Daily digest — 2026-09-23

This is the kind of binding-mode elegance you rarely see survive contact with an animal model — a stereoselective, cysteine-directed covalent lock on BAX's inactive conformation, in a target class that's almost entirely been reversible PPI chemistry. Holding up in both a liver-injury model and ALS-donor motor neurons argues for real engagement, not a lucky crosslinker; I'd still want the off-target cysteine panel.

FI6v3 and CR9114 are smart hotspots to steal from, and a compact L-peptide is far more tractable to make and dose than the antibody it's mimicking. But stem-directed HA binders have a rough history against escape mutants — durability of coverage across drift, not just initial fusion-blocking potency, is the question I'd want answered before getting excited about a therapeutic path.

Dihydropyridines have spent decades gating L-type calcium channels, so watching the core get repointed at caspase-1 is a nice reminder that a synthetically tractable, well-behaved scaffold doesn't have to stay in its original target class. The summary is thin on potency and selectivity numbers, though — as written, this reads as chemotype diversification for an inflammasome target, not yet a validated lead.

Forty-some analogs to land one clearly superior lead is a tight design-make-test cycle, and coupling ML-based ADMET prediction to fragment growing off NVR 3-778 is exactly where these models earn their keep — multi-parameter optimization, not potency-chasing alone. Next question: does 11B-6 hold up against the CAM-resistant HBV variants that have dogged this chemotype before.

Four pharmacophoric regions systematically walked — substituents, bioisosteres, linkers, and a scaffold hop — is a proper second-generation SAR campaign, not a token tweak. Pairing better cancer-cell potency with improved microsomal stability is what actually matters for a caPCNA-interaction disruptor, since these shallow, protein-interface-driven pockets are usually the first thing to fall apart metabolically.

Smarter to go after 'undruggable' mutant NRAS through the kinase that stabilizes it than to keep chasing the mutant protein directly, and a degrader — not a catalytic inhibitor — is what finally buys the CK1 paralog selectivity this target class has chronically lacked. Compound 8u reads like a real template, not just a tool compound to prove the hypothesis.

Masking an H-bond donor by intramolecular cyclization to dodge P-gp is a well-worn CNS trick, but pulling it off on an azaindole-quinazoline kinase core without bleeding DYRK1A potency is the part most efflux-liability campaigns fail at. If the microglial neuroinflammation readout replicates in a second model, this is a genuinely reusable template for other P-gp-limited kinase programs.

Picomolar affinity against an enzyme sitting in the gut lumen is a flex, but for pancreatic α-amylase the win isn't classic PK — it's durable target engagement at low dose. The translational question this conspectus glosses over: does that ultra-potency actually cut the GI side-effect burden that's dogged amylase inhibitors like acarbose, or just move the goalposts on Ki.