# Transcript: From Neanderthal DNA to the Dark Genome: Diversity Explained

**Date:** March 17, 2026 · 10:30 PM  
**Session:** [From Neanderthal DNA to the Dark Genome: Diversity Explained](/sessions/2026-03-17/pp1162389-from-neanderthal-dna-to-the-dark-genome-diversity-explained)

## Summary

This session explores the groundbreaking use of human brain organoids to understand neurodevelopment, neurological disorders, and even human evolution. The speaker details how these 'mini-brains' provide an accessible model to study complex processes like learning, disease progression, and the unique characteristics of the human brain, offering new avenues for therapeutic development and insights into our past and future. The main thesis is that brain organoids are revolutionizing neuroscience by overcoming the experimental inaccessibility of the human brain.

## Topics

`brain organoids` · `neurodevelopment` · `autism` · `rett syndrome` · `gene therapy` · `neanderthal dna` · `human evolution` · `space biology` · `aging` · `artificial intelligence` · `drug discovery`

## Key Takeaways

1. Brain organoids provide an invaluable, accessible model for studying human brain development and neurological conditions, enabling researchers to observe processes previously impossible to study directly in humans.
2. This technology allows for the rapid testing of drugs and gene therapies for neurological disorders, significantly accelerating the path from lab discovery to clinical trials, as demonstrated by the approved drug for Rett syndrome.
3. Organoids can be used to explore fundamental questions about human cognition, learning, and even our evolutionary history, offering insights into what makes the human brain unique compared to other species.
4. Space travel can accelerate brain aging in organoids, providing a unique model to study age-related neurological conditions like Alzheimer's and Parkinson's, and to screen for neuroprotective compounds.
5. The integration of organoid intelligence with AI algorithms holds the potential to create more human-like, energy-efficient, and unbiased learning systems, pushing the boundaries of artificial intelligence.

## Full Transcript

I'm originally from Brazil, and I came as a postdoc in the early 2000s, deciding to stay and pursue my career here. I now live in San Diego, California, where I have my own lab at UC San Diego. Today, I'll discuss new ideas on how we can leverage the power of the human brain for various applications, focusing on Neanderthals and how this technology can help us understand other human species' thinking, flexibility, intelligence, and their lessons on art, technology, and adaptation.

The main reason my lab focuses on this research, particularly within the School of Medicine, is to bring new treatments for people suffering from neurological conditions. My perspective shifted from pure academia to a more translational approach when my son, Yvonne, was diagnosed with severe autism and epilepsy. Autism is a broad spectrum, with some individuals having mild social difficulties and others, like my son, facing severe challenges that make it difficult for them to integrate into society, requiring 24/7 attention.

Understanding autism and other neurological conditions is challenging because while we know there's a strong genetic factor, we don't know how genes influence brain networks to affect behavior. This is a major dilemma in neuroscience, and my lab operates at the intersection of genetics and neurobiology. The primary reason for this lack of understanding is the experimental inaccessibility of the human brain, which develops inside the uterus and then within the skull.

Because of this inaccessibility, we cannot perform direct experiments on the human brain. We use non-invasive technologies like ultrasound, EEGs, and MRIs, but their resolution is insufficient to understand how brains form at a cellular level. We lack knowledge about the first neurons to fire, the initial cortical networks that enable communication, and how these processes unfold. Most neuroscience labs study animal brains, like mice, but extrapolating from a mouse to a human brain is often inaccurate due to fundamental differences in brain structure and social behavior.

In 2008, a colleague from Japan, Yoshiki Sasai, demonstrated that human pluripotent stem cells, which can specialize into any tissue, could self-organize into structures resembling the human brain when grown in suspension with the right cues. He called these 'brain organoids,' sometimes referred to as 'mini-brains.' This discovery was pivotal for my lab, as it offered a way to create pieces of the brain, focus on specific areas, or even develop multiple interconnected brain regions.

This technology, combined with cellular reprogramming, means we no longer need discarded embryos. We can reprogram cells from an individual—a piece of skin, blood, or hair—to become pluripotent stem cells. My lab then specializes in guiding these cells to differentiate into brain tissue. It's important to clarify that the term 'mini-brain' is misleading; these are not fully formed, mature human brains in a petri dish.

There are limitations to this technology: the neurons produced are immature, they are not yet vascularized (though we are learning how to achieve this), and not all cell types are represented. As we build better models, we also encounter ethical dilemmas, such as the potential for these tissues to become conscious or self-aware, raising questions about their moral status.

To create a brain organoid, we start with pluripotent stem cells and use a chemical cocktail to block their specialization into other tissues while instructing them to differentiate into brain cells. These cells form clusters in suspension, and we promote their proliferation with growth factors. This process is remarkable because the cells self-organize, following genetically encoded instructions, without us explicitly directing each cell.

We grow these organoids in floating media, where they start as tiny white dots and grow up to 2.5 centimeters, visible to the naked eye. They can last for several years, though maintaining them requires meticulous care. Our longest-kept organoid, nicknamed Hugo, lasted three years before an accidental drop.

