Polyamine dysregulation converges with RASopathies on RAS/MAPK and sensory processing phenotypes in Drosophila

RASopathy, Polyaminopathies, RAS/MAPK signaling, habituation learning, sensory 33 processing, Drosophila 34 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 2

35 RASopathies are developmental conditions associated with cognitive and sensory processing 36 impairments. They are caused by pathogenic variants in genes that result in overactivation of the 37 RAS/MAPK signaling pathway. Genes linked to this pathway have been reported to be enriched among 38 Drosophila models with habituation deficits, a behavioral phenotype reflecting sensory filtering. To 39 identify hidden RASopathies – monogenic disorders that converge on RAS/MAPK overactivation 40 without being classically linked to the pathway – we generated 89 and screened 41 viable habituation-41 deficient Drosophila RNAi models for RAS/MAPK overactivation, measured as an increased 42 phosphorylated ERK to ERK ratio. This screen identified Sms, the ortholog of human spermine synthase 43 (SMS), implicated in Snyder-Robinson syndrome. RAS/MAPK overactivation along with hyperreactivity 44 and habituation impairments are confirmed in a full loss-of-function mutant. A RNAi screen targeting 45 polyamine pathway genes identified Sat (human SAT1/2, SATL1) to reproduce these phenotypes. 46 Knockdown of Sms or Sat in GABAergic neurons impaired habituation, implicating polyamine 47 metabolism in inhibitory circuit function. These findings reveal previously unrecognized convergence 48 between polyamine dysregulation and RASopathies, suggesting shared therapeutic opportunities 49 through modulation of either pathway. 50 51 52 53 54 55 56 57 58 59 SUMMARY STATEMENT 60 Using Drosophila, we uncovered polyamine metabolism genes, Sms and Sat, as modulators of 61 RAS/MAPK and sensory processing, revealing a shared mechanism between polyaminopathies and 62 RASopathies that may inform unified therapies. 63 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 3

64 RASopathies are a group of developmental conditions caused by pathogenic variants in genes encoding 65 components of the RAS/mitogen-activated protein kinase (RAS/MAPK) signaling pathway, which plays 66 a key role in cell growth, differentiation, and development. These include Neurofibromatosis type 1 67 (NF1, OMIM #162200), Noonan syndrome (OMIM #163950) and Costello syndrome (OMIM #218040), 68 among others, and are characterized by a range of clinical features including intellectual disability (ID), 69 autism spectrum disorder (ASD), musculoskeletal abnormalities, and increased tumor risk (Alfieri et 70 al., 2014; Rauen, 2013; Rauen, 2022; Tartaglia et al., 2022; Zenker, 2022). Beyond its well-established 71 role in oncogenesis, dysregulation of the RAS/MAPK pathway has been robustly linked to cognitive 72 (dys)function. Recent studies suggest that pharmacological inhibition of this pathway using MEK 73 inhibitors, small -molecule compounds that block MEK, a kinase downstream of RAS, can lead to 74 improvements in cognitive performance in individuals with NF1 (Lalancette et al., 2024; Walsh et al., 75 2021). These findings suggest that cognitive impairments resulting from dysregulated RAS/MAPK 76 signaling may be amenable to treatment through pharmacological modulation of this pathway, 77 highlighting its promise as a therapeutic target. 78 Habituation is one of the most fundamental and evolutionarily conserved forms of learning, 79 enabling organisms to suppress responses to repetitive, irrelevant stimuli. By supporting attentional 80 filtering, habituation prevents sensory overload and allows cognitive resources to focus on salient 81 inputs (McDiarmid et al., 2017). Deficits in habituation have been reported in individuals with NF1 and 82 SYNGAP1-associated RASopathies and their animal models (Carreno-Munoz et al., 2021; Pride et al., 83 2023; Wolman et al., 2014). Drosophila melanogaster provides a powerful genetic model to investigate 84 the molecular basis of habituation learning and its disruption in neurodevelopmental disorders (NDDs) 85 (Blok et al., 2022). The light-off startle paradigm, which measures jump responses to repeated light-off 86 stimuli, enables efficient, quantitative assessment of habituation, and has been shown to capture 87 cognitive phenotypes across multiple fly models of ID and ASD. A previous large-scale screen identified 88 98 Drosophila orthologs of genes associated with ID with robust habituation deficits upon pan -89 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 4 neuronal knockdown (Fenckova et al., 2019) . Among these, genes operating in RAS/MAPK signaling 90 emerged centrally, and manipulations resulting in increased RAS/MAPK signaling in GABAergic neurons 91 alone was sufficient to impair habituation. These findings highlight habituation learning as a sensitive 92 functional readout to uncover convergent disease mechanisms and potentially treatments that are 93 applicable across genetically diverse NDDs, including those involving RAS/MAPK signaling. 94 Building on this foundation, we screened previously published and unpublished ID-associated 95 genes that were linked to habituation deficits in Drosophila using our light -off startle paradigm to 96 identify genes not previously linked to RAS/MAPK signaling that may converge on this pathway. This 97 approach revealed previously unrecognized RAS/MAPK dysregulation upon loss of several genes, 98 including spermine synthase (Sms), associated with Snyder-Robinson syndrome (SRS, OMIM #309583) 99 (Cason et al., 2003; Snyder and Robinson, 1969). Manipulation of additional polyamine pathway genes 100 revealed further mechanistic connections to RAS/MAPK signaling, specifically in the shared context of 101 disrupted sensory processing. This convergence underscores how genetically distinct rare disorders 102 may share common, druggable signaling disruptions and supports the development of unified 103 therapeutic strategies for both polyaminopathies and RASopathies, for which treatment options at 104 present remain limited. 105 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 5

106 A screen of habituation -deficient NDD models identifies RAS/MAPK overactivation upon Sms 107 knockdown 108 To uncover NDD genes that converge on RAS/MAPK pathway activity, we focused on 89 Drosophila 109 models ( Table S1) previously shown to exhibit habituation deficits following pan -neuronal RNAi -110 mediated knockdown. These genes were drawn from a larger pool ( Table S2, see Materials and 111 Methods) of over 100 habituation -deficient lines identified in previous ly published and unpublished 112 high-throughput behavioral screens (Fenckova et al., 2019; Stessman et al., 2017) and single gene 113 studies (Castells-Nobau et al., 2019; De Hayr et al., 2025; Dias et al., 2022) . Given that increased 114 RAS/MAPK signaling is sufficient to impair habituation in flies (Fenckova et al., 2019), we hypothesized 115 that elevated pathway activity might underlie the behavioral phenotypes in a subset of these models. 116 To test this hypothesis, we quantified phosphorylated ERK (pERK) and ERK levels via Enzyme-117 Linked Immunosorbent Assay (ELISA) and used increased ratios of pERK to ERK (pERK/ERK) as a proxy 118 for RAS/MAPK pathway overactivation characterizing RASopathies. To avoid dilution by non -119 manipulated tissues, we knocked down each gene ubiquitously by crossing an Act-Gal4 driver to the 120 respective UAS-RNAi and background control lines and measured pERK and ERK levels in adult fly 121 whole head lysates. 122 Of the 89 RNAi models, 41 were viable to adulthood and thus further investigated ( Fig. 1A). 