Studies of Developing Embryos and Transgenic Mice Help to Uncover the Function of Cancer-Critical Genes

Given a gene that is mutated in a cancer, we need to understand both how the gene functions in normal cells and how mutations in the gene contribute to the aberrant behaviors characteristic of cancer cells. When Rb was originally cloned, for example, all that was known was that it was deleted in a cancer. In the case of Ras, the mutant gene was known to direct cells to proliferate excessively and inappropriately in culture, but this observation did not reveal how the Ras protein functions in normal or cancer cells. In both cases, cancer research was the starting point for studies that revealed the key role of these gene products in normal cells—Rb as a cell-cycle regulator, Ras as a central component of cell-signaling pathways.

Today, we know much more about cells, so that when a new gene is identified as critical for cancer, it often turns out to be familiar from studies in another context. For example, many oncogenes and tumor suppressor genes are found to be homologs of genes already known for their role in embryonic development. Examples include components of practically all the major developmental signaling pathways through which cells communicate (discussed in Chapters 15 and 21): the Wnt, Hedgehog, TGFβ, Notch, and receptor tyrosine kinase signaling pathways all include important cancer-critical genes—with Ras being part of the last of these pathways.

In hindsight, this is no surprise. As we saw in Chapter 22, the same signaling mechanisms that control embryonic development operate in the normal adult body to control cell turnover and maintain homeostasis. Both the development of a multicellular animal and the maintenance of its adult structure depend on cell-cell communication and on regulated cell proliferation, cell differentiation, cell death, cell movement, and cell adhesion—in other words, on all the aspects of cell behavior whose derangement underlies cancer. Developmental biology, often using model animals such as Drosophila and C. elegans, thus provides a key to the normal functions of many cancer-critical genes.

Ultimately, however, we want to know what mutations in these genes do to cells in the tissues that give rise to the cancer. A certain amount of information can be obtained by studying cells in vitro or by examining human cancer patients. But to investigate how mutations in various cancer-critical genes affect tissues in a whole organism, the transgenic mouse has proved particularly useful.

Transgenic mice that carry an oncogene in all their cells can be generated by methods described in detail in Chapter 8. Oncogenes introduced in this way may be expressed in many tissues or in only a select few, according to the tissue specificity of the associated regulatory DNA. Studies of such transgenic animals reveal that, even in mice, a single oncogene is not usually sufficient to turn a normal cell into a cancer cell. Typically, in mice that are endowed with a Myc or Ras oncogene, some of the tissues that express the oncogene grow to an exaggerated size, and over time occasional cells undergo further changes to give rise to cancers. The vast majority of the cells in the transgenic mouse that express the Myc or Ras oncogene, however, do not give rise to cancers, showing that the single oncogene is not enough to cause malignancy. From the point of view of the whole animal, the inherited oncogene, nevertheless, is a serious menace because it increases the risk of developing cancer. Mice that express more than one oncogene can be generated by mating a pair of transgenic mice—one carrying a Myc oncogene, the other carrying a Ras oncogene, for example. These offspring develop cancers at a much higher rate than either parental strain (Figure 23-30), but again the cancers originate as scattered isolated tumors among noncancerous cells. Thus, even with these two expressed oncogenes, the cells must undergo further, randomly generated changes to become cancerous.

Figure 23-30

Oncogene collaboration in transgenic mice. The graphs show the incidence of tumors in three types of transgenic mice, one carrying a Myc oncogene, one carrying a Ras oncogene, and one carrying both oncogenes. For these experiments two lines of transgenic (more...)

Just as activated oncogenes can be introduced into mouse tissues, so tumor suppressor genes can be inactivated by ‘knocking out’ the gene in the mouse using reverse genetic techniques (see Figure 8-74). Several tumor suppressor genes have been knocked out in mice, including Rb. As anticipated, many of the mutant strains that are missing one copy of a tumor suppressor gene are cancer prone. Deletion of both copies often leads to death at an embryonic stage, reflecting the essential roles these genes play during normal development. To bypass this block and see the effect of homozygous mutations in an adult tissue, one can use the methods described in Chapter 8 to create conditional mutations, such that only one tissue—say the liver—displays the defect. In these ways, transgenic mice have become a key source of information as to the mechanisms of tumor formation. We shall see later that they also provide important models for the development of new cancer therapies.

