Different Methods Are Used to Identify Gain-of-Function and Loss-of-Function Mutations

Cancer-critical genes are grouped into two broad classes, according to whether the cancer risk arises from too much activity of the gene product, or too little. Genes of the first class, for which a gain-of-function mutation drives a cell toward cancer, are called proto-oncogenes; their mutant, overactive forms are called oncogenes. Genes of the second class, for which a loss-of-function mutation creates the danger, are called tumor suppressor genes.

As we will see, both kinds of mutations can have similar effects in enhancing cell proliferation and survival. Thus, from the point of view of a cancer cell, oncogenes and tumor suppressors—and the mutations that affect them—are flip sides of the same coin. The techniques needed to find these genes, however, are different, and are dictated by whether they are made overactive or underactive in cancer.

Mutation of a single copy of a proto-oncogene can have a dominant, growth-promoting effect on a cell (Figure 23-24A). Thus the oncogene can be detected by its effect when it is added—by DNA transfection, for example, or through infection with a viral vector—to the genome of a suitable type of tester cell. In the case of the tumor suppressor gene, on the other hand, the cancer-causing mutations are generally recessive: both copies of the normal gene must be removed or inactivated in the diploid somatic cell before an effect is seen (Figure 23-24B). This calls for a different approach, to discover what is missing.

Figure 23-24

Cancer-critical genes fall into two readily distinguishable categories, dominant and recessive. Oncogenes act in a dominant manner: a gain-of-function mutation in a single copy of the cancer-critical gene can drive a cell toward cancer. Tumor suppressor (more...)

In some cases, a specific gross chromosomal abnormality, visible under the microscope, is repeatedly associated with a particular type of cancer. This can give a clue to the location of either an oncogene that is activated as a result of the chromosomal rearrangement (as in the example of the chromosomal translocation responsible for chronic myelogenous leukemia, mentioned earlier) or a tumor suppressor gene that is deleted; but the types of chromosomal abnormality involved in the two cases, and the circumstances in which they are typically encountered, are again different.

Oncogenes Are Identified Through Their Dominant Transforming Effects

Traditionally, geneticists identify genes by studying patterns of inheritance in families of individuals who show some heritable trait. To find where a mutation responsible for the trait lies in the genome, one looks for genetic linkage between this trait and genetic markers whose chromosomal position is known. This approach exploits an effect of sexual reproduction: genes that lie close to one another tend to be inherited together, whereas genes that are further apart are often separated by recombination during meiosis. In this way, a disease-causing mutation can be located by determining how far it lies from other genetic markers (see Chapter 8 for a detailed discussion).

Cancer cells do not reproduce sexually, so geneticists must use other methods to track down the mutations that make cancer cells different from their normal neighbors in the body. For oncogenes, a direct and conceptually simple (though laborious) approach is to scan the genome of the cancer cell for segments of DNA that will, when introduced into cells of a suitable tester cell line, drive them toward cancerous behavior. A mouse-derived fibroblast cell line is a convenient source of tester cells for the assay; these cells, which were previously selected to thrive in culture, are thought to already contain genetic alterations that take them part of the way toward malignancy. For this reason, addition of a single oncogene can be enough to produce a dramatic effect.

To detect an oncogene in this way, DNA is extracted from tumor cells, broken into fragments, and introduced into these fibroblasts in culture. If any of the fragments contains an oncogene, small colonies of abnormally proliferating—so-called ‘transformed’—cells may begin to appear. Each of these colonies will be composed of a clone of cells that originated from a single cell whose growth was promoted by the added gene. Because they have been released from some of the social controls on cell division, the transformed cells outgrow normal ones, piling up in layer upon layer in the culture dish as they proliferate (Figure 23-25).

Figure 23-25

Loss of contact inhibition in cell culture. Most normal cells stop proliferating once they have carpeted the dish with a single layer of cells: proliferation seems to depend on contact with the dish, and to be inhibited by contacts with other cells. Cancer (more...)

This assay, applied to DNA from human tumors, led to the first direct identification of an oncogenic mutation in a human cancer. Several overactive growth-promoting genes were identified by isolating and sequencing the DNA fragments that had been transferred to these transformed cells. The first of these genes to be sequenced was a mutant version of the Ras gene, which is now known to be mutated in about one in four human tumors.

This discovery was all the more dramatic because, shortly beforehand, a mutated Ras gene had been found to be the tumor-causing gene in a retrovirus that causes sarcomas in rodents. Retroviruses, we now know, are capable of picking up, at random, fragments of genetic material from their animal hosts and ferrying them from one infected individual to another. Occasionally, a proto-oncogene is picked up, in a damaged or misregulated form that turns it into an oncogene. Infection with a virus that carries such a cargo can trigger tumors in some animal species, and a large number of oncogenes were first discovered in this way.

