IMAGINE fighting a war against a mysterious enemy. You don鈥檛 know what he looks like, where he is hiding, or how or when he plans to strike. Your adversary is highly coordinated, with sophisticated tactics and weaponry to attack from many angles at once. All you have are a few rusty hand grenades to lob into the darkness. Worse, the enemy has surrounded himself with civilians, so any move you make is bound to create innocent victims.
This is what the war against cancer has been like. But now, cancer researchers say the human genome project is going to tip the balance back in our favour. It should tell us a lot about the nature of this inherently genetic disease-after all, every cancer springs from faulty genes-and eventually give us elegant weapons to target cancer before it can take hold.
There are three main steps on the way to this goal. First, identify the enemy. Which genes, out of the 100,000 or so in the human genome, can cause cancer when they go wrong? Second, find out as much as you can about the nature and habits of your opponent. How will he attack? Who is helping him? As well as generating lists of genes, it鈥檚 important to know what they do, and how they interact to cause cancer. 鈥淭he more we understand about the enemy, the more we can find out what its weak points are,鈥� says Larry Brody of the US National Cancer Institute near Washington DC. Third, choose your weapon, and strike! The final move is to find drugs that can halt cancer鈥檚 progression without harming healthy cells.
Advertisement
The whole process starts with the genes. Sometimes cancer is triggered by mutations in the DNA we inherit from our parents, but most of the errors that lead to the disease simply accumulate during our lives. 鈥淎ll through life we are exposed to radiation, chemicals and viruses, which cause DNA changes in our cells,鈥� says cancer geneticist Michael Stratton, from the Institute of Cancer Research in London. Genes that normally keep our cells healthy go wrong, and cells divide uncontrollably.
For most cancers, we don鈥檛 yet know exactly what goes wrong, so we fight the disease blindly. 鈥淭he treatment of cancer is still very miserable in most cases-the drugs we use are very crude instruments,鈥� says Brody. 鈥淥ften people are more afraid of the chemotherapy than the cancer.鈥� Most cancer drugs are general cell poisons that target all dividing cells, killing healthy ones along with tumour cells, which is why chemotherapy can often have severe side effects. If we could understand exactly what sends normal cells off the rails and what happens inside tumour cells, we could devise drugs specifically targeted to intercept the errant biochemical pathways. Or alternatively, we could look for an 鈥淎chilles鈥� heel鈥� that would let us kill tumour cells while leaving healthy cells alone.
The drugs of the future won鈥檛 just be specific to a particular kind of cancer, they will be tailored to individual patients. At the moment we group cancers into broad categories depending on the tissue they arise from, and talk of 鈥渂reast cancer鈥� or 鈥減rostate cancer鈥�. But really, every cancer is unique. Each tumour has a particular pattern of genes that are affected, which influences how the tumour develops and whether it becomes resistant to therapy or perhaps never responds in the first place. Understand the genes that cause those differences and we can develop drugs targeted to specific tumours, which would mean treatments are more likely to work first time.
Until now though, finding those genes has usually involved years of hard work. Most methods involve narrowing down a gene鈥檚 location to a certain part of the genome-for example, by looking at families with cancer to see which regions of the chromosomes show the same pattern of inheritance as the disease, or by looking down the microscope for visible changes in chromosome structure. Then researchers have to study all the hundreds of genes in that area to work out which is the culprit.
But the number of genes we can find this way may be limited, as many of them don鈥檛 show up using traditional techniques. Genes which only increase the risk of cancer slightly, or which mutate during our lifetimes, can鈥檛 be traced down family lines. And many of the mutations that lead to cancer involve very subtle changes in the DNA. For example, in a cancer-causing gene called ras, the genetic code is only altered at a single DNA base. The abnormality would be invisible to methods designed to pick up large losses or rearrangements of DNA. So Stratton is using the genome sequence to step up the search for cancer genes, under the aegis of the Cancer Genome Project at the Sanger Centre in Cambridge. He and his colleagues lift normal sequences directly from the project and compare them with the DNA from cancer cells to look for differences that might point to cancer-causing genes. 鈥淲ith all the genes mapped, you can take on the whole genome,鈥� he says.
Stratton鈥檚 ultimate aim is to collect as many different cancer samples as possible, identify genes that are altered, and then work out how important each of these genes is. 鈥淚f the same gene is mutated 100 times in 100 different cancers, then that is a hint that it is causative,鈥� he says.
Until now, the main approach has been to concentrate on one gene at a time. Our cells are quite robust, however, and one defective gene is not usually enough to cause cancer-several genes have to go wrong before a cell鈥檚 control mechanisms are overwhelmed. In the wake of the Human Genome Project, a new device called a 鈥淒NA chip鈥� will let researchers screen for the activity of thousands of genes at once.
