Motor dysfunction phenotypes in a drosophila model validate DAO as an ALS gene Authors: Sylvana Tabone, Rebecca Cacciottolo, Ruben J Cauchi Published: 2026 ABSTRACT AbstractAmyotrophic lateral sclerosis (ALS) is a fatal neurodegenerative disorder characterised by progressive motor neuron (MN) loss. Although rare pathogenic variants in the D-amino acid oxidase (DAO) gene have been proposed to cause familial ALS, inconsistent findings across mammalian models have hindered functional validation. Here, we establish and characterise a Drosophila model of DAO loss-of-function by targeting its highly conserved orthologue, Daao1. Constitutive Daao1 knockdown caused early developmental lethality, while surviving adult escapers exhibited profound flight and climbing impairments. Contrary to expectations, Daao1 deficiency did not elevate D-serine levels in larval or adult brains, and simultaneous knockdown of serine racemase (Srr), the enzyme responsible for D-serine synthesis, failed to rescue locomotor phenotypes. Tissue-specific silencing revealed that Daao1 is essential within both MNs and muscles, but not glia, for maintaining motor performance. RNA-seq of Daao1-deficient flies uncovered widespread transcriptional dysregulation affecting metabolism, peptide transport, immune activation and developmental pathways, accompanied by alternative splicing changes in genes required for neuronal morphology and synaptic organisation. Consistent with these molecular alterations, neuromuscular junctions (NMJs) displayed reduced axonal branching and bouton numbers, indicating disrupted neuromuscular connectivity. Remarkably, ubiquitous expression of human DAO rescued lethality and motor defects, demonstrating functional conservation and supporting a role for DAO in maintaining neuromuscular health. Together, our findings reveal a previously unrecognised, D-serine-independent role for DAO in neuromuscular integrity and provide functional evidence connecting DAO dysfunction to ALS-relevant phenotypes. The DAO-ALS Drosophila model provides a powerful platform for mechanistic studies and for evaluating human DAO variants of uncertain significance in ALS.Keywords: amyotrophic lateral sclerosis, Drosophila model, neurogenetics, neurodegeneration Introduction Amyotrophic lateral sclerosis (ALS) is a relentless neurodegenerative disease characterised by the progressive loss of motor neurons (MNs) in the brain and spinal cord, leading to muscle atrophy, paralysis and ultimately death from respiratory failure, typically within 3–4 years from disease onset [1]. Genetics plays an important role in the aetiology of ALS, with over 40 genes now identified as a cause for monogenic forms of the disease [2]. However, for many of these genes, lack of replication or functional validation data raises questions about whether they are genuine ALS causative genes. Falling into this ambiguous category is the D-amino acid oxidase (DAO) gene, which encodes an enzyme that regulates levels of D-serine (D-ser) through its degradation [3]. While mutations in DAO were identified more than a decade ago as a rare cause of familial ALS [4], the findings were met with controversy due to a conflicting report [5], with this casting doubt on the link between DAO and ALS. D-Ser is an endogenous co-agonist of N-methyl-D-aspartate (NMDA) receptors expressed on MNs, and its levels are thought to critically regulate glutamate-dependent neurotransmission [6, 7]. Excessive NMDA receptor activity can lead to excitotoxic MN death [8]. This process is so strongly implicated in ALS pathophysiology that it remains the sole mechanism targeted by the only drug with universal indication for the disease (riluzole) [9]. An overproduction of D-ser due to suppression of DAO activity in addition to upregulation of its producing enzyme, serine racemase (SRR), has been observed in spinal cords of transgenic mice with mutations in ALS-causing gene superoxide dismutase 1 (SOD1) [10, 11]. This finding was crucially confirmed in plasma and postmortem spinal cords from ALS patients and is thought to augment glutamate excitotoxic stress in MNs [11, 12]. ALS-linked mutations in DAO are thought to cause loss of protein function, supported by evidence of markedly diminished or nearly abolished enzymatic activity observed both in postmortem spinal cord tissue from a mutation carrier and in cell cultures expressing the mutant construct [4, 13, 14]. Mice that lack DAO activity due to homozygosity for a naturally occurring missense mutation in the DAO gene have elevated D-ser levels in the central nervous system (CNS), enhanced NMDA receptor function and potentiated NMDA receptor-mediated excitatory postsynaptic currents in spinal cord dorsal horn neurons [15–18]. A similar increase in cerebellar D-ser levels was also reported upon knockdown of DAO in mice by lentivirus-delivered RNAi [19]. Importantly, although DAO mutant mice or transgenic mice expressing human DAO with an ALS-linked mutation displayed motor deficits, MN degeneration, and muscle atrophy [10, 15, 20], these phenotypes were relatively mild compared to the severe manifestations observed in ALS patients carrying pathogenic DAO variants. Also, when compared to single SOD1 transgenic mice, DAO and SOD1 double transgenic mice did not exhibit major differences in disease onset, progression or duration [20]. Furthermore, studies aimed at reducing D-ser levels in SOD1 mutant mice have yielded conflicting results. Hence, despite reducing spinal cord D-ser concentrations, deletion of the SRR gene in SOD1 mice exacerbated ALS onset [21], whereas adeno-associated virus–mediated delivery of DAO mitigated MN loss and extended survival [22]. It is also noteworthy that a study has showed that pharmacological inhibition of DAO elevates plasma D-ser levels in mice, but not in non-human primates, with this highlighting potential species-specific differences in D-ser metabolism and/or DAO function [23]. In this