By the time researchers had catalogued 102 genes linked to autism in a single landmark paper five years ago, a quietly uncomfortable question had settled over the field. How could so many genes, doing such different jobs, produce something that looks so much like the same disorder? Some govern how chromosomes are packaged. Others regulate the junctions between neurons. Still others keep brain cells dividing at the right pace. Their functions barely overlap. And yet losing any one of them tilts a developing brain in a recognisably similar direction. It is, frankly, one of the stranger puzzles in modern neuroscience.
A new Yale-led study published in Nature Neuroscience has an answer of sorts, though it may be less satisfying than the field hoped for, and simultaneously more useful. The genes themselves may matter less than the neural pathways they feed into.
Using a pooled CRISPR gene-editing approach, Kristen Brennand, the Elizabeth Mears and House Jameson Professor of Psychiatry at Yale School of Medicine, and her colleagues switched off 23 neurodevelopmental disorder genes in human brain cells derived from stem cells, all in parallel, tracked at single-cell resolution across different cell types as the cells matured. The team compared roughly 118,000 individual cells, looking for patterns of gene activity after each disruption. What they found was convergence: different genes, when silenced, produced strikingly similar downstream effects. Not the same effects in every detail, but the same broad destinations.
Converging on the Same Roads
“It’s been really challenging to put our finger on what causes autism,” Brennand said. “This research gives us a new target to study: Not the genes themselves but the way they converge along the same neural pathways.”
Convergence was strongest in mature glutamatergic neurons, the brain’s main excitatory cell type. These are the cells that fire signals forward, the workhorses of cognitive processing; and it’s here that the molecular consequences of losing very different autism-risk genes start to look most alike. Three broad categories of biology kept surfacing: synaptic communication, epigenetic regulation, and, unexpectedly, mitochondrial function. That last finding caught even the researchers off guard. Mitochondria are the cell’s energy factories, not traditionally the first place you’d look when thinking about autism. But across multiple gene knockouts, something about mitochondrial activity shifted in the same direction, suggesting the connection may be more than incidental.
The degree of convergence depended partly on how similar the genes were to begin with, in terms of co-expression patterns in the developing human brain. Genes more strongly associated with autism spectrum disorder specifically tended to converge in excitatory neurons, whereas genes linked more to broader developmental delay showed greater convergence in inhibitory neurons. The distinction matters: it implies different genetic subgroups of neurodevelopmental disorder might be targeting different neural circuits, with consequences for how you’d approach treatment.
Fish, Drugs, and Rescued Behaviours
To move beyond cell culture, co-lead author Ellen Hoffman at Yale’s Child Study Center ran parallel experiments in zebrafish engineered to carry the same mutations. Zebrafish have become a workhorse in autism research; they share a surprising amount of genetic architecture with humans, produce large broods, and their behaviour can be measured automatically at scale. The team tracked 24 parameters across sleep-wake cycles and visual-startle responses, finding that mutants clustered into four distinct groups based on their behavioural signatures.
Hoffman’s team then used the fish to test drugs. Starting from a computational screen of which compounds might reverse the convergent gene expression patterns seen in human neurons, they tested the best candidates in actual mutant animals. Ten of eleven drugs rescued at least one dysregulated behaviour. Paclitaxel, better known as a cancer treatment, normalised several parameters in zebrafish with kdm6b mutations. Pravastatin, a common cholesterol drug, partially corrected behaviour in chd2 mutants. The antidepressant fluvoxamine pushed one mutant line so far in the opposite direction it overcorrected, which in its own way confirms that the drugs are hitting real biological targets. The team also extended their analysis using a machine learning model trained on gene expression from hundreds of post-mortem human brains, pushing the comparison across all 102 known neurodevelopmental disorder genes. The convergence held.
A Longer View
None of this is close to clinical application. Zebrafish are not people, and the leap from correcting sleep disruption in a five-day-old fish larva to treating a neurodevelopmental disorder in a human child involves a great many steps that have defeated promising candidates before. What the research offers is a framework, a way of thinking about autism genetics that shifts the question from “which gene is broken?” to “which pathway is disrupted?”, and suggests that the pathway-level question might be the more tractable one.
Brennand offered a comparison that gestures at the scale of the ambition. We prescribe folic acid supplements to every pregnant woman to guard against neural tube defects, even though not every fetus faces that particular risk. She wonders whether something similar might one day be possible for neurodevelopmental disorders, a broadly applicable intervention targeting the convergent pathways rather than a separate treatment for each of the hundreds of gene variants that can lead there. “I like to say that we’re 20 to 30 years behind cancer research,” she added. “Researchers have been finding cancer genes since the 1980s, and we’ve been finding autism genes since about 2010. But we’re getting there.”
The number of autism-linked genes is still climbing, expected to reach roughly 250 as larger DNA datasets become available. That number, paradoxically, may matter less than researchers once feared. If the genes are many but the broken pathways are few, the therapeutic target has just become considerably smaller.
Source: https://doi.org/10.1038/s41593-026-02247-7
Frequently Asked Questions
Why do hundreds of different genes all seem to cause autism?
The research suggests the genes themselves are less important than the biological pathways they feed into. When 23 different autism-risk genes were silenced in human brain cells, the downstream effects converged on three broad systems: synaptic communication, epigenetic regulation, and mitochondrial function. Different starting points, roughly the same destination.
What is convergence, and why does it matter for treatment?
Convergence is the observation that diverse genetic mutations produce similar downstream effects in brain cells. It matters therapeutically because it suggests researchers don’t necessarily need a separate drug for each of the hundreds of autism-linked gene variants. If those variants all disrupt a smaller number of shared pathways, targeting those pathways could potentially help people regardless of which specific gene is mutated.
How did zebrafish help test these findings?
Zebrafish engineered to carry autism-risk mutations showed distinct behavioural abnormalities in sleep-wake cycles and responses to light. Researchers used computational analysis to identify drugs predicted to reverse the convergent gene expression seen in human neurons, then tested those drugs in the fish. Ten of eleven candidates rescued at least one disrupted behaviour, providing early, preliminary evidence that targeting convergent pathways might have therapeutic value.
Is mitochondrial dysfunction really connected to autism?
The link is stronger than it might seem. Around 5 percent of neurodevelopmental disorder cases meet diagnostic criteria for classic mitochondrial disorders, and several lines of evidence from human and animal studies have connected autism to mitochondrial deficits. The new study found that silencing multiple different autism genes all nudged mitochondrial activity in similar directions, suggesting the relationship is not coincidental, though what drives it remains an open question.
How far away is a treatment based on these findings?
Quite far, almost certainly. The zebrafish results are encouraging but a long way from clinical application; drugs that normalise sleep patterns in larval zebrafish face many hurdles before consideration for human use. The research is better understood as a conceptual advance, a new framework for which biological targets are worth pursuing, rather than an imminent therapy.
ScienceBlog.com has no paywalls, no sponsored content, and no agenda beyond getting the science right. Every story here is written to inform, not to impress an advertiser or push a point of view.
Good science journalism takes time — reading the papers, checking the claims, finding researchers who can put findings in context. We do that work because we think it matters.
If you find this site useful, consider supporting it with a donation. Even a few dollars a month helps keep the coverage independent and free for everyone.