Is ARIA the Best Hope for Breakthroughs in Severe Autism?

Спасибо, Сергей (Thanks Sergei!)

 

The IACC wants America to spend much more on autism research. Then, almost out of nowhere, a new initiative starts awarding $46 million grants for autism biology at UCSF, Harvard and Yale. Who is behind it — and could ARIA represent the best hope yet for genuine treatment breakthroughs in profound autism?

 

For years, families of people with severe and profound autism have been told that science is making progress.

Researchers have discovered hundreds of autism-associated genes. They can sequence genomes, study individual brain cells, map molecular pathways, grow patient-derived brain organoids. Artificial intelligence is beginning to analyse biological systems at a scale unimaginable a few years ago.

And yet there remains a perfectly reasonable question:

Where are the breakthroughs?

Where are the treatments that substantially change the lives of people with profound intellectual disability, severe epilepsy, dangerous self-injury, non-verbal communication difficulties, and lifelong dependency?

The Interagency Autism Coordinating Committee (IACC) — a US federal advisory committee that coordinates autism research and policy across government agencies — has argued that America needs substantially greater investment in autism research. But simply spending more money isn't enough. The real question is:

What kind of research should be funded — and how should it be organised?

In the last few weeks, something remarkable has started to happen. A new initiative has announced $46 million for molecular autism research at UCSF, $46 million for a Harvard-Yale project on human brain development, and major investment in clinical trial infrastructure.

So who is behind it?

The answer is Sergey Brin

The initiative is called ARIA — Aligning Research to Impact Autism, funded through CNS Quest, a programme of the Sergey Brin Family Foundation and Catalyst4, as part of a broader effort to accelerate discoveries, treatments and care for conditions affecting the central nervous system.

That immediately brings to mind another visionary builder.

First came Jim Simons

The Simons Foundation Autism Research Initiative (SFARI) changed autism research. Jim Simons, the mathematician and founder of Renaissance Technologies, directed enormous philanthropic resources toward genetics, sequencing, molecular neuroscience, rare genetic conditions, and the large collaborative datasets that built modern autism science.

The result was an explosion of knowledge. But that knowledge also revealed a paradox: there was no single autism gene, no single autism pathway. Instead, scientists found extraordinary polygenic complexity — hundreds of genes, thousands of variants, many different routes into altered brain development.

That was itself a major discovery. But it created a daunting question:

How do you turn hundreds of distinct genetic causes into something therapeutically manageable?

SFARI built the foundation. ARIA is asking what comes next.

ARIA: aligning science towards impact

Even the name is revealing. Not simply funding more autism research, but aligning research toward impact. ARIA has organised its work into interconnected research hubs, bringing multidisciplinary teams together to identify scientific bottlenecks — with explicit emphasis on people with profound autism who need lifelong, 24/7 care, and an explicit goal of developing usable treatment options.

Science is usually fragmented. One lab studies genes, another proteins, another neurons, another runs clinical trials. Discoveries don't automatically travel between them. ARIA is trying to build the bridge:

Genes → proteins → cellular networks → brain development → biological mechanisms → therapeutic targets → interventions

If autism research is going to produce real treatments, someone has to connect the links in that chain.

UCSF and molecular convergence

The first announcement that caught my attention came from the Quantitative Biosciences Institute at UCSF, awarded $46 million under ARIA's Protein-Protein Interactions Hub.

Genes produce proteins. Proteins interact and form the networks that do the actual work inside cells. UCSF researchers recently produced what they describe as the largest molecular interaction map of autism risk genes so far — and found something potentially very important:

Hundreds of apparently different genetic mutations may converge on a surprisingly small number of shared protein complexes.

Imagine Gene A, Gene B, Gene C and Gene D each causing a distinct-looking developmental disorder. At first glance, four unrelated problems requiring four unrelated treatments. But if all four eventually disrupt the same piece of molecular machinery, the therapeutic problem changes completely. Instead of needing four bespoke interventions for tiny sub-populations, researchers might be able to target a shared biological vulnerability.

This is not a universal autism drug. Autism biology is far too complicated for that. But it reframes the question from "what does autism look like?" to "what has actually gone wrong inside the cell — and can we do something about it?"

It's a question I've been asking on this blog for over a decade, in a much smaller and unfunded way: that the most productive way to think about autism may not be gene-by-gene, but as a set of shared downstream pathways — convergent biology that a relatively small number of interventions could plausibly touch. Seeing a $46 million, UCSF-led programme built explicitly around that same convergence hypothesis is, to put it mildly, encouraging.

Harvard, Yale, and the developmental movie

Shortly after, ARIA awarded another $46 million to establish a Human Developmental Neurobiology Hub at Harvard and Yale, co-led by Paola Arlotta (Harvard), Nenad Sestan (Yale), and biomedical AI researcher Marinka Zitnik.

The ambition: understand how the human brain develops, and where that development diverges in autism — one of the hardest questions in neuroscience, because you cannot repeatedly sample a developing child's brain, and every person has a different genetic background. Developmental neuroscience has mostly consisted of snapshots.

The hub combines four interconnected projects: high-resolution mapping of brain development from before birth through adolescence; cell-by-cell gene expression mapping across developmental stages; 150 patient-derived stem-cell lines used to grow brain organoids that preserve each donor's own genetic background; and a "virtual cell model" — an AI system intended to predict how cells carrying particular genetic variants develop over time, and eventually how an intervention might alter that trajectory.

The loop is meant to run: human developmental data → organoids → AI → predictions → experiments → improved models. Traditional developmental neuroscience gives snapshots; the ambition here is something closer to a movie.

