Protein-stabilizing and neurotransmission-potentiating activities of a synaptic chaperone modify spinal muscular atrophy in model mice

Spinal muscular atrophy (SMA) is an oft-fatal infantile-onset neuromuscular disease caused by low SMN protein. Administration of SMN-inducing agents to SMA newborns prevents early mortality, but therapeutic outcomes vary considerably, and disease mechanisms remain poorly understood. Genetic modifiers can provide clues to disease mechanisms and serve as targets for novel treatments. Here, we describe how one such modifier suppresses SMA in model mice. We show that the modifier, an Hspa8G470R synaptic chaperone variant we previously identified, functions beyond an already defined role as an SMN2 splice-switcher. Even in mice lacking the SMN2 gene, the modifier, whether expressed genetically or exogenously, potently suppressed disease, preventing motor neuron degeneration, ameliorating neuromuscular dysfunction and extending lifespan more than ten-fold. Unexpectedly, this was once again associated with incremental SMN increase – an outcome we discovered is linked to Hspa8G470R-mediated autophagy, effects of the modifier on autophagy-associated intermediate complexes and, ultimately, reduced SMN turnover. Interestingly, however, Hspa8G470R also stimulated neuromuscular transmission significantly, raising the effective, functional readily releasable pool of motor neuronal synaptic vesicles. This effect was not limited to mutants alone but apparent in healthy controls too and did not correlate with mere increase in SMN. Combined, these outcomes suggest that Hspa8 governs neuromuscular function in several ways including direct effects on synapses. Mechanisms revealed here shed additional light on pathways gone awry in SMA – ones that might be modulated to develop or refine therapies for neuromuscular disorders at large.

Spinal muscular atrophy, Survival Motor Neuron, genetic modifiers, Hspa8, autophagy (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 2

Homozygous mutations in the Survival Motor Neuron 1 (SMN1) gene result in low levels of its namesake SMN protein and trigger the frequently fatal pediatric-onset neuromuscular disease, spinal muscular atrophy1. Several SMN-inducing agents are now approved for the treatment of SMA2 and newborn screening for the disease routinely offered to new parents3. These developments have radically altered disease outcomes, particularly when infants are treated soon after birth. Still, none of the SMA therapies is curative, with many infants, especially those with 2 SMN2 copies – predictive of the severest form of the disease – failing to meet developmental motor milestones, notwithstanding early treatment4. Therapeutic intervention in these patients has also unmasked previously unappreciated phenotypes, a consequence of having prevented the early mortality associated with severe SMA5,6. These outcomes may be explained by insufficient SMN induction following postnatal therapy and/or critical requirements for the protein in utero. However, they are also impacted by an incomplete understanding of the most critical SMN-linked pathways that are perturbed in SMA and that must be normalized to prevent or reverse disease. Genetic modifiers of disease constitute important windows into the selective vulnerability of organ systems to deficiencies in essential proteins like SMN. Indeed, identifying SMA modifiers has been informative, linking SMN to neuromuscular junction (NMJ) organizers such as Agrin, MuSK and Dok7, to perturbed neuronal endocytosis and to factors essential for neurotransmission, e.g., Syt2 (Synaptotagmin2) and the P/Q type Ca2+ channels in motor neurons7-13. Still, exactly how SMN alters the expression or localization of these proteins is unclear. Consequently, the extent to which pathways associated with these proteins influence the SMA phenotype remains to be fully explained. We previously found that a G470R variant of the synaptic Hspa8 chaperone protein potently suppresses SMA in model mice14. Biochemical studies revealed that the modifier functions as an SMN2 splice modulator, modestly raising functional SMN protein. However, we also found that it enhanced synaptic SNARE complex levels significantly, an observation consistent with known Hspa8 functions15,16. This raised the prospect of a direct protective effect of Hspa8G470R on SMA NMJs but one confounded by the co-incidental SMN2 splice-switching properties of the modifier; increased SMN protein resulting from higher full-length (FL) SMN transcript in the presence of Hspa8G470R may be argued to be the exclusive driver of enhanced neurotransmission and healthier NMJs in modified mutant mice. Here, we attempted to dissociate any direct ameliorative effect of Hspa8G470R from one mediated by its role as an SMN2 splicing modulator by investigating if model mice devoid of the copy gene nevertheless derive benefit from the modifier. We found that Hspa8G470R is equally potent in suppressing disease in such mutants, preserving spinal motor (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 3 neurons, mitigating neuromuscular pathology and extending lifespan significantly. Intriguingly, however, the modifier once again raised SMN – by stabilizing the protein, via its effects as a cellular chaperone of clients destined for autophagic degradation. This effect was nevertheless accompanied by enhanced NMJ neurotransmission and, interestingly, a marked increase in the readily releasable pool (RRP) of synaptic vesicles that exceeded the RRP of even healthy controls – notwithstanding the lower SMN in modified mutants. Our results not only assign a new mechanism of action to the Hspa8G470R modifier in regulating SMN protein but also bolster the notion that it equally likely acts directly at synapses to potentiate neurotransmission and suppress disease. The multiple mechanisms of action we attribute to this variant in modulating the SMA phenotype make it and the pathways it affects appealing targets not just to refine treatments for this pediatric disorder but also to mitigate allied motor neuron conditions.

The Hspa8G470R variant suppresses overt disease and neuromuscular pathology in the Smn2B/- mouse model of SMA We previously employed ‘SMN7’ model mice17 harboring the human SMN2 gene to map and identify a potent suppressor of the SMA phenotype14. The Hspa8G470R disease suppressor raised SMN levels by functioning as an SMN2 splice-modulator. However, it also potentiated NMJ neurotransmission and stimulated synaptic SNARE complex assembly, suggesting a second, SMN-independent disease- modifying effect. Recognizing the challenge of discerning the relative contribution of each of these effects to disease rescue in SMN2-expressing mutants, we sought to examine modifier-driven outcomes in model mice that do not rely on the human copy gene. For this, we turned to the Smn2B/- SMA mutant, wherein the murine Smn allele is engineered to mis-splice and produce low SMN18,19. Because strain

influences disease outcomes, and considering that our earlier study was conducted on mutants derived on the FVB/N background, we made certain that our founder Smn2B breeders were also FVB/N. Single polymorphism nucleotide (SNP) analysis using 146 markers informative for the FVB/N and C57BL/6J strain confirmed that the breeders derived 99.31% of their genomes from the FVB/N strain (Table S1). We thus proceeded with these animals and bred them to Hspa8G470R mice on the same strain

to generate mutants for our study. We found that disease in Smn2B/- mutants bearing Hspa8G470R was markedly suppressed. Median survival of mutants with one Hspa8G470R allele increased from PND19 to PND238; mutants homozygous for the modifier survived even longer, indicative of dose dependency (Fig. 1A). Additionally, overall health as assessed by weight gain during the first three postnatal weeks of life improved (Fig. 1B and Fig. S1A) and agility of Smn2B/-;Hspa8G470R mutants between PND8 and PND10, as determined by righting ability, was significantly enhanced relative to that of (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 4 mutants absent the modifier (Fig. 1C). Finally, we also found that SMA mutants bearing the modifier displayed greater grip strength than cohorts without it (Fig. 1D). Neuromuscular pathology is a signature feature of SMA and is expected to be alleviated by disease suppressors. Accordingly, we began by examining muscle tissue of PND15 Smn2B/- mutants with and without the modifier. We found that myofibers in both the proximal triceps muscles as well as distal gastrocnemius of mutants harboring Hspa8G470R were significantly larger than those of mutants without the modifier (Fig. 2A, B and Fig. S1B), and this was reflected in frequency histograms revealing a shift toward fibers with greater cross-sectional areas in modified mutants (Fig. S1C, D). We next quantified spinal motor neurons in our various cohorts of mice. Expectedly, morphometric counts revealed fewer (~40%) cell bodies in lumbar spinal cord of Smn2B/- mutants relative to healthy controls. In contrast, we found no evidence of significant motor neuron loss in mutants expressing Hspa8G470R (Fig. 2C, D). To complete our assessment of neuromuscular health, we conducted a detailed anatomical assessment of the NMJs in the PND15 mutants and healthy controls. We found that NMJ abnormalities in Smn2B/- were either abolished or significantly mitigated by the modifier. Thus, for instance, endplate denervation in muscles of Smn2B/-;Hspa8G470R/Wt mutants, heterozygous for the modifier, was significantly reduced but not to control levels; NMJ innervation in mutants homozygous for Hspa8G470R was further improved and equivalent to that of controls (Fig. 2E, F and Figs. S1E). We observed similar patterns of rescue when NMJs were examined for nerve terminals abnormally swollen with neurofilament (NF) protein and when endplate size and complexity – as assessed by perforations – was determined (Fig. 2G – I). Curious to ascertain if disease suppression is limited to the neuromuscular system or extends to other organs too, we also examined liver tissue in PND15 controls and SMA mutants with and without the modifier; hepatic pathology has been reported in SMA model mice as well as human patients and is characterized by fatty liver and pronounced microvesicular steatosis20-23. Expectedly, liver tissue of Smn2B/- mutants displayed disrupted architecture and widespread microvesicular vacuolization, consistent with severe steatosis. In contrast, liver sections from Smn2B/-;Hspa8G470R mice revealed a normal cellular architecture, uniform hepatocyte morphology, and complete absence of steatosis (Fig. S2). Collectively, these results demonstrate that Hspa8G470R does not require the presence of the SMN2 gene to suppress the SMA phenotype. Moreover, disease suppression in Smn2B/- mutants is not restricted just to the neuromuscular system but observed in peripheral tissue too. Hspa8G470R effects SMN increase by retarding protein decay Considering how robustly Hspa8G70R protected against SMA, we sought to identify mechanisms of action of the modifier. Smn2B/- mice do not harbor SMN2. However, because the Smn2B allele was engineered to behave like SMN2 and generate a transcript lacking murine Smn exon 7, and given the (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 5 splice-switching property of Hspa8G470R, (refs. 14, 24) we commenced our assessment by examining Smn2B splice products in the various mouse cohorts. Predictably, healthy