We previously created a whole-brain single-cell RNAseq dataset from different Drosophila PD knock-in models. This dataset was generated from young 5-day-old flies to capture ‘early’ changes (Pech et al., 2025). We re-analyzed the data from Pink1^P399L knock-in loss-of-function mutant flies and isogenic controls and performed differential gene expression (DEG) analysis for each cell type using DESeq2 (Figure 1A). Glial cell subtypes show the highest degree of deregulation; particularly, ensheathing glia are strongly affected (Figure 1A). While flies do not myelinate neurons, ensheathing glia (EG) serve similar supporting functions as oligodendrocytes (Otto et al., 2018; Yildirim et al., 2019). These results suggest that ensheathing glia are deregulated in young Pink1^P399L mutants (and also Pink1^KO-WS, see below) at an age prior to dopaminergic neuron-dependent motor defects (Kaempf et al., 2026; Pech et al., 2025).

When neurons are damaged, activated EG invade the neuropil (Doherty et al., 2009; MacDonald et al., 2006). For example, when fly antennal lobes are removed, severing the Olfactory Receptor Neurons (ORNs), EG invade the antennal lobe neuropil (MacDonald et al., 2006), potentially as a protective response (Figure 1B’-C). To test if EG invaded the neuropil in Pink1 mutant flies, we expressed a transmembrane fluorescent protein (UAS-CD8GFP) under the control of a promoter specific to EG (MZ709-Gal4) (Doherty et al., 2009, Supplementary figure 1) in control and Pink1^KO flies. Samples were labeled with anti-GFP and the pre-synaptic protein Bruchpilot (NC82). Similar to ORN-severed controls, GFP signal, lining the membrane of EG, is visible inside the antennal neuropil in Pink1^KO, and severing the ORN in Pink1^KO does not further exacerbate this phenotype (Figure 1B”-C). Even though we cannot exclude that in Pink1 mutants the GFP signal is merely upregulated, Pink1 loss and neuron-severing that triggers EG activation seem to have a similar phenotype.

In order to determine whether this phenotype is cell-autonomous, we generated cell-type-specific Pink1 perturbations. We either downregulated Pink1 using a previously well-validated RNAi line specifically in EG (EG-specific Pink1 loss-of-function – Figure 1D’), or we re-expressed wild-type Pink1 specifically in EG in Pink1^KO-WS flies (all cells, but the EG, loss-of-function - Figure 1D”). When Pink1 is knocked-down in EG (but present in other cells, including neurons), no GFP signal is detected in the neuropil (Figure 1D’’-1E). However, when Pink1 is not present in neurons and expressed in EG, the GFP is present inside the antennal neuropil (Figure 1D”-E). This suggests that Pink1-defective neurons activate the EG in a cell-nonautonomous fashion to invade, and potentially protect, the neuropil.

Our finding that Pink1^KO-WS in neurons elicits a cell non-autonomous response in EG suggested that EG might, in turn, also functionally modulate neuronal integrity in mutant flies. To test this, we made use of a readily accessible model circuit in the fly visual system with an easy electrophysiological read-out, electroretinograms (ERG). Notably, whereas our functional readout here interrogates neuronal activity in the visual system, our analysis of EG morphology was performed in the antennal lobe, a distinct brain region. We therefore make the explicit assumption that EG perform comparable functions across fly brain regions. This assumption is supported by our single-cell sequencing dataset, in which we did not detect region-specific segregation of EG into distinct clusters that could be attributed to different brain areas. The ERG waveforms are susceptible to changes in intracellular signaling, neuronal function, and synaptic transmission, and have been well characterized and amply used to assess neuronal and cellular function (Hardie & Raghu, 2001; Praschberger et al., 2023; Soukup et al., 2016; C. F. Wu & Wong, 1977). We exploited this method to understand the role of Pink1 in EG on neuronal function and synaptic transmission in young flies.

Comparing control with young Pink1^KO-WS flies does not show a difference in the “depolarization-response”, indicating that -as expected-there is no neurodegeneration occurring in the photoreceptors at this stage (Figure 2A). However, when comparing the ON peak – which represents synaptic transmission from the photoreceptors and within the underlying neuronal circuit, and is a more robust readout of this than the OFF peak (Vilinsky & Johnson, 2012) – we detect a significant reduction in mutant compared to control animals, suggesting that in Pink1^KO-WS flies, synaptic communication in this circuit is partially impaired, already at 5-day-old (Figure 2A-B).

