First covalent PARG inhibitors join a widening toolkit for hard drug targets

A new way to hit PARG

The most striking result of the week is a first: in a preprint, researchers report the discovery of the first covalent PARG inhibitors using BCB-based warheads. Covalent inhibition of PARG is a viable therapeutic strategy, the preprint reports, and Cys872 is a tractable site for selective covalent targeting in PARG. BCB-hydrazide warheads enable selective covalent cysteine targeting, and FORX-425 exhibits nanomolar cellular activity in PARG-sensitive cancer cell lines. PARG is an enzyme involved in DNA repair, and a covalent inhibitor is one that forms a lasting chemical bond with its target rather than binding and releasing. The significance of a first covalent inhibitor is not that it is a drug, it is that it proves the target can be hit this way at all. A warhead is the reactive part of the molecule that forms the bond, and a new warhead class that works selectively is a tool other chemists can now pick up for other targets. The work is a preprint, not peer-reviewed, and the cellular activity is in cancer cell lines, not in patients, but the claim is precise: one site, one warhead class, one compound with nanomolar activity. m:…3160

The second paper in the week's toolkit is about a different bottleneck: what to do when a promising scaffold cannot be developed further. Scaffold hopping is the practice of replacing a molecule's core structure while keeping the parts that bind the target, and the preprint describes a docking-free method called SV-Hop that does it by conserving contacts rather than by simulating docking. SV-Hop achieves a 62.1% success rate in scaffold-held-out references, outperforming property-based methods at 50.3% and 2D similarity methods at 44.3%. The docking-free contact-conservation criterion recovers 80.6% of reference contacts across 40 crystal pairs. A 388-rule PDB-only lexicon reaches 81.9% success on target-held-out splits, outperforming the full 60,069-rule lexicon at 65.3%. The counterintuitive result is the last one: a smaller rulebook built only from deposited crystal structures beats the full rulebook. For a medicinal chemist, the practical read is that contact conservation, which parts of the molecule touch the target, is a stronger guide to a replacement scaffold than either similarity or docking scores, and that a curated set of rules can outperform a comprehensive one. m:…4276

Drugs designed to disappear

The third paper tackles the opposite problem: a drug that works but should not linger. In a preprint on soft-drug design for topical JAK1/TYK2 inhibition in atopic dermatitis, the authors define a target profile for soft JAK inhibitors with MW 420, logD 3.2 and plasma esterase clearance of 300 L/h, which achieves a skin partitioning index of 174. A soft drug is designed to act where it is applied and then be cleared quickly if it reaches the bloodstream. PBPK modeling predicts a plasma AUC of 20 µg·h/L for the target soft-drug profile, compared to 273 µg·h/L for ruxolitinib, which is the existing drug used as the comparator. The gap between those two numbers is the whole argument: similar action on the skin, far less exposure in the rest of the body. A five-step validation chain including a PON1-clearance gate is required to ensure systemic safety for topical JAK inhibitors, the preprint says, which is a reminder that a soft drug is not simply a weaker drug, it is a drug with a designed exit route, and the exit route has to be tested as carefully as the activity. m:…4611

The week's remaining two papers show the funnel working at both ends. In a preprint on USP5, a computational workflow generated 3,264 unique structures from 10 chemical-enumeration methods, and P06 is the top-ranked candidate for USP5 catalytic-cleft inhibition based on consensus analysis across SP, XP and MM-GBSA evaluations. P09 exhibited the strongest mean MM-GBSA binding energy among the tested panel, and P03 achieved the most favorable mean XP docking score. These are computational rankings, not experimental results, and the three different leaders across three different scores are themselves informative: no single candidate wins on every measure, which is why the authors use consensus. Separately, a preprint on triangulenium probes reports that the DM4 derivative exhibits increased fluorescence intensity and lifetime upon G4 DNA/RNA binding, that a semi-automated D2B workflow allows synthesis and evaluation of fluorescent probes directly from reaction mixtures, bypassing purification, and that DM4 demonstrates good cell permeability and low cytotoxicity in live-cell imaging. A probe that lights up when it binds a DNA structure, and can be made and tested without purification, is a tool for watching biology happen rather than a drug candidate. Together the five papers describe a field adding tools at every stage: a new warhead, a new way to swap scaffolds, a drug designed to clear, a screen that generates thousands of candidates, and a probe that lets researchers see what they are doing. m:…7328 m:…9585