Smn2B/+ controls produced copious quantities of FL (full-length) murine Smn – from both the wild-type (WT) and 2B alleles – and lower amounts of transcript lacking exon 7 – exclusively from the mutant allele (Figs. 3A, B). Moreover, as expected, the relative abundance of the two transcripts was reversed in Smn2B/- mutants. This ratio did not change in the presence of Hspa8G470R, (Figs. 3A, B and Fig. S3A) suggesting that notwithstanding having engineered the Smn2B allele to behave like SMN2, it is not subject to splice-switching by the modifier. We nevertheless proceeded to assess SMN protein levels in the different mouse cohorts. Consistent with past studies and the lower abundance of FL murine Smn transcript in Smn2B/- mutants, we found greatly reduced SMN protein in these mice relative to that in healthy controls (Figs. 3C, D). Protein levels in Hspa8G470R-expressing mutants remained very low but, surprisingly, did increase. The increase was modest or statistically insignificant in mutants heterozygous for the modifier but became noticeable – and significant – in mutants homozygous for Hspa8G470R (Figs. 3C, D and Fig. S3B). Nevertheless, mean protein amounts in Smn2B/-;Hspa8G470R/G470R mutants never exceeded 25% of WT levels. Considering unaltered FL murine Smn transcript levels but increased SMN protein in mutants harboring Hspa8G470R, we hypothesized a post-translational effect of the modifier. To test this, we assessed SMN turnover in mouse embryonic fibroblasts (MEFs) expressing either WT Hspa8 or the G470R variant. We found that SMN decay was indeed slowed in the presence of the variant (Figs. 3E, F). Moreover, consistent with a modifier-mediated effect on SMN protein, Hspa8G470R did not affect murine Smn transcript stability (Fig. S3C). These results extend our prior findings demonstrating an interaction between SMN and Hspa8 (ref. 14) and reveal the physiological relevance of the interaction, which has been independently verified25. Hspa8 regulates SMN levels via chaperone-assisted selective autophagy (CASA). Having established that Hspa8G470R stabilizes the SMN protein likely via a direct physical interaction, we explored how such stabilization might be effected. To do so, we first investigated how the two proteins interact. This was accomplished by generating a series of deletion constructs expressing specific regions of the two proteins and using these in co-immunoprecipitation (co-IP) assays to identify minimal interacting domains on each. Expectedly, an intact, WT Hspa8 protein bound efficiently to bead- immobilized FL-SMN. Moreover, consistent with previous findings examining the interaction between the variant and human protein14, the affinity of the chaperone for murine SMN was reduced by the G470R variant (Figs. S4A – C). We also found that the N-terminal ATPase and nucleotide binding domain (NBD) of Hspa8 failed to bind SMN, whereas truncated products containing the chaperone’s substrate binding (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 6 domain (SBD) did (Figs. S4A, B); deleting the C-terminal 104 amino acids of Hspa8, which constitutes the lid domain of the protein did not preclude it from binding SMN. In a similar manner, we determined that the Tudor, the Gemin-binding and the proline-rich domains of SMN are important for interacting with Hspa8 (Figs. S4D, E). The somewhat paradoxical observation that the N-terminal domain of SMN, which contains the first 151 amino acids of the protein and excludes the polyproline region binds Hspa8 while the SMNΔP fails to do so suggests that in the absence of the proline-rich region, the C-terminus of SMN sterically blocks Hspa8-SMN binding; neither region is present in the SMN_N construct, possibly enabling binding of the corresponding truncated protein to Hspa8. Notwithstanding this final observation, the overall binding data are consistent with SMN being one of several Hspa8 clients that interact with the latter’s SBD25. Hspa8 is best known for ensuring cellular homeostasis through the sequestration or degradation of nascent, misfolded or aggregated proteins via autophagic pathways26,27. Coincidentally, there are several reports not just of perturbed autophagy in SMA, but that SMN itself is subject to autophagic degradation28-33. Prompted by these reports and our own observations above, we investigated autophagy as a potential link between SMN and Hspa8G470R. To do so, we first subjected WT MEFs to activators or inhibitors of autophagy and examined the effect of such treatment on SMN protein. Consistent with the notion that SMN is at least partly turned over by autophagy and that such turnover is mediated by the autophagy adaptor, p62, we found that serum starvation, an activator of autophagy, reduced the levels of both proteins, and SMN significantly so, whereas a leupeptin + NH4Cl (NL) cocktail, which inhibits lysosomal degradation, raised their concentrations (Figs. 4A - C). Interestingly, the decrease in SMN in serum-starved WT fibroblasts was mitigated in similarly treated cells expressing Hspa8G470R, a result congruent with an inhibitory effect of the variant on autophagic degradation of SMN. This resistance to SMN degradation was also observed in variant-expressing cells treated with the protein synthesis inhibitor, cycloheximide – relative to similarly treated WT cells. Predictably, SMN reduction in serum- starved WT cells treated with cycloheximide was greater than it was in the same cells cultured without drug. These experimental results bolster earlier reports of autophagic control of SMN30-33 and prompted us to investigate how Hspa8 might modulate this process. Hspa8-mediated autophagy is effected several different ways34. We hypothesized that chaperone- assisted selective autophagy (CASA), a form of macroautophagy was the most likely pathway regulating SMN homeostasis, as the two other well-known forms of autophagy involving Hspa8, chaperone- mediated autophagy (CMA) and endosomal microautophagy (eMI), both require a KFERQ pentapeptide recognition motif on client proteins34; this motif is not found on SMN. Hspa8-assisted selective autophagy involves a large (CASA) complex, comprising among others the small heat shock protein, (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 7 Hspb8, a nucleotide exchange factor (NEF) and co-chaperone such as Bag3 and the Hspa8 interactor and E3 ubiquitin ligase, Chip/Stub1 (ref. 27). The assembly of the complex and its interaction with client proteins is summarized in cartoon form (Fig. S5A). To test the idea that SMN proteostasis involves CASA, SMN-FLAG was immobilized on agarose beads and immunoprecipitates from cells co-transfected with either WT Hspa8-Myc or a tagged version of the chaperone variant examined for evidence of CASA complex members. Predictably, Hspa8 was pulled down with SMN, and the G470R variant found to bind less efficiently than WT Hspa8 to SMN. However, we also found copious levels of endogenous Bag3 in the immunoprecipitates and detectable but lower amounts of endogenous Hspb8 and p62 (Fig. 4D). This indicates that SMN does indeed interact with CASA complex components. Because binding of Bag3 to Hspa8 regulates ADP to ATP exchange on the chaperone and thus determines how efficiently client proteins like SMN are released by the chaperone as they progress through the autophagic pathway (Fig. S5), we also sought to examine if and how Hspa8G40R affects Hspa8-Bag3 interaction. Accordingly, we immunoprecipitated tagged WT Hspa8 or Hspa8G470R and quantified levels of endogenous Bag3 that were pulled down by the respective proteins. Expectedly, Bag3 was pulled down with WT Hspa8. Interestingly, however, it bound significantly less well to Hspa8G470R (Figs. 4E, F). This suggests that client proteins like SMN, once bound to Hspa8G470R, are unlikely to be efficiently released as refolded proteins or, in instances where refolding fails, to phagosomes for degradation, instead becoming entrapped within the closed-conformation of the chaperone (Fig. S5B). To test this idea and furthermore avoid variability in interactions resulting from dynamic changes between ADP/ATP-bound Hspa8 states that can be affected by small changes in buffer or temperature, we repeated the SMN-FLAG immunoprecipitations with co- transfected Hspa8G470R or WT Hspa8. However, we did so following treatment of cells with ATPS, a modified form of ATP that is slow to hydrolyze, locks Hspa8 into an ATPS-bound state and is therefore suitable for detecting intermediate CASA complexes. Under this condition, not only did we detect significantly higher amounts of the SMN-Hspa8-Bag3 complex when cells over-expressed the G470R variant but also reduced amounts of SMN-Hspb8 and SMN-p62 complexes in the presence of the variant (Figs. 4G – J). A rise in the SMN-Hspa8G470R-Bag3 complex suggests that SMN does indeed become entrapped within this CASA intermediate. Congruently, reduced binding of SMN to Hspb8 and the autophagosome adaptor p62 in the presence of the G470R variant suggests that SMN is inefficiently released from the SMN-Hspa8G470R-Bag3 complex for lysosomal degradation. The net outcome is a modest but disease-relevant increase in SMN. Overall, our results shed additional light on SMN regulation via a specialized form of macroautophagy that involves the CASA complex. In so doing, they assign a second, novel SMA-modifying mechanism of action to the Hspa8G470R chaperone variant. (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 8 The Hspa8G470R variant enhances synaptic strength and the efficiency of synchronous NMJ neurotransmission Considering the improved agility of Smn2B/-;Hspa8G470R mutants and our prior study14 demonstrating that the variant contributes to motor recovery at least in part by stimulating neuronal SNARE complex formation and normalizing neurotransmission, we inquired if and how well Hspa8G470R had mitigated synaptic dysfunction in Smn2B/- mice. To do so, we focused on PND15 NMJs of the transverse abdominus (TVA), a muscle relevant to the righting reflex we employed (Fig. 1C) to assess motor function. We found that neither the amplitudes nor frequencies of miniature endplate potentials (mEPPs) in Smn2B/- mutants were altered compared to those of Smn2B/+ controls; the presence of Hspa8G470R in the mutants did not change these values, although mEPP frequency was modestly increased in controls heterozygous for the modifier (Figs. S6A, B). Expectedly, however, evoked potential (EPPs) and quantal content (m) in Smn2B/- mutants were significantly lower (65% and 54% respectively) than those of controls (Figs. 5A, B). Notably, Hspa8G470R raised these measures in mutants, with the increases not only became highly significant in Smn2B/-;Hspa8G470R/G470R mice – relative to Smn2B/- mice – but also, surprisingly, surpassing values in healthy Smn2B/+ controls (Figs. 5A, B and Figs. S6C, D). Additionally, Hspa8G470R also raised EPPs and quantal content in healthy controls (Figs. S6E, F). Together, these results suggest that Hspa8G470R is a potent effector of NMJ neurotransmission. Moreover, considering its ability to potentiate neurotransmission even in healthy controls and its propensity to raise EPPs and quantal content in Smn2B/-;Hspa8G470R/G470R mutants beyond those of Smn2B/+ mice – despite the markedly lower SMN levels in the former cohort (Figs. 3C, D) – our results suggest that the modifier’s effects on NMJs are direct and not merely because it stabilizes and raises SMN. The potentiated neurotransmission we observed at NMJs expressing the Hspa8G470R variant