Controlling Transcription

A research group has found a long-sought, generalizable way to modify nuclear hormone receptors so they respond with high specificity to synthetic agents.

Nuclear hormone receptors are hormone-activated proteins that turn genes on and off in cells; that is, they begin or terminate transcription of DNA segments that code for proteins. The new technique is a combinatorial strategy for engineering these receptors so their activity can be controlled at will with small molecules. The approach could have implications for gene therapy, artificial control of gene transcription, biosensor arrays, and enzyme engineering.

A number of groups have been trying for some time to engineer nuclear hormone receptors so they could be activated with small molecules. However, using rational design and site-directed mutagenesis to modify the receptors' binding sites has had only limited success. The synthetic ligands would frequently set off several types of cellular receptors, instead of just the modified one. Furthermore, the rational design/mutagenesis approach is tedious and not easily extended from one type of receptor to others.

Now, graduate student Lauren J. Schwimmer, assistant professor of chemistry and biochemistry Donald F. Doyle, and coworkers at Georgia Institute of Technology have developed a combinatorial approach that makes it possible to engineer nuclear hormone receptors quickly and easily and may also be useful for engineering other types of proteins [Proc. Natl. Acad. Sci. USA, 101, 14707 (2004)]. The researchers use random mutagenesis to create tens of thousands of receptor binding-site variants. They then apply a method called chemical complementation, developed by graduate student Bahareh Azizi, to carry out a yeast-based screen to identify variants activated by a synthetic small molecule.

They demonstrated the technique using retinoid X receptor (RXR), a nuclear hormone receptor important in cell development and differentiation. After creating a library of more than 30,000 binding-site variants of RXR, they identified those that recognize a synthetic retinoid called LG335 instead of one of RXR's natural ligands. One of those variants turns on gene transcription in yeast and mammalian cells when activated by only one-tenth as much synthetic ligand as the amount of natural ligand required to activate unmodified RXR.

"The technical obstacles that needed to be surmounted to achieve these results should not be underestimated," comments associate professor of chemistry and biochemistry John T. Koh of the University of Delaware, Newark. "We all knew that selection was the way to go, but the technical difficulties scared most of us away. Through careful analysis of the system and a true tour de force effort, Doyle and coworkers have shown us how to do it right. Using this technology, they could potentially create a family of selective receptors that could be used to independently regulate multiple genes in vivo. There are many exciting possible applications for such systems in biology, medicine, and bioengineering."

"The results are really very striking," says chemistry professor John A. Katzenellenbogen of the University of Illinois, Urbana-Champaign (UIUC). His group is currently collaborating with that of UIUC assistant professor of chemical and biomolecular engineering Huimin Zhao in an effort to achieve similar goals using rational design and directed evolution.

Doyle's former adviser, professor of pharmacology and biochemistry David R. Corey of the University of Texas Southwestern Medical Center, Dallas, says: "The screening is a breakthrough. It's remarkable that it works so well. The study is a classic that every student of chemical biology should be aware of. This is one of the rare papers that marks a real advance in the difficult field of protein engineering."