We have observed in several cases that energetic contribution of residues in a complex, in which either one or both sub-units are highly charged, are highly positive or negative. Some of the residues that are distant from binding site and also present on the surface also exhibit contribution to the binding. One cannot expect contribution of surface and distant residue to the binding, particularly when conformation is same in both bound and unbound forms. We here note that results of MM-PBSA should be carefully interpreted In case of highly charged complexes.
One of the major problem is that the system cannot be neutralized during the PB calculation, which is performed by APBS program. Therefore, energy of charged residues are likely to be highly positive or negative, which would further depend on the total charge of the system. For example, If total charge on the system is positive, residues LYS or ARG likely to have positive contribution energy, particularly when these are present at the surface of the protein.
We had tried to use two different values of positive and negative ion concentration to neutralize the system during PB calculations in the APBS program, but APBS does not run with the unequal concentrations. We had further removed this criteria from APBS at the source-code level to run using unequal concentrations, but APBS again used maximum equal concentration for both positive and negative ions during the calculation. Therefore, presently neutralization during PB calculation is not possible.
We would suggest followings:
a) Always use non-linear PBE solver; "PBsolver = npbe".
b) Increase value of "cfac" and "fadd". This will increase the volume of box and, will prevent any artifact on the surface of the protein. For example: "cfac = 3" and "fadd = 30". However, these values will slow down the calculation.
c) Because, protein is charged, its dielectric constant is most likely to be more than two. Recently, the dielectric constant of DNA was observed to be eight (http://www.pnas.org/content/111/35/E3624
). Therefore, I would suggest to increase the value of "pdie
". One may perform two calculations at "pdie = 8
" and "pdie = 20
". Then, after comparison, contribution energy of many surface residues could be significantly different, whereas energy of residues that really contribute to binding may not change with very large values.
We hope above information would be helpful.