In order to discern the role of EG in this defective ERG response, we (1) downregulated Pink1 specifically in EG and (2) we expressed Pink1 specifically in the EG of Pink1^KO-WS. When Pink1 is knocked down specifically in EG, we observe the same reduction in ON peak response as in the Pink1^KO-WS flies. This indicates that EG integrity is necessary for efficient neurotransmission in this circuit. Conversely, the ON peak defect of Pink1^KO-WS mutants is rescued when Pink1 is reintroduced in EG only. This suggests that activated EG are able to, at least partially, buffer the detrimental effects of Pink1 deficiency in neurons (Figure 2A-B).

To also test EG function in relation to PD-relevant dopaminergic (DA) synapses, we evaluated the loss of dopaminergic neuron afferents. In a previous paper from our laboratory, we have shown that the loss of Pink1 function causes a progressive loss of protocerebral anterior medial dopaminergic neuron (PAM DAN) afferents in the Mushroom Body (MB) neuropil in 25-day-old animals, but not when they are 5-day-old (Kaempf et al., 2026). Indeed, we confirm that 25-day-old animals show a decrease in the dopaminergic synaptic area in MB lobes (Figure 2 C-D). Hence, while ensheathing glia are already activated at a young age in Pink1^KO-WS, DA synapses are structurally still intact (at 5 days of age) and only deteriorate in older flies. We then knocked down Pink1 specifically in EG and found a similar strong decrease in DA synaptic area (Figure 2C’-D). Conversely, when Pink1 is reintroduced specifically in EG of Pink1^KO-WS animals, DA synapse-loss is significantly rescued (Figure 2C”-D and Supplementary figure 2). Together, these results indicate that Pink1 in EG is necessary to protect dopaminergic neurons from synapse loss.

To identify pathways in EG that are deregulated by Pink1 loss-of-function, we optimized a method to isolate EGs from the rest of the brain, enabling us to perform much higher-sensitivity transcriptomics than what we achieved with our 10x droplet-based single-cell sequencing (Pech et al., 2025). We expressed a fluorescently-tagged histone (UAS-HisTag-eGFP) in EG (GMR-56-Gal4) and used Fluorescence-activated Cell Sorting (FACS) to isolate EG. We sorted 300 cells per technical replicate and performed Bulk-RNA sequencing using a SMART-seq2 protocol (Figure 3A). The sequencing results of EG-sorted cells and of neuron-sorted cells (nSyb-Gal4 instead of GMR-56-Gal4) show strong enrichment of EG cell markers in the former and strong enrichment of neuronal markers in the latter (Figure 3B). We then compared the transcriptomic profile of EG in Pink1^KO-WS to that of control using differential expression analysis and found 617 deregulated genes in EG (Appendix 1, padj<0.05) (Figure 3C). To understand whether the deregulated genes were enriched for a specific pathway or associated with a cellular component compared to non-significantly deregulated genes, we performed Gene Ontology analysis of the deregulated genes in Pink1^KO-WS flies. Our analysis resulted in generic terms and very diverse pathways that did not allow us to draw further conclusions.

To test if the genes deregulated in EG are modifiers of the neuronal pink1^KO-WS phenotypes, we downregulated and upregulated (when possible) the 50 most deregulated genes (Figure 3C), for which we could find RNAi fly lines, specifically in the EG of Pink1^KO-WS flies and recorded ERGs in 5-day-old animals. Knock down of one of the genes (CG17660) resulted in a rescue of the ON transient defects in Pink1^KO-WS (suppressors at p<0.05), and a few exacerbated the defect (Figure 3D). Interestingly, the human orthologs of CG17660, TMEM87A/B, have an established role in endosomal sorting (Gaudet et al., 2011; Hirata et al., 2015). Hence, our genetic screening suggests that Pink1^KO-WS phenotypes might be suppressed by downregulating genes involved in vesicle trafficking in EG.

