissue-resident stem cells homed in an experimentally accessible niche that is susceptible to genetic manipulations and microscopic analysis. The coordinated activity of the stem cell populations in the niche fuels the generation of new egg gametes during oogenesis. Located in the germarium, a tapered structure found at the anterior tip of the ovariole, this niche is composed of extracellular matrix and a few mesodermal cell types that offer support to 2-4 germline stem cells (GSCs) and to a larger population of somatic stem cells termed follicle stem cells (FSCs) [11,12]. The GSC niche includes terminal filament cells (TFCs), cap cells (CpCs) and a special group of anterior escort cells (ECs; figure 1a). Both CpCs and the anterior ECs are in contact with GSCs and are necessary to maintain the pool of undifferentiated GSCs through adhesion and signalling events [14][15][16]. GSCs can be recognized by  [13]). In the apical part of the germarium, terminal filament cells (TFCs), cap cells (CpCs) and the escort cells (ECs) that surround the germline stem cells (GSCs) form the germline niche. Also shown are the differentiating cystoblasts (CB) and cysts, the GSC and CB spectrosomes and the cyst fusomes. (b) Quantification of GSCs/germarium (Y-axis, right) and the percentage of GSC reduction (Y-axis, left) in young (one week) compared with old (three, four and six weeks) females. (c) Percentage of germaria from the dataset in (b) containing 0, 1 or 2-4 GSCs. (c 0 -c 000 ) Niches containing two GSCs, one GSC and one CB, or two CBs stained with a DNA dye and with antibodies to Hts (red) and to Vasa (green) to visualize spectrosomes and germline cytoplasm, respectively. Dashed lines indicate CpC clusters. (d ) Pattern of bab1-Gal4 expression in TFCs and CpCs ( purple line, high levels of expression) and in escort cells (lower levels), as shown by the Tau::GFP reporter. Work flow for the transcriptomic study: sorting of bright (green fluorescent protein (GFP) signal, X-axis), alive ( propidium iodide (PI), Y-axis) niche cells from bab1 > tau::GFP ovaries was followed by mRNA pico-profiling and microarray analysis. Transcriptome comparisons of one-week-old cells with three-or fourweek-old samples yielded a number of downregulated or upregulated genes. In this and the following figures, boxes in the graphs indicate the treatment of the flies prior to dissection. In this and the rest of the figures, only p-values of two-tailed t-tests considered statistically significant between control and experimental samples are indicated (*p ≤ 0.05, **p ≤ 0.005, ***p ≤ 0.0005). Numbers in bars represent germaria analysed. Scale bars, 5 µm.
royalsocietypublishing.org/journal/rsob Open Biol. 9: 190127 their position adjacent to the CpCs and by the presence of a membranous structure called the spectrosome in their cytoplasm. Upon mitosis, a GSC can give rise to another GSC and to a differentiating cystoblast (CB), which divides synchronously to produce two-, four-, eight-and 16-cell cysts. Ovarian niche ageing in Drosophila conveys a significant reduction in the number of GSCs hosted in the niche [6,17,18] (figure 1a-c).
To try to identify new genes required in niche cells and involved in GSC depletion with ageing, we embarked upon a transcriptomic analysis of maturing niche support cells in Drosophila and found a discrete number of candidate genes whose transcription was regulated with age. We focused our efforts on the mastermind (mam) gene, a known cofactor involved in the canonical transduction of the Notch signal and in the transcription of target genes [19]. Mam forms a ternary complex with an intracellular processed form of the Notch receptor and Suppressor of Hairless (CSL in humans). The Notch pathway is required for the architecture of the GSC niche, where it controls the number of CpCs and thus niche size, and for the maintenance of the correct number of GSCs during adulthood [17,[20][21][22][23][24][25]. We found, however, that mam participates in ovarian niche ageing independent of the Notch pathway.