These organoids recapitulate human cortical development. Cells migrate from the inside out, guided by specialized radial glia cells, which are human-specific. This process forms layers of cells, mirroring how the human cortex develops in utero. We can observe these developmental stages in the lab, analyzing different cell populations as they emerge.

Initially, progenitor cells give rise to excitatory neurons, which are highly active during early gestation. By four months, inhibitory neurons begin to emerge, modulating this excitation to achieve a balance. This balance is crucial; imbalances can lead to conditions like seizures, autism, or schizophrenia. We can even add immune system cells to create neuro-immune modulation, allowing the organoids to respond to infections like a real brain.

A significant challenge was getting these neurons to fire and form networks. We developed a protocol to achieve this, using multi-electrode arrays to non-invasively monitor electrical activity. By placing organoids on these arrays and observing them for a year, we aimed to capture the full nine months of gestation plus postnatal stages.

We discovered that for the first two months, brain activity is random, without network formation. By four months, connections begin, and by six months, glutamatergic neurons form synchronized networks. Later, inhibitory neurons become active, modulating these networks into different frequencies, similar to the alpha, delta, and gamma waves seen in human EEGs.

This recapitulation of human brain wiring was exciting, but reviewers questioned its validity. Since we couldn't place electrodes in a healthy human fetus, we compared organoid data to EEGs from premature babies between 25 and 38 weeks. We trained a machine learning model to predict the age of premature baby brains from their EEG features, achieving 100% accuracy.

When we fed the machine data from our organoids, it became confused, unable to distinguish between real human brain data and artificial organoid data. This 'eureka moment' showed that organoids follow the same developmental trajectory as the human brain in gestation and postnatal stages. This finding garnered significant attention, prompting discussions about whether these lab-grown brains could be considered 'thinking' or 'conscious.'

These organoids serve as powerful tools for various purposes. For example, during the Zika virus outbreak, we used organoids to prove causation between the virus and microcephaly, observing blocked progenitor cell proliferation and shrinking cortical layers, which was not seen in mouse models. This led to targeted interventions against the mosquito vector.

Similarly, during the COVID-19 pandemic, we exposed organoids to coronavirus. We found it didn't kill cells but destroyed synapses in the cortex. This early observation, initially dismissed, predicted the neurological effects like brain fog and schizophrenia later associated with COVID-19. We are now testing other viruses, including those from the Arctic, on organoids to assess their potential threat.

Organoids also allow us to test the neurodevelopmental effects of environmental toxins and common substances. For instance, we are investigating the impact of PFAS chemicals, which are ubiquitous and linked to neurodevelopmental issues, on brain development. This provides a way to establish causation for substances where epidemiology only suggests correlation.

For autism research, collecting cells from severely affected children is challenging. To overcome this, we launched the 'Tooth Fairy Kid Collection' via social media, asking parents to send shed baby teeth. From these, we extracted dental pulp cells for reprogramming, collecting over 3,000 samples without requiring families to visit the lab.

In our first cohort, studying Rett syndrome, a syndromic form of autism, we observed a significant reduction in synapses compared to controls. Crucially, adding drugs like IGF-1 rescued the number of synapses. This allowed us to model the disease and test drugs in a dish, leading to the approval of a drug called trofinetide for Rett syndrome, accelerating the process by years.

We are building a pipeline for genetic disorders, creating organoids to test gene therapies, pharmacological treatments, or ASOs. For a condition called Pitt-Hopkins syndrome, we used gene therapy to replace a mutated gene in organoids. This treatment restored cell organization, growth, and neuronal function. Last year, the FDA approved testing this gene therapy in the first Pitt-Hopkins patient, marking the first time a brain organoid study directly led to a clinical trial.

After a year, the development of these organoid networks plateaus. I hypothesized that this might be due to a lack of environmental input and output, similar to how a newborn's brain develops rapidly with sensory experiences. My son, Yvonne, who enjoys disassembling and reassembling objects in unusual ways, inspired me to consider how to integrate sensory information into our organoid models.

We explored creating an eye organoid and connecting it to a thalamus and visual cortex organoid. To test if our cortex could receive visual inputs, we transplanted a human organoid into the visual cortex of a mouse whose own visual cortex had been removed. The animal regained sight, processing visual information through the human organoid. Remarkably, both the animal and the organoid showed synchronized brainwaves under anesthesia, raising questions about organoid consciousness.

Back in our in vitro system, we connected photoreceptors to a thalamus and cortex, stimulating them with light. We found that organoids responded to specific light frequencies, showing network reactions. We observed that synapses used in these circuits strengthened (LTP), while unused ones were eliminated (LTD), which is the basis of human cognition, learning, and memory. We successfully demonstrated that organoids could 'remember' specific light frequencies for up to 48 hours.

