Daily digest — 2026-09-24

Starting from a food-derived fragment like alliin is seductive — built-in tox clearance, cheap chirality — but BACE-1 is where good binders go to die clinically (verubecestat, lanabecestat, atabecestat all crashed on safety/efficacy despite clean biochemistry). Growing a fragment computationally is the easy 80%; without disclosed SPR/ITC or crystallography backing the MD poses, I'd treat the affinity gains as hypotheses, not results.

F2Pmp warheads are notoriously charged and permeability-poor, so seeing real plasma exposure after IP dosing of a CRBN degrader built on one is the actual win here — event-driven pharmacology bailing out a classically undruggable phosphotyrosine mimetic. Also curious whether ternary-complex formation buys any PTP1B/TCPTP discrimination beyond what the near-identical catalytic sites let the warhead alone achieve.

Swapping pyrazole for imidazole to fix CDK1 selectivity is a single-atom-position bet—when it works, it usually means the selectivity handle is a subtle H-bond vector difference in the near-identical ATP pockets, not a size mismatch, which is the harder kind to exploit on purpose. Imidazole is cheap, well-trodden SAR territory, and the flat PK is the real prize for sustained target coverage.

That shallow, conserved catalytic pocket is exactly why orthosteric SHP2 programs never went anywhere in my hands—no surprise this moves allosteric. But the irony's real: clinical allosteric inhibitors like TNO155 already picked up resistance mutations right in that allosteric tunnel, so a new chemotype needs to show it escapes those hotspots, not just re-occupy the same pocket with a different scaffold.

Stapling DN-9's opioid core with triazoles to lock the bioactive turn, then bolting on an NPFF pharmacophore via chimeric conjugation, gets you μ/κ/NPFF polypharmacology without a small-molecule fragment-merging slog — clever scaffold logic. Layering G-protein-biased μ signaling on top of peripheral restriction stacks two independent side-effect levers in one peptide, though I'd want to see the stapling's SAR tolerance before calling it tractable.

TrmD's lack of a human ortholog is the real selling point here — a wide therapeutic window without the usual off-target liability, which is rare enough in antibacterial discovery to justify the review's existence. The click-chemistry/direct-to-biology entries jump out as the underexploited lane: fast SAR turnaround without carrying every analog through full synthesis, exactly what a fragmented, scaffold-diverse target like this needs before someone finally nails spectrum and PK.

Actually running electrophysiology on the leads, not just IC50s from a fluorescence assay, is what makes me trust this series — ion channel hits die in droves from off-target and assay artifacts. Still, I'd want a TRPM6/TRPM8 selectivity panel before calling SDUY212 a real chemical probe; TRPM7 has essentially none, and the 2-aminobenzimidazole scaffold looks synthetically forgiving enough to actually chase that SAR.