The fine-tuning of synapse development by oxidative stress and autophagy requires presynaptic ATM kinase

Two processes held in delicate balance during the fine tuning of synapse development are oxidative stress and autophagy: each can promote synapse expansion yet in excess are toxic. How this balance is maintained is not fully understood. While ataxia-telangiectasia mutated (ATM) is recognized as a key regulator of the DNA damage response, there is increasing evidence of a neuronal-specific role for this ubiquitous kinase and deficiency causes early-onset neurodegeneration. We report a requirement for presyna ptic Drosophila ATM (dATM) in neurodevelopment that is independent of its functions in the DNA damage response. Reduction of presynaptic dATM expression causes hypersensitivity to raised oxidative stress and a failure to induce autophagy which leaves mitochondria in excess in neurons. We demonstrate that presynaptic dATM coordinates autophagy through the conserved ATM -AMPK axis. Similar ly to mammalian ATM, neuronal dATM is predominantly cytosolic and forms synaptic foci . dATM also colocalizes with autophagosomes. We propose a model wherein dATM responds to increased reactive oxygen species resulting from heightened neuronal activity by activating autophagy to induce synaptic growth, while protecting the neuron from excitotoxicity and oxidative stress. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 25, 2024. ; https://doi.org/10.1101/2024.04.25.591136doi: bioRxiv preprint Introduc2on As synapses mature and change over `me and with new experiences, the neuron must integrate different signals which, in other contexts, can be toxic. For example, oxida`ve species such as hydrogen peroxide can be used by neurons at physiological levels as an instruc`ve signal repor`ng increased presynap`c ac`vity levels to regulate synapse homeostasis, specifically increasing presynap`c arborisa`on and modula`ng synap`c output (1,2). However, at the upper end of its physiological concentra`on range, hydrogen peroxide is toxic and causes neurodegenera`on (3). A similar story is true for macroautophagy (referred to here as autophagy), the subcellular recycling programme which digests damaged cellular components and organelles. Autophagy has been iden`fied as necessary for normal expansion of synapses during development (4) but, like oxida`ve stress, an excess of autophagy can be deleterious (5). We do not fully understand the mechanisms through which neurons balance the compe`ng need for oxida`ve stress and autophagy with the threat of toxicity when in excess. The consequences when this balancing act goes awry can be observed in numerous neurodegenera`ve disorders, including Alzheimer’s, Parkinson’s and Hun`ngton’s diseases (6–10). Ataxia-telangiectasia (A-T), which is caused by muta`ons in ATM kinase, is an early -onset neurodegenera`ve disorder in which the cerebellum is par`cularly vulnerable – although it is s`ll an open ques`on as to why (11– 13). ATM kinase is a key regulator of the DNA damage response (DDR) but has other, non-nuclear func`ons that depend upon both its subcellular localisa`on and mechanism of ac`va`on , including s`mula`on of autophagy and mitophagy (macroautophagy of mitochondria) in the cytosol following ac`va`on of dimeric ATM by reac`ve oxygen species, which ac`vate the kinase ac`vity (14–16). There is increasing evidence of a neuronal -specific role for the ATM protein unrelated to its role in the DDR . From early studies of this kinase, it was no`ced that neurons have a par`cularly large pool of cytosolic ATM compared to other cell types (17–19) and a subset of ATM co-localises with presynap`c vesicles in synapses and is required for long-term poten`a`on (20). In A -T, various lines of evidence ranging from mouse models to human cell lines indicate autophagic flux is altered as a consequence of ATM deficiency (21–23). In addi`on, ATM deficient cells are vulnerable to increased oxida`ve stress (24). Together, these point to a poten`al role for ATM regula`ng how synapses respond to these instruc`ve signals and that the neurodegenera`on in A-T may be due to a failure to balance the risk of toxicity they bring. Rodent models of A-T recapitulate the immunological, fer`lity and radiosensi`vity aspects of the disease, with some evidence of cerebellar dysfunc`on, such as poorer performance on the rota-rod test or wider foot spacing in gait analysis tests (25–27). However, mouse models of A -T do not show evidence of cerebellar neurodegenera`on – a hallmark of the human disease – or other neurodevelopmental defects (26–28). We sought to define a role for ATM in synapse development and homeostasis using the well-characterised model of the Drosophila larval neuromuscular junc`on (NMJ). We find that presynap`c ATM is required for normal synapse development and func`on . Neurons depleted of ATM fail to induce autophagy via ac`va`on of AMP kinase, resul`ng in a failure to develop to the correct size. ATM-depleted neurons are on the cusp of neurodegenera`on and hypersensi`ve to a rise in oxida`ve stress. Our data suggest that presynap`c ATM is necessary for neurons to balance the posi`ve, pleiotropic effects of autophagy and reac`ve oxida`ve species with their inherent toxicity when in excess. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 25, 2024. ; https://doi.org/10.1101/2024.04.25.591136doi: bioRxiv preprint

Drosophila stocks and maintenance. Experimental crosses were reared on standard yeast-glucose-agar food using a 12 h:12 h light:dark cycle at 25°C throughout egg -laying and larval development unless otherwise stated. The dual -colour autophagy reporter UAS-GFP-Cherry::Atg8a was a kind gij of I. Nezis, University of Warwick, UK. UAS-dATM[msGFP2] was generated in this study through synthesis of the full-length dATM cDNA containing codon-op`mised msGFP2 (GenScript, USA), sub cloning into pUAST-a@B and injec`on into the a@P-3B landing site on chromosome 2 (BestGene, USA). The following were sourced from the Bloomington Drosophila Stock Centre (BDSC): w1118 (BDSC #5905); dATM-3 (BDSC #8625); dATM-6 (BDSC #8626); dATM-8 (BDSC #8624); elav-GAL4 (BDSC #25750); mef2-GAL4 (BDSC #27390); repo-GAL4 (BDSC #7415); OK371-GAL4 (BDSC #26160); dATM[TRiP .HMS02790] (BDSC #44073); dATM[TRiP .JF01422] (BDSC #31635); d MRE11[TRiP .HMC02995] (BDSC #50628); dATR[TRiP .HMS02331] (BDSC #41934); dCHK2[TRiP .HMC05499] (BDSC #64482); cat[TRiP .JF02173] (BDSC #31894); Atg18[TRiP .HMS01193] (BDSC #34714); AMPK[TRiP .JF01951] (BDSC #25931); UAS-ATG1 (BDSC #51654 and #51655); UAS-AMPK (BDSC #32108); UAS-GC3Ai (BDSC #84301 and #84313); UAS-TrpA1 (BDSC #23263 and #26264). 3rd instar larval dissec?on: NMJs - Drosophila wandering 3rd instar -stage larval NMJ “fillet” dissec`ons were performed as described elsewhere (29). The CNS was lej intact, except for electrophysiology experiments in which it was removed to prevent spontaneous muscle contrac`ons. The samples were fixed in 4% paraformaldehyde/PBS followed by washing in PBS prior to immunostaining. CNS and salivary glands - the CNS was exposed from wandering 3rd instar larvae by gentle applica`on of tension to the mouth-hooks. The salivary glands, eye discs and any excess `ssue was then separated from the CNS. The isolated CNS and salivary glands were fixed then washed in PBS prior to immunostaining or live imaging. For irradiaWon experiments , larvae were irradiated with 8 Gy (CellRad X -ray irradiator) followed by 30 min of recovery prior to dissec`on. All dissec`ons were performed in low Ca++ HL3.1, an isotonic buffer that mimics the larval haemolymph (30). Immunohistochemistry Dissected `ssues were permeabilised for 15 min in PBT (PBS + 0.1 % v/v Triton-X100) and blocked in 1 % BSA/PBS for 1 h. The primary an`body step was performed at 4°C for 1-3 days in blocking solu`on, before washing in PBS. Primary an`bodies were as follows: mouse an`-BRP (1:100, Developmental Studies Hybridoma Bank, University of Iowa [nc82]); chicken an`-GFP (1:400, Invitrogen, Cat# A10262); mouse an`-DLG (1:200, DSHB [4F3]); Alexa- 594 goat an`-HRP (1:400, Jackson Immuno, Cat# 123-585-021). Samples were incubated in secondary an`bodies in PBS 4 hr -overnight at 4°C. Secondary an`bodies included: Alexa -488 donkey an` -mouse (1:1000, Jackson Immuno, Cat# 715-545-150); Alexa-488 an`-chicken (1:400, Thermo, Cat# A32931); ToPro3 (1:2000, Invitrogen, Cat# T3605). Finally, the prepara`ons were washed in PBS and mounted on glass slides in either Fluoromount (with DAPI) or Prolong Gold (without DAPI). Electrophysiology 3rd instar larval electrophysiology was performed as described elsewhere (31). Larvae were dissected as for the larval NMJ preps in low Ca ++ HL3.1 to minimise muscle contrac`on during dissec`on. The motor neuron axons were severed at the base of the CNS, and the CNS was removed to prevent spontaneous muscle contrac`on .