Figure 18.6.1.1   Design of glutamate-binding specificity in glutamine binding protein

The number at each panel is the affinity (E_bound-E_free) of the protein for the ligand shown. These were calculated from the raw E_structure  values from the appropriate pdb files in examples/multistate/glu_bp/ and examples/multistate/glu_bp/wt_gln_bp/,  and not scaled by OVERALL_ENERGY_SCALE. As discussed in the ligand section, the quantitative accuracy of predicted affinities needs to be evaluated, and is dependent on number of ligamers and protein sidechains allowed to adjust; the values listed are for demonstration purposes only.

The arrows indicate the sidechain at position 10.

In the wt structure, Asp10 is near the sidechain amide of the gln-bound state and near the sidechain carboxylate of the glu-bound state (assuming that glu occupies the same conformation as gln if bound). The electrostatic repulsion from this drives the specificity towards the gln-bound by destabilizing the glu-bound form.

In the target-optimized sequence, Ser10 replaces Asp10. This mutation, along with the the other sequence changes, results in a tighter predicted affinity for the target glu than the wt sequence. However, this also results in a tighter affinity for the non-target gln. In fact, this sequence is predicted to bind gln slightly more tightly than glu, the opposite of the desired result.

In the multistate design, which takes into account the unbound state, as well as the unwanted gln-bound state, Asp10 is changed to Asn10. In the gln-bound structure, the NH2 of the Asn10 sidechain amide are placed near the NH2 of the bound gln's sidechain amide; this is unfavorable. In contrast, in the desired glu-bound state, the Asn10 NH2 group can hydrogen-bond with the glu carboxylate. These factors, as well as the other sequence changes, results in a binding specificity for glu over gln. However, this specificity comes at the cost of lowered absolute affinity compared to the target-optimized sequence.