Can you pick a lock with loose-fitting keys?

A systematic study reveals to which extend protein-ligand complexes form robust structures

Like Comment
Read the Paper

Emil Fischer’s lock-and-key is a well-known analogy to explain the specific recognition of small molecules (e.g. substrates, drugs) by proteins. The analogy is, of course, an oversimplification because proteins are malleable entities that can take multiple forms, adapting to differently-shaped ligands. Rivers of scientific literature have debated how the flexibility of the protein impacts the process of ligand binding.1,2 By contrast, there is little debate about what happens after the complex is formed. A commonly held view is that protein and ligand become interlocked by the chemical complementary of their respective features, forming a rigid complex. Such tight fit would be ideal for selectivity, but in the microscopic world, anything that becomes static pays a hefty entropic cost. So, are protein-ligand complexes really rigid? Some studies indicate that ligands retain some degree of residual mobility upon binding.3,4 But how general is this? A fascinating study demonstrated that two proteins can form a high affinity complex in the absence of structure.Could, in a similar way, a ligand populate the binding site of a protein without forming a well-structured complex?

We came to wonder about these questions after discovering that structural robustness of protein-ligand complexes can be used to discriminate between true binders (robust) and decoys (labile).This was an unexpected finding because robustness (defined as the free energy required to push a particular interaction away from its position of equilibrium) bears no relationship with any macroscopic observable (Figure 1). The method proved very effective as a virtual screening tool, but in the initial study we were only probing a single key interaction (a conserved hydrogen bond) that had to be predefined for each individual protein. Hence, we wanted to know if robustness is a universal property of protein-ligand complexes, and whether it concentrates on such key interactions or spreads across the protein-ligand interface. Using Dynamic Undocking (DUck), a tailored solution based on Molecular Dynamics, here we pull 469 individual interactions (Figure 2) in a diverse set of 106 protein-ligand complexes. 

Figure 1
Two hypothetical complexes with identical macroscopic observables KD, k1 and k-1 (related to ΔGBIND, ΔG*ON and ΔG*OFF, respectively) but different structural stability. The loose-binding compound (dotted line) has access to a much larger configurational space in the bound state (blue-shaded area) than the tight-binding one (solid line and green-shaded area).
Figure 2
Schematic representation of a protein-ligand complex (blue and grey areas, respectively). We have applied the DUck methodology to pull each and every hydrogen bond in the complex (red arrows), always considering the local environment provided by the protein.

We find that most complexes combine a tight fit region, which anchors the ligand to the protein, with other regions that are not structurally constrained. This may provide an optimal balance between the need for selectivity and a reduced entropic cost. Interestingly, only a small proportion of systems conform to the lock-and-key paradigm. Such static structures appear to fulfil a functional need, which may justify the higher entropic penalty. Intriguingly, some pairs lack any identifiable anchoring points. This can occur even in high-affinity complexes. Further studies will be necessary to fully understand the energetic and selectivity consequences of such loose interactions. But, considering that all our test cases originate from X-ray crystallography (a technique that works best for static systems) it should not come as a surprise if the actual proportion of loose-fitting ligands is significantly larger than what we have found here. 

From a practical standpoint, our results indicate that, for the majority of cases, structural robustness can be used as a guide in structure-based drug design. This not only explains our previous success in virtual screening, but also unlocks additional applications that we will be investigating in the future. 

(1)     Wei, G.; Xi, W.; Nussinov, R.; Ma, B. Protein Ensembles: How Does Nature Harness Thermodynamic Fluctuations for Life? The Diverse Functional Roles of Conformational Ensembles in the Cell. Chem. Rev.2016116, 6516–6551.

(2)     Stank, A.; Kokh, D. B.; Fuller, J. C.; Wade, R. C. Protein Binding Pocket Dynamics. Acc. Chem. Res.201649, 809–815.

(3)     Klebe, G. Applying Thermodynamic Profiling in Lead Finding and Optimization. Nat. Rev. Drug Discov.201514, 95–110.

(4)     van Zundert, G.; Hudson, B. M.; de Oliveira, S.; Keedy, D. A.; Fonseca, R.; Heliou, A.; Suresh, P.; Borrelli, K.; Day, T.; Fraser, J.; van den Bedem, H. QFit-Ligand Reveals Widespread Conformational Heterogeneity of Drug-like Molecules in X-Ray Electron Density Maps. J. Med. Chem.201861, 11183–11198.

(5)     Borgia, A.; Borgia, M. B.; Bugge, K.; Kissling, V. M.; Heidarsson, P. O.; Fernandes, C. B.; Sottini, A.; Soranno, A.; Buholzer, K. J.; Nettels, D.; Kragelund, B. B.; Best, R. B.; Schuler, B. Extreme Disorder in an Ultrahigh-Affinity Protein Complex. Nature2018555, 61–66.

(6)     Ruiz-Carmona, S.; Schmidtke, P.; Luque, F. J.; Baker, L.; Matassova, N.; Davis, B.; Roughley, S.; Murray, J.; Hubbard, R.; Barril, X. Dynamic Undocking and the Quasi-Bound State as Tools for Drug Discovery. Nat. Chem.20179, 201–206.

Xavier Barril

ICREA Research Professor, University of Barcelona

No comments yet.