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 25, 2024. ; https://doi.org/10.1101/2024.04.25.591136doi: bioRxiv preprint during recordings. The samples were washed in HL3.1 and recordings were performed in HL3.1 containing 1.5 mM Ca++. Recordings were taken from muscle 6/7 in segments A2-A5. Single-electrode current clamp recordings were performed using the bridge mode of an AxoClamp -2B amplifier with a HS -2A headstage (Axon Instruments), with s`mula`on provided by a DS2A Isolated Voltage s`mulator (Digi`mer Ltd). For the recording electrode, borosilicate glass capillaries (GC150F -10, Harvard Apparatus) were pulled using a Narishige PC-100 to a final resistance of 15-25 MΩ and filled with 3 M KCl. The s`mula`on electrode holder was constructed following a standard protocol (32). S`mula`on (suc`on) electrodes were made to a final resistance of 5 MΩ, the `p gently broken against a microscope lens `ssue, and backfilled with HL3.1 using nega`ve suc`on. The motor neuron innerva`ng the respec`ve segment was iden`fied and gentle applica`on of nega`ve suc`on brought the severed axon end into the s`mula`on electrode. Ajer inser`on of the recording electrode, various exclusion criteria were looked for: • A voltage drop to a Vm of at least -60 mV • A muscle Rin of at least 4 MΩ as measured by the voltage drop following a -1 nA current injec`on. • Correctly iden`fied segmental motor neuron – validated by manually s`mula`ng to check that an excitatory junc`on poten`al (EJP) was evoked. • Recruitment of both Ib and Is motor neuron inputs (see below). EJPs were evoked ini`ally by 200 μs s`mula`on at increasing voltages (ranging from 1-8 V) to find the minimum voltage required to recruit both Ib and Is motor neuron consistently, which could be seen by first a small EJP response at one threshold and th en a dis`nct, discrete increase with addi`on of extra voltage. Mean EJP amplitude was calculated from 10 evoked EJPs at 0.5 Hz. Mini excitatory junc`on poten`als (mEJPs) were observed by recording fluctua`ons in Vm for at least 2 minutes post-s`mula`on. Data were recorded in Spike2 v9.16. Larval locomo?on assay Individual wandering third instar larvae were transferred into a custom -made 3D printed arena with wells containing 2% agarose coloured with a small amount of Orange -G dye. Larval locomo`on while freely crawling was tracked for 5 minutes using EthovisionXT sojware. 8-12 larvae were recorded simultaneously. Only larvae with a mean speed and percentage `me moving > 0 were used for subsequent analysis. Confocal microscopy All images were taken using either a Zeiss LSM 780 or LSM 880 confocal microscope. For muscle size quan`fica`on, images were obtained using a 10X water immersion lens, using a single scan capturing at both 488 nm and 594 nm. For NMJ quan`fica`on, the N MJ on muscle 4 was imaged using a 63X water immersion lens; the images consisted of Z -stacks through the en`re NMJ with 0.25 μm step size. For dATM[sfGFP] and dATM[msGFP2] localisa`on, NMJ images were captured using a 100X oil immersion N.A. 1.46 lens. F or NMJ analysis, the resultant Zeiss raw (.czi) images were imported into FIJI for further analysis (see below). For live imaging of the GFP-mCherry-Atg8a reporter, dissected `ssues were maintained on ice in HL3.1 solu`on and transferred to 35 mm glass-bo{om dishes. CNS and salivary gland prepara`ons were allowed to sink to the bo{om these dishes prior to live imaging, which was performed on the inverted LSM 880 microscope. CNS and .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 25, 2024. ; https://doi.org/10.1101/2024.04.25.591136doi: bioRxiv preprint salivary glands were imaged using a 25X oil immersion lens, with a single track using both the 488 nm and 594 nm laser lines, with a z-step size of 2 μm. Analysis of NMJ images in FIJI Batch processing of larval NMJ images was performed using the Drosophila NMJ morphometrics plugin in FIJI (33). The surface area of muscle 4 was measured for each larva using the polygon selec`on tool and inbuilt measure func`on in FIJI. The mean muscle surface area (MSA) was then calculated for each genotype and the ra`o of this MSA to the control MSA for each experiment generated. All datapoints were then scaled using this ra`o to account for differences in muscle size between the genotypes. Autophagy quan?fica?on For quan`fica`on of autophagic flux using the GFP -mCherry-Atg8a reporter (34), z -stack maximum intensity projec`ons were analysed using a custom-made FIJI script. Essen`ally, the script would: iterate through each file in a directory; prompt the user to draw a freehand selec`on around the salivary glands; clear outside the selec`on; split the channels; process the GFP channel by thresholding using the auto thre shold mean func`on and selec`ng the salivary gland border as the region of interest (ROI); apply this ROI to the mCherry channel; automa`cally adjust contrast to a fixe d satura`on value (to ensure consistency between samples); auto threshold (using otsu); and finally analyse par`cles again to select and quan`fy the mCherry foci. The results were exported as an Excel file and imported into Rstudio for sta`s`cal tes`ng. Calcula?on of HRP-Dlg ra?o A custom FIJI script was wri{en which would iterate through max intensity projec`ons in a directory, select the HRP channel and ask the user to roughly draw a ROI around the NMJ. Auto threshold was used to detect the NMJ outline, which was then used to a utoma`cally select the en`re NMJ as a ROI. This ROI was re -applied to both the HRP and DLG channels, the mean intensity of the signal measured, and the ra`o of the two calculated. Processing of electrophysiology data Raw electrophysiology data from Spike2 v9 were analysed using custom ac`ve cursors. For detec`on and quan`fica`on of EJPs, one cursor detected the points automa`cally marked where each s`mulus was delivered, a second found the maximum value within one second of this event, while a third found the minimum. The max Vm, min Vm and difference was measured. For mEJP detec`on and quan`fica`on, an automa`c detec`on pipeline was set up. The data channel was duplicated, a low pass filter applied , and a DC removal filter applied with a `me constant of 50 ms. Finally, a smoothing filter with a `me constant of 0.65 ms was applied to the duplicated data channel to remove high frequency noise. Ac`ve cursors were then u`lised in a similar way as above, except the first cursor was set up to find peaks of at least 0.3 mV in amplitude in the DC -removed memory channel. The next cursor looked for the maximum Vm value within +/- 5 ms seconds of the original while the final cursor found the minimum V m value within 20 ms prior to the former. Frequency of mEJPs was calculated by taking the number of automa`cally detected mEJPs and dividing by the `me difference between the first and last observed mEJP . Amplitudes were corrected for differences in baseline Vm (i.e., correc`ons for non-linear summa`on) using a deriva`on of Mar`n’s rela`onship (35–37): v' = E(ln[E/(E-v)]) where: .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 25, 2024. ; https://doi.org/10.1101/2024.04.25.591136doi: bioRxiv preprint • v' = corrected amplitude • E = driving force (assumed to be equal to Vm given a reversal poten`al of 0 mV) • v = recorded amplitude Quantal content was calculated by dividing corrected EJP amplitude by corrected mEJP amplitude. Sta?s?cal analysis in Rstudio All sta`s`cal analysis and graph produc`on was performed in Rstudio v3.6.0, using the following libraries: readxl, ggplot2, dplyr, ggthemes, ggpubr, ggsignif, ggthemr, Wdyverse, ISLR, Rfast, rstaWx, ggtext, RolorBrewer, ggsci and MASS. An R script was made for each experiment type e.g., NMJ structural analysis, electrophysiology, autophagy quan`fica`on etc. If more than one datapoint was generated from one larva (i.e., the right-side NMJ vs the lej- side NMJ), the n the mean of the data for that larva was used to avoid infla`ng the n number with non - independent datapoints. Boxplots were generated using ggplot2 and sta`s`cal tests performed using func`ons within the rstaWx and multcomp packages. Data were checked for normality using the ‘shapiro.test()’ func`on. Student’s T-tests (with Welch’s correc`on) were used for pairwise comparisons of normally distributed data (Wilcoxon tests if not normally distributed), while mul`ple compari sons with the control genotype as the reference group were performed using Dunne{’s tests. If no group was selected as control (i.e., tes`ng every condi`on against every other condi`on) then Tukey’s Honest Significant Differences (parametric data) or Dunn’s tests (non-parametric data) were performed. Boxplots show individual data points. Boxes represent median plus interquar`le range (IQR). Whiskers represent 1.75x the IQR. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 25, 2024. ; https://doi.org/10.1101/2024.04.25.591136doi: bioRxiv preprint

Presynap?c dATM is required for NMJ development, expansion and func?on A-T is a loss-of-func`on disease and is classified as a neurodevelopmental and/or early-onset neurodegenera`ve disorder. The early-onset neurodegenera`on in A-T may well be underpinned by defec`ve neurodevelopment yet the role ATM plays in the development of the nervous system is not well described. Our recent work indicates that ATM may have a different role in the mature nervous system of adults, where deple`ng it from neurons can be protec`ve in neurological disease models (38), than it does during development. Drosophila is an ideal system to assess how ATM might func`on in neural development at the molecular level, par`cularly if the glutamatergic larval neuromuscular junc`on (NMJ) is used as the model, since it is amenable to a combina`on of gene`c manipula`on, high resolu`on microscopy and func`onal assays. Previous studies of Drosophila ATM (dATM) have focused predominantly on phenotypes in adult flies, such as a rough -eye phenotype or increased vacuoliza`on of brain sec`ons (39,40). A role for dATM in neural development remains unexplored. We started by asking if there are neurodevelopmental deficits in whole animal dATM mutants. dATM mutants are homozygous lethal but heterozygotes are viable and fer`le. The morphology of type Ib NMJ of body wall muscle 4 were quan`fied at the wandering 3 rd instar stage from confocal images ajer fixa`on and immunostaining. Significant reduc`ons in NMJ size, ac`ve zone number, and bouton count were observed in dATM-/+ mutant larvae compared to controls (Fig 1A), indica`ng that dATM is haploinsufficient for NMJ development. Introducing a 20 kb BAC encompassing the dATM locus into the dATM-8/+ background restored NMJ surface area and bouton count, confirming that the phenotype was due to haploinsufficiency of dATM, although ac`ve zone number was not fully restored. While confirming a requirement for neurodevelopment for dATM, these experiments do not address in which cell type dATM func`on is crucial for neurodevelopment, since the en`re animal is heterozygous. The larval NMJ is composed of the presynap`c neuron, the postsynap`c muscle, and perisynap`c glia. One key advantage of Drosophila as a model for neurodevelopment is the ability to interrogate spa`otemporal requirements of genes- of-interest through the GAL4-UAS system. We used neuronal, glial or muscle-specific Gal4 drivers to express UAS- shRNAi to knockdown dATM in a cell-type specific manner. Neuronal knockdown using elav-GAL4 with either of two different shRNA constructs phenocopied the dATM heterozygous phenotype, with marked reduc`ons in NMJ surface area, bouton number, and ac`ve zone count (Fig 1B). In contrast, neither glial nor muscle knockdown of dATM had any effect on the morphology of the NMJ (Fig 1D). This strongly suggests the requirement for presynap`c dATM during Drosophila larval NMJ development. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 25, 2024. ; https://doi.org/10.1101/2024.04.25.591136doi: bioRxiv preprint Figure 1. Presynap0c Drosophila ATM (dATM) is required for NMJ growth. (A) Representa-ve images of NMJ4b of the indicated dATM genotypes stained with an--HRP to visualised the neuronal membrane. Graphs below the images show quan-fica-on of 3 metrics (NMJ surface area, bouton count and ac-ve zone count) of control vs. 3 different heterozygous dATM null alleles plus a genomic rescue construct: 15C8;dATM-8/+, shown in pink. Tukey HSD test. (B) Representa-ve images of NMJ4b showing control or Elav-Gal4 driving two different shRNA constructs for pre-synap-c knockdown of dATM. Quan-fica-on of NMJ metrics for each genotype shown below. DunneS’s mul-ple comparisons test with elav > w1118 as the reference group. (C) Quan-fica-on of NMJ features from a screen of presynap-c (neuronal), postsynap-c (muscle) and perisynap-c (glia) dATM knockdown. DunneS’s mul-ple comparisons test with control as the reference group (Gal4>w1118 for each driver). Individual data points are shown with boxes represen-ng the median and interquar-le range. p≤0.05 *, p≤0.01 **, p≤0.001 ***, p≤0.0001 ****, ns = not significant. Scale bars = 10 μm. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 25, 2024. ; https://doi.org/10.1101/2024.04.25.591136doi: bioRxiv preprint To assess whether the structural developmental deficits we see have func`onal consequences for the neuron and for the animal, we recorded from the intersegmental neuron which innervates muscle 4 and, since the NMJs innervate the larval body wall muscles, we measured speed of crawling. Pan-neuronal dATM knockdown resulted in a significant decline in excitatory junc`on poten`al (EJP) and miniature EJP (mEJP) amplitudes (Fig 2A and 2B), while no overall change in mEJP frequency was observed (Fig 2C). There was also a small but significant decrease in quantal content (Fig 2D), and an increase in paired -pulse ra`o (PPR, Fig 2F), likely due to a reten`on of the readily-releasable pool of synap`c vesicles caused by an overall reduc`on of the amplitudes of ea ch EJP in the PPR test. Consistent with an electrophysiological deficit, dATM knockdown larvae crawled at reduced speed compared to controls (Fig 2H). There was, however, no overall change in behaviour of the knockdown larvae – they spend the same propor`on of `me moving as the controls (Fig 2I). Figure 2. Presynap0c dATM is required for NMJ func0on and larval locomo0on . (A-F) Quan-fica-on of electrophysiology parameters: (A) Corrected EJP amplitude; (B) Corrected mEJP amplitude; (C) mEJP frequency; (D) Quantal content; (E) Muscle resistance; (F) Paired-pulse ra-o. (G) Representa-ve electrophysiological traces of evoked EJPs from the indicated genotypes. (H-I) Quan-fica-on of larval locomo-on parameters: (H) Mean larval crawling speed; (I) Mean percentage -me moving during experiment. In all plots, dATMKD = presynap-c dATM knockdown. All p values from Student’s T tests with Welch’s correc-on, p≤0.05 *, p≤0.01 **, p≤0.001 ***, p≤0.0001 ****, ns = not significant. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 25, 2024. ; https://doi.org/10.1101/2024.04.25.591136doi: bioRxiv preprint Neurons deficient in dATM fail to expand during development and show signs of degenera?on at higher rearing temperatures Neuronal ac`vity plays a crucial role in shaping the forma`on and plas`city of connec`ons in developing nervous systems, including the larval neuromuscular system. As Drosophila are poikilothermic, external temperature substan`ally affects their development, aging, and ac`vity. Higher temperatures shorten their developmental period and enhance mobility, evident in both larval and adult stages. Consequently, there is an increase in neuronal ac`vity with increased temperature, and a concomitant increase in NMJ size and arborisa`on (41). To test the dependence of the neuronal dATM knockdown phenotype on developmental rearing temperature, presynap`c dATM knockdown was repeated with rearing temperatures of 19°C or 27°C (Fig 3A). Consistent with the work of others , e.g. (41,42) we observed that in the controls, increasing rearing temperature significantly increased the size and bouton number of NMJs, although ac`ve zone count appears to be unaffected by rearing temperature in our experiments. At both low and high rearing temperatures, presynap`c dATM knockdown

in significant deficits in NMJ development when compared to the controls: the knockdown larvae completely failed to expand their NMJs in response to increasing rearing temperature (Fig 3A). It appears that neuronal dATM may be required to sense the increased ac`vity, act on that signal and respond appropriately by driving expansion. Higher ac`vity levels in neurons places them under increased stress, par`cularly oxida`ve stress. Ataxia- telangiectasia features early-onset neurodegenera`on, principally in the cerebellum which causes the ataxia in children. We were interested to see if deple`ng dATM in neurons would destabilise them at higher temperatures and lead to signs of degenera`on. A marker of neurodegenera`on in the third instar larval model is withdrawal of the presynap`c membrane from the postsynap`c density, which remains in the muscle as a synap`c “footprint” (43). We shijed the flies from to the usual rearing temperature of 25 °C to 27 °C and knocked down dATM using the OK371 driver which drives in glutamatergic neurons, including the motor neurons innerva`ng the body wall muscles. In dATM knockdown animals reared at 27°C, there were no notable examples of DLG staining which completely lacked the cognate HRP signal. However, there was a clear reduc`on in intensity of the pre-synap`c HRP signal corresponding to the neuronal membrane (Fig. 3B,C) and a clear decrease in the ra`o of pre to postsynap`c signal in dATM knockdown neurons (Fig. 3C’), poten`ally an early indica`on of neurodegenera`on. We looked for localised caspase ac`vity in the motor neuron terminals, which might underpin retrac`on of the neuron. We expressed the GC3Ai system in motor neurons as a reporter of (44). GC3Ai u`lizes GFP molecules connected at the C- and N-termini by a DEVD caspase cleavage site. Without caspase ac`vity, the linker keeps GFP fluorescence suppressed, though the protein can s`ll be localised with GFP an`bodies. When ac`vated , caspases cleave the linker and GFP fluorescence is restored. Endogenous GFP ac`vity is therefore indica`ve of caspase ac`vity. In the long motor neurons innerva`ng the posterior segments, endogenous GFP fluorescence was seen in boutons of the NMJ when dATM was knocked down, but not in controls (Fig. 3D; controls shown in supplementary results). Altogether, these data suggest that, at higher rearing temperatures, dATM knockdown neurons are on the cusp of neurodegenera`on. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 25, 2024. ; https://doi.org/10.1101/2024.04.25.591136doi: bioRxiv preprint Figure 3. dATM-depleted neurons show signs of neurodegenera0on at higher rearing temperatures . (A) Quan-fica-on of NMJ features in control vs. presynap-c dATM knockdown larvae at low (19°C) and high (27°C) rearing temperatures. Tukey HSD test. (B) Representa-ve images of control vs. presynap-c dATM knockdown NMJs at 27°C rearing temperature. Note the thinning fragmenta-on of the presynap-c membrane as visualised by HRP staining. (C) Representa-ve images of the same genotypes from B co-stained with presynap-c (HRP) and postsynap-c (DLG) markers. (C’) Quan-fica-on of the ra-o of HRP to DLG staining from the indicated genotypes. P values from Student’s T tests with Welch’s correc-on. Individual data points are shown with boxes represen-ng the median and interquar-le range. p≤0.05 *, p≤0.01 **, p≤0.001 ***, p≤0.0001 ****, ns = not significant. (D) UAS- GC3Ai expression in posterior motor neurons in presynap-c dATM knockdown larvae. Le8 panel: immunoreac-vity from ⍺-GFP staining indicates expression of the GC3Ai reporter; righ panel: endogenous GC3Ai fluorescence indica-ng ac-vated caspase. Boxed sec-on is shown in 2x zoom in the inset. Arrows indicate sites of localised caspase ac-vity. Scale bars = 10 μm. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 25, 2024. ; https://doi.org/10.1101/2024.04.25.591136doi: bioRxiv preprint Localisa?on of neuronal GFP-dATM In mammals, neuronal ATM ojen shows strong cytosolic localisa`on and has been detected in synapses , colocalising with markers of presynap`c vesicles (20,45). Previous studies looking at dATM localisa`on overexpressed a FLAG-dATM construct in Drosophila S2 cells, revealing a predominantly nuclear localisa`on and foci forma`on upon irradia`on (46). Given this study was restricted to in vitro overexpression in a non-neuronal cell type, the specific neuronal localisa`on of dATM in Drosophila remains unclear. To address this gap in knowledge, a msGFP2 -tagged full-length dATM cDNA was synthesised and cloned into pUAST-a{B. msGFP2 was selected for its stability in oxidising condi`ons and the presence of point muta`ons in its dimeriza`on interface (47). This la{er point was crucial since if dATM’s func`on is orthologous to hATM, it may have differing downstream targets based on its dimeriza`on state. Thus, it was essen`al to prevent ar`ficial dimeriza`on caused by GFP-GFP interac`on. UAS-dATM[msGFP2] expression was driven in glutamatergic neurons with OK371-GAL4. GFP fluorescence was visible in both the CNS and salivary glands (where OK371-GAL4 drives off-target expression). Detailed examina`on revealed strong GFP expression in midline motor neuron cell bodies, peripheral motor neurons, and extending along axons from the ventral nerve cord (Fig. 4A). High magnifica`on confirmed that dATM[msGFP2] expression is predominantly cytosolic in the motor neuron cell bodies (Fig . 4B). However, irradia`on of larvae with X-irradia`on which generates double-strand breaks in the DNA leads to a relocaliza`on of dATM[msGFP2] into the nucleus, consistent with the role of ATM in the DDR (Fig. 4B). In the periphery, OK371-driven dATM[msGFP2] expression appears more punctate than diffuse (Fig . 4C). The axon exhibits bright dATM[msGFP2] puncta, which become more pronounced in distal segments compared to regions proximal to the VNC. There also appears to be foci of dATM[msGFP2] external to the HRP stain of the axon, poten`ally indica`ng shu{ling into the ensheathing glia. At the NMJ, presynap`c boutons display a notably higher density of msGFP2 puncta than the inter -bouton space (Fig. 4C, white arrowheads). Interes`ngly, low- level, punctate fluorescence was seen in the large nuclei of the muscle cells , possibly from some leaky expression. Taken together, neuronal dATM appears primarily cytosolic and localises to axonal and presynap`c puncta. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 25, 2024. ; https://doi.org/10.1101/2024.04.25.591136doi: bioRxiv preprint Figure 4. Motor neuron expression of msGFP2-tagged dATM. (A) Larval ventral nerve cord showing UAS-dATM[msGFP2] expression driven in glutamatergic neurons by OK371-GAL4. Scale bar = 20 μm. (B) Higher magnifica-on of a single motor neuron cell body where dATM[msGFP2] expression is predominantly cytosolic under basal condi-ons but some relocates to the nucleus aker irradia-on with 8 Gy of X -ray irradia-on (right). Scale bar = 1 μm. (C) dATM[msGFP2] expression becomes increasingly punctate at regions distal to the motor neuron cell body: Top – lower magnifica- on image of en-re muscle 4 NMJ of OK371-GAL4 driven dATM[msGFP2] expression, scale bar = 10 μm; Middle – higher magnifica-on of axon, scale bar = 5 μm; Lower – higher magnifica-on image of NMJ terminal, puncta of GFP appear to be concentrated within boutons (arrowheads). Scale bar = 5 μm. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 25, 2024. ; https://doi.org/10.1101/2024.04.25.591136doi: bioRxiv preprint The neurodevelopmental role of presynap?c dATM is independent of the DNA damage response Given ATM's key role in the DNA damage response (DDR) to double -stranded DNA breaks (DSBs), we hypothesized that the NMJ phenotype from presynap`c dATM knockdowns might stem from general DDR misregula`on, rather than a specific involvement of dATM in neurodeveloment per se . To explore this, the expression of two key components of the DDR upstream and downstream of ATM was knocked down in neurons using the pan neural elav-GAL4 driver to express a TRiP dsRNA construct. Specifically, the Drosophila homologs of the MRN complex component , MRE11, and the downstream checkpoint kinase 2 (CHK2 , loki in Drosophila) were targeted. However, no detectable changes in the NMJ's surface area or ac`ve zone count were observed compared to controls (Fig 5A. and 5B). Further, since there is evidence elsewhere showing an interac`on between neuronal ATM and its sister kinase, ATR, at synapses, we wanted to ask whether knockdown of dATR (mei-41 in Drosophila) would replicate the effect of dATM knockdown. However, as with the other DDR components, we observed no significant difference in NMJ surface area or ac`ve zone count compared to controls (Fig. 5A and 5B). Given that targe`ng of upstream, downstream, or parallel components of dATM do nor replicate the effect of dATM knockdown, and that the dATM protein is primarily cytosolic in neurons, it seems likely that the role for dATM in neurodevelopment is independent of its role in the DDR . This suggested to us that key to understanding the developmental role of presynap`c dATM lay with its cytosolic pathways and interac`ons. Presynap?c dATM knockdown sensi?ses larvae to excitotoxicity and oxida?ve stress In addi`on to its role in the DDR, ATM is a key player in oxida`ve stress signalling. Specifically, it is known that cytosolic, dimeric ATM can be directly oxidised by ROS, leading to an intermolecular disulphide bond forming at Cys 2991 and conversion of the dimer into an ac`ve state, with dis`nct downstream targets outside of the DDR (14,48). Given the signs of early degenera`on in dATM-depleted neurons, we considered that this may be caused by impaired oxida`ve stress signalling. In Drosophila, there is increasing evidence that ROS signalling regulates larval NMJ development and plas`city. For example, spinster mutants display elevated ROS levels and expanded NMJs, which can be rescued through increased ROS scavenging (49). Expression of the temperature -gated ca`on channel, TrpA1 , with a rearing temperature ≥25°C hyper-ac`vates neurons and results in increased mitochondrial ROS and consequent NMJ overgrowth. Consistent with this, reducing an`oxidant capability of neurons through catalase knockdown phenocopies this effect (1). However, excessive oxida`ve stress can lead to neurodegenera`on (50) and thus a delicate balance must be maintained. We hypothesised that the dATM knockdown phenotype could be a failure to sense normal changes in ROS or ac`vity levels during NMJ matura`on, resul`ng in a failure for the NMJ to expand appropriately. This could poten`ally be overcome through hyperac`va`ng the neuron with TrpA 1 throughout development to compensate. We combined knockdown of dATM in motor neurons with overexpression of the TrpA1 and reared the larvae at 27 °C, which result s in tonic neuronal firing (51). In larvae overexpressing TrpA1 without dATM knockdown, this produced a significant expansion of the NMJ (Fig . 5D), as reported previously (1) although no changes to ac`ve zone count were observed (Fig . 5E). In marked contrast, co-expression of TrpA1 with dATM knockdown was lethal before late larval stage , sugges`ng that dATM knockdown sensi`ses neurons to excitotoxicity. To explore whether this was the result of an increased ROS burden, we combined dATM and catalase knockdown. As reported previously, catalase knockdown at the standard 25 °C rearing temperature leads to an expansion of .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 25, 2024. ; https://doi.org/10.1101/2024.04.25.591136doi: bioRxiv preprint the NMJ due to the decreased ROS scavenging capacity (Fig. 5G,H). However, raising the temperature to 30 °C to increase neuronal ac`vity results in a significant undergrowth of the NMJ (Fig. 5G,H), poten`ally because the combina`on of increased ROS from ac`vity and reduced scavenging passes the threshold for toxicity. Here, as with TrpA1 overexpression, the combina`on of catalase and dATM knockdown is lethal, indica`ng that dATM- depleted neurons cannot cope with a reduc`on in the ROS-scavenging machinery. Taken together, these results indicate that dATM-depleted neurons are hypersensi`ve to oxida`ve stress. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 25, 2024. ; https://doi.org/10.1101/2024.04.25.591136doi: bioRxiv preprint Presynap?c dATM interacts with the autophagy machinery in NMJ development We noted that the dATM null heterozygotes and neuronal knockdowns phenocopied muta`ons and knockdowns of the autophagy machinery, which is itself another key regulator of NMJ development in Drosophila; atg1 mutants show significant underdevelopment of the NMJ, while synap`c overgrowth can be induced by overexpressing ATG1 in neurons (4). We had also noted that the localiza`on of dATM[msGFP2] in neurons i.e. a diffuse cytosolic localiza`on in the soma with discrete punctate along the axon and extending into the synapse, is consistent with the known localisa`on of neuronal autophagosomes. These typically form distal to the soma in axons and are transported retrogradely along the axons by dynein motors (52,53). Given the established role of cytosolic ATM linking oxida`ve stress signalling to autophagy (15,16), we therefore wanted to inves`gate whether dATM knockdowns would have altera`ons in autophagic flux, and whether any interac`on with the autophagy machinery would be observed. To quan`fy macroautophagic flux, we u`lised the tandem GFP-mCherry::Atg8a “traffic light” reporter, overexpressed in glutamatergic neurons using the OK371-GAL4 driver. This reporter relies o n the rela`ve pH sensi`vi`es of its cons`tuent fluorescent proteins. Atg8a localises to autophagosomes, where both GFP and mCherry are fluorescent, forming yellow puncta. Upon autophagosome matura`on into autolysosomes, the acidic environment results in quenching of the GFP signal, while mCherry fluorescence is ma intained, resul`ng in red puncta (34,54). To induce autophagy, feeding-stage third instar larvae were removed from the food and starved of amino acids for 4 h in a 20% sucrose solu`on before the ventral nerve cords were dissected for live imaging. Surprisingly the GFP signal remained diffuse although the mCherry signal was punctate, as expected (Fig. 6A). We used a custom FIJI rou`ne to measure the intensity of mCherry fluorescence in puncta plus the ra`o of GFP:mCherry fluorescence in each. This reported a lower intensity of mCherry fluorescence in the dATM knockdown neurons (Fig. 6B) and a considerably higher ra`o of GFP:mCherry, which we interpret to represent a failure of the autophagosomes to mature into autolysosomes (Fig. 6B). Imaging of autophagosomes in the CNS neurons proved difficult so for confirma`on of an autophagy deficit, we took advantage of the off-target driving of the Atg8 traffic light reporter by OK371-GAL4 in the salivary glands. In the fed state, both control and dATM knockdown salivary glands show diffuse GFP and mCherry signals. Ajer 4 h starva`on, significant numbers of mCherry+ puncta were visible in the control salivary gland cells, indica`ng increased autophagic flux. However, these were conspicuously absent from dATM knockdown cells (Fig. 6C,D) confirming that dATM knockdown cells are unable to induce autophagic flux in response to starva`on. Figure 5. (A-B) Presynap-c knockdown of other DNA damage response components has no effect on the structural development of the NMJ as measured by (A) NMJ surface area or (B) ac-ve zone count. DunneS’s mul-ple comparisons test with Control as the reference group. (C-E) Presynap-c overexpression of the temperature-gated ca-on channel TrpA1 leads to expansion of the NMJ but is lethal in combina-on with dATM knockdown at 27°C rearing temperature. (C) representa-ve muscle 4 NMJ images of the indicated genotypes; (D) NM J surface area quan-fica-on; (E) ac-ve zone count. Tukey HSD test. (F -H) Decreased ROS scavenging through catalase knockdown leads to NMJ expansion at moderate rearing temperatures (25°C) and reduced NMJ growth at high rearing temperatures (30°C), where it is lethal in combina-on with dATM knockdown. (F) representa-ve muscle 4 NMJ images of the indicated genotypes; (G) NMJ surface area; (H) ac-ve zone count. DunneS’s mul-ple comparisons test with Control as the reference group. p≤0.05 *, p≤0.01 **, p≤0.001 ***, p≤0.0001 ****, ns = not significant. All scale bars = 10 μm. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 25, 2024. ; https://doi.org/10.1101/2024.04.25.591136doi: bioRxiv preprint We next asked whether dATM may be present in autophagosomes. We expressed dATM[msGFP2] with OK371- GAL4 and starved the larvae for 4 h , as before. We could see detect clear colocaliza`on of GFP-dATM with the autophagosome marker, an`-GABARAP (Fig. 6E) in the salivary gland cells. Our results pointed to a requirement for dATM for the induc`on of autophagy but we sought to strengthen this

by looking for gene`c interac`ons between dATM and the autophagy machinery. Consistent with a previous study of autophagy in NMJ development (4), w e found that presynap`c knockdown of Atg18 phenocopies presynap`c dATM knockdown, and that overexpression of ATG1 from a weaker UAS-line leads to significant synapse expansion (Fig. 6F,G). However, strong overexpression of ATG1 is lethal: as with oxida`ve stress, the levels of autophagy appears to be held in a delicate balance during synapse development . Significantly, combining dATM knockdown with strong ATG1 overexpression is no longer lethal and rescues the synapse development deficits of dATM knockdown larvae (Fig . 6F,H). These data are a clear indica`on of an interac`on between dATM and the induc`on of the autophagy machinery. ATM has been reported previously to s`mulate mitophagy via PINK1 /Parkin (55–57). We asked whether mitophagy was affected in motor neurons by knockdown of dATM driving expression of the reporter, UAS - mitoGFP , concurrently with shRNA to dATM. Mitochondria in the NMJ were then counted using a FIJI rou`ne . Knockdown of dATM resulted in a significant increase in mitochondrial density, indica`ng that mitophagy in these neurons may indeed be defec`ve (Fig. 6I,J). A failure in mitophagy has the poten`al to increase oxida`ve stress since defec`ve mitochondria are not recycled. This may contribute to the suscep`bility of the dATM knockdown neurons to increased ROS, seen earlier (Fig. 5). Finally, we asked whether chemical induc`on of autophagy could rescue the locomotor deficit exhibited by the dATM knockdown larvae. We supplemented food with 5 mM me‚ormin, a potent inducer of macroautophagy thought to func`on via ac`va`on of AMP kinase (58). Interes`ngly, me‚ormin supplementa`on significantly diminished locomotor func`on of control larvae compared to larvae raised on standard food (Fig. 6K), However, me‚ormin was beneficial to the locomotor performance of presynap`c dATM knockdown larvae whose crawling speed was raised to the same level as the me‚ormin-fed control larvae. (Fig. 6K). These results further support the no`on that autophagy levels must be held in delicate balance and that presynap`c dATM is required for neurons to maintain this balance. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 25, 2024. ; https://doi.org/10.1101/2024.04.25.591136doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 25, 2024. ; https://doi.org/10.1101/2024.04.25.591136doi: bioRxiv preprint AMP kinase acts downstream of pre-synap?c dATM to induce autophagy If dATM is ac`ng downstream of an oxida`ve stress signal to regulate autophagic flux and expand the synapse, we reasoned that this may be occurring through the canonical ROS -ATM-AMP kinase (AMPK) signalling axis previously iden`fied in mammalian cell studies (15). We overexpressed or knocked down AMPK in neurons with, or without, concurrent dATM knockdown. Interes`ngly, AMPK knockdown alone did not significantly alter NMJ development (Fig. 7A -C) and the combin a`on of AMPK and dATM knockdown phenocopied the of dATM knockdown alone and the NMJs failed to expand. There was , however, a marked fragmenta`on of the presynap`c membrane in double knockdown NMJs with o ccasional bright spots, reminiscent of “retrac`on bulbs” seen with degenera`ng neurons. AMPK overexpression resulted in a small but non-significant increase in NMJ size (Fig. 7D-F) and overexpression combined with dATM knockdown rescue the undergrowth phenoptype and expanded the synapse to a significant degree vs. controls (Fig. 7D-F). These gene`c epistasis experiments are consistent with dATM ac`ng through AMPK to induce autophagy in response to increasing ROS levels in NMJ development, thereby regula`ng synapse expansion. Figure 6. The interac0on of dATM with autophagy. (A,B) Expression of the tandem GFP-mCherry::Atg8 reporter of autophagic flux in feeding-stage larvae starved for 4 h. The reporter is expressed in glutamatergic neurons of the ventral nerve cord under the control of OK371-Gal4 and imaged live . GFP fluorescence reports early autophagosomes but remains diffuse. mCherry reports both autophagosomes and low pH late autophagolysosomes and is punctate. The fluorescence intensity of mCherry puncta is significantly reduced by knockdown of dATM (B) and the ra-o of GFP (488) to mCherry (594) fluorescence is signifi cantly increased (B). (C,D) Localisa-on of the mCherry fluorescence of the tandem reporter expressed and imaged live in salivary gland cells. mCherry fluorescence is diffuse in salivary gland cells from feeding-stage larvae (C, upper panels). Aker 4 h starva-on mCherry+ puncta are present in Control cells represen-ng induc-on of autophagy but not in dATM knockdown cells where fluorescence remains diffuse (C, lower panels). mCherry+ puncta are quan-fied rela-ve to area of the salivary gland cells in (D). (E) dATM -sfGFP (green) co -localises with autophagosomes labelled with an- -GABARAP (magenta) in starved salivary gland cells. Arrowheads point to examples of colocalized foci. (F-H) Gene-c interac-ons between dATM and components of the autophagy machinery. Knockdown of atg18 leads to undergrowth of the NMJ and phenocopies knockdown of dATM. Overexpression of ATG1 leads to overgrowth but concurrent overexpression of ATG1 with knockdown of dATM restores the NMJ to the size of Controls. (F) Representa-ve muscle 4 NMJ images of the indicated genotypes; (G) NMJ surface area; (H) ac-ve zone count. (I,J) Mitochondrial density increases in the NMJ aker dATM knockdown. (I) Representa-ve images of NMJ4 of Control (upper row) and dATM knockdown (lower row) expressing the mitoGFP reporter then stained for HRP to visualize the neuronal membrane and GFP for mitochondria. (J) Quan-fica-on of NMJ surface area, boutons number per NMJ and the density of mitochondria per µm2 of NMJ surface. (K) Supplementa-on of food with 5 mM meqormin which induces autophagy via ac-va-on of AMP kinase significantly reduces the locomo-ve speed of Control larvae but increases the speed of dATM knockdown larvae to the same speed as Controls. Percentage -me larvae spend moving is not affected by meqormin supplementa-on. p<0.0 5 *, p≤0.01 **, p≤0.001 ***, p≤0.0001 ****, ns = not significant. Scale bars = 40 μm (A), 30 μm (C), 40 μm (E) and 10 μm (I). .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 25, 2024. ; https://doi.org/10.1101/2024.04.25.591136doi: bioRxiv preprint Figure 7. AMP kinase acts downstream of dATM to regulate NMJ development. (A-C) Single and concurrent presynap-c knockdowns of dATM and AMPK. Presynap-c AMPK knockdown alone has no impact on NMJ structural development. Concurrent knockdown of AMPK with dATM phenocopies single dATM knockdown: (A) Representa-ve muscle 4 NMJ images of the indicated genotypes; (B) NMJ surface area; (C) ac-ve zone count. Tukey HSD test. (D-F) Overexpression of AMPK singly and concurrently with dATM knockdown. Overexpression of AMPK has a non -significant effect on NMJ development. In combina-on with dATM knockdown NMJs expand significantly: (D) Representa-ve muscle 4 NMJ images of the indicated genotypes; (E) NMJ surface area; (F) ac-ve zone count. Tukey HSD test. p≤0.05 *, p≤0.01 **, p≤0.001 ***, p≤0.0001 ****, ns = not significant. Scale bars = 10 μm. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 25, 2024. ; https://doi.org/10.1101/2024.04.25.591136doi: bioRxiv preprint

