Barak Cohen (left) and Roger Brent

In the bustling world of the cell, few sights are more common than the coupling of two proteins: hormones bind to receptors, enzymes fasten onto substrates, neurotransmitters cling to receptors on the surface of the cells of the brain.

How proteins interact with one another inside and outside of the cell is one of the central questions in biology. As researchers pinpoint key protein-protein interactions inside the cell, they are finding that proteins are more promiscuous than previously believed.

"Different combinations of proteins are ganging up at different times inside the cell to do different things," says Roger Brent, associate professor of genetics at MGH.

Brent and his colleagues have recently invented a way to tease apart a protein's interactions. Their method consists of placing a tiny molecular wedge on one side of a protein, while leaving its other sides intact.

The molecular wedge consists of a short peptide, composed of a string of about 20 amino acids, fused to a protein. The fused molecule, called a "peptide aptamer," is designed to block a single protein-protein interaction, making it possible to see what the protein pair normally does.

Brent, together with Barak Cohen, an HMS graduate student, and Pierre Colas, an HMS research fellow, have recently used the peptide aptamer approach to disrupt a specific interaction of a protein, Cdk2, involved in the proliferation of cells. Their findings appear in the April 11 issue of Nature.

Block that Coupling

The researchers hope to pinpoint other protein interactions with Cdk2, including those that might be involved in cancer. If they can do so, their research could lead to new approaches to fighting cancer that involve blocking specific cancer-causing protein-protein interactions, while preserving those that are beneficial.

More immediately, the peptide aptamer tool they have developed may help scientists piece together a picture of the larger network of protein-protein connections occurring in cells. Brent is currently working with scientists at other institutions to create such a protein network map.

Panel A: In one experiment, Brent and his colleagues looked for peptide aptamers that could bind Cdk2. Peptide aptamers that bind Cdk2 bring an activation domain close enough to the DNA to turn on a gene.

Panel B: Next, they looked for peptide aptamers that could block the normal interaction between Cdk2 and one of its substrates. Normally, Cdk2 adds a phophate to the substrate (top). The researchers found peptide aptamers that could block this interaction (bottom).

"Analyzing the patterns made by these networks has the potential to give us important, unexpected information about how proteins behave in normal situations and disease," says Brent.

Brent's interest in how proteins behave was first sparked over twenty-five years ago. "While I was at the University of Southern Mississippi as an undergraduate, there was a friend of the family who was a kind of wastrel, rich twit, eternal graduate student at Stanford who used to go skiing in Chile under Allende. He gave me a copy of Molecular Biology of the Gene," Brent says.

Inspired by James Watson's classic, Brent-a math and computer major-came to the biochemistry and molecular biology department at Harvard in 1973. Working in the lab of Harvard scientist Mark Ptashne, Brent became interested in the question of how proteins turn on genes in yeast and other higher organisms.

In bacteria and other lower organisms, proteins usually bind directly to the genes they regulate. But, in higher organisms such as yeast, proteins typically bind some distance away from the genes they activate.

To find out how this "activation-at-a-distance" happens, Brent took yeast regulatory proteins and removed the sections used to bind DNA. He then attached the remaining portions to the DNA-binding domain of a bacterial protein, Lex A, essentially creating hybrid proteins. These hybrid proteins were able to turn genes on in yeast cells.

"That defined DNA-binding and activation as two separable functions in two separable domains of the protein," Brent says. "It also showed there was nothing special about the way [proteins] bind DNA in order to make activation happen. All that matters is that you drag the activating portion to striking distance of the gene."

Suspecting that Lex A's ability to "drag" proteins to the DNA could provide a tool for understanding how proteins interact with other proteins, as well as with genes, Brent, now at HMS, embarked on a new set of yeast hybrid experiments.

First, Brent and his colleagues fused the Lex A binding domain to a protein whose partner they wanted to find (bait). They then created a pool of potentially interacting proteins (prey), and fused each one to an all-purpose activator.

Both hybrids-bait and prey-were introduced into yeast cells. The researchers found that when the bait hooked a prey protein, the activator portion fused to the latter was brought close enough to the DNA to turn on a gene. (The gene, known as a reporter gene, was one designed to have a clearly observable effect, such as turning the yeast cell blue or enabling it to grow in a special medium).

Protein Pairs

The availability of this tool has also created a dilemma for Brent and his colleagues. "Using these newly- minted two hybrid yeast methods as well as a number of other new-age yeast techniques, we-the lab-found ourselves in a position that is familiar to a lot of geneticists. We have an embarrassingly large number of proteins we think are important. And we're not sure what they do," says Brent.

The researchers began searching in 1993 for a method that would not only pick out interacting proteins but also show what the interacting protein pairs actually do. "We started with the proverbial blank sheet of paper," Brent says. Gradually, a plan emerged to design a molecule that would drive a wedge between a protein and one of its protein partners, while leaving other interactions alone. They could then observe the effect of blocking a single protein-protein interaction.

To create the wedge, Brent and his colleagues took short strings of amino acids, each one different, and fused each end of the peptide to a large protein platform. "Instead of having a small floppy peptide, you would have a peptide within the context of a larger molecule and that would help it to bind tighter to the target protein," Cohen says.

The researchers then tested their model aptamers against a protein known to be involved in cell proliferation in humans. The first step was to see if the peptide aptamers could bind to the Cdk2 target protein in test tubes-which they could (figure 1).

The researchers then looked for bound peptide aptamers that could actually prevent the Cdk2 target protein from interacting with one of its substrates (again in test tubes). "We found some of them do what we want them to; they screw up interactions between the substrate and the target molecule," says Brent. Specifically, the peptide aptamers prevented Cdk2 from adding phosphates to a particular substrate, Histone H1 (figure 2).

The researchers are trying to see if the aptamers work in cells as well as in test tubes. "We know the aptamers can hit their targets-bind to proteins-inside of cells, but we have not yet shown that they can inhibit interactions with substrates," Cohen says.

Meanwhile, Brent envisions designing more intricate molecules to serve as research tools. "Let's make more complicated machines to help us learn about biology inside cells," he says. For example, an aptamer with two different peptides instead of one could bring protein partners together instead of driving them apart, much in the fashion of a "matchmaker" molecule. "So we can have fun studying cell proliferation," Brent says.

"If you want to find out how genes
affect other genes, you have to find
out how proteins affect other proteins."
--Roger Brent

While use of peptide aptamers for fighting disease may be many years away, they should quickly find a place in the laboratory as a new tool for molecular biologists. New genes and their protein products are being identified every day, with little clue as to what they actually do. "We're working with too many proteins for our liking and the world is working with way too many proteins. The genome projects are gushing forth with data now," Brent says.

He hopes the peptide aptamers will help researchers sort through the growing morass. "The thing we built the peptide aptamers for is this: If we have increasing data about which proteins touch each other and the patterns of connections inside of cells, then let's get a tool that will allow us to determine the importance of a particular protein- protein connection. And let's ready it in time for the cresting Human Genome Project," says Brent.

Brent is currently working with childhood mathematician friends at other institutions to analyze hundreds of thousands of protein-protein interactions. "Does protein A interact with B, C, D, E-yes or no? By saving the data we begin to build up a map of connections that proteins make to one another," Brent says. "So we're building up these big- old maps. I can't promise they'll come through, but it's a lot of fun."

--Misia Landau