123 We identified eight models with increases in pERK/ERK ratios exceeding a 1.2-fold threshold, three of 124 which had an FDR < 0.05 (Fig. 1B, Table S1). Two of these three are Nf1 (2.2-fold) and Spred (1.6-fold), 125 which encode established negative regulators of RAS and whose human orthologues are causally 126 implicated in the classic RASopathies NF1 and Legius syndrome (OMIM #611431), respectively. Their 127 identification provides proof of principle that our screen can detect biologically meaningful RAS/MAPK 128 overactivation. The third significant hit was Sms (1.7-fold), which encodes the enzyme spermine 129 synthase. In humans, pathogenic variants in SMS cause Snyder-Robinson syndrome, an X-linked NDD 130 characterized by ID, muscle hypotonia, and skeletal abnormalities (Cason et al., 2003; Snyder and 131 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 6 Robinson, 1969) . A significant but 1.2 -fold, including Set2 (SETD2), Nmdar2 133 (GRIN2A/B), β-Man (MANBA), ScpX (SCP2), and dnc (PDE4A-D). In contrast to these modest effects, 134 the significant elevation in Sms, a gene not previously molecularly linked to RAS/MAPK signaling yet 135 recovered alongside the well-established RAS/MAPK regulators Nf1 and Spred, revealed a potentially 136 novel pathway connection. 137 138 Figure 1. ELISA-based screen for RAS-MAPK activation in Drosophila NDD models with habituation 139 deficits. (A) Schematic overview of the screening strategy. A total of 89 RNAi lines targeting NDD -140 associated genes and previously shown to cause habituation learning deficits when expressed pan -141 neuronally were selected. Ubiquitous knockdown ( Act-Gal4 > UAS -RNAi) yielded 41 viable models. 142 Head lysates were subjected to ELISA to quantify pERK and ERK levels; their ratio serves as a readout 143 of RAS/MAPK activity. Nf1 and Spred are highlighted as established negative regulators, whose loss-of-144 function increases RAS/MAPK signaling and thus elevates pERK/ERK ratios. (B) Results from the ELISA 145 screen. Mean pERK/ERK fold change of knockdown models normalized to controls (Act-Gal4 / control) 146 are plotted against the FDR -corrected p -values estimated using a linear model as described in 147

and Methods. Per genotype ≥3 biological replicates are included (except for Set2, n=1 due 148 to lethality). Red line indicates p = 0.05, blue line indicates fold change > 1.2. 149 150 Loss of Sms phenocopies Nf1 with increased RAS/MAPK signaling 151 To validate the ELISA screen findings and further characterize Sms as a potential regulator of 152 RAS/MAPK signaling, we performed western blot analysis on adult head extracts from independent 153 Sms knockdown flies (Act-Gal4 > SmsRNAi). This confirmed a significant increase in the pERK/ERK ratio, 154 consistent with RAS/MAPK pathway overactivation (Fig. 2A, A’). Interestingly, based on normalization 155 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 7 with β-tubulin, this elevated pERK/ERK ratio may result from a reduction in ERK rather than an increase 156 in pERK, as reflected by decreased ERK/β-tubulin but unaltered pERK/β-tubulin levels. 157 158 Figure 2. Loss -of-Sms dysregulates RAS-MAPK in a similar way as Nf1. (A) Western blots of head 159 lysates from Sms knockdown models ( Act-Gal4 > Sms RNAi) and controls ( Act-Gal4 / control ). (A’) 160 Quantification and normalization to controls reveal that elevated pERK/ERK is driven by a decrease in 161 ERK/β-tubulin (6 biological replicates, 2 blots). (B) Western blots of head lysates from Sms and Nf1 162 complete loss-of-function mutants. (B’) Quantification and normalization to controls show the same 163 pattern as Sms knockdown models (6 biological replicates, 2 blots). Graphs indicate mean ± SEM. 164 Statistical significance was assessed using linear models as described in Materials and Methods. 165 Corrected p-values are indicated as follows: * p < 0.05, ** p < 0.01, **** p < 0.0001. 166 167 To confirm these findings in an independent genetic loss -of-function model, we analyzed a 168 previously characterized, viable Sms null mutant ( Sms-/-) (Li et al., 2017) . Western blots from Sms-/- 169 head lysates recapitulated the increased pERK/ERK and decreased ERK/ β-tubulin ratios observed in 170 the knockdown animals (Fig. 2B, B’). Sms heterozygotes (Sms+/-) showed intermediate pERK/ERK and 171 ERK/β-tubulin levels, illustrating a dose -dependent relationship between Sms levels and ERK 172 homeostasis. 173 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 8 To facilitate interpretation of the observed ERK levels and activity, we include d a classic 174 RASopathy model into our analysis. Nf1E1/E1 mutants, which carry homozygous alleles containing loss-175 of-function variants of Nf1 generated through ethyl methanesulfonate (EMS) mutagenesis (Walker et 176 al., 2006), displayed the same, even more pronounced pattern of increased pERK/ERK ratios in the 177 presence of lower ERK levels (Fig. 2B, B’). The fact that loss of Sms recapitulates Nf1 supports that Sms 178 may act as a novel upstream modulator of ERK signaling. 179 180 Loss of Sms causes hyperreactivity and habituation deficits 181 Building on these molecular findings, we next asked whether loss of Sms also recapitulates behavioral 182 phenotypes previously associated with RAS/MAPK dysregulation. Because sensory processing 183 alterations in individuals with RASopathies and their models encompass both baseline sensory 184 reactivity and habituation learning (Carreno-Munoz et al., 2021; Pride et al., 2023; Wolman et al., 185 2014), our goal here was to systematically evaluate both components in Sms mutants. Prior work 186 suggested that pan-neuronal Sms knockdown impairs habituation learning in the Drosophila light-off 187 startle paradigm (Fenckova et al., 2019), together raising a possibly conserved role for Sms in sensory 188 filtering in flies. To determine whether these effects extend to a genetic full loss-of-function model and 189 to characterize the sensory profile, we assessed Sms mutants using two inter-trial intervals in the light-190 off startle paradigm (Fig. 3A). In this paradigm, flies jump in response to abrupt lights-off stimuli. When 191 stimuli are presented at five -second intervals, flies do not habituate, and the resulting jump rate 192 quantifies baseline sensory reactivity, though potentially masked by motor impairments (Fenckova et 193 al., 2019; Stessman et al., 2017). In contrast, one-second interval stimulation induces a gradual decline 194 in jump responses, reflecting habituation learning. Following suggested analytical approaches 195 (McDiarmid et al., 2017) , we quantified habituation by calculating a habituation ratio: the average 196 jump rate during trials 51–100 (the habituated plateau) normalized to each fly’s unhabituated baseline 197 reactivity measured under the five -second interval condition ( Fig. 3A). This approach estimates the 198 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 9 extent to which flies reduce their responses during the habituation assay, independent of differences 199 in baseline reactivity. 200 201 Figure 3. Loss of Sms leads to hyperreactivity and habituation deficits in the light -off startle 202 paradigm. (A) In the light-off startle paradigm, flies jump in response to brief (15 ms) light-off stimuli. 203 Reactivity was measured as jump rate under 5 s interval stimulation (reactivity assay). Habituation was 204 quantified as the ratio of jump rate during trials 51–100 under 1 s interval stimulation to jump rate in 205 the reactivity assay. (B) Sms-/- mutants show elevated jump rates in the reactivity assay compared to 206 controls, with Sms+/- flies displaying intermediate phenotypes (4 -5 replicates, 105-153 total flies per 207 genotype). (B’) Habituation ratios are significantly elevated in Sms-/- and Sms+/- mutants, indicating 208 impaired habituation even after accounting for increased baseline reactivity (4 -5 replicates, 93 -151 209 total flies per genotype). (C) Nf1 heterozygous mutants have no altered baseline jump rates, but 210 Nf1E1/E2 flies show strongly reduced jump rates, potentially indicating motor impairment (4 replicates, 211 70-128 total flies per genotype). (C’) Habituation ratios are significantly elevated in hetero - and 212 transheterozygous Nf1 mutants, indicating impaired habituation (4 replicates, 66 -126 total flies per 213 genotype). Line graphs indicate mean ± SEM jump responses per bins of 5 stimuli. Boxes indicate 214 median ± 25th and 75th percentiles. Statistical significance was assessed using linear models as 215 described in Materials and Methods. p-values are indicated as follows: * p < 0.05, ** p < 0.01, **** p 216 < 0.0001. 