could derive from increased vesicle release probability (p), a greater number of functional release sites (n), or both. To estimate which, if either, mechanism might be responsible for the enhanced synaptic function in the presence of the modifier, we employed a simplified binomial model of synaptic transmission in which m = p * n (Ref. 35). This assessment failed to reveal differences in release probability in the various mouse cohorts whether or not they expressed Hspa8G470R (Fig. 5C). In contrast, we found that n values at mutant NMJs rose significantly in the presence of the modifier and, in fact, were normalized to control (Smn2B/+) levels (Fig. 5D). This effect of Hspa8G470R on n was observed at control (Smn2B/+) NMJs too, raising the number of functional release sites at these synapses to supraphysiological levels and mirroring outcomes of EPP and quantal content estimates (Fig. 5D). To show experimentally that the calculated increase in n by Hspa8G470R did indeed reflect greater numbers of effective neurotransmitter release sites, we subjected the NMJs of the various cohorts of mice to high-frequency (20 Hz, 2.5s) nerve stimulation (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 9 trains. Under this protocol, EPP values experience a gradual decrement and then stabilize, reflecting the dynamics – vesicle mobilization, depletion and replenishment – of the RRP. Akin to outcomes following low-frequency (0.5Hz) stimulation (see Figs. 5C, D), we found that neurotransmitter release in response to high-frequency trains was consistently higher at SMA NMJs in the presence of Hspa8G470R (Fig. 5E). This pattern of enhanced release was also seen at control NMJs (Fig. 5F). Plots of cumulative neurotransmitter release (Fig. 5G), the sum total m (Fig. 5H) and plateau amplitudes (Fig. S6G) reflected these findings, confirming the estimated n values (see Fig. 5D) and suggesting that even a single Hspa8G470R allele is sufficient to restore neurotransmission in Smn2B/- mutants to Smn2B/+ control levels. What is the physiological correlate of increased n in the presence of Hspa8G470R and how might one explain the variant’s effects during high-frequency stimulation? We first investigated possible alterations in short-term plasticity. However, we did not observe significant differences in either paired-pulse facilitation (PPF) or short-term depression (STD) among Smn2B/- or Smn2B/+ mice expressing zero, one, or two copies of the modifier (Figs. S6H, I). An alternative explanation for increased neurotransmission in the presence of the G470R variant is an increase in the effective size of the RRP of vesicles. The RRP size is critically influenced by the availability and assembly of SNARE complexes, which mediate vesicle docking and confer fusion competence36. Importantly, Hspa8 is an established component of a tripartite chaperone complex known to be important for SNARE complex assembly, and the G470R variant is known to stimulate the formation of such SNARE complexes14,16. Accordingly, we investigated functional RRP size in the presence or absence of Hspa8G470R. To do so while simultaneously assessing the dynamics of vesicle mobilization, depletion and replenishment, we employed a sequential kinetic model37,38. This model (Figs. S7A, B) posits that docked vesicles in the RRP are the first to be released during repetitive stimulation, that the pool undergoes exponential depletion, and that vesicle recruitment increases over time to sustain release39. Predictably, we estimated that RRP size was substantially lower in Smn2B/- mutants than it was in healthy controls. Significantly, Hspa8G470R restored this to control levels, with RRP size rising in a dose-dependent manner (Fig. 5I). Equally notably, the variant similarly raised RRP size in controls (Fig. 5I), emphasizing its broad impact on synaptic vesicle availability. An increase in effective RRP size by Hspa8G470R might be effected by inducing SNARE complex assembly or slowing disassembly of these complexes, e.g., when N-ethylmaleimide-sensitive factor (NSF) is inhibited36. To rule out impairments in SNARE complex disassembly, we estimated the mean vesicle refilling rate in Smn2B/- and Smn2B/+ mice with or without the modifier. We found that refilling rates remained unchanged irrespective of genotype and modifier copy number (Fig. 5J). These results imply that the modifier does not significantly affect the kinetics of SNARE complex disassembly, endocytosis, or the recruitment of new vesicles to release sites. Instead, the observed RRP increase is most likely due to enhanced SNARE (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 10 complex assembly by the G470R variant. Collectively, our physiological assessments of neuromuscular activity bolster notions of a direct, SMN-independent effect of Hspa8G470R on synaptic strength and NMJ function. Hspa8G470R from an exogenous source mitigates the SMA phenotype in model mice Rescue of the SMA phenotype when Hspa8G470R is genetically expressed raises the possibility of using this specific variant or targeting endogenous Hspa8 to treat the disease. To test this idea, we packaged an expression cassette consisting of the modifier under the control of a chicken -actin promoter into AAV- PHP .eB, a serotype that transduces neurons especially robustly40, and introduced it, or vehicle, via the cerebral ventricles into PND0 Smn2B/- mutants. Predictably, vehicle-treated mutants began exhibiting overt disease at two weeks of age and succumbed to SMA by the third week of life. In contrast, median survival of AAV-PHP .eB-Hspa8G470R-treated mutants increased by ~100% to 33 days; the longest living mutants survived beyond 100 days (Fig. 6A). Consistent with this observation, overall health of the mutants receiving AAV-PHP .eB-Hspa8G470R, as assessed by body weight, was significantly improved by the second postnatal week of life (Fig. 6B and Figs. S8A, B) relative to vehicle-treated Smn2B/- cohorts. This was further reflected in greater agility of these mutants (Fig. 6C). Additionally, we found that overt phenotypic rescue following delivery of AAV-PHP .eB-Hspa8G470R to the mutants was accompanied by significantly less pathology at nerve-muscle connections. At PND18, mean endplate size in mutants receiving the Hspa8G470R construct was significantly larger than that of age-matched vehicle-treated mutants in both the proximal triceps as well as the distal gastrocnemius muscles (Fig. 6D). Correspondingly, denervation of the muscles was suppressed (Figs. 6E, G) and the characteristic, abnormal accumulation of NF protein in SMA nerve terminals markedly reduced (Figs. 6F , G). While the overall rescue of disease in AAV-PHP .eB-Hspa8G470R-treated mutants is more modest than that seen when the variant is expressed endogenously, these results nevertheless constitute compelling evidence that the modifier has a potent SMA-suppressing effect even when expressed from an exogenous source delivered postnatally. Delayed-onset neuromuscular disease, in Smn2B/-;Hspa8G470R/Wt mutants, models intermediate (type II) SMA One copy of the Hspa8G470R allele mitigates the severe phenotype of SMN7 SMA model mice but fails to prevent disease onset before weaning or extend lifespan beyond 4 weeks of age14. In contrast, we were unable to detect overt disease in 2-week-old Smn2B/-;Hspa8G470R/Wt mice (Fig. 1B, C). Still, because our anatomical studies uncovered neuromuscular pathology in these mutants, we continued to monitor them closely through early adulthood. At ~4 months of age, we did not detect any difference in forelimb strength in the mutants (Fig. 7A). Yet, ambulation was perturbed in a proportion of these animals and (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 11 visually linked to altered hindlimb function. This defect worsened until paralysis of the hindquarters occurred between 5 and 6 months of age (Supplemental Video 1). Altered gait by 4 months of age in these mutants was reflected in an abnormal hindlimb splay reflex, whereas no such defect was detected in mutants homozygous for Hspa8G470R (Fig. 7B); ambulation in Hspa8G470R homozygous mutants remained normal even when their heterozygous counterparts were paralyzed (Supplemental Video 1). Hindlimb weakness and paralysis are frequently preceded by neuromuscular pathology. Expectedly, an examination of NMJs in the gastrocnemius of Smn2B/-;Hspa8G470R/Wt mutants at 4 months revealed significantly greater denervation accompanied by abnormal amounts of NF protein in nerve terminals (Figs. 7C – E). Interestingly, AChR clusters in these mutants were also frequently disassembled and fragmented (Fig. 7E), no longer conforming to the normal, elaborate pretzel-shaped endplates we observed in healthy controls and mutants homozygous for the modifier. Consistent with the NMJ defects, we found fewer motor neurons in the lumbar spinal cord of 4-month-old mutants with a single Hspa8G470R allele (Fig. 7F); mutants with two copies of the modifier retained a normal complement of spinal motor neurons. Electrophysiological analysis of NMJ function in the TVA muscle of these two cohorts of mice and age-matched controls reflected many of our anatomical findings. At low frequency (0.5Hz) stimulation, mean evoked potentials trended lower in Smn2B/-;Hspa8G470R/Wt mice but were not statistically different from those in controls (Fig. 7G). In contrast, quantal content in these mutants did differ significantly from control values and was markedly lower (Fig. 7H). mEPP amplitudes were not altered in mutant mice, but interestingly, mEPP frequency was (Figs. S9A, B). The lower quantal content assessed at 0.5Hz stimulation was also apparent in response to high-frequency (20 Hz, 2.5 s) trains (Fig. 7I and Figs. S9C, D). Finally, we inquired if and how Hspa8G470R had affected short-term plasticity and the RRP in young adult mutants. We found that whereas RRP in mutants homozygous for the G470R variant remained normal, RRP in mutants heterozygous for the modifier was statistically lower than control values (Fig. 7J). Paired-pulse facilitation remained unaltered in the mutants (Fig. S9E). However, interestingly, short-term depression rose in both sets of animals relative to that estimated in healthy controls (Fig. S9F), implying altered neurotransmission. Collectively, these results suggest that notwithstanding a relatively normal phenotype assessed at two weeks of age, mutants with just one copy of the modifier develop late-onset neuromuscular disease accompanied by functional and anatomical deficits of the NMJs. The steady deterioration of neuromuscular health in these mutants between PND14 and ~6 months of age, followed by a slower decline that resulted in several such animals surviving beyond 10 months is reminiscent of type II SMA, making this line of mice a unique resource to investigate intermediate forms of the human disease. (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 12