study, we make use of Drosophila to evaluate its suitability as an alternative model system for investigating DAO-linked ALS. Drosophila has been used extensively to model motor neuron disease, as it combines extensive genetic tractability with a high degree of conservation in disease-implicated cellular pathways [24–27]. Here, we show that constitutive knockdown of the fly’s highly homologous DAO orthologue, Daao1, induced lethality during early development. Nonetheless, the genetic background of one constitutive driver line produced escapers, allowing us to show that global Daao1 downregulation causes profound motor defects. Interestingly, unlike what might be expected, a reduction in Daao1 did not lead to elevated D-ser levels in fly brains. Consistent with this, silencing the D-ser-producing enzyme, Srr, in Daao1 knockdown flies did not revert the motor phenotypes. We also find that motor dysfunction could be reproduced by selective Daao1 silencing in MNs or muscles, but not glia. In response to Daao1 loss, RNA sequencing (RNA-seq) revealed transcriptional alterations across several processes, notably metabolism, immune response, peptide transport and development. In addition, several genes with roles in neuronal formation and maintenance were found to be differentially spliced downstream of Daao1 deficiency. This ties well with the reduced axonal projections observed at neuromuscular junctions (NMJs) of Daao1 knockdown flies. Remarkably, expressing human DAO rescued the viability and motor defects in the Daao1 knockdown flies, underscoring the functional conservation between the fly and human genes. Our findings not only validate DAO as an ALS gene but also provide insights into how loss of DAO function drives neuromuscular degeneration independent of D-ser metabolism defects. Results Global loss of the Drosophila DAO orthologue, Daao1, induces early lethality The Daao1 (CG12338) gene in Drosophila is the top-ranked orthologue of human DAO, achieving a high score of 13 out of 18 in the Drosophila RNAi Screening Center Integrative Ortholog Prediction Tool (DIOPT, version 9.0). Notably, Daao1 is highly conserved, displaying 82% amino acid similarity and 33% identity (with 85% coverage) compared to the human protein (Fig. 1A). To develop a DAO-ALS fly model, we sought to mimic the loss-of-function observed with ALS-linked DAO pathogenic variants by employing RNAi interference (RNAi) to silence Daao1. To this end, the GAL4/upstream activation sequence (UAS) system was utilised to activate 2 RNAi transgenes targeting a similar sequence within the 3′ coding region that is common to the two predicted Daao1 mRNA transcripts (Fig. 1B). We first assessed the efficiency and specificity of gene knockdown by performing quantitative RT-PCR (qRT-PCR) on RNA extracted from wandering third instar (L3) larvae with constitutive expression (Act5c-GAL4) of each transgenic construct from early development. Activation of the Daao1-IR1 transgene resulted in a strong reduction in Daao1 transcript levels (12% residual expression), whereas the Daao1-IR2 transgene produced a more moderate knockdown (24% residual expression) (Fig. 1C). Global or muscle-selective Daao1 knockdown induces early lethality. (A) Clustal omega alignment of human DAO (Hs) and its Drosophila orthologue Daao1 (Dm), showing strong evolutionary conservation of amino acids throughout the entire length of the protein sequence. Amino acid colouring denotes the degree of conservation: black, non-conserved; green, weakly similar; orange, strongly similar; white on red, identical residues. (B) Predicted Daao1 mRNA transcripts in Drosophila and the regions targeted by RNAi constructs. Daao1-IR1 and Daao1-IR2 target overlapping sequences within the 3′ coding region of both transcripts, although the targeting sequence of Daao1-IR2 is shorter. (C) Daao1 expression relative to the housekeeping RpL32 gene in L3b larvae in which the indicated RNAi transgene was constitutively expressed (Act5C-GAL4) as determined by qRT-PCR. Each bar represents the mean ± SEM of at least three biological replicates with the respective data points superimposed on the bars. (D) Viability outcomes of flies in which each RNAi transgene was activated either in whole organisms or in specific tissues from early development by the indicated GAL4 drivers. Co-expression of Dcr-2 was employed to achieve stronger gene knockdown. We next assessed the impact of Daao1 gene silencing on fly viability. Ubiquitous knockdown using either the Daao1-IR1 or the Daao1-IR2 RNAi construct driven by the strong tubP-GAL4 driver resulted in lethality prior to the third instar larval (L3) stage. By contrast, the Act5C-GAL4 driver, despite exhibiting a similar expression profile and strength, produced a later-onset lethality and, in the case of Daao1-IR1, even allowed adult escapers (Fig. 1D, Supplementary Materials Fig. 1). We hypothesise that this difference may reflect partial rescue conferred by the genetic background in the Act5C-GAL4 driver. Tissue-specific knockdown revealed that Daao1 silencing in muscles, but not neurons (even on co-expression of Dcr-2 to enhance RNAi), was sufficient to cause lethality before the L3 stage, indicating that global lethality was recapitulated by muscle-restricted depletion (Fig. 1D, Supplementary Materials Fig. 1). Together, these findings demonstrate that Daao1 is an essential gene in Drosophila, and adult escapers with constitutive loss of Daao1 function presented a valuable opportunity to investigate ALS-like phenotypes. Downregulation of Daao1 induces motor defects that are not rescued by loss of Srr Adult fly escapers with ubiquitous Daao1 knockdown were assessed for an impairment in motor function, given that this is the most prominent outward manifestation of ALS. By week 1 post-eclosion, flies had marked difficulties with flying and were found to be completely flightless in a flight assay compared with controls (Fig. 2A). They also showed