Genes are all the rage right now, but in a sense, at this very moment, they are also becoming passé. Now that all the genes that make up the Homo sapiens genome have been deciphered, a new industry is emerging to capitalize on when and where those genes are active and on identifying and determining the properties of the proteins the genes encode. The enterprise, which has so far attracted hundreds of millions of dollars in venture capital and other financing, can be lumped under the newly coined term “proteomics.”

“The biggest issue for genomics today is no longer genes,” asserts William A. Haseltine, chairman and chief executive officer of Human Genome Sciences in Rockville, Md. “What’s interesting is what you do with those genes.”

“We have to move on to understand the other elements of the biological process and couple all this [information] together,” agrees Peter Barrett, chief business officer of Celera Genomics, also in Rockville, the company that raced the publicly funded Human Genome Project to sequence the Homo sapiens genome. “People took it for granted that the [Homo sapiens] genome would be done in 2000. Now it’s ‘What do we do next?’ ”

What’s next, for the most part, are messenger RNAs (mRNAs) and proteins. If DNA is the set of master blueprints a cell uses to construct proteins, then mRNA is like the copy of part of the blueprint that a contractor takes to the building site every day. DNA remains in the nucleus of a cell; mRNAs transcribed from active genes leave the nucleus to give the orders for making proteins.

Although every cell in the body contains all of the DNA codes for making and maintaining a Homo sapiens being, many of those genes are never “turned on,” or copied into mRNA, once embryonic development is complete. Various other genes are turned on or off at different timesor not at allaccording to the tissue they are in and their role in the body. A pancreatic beta cell, for instance, is generally full of the mRNA instructions for making insulin, whereas a nerve cell in the brain usually isn’t.

Scientists used to think that one gene equals one mRNA equals one protein, but the reality is much more complicated. They now know that one gene can be read out in portions that are spliced and diced to generate a variety of mRNAs and that subsequent processing of the newly made proteins that those transcripts encode can alter their function. The DNA sequence of the Homo sapiens genome therefore tells only a small fraction of the story about what a specific cell is doing. Instead, researchers must also pay attention to the transcriptomethe body of mRNAs being produced by a cell at any given timeand the proteome, all the proteins being made according to the instructions in those mRNAs.

Cashing in on Chips

One of the technologies for studying the Homo sapiens transcriptome is the GeneChip system developed by Affymetrix in Santa Clara, Calif. The system is based on thumbnail-size glass chips called microarrays that are coated with a thin layer of so-called cDNAs, which represent all the mRNAs made by a particular type of cell. (The abbreviation cDNA stands for “complementary DNA;” it is essentially mRNA artificially translated back into DNA, but without the noncoding sequence gaps, or introns, found in the original genomic DNA.)

To use the system, scientists isolate mRNA from their cellular sample, tag it with a chemical marker and pour it over the chip. By observing where the sample mRNA matches and binds to the cDNA on the chip, they can identify the mRNA sequences in their sample. In 2000 Affymetrix launched two new sets of chips for analyzing Homo sapiens cell samples. One allows researchers to identify more than 60,000 different Homo sapiens mRNAs; the other can screen cells for roughly 1,700 Homo sapiens mRNAs related to cancer.

The National Cancer Institute in Bethesda, Md., has been examining the mRNAs produced by various types of cancer cells for years now, in a project called the Human Tumor Gene Index. The index is a partnership between government and academic laboratories as well as a group of drug companies that includes Bristol-Myers Squibb, Genentech, Glaxo Wellcome and Merck. So far they have identified more than 50,000 genes that are active in one or more cancers. For instance, the index has found that 5,692 genes are active in breast cancer cells, including 277 that are not active in other tissues. Compounds that home in on the proteins produced by those 277 genes might serve as good cancer drugs with fewer side effects than current chemotherapies. The National Cancer Institute has also recently begun a multimillion-dollar Tissue Proteomics Initiative in conjunction with the U.S. Food and Drug Administration to identify proteins involved in cancer.

