Impaired Activity-Regulated Proteostasis: A Shared Mechanism in Neurodevelopmental Disorders

mRNA Translation

Protein synthesis, the first key node of the neuronal proteostasis network, proceeds through 3 stages: initiation, elongation, and termination. Initiation is rate limiting and controlled by mTORC1 (mechanistic target of rapamycin complex 1), which promotes cap-dependent translation (25). During elongation, eEF2 (eukaryotic elongation factor 2) kinase phosphorylates eEF2, providing a dynamic brake on protein production that can be rapidly relieved by neuronal activity (26). A parallel regulatory axis is the integrated stress response (ISR), in which stress-activated kinases phosphorylate eIF2α (eukaryotic translation initiation factor 2 subunit 1), thereby suppressing global translation while selectively permitting translation of stress-response mRNAs (27). Neurons have repurposed this stress mechanism for plasticity. BDNF (brain-derived neurotrophic factor) and neuronal activity upregulates a negative ISR regulator to promote eIF2α dephosphorylation and activity-induced protein synthesis (28). In addition to soma, thousands of mRNAs are trafficked into dendrites and translated locally at synapses (29,30), where FMRP (fragile X messenger ribonucleoprotein) acts as a key brake on translation by suppressing target mRNA expression (31). To ensure the fidelity of this spatially distributed synthesis, RQC detects stalled ribosomes and eliminates aberrant nascent proteins before they can misfold.

Ribosome Quality Control

The RQC detects and resolves ribosome stalling at damaged mRNAs or rare codons, targeting the aberrant nascent polypeptide for degradation (32). The process is coordinated sequentially as ZNF598 recognizes colliding ribosomes and ubiquitinates ribosomal proteins, PELO/HBS1L dissociates the stalled complex, and the E3 ligase listerin ubiquitinates the nascent peptide on the 60S subunit, targeting it for proteasomal degradation (33,34). Furthermore, the conserved quality control factor NEMF adds CAT-tail extensions that expose lysine residues for ubiquitination, but when RQC is overwhelmed, these extensions instead drive protein aggregation and cause neurological diseases in both mice and humans. Because RQC substrates are ultimately degraded by the proteasome, RQC directly links translation fidelity to UPS function.

Ubiquitin-Proteasome System

The UPS operates through a cascade of 3 enzymes (E1, E2, and E3) that tag target proteins with polyubiquitin chains for degradation by the 26S proteasome (35). Deubiquitinating enzymes add another layer of regulation and reversibility to the system. In neurons, CaMKII (calcium/calmodulin-dependent protein kinase II) phosphorylates proteasome subunit in response to calcium influx, directly stimulating proteasome activity at synapses (36). Additionally, neurons have a specialized membrane proteasome (NMP) that degrades proteins co-translationally, acting as a real-time quality control mechanism that couples synthesis and degradation with temporal precision (37).

Autophagy-Lysosomal Pathway

The ALP is the second major degradation arm of the proteostasis network (38). Macroautophagy, the most studied form, proceeds through 3 coordinated steps. First, a double-membrane phagophore forms around cytoplasmic cargo, elongates, and seals to form the autophagosome (39). Second, autophagosomes fuse with lysosomes to form autolysosomes (40). Third, lysosomal hydrolases degrade the cargo in the acidic autolysosomal lumen. TFEB (transcription factor EB) acts as a transcriptional regulator of the ALP and is suppressed by mTORC1 under nutrient-rich conditions, linking the ALP to the cell’s metabolic state (39). In neurons, autophagosomes form distally in axons and are retrogradely transported to lysosomes in the soma, while dendritic autophagosomes are stationary and regulate local proteostasis (41,42). Neurons also use chaperone-mediated autophagy and microautophagy for selective degradation of proteins (43); however, their roles in NDDs remain largely unexplored.

Dysregulated mRNA Translation

Translational dysregulation in NDDs is bidirectional. Some disorders are characterized by a pathological excess of protein synthesis, while others show translational arrest. Understanding these mechanisms is important because therapeutic strategies that are beneficial in one context may be harmful in the other.