217 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 10 Sms-/- mutants displayed significantly increased jump rates in the reactivity assay compared to 218 controls, with Sms+/- heterozygotes showing an intermediate, still significant phenotype ( Fig. 3B). In 219 the habituation assay, both Sms-/- and Sms+/- mutants exhibited elevated habituation ratios relative to 220 controls (Fig. 3B’). This indicates that even after accounting for their increased baseline reactivity, 221 mutant flies maintained elevated reactivity during the habituated phase, suggesting impaired adaptive 222 filtering. 223 To compare these findings with a classic RASopathy model, we again tested Nf1 mutant flies. 224 Baseline reactivity of heterozygous loss-of-function (Nf1E1/+ and Nf1E2/+) flies was unaffected, whereas 225 transheterozygous Nf1E1/E2 null mutants showed reduced baseline reactivity, indicative of motor 226 impairments (Fig. 3C ). Despite this limitation, all Nf1 mutant conditions exhibited elevated jump 227 responses in the habituated state when normalized to their baseline reactivity, revealing habituation 228 deficits (Fig. 3C’). While the hyperreactive phenotype of Sms mutants was not recapitulated in Nf1 229 mutants, the shared habituation deficits could suggest a converging role in adaptive sensory 230 processing. 231 232 Sat knockdown mirrors Sms-associated sensory and RAS/MAPK phenotypes 233 Sms encodes spermine synthase, an enzyme that converts spermidine into spermine within the 234 evolutionarily conserved polyamine pathway ( Fig. 4A ) (Burnette and Zartman, 2015; Wu and Liu, 235 2024). To determine whether the sensory processing phenotypes observed in Sms-deficient models 236 reflect a broader consequence of polyamine pathway disruption, we ubiquitously knocked down and 237 tested five additional pathway components. We included RNAi lines (RNAi -1) targeting Sat, Odc1, 238 Odc2, SamDC, and SpdS from the VDRC GD library. We were able to include a second, independent 239 RNAi line (RNAi-2) for Sat, Odc2, and SpdS from the VDRC KK library. 240 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 11 241 Figure 4. Knockdown of polyamine metabolism genes identifies sensory processing and RAS-MAPK 242 phenotypes upon loss-of-Sat. (A) Drosophila (left) and human (right) genes mapped onto the enzymes 243 involved in polyamine metabolism, based on Burnette and Zartman (2015); Wu and Liu (2024) . (B) 244 Jump rates in reactivity assay and (C) habituation ratios of knockdown models ( Act-Gal4,GMR-wIR > 245 UAS-RNAi) normalized to controls ( Act-Gal4,GMR-wIR / control; 3-4 replicates, 62-128 total flies per 246 genotype). Boxes indicate median ± 25th and 75th percentiles. Fig. S1 for line graphs and non -247 normalized data. (D) Illustrative western blot bands and (D’) quantifications of pERK/ERK ratios in 248 lysates of 10 pooled fly heads collected after behavioral testing (5-7 samples per genotype, Fig. S2 for 249 all blots and ratios). Graphs indicate mean ± SEM. Statistical significance was assessed using linear 250 models as described in Materials and Methods. p -values are indicated as follows: * p < 0.05, ** p < 251 0.01, *** p < 0.001, **** p < 0.0001. 252 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 12 Knockdown of Sat (RNAi-1 and -2), Odc1, Odc2, and SpdS led to elevated jump responses in 253 the reactivity assay, sharing increased sensory reactivity as phenotypes with Sms (Fig. 4B). Impaired 254 habituation, as reflected by significantly elevated habituation ratios, was observed in single RNAi lines 255 targeting Sat, Odc1, and SpdS (Fig. 4C). After being assayed for reactivity and habituation, fly heads 256 from all genotypes were used for determining pERK, ERK and β-tubulin by western blot analysis. SatRNAi-257 2 showed a significant increase in pERK/ERK ratios ( Fig. 4D, D’). The increase was driven by elevated 258 pERK/β-tubulin levels, with no changes in ERK/ β-tubulin (Figs. 4D and S2). This pattern is consistent 259 with RAS/MAPK pathway overactivation, although it exhibits a distinct molecular signature compared 260 to the decreased ERK/β-tubulin levels observed in Sms- and Nf1-deficient models. 261 Taken together, knockdown of Sat induces hyperreactivity across two independent RNAi lines, 262 while habituation deficits and RAS/MAPK pathway overactivation are each observed in one seperate 263 line. These findings highlight Sat as the only gene whose knockdown recapitulates all three phenotypes 264 seen in Sms-deficient models. Sat encodes spermidine/spermine acetyltransferase, the rate-limiting 265 enzyme in spermidine catabolism ( Fig. 4A ). The fact that Sms and Sat both use spermidine as a 266 substrate to generate different products (spermine versus acetyl-spermidine) suggests that spermidine 267 accumulation may underlie the convergent behavioral and molecular phenotypes. 268 269 GABAergic origin of sensory processing deficits in Sms and Sat models 270 Given the ubiquitous role of polyamines (Sagar et al., 2021), we next sought to determine whether the 271 sensory phenotypes observed in Sms and Sat models arise from a shared cellular context. Previous 272 work has shown that pan-neuronal knockdown of Sms impairs habituation (Fenckova et al., 2019). We 273 therefore first used Elav-Gal4 to drive pan -neuronal RNAi-mediated knockdown of both enzymes. 274 Unlike seen in the Sms mutant, pan -neuronal knockdown of Sms does not significantly increase 275 baseline reactivity (Fig. 5A). However, consistent with findings from Fenckova et al. and Sms mutant 276 flies, the habituation ratio is significantly increased (Fig. 5B). Looking at the response curves, both Sat 277 RNAi lines exhibited elevated jump responses during the habituated phase (Fig. 5B). However, because 278 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 13 these lines also showed increased baseline reactivity ( Fig. 5A ), the habituation ratio remained 279 unchanged, suggesting that deficits in habituation may be secondary to a heightened sensory 280 reactivity. 281 282 Figure 5. Sms and Sat cause habituation deficit when knocked down specifically in GABAergic 283 neurons. (A) Jump rates in reactivity assay and (B) habituation ratios of pan-neuronal knockdown 284 models (Elav-Gal4, GMR-wIR > UAS-RNAi) normalized to controls (Elav-Gal4,GMR-wIR / control) reveal 285 significantly elevated reactivity in SatRNAi-2 and habituation ratios in SmsRNAi (4 replicates, 116-124 total 286 flies per genotype). (C) Jump rates in reactivity assay and (D) habituation ratios of GABAergic 287 knockdown models (Gad1-Gal4, GMR-wIR > UAS-RNAi) normalized to controls (Gad1-Gal4,GMR-wIR / 288 control) reveal significantly elevated reactivity in SatRNAi-2 and habituation ratios in SmsRNAi, SatRNAi-1 and 289 SatRNAi-2 (2 replicates, 61-63 total flies per genotype). Line graphs indicate mean ± SEM jump responses 290 per bins of 5 stimuli. Boxes indicate median ± 25th and 75th percentiles. Statistical significance was 291 assessed using linear models as described in Materials and Methods. p-values are indicated as follows: 292 *** p < 0.001, **** p < 0.0001. 293 294 Habituation deficits in RASopathy models, including Nf1, have previously been attributed to 295 dysfunction within GABAergic neurons (Fenckova et al., 2019) . To test whether perturbation of 296 polyamine metabolism in these neurons is also sufficient to induce cognitive phenotypes, we 297 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 14 selectively knocked down Sms and Sat using the GABAergic driver Gad1-Gal4. Baseline reactivity 298 remained unchanged for SmsRNAi and SatRNAi-1, whereas SatRNAi-2 showed a significant increase (Fig. 5C). 299 Notably, despite elevated baseline reactivity under GABAergic-specific knockdown, SatRNAi-2 displayed 300 habituation deficits ( Fig. 5D), unlike when crossed to ubiquitous ( Fig. 4C) or pan -neuronal (Fig. 5B) 301 drivers. Moreover, increased habituation ratios were also observed in SmsRNAi and SatRNAi-1, indicating 302 that all three RNAi lines exhibited impaired habituation when crossed to the GABAergic driver ( Fig. 303 5D). Together, this identifies an essential role for both Sms and Sat in GABAergic neurons for sensory 304 processing, with Sms appearing to be primarily involved in habituation learning, while Sat contributes 305 to both increased sensory reactivity and impaired habituation. This suggests that knockdown models 306 of polyamine metabolism genes can recapitulate the cellular origin of habituation deficits in classic 307 RASopathy models. 308 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 15