A transcriptomic analysis identifies mastermind
as a candidate gene involved in ovarian stem cell niche ageing In order to identify genes involved in the ageing of stem cell microenvironments, we first confirmed that ovaries from aged, mated Drosophila females contain germaria with fewer GSCs than controls [6,17,18]. In addition, the proportion of niches devoid of GSCs in four-or six-week-old adults was significantly higher than that in one-week-old controls (figure 1b,c). Next, we used fluorescence-activated cell sorting (FACS) to isolate green fluorescent protein (GFP)-labelled TFCs and CpCs from one-, three-and four-week-old females and performed a transcriptomic analysis using DNA microarrays (approx. 400 cells/ array). We identified 30 upregulated and 32 downregulated genes in four-week-old females compared with one-week-old controls (false discovery rate < 0.05; linear fold change >2; figure 1d; electronic supplementary material, figure S1 and table S1; because of technical limitations, we could not isolate enough niche cells from six-week-old females to perform a similar analysis). Among the upregulated genes, mastermind (mam) was expressed 3.58-and 2.47-fold higher in three-and fourweek-old TFCs and CpCs, respectively, than in one-week-old controls. This was further confirmed by measuring directly mam RNA levels corresponding to the long and short isoforms using digital polymerase chain reaction (PCR) in isolated one-, three-and four-week-old niche cells (figure 2a). Considering that mam was the only gene upregulated in the three-week-old sample and that it maintained a higher expression in older cells, we decided to analyse in deeper detail the role of mam during ovarian niche ageing.

High mam levels in niche cells induce GSC loss
To validate a possible role for mam in aged niches, we first looked at mam expression, making use of an enhancer of the mam gene (GMR28A08; [26]) to drive the expression of a GFP reporter and found that the enhancer is active in TFCs, CpCs and ECs (figure 2b). Second, and considering that an increase of mam levels in niche cells correlated with a decrease in GSC numbers, we specifically knocked down mam mRNA amounts in adult niche cells to check if lower mam levels could prevent the drop in GSC numbers typical of ageing niches. We used the bab1-Gal4 driver combined with the Gal-80 ts to overexpress after eclosion an RNA interference (RNAi) version of the mam gene. bab1-Gal4 is expressed at high levels in adult TFCs and CpCs; it is also expressed, albeit weakly, in ECs [27]. We raised flies at 22°C and 25°C for four weeks to achieve different degrees of RNAi and found that, while experimental flies grown at 22°C behaved like controls in terms of GSCs/germarium (controls, 2.21 GSCs/germarium on average, number of germaria analysed (n) = 38; bab1 > mam RNAi, 2.23 GSCs/germarium, n = 30) and of Hedgehog and DE-cadherin levels (see below), raising the flies at 25°C induced GSC loss (controls, 1.74 GSCs/germarium, n = 31; bab1 > mam RNAi, 1.21 GSCs/germarium, n = 29; electronic supplementary material, figure S2A, S2B). We interpret this result as indicating that a certain activity of mam is required for GSC maintenance (in accordance with its known role in Notch signalling; see below in the Notch section). Third, to determine if the mere increase of mam levels could mimic niche ageing and thus GSC loss, we used a similar approach to overexpress a full-length version of the mam gene (mam Long ; see below) in the somatic cells of the adult niche (hereafter referred to as the bab1 > mam condition). Overexpressing mam for one or two weeks at consistent levels in the ovarian niche of freshly eclosed females gradually induced a significant GSC loss (one-week-old controls, 2.52 GSCs/germarium, n = 79; one-week-old bab1 > mam, 2.20 GSCs/germarium, n = 114; two-week-old controls, 2.45 GSCs/germarium, n = 65; twoweek-old bab1 > mam, 1.78 GSCs/germarium, n = 66; figure 2c). In addition, we also observed that the proportion of niches containing one or zero GSCs rose from 15% to 30% in females overexpressing mam for two weeks (figure 2d; see also controls in electronic supplementary material, figure S2C). These changes in GSC behaviour were not a consequence of modifying CpC numbers or fate, as determined by Lamin-C staining of control and experimental niches (figure 6c,d).
The mam gene encodes two different protein products, Mam-Long and MamShort . The latter lacks some C-terminal sequences essential for Notch signalling and it is known to act as a dominant-negative version that interferes with Mam Long and that represses the Notch pathway in certain cell types [28]. Since aged niche cells showed higher mRNA levels of both mam isoforms (figure 2a), we wished to check whether mam Short affected the GSC niche. We thus overexpressed mam Short for one week and found that GSC numbers did not change compared with controls (bab1 > mam Short ; electronic supplementary material, figure S2D). This result suggests that mam Short does not have a critical function in this context. Finally, we determined that one-week overexpression of full-length mam in somatic cells of the ovary with the traffic jam-Gal4 driver did not affect mam Short levels (electronic supplementary material, figure S2E), strongly suggesting that inducing high levels of mam Long in the ovary represents a true mam gain-of-function situation.
Because bab1-Gal4 is expressed at high levels in adult TFCs and CpCs and in ECs [27], we wished to determine if mam overexpression in TFCs/CpCs or ECs would provoke GSC depletion. To this end, we overexpressed mam for two weeks with the hedgehog-Gal4 line, expressed in adult TFCs and royalsocietypublishing.org/journal/rsob Open Biol. 9: 190127 royalsocietypublishing.org/journal/rsob Open Biol. 9: 190127 CpCs. Although the tendency was to find fewer GSCs in these experimental niches, there was no significant GSC loss compared with controls (two-week-old controls, 2.28 GSCs/ germarium on average, n = 29; two-week-old hh > mam, 1.93 GSCs/germarium on average, n = 28; figure 2e,f; see also controls in electronic supplementary material, figure S2F). Next, we drove full-length mam expression in adult ECs for two weeks using the patched-Gal4 line and found that the number of GSCs present in experimental niches, even though significantly lower in ptc > mam (two-week-old controls, 2.24 GSCs/germarium on average, n = 30; two-week-old ptc > mam, 1.79 GSCs/germarium, n = 29; figure 2g,h; see electronic supplementary material, figure S2F for additional controls), was not as reduced as in bab1 > mam. These data thus suggest that high levels of mam are necessary in TFCs/CpCs and ECs to account for the total GSC loss typical of aged niches. In all, from the above results, we conclude that mam levels increase with time in the niche, and that this increase is sufficient to induce GSC loss.

