Team:ULaVerne Collab/Model
I. Single Chain Insulin Protein Models
(Hover on image for 360° 3D animation)
Figure 1. Simulated models of human insulin and our 3 single chain insulin mutants using Swiss-Model online protein modeling program: Images are a, b, c, d from left to right.(a) Human proinsulin, (b) native insulin + linker (pI 5.50), (c) fast-acting insulin + linker (pI 5.50), and (d) long-lasting insulin + linker (pI 6.46). The A chain is indicated as orange, B chain is green, and the linker is red.
According to Swiss-Model all our proteins were modeled under pH 7 in vacuo. The compartment where insulin works and binds to the human insulin receptor is in the liver (pH 7), skeletal muscle (pH 6.9), and adipose tissues (7-7.35). Since the pH they were modeled in is similar to the pH of the compartment where the proteins are active, these models should show the protein structure at its active conformation.
II. Single Chain Insulin and Human Insulin Protein Overlapping Models
(Hover on image for 360° 3D animation)
Figure 2. Simulated models of our 3 single chain insulin mutants overlapped with the original human insulin using Chimera Matchmaker: Images are a, b, c from left to right. (a) Native insulin + linker (pI 5.50) with human insulin (b) fast-acting insulin + linker (pI 5.50) with human insulin (c) long-lasting insulin + linker (pI 6.46). For human insulin, the A chain is indicated as orange, and B chain is green. For our single insulin mutants, the A chain is indicated as magenta, the B chain as cyan, the linker as red, and mutated sites as black.
Overall, the backbone of the native insulin A and B chains almost perfectly aligns with the A and B chains of each of the mutated insulin. There seems to be backbone conformational differences in the first few amino acids of the native insulin B chain to the B chain of each of the SCI. We assume this could be due to the different conformational results produced by different programs with different modeling algorithms. Another possible assumption might be due to the interactions with the surrounding side chains, causing torsion and shifting the backbone at the first few amino acids of the B chain. The added linker to each SCI appear to be sticking out of the protein perimeter, especially in the fast-acting model. The native and long-lasting SCI shows formation of alpha-helix sheets in addition to the beta-pleated sheets.
III. Simple Docking and Interaction Energy of our Insulin and the Ectodomain of the Human Insulin Receptor
Note: Figure 3 models and interaction energy calculation courtesy of the 2019 iGEM Moscow Team and their protein modelers who collaborated with us and did this docking simulation and calculation for us per our input.
Figure 3. Simulated models of our 3 single chain insulin mutants binding to the human insulin receptor ectodomain 6ce7 from PDB using the program FoldX and values of interaction energy (DG) between the insulin molecule and the receptor. Each insulin molecule has the A and B chain connected as a single chain instead and does not include our 12AA linker. This is because there is no similar insulin structure in the PDB database with a 12AA linker (the nearest single chain insulin structure has 6AA linkers), and the program requires an existing similar PDB structure in order to perform docking. (a) Single chain native insulin (red) binding to the human insulin receptor ectodomain 6ce7 (cyan). The interaction Energy (DG) between them is calculated to be 2.04839 kcal/mol. (b) Single chain long-lasting insulin (red) binding to 6ce7 (green), DG = 1.84606 kcal/mol. (c) Single chain fast-acting insulin (red) binding to 6ce7 (pink), DG = 1.49372 kcal/mol.
The simulated docking only allowed binding of single chain insulin without our added 12AA linker due to the reason listed in the figure description. However, from the calculated interaction energy, we are able to predict and make assumptions about the effect of the mutations we made on the insulin binding affinity. Overall, the interaction energy (DG) decreases as we go from the SCI native insulin to long-lasting and fast-acting. Since interaction energy is measured in kcal/mol, the higher the kcal the more energy it takes for one mole of insulin to bind to the receptor. Therefore, the fast-acting SCI with the lowest DG value would have the most binding affinity compared the other two. Future directions is to observe the side chains interactions between the mutant insulin and the receptor and find out if the linker interfere with the receptor. Also, we would be able assume the affinity of these single chain insulin to the receptor if we also had the interaction energy of the native insulin and receptor complex.
IV. Predicted Measurement of the RMSD and Radius of Gyration
Note: Figure 4-7: measurements of RMSD, radius of gyration, and energy minimization courtesy of the 2019 iGEM Moscow Team and their protein modelers who collaborated with us and did these for us per our input.