pronounced climbing deficits, with the first fly in each group assessed requiring significantly more time to reach the predetermined threshold than controls (Fig. 2B). We next questioned whether the observed motor dysfunction was a consequence of elevated D-ser levels in the fly brain. Unexpectedly, immunofluorescence imaging and analysis revealed no detectable differences in D-ser levels in either larvae or adult flies with Daao1 deficiency (Fig. 2C). Loss of Daao1 function induced severe motor defects independent of D-ser levels. (A) Heat maps comparing distribution of final landing position (Sector 4, top; Sector 1, bottom) of 1-week-old female flies with the indicated genotype (n = 45–60/genotype). Constitutive Daao1 knockdown caused profound flightlessness compared to the driver-only control, which is not rescued by the simultaneous depletion of Srr. (B) Bar chart showing the time taken for the first fly in a sampled population, with the indicated genotype, to climb to a pre-determined threshold (n = 45–60/genotype). The sluggish behaviour caused by global Daao1 downregulation in 1-week-old female flies was maintained upon loss of Srr activity. (C) Left, representative L3 (top) and adult (bottom) brains stained for D-Ser and nuclear stain Hoechst. On visual inspection, there was no overall difference in D-Ser fluorescence intensity between flies with global Daao1 gene silencing and control. Right, bar chart showing quantification of D-Ser fluorescence intensity normalised to Hoechst (used as a proxy for cell count). There was no significant difference in D-ser levels between Daao1 knockdown flies and control (n ≧ 8). For heat maps, exact percentage values are provided within each box. For graphs, each bar represents the mean ± SEM of several independent experiments. Statistical significance was tested using a two-way ANOVA with Bonferroni’s post hoc test for multiple groups or an unpaired t-test for two groups. For all data, ****p < .0001. To confirm that the motor impairment was independent of D-ser levels, we investigated whether loss of Srr, the enzyme responsible for D-ser synthesis, would revert the behavioural abnormalities observed in Daao1 knockdown flies. Drosophila harbours a highly homologous SRR orthologue, exhibiting 90% amino acid similarity and 37% identity (94% coverage, DIOPT) (Supplementary Materials Fig. 2A). Srr (CG8129) knockdown was achieved by targeting a region within the 3′ coding sequence of its two predicted mRNA transcripts (Supplementary Materials Fig. 2B). We first validated that loss of Srr activity effectively reduced D-ser levels in fly brains, with this also confirming the robustness of our D-ser immunofluorescence protocol (Supplementary Materials Fig. 3). Next, we generated flies with ubiquitous knockdown of both Daao1 and Srr. qRT-PCR confirmed strong Srr transcript reduction alongside sustained Daao1 knockdown, despite GAL4 dilution (Supplementary Materials Fig. 2C). Crucially, the Daao1-associated flight and climbing deficits were not rescued by Srr loss-of-function (Fig. 2A and B). This finding corroborates our earlier D-ser quantification analysis. Overall, our results strongly indicate that the ALS-like phenotypes in Daao1 knockdown flies do not arise from elevated D-ser levels in the brain and, consequently, cannot be reverted by reducing D-ser synthesis. Motor ability is impaired by either muscle or neuron-selective Daao1 gene silencing We next wished to determine whether the motor abnormalities observed with constitutive Daao1 knockdown arose from disruption of Daao1 function in specific components of the motor system. Considering that neuronally-restricted Daao1 gene silencing resulted in viable adult flies, we first assessed these flies and found that, as early as week 1 post-eclosion, reduced Daao1 levels specifically in neurons significantly impaired flight capacity (Fig. 3A). While climbing ability was initially unaffected, brain-selective Daao1 knockdown induced dramatic climbing difficulties in late adulthood compared to control flies (Fig. 3B). To test if enhanced Daao1 knockdown could induce climbing defects at an earlier timepoint, we co-expressed the Dcr-2 transgene to boost the RNAi effect. We found that enhanced neuronal knockdown significantly impaired locomotor behaviour by week 1 post-eclosion (Fig. 3C). To further define which subset of neurons contributes to the motor dysfunction, we specifically targeted knockdown of Daao1 to MNs, given their critical relevance to ALS. Interestingly, this resulted in a progressive, age-dependent decline in both flight behaviour and climbing ability (Fig. 3D and E). Neuron- or muscle-selective Daao1 silencing impaired motoric ability. (A) Heat maps showing the distribution of final landing positions (Sector 4, top; Sector 1, bottom) of 1-week-old male flies with the indicated genotypes (n = 60/genotype). Neuron-selective Daao1 knockdown from early development resulted in complete flightlessness. (B) Bar chart showing the time taken for the first fly in each population to climb to a pre-determined threshold (n = 60–300/genotype), demonstrating that brain-selective Daao1 reduction induced pronounced late-onset climbing difficulties. (C) Bar chart showing the time taken for the first fly in each population to reach the climbing threshold. Performance was significantly delayed for female flies with enhanced, neuronally-restricted Daao1 knockdown (n = 45/genotype). (D) Heat maps comparing the distribution of final landing positions (Sector 4, top; Sector 1, bottom) for flies with MN-selective Daao1 knockdown, assessed at different timepoints (n = 60–300/genotype). Flight defects were apparent as early as week 1 post-eclosion. (E) Bar chart showing the time required for the first fly in the sampled population to reach the climbing threshold as determined at different timepoints. At week 4 post-eclosion, MN-selective Daao1 reduction significantly delayed the time required for flies