mam overexpression decreases DE-cadherin levels in niche cells
Next, we sought to determine the reason(s) for GSC loss upon induction of high mam levels. We dismissed cell death as oneweek-old bab1 > mam germaria did not increase expression of the Dcp-1 caspase in GSCs (electronic supplementary material, figure S3). Considering the role of GSC-niche cell adhesion in stem cell maintenance [29], we decided to analyse cadherinmediated adhesion in control and experimental germaria. It has been previously reported that ageing affects adhesion in GSC niches in both Drosophila males and females, and that this is due to a decrease in cadherin levels [2,6]. To test if the GSC loss characteristic of mam-overexpressing niches could be due to a reduction in cadherin levels, we studied the distribution of DE-cadherin in CpC-CpC and CpC-GSC interfaces of aged and bab1 > mam niches. As a reporter for DE-cadherin expression, we used a DE-cadherin::GFP fusion protein under the control of the endogenous regulatory sequences of the gene [30]. When analysing one-week-old versus six-weekold germaria, we found a significant decrease in DE-cadherin::GFP levels, with aged niches containing on average DE-cadherin levels of 74% of the controls (figure 3a,b). We then examined the effect on DE-cadherin::GFP levels of overexpressing mam for one or two weeks in experimental germaria and found that, while one-week-old overexpressing germaria behaved like controls, DE-cadherin::GFP levels were significantly reduced after two weeks of overexpression (figure 3c, d; experimental germaria contained on average 86% DE-cadherin::GFP levels of controls). To further confirm this result, we measured endogenous, untagged DE-cadherin protein levels by immunostaining and found that they were reduced by nearly 50% after two weeks of mam overexpression (electronic supplementary material, figure S4A,B). Together, the above data demonstrate that mam overexpression in young niches recapitulates the adherens junction defects typical of older niches. They also suggest that high levels of mam are sufficient to induce premature niche ageing and the concomitant loss of GSCs by impairing cadherin levels and thus niche cell-GSC adhesion. Finally, to test if the reduction in DEcadherin levels upon ectopic expression of mam is functionally relevant, we checked whether DE-cadherin overexpression could rescue the GSC loss characteristic of bab1 > mam germaria. We found that a two-week induction of high DE-cadherin amounts in bab1 > mam niche cells did not rescue the decrease in GSC contents characteristic of high mam levels (electronic supplementary material, figure S4C).

