Atg1 modulates mitochondrial dynamics to promote germline stem cell maintenance in Drosophila

Mitochondrial dynamics (fusion and fission) are necessary for stem cell maintenance and differentiation. However, the relationship between mitophagy, mitochondrial dynamics and stem cell exhaustion is not clearly understood. Here we report the multifaceted role of Atg1 in mitophagy, mitochondrial dynamics and stem cell maintenance in female germline stem cells (GSCs) in Drosophila. We found that depletion of Atg1 in GSCs leads to impaired autophagy (mitophagy) as measured by reduced formation of autophagosomes, increased accumulation of p62/Ref (2)P and accumulation of damaged mitochondria. Disrupting Atg1 function led to mitochondrial fusion in developing cysts. The fusion was a result of an increase in Marf levels in both GSCs and cysts, and the fusion phenotype could be rescued by overexpression of Drp1 or by depleting Marf via RNAi in Atg1-depleted cyst cells. Interestingly, double knockdown of both Atg1:Marf affected ovariole size and the number of vitellogenic oocytes. While Atg1:Marf knockdown led to decrease in germ cell number. Strikingly, Atg1:Marf double knockdown leads to a dramatic loss of GSCs, GCs and a total loss of vitellogenic stages, suggesting a block in oogenesis. Overall, our results demonstrate that Drp1, Marf and Atg1 function together to influence female GSC maintenance and their differentiation into cysts. Research Highlights Atg1, in addition to its role in mitophagy, influences mitochondrial dynamics during oogenesis through modulation of Marf. Atg1 and Marf promote Germline stem cell maintenance in Drosophila.


Introduction
Female GSCs have the ability to self-renew and differentiate into the egg. This ability depends on their interaction with the niche cells in Drosophila. GSC maintenance affects reproductive fitness and is crucial for the survival of the species [1]. Autophagy is an important homeostasis mechanism operational within all cells, including GSCs. Autophagy maintains proteostasis by removing toxic aggregates and redox homeostasis by removing damaged mitochondria via mitophagy. However, the molecular role of autophagy in the maintenance of GSCs is not understood [2]. Autophagy is a highly regulated and complex process that is initiated by Ulk1/Atg1, a serine-threonine kinase that phosphorylates downstream proteins that catalyze the formation of the autophagosome. Autophagosome fuses with the lysosome to form autolysosome, wherein the hydrolytic enzymes degrade the cargo and the contents released in the cytoplasm for recycling [3].
In metazoans, mitochondria are inherited maternally and are important for generating energy in the form of ATP. Mitophagy is crucial for the maintenance of GSCs by sensing and removal of dysfunctional mitochondria. Mitophagy is influenced by mitochondrial biogenesis, dynamics (fission and fusion), translocation (distribution) and degradation [4]. Perturbation in any of these processes can lead to an imbalance in redox homeostasis and mitochondrial dysfunction. If mitochondrial damage is beyond repair, the fusion machinery is inactivated, and mitochondrial fragmentation is promoted to facilitate mitophagy. Mitochondrial dynamics is controlled by activities of Dynamin-related protein 1 (Drp1), Mitochondrial assembly regulation factor (Marf) (Mitofusin 2 Mfn2 in mammals) and Optic atrophy 1 (Opa1). These GTPases hydrolyse GTP and use the energy to execute changes in mitochondrial morphology.
Mfn2 is required for the fusion of outer mitochondrial membranes of mitochondria. Opa1 is important for the fusion of the inner mitochondrial membrane [5]. Fission requires GTPase Dynamin-related protein 1 (Drp1), which forms oligomers surrounding the mitochondria and cleaves the mitochondria. In Drosophila, loss of Marf and Drp1 can lead to progressive loss of the female GSCs. Loss of Marf results in a reduction of pMad signaling, which is a self-renewal factor for female GSCs [6,7]. Mitophagy has also been demonstrated to be crucial for the development of the ovariole and oocyte. Mitochondria exhibit maternal inheritance, and it is vital for proper embryogenesis. Mitochondria with mutated mtDNA are eliminated during oogenesis by mitophagy and requires Atg1 and BNIP3 [8]. Despite these interesting observations, the relation between mitophagy, mitochondrial dynamics and stem cell exhaustion is not clearly understood.
We used Drosophila female GSCs to address the question of how mitophagy, mitochondrial dynamics, and stem cell maintenance are related. Female Drosophila harbor a pair of ovaries with 16-20 individual units called ovarioles within each ovary. The GSCs are located within the anteriormost structure called the germarium in each ovariole. It is surrounded by the niche cells that secrete factors required for maintenance of the 'stemness' while the daughter cells that exit the niche differentiate into cystoblast (CB). A single CB undergoes four asymmetric cell divisions to generate interconnected 16-celled cysts. One of the 16-celled cysts is chosen as the future oocyte, which is then sustained by the remaining 15 nurse cells [1]. The Drosophila ovary presents a simple and genetically tractable model to study mitophagy, mitochondrial dynamics and their role in stem cell maintenance.
Here we demonstrate the importance of Atg1 in the maintenance of basal autophagy and its role in mitophagy using Atg1RNAi and loss-of-function clones of Atg1. Atg1 knockdown cyst cells also show a strikingly different mitochondrial morphology which implies a relation between the mitophagy machinery and the mitochondrial dynamics. Using a combination of genetics and cell biology, we demonstrate that Drp1, Marf and Atg1 collaborate to regulate stem cell maintenance in Drosophila.

