←
Robert Y. Chen4,503 posts
Robert Y. Chen avatar

Robert Y. Chen

@therealRYC

PGY-4 @UW Psychiatry. Prev MD-PhD @WUSTL. Biomarker and drug discovery for psychiatry, starting with Schizophrenia. Dog Daddy x2 to Taro and Azuki.

◷ Joined February 2020▧ 1,169 Photos and videos
Robert Y. Chen avatar
Robert Y. Chen@therealRYC·Nov 10

🚨 [ONE GENE PROTECTS AGAINST AUTISM] Autism is 3 to 4 times more common in boys than girls. For years, we've known females have a "protective effect" against autism mutations, but we didn't know how it worked. A new @Nature paper just found the molecular buffer: a gene on the X chromosome called ZFX. 🧵

Post attachment
Robert Y. Chen avatar
Robert Y. Chen@therealRYC·Nov 10

💡 Takeaways For the first time, we have a concrete molecular mechanism explaining the female protective effect in autism. Rather than pointing to sex hormones, this paper identifies a transcriptional buffer: ZFX, which is expressed higher in females because it escapes X-inactivation. By demonstrating that 18 distinct autism genes all collapse the same downstream network, they showed that ZFX acts as a structural reinforcement for that exact network, allowing the female brain to absorb genetic hits that would push a male brain into clinical disease. The multi-billion-dollar question is whether we can therapeutically upregulate ZFX or its downstream network in males to mimic this resilience and alter the developmental trajectory of autism. Just think about that for a second: we have a SINGLE gene that could potentially SHIELD against Autism. Despite Autism being caused by hundreds of different genes that can go wrong, we can make ONE genetic medicine, NOT hundreds. The implications of convergent biology is profound. One of the reasons I shied away from researching ASD was because I didn't see how a therapy could be reasonably developed when the genetics were so pleitropic. This paper has changed my mind. Source: Andersen et al. Nature 2026 — doi.org/10.1038/s41586-026-1…

Post attachment
Robert Y. Chen avatar
Robert Y. Chen@therealRYC·Nov 10

🔎 Only modeled in a male cell line This study proposes a brilliant mechanism for female protection, but the entire experimental platform rests on a single male (XY) iPSC line. Because they used an isogenic male line to reduce genomic background noise, they could not directly measure whether female (XX) cells actually resist these transcriptomic ripples better than male cells do in a dish. The buffering effect of ZFX is elegantly deduced from network analysis, gene expression databases, and male autism brain tissue, but the functional "stress test" comparing XX versus XY organoid resilience hasn't been run yet. Furthermore, they targeted rare, highly penetrant single-gene causes of autism. It remains to be seen if this network also buffers against the common, polygenic risk that drives the vast majority of clinical autism cases.

Post attachment
Robert Y. Chen avatar
Robert Y. Chen@therealRYC·Nov 10

🧬 The X chromosome buffer When they looked for genes that were consistently downregulated by these knockdowns and also had female-biased expression in the brain, one stood out: ZFX. » ZFX is a potent transcription factor located on the X chromosome » It escapes X-inactivation, meaning females express it from both of their X chromosomes, giving them higher baseline levels than males » ZFX was found to be a massive activator of autism genes—its targets were highly enriched for the top drivers of transcriptomic convergence Because female brains have higher baseline ZFX expression, they are structurally buffered against the cascading downregulation caused by autism mutations.

Post attachment
Robert Y. Chen avatar
Robert Y. Chen@therealRYC·Nov 10

🧠 Mapping the regulatory network The team built a gene regulatory network from this RNA-seq data to find the master switches controlling these widespread transcriptomic ripples. » 70 transcription factors and epigenetic regulators were identified as "top drivers" of this convergence » Nearly half of these drivers were known autism risk genes themselves, including the prominent gene CHD8 » Non-autism genes like REST (a master negative regulator of neurogenesis) also emerged as highly connected central hubs When you disrupt a single upstream regulator, the changes cascade through this highly interconnected web. These shared downstream targets dictate the shared neurodevelopmental phenotypes.