A clear understanding of the precise mechanisms linking low SMN protein to neuromuscular dysfunction in SMA has lagged the development of therapies for the disease. While the therapies, which focus primarily on SMN repletion, are major breakthroughs, having profoundly altered the prognosis for most newborns diagnosed with the disease, they have done comparatively little for patients who received treatment late. Moreover, in very severely affected patients, newborn treatment prevents early mortality but is increasingly associated with novel disease phenotypes4-6. These observations reinforce arguments for a renewed focus on basic SMA mechanisms. One means of revealing such mechanisms is through the identification of novel disease modifiers. Here, we report on one such modifier – a G470R variant of the widely recognized proteostasis factor and synaptic chaperone, Hspa8. Our overall results reveal new mechanisms of action of the modifier. Correspondingly, they shed additional light on the basic biology of SMN and specific pathways relevant to neuromuscular dysfunction in SMA. Five principal findings emerge from our study. First, we demonstrate that the modifier mitigates the SMA phenotype just as potently in model mice devoid of SMN2 as it does in mutants bearing the SMN1 copy gene. This attests to the bona fide SMA-modifying effects of Hspa8G470R and demonstrates that its disease suppressing actions are possible without necessarily altering SMN2 splicing. Second, we show that despite the absence of SMN2, disease modification is effected, at least in part, by augmenting the SMN protein. Our findings link the increase in the protein to effects of Hspa8G470R on SMN turnover, specifically via autophagy- associated pathways. Third, our study provides additional evidence of a direct and SMN-independent effect of the SMA modifier in mitigating disease. This effect, likely related to the role of Hspa8 in chaperoning synaptic SNARE complexes, involves an increase in the readily releasable pool of synaptic vesicles and a corresponding potentiation of neurotransmission not only at SMA NMJs but also at healthy synapses. Enhanced neurotransmission in the presence of Hspa8G470R or agents that similarly modulate Hspa8 function is potentially useful for myriad neuromuscular diseases and thus relevant beyond just SMA. Fourth, we provide proof-of-concept data of the feasibility of employing either Hspa8G470R or factors that target Hspa8 as therapeutic agents. In our study, AAV-mediated expression of Hspa8G470R robustly suppressed disease in model mice. Finally, and serendipitously, we have discovered that a single, G470R variant allele on the Smn2B/- background is sufficient to delay but not prevent neuromuscular dysfunction. Onset of disease in young adult mutants culminated in total hindlimb dysfunction by 6 months of age and constitutes, to our knowledge, the first truly paralytic model of SMA and thus a useful representation of intermediate disease in humans. The SMN2 splice-modulating function we previously attributed to Hspa8G470R confounded inferring any potential direct and SMN-independent SMA-ameliorating effects of the modifier14,41. Accordingly, we (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 13 sought to reveal such mechanisms by testing the modifier in Smn2B/- model mice, which are devoid of SMN2. Reassuringly, we found that the modifier also potently suppresses disease in such mice. Yet, unexpectedly, SMN protein once again rose modestly – not because of a gene splice-switching event but rather owing to reduced SMN turnover. Our results implicate chaperone-assisted selective autophagy (CASA) in this outcome and suggest that normal processing of SMN through this pathway is thwarted not only by weakened interactions between SMN and the G470R variant but also by altered affinity of the variant for the NEF and CASA co-chaperone, Bag3. Because this latter interaction is critically important for release of Hspa8 client proteins, either as a functional, refolded molecule or, in the instance that refolding fails, via the multi-protein CASA complex to the autophagy adaptor, p62, for lysosomal degradation27,32, we infer that SMN processing through the CASA pathway is retarded, and residual protein trapped in Hspa8G470R-SMN-Bag3 intermediate complexes. The apparently paradoxical relative increase in this complex vis-à-vis complexes containing WT Hspa8, considering reduced affinity of Hspa8G470R for both SMN and Bag3 could be an effect of the variant on the overall 3-D structure of Hspa8, rendering its lid domain unable to respond appropriately to Bag3 binding, revert to an open conformation and release bound client protein. While additional work will be needed to reveal such details, it is clear from our work that SMN engages less with both Hspb8, a member of the CASA complex that is important for delivery of misfolded proteins to phagosomes, and p62 – observations strongly suggestive of reduced autophagy turnover of the protein. Collectively, these results have two important implications. First, they link Hspa8 to a second, SMN-modulating function – one directed at the protein. To our knowledge, this is the first factor to have dual effects on SMN, functioning not only as an SMN2 splice-switcher but also serving to stabilize SMN protein. Second, they suggest that selectively targeting autophagy could be of important therapeutic value in the treatment of SMA. The unexpected discovery that Hspa8G470R can bypass SMN2 and nevertheless raise SMN, once again confounds unequivocally assigning to the variant an SMN-independent effect in SMA suppression. Yet, our findings here strengthen notions of such an effect. Given its SMN-stabilizing role, rescue of the SMA phenotype of Smn2B/- mice is not surprising. However, the degree of phenotypic rescue – greater than10- fold increase in lifespan, particularly in Smn2B/-;Hspa8G470R/Wt mice – is. In these Hspa8G470R heterozygous mutants, SMN generally did not exceed that expressed by Smn2B/- mutants, yet disease rescue was significant. In contrast, in a previous study which involved p62 knockdown in SMA mutants and resulted in an increase in SMN of ~75%, lifespan of the mutants was merely enhanced by half32. Direct, Hspa8G470R-mediated effects at NMJs provide one explanation for the more potent disease-mitigating effects of the chaperone variant. Indeed, not only did the modifier potentiate neurotransmission significantly in SMA mutants but also did so in healthy controls. Moreover, direct comparisons of NMJ (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 14 function in Smn2B/-;Hspa8G470R/G470R mutants and healthy (Smn2B/+) controls revealed a modest yet significant potentiation of neurotransmission in the former – notwithstanding markedly lower SMN levels in the mutants. Our results suggest that the enhanced neurotransmission derives less from modest increase of SMN by the modifier than from a primary, synapse-specific Hspa8G470R role – in augmenting the readily releasable pool of synaptic vesicles. This pool of synaptic vesicles is known to be influenced by SNARE complex levels36, which in turn are critically dependent on an Hspa8-containing synaptic chaperone complex15. Modulation of the SMA phenotype via direct effects of Hspa8G470R at NMJs on SNAREs or related factors whose activities are subject to fine-tuning by CASA would assign it a surprising third mechanism of action in mitigating the SMA phenotype. More generally, the effects of the variant on the NMJ would be relevant for motor neuron diseases at large. What is the therapeutic potential of Hspa8G470R? Our AAV-mediated results clearly attest to its therapeutic potential. The more modest rescue we observed when the variant was virally delivered is a likely consequence of at least three factors. First, there is a well-established lag between administration of AAV constructs and their physiological effects42. In severe SMA mice, this mutes phenotypic rescue. Second, disease modification is influenced by AAV biodistribution; the level of vector spread and, consequently, payload expression is invariably inferior to the uniform expression from endogenous alleles. This too is expected to reduce treatment outcomes. Finally, we posit that the presence of endogenous WT Hspa8 in our mutants contributes to the sub-optimal disease-modifying effects of the variant by competing with the latter in relevant SMA-suppressing pathways. Future work will assess if this last handicap can be overcome and treatment outcomes improved by editing the endogenous Hspa8 gene to bring about the single g1489c nucleotide conversion that underlies the G470R variant. Considering the abolition of neuromuscular disease by Hspa8G470R in the 7 line of SMA model mice14, an unexpected outcome of this study was the emergence of a late-onset phenotype, culminating in total hindlimb paralysis of Smn2B/-;Hspa8G470R/Wt mutants. Our analysis of these mutants indicated normal neuromuscular function at PND14. Yet, we did detect modest NMJ pathology in limb muscles of such mutants. These defects grew inexorably, particularly in hindlimbs, eventually causing complete hindlimb immobility by 6 months; age-matched mutants homozygous for the G470R variant remained asymptomatic and overtly normal, although we lost a small (~15%) proportion of such animals at 12 months of age to undetermined causes. What accounts for the severe, albeit delayed-onset phenotype of Smn2B/-;Hspa8G470R/Wt mutants considering abolition of neuromuscular disease by the variant in 7 SMA mice? One intriguing explanation stems from the SMN alleles in the two mouse models. 7 SMA mice17 are endowed with 2 WT copies of human SMN2 and therefore express low but fully functional WT FL-SMN (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 15 protein. In contrast, Smn2B/- mutants do not express any WT SMN; the Smn2B allele is engineered to mis- splice but does so by relying on an SmnG282F mutation in an exon splice enhancer of murine Smn exon 7 (ref. 43). In humans, this residue is conserved and corresponds to position 287 of the SMN protein. Although missense mutations at SMNG287 have not been linked unequivocally to SMA, the following residue at position 288 has (www.ncbi.nlm.nih.gov/clinvar/). Notably, G287 also lies within 7 amino acids of the functionally important YG box, which is implicated in SMN oligomerization44-47. Consequently, it is plausible that SmnG282F in Smn2B/- mutants is a mild, oligomerization-incompetent protein whose effects at low concentrations only become apparent over time. The delayed onset phenotype of Smn2B/- ;Hspa8G470R/Wt mice and the much milder phenotype of Smn2B/-;SMN2Tg/+ mutants48 wherein WT FL-SMN is expected to oligomerize with and stabilize Smn2B protein are consistent with this hypothesis. Notwithstanding these observations, Smn2B/-;Hspa8G470R/Wt mutants with their paralytic phenotype represent a useful and novel model of intermediate SMA. In summary, this study cements the relevance of the synaptic chaperone, Hspa8, and its associated biochemical pathways to SMA biology and neuromuscular health. Modulating these pathways by either directly targeting Hspa8 or via linked factors could be broadly useful for the treatment of human neuromuscular disease. (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 16

