To learn more about how this study fits into the broader scientific vision of the Levy Lab, read a Q&A with Innovation Investigator @maayanlevy.bsky.social: arcinstitute.org/news/meet-ma...
Metabolite supplementation could offer a simple, low-cost way to improve CAR T cell therapy without any changes to the cells themselves. A clinical trial in r/r DLBCL is now enrolling (NCT06610344).
See the paper: www.cell.com/cell/fulltex...
BHB also reshapes the CAR T cells themselves. Single-cell sequencing showed BHB-treated cells shift toward more active, tumor-killing subtypes & reduced expression of exhaustion markers, suggesting BHB helps CAR T cells stay effective for longer.
BHB provides CAR T cells with a better energy supply. When both fuels are available, activated CAR T cells preferentially burn BHB over glucose, directly boosting the mitochondrial output that drives proliferation and tumor killing.
Metabolomic analysis of keto-fed mice pointed to Ξ²-hydroxybutyrate (BHB) as the key factor. Of 6 enriched metabolites tested, BHB was the only one to improve CAR T cell proliferation in vitro. BHB supplementation alone matched the benefits of a full keto diet across 3 cancer models.
The team screened 6 diets in a mouse B-cell lymphoma modelβhigh fiber, high protein, Western, high fat, ketogenic & control. Only the ketogenic diet consistently improved tumor control & survival in CAR T cell-treated mice. The diet had no effect without CAR T cells.
CAR T cell therapy has shown strong results against certain cancers, but metabolic exhaustion limits its effectiveness over time.
In Cell, the @maayanlevy.bsky.social & @christophthaiss.bsky.social labs report that a common supplement can fuel CAR T cells to fight cancer more effectively.
This was a massive effort across Arc, @NVIDIAHealth, @Stanford, & many collaborators. Congratulations and thank you to everyone who contributed to this project.
Learn more from @pdhsu and @brianhie about Evo 2βs first year: arcinstitute.org/news/evo-2-o...
Try it for yourself.
Using Evo Designer, you can paste in a DNA sequence to score variants or generate new sequences in real time.
See how it works: youtu.be/6P0L1qG_PhU
In the updated paper, the team used Evo 2 to generate synthetic DNA that produced experimentally confirmed changes in chromatin accessibility in both mouse & human cells (AUROCs 0.92β0.95).
Code, weights, and resources available here: arcinstitute.org/tools/evo
Since last February, Evo 2 has seen 200+ citations, 88K+ downloads, and 8M+ API requests. Groups at Yale, JohnsHopkins, NCState & UW have applied Evo 2 to problems from Alzheimer's risk prediction to 3D genome structure
First released one year ago, Evo 2 predicts variant effects & generates functional DNA across all domains of life, from bacteria to humans, from a single set of weights, with no fine-tuning required.
www.nature.com/articles/s41...
Evo 2, the largest fully open biological AI model to date, is now published in @nature.com
youtu.be/o_w-E--u7GQ
This new review from Core Investigator @christophthaiss.bsky.social & Innovation Investigator @maayanlevy.bsky.social maps how gut metabolites shape diseases from IBD to cancer, and lays out a framework for using them as therapies.
Congratulations to Arc Innovation Investigator Theo Roth on receiving a 2026 Michelson Medical Research Foundation Next Generation Grant.
Theo's team is using combinatorial genetic engineering to create immune cell states not found in nature.
www.michelsonmedicalresearch.org/news/michels...
We hope this work inspires other labs to apply this framework to other micronutrients.
The Jain Lab is currently looking for a postdoctoral fellow: job-boards.greenhouse.io/arcinstitute...
Go behind the scenes of the paper with the authors: arcinstitute.org/news/vitamin...
By starting with a potential cure & working backward, this work identified a fatal genetic disease that may be treatable with timely vitamin B3 intervention & an approach that could be applied across dozens of other micronutrients.
Read the full paper: www.cell.com/cell/fulltex...
Dietary B3 levels dramatically affected disease. On a B3-deficient diet, no KO mice were born. But B3 supplementation starting at birth extended lifespan more than 40-fold. Treated mice were indistinguishable from littermates at 50+ days.
The team generated the first mouse model of NAXD deficiency. These mice recapitulated the human disease, appearing normal at birth but dying within days. Brain pathology & biochemical changes matched what's seen in patients.
NAXD repairs an aberrant form of NADH called NADHX. Mutations in NAXD cause a devastating neurodevelopmental disorder in humans, typically fatal within the first months of life.
Using genome-wide CRISPR screens under varying vitamin B2 & B3 conditions, the team identified dozens of genetic diseases responsive to high-dose vitamin supplementation. NAXD emerged as the top hit from the B3 screen.
Most drug discovery starts with a disease & searches for a cure. New research out in Cell from Arc investigator Isha Jainβs lab flips this model, starting with a potential cure (vitamins) & systematically identifying which diseases it could treat.
Weβre looking for people to join us on this mission to accelerate scientific progress and find the root causes of disease: job-boards.greenhouse.io/arcinstitute...
See the full announcement: arcinstitute.org/news/the-aud...
We'll test predictions against real experimental results, starting with Alzheimer's disease, building a lab-in-the-loop to connect model outputs to potential drug targets.
Read more about our Alzheimer's Disease Initiative: arcinstitute.org/alzheimers-d...
This means generating large-scale, causal experimental data and using it to continuously train and improve our virtual cell models.
This is our Institute-wide effort to build and validate AI models that accurately predict how cells respond to genetic, chemical, environmental, and metabolic perturbations across cell types and disease states.
Learn more: arcinstitute.org/virtual-cell...
We're honored to be named a grantee in The Audacious Project's 2025 cohort, supporting our Virtual Cell Initiative.
Weβre looking for people to join us on this mission to accelerate scientific progress and find the root causes of disease: job-boards.greenhouse.io/arcinstitute...
See the full announcement:
arcinstitute.org/news/the-aud...
We'll test predictions against real experimental results, starting with Alzheimer's disease, building a lab-in-the-loop to connect model outputs to potential drug targets.
Read more about our Alzheimer's Disease Initiative: arcinstitute.org/alzheimers-d...
This means generating large-scale, causal experimental data and using it to continuously train and improve our virtual cell models.
