The research team, led by Rebekah E. Townsley, Jonathan C. Andrews, and colleagues in the laboratories of Shinya Yamamoto and Michael F. Wangler, focused on three human NLGN3 variants that had been identified in people with autism. The first, p.R451C, is historically significant: it was the very first variant that connected NLGN3 to autism spectrum disorder, discovered in two brothers, and it has been studied intensively ever since. The second, p.R175W, is a de novo variant, meaning it arose fresh in the affected individual rather than being inherited; it was found in a heterozygous female with autism. The third, p.R597W, was inherited from an unaffected mother by a male patient who had both autism and a diagnosed sleep disorder. Despite decades of work on p.R451C, the functional consequences of the other variants had remained largely unexplored in a living animal.

To fill that gap, the researchers turned to Drosophila, the fruit fly, whose nervous system offers a powerful combination of genetic tractability and well-mapped synaptic anatomy. Flies carry a homolog of the human gene called Nlg3, and when the team removed it, the flies showed measurable disruptions in three domains: their sleep patterns changed, the structure of their synapses was altered, and the dynamics of synaptic vesicles, the tiny membrane sacs that store and release neurotransmitters, were disturbed. Crucially, when the researchers introduced the normal human reference version of NLGN3 into these mutant flies, all of these defects were rescued. That result confirmed two important things at once: the fly gene and the human gene are functionally equivalent enough to stand in for each other, and the observed phenotypes genuinely trace back to loss of neuroligin-3 function rather than to experimental side effects.

Neuroligin-3 itself is a synaptic adhesion molecule, a protein embedded in the postsynaptic membrane that reaches across the synaptic cleft and physically locks onto partners on the presynaptic side. This molecular handshake helps organize the synapse, aligning the machinery for neurotransmitter release on one side with receptor arrays on the other. Because neuroligins sit at this structural and functional crossroads, mutations in their coding sequence can plausibly disturb either the building of synapses or the moment-to-moment signaling they carry out, and the new study shows that different mutations do indeed split along those lines.

When the researchers expressed each human variant in flies lacking endogenous Nlg3, a clear division of labor emerged. The p.R175W variant and the maternally inherited p.R451C variant both altered synapse morphology, reshaping the physical structure of neural junctions, while leaving vesicle dynamics largely intact. In sharp contrast, the p.R597W variant, the one associated with a sleep disorder in its human carrier, produced the opposite profile: it disturbed sleep and vesicle dynamics but had minimal impact on synapse shape. This dissociation is the conceptual heart of the paper. It demonstrates that a single gene can be broken in different ways by different mutations, with each break producing a different downstream signature in the nervous system.

The team then pushed further by asking what happens when NLGN3 is present in excess rather than absent. In overexpression models, even the normal human reference allele changed sleep patterns and synaptic morphology, indicating that neuroligin-3 dosage itself matters for nervous system function. Against this background, the variants revealed additional layers of difference. The p.R175W variant worsened the sleep phenotypes beyond what simple overexpression produced, and both p.R175W and p.R451C exacerbated the synapse morphology defects. Taken together with the rescue experiments, these results led the authors to a specific mechanistic conclusion: the de novo p.R175W variant, found in a heterozygous female, behaves as a gain-of-function mutation, actively adding disruptive activity rather than merely removing normal function.

The picture for the maternally inherited variants is more nuanced. Because most reported NLGN3 variants have been found in affected males who inherited them from unaffected mothers, the researchers note that these variants appear to have context-dependent functional impact. A mother carrying the variant on one of her two X chromosomes can remain unaffected, while a son who receives it and has no second X chromosome copy to buffer the effect may develop autism. This pattern fits the well-known sex bias in X-linked neurodevelopmental conditions and helps explain why NLGN3 variants cluster in male patients, even as de novo variants also surface in affected females.

For the broader autism field, the study carries a message that goes beyond NLGN3 itself. Autism spectrum disorder is defined by its heterogeneity: hundreds of genes have been implicated, and no two patients present identically. Much of the research effort has focused on establishing whether a given gene is associated with autism at all, treating each gene as a single entity. This work shows that within one gene, the variants are not interchangeable. Some NLGN3 variants disrupt synaptic development, the slow process of building proper connections, while others alter synaptic function, the fast process of transmitting signals, including the vesicle cycling that underlies neurotransmitter release. If this principle generalizes, then variant-level functional characterization, not just gene-level association, will be essential for understanding why one patient sleeps badly and another does not, or why synaptic architecture is disrupted in one brain and signaling dynamics in another.

The choice of Drosophila as the experimental platform also deserves attention. Sleep in flies is a well-established model system, governed by conserved neural circuitry, and synaptic vesicle dynamics can be imaged and quantified with precision in the fly nervous system. By establishing that the human reference NLGN3 can rescue the fly Nlg3 loss-of-function phenotype, the researchers created a platform in which any NLGN3 variant, including ones not yet discovered, can be tested for its functional signature in a living animal within a manageable timeframe. That kind of scalable variant-to-function pipeline is increasingly seen as necessary as human genetics keeps identifying variants faster than traditional animal models can evaluate them.

The study leaves open questions that future work will need to address, including how the distinct synaptic and behavioral phenotypes produced by different variants map onto the specific clinical profiles of the patients who carried them, and whether the gain-of-function behavior of p.R175W points to therapeutic strategies different from those suited to loss-of-function variants. But the central finding stands on its own: three variants in the same autism risk gene, in the same animal model, produce three distinguishable patterns of synaptic and behavioral disruption. The heterogeneity of autism, the authors suggest, may begin at the level of the individual variant, and understanding that variant-level diversity may be a prerequisite for understanding the disorder itself.

Subject of Research: Functional differences among NLGN3 autism variants in synapse morphology, vesicle dynamics, and sleep behavior in Drosophila

Article Title: NLGN3 autism variants have distinct functional impact on synapses and sleep behavior in Drosophila

Article References: Townsley, R. E., Andrews, J. C., Srivastav, S., Jangam, S. V., Hannan, S. B., Kanca, O., Marcogliese, P. C., Yamamoto, S., & Wangler, M. F. (2026). NLGN3 autism variants have distinct functional impact on synapses and sleep behavior in Drosophila. PLOS Genetics, 22(10), e1012341. https://doi.org/10.1371/journal.pgen.1012341

Image Credits: AI Generated

DOI: 10.1371/journal.pgen.1012341

Keywords: autism spectrum disorder, NLGN3, neuroligin-3, Drosophila, synapse morphology, synaptic vesicle dynamics, sleep behavior, gain-of-function, X-linked gene, de novo variant, PLOS Genetics, neurodevelopment