We thank members of the Monani lab for comments and suggestions. We are also very appreciative of D.C. De Vivo, A.H.M. Burghes and A.P . Lieberman for critically reading this manuscript and providing feedback. Finally, Manuel de Miguel and Marcos Ortega Medina are recognized for providing technical assistance and use of their histological facilities. Funding: This study was supported by grants from AFM- France, Cure SMA, the Muscular Dystrophy Association (MDA 1064312), the Hope for Children Research Foundation and NIH (R01 NS123292) to U.R.M., the MCIN/AEI/10.13039/501100011033 (PID2023- 150602NB-100) to L.T., FPU22/02843 to A.F-M., and MDA (10.55762/pc.gr.157042) and the Canadian Institutes of Health Research (PJT-186300) to R.K. Author contributions: Conceptualization: Y-R.H, L.T., U.R.M.; Methodology: Y-R.H, A.F-M.; Investigation: Y-R.H, A.F-M.; Supervision: R.K., L.T., U.R.M.; Writing: Y-R.H. & U.R.M. with input from all co-authors. Competing interests: U.R.M. is an inventor on a patent application filed by Columbia University on the use of Hspa8 to treat neurodegenerative disease. Figure Legends (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 17 Figure 1 – The Hspa8 G470R variant mitigates overt disease in SMA model mice. (A) Kaplan-Meier survival curves depicting significant enhancement of mutant lifespan in the presence of the Hspa8 G470R modifier. P < 0.0001 between SMA and SMA;Hspa8 G470R mutants, log-rank test, n = 10 – 32 mice of each cohort. (B) Mutants continue to gain weight in the presence of the modifier. ** , P < 0.01, between SMA and SMA;Hspa8 G470R mutants, one-way ANOVA, n = 14 – 22 mice. (C) Quantified results of righting time depicts improved motor performance of mutants expressing Hspa8 G470R . ** , P < 0.05, between SMA and SMA;Hspa8 G470R mutants, two-way ANOVA, n = 15 mice of each genotype. (D) Graph showing that Hspa8 G470R restores grip strength to SMA mutants. Note: ** , P < 0.01 between SMA and SMA;Hspa8 G470R mutants; two-way ANOVA, n = 3 – 19 mice. Data: mean ± SEM Figure 2 – The Hspa8 G470R variant lessens neuromuscular pathology in SMA model mice. (A) Representative H&E-stained sections of gastrocnemius muscles from PND15 controls and mutants with and without the Hspa8 G470R modifier. Scale bar – 25m. (B) Quantified myofiber sizes in controls and SMA mutants with or without Hspa8 G470R . *** , P 1000 fibers from N = 3 mice of each cohort. (C) Representative lumbar spinal cord sections from the various mouse cohorts depicting ventral horn cells. Scale bar – 20m. (D) Morphometric counts of ventral horn cells in controls and SMA mutants with or without Hspa8 G470R . * , *** , P < 0.05 and P < 0.001, respectively one-way ANOVA, n = 5 mice of each genotype. (E) Immunostains of NMJs in muscle from controls and mutants bearing or devoid of the Hspa8 G470R modifier. Arrows denote denervated endplates. Scale bar – 20m. Quantification of (F) denervated endplates, (G) endplates exhibiting abnormal swellings of NF protein in nerve terminals, (H) endplate complexity and (I) endplate size, in controls and mutants with or without the SMA modifier. Note: * , ** , *** , P < 0.05, P < 0.01 and P 100 endplates from N = 3 – 5 mice of each cohort (panels F – H); Kruskal-Wallis test, n > 300 NMJs from N = 3 mice each for panel I. Data: mean ± SEM Figure 3 – The Hspa8 G470R variant raises SMN levels by slowing protein decay. (A) Representative polyacrylamide gel electrophoresis of the FL and 7 RNA isoforms of the Smn 2B allele in controls and SMA mutants with or without the Hspa8 G470R disease modifier. (B) Quantified results of the two RNA isoforms in diverse tissues of the four cohorts of mice. Kruskal-Wallis tests, N = 7 mice of each cohort. (C) Representative western blot of SMN protein in brain tissue of controls and mutants expressing either the WT or variant form of Hspa8. (D) Quantification of SMN protein in diverse tissues of the four cohorts of mice. * , *** , P < 0.05 and P < 0.001, respectively, one-way ANOVA (muscle and brain samples), (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 18 Kruskal-Wallis test (liver and spinal cord samples), N = 7 mice of each genotype. (E) Representative western blot of SMN protein at indicated time point in MEFs expressing either Hspa8 WT or Hspa8 G470R following treatment with cycloheximide (CHX) to halt protein synthesis. (F) Graphical representation of SMN protein decay in MEFs either WT for Hspa8 or expressing the variant form of it. Note: ***, P < 0.001, AUC, results compiled from 3 independent experiments. Data: mean ± SEM Figure 4 – The Hspa8 G470R variant regulates SMN via chaperone-assisted selective autophagy. (A) Representative western blot of MEFs of indicated genotypes left untreated or treated to induce (serum starvation = S-) or inhibit (NL) autophagy. Quantification of (B) SMN and (C) p62 in fibroblasts depicted in panel A. Note: * , ** , *** , P < 0.05, P < 0.01 and P < 0.001 respectively, one-way ANOVA, n = 6 replicates. (D) Representative co-IP and immunostains of CASA complex components bound to SMN in HEK293 cells co-transfected with either WT Hspa8 or the G470R variant of the chaperone. (E) Representative co- IP and western blot depicting reduced affinity of Bag3 for Hspa8 G470R compared to WT Hspa8. (F) Quantification of Bag3 protein bound to WT or the variant form of Hspa8. Note: ** , P < 0.01, t test, n = 3 replicates. (G) Representative co-IP and immunostains of CASA component proteins pulled down with SMN in the presence of either the WT or the G470R variant form of Hspa8, following treatment with ATPS. Right hand panel shows blots of endogenous or transfected proteins from cells used for the IPs. Quantified amounts of (H) Bag3, (I) p62 and (J) Hspb8 bound to SMN in the presence of WT or the G470R variant form of Hspa8 in experiments represented by panel G. Note: Comparisons made using the t test, n = 3 independent experiments. Also note: ** , *** , P < 0.01 and P < 0.001 respectively; asterisks in panels D and G denote Hspa8-Myc bands and distinguish them from the larger, non-specific band in the gels. Data: mean ± SEM Figure 5 – Hspa8 G470R potentiates neurotransmission in SMA and control mice. Quantified results of (A) evoked potentials and (B) quantal content in PND15 controls and SMA mice with or without the G470R variant. Note: ** , *** , P < 0.01 and P < 0.001 respectively, Kruskal-Wallis test (panel A) and one- way ANOVA (panel B). Estimates of (C) release probability and (D) number of neurotransmitter release sites at NMJs of SMA and control mice with or without Hspa8 G470R . Note: * , ** , *** , P < 0.05, P < 0.01 and P < 0.001 respectively, one-way ANOVA (panel C) and Kruskal-Wallis test (panel D). Mean quantal content at each of 40 stimuli in response to high-frequency stimulation in (E) SMA mutants and (F) controls with 0, 1 or 2 copies of the Hspa8 G470R modifier. Note: means estimated from n ≥ 25 NMJs from N = 3 – 4 mice of each cohort. (G) Traces depicting accumulating quantal content over 2.5s of the high-frequency stimulation in each of the six cohorts of mice analyzed in panels E and F . Note higher quantal content in (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 19 mutants and controls with the modifier relative to their respective cohorts expressing Hspa8 WT . (H) Cumulative quantal content in the various mouse cohorts at the conclusion of the train. Note: * , ** , *** , P < 0.05, P < 0.01 and P < 0.001 respectively, Kruskal-Wallis test. (I) Hspa8 G470R increases the RRP of NMJ vesicles in SMA and control mice. Note: * , ** , *** , P < 0.05, P < 0.01 and P < 0.001 respectively, Kruskal- Wallis test. (J) Neurotransmitter refilling rates in the various mouse cohorts remain unaltered by the modifier; Kruskal-Wallis test. Note: All quantifications here are based on n ≥ 25 NMJs from N = 3 – 4 mice of each cohort. Data: mean ± SEM Figure 6 – AAV-mediated delivery of Hspa8 G470R mitigates disease in SMA mice. (A) Kaplan-Meier survival curves depicting increased lifespan of SMA mutants administered AAV-PHP .eB-Hspa8 G470R . P < 0.0001 between vehicle-treated and AAV-treated mutants, log-rank test, n = 10 – 25 mice of each cohort. Mutants treated with AAV-PHP .eB-Hspa8 G470R (B) gain significantly more weight and (C) perform better in the righting reflex assay than vehicle-treated SMA mice. Note: * , ** , P < 0.05, P < 0.01, t tests between the two groups, n = 10 – 23 mice. Quantified (D) mean NMJ size, (E) degree of denervation and (F) pathology in nerve terminals assessed by NF swellings in the three cohorts of mice. Note: * , ** , *** , P < 0.05, P < 0.01 and P < 0.001 respectively, Kruskal-Wallis tests (panels D, F) and one-way ANOVA (panel E), n = 500 – 800 AChR clusters from N = 4 mice of each cohort. (G) Representative photomicrographs of NMJs in triceps muscle from healthy controls or mutants administered either AAV-PHP .eB-Hspa8 G470R or vehicle. Asterisks denote denervated endplates; arrows highlight relatively simplified nerve terminals abnormally swollen with NF protein. Note fewer such terminals and denervated endplates in muscle from the mutant administered AAV-Hspa8 G470R . Scale bar – 20m. Data: mean ± SEM Figure 7 – Delayed-onset disease in Smn 2B/-; Hspa8 G470R/Wt mutants models type II SMA. (A) Forelimb function remains unaffected in 4-month-old mutants bearing the G470R variant. Comparisons done using one-way ANOVA, n = 7 – 9 mice. (B) A reduced hindlimb splay reflex (arrow) in 4-month-old mutants with one but not two copies of the variant signifies neuromuscular weakness in these limbs. Four-month- old mutants heterozygous for Hspa8 G470R exhibit greater (C) denervation and (D) nerve terminal pathology. Note: * , ** , P < 0.05 and P < 0.01 respectively, one-way ANOVA, n = 135 NMJs from N = 5 mice of each cohort. (E) Representative NMJs in 4-month-old controls and mutants bearing one or two copies of Hspa8 G470R . Note denervation (arrowheads), fragmented endplates (asterisks) and simplified nerve terminals in the form of endbulbs (arrows) in the mutant with one G470R variant allele. Scale bar – 20m. (F) Graph of morphometric counts of lumbar spinal motor neurons in 4-month-old controls and mutants with the modifier. Note: * , ** , P < 0.05 and P < 0.01 respectively, one-way ANOVA, N = 5 mice of each (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 20 cohort. Quantified results of NMJ electrophysiology performed at 0.5Hz stimulation on the three mouse cohorts depict (G) unaltered EPPs and (H) reduced quantal content in mutants heterozygous for the modifier. Note: * , ** , P < 0.05, P < 0.01, Kruskal-Wallis tests (panels G, H). (I) Sum total of quantal content values at the end of 40 stimuli along a high-frequency (20Hz, 2.5s) stimulation train depicts a significantly lower value in 4-month-old mutants heterozygous for the G470R variant relative to the two other mouse cohorts. Note: *** , P < 0.001, Kruskal-Wallis test. (J) Fewer readily releasable vesicles were estimated in SMA mice heterozygous for the modifier. Note: * , P < 0.05, one-way ANOVA. Also note that samples sizes of n = 50 – 60 NMJs from N = 5 – 6 mice were employed (panels G – J). Data: mean ± SEM. (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 21