to reach the pre-determined threshold (n = 60–80/genotype). (F) Heat maps comparing distribution of final landing position (Sector 4, top; Sector 1, bottom) for male flies with adult-onset, muscle-specific Daao1 reduction assessed across timepoints (n = 60–80/genotype). Flight capacity was significantly diminished by week 4 post-eclosion. (G) Bar chart showing the time required for the first fly in each population to reach the climbing threshold (n = 60–80/genotype). At week 4, flies with adult-onset, muscle-restricted Daao1 deficiency took significantly longer to reach the threshold. For graphs, each bar represents the mean ± SEM of several independent experiments. For heat maps, exact percentage values are provided within each box. Statistical significance was assessed using two-way ANOVA followed by Bonferroni’s post hoc test. Asterisks denote comparisons with the driver-only control group (GAL4 > WT), where *p < 0.05, **p < 0.01, and ****p < 0.0001. Turning to muscle, we noted that muscle-selective loss of Daao1 starting from early development resulted in lethality prior to L3 (Fig. 1D), which precluded us from assessing mobility defects. To bypass this early developmental lethality, we utilised the GAL80-GAL4 system to temporarily supress the Daao1 RNAi transgene during development and switching it back on only in adult flies after eclosion. Aging adult flies in which Daao1 was specifically silenced in muscles were subsequently found to exhibit significant flight and climbing deficits in contrast to control (Fig. 3F and G). Finally, we wished to determine whether Daao1 is also required in glia for motor function. To this end, we induced enhanced Daao1 knockdown starting from early development and assessed adult flies at different timepoints following eclosion. We did not observe differences in motor behaviour in flies with glial-restricted Daao1 loss compared to controls (Supplementary Materials Fig. 4A and B). In sum, these findings underscore that reducing Daao1 levels selectively in neurons, particularly MNs or muscles is sufficient to replicate the motor defects observed upon knockdown in all tissues. Transcriptional alterations in response to loss of Daao1 To elucidate the molecular consequences of Daao1 loss and gain insights into the mechanisms underlying the observed defects in viability and neuromuscular behaviour, we performed RNA-seq on L3 larvae with constitutive Daao1 gene silencing. Differential expression analysis identified 272 differentially expressed genes (DEGs) relative to controls, including 172 downregulated and 100 upregulated transcripts (Fig. 4A, Supplementary Materials Dataset S1). As expected, Daao1 itself was among the most significantly downregulated genes, confirming efficient RNAi-mediated knockdown. Only a small subset of transcripts, 4 upregulated and 1 downregulated, were annotated as novel. To explore the biological pathways impacted by Daao1 reduction, we next conducted gene set enrichment analysis (GSEA). Enrichment of Gene Ontology (GO) biological processes among DEGs revealed a marked downregulation of pathways associated with metabolism, development and peptide transport (Fig. 4B). In contrast, upregulated DEGs were significantly enriched for pathways related to the immune response, lipid metabolism and the response to insecticides (Fig. 4C). Several key genes in different pathways that were found to have altered expression can be highlighted (Supplementary Materials Fig. 5). Notably, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis identified significant enrichment of the Toll and Imd signalling pathway among upregulated genes (FDR = 0.048; fold enrichment = 7.2), implicating increased innate immune activity downstream of Daao1 loss of function. Differentially expressed transcripts and altered pathways following Daao1 gene silencing. (A) Volcano plot showing differentially expressed genes (DEGs) in L3 fly larvae with ubiquitous Daao1 knockdown (Act5C-GAL4 > Daao1-IR1) compared to the driver-only control (n = 3 biological replicates, sex = females). Topmost significant DEGs, including Daao1, are annotated. (B) Lollipop plot depicting the most significant GO biological process terms enriched in downregulated DEGs, highlighting pathways related to metabolism, development and peptide transport. (C) Lollipop plot showing the most significant GO biological process terms enriched in upregulated DEGs, including those associated with the immune response, lipid metabolism and response to insecticides. In B and C, GO terms are sorted by FDR (<0.05) with the colour of the lollipops representing the values of the enrichment analysis relative to the other displayed terms (brighter red is more significant) and the size of the dots indicating the number of genes that comprise the term. GO terms that are tagged with a colour-coded star indicate pathway overlap. RNA-seq also revealed that Daao1 knockdown flies had 61 differentially spliced genes (DSGs) including 14 with an alternative 3′ splice site (A3SS), 15 with an alternative 5′ splice site (A5SS), 6 with a mutually exclusive exon (MXE), 17 with a retained intron (RI) and 23 with a skipped exon (SE) (Supplementary Materials Dataset S2). Several transcripts exhibited more than one mode of alternative splicing, although none were affected by all modes (Fig. 5A, Supplementary Materials Dataset S3). GSEA of DSGs revealed a disruption of biological processes involved in neuronal development and synaptic connectivity (Fig. 5B). Several key genes that are involved in shaping neuronal architecture including synaptic connectivity were found to be alternatively spliced (Supplementary Materials Fig. 6). Correspondingly, GO terms for cellular components highlighted several key neuronal structures including the synapse, the NMJ, the axon and the terminal boutons (Fig. 5B). In summary, transcriptome profiling uncovered alterations across several processes downstream of Daao1 deficiency including