Paola Arlotta awarded ARIA grant for autism research

The human brain has long posed unique challenges for researchers. The very organ that most differentiates humans from the rest of the animal kingdom is arguably the hardest to study. For one thing, it develops over many years, expanding rapidly in utero and in early childhood, but continuing to grow through adolescence and rewiring into early adulthood.

Brain biopsies risk neurological damage to the patient, and unlike some organs, the brain doesn’t regenerate — meaning that a removed section would not replenish for the donor, but also that researchers couldn’t grow more brain tissue in the lab from a small sample. On top of that, there is no “single human brain,” as Paola Arlotta, Golub Family Professor of Stem Cell and Regenerative Biology (SCRB) explained, because of genetic and developmental differences and divergences across each person.

But cutting-edge technologies — including artificial intelligence and the study of brain-mimicking “organoids” that Arlotta has pioneered, growing stem cells in a lab into clusters of cerebral cortex cells with firing neurons — are giving scientists new tools for understanding the brain.

Now, a $46 million grant from a new autism research initiative, Aligning Research to Impact Autism (ARIA), will turbocharge that mission at Harvard and Yale, scaling up the efforts of individual labs like Arlotta’s while aligning their work, with a focus on two shared goals: understanding and mapping how the human brain develops — at the neural, cellular, genetic, and molecular levels — to understand the divergences that occur in the developing brains of children with autism.

“This is really a dream-come-true opportunity and collaboration, because of the scale of the investment, and therefore the scale of the science that we will be able to do, and because it brings together experts from a variety of different fields who normally wouldn’t necessarily work together,” said Arlotta, who is also an associate member of the Broad Institute of Harvard and MIT. “We have always studied things one experiment at a time, while here we’re aiming for the full picture.”

Arlotta is one of two lead investigators, along with Yale School of Medicine’s Nenad Sestan, on a set of four coordinated projects that will form the Human Developmental Neurobiology Hub for the ARIA initiative. This initial grant will last three years and be split evenly and tap multiple investigators at both institutions.

Arlotta will lead one project in the Department of Stem Cell and Regenerative Biology (SCRB) and work closely with Marinka Zitnik of Harvard Medical School on a second. Arlotta’s lab is a pioneer in the field of organoids, deriving stem cells from blood samples and carefully guiding them over months and years — with what she calls “knowledge of embryonic development combined with a bit of magic” in spinning flasks — into tiny clusters of specialized brain cells that progressively develop and mature, following similar mechanisms to those of a budding human brain.

Arlotta’s human brain organoids, or “avatars of the brain,” measure about 5 millimeters — roughly the size of an apple seed, but whitish and rounder — and represent tiny clusters of cerebral cortex cells, the part of the brain that controls high-level functions such as language, fine motor skills, and sensory perception. They’re each coded with the unique genetic information of the person who provided the blood sample that yielded the undifferentiated stem cells from which those respective organoids were grown.

Her lab has already demonstrated at a small scale that brain organoids carrying mutations linked to autism present with neurodivergences of development and demonstrate distinct activity patterns from organoids grown from stem cells of people without autism.

Working with ARIA, Arlotta’s lab will generate 150 stem cell lines derived from autism patients and control subjects, both for their own research and to be made available for other researchers in the ARIA initiative. Some of these cells will be turned into brain organoids to investigate any molecular and activity-related divergences between the two groups. These brain-avatar models will help researchers attain a mechanistic understanding of autism.

At Yale, Sestan, a professor of comparative medicine, of genetics, and of psychiatry, will oversee two projects. One will develop a “connectome” of the developing brain, studying histological tissue sections from more than 1,200 human and primate neuroembryological samples to construct a series of high-resolution models showing how the brain develops from womb to adolescence, akin to detailed planetary maps. The other project will entail single-cell genomics from human brain-cell samples to examine which genes are expressed and in which cells at different stages of development.

All three of these projects will feed data to the other project at Harvard, where Zitnik, already a collaborator with Arlotta, will develop AI models at the cellular level to predict how brain cells carrying a particular set of genetic variants develop over time, and how a drug or other intervention could alter that course.

“Our goal in this project is to build a set of computation models that will give rise to what we call a ‘virtual cell model’ of the developing human brain,” said Zitnik, associate professor of biomedical informatics at HMS as well as an associate faculty member of Harvard’s Kempner Institute for the Study of Natural and Artificial Intelligence, who works at the intersection of machine learning and biomedicine.

She likened her work to building a simulation of a spacecraft that engineers cannot test in flight. They check the simulation against a physical model of the vehicle built for testing on the ground, and they revise it wherever the two disagree. In this case, the organoids will play a similar role. Zitnik’s team will work with neuroscientists at every step to gauge whether the AI models are capturing experimental data and drawing reliable inferences — and whether the predictions from those virtual cell models match the continued development of comparison organoids in Arlotta’s lab.

The end goal, Zitnik said, is to build “a faithful model of a virtual cell that is representative of actual human biology.” That would solve a fundamental problem, “because we see human brain development only in snapshots, never as a continuous process. Generative models can learn the process behind those snapshots, which gives us a simulator of the developing cell. We can run it forward, change a single gene, and ask what the cell becomes — then go back to the organoids and test whether the model was right.”

The goal is to eventually simulate “in silico,” or via computer simulation, how a person with a specific set of traits would respond to a particular drug or therapeutic treatment — a watershed moment for a condition that exists on a broad spectrum and for which there are no pharmaceutical therapies.

One in 31 American children under age 8 have been diagnosed with autism spectrum disorders, according to the most recent statistics from the U.S. Centers for Disease Control and Prevention. ARIA’s work is aimed at accelerating discovery for people with profound autism, who require lifelong 24/7 care, but it will ultimately benefit people across the entire autism spectrum.

The Human Developmental Neurobiology Hub is one of six research hubs in the ARIA initiative. Boston Children’s Hospital was recently awarded up to $17.25 million as one of several clinical sites within the IMPACT Network. These IMPACT Network sites will collect patient samples that will fuel part of Arlotta’s efforts.

Lead investigators for the Human Developmental Neurobiology Hub likened the project to a “coordinated moon mission” — in place of fragmented, siloed work — to map the developing brain. Arlotta invoked an even deeper-space analogy.

“How do we train an AI model to make predictions about how the disease unfolds and how to intervene in the future to modify the outcome of a neurodivergent trajectory?” she said. “This has never been possible before, because — as you can see — it requires a sort of intersection of the stars.”