A rare neurodevelopmental disorder linked to loss of one working copy of the WDFY3 gene is beginning to reveal how disrupted cellular housekeeping can alter the way neurons grow. In a new study published in Molecular Psychiatry on 27 September 2026, researchers combined clinical data from 32 people with laboratory experiments in human neuronal cells and mouse cortical neurons to trace the consequences of WDFY3 loss and test whether lithium chloride could reverse some of them.
The results connect several levels of biology. People carrying protein-truncating WDFY3 variants commonly had neurodevelopmental and neuropsychiatric features. In cells, reducing WDFY3 disrupted selective autophagy, altered developmental signalling, increased proliferation and reduced neurite growth. When researchers exposed WDFY3-deficient neurons to lithium chloride, WDFY3 expression increased and dendritic branching improved markedly.
That last finding is intriguing because lithium is already used clinically for psychiatric conditions. It is not, however, evidence that lithium treats WDFY3-associated neurodevelopmental disorder. The intervention was tested mainly in cultured cells, at experimental concentrations and under tightly controlled laboratory conditions. The study therefore offers a mechanistic lead and a possible therapeutic direction, not a treatment recommendation.
A larger picture of a rare genetic disorder
WDFY3 encodes a scaffold protein involved in selective autophagy, the process cells use to recognise and clear particular unwanted proteins and other material. The protein is also involved in brain development, neuronal progenitor behaviour and neuronal migration. Previous research had linked damaging WDFY3 variants to developmental difficulties and altered brain size, but the number of described patients remained small.
The new study assembled 32 people with WDFY3 loss-of-function variants, including nine previously reported cases and 23 additional cases. Participants ranged in age from 2 to 49 years. Twenty-three were male and nine were female. The variants were distributed across the protein and were classified as pathogenic under the researchers’ stated ACMG criteria.
Neurobehavioural diagnoses were common. Autism spectrum disorder was reported in 23 of the 32 participants, or 72%. Nine of 32, or 28%, showed signs of attention deficit hyperactivity disorder, while two, or 6%, had obsessive-compulsive disorder. The broader clinical picture included mild to moderate neurodevelopmental delay and variable head and brain growth, most commonly increased head size. One participant instead had reduced head circumference in the context of broader growth restriction.
These clinical observations refine the phenotype associated with WDFY3 loss, but they do not by themselves explain how a disrupted gene produces the neurological features. The researchers therefore moved from patients to experimental models.
Reducing WDFY3 changed how neuronal cells handled proteins
The team reduced WDFY3 expression in SH-SY5Y human neuroblastoma cells, a widely used laboratory model for neuronal biology. They then challenged the cells with an aggregation-prone fragment of huntingtin protein that is normally cleared through selective autophagy.
Cells with reduced WDFY3 accumulated more of the aggregation-prone protein. The researchers also observed lower levels of p62, a ubiquitin-binding protein involved in directing material toward autophagic degradation. In western blot experiments, the aggregation-prone protein increased by about 1.20-fold after WDFY3 knockdown. The reduction in p62 showed a trend but did not reach conventional statistical significance, with a reported p value of 0.095.
The distinction matters. The study provides several converging indications that selective autophagy was impaired, but not every individual molecular measure crossed a significance threshold. The broader interpretation rests on the pattern across imaging, protein measurements and downstream cellular behaviour rather than on one isolated result.
The gene knockdown shifted developmental signalling
RNA sequencing offered a wider view. WDFY3 knockdown reduced WDFY3 expression with a log2 fold change of -1.85 and an adjusted p value below 0.001. The analysis identified 130 genes whose expression differed significantly between WDFY3-deficient and control cells.
The affected networks included processes related to nervous-system development, axon formation and cell growth, along with WNT/β-catenin and MAPK/ERK1/2 signalling. These pathways help regulate when neural progenitors divide, differentiate and extend the cellular processes required to form networks.
Follow-up measurements supported the transcriptomic signal. WDFY3 expression fell to 0.271 times the control level in one quantitative PCR experiment, with p = 0.002. β-catenin protein fell to 0.794 times control levels, p = 0.023. GSK3B expression increased to 1.182 times control levels, p = 0.008, while total ERK1/2 protein rose to 1.67 times control levels five days after transfection, p = 0.023.
The timing was not uniform. Some signalling differences appeared early and others later, suggesting that WDFY3 loss triggers a changing cellular response rather than a single fixed molecular defect. The authors also note possible compensatory responses within these interconnected pathways.
More cell division, less neuronal maturation
The functional consequences were visible in the cells themselves. Four days after WDFY3 knockdown, cell counts were 1.438 times higher than in controls, p = 0.01. The proportion of cells positive for the proliferation marker Ki-67 was 1.129 times higher, p = 0.011. Expression of another proliferation marker, PCNA, was 1.17 times higher, although that comparison fell short of statistical significance at p = 0.065.
At the same time, the cells became worse at taking on a mature neuronal form. After five days of differentiation, neurite-like protrusions in WDFY3 knockdown cells were 0.736 times the length seen in controls, corresponding to an average reduction of 30.9 micrometres. The difference was statistically significant at p = 0.005.