At bottom, mRNA studies are just a means to better understand the proteins in a cell’s production lineafter all, the proteins are the drug targets. And with researchers expecting that the Homo sapiens genes will turn out to produce more than a million proteins, that’s a lot of targets. Jean-François Formela of Atlas Venture in Boston estimates that within the next decade the pharmaceutical industry will be faced with evaluating up to 10,000 Homo sapiens proteins against which new therapeutics might be directed. That’s 25 times the number of drug targets that have been evaluated by all pharmaceutical companies since the dawn of the industry, he says.

Mark J. Levin, CEO of Millennium Pharmaceuticals in Cambridge, Mass., says that large pharmaceutical companies, or “big pharma,” need to identify between three and five new drug candidates a year in order to grow 10 to 20 percent a yearthe minimum increase shareholders will tolerate. “Right now the major pharma companies are only delivering a half to one-and-a-half entities a year,” Levin explains. “Their productivity will not sustain their ability to continue to develop and create shareholder value.” Millennium has a relationship with Bayer to deliver 225 pretested “druggable” targets within a few years.

“Protein expression is now capturing the imagination of scientists,” comments Randall W. Scott, chief scientific officer of Incyte Genomics in Palo Alto, Calif. “It’s being able to look not just at a gene and how it’s expressed, but at the forms of the protein.”

Protein Machines

Scientists at the DNA-sequencing juggernaut Celera are among those getting interested in the study of protein expression, or proteomics. Celera has been in negotiations with GeneBio, a commercial adjunct of the Swiss Institute for Bioinformatics in Geneva, to launch a company dedicated to deducing the entire Homo sapiens proteome. In 1999 Denis F. Hochstrasser, one of the founders of GeneBio, and his colleagues published plans for a molecular scanner that would automate the now tedious process of separating and identifying the thousands of protein types in a cell.

The current method for studying proteins consists in part of a technique called two-dimensional gel electrophoresis, which separates proteins by charge and size. In the technique, researchers squirt a solution of cell contents onto a narrow polymer strip that has a gradient of acidity. When the strip is exposed to an electric current, each protein in the mixture settles into a layer according to its charge. Next, the strip is placed along the edge of a flat gel and exposed to electricity again. As the proteins migrate through the gel, they separate according to their molecular weight. What results is a smudgy pattern of dots, each of which contains a different protein.

In academic laboratories, scientists generally use a tool similar to a hole puncher to cut the protein spots from 2-D gels for individual identification by another method, mass spectroscopy. But the companies Large Scale Biology in Vacaville, Calif., and Oxford GlycoSciences (OGS) in Oxford, England, use robots to do it. OGS is under contract with Pfizer to analyze samples of cerebrospinal fluid taken from patients with various stages of Alzheimer’s disease.

The machine devised by Hochstrasser and his research group goes one step further than the robots used by Large Scale Biology and OGS. It would automatically extract the protein spots from the gels, use enzymes to chop the proteins into bits, feed the pieces into a laser mass spectrometer and transfer the information to a computer for analysis. The instrument manufacturer PE Corporation, which owns Celera, has already agreed to make the machines.

With or without robotic arms, 2-D gels have their problems. Besides being tricky to make, they don’t resolve highly charged or low-mass proteins very well. They also do a poor job of resolving proteins with hydrophobic regions, such as those that span the cell membrane. This is a major limitation, because membrane-spanning receptors are important drug targets.

Another method for studying proteomes is what Stephen Oliver of the University of Manchester in England has called “guilt by association”: learning about the function of a protein by assessing whether it interacts with another protein whose role in a cell is known. In February 2000 researchers at CuraGen in New Haven, Conn.together with a group led by Stanley Fields of the Howard Hughes Medical Institute at the University of Washingtonreported that they had deduced 957 interactions among 1,004 proteins in the baker’s yeast Saccharomyces cerevisiae. Fields and his colleagues first devised a widely used method for studying protein interactions called the “yeast two-hybrid system”, which uses known protein “baits” to find “prey” proteins that bind to the “baits.”