309 RASopathies are developmental conditions marked by elevated RAS/MAPK signaling, a pathway 310 previously identified as a central node in impaired habituation learning across Drosophila models of ID 311 (Fenckova et al., 2019) . We reasoned that some of the habituation deficit disease models may 312 represent “hidden RASopathies”: genetic diseases not classically linked to RAS/MAPK signaling but 313 sharing molecular and clinical features. Here, we report several disorders whose Drosophila model, in 314 addition to habituation deficits, exhibit s RAS/MAPK overactivation. Notably, this screen highlighted 315 polyamine metabolism as a modulator of RAS/MAPK activity. Manipulating additional polyamine 316 pathway enzymes further demonstrated a direct influence on both RAS/MAPK activity and sensory 317 processing behavior, with perturbation in inhibitory neurons alone being sufficient to recapitulate the 318 phenotypes. These findings extend the molecular framework underlying habituation deficits and 319 position polyamine metabolism as a novel regulator of RAS/MAPK -driven sensory processing, with 320 implications for genetic disorders linked to RAS and polyamine pathways. 321 322 Relevance of the screening approach for rare genetic neurodevelopmental disorders 323 Our screening approach was designed to measure RAS/MAPK activity in fly heads of habituation-324 deficient models, providing disease -relevant resolution. Using this strategy, we identified modifiers, 325 including most notably Sms, that were not detected in previous in vitro RAS/MAPK screens (Ashton-326 Beaucage et al., 2014; Sawyer et al., 2020) . We report three genes, Nf1, Spred and Sms, whose 327 knockdown produced pERK/ERK ratios above the significance threshold. Sms showed an intermediate 328 pERK/ERK level relative to the classic RASopathy genes Nf1 and Spred, supporting its identification as 329 a genuine RAS/MAPK modulator. Sms encodes spermine synthase, an enzyme involved in polyamine 330 metabolism. In humans, pathogenic variants in SMS cause Snyder-Robinson syndrome, an X -linked 331 NDD characterized by ID, muscle hypotonia, and skeletal abnormalities (Cason et al., 2003; Snyder and 332 Robinson, 1969). In addition, five further genes, Set2 (SETD2 in human), Nmdar2 (GRIN2A/B), β-Man 333 (MANBA), ScpX (SCP2), and dnc (PDE4A-D), showed more than 1.2-fold increases in pERK/ERK levels. 334 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 16 Notably, two of these genes, Nmdar2 and dnc, have known mechanistic links to RAS/MAPK signaling, 335 reinforcing their relevance even below statistical cutoffs. Nmdar2 encodes an NMDA receptor subunit 336 essential for synaptic plasticity. NMDA activation can both stimulate Ras and limit its activity via 337 SynGAP1, a RASopathy-associated protein (Jeyabalan and Clement, 2016; Kim et al., 2005; Wang et al., 338 2007). Dnc encodes the fly orthologue of mam malian PDE4 family of phosphodiesterases (PDE4s), 339 which degrade cAMP , a second messenger that modulates signaling pathways including RAS/MAPK 340 (Donders et al., 2024). Further support for a role of cAMP in the RAS‑associated habituation phenotype 341 comes from the fact that Nf1 can directly control both cAMP and RAS/MAPK pathways (Botero and 342 Tomchik, 2024), and inhibition of PDE4 improved habituation in NF1 zebrafish models (Wolman et al., 343 2014). Considering these established mechanistic links, the increase in RAS/MAPK activation observed 344 for these genes – albeit non-significant – may reflect subtle or potentially context-dependent 345 contributions to pathway regulation. Our identification of these associations in genetic in vivo disease 346 models warrants further investigation. 347 Sms knockdown increased pERK/ERK ratios , driven by reduced ERK /β-tubulin, a pattern 348 consistent with RAS/MAPK overactivation as it is resembling the phenotype observed in Nf1 mutants. 349 Although prior work on the same Nf1 allele reported elevated pERK with stable ERK and β-tubulin 350 levels in adult head samples (Walker et al., 2006) , this difference may reflect time- and context-351 dependent negative feedback on ERK expression after sustained pathway activation (Lake et al., 2016). 352 One potential mechanism for such feedback involves pERK stimulated, RREB1-dependent induction of 353 miR-143/145, which has been shown to attenuate ERK expression (Kent et al., 2013). Eventually, the 354 balance between pERK and ERK is evidently disturbed , with predictably detrimental consequences, 355 given the pathway’s reliance on tight spatiotemporal control (Ram et al., 2023). 356 357 Spermidine as a regulator of RAS/MAPK signaling 358 We investigated additional components of the polyamine pathway to further establish their 359 connection to sensory processing and RAS/MAPK dysregulation. In addition to Sms, genetic 360 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 17 knockdown of Sat (SAT1, SAT2, SATL1 in human) increased RAS/MAPK signaling, strengthening the link 361 between this pathway and polyamines —molecules which are present across all cell types and living 362 organisms, essential for nucleic acid stabilization, protein translation, and signal transduction (Xuan et 363 al., 2023). Since both Sms and Sat use spermidine (SPD) as a substrate, loss of either gene product is 364 expected to cause SPD accumulation, providing a possible pathogenic mechanism for the observed 365 phenotypes. Interestingly, increased levels of polyamines are a hallmark of cancer (Sagar et al., 2021), 366 and increased levels of spermidine specifically have been associated with elevated pERK in cancer cell 367 cultures (Bachrach et al., 2001). Pharmacological reduction of polyamine synthesis via ODC inhibition 368 (DMFO) has shown to be an effective cancer treatment (Schramm et al., 2025) . Conversely, while 369 DMFO is effective in reducing putrescine (PUT) and SPD levels, spermine (SPM) levels go up (Flamigni 370 et al., 1999). Indeed, SPM is proposed to have negative regulating effects on RAS/MAPK as evidenced 371 from its capability to disturb phosphorylation of DRaf (Stark et al., 2011). 372 In addition, our findings propose that increased RAS/MAPK signaling may contribute to 373 cognitive impairments in Snyder-Robinson syndrome, caused by pathogenic variants in SMS. In fly 374 models of SRS, elevated SPD has been linked to cellular toxicity through ROS -generating acetylation 375 and oxidation (Li et al., 2017) . Notably, recent work in human bone ‑marrow–derived pluripotent 376 stromal cells demonstrated that upregulated SAT1 strongly reduced SPD levels (Cressman et al., 2024). 