**4.2 An integrated picture of galactose binding to Arabinose Binding Protein (ABP)**

The arabinose binding protein is found in the periplasm of Gram-negative bacteria. It belongs to the family of proteins, which extert their biological function as components of osmotic shock-sensitive transport systems for sugars and amino acids (Vyas *et al.*, 1991). Besides its biological context, ABP is a very well-defined model system for structure-activity relationships in the hydrophilic ligand binding systems. As demonstrated by ITC, ABP interacts with its natural ligands, namely L-arabinose and D-galactose, and their deoxy derivatives (Daranas *et al.*, 2004). The interactions are enthalpy-driven. The galactose-ABP interactions served as a model for interactions between hydrophilic ligands and hydrophilic binding pockets. Such choice of the model system was made mainly because of the large unfavourable entropic contribution to binding, whose origin was difficult to understand in the framework of the current ligand-protein interaction paradigms. Such a large entropy change upon ligand binding is frequently observed in proteins interacting with carbohydrate ligands, and it contributes to the fact that these interactions are notoriously challenging for predictions and design. In order to address those issues, we employed a combination of solution NMR and MD simulations.

### **4.2.1 NMR measurements of ABP**

An essential prerequisite to NMR studies of protein dynamics was the establishment of the 1H, 15N and 13C resonances that contributed to the investigated spectra. These assignments have been obtained for the complex of ABP with the ligand D-galactose using conventional triple-resonance assignment strategies (Daranas *et al,* 2004). Assignments for the unbound (apo) ABP were determined from these results, using an approach that combined conventional triple-resonance assignment strategies and 1H-15N heteronuclear single

ligand degrees of freedom upon binding. The favourable entropic contribution from desolvation of the protein binding pocket that one would predict in a "classical" hydrophobic interaction is absent in MUP, since the occluded binding pocket is substantially desolvated prior to binding. This phenomenon has subsequently been observed in other

As suggested by MD simulations, there are around 6 water molecules (on average) occupying the binding site of HBP prior to ligand binding. As already stated, no significant change in entropic contribution should arise from displacing these waters upon ligand binding, since their dynamic behaviour inside the pocket is similar to the behaviour of the bulk water. However, four water molecules are sequestered in the binding pocket in the histamine-HBP complex. These waters are significantly more ordered than the bulk water (Syme *et al.*, 2010), which contributes negatively to the entropic term of binding free energy. This unfavourable entropic contribution can be estimated as about 30 to 40 kJ/mol,

In conclusion, in the case of HBP, we found favourable entropic contributions to binding from desolvation of the ligand. However, the overall entropy of binding was unfavourable due to a dominant unfavourable contribution arising from the loss of ligand degrees of freedom, together with the sequestration of solvent water molecules into the binding pocket in the complex. This can be contrasted with MUP, where desolvation of the protein binding pocket made a minor contribution to the overall entropy of binding given that the pocket is

**4.2 An integrated picture of galactose binding to Arabinose Binding Protein (ABP)**  The arabinose binding protein is found in the periplasm of Gram-negative bacteria. It belongs to the family of proteins, which extert their biological function as components of osmotic shock-sensitive transport systems for sugars and amino acids (Vyas *et al.*, 1991). Besides its biological context, ABP is a very well-defined model system for structure-activity relationships in the hydrophilic ligand binding systems. As demonstrated by ITC, ABP interacts with its natural ligands, namely L-arabinose and D-galactose, and their deoxy derivatives (Daranas *et al.*, 2004). The interactions are enthalpy-driven. The galactose-ABP interactions served as a model for interactions between hydrophilic ligands and hydrophilic binding pockets. Such choice of the model system was made mainly because of the large unfavourable entropic contribution to binding, whose origin was difficult to understand in the framework of the current ligand-protein interaction paradigms. Such a large entropy change upon ligand binding is frequently observed in proteins interacting with carbohydrate ligands, and it contributes to the fact that these interactions are notoriously challenging for predictions and design. In order to address those issues, we employed a

An essential prerequisite to NMR studies of protein dynamics was the establishment of the 1H, 15N and 13C resonances that contributed to the investigated spectra. These assignments have been obtained for the complex of ABP with the ligand D-galactose using conventional triple-resonance assignment strategies (Daranas *et al,* 2004). Assignments for the unbound (apo) ABP were determined from these results, using an approach that combined conventional triple-resonance assignment strategies and 1H-15N heteronuclear single

*CP* on binding.

proteins, such as streptavidin and HIV-protease receptors (Young *et al.*, 2007).

which is qualitatively consistent with the observed sign of

substantially desolvated prior to binding.

combination of solution NMR and MD simulations.