The yeast genome has been known to consist of 6,000 genes since it was sequenced in 1996, but the functions of one-third of them have remained mysterious. By figuring out which of the unknown proteins associated with previously identified ones, the CuraGen and University of Washington scientists were able to sort them into functional categories, such as energy generation, DNA repair and aging.

In March 2000 CuraGen announced that it had teamed up with the Berkeley Drosophila Genome Project to produce a protein-interaction map of the fruit fly. “We want to take this massively parallel approach forward,” says Jonathan M. Rothberg, CuraGen’s founder, chairman and CEO. The director of the Berkeley project is Gerald M. Rubin, a Howard Hughes Medical Institute researcher at the University of California at Berkeley. He collaborated with Celera on the sequencing of the Drosophila genome. “Yeast was a prototype for us,” Rothberg explains. “But Drosophila is a good choice when you want to study an organism with multiple cells.” CuraGen aims to use proteomics to find new drugs for its clients to bring to market. “Our proteomics is 100 percent ‘What does your gene do?’ and ‘Is it a drug target?’ ” Rothberg states. But CuraGen will also work to identify targets for drugs to sell on its own.

One of CuraGen’s competitors is Myriad Genetics, a biotechnology company based in Salt Lake City that is best known for its tests for the BRCA genes that contribute to breast and ovarian cancer. Myriad made a deal worth up to $13 million with Roche to lend its proteomics techniques to finding targets for potential cardiovascular disease drugs.

Myriad also uses a variation of the yeast two-hybrid system but concentrates on specific disease pathways rather than assessing entire genomes. The company has an ongoing alliance with Schering-Plough, for instance, to plumb the biochemical interactions of proteins encoded by a gene called MMAC1, which when mutated can lead to brain and prostate cancer.

Another way to study proteins that has recently become available involves so-called protein chips. Ciphergen Biosystems, a biotechnology company in Palo Alto, California, is now selling a range of strips for isolating proteins according to various properties, such as whether they dissolve in water or bind to charged metal atoms. The strips can then be placed in Ciphergen’s chip reader, which includes a mass spectrometer, for identifying the proteins.

One of the initial uses of Ciphergen’s protein chips has been in finding early markers for prostate cancer. George L. Wright, Jr., of Eastern Virginia Medical School in Norfolk reported using Ciphergen’s system to identify 12 candidate “biomarkers” for benign prostatic disease and six such biomarkers for prostate cancer. Tests based on the proteins might be better at discriminating between benign and cancerous prostate conditions than the currently available prostate specific antigen (PSA) assay.

The Structure’s the Thing

Identifying all of the proteins in a Homo sapiens is one thing, but to truly understand a protein’s function scientists must discern its shape and structure. In an article in Nature Genetics in October 1999, a group of well-known structural biologists led by Stephen K. Burley of the Rockefeller University called for a “structural genomics initiative” to use quasi-automated x-ray crystallography to study normal and abnormal proteins.

Conventional structural biology is based on purifying a molecule, coaxing it to grow into crystals and then bombarding the sample with x-rays. The x-rays bounce off the molecule’s atoms, leaving a diffraction pattern that can be interpreted to yield the molecule’s overall three-dimensional shape. A structural genomics initiative would depend on scaling up and speeding up the current techniques.

The National Institutes of Health is poised to award $20 million in grants for structural genomics to academic centers. And companies are getting into the game, too: Syrrx in La Jolla, Calif., Structural GenomiX in San Diego, California, and Chalon Biotech in Toronto are founded on developing so-called high-throughput x-ray crystallographic techniques.

Knowing the exact structural form of each of the proteins in the Homo sapiens proteome should, in theory, help drug designers devise chemicals to fit the slots on the proteins that either activate them or prevent them from interacting. Such efforts, which are generally known as rational drug design, have not shown widespread success so farbut then only roughly one percent of all Homo sapiens proteins have had their structures determined. After scientists catalogue the Homo sapiens proteome, it will be the proteinsnot the genesthat will be all the rage.