Persistent neuronal stress signalling underlies Angelman syndrome phenotypes

Angelman syndrome (AS) occurs in 1:20,000 births and is characterized by intellectual disability, developmental delay, epilepsy, microcephaly, dysmorphic facial features and a unique behavioural profile^1,2. It is caused by deletion or mutation of the maternally inherited copy of UBE3A, a gene also linked to mood disorders^7,8, autism^4,5,7 and cancer^9,10. It is widely accepted that phenotypes in AS arise due to the dysregulation of UBE3A substrates during development. However, although numerous substrates have been proposed³, establishing their link to AS phenotypes has proved difficult, limiting researchers’ mechanistic understanding of this disorder.

UBE3A is the founding member of the HECT (homologous to E6-AP C terminus) domain family of E3 ubiquitin ligases, all of which have a conserved HECT domain that contains the necessary biochemical machinery to accept activated ubiquitin from E2 enzymes and to transfer it to target proteins⁹. UBE3A expression is imprinted specifically in neurons such that only the maternally inherited allele is expressed, with the paternal allele being epigenetically silenced¹¹. This imprinting mechanism, along with many disease-related phenotypes, is conserved in AS model mice that lack a maternal copy of Ube3a (Ube3a^m−/p+)¹². These phenotypes include motor and behavioural deficits, and microcephaly, the most prominent neuroanatomical deficit in AS¹³. Microcephaly is not present at birth in mice but is acquired postnatally, driven by a failure of neuronal growth that results in shorter and simplified dendritic arbors and smaller-diameter axons^14,15.

The integrated stress response (ISR) is a conserved pathway that helps maintain cellular homeostasis in response to noxious stimuli, including proteotoxic stress, nutrient deprivation, viral infection and oxidative stress¹⁶. These stimuli are sensed by four specialized kinases (PERK, GCN2, PKR and HRI) that converge on phosphorylation of the α-subunit of eukaryotic translation initiation factor 2 (p-eIF2α) to suppress global protein synthesis and initiate compensatory transcription programs. Asparagine synthetase (ASNS) is a conserved metabolic enzyme that is induced by the ATF4 transcription factor in response to amino acid deprivation¹⁷. ASNS catalyses the ATP-dependent synthesis of asparagine and glutamate from aspartate and glutamine, but its role in neuronal development remains poorly understood.

To identify substrates of UBE3A, we used cell lines expressing gain-of-function (GOF) and loss-of-function (LOF) mutants^4,5 and subjected them to stable isotype labelling by amino acids in cell culture (SILAC), tryptic remnant immunoaffinity enrichment (di-Gly purification) and mass spectrometry¹⁸ (Fig. 1a). We performed this experiment in quadruplicate and identified proteins whose ubiquitination levels were enriched by at least 20% in cells expressing a hyperactivating mutation in UBE3A (Supplementary Table 1). Because of technical issues in this experiment (see Methods for details), the total numbers of detected proteins varied across replicates. However, comparing unique proteins across replicates allowed us to compile a list of 23 proteins that could serve as potential substrates (Fig. 1b). Within this list were novel substrates, as well as many proteins that are known to bind to or become ubiquitinated by UBE3A (Fig. 1c).

To validate these candidates, first we compared protein abundance by western blot analysis in HEK293T cells expressing (1) wild-type (WT) or ligase-dead UBE3A (C820A; UBE3A(LD)), and (2) WT UBE3A or the GOF UBE3A(T485A) mutant⁵. In both experiments, the abundance of the catalytic subunit of DNA-dependent protein kinase (DNA-PKc), ASNS and the ribosomal protein RPL3 was affected by UBE3A in a manner consistent with these proteins being substrates of UBE3A (Fig. 1d,e and Extended Data Fig. 1a–d). Second, we generated knockdown (KD) cells stably expressing a short hairpin RNA (shRNA) targeting UBE3A (UBE3A KD), and analysed candidate substrate abundance. A moderate reduction of UBE3A (25%) increased the abundance of ASNS, DNA-PKc and RPL3 in these cells (Extended Data Fig. 1e,f). Third, we treated cells with the proteasome inhibitor MG132 (15 μM, 12 h) to test proteasome dependence. MG132 reversed the effect of WT UBE3A on DNA-PKc and ASNS abundance but had no effect on RPL3, suggesting that the UBE3A-elicited changes in RPL3 were proteasome independent (Extended Data Fig. 1g,h).

We next compared the results of our screen with the findings of a previous co-immunoprecipitation experiment using UBE3A(LD) overexpressed in HEK293T cells¹⁸. Fifteen of the 23 candidates, including ASNS and DNA-PKc, were also identified as binding partners in this previous experiment (Fig. 1f). Finally, we compared a list of 55 proposed UBE3A substrates, including ASNS and DNA-PKc, with a previously reported dataset of upregulated proteins identified in two independent patient-derived AS neuron lines⁶ (Supplementary Table 2). Of the 532 proteins that were commonly upregulated in both lines, only 4 proposed substrates were on this list, including annexin A1¹⁹, peroxiredoxin 1²⁰, PEG10⁶ and ASNS (Fig. 1g). Because every validation step pointed to ASNS, we expanded our analysis of this protein.

ASNS is composed of a glutamine hydrolysing domain and an asparagine synthase domain, reflecting its two catalytic activities (Extended Data Fig. 2a). We first performed co-immunoprecipitation experiments and found that overexpressed UBE3A co-purifies with endogenous ASNS in HEK293T cells, consistent with our previous mass-spectrometry results¹⁸ (Extended Data Fig. 2b). We next reconstituted an in vitro ubiquitin reaction consisting of recombinant human ASNS, UBE1 (E1), UBCH7 (E2), UBE3A, ubiquitin and ATP. We found that UBE3A was required to catalyse the time-dependent formation of high-molecular-weight ubiquitin chains on ASNS while simultaneously reducing the level of unmodified ASNS (Extended Data Fig. 2c–e). Finally, we examined ASNS ubiquitination in UBE3A-KD cells expressing HA-tagged ASNS and Flag-tagged ubiquitin and treated with MG132 (30 μM) for 1 h. A modest decrease in UBE3A produced a proportional 25% decrease in ASNS ubiquitination (Extended Data Fig. 2f,g), suggesting that UBE3A is the principal E3 enzyme targeting ASNS for proteasomal degradation in these cells.

Our proteomics experiment identified five candidate ubiquitination sites on ASNS: K83 and K168, which are poorly conserved and located in the glutamine hydrolysing domain; and K467, K478 and K504, which are more highly conserved and are found in the ASNS domain (Extended Data Fig. 2h). To confirm UBE3A-dependent ubiquitination at these sites, we generated ASNS lysine-to-arginine (KR) mutants and co-transfected them with Flag-tagged ubiquitin and UBE3A into HEK293T cells. Whereas ubiquitination of the K83R and K168R mutants resembled what was observed with WT ASNS, it was reduced or altered for the K467R, K478R and K504R mutants (Extended Data Fig. 2i). In steady-state experiments, WT UBE3A reduced the abundance of WT ASNS and the K83R and K168R mutants, but no changes were observed for the K467R, K478R and K504R mutants (Extended Data Fig. 2j,k). Because degradative ubiquitination is rarely directed by a single lysine, we cannot rule out changes in cooperativity, ubiquitination linkage or ubiquitination kinetics arising from K467R, K478R and K504R mutations. Nonetheless, these results strongly suggest that ubiquitination in the asparagine synthase domain is crucial for ASNS degradation.

We next assessed UBE3A-mediated ASNS regulation in vivo by analysing ASNS protein levels and ubiquitination in the brains of Ube3a^m−/p+ mice (referred to as Angelman mice). In hippocampal tissue, immunoprecipitated ASNS showed a reduction in ubiquitination of about 50% in Angelman mice, compared with WT controls (Extended Data Fig. 3a,b). Similarly, ASNS abundance was increased in the hippocampus (130.63 ± 3.92% above WT) and cortex (114.55 ± 3.60%) with a trend in the midbrain, but not in cerebellar samples (Extended Data Fig. 3c). These changes were not due to an increase in Asns transcription, because quantitative PCR (qPCR) of hippocampal RNA showed that there was no difference in Asns mRNA levels (Extended Data Fig. 3d). Finally, ASNS protein levels in hippocampal lysates from mice with an equivalent maternal GOF T485A mutation in UBE3A (mT503A in mice²¹) showed a subtle but consistent decrease (Extended Data Fig. 3e,f). Altogether, our biochemical experiments show that UBE3A ubiquitinates ASNS both in vitro and in vivo.

ASNS is expressed widely across tissues and cell types²¹; however, its expression in the brain is not well characterized. We visualized Asns transcripts in mouse glutamatergic neurons, GABAergic neurons and non-neuronal cells using whole-brain single-cell RNA sequencing (scRNA-seq) and spatial transcriptomic data²². Notably, Asns was expressed broadly in both excitatory and inhibitory neurons but was mostly absent in non-neuronal cells (Fig. 2a). When we extended this analysis to cell subclasses across the brain, we found that Asns was expressed in most neuronal subclasses but was undetectable in nearly all non-neuronal cells (Fig. 2b and Supplementary Table 3). We further confirmed this using a published scRNA-seq dataset from the mouse hippocampus²³, which again showed that Asns expression was enriched in neurons (Fig. 2c), a pattern that matches the neuron-specific imprinting of Ube3a.