Fly maintenance
All the transgenic fly stocks were maintained at 25°C on standard cornmeal sucrose malt agar.

Generation of Atg1 clones
FRT80BAtg1∆3D mutants crossed to FRT80BmRFP or FRT80BGFP were collected and fed on yeast for 2 days. The clones were generated using the FLP-FRT technique described previously .

Immunostaining
Immunostaining was performed according to the protocol described in [2]. Antibodies and

Redox chemical treatment
Ovaries of mito-roGFP2-Orp1 and mito-roGFP2-Grx1 flies were dissected in Grace's medium and washed in 1xPBS for 2 minutes. The redox treatment procedure was followed according to [2] Imaging and Analysis All imaging was performed on Leica SP8 Confocal microscope using 63x oil objective.
Images acquired were 8 bit, 1024x1024 pixel resolution at 100Hz scanning. Frame accumulation was performed with 6 frames for mCherry-Atg8a and GFP-Ref(2)P. Mito-roGFP2-Grx1 imaging and image analyses for all markers were performed as described in [2].
For mCherry-Atg8a puncta measurements in Atg1∆3D (Atg1-/-) clones, an ROI was drawn around the GSCs and Atg1-/-clones. GSCs were identified by their close association with the niche cells and size of their nuclei while clones were recognized by absence of GFP. mCherry-Atg8a and GFP-Ref(2)P punctae were counted manually in RNAi mediated knockdown studies and the area of germarium were measured using ImageJ. Analysis of mitochondrial size was done by the MitoAnalyzer plugin. For colocalization analysis, we analyzed overlap of mCherry-Atg8a and CathepsinL, and GFP-Ref(2)P with CathepsinL puncta. Mitophagy flux was measured as colocalization between mitochondria (ATP5α) and lysosomes (Cathepsin L).
Microsoft Excel was used for statistical analysis. Student's T-Test of two samples assuming unequal variance was performed for all comparisons. GraphPad Prism7 was used to plot the graphs.

DHE staining protocol
DHE staining and live imaging were performed according to manufacturer's instruction (Invitrogen, USA).

RNA isolation
20 pairs of ovaries were used for RNA extraction using the Trizol method as per manufacturer's instruction (Invitrogen, USA).

cDNA synthesis
1µg of tissue specific (ovary) total RNA was used for First strand cDNA synthesis using PrimeScript TM 1st strand cDNA Synthesis Kit (TakaraBio, India, Cat# 6110A).

qPCR:
qPCR was performed on EcoMax (Bibby Scientific, UK). The mean of housekeeping gene Tubulin was used as an internal control for normalization. The expression data was analysed using 2-ΔΔCT described by [10]