This ability of organoids to learn suggests a new way to study how the brain learns. We've teamed up with the Center for Engineered Natural Intelligence at UC San Diego to create new AI tools inspired by organoid learning, a field we call 'organoid intelligence.' This approach aims to develop algorithms that are more human-like, potentially complementing or even replacing current AI models, and addressing issues like energy consumption, data storage, and continuous learning.

One fascinating application is the creation of a 'cyborg' robot whose four legs are coordinated by the electrical activity of a human brain organoid in an incubator. The organoid receives sensory input from the robot's infrared 'eyes' and its network response guides the robot's movements. The organoid itself is unaware it's controlling a robot; it's our interface that translates its responses into actions.

We are also teaching organoids to be more 'aware' by enabling them to sense nutrient depletion and 'feed themselves.' This research aims to understand autism better by studying learning and cognition in different subtypes. We've observed that networks from severe autism cases are more rigid, lacking the LTP and LTD seen in neurotypical brains. Interestingly, exposing these rigid networks to psychedelics like 5-MeO-DMT made them more flexible and improved learning, suggesting new therapeutic avenues.

Another area of my lab explores human brain complexity and evolution. We know Neanderthals lived on Earth longer than modern humans, possessing tools and fossil records dating back 0.6 million years. Modern humans, though younger, rapidly developed advanced science, technology, art, and adaptation. I've always wondered what makes the human brain so unique, hypothesizing genetic differences.

By aligning the Neanderthal genome with modern human populations, we identified 61 protein-coding gene alterations unique to modern humans. One gene, NOVA1, a master regulator of neurodevelopment, caught our attention. We used 'organoid neanderthalization' to replace the modern human NOVA1 gene with its Neanderthal variant in organoids, asking what makes us human.

These 'Neanderthalized' organoids looked different, with a 'popcorn shape,' and showed faster maturation but lower overall complexity compared to modern human organoids. This suggests that the modern human NOVA1 variant contributes to slower development, leading to higher complexity, aligning with the extended human gestation and postnatal development compared to other primates.

The question remained: why was this NOVA1 mutation selected in humans? Last year, we found a potential answer. NOVA1 also responds to lead exposure. We know lead poisoning causes neurological problems, as seen in ancient Romans. Collaborating with researchers from Mount Sinai, we analyzed heavy metal levels in Neanderthal and other hominin teeth, finding widespread lead contamination during a period when both species coexisted.

We hypothesized that lead contamination, possibly from volcanic ashes or contaminated water in caves, played a role. When we exposed both Neanderthal and modern human organoids to lead, only the Neanderthal version showed neuronal death in FoxP2-expressing neurons. FoxP2 is a gene crucial for language. This suggests that the modern human NOVA1 variant protected these language-related neurons from lead toxicity, preserving our ability to communicate, which we postulate is our superpower.

This hypothesis suggests that lead poisoning may have inadvertently driven the selection for the modern human NOVA1 variant, contributing to the development of human language. While organoids cannot talk, we are developing tests to observe network differences upon lead exposure, further exploring this idea.

Beyond reconstructing the past, we are also exploring accelerated evolution to create brains resistant to synaptic loss, dementia, and other diseases. However, our current organoid models recapitulate early development, not aging. Inspired by NASA's findings that astronauts experience accelerated cellular aging in space, we sought to use space as a model to age our brain organoids.

Our first mission to space in 2019 confirmed that organoids aged, showing changes in genes and pathways related to aging. One month in space resulted in about 10 years of aging for the organoids. This is bad news for astronauts but provides a unique opportunity for us to study aging and find neuroprotective treatments, leading to collaboration with NASA.

We sent Rett syndrome cells to space to study their aging. We discovered that these aged organoids showed pathways related to viral infection, specifically from endogenous retroviruses (LINE elements) that are usually silent in our genome. As we age, or in conditions like Rett syndrome, these 'dark genome' elements become re-expressed, causing inflammation.

This led us to hypothesize that antiretroviral drugs, typically used for HIV, could treat Rett syndrome. We tested this in animal models, and the animals not only survived longer but also regained motility and other functions. This evidence was sufficient for the FDA to approve a clinical trial for antiretrovirals in Rett syndrome patients.

For age-related conditions like Alzheimer's, we are collaborating with indigenous tribes in the Amazon, studying their ancestral knowledge of plants used for aging and memory. We are cataloging these plants and using AI to identify potential therapeutic molecules. These molecules are then tested in our lab and, if promising, will be sent to space on future missions.

In space, astronaut scientists will test these molecules on organoids to identify neuroprotective compounds that can revert the detrimental effects of aging. This approach aims to find blockbuster drugs for Alzheimer's, dementia, and Parkinson's by focusing on neuroprotection rather than specific diseases. This ambitious project, supported by a significant donation from Denny Sanford, includes creating an orbiting lab for organoid production and testing, with future plans for private space stations to scale up production.

---

*Source: stt · Language: en · Model: google-vertex/gemini-2.5-flash*

[← Back to session](/sessions/2026-03-17/pp1162389-from-neanderthal-dna-to-the-dark-genome-diversity-explained)