Animals All mouse work was performed in accordance with the NIH Guidelines on the Care and Use of Lab Animals, complied with all ethical regulations, and was approved by the IACUC of Columbia University (Protocol Number AC-AABE8550). Mice were housed in a controlled environment on a 12h light/12h dark cycle with food and water, and all efforts were made to minimize suffering. Smn2B/2B and Hspa8G470R mice on the FVB/N genetic background were previously described14,49. Smn2B/2B; Hspa8G470R/G470R mice were crossed with Smn+/− homozygous for the G470R variant or bearing just WT Hspa8 to obtain mutants with one or two copies of the modifier. Smn2B/2B x Smn+/− crosses generated Smn2B/− mice. Smn2B/+ mice served as controls in our experiments. Genotyping was performed by PCR on tail DNA using primers listed in Table S2. Mouse work in the Tabares lab was conducted in accordance with the European Council Directive for the Care and Use of Laboratory Animals and was approved by the Animal Care and Ethics Committee of the University of Seville. Righting ability of the various mouse cohorts was assessed as previously described50. Briefly, righting reflex was assessed daily from P5 to P10 by placing the mouse on its back and measuring the time it took to turn upright on its four paws (righting time). For each testing session, the assay was performed thrice and the mean reported. Grip strength was performed using a force transducer (Bioseb #GT3) by positioning the animals on a wire grid and gently pulling the subject horizontally along the grid. Each subject was assessed 3 – 5 times with 1min. rest intervals between tests and means reported. Transcript and protein assessments Total RNA was isolated using TRIzol reagent (Invitrogen) according to the manufacturer’s instructions and subsequently reversed transcribed into cDNA with the SuperScript™ III First-Strand Synthesis System (Invitrogen). Quantitative PCR was performed in triplicate on a CFX96 Real-Time System (Bio-Rad) using the SsoAdvanced Universal SYBR Green Supermix (Bio-Rad) and raw values normalized to endogenous Gapdh or β-Actin mRNA levels; for primer sequences, see Table S2. For immunoblot analysis, tissue or cell lysates from Mouse Embryonic Fibroblasts (MEFs) were used. Briefly, lysates were made with modified RIPA buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS and 1X protease/phosphatase inhibitors cocktail (Roche) and boiled at 95°C for 5 minutes in 1X SDS sample buffer. For tissue-derived protein, tissue was homogenized and incubated on ice for 20 min. Lysates were then centrifuged for 30 min at 14,000g to remove the insoluble debris. Protein amounts were assessed using the BCA protein assay kit (Thermofisher) and concentrations estimated on an Infinite F50 Plus plate reader supplied with Magellan data analysis software (Tecan). Western blotting (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 22 was conducted by separating proteins (10-50g) on 10% or 12% SDS-PAGE gels and then transferring the proteins to a PVDF membrane (Millipore Inc.) as previously described51. Blots were blocked in 1X TBST (Tris-buffered saline with 0.1% Tween 20) containing 5% nonfat milk, and primary antibody used to probe blots for 1hr at room temperature or for 18hrs at 4°C according to the supplier’s recommendations. Following incubation with primary antibody membranes were washed (1X TBST) 3 times with agitation for 5 minutes at room temperature and then incubated with appropriate secondary antibodies for 1hr at room temperature. Antibodies used for the western blotting are as follows: SMN (1:1000, BD Biosciences, Cat#610647), p62 (1:1000, Cell Signaling, Cat#5114), β-tubulin (1:1000, Santa Cruz, Cat#SC53140), - tubulin (1:1000, Santa Cruz, Cat#SC-398103), β-actin (1:1000, Santa Cruz, Cat#SC47778), Goat anti- Rabbit IgG-HRP (1:10,000, Jackson ImmunoResearch, Cat#111-035-003) and Goat anti-Mouse IgG-HRP (1:10,000, Jackson ImmunoResearch, Cat# 115-035-003). Protein bands were visualized using the ECL kit (Bio-Rad Inc.) and images captured on a ChemiDoc Imaging System (Bio-Rad Inc.). Band intensities were assessed using the ImageJ software (NIH, Bethesda, MD, USA) or Image Lab (Bio-Rad Inc.). Plasmids and constructs The SMN-FLAG and Hspa8-Myc expression vectors were generated by inserting the full-length murine Smn and murine Hspa8 cDNAs into the Ecor1-Xho1 and Kpn1-Xho1 sites of pcDNA3.1-FLAG and pcDNA3.1-Myc respectively. The full-length constructs were used as templates to generate all deletion, truncation and substitution mutants (Gene Synthesis system, GeneScript). The various constructs used are as follows: SMN(FL)-FLAG, SMNΔG-FLAG, SMNΔT-FLAG, SMNΔP-FLAG, SMNΔY-FLAG, SMN_N-FLAG, Hspa8WT-Myc, Hspa8G470R-Myc, Hspa8_NBD-Myc, Hspa8_SBD-Myc and Hspa8ΔLid-Myc. The amino acid boundaries corresponding to each deletion or truncation mutant are detailed in Fig. S4. All constructs were verified by DNA sequencing. Cell cultures, transfections and co-immunoprecipitation studies Mouse embryonic fibroblasts were obtained from either WT or Hspa8G470R/G470R E12.5-day-old embryos. Cells were cultured in DMEM (5% CO2 at 37°C) supplemented with 10% fetal bovine serum, 100 units/ml penicillin and 100μg/ml streptomycin. To measure the half-life of SMN protein and mRNA, cells were treated with 80μg/ml of cycloheximide and 7ug/ml actinomycin D (Sigma-Aldrich), respectively for the indicated periods of time. To inhibit lysosomal activity, cells were treated with 100μM leupeptin (Fisher BioReagents, Thermo Fisher Scientific) and 20mM ammonium chloride (Sigma-Aldrich) for 6 hours with, or without 80ug/ml of cycloheximide. To identify domains in Hspa8 and SMN that mediate their interactions with each other, Myc and FLAG-tagged versions of these cDNAs were co-transfected [Lipofectamine 3000 (Life Technologies, Thermo Fisher Scientific)] into and over-expressed in HEK293 cells. Twenty-four hours later, cells were (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 23 washed once with ice-cold PBS then harvested in NP-40 lysis buffer containing 50mM Tris-HCl pH 7.5, 150mM NaCl, 1% NP-40, 10% glycerol and 1X protease/phosphatase inhibitors cocktail (Roche). The resulting lysates were clarified by centrifugation at 16,000g for 20 min at 4°C, then pre-cleared by incubating with 20l of Protein A/G Plus Agarose (Thermofisher Inc.) for 1hr. Next, SMN-FLAG was immunoprecipitated using 20l of anti-DYKDDDDK IP Resin (GenScript, Cat#L00425) by end-over-end rotation (8hrs at 4°C). The immunoprecipitated proteins were washed (3X) with ice-cold lysis buffer and following the final wash, the supernatant was removed and the beads resuspended in 1X SDS sample, boiled for 5mins., resolved by SDS-PAGE and then subjected to immunoblot analysis as described above. Blots were probed using anti-DYKDDDDK (1:1000, GenScript, Cat#A00187), anti-MYC (1:1000, Cell Signaling, Cat#2276), anti-Bag3 (1:1000, Proteintech, Cat#10599-1-AP) and/or anti-Hspb8 (1:1000, Proteintech, Cat# 15287-1-AP) antibodies as indicated in Fig. 4. All immunoblots depicted are representatives of at least three experiments that demonstrated similar results. Note that detection of SMN-p62 and SMN-Hspb8 complexes was enhanced by incubating SMN-FLAG immunoprecipitate with ATPS, a modified form of ATP relatively slow to undergo hydrolysis and therefore suitable for identifying ATP-bound intermediate CASA complexes. For this, following the final wash of immunoprecipitated SMN- FLAG complexes, proteins were resuspended in nucleotide buffer (25mM HEPES pH 7.4, 100mM KCl, 5mM MgCl₂, 0.1% NP-40) containing 1mM ATPS and 5mM MgCl2. Bead suspensions were incubated for 5min on ice with occasional gentle mixing. Following incubation, beads were briefly centrifuged, washed once with NP-40 wash buffer to remove excess nucleotide, supernatant removed and beads boiled in 2X SDS sample buffer for 5 min. Eluates were analyzed by SDS–PAGE and immunoblotting as described above. Anatomical studies Motor neuron, NMJ and muscle histology were essentially carried out as previously described14,50 and detailed below. Motor neurons: Mice were euthanized, subjected to transcardial perfusion [4% PFA in PBS (pH 7.4)] and lumbar spinal cord (L1-L5) extracted. The tissue was then post-fixed in the same fixative, cryo-protected initially in 20% and then 30% sucrose before embedding it in Tissue-Tek Optimal Cutting Temperature (OCT) compound (Fisher) for cryostat sections. All embedded tissue was frozen on dry ice and stored at -80°C until processing. Cryosections (20μm) were collected onto Superfrost Plus glass slides (Fisher) using a CM3050S cryostat (Leica). The sections were washed once with 1X PBS for 5 minutes to remove OCT, overlaid with 4% PFA for 10 min, washed again (1X PBS) and then permeabilized with 0.5% Triton X- 100 (5 min.). To stain the motor neurons, sections were placed in blocking buffer (5% normal donkey (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 24 serum, 3% BSA, 0.1% Triton X-100 in PBS) for 1h, then incubated (4°C, overnight) with anti-ChAT primary antibody (1:100, Millipore, Cat#AB144P) diluted in blocking buffer, following which they were washed (4 X 15 min) in 1X PBST. The following day, the sections were incubated with secondary antibody, Alexa Fluor- 594 conjugated donkey anti-goat IgG (1:1000, Invitrogen), for 2h. After a second round of washing (4 X 15 min.) in 1X PBST, the sections were mounted in Vectashield (Vector Labs, Burlington, VT, USA) containing DAPI. Motor neurons were visualized either or a Nikon 80i fluorescent microscope (Nikon) or a Leica TCS SP8 laser scanning confocal microscope (Leica). ChAT-positive motor neurons soma in the ventral horns were enumerated between L1 and L5 regions of the spinal cord; sections were analyzed at an effective spacing of ~50m to avoid counting soma twice. NMJs: NMJs of triceps brachii or gastrocnemius muscles were stained as previously described49. Briefly, muscle tissue was fixed and permeabilized with 100% methanol (10 min, −20°C), then teased longitudinally into small bundles under a dissecting microscope before incubating (1h) with blocking buffer (5% normal serum, 0.1% Triton X-100 in PBS). It was then sequentially incubated for overnight periods at 4°C with an anti-neurofilament antibody (1:1000, Millipore), followed by Alexa Fluor-488 conjugated donkey anti-rabbit IgG secondary antibody (1:1000, Invitrogen) and rhodamine-α- bungarotoxin (1:1000, Invitrogen). Each incubation was followed by a washing step (4 X in 1X PBST for 15 min). Tissues were then mounted in anti-fade medium (Vector Labs), and NMJs imaged using a Leica SP8 confocal microscope equipped with LAS X software (v1.9.0.13747). NMJs exhibiting no innervation or, <50% overlap between NF and α-bungarotoxin–label, were classified as denervated. Characterization and quantification of innervated NMJs, defective terminals, and motor endplates was conducted as previously described14,50. All images were analyzed off-line using the Leica LAS X software (v1.9.0.13747). Muscles: Muscle was flash-frozen in isopentane chilled with liquid nitrogen, then embedded in Tissue- Tek OCT medium, and 12-µm transverse cryosections cut on a Leica cryostat (Leica, Deerfield). Sections were stained with hematoxylin and eosin (Sigma-Aldrich) and mounted using Cytoseal XYL (Fisher Scientific). Muscle morphology, fiber size, and fiber number were evaluated using a Nikon Eclipse 80i fluorescence microscope equipped with a SPOT 4.5 camera (Diagnostic Instruments, Sterling Heights, MI, USA) and analyzed with ImageJ software (NIH, Bethesda, MD, USA). Fiber size measurements were obtained from ≥100 fibers per sample. Liver: Livers were fixed in 4% paraformaldehyde for 48-72 h at 4 °C and then processed at the Department of Normal and Pathological Cytology and Histology at the University of Seville. All the samples were dehydrated, cleared with xylol, and embedded in paraffin wax using a Spin Tissue Processor STP 120 (Myr). Paraffin block tissues were cut with a microtome (HM 310, Thermo Scientific) at 4 µm thickness. (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 25 Sections were stained with Haematoxylin and Eosin (H&E) using standard methods and assessed by light microscopy (Zeiss Axiovert 40C, x10, x40 objectives). AAV-mediated Hspa8G470R treatment Mice were administered AAV-PHP .eB-Hspa8G470R (6X1010 GC per pup in a volume of 2μl) at PND0 via the cerebral ventricles. Briefly, pups were cryo-anesthetized on a paper towel on wet ice for 3 mins until no pedal withdrawal reflex was observed. ICV injections were conducted using a 1-mL insulin syringe with a pulled glass capillary needle tip (1μm diameter). Vectors were