Daily digest — 2026-11-01

Swapping NK-mediated ADCC for T cell redirection is the bet here, and it's a double-edged one: CD3 engagers bring cytokine release and on-target hits on any CCR8+ activated T cell. Syngeneic wins say little about human Treg biology, since mouse CCR8 expression differs. Without the primate PK and tolerability data, I wouldn't move forward.

I would not bank on the mouse genetics alone: myeloid CD74 deletion is a clean target-validation experiment, but OA mice are mostly surgical models that skew toward inflammation more than human disease does. The harder question is delivery. A MIF-mimetic ligand on an LNP may hit CD74+ macrophages, but I want synovial clearance, off-target uptake in other CD74+ antigen-presenting cells, and knockdown duration in a human-sized joint.

Human CHI3L2 knock-in worsening two induced models is gain-of-function evidence only; I'd want knockout or neutralization to show necessity, and psoriasis mouse models are IL-23/imiquimod-driven and notoriously forgiving. CHI3L2 has no known enzymatic activity, so a druggable handle is unclear. Does the phospholipid readout shift in patient skin, and which species?

The B7-H3 arm is the lesson: a gate on CAR expression doesn't help when the gate leaks in the wrong compartment. Fatal neurotoxicity after intracerebroventricular dosing, yet efficacy intravenously, says route and exposure kinetics set the safety margin, not just the logic circuit. Before any IND talk I'd want immunocompetent safety data with human-relevant antigen expression, and brevican-gate leakage measured outside the brain.

A brighter sensor is only as good as its pharmacology. Before I trust the new Gβγ pairing preferences, I want them checked against orthogonal readouts, because overexpressed, tagged subunits can manufacture selectivity. The inverse-agonist detection at the ghrelin receptor is the more useful result: constitutive activity is where TRUPATH went blind, and it matters for screening.

I'd take this as a diagnostic lens, not a new parameter set. Flux plots are handy for explaining why a compound overshoots transiently or why a closing lid slows off-rate, but the evidence here is simulation only. Until someone fits real association/dissociation time courses with it, I'd keep reading kon/koff and residence time first.

Gut-first timing in A53T mice is appealing, but it's an overexpression model, so I wouldn't read it as human early PD. Before touching a LRRK2 inhibitor here, I'd want kinase activity (pRab10, not just protein level) tied to the barrier defect, and rescue by inhibition in the glia-epithelial co-culture. Upregulation alone is correlative, and the secretory autophagy link looks inferred.