Because our ERG-based modifier screen for the Pink1^KO-WS phenotype identified a gene with a known role in vesicle trafficking, we next asked whether Pink1 in EG also genetically interacts with other PD-causative genes that function in vesicle trafficking, specifically vesicle trafficking components affecting membrane interactions between mitochondria and the endoplasmic reticulum (ER), like Vps35 and Vps13 (Brickner & Fuller, 1997; Lesage et al., 2016; Vilariño-Güell et al., 2011; Zimprich et al., 2011). We find that the knock down of each of these genes in EG also rescues the ON transient defect of Pink1^KO-WS mutants (Figure 4A-B).

To determine if manipulation of this pathway in EG can also rescue the synaptic innervation defect of PAM DAN onto the mushroom bodies in old (20-25-day-old) Pink1^KO-WS mutants, we expressed RNAi to Vps13 in EG using MZ709-Gal4. Anti-TH labeling (marking DAN synapses) in the MB area (marked by anti-DLG) is significantly reduced in aged Pink1^KO-WS, but is rescued to control levels when Vps13 is knocked down in EG (Figure 4C-D, Supplementary figure 3). These results suggest that manipulation of vesicle trafficking components affecting membrane interactions between mitochondria and the ER in EG regulate glia-neuron crosstalk to maintain synaptic integrity of dopaminergic neuron synapses in the fly brain of Pink1 mutants.

The fruit flies were maintained in an incubator at 25°C under a 12h:12h light-dark cycle and provided with a standard diet consisting of corn meal and molasses. For experiments, only male flies were used. Flies were raised in parallel on the same batch of food, which was exchanged every three to four days and kept in only male population of similar density. Flies were aged to 5 ± 1-day-old, and 22 ± 2-day-old, as indicated for immunohistochemistry and ERGs, respectively.

To create a set of Drosophila models for Parkinson’s disease, CRISPR/Cas9-based gene editing was utilized, as outlined in (Kaempf et al., 2026; Pech et al., 2025). In brief, each knock-out line contained an attP-flanked w+ reporter cassette that replaced the first shared exon across different isoforms of the targeted gene. For this study, we used knockout flies for Pink1 from this collection (Kaempf et al., 2026). The “control” strain denotes a semi-isogenized w¹¹¹⁸ strain that underwent backcrossing to Canton-S for 10 successive generations, resulting in a strain termed Canton-S-w1118. To generate the fly line yw; UAS-His2Av::eGFP.VK27/TM3Sb, we created the pUASTattB_His2AV-GFP plasmid, linearizing the pUAST.attB (Bischof et al., 2007) with EcoRI-BamHI. A gBlock, containing the His2AV, GS linker and eGFP sequence, was cloned into this linearized plasmid with Gibson Assembly. The plasmid was inserted into the VK27 landing site by BestGene. The sequence His2AV-GS-eGFP gblock is:

TGAATAGGGAATTGGGcAAacATGGCTGGCGGTAAAGCAGGCAAGGATTCGGGCAAGGCCAAGGCGAAGGC GGTATCGCGTTCCGCGCGCGCGGGTCTTCAGTTCCCCGTGGGTCGCATCCATCGTCATCTCAAGAGCCGCACTA CGTCACATGGACGCGTCGGAGCCACTGCAGCCGTGTACTCCGCTGCCATATTGGAATACCTGACCGCCGAGGT CCTGGAGTTGGCAGGCAACGCATCGAAGGACTTGAAAGTGAAACGTATCACTCCTCGCCACTTACAGCTCGCC ATTCGCGGAGACGAGGAGCTGGACAGCCTGATCAAGGCAACCATCGCTGGTGGCGGTGTCATTCCGCACATA CACAAGTCGCTGATCGGCAAAAAGGAGGAAACGGTGCAGGAcCCGCAGCGGAAGGGCAACGTCATTCTGTCG CAGGCCTACGGTTCAGGCGGAGGTGGCAGCGGCGGTGGCGGATCCATGGTGAGCAAGGGCGAGGAGCTGTT CACCGGGGTGGTGCCCATCCTGGTCGAGCTGGACGGCGACGTAAACGGCCACAAGTTCAGCGTGTCCGGCGA GGGCGAGGGCGATGCCACCTACGGCAAGCTGACCCTGAAGTTCATCTGCACCACCGGCAAGCTGCCCGTGCC CTGGCCCACCCTCGTGACCACCCTGACCTACGGCGTGCAGTGCTTCAGCCGCTACCCCGACCACATGAAGCAG CACGACTTCTTCAAGTCCGCCATGCCCGAAGGCTACGTCCAGGAGCGCACCATCTTCTTCAAGGACGACGGCA ACTACAAGACCCGCGCCGAGGTGAAGTTCGAGGGCGACACCCTGGTGAACCGCATCGAGCTGAAGGGCATCG ACTTCAAGGAGGACGGCAACATCCTGGGGCACAAGCTGGAGTACAACTACAACAGCCACAACGTCTATATCAT GGCCGACAAGCAGAAGAACGGCATCAAGGTGAACTTCAAGATCCGCCACAACATCGAGGACGGCAGCGTGCA GCTCGCCGACCACTACCAGCAGAACACCCCCATCGGCGACGGCCCCGTGCTGCTGCCCGACAACCACTACCTG AGCACCCAGTCCGCCCTGAGCAAAGACCCCAACGAGAAGCGCGATCACATGGTCCTGCTGGAGTTCGTGACCG CCGCCGGGATCACTCTCGGCATGGACGAGCTGTACAAATAAGGGTACCTCTAGAGGATCTT.