Translational excess is a hallmark of fragile X syndrome (FXS), tuberous sclerosis complex (TSC), and CUL3-related NDD, despite their distinct genetic origins. In FXS, loss of FMRP removes the translational brake on hundreds of dendritic mRNAs, leading to excessive protein synthesis (62) and exaggerated mGluR-LTD (23). When FXS neurons are activated, they fail to sense the already elevated basal protein levels and further increase synthesis (23,63). Similarly, TSC1 or TSC2, which encode the hamartin-tuberin complex, normally limits mTORC1 activation. Loss of TSC1 or TSC2 causes constitutive mTORC1 hyperactivation, bypassing the normal requirement for growth factor or synaptic activity to activate the translational machinery (64). CUL3-related NDD converges on the same outcome but through a distinct mechanism. CUL3 ubiquitinates and degrades eIF4G1, a scaffold protein of the eIF4F complex that drives cap-dependent translation (65). Loss-of-function mutations in CUL3 cause eIF4G1 to accumulate, promoting excess protein synthesis and producing social deficits, anxiety, and cognitive impairment (65–67). Although CUL3 is an E3 ligase, its primary proteostatic failure is translational excess rather than degradation failure, and it converges with FXS and TSC on elevated basal synthesis that overwhelms the degradation arm.

Translational arrest defines a different set of NDDs. In Down syndrome, hyperactivation of the ISR through increased phosphorylation of eIF2α suppresses global cap-dependent translation while selectively allowing translation of specific stress-response mRNAs (68). In Rett syndrome, loss of MECP2 (methyl-CpG binding protein 2) reduces mTOR signaling and BDNF expression, producing a translational deficit opposite in direction to TSC (69). Despite differences in their dysregulated proteostasis nodes, Down syndrome and FXS share overlapping symptoms and pathologies (70). Similarly, mutations in EEF1A2 disrupt translational elongation, causing early epileptic seizures and ID (71). In Dravet syndrome, which is caused by loss-of-function mutations in the SCN1A gene, hyperactivation of eEF2K disrupts translation (72). Mutations in ELP1 (elongator acetyltransferase complex subunit 1) impair transfer RNA (tRNA) modification by the elongator complex, disrupting protein synthesis and causing familial dysautonomia (73,74).

RQC dysfunction has also been linked to neurological diseases in both mice and humans. Loss of Pelo and Hbs1l impairs mouse embryonic and cerebellar development (75). Mutations in LTN1, which encodes listerin protein, cause early-onset progressive motor and sensory neurodegeneration in mice (76), and NEMF mutations cause juvenile-onset neuromuscular disease in humans (77). When RQC is overwhelmed, stalled polypeptides acquire CAT-tail extensions. Failure to degrade these aberrant CAT-tailed proteins disrupts neuronal morphogenesis and triggers apoptosis (78).

Overall, these findings suggest that the proteostasis network can be destabilized by either excess or insufficiency at the same node. In each case, it is likely that the failure activates the translation node that drives pathology.

UPS Impairments

Although UPS dysfunction in NDDs arises through diverse mechanisms, including loss of substrate-specific E3 ligases, disruption of ubiquitin chain assembly scaffolds, and impaired proteasome activation, all converge on a shared outcome of abnormal protein accumulation.

Angelman syndrome is caused by loss of E3 ligase UBE3A, leading to constitutive ARC accumulation, disrupted AMPAR trafficking, impaired synaptic plasticity, and ID (60,79). ARC persists abnormally in dendrites and accumulates at the nucleus of Angelman syndrome neurons (80). Because ARC turnover is normally driven by neuronal activity, its accumulation reflects a dual failure, overaccumulation at baseline and an inability to engage activity-regulated proteolysis in response to stimulation.

UPS dysfunction in NDDs can also arise indirectly, when aberrant upstream regulators overwhelm degradation capacity rather than disrupting the UPS machinery itself. In Rett syndrome, MECP2 mutations alter the association of RNA polymerase II with promoters of immediate-early genes such as ARC, impairing their activity-regulated transcription (81), contributing to synaptic dysfunction (82). In FXS, absence of FMRP leads to excess production of ARC protein at synapses, overwhelming the local UPS capacity and contributing to the synaptic pathology (62,83). In both cases, the UPS is not intrinsically defective but rather is rendered insufficient by the pathological substrate load imposed on it, following increased neuronal activity.

