Contribution of buried hydrogen bonds to protein stability. The crystal structures of two barnase mutants.

The crystal structures of two barnase mutants, Tyr78-->Phe and Ser91-->Ala, have been determined to 2.2 A resolution. In both cases, a buried hydroxyl group that makes two hydrogen bonds within the protein was replaced by a hydrogen atom. It is found that neither mutation causes any structural changes, within the limits of error, compared with wild-type and so are confirmed to be non-disruptive. Solvent molecules are not observed in the cavities created by removal of the respective hydroxyl groups and no new interactions are introduced. The local water structure surrounding both sites of mutation is well conserved and resembles that of the wild-type. All four water molecules making contacts with the side-chain of residue 78 and two water molecules nearest to residue 91 in the wild-type are found within a sphere of 0.5 A radius, at the equivalent positions of the respective mutant. No new water molecules are found bound to any of the hydrogen bond donor or acceptor residues involved in these two mutation sites. Previous protein engineering experiments established that the solvent-inaccessible phenolic OH of Tyr78 that makes hydrogen bonds with two uncharged groups (main-chain NH and CO) contributes 1.4 kcal mol-1 to protein stability, while the solvent-inaccessible OH of Ser91 that makes hydrogen bonds with an uncharged main-chain NH and a charged group (O gamma 1) contributes 1.9 kcal mol-1. These stability measurements can now be attributed primarily to the loss of the hydrogen bonding interactions because both mutations neither disrupt the respective protein and local solvent structures, upset the overall hydrogen bonding pattern nor introduce new interactions. The mutations Tyr78-->Phe and Ser91-->Ala are thus good examples of "non-disruptive deletions" and the results of mutagenesis can be analysed at the simplest level.