Flies were backcrossed for five generations into an inbred Canton-S strain harboring the w¹¹¹⁸ mutation. The genotypes used in this study are listed in the key resource table and Supplementary table 1.

ORN bilateral axotomy was performed on 4±1 day old flies by surgical ablation of the third antennal segment. Briefly, flies were anesthetized with CO₂, and then both antennae were removed with Dumont #5 forceps and returned to the vial (M. Purice, 2020; M. D. Purice et al., 2017). Flies recovered for 24 h on food at 25°C before dissection.

Immunohistochemistry was performed on adult fly brains of 5±1-day-old, with at least two independent experiments. The brains were dissected in ice-cold PBS and fixed for 20 min in freshly prepared 3.7% paraformaldehyde (in 1x PBS, 0.3% Triton X-100 [PBX] [Sigma]) at room temperature (RT), followed by three 15-min washes in PBX at RT on a shaker. Then, the brains were incubated for 1 h in blocking solution (PBX, 10% normal goat serum [NGS]) at RT. Following blocking, the brains were incubated with primary antibodies (rabbit anti-GFP [Thermo Fisher Scientific], 1:1000, mouse anti-brp [DSHB, nc82], 1:100) in blocking solution at 4°C overnight, followed by three 15-min washes in PBX at RT. Secondary antibodies (goat anti-rabbit Alexa488, goat anti-mouse Alexa555, both at 1:1000 [Thermo Fisher Scientific]) in PBX with 10% NGS were applied overnight at 4°C. Afterwards, the brains were washed three times for 15 min in PBT at RT on a shaker and mounted with the anterior facing up in RapiClear 1.47 (SUNJin Lab). Imaging was performed using a Nikon A1R confocal microscope with a 40x (NA 1.15) water immersion lens; Z-stacks of the entire antennal lobes were acquired. The acquisition was carried out using a Galvano scanner, with a zoom factor of 1, scan speed of 0.5, and line averaging set to 2. All images were captured with a pinhole of 0.9 Airy units and a resolution of 1024 × 1024. Z-stacks (with 0,5 μm step intervals) were obtained for data acquisition, and the same imaging settings were applied across all genotypes and sessions. Image analysis was performed using Fiji (Schindelin et al., 2009). Image analysis was performed using Fiji (Schindelin et al., 2009). Each antennal lobe is composed of multiple glomeruli, and EG separate them and invade them when active (Doherty et al., 2009; Pogodalla et al., 2021). To quantify EG invasion in the antennal lobe, we calculated the invasion in every single glomerulus. To do so and be consistent with the section of the antennal lobe analyzed, anti-brp was used to identify the 3 Z-planes where DM6 and DM1 glomeruli were present (B. Wu et al., 2017). Then, to automatically detect and segment each glomerulus of the antennal lobe section, regions of interest (ROIs) were defined in the SUM projection of the 3 Z-planes by using the Fiji plugin Stardist (Weigert et al., 2019).