Atg1 is necessary for basal autophagy and mitophagy during oogenesis.
Atg1 is necessary and sufficient to induce autophagy in several tissues in Drosophila [3].
However, if Atg1 is required for autophagy in female GSCs is unclear. To address this, we Autophagy is necessary for the removal of damaged mitochondria through mitophagy. p62/Ref(2)P is recruited to damaged mitochondria and has been shown to be necessary for mitophagy in mammals and Drosophila by facilitating the recruitment of damaged mitochondria to autophagosomes. Further, p62/Ref(2)P has been shown to be essential for PINK1-Parkin-mediated mitophagy, wherein decline of mitophagy was shown to be associated with increased co-localization with p62/Ref(2)P. As expected, we observed a significant number of punctate p62/Ref(2)P structures were found to colocalize with mitochondria in Atg1RNAi cells as compared to controls consistent with disruption of mitophagy ( Figure 1E-H, Supplementary Figure 1G-G"). Further, we measured mitophagy flux using colocalization of ATP5 alpha (red staining) and CathepsinL (green staining), followed by Pearson's coefficient correlation. Mitophagy flux was found to be significantly reduced in Atg1RNAi expressing GSCs (and GCs) as compared to the controls, evident by reduced Pearson's coefficient correlation (Supplementary Figure 1L-O). Taken together, these data suggest that mitophagy is disrupted in GSCs and cysts in Atg1KD.
Mitochondria within GSCs are elongated and become fragmented as cystoblasts differentiate into cysts cells [11]. In Atg1KD, mitophagy is disrupted in all cells within the germarium, but mitochondria were predominantly fused, particularly in the cyst cells, which was very surprising (

Atg1 modulates Marf levels during oogenesis.
Mitochondrial dynamics are regulated primarily by specific GTPases. Marf (Mitochondrial assembly regulatory factor) or Mitofusin 2, which regulates mitochondrial outer-membrane fusion, while Opa1 mediates fusion of the inner membrane of mitochondria.
Mitochondrial fission is catalyzed by Drp1 (Dynamin-related protein 1), which associates itself with the mitochondrial outer membrane and constricts mitochondria [6,7]. Mitochondrial fusion events can occur when Drp1 is reduced or when Marf and Opa1 levels increase. It is possible that cysts expressing Atg1RNAi have reduced activity of Drp1 or increased Marf activity. To test these possibilities, we monitored the levels of both Drp1 and Marf in Atg1KD cells. We utilized protein fusion reporters, Drp1-HA and Marf-GFP (Mfn2-GFP), that are expressed from their respective endogenous promoters [12]. No detectable difference in Drp1-  Figure 2E). Taken together, our results indicate that Atg1 is necessary to maintain fissed mitochondria in the cyst cells by regulating Marf levels in these cells. Further, we interfered with Marf function using RNAi-mediated KD and tested whether this affects mitochondrial morphology [6,7]. MarfKD caused fragmentation of mitochondria within germarium, including GSCs which is consistent with the reduction in fusion events upon depletion of fusion regulators ( Figure 2I, L and Supplementary Figure   2G). Taken together, our data show that Atg1KD disrupts mitochondrial dynamics leading to fused mitochondrial morphology during early oogenesis.

Atg1 genetically interacts with mitochondrial fusion machinery components.
We next sought to test whether enhancing fission can rescue the Atg1KD mitochondrial fusion phenotype in cysts, given that Marf levels are elevated in Atg1RNAi-expressing germline cells.
Fission could be enhanced by either elevating levels of Drp1 protein or by reducing Marf levels [6,7]. We hypothesized that by forcing mitochondrial fragmentation by overexpressing Drp1  Figure 3A, B, F). The mitochondrial area was significantly reduced as compared to GCs expressing Atg1RNAi alone, suggesting that Atg1 and Drp1 genetically interact with each other to control mitochondrial dynamics. However, it is important to note that as compared to controls, the complete rescue of the mitochondrial phenotype was not observed.
We further tested if the mitochondrial fusion phenotype can be rescued by reducing Marf levels in Atg1RNAi. We co-expressed Atg1RNAi and MarfRNAi in all GCs. Previous studies have shown that a reduction in Marf levels within the GSCs leads to the fragmentation of mitochondria [6,7]. GSCs/GCs co-expressing Atg1RNAi and MarfRNAi exhibited fragmented mitochondria suggesting suppression of fused mitochondrial phenotype in GCs

