Molecular Determinants of Host–Guest Retention: A Pose-Resolved Molecular Dynamics Study of Celecoxib in β-Cyclodextrin

Abstract

Getting hydrophobic drugs to dissolve in water is not just a technical hurdle. It’s often the make-or-break factor in formulation. That’s where cyclodextrin-based host–guest systems come in, and among them, β-cyclodextrin (β-CD) has earned its place. Its truncated-cone shape hydrophobic cavity inside, hydrophilic surface outside creates a natural “pocket” for poorly soluble guests, and the resulting non-covalent complexes can meaningfully improve dissolution profiles. Celecoxib was selected as the test case, not because it’s easy (it is not; its water solubility is notoriously low), but because it’s a well-documented COX-2 inhibitor that has already been studied from multiple angles. Even so, the literature does not quite agree on how stably it sits inside β-CD. Small details, initial orientation, and solvation setup seem to sway the outcome, and in our experience, that sensitivity is easy to overlook if you rely on a single starting pose. That is exactly what happened in our first run (ISA), hereafter referred to as V3: a 5-ns simulation launched from one docking configuration, and within nanoseconds, celecoxib was already drifting away. At first glance, you might write it off as weak binding. But we wondered: what if the geometry, not the pairing itself, was the issue? To test that, we used AutoDock Vina to generate several plausible poses, scored them both structurally and energetically, and picked Pose 2 as the most promising candidate. From there, CHARMM-GUI helped us assemble a fully solvated, charge-neutral system, which we then simulated in GROMACS minimization first, then NVT and NPT equilibration, followed by a 5-ns production run. The contrast was hard to miss. Pose 2 held together: host and guest remained associated, with a center-of-mass (COM) distance averaging 0.332 ± 0.039 nm across 51 frames. ISA, meanwhile, continued to fall apart. What we take from this, and it aligns with earlier suggestions by others, is that celecoxib’s retention is not an all-or-nothing trait. It is conditional, emerging only when steric complementarity, van der Waals contacts, electrostatic alignment, and solvent-driven hydrophobic effects happen to reinforce one another. Practically speaking, that means one pose rarely tells the whole story. Validating multiple starting configurations in explicit-solvent MD is not merely prudent; in our view, it’s essential for reliable predictions. We hope the workflow outlined here, systematic, reproducible, and grounded in physical chemistry, can serve as a practical template for others working on cyclodextrin-based formulation design.