Muta`ons in ATM kinase result in the progressive early-onset neurodegenera`ve disorder ataxia-telangiectasia (A-T), although the underlying disease mechanism is not well understood (59,60). ATM has well-characterized nuclear func`ons in the DNA damage response but there is debate around the extent to which neuronal ATM is primarily cytosolic or nuclear, and thus which of its nuclear vs. cytosolic func`ons are most relevant to the vulnerability of neurons in A-T (18,19,61). In addi`on, there is an increasing understanding that ATM may have unique roles in neurons compared to other cell types (20,45,62). Here, we have demonstrated that the Drosophila homologue, dATM, is required specifically presynap`cally for normal synapse development, func`on and homeostasis. The structural and func`onal phenotypes of both dATM null heterozygosity and presynap`c dATM knockdown are indica`ve either of an under-grown synapse which has failed to respond to growth signals, or of a synapse in the early stages of degenera`on. The failure of dATM-deficient motor neurons to expand in response to increases in developmental rearing temperature suggests a failure of the neuron to transduce ac`vity - dependent growth signals (42). However, these larvae were also vulnerable to ar`ficial chronic neuronal over - ac`va`on or a reduc`on in an`oxidant protec`on. Further, at higher rearing temperatures, there were indica`ons of presynap`c retrac`on from the postsynap`c density, and local caspase ac`vity within boutons, sugges`ng that these neurons are also on the cusp of degenera`on. As a neuron matures, it con`nually processes signals from numerous interconnected pathways, influencing its synap`c connec`vity, strength, and structure. Our study has concentrated on the processes of autophagy and oxida`ve stress signalling; both pathways are known to posi`vely regulate synapse expansion in Drosophila, yet when overly ac`ve, are detrimental to the health of neurons (1,4,49). Similarly, neuronal ac`vity is essen`al for normal synapse development and maintenance, but excessive ac`vity leads to excitotoxicity (41,42) which highlights the delicate balance the neuron faces during development and homeostasis. Our findings suggest that neuronal dATM plays a key role in transducing ROS-autophagy signalling and in maintaining a balance between these interconnected processes. For instance, presynap`c dATM deple`on sensi`zed larvae to excitotoxicity and decreased ROS-scavenging yet was protec`ve against chronic autophagy upregula`on. When autophagy was pharmacologically induced in control larvae, their locomotor performance declined, but this interven`on proved beneficial for larvae with presynap`c dATM knockdown. This aligns with other Drosophila research, both in neuronal and non -neuronal `ssues, indica`ng an op`mal level of autophagy in promo`ng lifespan and health (63,64). This finding also correlates with broader mammalian studies underscoring the importance of balanced autophagy for maintaining neuronal health (6). Enhanced autophagy has been shown to help clear toxic proteins that aggregate in disorders like Parkinson’s (65) and Alzheimer’s disease (66). However, excessive autophagy can itself result in neurodegenera`on in different contexts (5,67). In our model, p resynap`c ATM responds to local ROS produc`on generated through neuronal ac`vity by ac`va`ng the autophagic machinery through the conserved ATM-AMPK axis (15): an increase in neuronal ac`vity s`mulates expansion, a reduc`on in ac`vity causes the reverse (4,41). This pathway may then be held in balance with other redox -sensi`ve ATM pathways, such as p53 -mediated pro -apopto`c signalling, upregula`on of mitophagy through Parkin, or promo`on of the pentose phosphate pathway (PPP) to buffer oxida`ve stress . If ATM is a nexus for redox signalling , this would explain why dATM-depleted neurons are vulnerable to excitotoxicity or decreased ROS scavenging. Other work has shown redox-ac`vated mammalian ATM upregulates the PPP via Hsp70 phosphoryla`on, increasing an` oxidant capability (68,69), so this is a poten`al mechanism .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 25, 2024. ; https://doi.org/10.1101/2024.04.25.591136doi: bioRxiv preprint for how ATM ac`va`on at synapses could provide local homeosta`c feedback to buffer ROS levels and prevent toxicity. Further epistasis work overexpressing or knocking down Drosophila PPP components may help to elucidate the precise nature of this feedback mechanism. There is some controversy about the role of autophagy in the ae`ology of the A-T. For example, neuronal precursor cells derived from A-T pa`ents exhibit impaired autophagic flux and disrupted mitophagy (23), while pharmacological inhibi`on of autophagy rescued survival and synapse loss of ATM -deficient mouse cor`cal neurons (22). Given that the mouse A-T model lacks cerebellar degenera`on, the fact that our Drosophila model recapitulates the deficient autophagic flux of human A -T neuronal precursors and shows evidence of neurological altera`ons suggest that it could prove to be a useful screening tool to iden`fy poten`al treatments that s`mulate autophagic flux. While other studies have used Drosophila to examine the effects of dATM muta`ons and knockdowns on the structure of the adult brain, adult locomo`on, and lifespan (70,71), this is the first to inves`gate the consequences on neurodevelopment and func`on. We believe this approach shows greater relevance to the progression of A -T, given its early-onset nature (12) and evidence of developmental pa{erning defects in the cerebellar architecture (72). With adult flies, there is a risk of failing to dissociate between the role of dATM in neural progenitors or in coordina`ng some aspect of neurodevelopment in metamorphosis, which would be dis`nct from mammalian ATM. Inconsistency in the literature exists as to the specific consequence of neuronal knockdown of dATM: some studies describe photoreceptor degenera`on and temperature-dependent lethality (71); others report that it is glia which are vulnerable to dATM-deficiency and not neurons (40). Recent work in our lab has demonstrated that limi`ng dATM knockdown to adult neurons was neuroprotec`ve and extended lifespan in different Drosophila models of neurodegenera`ve disorders (73). It seems likely that ATM plays different roles in cycling vs. non-cycling cells of the nervous system and in developing vs. matured neurons. We found that neuronal knockdown of other DDR components did not recapitulate the dATM knockdown phenotype. While muta`ons in DDR proteins, such as components of the MRE11 -Rad50-NBS1 complex, are associated with microcephaly (74–76) and neurodegenera`on (77,78), a probable mechanism for the pathology in these DDR -related condi`ons is the death of neuronal precursors. This should not be a factor in our experiments; the shRNA to each component is almost exclusively being expressed in differen`ated, post-mito`c neurons. Clearly there is a di ssocia`on of the necessity of different DDR proteins in neuronal precursors vs. differen`ated neurons, especially given our recent findings that knockdown of DDR components in a mature nervous system can be neuroprotec`ve (38,73). Neurons expressing dATM[msGFP2] showed a predominantly cytosolic localiza`on of GFP. which became more nuclear only ajer DNA damage was induced . This localiza`on, coupled with the epista`c interac`ons of ATM with the autophagy machinery and AMPK we demonstrate here, supports the idea that the extranuclear, redox- dependent signalling pathways of ATM are cri`cal for its func`ons in neurons. We have a growing understanding of the unique role for cytosolic ATM in neurons, separable from the DDR. This includes the physical associa`on of ATM with synap`c vesicle proteins VAMP2 and synapsin-I (45), the regula`on of excitatory vs. inhibitory neurotransmi{er release (62), its role in LTP (20), and associa`on with mitochondria and concomitant regula`on of mitophagy (79). Our results show that cytosolic ATM is cri`cal for neurodevelopment, ac`ng to regulate the homeosta`c expansion of synapses in response to changes in ac`vity. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 25, 2024. ; https://doi.org/10.1101/2024.04.25.591136doi: bioRxiv preprint

The authors would like to thank Dr Ioannis Nezis for sharing the GFP-mCherry-Atg8 reporter line and for advice on visualising autophagy and the West Midlands Drosophila community for support and advice. MJT was funded by the Biotechnology and Biological Sciences Research Counci l Midlands Integra`ve Biosciences Training Partnership. Conflict of Interest Statement The authors declare no conflicts of interest.