Atg1, Drp1 and Marf collaborate to regulate oogenesis in Drosophila.
GSCs undergo differentiation and proceed through multiple stages of development termed as egg chambers. Egg chambers undergo further development and they are supplemented with factors necessary during embryogenesis. Previous reports suggest that both Drp1 and Marf are crucial for oogenesis, and their absence leads to detrimental effects on the development of the ovary [6]. Atg1 has been demonstrated to be dispensable for oogenesis as Atg1KD, or null mutants, do not disrupt egg development [13]. Our data also suggests that overall egg development is not disrupted in Atg1KD. However, the average area of the ovary is significantly reduced in Atg1KD (Figure 4A Figure 4D-E"', H and I).
Depletion of GSCs and their differentiated progeny GCs has been previously linked to the failure of oogenesis.
We next asked if GSC maintenance was affected in Atg1KD and Atg1:Marf double KD.
We did not consider Drp1KD and Atg1:Drp1 double KD for GSC maintenance assay as Drp1 levels did not alter in Atg1KD. Also, previous studies have shown that Drp1 mutants exhibit GSC loss or gain based on the use of Drp1 mutants and Drp1RNAi and the assay used [6,7].
We expected GSC loss in Atg1KD as compared to the controls. Unexpectedly, average GSCs counts were higher in Atg1KD at midlife as compared to the controls (Figure 5A-D').
Increased fusion of mitochondria was shown to reduce GSC loss due to aging and was observed case of decreased Drp1 or increased Marf activity. Atg1:Drp1 double KD did exhibit a GC loss phenotype. Interestingly, however, the double KD of Atg1:Marf exhibited a dramatic loss of GSCs as early as 1week and a complete loss of GSCs at midlife of the flies. To further assess the GSCs phenotypes in Atg1KD and Atg1:Marf double KD, we measured pMad levels in GSCs. pMad, a self-renewal signal in GSCs is reduced in Marf mutants [7]. We detected a significant loss of pMad intensity in both Atg1KD GSCs ( Figure 6A-A', B-B' and E). and Atg1:Marf double KD suggesting that the reduction in pMad may lead to GSC loss ( Figure   6C-C', D-D' and E). In summary, these data suggest that Drp1, Marf and Atg1 function together to regulate both the maintenance of GSCs and differentiation of GCs during Drosophila oogenesis.