The translational catch

There is, however, a real bottleneck this framework has to overcome, and it's worth naming plainly: timing.

Brain organoids and fetal tissue maps primarily capture prenatal neurogenesis and early cell migration. But profound autism is usually recognised and diagnosed postnatally, typically between ages two and four. If the biological divergences this programme maps turn out to require intervention before birth to change, their practical value for a living child will be limited.

For this architecture to deliver breakthroughs that matter to families now, it needs to identify mechanisms that stay malleable after birth — ongoing synaptic plasticity, channelopathies, neuroinflammation, metabolic signalling in mature circuits — not just the embryonic events that set development on its initial course. That's not a reason to be cynical about the programme. It's the specific scientific test it will have to pass.

NIH, SFARI and ARIA

America now has an unusual tripartite architecture: the National Institutes of Health (NIH) — the US government's main biomedical research agency, funding basic and clinical science across virtually every disease area — providing scale and infrastructure; SFARI providing deep genetic cataloguing and cohort-building; and ARIA acting as a fast-moving, high-risk engine explicitly linking molecular discovery to therapeutic targets. Government provides stability, foundations provide specialist depth, and visionary philanthropy takes the strategic risks neither of the others can.

Which raises an uncomfortable question: why is there so little comparable activity in Europe?

Cambridge, and a structural paradox

I have nothing but respect for Cambridge — both my parents studied there, as did two of my siblings, and as a STEM university it is extraordinary, with world-class strength in genetics, molecular biology, developmental biology, neuroscience, computing and AI. Which is exactly why I find its traditional approach to autism research so frustrating.

Cambridge's Autism Research Centre has historically been associated far more with psychology, cognition, autistic traits, diagnosis and social experience than with the molecular and developmental biology ARIA is now funding. Its own position has been explicit: it says it does not seek a cure for autism itself, since autism is part of who a person is — though it does support treatment of specific symptoms and associated problems that cause distress. That distinction is real and deserves to be represented fairly. But so does the philosophical difference between describing autism and investigating whether its most severe biological consequences can be prevented or reduced.

In July 2026, Cambridge received one of the largest philanthropic gifts for autism research ever made to a British university: $34.5 million from K. Lisa Yang, $28 million of it establishing the K. Lisa Yang Centre for Autism Research, the rest supporting a new clinical centre at Cambridge Children's Hospital.

Cambridge was not the only recipient. Yang's gift established matching K. Lisa Yang Centres for Autism Research at Harvard and MIT as well — institutions that, unlike Cambridge's ARC, already have deep infrastructure in exactly the kind of molecular and developmental biology ARIA is now funding. That makes the Cambridge leg of the gift more striking, not less: the same donor, with the same stated ambition, chose to make an equivalent bet on an institution whose autism centre has historically taken a markedly less scientific approach.

Lisa Yang's philanthropy is deeply personal — she has written about her own children, one more profoundly autistic than the other, and about wanting better therapies for those most severely affected. Announcing the gift, she called for centres that would "uncover therapeutics and interventions" through "daring out-of-the-box and innovative thinking."

That sets up a genuine structural paradox: a donor explicitly asking for actionable therapeutics and biological breakthroughs, given to an institution whose autism centre has, for decades, operated within a descriptive and psychological framework that has explicitly said it isn't seeking a cure. The two positions aren't necessarily incompatible — treating distressing symptoms is not the same as denying identity — but the tension is real, and whether $34.5 million is enough to pivot a legacy institution's scientific orientation toward hard translational biology is an open question. Cambridge was chosen, by its own account, because of its Autism Research Centre and its nearly thirty-year history — so this isn't a case of a donor stumbling in unaware. The question is whether she expects, and can get, a shift.

Why not East Asia?

Every major gift also carries an opportunity cost. Western prestige institutions — Harvard, MIT, Cambridge — naturally attract this kind of philanthropy, but some of the most exciting functional genomics and developmental neuroscience infrastructure right now is in East Asia: Japan's RIKEN in developmental biology and stem-cell models; South Korea in multi-omics and rapid biotech translation; Taiwan in computational biology, medical AI and precision medicine. Building equivalent hubs there would also help correct a real problem — autism genetics has been built overwhelmingly on populations of European ancestry.

None of this is an argument that Japan, Korea or Taiwan are scientifically behind — quite the opposite, they're already producing world-class science. It's a question about whether autism philanthropy is still too automatically drawn to a familiar Western prestige hierarchy, when prestige and maximum scientific impact aren't the same thing.

There's a genuine irony worth sitting with here. Lisa Yang grew up in Singapore; her former philanthropic partner Hock Tan is from Malaysia. Two donors with direct personal ties to a region now producing some of the world's sharpest developmental biology and genomics infrastructure chose, when the moment came, to route one of the largest autism gifts in British history toward a psychology-oriented centre at a famous UK university rather than toward the scientific capability that already exists closer to their own roots. Whatever the reasoning behind it, it's a missed opportunity — and a reminder of how strong the pull of the traditional Western prestige hierarchy still is, even for donors best positioned to see past it.

Is ARIA the best hope?

It's too early to know. Autism biology is notoriously resistant to easy answers, and turning a promising convergence finding in a dish into a safe, blood-brain-barrier-crossing therapy is an enormous distance to travel. Many of these projects will fail.

But ARIA is doing something genuinely different from most autism funding announcements. It isn't cataloguing more genes, and it isn't purely descriptive psychology. It connects protein complexes and organoids to AI predictive modelling and clinical trial infrastructure — asking not just what autism looks like, but what has actually gone wrong inside the cell, and where it might still be possible to intervene.

For families living with profound autism, that shift in ambition isn't just welcome. It's long overdue.