Chapter 26Ligand Configurational Entropy

To understand protein and ligand flexibility, it is necessary to have a basic grounding in statistical mechanics and the theory of entropy. This chapter presents work on the characterization of the components of ligand conformational entropy [1] and its implications for protein design and drug design. This chapter builds on the understanding of free energy developed in chapter 25.

Experimental Input

This method in this chapter is not heavily dependent on experimental data. The required inputs consist of structures of a ligand of interest and of the protein it binds to. Standard energy functions are used with mathematical decompositions to characterize changes in entropy on binding based on these structures. The structures used are already solved, so the raw NMR or X-ray data used in structure determination is not considered by the technique in this chapter.

Entropy

Entropy, in the broad sense, reflects the thermodynamic favorability of disorder in a system. The familiar Gibbs free-energy equation (chapter 25) shows this effect:

ΔG=ΔH−TΔS.(26.1)\Delta G = \Delta H - T \Delta S. \quad (26.1)

Thus, a change to a more disordered state (positive ΔS\Delta S) results in a more negative, more favorable ΔG\Delta G . One interpretation of disorder is the multiplicity WW of a system, that is, the number of states readily available to the system. A toy example of multiplicity is a standard 6-sided die. For such a die, W=6W = 6 is the number of states possible, that is, the number of possible rolls. From the multiplicity WW of a system, we can determine a measure of the system’s entropy:

S=−Rln⁡(W),(26.2)S = -R \ln(W), \quad (26.2) where RR is the universal gas constant. If not all states of a system are equally accessible or equally populated, a modified version of equation (26.2) is used:

S=−R∑jpjln⁡(pj),(26.3)S = -R \sum_j p_j \ln(p_j), \quad (26.3) where pjp_j is the probability of state jj being occupied, and the summation is over all states.

Thus, the value of ΔS\Delta S can be determined by comparing multiplicity-derived entropies before and after a reaction.

Entropy in Ligand Binding

Ligand configurational entropy is an entropy value derived from the total multiplicity of a ligand molecule. Configurational entropy can be considered as the combination of two other types of entropy: conformational entropy and vibrational entropy.

Conformational Entropy

Conformational entropy has also been covered, in some sense, in our chapters on K*K^* , belief propagation, and rotamer search/pruning. Conformational entropy reflects the number of conformations that a molecule can assume. In proteins, these conformations are often represented with rotamer libraries, reflecting the rotations that side-chains can undergo. Ligands may be treated similarly, with flexible functional groups or other components. In terms of an energy landscape, conformational entropy reflects the number of wells or minima.

Conformational entropy is lost on binding when proximity to a bound molecule causes interactions (most often, steric clashes) that prevent previously available conformations from being assumed. It has often been assumed that this is the main contributor to entropy change on binding.

Vibrational Entropy

Vibrational entropy reflects the motions available, not to side-chains or functional groups, but to a molecule as a whole or to the individual atoms that compose it. Vibrational entropy can be further subdivided. Rotation/translation accounts for motions of the entire molecule. Torsion, angle bending, and angle stretching account for motions of pairs of individual atoms. These latter three entropies may be recognizable as motions important in infrared (IR) spectroscopy. In terms of an energy landscape, vibrational entropy reflects the width of wells or minima.

Vibrational entropy is lost on binding when proximity to a bound molecule causes interactions (most often steric clashes) that restrict previously possible motions. While rotation and translation are clearly restricted on binding, the entropy loss from atomic motions is less intuitive.

Entropy and Amprenavir

Gilson and coworkers [1] calculated the entropies for amprenavir, an inhibitory drug molecule that binds to HIV protease. They calculated its overall loss of configurational entropy on binding to HIV protease and the individual components of that entropy as discussed earlier. (For the mathematical decomposition of entropy components, see the appendix of [1].)

The authors determined the overall configurational entropy loss on binding to be 26.4 kcal/mol. Of this, at most 4.1 kcal/mol can be accounted for by conformational entropy loss. While 4.1 kcal/mol is not negligible (the free energy difference between a folded and an unfolded protein being typically approximately that value), it is less than a sixth of the total entropy lost. Therefore, vibrational entropy appears to have an unexpectedly dominant role in binding.

The authors also found that the individual components of vibrational entropy (rotation/translation, torsion, bending, stretching) summed to more than the total vibrational entropy. They conclude that there is crosstalk among these components. This means that certain combinations of vibrational motions are allowed or disallowed together, which is reasonably intuitive.

Table 26.1 is taken from Gilson and coworkers [1]. Here the authors presented the correlations among components of vibrational entropy, as determined by a quasiharmonic analysis (essentially a molecular dynamics approach). Rotation/translation, torsion, and angle bending are all fairly strongly correlated to each other. Bond stretching, which contributes vastly to total entropy, is almost unaffected by binding, and is less strongly correlated to the other components.

Implications for Design

Entropy loss on binding is of interest to drug designers because a negative ΔS\Delta S value in Eq. ( 26.1 ) renders the reaction, in this case inhibition of a target protein, less favorable. Understanding the components of the inevitable loss in entropy may allow the design of drugs that minimize that loss.

Table 26.1 Decomposition of amprenavir’s change in configurational entropy upon binding, calculated via submatrices of the covariance matrices of the dominant free and bound energy well

R/T Tors Angle Stretch
Free
R/T -7.07 0 0 0
Tors 54.5 3.6 0.1
Angle 78.1 1.0
Stretch 109.52
Bound
R/T 7.02 1.5 0.8 0.08
Tors 61.6 4.1 0.13
Angle 80.1 1.2
Stretch 109.55
Difference
R/T 12.3 1.5 0.8 0.08
Tors 7.1 0.5 0.03
Angle 2.0 0.2
Stretch 0.04

Entropy values multiplied by −T-T to yield free energy contributions in kcal/mol. R/T-rotational and translation part, including RTln⁡(8π2/C∘)RT \ln(8\pi^2/C^\circ) for the free ligand; Tors-torsional; Ang-angle bend; Stretch-bond stretch. Off-diagonal terms are entropy changes caused by changes in correlations between classes of motions, for example, Stors,stretchcorr=Stors,stretch−Stors−SstretchS_{\text{tors,stretch}}^{\text{corr}} = S_{\text{tors,stretch}} - S_{\text{tors}} - S_{\text{stretch}} . Lower triangle terms are not listed because the matrix is symmetric [1]. Copyright 2006 National Academy of Sciences.

The analysis presented here only measures the entropy change of the ligand amprenavir. A true total entropy change on binding would also have to take into account entropy loss in the protein it binds to. However, in considering only the ligand, this study indicates that the entropy loss of the ligand alone can have a substantial effect on binding.

The most substantial contributor to entropy loss appears to be rotation/translation. Unfortunately, it would be very difficult to change the ligand structure to address this. However, both conformational entropy and torsional vibration can be addressed. Fewer or less mobile side groups would reduce the loss of conformational entropy on binding. In addition, any bond with multiple-bond character would reduce the overall torsional entropy of a ligand and thus reduce the potential for entropy loss.

It may also be necessary to reconsider the weighting in energy functions to reflect the dominant role of vibrational entropy more accurately.

Reference

  • [1] Chia-en A. Chang, Wei Chen, and Michael K. Gilson. Ligand configurational entropy and protein binding. PNAS 104;5(2006):1534–1539.