The hybrid coarse graining method was put to the test to investigate the confor- mational space of yeast PDI protein, see chapter 5. Each normal mode defined a direction of motion that was explored to identify the motion limits and the effects of using different networks of hydrogen bonds by using differentEcutfrom the rigid- ity analysis. The rigidity analysis identifies the modular nature of the protein and the geometric simulations identifies the large conformational change suggested by previous experimental data [63]. The distance between the active sites of yeast PDI provided a rough measure to quantify PDI motion and confirms that the protein undergoes a large conformational change. Further, the use of differentEcut showed that the protein mobility is impaired at highEcut. Although initially it was thought that such limitation was due to the high density of hydrogen bonds constraining protein motion, recent work [38] highlighted several limitations ofFroda. For ex-
ample, the enforcement of constraints procedure inFroda does not guaranty that the number of constraint violations are reduced at each step. Therefore, although projecting the initial structure along the normal modes has confirmed yeast PDI large conformational change, further investigation will be advisable to prove the robustness and limitations of the enforcement of constraints procedure byFroda.
Furthermore, in chapter 6 the MD simulations showed that the active sites distance betweenα-carbons atoms (599−5979) vary approximately between amin- imum of d(minM D) ≃ 22˚A and a maximum of d(maxM D) ≃70˚A . Since the motion along
normal modes using Froda is not modulated by force fields but limited by the stereochemical constraints it is expected that protein motion will reach larger con- formational changes. However, in the series of simulations here presented the max- imum inter-cysteine distance appears to be bigger during the MD simulations. On the contrary, the minimum intra-cysteine distance is smaller for the HCG than for the MD simulations, which is to be expected. MD will seldom guide the protein to energetically not favourable conformations whereas HCG method will explore the stereochemically accessible space as is. The experiments presented in chapter 7 using cross-linkers to identify the minimum distance between active sites revealed that the minimum intra-cysteine distance for yeast PDI is longer that the largest cross-linker used, i.e. a distance of 12˚A between the two maleimide groups. Longer cross-linkers are needed to identify such minimum intra-cysteine distance.
The HCG method has proven to reveal interesting and useful results. It allows to project a protein along the pathway defined by a normal mode and generate the corresponding conformers along the pathway by using a minimum of computer resources, typically hours of CPU time. However, it will be revealing to investigate further whether the more restrictive motions displayed by the HCG method are due to Froda limitations and to which extend do they affect protein mobility for simulations at different Ecut. Hence, new simulations to test the robustness of the mobility limits will be advisable. In regard to yeast PDI mobility predictions, it will be interesting to perform mobility simulations with other methods which can circumvent Froda’s issues and compare with new experiments using longer cross- linkers or FRET experiments to determine the accuracy of yeast PDI mobility predictions.
The minimum and maximum inter-cysteine distance predicted by the HCG and MD, and the predicted most likely inter-cysteine distance by the MD simulation are very useful results to provide a good indication for the type of fluorophore pairs that are functional within the distance ranges that we wish to explore. The distance range that a given pair of fluorophores can explore varies depending on
the fluorophore pairs, e.g. a given set of pairs will be able to explore motion for distances between 45 to 55˚A. Hence, using the inter-cysteine distance ranges will be very useful to narrow down the number of useful fluorophore pairs. Single-molecule FRETexperiments with alternating laser excitations (ALEX) is a new and suitable technique to investigate the relative motion between yeast PDI cysteine groups. Currently, members of my co-supervisor’s group, Prof. R.B. Freedman, are in the process of bringing forward such experiments. Their choice of fluorophore pairs based on the HCG and MD simulations is Atto550–Atto647N, Atto550–Atto665 and Dylight488–Atto665. Each pair has a mid point distance where the resolution is optimal. The mid points of the effective ranges are 65˚A, 60˚Aand 39˚Arespectively and they are quantitative approximately 10˚A either side of the mid point. Hence, with the three fluorophore pairs it is possible to explore inter-cysteine distances approximately between 29˚A to 75˚A. This will allow to put to the text the maximum inter-cysteine distance identified by the HCG and MD methods.
In summary, he HCG method provides a quick and versatile approach to explore protein conformational changes that would be very computationally costly for atomistic methods. Besides its advanced performance, it also provides the user with data that was not revealed by MD simulations and that could be of biological importance. Further, the HCG method advice on the type of experiments and experimental probes that could be most useful to corroborate the simulations.