Post attachment
Robert Y. Chen avatar
Robert Y. Chen@therealRYC·Nov 10

📈 Transcriptomic convergence Despite targeting 18 distinct genes that perform different cellular functions, the downstream effects on the transcriptome strongly overlapped. » 83% of the knockdown perturbations significantly dysregulated other known autism genes » 27 of the highest-confidence autism genes were downregulated in at least 10 of the 18 knockdown conditions » This overlap was specific to autism: there was no enrichment for genes that exclusively cause intellectual disability This suggests that disrupting one node of the autism network pulls down the rest of it, explaining why so many different mutations can lead to the exact same clinical diagnosis. Think of it as a NETWORK of Autism genes - pulling on one pulls on the others. One goes down, they all go down.

Post attachment
Robert Y. Chen avatar
Robert Y. Chen@therealRYC·Nov 10

📋 Knocking down autism genes To find the point of convergence, researchers used CRISPR (specifically an RNA-targeting enzyme called CasRx) to suppress high-confidence autism risk genes. They did this in human neural progenitor cells and 3D mosaic cerebral organoids using an inducible system called FLICK. » N = 20 individual ASD risk genes targeted, with 18 analyzed » Model: Isogenic male human induced pluripotent stem cells (iPSCs) » Assays: Bulk RNA-seq at 24h for all genes, plus scRNA-seq in 1- and 2-month organoids for a subset of 7 genes The goal was to see if disrupting totally different genetic targets produced the same downstream ripple effects in brain development.

Post attachment
Robert Y. Chen avatar
Robert Y. Chen@therealRYC·Nov 10

🤔 The female protective effect Autism is highly heritable, with HUNDREDS of different rare genes implicated in its development. Yet, females diagnosed with ASD consistently carry a significantly higher burden of these severe mutations than males do. The ratio of autism in boys versus girls sits at 3 to 4 to 1. This implies that a female brain can tolerate more genetic hits before crossing the threshold into clinical autism. But how does a developing brain actually buffer against so many diverse, functionally unrelated genetic mutations? We have lacked a unified molecular explanation for this resilience.

Post attachment
Robert Y. Chen avatar
Robert Y. Chen@therealRYC·Nov 7

💡 Takeaways For decades, DREADDs have been a staple of neuroscience animal labs. This paper successfully translates that chemogenetic control into human patients. It is a true paradigm shift for neuromodulation. Instead of continuous electrical stimulation or resecting brain tissue, we can now genetically tag a hyperactive network and silence it with a targeted oral drug. The magnitude of the early response is remarkable, effectively halting refractory seizures in a tiny cohort that had failed multiple standard therapies for decades. The open question is scalability. Whether this exquisite localized control can be reliably replicated across larger populations and more complex seizure architectures will define its future in clinical neurology. But here’s where I’m really getting excited. DREADDS are a general platform for homing a small molecule to a specific target. Let me break it down: » Small molecules are messy and go everywhere which means off target effects are all but inevitable. » Genetics are specific: you can have a cell-type specific promoter drive the expression of an on/off switch (an excitatory or inhibitory channel) that only flips when given a small molecule that is highly specific to that receptor. SO: that means instead of blocking all sodium channels in all neurons, you can block them in just the subset you want. AND: because we are making single cell RNA atlases of different neuronal cell types with different promoter specificities, we will be able to target a huge diversity of neurons - with SMALL MOLECULES!! The implications are vast and go far beyond neuropsychiatric diseases. Imagine treating arrhythmias with DREADDs. Or cystic fibrosis, which is also primarily a channel disorder? Lots to leave to the imagination. Source: Dai et al. N Engl J Med 2026 — doi.org/10.1056/NEJMc2606815

Post attachment
Robert Y. Chen avatar
Robert Y. Chen@therealRYC·Nov 7

🔎 An open-label surgical trial in three people Okay. Yes it’s exciting. But this is early days: The dramatic 92% reduction rests on just three patients, which means we are looking at proof-of-concept safety data, not a powered efficacy measure. Because this was an open-label surgical intervention, patient placebo response is uncontrolled, though blinded evaluators controlled for rater expectancy. Furthermore, the targeted delivery requires precise, MRI-guided intracranial infusion into a known structural lesion (hippocampal sclerosis). That limits the immediate generalizability. We do not yet know if this approach will work in non-lesional epilepsy or multifocal disease where the epileptogenic zone is harder to neatly bracket with an infusion trajectory.

