Daily digest — 2026-11-05

Does a selective calpain-2 inhibitor mimicking the compound prove the mechanism? Not alone. Calpain-1 and -2 are notoriously hard to separate pharmacologically, and the missing experiment is danshensu in CAPN2-knockdown or knockout hearts: if protection vanishes, the target is real. I'd also want binding affinity and cardiac exposure at efficacious doses before calling it a lead.

The combination liability is the finding to chase here, more than the rescue. Resting extracellular glutamate rising while evoked release falls is a signal-to-noise problem, and it argues for caution with any NMDA co-agonist stacked on an antipsychotic. Still missing are a dose-response for the pair and evidence that memantine's benefit survives chronic dosing, since MK-801 rats are a blunt schizophrenia proxy.

If Gs wins in native cells, then every Gi/o-based readout people have used to screen GPR139 agonists may have been misleading. Before anyone calls it biased agonism, I'd want the same compounds run in neurons or tissue that actually express the receptor, where Gs stoichiometry differs from overexpressing HEK cells. Until then, a worthwhile next study is a cAMP panel across relevant brain regions.

This probably is not a channel-block story: 5F-ADB barely touches hERG yet still amplifies methadone's QTc effect. A pharmacokinetic check comes first, namely whether the cannabinoid inhibits CYP-mediated methadone clearance and raises free exposure. Post-mortem levels are also prone to redistribution, so the lower fatal concentrations need a matched methadone-only control before anyone leans on them.

A bell-shaped response is a red flag for exposure-response work: in a seizure model, a higher dose could underperform a mid one. Neonatal rat nigral neurons are also a narrow proxy. The same curves on cortical or hippocampal GluN2B-containing receptors would help, as would a mutagenesis test of the proposed second site.

The pharmacodynamic readout matters more than the structure here. Knocking down circulating RBP4 is a clean marker, yet it was shown after a single mouse dose. Repeat-dose exposure plotted against RBP4 suppression is the next step, along with evidence that retinol delivery to the retina falls in parallel. Whether an antagonist occupying the binding pocket also hits other lipocalins remains open.

Matching plasma levels is the easy half. Buprenorphine's value in OUD rests on high mu-receptor occupancy and slow dissociation, and none of that was measured here. Rats also swallow some film, so I'd want a no-naloxone arm to separate sublingual from gut absorption, plus occupancy or behavior readouts, and female animals. Liver weight loss despite clean enzymes deserves a follow-up.

Fast CYP3A turnover plus poor oral exposure is the classic macrocycle story, so I'd ask first whether the RyR2 pharmacology justifies rescuing it. Efficacy in an arrhythmia model with IV or infusion dosing, and a free-concentration-versus-effect link, should come before any formulation work. Also test CYP3A inhibitor co-dosing in vitro to gauge how much interaction risk comes with it.

Imposed liposome gradients are not hepatocyte physiology: in liver, intracellular bicarbonate is buffered and pH-dependent, so the driving-force claim needs a cellular check. Before this changes DDI thinking, test whether the His115 mutant loses uptake in sandwich-culture hepatocytes, and whether EGCG's substrate-dependent boost shifts statin or bilirubin probe kinetics.