Discussion
Our work demonstrates that Autophagy related gene-1 (Atg1) in addition to its role in the regulation of autophagy (and mitophagy) in Drosophila, also regulates mitochondrial dynamics during oogenesis. Atg1 is a core autophagy protein in Drosophila that exerts its function through its kinase activity by phosphorylating substrates necessary for the upregulation of autophagy in several tissues. Loss of Atg1 has been previously reported to disrupt autophagosome formation, and we observed disruption of autophagosome formation in Atg1RNAi expressing cells as well as Atg1-/-clones in the germarium [3,13]. Additionally, we also observed the accumulation of Ref (2) Figure 2A). Thus, these data suggest that Atg1 influences Drp1 expression.
Interestingly though, this did not lead to increased mitochondrial fragmentation. Instead, mitochondria were fused suggesting that fusion phenotype is independent of Drp1 in this context. Marf levels exhibited a significant increase in both GSCs and particularly in the cyst cells, where the mitochondrial fusion phenotype was most prominent. This suggested that the fusion of mitochondria in Atg1RNAi is primarily mediated by elevated Marf and not through reduced Drp1. Marf is downregulated in wild-type germline cysts to facilitate mitochondrial selection by enabling the fragmentation of mitochondria. Interestingly, we observed that impaired Drp1 function could cause mitochondrial fusion leading to the formation of "mitochondrial islands" similar to Atg1RNAi (Figure 4). Thus, it is possible that the mitochondrial fusion phenotype could be due to changes in the ratio of expression Drp1 to Marf rather than changes in their expression. These findings need to be further explored in the context of mitophagy and mitochondrial dynamics during Drosophila oogenesis.
The Atg1 mitochondrial fusion phenotype could be rescued by Drp1 OE, suggesting that These studies need to be conducted to understand the relationship between Atg1, Drp1 and Marf.
Previous studies and our work suggest that Atg1 is not necessary in GCs for oogenesis [13]. This could be due to a redundant function by another kinase, Aduk (Ulk3 homologue in Drosophila) in the absence of Atg1. It is also possible that since basal autophagy levels are very low in GSCs loss of Atg1 is not detrimental [14]. However, it remains to be tested if combined loss of Atg1 and Aduk function is detrimental during oogenesis. We observed a significant reduction in ovary size in Atg1KD indicating that Atg1 might control growth of the ovary. A previous study shows that Atg1 is required in follicle cells during oogenesis. Thus, an important question arises "What is the precise role of Atg1 in GCs, including GSCs?". To address this, we tested if Atg1 depletion has any effect on GSC maintenance. It was demonstrated that mitochondrial fission promotes GSC loss in the wild-type. Thus, we hypothesized that the depletion of Atg1 could be beneficial for the maintenance of GSCs and may lead to their longer retention in the GSC niche. Our data support this hypothesis. We observed an increase in average GSC counts at midlife. Although, Atg1 is not necessary for oogenesis per se, it is essential for embryogenesis [13,15]. Embryos with depleted Atg1 do not complete embryogenesis, and this appears to be partly due to inefficient utilization of lipids [15]. It is not known if the disruption of embryogenesis is due to the inheritance of damaged mitochondria within the developing egg [8]. Drp1 and Marf have been reported to regulate germline stem cell maintenance, differentiation, and ovariole development for the development of vitellogenic eggs and fecundity [6,7]. Atg1 and Drp1 double RNAi did not enhance the mitochondrial fusion phenotype of Drp1, suggesting that Atg1 acts downstream of Drp1. The double KD of Drp1:Atg1 exhibited an increase in the average area of the ovary and an increased number of vitellogenic stages as compared to KD of Drp1 alone, suggesting a rescue of Drp1 phenotype ( Figure 4H). However, we observed a significant reduction in the number of GCs double KD of Drp1:Atg1, indicating that Drp1 and Atg1 together regulate GSC and GC differentiation.
Marf and Atg1 KD led to a dramatic phenotype leading to an arrested growth of ovariole, a drastic reduction in both GSC and GC number. This correlated with a significant reduction in the mitochondrial mass and GSCs were rapidly lost from the niche as early as 7 days. Our data suggest that pMad, one of the self-renewal signals, is significantly decreased in GSCs with Marf:Atg1 double KD. These data indicate that Atg1 and Marf interact genetically in a complex manner to regulate GSC maintenance through pMad signaling [7]. Thus, our data suggest a complex network formed by Atg1, Drp1 and Marf during oogenesis that connects mitophagy and mitochondrial dynamics in the maintenance of GSCs in Drosophila.

Conflict of Interest Statement
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Acknowledgments
We would like to thank Ms. Amruta Nikam for assisting with fly food preparation. Kiran Nilangekar for help with setting up qPCR. We acknowledge the confocal facility at ARI for assistance with imaging. Thanks to members of the Shravage lab for helpful discussions. We would like to thank Developmental Studies Hydridoma Bank, United States, for providing antibodies and plasmid constructs. Thanks to Prof. L. S. Shashidhara and IISER Fly community and Indian Drosophila community for providing fly stocks. Dr. Richa Rikhy, IISER Pune, for fly stocks, reagents, and helpful discussions. Dr. Manish Jaiswal, TCIS, Hyderabad for fly stocks, reagents, and valuable comments on the mitochondrial phenotypes. We would also like to thank Dr. P. K. Dhakephalkar, Director, Agharkar Research Institute, Pune, and entire ARI fraternity for support and access to facilities.

Data Availability Statement
Data available on request from the authors    Dotted ovals mark the GSCs and asterisk mark the cap cells. Scale bar-10μm. Error bars represent SD in red and the mean is represented in blue. n=20, **p < 0.05, **p < 0.01, ****p < 0.0001.         Scale bar-10μm. Error bars represent SD in red and the mean is represented in blue. n=20, **p < 0.05, **p < 0.01, ****p < 0.0001.