377 This aligns with our observation that both Sms loss and Sat loss lead to phenotypes associated with 378 SPD accumulation. The SmsG56S mutant mouse model exhibits mitochondrial respiration defects and 379 increased energy expenditure , along with abnormal growth and body composition , and behavioral 380 phenotypes such as an increased and maintained fear response (Akinyele et al., 2024). Upon reanalysis 381 of the RNA-seq data from this mouse model, we identified an enrichment of dysregulated genes within 382 RAS- and ERK -related Gene Ontology categories, including “Ras protein signal transduction” 383 (GO:0007265) and “ERK1 and ERK2 cascade” (GO:0070371) ( Table S3). This supports our finding of 384 dysregulated RAS/MAPK signaling in a model of SRS and extends it across species. While polyamine 385 imbalance could drive RAS/MAPK activation, the pathway may also feed back onto polyamine 386 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 18 metabolism. In cancer cells, inhibition of MEK (PD98059) can suppress expression of ODC1 and thus 387 restore polyamine balance (Flamigni et al., 1999) , suggesting that restoring RAS/MAPK holds 388 therapeutic potential to restore polyamine balance in SRS. 389 390 Polyamine dysregulation in GABAergic neurons drives habituation deficits 391 Behavioral assays revealed that Sms and Sat mutants display impairments in both baseline sensory 392 reactivity and habituation learning, as assessed using the light -off startle paradigm. Although 393 habituation and sensory reactivity represent distinct aspects of sensory processing, namely adaptation 394 to repeated stimuli versus general responsiveness, they are both fundamental components of how 395 organisms respond to sensory input (He et al., 2023; McDiarmid et al., 2017; Schauder and Bennetto, 396 2016). Importantly, both traits are clinically relevant. Altered sensory responsiveness and habituation 397 are frequently observed in individuals with neurodevelopmental disorders and RASopathies, as well as 398 in their animal models (Carreno-Munoz et al., 2021; Ethridge et al., 2019; Pride et al., 2023; Wolman 399 et al., 2014) . Their relative importance to atypical behavioral responses and problems in daily living 400 skills is however unknown. 401 Neuronal subtype -specific knockdown experiments demonstrated that both Sms and Sat 402 operate in GABAergic neurons, pinpointing inhibitory circuitry as the origin of dysfunction. Polyamines 403 intersect with inhibitory transmission via putrescine, a precursor for GABA (Makletsova et al., 2022). 404 While increased putrescine availability can enhance GABA synthesis and tonic inhibition in epilepsy 405 models (Kovács et al., 2021) , chronic accumulation of spermidine and putrescine in Sms and Sat 406 mutants may destabilize inhibitory balance, paralleling findings in NF1 models where enhanced GABA 407 release contributes to cognitive deficits (Costa et al., 2002; Cui et al., 2008; Omrani et al., 2015) . 408 Furthermore, it aligns with observed habituation defects when manipulating RASopathy genes , 409 including Nf1 in GABAergic neurons (Fenckova et al., 2019) . Together, these results suggest that 410 polyamine dysregulation and RAS/MAPK hyperactivation converge on inhibitory circuit dysfunction, 411 producing deficits in sensory reactivity and habituation learning. 412 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 19 Shared pathophysiology of polyaminopathies and RASopathies creates therapeutic opportunities 413 The convergence of polyaminopathies and RASopathies extends beyond molecular and behavioral 414 phenotypes. Snyder -Robinson syndrome shares clinical features with NF1, Noonan, and Costello 415 syndromes, including cognitive/learning impairments, but also short stature, reduced bone mineral 416 density, and skeletal abnormalities such as scoliosis (Aftab and Dattani, 2019; Friedman, 1993; Gripp 417 et al., 2019; Rauen and Tidyman, 2024; Reynolds et al., 2025) , further supporting shared aspects of 418 pathophysiology. This overlap has important therapeutic implications, as inhibitors of MEK are FDA -419 approved for tumors in NF1 (Cook, 2025; Lalancette et al., 2024; Walsh et al., 2021). At the same time, 420 polyamines can be targeted by ODC inhibitor Eflornithine (DFMO), which was developed to treat 421 cancer and repurposed for Bachmann-Bupp syndrome (OMIM #619075), caused by activating variants 422 in ODC1 (Bachmann et al., 2024) . Eflornithine is also proposed as a treatment in SRS, as it restored 423 SPD:SPM balance and improved several readouts in human bone marrow-derived pluripotent stromal 424 cells and longevity in fly models (Cressman et al., 2024; Stewart et al., 2023). 425 By highlighting shared signaling disruptions across genetically distinct but functionally 426 convergent rare neurodevelopmental disorders, these findings raise the potential for unified 427 therapeutic strategies targeting both polyamine metabolism and RAS/MAPK signaling. 428 429 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 20

430 Drosophila stock selection for ELISA screen 431 146 genes, when pan-neuronally knocked down, have been associated with habituation defects in our 432 light-off startle paradigm. Before screening, 57 genes were excluded because the gene was either not 433 associated with monogenic NDD in the SysNDD database curated at 12 -nov-2025 (13 genes), the 434 habituation deficit fly line was from another source than the Vienna Drosophila Resource Center 435 (VDRC) GD or KK collection (6 genes), or was not readily available (2 genes), the UAS -RNAi construct 436 was located on the X-chromosome (13 genes) or at the 40D landing-site (23 genes). All excluded RNAi 437 lines are listed in Table S1, and included RNAi lines are listed in Table S2 including a reference to the 438 study where a habituation deficit is shown with a pan-neuronal driver, if published. 439 440 Drosophila stocks and maintenance 441 Fly stocks were maintained on standard cornmeal -yeast-sugar-agar medium supplemented with the 442 mold inhibitors methyl paraben and propanoic acid, at 25 °C and 60% humidity under a 12:12 h light-443 dark cycle. Genotypes are referred to by shorthand labels in figures and text, with full genotypes 444 provided in Table S4. Gene knockdown was achieved using the Gal4/UAS system (Brand and Perrimon, 445 1993). Knockdown animals were generated by crossing Gal4 driver lines to UAS -RNAi lines from the 446 GD and KK libraries of VDRC (Dietzl et al., 2007) (Table S1 and S4). Control animals were generated by 447 crossing each Gal4 driver line to the genetic background control lines of the corresponding library 448 (VDRC #60000 for GD lines and #60100 for KK lines). The Sms mutant line (Table S4) has a P-element 449 insertion resulting in less than 0.05% transcript levels of Sms (Li et al., 2017) and was backcrossed 7 450 times to the w- control line (VDRC #60000), which served as control. EMS generated Nf1E1 and NF1E2 451 mutant and isogenic control lines were kindly provided by James Walker (Walker et al., 2006). In several 452 models (Table S4), a GMR -wIR element (eye -specific RNAi knockdown of white) was introduced to 453 suppress eye pigmentation which is required for light-off startle responses (Fenckova et al., 2019). 454 455 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 21 Enzyme-Linked Immunosorbent Assay (ELISA) 456 Phospho-ERK (pERK) and ERK levels were quantified in fly head extracts using the ERK1/2 (pT202/Y204) 457 + Total ERK1/2 ELISA kit (Abcam, ab176640). Twenty 1 –3-day-old flies were anesthetized with CO₂, 458 snap-frozen in liquid nitrogen, and decapitated by flicking. Heads were collected on ice using a brush. 459 Heads were homogenized in 100 µL of the provided lysis buffer using a plastic pestle, incubated on ice 460 for 20 minutes, and centrifuged at 12,000 rpm for 20 minutes at 4 °C. Supernatants were diluted 1:10 461 to approximately ~500 µg/mL protein. A positive control was prepared per kit instructions. For each 462 well, 50 µL of diluted sample or control and 50 µL of antibody cocktail were added. Plates were 463 incubated for 1 hour at room temperature (400 rpm), washed 3× with provided wash buffer, then 464 incubated with 100 µL TMB substrate for 15 minutes in the dark. Reactions were stopped with 100 µL 465 stop solution, and absorbance was read at 450 nm. 466 Samples were run in technical duplicates for both pERK and ERK and their respective mean 467 absorbance was included only if the coefficient of variation (CV, calculated as standard deviation / 468 mean) was ≤0.2 for both pERK and ERK. For each RNAi model ≥3 biological replicates (except for Set2, 469 n=1 due to lethality) were tested, meaning independent crosses and plates, but always measured on 470 the same plate together with a background control. 471 Statistical significance was assessed using a linear mixed -effects model (lme4 package) in R 472 (v4.3.3) (Bates et al., 2015). Log₁₀-transformed pERK/ERK absorbance ratios were used as the response 473 variable, with genotype as a fixed effect and ELISA plate as a random effect. Pairwise comparisons 474 between disease models and their respective controls were performed using the emmeans R package 475 (DOI: 10.32614/CRAN.package.emmeans), with multiple testing correction using the Benjamini -476 Hochberg false discovery rate (FDR). Data figures are generated using the ggplot2 R package (Wickham, 477 2016). 478 479 480 481 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 22 Western blotting 482 To quantify pERK, ERK, and β-tubulin levels, western blotting was performed. For validation of Sms 483 phenotypes ( Fig. 2 ), heads from ten 1 –3-day-old flies were collected as described for ELISA. For 484 polyamine pathway knockdown models ( Fig. 4 ), heads from ten 7 –10-day-old flies were collected 485 following light-off startle assays, anesthetized on ice, and snap -frozen in liquid nitrogen. Heads were 486 homogenized in 60µL RIPA buffer with protease and phosphatase inhibitors, incubated on ice for 20 487 minutes, and centrifuged at 12,000 rpm for 20 minutes at 4 °C. 15µL of supernatant was mixed with 488 5µL sample buffer (100 µM DTT in NuPAGE LDS Sample Buffer, Invitrogen) and heated at 95 ℃ for 5 489 minutes. Samples (12 µL) or 3 µL of Odyssey molecular weight marker were loaded onto 4 –12% Bis-490 Tris gels (NuPAGE 15 wells, Invitrogen) and electrophoresed in MOPS buffer (NuPage MOPS SDS 491 Running Buffer, Invitrogen) at 120 V for ~2 hours. Proteins were transferred to nitrocellulose 492 membranes (Trans-Blot Turbo Transfer Pack, Bio-Rad) using the Trans-Blot Turbo system (Bio-Rad) at 493 25V for 7 minutes. Membranes were blocked for one hour in 10% bovine serum albumin (BSA, Sigma-494 Aldrich, A8806-1G), and then incubated overnight at 4°C in primary antibodies diluted in a 1:1 mix of 495 10% BSA and tris buffered saline buffer with 0.2% tween (TBST). The antibody against pERK (M8159, 496 Sigma) was diluted 1:2500, and total ERK (9102S, Cell signaling) was diluted 1:500. After three 5-minute 497 washes in TBST, membranes were incubated with secondary antibodies (goat anti-mouse, Alexa Fluor 498 680, A-21057, Thermo Fisher Scientific and goat anti -rabbit, IRDye 800, LI -COR, each 1:10000) for 1 499 hour at room temperature in the dark. Membranes were then washed, rinsed in TBS, and imaged on 500 an Odyssey Infrared Scanner (Odyssey® DLx Imaging System, Li-cor). Membranes were reprobed with 501 anti-β-tubulin (dilution 1:10000, AB_2315513, DSHB) for 1 hour at room temperature, followed by 502 secondary antibody incubation and imaging as above. 503 Quantification was conducted using Image Studio software (version 4.0.21). Local background 504 was subtracted to obtain signal intensity of each band. Genotypes are always compared against control 505 samples on the same gel. Samples are always from independent biological replicates. Statistical 506 significance was assessed using a linear mixed-effects model (lme4 package) in R (v4.3.3) (Bates et al., 507 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 23 2015). Log₁₀ -transformed pERK/ERK, pERK/ β-tubulin, or ERK/ β-tubulin ratios were used as the 508 response variable, with genotype as a fixed effect and gel as a random effect. Pairwise comparisons 509 between groups were performed using the emmeans R package (DOI: 510 10.32614/CRAN.package.emmeans), with multiple testing correction via Šidák method. Data figures 511 are generated using the ggplot2 R package (Wickham, 2016). 512 513 Light-off startle paradigm 514 Sensory processing phenotypes, reactivity and habituation, were assessed using the light -off startle 515 paradigm (Aktogen Ltd.) as previous described (Fenckova et al., 2019). In each round, 32 flies aged 7–516 10 days were individually placed in tubes. Following a 5 -minute acclimatization period, flies were 517 subjected to the habituation paradigm, receiving 100 light-off stimuli (15 ms) with a 1-second intertrial 518 interval (ITI). Then, following a 2-minutes rest period the reactivity assay was started where 50 stimuli 519 were delivered with a 5 -second ITI. Wing vibrations during jump responses were recorded by 520 microphones at both ends, and a jump was concluded when a threshold of 0.8 V was exceeded. Data 521 were collected using custom LabVIEW software (National Instruments, Austin, TX). Reactivity was 522 quantified for each fly as the proportion of trials in which a jump occurred out of the 50 trials, including 523 only flies that responded more than once. Habituation was quantified for each fly as the jump rate 524 during the presumed habituated phase (trials 51 –100 of the habituation assay), normalized to that 525 fly’s jump rate in the reactivity assay, and included only flies with more than one response in both 526 assays. 527 Statistical significance was assessed using a linear mixed -effects model (lme4 package) in R 528 (v4.3.3) (Bates et al., 2015) . Reactivity or habituation ratios were used as the response variable in 529 separate linear models, with genotype as a fixed effect and light -off chamber identity and date of 530 testing as random effects. Pairwise comparisons between groups were performed using the emmeans 531 R package (DOI: 10.32614/CRAN.package.emmeans), with multiple testing correction via Šidák 532 method. Data figures are generated using the ggplot2 R package (Wickham, 2016). 533 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 24 Re-analysis of SMS mouse model transcriptome data 534 Processed RNA-seq data for the SmsG56S mouse model, which carries a missense variant in the Sms 535 gene, and wild -type controls were obtained from the original study (GSE226413) (Akinyele et al., 536 2024). Differentially expressed genes (DEGs) were identified using the following criteria: false discovery 537 rate (FDR) 0.5 or < -0.5. This thresholding yielded 4,760 DEGs. 538 Gene ontology (GO) biological process (BP) enrichment analysis was performed to identify 539 overrepresented functional categories. Statistical comparison between groups of genes was conducted 540 using the BinfTools R package (https://github.com/kevincjnixon/BinfTools), specifically employing the 541 GO_GEM function. The background set for enrichment analysis consisted of all annotated mouse 542 genes. Significant GO terms related to “RAS protein signal transduction” and the “ERK1 and ERK2 543 cascade,” along with their corresponding up- and down-regulated genes (log2 fold change > 0.5 or < -544 0.5, respectively), are provided in Table S3. 545 546