Post attachment
Robert Y. Chen avatar
Robert Y. Chen@therealRYC·Nov 7

🔄 An accidental proof of mechanism Long-term follow-up extended to 12 months for Patient 1, and 8 months for the others. The benefit was sustained, and safety data held up perfectly. All adverse events were mild (grade 1 or 2), with no vector shedding or clinical cognitive decline observed. But the most fascinating data point came from an unplanned interruption. Patient 3 temporarily stopped taking the clozapine pills. Without the ligand to activate the engineered receptors, the native epileptic activity returned, resulting in two breakthrough seizures. When clozapine was resumed, seizure control was immediately restored. It is a perfect human demonstration of the chemogenetic on/off switch.

Post attachment
Robert Y. Chen avatar
Robert Y. Chen@therealRYC·Nov 7

📉 Near-complete seizure suppression The clinical response during the 12-week efficacy assessment was stark, with a mean seizure frequency reduction of 92.6% compared to the two months prior to surgery. » Patient 1: 4.5 seizures/month down to 0.5 » Patient 2: 4.0 seizures/month down to 0 (seizure-free) » Patient 3: 4.5 seizures/month down to 0.5 » Liverpool Seizure Severity Scale scores decreased in all three patients » Interictal EEG spikes dropped dramatically (e.g., Patient 3 went from 348 to 128 spikes/hour) By placing the receptors directly in the seizure focus, the low-dose clozapine acts as a targeted brake, achieving local seizure control with all adverse events limited to grade 1 or 2.

Post attachment
Robert Y. Chen avatar
Robert Y. Chen@therealRYC·Nov 7

📋 First-in-human trial design This was a phase 1 safety and proof-of-concept study targeting mesial temporal lobe epilepsy with hippocampal sclerosis (MTLE-HS). » N = 3 adult males (ages 28–37) with drug-resistant epilepsy for 10–25 years » Intervention 1: Intraoperative MRI-guided convection-enhanced delivery of the AAV vector (GA002) directly into the sclerotic hippocampus » Intervention 2: Two weeks later, patients started oral clozapine (12.5 mg every 8 hours) to activate the receptors, continuing their baseline antiseizure medications » Primary endpoint was safety; secondary included seizure frequency at 12 weeks and EEG spike rates The infusion volumes were highly individualized based on 3D-MRI to maximize coverage of the target region without spilling over.

Post attachment
Robert Y. Chen avatar
Robert Y. Chen@therealRYC·Nov 7

🧠 Engineering an on/off switch About 30% of epilepsy patients are drug-resistant, and surgical resection isn't always viable. Enter chemogenetics. Combining the best parts of small molecules and genetic targeting. Instead of broadly dampening the whole brain, you inject a viral vector into the specific seizure focus. This vector (AAV with a CaMKIIα promoter) installs mutated hM4D(Gi) receptors exclusively on excitatory neurons. These "designer receptors" don't respond to the brain's natural neurotransmitters. They sit dormant until the patient swallows a specific drug (in this case, low-dose clozapine). The clozapine binds, hyperpolarizes the neuron, and shuts down the epileptic firing.

Post attachment
Robert Y. Chen avatar
Robert Y. Chen@therealRYC·Nov 7

🚨 [SCIENCE BREAKTHROUGH] DREADDs have officially reached human trials. Chemogenetic neuromodulation aka using viral vectors to install designer receptors in the brain, then activating them with an oral pill, just showed proof-of-concept for refractory epilepsy. Seizures dropped 92% in 3 patients. But the real potential is the PLATFORM - I break it down at the end🧵

Post attachment
Robert Y. Chen avatar
Robert Y. Chen@therealRYC·Nov 6

One of the most interesting stories I’ve seen of citizen science being translated into therapeutic potential - thanks for sharing your journey! The last story like this I can remember was aducanumab (Alzheimer’s antibody) but even then that was an explicit search for samples in people who were older without Alzheimer’s - this is like 100x cooler!

I think I just received the most statistically improbable good news in human history and if you think I am exaggerating read the tldr TLDR: My blood sample was used by researchers to develop a *successful* anti-aging drug since scientists discovered that my extremely rare (<100 diagnosed worldwide) hereditary bleeding disorder, Pai-1 deficiency, has for side effect to increase life expectancy by ≈10% … bruh lmfao am never beating the allegations 1/x