To characterize ASNS protein levels in situ, we validated an antibody against mouse ASNS protein using mouse Neuro2a cells transfected with a plasmid that expresses EGFP along with an shRNA that targets mouse Asns. We confirmed the specificity of the antibody through immunofluorescence experiments (Extended Data Fig. 4a–d). We first examined ASNS protein in the developing brain, focusing on the cortex and hippocampus. In mice at postnatal day 0 (P0), ASNS staining was most prominent in the cortical neuronal layers. Cell bodies and apical dendrites were prominently labelled, whereas axonal fibres showed subtle labelling, and staining was mostly absent from proliferative zones adjacent to the lateral ventricles (Fig. 2d). By P10, ASNS immunoreactivity was diffuse throughout the cortex and hippocampus, with more prominent somatic labelling in deeper-layer cortical neurons and the pyramidal layers of the developing hippocampus. By P21, ASNS staining was intermittent throughout the brain and enriched in deep layers of the cortex, the dentate gyrus, neuronal layers of the hippocampus, the amygdala, and thalamus (Fig. 2d,e). In other regions, ASNS strongly labelled most serotonin (5-hydroxytryptamine; 5-HT)-positive serotonergic neurons of the dorsal raphe (88.20 ± 5.77% of cells) and a subset of tyrosine hydroxylase (TH)-positive dopaminergic neurons in the substantia nigra pars compacta (62.36 ± 3.83% of cells), suggesting that ASNS protein is heterogeneously maintained across the brain (Fig. 2f,g).

To test whether these ASNS-enriched cells were neurons, we co-stained for markers of neurons (NeuN), astrocytes (GFAP) and microglia (IBA1) in the cortex and hippocampus. Virtually all ASNS-positive cells were co-localized with NeuN, confirming that ASNS transcript and protein are predominantly neuronal (Fig. 2h–j). We next asked whether the heterogeneous labelling was stochastic or systematic by examining ASNS in inhibitory cells of the stratum oriens and stratum radiatum. If stochastic, ASNS-enriched cells should be represented equally across inhibitory neuron subclasses. If systematic, then ASNS enrichment should be biased towards specific subclasses. We co-stained brain slices for ASNS and several widely used markers of interneurons, including parvalbumin (PV), somatostatin (SST), calretinin (CR) and neuropeptide Y (NPY) (Extended Data Fig. 4e–g). Of note, all ASNS-positive cells were NPY-positive (PV, 56.94 ± 6.91%; SST, 57.5 ± 3.82%; CR, 22.62 ± 12.43%; NPY, 100%), suggesting that ASNS enrichment is not stochastic but maintained by functionally distinct classes of cells.

Finally, we asked whether loss of UBE3A increases ASNS protein abundance in selected neuronal populations. Using immunofluorescence, we examined ASNS abundance in NYP-positive interneurons and CaMKIIα-positive excitatory neurons in the hippocampus (Extended Data Fig. 4h–k). We saw a subtle increase in in both excitatory and inhibitory cells, which suggests that the loss of UBE3A affects ASNS abundance broadly rather than in specific cell populations.

We next wondered whether correcting ASNS overabundance could affect aberrant phenotypes in Angelman mice. So far, only one Asns hypomorphic mouse line has been published²⁴. We therefore generated a new null line that contains a CRISPR–Cas9-targeted nonsense mutation in Asns. We crossed heterozygous Asns-null males (Asns^+/−) with paternal-null Ube3a females (Ube3a^m+/p−) to generate four genotypes: WT (Ube3a^m+/p+, Asns^+/+), Het (Ube3a^m+/p+, Asns^+/−), Angelman (Ube3a^m−/p+, Asns^+/+) and Rescue (Ube3a^m−/p+, Asns^+/−) (Fig. 3a). As expected, western blotting and immunostaining analysis showed that Het mice had roughly 50% of WT ASNS protein levels, and ASNS abundance in Rescue mice returned to near WT levels (Fig. 3b and Extended Data Fig. 5a–c).

Behavioural phenotypes in Angelman mice are well characterized¹² and we used a range of assays to examine these established phenotypes (Fig. 3c). Angelman mice are known to exhibit adult-onset obesity, which becomes apparent around 3 months of age¹⁴. We observed a similar obesity phenotype that was greatly improved in Rescue mice, independent of sex (Fig. 3d and Extended Data Fig. 5d). In the open field assay, Angelman mice ambulated less and spent more time in the centre of the arena than control mice did²⁵. Notably, Rescue mice were indistinguishable from WT mice in terms of time spent in the centre of the arena (Fig. 3e), even though they showed deficits in the total distance travelled (Fig. 3f). We found additional phenotypic improvements in Rescue mice in the hindlimb clasping assay (Fig. 3g) and the marble bury assay (Fig. 3h,i). Notably, Het mice performed indistinguishably from WT mice in every assay.

Reducing the overabundance of ASNS did not rescue all phenotypes. For example, Rescue mice showed no improvement in the rotarod assay (Extended Data Fig. 5e). Angelman mice are known to skip a period of sleep, known as the siesta, that normally occurs during a mouse’s active phase²⁶. We observed a similar phenotype in our Angelman mice, but sleep measures did not improve in Rescue mice (Extended Data Fig. 5f–h). These results indicate that many—but not all—AS phenotypes in mice result from the dysregulation of ASNS, suggesting that there are additional disease-relevant substrates.

Given that ASNS maintains amino acid homeostasis, we investigated the effects of UBE3A loss on the levels of amino acids in the brains of Angelman mice. We isolated hippocampi from all four genotypes at P21 for metabolomic profiling (Fig. 4a). As expected, levels of asparagine were decreased in Het and Rescue mice, both of which lack one copy of Asns (Fig. 4b and Supplementary Table 4). Conversely, asparagine levels were increased in Angelman mice, as compared with WT controls, functionally confirming the overabundance of ASNS in these mice. Of note, no amino acid was decreased, but rather, we observed significantly increased levels of phenylalanine, glycine and proline, as well as a trend towards increased levels of glutamine, histidine, leucine and tyrosine (Fig. 4b). All of these changes returned to baseline in Rescue mice, suggesting that they resulted from ASNS overabundance.

Loss of amino acid homeostasis is closely linked to the ISR, a conserved pathway that ultimately halts protein synthesis. Activation of the ISR is typically detected through increased phosphorylation of the eIF2α translation initiation factor at serine 51 (p-eIF2α)²⁷. To determine whether this pathway was engaged in Angelman mice, we assessed the levels of p-eIF2α by western blot in brain hemisphere lysates. We observed a significant increase in p-eIF2α in Angelman mice (176.43 ± 7.13% of WT), indicating increased ISR signalling (Fig. 4c). Notably, this increase was improved by correcting ASNS overabundance in Rescue mice (p-eIF2α; 140.56 ± 10.94% of WT).

We next asked whether simple overexpression of ASNS is sufficient to activate the ISR in neurons. We used an adeno-associated virus (AAV) encoding EGFP or ASNS-T2A-EGFP (in which a 2A self-cleaving peptide (T2A) is inserted at the 5′ end of Asns) in cultured mouse cortical neurons, and assessed their effect on ISR signalling. Overexpression of ASNS was sufficient to increase the levels of p-eIF2α above EGFP-alone controls (129.27 ± 2.20% of control) (Extended Data Fig. 6a,b), indicating that ASNS overabundance alone is sufficient to activate the ISR.

Amino acid starvation activates a branch of the ISR known as the amino acid response (AAR). Because our metabolomic studies showed that free amino acid concentrations were altered in Angelman mice, we performed RNA-seq to determine whether the AAR is genuinely engaged in these mice. We sequenced RNA extracted from the cortex and hippocampus of P21 WT, Angelman and Rescue mice. We found few differentially expressed genes (DEGs) in the hippocampus of Angelman mice relative to WT controls (51 DEGs; false discovery rate (FDR) < 0.05) with only eight highly changed genes (log₂-transformed fold expression < −0.75 or > 0.75; Extended Data Fig. 6c and Supplementary Table 5). By contrast, we found a much stronger signal in the cortex (1,717 DEGs, 777 up, 940 down; FDR < 0.05) with 189 highly changed genes (Fig. 4d,e and Supplementary Table 5). Given this more robust transcriptional response, we focused subsequent analyses on the cortex dataset.

We found only 22 DEGs (FDR < 0.05) in Het mice and 1,214 DEGs (623 up, 591 down; FDR < 0.05) in Rescue mice, 805 of which overlapped with Angelman mice (Fig. 5e). Gene Ontology (GO) analysis showed that downregulated genes implicated pathways related to mitochondrial function and protein synthesis, whereas upregulated genes were enriched for pathways related to neuronal growth and synaptic function (Fig. 4f,g). Notably, the significance of these changes was uniformly weakened in Rescue mice (Fig. 4f–i), strongly suggesting that they occurred as a direct result of ASNS overabundance. We specifically examined genes related to the ISR (GO: 0140467) and found only a weak enrichment, with just 7 of 51 genes in this set differentially expressed in Angelman mice (Extended Data Fig. 6d). We further compared our data with RNA-seq data from mice that lack Slc7a5, a gene encoding an essential amino acid transporter at the blood–brain barrier²⁸. These mice show increased expression of genes encoding amino acid transporters (Slc7a1 and Slc7a3), transcription factors that are central to the ISR (Atf4 and Atf5) and transfer RNA (tRNA) synthases (Extended Data Fig. 6e). However, we did not detect changes in any of these genes in any of our genotypes, suggesting that these pathways are not activated in Angelman mice.

Of note, we found that Eif2ak1 is upregulated in Angelman mice and returned to baseline in Rescue mice (Extended Data Fig. 6d). Eif2ak1 encodes a kinase called HRI (haem-regulated inhibitor of translation), which is classically known to phosphorylate eIF2α in response to low levels of haem in the cell²⁹. More recent work has found that HRI promotes mitophagy³⁰, and that its expression is induced in neurons in response to proteasome inhibition³¹. Consistent with our RNA-seq results, HRI protein abundance was increased in the cortex of Angelman mice (Extended Data Fig. 6f,g), and in addition, we found that overexpression of ASNS in neurons was sufficient to increase HRI abundance (Extended Data Fig. 6j,k). By contrast, overexpression of ASNS did not activate GCN2 (general control nonderepressible 2), the kinase that phosphorylates eIF2α during the AAR (Extended Data Fig. 6h,i), nor did it induce changes in the two other upstream kinases PERK and PKR (Extended Data Fig. 6j,k). Together, these observations suggest that ASNS overabundance in Angelman mice drives a downregulation of mitochondrial and protein synthesis pathways, and that upregulation of the HRI kinase leads to increased phosphorylation of eIF2α.

