By now most people are familiar with philosopher Daniel Dennett’s characterization of natural selection as Darwin’s dangerous idea-- dangerous because it acted as a corrosive acid capable of dissolving the established structures of human society. That acid can be just as corrosive of scientific structures, which one might have thought more impervious to the damage. Thus a Darwinian idea has eaten away at some of the foundations of my own field of research for the past half-century, tumor biology, and forced cancer researchers to reexamine some cherished notions about the origins of cancer that were current during the first half of the century. Today, with the discovery of new genes that contribute to the development of cancerous cells, we are keenly aware that cancer is, above all, a disease of DNA. But more important, we know that this disease does not occur in a preprogrammed manner. Only through the gradual emancipation of a cell from the controls that govern its normal process of division does a cell turn cancerous. And that emancipation, it turns out, proceeds by the mechanics of Darwinian evolution.
In hindsight, perhaps, that is not surprising. Since Darwin’s day we have known of the power of natural selection to shape the organisms of the world. And over the past 50 years biologists have come to understand how mutations in DNA provide the array of genetic variation through which natural selection operates. Yet the importance of evolution has only slowly crept into the field of cancer research. To be sure, the process by which cells of the body turn malignant is a very limited one compared with the evolution of a species. But just as we have come to understand that microorganisms evolve resistance to drugs, we now know that cancerous cells evolve to become unresponsive to the growth-controlling forces of the body. How those genetic changes occur is based on Darwinian principles of variation and selection.
That insight changes our understanding of cancer. It lays to rest the hopes of finding a single key change or infectious agent that can explain all forms of the disease. When I started working in cancer research, in the late 1940s, the search for that key change was still in full swing, and it wasn’t long before one prominent theorist--the famed biochemist Otto Warburg, of the Max Planck Institute in Berlin--thought he had found it. Warburg proposed that what made cancer cells different from others was their unusual use of the cell’s energy sources--sugar and oxygen. As it happened, Warburg devised his theory, in part, based on cells I had worked with. When I gave my first talk at an international congress in 1950, I was one of the most junior participants. I spoke about ascites tumors in mice, which are generated by the growth of freely floating cancer cells in animals’ abdominal fluid. Unbeknownst to me, an assistant of Warburg’s was in the audience. A week later the great man sent me a letter requesting the cells, which I promptly sent.
In the following year or two, Warburg published several papers stating that ascites tumor cells preferred to burn sugar as if oxygen was not available, even if it was. He concluded that cancer cells, unlike normal cells, could thrive under conditions of great oxygen shortage. Some years later Warburg wrote that I had made a very important contribution to cancer research by sending him the cells with which he had solved the cancer problem.
Unfortunately, I was not impressed. One of the recurring problems in cancer research is that cells in the laboratory often behave in ways that would be far-fetched for cells in the body. It seemed to me that Warburg’s choice of a cancer cell that had passed through a large number of mice over more than half a century--under highly crowded conditions in a fluid containing little free oxygen--was like choosing the whale for studying the mechanisms of walking in quadrupeds. In the laboratory, it is easy to create phenomena that have no parallel in the natural world.
Still, such experiments can yield historically important results. The ever-recurring theme in science is that what appears to be most important can turn out to be trivial, and vice versa: a seemingly unimportant discovery may later acquire paramount significance. In this process, theories are like the scaffolding around a building that’s under construction: it exists only to be removed as the building grows.
The experimental results of Peyton Rous in 1911, for example, provided unexpected insights into tumor growth. Rous, a young researcher at the Rockefeller Institute in New York, suspected that cancer was caused by a virus--at the time a rather new and poorly understood entity. He soon had the opportunity to test his idea when a Long Island farmer sought his help in treating a prize hen with a tumor. In an attempt to isolate the cancer- causing pathogen, Rous removed the tumor, ground it up, filtered out the cells, and then injected the remaining cell-free material into a young chicken. The result: A cancerous growth. Rous concluded that the cells from the tumor produced an infectious agent that could transmit cancer.
Over the next four decades, many researchers tried to repeat Rous’s experiments in mice and rats, without success. However, in the 1950s that changed. Ludwik Gross, a Jewish refugee from Poland working at a va hospital in the Bronx, successfully isolated a virus that caused leukemia in mice. Soon after his discovery, other researchers began isolating viruses that, when injected into different types of experimental animals, could cause tumors. Some of these viruses could also turn normal cells in culture into cancer cells. By the 1960s and 1970s, the theory that cancer had a viral cause had developed a strong following.
These studies eventually identified two families of tumor viruses--dna viruses and rna viruses--with different modi operandi. When a DNA tumor virus inserts its genes among the genes of the host cell, it can disrupt the regulation of cell division, causing tumor growth. (Fortunately, the immune system usually recognizes--and eliminates--these altered cells.) More puzzling were the insidious reproductive habits of rna tumor viruses. It turned out that these viruses copy their own genetic material, which exists as rna, into double-stranded DNA. They then splice this DNA into the host cell’s DNA. In the cell’s DNA the virus can lie low and hide from the immune system. Because the researchers were ignorant of these things, they did not yet realize that the cancer-inducing effect of these viruses was merely a side effect of their lifestyle.