Many Cancer-Critical Genes Regulate Cell Division

Most cancer-critical genes code for components of the pathways that regulate the social behavior of cells in the body—in particular, the mechanisms by which signals from a cell's neighbors can impel it to divide, differentiate, or die (Figure 23-31). In fact, many of the components of cell-signaling pathways were first identified through searches for cancer-causing genes, and a full list of proto-oncogene products and tumor suppressors includes examples of practically every type of molecule involved in cell signaling—secreted proteins, transmembrane receptors, GTP-binding proteins, protein kinases, gene regulatory proteins, and so on (discussed in detail in Chapter 15). Many cancer mutations alter signal pathway components in a way that causes them to deliver proliferative signals even when more cells are not needed, switching on cell growth, DNA replication, and cell division inappropriately. Mutations that inappropriately activate a receptor tyrosine kinase, such as the EGF receptor, or proteins in the Ras family, which lie downstream from such growth factor receptors, act in this way.

Figure 23-31

Chart of the major signaling pathways relevant to cancer in human cells, indicating the cellular locations of some of the proteins modified by mutation in cancers. Products of both oncogenes and tumor suppressor genes often occur within the same pathways. (more...)

Other signaling pathways can function to inhibit cell division, the best known example being the antigrowth effect of the TGFβ family of signaling proteins. Loss of growth inhibition through TGFβ-mediated pathways contributes to the genesis of several types of human cancers. The receptor TGFβ-RII is found to be mutated in some cancers of the colon and Smad4—a key intracellular signal transducer in the pathway—is inactivated in cancers of the pancreas and some other tissues.

Ultimately, the cancer-critical genes that regulate cell division exert their effects by acting on the central cell-cycle control machinery (see Figure 17-41). Not surprisingly, mutations in this machinery feature prominently in many cancers. As described in Chapter 17, a key point at which cells make the decision to replicate their DNA and enter the cell division cycle is thought to be controlled by the Rb protein, the product of the tumor suppressor gene Rb. Rb serves as a brake that restricts entry into S phase by binding to gene regulatory proteins needed to express genes whose products are required for progress round the cycle. Normally, this inhibition by Rb is relieved at the appropriate time by phosphorylation of Rb, which causes it to release its inhibitory grip.

Many cancer cells proliferate inappropriately by eliminating Rb entirely, as we have already seen. Other tumors achieve the same endpoint by acquiring mutations in other components of the Rb regulatory pathway (Figure 23-32). Thus, in normal cells, a complex of cyclin D1 and the cyclin-dependent kinase Cdk4 (G₁-Cdk) stimulates progression through the cell cycle by phosphorylating Rb (see Figure 17-30). The p16 (INK4) protein—which is produced when cells are stressed—inhibits cell-cycle progression by preventing the formation of an active cyclin D1-Cdk4 complex. Some glioblastomas and breast cancers are found to have amplified the genes encoding Cdk4 or cyclin D1, thus favoring cell proliferation. And deletion or inactivation of the p16 gene is common in many forms of human cancer. In cancers where it is not inactivated by mutation, this gene is often silenced by methylation of its regulatory DNA.

Figure 23-32

The pathway that controls cell cycling via Rb protein. All the components of this pathway have been found to be altered by mutation in human cancers (products of proto-oncogenes, green; products of tumor suppressor genes, red; E2F shown in blue because (more...)

The variety of ways in which the machinery of cell-cycle control can be altered in cancer illustrates two important points. First, it explains why individual cases of a particular cancer showing the same symptoms may arise from different mutations: in many cases several alternative mutations will have much the same effect on cell proliferation. Second, it reinforces the point that there is no fundamental difference in the processes that are affected by oncogenes—which become activated by mutation—and those affected by tumor suppressor genes—which become inactivated. These two classes of cancer-critical genes merely differ in whether they play a stimulatory or inhibitory role in a pathway (see also Figure 23-31).