We next examined protein synthesis directly, using two different strategies. First, we measured new protein synthesis in vivo by using intraperitoneal injections of l-azidohomoalanine (AHA). AHA is a cell- and blood–brain-barrier-permeable synthetic amino acid that is charged onto methionine tRNAs and incorporates into newly synthesized proteins³². Proteins containing AHA can be labelled and visualized using a process known as bioorthogonal noncanonical amino acid tagging (BONCAT)³². We extracted proteins from mouse brains 16 h after injection (Fig. 5a), and found that on average there was a four-to-five-fold reduction in the rates of new protein synthesis in Angelman mice, compared with WT mice (Fig. 5b,c). This deficit was greatly improved in our Rescue mice (65.949 ± 7.569 of WT). Second, we performed polysome profiling in all four genotypes by isolating actively translating ribosomes from the whole brain at P21 (Fig. 5d). Quantification of polysome/monosome ratios revealed a significant shift towards monosome fractions in Angelman mice, compared with WT mice (Angelman 72.94 ± 2.83% of WT; Het, 95.57 ± 4.32% of WT; Rescue, 95.08 ± 5.12% of WT), indicating decreased translational efficiency by polysomes (Fig. 5e). Again, this deficit was improved in our Rescue mice. We further examined whether ASNS overabundance alone is sufficient to slow new protein synthesis in neurons. Cultured neurons overexpressing ASNS were treated with AHA, and fluorescent noncanonical amino acid tagging (FUNCAT)³³ was performed to visualize the incorporation of AHA. AHA fluorescence was reduced by around 50% in neurons overexpressing ASNS (Extended Data Fig. 7a,b), suggesting that ASNS overabundance causes a global slowing of protein synthesis.

Finally, we sought to understand how reduced protein synthesis contributes to disease-relevant phenotypes. Previous studies showed that microcephaly in Angelman model mice arises from impaired neuronal growth that stunts axon and dendrite development^14,15. We hypothesized that a reduction in neuronal protein synthesis contributes to poor neuronal growth, underlying the microcephaly observed in these mice. We measured brain weights in all four genotypes at P21. Consistent with previous reports, we observed a significant reduction in brain weights in Angelman mice¹⁴. This deficit was greatly improved in Rescue mice (Fig. 5f), independent of sex (Extended Data Fig. 7c). A previous study reported that smaller neurons in Angelman model mice cause greater neuronal packing in the cortex, leading to increased neuronal densities¹⁴. We therefore analysed neuronal density, and found that this measure was also improved in our Rescue mice (Extended Data Fig. 7d,e). Altogether, these data suggest that ASNS overabundance triggers the ISR and reduces the rate of protein synthesis, which, in turn, negatively affects neuronal growth and development in AS.

Identifying disease-relevant substrates in AS remains difficult, given the promiscuity of UBE3A promiscuity and its modest effect on substrate abundance. Here, we addressed this challenge by using disease-linked UBE3A mutants^4,5,34, combined with di-Gly purification and mass spectrometry. Although ASNS seems to be a major contributor to AS, not all phenotypes were rescued by ASNS correction, suggesting that other substrates contribute to overall pathology. Two limitations of our proteomics experiments are worth noting. First, although it is unlikely, we cannot rule out that endogenous UBE3A competed with overexpressed UBE3A for binding substrate proteins. Second, because our workflow relied on affinity purification of ubiquitinated proteins, low abundance or rapidly degraded proteins might be under-represented in our data. Future studies using deeper proteomic coverage or complementary approaches will be needed to fully define the UBE3A substrate landscape relevant to AS pathology.

Our study found that Asns expression is limited mainly to neurons (Fig. 2), matching the neuron-specific imprinting pattern of Ube3a. Asparagine is a non-essential amino acid that is readily attained through the diet, but at the blood–brain barrier, it must be transported using a bidirectional facilitated transporter system (System N)³⁵. However, rather than driving the accumulation of asparagine into the brain, this system seems to remove it from the cerebrospinal fluid³⁶. We hypothesize that neurons rely on ASNS to synthesize most, if not all, of the asparagine they need, and that perturbations to ASNS represent a point of susceptibility in development.

Homozygous loss of ASNS causes asparagine synthetase deficiency, a rare congenital condition characterized by microcephaly, global developmental delay, hypotonia, spastic quadriplegia and intractable seizures³⁷. However, contrary to a loss of ASNS function, our study found increased levels of ASNS protein in Angelman model mice, suggesting that a gain in ASNS function causes aberrant phenotypes. In support of this idea, we found that overexpression of ASNS alone in neurons was sufficient to increase both p-eIF2α and HRI levels (Extended Data Fig. 6a,j), and that genetic reduction of ASNS expression rescued protein synthesis as well as many aberrant behavioural phenotypes (Figs. 3 and 5). GOF mutations in ASNS have yet to be established as a cause of neurodevelopmental disorders; however, our results suggest that such a mechanism is plausible.

Our finding that free amino acids in the brains of Angelman mice are elevated (Fig. 4b) is consistent with our observation that neuronal protein synthesis is reduced with the loss of UBE3A. However, contrary to our original hypothesis, we did not find an activation of the AAR in Angelman mice. The only ISR-associated gene that was upregulated in Angelman mice and rescued with Asns modulation was Eif2ak1, the gene that encodes HRI. It is notable that HRI is induced in neurons upon proteostatic stress³¹, and this observation, combined with UBE3A’s known role at the proteasome^18,38,39,40, suggests that proteostatic stress has a crucial role in AS pathology. Moreover, HRI mediates both mitochondrial maintenance³⁰ and protein synthesis³¹, two pathways that are highly implicated in the current study. Whether these changes result from increased asparagine production, or through an alternative unidentified activity of ASNS, remains to be elucidated.

Our study found a reduction in new protein synthesis and a reduction in polysome, but not monosome, fractions in neurons (Fig. 5d). Previous work has shown that monosomes are predominantly localized to neuronal processes and are capable of synthesizing proteins from at least a subset of transcripts⁴¹. Future studies will be required to address whether monosome-dependent forms of translation are preserved in Ube3a^m−/p+ neurons.

Finally, our results have therapeutic implications. Our finding that persistent stress signalling is central to AS pathology places it among a growing list of neurodevelopmental disorders in which disrupted proteostasis and heightened stress signalling are a common mechanism of pathology^28,42,43. This suggests that targeting ASNS, HRI or the ISR could alleviate at least some symptoms in AS and could complement genetically targeted therapies to improve their safety and efficacy.

Mice

All animal procedures were approved by and performed in accordance with the guidelines of the Institutional Animal Care and Use Committee (IACUC) at Washington University in St. Louis. Mice were housed in a specific-pathogen-free (SPF) facility managed by the Division of Comparative Medicine at Washington University School of Medicine, which is fully accredited by AAALAC International. Mouse holding rooms were maintained in a controlled environment with an ambient temperature of 20–24 °C and a relative humidity of 40–60%. Mice were kept on a regulated 12-h light–12-h dark cycle and provided with standard laboratory chow and water ad libitum. Ube3a-deficient mice were generated previously⁴⁴ and obtained through Jackson Laboratory (016590). The mice used throughout this study were all maintained on a C57BL6/J background. Angelman mice (Ube3a^m−/p+) were generated by crossing Ube3a^m+/p− heterozygous females with WT males. An Asns-heterozygous mouse was generated at the Washington University Mouse Genetics Core using a guide RNA to generate a deletion of a single nucleotide at the Q17 codon on exon 2 of Asns (chr. 6: 7689452, GRCm39 assembly). Mice were backcrossed at least seven generations with WT C57Bl/6J mice (Ube3a^m+/p+; Asns^+/+) before use in the study. mRNA of Cas9 and the selected gRNA were injected into fertilized hybrid C57Bl/6J/JxCBA eggs at the Mouse Genetics Core at Washington University School of Medicine and validated by targeted Sanger sequencing. Mice were genotyped by incubating tail-tips overnight at 55 °C in tail digest buffer (0.1 M Tris pH 8.8, 0.5 M NaCl, 0.005 M EDTA and 0.5% SDS) containing 0.4 mg ml⁻¹ proteinase K (Gold Biotechnology, P-480-SL10). Afterwards, proteinase K was inactivated by incubation at 95 °C for 10 min and genomic DNA was purified by isopropanol precipitation. PCR was performed by amplifying a 486-bp fragment using forward primer 5′-GGTGCAGGGTTTATGGTTGGA-3′ and reverse primer 5′-GCTTACCGCCTTGTGGTTGTAG-3′. The products were purified using a PCR purification kit (QIAGEN, 28106) and submitted for Sanger sequencing (GENEWIZ). PCR was performed using MyTaq DNA polymerase (Meridian Bioscience, BIO-25043) with an initial denaturation step of 95 °C for 1 min, followed by 30 cycles of denaturation at 95 °C for 15 s, annealing at 60 °C for 15 s and an extension step of 72 °C for 55 s.

Cell lines and culture

HEK293T cells and Neuro2a cells, obtained from ATCC, were cultured in a 5% CO₂ humidified incubator and tested regularly for mycoplasma contamination. Cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) pre-supplemented with GlutaMAX (Gibco, 10569044), 5% fetal bovine serum (FBS; Gibco, 16140071) and 1% antibiotic–antimycotic solution (Gibco, 15240096) containing 10,000 units ml⁻¹ penicillin, 10,000 µg ml⁻¹ streptomycin and 25 µg ml⁻¹ of amphotericin B. For biochemical analysis, transfections were done using the FuGENE 6 transfection reagent (Promega, E2691), according to the manufacturer’s instructions.