Aged niches show defective DE-cadherin trafficking
Once we had confirmed that the overexpression of mam induced a precocious decrease in DE-cadherin levels in niche cells, we determined if membrane DE-cadherin turnover was also affected by age and/or mam overexpression. We set out to measure the dynamics of protein replacement in CpC cytoplasmic membranes by performing FRAP (fluorescence recovery after photobleaching) experiments to look at the recovery of DE-cadherin::GFP signal in one-week-old versus four-week-old females (the DE-cadherin::GFP signal in females older than four weeks was too low to allow consistent measurements). In young samples from DE-cadherin::GFP females, bleached CpC membranes reached a maximum of fluorescence recovery close to 30% of the pre-bleached signal 30 min after photobleaching, which was in the same range as previously published data involving DE-cadherin::GFP FRAP analyses in embryos [31]. By contrast, four-week-old cells subjected to the same experimental conditions recovered less than 10% of fluorescence 30 min after photobleaching, suggesting that ageing jeopardizes trafficking and replacement of cell-surface DE-cadherin (figure 4a,b). Next, we analysed the membranebound fusion protein Resil::GFP as a turnover control, and noticed that one-week-old samples recovered 40% of the pre-bleached signal 30 min after photobleaching, while fourweek-old TFCs and CpCs recovered around 20% (figure 4c, d). Thus, the mere ageing of niche cells can affect membrane protein replacement dynamics. It remains to be determined if the small difference in fluorescence recovery observed on DEcadherin::GFP trafficking in aged niches compared with that on Resil::GFP implies that ageing impairs DE-cadherin recycling to a larger extent than other membrane proteins. royalsocietypublishing.org/journal/rsob Open Biol. 9: 190127 The Rab11 GTPase has been implicated in DE-cadherin recycling from internalized endosomes [32]. Nuclear fallout (Nuf ) is an adaptor protein that interacts with Rab11 to control the recycling endosome [33]. In order to test if niche cell ageing affected membrane protein trafficking by changing Rab11-Nuf levels, we determined whether Rab11 and Nuf protein amounts in CpCs varied during ageing conditions. We compared one-and six-week-old germaria and observed a significant reduction in both Rab11 and Nuf contents in the aged cells (figure 4e,f; electronic supplementary material, figure S5C,D).
To address the role of high mam levels in DE-cadherin recycling, we first performed FRAP on DE-cadherin::GFP TFCs and CpCs overexpressing mam for one week but could not detect any differences in fluorescence recovery with control flies (electronic supplementary material, figure S5A,B), indicating that DE-cadherin::GFP dynamics are not changed in one-week-old bab1 > mam flies. This result is in line with our previous observation that ovarian niches in these experimental females show normal levels of DE-cadherin::GFP. Unfortunately (as shown in figure 3c,d), overexpression of mam for two weeks reduces DE-cadherin::GFP levels to the extent that we could not perform FRAP analyses reliably. Second, we observed that after two weeks of mam overexpression Rab11 levels were decreased, suggesting that high mam quantities affected membrane protein recycling. However, we failed to detect changes in Nuf amounts upon mam overexpression for two weeks (figure 4g-h; electronic supplementary material, figure S5E,F). In all, our results demonstrate that physiologically aged niche cells recycle membrane proteins (such as DE-cadherin or Resil) less efficiently than younger ones. Moreover, considering the reduction in DE-cadherin, Rab11 and Nuf levels in aged niches (and of DE-cadherin and Rab11 in high mam conditions), our results may suggest that ageing favours the sorting of internalized DE-cadherin (and probably many other membrane proteins such as Resil) towards the lysosomal degradation pathway rather than towards the Rab11-mediated recycling [34,35], and that mam overexpression induces a similar scenario. royalsocietypublishing.org/journal/rsob Open Biol. 9: 190127 niche cell numbers ( figure 6c,d), the decrease of GSC numbers in ovarian niches expressing high levels of mam could be due to flawed niche signalling that, in turn, induced GSC differentiation. In this context, it has been shown that aged niches are less effective in transducing the BMP pathway or in keeping constant levels of the Hedgehog (Hh) protein, two events that act normally in the niche to maintain a proliferative GSC pool [6,15,18,36,37]. We thus decided to analyse whether mam overexpression had an effect on the signalling properties of the GSC niche. First, we quantified the activity of the BMP pathway by scoring the intensity of the BMP activity reporter phosphorylated Mothers against dpp ( pMad) in aged GSCs containing normal-looking spectrosomes. In contrast to published results [18], we could not detect a significant decrease in pMad levels in six-week-old GSCs (electronic supplementary material, figure S6A,B), suggesting that, at least under our conditions, BMP signalling might not be affected by ageing. We found the same scenario when looking at bab1 > mam niches, which did not show a decrease in pMad levels even after a two-week exposure to high mam levels (electronic supplementary material, figure S6C,D). Second, we looked at the Hedgehog pathway. The Hh ligand is expressed in TFCs and CpCs, where it acts to maintain the GSC pool in adult ovaries [15,36]. Interestingly, the analysis of six-week-old niches provided a clear difference in Hh signal when compared with one-week-old controls (figure 5a,b). This finding is in agreement with a previous report showing that aged TFCs/CpCs exhibited lower activation of a hh-lacZ enhancer trap [18]. Similarly, niches overexpressing mam (bab1 > mam, two weeks at 29°C) also displayed significantly lower levels of Hh amounts (figure 5c,d).
To test if ageing and/or mam overexpression affected not only Hh levels in CpCs but also Hh signalling in the target cells, we used an antibody to Patched that gives a clean, specific staining in the wing disc and we scored Patched amounts in the anterior half of the germarium. The Patched (Ptc) receptor is a known target of Hh signalling and its transcription is a direct readout of the pathway [38,39]. In control germaria, Ptc is found in puncta in the areas that the separate the different germline cysts and that are occupied by the ECs, similar to the expression reported for a ptc-lacZ construct [15]. CpCs did not show any specific visible signal. This pattern of expression recapitulates that described for a reporter construct known to mimic hh-pathway signalling, Ptc-pelican-GFP [40]. However, and in spite of the reduced Hh amounts under both experimental conditions, we found that ECs of aged and bab1 > mam germaria kept for two weeks at 29°C contained Ptc levels similar to their respective controls (figure 5e-g). From the above experiments, we conclude that both ageing and mam overexpression can induce a decrease in Hh levels in niche cells. Surprisingly, we also found that the activity of the pathway-at least as indicated by Ptc levels-seemed to be maintained in the above conditions, suggesting that other modulators apart from the Hh ligand may act during its signalling in the germarium. One such modulator reported to act in other tissues such as imaginal discs is dachshund [41]; however, in our experiments, Dachshund was not expressed in the germarium. Thus, the mechanism(s) responsible for the enhanced Hh-independent signalling in aged ovarian niches remains to be identified. In this context, we also tested whether the overexpression of a functional version of the Hh protein, Hh::GFP [42], could rescue the loss of GSC numbers induced by high mam levels and found that it did not (electronic supplementary material, figure S6E).