Mutations in Genes That Regulate Apoptosis Allow Cancer Cells to Escape Suicide

To achieve net cell proliferation, it is necessary not only to drive cells into division, but also to keep cells from committing suicide by apoptosis. There are many normal situations in which cells proliferate continuously, but the cell division is exactly balanced by cell loss. In the germinal centers of lymph nodes, for example, B cells proliferate rapidly but most of their progeny are eliminated by apoptosis. Apoptosis is thus essential in maintaining the normal balance of cell births and deaths in tissues that undergo cell turnover. It also has a vital role in the cellular reaction to damage and disorder. As described in Chapter 17, cells in a multicellular organism commit suicide when they sense that something has gone wrong—when their DNA is severely damaged or when they are deprived of survival signals that tell them they are in their proper place. Resistance to apoptosis is thus a key characteristic of malignant cells, essential for enabling them to increase in number and survive where they should not.

A number of mutations that inhibit apoptosis have been found in tumors. One protein that blocks apoptosis, called Bcl-2, was discovered because it is the target of a chromosome translocation in a B-cell lymphoma. The effect of the translocation is to place the Bcl-2 gene under the control of a regulatory element that drives overexpression, which allows survival of B lymphocytes that would normally have died.

Of all the cancer-critical genes involved in control of apoptosis, however, there is one that is implicated in cancers in an exceptionally wide range of tissues. This gene, called p53, stands at a crucial intersection in the network of pathways governing a cell's responses to DNA damage and other stressful mishaps. Control of apoptosis is only part of the gene's function—though a very important part. As we shall now explain, when p53 is defective, genetically damaged cells do not merely fail to die; worse still, they wantonly continue to proliferate, accumulating yet more genetic damage that can lead toward cancer.

Mutations in the p53 Gene Allow Cancer Cells to Survive and Proliferate Despite DNA Damage

The p53 gene—named for the molecular mass of its protein product—may be the most important gene in human cancer. This tumor suppressor gene is mutated in about half of all human cancers. What makes p53 so critical? The answer lies in its triple involvement in cell-cycle control, in apoptosis, and in maintenance of genetic stability—all aspects of the fundamental role of the p53 protein in protecting the organism against cellular damage and disorder.

In contrast with Rb, very little p53 protein is found in most of the cells of the body under normal conditions. In fact, p53 is not required for normal development: transgenic mice in which both copies of the gene have been knocked out appear normal in all respects except one—they usually develop cancer by the age of 3 months. These observations suggest that p53 may have a function that is required only occasionally or in special circumstances. Indeed, when normal cells are deprived of oxygen or exposed to treatments that damage DNA, such as ultraviolet light or gamma rays, they raise their concentration of p53 protein by reducing the normally rapid rate of degradation of the molecule. The p53 response is seen also in cells where oncogenes such as Ras and Myc are active, generating an abnormal stimulus for cell division.

In all these cases, the high level of p53 protein acts to limit the harm done. Depending on circumstances and the severity of the damage, the p53 may either drive the damaged or mutant cell to commit suicide by apoptosis (see p. 1013)—a relatively harmless event for the multicellular organism—or it may trigger a mechanism that bars the cell from dividing so long as the damage remains unrepaired. The protection provided by p53 is an important part of the reason why mutations that activate oncogenes such as Ras and Myc are not enough by themselves to create a tumor.

As discussed in Chapter 17, the p53 protein exerts its cell-cycle effects, in part at least, by binding to DNA and inducing the transcription of p21—a regulatory gene whose protein product binds to Cdk complexes required for entry into and progress through S-phase. By blocking the kinase activity of these Cdk complexes, the p21 protein prevents the cell from entering S phase and replicating its DNA.

Cells defective in p53 fail to show these responses. They tend to escape apoptosis, and if their DNA is damaged—by radiation or by some other mishap—they carry on dividing, plunging into DNA replication without pausing to repair the breaks and other DNA lesions that the damage has caused. As a result, they may either die or, far worse, survive and proliferate with a corrupted genome. A common consequence is that chromosomes become fragmented and incorrectly rejoined, creating, through further rounds of cell division, an increasingly disrupted genome as explained in Figure 23-33. Such chromosomal mayhem can lead to both loss of tumor suppressor genes and activation of oncogenes, for example by gene amplification. In addition to being an important mechanism for activating oncogenes, gene amplification can also enable cells to develop resistance to therapeutic drugs, as we see below.