We established a stable UBE3A-KD cell line (HEK293T) by using an shRNA (Sigma, TRCN0000012894; target sequence: CCCAATGATGTATGATCTAAA) against the UBE3A gene. A lentiviral construct containing a scrambled shRNA sequence (Sigma, SHC002) was used as a control. pLKO.1-puro-UBE3A shRNA and scrambled shRNA were transfected into HEK293T cells and stable clones were selected with 1 μg ml⁻¹ puromycin. Despite consecutive optimization trials to maximize silencing efficiency, the viable stable clones consistently exhibited a moderate reduction of about 25% in steady-state UBE3A expression levels. However, this level of knockdown proved biologically sufficient to elicit robust biochemical alterations in downstream target proteins. For proteasome inhibition assays, cells were treated with 15 μM MG132 (Calbiochem, 474790) for 12 h before collection, and cells treated with an equal volume of dimethyl sulfoxide (DMSO) served as the vehicle control.

Primary cortical neurons were cultured and maintained using the B-27 Plus neuronal culture system (Gibco, A3653401) supplemented with N2 (Gibco, 17502048) and penicillin–streptomycin (Gibco, 15140122). In brief, embryos were isolated at embryonic day 14.5 (E14.5) and the cortex of WT and Angelman mice was dissected. The tissue was dissociated using a trypsin solution and mechanical trituration to obtain a single-cell suspension. After dissociation, cells were seeded onto cell culture dishes pre-coated with poly-d-lysine (Sigma-Aldrich, P0899). Half of the neuronal culture medium was replaced every 2 days until the desired day of the experiment.

Plasmids and AAVs

MYC-tagged human UBE3A isoform 2 constructs used in this study, including WT, ligase dead (LD; C829A), UBE3A(T485A) and UBE3A(T485E), were cloned into the pCIG2 plasmid DNA as previously described^4,5. This vector is driven by a chicken β-actin promoter with bicistronic GFP expression via an internal ribosome entry site (IRES). The WT HA-tagged ASNS construct, derived from human ASNS and containing the complete reading frame (accession: BC014621.2), was cloned into the pCIG2 plasmid using PCR to introduce SacI and XmaI sites. The constructed plasmids were verified by DNA sequencing (GENEWIZ) and purified using a HiSpeed Plasmid Midi Kit (QIAGEN, 12643). The oligonucleotide sequences used to generate WT HA–ASNS and five lysine mutants of ASNS (K83R, K168, K467, K478 and K504) are provided in Supplementary Table 6. The HA codons are shown in red, and the blue colour indicates the presence of restriction sites. Lysine mutations in the oligonucleotides used for mutagenesis are shown in bold. To create viral expression vectors, WT HA–ASNS was subcloned into pAAV-hSyn-EGFP plasmid DNA (Addgene, 50465) by PCR using a KpnI site. To separate ASNS from EGFP, we inserted a 2A self-cleaving peptide (T2A) at the 5′ end of ASNS. Control virus and AAV-hSyn-WT-HA-ASNS-T2A-EGFP were packaged into AAV2/9n capsids (Addgene, 112865) with an AAV helper plasmid and pAdDeltaF6 (Addgene, 112867), and transfected into HEK293T cells at a 1:1:1 molar ratio using PEI MAX (Polysciences, 24765-100). AAVs were purified using the AAVpro Purification Kit (Takara, 6666) and titred using a qPCR AAV titration kit (Applied Biological Materials, G931). Primary neurons were transduced with AAVs at a multiplicity of infection of 7.5.

Ubiquitin remnant immunoaffinity profiling with SILAC and LC–MS/MS

HEK293T cells expressing UBE3A(T485A) and UBE3A(T485E) constructs were cultured in DMEM supplemented with high glucose, GlutaMAX supplement and pyruvate (Gibco, 10569044), 10% dialysed FBS (Gibco, 26400044), 1% antibiotic–antimycotic solution containing 10,000 units ml⁻¹ penicillin, 10,000 µg ml⁻¹ streptomycin and 25 µg ml⁻¹ amphotericin B (Gibco, 15240096). For heavy labelling, 50 µg ml^{−1 13}C₆, ¹⁵N₂ l-lysine (Lys8) and 40 µg ml^{−1 13}C₆, ¹⁵N₄ l-arginine (Arg10) (Cambridge Isotope Laboratories, CNLM-291-H-PK and CNLM-539-H-PK, respectively) were added to the medium for cells expressing UBE3A(T485A). For cells expressing UBE3A(T485E) in light medium, standard 50 µg ml⁻¹ l-lysine and 40 µg ml⁻¹ l-arginine (Thermo Fisher Scientific, J62225.36 and A15738.36, respectively) were used. Cells were maintained in their respective labelled or standard medium for at least seven passages to ensure complete labelling. Subsequently, the cells were lysed using a lysis buffer (8 M urea in 50 mM Tris-HCl, 150 mM NaCl, 1 mM EDTA, 2 µg ml⁻¹ aprotinin, 10 µg ml⁻¹ leupeptin, 1 mM PMSF, 50 µM deubiquitinating enzyme (PR-619) and 1 mM iodoacetamide). The lysates were then centrifuged at 20,000g at 4 °C for 15 min, and the protein concentration was determined using a Pierce BCA protein assay kit (Invitrogen, 23227). Equal amounts of lysates from cells cultured in heavy and light media were mixed in a 1:1 ratio (mg:mg). The mixture was diluted in 50 mM Tris (pH 7.5) to a final concentration of 2 M urea. Total protein was enzymatically digested overnight at a 1:100 ratio (trypsin:protein) using sequencing-grade modified trypsin (Promega, V5111). The resulting peptides were purified on a Sep-Pak column (Waters) and eluted with a 40-ml step gradient of 10–50% acetonitrile and 0.1% trifluoroacetic acid (TFA) (10%, 15%, 20%, 23%, 30%, 35%, 40% and 50%: 5 ml each). Each fraction was lyophilized overnight. The lyophilized peptides were subjected to immunoaffinity purification using a PTMScan Ubiquitin Remnant Motif (K-ɛ-GG) Kit (Cell Signaling Technology, 5562). The peptide solution was transferred into the vial containing motif antibody beads, and the mixtures were incubated on a rotator for 4–6 h at 4 °C. K-ɛ-GG peptides were eluted with 50 μl of 0.15% TFA after washing the bead slurry with phosphate-buffered saline (PBS) and water. Eluted peptides were further purified using Pierce C18 spin columns (Thermo Fisher Scientific, 89870) following the manufacturer’s instructions. The peptides were eluted with 70% acetonitrile and 0.1% TFA solution in 50-μl volumes twice and then lyophilized using a SpeedVac.

For mass-spectrometry analysis, lyophilized peptides were resuspended in 1% acetonitrile and 0.1% formic acid. Each sample was analysed using LC–MS/MS (EASY nLC 1000 coupled with QExactive HF mass spectrometer; Thermo Fisher Scientific). The samples were injected onto an EASY-Spray PepMap C18 column (75 μm inner diameter × 25 cm; 2 μm particle size) and separated using a gradient consisting of 5–35% mobile phase B at a flow rate of 250 nl per min flow rate, with mobile phase A consisting of 0.1% formic acid in water and mobile phase B consisting of 0.1% formic acid in acetonitrile. The QExactive HF was operated in data-dependent mode, selecting the 15 most intense precursors for subsequent fragmentation. The resolution (m/z 400–1,600) for the precursor scan was set to 120,000, with a target value of 3 × 10⁶ ions. The resolution of the MS/MS scan was set to 30,000, with a target value of 1 × 10⁵ ions. For higher-energy collision dissociation, we set the normalized collision energy to 27% and the separation window to 1.6 m/z. We set peptide matches as preferred and excluded precursors with unknown charge or charge states of 1 and ≥7. Raw data files were processed using Proteome Discoverer (v.2.1) (Thermo Fisher Scientific). The peak list was searched against the reviewed human UniProt database (containing 20,203 entries) with the contaminant database added using Sequest (v.2.1). For protein identification, the following parameters were used to identify the trypsin peptide: 10 ppm precursor ion mass tolerance, 0.02-Da product ion mass tolerance, a maximum of two missing trypsin cleavage sites, carbamidomethylation of Cys set as a fixed modification and oxidation of Met, gly-gly of Lys, Ser/Thr/Tyr, [¹³C₆]Arg, [¹³C₆]Lys and acetylation of the N terminus set as variable modifications. We used Percolator (v.2.05) to estimate the number of false positive identifications and assigned a q-value of 0.05 (5% FDR), which was used to filter all results. We used the ptmRS node (v.2.1) to assign a probability of post-translational modification site localization. To determine the SILAC rate for ubiquitin-modified peptides, we enabled peptide quantification. For two replicates (experiments 2 and 3), we note that there was a polymer present in our di-Gly motif pull-downs that caused the LC column to become partially clogged, leading to a reduction in the number of total detected proteins. Consequently, owing to these replication-specific technical issues, the total numbers of detected proteins varied across individual replicates, and candidates were filtered on the basis of unique protein comparisons across all valid trials.

Cross-referencing of patient-derived proteomic data

To assess how the identified candidate substrates are altered in human neurons, the compiled list of 55 previously proposed substrates, along with ASNS and DNA-PKc, was cross-referenced against a previously published embryonic-stem-cell-derived proteomic dataset from patient-derived AS neurons (Supplementary Table 2). This clinical dataset comprised upregulated protein profiles from two independent patient-derived backgrounds: a line containing a large maternal deletion encompassing UBE3A (ΔChr. 15q11.2–q12) and a line containing a premature truncating nonsense mutation in the maternal UBE3A allele (Q322X). Overlapping proteins between our candidate screen and the 532 commonly upregulated proteins in both patient lines were identified to extract high-confidence disease-relevant substrates.