Increased antioxidant activity can rescue premature
ageing due to mam overexpression ROS have been suggested to cause ageing, and previous results indicate that reduction of ROS in either GSCs or their niche improves GSC proliferation and lifespan, pointing towards an age-dependent increase in ROS levels in niches and stem cells [6]. In order to determine if high mam levels in the niche could induce premature ROS-dependent GSC loss, we tested whether reducing oxidative species in bab1 > mam niches could improve GSC maintenance and its effect on Hh and DE-cad amounts. To this end, we increased the amounts of the well-known antioxidant enzyme superoxide dismutase (SOD1) in the cytoplasm of niche cells and determined that SOD overexpression rescued the GSC loss characteristic of bab > mam niches (two-week-old controls, 2.91 GSCs/germarium, n = 35; two-week-old bab1 > mam + GFP, 2.05 GSCs/ germarium, n = 37; two-week-old bab1 > mam + SOD1, 2.93 GSCs/germarium, n = 40; figure 6a). We also confirmed that Hh and DE-cad levels in experimental flies were similar to those in controls (electronic supplementary material, figure  S7A). Since overexpression of SOD1 on its own in niche cells does not affect GSC numbers (electronic supplementary material, figure S7B), we thus propose that mam overexpression in niche cells increases ROS levels and that this contributes to the reduction in GSC numbers found in bab1 > mam niches.