This combination is biologically important. Brain development depends not only on producing enough cells but also on getting those cells to exit proliferative states, differentiate appropriately and build branching connections. A shift toward proliferation alongside weaker outgrowth offers a plausible cellular route through which WDFY3 disruption could influence brain growth and connectivity.
Lithium increased WDFY3 expression
The researchers next asked whether they could push the altered signalling system in the opposite direction. Lithium chloride can inhibit GSK-3β and influence canonical WNT/β-catenin signalling, making it a logical experimental probe.
In control SH-SY5Y cells, WDFY3 expression was already 2.672 times higher after neuronal differentiation than in non-differentiated cells, p = 0.018. That observation supported the idea that WDFY3 participates in neuronal maturation.
When non-differentiated cells were treated with 5 millimolar lithium chloride for 24 hours, WDFY3 expression increased 1.448-fold, p = 0.045. AXIN2 increased 1.213-fold, p = 0.024, and GSK3B increased 1.512-fold, p = 0.022. A shorter 15-minute exposure also increased inhibitory phosphorylation of GSK-3β, consistent with lithium engaging the expected molecular target.
The researchers then repeated the intervention in primary cortical neurons from mice carrying one disrupted copy of Wdfy3. After eight days of 5 millimolar lithium chloride, Wdfy3 expression increased 9.027-fold in wild-type neurons and 100.515-fold in heterozygous neurons. The latter estimate was highly variable, with a reported spread of ±182.895-fold, so its large numerical magnitude should be interpreted cautiously rather than as a precise estimate of effect size.
Branching recovered beyond the untreated wild-type level
The most visually and functionally striking experiment involved dendritic arborisation. Cortical neurons from embryonic mice were cultured and analysed with Sholl profiling, which counts how often neuronal branches intersect a series of imaginary circles extending outward from the cell body.
Untreated Wdfy3-heterozygous neurons showed fewer intersections than untreated wild-type neurons at distances of 35 to 55 micrometres from the cell body. Lithium treatment increased branching in the heterozygous neurons across 15 to 55 micrometres compared with untreated heterozygous cells. At 10 to 20 micrometres, the treated heterozygous neurons even showed more intersections than untreated wild-type cells.
Importantly, lithium did not significantly change branching in wild-type neurons. The response therefore appeared particularly pronounced in the Wdfy3-deficient cells under these culture conditions.
Yet the mechanism became more complicated rather than simpler. Although lithium robustly increased Wdfy3 expression and branching, the researchers did not find evidence that sustained canonical β-catenin stabilisation was the main driver of the branching effect in heterozygous neurons. That leaves open the possibility that other GSK3β substrates or parallel signalling pathways are responsible.
Why this is not evidence to treat patients with lithium
The study is strongest as a bridge between rare-disease genetics and cellular mechanism. It expands the clinical cohort, reproduces aspects of WDFY3 biology in multiple experimental systems and identifies a pharmacological manipulation that changes a measurable neuronal phenotype.
But several limits are crucial. The human clinical component is observational and relatively small because the disorder is rare. The mechanistic experiments rely on a neuroblastoma cell line and mouse neurons rather than neurons derived directly from affected patients. Lithium chloride was tested in vitro at 5 millimolar, and cellular exposure cannot be translated directly into a safe or effective human dose. The branching experiments also used small biological sample sizes, while some expression estimates showed substantial variability.
Most importantly, improved neurite branching in cultured neurons is not the same outcome as improved cognition, development or behaviour in a person. Lithium has a narrow therapeutic window in clinical use and requires medical monitoring. The present research does not establish efficacy, dosing or safety for people with WDFY3 variants.
What it does provide is a testable path forward. Patient-derived neuronal models could establish whether the same molecular and morphological abnormalities occur in human cells carrying naturally occurring WDFY3 variants. Animal studies could examine whether intervention changes developmental or behavioural outcomes. Only after those steps would a clinical trial become a meaningful question.
A rare disorder points to a broader question about neuronal development
Rare genetic disorders often function as unusually clear biological experiments. When a single disrupted gene produces a recognisable developmental pattern, researchers can work backward from the phenotype toward the cellular processes that shape the brain.
In this case, WDFY3 sits at an intersection between cellular waste handling, developmental signalling, cell proliferation and neuronal branching. The new results suggest that losing part of its function may keep neuronal cells dividing while making it harder for them to mature and extend normal processes. Lithium chloride altered part of that phenotype in culture, but the study also showed that the route to recovery may not follow the signalling pathway researchers initially expected.
That complexity is scientifically useful. A potential intervention that works through an unexpected mechanism can expose new biology as well as new therapeutic targets. For WDFY3-associated neurodevelopmental disorder, the next challenge is to determine whether the striking cellular rescue survives the much harder tests of intact organisms and, eventually, carefully designed human research.
Source Information
Study: Lithium chloride in vitro treatment shows potential to rescue the neuronal phenotype caused by WDFY3 haploinsufficiency
Authors: M. J. Paha, A. Mustafa, L. Tat and colleagues
Journal: Molecular Psychiatry
Published: 27 September 2026
DOI: 10.1038/s41380-026-03844-5
Study type: Clinical phenotype characterisation combined with human neuronal cell experiments and mouse cortical neuron models
Primary source: Springer Nature, Molecular Psychiatry