Pass the chips
Each chip is a thin slice of glass or silicon, about the size of a postage stamp, on which fragments of up to 20,000 different genes fit in a grid (see Diagram). Researchers take mRNAs-molecules only produced by active genes in the first step of protein synthesis-and wash them over the chip. Each mRNA only binds to the fragment of the gene from which it came, and others are rinsed away. By labelling each mRNA with a fluorescent tag, it is possible to measure the level of activity of every gene represented on the chip. Researchers can use the chips to compare patterns of gene activity in normal cells and cancer cells, and in different kinds of cancers. They try to match the pattern they see to the way the tumour behaves and how it responds to treatment.
Earlier this year, for example, Louis Staudt and his colleagues from the NCI used a 鈥渓ymphochip鈥� carrying 18,000 gene fragments to study diffuse large B-cell lymphoma, a cancer that affects the lymphatic system and blood. The researchers could separate the tumours into two new classes, based on their gene activity, even though the tumours looked the same under the microscope. The distinction was an important one: patients from one class responded well to chemotherapy, while the others did not.
Staudt is now involved in a study called the Lymphoma/Leukemia Molecular Profiling Project, looking at hundreds of patients with B-cell lymphoma to see if those results hold up. When cancers can be diagnosed more precisely, Staudt says, we will customise therapies to specific subclasses of tumours. Certain patients could receive bone marrow transplants when first diagnosed, for example, instead of wasting time on chemotherapy that was unlikely to work.
In Britain, scientists at the Medical Research Council鈥檚 Clinical Sciences Centre in London are using chips to study why some tumours become resistant to drugs and others don鈥檛. 鈥淲e know the proteins that make cancers resistant to chemotherapy,鈥� says centre head Chris Higgins. 鈥淏ut we don鈥檛 know why there is expression [of the proteins] at high level in some tumours and not in other similar tumours.鈥� He wants to be able to screen tumours before starting chemotherapy to see whether or not they will be resistant, and buy time for the patient by using the right drug straight away. The ultimate aim is to design drugs to prevent resistance.
To do that, though, researchers need to do more than just screen for gene activity-they need to find out exactly what the genes are doing. Researchers are using DNA chips for this, too. Cancer鈥檚 most deadly attribute is metastasis, which happens when cells from the original tumour invade other parts of the body. Eric Lander and Todd Golub of the Whitehead Institute in Boston compared gene activity in mouse and human melanoma cells that readily metastasised with those that made poor invaders. They found three genes that were much more active in all the invasive cells they studied. When they ramped up the activity of one of the genes, rhoC, in a cell line that wasn鈥檛 good at metastasis, the cells became highly invasive. Inhibiting rhoC鈥榮 activity in aggressive cells had the reverse effect, suggesting the gene could be a good target for drugs to stop cancer spreading.
Meanwhile, Brody is looking for the function of BRCA1. Women who inherit this gene have a high risk of developing breast cancer. Brody is switching the gene on in normal cells, and using a sequence of chips over time to see what other genes it affects. 鈥淓ach chip gives you a snapshot,鈥� he says. 鈥淭his helps to assemble the pathway.鈥� He and his colleagues are building a cell line where they can turn BRCA1 on and off at will, to study the effects of the gene, and also see whether switching the gene off at different stages of the cancer can help halt its progression.
The work doesn鈥檛 stop at human genes. Researchers are also looking to familiar lab workhorses to help them understand the tricks that cancer plays. At the Genomics Institute of the Novartis Research Foundation (GNF) in San Diego, mice are the animal of choice. Colin Fletcher and his team aim to create random DNA mutations in hundreds of different mice to see which cause diseases like those in humans. This information, combined with the mouse genome sequence, which Celera Genomics will soon complete, should help them identify genes that might also cause human disease.
At the biotech company Exelixis in San Francisco, researchers are studying cancer using a range of organisms, including nematode worms, zebrafish and fruit flies. 鈥淲hen I explain to my mother what we are doing, I say we are using little insects to attack cancer,鈥� says Geoffrey Duyk, the company鈥檚 chief scientific officer. 鈥淚t sounds absurd, but actually nature is using the same biochemical strategies again and again.鈥� By studying the effects of cancer genes in simpler systems, researchers get valuable clues to their function, which they can then go back and try out in human cells. Duyk says that genome sequence data collected from many different organisms can help us see the links between species. 鈥淚t is a leap of faith that aspects of biochemistry are conserved across millions of years of evolution,鈥� he says. 鈥淪o it is reassuring to see the similarities on the databases.鈥�
Beyond genes
Other researchers are looking further than genes. While genes carry the instructions for making proteins, it鈥檚 the proteins that make things happen in the cell. So the US鈥檚 Food and Drug Administration (FDA) and National Cancer Institute (NCI) have teamed up in a 鈥淭issue Proteomics Initiative鈥�, applying the latest technology to see what proteins are doing in cancer cells. 鈥淚t is the proteins that really tell you what is going on in the functioning cell, moment to moment,鈥� says tumour pathologist Lance Liotta of the NCI. He and Emanuel Petricoin of the FDA are studying cells taken from thousands of different cancer samples to see which proteins are present at different stages of the disease.