High mam levels induce GSC loss in a Notchindependent manner
Mam is a transcriptional co-activator that supports the activity of the Notch receptor by forming a ternary complex with the DNA-binding protein CSL (Su(H) in Drosophila) and the Notch intracellular domain, NICD or N intra [19,43]. Thus, Mam is proposed to act as an essential component of   royalsocietypublishing.org/journal/rsob Open Biol. 9: 190127 GSCs, it decreases in seven-week-old niche cells [17,21], suggesting that low levels of Notch pathway activity in niche cells correlate with the GSC loss characteristic of aged niches, in line with our finding that bab1 > mam RNAi females grown at 25°C contained fewer GSCs than their controls (electronic supplementary material, figure S2A). However, we have found that mam levels augment at least during the first four weeks in aged niche cells and that an increase in mam mRNA in somatic niche cells induces GSC loss, results that are seemingly contradictory to Mam's role as a cofactor of the Notch pathway. Hence, we set out to investigate whether the observed higher mam levels in three-and fourweek-old niche cells may act Notch-independently. First, we confirmed that loss of function of the Notch pathway (by reducing the amount of the Delta ligand in niche cells) reduced GSC numbers (one-week-old controls, 2.53 GSCs/ germarium, n = 30; one-week-old bab1 > Dl RNAi, 2.13 GSCs/germarium, n = 27; figure 6b), corroborating that royalsocietypublishing.org/journal/rsob Open Biol. 9: 190127 adult germaria depended on Notch signalling in niche cells for GSC maintenance [17,21]. Second, we showed that increasing Notch signalling in niche cells (by overexpressing N intra in niche cells) did not induce GSC loss, a finding in sharp contrast to the decline in GSC numbers typical of mam overexpression (one-week-old controls, 2.98 GSCs/germarium, n = 29; one-week-old bab1 > N intra , 3.16 GSCs/ germarium, n = 29; figure 6b). To prove that the Notch pathway was activated in bab1 > N intra , we determined the number of CpCs present in control and experimental niches. It has been previously shown that ectopic activation of the Notch pathway in niche cells induces extra CpCs [23]. In our experimental conditions, two weeks of N intra overexpression nearly duplicated the number of CpCs present in experimental germaria (bab1 > N intra ). By contrast, ectopic expression of mam for two weeks (bab1 > mam) did not affect CpC numbers ( figure 6c,d).
Thus, the fact that increasing mam mRNA levels in niche cells induced GSC loss (whereas ectopic Notch signalling did not) and the observation that overexpression of N intra augmented CpC numbers (in contrast to mam overexpression, which did not) led us to conclude that the induction of high mam mRNA levels during niche ageing affects the GSC pool in a Notch-independent pathway and that high Mam levels in aged niche cells might interact with partners different from the CSL/NICD complex figure 7.

mam, a novel factor in stem cell niche ageing
The ageing of tissues correlates with a decline in regenerative capacity and in homeostasis, linking the activity of tissuespecific stem cells to age-related organ dysfunction. Several factors had been reported to cause stem cell waning during ageing, mainly increased ROS production, decreased signalling efficiency and compromised adhesion to niche support cells, all of which can jeopardize stem cell maintenance within the niche. These 'ageing' factors can be intrinsic to the stem cells, or affect their surrounding microenvironments, which also contribute to impaired tissue activity in aged organisms. We have identified mam as a novel factor involved in the ageing of a particular stem cell microenvironment, the ovarian niche of Drosophila melanogaster.
Our transcriptomic analysis reveals a discrete number of genes whose expression is either upregulated or downregulated in niche support cells aged for three or four weeks, which represents around the half-life of a laboratory fly. The fact that the identified genes in the three-week-old samples are only a small fraction of those in four-week-old cells (and mam is a good example of this, as it is the only gene upregulated in threeweek-old samples whereas in four-week-old niches there are 30+ upregulated genes), and that all but one candidate maintain their differential expression one week later, suggests that ageing is a gradual process in which few genes implicated in seemingly unrelated biological processes contribute to the progressive loss of stem cells from the niche.
mam is upregulated in three-and four-week-old niche cells, albeit with a smaller increase in the latter. Interestingly, when artificially induced, mam overexpression in niche cells is enough to bring about stem cell loss in young tissues in a manner that recapitulates niche ageing, as it can induce DE-cadherin loss; it decreases the levels of a key niche signalling molecule (Hh); and it most likely increases ROS amounts in niche support cells. The link between these important events in stem cell maintenance is currently missing, as it relates to why higher mam levels would induce oxidative stress most likely by increasing ROS quantities. We can only speculate that overexpression of mam in otherwise young niches might induce an accumulation of ROS-dependent damage in the cell, which ultimately results in impaired signalling, recycling and adhesion within the niche. Perhaps high mam amounts accelerate the cell's metabolism, thus increasing the generation of free radicals to the extent that the cell cannot cope with them efficiently. Alternatively, or in addition, mam overexpression might impede the normal activity of the antioxidant defence system present in niche cells. Whatever the mechanism, this places the physiological increase in mam levels in a preeminent position in the ageing of the ovarian niche, at least during the first half of adult life. Furthermore, the fact that 'young' stem cells can be lost from artificially aged niches indicates that support niche cells are fundamental to regulating stem cell waning from old niches and that the physiological condition of the microenvironment may determine the maintenance of stem cells independently of stem cell age. This observation emphasizes the constraints that niche physiology poses on stem cell-based therapies.