1. Oswald MC, Brooks PS, Zwart MF, Mukherjee A, West RJ, Giachello CN, et al. ReacBve oxygen species regulate acBvity-dependent neuronal plasBcity in Drosophila. Davis GW, Marder E, editors. eLife. 2018 Dec 12;7:e39393. 2. Sobrido-Cameán D, Oswald MCW, Bailey DMD, Mukherjee A, Landgraf M. AcBvity-regulated growth of motoneurons at the neuromuscular juncBon is mediated by NADPH oxidases. Front Cell Neurosci. 2023 Jan 13;16:1106593. 3. Sies H, Jones DP . ReacBve oxygen species (ROS) as pleiotropic physiological signalling agents. Nat Rev Mol Cell Biol. 2020 Jul;21(7):363–83. 4. Shen W , Ganetzky B. Autophagy promotes synapse development in Drosophila. J Cell Biol. 2009 Oct 5;187(1):71–9. 5. Liu Y , Levine B. Autosis and autophagic cell death: the dark side of autophagy. Cell Death Differ. 2015 Mar;22(3):367– 76. 6. Cherra SJ, Chu CT. Autophagy in neuroprotecBon and neurodegeneraBon: A quesBon of balance. Future Neurol. 2008 May;3(3):309–23. 7. Fracassi A, Marcae M, Zolochevska O, Tabor N, Woltjer R, Moreno S, et al. OxidaBve Damage and AnBoxidant Response in Frontal Cortex of Demented and Nondemented Individuals with Alzheimer’s Neuropathology. J Neurosci. 2021 Jan 20;41(3):538–54. 8. Domise M, Sauvé F, Didier S, Caillerez R, Bégard S, Carrier S, et al. Neuronal AMP-acBvated protein kinase hyper- acBvaBon induces synapBc loss by an autophagy-mediated process. Cell Death Dis. 2019 Mar 4;10(3):1–15. 9. Dias V, Junn E, Mouradian MM. The Role of OxidaBve Stress in Parkinson’s Disease. J Parkinsons Dis. 2013;3(4):461– 91. 10. Kumar A, Ratan RR. OxidaBve Stress and HunBngton’s Disease: The Good, The Bad, and The Ugly. J HunBngtons Dis. 2016 Oct 1;5(3):217–37. 11. Boder E, Sedgwick RP . ATAXIA-TELANGIECTASIA: A Familial Syndrome of Progressive Cerebellar Ataxia, Oculocutaneous Telangiectasia and Frequent Pulmonary InfecBon. Pediatrics. 1958 Apr 1;21(4):526–54. 12. McKinnon PJ. ATM and ataxia telangiectasia. EMBO Rep. 2004 Aug;5(8):772–6. 13. Taylor AMR, Lam Z, Last JI, Byrd PJ. Ataxia telangiectasia: more variaBon at clinical and cellular levels. Clinical GeneBcs. 2015;87(3):199–208. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 25, 2024. ; https://doi.org/10.1101/2024.04.25.591136doi: bioRxiv preprint 14. Guo Z, Kozlov S, Lavin MF, Person MD, Paull TT. ATM AcBvaBon by OxidaBve Stress. Science. 2010 Oct 22;330(6003):517–21. 15. Alexander A, Cai SL, Kim J, Nanez A, Sahin M, MacLean KH, et al. ATM signals to TSC2 in the cytoplasm to regulate mTORC1 in response to ROS. PNAS. 2010 Mar 2;107(9):4153–8. 16. Zhang J, Tripathi DN, Jing J, Alexander A, Kim J, Powell RT, et al. ATM FuncBons at the Peroxisome to Induce Pexophagy in Response to ROS. Nat Cell Biol. 2015 Oct;17(10):1259–69. 17. Oka A, Takashima S. Expression of the ataxia-telangiectasia gene (ATM) product in human cerebellar neurons during development. Neuroscience Lemers. 1998 Aug 21;252(3):195–8. 18. Barlow C, Ribaut-Barassin C, Zwingman TA, Pope AJ, Brown KD, Owens JW, et al. ATM is a cytoplasmic protein in mouse brain required to prevent lysosomal accumulaBon. Proc Natl Acad Sci U S A. 2000 Jan 18;97(2):871–6. 19. Boehrs JK, He J, Halaby MJ, Yang DQ. ConsBtuBve expression and cytoplasmic compartmentalizaBon of ATM protein in differenBated human neuron-like SH-SY5Y cells. Journal of Neurochemistry. 2007;100(2):337–45. 20. Vail G, Cheng A, Han YR, Zhao T, Du S, Loy MMT, et al. ATM protein is located on presynapBc vesicles and its deficit leads to failures in synapBc plasBcity. Journal of Neurophysiology. 2016 Apr 20;116(1):201–9. 21. Lee JH, Mand MR, Kao CH, Zhou Y , Ryu SW , Richards AL, et al. ATM directs DNA damage responses and proteostasis via geneBcally separable pathways. Sci Signal. 2018 Jan 9;11(512):eaan5598. 22. Cheng A, Tse KH, Chow HM, Gan Y , Song X, Ma F, et al. ATM loss disrupts the autophagy-lysosomal pathway. Autophagy. 2020 Aug 5;0(0):1–13. 23. Sunderland P , Augustyniak J, Lenart J, Bużańska L, Carlessi L, Delia D, et al. ATM-deficient neural precursors develop senescence phenotype with disturbances in autophagy. Mechanisms of Ageing and Development. 2020 Sep;190:111296. 24. Shackelford RE, Innes CL, Sieber SO, Heinloth AN, Leadon SA, Paules RS. The Ataxia telangiectasia Gene Product Is Required for OxidaBve Stress-induced G1 and G2Checkpoint FuncBon in Human Fibroblasts. J Biol Chem. 2001 Jun 15;276(24):21951–9. 25. Barlow C, Hirotsune S, Paylor R, Liyanage M, Eckhaus M, Collins F, et al. Atm-Deficient Mice: A Paradigm of Ataxia Telangiectasia. Cell. 1996 Jul 12;86(1):159–71. 26. Campbell A, Krupp B, Bushman J, Noble M, Pröschel C, Mayer-Pröschel M. A novel mouse model for ataxia- telangiectasia with a N-terminal mutaBon displays a behavioral defect and a low incidence of lymphoma but no increased oxidaBve burden. Hum Mol Genet. 2015 Nov 15;24(22):6331–49. 27. Lavin MF. The appropriateness of the mouse model for ataxia-telangiectasia: Neurological defects but no neurodegeneraBon. DNA Repair. 2013 Aug 1;12(8):612–9. 28. Perez H, Abdallah MF, Chavira JI, Norris AS, Egeland MT, Vo KL, et al. A novel, ataxic mouse model of ataxia telangiectasia caused by a clinically relevant nonsense mutaBon. Sillitoe RV, Westbrook GL, Sillitoe RV, Wam AJ, editors. eLife. 2021 Nov 1;10:e64695. 29. Brent JR, Werner KM, McCabe BD. Drosophila Larval NMJ DissecBon. J Vis Exp. 2009 Feb 4;(24):1107. 30. Feng Y , Ueda A, Wu CF. A Modified Minimal Hemolymph-Like SoluBon, Hl3.1, for Physiological Recordings at the Neuromuscular JuncBons of Normal and Mutant Drosophila Larvae. Journal of NeurogeneBcs. 2004 Jan 1;18(2):377– 402. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 25, 2024. ; https://doi.org/10.1101/2024.04.25.591136doi: bioRxiv preprint 31. Zhang B, Stewart B. Electrophysiological Recording from Drosophila Larval Body-Wall Muscles. Cold Spring Harb Protoc. 2010 Jan 9;2010(9):pdb.prot5487. 32. Zhang B, Stewart B. Equipment Setup for Drosophila Electrophysiology. Cold Spring Harb Protoc. 2010 Jan 9;2010(9):pdb.ip80. 33. Nijhof B, Castells-Nobau A, Wolf L, Gooyert JMS de, Monedero I, Torroja L, et al. A New Fiji-Based Algorithm That SystemaBcally QuanBfies Nine SynapBc Parameters Provides Insights into Drosophila NMJ Morphometry. PLOS ComputaBonal Biology. 2016 Mar 21;12(3):e1004823. 34. Nezis IP , Shravage BV, Sagona AP , Lamark T, Bjørkøy G, Johansen T, et al. Autophagic degradaBon of dBruce controls DNA fragmentaBon in nurse cells during late Drosophila melanogaster oogenesis. J Cell Biol. 2010 Aug 23;190(4):523–31. 35. Stevens CF. A comment on MarBn’s relaBon. Biophysical Journal. 1976 Aug;16(8):891–5. 36. McLachlan EM, MarBn AR. Non-linear summaBon of end-plate potenBals in the frog and mouse. The Journal of Physiology. 1981 Feb 1;311(1):307–24. 37. Feeney CJ, Karunanithi S, Pearce J, Govind CK, Atwood HL. Motor nerve terminals on abdominal muscles in larval flesh flies,Sarcophaga bullata: Comparisons withDrosophila. J Comp Neurol. 1998 Dec 14;402(2):197–209. 38. Tuxworth RI, Taylor MJ, MarBn Anduaga A, Hussien-Ali A, Chatzimamhaiou S, Longland J, et al. AmenuaBng the DNA damage response to double-strand breaks restores funcBon in models of CNS neurodegeneraBon. Brain CommunicaBons. 2019 Jan 1;1(1):fcz005. 39. Pedersen M, Tiong S, Campbell SD. Molecular geneBc characterizaBon of Drosophila ATM conserved funcBonal domains. Genome. 2010 Sep 8;53(10):778–86. 40. Petersen AJ, Rimkus SA, Wassarman DA. ATM kinase inhibiBon in glial cells acBvates the innate immune response and causes neurodegeneraBon in Drosophila. PNAS. 2012 Mar 13;109(11):E656–64. 41. Sigrist SJ, Reiff DF, Thiel PR, Steinert JR, Schuster CM. Experience-Dependent Strengthening of Drosophila Neuromuscular JuncBons. J Neurosci. 2003 Jul 23;23(16):6546–56. 