Transcription by Design

(ZINC) FINGER POINTING

Transcription factors generally have two domains, one that recognizes and binds to a specific DNA sequence (DNA-binding domain, or DBD), and another that actually regulates transcription through protein-protein interactions. These domains are exchangeable; a DNA-binding motif from one transcription factor can be coupled to the regulatory motif of another to create a molecule with hybrid properties.¹

Though the approach works, most DBDs are not sufficiently malleable for engineering purposes, as it's difficult to modify the protein's recognition sequence. Says Aseem Ansari, a biochemist at the University of Wisconsin, "They couldn't de novo design a DNA-binding protein that would go bind whatever sequence they wanted, and whatever gene they wanted."

This is not the case, however, with zinc finger DBDs.² Zinc finger proteins contain several 30-residue-long zinc finger motifs, each of which recognizes a specific three-base sequence. Like transcription factors themselves, these fingers are modular; by shuffling them, scientists could design novel transcription factors that could recognize any desired sequence.

Combining rational design with library screening of zinc finger domains, researchers such as Carlos Barbas at the Scripps Research Institute in La Jolla, Calif., and Carl Pabo, formerly at Sangamo Biosciences, Richmond, Calif., have designed proteins that can target specific DNA elements and regulate proximal genes. One such construct, designed to activate vascular endothelial growth factor, currently is in clinical trials, according to Barbas.

SMALL-MOLECULE MIMICS

But introducing such protein constructs into cells (or patients) can be difficult, requiring a gene-or protein-delivery technique. Small-molecule-based transcription factors, on the other hand, are easily delivered both to cells and patients, though the effects of such drugs are transient. "In terms of therapeutics, [small molecules] might be better in terms of getting things into cells," says Pamela Silver, a systems biologist at Harvard University.

Some nonpeptide components have already been developed. Peter Dervan and colleagues at the California Institute of Technology developed polyamides, based on the antibiotic distamycin, that recognize base pairs in the minor groove of DNA. Others have built activation domains out of RNA. Now Mapp and her colleagues have designed and synthesized small-molecule activation domains for the first time using rigid isoxazoladine-based molecules as scaffolds.³

The team conducted in vitro transcription reactions using a LexA-DHFR (dihydrofolate reductase) DBD. DHFR binds methotrexate, and as each synthetic activator molecule contained a methotrexate moiety, the activator and DBD domains became coupled. The team found that one of the small heterocyclic molecules functioned almost as well as a modified version of an endogenous transcription activator. "It's been difficult to develop effective strategies for finding molecules [that] could reconstitute the function of a particular [activation domain]. We took a slightly different approach and found that to be successful," says Mapp. Mapp next plans to develop molecules that mimic the activity of endogenous activators and to test the strategy in vivo. "There are certainly changes that could be made to make them more potent and we're already working on that, but I think the general idea of finding a rigid molecule and appending these key functional groups to it is going to be pretty general for finding artificial activators," she says.

Ultimately both small-molecule based factors such as Mapp's and their protein-based counterparts may help guide future drug development. With improvements in gene therapy, says Barbas, "there may be fewer and fewer distinguishing features between the [chemical and protein-based] approaches."

- Aileen Constans