An unconventional function for high mam levels independent of Notch in aged niches
The Notch signalling pathway operates in a myriad of evolutionarily conserved developmental and disease processes. It ensures the release of NICD upon ligand binding. Partnered with the DNA-binding protein CSL and the co-activator Mam, the NICD/CSL/Mam complex regulates expression of target genes. Modulation of this process involves several factors, including ligand and receptor expression and levels, regulation of the processing steps required to release NICD and crosstalk with other signalling pathways [45,46]. There are, however, examples in which CSL or Mam can act independently of their interaction with Notch to induce different cellular phenotypes [47]. For instance, CSL can engage the pancreas transcription factor PTF-1 to regulate exocrine pancreatic genes [47,48]. In Drosophila, Su(H) acts as a Notch-independent activator to allow proper bristle formation in the adult notum royalsocietypublishing.org/journal/rsob Open Biol. 9: 190127 [49]. As for Mam, it is known that it can interact with other factors not included in Notch signalling, such as MEF2C, p53, β-catenin and NF-κB [50][51][52][53]. In this context and since Mam functions as a co-activator in numerous cellular processes, Mam family proteins have been proposed to mediate crosstalk among multiple signalling pathways involving Notch signalling or independent of it [54]. In the case of Drosophila, several studies have also reported a non-canonical role for Mam in different systems such as the central nervous system, where Notch and mam mutants display distinct phenotypes [55], or the wing and eye imaginal discs, where Mam regulates Wnt pathway activity independently of Notch through the activity of specific, additional factors [56]. Interestingly, considering that Wnt signalling controls the redox state of germline and somatic cells of the germarium [57], mam may provide a link between mam, Wnt signalling and ROS in the ovarian niche. Finally, a previous report demonstrated that Mam does not act in the conventional Notch-transduction pathway in the response of FSCs to the Hedgehog ligand in the Drosophila ovary [58]. These authors also showed that overexpression of a dominant-negative, truncated Mam protein lacking an acidic cluster involved in transcriptional activation could interfere with Notch signalling in the ovary but did not affect Mam Long activity in FSCs, where mam acts in a Notch-independent fashion. In our work, we demonstrate that high mam levels and the overactivation of the Notch pathway have distinct outcomes, supporting our claim that the naturally occurring upregulation of mam mRNA levels acts independently of Notch signalling. Furthermore, while the nature of the partners in niche cells containing elevated Mam amounts is currently unknown, they are most likely not the canonical effectors of the Notch pathway Su(H) and N intra , suggesting that fulllength Mam is able to interact with other partners to regulate GSC numbers during ageing. We envision a situation in which increased levels of the Mam co-activator could be acting in the niche to integrate multiple signalling pathways that ultimately lead to niche ageing. The consequences of this regulatory cross-talk are the reduction in niche cell-stem cell adhesion and niche signals such as Hedgehog and the increase in oxidative stress characteristic of aged niche cells.