Figure 23-33

How the replication of damaged DNA can lead to chromosome abnormalities, gene amplification and gene loss. The diagram shows one of several possible mechanisms. The process begins with accidental DNA damage in a cell that lacks functional p53 protein. (more...)

In summary, p53 helps a multicellular organism to cope safely with DNA damage and other stressful cellular events, acting as a check on cell proliferation in circumstances where it would be dangerous. Many cancer cells contain large quantities of mutant p53 protein (of an ineffectual variety), suggesting that the genetic accidents they undergo, or the stresses of growth in an inappropriate environment, have created the signals that normally call the p53 protein into play. The loss of p53 activity can thus be trebly dangerous in relation to cancer. First, it allows faulty mutant cells to continue through the cell cycle. Second, it allows them to escape apoptosis. Third, it leads to the genetic instability characteristic of cancer cells, allowing further cancer-promoting mutations to accumulate as they divide. Many other mutations can contribute to each of these types of misbehavior, but p53 mutations contribute to them all.

DNA Tumor Viruses Activate the Cell's Replication Machinery by Blocking the Action of Key Tumor Suppressor Genes

DNA tumor viruses cause cancer mainly by interfering with cell-cycle controls, including those that depend on p53. To understand this type of viral carcinogenesis, it is important to understand the life history of the virus.Viruses use the DNA replication machinery of the host cell to replicate their own genomes. To make many infectious virus particles from a single host cell, a DNA virus has to commandeer this machinery and drive it hard, breaking through the normal constraints on DNA replication and usually killing the host cell in the process. Typically, however, the virus also has another option: it can propagate its genome as a quiet, well-behaved passenger in the host cell, replicating in parallel with the host cell's DNA in the course of ordinary cell division cycles. The virus can switch between these two modes of existence, remaining latent and harmless or proliferating to generate infectious particles according to circumstances. No matter which way of life the virus is following, it is not in its interests to cause cancer. But genetic accidents can occur, such that the virus misuses its equipment for commandeering the DNA replication machinery, and instead of switching on rapid replication of its own genome, switches on persistent proliferation of the host cell.

DNA viruses are a diverse group, but something of this sort seems to happen with most of those that are involved in cancer. The papillomaviruses, for example, are the cause of human warts and are especially important as a key causative factor in carcinomas of the uterine cervix (about 6% of all human cancers). Papillomaviruses infect the epithelium, and are retained in the basal layer of cells as extrachromosomal plasmids that replicate in step with the chromosomes. Infectious virus particles are generated in the outer epithelial layers, as cells begin to differentiate before being sloughed from the surface. Here, cell division should normally be arrested, but the virus interferes with this arrest so as to allow rapid replication of its own genome. Usually, the effect is restricted to the outer layers of cells and relatively harmless, as in a wart. Occasionally, through a genetic accident causing misregulation of the viral genes whose products prevent cell-cycle arrest, the control of cell division is subverted in the basal layer also, in the stem cells of the epithelium. This can lead to cancer, with the viral genes acting as oncogenes (Figure 23-34).

Figure 23-34

How certain papillomaviruses are thought to give rise to cancer of the uterine cervix. Papillomaviruses have double-stranded circular DNA chromosomes of about 8000 nucleotide pairs. In a wart or other benign infection these chromosomes are stably maintained (more...)

In papillomaviruses, the viral genes that are mainly to blame are called E6 and E7. The products of these viral oncogenes interact with many host cell proteins, but in particular they bind to the protein products of two key tumor suppressor genes of the host cell, putting them out of action and so permitting the cell to replicate its DNA and divide in an uncontrolled way. One of these host proteins is Rb: by binding to Rb, the viral E7 protein prevents it from binding to its normal associates in the cell. The other host protein inactivated by the virus is the tumor suppressor p53, which is bound by the viral E6 protein, triggering p53 destruction (Figure 23-35). Elimination of p53 allows the abnormal cell to survive, divide, and accumulate yet more abnormalities.