To quantify the EG invasion in the ROIs, anti-GFP SUM projection of the same 3 Z-planes was used. The GFP intensity was detected in each ROI, and it was normalized to the area of each ROI (GFP intensity/glomeruli area) to calculate the EG invasion in each glomerulus. Then, the normalized GFP intensity of each glomerulus of an antennal lobe was summed and normalized for the number of glomeruli detected in the antennal lobe. For each experiment, the EG antennal lobe invasion of each fly was normalized to the mean of the control. Representative images show the SUM projections of the 3 Z-planes.

Immunohistochemistry was performed on adult fly brains of 22±2-day-old, with at least two independent experiments. The brains were dissected in ice-cold PBS and fixed for 20 min in freshly prepared 3.7% paraformaldehyde (in 1x PBS, 0.2% Triton X-100 [PBX]) at room temperature (RT), followed by three 15-min washes in PBX at RT on a shaker. Then, the brains were incubated for 1 h in blocking solution (PBX, 10% normal goat serum [NGS]) at RT. Following blocking, the brains were incubated with primary antibodies (rabbit anti-TH [Sigma], 1:200, mouse anti-DLG [DSHB], 1:100) in blocking solution at 4°C for 1.5 to 2 days, followed by three 15-min washes in PBX at RT. Secondary antibodies (goat anti-rabbit Alexa488, goat anti-mouse Alexa555, both at 1:500 [Thermo Fisher Scientific]) in PBX with 10% NGS were applied overnight at 4°C. Afterwards, the brains were washed three times for 15 min in PBT at RT on a shaker and mounted with the anterior facing up in RapiClear 1.47 (SUNJin Lab). Imaging was performed using a Nikon A1R confocal microscope with a 20x (NA0.95) water immersion lens; Z-stacks of the entire brain were acquired. The acquisition was carried out using a Galvano scanner, with a zoom factor of 1, scan speed of 0.5, and line averaging set to 2. All images were captured with a pinhole of 2.3 Airy units and a resolution of 1024 × 1024. Z-stacks (with 3 μm step intervals) were obtained for data acquisition, and the same imaging settings were applied across all genotypes and sessions. Image analysis was performed using Fiji (Schindelin et al., 2009). To quantify dopaminergic neuron innervation of the mushroom body (MB), anti-DLG was used to identify the five Z-planes containing the synaptic region of the MB lobes. The ROI for the MB was defined in the sum projection of the five z-planes. To exclude background signal comparable to control, the area of anti-TH fluorescence was thresholded (using the default threshold for all z-planes) within the selected z-stacks. Quantification of the thresholded area within the ROI was performed in each z-plane, then summed and normalized to the MB ROI area for each brain individually (TH+ area/MB area). For each experiment, the individual TH+ area/MB area values were normalized to the mean of the control. Representative images show the maximum projection of five z-planes and the thresholded middle z-plane.

ERGs were recorded from flies 5±1-day-old as previously described (Heisenberg, 1971; Slabbaert et al., 2016). Flies were immobilized on glass microscope slides using double-sided tape. For recordings, glass electrodes (borosilicate, 1.5 mm outer diameter) filled with 3 M NaCl were placed in the thorax as a reference and on the fly eye for recordings. Each fly was exposed to 5 cycles of 3 s of darkness, followed by a 1-s of light stimuli with LED illumination. Response to the stimuli was recorded using Axoscope 10.7 and analyzed using Clampfit 10.7 software (Molecular Devices). ERG traces were analyzed with Igor Pro 6.37 (Wave Metrics) using a custom-made macro.

To test that EG could be isolated from the rest of the brain, we used two cohorts GMR56F03> UAS-His2Av::eGFP.VK27 and nSyb-Gal4> UAS-His2Av::eGFP.VK27. To identify differentially expressed genes in EG, two cohorts were used: the control w¹¹¹⁸ expressing GMR56F03> UAS-His2Av::eGFP.VK27 and Pink1^KO-WS expressing GMR56F03> UAS-His2Av::eGFP.VK27. For each genotype, ten brains of flies 5±1-day-old were collected, with most of the laminae removed, from each experimental repeat. The dissections were performed in ice-cold PBS that contained 5 μM Actinomycin D (Sigma) to inhibit changes in gene expression during tissue processing(Y. E. Wu et al., 2017). To prevent any bias caused by different experimental batches, all the genotypes were processed in parallel. To minimize variability across batches, the same reagents and procedures were used throughout the experiment. Dissections were completed within one hour. The dissection order and the assignment of genotypes to dissectors were alternated to reduce any experimenter-related variation.