ALP Dysfunction

ALP dysfunction is a recurring feature across NDDs. In TSC and in phosphatase and tensin homolog (PTEN) hamartoma syndromes, constitutive mTORC1 hyperactivation suppresses TFEB and autophagy initiation, causing accumulation of undegraded proteins and ASD-like behaviors (64). In contrast, our own work with Angelman syndrome mice showed that hippocampal neurons have impaired ALP and accumulation of immediate-early gene products, and these neurons fail to register any further increases following learning (24).

In Rett syndrome, in addition to reduced protein synthesis, MECP2-deficient neurons also show reduced autophagosome formation, marked by reduced MAP1LC3B lipidation, and impaired autophagic flux that can be partially rescued by autophagy-enhancing compounds (84). Whether Rett neurons specifically fail to upregulate autophagy in response to stimulation remains to be established.

BPAN (beta-propeller protein-associated neurodegeneration) provides a particularly interesting example of how ALP dysfunction bridges NDDs and neurodegeneration. Caused by loss-of-function mutations in WDR45 (WD repeat domain 45), which encodes WIPI4, a scaffolding protein required for autophagosome biogenesis, BPAN impairs autophagic flux and causes accumulation of undegraded proteins in neurons (85). The resulting developmental delay and early-onset epilepsy overlaps with many other NDDs. BPAN also leads to progressive Parkinsonism-dystonia and dementia in early adulthood (85), which represent different temporal manifestations of the same underlying proteostatic vulnerability.

NDDs and the Neurodegeneration Proteostasis Continuum

Several gene products implicated in neurodegeneration are now recognized to also cause NDDs, suggesting greater pathological overlap between these diseases than was previously recognized. VCP (valosin containing protein), an AAA+ ATPase required for protein degradation (86), demonstrates this overlap. VCP mutations cause mislocalization of TDP-43 and are a well-established risk factor for amyotrophic lateral sclerosis and frontotemporal dementia, but the same mutations have also been linked to an NDD characterized by motor deficits, ID, and epilepsy (86–88). The same mutation can produce either an NDD or a neurodegenerative phenotype depending on context and timing, suggesting a shared proteostatic vulnerability continuum.

One way to predict whether proteostatic failure will appear early or late in life is through the concept of proteostatic reserve. Proteostatic reserve refers to the difference between a neuron’s ability to handle protein turnover under basal conditions and following activity. The severity of the mutation determines timing. Severe mutations exhaust proteostatic reserve during development, when neuronal activity and protein turnover demands peak, producing early symptoms. Milder mutations allow neurons to cope initially, but aging gradually depletes the remaining reserve, culminating in late-onset neurodegeneration. Consistent with this view, several late-onset neurodegenerative diseases show early synaptic and translational abnormalities years before neuronal death. Therefore, interventions that enhance proteostatic response may be beneficial across both NDDs and neurodegenerative diseases. However, the timing also matters. The brain’s tolerance for proteostatic perturbation differs markedly between development and adulthood.

Strategies That Reduce Translation Excess

In disorders of translational excess, the mTORC1-TFEB axis is the most clinically advanced target. Rapamycin analogs such as everolimus are U.S. Food and Drug Administration approved or in trials for TSC (89,90), targeting both arms of the proteostatic imbalance by simultaneously reducing translation and enhancing autophagy. Metformin, an AMPK activator that inhibits mTORC1 through a separate pathway, also enhances autophagic clearance and has shown benefit in FXS and ASD mouse models and in children with ASD (91–95).

ISRIB (integrated stress response inhibitor) works by stabilizing the eIF2B complex, reducing its sensitivity to the inhibitory effect of phosphorylated eIF2α, and allowing normal translation to resume without switching off the stress response (96). Thus, ISRIB is effective across disorders with opposite translational dysregulation, including Down syndrome and FXS (97,98). In both cases, it recalibrates ISR without stimulating global synthesis and restores translational control (96,97).

In FXS, where translational excess is driven by loss of FMRP rather than mTOR hyperactivation, the therapeutic target shifts to the eIF4E-eIF4G axis. 4EGI-1, which blocks this interaction to reduce cap-dependent translation, rescues hippocampus-dependent memory and spine morphology in FXS mice (99) and rescues social deficits and synaptic phenotypes in CUL3-deficient mice (67). Together these findings establish eIF4E-eIF4G inhibition as a shared therapeutic strategy across NDDs of translational excess.