547 We thank James Walker, the VDRC and the Bloomington Drosophila Stock Center for providing 548 Drosophila strains. We are grateful to F. Kampshoff, S. Letteboer, M. Aslanyan and S. Beersum for 549 experimental or logistic support, and to all members of the Schenck lab for helpful discussion. 550 551 Competing interests 552 The authors declare that they have no competing interests. 553 554 Funding 555 This work was in part supported by a Vici grant from the Netherlands Organization for Health Research 556 and Development (ZonMw, 09150181910022) to A.S, and a Radboud Excellence Fellowship to S.G.J. 557 558 559 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 25 Data and resource availability 560 All relevant data and resources can be found within the article and its supplementary information. 561 562 Author contributions statement 563 Conceptualization: B.v.R., S.G.J., A.S. 564 Methodology: B.v.R. 565 Software: B.v.R., M.Bo., S.G.J. 566 Formal analysis: B.v.R., S.G.J. 567 Investigation: B.v.R., M.d.W., K.P ., Z.-A.G., P .S., M.Be., S.G.J. 568 Data curation: B.v.R., S.G.J. 569 Visualization: B.v.R. 570 Supervision: B.v.R., S.G.J., A.S. 571 Project administration: B.v.R., A.S. 572 Funding acquisition: S.G.J., A.S. 573 Writing – original draft: B.v.R., S.G.J., A.S. 574 Writing – review & editing: All authors 575 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 26

576 Aftab, S. and Dattani, M. T. (2019). Pathogenesis of Growth Failure in Rasopathies. Pediatr 577 Endocrinol Rev 16, 447-458. 578 Akinyele, O., Munir, A., Johnson, M. A., Perez, M. S., Gao, Y ., Foley, J. R., Nwafor, A., Wu, Y ., 579 Murray-Stewart, T., Casero, R. A., Jr. et al. (2024). Impaired polyamine metabolism causes 580 behavioral and neuroanatomical defects in a mouse model of Snyder-Robinson syndrome. Dis 581 Model Mech 17. 582 Alfieri, P., Piccini, G., Caciolo, C., Perrino, F., Gambardella, M. L., Mallardi, M., Cesarini, L., Leoni, C., 583 Leone, D., Fossati, C. et al. (2014). Behavioral profile in RASopathies. Am J Med Genet A 584 164A, 934-42. 585 Ashton-Beaucage, D., Udell, C. M., Gendron, P ., Sahmi, M., Lefrancois, M., Baril, C., Guenier, A. S., 586 Duchaine, J., Lamarre, D., Lemieux, S. and Therrien, M. (2014). A functional screen reveals 587 an extensive layer of transcriptional and splicing control underlying RAS/MAPK signaling in 588 Drosophila. PLoS Biol 12, e1001809. 589 Bachmann, A. S., VanSickle, E. A., Michael, J., Vipond, M. and Bupp, C. P. (2024). Bachmann-Bupp 590 syndrome and treatment. Dev Med Child Neurol 66, 445-455. 591 Bachrach, U., Wang, Y . C. and Tabib, A. (2001). Polyamines: new cues in cellular signal transduction. 592 News Physiol Sci 16, 106-9. 593 Bates, D., Mächler, M., Bolker, B. M. and Walker, S. C. (2015). Fitting Linear Mixed-Effects Models 594 Using lme4. Journal of Statistical Software 67, 1-48. 595 Blok, L. E. R., Boon, M., van Reijmersdal, B., Hoffler, K. D., Fenckova, M. and Schenck, A. (2022). 596 Genetics, molecular control and clinical relevance of habituation learning. Neurosci Biobehav 597 Rev 143, 104883. 598 Botero, V. and Tomchik, S. M. (2024). Unraveling neuronal and metabolic alterations in 599 neurofibromatosis type 1. J Neurodev Disord 16, 49. 600 Brand, A. H. and Perrimon, N. (1993). Targeted gene expression as a means of altering cell fates and 601 generating dominant phenotypes. Development 118, 401-15. 602 Burnette, M. and Zartman, J. J. (2015). Spatiotemporal patterning of polyamines in Drosophila 603 development. Amino Acids 47, 2665-70. 604 Carreno-Munoz, M. I., Chattopadhyaya, B., Agbogba, K., Cote, V., Wang, S., Levesque, M., Avoli, M., 605 Michaud, J. L., Lippe, S. and Di Cristo, G. (2021). Sensory processing dysregulations as 606 reliable translational biomarkers in SYNGAP1 haploinsufficiency. Brain. 607 Cason, A. L., Ikeguchi, Y ., Skinner, C., Wood, T. C., Holden, K. R., Lubs, H. A., Martinez, F., Simensen, 608 R. J., Stevenson, R. E., Pegg, A. E. and Schwartz, C. E. (2003). X-linked spermine synthase 609 gene (SMS) defect: the first polyamine deficiency syndrome. Eur J Hum Genet 11, 937-44. 610 Castells-Nobau, A., Eidhof, I., Fenckova, M., Brenman-Suttner, D. B., Scheffer-de Gooyert, J. M., 611 Christine, S., Schellevis, R. L., van der Laan, K., Quentin, C., van Ninhuijs, L. et al. (2019). 612 Conserved regulation of neurodevelopmental processes and behavior by FoxP in Drosophila. 613 PLoS One 14, e0211652. 614 Cook, S. J. (2025). Second MEK Inhibitor Shows Promise in Neurofibromatosis Type 1 With 615 Symptomatic Plexiform Neurofibromas Signaling Hope for Other RASopathies. J Clin Oncol 43, 616 730-734. 617 Costa, R. M., Federov, N. B., Kogan, J. H., Murphy, G. G., Stern, J., Ohno, M., Kucherlapati, R., Jacks, 618 T. and Silva, A. J. (2002). Mechanism for the learning deficits in a mouse model of 619 neurofibromatosis type 1. Nature 415, 526-30. 620 Cressman, A., Morales, D., Zhang, Z., Le, B., Foley, J., Murray-Stewart, T., Genetos, D. C. and Fierro, 621 F. A. (2024). Effects of Spermine Synthase Deficiency in Mesenchymal Stromal Cells Are 622 Rescued by Upstream Inhibition of Ornithine Decarboxylase. Int J Mol Sci 25. 623 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 27 Cui, Y ., Costa, R. M., Murphy, G. G., Elgersma, Y ., Zhu, Y ., Gutmann, D. H., Parada, L. F., Mody, I. and 624 Silva, A. J. (2008). Neurofibromin regulation of ERK signaling modulates GABA release and 625 learning. Cell 135, 549-60. 626 De Hayr, L., Blok, L. E. R., Dias, K. R., Long, J., Begemann, A., Moir, R. D., Willis, I. M., Mocera, M., 627 Siegel, G., Steindl, K. et al. (2025). Biallelic variants in GTF3C3 result in an autosomal 628 recessive disorder with intellectual disability. Genet Med 27, 101253. 629 Dias, K. R., Carlston, C. M., Blok, L. E. R., De Hayr, L., Nawaz, U., Evans, C. A., Bayrak-Toydemir, P., 630 Htun, S., Zhu, Y ., Ma, A. et al. (2022). De Novo ZMYND8 variants result in an autosomal 631 dominant neurodevelopmental disorder with cardiac malformations. Genet Med 24, 1952-632 1966. 633 Dietzl, G., Chen, D., Schnorrer, F., Su, K. C., Barinova, Y ., Fellner, M., Gasser, B., Kinsey, K., Oppel, S., 634 Scheiblauer, S. et al. (2007). A genome-wide transgenic RNAi library for conditional gene 635 inactivation in Drosophila. Nature 448, 151-6. 636 Donders, Z., Skorupska, I. J., Willems, E., Mussen, F., Broeckhoven, J. V., Carlier, A., Schepers, M. 637 and Vanmierlo, T. (2024). Beyond PDE4 inhibition: A comprehensive review on downstream 638 cAMP signaling in the central nervous system. Biomed Pharmacother 177, 117009. 639 Ethridge, L. E., De Stefano, L. A., Schmitt, L. M., Woodruff, N. E., Brown, K. L., Tran, M., Wang, J., 640 Pedapati, E. V., Erickson, C. A. and Sweeney, J. A. (2019). Auditory EEG Biomarkers in Fragile 641 X Syndrome: Clinical Relevance. Front Integr Neurosci 13, 60. 642 Fenckova, M., Blok, L. E. R., Asztalos, L., Goodman, D. P., Cizek, P ., Singgih, E. L., Glennon, J. C., 643 IntHout, J., Zweier, C., Eichler, E. E. et al. (2019). Habituation Learning Is a Widely Affected 644 Mechanism in Drosophila Models of Intellectual Disability and Autism Spectrum Disorders. 645 Biol Psychiatry 86, 294-305. 646 Flamigni, F., Facchini, A., Capanni, C., Stefanelli, C., Tantini, B. and Caldarera, C. M. (1999). p44/42 647 mitogen-activated protein kinase is involved in the expression of ornithine decarboxylase in 648 leukaemia L1210 cells. Biochem J 341 ( Pt 2), 363-9. 649 Friedman, J. M. (1993). Neurofibromatosis 1. In GeneReviews®, (eds. M. P . Adam J. Feldman G. M. 650 Mirzaa R. A. Pagon S. E. Wallace and A. Amemiya). University of Washington, Seattle. 651 Gripp, K. W., Morse, L. A., Axelrad, M., Chatfield, K. C., Chidekel, A., Dobyns, W., Doyle, D., Kerr, B., 652 Lin, A. E., Schwartz, D. D. et al. (2019). Costello syndrome: Clinical phenotype, genotype, and 653 management guidelines. Am J Med Genet A 179, 1725-1744. 654 He, J. L., Williams, Z. J., Harris, A., Powell, H., Schaaf, R., Tavassoli, T. and Puts, N. A. J. (2023). A 655 working taxonomy for describing the sensory differences of autism. Mol Autism 14, 15. 656 Jeyabalan, N. and Clement, J. P . (2016). SYNGAP1: Mind the Gap. Front Cell Neurosci 10, 32. 657 Kent, O. A., Fox-Talbot, K. and Halushka, M. K. (2013). RREB1 repressed miR-143/145 modulates 658 KRAS signaling through downregulation of multiple targets. Oncogene 32, 2576-85. 659 Kim, M. J., Dunah, A. W., Wang, Y . T. and Sheng, M. (2005). Differential roles of NR2A- and NR2B-660 containing NMDA receptors in Ras-ERK signaling and AMPA receptor trafficking. Neuron 46, 661 745-60. 