For the dissociation of brain tissue, a mix of Dispase I (0.6 mg/ml) (Sigma), Collagenase I (30 mg/ml) (Thermo Fisher), Trypsin (0.5x) (Thermo Fisher), and 5 μM Actinomycin D was used. The brains were incubated for 20 min at 25°C with 1000 rpm shaking, performing four triturations at 5-min intervals. After dissociation, the cell suspension was washed with PBS containing 5 μM Actinomycin D and then filtered through a 10 μm strainer (Pluriselect), using 300 μl of PBS EDTA (Sigma) 1μM. DAPI (Sigma) (1:300.000) was added after filtration.

Immediately after the dissociation, cells were sorted using a FACSAria Fusion (BD Biosciences) flow cytometer equipped with 4 lasers (405nm, 488nm, 561nm and 640nm) and a 100 µm nozzle at 20 psi. Live (DAPI-negative), GFP-positive cells were sorted into 96-well PCR plates at 300 cells per well (3 wells per genotype) using a 4-way purity mask with the FACSDiva software v9.0.1 (BD Biosciences).

Library preparations and sequencing for the Bulk RNA seq of the EG cells were performed using a modified Smart-seq2 protocol based on a previously published protocol (Picelli et al., 2013). Briefly, 3 μl of cell lysis buffer (0.1% Triton X-100, Sigma Aldrich; 1U/ul of RNase Inhibitor, RNase OUT, Thermo Fisher; 2.5 μM Oligo dT(25), IDT; and 2.5 mM dNTP, Promega) was aliquoted in triplicate to 96-well plates (4titude, Cat. No. 4TI-0960/C). Roughly 300 cells were sorted into the wells with lysis buffer and the plate was spun down at 2000xg for 1 min prior to storage at -80 °C. For the first strand synthesis, the Smart-Seq 2 plates were thawed to RT for a minute and subsequently spun down at 2000 xg for a minute. The RNA denaturation is carried out at 72°C for 10 min and then flash cooled on ice for 5 min. First strand synthesis mix (1X SuperScript II first strand synthesis buffer, Thermo Fisher; 5 mM DTT, Thermo Fisher; 100 U SuperScript II reverse transcriptase enzyme, Thermo Fisher; 10 U RNAse OUT, Thermo Fisher; 1M Betaine, Sigma Aldrich; 6 mM MgCl2, Thermo Fisher; and 1 uM TSO, IDT Technologies) is added to the cell lysis mix to a total volume of 10 μl. The First strand synthesis reaction was carried out with following program: 42°C for 90 min; 10 cycles of {50°C for 2 min; 42°C for 2 min}; 72°C for 15 min; 4°C indefinite hold. PCR amplification of the first strand product was performed by adding 15 μl of the PCR mix to the first strand product (1X KAPA HiFi HotStart Ready Mix, Roche; and 0.1 μM of IS PCR primer, IDT Technologies). 22 cycles of the following PCR program were used to amplify the first strand product: 98°C for 3 min; 22 cycles of {98°C for 20 sec; 67°C for 15 sec; 72°C for 6 min}; 72°C for 5 min; 4C indefinite hold. 20 μl of Ampure XP was added to each well and mixed. Standard Ampure XP purification was carried out as per manufacturer’s recommendation, and the cDNA library was eluted in 17.5 μl of Elution buffer. 17 μl of the eluted library was,transferred to a fresh 96-well plate (4titude, Cat. No. 4TI-0960/C). 5 μl of Tn5 tagmentation mix (1X Tagment DNA buffer, Illumina; ATM mix, Illumina, and cDNA library 1 ng) was prepared for cDNA fragmentation. Tn5 tagmentation was performed with the following program: 55°C for 10 min; 4°C indefinite hold. Tagmentation reaction was stopped by quenching the reaction with 1.25 μl of NT buffer for 5 min (Illumina). 1.25 μl of the i5 and i7 Illumina indexes were added to the plates. Finally 3.75 μl of NPM master mix was added to the plate and mixed well and put for the index PCR amplification: 72°C for 3 min; 95°C for 30 sec; 12 cycles of {95°C for 10 sec; 55°C for 30 sec; 72°C for 30 sec}; 72°C for 5 min, and 4°C indefinite hold.