injected into the left and right hemisphere 2mm lateral to the midline, midway between bregma and lambda, and to a depth of 1mm below the surface of the skull. Following injection, pups were placed on a warming blanket and returned to the dam in their own cage once they were visibly active. Pups were monitored every day following virus delivery and were weaned at 21days. Acute neuromuscular preparations Mice were euthanized with 100% CO₂ and the TVA muscle dissected as previously described10. Preparations were continuously perfused with a solution containing (in mM): 135 NaCl, 4 KCl, 2 CaCl₂, 1 MgCl₂, 15 NaHCO₃, 0.33 NaH₂PO₄, and 10 glucose. The solution was equilibrated with a gas mixture of 95% O₂ and 5% CO₂. Electrophysiology The motor nerve was stimulated using a suction electrode with square wave pulses of 0.15 ms duration and 2–10 mV amplitude. Intracellular recordings from single muscle fibers near motor endplates were obtained using glass microelectrodes (10–20 MΩ) filled with 3 M KCl and connected to an intracellular amplifier (TEC-05X; NPI Electronic, Germany). Evoked endplate potentials (EPPs) and miniature EPPs (mEPPs) were recorded at room temperature (22–23°C), as previously described38. Muscle contractions were blocked by adding 2 μM μ-conotoxin GIIIB (Alomone Labs, C-270, Israel) a selective inhibitor of voltage-gated sodium channels in skeletal muscle, to the bath. Recordings were sampled at 20 kHz. EPP and mEPP amplitudes were normalized to a resting membrane potential of −70 mV , and EPPs were corrected for nonlinear summation52. The quantal content (m) was calculated using the direct method: m=Average mEPP amplitude/Average EPP amplitude. During high-frequency stimulation trains, m was calculated as the ratio between each EPP and the mean mEPP amplitude under each experimental condition. To estimate the readily releasable pool (RRP) size, m values during the train were plotted over time and fitted to a sequential model as described37. The model assumes that quanta released upon stimulation originate from the RRP , which is depleted exponentially. Recruitment begins after the first (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 26 stimulus and rises sigmoidally to a plateau as the RRP depletes. The curve of quantal content over time, m(t), was fitted to the following function: where, A = Initial quantal content, B = Time constant of RRP depletion, C = Plateau amplitude, D = Vesicle recruitment half-time, and E = Steepness of the recruitment curve. Note: The sigmoid component was constrained to begin at zero. Integration of the first exponential component provided an estimate of the RRP size (also see Fig. S7). Statistical analyses Sample sizes for the various experiment are detailed in the figure legends. Kaplan-Meier survival curves were assessed for differences using the log-rank test equivalent to the Mantel-Haenszel test. Two-tailed unpaired Student’s t-test or Mann-Whitney tests – when data was not normally distributed – were employed to compare pairs of cohorts with one another. One-way ANOVA followed by Tukey’s post- hoc comparison, or the Kruskal-Wallis test followed by Dunn’s post-hoc comparison – when data was not normally distributed – were used to compare three or more cohorts for statistical differences. For protein and RNA stability comparisons, AUC (Area Under Curves) were calculated and an overall P value assigned based on t-tests of the AUCs. Data are represented as mean ± SEM unless otherwise indicated. P < 0.05 was considered significant. Statistical analyses were performed with GraphPad Prism v9.5.1 (GraphPad Software). Data availability Underlying values associated with data presented in the study may be requested and will be made available by the lead contact, Umrao R. Monani ([email protected]). Supplemental Information Table S1 – Single nucleotide polymorphism (SNP) analysis of Smn2B/- breeders used to generate mutants for the study. Table S2 – Key resources used for the study. Video S1 – Video depicting hindlimb paralysis in Smn2B/-;Hspa8G470R/Wt mutants but not a littermate homozygous for the modifier (Smn2B/-;Hspa8G470R/G470R). (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 27

1. Lefebvre, S., Burglen, L., Reboullet, S., Clermont, O., Burlet, P., Viollet, L., Benichou, B., Cruaud, C., Millasseau, P., Zeviani, M., and et al. (1995). Identification and characterization of a spinal muscular atrophy-determining gene. Cell 80, 155-165. 10.1016/0092-8674(95)90460-3. 2. Oskoui, M., & Servais, L. (2023). Spinal Muscular Atrophy. Continuum (Minneapolis, Minn.) 29, 1564–1584. https://doi.org/10.1212/CON.0000000000001338 3. Vrščaj, E., Dangouloff, T., Osredkar, D., Servais, L., & SMA NBS World Study Group (2024). Newborn screening programs for spinal muscular atrophy worldwide in 2023. J. Neuromusc Dis. 11, 1180–1189. https://doi.org/10.1177/22143602241288095 4. Aragon-Gawinska, K., Mouraux, C., Dangouloff, T., & Servais, L. (2023). Spinal Muscular Atrophy Treatment in Patients Identified by Newborn Screening-A Systematic Review. Genes 14, 1377. https://doi.org/10.3390/genes14071377 5. Baranello, G., & Neurodevelopment in SMA Working Group (2024). The emerging spectrum of neurodevelopmental comorbidities in early-onset Spinal Muscular Atrophy. Eur J Paediatr Neurol. 48, 67–68. https://doi.org/10.1016/j.ejpn.2023.11.006 6. Buchignani, B., Coratti, G., Cutrì, C., Scattoni, M. L., Pane, M., Palermo, C., Leone, D., De Sanctis, R., Stanca, G., Antonaci, L., et al (2025). Neurodevelopmental and mental disorders in children with type I and presymptomatic spinal muscular atrophy. Sci Rep. 15, 26984. https://doi.org/10.1038/s41598-025-12484-8 7. Delers, P., Sapaly, D., Salman, B., De Waard, S., De Waard, M., & Lefebvre, S. (2022). A link between agrin signalling and Cav3.2 at the neuromuscular junction in spinal muscular atrophy. Sci Rep 12, 18960. https://doi.org/10.1038/s41598-022-23703-x 8. Feng, Z., Lam, S., Tenn, E. S., Ghosh, A. S., Cantor, S., Zhang, W., Yen, P. F., Chen, K. S., Burden, S., Paushkin, S., et al. (2021). Activation of Muscle-Specific Kinase (MuSK) Reduces Neuromuscular Defects in the Delta7 Mouse Model of Spinal Muscular Atrophy (SMA). Int J Mol Sci. 22, 8015. https://doi.org/10.3390/ijms22158015 9. Kaifer, K. A., Villalón, E., Smith, C. E., Simon, M. E., Marquez, J., Hopkins, A. E., Morcos, T. I., & Lorson, C. L. (2020). AAV9-DOK7 gene therapy reduces disease severity in Smn2B/- SMA model mice. Biochem Biophy Res Commun. 530, 107–114. https://doi.org/10.1016/j.bbrc.2020.07.031 10. Tejero, R., Lopez-Manzaneda, M., Arumugam, S., & Tabares, L. (2016). Synaptotagmin-2, and -1, linked to neurotransmission impairment and vulnerability in Spinal Muscular Atrophy. Hum Mol Genet. 25, 4703–4716. https://doi.org/10.1093/hmg/ddw297 11. Jablonka, S., Beck, M., Lechner, B. D., Mayer, C., & Sendtner, M. (2007). Defective Ca2+ channel clustering in axon terminals disturbs excitability in motoneurons in spinal muscular atrophy. J Cell Biol. 179, 139–149. https://doi.org/10.1083/jcb.200703187 12. Hosseinibarkooie, S., Peters, M., Torres-Benito, L., Rastetter, R. H., Hupperich, K., Hoffmann, A., Mendoza-Ferreira, N., Kaczmarek, A., Janzen, E., Milbradt, J., et al. (2016). The Power of Human Protective Modifiers: PLS3 and CORO1C Unravel Impaired Endocytosis in Spinal Muscular Atrophy and Rescue SMA Phenotype. Am J Hum Genet. 99, 647–665. https://doi.org/10.1016/j.ajhg.2016.07.014 13. Walsh, M. B., Janzen, E., Wingrove, E., Hosseinibarkooie, S., Muela, N. R., Davidow, L., Dimitriadi, M., Norabuena, E. M., Rubin, L. L., Wirth, B., et al. (2020). Genetic modifiers ameliorate endocytic (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 28 and neuromuscular defects in a model of spinal muscular atrophy. BMC Biol. 18, 127. https://doi.org/10.1186/s12915-020-00845-w 14. Kim, J. K., Jha, N. N., Awano, T., Caine, C., Gollapalli, K., Welby, E., Kim, S. S., Fuentes-Moliz, A., Wang, X., Feng, Z., et al. (2023). A spinal muscular atrophy modifier implicates the SMN protein in SNARE complex assembly at neuromuscular synapses. Neuron 111, 1423–1439.e4. https://doi.org/10.1016/j.neuron.2023.02.004 15. Bhasne, K., Bogoian-Mullen, A., Clerico, E. M., & Gierasch, L. M. (2024). The Hsc70 system maintains the synaptic SNARE protein SNAP-25 in an assembly-competent state and delays its aggregation. J Biol Chem. 300, 108001. https://doi.org/10.1016/j.jbc.2024.108001 16. Sharma, M., Burré, J., & Südhof, T. C. (2011). CSPα promotes SNARE-complex assembly by chaperoning SNAP-25 during synaptic activity. Nat Cell Biol. 13, 30–39. https://doi.org/10.1038/ncb2131 17. Le, T. T., Pham, L. T., Butchbach, M. E., Zhang, H. L., Monani, U. R., Coovert, D. D., Gavrilina, T. O., Xing, L., Bassell, G. J., & Burghes, A. H. (2005). SMNDelta7, the major product of the centromeric survival motor neuron (SMN2) gene, extends survival in mice with spinal muscular atrophy and associates with full-length SMN. Hum Mol Genet 14, 845–857. https://doi.org/10.1093/hmg/ddi078 18. Hammond, S. M., Gogliotti, R. G., Rao, V., Beauvais, A., Kothary, R., & DiDonato, C. J. (2010). Mouse survival motor neuron alleles that mimic SMN2 splicing and are inducible rescue embryonic lethality early in development but not late. PloS One 5, e15887. https://doi.org/10.1371/journal.pone.0015887 19. Bowerman, M., Murray, L. M., Beauvais, A., Pinheiro, B., & Kothary, R. (2012). A critical smn threshold in mice dictates onset of an intermediate spinal muscular atrophy phenotype associated with a distinct neuromuscular junction pathology. Neuromusc Disord. 22, 263–276. https://doi.org/10.1016/j.nmd.2011.09.007 20. Deguise, M. O., Baranello, G., Mastella, C., Beauvais, A., Michaud, J., Leone, A., De Amicis, R., Battezzati, A., Dunham, C., Selby, K., et al. (2019). Abnormal fatty acid metabolism is a core component of spinal muscular atrophy. Ann Clin Transl Neurol. 6(8), 1519–1532. https://doi.org/10.1002/acn3.50855 21. Reilly, A., Beauvais, A., Al-Aarg, M., Yaworski, R., Sutton, E. R., Thebault, S., & Kothary, R. (2024). Peripheral defects precede neuromuscular pathology in the Smn2B/- mouse model of spinal muscular atrophy. J Neuromusc Dis. 11, 1200–1210. https://doi.org/10.1177/22143602241288036 22. Sutton, E. R., Beauvais, A., Yaworski, R., De Repentigny, Y., Reilly, A., Alves de Almeida, M. M., Deguise, M. O., Poulin, K. L., Parks, R. J., Schneider, B. L., et al. (2024). Liver SMN restoration rescues the Smn2B/- mouse model of spinal muscular atrophy. EBioMedicine 110, 105444. https://doi.org/10.1016/j.ebiom.2024.105444 23. Leow, D. M., Ng, Y. K., Wang, L. C., Koh, H. W., Zhao, T., Khong, Z. J., Tabaglio, T., Narayanan, G., Giadone, R. M., Sobota, et al. (2024). Hepatocyte-intrinsic SMN deficiency drives metabolic dysfunction and liver steatosis in spinal muscular atrophy. J Clin Invest. 134, e173702. https://doi.org/10.1172/JCI173702 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 29 24. Zheng, S., Damoiseaux, R., Chen, L., & Black, D. L. (2013). A broadly applicable high-throughput screening strategy identifies new regulators of Dlg4 (Psd-95) alternative splicing. Genome Res. 23, 998–1007. https://doi.org/10.1101/gr.147546.112 25. Ryu SW, Stewart R, Pectol DC, Ender NA, Wimalarathne O, Lee JH, Zanini CP , Harvey A, Huibregtse JM, Mueller P , and Paull TT (2020). Proteome-wide identification of HSP70/HSC70 chaperone clients in human cells. PLoS Biol. 18, e3000606. 10.1371/journal.pbio.3000606 26. Liu T, Daniels CK, and Cao S (2012). Comprehensive review on the HSC70 functions, interactions with related molecules and involvement in clinical diseases and therapeutic potential. Pharmacol Ther. 136, 354–374. 