Twenty years ago, the discovery of the same mutant gene in human tumor cells and in an animal virus that causes tumors was electrifying. The suggestion that cancers are caused by mutations in a limited number of cancer-critical genes made unraveling the exact nature of cancer seem a soluble scientific problem, and it helped to launch a transformation in our understanding of the molecular biology of cancer.

As we saw in Chapter 15, normal Ras proteins are monomeric GTPases that help to transmit signals from growth factor receptors on the cell surface. The point mutations found in Ras genes isolated from human tumors generate a hyperactive Ras protein that persists abnormally in its active state—transmitting an inappropriate signal for cell proliferation. Because this type of mutation makes a gene product hyperactive, the effect is dominant—only one of the cell's two gene copies needs to undergo the change. The Ras genes are mutated in a wide range of human cancers, and they remain one of the most important examples of cancer-critical genes.

Tumor Suppressor Genes Can Sometimes Be Identified by Study of Rare Hereditary Cancer Syndromes

Given a cancer cell, identifying a gene that has been inactivated requires a different strategy than finding a gene that has become hyperactive: one cannot, for example, use a cell transformation assay to identify something that simply is not there. Thus, the search for tumor suppressor genes has taken a route quite different from that followed in the hunt for oncogenes. The key insight that led to the discovery of the first tumor suppressor gene came from studies of a rare type of human cancer, retinoblastoma, which arises from cells in the body that are converted to a cancerous state by an unusually small number of mutations. As often happens in biology, the discovery arose from examination of a special case, but it turned out to be of universal relevance.

Retinoblastoma occurs in childhood; tumors develop from neural precursor cells in the immature retina. About one child in 20,000 is afflicted. There are two forms of the disease, one hereditary, the other not. In the hereditary form, multiple tumors usually arise independently, affecting both eyes; in the non-hereditary form only one eye is affected, and by only one tumor. Some individuals with hereditary retinoblastoma have a visibly abnormal karyotype, with a deletion of a specific band on chromosome 13. Deletions of this same locus are also encountered in tumor cells from some patients with the nonhereditary disease, suggesting that the cancer may be caused by loss of a critical gene in that chromosomal region.

Using the known location of the chromosomal deletion associated with retinoblastoma, it was possible to clone and sequence the gene whose loss appears to be critical for development of the cancer—the Rb gene. As would be predicted, in those who suffer from the hereditary form of the disease, a deletion or loss-of-function mutation occurs in one copy of the Rb gene in every cell of the body. Thus, these cells are predisposed to becoming cancerous, but are not actually cancerous so long as they retain one good copy of the gene. The retinal cells that do become cancerous are defective in both copies of Rb because a somatic mutation has occurred, in addition to the original inherited mutation, and has eliminated the remaining good copy. In patients with the nonhereditary form of the disease, by contrast, the noncancerous cells show no defect in either copy of Rb, while the cancerous cells are again defective in both copies. These nonhereditary retinoblastomas are very rare, because they require the coincidence of two somatic mutations in a single retinal cell lineage, so as to destroy both copies of the Rb gene (Figure 23-26).

Figure 23-26

The genetic mechanisms underlying retinoblastoma. In the hereditary form, all cells in the body lack one of the normal two functional copies of the Rb tumor suppressor gene, and tumors occur where the remaining copy is lost or inactivated by a somatic (more...)

The Rb gene subsequently turned out to much more than a gene mutated in a rare childhood tumor: it is also missing in several common types of cancer, including carcinomas of lung, breast, and bladder. These more common cancers arise by a more complex series of genetic changes than does retinoblastoma, and they make their appearance later in life and in other tissues of the body. But in all of them, it seems, loss of Rb is frequently a major step in the progression toward malignancy. The Rb gene encodes the Rb protein, which is a universal regulator of the cell cycle that is normally expressed in almost all the cells of the body (see Figure 17-30). Because it acts as one of the main brakes on progress through the cell-division cycle, the loss of Rb can allow cells to enter the cell division cycle inappropriately, as we shall discuss in the section on the molecular basis of cancer cell behavior.