Western blot and immunoprecipitation

Cultured cells or mouse brain tissue were lysed in ice-cold RIPA buffer for western blotting or cold modified RIPA buffer for immunoprecipitation. The composition of the RIPA buffer was as follows: 50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1% Triton X-100, 0.1% sodium dodecyl sulfate and 0.25% sodium deoxy-cholate. For western blotting, an EDTA-free protease (Thermo Fisher Scientific, 78437) and phosphatase (Thermo Fisher Scientific, 78428) inhibitor cocktail was added to the RIPA buffer. For immunoprecipitation, 5 mM EDTA and protease and phosphatase inhibitor cocktail were included in the modified RIPA buffer. After lysis, the samples were centrifuged, and protein concentrations were determined using a Pierce BCA protein assay kit (Invitrogen, 23227). Equal amounts of protein were resuspended in 1× LDS sample buffer (Thermo Fisher Scientific, 84788), separated by SDS–PAGE, and transferred onto nitrocellulose membranes (Cytiva, 10600002). Membranes were blocked in 1× fish gelatin blocking agent (Biotium, 22010) in Tris-buffered saline (TBS) containing 0.1% Tween 20 (TBST) for 1 h at room temperature. Afterwards, membranes were incubated with primary antibodies at 4 °C overnight. Bands were visualized using secondary antibodies, either IRDye 680RD donkey anti-mouse IgG (Li-Cor, 926-68072) or IRDye 800CW donkey anti-rabbit IgG (Li-Cor, 926-32213), used at a dilution of 1:10,000. Protein bands were analysed using the Odyssey CLx infrared imaging system and Image Studio v.5.2 software (Li-Cor Biosciences).

For immunoprecipitation analysis, lysates were pre-cleared using protein G Sepharose beads (Sigma-Aldrich, GE17-0618-01), for 1 h at 4 °C on an orbital shaker. An equal amount of pre-cleared lysate was incubated with the primary antibody for 12 h, followed by an additional 4-h incubation with protein G sepharose beads. The resulting immune complexes were isolated by centrifugation, washed at least three times with ice-cold RIPA buffer and resuspended in 2× SDS sample buffer. Protein complexes were resolved by SDS–PAGE and western blot analysis. The primary antibodies used were mouse anti-MYC (1:3,000, EMD Millipore, 05-724), rabbit anti-ASNS (1:3,000 or 1:7,500, Proteintech, 14681-1-AP), mouse anti-DNA-PKcs (1:1,000, Cell Signaling Technology, 12311), rabbit anti-RPL3 (1:3,000, Proteintech, 11005-1-AP), mouse anti-β-actin (1:5,000, Santa Cruz Biotechnology, sc-47778), mouse anti-GFP (1:5,000, Santa Cruz Biotechnology, sc-9996), rabbit anti-FASN (1:2,000, Proteintech, 10624-2-AP), rabbit anti-FANCI (1:3,000, Proteintech, 20789-1-AP), mouse anti-VCP (1:1,000, Invitrogen, MA3-004), rabbit anti-PYGL (1:1,000, Proteintech, 15851-1-AP), rabbit anti-TUBGCP2 (1:1,000, ABclonal, A10283 and 1:3,000, Proteintech, 25856-1-AP), mouse anti-KPNB1 (1:1,000, EMD Millipore, 05-1530), rabbit anti-WRNIP1 (1:1,000, ABclonal, A9170 and 1:2,000, Santa Cruz Biotechnology, sc-376438), rabbit anti-ALDOA (1:3,000, Proteintech, 11217-1-AP), rabbit anti-FBL (1:1,000, ABclonal, A1136), rabbit anti-EIF4A1 (1:2,000, ABclonal, A5294), mouse anti-Flag (1:3,000, Cell Signaling Technology, 8146), rabbit anti-Flag (1:3,000, Cell Signaling Technology, 14793), rabbit anti-E6AP (1:1,000, Bethyl, A300-351A), rabbit anti-LDHB (1:3,000, Proteintech, 14824-1-AP), rabbit anti-eIF2α (1:1,000, Cell Signaling Technology, 5324), rabbit anti-phospho-eIF2α (1:1,000, Cell Signaling Technology, 3398), mouse anti-HA (1:3,000, BioLegend, 901501), rabbit anti-HA (1:3,000, Abcam, ab9110), rabbit anti-ubiquitin (1:3,000, Cell Signaling Technology, 58395), mouse anti-ubiquitin (P4D1) (1:1,000, Santa Cruz, sc-8017), rabbit anti-GCN2 (1:1,000, Cell Signaling Technology, 3302), rabbit anti-GCN2 (phospho T899) (1:1,000, Abcam, ab75836), rabbit anti-PERK (1:1,000, Cell Signaling Technology, 3192), rabbit anti-phospho-PERK (1:1,000, Cell Signaling Technology, 3179), rabbit anti-EIF2AK2 (PKR) (1:1,000, Proteintech, 18244-1-AP) and rabbit anti-EIF2AK1 (HRI) (1:3,000, Proteintech, 20499-1-AP).

In vitro and in vivo ubiquitination assays

In vitro ubiquitination assays were done using recombinant human ubiquitin-activating enzyme UBE1 (R&D Systems, E-305), recombinant UBCH7 (UBE2L3; R&D Systems, E2-640), recombinant human E6AP (UBE3A; R&D Systems, E3-230), recombinant human ubiquitin (R&D Systems, U-100H), recombinant human ASNS (R&D Systems, 10193-AS-020), 10× E3 ligase reaction buffer (500 mM HEPEs (pH 8.0), 500 mM NaCl and 10 mM TCEP)) and ATP (Sigma, A9187). Assays were performed as described previously. In brief, a single reaction contained 0.1 µM UBE1, 1 µM UBE2L3 (UbcH7), 50 µM ubiquitin, 2 mM Mg²⁺-ATP, 0.25 µM UBE3A and 0.125 μM ASNS. The reaction was allowed to proceed at 37 °C for the indicated time. Reactions were stopped by adding SDS sample buffer and boiling for 5 min at 95 °C. The samples were subjected to western blotting using an anti-ASNS antibody (Proteintech, 14681-1-AP).

To detect in vivo ASNS ubiquitination, hippocampi were collected from P43 WT and Angelman mice and lysed under denaturing conditions using a lysis buffer containing 25 mM HEPES (pH 7.4), 1% Triton X-100, 15 mM KCl, 5 mM EDTA, 1% SDS and EDTA-free protease inhibitor cocktail (Thermo Fisher Scientific, 78437). The resulting lysates were boiled for 30 min, and cellular debris was removed by centrifugation at 15,000g. Equal amounts of lysate from each genotype were diluted 1:10 (v/v) in a lysis buffer without SDS. Lysates were pre-cleared with protein G sepharose beads and endogenous ASNS protein was immunoprecipitated using an ASNS antibody (Proteintech, 14681-1-AP) for 12 h at 4 °C. The final immune complexes were washed three times with a lysis buffer containing 125 mM NaCl and then boiled for 10 min at 95 °C in LDS sample buffer. The resulting samples were visualized using an anti-ubiquitin antibody (1:1,000, Cell Signaling Technology, 3936).

To monitor the ubiquitination of ASNS lysine mutants, HEK293T cells transfected with HA-tagged ASNS mutants and Flag-tagged ubiquitin were treated with 30 μM MG132 (Calbiochem, 474790) for 4 h to inhibit proteasome activity. Cells were then lysed in a lysis buffer containing 25 mM HEPES (pH7.4), 1% Triton X-100, 15 mM KCl, 5 mM EDTA, 1% SDS and EDTA-free protease inhibitor cocktail (Thermo Fisher Scientific, 78437). Lysates were diluted in lysis buffer without SDS and pre-cleared with protein G sepharose beads, and ASNS was immunopurified using an HA antibody (EMD Millipore, 05-904) for 4 h at 4 °C, followed by a further 12-h incubation with protein G Sepharose beads. Ubiquitination was detected using a Flag antibody in western blot analysis.

Immunofluorescence

For brain tissue staining, WT and Angelman mice were anaesthetized and transcardially perfused with ice-cold PBS, followed by 4% paraformaldehyde (PFA), and the brains were dissected. Brains were further fixed overnight at 4 °C in 4% PFA. After fixation, the brains were cryopreserved and sectioned into 20–30-µm-thick slices. Sections were permeabilized with 0.5% Triton X-100 in PBS for 30 min, followed by blocking with a solution containing 5% bovine serum albumin (BSA), 1% normal donkey serum and 0.2% Triton X-100 in PBS for 1 h. Sections were incubated with primary antibodies (rabbit anti-ASNS, 1:800, Proteintech, 14681-1-AP; guinea pig anti-GABA, 1:500, EMD Millipore, AB175; guinea pig anti-NeuN; 1:3,000, EMD Millipore, ABN90; mouse anti-NeuN, 1:3.000, Abcam, ab104224; mouse anti-GFAP, 1:3,000, EMD Millipore, MAB360; goat anti-IBA1, 1:3,000, Novus Biologicals, NB100-1028; goat anti-5-HT, 1:3;000, ImmunoStar, 20079; mouse anti-TH, 1:3,000, ImmunoStar, 22941; sheep anti-neuropeptide Y, 1:1,000, EMD Millipore, AB1583; mouse anti-parvalbumin, 1:1,000, Swant, PV235; goat anti-calretinin, 1:1,000, Swant, CG1; rat anti-somatostatin, 1:1,000, EMD Millipore, MAB354; and mouse CaMKII-α, 1:1,500, Cell Signaling Technology, 50094) overnight for 4 °C. The following day, sections were incubated with secondary antibodies conjugated with Alexa Fluor dyes (488 AffiniPure donkey anti-rabbit, 1:2,000, Jackson ImmunoResearch Laboratories, 711-545-152; 647 AffiniPure donkey anti-guinea pig, 1:1,000, Jackson ImmunoResearch Laboratories, 706-605-148; 647 AffiniPure donkey anti-goat, 1:2,000, Jackson ImmunoResearch Laboratories, 705-605-147; 647 donkey anti-mouse, Thermo Fisher Scientific, A-31571; 647 AffiniPure donkey anti-rat, 1:1,000, Jackson ImmunoResearch Laboratories, 712-605-153; 568 donkey anti-mouse, 1:2,000, Invitrogen, A10037; and 568 donkey anti-sheep, 1:1,000, Thermo Fisher Scientific, A-21099) for 2 h at room temperature with gentle rocking. In the last 10 min of the incubation, nuclei were stained with Hoechst 33342 (Thermo Fisher Scientific, H3570) or DAPI (Sigma-Aldrich, D9542) for 10 min. Afterwards, sections were washed three times with PBST (PBS with 0.1% Tween 20), mounted on slides using Vectashield mounting medium (Vector Laboratories, H-1000) and imaged using an Olympus FV1299 confocal microscope. Fluorescence images were analysed using ImageJ software (v.1.53c).