**4.2.1 NMR measurements of ABP** 

quantum coherence (HSQC) titrations of ABP with 1-deoxy-galactose, which is a fastexchanging ligand (MacRaild *et al.*, 2007).

From these assignments, a comparison was made of the chemical shifts of the backbone amide resonances of ABP in the unbound state and in the complex (Figure 7). Large chemical shift changes were observed in the binding site area and in the region linking the two domains of ABP. This suggested that ligand binding might be associated with a substantial conformational change in ABP (Figure 8). Such conformational change (domain reorientation) upon binding is observed in other members of the periplasmic-binding protein family, and have been proposed for ABP on the basis of the results of small-angle Xray scattering and theoretical studies (Mao *et al.*, 1982, Newcomer *et al.*, 1981). Smaller changes in chemical shift were also observed at sites distal to the binding site, which suggests that small conformational changes, resulting from protein dynamic behaviour, occur upon the binding event.

Fig. 7. Chemical shifts for galactose binding to ABP. Changes in backbone amide chemical shifts are plotted against protein residue number.

Fig. 8. Conformational changes between apo-ABP (light blue) and galactose-ABP complex. The apo protein is much more 'open' than the complex. Large conformational changes are observed in the hinge region (coloured red), and the reorientation of N-domain (yellow) and C-domain (dark blue) towards each other is quite pronounced. Bound galactose is coloured cyan and displayed as VDW spheres. Side chains of several residues involved in direct interactions with ABP are showed and coloured dark cyan.

assumption of un-correlated motion is not likely to hold for all residues of the protein. It is evident, however, that the entropy change associated with changes in 'fast' dynamics contributes favourably to the free energy of binding. As the result, it allows the reduction of

Fig. 9. Backbone amide order parameters for apo-ABP and galactose-ABP complex. <sup>2</sup> *S* order parameters were obtained by NMR relaxation measurements (red dots) and MD simulations (black lines). Data obtained for apo-ABP are showed in panel (a), while results for galactose-ABP complex are displayed in panel (b). Panel (c) shows changes in order parameters induced by galactose binding to ABP, with protein secondary structure elements displayed above. The N-domain of ABP is coloured is yellow, the C-domain is coloured blue, and the hinge region is coloured red. Residues interacting with the ligand are coloured cyan. This colouring scheme is consistent with colouring in Figure 10. (figure taken from

We investigated the origin of the large and unfavourable entropic contribution to the binding free energy of galactose-ABP, observed by ITC (Daranas *et al.*, 2004), in terms of the

It is clear that formation of a ligand-protein complex will involve the loss of entropy associated with constraining the translational and rotational degrees of freedom of one binding partner with respect to the other. The magnitude of these unfavourable contributions to the ligand-protein interaction can be approximated. In the case of galactose-ABP interactions, we took an estimate of the loss of ligand translational and rotational entropy from the work by Turnbull *et al* (2004) and Lundquist and his coworkers (Lundquist and Toone, 2002), which gave at approximately 25 kJ/mol for the free energy penalty. It is assumed that the bound ligand will experience a loss of entropy reflecting the loss of conformational flexibility of the ligand in solution. On the assumption that conformational degrees of freedom are substantially restricted upon ligand-protein binding, the entropic

MacRaild *et al.,* 2007)

different contributions.

**4.2.3 Origins of entropic costs of binding** 

the unfavourable entropic contribution associated with ligand binding.

Full sets of 15N relaxation measurements were made for ABP in its apo form and in complex with D-galactose, L-arabinose and D-fucose. The analysis of these measurements, assessed by means of Lipari–Szabo model-free approach (1982), allowed for extraction of the information on the extent of 'fast' (ps-ns time scale) motion of the protein backbone and 15N-containing side chains. Differences in Lipari-Szabo generalised order parameters ( <sup>2</sup> *S* ) between apo- and holo-ABP and were interpreted in terms of changes in dynamics accompanying the binding event. Thus, observed dynamic changes could be related to binding thermodynamics by means of the relationship between changes in order parameters derived from NMR relaxation and changes in conformational entropy. Because <sup>2</sup> *S* parameters are specific to individual bond vectors, the described approach offered an unprecedented degree of structural resolution in thermodynamic analysis of protein function.

Surprisingly, generalised order parameters for apo ABP were, in general, larger than for the ABP–galactose complex. This suggests that 'fast' (pico- to nanosecond time scale) motions are more extensive in the ligand-protein complex than in the unbound protein.