The indexed PCR products were pooled in a single tube, and 0.8X Ampure XP purification (Beckman Coulter) was carried out as per the manufacturer’s recommendation, and finally, the sequencing library was eluted in 35.5 μl of Elution buffer (Qiagen).

SMART seq 2 libraries were sequenced on the NextSeq 500 (Illumina) sequencing platform. Sequencing was done as per the protocol recommendations: paired-end read of 76 bps (read 1), 76 bps (read 2), 8 bps (index 1) and 8 bps (index 2). For a targeted sequencing depth of ∼30 million reads per sample.

Raw reads from neuron- vs. glia-specific FACsorting experiments were processed with the nf-core/rnaseq pipeline (Patel et al., 2020) with default parameters and the BDGP6 reference genome. The STAR-aligned and salmon-quantified expression matrix was then subset to genes known to be enriched in neurons or glia and displayed as a per gene scaled heatmap.

After FACsorting of Pink^KO-ws vs. control ensheathing glia, library preparation and sequencing resulting FASTQ files were cleaned with fastp (Chen et al., 2018) with default settings. Aligning and counting were carried out with STAR (Dobin et al., 2013) and the 4th 2020 FlyBase Drosophila melanogaster release (r6.35) with the –quantMode flag set as GeneCounts. For differential gene expression testing we used DESeq2 (Love et al., 2014), where we fit a negative binomial model and carried out the Wald-test. We removed the experimental batch as a covariate according to the design formula ∼date+genotype. All genes below a Benjamini-Hochberg corrected p-value of 0.05 were considered deregulated.

Raw counts data from (Pech et al., 2025), alongside cell annotations provided by the authors, was loaded into Scanpy (v1.9.1) (Wolf et al., 2018) and subset to only include cells in the young control (<= 6-day-old) or Pink1^KO-WS samples. After subsetting the data, filtering was performed to remove any cell with less than 250 genes expressed or a percentage of counts coming from mitochondrial genes greater than 15%, the data was then normalized to a total of 10.000 counts and log transformed. Highly variable genes were detected, total_counts and percent of mitochondrial reads were regressed out, and finally the data was scaled to unit variance with a zero mean, clipped to a max of 10.

After data pre-processing, a PCA was performed, Harmony (Korsunsky et al., 2019) was used to correct batches using “sample_id” as a batch key, and 122 components were used for dimensionality reduction and clustering analyses. Leiden clustering was performed with a resolution of 8.0 and using the annotations from (Pech et al., 2025), all cells within each cluster were labeled by taking the most common, original annotation per cluster.

Per cluster, a differential analysis was performed. First, pseudobulk counts were generated by summing the counts per cell from all cells per sample. These counts were then used in DESeq2 (v1.44.0) (Love et al., 2014) comparing Control vs Pink1^KO-WS samples. The total number of genes significantly up- or downregulated per cluster (padj <= 0.05 and an abs (log2foldchange) >= 1.5) was counted and plotted.

GraphPad Prism was used for visualization and to determine statistical significance. Datasets were tested for normal distribution using the D’Agostino-Person Omnibus and the Shapiro-Wilk normality test. For a normally distributed dataset ordinary one-way ANOVA was used, followed to correct for multiple comparisons by a post hoc Tukey’s test when comparing all the datasets with each other or Dunnett’s test when comparing all the datasets to a general control. For non-normally distributed datasets Kruskal-Wallis test is used, followed by a post hoc Dunn’s test to correct for multiple comparisons. Significance levels are defined as * is p<0.05, ** is p<0.01, *** is p<0.001, and ns, not significant. Effect size is calculated with eta-squared when one-way ANOVA is used, indicated in the figure legend as η. ‘N’ in the legends is used to indicate how many animals were analyzed. Data are plotted as mean ± SD. Specifics on the statistical test used for each analysis are reported in the figure legends.