metabolism, immune activation, peptide transport, development, and neuronal morphogenesis. Differential splicing in Daao1 deficient flies affects genes involved in neurodevelopmental processes. (A) Venn diagram showing the overlap of differentially spliced genes (DSGs) across the 5 modes of alternative splicing identified in the Daao1 RNA-seq dataset including number of mRNA transcripts with an alternative 3′ splice site (A3SS), an alternative 5′ splice site (A5SS), a mutually exclusive exon (MXE), a retained intron (RI) and/or a skipped exon (SE). (B) Lollipop plots displaying the most significantly enriched GO terms for biological processes (top) and cell components (bottom) among DSGs. Pathways related to neurodevelopment and neuronal structural components are highlighted. GO terms are sorted by FDR (<0.05) with the colour of the lollipops representing the values of the enrichment analysis relative to the other displayed terms (brighter red is more significant) and the size of the dots represent the number of genes that comprise the term. GO terms tagged with a colour-coded star indicate pathway overlap. Loss of Daao1 function compromises axonal projections at the NMJ Given the predicted effects of Daao1-associated transcriptional alterations on neuronal architecture, we next investigated whether loss of Daao1 impairs the formation and organisation of MNs in L3 larvae. The Drosophila larval NMJ serves as one of the most extensively characterised synapses for studying axon guidance, synaptic growth and neuromuscular connectivity, offering a powerful system to dissect genetic influences on motor circuit architecture [28]. Immunofluorescence analysis of NMJs in Daao1 knockdown larvae revealed noticeable structural differences compared with controls. Visual inspection showed an obvious reduction in NMJ span and complexity (Fig. 6A). Quantitative assessment confirmed that, the total number of branches per NMJ, which reflect the ability of MNs to extend and elaborate axonal projections, was significantly reduced in Daao1 deficient flies (Fig. 6B). We also observed a decrease in the number of synaptic boutons per NMJ in the absence of changes in both overall NMJ area and individual bouton area. Importantly, these changes were not attributable to differences in muscle size, as no significant alterations in muscle surface area were detected between Daao1 knockdown and control flies (Supplementary Materials Fig. 7). Collectively, these findings demonstrate that Daao1 is essential for proper axonal branching and NMJ development. The observed reduction in synaptic complexity and bouton number points to a requirement for Daao1 in maintaining normal neuromuscular connectivity, and these structural abnormalities likely underlie the motor dysfunction phenotypes observed upon Daao1 loss. Daao1 depletion reduces axonal projections at the NMJ. (A) Representative images of DLG- and HRP-stained NMJs innervating ventral longitudinal muscles 6 and 7 in L3b larvae. Daao1-deficient larvae display a clear reduction in NMJ span and complexity compared with controls. (B) Bar charts showing quantification of NMJ morphology parameters. Total number of branches and bouton number per NMJ are significantly lower in Daao1 knockdown flies (n ≥ 23 NMJs derived from 8 larvae/genotype). For graphs, each bar represents the mean ± SEM of several independent experiments. Significance was tested by the unpaired t-test and for all data, *p < 0.05, and ***p < 0.001. Expression of human DAO rescues survival and motor defects in Daao1 knockdown flies Finally, we asked whether ubiquitous expression of human DAO (hDAO) could rescue the phenotypes observed in Daao1 knockdown flies. To test this, we generated flies expressing Daao1-IR2 together with hDAO under the control of Act5C-GAL4 (Fig. 7A). hDAO expression alone did not impair viability. In contrast to Daao1 knockdown flies, which failed to eclose, co-expression of hDAO resulted in a modest but statistically significant increase in the percentage of flies that successfully eclosed (Fig. 7B). Furthermore, these rescued flies exhibited no observable climbing defects compared to control flies expressing hDAO in a wild-type background (Fig. 7C). This effect cannot be attributable to GAL4 dilution, as residual Daao1 transcript levels (29%) were similar to those measured in the absence of hDAO expression (Fig. 7D, Fig. 1C). Together, these findings demonstrate that hDAO expression is sufficient to partially restore viability and preserve motor performance in Daao1-deficient flies. hDAO expression partially rescues survival in Daao1 knockdown flies. (A) Schematic of the genetic cross utilised to introduce hDAO in Daao1 knockdown flies. (B) Bar chart showing percentage number of eclosed flies for the indicated genotypes. Co-expression of hDAO with Daao1-IR2 produced a modest but statistically significant increase in the proportion of viable adults, whereas Daao1-IR2 alone prevented eclosion. Ubiquitous expression of hDAO alone did not affect viability. Significant deviation from 0% viability was tested by the one sample t-test (*p < 0.05, **p < 0.01, and **** p < 0.0001). (C) Bar chart showing the time taken for flies of the indicated genotype to climb to a pre-determined threshold (n = 3–7 flies/genotype). Motor performance of hDAO-rescued Daao1-knockdown flies is similar to that of hDAO-expressing wild-type controls. Statistical significance was tested by the unpaired t-test. (D) qRT-PCR analysis showing the relative expression of Daao1 and hDAO transcripts compared to the housekeeping gene RpL32. RNA was extracted from L3b larvae following constitutive simultaneous Daao1 knockdown and hDAO overexpression (Act5C-GAL4 > Daao1-IR2 + hDAO). Each bar represents the mean ± SEM of at least three independent biological replicates, with individual data points superimposed. Discussion Since the initial discovery of a rare missense mutation that segregates with ALS [4], we and