Cracking Cancer's Code
Internationally respected breast cancer researcher Joe Gray '68 pioneers individualized therapy using DNA sequencing technology.

He heads the Life Science Division of Lawrence Berkeley National Laboratory, and he is a highly respected figure in the international cancer research community, yet Joe Gray ’68 doesn’t have a formal biological-anything degree to his name.

Recipient of a Distinguished Achievement Medal from Mines in 2005, Gray is in fact a physicist.

How highly regarded is he in the cancer research community? In 2008, when the much-publicized Stand Up To Cancer initiative raised $100 million for research, the bulk of the proceeds were divided among five “Dream Teams” of scientists to fund some of the most ambitious research being conducted on cancer today. Gray was tapped to co-lead one of these teams.

He’s achieved such standing through a long series of accomplishments, beginning in the ’70s and ’80s when he made pivotal contributions to cell analysis technology and cytogenetic research. In the ’90s, he helped develop an inexpensive diagnostic now used to identify breast cancer patients whose tumors carry a particular genetic abnormality—an important indicator for selecting treatment. In the mid-nineties, he also helped develop a technology for analyzing genetic aberrations in cancer cells that is now used worldwide. Since then, he has continued to operate on the very front lines of the fight against cancer, organizing teams, rallying support and pushing forward with innovative research.

The clinical study funded by Stand Up To Cancer is a good example. It’s a new approach to cancer; one that literally maps out the mangled DNA sequence of each patient’s tumor (no two cancers are the same) and individualizes treatment based on this unique signature. If successful, it could mark the beginning of a fundamental shift in cancer treatment.

He is also a key player in another program with game-changing implications. The Cancer Genome Atlas Project is a broad collaboration among many researchers involving massive volumes of data. Their objective is to map the fundamental genomic characteristics of 20 different cancer types. Gordon Mills, chair of the Department of Systems Biology at MD Anderson Cancer Center, the world’s largest cancer research center, describes the first scientific paper to come out of the study as “spectacular … It’s changing how we are thinking about cancer, going from thinking about single molecules to pathways and networks… it’s had a major impact.” He adds that “Joe was an important force in convincing the community to do this study. His vision and support were enough to swing a lot of people who would have said no otherwise.”

Given that Joe Gray is such a pillar of modern genomic research—a pioneer who has helped pave the way for today’s explosion of knowledge in genetics and cancer research—it’s ironic that he began as a physics major at Colorado School of Mines, where there wasn’t a single biology course on offer.

“I sometimes refer to my scientific career as a random walk through science,” Gray says with a smile. But in fact, he’s had a purpose: “One of the things I’ve done in my career, more or less deliberately, is to change fields every now and then, so I could bring my skill set to bear on a new problem,” he says.

After graduating from Mines, Gray knew he wanted to study particle physics. Explaining why he chose Kansas State for graduate school, he says, “Being a Mines engineer type, I didn’t just want to study accelerator physics, I wanted to build the accelerator. Everyone else in the country had one built, and K-State had a hole in the ground, so I decided to go see how it went together.”

However, soon after building the accelerator, he found he was tiring of low-energy particle physics: the theory was already worked out; his experiments went exactly as expected; and, as he puts it, “once you’ve looked inside one nucleus, the next one looks very much the same.” At the same time, during his final year at Kansas State, an entirely different field of science had caught his attention.

He was sharing an office with several yeast geneticists. “They were having a grand old time,” Gray recalls. “It was day one of yeast genetics, and they had thousands of different things they were thinking about. Also, the experiments they were doing had long-term relevance to the human condition.”

So when an opportunity to work on a biomedical science project at the Lawrence Livermore National Laboratory came up, he jumped at the chance. They needed a cell analysis and sorting device built, says Gray, explaining that the new technology had not yet become commercially available. He was qualified for the job because, he explains, cell-sorting technology is very similar to the technology of particle accelerators.

