Daily digest — 2026-09-17

Three degron classes, one receptor — that's the headline for degrader hunters tired of CRBN/VHL. The Kelch channel reading C-terminal Arg is the most druggable-looking piece: Kelch domains (think KEAP1) have a track record of taking small, polar ligands into narrow grooves, so a compact peptidomimetic anchor there feels plausible. The acetyl-N-degron tunnel is murkier — adaptor-dependent, no small molecule has touched an acetyl-N-degron pocket yet, so that's fragment-screening territory, not a near-term glue. Biggest catch: MKLN1 self-dimerizes through the same discoidin-Kelch interface, meaning any exploiting ligand has to outcompete an endogenous, high-avidity PPI — selectivity and target engagement won't be trivial. Pure structural biology here, no chemical matter yet, but a legitimate new hub to diversify the E3 ligase toolbox beyond the same two receptors everyone's already crowded into.

The caged VHL ligand is the clever bit — redox-triggered decaging means the warhead sits as a prodrug until it hits oxidative-stress tissue, which could finally address the peripheral-toxicity problem that's dogged systemic PROTAC dosing. I want the counter-screen though: how much decaging leaks in non-lesion CNS tissue over chronic dosing, and does 'activated DAPK1' selectivity hold in human proteoform or is that mouse-conformer-specific? The macrocyclic peptide-small molecule hybrid is a brutal synthesis — SPPS, macrocyclization, PROTAC conjugation, yield bleeding at every step — this is proof-of-concept, not a scalable series. Exploiting age-upregulated transcytosis instead of fixed RMT receptors (transferrin, insulin) is the real advance over TfR-shuttle degraders; the open question is whether that route is druggable across disease stages, not just aged models.

H-phosphonate diesters as bench-stable P(III) surrogates is the headline: no more babysitting moisture-sensitive phosphitylating reagents under inert atmosphere just to install a novel backbone linkage. Backbone modification has been the stubborn bottleneck in oligo SAR — everyone can vary the sugar or the base, but stepping outside phosphorothioate/phosphorodiester space usually means bespoke phosphoramidite synthesis for every single analog. Pre-made dinucleotide blockmers flip that: build a small library of linkage types once, then mix-and-match them like ordinary phosphoramidites on standard solid-phase cycles, placing modifications at any position in the sequence. For a team running iterative backbone SAR, that's the difference between a multi-week custom-monomer detour and a same-week analog set. Real question is whether coupling efficiencies and diastereomer control hold up across the full modification set, not just the showcase examples.

Maleimide-cysteine chemistry's dirty secret is retro-Michael instability — payload leaches back into circulation, and DAR isn't as fixed as the spec sheet claims. This bifunctional pyridine/thianthrenium scaffold attacks that head-on: the resulting aryl-thioether is non-chiral and hydrolytically stable, so no diastereomer mixtures and no slow linker exchange with serum thiols. Sequential chemoselectivity — amine coupling first, cysteine second, off one handle — is a real practical upgrade for installing orthogonal payloads/tags, not just another NHS-ester variant. What earns my attention is that they didn't stop at peptide-stapling demos: a tunable-DAR HER2 ADC analog of MRG002 with improved serum payload retention versus conventional linkage is the kind of proof point that actually moves a methodology from academic curiosity toward something worth piloting. Thianthrenium reagent scale-up and cost will decide if it ever displaces maleimide on a GMP line.

A noninhibitory DUB ligand is a harder ask than it sounds - you need a binder that occupies USP7 without touching the catalytic cleft, so a standard activity-inhibition screen is the wrong tool entirely. That's what makes the on-bead, proximity-driven functional assay the real story here: it's a hit-finding method, not just a molecule, and if it generalizes across other DUBs it addresses the actual bottleneck in this space. DUBTACs have quietly been the underdeveloped mirror image of degrader chemistry for years - everyone chasing PROTACs, almost nobody with tractable DUB recruiters in hand. WJ045 stabilizing CFTR is a nice proof point tying this to real biology, though it's a single cellular readout so far - want to see USP-family selectivity and whether the platform yields a second, unrelated ligand.

WBP4 is a spliceosome scaffold with no ligandable pocket — this isn't a kinase-with-a-degron-tail story, it's proof CRBN can be steered onto a genuinely undruggable surface via a G-loop contact. That binding hypothesis is worth stealing for other scaffold targets sitting in my own tractability triage pile. The selectivity data is the real flex: clean proteome-wide profile AND minimal splicing perturbation, exactly where glues usually get you in trouble (neosubstrate promiscuity, cryptic RBM39-style splicing hits). Getting both together is rare. wTAG might matter more than the degrader itself — a 41-residue portable degron with no hook effect is a cleaner dTAG-class tool, and I'd rather have that in my assay toolbox than another single-target lead.

Solid probe engineering, not just a screening reagent: coaxing ER-selective localization and a large Stokes shift out of the HDCI scaffold gives clean signal separation in situ — photophysics that holds up in cells, not just a cuvette. The real value-add is methodological. A mechanistic imaging tool doubling as an HTS readout turns hCES2A activity into a screenable phenotype in living systems, closing the usual gap between biochemical assays and cellular reality. Target-wise, nothing novel here — hCES2A is a known off-target liability that bioactivates irinotecan into its gut-toxic metabolite, so this is a 'fix the side effect' play, not a new mechanism. A4 is an HTS-derived covalent hit with encouraging mouse data, but still needs the standard covalent-hit workup: reactivity profiling, esterase-family selectivity, and SAR.