Nearly one in ten human cancers carries a mutation that disables ARID1A, a chromatin-remodeling protein critical for suppressing tumor growth. When ARID1A is lost, cancer cells become unusually dependent on its close structural relative, ARID1B — a vulnerability known as synthetic lethality. Targeting ARID1B selectively could, in principle, kill those cancer cells while leaving normal tissue unharmed. The central obstacle has been that ARID1B's surface, at least as revealed by conventional structural analysis, appears frustratingly smooth and featureless — offering no obvious pocket where a small molecule could bind with selectivity.
Researchers publishing in PNAS used a combination of computational and biophysical techniques to identify so-called cryptic binding sites within the ARID1B DNA-binding domain — pockets that are largely hidden in the protein's resting conformation but become accessible during natural conformational fluctuations. These transient structural openings, invisible to standard static crystallography, were mapped and characterized as genuine small-molecule binding candidates rather than artifacts of modeling, representing a meaningful advance in defining the druggable landscape of this historically intractable target.
This finding matters because synthetic lethality in ARID1A-mutant cancers has been an appealing but chemically underexplored therapeutic concept for over a decade. The SWI/SNF chromatin-remodeling complex — of which both ARID1A and ARID1B are components — has proven notoriously difficult to drug, largely due to protein-protein interaction surfaces and DNA-binding domains that lack conventional small-molecule pockets. The cryptic site approach, pioneered most visibly in oncology by covalent KRAS inhibitors, is emerging as a legitimate strategy for similarly flat or featureless targets. The current work is early-stage structural biology, not a clinical candidate, and cryptic sites identified computationally still face substantial hurdles — including whether hit compounds can achieve sufficient affinity, selectivity, and cellular penetration. Nonetheless, this represents a potentially important framework shift: from treating ARID1B as undruggable to actively mapping its exploitable geometry.