In vivo and in vitro protein synthesis assay (BONCAT and FUNCAT)

To assess in vivo protein synthesis, we dissolved AHA (Thermo Fisher Scientific, C10102) in PBS. Mice were intraperitoneally injected at a dose of 50 μg AHA per g body weight. After 16 h, brains were collected (excluding the cerebellum) and a hemisphere was processed for BONCAT⁴⁵. Proteins were extracted in RIPA buffer (20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1% Triton X-100, 0.1% sodium dodecyl sulfate, 1 mM EDTA, 1 mM EGTA and 1% sodium deoxy-cholate), supplemented with a protease (Thermo Fisher Scientific, 78437) and phosphatase (Thermo Fisher Scientific, 78428) inhibitor cocktail. To collect AHA-labelled proteins, 200 μg protein was incubated with 100 µM sDIBO-alkyne-conjugated biotin (Thermo Fisher Scientific, C20023) to biotinylate AHA-containing proteins through a copper-free click chemical reaction at room temperature for 2 h. Biotinylated proteins were then purified using streptavidin-conjugated dynabeads (50 μg; Thermo Fisher Scientific, 11205D). Purified proteins were detached from dynabeads by boiling in 1× SDS sample buffer. Eluted samples were processed for western blot analysis and visualized with the Revert 700 Total Protein Stain Kit (Li-Cor Biosciences, 926-11010).

Neurons cultured on coverslips were treated with 1 mM AHA (Click Chemistry Tools, 1066-100) for 4 h. Cells were then fixed in 4% PFA for 15 min at room temperature and processed for FUNCAT³³. Fixed cells were blocked in a 5% BSA solution and incubated with 10 μM Click-iT Alexa Fluor 555 sDIBO alkyne (Invitrogen, C20021), a fluorescent compound that reacts with the azido group of proteins containing AHA, at room temperature for 2 h. Cell nuclei were stained with DAPI (Sigma-Aldrich, D9542). Coverslips were mounted in Vectashield (Vector Laboratories, H-1000) and imaged using an Olympus FV1200 scanning confocal microscope.

LC–MS metabolomics

All snap-frozen hippocampal tissue from WT, Angelman, Het and Rescue mice were weighed at the start of the extraction. Metabolites were extracted from frozen hippocampus samples using an Omni Bead Ruptor Elite Homogenizer with an acetonitrile:methanol:water ratio of 2:2:1 at a volume of 40 μl per mg of tissue. Samples were incubated at −20 °C for 1 h, then centrifuged for 10 min at 20,000g at 4 °C. The supernatant was transferred into a LC–MS vial for analysis on the same day⁴⁶. Ultra-high-performance LC (UHPLC)–MS was performed with a Thermo Fisher Scientific Vanquish Horizon UHPLC system interfaced with a Thermo Fisher Scientific Orbitrap ID-X Tribrid mass spectrometer. Polar metabolites were separated on a HILICON iHILIC-(P)-Classic column (100 × 2.1 mm, 5 μm). The mobile-phase solvents were composed of: A = 20 mM ammonium bicarbonate, 2.5 μM medronic acid and 0.1% ammonium hydroxide in water:acetonitrile 5:95. The column compartment was maintained at 45 °C. The following linear gradient was applied at a flow rate of 0.25 ml min⁻¹: 0 min, 90% B; 12 min, 35% B; 12.5–14.5 min, 25% B; and 15 min, 90% B. The injection volume was 4 μl. Data were acquired in both positive- and negative-ion mode. LC–MS data were processed and analysed with the open-source Skyline software (v.23.1)⁴⁷.

qPCR

Total RNA was isolated using TRIzol reagent (Thermo Fisher Scientific, 15596026) according to the manufacturer’s instructions. cDNA synthesis was performed with a SMARTScribe reverse transcriptase (Takara, 639536). The following primers were used: forward, 5′-GCAGTGTCTGAGTGCGATGAA-3′ and reverse, 5′-TCTTATCGGCTGCATTCCAAAC-3′ for mouse Asns; forward, 5′-GGCTGTATTCCCCTCCATCG-3′ and reverse, 5′-CCAGTTGGTAACAATGCCATGT for mouse β-actin. For qPCR, we used the PowerUp SYBR Green master mix (Applied Biosystems, A25742) and reactions were performed on a QuantStudio 6 Flex Real-Time PCR machine (Applied Biosystems). Calculations of relative expression were performed using the comparative C_t method.

Behavioural testing

Before each test, mice were acclimatized to the testing room and the experimental set-up for several days. All tests were performed during the light period of the light–dark cycle, except for sleep and circadian rhythm experiments. We chose the number of mice used in our study according to published guidelines for Ube3a^m−/p+ mice¹² and aimed to achieve a similar ratio of males and females in each genotypic group. Mice were tested by random distribution of genotypes and data analyses were performed by experimenters who were blinded to genotype.

In this study, a total of 138 mice were used for hindlimb clasping, marble burying and brain weight measurements (at P90 and P150): 33 WT (12 males, 21 females), 26 Angelman (15 males, 11 females), 40 Het (16 males, 24 females) and 39 Rescue (22 males, 17 females). For the open field and accelerating rotarod tests, a total of 71 mice were used: 16 WT (5 males, 11 females), 13 Angelman (5 males, 8 females), 19 Het (12 males, 7 females) and 23 Rescue (12 males, 11 females).

Hindlimb clasping assay

Mice at P35 were manually suspended by the tips of their tails at a distance of 30.5 cm from the surface of the bench. Each mouse underwent two trials with a 5-min interval between trials. The presence of hindlimb clasping behaviour was recorded for 30 s during each trial using a Sony HDR-CX405 camera. Hindlimb clasping behaviour was measured by the movement of the hindlimbs (one or both) retracting and curling inward toward the mouse’s body. Data are presented as the total time spent clasping across the two trials.

Open field

To assess general locomotor activity and approach or avoidance (anxiety-like) behaviours, P56 mice were placed individually in a 40 × 40-cm square chamber illuminated by a 30-lx LED light source. Locomotor activity was measured in the centre of the arena for 1 h, following published protocols⁴⁸. The chamber was divided into specific zones: a central zone occupying the middle 50% of the total chamber area and a perimeter zone encompassing the outer 50%. Movement in the arena was quantified in terms of distance travelled, as well as time spent in and the number of entries into the centre and perimeter zones.

Accelerating rotarod

The accelerating rotarod test was performed at speeds ranging from 4 to 40 rpm over a duration of 300 s. Mice underwent a total of eight trials per day for four consecutive days, with all trials of the first day performed on a stationary rod for habituation. To assess their performance, the average time spent on the rotarod across all trials was calculated.

Marble burying

Polycarbonate cages with open tops were filled with a layer of aspen wood chips, approximately 4 cm deep, provided by the Division of Comparative Medicine at Washington University School of Medicine in St. Louis. On top of the aspen wood chips, a total of 20 black marbles were evenly arranged in a grid format of 5 rows and 4 columns. A mouse at P85 was allowed to interact with these marbles for a period of 45 min. Afterwards, the mouse was transferred back to its home cage and the cage with marbles was photographed. Marbles that were covered by aspen wood chips to an extent greater than 50% were counted as buried.

Measurement of sleep and wake states

The Adapt-A-Base sleep monitoring system for mice (Signal Solutions, model MB-ACN10) is an automated and highly sensitive piezoelectric sleep monitor that uses short-duration (2-s) pressure-signal segments to ascribe a sleep or wake statistic. The non-invasive and high-throughput system detects breathing and gross body movements to analyse unsupervised sleep–wake patterns with 90% accuracy when compared with electroencephalogram (EEG) and electromyogram (EMG)-based methods^49,50. Mice (n = 6 per group) were individually housed in Ancare N10 mouse cages (Ancare) and placed on the Adapt-A-Base systems in sound- and light-proof circadian cabinets (Actimetrics) under 12-h light–12-h dark cycle (light on from 06:00 to 18:00; 300 lx) with ad libitum access to food, water and nesting material. Each cage was furnished with 160 g corn-cob bedding, 220 g food pellets and 380 ml water so that each cage weighed the same except for a small variability between mouse body weight. After 5 days of acclimation, the sleep–wake states were recorded without disturbance over 7 days using PiezoSleep v.2.08 (Signal Solutions). Sleep data were analysed for multiple sleep features of individual mice using SleepStats v.2.18 (Signal Solutions), and 30-s epochs were used for scoring sleep or wake bout lengths.

