A DNA-encoded library uncovers a new way to drug NOS1AP in Alzheimer’s disease
A preprint from Columbia investigators reports a different strategy for Alzheimer’s disease: instead of blocking NOS1AP, they designed a small molecule that stabilizes one of its protein interactions. The work centers on a benzimidazole-5-carboxamide scaffold identified through affinity selection of a 1.4-billion-member DNA-encoded library and then refined into a compound called NOSTRA. In the authors’ hands, the molecule bound NOS1AP with submicromolar affinity and behaved as a first-in-class molecular glue for the NOS1AP-SCRIB complex.
The concept is important because NOS1AP sits in a network of neuronal protein partners implicated in Alzheimer’s biology. Many drug-discovery programs try to prevent protein-protein interactions from forming. Here, the direction was reversed. A small molecule was used to make one interaction stronger, not weaker.
Selective reinforcement of SCRIB binding drove the biochemical signal
NOSTRA did not behave like a broad-spectrum stickiness agent. Of six tested protein interactions, it enhanced only SCRIB binding, and the reported cooperativity factor reached 30. The kinetic effect leaned heavily on slower dissociation rather than faster association: the compound mainly prolonged the lifetime of the NOS1AP-SCRIB complex by a 14-fold reduction in dissociation rate. Native mass spectrometry then resolved a ternary NOS1AP:NOSTRA:SCRIB complex, giving direct structural support to the interaction-stabilization model.
That kinetic profile matters at the bench and at the bedside. A molecule that keeps a complex together longer can shift downstream signaling without requiring high occupancy at every moment, and that can be particularly relevant in the brain where exposure, selectivity, and tolerability all constrain therapeutic design.
The study’s language is careful about mechanism, and the biology follows that caution. NOS1AP is an adaptor for neuronal nitric oxide synthase, and the SCRIB interaction appears to sit within a signaling axis capable of altering neuronal behavior when stabilized. The report does not claim that the entire Alzheimer’s cascade has been solved. It does show that one protein complex can be pharmacologically strengthened in a controlled way.
Patient-derived neurons showed biomarker shifts without overt toxicity
The functional experiments moved beyond purified proteins and into human neurons. In healthy human iPSC-derived cortical neurons, NOSTRA engaged the endogenous complex and activated Rac1 without cytotoxicity. In Alzheimer’s patient-derived neurons, the compound lowered the Aβ42/Aβ40 ratio and reduced p-tau217. It also reduced neurofilament light, a marker tied to neuronal injury, and improved network activity.
Those changes are notable because they span multiple disease-relevant readouts rather than a single assay endpoint. A lower Aβ42/Aβ40 ratio suggests a shift in amyloid processing or handling, while reduced p-tau217 points toward a parallel effect on tau-related pathology. Neurofilament light and network activity extend the observation into neuronal health and function, which is the level at which Alzheimer’s disease ultimately matters clinically.
One sentence captures the central tension in the preprint: a chemistry program aimed at a protein-protein interface produced biomarker movement in diseased human neurons.
CNS exposure potential and what still needs confirmation
The compound’s in vitro ADME profile was described as consistent with central nervous system exposure, an essential requirement for any candidate aimed at a neurodegenerative target. Even so, the current evidence stops at preclinical cellular systems. No animal efficacy data, human pharmacokinetics, or clinical outcomes are included in the material provided here, and the findings remain unreviewed as a bioRxiv preprint.
The translational promise lies in the target class as much as the molecule itself. Protein-protein interactions have long been considered difficult to drug, especially in neuroscience, where selectivity and brain exposure are unforgiving constraints. A molecular glue offers a different pharmacologic logic: not simply occupying an active site, but reshaping a native complex in a direction that favors a desired signaling state. If that can be replicated in vivo, the approach could expand how researchers think about NOS1AP and similar adaptor proteins.
Yet the same features that make the work intriguing also limit how far it can be pushed today. The experiments were performed in cell-based systems, and the disease biology of Alzheimer’s in patients is broader than any one neuronal pathway. Biomarker improvements in cultured neurons do not guarantee durability, safety, or clinical benefit in the human brain. The next steps will have to establish whether the NOS1AP-SCRIB interaction can be modulated in vivo at exposures that remain compatible with long-term CNS dosing.
Even with those constraints, the preprint offers a concrete pharmacologic advance. It identifies a small molecule that stabilizes a previously unexploited protein complex, links that stabilization to Rac1 activation, and connects the signal to disease-relevant neuronal readouts in Alzheimer’s patient-derived cells. The central unanswered issue is whether this molecular glue can preserve those effects in living systems without losing the selectivity that made it attractive in the first place.