Tumor Suppressor Genes Can Be Identified Even Without Clues from Heritable Cancer Syndromes

Hereditary cancer syndromes such as retinoblastoma are very rare, as we have already emphasized, and only a few—though important—tumor suppressor genes have been discovered by studying them. Where there is no such clue from a hereditary syndrome, we face the arduous task of identifying tumor suppressor genes simply by virtue of their absence from tumor cells. This involves comparing the tumor cells with non-cancerous cells from the same patient and discovering what exactly, out of the 3 billion nucleotides of the human genome, is missing or functionally defective. Because of the genetic instability of cancer cells, there is usually a great deal missing. Most of the defects are random and accidental by-products of the genetic instability. The tumor suppressor genes can only be identified by the criterion that they are repeatedly missing or defective in many independent cases of the cancer.

Gene loss often occurs by deletion of a relatively large segment of a chromosome. As a result, one copy of a tumor suppressor gene in a cancer cell may, for example, undergo an inactivating point mutation, while a gross deletion eliminates the other copy along with some neighboring genes (deletion of a large cluster of genes on both chromosomes is likely to kill the cell). The large defect on one chromosome makes detection of the loss much easier.

One deletion detection strategy takes advantage of the normal human genetic variation that makes the maternal and paternal chromosome sets distinguishably different. On average human DNA sequences differ—that is, we are heterozygous—at roughly one in every thousand nucleotides. Where a large segment of one chromosome has been lost, there is consequently a loss of heterozygosity: only one version of each variable DNA sequence in that neighborhood remains. A huge number—over a million—common sites of heterozygosity in the human genome have been mapped as part of the Human Genome Project: each of these sites is characterized by a specific DNA sequence that is known to be polymorphic—that is, to occur commonly in two or more slightly different versions in the human population. Given a sample of tumor DNA, one can check which of the versions of these polymorphic sequences are present. The same can be done with a sample of DNA from non-cancerous tissue from the same patient, for comparison. Absence of heterozygosity throughout a region of the genome containing many polymorphic sites, or loss of a genetic marker sequence that is seen in the non-cancerous control DNA, points the way toward a chromosomal region that has been deleted.

Techniques for large-scale DNA analysis and for the detection of deletions and other mutations are advancing rapidly, and they can be expected to add many more tumor suppressor genes to the present catalog.

Genes Mutated in Cancer Can Be Made Overactive or Underactive in Many Ways

We now know of at least 100 cancer-critical genes that can be converted into oncogenes by an activating mutation. The collection of genes whose absence or inactivation leads toward cancer, the tumor suppressor genes, is smaller but also growing. The ways in which genes of either class can be mutated to make them more—or less—active are enormously varied: as we might expect, any genetic accident that can increase, decrease, or change the activity of a gene is likely to be found somewhere in the increasingly complex catalogue of gene changes that occur in cancer.

The types of genetic alterations that can make a cancer-critical gene into an oncogene fall into three basic categories, as summarized in Figure 23-27. The gene may be altered by a small change in sequence such as a point mutation, by a larger-scale change such as a partial deletion, or by a chromosomal translocation that involves the breakage and rejoining of the DNA helix. These changes can occur in the protein-coding region so as to yield a hyperactive product, or they can occur in adjacent control regions so that the gene is simply expressed at concentrations that are much higher than normal. Alternatively, the cancer-critical gene may be overexpressed because extra copies are present due to gene amplification events caused by errors in DNA replication (Figure 23-28).

Figure 23-27

Three ways in which a proto-oncogene can be made overactive to convert it into an oncogene.

Figure 23-28

Chromosomal changes in cancer cells reflecting gene amplification. In these examples, the numbers of copies of a myc proto-oncogene have been amplified. Amplification of oncogenes is common in carcinomas and is often visible as a curious change in the (more...)

Specific types of abnormality are characteristic of particular genes and of the responses to particular carcinogens. For example, 90% of the skin tumors evoked in mice by the tumor initiator dimethylbenz[a]anthracene (DMBA) have an A-to-T alteration at exactly the same site in a mutant Ras gene; presumably, of the mutations caused by DMBA, only those at this site efficiently activate skin cells to form a tumor.

The receptor for epidermal growth factor (EGF), on the other hand, can be activated by a deletion that removes part of its extracellular domain. These mutant receptors are able to form active dimers even in the absence of EGF, and thus they produce a stimulatory signal inappropriately, like a faulty doorbell that rings even when nobody is pressing the button. Such mutations are found in many human brain tumors of the type called glioblastomas.