Measurement of circadian locomotor activity

Mice (n = 7–9 per group) were individually housed in Ancare N10 mouse cages (Ancare) mounted with passive infrared sensor wireless nodes (Actimetrics) and acclimated for 5 days. Daily locomotor activity was recorded at 1-min intervals using the ClockLab Wireless data collection program v.4.129 (Actimetrics) under 12-h light–12-h dark cycle (light on from 06:00 to 18:00; 300 lx) for the first 7 days, followed by constant dark for the next 21 days with ad libitum access to food, water and nesting material. Period length (tau) was determined by χ2 periodogram analysis, rhythm strength was measured by relative fast Fourier transform analysis and other features of circadian locomotor activity were measured using ClockLab Analysis v.6.1.02 (Actimetrics). All routine mouse care during the dark phase was performed under dim red light.

RNA-seq analysis

Cortices and hippocampi were dissected on ice in PBS from WT, Angelman, Het and Rescue littermates at P21 (n = 6 mice per genotype; WT, 3 males, 3 females; Angelman, 4 males, 2 females; Het, 3 males, 3 females; and Rescue, 3 males, 3 females). Tissue was flash-frozen in liquid nitrogen, lysed and the RNA was extracted using the RNeasy Plus Mini Kit (QIAGEN, 74134) following the manufacturer’s instructions. Libraries were generated from 10 ng of RNA using the SMART-Seq Stranded Kit (Takara, 634443) for Illumina and sequenced on an Illumina NextSeq 550 1×75 bp at the Edison Family Center for Genomic Sciences and Systems Biology at Washington University School of Medicine. Our sequencing yielded an average of 8.20 × 10⁶ 75-bp single-end reads per sample. RNA-seq quality control was performed using FastQC⁵¹ (v.0.11.9) and MultiQC⁵² (v.1.9), followed by adapter and quality trimming with Trim Galore (v.0.6.7; Q ≥ 20, minimum length 36 bp). Ribosomal RNA reads were removed by Bowtie2⁵³ (v.2.4.4) alignment to mouse rRNA sequences (NR_003278.3, NR_003279.1 and NR_046144.1), and the remaining reads were aligned to the GRCm39 genome (GCF_000001635.27) using STAR⁵⁴ (v.2.7.11b), allowing up to ten multi-mapping locations per read. Gene-level counts were generated with featureCounts⁵⁵ (v.2.1.1; Subread) using reverse strand specificity (-s 2), counting exon-mapped reads and excluding multi-mapping reads. Differential expression analysis was performed using edgeR⁵⁶ (v.4.0.16) in R after retaining genes with counts per million (CPM) > 1 in fewer than three samples and TMM normalization; generalized linear models included sex (~Sex + Genotype) for tissue-specific analyses with significance assessed by quasi-likelihood F-tests (FDR < 0.05). For exon-level analysis of Ube3a (NM_011668.3) and Asns (NM_012055.5), featureCounts was run in exon mode (-f) with reverse strand specificity (-s 2) and fractional assignment of multi-mapping reads (-M–fraction), and exon counts from RefSeq Select transcripts were normalized by exon length to obtain counts per kilobase.

Spatial transcriptomic analysis

Using the Allen Institute’s whole-mouse-brain single-cell RNA-seq dataset, we calculated the mean and variance of Asns log₂ expression and its corresponding log₂ expression z-score (computed using the scanpy.tl.score_genes function⁵⁷) for each annotated cell type identified⁵⁸. These gene-expression scores were then mapped onto the corresponding cell types in the Allen Institute’s annotated MERFISH spatial transcriptomics dataset⁵⁸. Spatial maps were generated by colouring each cell type according to its Asns expression using the ggplot2 package⁵⁹ in R (v3.4.4).

Single-cell analysis

The single-cell genomics tool Seurat (v.4)⁶⁰ was used for scRNA-seq data analysis. We followed the workflow described in the Seurat guided analysis on datasets normalized with SCTransform to normalize gene-expression levels. Principal component analysis was performed using the top 30 principal components. Uniform manifold approximation and projection (UMAP) analysis was performed with the top 30 dimensions. Clusters were identified with the functions FindNeighbors (dims = 1:30) and FindClusters (resolution = 0.8). A resolution of 0.8 was selected for the downstream analysis because clusters were clearly separated and matched visual inspection. Default parameters were used unless noted. For identified clusters, cluster marker analysis confirmed that they were transcriptionally distinct, with their unique marker genes. Major cell types and five different neuronal cell populations were annotated on the basis of the expression of unique markers as described in the original publication²³, including astrocytes (Gfap, Aldh1l1 and Aqp4), endothelial cells (Ly6c1), oligodendrocytes (Mobp, Mog and Plp1), perivascular macrophages (Mrc1 and Lyve1), mural cells (Acta2), microglia (Tmem119, P2ry12 and Cx3cr1), ependymoma cells (Foxj1) and neuronal subpopulations N-Tbr1, N-Grp, N-Lhx6, N-Synpr and N-Spink8.

Statistical analysis

Statistical analyses were performed using GraphPad Prism (v.10).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

1. Kishino, T., Lalande, M. & Wagstaff, J. UBE3A/E6-AP mutations cause Angelman syndrome. Nat. Genet. 15, 70–73 (1997).

   Article  CAS  PubMed  Google Scholar 

2. Matsuura, T. et al. De novo truncating mutations in E6-AP ubiquitin-protein ligase gene (UBE3A) in Angelman syndrome. Nat. Genet. 15, 74–77 (1997).

   Article  CAS  PubMed  Google Scholar 

3. Krzeski, J. C., Judson, M. C. & Philpot, B. D. Neuronal UBE3A substrates hold therapeutic potential for Angelman syndrome. Curr. Opin. Neurobiol. 88, 102899 (2024).

   Article  CAS  PubMed  PubMed Central  Google Scholar 

4. Weston, K. P. et al. Identification of disease-linked hyperactivating mutations in UBE3A through large-scale functional variant analysis. Nat. Commun. 12, 6809 (2021).

   Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

5. Yi, J. J. et al. An autism-linked mutation disables phosphorylation control of UBE3A. Cell 162, 795–807 (2015).

   Article  CAS  PubMed  PubMed Central  Google Scholar 

6. Pandya, N. J. et al. Secreted retrovirus-like GAG-domain-containing protein PEG10 is regulated by UBE3A and is involved in Angelman syndrome pathophysiology. Cell Rep. Med. 2, 100360 (2021).

   Article  CAS  PubMed  PubMed Central  Google Scholar 

7. Noor, A. et al. 15q11.2 duplication encompassing only the UBE3A gene is associated with developmental delay and neuropsychiatric phenotypes. Hum. Mutat. 36, 689–693 (2015).

   Article  CAS  PubMed  Google Scholar 

8. Salminen, I., Read, S., Hurd, P. & Crespi, B. Genetic variation of UBE3A is associated with schizotypy in a population of typical individuals. Psychiatry Res. 275, 94–99 (2019).

   Article  CAS  PubMed  Google Scholar 

9. Scheffner, M., Huibregtse, J. M., Vierstra, R. D. & Howley, P. M. The HPV-16 E6 and E6-AP complex functions as a ubiquitin-protein ligase in the ubiquitination of p53. Cell 75, 495–505 (1993).

   Article  CAS  PubMed  Google Scholar 

10. Wolyniec, K. et al. E6AP ubiquitin ligase regulates PML-induced senescence in Myc-driven lymphomagenesis. Blood 120, 822–832 (2012).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

11. Albrecht, U. et al. Imprinted expression of the murine Angelman syndrome gene, Ube3a, in hippocampal and Purkinje neurons. Nat. Genet. 17, 75–78 (1997).

    Article  CAS  PubMed  Google Scholar 

12. Sonzogni, M. et al. A behavioral test battery for mouse models of Angelman syndrome: a powerful tool for testing drugs and novel Ube3a mutants. Mol. Autism 9, 47 (2018).

    Article  PubMed  PubMed Central  Google Scholar 

13. Tan, W. H. et al. Angelman syndrome: mutations influence features in early childhood. Am. J. Med. Genet. A 155A, 81–90 (2011).

    Article  PubMed  PubMed Central  Google Scholar 

14. Judson, M. C. et al. Decreased axon caliber underlies loss of fiber tract integrity, disproportional reductions in white matter volume, and microcephaly in Angelman syndrome model mice. J. Neurosci. 37, 7347–7361 (2017).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

15. Miao, S. et al. The Angelman syndrome protein Ube3a is required for polarized dendrite morphogenesis in pyramidal neurons. J. Neurosci. 33, 327–333 (2013).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

16. Costa-Mattioli, M. & Walter, P. The integrated stress response: from mechanism to disease. Science 368, eaat5314 (2020).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

17. Gong, S. S., Guerrini, L. & Basilico, C. Regulation of asparagine synthetase gene expression by amino acid starvation. Mol. Cell. Biol. 11, 6059–6066 (1991).

    CAS  PubMed  PubMed Central  Google Scholar 

18. Yi, J. J. et al. The autism-linked UBE3A T485A mutant E3 ubiquitin ligase activates the Wnt/β-catenin pathway by inhibiting the proteasome. J. Biol. Chem. 292, 12503–12515 (2017).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

19. Shimoji, T. et al. Identification of annexin A1 as a novel substrate for E6AP-mediated ubiquitylation. J. Cell. Biochem. 106, 1123–1135 (2009).

    Article  CAS  PubMed  Google Scholar 

20. Nasu, J. et al. E6AP ubiquitin ligase mediates ubiquitin-dependent degradation of peroxiredoxin 1. J. Cell. Biochem. 111, 676–685 (2010).