10.1016/j.pharmthera.2012.08.014. 27. Tedesco, B., Vendredy, L., Timmerman, V., & Poletti, A. (2023). The chaperone-assisted selective autophagy complex dynamics and dysfunctions. Autophagy 19, 1619–1641. https://doi.org/10.1080/15548627.2022.2160564 28. Garcera, A., Bahi, N., Periyakaruppiah, A., Arumugam, S., & Soler, R. M. (2013). Survival motor neuron protein reduction deregulates autophagy in spinal cord motoneurons in vitro. Cell Death Dis. 4, e686. https://doi.org/10.1038/cddis.2013.209 29. Custer, S. K., & Androphy, E. J. (2014). Autophagy dysregulation in cell culture and animals models of spinal muscular atrophy. Mol Cell Neurosci. 61, 133–140. https://doi.org/10.1016/j.mcn.2014.06.006 30. Periyakaruppiah, A., de la Fuente, S., Arumugam, S., Bahí, N., Garcera, A., & Soler, R. M. (2016). Autophagy modulators regulate survival motor neuron protein stability in motoneurons. Exp Neurol. 283, 287–297. https://doi.org/10.1016/j.expneurol.2016.06.032 31. Piras, A., Schiaffino, L., Boido, M., Valsecchi, V., Guglielmotto, M., De Amicis, E., Puyal, J., Garcera, A., Tamagno, E., Soler, R. M., et al. (2017). Inhibition of autophagy delays motoneuron degeneration and extends lifespan in a mouse model of spinal muscular atrophy. Cell Death Dis. 8, 3223. https://doi.org/10.1038/s41419-017-0086-4 32. Rodriguez-Muela, N., Parkhitko, A., Grass, T., Gibbs, R. M., Norabuena, E. M., Perrimon, N., Singh, R., & Rubin, L. L. (2018). Blocking p62-dependent SMN degradation ameliorates spinal muscular atrophy disease phenotypes. J Clin Invest. 128, 3008–3023. https://doi.org/10.1172/JCI95231 33. Sansa, A., Hidalgo, I., Miralles, M. P., de la Fuente, S., Perez-Garcia, M. J., Munell, F., Soler, R. M., & Garcera, A. (2021). Spinal Muscular Atrophy autophagy profile is tissue-dependent: differential regulation between muscle and motoneurons. Acta Neuropathol Commun. 9, 122. https://doi.org/10.1186/s40478-021-01223-5 34. Tekirdag, K., & Cuervo, A. M. (2018). Chaperone-mediated autophagy and endosomal microautophagy: Joint by a chaperone. J Biol Chem. 293, 5414–5424. https://doi.org/10.1074/jbc.R117.818237 35. Ruiz, R., Biea, I. A., & Tabares, L. (2014). α-Synuclein A30P decreases neurodegeneration and increases synaptic vesicle release probability in CSPα-null mice. Neuropharmacology 76, 106– 117. https://doi.org/10.1016/j.neuropharm.2013.08.032 36. Lonart, G., & Südhof, T. C. (2000). Assembly of SNARE core complexes prior to neurotransmitter release sets the readily releasable pool of synaptic vesicles. J Biol Chem. 275(36), 27703–27707. (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 30 37. Ruiz, R., Cano, R., Casañas, J. J., Gaffield, M. A., Betz, W. J., & Tabares, L. (2011). Active zones and the readily releasable pool of synaptic vesicles at the neuromuscular junction of the mouse. J Neurosci. 31, 2000–2008. https://doi.org/10.1523/JNEUROSCI.4663-10.2011 38. Lopez-Manzaneda, M., Franco-Espin, J., Tejero, R., Cano, R., & Tabares, L. (2021). Calcium is reduced in presynaptic mitochondria of motor nerve terminals during neurotransmission in SMA mice. Hum Mol Genet. 30, 629–643. https://doi.org/10.1093/hmg/ddab065 39. Elmqvist, D., & Quastel, D. M. (1965). A quantitative study of end-plate potentials in isolated human muscle. J Physiol 178, 505–529. https://doi.org/10.1113/jphysiol.1965.sp007639 40. Chan, K. Y., Jang, M. J., Yoo, B. B., Greenbaum, A., Ravi, N., Wu, W. L., Sánchez-Guardado, L., Lois, C., Mazmanian, S. K., Deverman, B. E., et al. (2017). Engineered AAVs for efficient noninvasive gene delivery to the central and peripheral nervous systems. Nat Neurosci. 20, 1172– 1179. https://doi.org/10.1038/nn.4593 41. Blatnik, A. J., 3rd, & Macleod Burghes, A. H. (2023). An Hspa8 variant is a shocking modifier of spinal muscular atrophy in mice. Neuron 111, 1349–1350. https://doi.org/10.1016/j.neuron.2023.03.025 42. McCarty D. M. (2008). Self-complementary AAV vectors; advances and applications. Mol Ther. 16, 1648–1656. https://doi.org/10.1038/mt.2008.171 43. DiDonato, C. J., Lorson, C. L., De Repentigny, Y., Simard, L., Chartrand, C., Androphy, E. J., & Kothary, R. (2001). Regulation of murine survival motor neuron (Smn) protein levels by modifying Smn exon 7 splicing. Hum Mol Genet. 10, 2727–2736. https://doi.org/10.1093/hmg/10.23.2727 44. Praveen, K., Wen, Y., Gray, K. M., Noto, J. J., Patlolla, A. R., Van Duyne, G. D., & Matera, A. G. (2014). SMA-causing missense mutations in survival motor neuron (Smn) display a wide range of phenotypes when modeled in Drosophila. PLoS Genet. 10, e1004489. https://doi.org/10.1371/journal.pgen.1004489 45. Gupta, K., Wen, Y., Ninan, N. S., Raimer, A. C., Sharp, R., Spring, A. M., Sarachan, K. L., Johnson, M. C., Van Duyne, G. D., & Matera, A. G. (2021). Assembly of higher-order SMN oligomers is essential for metazoan viability and requires an exposed structural motif present in the YG zipper dimer. Nucleic Acids Res. 49, 7644–7664. https://doi.org/10.1093/nar/gkab508 46. Mourelatos, Z., Abel, L., Yong, J., Kataoka, N., & Dreyfuss, G. (2001). SMN interacts with a novel family of hnRNP and spliceosomal proteins. EMBO J. 20, 5443–5452. https://doi.org/10.1093/emboj/20.19.5443 47. Martin, R., Gupta, K., Ninan, N. S., Perry, K., & Van Duyne, G. D. (2012). The survival motor neuron protein forms soluble glycine zipper oligomers. Structure. 20, 1929–1939. https://doi.org/10.1016/j.str.2012.08.024 48. Deguise, M. O., De Repentigny, Y., Tierney, A., Beauvais, A., Michaud, J., Chehade, L., Thabet, M., Paul, B., Reilly, A., Gagnon, S., Renaud, J. M., & Kothary, R. (2020). Motor transmission defects with sex differences in a new mouse model of mild spinal muscular atrophy. EBioMedicine. 55, 102750. https://doi.org/10.1016/j.ebiom.2020.102750 49. Eshraghi, M., McFall, E., Gibeault, S., & Kothary, R. (2016). Effect of genetic background on the phenotype of the Smn2B/- mouse model of spinal muscular atrophy. Hum Mol Genet, 25, 4494- 4506. doi:10.1093/hmg/ddw278 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 31 50. Kariya, S., Park, G. H., Maeno-Hikichi, Y ., Leykekhman, O., Lutz, C., Arkovitz, M. S., Landmesser, L. T., & Monani, U. R. (2008). Reduced SMN protein impairs maturation of the neuromuscular junctions in mouse models of spinal muscular atrophy. Hum Mol Genet, 17, 2552-2569. doi:10.1093/hmg/ddn156 51. Monani, U. R., Pastore, M. T., Gavrilina, T. O., Jablonka, S., Le, T. T., Andreassi, C., DiCocco, J. M., Lorson, C., Androphy, E. J., Sendtner, M., et al. (2003). A transgene carrying an A2G missense mutation in the SMN gene modulates phenotypic severity in mice with severe (type I) spinal muscular atrophy. J Cell Biol, 160, 41-52. doi:10.1083/jcb.200208079 52. McLachlan, E. M., & Martin, A. R. (1981). Non-linear summation of end-plate potentials in the frog and mouse. J Physiol, 311, 307–324. https://doi.org/10.1113/jphysiol.1981.sp013586 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 56789 1 0 0 1 2 3 4 5 6 Age (Days) A Ctrl (Smn2B/+) SMA;Hspa8G470R/Wt SMA;Hspa8G470R/G470R ** SMA B * C D ** Ctrl (Smn2B/+) SMA;Hspa8G470R/Wt SMA;Hspa8G470R/G470R SMA Ctrl (Smn2B/+) SMA;Hspa8G470R/Wt SMA;Hspa8G470R/G470R SMA Ctrl (Smn2B/+) SMA;Hspa8G470R/Wt SMA;Hspa8G470R/G470R SMA 32 Figure 1 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint Control SMA SMA;Hspa8G470R/Wt SMA;Hspa8G470R/G470R Gastrocnemius *** *** *** *** anti-ChAT Ctrl (Smn2B/+) SMA;Hspa8G470R/Wt SMA;Hspa8G470R/G470R SMA **** N.S. *** A B C D E NF + BTX Triceps Control SMA SMA;Hspa8G470R/Wt SMA;Hspa8G470R/G470R N.S. -10 10 30 50 70 90 *** * N.S. N.S. *** ** ****** * N.S. **** ** *** P=0.05 ** *** *** Ctrl (Smn2B/+) SMA;Hspa8G470R/Wt SMA;Hspa8G470R/G470R SMA Triceps Gastrocs Triceps Gastrocs Triceps Gastrocs Triceps Gastrocs F G H I *** N.S. Gastrocs Triceps 33 Figure 2 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 0 1 2 3 4 5 Relative Protein Levels (a.u.) FL-Smn Smn7 Spinal cord Brain Liver Muscle N.S.N.S. N.S. N.S. Ctrl (Smn2B/+) SMA;Hspa8G470R/Wt SMA;Hspa8G470R/G470R SMA Brain Brain SMN -actin *** *** *** *** * * * P=0.06 N.S. N.S. N.S. Spinal cord Brain Liver Muscle Ctrl (Smn2B/+) SMA;Hspa8G470R/Wt SMA;Hspa8G470R/G470R SMA SMN -Tubulin Hspa8WT Hspa8WT Hspa8G470R Hspa8G470R 013 5 7 9 hrsCHX (50mg/ml) Hspa8WT Hspa8G470R *** A B C D E F 34 Figure 3 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint Ctrl Ctrl Ctrl CtrlS- S- S- S-NL NL NL NL Hspa8Wt Hspa8WtHspa8G470R Hspa8G470R CHX SMN p62 -Tub A ** ** ** * Hspa8Wt Hspa8Wt Hspa8G470RHspa8G470R Ctrl NL treated Serum starved - CHX + CHX N.S. N.S.* * *** *** *** ** Ctrl NL treated Serum starved - CHX + CHX Hspa8G470R BC Hspa8Wt Hspa8WtHspa8G470R Hspa8G470R-Myc: Hspa8WT-Myc: SMN-FLAG: - + + - + - - + - - - + 100 70 * 70 70 55 25 15 100 40 WB: Bag3 WB: p62 WB: Hspb8 WB: Myc (Hspa8) WB: FLAG (SMN) IP: FLAG IP: FLAG WB: Bag3 WB: FLAG WB: Bag3 100 70 Hspa8WT-FLAG: Hspa8G470R-FLAG: - - + -+ - * E F ** Hspa8WT Hspa8G470R D Input (1%) Hspa8WT-Myc: + - SMN-FLAG: + - + + + + Hspa8G470R-Myc: ATPS: 100 70 70 70 55 25 15 100 40 * WB: Bag3 WB: p62 WB: Hspb8 WB: Myc (Hspa8) WB: FLAG (SMN) IP: FLAG Bag3 p62 Hspb8 Hspa8-Myc SMN-FLAG Hspa8WT-Myc: + - SMN-FLAG: + - + + Hspa8 G470R-Myc: H Western blots ++ATPS:G Hspa8WT Hspa8G470R P = 0.05 ** *** Relative SMN-Bag3 interaction (a.u.) I J Hspa8WT Hspa8G470R Hspa8WT Hspa8G470R Figure 4 35 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 0 75 150 225 EPP Amplitude (mV) Ctrl (Smn2B/+) SMA;Hspa8G470R/Wt SMA;Hspa8G470R/G470R SMA ** ** *** Ctrl (Smn2B/+) SMA;Hspa8G470R/Wt SMA;Hspa8G470R/G470R SMA Ctrl;Hspa8G470R/G470R Ctrl;Hspa8G470R/Wt ** *** * N.S. ** SMA SMA;Hspa8G470R/Wt SMA;Hspa8G470R/G470R Ctrl;Hspa8G470R/Wt Ctrl;Hspa8G470R/G470R Ctrl (Smn2B/+) SMA SMA;Hspa8G470R/Wt SMA;Hspa8G470R/G470R Ctrl;Hspa8G470R/Wt Ctrl;Hspa8G470R/G470R Ctrl (Smn2B/+) * *** ***** Ctrl (Smn2B/+) SMA;Hspa8G470R/Wt SMA;Hspa8G470R/G470R SMA Ctrl;Hspa8G470R/G470R Ctrl;Hspa8G470R/Wt 0 250 500 750 1000 1250Readily releasable vesicles 0 100 200 300 400 500 * *** ***** *** N.S. A E H 0 10 20 30 40 50 *** ****** B C D FG IJ 36 Figure 5 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint 0 400 800 1200Endplate area ( m2) 56789 1 0 0 1 2 3 4 5 6 Age (Days) Time to right (sec) 0 20 40 60 80 100 120 0 25 50 75 100 Age (Days) Percent surviving 2 4 6 8 10 12 14 16 18 20 22 Body Weight (g) Ctrl (Smn2B/+) Ctrl (Smn2B/+) SMA SMA SMA (AAV-G470R) SMA (AAV-G470R) Control Triceps * * * * SMA SMA (AAV-G470R) NF + BTX Triceps Gastrocs Triceps Gastrocs Triceps Gastrocs *** ****** *** *** *** *** ****** *** ****** *** *** *** *** * * Ctrl (Smn2B/+) SMA (AAV-G470R) SMA Ctrl (Smn2B/+) SMA (AAV-G470R) SMA Ctrl (Smn2B/+) SMA (AAV-G470R) SMA AB Ctrl (Smn2B/+) SMA SMA (AAV-G470R) C D E F G ** * 37 Figure 6 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint Readily releasable vesicles N.S. * * N.S. ** * N.S. Motor neuron number ** * N.S. N.S. *** N.S. ** * N.S. N.S. Ctrl (Smn2B/+) SMA;Hspa8G470R/Wt SMA;Hspa8G470R/G470R Ctrl (Smn2B/+) SMA;Hspa8G470R/Wt SMA;Hspa8G470R/G470R Ctrl (Smn2B/+) SMA;Hspa8G470R/Wt SMA;Hspa8G470R/G470R A CD F GH J Control SMA;Hspa8G470R/G470R SMA;Hspa8G470R/Wt NF + BTX * * Gastrocnemius muscle E B SMA;Hspa8G470R/Wt SMA;Hspa8G470R/G470R Ctrl (Smn2B/+) SMA;Hspa8G470R/Wt SMA;Hspa8G470R/G470R *** N.S. *** Ctrl (Smn2B/+) SMA;Hspa8G470R/Wt SMA;Hspa8G470R/G470R I * N.S. 38 Figure 7 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 24, 2026. ; https://doi.org/10.64898/2026.02.23.707472doi: bioRxiv preprint