Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2017 Dec 22;8(1):2276.
doi: 10.1038/s41467-017-02258-w.

Protein Conformational Flexibility Modulates Kinetics and Thermodynamics of Drug Binding

Affiliations
Free PMC article

Protein Conformational Flexibility Modulates Kinetics and Thermodynamics of Drug Binding

M Amaral et al. Nat Commun. .
Free PMC article

Abstract

Structure-based drug design has often been restricted by the rather static picture of protein-ligand complexes presented by crystal structures, despite the widely accepted importance of protein flexibility in biomolecular recognition. Here we report a detailed experimental and computational study of the drug target, human heat shock protein 90, to explore the contribution of protein dynamics to the binding thermodynamics and kinetics of drug-like compounds. We observe that their binding properties depend on whether the protein has a loop or a helical conformation in the binding site of the ligand-bound state. Compounds bound to the helical conformation display slow association and dissociation rates, high-affinity and high cellular efficacy, and predominantly entropically driven binding. An important entropic contribution comes from the greater flexibility of the helical relative to the loop conformation in the ligand-bound state. This unusual mechanism suggests increasing target flexibility in the bound state by ligand design as a new strategy for drug discovery.

Conflict of interest statement

The authors declare no competing financial interests.

Figures

Fig. 1
Fig. 1
Models of drug–target binding. a Schematic diagram of a one-barrier drug–target binding free energy profile. A one-step model with one free energy barrier is used to derive the experimental rate constants. The figure and equations show how the steady-state rate constants relate to the free energy differences shown. The residence time of a drug bound to its target, τ (which is the reciprocal of the rate constant for dissociation of the drug–target complex, k off), results from the “difference” in free energy between the transition state (TS) and the bound ground state (GS), ΔG off. The red arrows indicate that prolongation of the τ can be achieved by stabilizing the GS (increasing the magnitude of ΔG D), destabilizing the TS (increasing ΔG on) or a combination of both (i.e., Koffe-ΔGoffkT=KDe-ΔGonkT). b Diagram schematically illustrating different mechanisms of drug binding involving protein conformational changes. R and RC denote two different conformations of the protein, the latter requires conformational changes for ligand binding. These may occur by conformational selection (blue path) or by induced fit upon formation of an encounter complex [RL]# (red path), or by a combination of the two mechanisms. Binding proceeds through an energetically unfavorable intermediate state (TS in panel A or a local minimum in a 2 (or more)-step binding free energy profile) that, in the conformational selection and induced fit mechanisms, corresponds, respectively, to the R+L or [RL]# state of the system); the final complex is denoted by [RL]. k C/–C and k 2/−2 are the rates of intrinsic and ligand-induced protein conformational transitions, respectively; k 1/−1 and k C1/−C1 are rates of formation of the bound and encounter complexes, assuming that the protein is in conformations R and RC, respectively; k off and k on are experimentally observed off- and on–binding rates. The gray path and third equation describe the pseudo-one-step binding process shown in (a) is used to derive the experimental rate constants
Fig. 2
Fig. 2
Comparison of different conformations of N-HSP90. a Overlay of three N-HSP90 crystal structures in complex with compounds 1 (black), 20 (red) and geldanamycin (wheat, PDB 1YET), representing loop-in, helical and loop-out conformations, respectively. The protein structures are shown in gray except α-helix3 (residues 101–123) and the lid segment (residues 107–141). A detailed view of the different conformations of α-helix3 is given in the inset. b Protein-ligand interactions representative of the loop-in conformation (compound 1). c Protein-ligand interactions representative of the helical conformation (compound 20). Dashed lines indicate interactions (blue: hydrogen bonds, yellow: aromatic, brown: hydrophobic). 2Fo-Fc electron density maps, contoured at 1.5σ, are shown in gray around each ligand
Fig. 3
Fig. 3
Thermodynamic profiles of N-HSP90 inhibitors measured by ITC. The enthalpic and entropic components of the binding free energy are shown in a and b for for WT N-HSP90, and the L107A mutant, respectively. The dashed diagonal line (ΔH=−TΔS) divides the plot into two main areas where the enthalpy (gray) or the entropy (red) dominate the binding free energy (ΔG). Compounds 16 are loop binders and are colored black and compounds 720 are helix-binders and are colored red. c, d Isothermal titration calorimetry fitting curves for compounds 1 (c) and 16 (d) bound to WT N-HSP90 (full circles) and to the L107A mutant (open circles) with thermodynamic parameters shown in the respective tables. e Box plot showing the difference in the enthalpy (ΔΔH), entropy (Δ(−TΔS)) and binding free energy (ΔΔG) for L07A relative to WT for loop- (colored black) and helix-binders (colored red). The boxes denote the 25th and 75th percentiles and the error bars the 5th and 95th percentiles
Fig. 4
Fig. 4
Simulation of the protein and ligand hydration effects. a Relation between the computed desolvation free energy of the inhibitors (see Methods section) and their measured binding entropy in ITC experiments. Compounds assigned as loop-binders are colored black and compounds assigned as helix-binders are colored red. Error bars show the root mean squared error of 3D-RISM predictions against experiment (RMSE=5.4 kJ mol−1 as reported in ref. ). Black and red dashed lines indicate the average values of the desolvation energy and binding entropy for loop- and helix-binders, and the arrows show the corresponding differences between loop- and helix-binding compounds, as observed in experiment (gray) and in computations (light red). b, c Conserved water sites observed in loop-in (b) and helical (c) crystal structures (listed in Supplementary Table 5). The degree of conservation is visualized by increasing size and color; only water sites within 0.8 nm of N106 are shown. In the insets, water sites predicted by GIST are depicted by blue mesh iso-surfaces at a water density value twice that of bulk water; the oxygen atoms of the crystallographic water sites are represented by red spheres; red arrows indicate the positions of stable water sites predicted by 3D-RISM simulations (for details, see Supplementary Information)
Fig. 5
Fig. 5
Simulations of the flexibility of WT-N-HSP90 and the L107A mutant. a Crystal structures of the loop-in (left) and helical (right) conformations with typical crystallographic B-factor values shown by cartoon ribbon radius and color, increasing from blue to red. b Variations of the RMSD of α-helix3 (residues 96–126) along 1 μs MD trajectories in loop-in and helical conformations of several complexes; the bold curves represent cubic splines of the original (gray, light-red, light-green, or light-blue) curves (the corresponding RMSF variations are shown in Supplementary Fig. 6). c Difference in the binding entropy of the helix- with respect to the loop-binders (averaged over complexes with compounds 8, 14, 16, 20, and 1,6, respectively) as observed in experiments (TΔΔS EXP) and in computations (TΔΔS binding) from the sum of the conformational entropy of the α-helix1 and α-helix3 segments (TΔS l-h P obtained using QH and CC-MLA approaches; see Methods section and Supplementary Figs. 7, 8b, and 9e); the binding entropy arising from the ligand and protein motion, but assuming that the protein has the same conformation in apo- and holo-states (TΔΔS, Supplementary Fig. 8b); the ligand desolvation energy (TΔΔS desolv-l, Fig. 4a), and the protein desolvation energy obtained from the analysis of water sites around the flexible region of the α-helix3 in crystal structures (TΔΔS desolv-p). Error bars indicate the standard deviations within an ensemble of complexes
Fig. 6
Fig. 6
Contributions of thermodynamic affinity and association kinetics to the modulation of dissociation rate constants. a-d Logarithmic plots showing correlation of dissociation rate constant k off (x axis) with the association rate constant, k on, and the dissociation constant, K D, (y axis) of compounds 120 determined by SPR for N-HSP90 WT (a, b) and L107A mutant (c, d). Points representing compounds assigned as loop-binders are colored black and compounds assigned as helix-binders are colored red. The black line is the linear regression with R 2 representing the coefficient of determination and R the correlation coefficient. The gray lines represent the 99% upper and lower confidence intervals. The error bars represent the standard deviation of at least three measurements. The red and black shaded regions highlight the different kinetic profiles of helix- and loop-binders, respectively. k off is not strongly correlated either with k on or with K D for N-HSP90 WT. Thus, an increase of residence time is driven by a combination of GS stabilization and TS destabilization. For the L107A mutant, k off is strongly correlated with K D (R=0.69 for WT and R=0.93 for L107A) and not correlated with k on (R=0.48 for WT and R=0.017 for L107A), indicating that residence time is mainly driven by GS stabilization. These relations are shown on the right in schematic pseudo 1-step free energy profiles for the binding reaction of helix- and loop-binders (shown in red and black, respectively; the filled area indicates the energy distribution among the entire compound series) Red and black dashed lines indicate average free energy values for the helix-and loop-binders, respectively. e, f The binding pocket shape observed in the crystal structures of N-HSP90 WT and L107A mutant for loop- and helical- complexes (e, f, respectively). Two alternative conformations observed in the crystal structures are shown for the L107A mutant co-crystallized with compound 6 in e The molecular surface of the protein is colored from red to white indicating increasing hydrophobicity

Similar articles

See all similar articles

Cited by 16 articles

See all "Cited by" articles

References

    1. Tummino PJ, Copeland RA. Residence time of receptor-ligand complexes and its effect on biological function. Biochemistry. 2008;47:5481–5492. doi: 10.1021/bi8002023. - DOI - PubMed
    1. Copeland RA, Pompliano DL, Meek TD. Drug-target residence time and its implications for lead optimization. Nat. Rev. Drug. Discov. 2006;5:730–739. doi: 10.1038/nrd2082. - DOI - PubMed
    1. Maschera B, et al. Human immunodeficiency virus: Mutations in the viral protease that confer resistance to saquinavir increase the dissociation rate constant of the protease-saquinavir complex. J. Biol. Chem. 1996;271:33231–33235. doi: 10.1074/jbc.271.52.33231. - DOI - PubMed
    1. Swinney DC. Biochemical mechanisms of drug action: what does it take for success? Nat. Rev. Drug. Discov. 2004;3:801–808. doi: 10.1038/nrd1500. - DOI - PubMed
    1. Copeland, R. A. Evaluation of Enzyme Inhibitors in Drug Discovery: A Guide for Medicina Chemists and Pharmacologists. (Wiley, New York, 2013). - PubMed

Publication types

LinkOut - more resources

Feedback