Questions: Structure-Based Drug Design

Question 1

A drug designer has a crystal structure of a target protein with a lead compound bound. The structure shows that the compound's methyl group sits in a larger hydrophobic pocket with unfilled space. What optimization strategy does this suggest?

• A. Remove the methyl group to reduce molecular weight
• B. Replace the methyl group with a larger hydrophobic group (ethyl, isopropyl, or cyclopropyl) that better fills the pocket — improved shape complementarity with the pocket increases van der Waals contacts and binding affinity
• C. Add a charged group to the methyl position to form a salt bridge
• D. This observation has no implications for drug design

Answer: (B)

Filling unfilled hydrophobic pockets is one of the most reliable strategies for improving binding affinity in SBDD. The larger group increases van der Waals contacts with the pocket walls and may also improve the entropy of binding by displacing ordered water molecules from the pocket. The structural data directly suggests the modification and predicts which groups will fit. This 'grow into the pocket' strategy has been used successfully in the optimization of many drug candidates, and the co-crystal structure confirms whether the modified compound adopts the predicted binding mode.

Question 2

Virtual screening using molecular docking can reliably predict the binding affinity of any compound to any target with quantitative accuracy.

• T. True
• F. False

Answer: False

Molecular docking is good at predicting binding poses (where and how a molecule binds) but poor at quantitatively predicting binding affinity (how tightly it binds). Scoring functions used in docking are approximate — they estimate binding energy from simple terms (shape complementarity, hydrogen bonds, electrostatics) but miss important contributions (entropy changes, water displacement, protein conformational change, strain energy). Docking is most useful for virtual screening (enriching a library for likely binders from among millions of compounds) and for predicting binding mode, not for ranking compounds by affinity. More rigorous methods (free energy perturbation, MM-GBSA) provide better affinity predictions but at much higher computational cost.

Question 3

Why is the iterative cycle of structure determination, design, synthesis, and testing essential in SBDD rather than a single round of computational design?

Model answer: Computational predictions — docking scores, predicted binding modes, designed compounds — are approximations that frequently differ from experimental reality. The protein may adopt a different conformation upon binding (induced fit), the designed compound may bind in an unexpected orientation, water molecules may mediate interactions not captured by the computational model, or the compound may have unfavorable properties (solubility, metabolic stability) not predicted from the structure alone. Each round of co-crystal structure determination reveals these discrepancies, enabling corrections in the next design cycle. The iterative cycle converges on potent, selective compounds by alternating between structure-guided hypothesis and experimental validation — typically requiring 3-10 cycles for lead optimization.

The HIV protease inhibitor program at Merck (leading to indinavir/Crixivan) exemplified this iterative cycle: each co-crystal structure revealed unexpected binding features that guided the next round of medicinal chemistry optimization, ultimately producing a potent, orally bioavailable drug after multiple structure-guided redesign cycles.