The energetic cost of domain reorientation in maltose-binding protein as studied by NMR and fluorescence spectroscopy

Abstract

Maltose-binding protein (MBP) is a two-domain protein that undergoes a ligand-mediated conformational rearrangement from an "open" to a "closed" structure on binding to maltooligosaccharides. To characterize the energy landscape associated with this transition, we have generated five variants of MBP with mutations located in the hinge region of the molecule. Residual dipolar couplings, measured in the presence of a weak alignment medium, have been used to establish that the average structures of the mutant proteins are related to each other by domain rotation about an invariant axis, with the rotation angle varying from 5 degrees to 28 degrees. Additionally, the domain orientations observed in the wild-type apo and ligand-bound (maltose, maltotriose, etc.) structures are related through a rotation of 35 degrees about the same axis. Remarkably, the free energy of unfolding, measured by equilibrium denaturation experiments and monitored by fluorescence spectroscopy, shows a linear correlation with the rotation angle, with the stability of the (apo)protein decreasing with domain closure by 212 +/- 16 cal mol-1 per degree of rotation. The apparent binding energy for maltose also shows a similar correlation with the interdomain angle, suggesting that the mutations, as they relate to binding, affect predominantly the ligand-free structure. The linearity of the energy change is interpreted in terms of an increase in the extent of hydrophobic surface that becomes solvent accessible on closure. The combination of structural, stability, and binding data allows separation of the energetics of domain reorientation from ligand binding. This work presents a near quantitative structure-energy-binding relationship for a series of mutants of MBP, illustrating the power of combined studies involving protein engineering and solution NMR spectroscopy.

Fig. 1.

Structural changes associated with domain reorientation in MBP. (A) Correlation between experimental ¹D_HN values recorded on I329W and those predicted on the basis of the 1OMP starting x-ray structure, rotated first to best satisfy the experimental dipolar coupling values. Dipolar couplings from the N- and C-terminal domains are shown in blue and red, respectively. (B Left to Right) Wild-type open structure (0° closure), I329C (5.5° closure), I329W (9.5° closure), I329W/A96W (28.4° closure), and the closed conformation for wild type in the absence of ligand (35° closure). The open wild-type structures are identical in solution and in the crystal, as are the closed wild-type structures (12, 36). The closure, bend, and twist (C, B, T) coordinate axes are displayed. The residues that have been mutated are shown in blue.

Fig. 2.

The interdomain closure angle correlates with both the free energy of unfolding and the affinity for maltose. (A) Experimental values for the free energy of unfolding (ΔG_U-F). (B) Values in A corrected for the changes in stability introduced with the mutation (); a linear dependence arises with a slope of 212 ± 16 cal·mol^–1 per degree of rotation. (C) Correlation between the dissociation constant for maltose-binding and closure angle (slope = 151 ± 38 cal·mol^–1·deg^–1). Contributions arising from differences between the wild-type and mutant residues have been corrected, following a procedure analogous to that described for obtaining the correlation in B.

Fig. 3.

Energy level diagram showing the relative stabilities of the apo wild-type open, apo wild-type closed, and mutant apo conformations of MBP along with the wild-type closed bound form of the protein. The dashed axis indicates the point at which the unfolded apo state is as stable as the folded. For all mutants, the energy has been corrected according to Eq. 1 so that the stabilities plotted are those associated with the wild-type sequence with the closure angle indicated along the x axis. The closed-bound state is arbitrarily assigned a free energy value of 0.

Fig. 4.

(A) Nonpolar solvent accessible surface area, as estimated by rolling a ball of radius 1.4 Šalong the protein surface (37), as a function of interdomain closure angle. (B) Hydrophobic residues with increased (decreased) solved accessibility on domain closure are shown in green (pink). Only those residues with a change >10 Ų between open and closed conformers are displayed. There is a net destabilization of the protein on closure from this effect. (C) Residues that form van der Waals interactions with maltose in the bound form are shown in blue, whereas those hydrogen bonding to sugar are in red. The orientations of the protein are different in B and C, with the axis system showing how the view in B is transformed to that in C indicated. In B and C, the structures are drawn in stereo.