Embedded Files
In this project, the antibodies were taken from the crystal structures in the protein data bank to give the fab portions as the current structure does not keep CH1 and CL domains. The PDB codes are (7LYV) and (7M1C) for the antibodies. The starting HCMV pentamer structure taken from the protein data bank is (7T4Q) as it was the only full structure of the HCMV pentamer, and the reference structures from the McLellan lab are (7M30) and (7M22) which were locally refined to give a better resolution by cutting away flexible regions of the pentamer in the iterative process to receive a cryo em structure. To ensure the ending RMSD not was too far off, the backbones of 7T4Q(green) and 7M30(red) were superpositioned to give an RMSD of 1.44 Angstroms as shown below. The region where the two backbones do not interlap well are where the protein gL which in both 7T4Q and 7M30 have as a more flexible region as no antibodies are bound in both crystal structures to make this region more rigid.
After creating a base construct to make a reference for the ending homology model, the antibody 7LYV was then docked in the Rosie server using the beta Rosetta dock 3. To not bias the results I aligned the antibody to the reference structure 7M30 and then shifted the antibody close enough for local docking to work but far away enough to not bias the results by entering the exact three dimensional coordinates. Below shows the 100 structures predicted by the server and the structure used in the docking protocol (the antibody is shown in dark blue and grey binding to the top UL protein of the pentamer. Normally with more computational power 10,000 structures would be computed and clustered however with this plot as long as a shape of a pseudo funnel is found, the top ten lowest Interface score rosetta energy units should give the best homology model. It is important to note the energy units do not reflect free energy or gibs energy but rather the score represents the existence of a protein in a given metastate.
After the first antibody was docked, the second antibody 7M1C (colored in gray) was docked into the top-scoring model in a similar fashion. Below I show the 100 structures predicted and the structure I used in the docking protocol. It is important to note that this graph gave no direct funnel shape and would usually be considered a failure job, however the failing job would then be further computed in a more robust docking protocol specifically for antibodies called Rosetta snugdock which has CDR loop modeling. The reasoning for this failure job in the Rosetta beta dock 3 is because the CDR loops in the light chain where not all determined by crystallography and needed Rosetta loop modeling and or Antibody modeling to work as intended. The reason this was not done will be mentioned in the Results and Discussion sections. With all this said the final structure was chosen by comparing the top ten interface scores vs the top total score. Out of the top ten of each, one structure was the highest ranked in both and was used in the final structure.
Lastly, NRP2 was docked into the starting structure 7T4Q with Rosetta beta dock 3. NRP2 was moved away from its starting position to unbias the results. The reference structure is on the left with NRP2 colored in a series of blue while the input structure is on the right with NRP2 colored red, orange, green, and yellow. Below as well are the 100 predicted structures for the docking of NRP2.
Page updated
Google Sites
Report abuse