Shortly after finding this unusual niche in the biological sciences, he made his first major contribution to the field. “It was literally a Friday afternoon eureka experiment,” says Gray, recalling the first time he used cell-sorting technology to separate chromosomes. “I borrowed a cup of chromosomes from a friend at Berkeley, stained them with a DNA dye, ran them through the system and bang, the chromosomes were resolved and separable! All of this happened in one experiment.” Prior to this discovery, separating chromosomes was a laborious process and didn’t produce a very pure supply. By providing a means of sorting them quickly, Gray made an immediate and significant impact on the pace of research in numerous branches of genetic science.

In the years that followed, Gray continued to work on chromosomes, notably through his involvement in the National Laboratory Gene Library Project. Though he describes himself as a foot soldier in the initiative, it laid important groundwork for the startling contribution he made to the biosciences in the mid-’80s.

Essentially, the gene library project involved compiling and organizing various collections of DNA from all 23 human chromosomes. The general objective was to make the human genome more accessible and accelerate the pace of research. Isolating a specific chromosome was hard enough; teasing one apart to study a specific portion was even more complex. However, using newly developed cloning technology and propagating copies in the gene library, it now only had to be done once.

Gray knew this was important work, but it didn’t suit him; a man of action, he began thinking ahead, musing about novel ways the gene library could be used. The technique that he ultimately came up with was revolutionary, illuminating the geography of the human chromosome like never before. Developed in collaboration with his colleague, Dan Pinkel, the technique utilized an emerging technology called “chromosome painting” or more formally, fluorescence in situ hybridization (FISH)—a way of using fluorescently labeled fragments of DNA as probes to label a specific region of a whole chromosome.

The impact of their research was significant. “Prior to this, it took a very highly skilled person to recognize an individual chromosome. With this new technique, it became trivial, even for me,” says the unpretentious Gray.

After the 1988 publication of their paper, chromosome painting took off. Multicolored processes were developed so that distinct areas of all 23 chromosomes could be rendered in contrasting colors. “It revolutionized the field of cytogenetics,” says Gray. The ability to observe chromosomes in such sharp relief made prenatal diagnosis of certain diseases easier, and allowed other diseases to be linked to chromosomal abnormalities for the first time.

Over time, as the 3 billion base-pairs of nucleotide molecules that comprise human DNA have become better understood, FISH probes have become more specific, their value being particularly felt in the fight against cancer, where runaway replication of specific genes is an underlying problem. With a specific enough probe, FISH analysis can indicate if a single gene is running amok, facilitating a more targeted approach to therapy.

Top: Fluorescent probes mark numerous copies of oncogene ErB-B2 in these breast cancer cell nuclei-normal nuclei should only have two.

Bottom: Probes make chromosomes 7 (red) and 11 (green) distinguishable from others (dyed blue) and the patch of red on one copy of chromosome 11, and patch of green on chromosome 7 indicate an aberrant exchange of DNA across chromosomes.

An example is breast cancer, where about 20 percent of cases are associated with excessive replication of a gene called ErB-B2, located in chromosome 17. If this is the case, the drug Herceptin is generally considered the most effective treatment; but if ErB-B2 is not a factor, Herceptin is ineffective and may harm the patient. To positively determine the status of the ErB-B2 gene in tumors, tens of thousands of patients each year have tumor cells checked using a patented FISH-analysis that Gray and Pinkel developed.

The ability to zero in on a single gene with FISH is clearly of great value; but it also illustrates a drawback of the technique. FISH is valuable when you know what you need to study; it’s not so helpful when you don’t, particularly with regard to cancer.

Tackling this problem, Pinkel, Gray and colleagues came up with a different technique, turning FISH around in such a way that with one procedure, they could measure abnormalities across a cell’s entire genome. Gray explains the modified approach, which they named comparative genomic hybridization (CGH): “Let’s say we want to study a tumor genome, and we don’t know which part of the genome is aberrant,” he begins. “We grind up the tumor genome and label it one color, say red, and grind up a genome from a normal cell and label it a different color, say green, and then hybridize both of those back to normal [intact] chromosomes.” The outcome is concentration-dependent: where the tumor cell has duplicated extra genes, the corresponding area of the normal chromosome will turn red; the normal chromosome will turn green where the tumor has lost DNA. “The beauty of this approach is you don’t have to know where the tumor is aberrant,” Gray points out. “It just maps it out for you.”