Figure 23-35

Activation of cell proliferation by a DNA tumor virus. Papillomavirus uses two viral proteins, E6 and E7, to sequester the host cell's p53 and Rb respectively. The SV40 virus (a related virus that infects monkeys) uses a single dual-purpose protein called (more...)

Telomere Shortening May Pave the Way to Cancer in Humans

The mouse is the most widely used model organism for the study of cancer, yet the spectrum of cancers seen in mice differs dramatically from that seen in humans. The great majority of mouse cancers are sarcomas and leukemias, whereas more than 80 percent of human cancers are carcinomas—cancers of epithelia where rapid cell turnover occurs (see Figure 23-2). Many therapies have been found to cure cancers in mice; but when the same treatments are tried in humans, they usually fail. What could be the reason for the difference between mouse and human cancer, and what can it tell us about the molecular mechanisms of the disease? An important part of the answer may lie in the behavior of telomeres and the relationship between telomere shortening, replicative cell senescence, and genetic instability.

As we saw earlier, most human cells seem to have a built-in limit to their proliferation: they show replicative senescence, at least when grown in culture. Replicative cell senescence in humans is thought to be caused by changes in the structure of telomeres—the repetitive DNA sequences and associated proteins that cap the ends of each chromosome. These telomeric DNA sequences are synthesized and maintained by a special mechanism that requires the enzyme telomerase, as explained in Chapter 5. In most human cells, other than those of the germ line and some stem cells, expression of the gene coding for the catalytic subunit of telomerase is switched off, or at least not fully activated. As a result, the telomeres in these cells tend to become a little shorter with each round of cell division. Eventually, the telomeric cap on the chromosome end can become shortened to the point where a danger signal is generated, arresting the cell cycle. The signal is similar, in function at least, to the one that arrests the cycle when an uncapped DNA end is created by an accidental double-strand chromosome break. The effect in both cases is to prevent cell division so long as the cell contains broken or inadequately capped DNA. In the cell with the chromosome break, this allows time for DNA repair; in the normal senescent cell, it seems that it simply puts a stop to cell proliferation. As we discussed at the beginning of the chapter, it is not clear how often cells in normal human tissues run up against this limit; but if a self-renewing cell population does undergo replicative senescence, any rogue cell that undergoes a mutation that lets it carry on dividing will enjoy a huge competitive advantage—much more than if the same mutation had occurred in a cell in a nonsenescent population. Viewed in this light, replicative senescence might be expected to favor the development of cancer.

Mice have telomeres much longer than those of humans. Moreover, unlike humans, they keep telomerase active in their somatic cells, and mouse telomeres therefore do not tend to shorten with increasing age of the organism. It is possible, however, to use gene knockout technology to make mice that lack functional telomerase. In these mice, the telomeres become shorter with every generation, but no untoward consequences are seen until, in the great-great-grandchildren of the initial mutants, the telomeres become so short that they disappear or cease to function. Beyond this point, the mice begin to show various abnormalities, including an increased incidence of cancer. This raises the possibility that natural telomere shortening helps to engender many human tumors.

In a Population of Telomere-Deficient Cells, Loss of p53 Opens an Easy Gateway to Cancer

In contrast with most normal cells in humans, most human cancer cells express telomerase. This is thought to be the reason why, unlike normal cells, they tend to divide without limit in culture, and it is added evidence that telomere maintenance has a significant role in cancer.

Given that cancer cells contain telomerase, an obvious suggestion is that they arise from mutant precursors that have simply avoided shortening their telomeres, and so have never encountered the telomeric limit to cell division. There is, however, another possibility, highlighted by the observations in the telomerase-deficient mice. A cancer may derive from a cell that has experienced telomere shortening, but has suffered a mutation that lets it disregard the signals that normally arrest cell division when telomeres are too short. A mutation causing loss of p53 activity can have just this effect: occurring in one cell within a population arrested by telomere shortening, it can give that cell and its progeny an immediate competitive advantage over their nondividing neighbors. At the same time, as explained in Figure 23-36, the absence of p53 can bring on the gross chromosomal instability that is characteristic of human carcinomas, allowing the mutant cells to accumulate more mutations and evolve rapidly toward cancer. This is what appears to happen in the telomerase-deficient mice. If they lack not only telomerase but also one of their two copies of the p53 gene, carcinomas become even more frequent. It is striking that these additional cancers are predominantly carcinomas—cancers of self-renewing epithelia—rather than the sarcomas and lymphomas usually seen in mice. Carcinomas constitute by far the commonest class of human tumors. The differences in telomere behavior could thereby account for the major difference observed between normal mice and humans in the predominant types of cancers that arise. Like human carcinoma cells, the mouse tumor cells are usually found to have inactivated their last remaining p53 gene; they also show gross chromosomal abnormalities, with many breaks, fusions, and broken chromosome ends.

