Several years ago, when researchers discovered intramembrane proteases—a class of hydrophilic enzymes that seemed to work despite being surrounded by cells’ oily membranes—many scientists were perplexed. Some were downright skeptical. After all, oil and water don’t mix.

Ya Ha, Ph.D., assistant professor of pharmacology, and colleagues may have revealed the intramembrane proteases’ recipe for success with their publication in October in the journal Nature of the protein’s structure. In addition to providing a solution to a slippery biological mystery, Ha’s work could shed light on the mechanisms underlying Alzheimer’s disease.

The chewing up of proteins, a process known as proteolysis, is the job of proteases. But protein-splitting reactions require water, not normally found in the greasy interior of cell membranes. In 1997, Nobel laureates Joseph L. Goldstein, Ph.D., and Michael S. Brown, Ph.D., suggested in an article in Cell that a protease involved in regulating cholesterol somehow did its work within the cell membrane. This protease must be “unusual,” they acknowledged, but they proposed that gamma-secretase, the enzyme that cleaves amyloid protein into the toxic fragments seen in the brains of Alzheimer’s patients, might also operate in the same manner.

When he came to the School of Medicine from Harvard University five years ago, structural biologist Ha was convinced that gaining structural information through X-ray crystallography was the key to understanding intramembrane proteases. He started with gamma-secretase. But despite his best efforts, it could not be coaxed into forming crystals, the first step in determining a protein’s molecular structure by X-ray crystallography.

When a family of bacterial enzymes with similar activity known as rhomboid proteins was discovered in 2001, Ha seized on those as an alternative. Over the next four years, Ha worked with postdocs Yongcheng Wang, Ph.D., and Yingjiu Zhang, Ph.D., to obtain the first X-ray data of the rhomboid enzyme molecules and created a visualization of the structure.

At last, they saw how an intramembrane enzyme is built: in the middle of a sea of fat, the rhomboid protease creates a protective bubble to shelter water molecules (whose source is unknown; they may come from a surrounding aqueous solution) in its active site. The enzyme is serpentine, crisscrossing the cell membrane six times. Five of these segments bundle together to create a water-filled chamber within the membrane.

However improbable this enzyme’s mechanics, they are medically important because the enzyme belongs to the family that includes human gamma-secretase. “Compounds that inhibit the production of toxic amyloid peptides are now believed to be one of the most promising approaches to the development of drugs for Alzheimer’s disease,” says Vincent T. Marchesi, M.D., Ph.D., the Anthony N. Brady Professor of Pathology and an expert on both membrane protein structure and Alzheimer’s disease. “Ha’s findings are an important contribution to this effort.”

Ha says that the rhomboid protease is a good model system for intramembrane proteases in general, but he confesses that he still has his eye on gamma-secretase. “I would love to see it,” he says. While the two enzymes are not related by their protein sequence or by evolution, Ha believes that they share common features because they face the same challenge of mixing water with oil. “Once you have a few structures, you’ll see a pattern start to emerge,” Ha explains. “That will give us a better understanding of how inhibitors might work. And then maybe we can design better inhibitors, and maybe those inhibitors can be used as drugs.”