To do this they are developing 鈥減rotein chips鈥� that hunt for thousands of proteins at once (see 鈥淗igh in protein鈥�). As well as monitoring the changes in cells that become cancerous, they are studying the effects of different drugs on tumour cells, and linking the information back to the patients the samples came from. 鈥淲e have ways of fingerprinting the individual patient, and we can monitor in their tissue the effects of the drug,鈥� says Petricoin. Ultimately, he and Liotta aim to develop a 鈥渨iring diagram鈥� of how all the proteins in a cell interact. It will tell us who is talking to whom, says Liotta.
But none of these efforts will matter much unless we can translate our knowledge into drugs that will send cancer into retreat. The big difference that genomics will make to this fight is specificity. It is the difference between a hand grenade and a sharpshooter when trying to pick out the bad guys from a crowd of innocents. And there will be many more different drugs around, each tailored to a specific tumour type. 鈥淚n the future, perhaps you will take a little biopsy from the tumour, do molecular or genomic profiling, say `this is tumour 61/3-B鈥�, and say at the drugstore, `I want the drug that kills this tumour鈥�,鈥� predicts Brody.
One example of this approach is a small molecule called STI-571. It is a drug for chronic myeloid leukaemia (CML), a blood cancer caused when part of chromosome 9 swaps with part of chromosome 22 in a white blood cell, creating a new structure called a 鈥淧hiladelphia chromosome鈥�. Cytologists looking through microscopes at the cancerous cells discovered the abnormality back in the 1960s, but it wasn鈥檛 until the late 1990s that geneticists realised that Philadelphia chromosomes make cells cancerous because the mutation creates a new, active gene.
Targeting tumours
This gene encodes an enzyme that makes cells grow out of control, so scientists at Novartis designed a chemical to block the action of the aberrant enzyme. The drug went into clinical trials a year ago, and so far has been an unprecedented success. 鈥淲e can see the cells carrying the mutation being eliminated from the blood,鈥� says Paul Herrling, head of research at Novartis Pharma in Basle. 鈥淭he drug hits only the cells that carry the translocation. It brings a completely new quality to the therapy.鈥� With the human genome sequence, the drug discovery process that has taken nearly forty years for CML will be much accelerated.
Automation should speed things up even more dramatically. At GNF, for example, a team led by Jeremy Caldwell is developing a system to rapidly find drugs that have a specific action. For example, the team might engineer cultured cells so that they fluoresce green whenever a particular cancer-causing gene turns on. By using these telltale cells in a machine that automatically exposes them to many different potential drugs, the team expects to screen 100,000 chemicals per day for their ability to turn the gene in question on or off.
Because the new generation of cancer drugs will be so specific, they should be safer, as well as more effective. So once researchers identify the genes that predispose people to cancer (see 鈥淐lose cousins鈥�), they should be able to design preventative drugs that stop tumours before they strike. 鈥淭he drugs would have to be extremely safe because you start to take them before you have any problems, and you take them for a long period of time,鈥� explains Herrling. In the future, we could be screened early in life to see which cancers we are likely to develop later on, and doctors could prescribe preventative measures-either lifestyle changes or drugs to protect our cells. 鈥淚t would work in a similar way to cholesterol-lowering drugs today,鈥� says Duyk.
For example, oncologist Kenneth Kinzler at Johns Hopkins University in Baltimore and his colleagues are working on blocking colon cancer in mice bred to be susceptible to the disease. By using a combination of drugs to attack two of the pathways leading to cancer in the mice, they reduced tumour development by 95 per cent, and half the mice developed no tumours at all. The researchers are now planning clinical trials in humans, to begin in a year鈥檚 time.
But while the future seems bright, we may not see the full benefits of the human genome project for many years. Cancer research will be vastly accelerated, but any candidate drugs will still have to go through stringent clinical trials before reaching the market. For preventative drugs in particular, trials won鈥檛 be easy. 鈥淎t what age do you give the drug, how long for, and how long do you have to wait before you know if it has worked?鈥� asks oncologist Bruce Ponder of Cambridge University. Trials will last years, and involve huge numbers of people. 鈥淓ven if we had the ideal drug target today, it would still take 10 to 15 years and $100 million to turn that into a drug,鈥� says Higgins. Then maybe we鈥檒l win that war.