Figure 23-36

A view of how shortened telomeres may lead to chromosomal instability and cancer. Most human cells lack telomerase. As such cells divide, the telomeres that cap the ends of their chromosomes shrink. After many divisions, the telomeres become so short (more...)

If this scenario applies to cancers in humans, the suggestion would be that telomerase becomes reactivated after, not before, the genetic catastrophe. The progressive genetic disruption following loss of p53 may be so severe that cells can rarely survive it for more than a few generations. A cell that reactivates telomerase expression will be able to halt the catastrophic cycle and regain enough chromosomal stability to survive (see Figure 23-36). Its progeny will inherit a highly abnormal chromosome set, with many mutations and alterations; these damaged cells can then continue to accumulate further mutations at a more moderate rate, driving tumor progression. This model tallies with the findings that, in breast and colorectal cancers, gross chromosomal abnormalities appear to arise early during tumor development, before telomerase is reactivated. With telomerase turned on, these carcinomas still possess enough genetic instability—due to a loss of p53 or other mutations—to continue to evolve, thereby tending to become metastatic.

It is possible that many of the most common types of human carcinomas originate in the way we have just described. But there are certainly many other ways in which cancers can arise, and the true importance of replicative senescence and telomere behavior in human cancer remains to be determined. The uncertainties highlight how little we still know about the natural history of cancer, despite the dramatic advances in understanding of cancer molecular genetics. At the very least, however, the results suggest that telomere-deficient mice can provide a reasonable model for the study of common human carcinomas. Perhaps, this mouse model will also help us to devise cancer treatments that work as well in humans as they do in mice.

The Mutations That Lead to Metastasis Are Still a Mystery

Perhaps the most serious gap in our understanding of cancer concerns malignancy and metastasis. We have yet to clearly identify mutations that specifically permit cells to invade surrounding tissues, spread through the body, and form metastases. Indeed, it is not even clear exactly what properties a cancer cell must acquire to become metastatic. One extreme view would be that the ability of cancer cells in the body to metastasize requires no further genetic changes beyond those needed for loss of cell division control. An opposite view, more commonly espoused, is that metastasis is a difficult and complex task for a cell, requiring a mass of new mutations—so many, and so varied according to circumstance, that it is hard to discover what they are individually.

Which steps in metastasis are the most difficult is still a matter of debate. But there are experimental findings that throw some light on the issue. It is obvious, first of all, that metastasis presents different problems for different types of cells. For a leukemia cell, already roaming the body via the circulation, metastasis should be easier than for a carcinoma cell that has to escape from an epithelium. As we saw at the beginning of the chapter, it is helpful to distinguish two degrees of malignancy for carcinomas, representing two phases of tumor progression. In the first phase, the tumor cells escape the normal confines of their parent epithelium and begin to invade the tissue immediately beneath—becoming locally invasive. In the second phase, they travel to distant sites and settle to form colonies, a process known as metastasis.

Local invasiveness requires a breakdown of the mechanisms that normally hold epithelial cells together. In some carcinomas of the stomach and of the breast, the E-cadherin gene has been identified as a tumor suppressor gene. The primary function of the E-cadherin protein is in cell-cell adhesion, where this protein is embedded in two adjacent plasma membranes to bind epithelial cells together (see Figure 19-24). When tumor cells lacking this adhesion molecule are placed in culture, and a functional E-cadherin gene is put back into them, they lose some of their invasive characteristics and begin to cohere more like normal cells. Loss of E-cadherin, therefore, may favor cancer by specifically contributing to local invasiveness.