others have identified additional rare, potentially deleterious variants in the DAO gene [29–32]. Although these findings support a role for DAO in ALS aetiology, functional validation in experimental models remains essential to establish causality and delineate underlying mechanisms. To reach these aims and to overcome both the limitations of existing mouse models and the inconsistences reported across studies, we have developed a Drosophila model of DAO loss of function. We demonstrate that flies deficient in the DAO orthologue, Daao1, exhibit phenotypes that recapitulate key clinical and pathological features of ALS including motor impairment and structural abnormalities at the NMJ. These findings mirror those observed in classic fly models of ALS including those targeting genes like C9orf72, SOD1, TARDBP and FUS [26], which are a major cause of familial ALS, and therefore provide definitive functional evidence validating DAO as an ALS-linked gene. Unexpectedly, we found that motor dysfunction occurred independently of D-ser metabolism defects and was associated with disruption of Daao1 activity within both MNs and muscles. RNA-seq further revealed transcriptional alterations in genes involved in metabolism, peptide transport, immune activation, and in the formation and eventual fine-tuning of neural connections (circuits), potentially explaining the ALS-like deficits observed in Daao1 knockdown flies. Together, these data point to a previously unrecognised, non-canonical role for DAO that is partially conserved in humans, as supported by our rescue experiments. It is noteworthy that in flies, excitotoxic neuron dysfunction is unlikely to account for the motor impairments resulting from loss of Daao1. This is supported by the absence of detectable alterations in brain D-ser levels. While fluorescence quantification relied on nuclear Hoechst normalisation, which may not account for cytoplasmic volume differences, the absence of motor phenotype rescue upon Srr co-silencing functionally confirms that D-ser accumulation is not the primary driver of pathology in this model. Furthermore, we also observed that Daao1 function is not only required in neurons but also in muscles, which themselves do not express NMDA receptors. Our findings appear to contradict earlier reports in mice in which lack of DAO activity was found to induce elevated D-ser levels in the CNS [16, 17, 19]. However, it must be underscored that while D-ser was found to be overproduced in the cerebellum of DAO-null mice, D-ser levels remained unchanged in the forebrain [33]. This brain region is known to contain several glutamatergic neurons including ALS-linked MNs in the cerebral cortex. Elevated D-ser plasma levels were also undetectable in monkeys that were administered a DAO inhibitor [23]. We therefore argue against a mechanism primarily driven by glutamatergic excitotoxicity and instead point towards alternative, D-ser-independent pathways through which DAO regulates neuromuscular integrity. Transcriptomic profiling of the DAO-ALS fly model has provided hints about possible non-canonical roles for Daao1 in muscles and neurons. We show transcriptional alterations in genes associated with metabolism, peptide transport, and the immune response. Downregulation of metabolic pathways likely reflects an energetic imbalance that compromises neuromuscular function, whereas an upregulation of fatty acid synthesis and lipid metabolism could be a compensatory reaction to disruption of these processes in ALS as recently revealed in C9orf72 ALS Drosophila models and induced pluripotent stem cell neurons [34]. Upregulation of immune response pathways may indicate secondary inflammatory activation, and is a feature commonly observed in ALS animal models and human samples upon similar transcriptome-based investigations [35–43]. Importantly, the enrichment of differentially spliced genes linked to neuronal morphology and synaptic assembly suggests that DAO also contributes to the refinement of neural circuit connectivity, a novel finding whose underlying pathway warrants further exploration. This is corroborated by data showing maximal Daao1 expression during the pupal stage [44], a critical window of neuronal refinement, synaptic pruning and rewiring [45]. Notably, we showed that constitutive Daao1 knockdown halted developmental progression prior to or during metamorphosis. The partial rescue of phenotypes in the DAO-ALS fly model by human DAO expression supports functional conservation and suggests that certain disease-linked variants may disrupt these non-canonical activities rather than enzymatic function alone. The transcriptomic alterations presented here provide a good foundation for future mechanistic work, including genetic modifier approaches, to establish the key pathways that mediate the DAO-ALS fly model phenotypes. It is worth highlighting that our findings do not support a role for Daao1 in glial cells. To this end, flies with glia-specific Daao1 knockdown exhibited no measurable defects in locomotor performance or flight ability, suggesting that Daao1 function is dispensable in this cell type. This observation aligns with previous mammalian studies reporting minimal or undetectable DAO expression in glial populations [4, 22, 46], reinforcing the notion that DAO acts primarily within neurons and muscles to maintain neuromuscular function. In summary, our study establishes Drosophila as a powerful and genetically tractable model for studying DAO-linked ALS. We show that DAO loss of function leads to motor and synaptic phenotypes analogous to those seen in ALS, accompanied by significant transcriptional and splicing alterations implicating metabolism, immune activation and processes that are key for neuronal remodelling and refinement. These findings not only reinforce DAO as an ALS-associated gene but also uncover a previously unrecognised role for DAO in maintaining neuromuscular integrity through pathways