Members of the Myc proto-oncogene family, on the other hand, are not usually activated by mutations in the protein coding region; instead the genes are overexpressed or amplified (see Figure 23-28). In normal cells, the Myc protein acts in the nucleus as a signal for cell proliferation, as discussed in Chapter 17; excessive quantities of Myc can cause the cell to proliferate in circumstances where a normal cell would halt. Although Myc is frequently amplified in cancers, it can also be made active by a chromosomal translocation. As a result of this rearrangement, powerful gene regulatory sequences are placed inappropriately next to the Myc protein coding sequence, producing unusually large amounts of Myc mRNA. For example, in Burkitt's lymphoma a translocation brings the Myc gene under the control of sequences that normally drive the expression of antibodies in B cells. As a result, the mutant B cells proliferate to excess and form a tumor. Similar specific chromosome translocations are common in lymphomas and leukemias.

Tumor suppressor genes can also be inactivated in many different ways, with different combinations of genetic mishaps coming together to eliminate or cripple both gene copies. The first copy may, for example, be lost by a small chromosomal deletion or inactivated by a point mutation. Even epigenetic changes can inactivate a tumor suppressor gene: for example, its promoter can become methylated or the gene packed into heterochromatin, effectively shutting down gene expression. The second copy may be inactivated in a similar way, but more commonly it is eliminated by a less specific mechanism: the chromosome carrying the remaining normal copy may be lost from the cell, or the normal gene can be replaced by a mutant version through mitotic recombination or gene conversion. The range of possibilities for losing the remaining good copy of a tumor suppressor gene is summarized, using the Rb gene as an example, in Figure 23-29.

Figure 23-29

Six ways of losing the remaining good copy of a tumor suppressor gene. A cell that is defective in only one of its two copies of a tumor suppressor gene—for example, the Rb gene—usually behaves as a normal, healthy cell; the diagrams show (more...)

The Hunt for Cancer-Critical Genes Continues

The sequencing of the human genome has opened up new avenues toward the systematic discovery of cancer-critical genes. It is now possible in principle to draw up a practically complete list of the 30,000-odd genes in the human genome and to examine every one of them in a given cancer cell line, looking for potentially significant abnormalities, through automated analysis of either the mRNA that the cells produce or their genomic DNA. By applying this procedure to a reasonable number of different cancers, it should be possible to identify the genes that repeatedly undergo mutation in cancer. To carry out such a project in an exhaustive and systematic way is an enormously costly proposition, but not impossibly so, since less than 2% of the human genome actually codes for proteins. Such an effort is already under way. At the same time, a collaborative enterprise has begun to create an accessible central clearing-house for information about cancer-critical genes—their sequences, mutations, and expression profiles in a variety of cancerous and normal cells. This accumulation of data should lead eventually to an exhaustive catalog of the genes whose mutations contribute to cancer.

To search for cancer-critical genes in the ways we have described, there is in principle no need at the outset to know their normal functions or to understand how mutations in them cause cancer. The strategy, as in a genetic screen for genes involved in any other process, is first to find the culprits, then to work out how they commit their crimes. As our understanding of cell biology improves, however, it becomes easier to guess which genes are likely suspects, and to use these functional clues to track them down. The task of finding the genes is therefore closely entangled with the problem of discovering what they do. This is the topic of the next section.

Summary

Cancer-critical genes in general can be classified into two groups, according to whether the dangerous mutations in them are those that cause loss of function or those that cause gain of function. Loss-of-function mutations of tumor suppressor genes relieve cells of inhibitions that normally help to hold their numbers in check; gain-of-function mutations of proto-oncogenes stimulate cells to increase their numbers when they should not. These latter mutations have a dominant effect, and the mutant genes, known as oncogenes, are sometimes identified by their ability to drive a specialized line of tester cells toward cancerous proliferation. Many were first discovered because they cause cancer in animals when they are introduced by infection with a viral vector that has picked up genetic material from a previous host cell. Oncogenes can also be located by examining human cancer cells for genes targeted by activating mutations or by the chromosomal translocations that can signal the presence of a cancer-critical gene.

Mutations in tumor suppressor genes are generally recessive in their effects on the individual cell: there is no loss of control until both gene copies are put out of action. Current methods for finding these genes depend on scanning the genomes of cancer cells for signs of gene loss, often manifest as a loss of heterozygosity in a specific chromosomal region. Another approach has been to study cancers that run in families. Though these hereditary forms of cancer are rare, they have led to the discovery of tumor suppressor genes whose loss is a common feature of many cancers. Such cancer-prone individuals often inherit one defective and one functional gene copy of a tumor suppressor gene; they have an increased predisposition toward developing cancer because a single mutation in a somatic cell is enough, with the inherited mutation, to create a cell that totally lacks the tumor suppressor gene function. The recent sequencing of the human genome and the availability of increasingly powerful tools for systematically searching DNA for significant mutations should soon lead to a much more complete catalog of cancer-critical genes.