Since the 1960s, chemists have known about an unusual type of hydrogen bond: one that can form between the amide hydrogen and the carbonyl oxygen in the same amino acid. But it’s been hard to show that this interaction, called the C5 hydrogen bond, matters beyond amino acids, in proteins.
Now, biochemistry professor Ronald T. Raines and graduate student Robert W. Newberry of the University of Wisconsin, Madison, report experimental and computational evidence that this hydrogen bond indeed helps stabilize proteins—significantly in some cases (Nat. Chem. Biol. 2016, DOI: 10.1038/nchembio.2206).
“It has been problematic to quantify the energetic consequence of this H-bond, particularly in systems larger than a dipeptide where numerous other factors come into play,” says Steve Scheiner, a chemistry professor at Utah State University who studies hydrogen bonding. The new work “takes a major step toward unambiguously verifying the presence of this H-bond, as well as offering some quantitative measures of its influence.”
Raines and Newberry used spectroscopic measurements and quantum mechanical calculations of model peptides called tryptophan zippers, which are β hairpins that are 12 amino acids long, to show that the C5 interaction behaves like a hydrogen bond. They then used bioinformatics to analyze a database of protein structures, revealing that the interaction is widespread, affecting ~5% of amino acids and ~94% of all proteins.
The C5 H-bond is strongest in flat structures, which limits where it can occur. “It has to be in β sheets,” Raines says. “Otherwise, the oxygen and hydrogen are too far away from each other.” Raines speculates that this interaction could be important for stabilizing amyloid-forming proteins rich in β sheets, such as those involved in Alzheimer’s disease.
Incorporating this stabilization energy into models of protein folding could help improve predictions of protein structures, Raines says. “We’re very good at predicting the structures of things that are similar to what’s in the database,” he says. Predicting structures from scratch is much harder. “To truly solve the protein folding problem, we need to be much better in our accounting. We need to understand the fundamental forces that lead to a stable structure.”