independent of D-serine metabolism. Future work should aim to define the precise molecular mechanisms underlying this function and explore its therapeutic implications in ALS. This work also establishes the DAO-ALS fly model as a valuable tool for studying DAO variants of uncertain significance (VUSs), thereby holding the potential to accelerate diagnoses and end the diagnostic odyssey for patients. Materials and methods Protein sequence alignment Protein sequences of human genes and their Drosophila orthologues were retrieved from the Ensembl database. Pairwise protein sequence alignment was performed using the Clustal Omega algorithm available through the European Bioinformatics Institute (EBI) web server with default parameters. Drosophila husbandry and stocks Flies were reared at 25°C on a 12-hour light/dark cycle using a standard sugar-cornmeal-yeast-agar medium. The wild-type strain was w1118. The bipartite GAL4/UAS was used to express UAS-linked transgenes (reviewed in [47]). The RNAi transgenic constructs Daao1-IR1 (ID: 41927), Daao1-IR2 (ID: 108398), and Srr-IR (ID: 24201) were obtained from the Vienna Drosophila Resource Centre (VDRC), Austria [48]. The following stocks were obtained from the Bloomington Drosophila Stock Center (BDSC; NIH P40OD018537): UAS-Dcr-2 (ID 24648, 24650), UAS-hDAO (ID 97462), the ubiquitous drivers Act5C-GAL4 (ID 4414) and tubP-GAL4 (ID 5138), the muscle-specific drivers Mef2-GAL4 (ID 27390) and how-GAL4 (ID 1767), the pan-neuronal driver elav-GAL4 (ID 458) and the glial-restricted driver Repo-GAL4 (ID 7415). The MN driver OK6-GAL4 was a generous gift from Cahir O’Kane (University of Cambridge, UK). All experimental genotypes were generated using standard genetic crossing schemes. Adult fly viability Adult viability was assessed by collecting and genotyping progeny at regular intervals following eclosion. Viability was calculated as the percentage of eclosed adults of the experimental genotype relative to the number expected based on Mendelian segregation ratios for the corresponding cross. Motor function assays To uncover ALS-like motor impairments, motor function was assessed using climbing and flight performance assays. All experiments were conducted during daylight to minimise potential effects of circadian variation on locomotor behaviour. For the climbing assay, groups of adult flies were gently tapped to the bottom of an empty plastic vial, and the time taken for the first fly in each group to climb past the 8 cm mark was recorded. Each group underwent four independent trials, with at least three groups tested per genotype. For hDAO rescue experiments, the climbing assay was adapted to evaluate individual flies. Flight performance was assessed using the Droso-Drome apparatus, as previously described [49]. The device consists of a 1 L glass cylinder coated internally with a sticky, alcohol-based solution and divided into four 5 cm vertical sectors over a total height of 20 cm. Prior to testing, flies were conditioned by inducing at least three consecutive negative geotaxis responses in an empty vial. Conditioned flies were then released from the top of the apparatus, and their landing positions were recorded. The percentage of flies adhering to each sector was calculated, with a distribution skewed toward the lower sectors indicating impaired flight ability. Each group was tested in four independent trials, and a minimum of four groups per genotype were assayed. For all age-dependent assessments, assays were conducted at specific post-eclosion time points as follows: week 1 (day 5 post-eclosion), week 3 (day 15 post-eclosion), week 4 (day 25 post-eclosion), and week 5 (day 35 post-eclosion). D-serine immunostaining and quantification L3 larvae and adult flies were dissected in phosphate-buffered saline (PBS) to expose the CNS. Dissected tissues were fixed for 20 min in 4% paraformaldehyde (PFA) in PBS, followed by several washes in PBS containing 0.1% Triton X-100 (PBT). Samples were incubated overnight at room temperature with a rabbit anti-D-ser antibody (1:80; ImmunoSmol, France). After washing in PBT, brains were incubated overnight at room temperature with a goat anti-rabbit Alexa Fluor 488 secondary antibody (1:500; Invitrogen, Thermo Fisher Scientific, USA) and Hoechst 33342 (1:500; Sigma-Aldrich, Germany) to label nuclei. Following final washes, samples were mounted in 90% glycerol containing an anti-fade reagent. Fluorescence images were acquired using a Leica Thunder Imager 3D Tissue System at 200× magnification under identical acquisition settings for all samples. D-ser levels were quantified using the LAS X life science imaging software by measuring the D-ser fluorescence-stained areas within the brain, followed by normalisation using Hoechst-stained areas to control for differences in cell numbers and imaging variability. NMJ immunostaining and quantification of morphological parameters L3 larvae were dissected in PBS to expose the body wall musculature. Dissected tissues were fixed for 20 min in 4% PFA in PBS, followed by several washes in PBT. Body wall preparations were then incubated overnight at room temperature with a mouse anti-Discs large (DLG) antibody (1:1000; Developmental Studies Hybridoma Bank, DSHB), as described previously [40]. After washing in PBT, samples were incubated overnight at room temperature with a goat anti-mouse Alexa Fluor 488 secondary antibody (1:500; Invitrogen, Thermo Fisher Scientific, USA), Cy3 AffiniPure goat anti-horseradish peroxidase (HRP; 1:50; Biozol, Germany) and Alexa Fluor 647–conjugated Phalloidin (1:50; Invitrogen) to visualise muscle actin filaments. Following final washes, samples were mounted on glass slides in 90% glycerol containing an anti-fade reagent. Fluorescence images were acquired using a Leica Thunder Imager 3D Tissue System at 200× or 400× magnification, with identical imaging parameters applied across all samples. Quantitative analysis of neuromuscular morphology