Computational study of the activity, dynamics, energetics and conformations of insulin analogues using molecular dynamics simulations: Application to hyperinsulinemia and the critical residue B26

Abstract

Due to the increasing prevalence of diabetes, finding therapeutic analogues for insulin has become an urgent issue. While many experimental studies have been performed towards this end, they have limited scope to examine all aspects of the effect of a mutation. Computational studies can help to overcome these limitations, however, relatively few studies that focus on insulin analogues have been performed to date. Here, we present a comprehensive computational study of insulin analogues-three mutant insulins that have been identified with hyperinsulinemia and three mutations on the critical B26 residue that exhibit similar binding affinity to the insulin receptor-using molecular dynamics simulations with the aim of predicting how mutations of insulin affect its activity, dynamics, energetics and conformations. The time evolution of the conformers is studied in long simulations. The probability density function and potential of mean force calculations are performed on each insulin analogue to unravel the effect of mutations on the dynamics and energetics of insulin activation. Our conformational study can decrypt the key features and molecular mechanisms that are responsible for an enhanced or reduced activity of an insulin analogue. We find two key results: 1) hyperinsulinemia may be due to the drastically reduced activity (and binding affinity) of the mutant insulins. 2) Y26_BS and Y26_BE are promising therapeutic candidates for insulin as they are more active than WT-insulin. The analysis in this work can be readily applied to any set of mutations on insulin to guide development of more effective therapeutic analogues.

Keywords: Diabetes; Insulin; Molecular dynamics; Mutations.

Fig. 1

Illustration of the distances between the C_α atoms of the BC-CT residues B25-B29, and their closest partner in the B-chain α-helix. The two critical distances, namely B26(C_α)-B12(C_α) and B28(C_α)-B8(C_α), that provide the best criterion for the BC-CT opening and are used in the PDF and PMF calculations are shown in solid lines.

Fig. 2

Convergence analysis of the PMF calculation in (A) WT-insulin (B) the V3_AL analogue and (C) the Y26_BA analogue. In WT-insulin and V3_AL analogue, the PMF calculation converges after 500 ns, whereas in Y26_BA, it converges after 1000 ns, suggesting that a total simulation time of 600 ns and 1200 ns is adequate for sampling in WT-insulin/V3_AL analogue and Y26_BA analogue, respectively. All the other analogues display similar behavior to V3_AL.

Fig. 3

Time series for the hyperinsulinemia and critical residue analogues using (A) the B26(C_α)-B12(C_α) distance and (B) the B28(C_α)-B8(C_α) distance. The corresponding plots for WT-insulin are also shown for comparison in the bottom row. The threshold from the closed to the open conformation and from the open to the wide-open conformation are represented by yellow and green dashed lines in the plots, respectively.

Fig. 4

Comparison of PMFs for the hyperinsulinemia and critical residue analogues against those of WT-insulin using (A, C) B26(C_α)-B12(C_α) distance and (B, D) B28(C_α)-B8(C_α) distance, respectively.

Fig. 5

Time series of the B24(C_α)-B15(C_α) distance which forms a hinge between BC-CT and B-chain α-helix (A) in the hyperinsulinemia analogues and (B) in the critical residue analogues, compared to WT-insulin in both cases.

Fig. 6

Correlation maps of the (A) hyperinsulinemia and (B) critical residue analogues compared to WT insulin. WT-insulin is comprised of 3 distinct groups of positively correlated residues, while similar correlation pattern can only be seen in V3_AL and to a lesser extent in F24_BS. The Y26_BA analogue shows the strongest positively correlated behavior, while F25_BL and Y26_BS exhibit strong positively correlated behavior but to a lesser extent and Y26_BE does not show any pattern in its correlation map.