RNA viruses are effective but sloppy reproducers. Unlike the host cell, the virus doesn’t have any mechanism for proofreading what it copies into DNA. It can afford to produce a vast number of incorrect copies, including some that have accidentally picked up genes from the host DNA. Usually when this pickup occurs, other viral genetic information is lost. The resulting virus particles are so defective and disadvantaged that they could not survive in nature. But the tumor virologist, motivated by the desire to show that viruses can cause tumors, may save some of them from extinction.
Consider what Peyton Rous did back in 1911. He ground up the hen’s tumor, passed the material through a very fine filter that would not allow cells through, and injected the filtered material into newly hatched chicks. Then he looked for tumor development. What Rous could not have realized is that he was selecting virus particles that had accidentally picked up a host gene that promotes cell growth. He had selected the viruses that were capable not only of infecting new cells in the recipient chick but of prompting them to divide incessantly.
The key to unlimited growth was the stolen cellular gene, switched on by the virus, that forced the cells to divide without having been instructed by the normal signals of the organism. Not until some 60 years after Rous’s experiment did researchers realize that the cancer- inducing gene from Rous’s virus was in fact derived from a normal chicken cell. Later, other rna tumor viruses picked up from chicken, mouse, rat, or monkey tumors were found to harbor similar growth-promoting cellular genes. These genes were also found to play important roles in the spontaneous development of human tumors.
The hunt for virally encoded genetic information that could turn normal cells cancerous had instead led to the discovery that viruses could hijack and alter growth-regulating cellular genes. That finding highlighted the importance of DNA in tumor development, and later studies showed that mutations could turn on these genes in normal cells and promote cancer even without any viral intervention. Recognizing the role of mutations in the cellular DNA helped make sense of the emerging picture of steplike cancer development. Studies of the natural history of human cancers strongly hinted that they proceeded through a number of distinct stages, which emerged through a series of multiple changes that occurred at unpredictable intervals. In fact, back in the 1930s, Peyton Rous had begun documenting the changes in the tissue as cancer developed. He coined the term tumor progression to describe the process whereby tumors went from bad to worse.
Some 20 years later, Leslie Foulds, an experimental pathologist at the Chester Beatty Research Institute in London, formulated a set of rules to describe this process. He stressed the importance of distinguishing each of the traits that characterize the cells as they progress, step-by-step, toward cancer. Foulds’s work was critical for our later understanding of the role mutations play in the disease--one may, in fact, refer to the stepwise evolution of malignant tumors by sequential changes as Foulds’s dangerous idea. Foulds spoke about traits such as growth rate, hormone dependence, and the ability to invade surrounding tissues or to spread by metastasis. Moreover, he pointed out that these properties could change independently of each other as the tumor progresses. In other words, there didn’t appear to be one straight line a cell had to take to become cancerous.
Over the last four decades, research has fully vindicated Foulds’s ideas, with one important exception. Foulds believed that the changes were not caused by mutations. Instead, he hypothesized that the genes of a cancerous cell were normal; only their expression was disturbed. Thought of in this way, cancer was a disease of abnormal development, in which the wrong genes were being turned on and off. In this respect, Foulds was clearly wrong. Today we know that cancer is not only a disease of abnormal gene regulation but a disease of the DNA itself.
To Foulds, it seemed highly unlikely that mutations could be responsible for the steps of tumor development and progression. Each cell carries two copies of every gene, one from each parent. The two genes sit on two different chromosomes, and if one gene loses its function by mutation, its normal counterpart on the other chromosome can usually do the job. Random mutations were expected to affect only one of the two copies-- mutations of both genes seemed highly improbable.
We have learned, however, that it is easy to lose a second copy of a gene during cell division if the first copy is already impaired. Often the whole chromosome on which the second gene sits is lost. It turns out that cancer cells tolerate such losses very well because, unlike normal cells, they do not need to perform any specialized function. All they have to do is reproduce themselves.
In other words, the rules of their game have changed. Ordinary cells in multicellular organisms abide by rules that regulate their growth and ensure that they perform particular metabolic tasks. But as mutations accumulate, a cell stops being a team player and plays instead by the rules of natural selection. And those rules favor the fastest-growing cells. Many mutations may crop up during the evolution of a tumor, but it is the cell that has acquired the most growth-promoting mutations that will thrive and spread.
The genes involved in this gradual evolution fall into three somewhat overlapping categories. The first group are the oncogenes, the cancer-causing mutated genes that virologists first stumbled upon in the 1970s. All oncogenes urge cells to divide, and they can do so with a change in only one of the two gene copies.