was performed using the LAS X life science imaging software. Measurements were obtained from ventral longitudinal muscles 6 and 7 within abdominal segment 2 (A2). Muscle area was calculated to control for potential size-related variability. On these same muscles, the NMJ was characterised by quantifying the total area of the DLG-positive postsynaptic terminal, the number of branches (defined as arborisations containing at least two boutons), the total number of individual synaptic boutons per NMJ and individual bouton area per NMJ. RNA extraction Total RNA was extracted from pools of 12–15 L3 larvae per genotype using the RNeasy Plus Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer’s protocol. Briefly, whole larvae were homogenised in lysis buffer, and the resulting lysate was cleared of genomic DNA (gDNA) using the provided gDNA Eliminator spin columns. Total RNA was then bound to and purified using RNeasy Mini spin columns. Quantitative RT-PCR (qRT-PCR) Transcript levels of Daao1 or Srr were quantified from cDNA using the SOLIScript 1-step SolisGreen Kit (Solis Biodyne, Tartu, Estonia) following the manufacturer’s protocol. Gene-specific primers were sourced from Integrated DNA Technologies (Leuven, Belgium) with the following sequences: Daao1 (forward: 5′–GATGACCGATGCCTTCAACT–3′; reverse: 5′– GGCATGTGGTGAAGAAGGAT–3′), Srr (forward: 5′–GCCGCTAATAGATCGCATGGT–3′; reverse: 5′–CCAAACCTCGCTCCAAACAA–3′), hDAO (forward: 5′–GCCATGTCCATTCTCCCAAC–3′; reverse: 5′–ATACCCCAGTGCAGTTGACA–3′) and housekeeping gene RpL32 (forward: 5′–TACAGGCCCAAGATCGTGAA–3′; reverse: 5′–GACAATCTCCTTGCGCTTCT–3′). Relative gene expression for Daao1 or Srr knockdown experiments was calculated using the Pfaffl method. RNA-sequencing and bioinformatic analysis RNA samples, extracted from female L3 larvae, were submitted to the Beijing Genomics Institute (BGI, Denmark) for library preparation and sequencing. Poly(A) + mRNA was enriched from total RNA using oligo(dT)-conjugated magnetic beads, followed by fragmentation. First-strand cDNA was synthesised using random hexamer primers. Libraries were prepared through end-repair, adaptor ligation, and PCR amplification, followed by purification. Single-stranded DNA circles were generated, and amplified by rolling circle replication to produce DNA nanoballs, which were sequenced as 100 bp paired-end reads on the BGI DNBseq platform. Raw sequencing reads were processed for quality control using SOAPnuke [50], and high-quality clean reads were aligned to the Drosophila melanogaster reference genome using HISAT2 [51]. Transcript abundance was quantified with RSEM and normalised as Fragments Per Kilobase of transcript per Million mapped reads (FPKM) [52]. Differentially expressed genes (DEGs) were identified using the DESeq2 [53], applying Benjamini-Hochberg correction for multiple testing. Genes were considered significantly differentially expressed at a false discovery rate (FDR) < 0.05 and a log2 fold change > ±1. DSGs were identified using rMATS [54], to detect five categories of alternative splicing events: SE, A5SS, A3SS, MXE, and RI. KEGG pathway, GO biological pathway, and GO cell location analysis was performed on both DEGs and DSGs using ShinyGO [55] to identify enriched terms. Statistical analysis All quantitative data are presented as mean ± standard error of the mean (SEM) unless otherwise indicated. Statistical analyses were performed using GraphPad Prism version 10.6.0 (GraphPad Software, USA). Comparisons between two independent groups were conducted using an unpaired two-tailed t-test, whereas comparisons against a baseline value were evaluated using a one-sample t-test. For analyses involving multiple groups, a two-way ANOVA followed by Bonferroni’s post hoc test was employed. Statistical significance was defined as p < 0.05. Sample sizes (n), and the number of independent replicates are indicated in the respective figure legends. Supplementary Material Acknowledgements The authors are grateful to Matthew Camilleri for his unwavering technical support with Drosophila work and to all members of the laboratory for their valuable discussions, insightful feedback, and continued support throughout the course of this study. Contributor Information Sylvana Tabone, Department of Physiology and Biochemistry, Faculty of Medicine and Surgery, Biomedical Sciences Building, University of Malta, Msida MSD 2080, Malta; Centre for Molecular Medicine and Biobanking, Biomedical Sciences Building, University of Malta, Msida MSD 2080, Malta. Rebecca Cacciottolo, Department of Physiology and Biochemistry, Faculty of Medicine and Surgery, Biomedical Sciences Building, University of Malta, Msida MSD 2080, Malta; Centre for Molecular Medicine and Biobanking, Biomedical Sciences Building, University of Malta, Msida MSD 2080, Malta. Ruben J Cauchi, Department of Physiology and Biochemistry, Faculty of Medicine and Surgery, Biomedical Sciences Building, University of Malta, Msida MSD 2080, Malta; Centre for Molecular Medicine and Biobanking, Biomedical Sciences Building, University of Malta, Msida MSD 2080, Malta. Author contributions Sylvana Tabone (Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Visualisation, Writing—review & editing), Rebecca Cacciottolo (Funding acquisition, Investigation, Methodology, Supervision, Writing—review & editing), and Ruben J Cauchi (Conceptualisation, Formal analysis, Funding acquisition, Project administration, Supervision, Visualisation, Writing—original draft) Conflicts of interest None declared. Funding This work was partly supported by the University of Malta Research Seed Fund, the Xjenza Malta Fusion R&I Research Excellence Programme 2023 (REP-2023-018), and the Xjenza Malta Research Networking Scheme 2026 (RNS-2026-0030). S.T. is supported by a Tertiary Education Scholarship Scheme (Malta), and R.C. is supported by a Reach High II Scholarship (co-funded by ESF+ 2021–2027). References Associated Data Supplementary Materials