Daily digest — 2026-09-21

Caging panobinostat's hydroxamate behind an Fe(II) trigger is a sensible fix for the isocyanate-genotoxicity liability that's dogged this chemotype, and the bioisosteric control actually showing lower genotoxicity than free drug is the kind of data I'd want before calling any HDACi "safer." But what grabs me is the CES1 finding: an esterase chewing through the conjugate in vitro and in vivo is the failure mode a clean MM cell assay never warns you about. That's not a footnote — it's the design problem for the next iteration, whether you engineer around the CES1 recognition motif or add orthogonal masking. I'd rather see this reported now than discovered in an IND package; it's a useful cautionary data point for anyone building Fe(II)-activatable payloads beyond this one hydroxamate.

A degradation route that bypasses CRBN/VHL entirely is worth my attention regardless of target — CMA recruitment sidesteps the ubiquitin-proteasome toolbox, so it should be orthogonal to the resistance mechanisms (neosubstrate mutation, E3 downregulation) that kill PROTAC leads in the clinic. PD-L1 is the result I'd chase: pulling a membrane protein through chaperone-mediated autophagy via LAMP2A translocation is something UPS-based PROTACs structurally can't do. My skepticism sits with the Hsc70 arm — KFERQ-motif engagement is promiscuous by design, so selectivity and off-target chaperone-client disruption need real scrutiny before I trust the SAR. B-2's in vivo PK and clean BRD4 knockdown at least de-risks synthesis/linker chemistry enough to treat this as a platform, not just a proof of concept.

Call this a metabolism play more than a new-target story. With VX-548's clearance liabilities already mapped in human and rat, swapping to a 2-benzimidazole-tetrahydrofuran scaffold reads as targeted scaffold-hopping around known soft spots rather than blind SAR exploration. That's the efficient way to run a fast-follower program: let the first approved drug's ADME data write your design brief. A >10x jump in rat AUC is the number that matters here — that's the difference between a compound needing frequent dosing with exposure gaps and one you can actually push into efficacy studies with margin. Clean hERG and no genotox flags this early derisks the biggest attrition risks before you've spent tox budget. The THF-benzimidazole core isn't exotic chemistry, nothing hard to scale — this looks like a real backup/second-gen candidate, not just a paper compound.

Alpelisib's hyperglycemia problem is the elephant in every PI3Kα SAR discussion — it's the reason dose reductions dog that drug in clinic, tracing straight back to wild-type inhibition in normal insulin signaling. A slow-on, slow-off kinetic selectivity mechanism here is more interesting than a static pocket-shape argument: if H1047R's altered conformational dynamics let this quinazolinone linger on mutant enzyme while it washes out of WT, that's a residence-time story I'd want kinITC and washout kinetics to confirm before fully believing. Tetrasubstituted quinazolinones aren't an exotic scaffold to make, which helps for a program needing heavy analoging around that pocket. Clean xenograft efficacy without glucose/insulin perturbation is the proof-of-concept that matters — but the authors' own admission that compound 34 needs a lower efficacious dose and better tolerability is the real tell: promising probe, not yet a candidate.

Lysine-directed covalency is the real bet here, more than the new pocket itself: warhead chemistry and selectivity triage are far less mature than for cysteine acrylamides, so the first thing I'd pull up is proteome-wide lysine reactivity data before trusting the cell activity. The latch site itself is also genuinely new — it sits apart from where SHP099/TNO155-type allosteric inhibitors already bind, so this isn't retreading crowded IP. Getting cell-active hits straight out of fragment screening is a strong signal for this pocket's shape or a hyperreactive lysine — but fragment-to-lead on Lys chemistry means slower SAR cycles than we're used to with Cys warheads. Real value: a fresh chemotype and site for a target whose allosteric space was starting to feel exhausted.

D4 clearing both the oral bioavailability and blood-brain-barrier bars is the real headline here, more than the VHL-EGFR chemistry itself — most VHL degraders are 800-1000 Da PEG-linked bricks, gorgeous in a dish, DOA as an oral CNS drug. That's the developability wall this class has been stuck at. C4's profile reads more like a conventional IV PROTAC: subnanomolar antiproliferative potency and regression at low twice-weekly doses is a solid therapeutic index, but the reduced dosing schedule signals they're still managing catalytic turnover and clearance, not solving it outright. Degrading Del19/L858R instead of inhibiting them sidesteps the secondary-mutation resistance that keeps eroding osimertinib responses. If selectivity holds against WT EGFR in vivo, this is a real next-gen chemotype for NSCLC, not just a mechanistic proof-of-concept.