The second group of genes are the so-called tumor suppressor genes. The first sign that normal cells may contain genes that can inhibit cancerous growth came from experiments performed nearly three decades ago by Henry Harris in Oxford in collaboration with our group, at the Karolinska Institute in Stockholm. When we fused normal and malignant cells, the resulting hybrid cell--and its progeny--were nonmalignant. But when some chromosomes from the normal parent cell were lost in the course of cell division in culture, the cells become malignant again. This indicated that tumor cells have suffered a genetic loss, and that normal genes can compensate for the loss.
Other researchers later identified the individual tumor suppressor genes. Eventually it became clear that suppressor genes make proteins that prevent inappropriate cell division. One of the best-known examples is the gene that produces the protein p53. Normal cells make very little p53. But whenever DNA is damaged--either by radiation, chemicals, or lack of oxygen--p53 levels rise dramatically. The p53 binds to DNA and prevents the cell from dividing--and thereby makes time for the DNA repair enzymes to perform their task. After the DNA has been repaired, p53 levels decline and cell division can go on. But if the damage has been too extensive, the cell undergoes programmed cell death, called apoptosis.
More than half of all human tumors contain mutated p53 that cannot bind to DNA and cannot, therefore, arrest the growth of cells with damaged DNA. The mutation does more than just impair the cell-death program. In cells where both copies of p53 are lost or mutated, damaged DNA doesn’t elicit the signals that halt growth long enough for it to be repaired. These cells nevertheless survive and are therefore prone to other mutations, including mutations in oncogenes and suppressor genes. This is why an inherited p53 mutation can lead to Li-Fraumeni syndrome, a condition in which patients often develop multiple tumors, arising in different tissues.
The third group of cancer-causing genes are the DNA repair genes themselves--the genes that ensure each strand of genetic information is correctly copied during cell division. Mutations in these genes predispose humans to hereditary nonpolyposis colon cancer syndrome. Families with this syndrome tend to be at risk for cancer in the colon, the rest of the gastrointestinal tract, the ovaries, the uterus, the urinary epithelium, and the skin. Mutations in at least five other DNA repair genes have now been discovered as well, and they are associated with other cancer syndromes.
The destabilizing effects of this set of mutations were first identified in organisms like bacteria and yeast. Because mutations in DNA repair genes increase the frequency of other mutations, they may enhance these single-celled creatures’ ability to survive in a stressful environment. But the same phenomenon in multicellular creatures like ourselves may result in cancer. The more cancer cells break away from the rules that ensure cooperation among the many cells of the body, the more they resemble populations of microorganisms. Among free-living bacteria, yeasts, and amoebas, for example, natural selection favors variants that can use nutrients and other resources more effectively. Among cancer cells, natural selection favors cells that are less and less responsive to the growth-controlling forces of the organism. And much as natural selection favors bacteria that can adapt to a new environment, so too does it favor cancer cells with mutations that help them survive in the low-oxygen environment of a growing tumor. In a way, Warburg was right, after all. But his observation about the altered energy needs of cancer cells only makes sense when viewed through the lens of Darwinian evolution.
Fortunately, it takes more than one genetic change to emancipate an ordinary cell from growth control. No single mutation is in and of itself cancer-causing. As Foulds suspected nearly four decades ago, cancer progression does not unfold in a rigid, predetermined manner. It unfolds slowly through a string of mutations, changes that provide a series of green lights to cellular growth.
How complete is our present picture of the three gene worlds that may influence cancer development? Are there others? Yes, certainly, but their study is at an earlier stage. Some genes influence the ability of the fledgling tumor cell to attract the blood vessels that bring it nutrients, a precondition for tumor growth. Others interfere with the process of normal cellular aging, helping to make the precancerous cell immortal. Still others help disguise the cancerous cell from the surveillance of the immune system.
What does our new understanding of tumor evolution hold for the future? Some of our clinical colleagues and most of the lay public have long awaited a cancer cure. Some say the investment in cancer research has been a waste, or in its nasty version, that cancer supports more people than it kills.
For the cancer biologists who have followed in Darwin’s and Foulds’s footsteps, there is no return. We must not only live with this new complexity but embrace it. Even though tumor development represents an evolutionary process on a very small scale, it is nonetheless an evolutionary process, with many subtle, seemingly unconnected steps, and with almost infinite variability. This does not mean that we have to know all the steps in minute detail before we can control the cancer cell. The new cancer biology may also provide methods to stop a multiply changed tumor cell in a single step.
Gene therapy may halt the growth of cancerous cells by introducing a powerful suppressor gene or a gene that promotes cell death. Still other approaches include cutting off a tumor’s blood supply--the source of its nutrients. If the capillary blood supply of the tumor is cut off, it will die. Yet another approach is to construct immune missiles composed of a toxin or a radioactive tag along with a specific antibody. Although the origins of cancer are far more tangled than Rous or Warburg ever imagined, the light of Darwin may yet let us find our way through the thicket.