    Article  CAS  PubMed  Google Scholar 

21. Balasubramanian, M. N., Butterworth, E. A. & Kilberg, M. S. Asparagine synthetase: regulation by cell stress and involvement in tumor biology. Am. J. Physiol. Endocrinol. Metab. 304, E789–799 (2013).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

22. Yao, Z. et al. A taxonomy of transcriptomic cell types across the isocortex and hippocampal formation. Cell 184, 3222–3241 (2021).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

23. Zeisel, A. et al. Cell types in the mouse cortex and hippocampus revealed by single-cell RNA-seq. Science 347, 1138–1142 (2015).

    Article  ADS  CAS  PubMed  Google Scholar 

24. Ruzzo, E. K. et al. Deficiency of asparagine synthetase causes congenital microcephaly and a progressive form of encephalopathy. Neuron 80, 429–441 (2013).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

25. Wolter, J. M. et al. Cas9 gene therapy for Angelman syndrome traps Ube3a-ATS long non-coding RNA. Nature 587, 281–284 (2020).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

26. Ehlen, J. C. et al. Maternal Ube3a loss disrupts sleep homeostasis but leaves circadian rhythmicity largely intact. J. Neurosci. 35, 13587–13598 (2015).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

27. Harding, H. P. et al. Regulated translation initiation controls stress-induced gene expression in mammalian cells. Mol. Cell 6, 1099–1108 (2000).

    Article  CAS  PubMed  Google Scholar 

28. Tarlungeanu, D. C. et al. Impaired amino acid transport at the blood brain barrier is a cause of autism spectrum disorder. Cell 167, 1481–1494 (2016).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

29. Levin, D., Ranu, R. S., Ernst, V. & London, I. M. Regulation of protein synthesis in reticulocyte lysates: phosphorylation of methionyl-tRNA_f binding factor by protein kinase activity of translational inhibitor isolated from hemedeficient lysates. Proc. Natl Acad. Sci. USA 73, 3112–3116 (1976).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

30. Chakrabarty, Y., Yang, Z., Chen, H. & Chan, D. C. The HRI branch of the integrated stress response selectively triggers mitophagy. Mol. Cell 84, 1090–1100 (2024).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

31. Alvarez-Castelao, B. et al. The switch-like expression of heme-regulated kinase 1 mediates neuronal proteostasis following proteasome inhibition. eLife 9, e52714 (2020).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

32. Dieterich, D. C., Link, A. J., Graumann, J., Tirrell, D. A. & Schuman, E. M. Selective identification of newly synthesized proteins in mammalian cells using bioorthogonal noncanonical amino acid tagging (BONCAT). Proc. Natl Acad. Sci. USA 103, 9482–9487 (2006).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

33. Dieterich, D. C. et al. In situ visualization and dynamics of newly synthesized proteins in rat hippocampal neurons. Nat. Neurosci. 13, 897–905 (2010).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

34. Gunelson, A. M. et al. Autism and intellectual disability due to a novel gain-of-function mutation in UBE3A J. Hum. Genet. 70, 439–442 (2025).

    Article  PubMed  PubMed Central  Google Scholar 

35. Pacitti, A. J., Inoue, Y. & Souba, W. W. Characterization of Na⁺-independent glutamine transport in rat liver. Am. J. Physiol. 265, G90–98 (1993).

    CAS  PubMed  Google Scholar 

36. Lee, W. J., Hawkins, R. A., Vina, J. R. & Peterson, D. R. Glutamine transport by the blood-brain barrier: a possible mechanism for nitrogen removal. Am. J. Physiol. 274, C1101–1107 (1998).

    Article  CAS  PubMed  Google Scholar 

37. Alfadhel, M. & El-Hattab, A. W. Asparagine synthase deficiency. In GeneReviews (eds Adam, M. P. et al.) (University of Washington, 2025).

38. Avagliano Trezza, R. et al. Loss of nuclear UBE3A causes electrophysiological and behavioral deficits in mice and is associated with Angelman syndrome. Nat. Neurosci. 22, 1235–1247 (2019).

    Article  CAS  PubMed  Google Scholar 

39. Martinez-Noel, G. et al. Identification and proteomic analysis of distinct UBE3A/E6AP protein complexes. Mol. Cell. Biol. 32, 3095–3106 (2012).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

40. Yasuda, S. et al. Stress- and ubiquitylation-dependent phase separation of the proteasome. Nature 578, 296–300 (2020).

    Article  ADS  CAS  PubMed  Google Scholar 

41. Biever, A. et al. Monosomes actively translate synaptic mRNAs in neuronal processes. Science 367, eaay4991 (2020).

    Article  CAS  PubMed  Google Scholar 

42. Hooshmandi, M. et al. Excitatory neuron-specific suppression of the integrated stress response contributes to autism-related phenotypes in fragile X syndrome. Neuron 111, 3028–3040 (2023).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

43. Borck, G. et al. eIF2γ mutation that disrupts eIF2 complex integrity links intellectual disability to impaired translation initiation. Mol. Cell 48, 641–646 (2012).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

44. Jiang, Y. H. et al. Mutation of the Angelman ubiquitin ligase in mice causes increased cytoplasmic p53 and deficits of contextual learning and long-term potentiation. Neuron 21, 799–811 (1998).

    Article  CAS  PubMed  Google Scholar 

45. Dieterich, D. C. et al. Labeling, detection and identification of newly synthesized proteomes with bioorthogonal non-canonical amino-acid tagging. Nat. Protoc. 2, 532–540 (2007).

    Article  PubMed  Google Scholar 

46. Naser, F. J. et al. Isotope tracing in adult zebrafish reveals alanine cycling between melanoma and liver. Cell Metab. 33, 1493–1504 (2021).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

47. Adams, K. J. et al. Skyline for small molecules: a unifying software package for quantitative metabolomics. J. Proteome Res. 19, 1447–1458 (2020).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

48. Chen, J. et al. A MYT1L syndrome mouse model recapitulates patient phenotypes and reveals altered brain development due to disrupted neuronal maturation. Neuron 109, 3775–3792 (2021).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

49. Donohue, K. D., Medonza, D. C., Crane, E. R. & O’Hara, B. F. Assessment of a non-invasive high-throughput classifier for behaviours associated with sleep and wake in mice. Biomed. Eng. Online 7, 14 (2008).

    Article  PubMed  PubMed Central  Google Scholar 

50. Mang, G. M. et al. Evaluation of a piezoelectric system as an alternative to electroencephalogram/ electromyogram recordings in mouse sleep studies. Sleep 37, 1383–1392 (2014).

    Article  PubMed  PubMed Central  Google Scholar 

51. Wingett, S. W. & Andrews, S. FastQ Screen: a tool for multi-genome mapping and quality control. F1000Res 7, 1338 (2018).

    Article  PubMed  PubMed Central  Google Scholar 

52. Ewels, P., Magnusson, M., Lundin, S. & Kaller, M. MultiQC: summarize analysis results for multiple tools and samples in a single report. Bioinformatics 32, 3047–3048 (2016).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

53. Langmead, B. & Salzberg, S. L. Fast gapped-read alignment with Bowtie 2. Nat. Methods 9, 357–359 (2012).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

54. Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21 (2013).

    Article  CAS  PubMed  Google Scholar 

55. Liao, Y., Smyth, G. K. & Shi, W. featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics 30, 923–930 (2014).

    Article  CAS  PubMed  Google Scholar 

56. Robinson, M. D., McCarthy, D. J. & Smyth, G. K. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26, 139–140 (2010).

    Article  CAS  PubMed  Google Scholar 

57. Wolf, F. A., Angerer, P. & Theis, F. J. SCANPY: large-scale single-cell gene expression data analysis. Genome Biol. 19, 15 (2018).

    Article  PubMed  PubMed Central  Google Scholar 

58. Yao, Z. et al. A high-resolution transcriptomic and spatial atlas of cell types in the whole mouse brain. Nature 624, 317–332 (2023).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

59. Wickham, H. ggplot2: Elegant Graphics for Data Analysis (Springer, 2016).

60. Hao, Y. et al. Integrated analysis of multimodal single-cell data. Cell 184, 3573–3587 (2021).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

61. Kim, K.-S. Automated bioinformatics pipeline for mouse brain RNA-seq: alignment filtering, exon-level quantification, and sex-adjusted differential expression analysis. Zenodo https://doi.org/10.5281/zenodo.22259707 (2026).

Download references

Author notes

1. Deceased: Guoyan Zhao

Authors and Affiliations

1. Department of Neuroscience, Washington University School of Medicine, St. Louis, MO, USA

   Kwang-Soo Kim, Alissa Cullen, Jalin A. Stelzer, Mati T. Nemera, Travis Law, Harrison W. Gabel & Jason J. Yi

2. Department of Computer Science, Saint Louis University, St. Louis, MO, USA

   Cory Gardner & Tae-Hyuk Ahn

3. Department of Chemistry, Washington University in St. Louis, St. Louis, MO, USA

   Kevin Cho & Gary J. Patti

4. Department of Neurology, Washington University School of Medicine, St. Louis, MO, USA

   Ashish Sharma & Erik S. Musiek

5. Department of Genetics, Washington University School of Medicine, St. Louis, MO, USA

   Huma Naz & Guoyan Zhao

Contributions

J.J.Y. and K.-S.K conceived this project. A.C., C.G. and T.-H.A. performed transcriptomic analysis. M.T.N., T.L. and H.W.G. performed spatial transcriptomic analysis. H.N. and G.Z. analysed single-cell transcriptomic data. J.A.S. assisted with animal behaviour analysis. K.C. and G.J.P. performed metabolomic analysis. A.S. and E.S.M. performed the sleep studies. J.J.Y. and K.-S.K. wrote the manuscript with input from the other co-authors. J.J.Y. supervised the work.

Corresponding author

Correspondence to Jason J. Yi.

Competing interests

The authors declare no competing interests.

Peer review information

Nature thanks Lawrence Reiter and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions