**5. Conclusions**

The notion of the binding event being the result of shape complementarity between ligand and protein binding site (key-and-lock model) has been a paradigm in the description of binding events and molecular recognition phenomena for a long time. The recent discovery of the important role played by protein dynamics and solvent effects, as well as the enthalpy-entropy compensation phenomenon, challenged this concept, and demanded the thorough examination of entropic contributions and solvent effects. Assessment of all these contributions to the thermodynamics of ligand-protein binding is a challenging task. Although understanding the role of each contribution and methods allowing for a complete dissection of thermodynamic contributions are tasks far from being completed, significant progress has been made in recent years. For instance, development of high-resolution heteronuclear NMR methods allowed for assessment of the contribution from protein degrees of freedom to the intrinsic entropy of binding. The usefulness of such approach has been demonstrated in the course of this chapter on several ligand-protein examples. In addition, progresses in the development of MD-related methodologies and advanced force fields enabled the application of the NMR-derived formalism on relevant time scales and the assessment of the intrinsic entropic contributions solely using computational methods. Development of QM methods allows the study of larger and larger systems, while advances in ITC calorimetry allow the use of very small amounts of reagents for a single experiment. Despite this progress, much remains to be done. The enthalpy-entropy compensation

phenomenon seems to be widespread among ligand-protein systems. It seems universal: binding restricts motions, while motions oppose tight confinement. However, our current knowledge about intrinsic protein dynamics is still insufficient to allow us to predict this phenomenon and hence to exploit it for the purposes of rational molecular design. Another challenge lies within the quantification of solvation contributions. There seem to be conflicting data regarding the contributions from confined water molecules. Their influence on binding can be favourable or unfavourable, enthalpy- or entropy- driven. Bound water molecules can be released upon ligand binding or – on the contrary – bind tighter (Poornima CS and Dean, 1995a-c). Their presence can make the protein structure more rigid (Mao *et al.*, 2000), or more flexible (Fischer and Verma, 1999). Finally, protein binding sites can be fully solvated prior to binding, or fully desolvated (Barratt *et al.*, 2006, Syme *et al.*, 2010). The only common feature that seems to exist is that the contribution of the solvation effects to the ligand-protein binding thermodynamics can be – and often is – significant.

Last but not least, intrinsic entropic contributions are notoriously difficult to quantify. A handful of experimental and theoretical methods can be employed to quantify these contributions, as have been described. However, all of these methods have their limitations, and one should be aware of these and of the assumptions that are being made. Theoretical results should be treated with caution, experimental data likewise, as they are based on many approximations and heavily dependent on the conditions applied. Care must be taken not to over-extrapolate data, and not fall the victim to confirmation bias.

Fundamentally, in order to predict free energy of binding accurately, it would be necessary to go beyond predicting a single 'dominant' conformation of the ligand-protein complex. It should be emphasised that the overall shape of the free energy landscape controls the binding free energy. This shape is affected by the depth and width of the local minima, and the height and breadth of the energy barriers. The factors that shape that landscape include intrinsic entropic contributions of both interacting partners, ligand poses, protein conformations, solvent effects, and protonation states. Computational and experimental approaches combined together can provide insight into this crucial but otherwise hidden landscape, which is pivotal not only to understand the origin of each contribution and its role in the binding event, but which can allow a truly rational molecular design.