Figure 4. (a) Line graph showing the RMSD values (nm) over time (ns) for our Single Chain Native Insulin protein structure. (b) Line graph showing the radius of gyration (nm) over time (ns) for the Single Chain Native Insulin. (c) Line graph showing the potential energy minimization (kJ/mol) throughout the energy minimization step for the Single Chain Native Insulin.
Figure 5. (a) Line graph showing the RMSD values (nm) over time (ns) for our Single Chain Long-lasting Insulin protein structure. (b) Line graph showing the radius of gyration (nm) over time (ns) for the Single Chain Long-lasting Insulin.
Figure 6.(a) Line graph showing the RMSD values (nm) over time (ns) for our Single Chain Fast-acting Insulin protein structure. (b) Line graph showing the radius of gyration (nm) over time (ns) for the Single Chain Fast-acting Insulin.
Figure 7. (a) Line graph comparing the RMSD values (nm) over time (ns) of our Single Chain Native, Long-lasting, and Fast-acting Insulin protein structure. (b) Line graph comparing the radius of gyration (nm) over time (ns) for the Single Chain Native, Long-lasting, and Fast-acting Insulin protein structure.
Analysis:
The RMSD (root-mean-square-deviation) was calculated in order to observe the stability of the our single chain insulin mutants. If the RMSD values are stable over time, it means that the protein from its 3D model stays similar its initial optimal rigid body superposition, suggesting that the protein is stable. On the other hand, the radius of gyration (Rg) was measured in order to observe the protein structure compactness or distribution of atoms of the protein around its axis. However, we are observing the stability of the protein more than the structure compactness, so we observed the Rg over time. If the Rg values is relatively steady over time, the protein should be stably folded. If the values fluctuate, that means the protein is unstable, meaning it unfolds and refolds. Figure 4c is a graph showing energy minimization of one of the protein models. Energy minimization was done on all the protein models before conducting the RSMD and Rg measurements in order to obtain the best conformation and convenience for running molecular dynamics on the computer.
Figure 4-6 shows the RMSD and Rg values and trend of each insulin individually. The single chain native insulin (4a) shows a plateau for its RMSD over time, and its Rg (4b) does not show very significant fluctuation either, suggesting that its structure is somewhat stable. The single chain long-lasting insulin shows a more significant fluctuation in both the RMSD and Rg values over time (5a,b). The single chain fast-acting insulin shows a plateau in its RMSD values (6a) but significant fluctuation in the radius of gyration (6b). Therefore, the single chain native and fast-acting insulin have the most structural stability out of the 3, which might support the DG value calculated for the fast-acting in figure 3c. However, the protein folding instability in the Rg values of the fast-acting might be due to intramolecular interactions either caused by the distinct 12AA linker loop structure that significantly protrudes out from the structure (1c, 2b) or just by the mutation we made. It is important to notice, though, that although there were fluctuating values for our proteins, none of them decreased to the point of protein denaturation. A future direction to discuss this hypothesis for the single chain fast-acting insulin would be to study the intramolecular side chain interactions within the structures in comparison to the native human insulin or Lispro insulin. In figure 7, the RMSD and Rg values for all 3 insulin are stacked and compared. The single chain long-lasting insulin had lower RMSD values over time compared to the single chain native and long-lasting (7a). The Rg values are decreasing from the single chain long-lasting to native and to fast-acting, but again the difference is minimal at just 0.1 nm (7b).
Conclusion
The single chain native and fast acting insulin has a more stable structure throughout time than the long-lasting insulin. The RMSD values ranges that these proteins are at could not be analyzed to validate the simulated docking that was done and shown in figure 3 because there are no crystallography structures of the protein and receptor complex for all of them to compare the RMSD values with (Coutsias, 2004). The single chain native insulin show the most stable Rg value or protein folding out of the 3. However, we are unable to analyze the Rg values of each of the protein because there is no Rg values of the protein in the database to compare to. A future direction would be to calculate and graph the RMSD and Rg values for the native insulin protein, in order to be able to analyze the values of each of the protein.
References:
Coutsias, Evangelos A., et al. “Using Quaternions to Calculate RMSD.” Journal of Computational Chemistry, vol. 25, no. 15, 2004, pp. 1849–1857., doi:10.1002/jcc.20110.