At the moment it is only possible to do it one by one. We are currently working on such an implementation that we must do the final tests in the coming week before merging it with the stable branch (it would be version 1.5.0).
Although you can do it for all residues in the protein, we do not recommend it. Keep in mind that as many calculations will be carried out as residues (minus the terminals) your protein has, which is very inefficient. We recommend doing a per-residue analysis to determine each residue's energy contribution and then choose the set you consider and perform the CAS. In any case, gmx_MMPBSA and gmx_MMPBSA_ana will be able to deal with all the residues that you mutate.
Cheers!
Mario S.