CGH proved extremely effective, serving for well over a decade as the cancer research community’s most comprehensive source of information on the genetic workings of cancer. After being developed in the mid-’90s, CGH was significantly refined: Today, slides with as many as a million probes bonded in a microarray can be digitally scanned to map abnormalities at a million different points within the human genome.

Amazingly, DNA mapping technology has now advanced past even the latest CGH arrays: the latest devices can effectively sequence the human genome, nucleotide by nucleotide, in just a few days. (The Human Genome Project took 12 years to complete the task.) The Gray Lab recently acquired one. Pointing it out, he shakes his head, saying, “That thing is unbelievable.”

After developing CGH in the mid-’90s, the focus of Gray’s research turned more exclusively to cancer. It is no coincidence that after losing his father to the disease in the mid-’70s, his work on chromosome analysis technologies meant he was contributing toward understand the disease. It wasn’t until CGH revealed the genetics of cancer in such detail and complexity that the disease itself became the primary focus of his work. He says he was struck by two things: how utterly scrambled cancer-cell DNA becomes, and how few similarities exist between patients’ tumors.

Suspecting the latter might explain why not all patients respond the same way to a particular therapy, his most recent work has looked for ways to determine effective treatment, individualizing it based on the DNA profile a tumor cell.

One of Gray’s studies has systematically analyzed the sensitivity of 50 different breast cancer cell lines (colonies of self-propagating cancer cells that can be kept alive in the lab) to varying concentrations of almost a hundred approved or experimental cancer drugs. The genomic signature of each cell line—determined using CGH and, more recently, next-generation sequencing technology—can then be compared to their drug responses in order to identify signatures that are strongly associated with response.

Although the study is not over yet—there are 400 cancer drugs on the market—Gray says patterns are emerging. “This drug works better when one set of genes are most active, that drug is more effective when another set of genes are most active,” he explains. And the results are compelling enough to warrant clinical testing, which is the objective of the trials Gray is preparing to launch, with support from the National Cancer Institute and Stand Up To Cancer.

Breast cancer patients who choose to participate in these studies will first have their tumors comprehensively analyzed. “The concept is to fingerprint the patient’s tumor,” explains Gray, “and then use the database of associations between genotype and drug response to select drugs that should be effective against the individual tumor.”

It’s a tremendously complex approach—the DNA sequence of a single cancer can occupy about a terabyte of memory (the hard drive capacity of about four average desktops). Given the volume of information, a central challenge of the study is managing and mining the information. “The computational area is a major weak link in the field right now,” says Gray. “We have way too few people who are computationally sophisticated enough to manage the data we are talking about. … I’m hiring as many of those people as I can find.”

But despite this limitation, Gray is confident that there’s a more effective approach to drug selection than is used today. Pointing to two graphs in a breast cancer textbook he’s authored, he explains that each one corresponds to the genomic abnormalities in tumor cells from two different breast cancer patients. Both patients were given exactly the same clinical diagnosis, yet the graphs bear no resemblance to one another. “There’s no reason to believe that these two cancers are going to respond the same way to therapy, but they are treated the same,” says Gray. “Our goal is to fix that.”

Asked whether he envisions a day when cancer is cured, Gray says, “I don’t think there are any silver bullets.” But he does envision a day when it’s highly treatable; when the disease is so thoroughly understood that individualized treatment plans, adjusted over time to track cellular changes in tumors, can keep cancer in check for many years. “We can do things today that just two years ago seemed impossible—not to be dreamed of,” says Gray.

Chromosome Painting

1. Millions of probes (strands of DNA from a gene library) are fluorescently dyed and mixed with chromosomes to be labeled.

2. Heat is used to denature DNA, causing nucleotide chains to separate.

3. As the mixture cools,
double-helix structure of DNA is restored, but with probes bonded to unique sections of the chromosomes to which they correspond.

4. Once unbound probes are rinsed away, the ones insinuated into the structure of the chromosome make that region easily distinguishable under a microscope.