Defective DE-cadherin trafficking in aged niches
A landmark of aged ovarian niches is a characteristic reduction in the cadherin-mediated adhesion between stem cells and the supporting CpCs. In addition to confirming that ageing and mam overexpression reduce DE-cadherin amounts, we report that the decreased DE-cadherin levels in aged niches could be due to a less effective trafficking and/or to a differential balance between recycling and endocycling for destruction via the lysosomal compartment. This idea is supported both by the FRAP experiments-which indicate an age-dependent decline in recycling capacities of the CpCs-and by the decrease in Rab11 (also observed in mam-overexpressing niches) and in Nuf contents in aged cells, which could be indicative of unbalanced DE-cadherin recycling/degradation. Together with the finding that maturing niches present a reduced capacity to replace the fusion protein Resil::GFP, our results thus point towards a general decay in membrane protein turnover with age, hence implicating ageing in the correct trafficking of membrane-associated proteins. Whether this effect is of general importance and whether it is due to the increasing levels of oxidative stress in old niche cells remains to be seen, but the fact that protein replacement dynamics are related to ageing and that DE-cadherin recycling seems more sensitive to ageing than a control protein is of particular interest to the field, in view of the importance of stem cell-niche cell adhesion for tissue homeostasis [59,60].

Material and Methods
The following is a short version of the material and methods used in this study. A complete account can be found in the electronic supplementary material.

Germline stem cell quantifications
GSCs were identified by their position within the germarium (they are located just underneath the CpCs), by their size (they are larger that differentiating germline cells) and by the presence of a spectrosome in their cytoplasm. GSC spectrosomes adopt different morphologies depending on the cellcycle phase. The number of GSCs found in a niche shows some variability, even in flies of the same genotype and raised in the same vial, and it ranges from 2 to 3 GSCs/germarium in the different control females used in this study. Because of this variability, in all of the experiments where we scored GSC numbers, we analysed control flies in parallel with their experimental siblings. To normalize the changes observed, we show the variation in the percentage of GSCs/germarium in all pertinent graphs. Thus, comparisons should be made between control and experimental niches of the same experiment and not between genotypes of different treatments and/or conditions. All quantifications were obtained from at least five different females/replicate of a given phenotype and from a minimum of two biological replicates. We also performed a number of controls to make sure that the chromosomes used in this study did not affect GSC numbers specifically. Thus, we quantified GSCs in females bearing the TM6B balancer or chromosomes carrying the bab1-Gal4, tub-G80 ts , ptc-Gal4 or hh-Gal4 constructs and determined that they did not have a significant effect on the GSC pool present in the niches.

Image analysis
Confocal images of control and experimental samples processed in parallel were obtained with exactly the same settings. Average fluorescence intensity was quantified with ImageJ. See electronic supplementary material and [65] for further details.

Fluorescence recovery after photobleaching
Regions of interest per niche were bleached with the 488 nm and 405 nm laser lines at 100% power for three 2.6 s long scans (400 Hz). Three Z-stacks were then collected at 10 min time intervals.

Isolation of niche cells and microarray analysis
Preparation of single-cell suspensions was done by optimizing a previous protocol [66]. Niche cells were selected by FACS according to size, viability and GFP-signal intensity, which allowed TFCs and CpCs to be separated from ECs. RNA isolated from three biological replicates for each of the three time points was used to prepare amplified cDNA libraries. The nine cDNAs were hybridized to GeneChip Drosophila Genome 2.0 Arrays from Affymetrix.

Quantification of mRNA levels by qPCR and ddPCR
The relative mRNA amounts of mam long and short isoforms were determined by real-time quantitative PCR (qPCR; CFX Connect; Bio-Rad) using the comparative cycle threshold (C T ) method [67]. Pico-profiled samples from sorted niche cells (TFCs and CpCs; see above) were used to quantify mRNA levels of mam long and short isoforms by Droplet Digital PCR (ddPCR).

Statistical analysis
Samples were collected from at least five different adult females. Graphs represent the mean values and standard errors. Significant differences were determined using two-tailed Student's t-tests. In all figures, only p-values of two-tailed t-tests considered statistically significant ( p ≤ 0.05) are indicated. Thus, the absence of asterisks indicates no statistical significance between control and experimental values.
Data accessibility. The article's supporting data and materials are referred to in the manuscript. The electronic supplementary material contains data relevant to our discoveries.