42. Zhong Y , Wu CF. Neuronal AcBvity and Adenylyl Cyclase in Environment-Dependent PlasBcity of Axonal Outgrowth in Drosophila. J Neurosci. 2004 Feb 11;24(6):1439–45. 43. Koch I, Schwarz H, Beuchle D, Goellner B, Langegger M, Aberle H. Drosophila Ankyrin 2 Is Required for SynapBc Stability. Neuron. 2008 Apr;58(2):210–22. 44. Schom S, Ambrosini A, Barbaste A, Benassayag C, Gracia M, Proag A, et al. A fluorescent toolkit for spaBotemporal tracking of apoptoBc cells in living Drosophila Bssues. Development. 2017 Jan 1;dev.149807. 45. Li J, Han YR, Plummer MR, Herrup K. Cytoplasmic ATM in neurons modulates synapBc funcBon. Curr Biol. 2009 Dec 29;19(24):2091–6. 46. Hong ST, Choi KW. TCTP directly regulates ATM acBvity to control genome stability and organ development in Drosophila melanogaster. Nature CommunicaBons. 2013 Dec 19;4(1):1–14. 47. Valbuena FM, Fitzgerald I, Strack RL, Andruska N, Smith L, Glick BS. A photostable monomeric superfolder green fluorescent protein. Traffic. 2020;21(8):534–44. 48. BareBć D, Pollard HK, Fisher DI, Johnson CM, Santhanam B, Truman CM, et al. Structures of closed and open conformaBons of dimeric human ATM. Sci Adv. 2017 May;3(5):e1700933. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 25, 2024. ; https://doi.org/10.1101/2024.04.25.591136doi: bioRxiv preprint 49. Milton VJ, Jarrem HE, Gowers K, Chalak S, Briggs L, Robinson IM, et al. OxidaBve stress induces overgrowth of the Drosophila neuromuscular juncBon. Proc Natl Acad Sci U S A. 2011 Oct 18;108(42):17521–6. 50. Rego AC, Oliveira CR. Mitochondrial DysfuncBon and ReacBve Oxygen Species in Excitotoxicity and Apoptosis: ImplicaBons for the Pathogenesis of NeurodegeneraBve Diseases. Neurochem Res. 2003 Oct 1;28(10):1563–74. 51. Pulver SR, Pashkovski SL, Hornstein NJ, Garrity PA, Griffith LC. Temporal Dynamics of Neuronal AcBvaBon by Channelrhodopsin-2 and TRPA1 Determine Behavioral Output in Drosophila Larvae. Journal of Neurophysiology. 2009 Jun;101(6):3075–88. 52. Maday S, Holzbaur ELF. Autophagosome biogenesis in primary neurons follows an ordered and spaBally regulated pathway. Dev Cell. 2014 Jul 14;30(1):71–85. 53. Cheng XT, Zhou B, Lin MY , Cai Q, Sheng ZH. Axonal autophagosomes recruit dynein for retrograde transport through fusion with late endosomes. J Cell Biol. 2015 May 11;209(3):377–86. 54. Klionsky DJ, Abdel-Aziz AK, Abdelfatah S, AbdellaBf M, Abdoli A, Abel S, et al. Guidelines for the use and interpretaBon of assays for monitoring autophagy (4th ediBon) 1. Autophagy. 2021 Jan 2;17(1):1–382. 55. Qi Y , Qiu Q, Gu X, Tian Y , Zhang Y . ATM mediates spermidine-induced mitophagy via PINK1 and Parkin regulaBon in human fibroblasts. Sci Rep [Internet]. 2016 Apr 19 [cited 2020 May 4];6. Available from: hmps://www.ncbi.nlm.nih.gov/pmc/arBcles/PMC4835770/ 56. Gu X, Qi Y , Feng Z, Ma L, Gao K, Zhang Y . Lead (Pb) induced ATM-dependent mitophagy via PINK1/Parkin pathway. Toxicology Lemers. 2018 Jul 1;291:92–100. 57. Aloke Sarkar, ChrisBne M. Stellrecht, Hima V. Vangapandu, Mary Ayres, Benny A. Kaipparemu, Jun Hyoung Park, et al. Ataxia-telangiectasia mutated interacts with Parkin and induces mitophagy independent of kinase acBvity. Evidence from mantle cell lymphoma. haematol. 2020 Feb 6;106(2):495–512. 58. Lu G, Wu Z, Shang J, Xie Z, Chen C, zhang C. The effects of mezormin on autophagy. Biomedicine & Pharmacotherapy. 2021 May 1;137:111286. 59. Lee JH, Paull TT. Cellular funcBons of the protein kinase ATM and their relevance to human disease. Nat Rev Mol Cell Biol. 2021 Aug 24;1–19. 60. Rothblum -Oviam C, Wright J, Le|on-Greif MA, McGrath-Morrow SA, Crawford TO, Lederman HM. Ataxia telangiectasia: a review. Orphanet Journal of Rare Diseases. 2016 Nov 25;11(1):159. 61. Biton S, Gropp M, Itsykson P , Pereg Y , Mimelman L, Johe K, et al. ATM-mediated response to DNA double strand breaks in human neurons derived from stem cells. DNA Repair. 2007 Jan 4;6(1):128–34. 62. Cheng A, Zhao T, Tse KH, Chow HM, Cui Y , Jiang L, et al. ATM and ATR play complementary roles in the behavior of excitatory and inhibitory vesicle populaBons. Proc Natl Acad Sci USA. 2018 Jan 9;115(2):E292–301. 63. Ling D, Song HJ, Garza D, Neufeld TP , Salvaterra PM. Abeta42-Induced NeurodegeneraBon via an Age-Dependent Autophagic-Lysosomal Injury in Drosophila. Cookson MR, editor. PLoS ONE. 2009 Jan 15;4(1):e4201. 64. Bjedov I, Cochemé HM, Foley A, Wieser D, Woodling NS, CasBllo-Quan JI, et al. Fine-tuning autophagy maximises lifespan and is associated with changes in mitochondrial gene expression in Drosophila. PLOS GeneBcs. 2020 Nov 30;16(11):e1009083. 65. Malagelada C, Jin ZH, Jackson-Lewis V, Przedborski S, Greene LA. Rapamycin Protects against Neuron Death in In Vitro andIn Vivo Models of Parkinson’s Disease. J Neurosci. 2010 Jan 20;30(3):1166–75. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 25, 2024. ; https://doi.org/10.1101/2024.04.25.591136doi: bioRxiv preprint 66. Spilman P , Podlutskaya N, Hart MJ, Debnath J, GorosBza O, Bredesen D, et al. InhibiBon of mTOR by Rapamycin Abolishes CogniBve Deficits and Reduces Amyloid-β Levels in a Mouse Model of Alzheimer’s Disease. PLoS One. 2010 Apr 1;5(4):e9979. 67. Chang YY , Neufeld TP . An Atg1/Atg13 Complex with MulBple Roles in TOR-mediated Autophagy RegulaBon. Mol Biol Cell. 2009 Apr 1;20(7):2004–14. 68. CosenBno C, Grieco D, Costanzo V. ATM acBvates the pentose phosphate pathway promoBng anB-oxidant defence and DNA repair: ATM acBvates the pentose phosphate pathway. The EMBO Journal. 2011 Feb 2;30(3):546–55. 69. Zhang Y , Lee JH, Paull TT, Gehrke S, D’alessandro A, Dou Q, et al. Mitochondrial redox sensing by the kinase ATM maintains cellular anBoxidant capacity. Sci Signal. 2018 Jul 10;11(538):eaaq0702. 70. Song YH, Mirey G, Betson M, Haber DA, Semleman J. The Drosophila ATM Ortholog, dATM, Mediates the Response to Ionizing RadiaBon and to Spontaneous DNA Damage during Development. Current Biology. 2004 Aug 10;14(15):1354–9. 71. Rimkus SA, Katzenberger RJ, Trinh AT, Dodson GE, Tibbems RS, Wassarman DA. MutaBons in String/CDC25 inhibit cell cycle re-entry and neurodegeneraBon in a Drosophila model of Ataxia telangiectasia. Genes Dev. 2008 May 1;22(9):1205–20. 72. Vinters HV, Gae RA, Rakic P . Sequence of cellular events in cerebellar ontogeny relevant to expression of neuronal abnormaliBes in ataxia-telangiectasia. Kroc Found Ser. 1985;19:233–55. 73. Taylor MJ, Thompson AM, Alhajlah S, Tuxworth RI, Ahmed Z. InhibiBon of Chk2 promotes neuroprotecBon, axon regeneraBon, and funcBonal recovery a|er CNS injury. Sci Adv. 2022 Sep 16;8(37):eabq2611. 74. Barbi G, Scheres JMJC, Schindler D, Taalman RDFM, Rodens K, Mehnert K, et al. Chromosome instability and X-ray hypersensiBvity in a microcephalic and growth-retarded child. American Journal of Medical GeneBcs. 1991;40(1):44– 50. 75. Waltes R, Kalb R, Gatei M, Kijas AW, Stumm M, Sobeck A, et al. Human RAD50 Deficiency in a Nijmegen Breakage Syndrome-like Disorder. The American Journal of Human GeneBcs. 2009 May 15;84(5):605–16. 76. Matsumoto Y , Miyamoto T, Sakamoto H, Izumi H, Nakazawa Y , Ogi T, et al. Two unrelated paBents with MRE11A mutaBons and Nijmegen breakage syndrome-like severe microcephaly. DNA Repair. 2011 Mar 7;10(3):314–21. 77. Digweed M, Reis A, Sperling K. Nijmegen breakage syndrome: consequences of defecBve DNA double strand break repair. BioEssays. 1999;21(8):649–56. 78. Stewart. The DNA Double-Strand Break Repair Gene hMRE11 Is Mutated in Individuals with an Ataxia-Telangiectasia- like Disorder. Cell. 1999; 79. ValenBn-Vega YA, MacLean KH, Tait-Mulder J, Milasta S, Steeves M, Dorsey FC, et al. Mitochondrial dysfuncBon in ataxia-telangiectasia. Blood. 2012 Feb 9;119(6):1490–500. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 25, 2024. ; https://doi.org/10.1101/2024.04.25.591136doi: bioRxiv preprint