10/n: Reflection:
Incredible collab between my lab (DMS), Ntranos Lab (PLM), @amanglik.bsky.social's Lab (structural biology), & Cyster Lab (lymphocyte trafficking). All led by Taylor LaFlam, a Pediatric Rheumatology Fellow. I learned so much and can't wait for what's next...

9/n: Broader Impact for Genetics:
* With synbio & genome engineering, we can test in vitro variant effects at scale, many of which may rarely segregate in humans.
* PLMs can be fine-tuned with experimental data to improve in silico prediction of GoF mutations which are depleted in training data.

8/n: Lymphoma Connections:
* P2RY8 is mutated frequently in GC-derived B cell lymphomas.
* Almost all such tumor variants reduce P2RY8 function, suggesting that escaping confinement leads to uncontrolled growth.
* Potential path toward novel therapeutic strategies targeting P2RY8 & related pathways.

7/n: Computational Twist:
* We compared DMS results with zero shot variant effect predictions using protein language models (PLMs).
* Found decent correlation (~Spearman 0.6)—but much better after “fine-tuning” (~Spearman 0.8) with ~20% of real experimental data points. Thanks Vasilis Ntranos!

6/n: In collaboration with Jason Cyster, who has done seminal work characterizing the gene, we validated variant effects in vivo. Mice expressing human P2RY8 variants (they don't natively express the gene) showed differences in GC confinement vs. B cell expansion.

5/n: Christian Billesboelle from the Manglik lab solved the structure by Cryo-EM:
* Captured P2RY8 in its active conformation bound to its ligand.
* Showed key contacts needed for ligand binding vs. G-protein interaction.
* Helped interpret which amino acids are especially critical for function.

4/n: We also found distinct classes of mutations:
* Loss-of-function (LoF): reduce expression -> increase migration (less constraint)
* Gain-of-function (GoF): boost expression -> decreased migration (more constraint)
* Subtle “bias” mutants: normal in one assay, impaired in another

3/n: Taylor introduced nearly every possible missense substitution (7,045 variants) into B cells lacking endogenous P2RY8. Effects on 3 phenotypes were measured: expression, migration, & proliferation. We observed an inverse relationship between expression & migration across the allelic series.

2/n: Why P2RY8?
* It’s a GPCR critical for restraining GC B cell migration and proliferation.
* Mutations in P2RY8 are linked to lymphomas like DLBCL and Burkitt.
* But how these mutations affect expression or function wasn’t fully understood.

We also saw an enrichment of transitional B cells that do not seem to fully mature in SSA+ samples. These cells have a less diverse repertoire and their frequency is associated with type-1 interferon response. 5/n

Really striking to see SSA+ and SSA- are molecularly distinct with a signature of Type-1 interferon response, often seen in lupus, in SSA+ patients. 4/n

We analyzed PBMCs using multiplexed VDJ+CITE-seq from 333 patients from the SICCA cohort (an amazing resource assembled by Caroline Shiboski here at UCSF. 3/n

Sjogrens is an autoimmune disease marked by immune infiltration of exocrine glands. Anti-SSA antibodies are a hallmark (SSA+), but nearly half of patients lack them, raising big questions about disease mechanisms. 2/n