Enhancing degradation is a complementary approach. TAT-beclin 1, a peptide activator of autophagy, rescues synaptic protein overaccumulation and memory deficits in Angelman syndrome mice (24), demonstrating that boosting clearance can compensate even when the primary defect lies upstream in translation.

Trehalose, a natural disaccharide, stimulates autophagy by activating TFEB dephosphorylation and nuclear translocation, upregulating lysosomal hydrolases and autophagy components (100). Oral trehalose supplementation also reduced intranuclear inclusions and improved motor function in an inducible mouse model of fragile X-associated tremor/ataxia syndrome, a late-onset progressive neurodegenerative disorder caused by CGG repeat expansion in the FMR1 gene (100). However, once aggregates are established, they cannot be cleared by trehalose, possibly because clearance requires not only autophagy induction but also upregulation of nuclear export machinery, which TFEB activation alone may not achieve.

Strategies That Relieve Translational Arrest or Degradation Failure

In NDDs with degradation failure, the strategy is to restore UPS or ALP function. For Angelman syndrome, 2 complementary strategies are being pursued. UBE3A reinstatement by unsilencing paternal UBE3A allele using antisense oligo-nucleotides (ASOs), which showed promise in mouse models and early clinical trials (101,102), does not reverse the proteostatic dysregulation. In contrast, our findings showed that TAT-beclin 1 rescues both memory deficits and proteostatic imbalance in Angelman syndrome mice by reinstating activity-regulated autophagic flux (24).

Rett syndrome has also shown a promising response to autophagy enhancer (84). MECP2-deficient neurons show autophagic imbalance and synaptic dysfunction that can be partially rescued by trehalose (84). These findings suggest that restoring degradation can improve synaptic function even when the primary defect is transcriptional.

Strategies That Improve Overall Protein Quality

Chemical chaperones stabilize misfolded proteins in the endoplasmic reticulum (ER) and act independently of the direction of proteostatic failure. 4-phenylbutyrate (4-PBA) reduces ER stress through 2 complementary mechanisms: by stabilizing misfolded proteins in the ER lumen to reduce aggregation and facilitate correct folding and by enhancing HSP70 and HSP90 chaperone activity through increased acetylation (103). Although the precise mechanism remains debated, the beneficial effects of 4-PBA generalize across NDDs with ER trafficking defects, including SLC6A1-related epilepsy (103), a Dravet syndrome mouse model carrying a GABRG2 mutation (104), STXBP1 encephalopathy (105), and Down syndrome (106). Notably, 4-PBA is most effective when misfolding is partial rather than complete, explaining its efficacy in NDDs where pathology is driven by trafficking failure rather than complete loss of protein function.

No drugs targeting RQC have been reported in clinical use. However, because RQC substrates are degraded by proteasome, proteasome-enhancing strategies and approaches that reduce ribosome stalling such as tRNA supplementation or reduction of oxidative damage to mRNA can indirectly restore RQC.

The Challenge of Specificity and Developmental Timing

A recurring challenge is that available compounds act broadly rather than in a cell type– or circuit-specific manner, raising concern that they could disrupt global protein turnover and impair the plasticity mechanisms they aim to restore. This risk is partially offset by the nature of the compounds themselves. These drugs recalibrate rather than block their target pathways, restoring normal signaling range without eliminating it, thereby reducing the risk of disrupting global proteostasis. Rapamycin derivatives and 4-PBA have established pediatric safety profiles, accelerating clinical translation.

Targeted delivery offers a more precise solution. Lipid nanoparticles enable brainwide delivery with cell-type specificity (107), adeno-associated viral vectors with cell type–specific promoters such as DLX2.0 allow interneuron-selective delivery (108), and intrathecal ASO delivery is already in clinical use (109). The strategy going forward is to use broadly acting proteostasis modulators delivered to specific cell types or developmental windows.

The timing of intervention is equally critical. Restoring activity-regulated proteostasis may be especially effective during early postnatal periods when circuits remain plastic, but broadly suppressing protein turnover during this window carries developmental risk.