Abstract B088: A cryptic alternative bound state of HDAC6 drives mutation- and ligand-dependent remodeling of inhibitor interaction
HDAC6-selective hydroxamate inhibitors, including ricolinostat and citarinostat, have advanced clinically in multiple myeloma and other malignancies, yet resistance mechanisms remain structurally unresolved. We hypothesized that evolutionarily tolerated mutations identified by a protein language model selectively perturb inhibitor binding through access to alternative conformational states. Per-position log-likelihoods from the ESM2 protein language model across the HDAC6 CD2 domain nominated ten candidate resistance mutations. Each mutant and the wild-type tandem was simulated in complex with five hydroxamate inhibitors across three docking-derived poses (159 × 100-ns all-atom MD trajectories; GROMACS; ff14SB/GAFF2; bonded Zn2+). Paired within-system analysis of matched receptor–ligand conditions isolated state-dependent effects. We identified a kinetically stable alternative bound state in which the catalytic histidine H651 is displaced from canonical Zn2+ coordination. The H651 ND1–Zn2+ distance is bimodal (2.16–2.56 Å vs. 2.83–3.34 Å) with no interconversion on the 100-ns timescale, indicating a high barrier and early pose-dependent state selection. The displaced state involves primarily whole-residue backbone translation (0.6–1.8 Å Cα shift) with partial imidazole counter-rotation, not a simple rotamer flip. It is accessed by both active-site-proximal and distal mutations up to 23 Å from H651, demonstrating long-range structural coupling and indicating that mutations shift access probability rather than state identity. The displaced conformation consistently rewires protein–ligand interactions, replacing canonical active-site hydrogen bonds with alternative contacts in the L1 and L7 loops. Critically, bulky-cap hydroxamates (cay10603, citarinostat, ricolinostat) preferentially occupy the displaced state (11 of 12 trajectories), whereas smaller-cap inhibitors largely do not, implicating cap geometry as a determinant of conformational basin selection. These findings reveal a cryptic, mutation-accessible conformational state of HDAC6 that fundamentally alters inhibitor binding. Resistance emerges not from discrete structural disruptions but from shifts in the probability of accessing alternative binding geometries. This mechanism challenges single-structure drug design paradigms and suggests that HDAC6 inhibitor optimization—and resistance prediction—must explicitly account for multiple bound states. The integrated language-model/MD framework provides a general strategy for identifying resistance-relevant conformational landscapes in Zn2+-dependent oncology targets. AI Disclosure: Generative AI assisted in drafting and editing the text of this abstract. Paul Fong, Fan Yuan. A cryptic alternative bound state of HDAC6 drives mutation- and ligand-dependent remodeling of inhibitor interaction [abstract]. In: Proceedings of AACR Drug Discovery and Development (AACR D3) Conference; 2026 Jul 21-24; Boston, MA. Philadelphia (PA): AACR; Clin Cancer Res 2026;32(14_Suppl):Abstract nr B088.