662 Kovács, Z., Skatchkov, S. N., Veh, R. W., Szabó, Z., Németh, K., Szabó, P . T., Kardos, J. and Héja, L. 663 (2021). Critical Role of Astrocytic Polyamine and GABA Metabolism in Epileptogenesis. Front 664 Cell Neurosci 15, 787319. 665 Lake, D., Corrêa, S. A. and Müller, J. (2016). Negative feedback regulation of the ERK1/2 MAPK 666 pathway. Cell Mol Life Sci 73, 4397-4413. 667 Lalancette, E., Cantin, E., Routhier, M. E., Mailloux, C., Bertrand, M. C., Kiaei, D. S., Larouche, V., 668 Tabori, U., Hawkins, C., Ellezam, B. et al. (2024). Impact of trametinib on the 669 neuropsychological profile of NF1 patients. J Neurooncol 167, 447-454. 670 Li, C., Brazill, J. M., Liu, S., Bello, C., Zhu, Y ., Morimoto, M., Cascio, L., Pauly, R., Diaz-Perez, Z., 671 Malicdan, M. C. V. et al. (2017). Spermine synthase deficiency causes lysosomal dysfunction 672 and oxidative stress in models of Snyder-Robinson syndrome. Nat Commun 8, 1257. 673 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 28 Makletsova, M. G., Rikhireva, G. T., Kirichenko, E. Y ., Trinitatsky, I. Y ., Vakulenko, M. Y . and Ermakov, 674 A. M. (2022). The Role of Polyamines in the Mechanisms of Cognitive Impairment. 675 Neurochemical Journal 16, 283-294. 676 McDiarmid, T. A., Bernardos, A. C. and Rankin, C. H. (2017). Habituation is altered in 677 neuropsychiatric disorders-A comprehensive review with recommendations for experimental 678 design and analysis. Neurosci Biobehav Rev 80, 286-305. 679 Omrani, A., van der Vaart, T., Mientjes, E., van Woerden, G. M., Hojjati, M. R., Li, K. W., Gutmann, 680 D. H., Levelt, C. N., Smit, A. B., Silva, A. J. et al. (2015). HCN channels are a novel therapeutic 681 target for cognitive dysfunction in Neurofibromatosis type 1. Mol Psychiatry 20, 1311-21. 682 Pride, N. A., Haebich, K. M., Walsh, K. S., Lami, F., Rouel, M., Maier, A., Chisholm, A. K., Lorenzo, J., 683 Hearps, S. J. C., North, K. N. and Payne, J. M. (2023). Sensory Processing in Children and 684 Adolescents with Neurofibromatosis Type 1. Cancers (Basel) 15. 685 Ram, A., Murphy, D., DeCuzzi, N., Patankar, M., Hu, J., Pargett, M. and Albeck, J. G. (2023). A guide 686 to ERK dynamics, part 1: mechanisms and models. Biochem J 480, 1887-1907. 687 Rauen, K. A. (2013). The RASopathies. Annu Rev Genomics Hum Genet 14, 355-69. 688 Rauen, K. A. (2022). Defining RASopathy. Dis Model Mech 15. 689 Rauen, K. A. and Tidyman, W. E. (2024). RASopathies - what they reveal about RAS/MAPK signaling 690 in skeletal muscle development. Dis Model Mech 17. 691 Reynolds, G., Gazzin, A., Carli, D., Massuras, S., Cardaropoli, S., Luca, M., Defilippi, B., Tartaglia, M., 692 Ferrero, G. B. and Mussa, A. (2025). Update on the Clinical and Molecular Characterization of 693 Noonan Syndrome and Other RASopathies: A Retrospective Study and Systematic Review. Int 694 J Mol Sci 26. 695 Sagar, N. A., Tarafdar, S., Agarwal, S., Tarafdar, A. and Sharma, S. (2021). Polyamines: Functions, 696 Metabolism, and Role in Human Disease Management. Med Sci (Basel) 9. 697 Sawyer, J. K., Kabiri, Z., Montague, R. A., Allen, S. R., Stewart, R., Paramore, S. V., Cohen, E., 698 Zaribafzadeh, H., Counter, C. M. and Fox, D. T. (2020). Exploiting codon usage identifies 699 intensity-specific modifiers of Ras/MAPK signaling in vivo. PLoS Genet 16, e1009228. 700 Schauder, K. B. and Bennetto, L. (2016). Toward an Interdisciplinary Understanding of Sensory 701 Dysfunction in Autism Spectrum Disorder: An Integration of the Neural and Symptom 702 Literatures. Front Neurosci 10, 268. 703 Schramm, J., Sholler, C., Menachery, L., Vazquez, L. and Saulnier Sholler, G. (2025). Polyamine 704 Inhibition with DFMO: Shifting the Paradigm in Neuroblastoma Therapy. J Clin Med 14. 705 Snyder, R. D. and Robinson, A. (1969). Recessive sex-linked mental retardation in the absence of 706 other recognizable abnormalities. Report of a family. Clin Pediatr (Phila) 8, 669-74. 707 Stark, F., Pfannstiel, J., Klaiber, I. and Raabe, T. (2011). Protein kinase CK2 links polyamine 708 metabolism to MAPK signalling in Drosophila. Cell Signal 23, 876-82. 709 Stessman, H. A., Xiong, B., Coe, B. P ., Wang, T., Hoekzema, K., Fenckova, M., Kvarnung, M., Gerdts, 710 J., Trinh, S., Cosemans, N. et al. (2017). Targeted sequencing identifies 91 711 neurodevelopmental-disorder risk genes with autism and developmental-disability biases. 712 Nat Genet 49, 515-526. 713 Stewart, T. M., Foley, J. R., Holbert, C. E., Khomutov, M., Rastkari, N., Tao, X., Khomutov, A. R., Zhai, 714 R. G. and Casero, R. A., Jr. (2023). Difluoromethylornithine rebalances aberrant polyamine 715 ratios in Snyder-Robinson syndrome. EMBO Mol Med 15, e17833. 716 Tartaglia, M., Aoki, Y . and Gelb, B. D. (2022). The molecular genetics of RASopathies: An update on 717 novel disease genes and new disorders. Am J Med Genet C Semin Med Genet 190, 425-439. 718 Walker, J. A., Tchoudakova, A. V., McKenney, P . T., Brill, S., Wu, D., Cowley, G. S., Hariharan, I. K. and 719 Bernards, A. (2006). Reduced growth of Drosophila neurofibromatosis 1 mutants reflects a 720 non-cell-autonomous requirement for GTPase-Activating Protein activity in larval neurons. 721 Genes Dev 20, 3311-23. 722 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 29 Walsh, K. S., Wolters, P . L., Widemann, B. C., Del Castillo, A., Sady, M. D., Inker, T., Roderick, M. C., 723 Martin, S., Toledo-Tamula, M. A., Struemph, K. et al. (2021). Impact of MEK Inhibitor 724 Therapy on Neurocognitive Functioning in NF1. Neurol Genet 7, e616. 725 Wang, J. Q., Fibuch, E. E. and Mao, L. (2007). Regulation of mitogen-activated protein kinases by 726 glutamate receptors. J Neurochem 100, 1-11. 727 Wickham, H. (2016). ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag New York. 728 Wolman, M. A., de Groh, E. D., McBride, S. M., Jongens, T. A., Granato, M. and Epstein, J. A. (2014). 729 Modulation of cAMP and ras signaling pathways improves distinct behavioral deficits in a 730 zebrafish model of neurofibromatosis type 1. Cell Rep 8, 1265-70. 731 Wu, B. and Liu, S. (2024). Structural Insights into the Mechanisms Underlying Polyaminopathies. Int J 732 Mol Sci 25. 733 Xuan, M., Gu, X., Li, J., Huang, D., Xue, C. and He, Y . (2023). Polyamines: their significance for 734 maintaining health and contributing to diseases. Cell Commun Signal 21, 348. 735 Zenker, M. (2022). Clinical overview on RASopathies. Am J Med Genet C Semin Med Genet 190, 414-736 424. 737 738 739 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 30 740 Figure S1. Sensory processing profiles upon knockdown of polyamine metabolism genes, related to 741 Figure 4. (A) Jump rates in reactivity assay and (B) habituation ratios of knockdown models ( Act-742 Gal4,GMR-wIR > UAS-RNAi) and controls (Act-Gal4,GMR-wIR / control; 3-4 replicates, 62-128 total flies 743 per genotype). Line graphs indicate mean ± SEM jump responses per bins of 5 stimuli. Boxes indicate 744 median ± 25th and 75th percentiles. 745 746 747 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 31 748 Figure S2. All western blots and ratios evaluating pERK, ERK and β-tubulin profiles upon knockdown 749 of polyamine metabolism genes, related to Figure 4. (A) All western blot bands (labeled bands are 750 excluded from quantification) and (B) quantifications of pERK/β-tubulin and (C) ERK/β-tubulin ratios in 751 lysates of 10 pooled fly heads collected after behavioral testing (5-7 samples per genotype). Graphs 752 indicate mean ± SEM. 753 754 755 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint 32 Supplementary table titles 756 Table S1. Results of ELISA screen on habituation deficit fly lines, related to Figure 1. 757 Table S2. List of excluded habituation deficit fly lines. 758 Table S3. RAS and ERK-related GO-terms enriched for differentially expressed genes in SmsG56S mouse 759 model. 760 Table S4. List of fly lines used in this study. 761 762 763 764 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 24, 2026. ; https://doi.org/10.64898/2026.01.23.701032doi: bioRxiv preprint