The second stage of malignancy, involving entry into the bloodstream or lymphatic vessels, travel via the circulation, and colonization of remote sites, is more enigmatic. The cells must, for example, cross barriers such as the basal lamina of the parent epithelium and of blood vessels. Do they need to acquire additional mutations to become able to do this? The answer is not clear. To discover what steps in metastasis present cells with the greatest difficulties—and thus might be aided by the acquisition of additional mutations—one can label cancer cells with a fluorescent dye, inject them into the circulation of a living animal, and monitor their fate. In these experiments, many cells are found to survive in the circulation, to become lodged in small vessels, and to get out of these into the surrounding tissue, regardless of whether they come from a metastatic or a nonmetastatic tumor. Most of the losses occur after this. Some cells die immediately; others survive entry into the foreign tissue but fail to grow; still others divide a few times and then stop. Here the metastasis-competent cells outperform their nonmetastatic relatives, suggesting that the ability to grow in the foreign tissue is a key property that cells must acquire to become metastatic (Figure 23-37).

Figure 23-37

The barriers to metastasis. Studies of labeled tumor cells leaving a tumor site, entering the circulation, and establishing metastases show which steps in the metastatic process, outlined in Figure 23-15, are difficult or ‘inefficient,’ (more...)

To discover the changes that confer metastatic potential, one can use DNA microarrays to look for genes that are selectively switched on in cancer cells that have become highly malignant. These microarrays allow one to monitor the expression of thousands of genes at a time (see Figure 8-64). One such study took human and mouse melanoma cells that had been selected for high metastatic potential and compared them with their poorly metastatic counterparts. Of the dozen or so genes that appeared to be selectively active in the malignant cells, one that showed increased expression repeatedly in the metastatic cells was RhoC—a member of a family of genes known to regulate cell motility (discussed in Chapter 16). A greatly expanded use of such methods, made much more powerful with the availability of the human genome sequence, should soon give us a clearer picture of the molecular changes that allow tumor cells to metastasize.

Colorectal Cancers Evolve Slowly Via a Succession of Visible Changes

At the beginning of this chapter, we saw that most cancers develop gradually from a single aberrant cell, progressing from benign to malignant tumors by the accumulation of a number of independent genetic accidents. We have discussed what some of these accidents are in molecular terms and seen how they contribute to cancerous behavior. We now examine one particular class of common human cancers more closely, using it to illustrate and enlarge upon some of the general principles and molecular mechanisms we have introduced, and to see how we can make sense of the natural history of the disease in terms of them. We take colorectal cancer as our example, where the steps of tumor progression have been followed in vivo and carefully studied at the molecular level.

Colorectal cancers arise from the epithelium lining the colon and rectum (the lower end of the gut). They are common, currently causing over 60,000 deaths a year in the United States, or about 11% of total deaths from cancer. Like most cancers, they are not usually diagnosed until late in life (90% after the age of 55). However, routine examination of normal adults with a colonoscope (a fiber-optic device for viewing the interior of the colon and rectum) often reveals a small benign tumor, or adenoma, of the gut epithelium in the form of a protruding mass of tissue called a polyp (Figure 23-38A). These adenomatous polyps are believed to be the precursors of a large proportion of colorectal cancers. Because progression of the disease is usually very slow, there is typically a period of 10–35 years in which the slowly growing tumor is detectable but has not yet turned malignant. Thus, when people are screened by colonoscopy in their fifties and the polyps are removed—a quick and easy surgical procedure—the subsequent incidence of colorectal cancer is very low—according to some studies, less than a quarter of what it would be otherwise.

Figure 23-38

Cross-sections showing the stages in development of a typical colon cancer. (A) An adenomatous polyp from the colon. The polyp protrudes into the lumen—the space inside the colon. The rest of the wall of the colon is covered with normal colonic (more...)

Colon cancer provides a clear example of the phenomenon of tumor progression discussed previously. In polyps smaller than 1 cm in diameter, the cells and the local details of their arrangement in the epithelium usually appear almost normal. The larger the polyp, the more likely it is to contain cells that look abnormally undifferentiated and form abnormally organized structures. Sometimes, two or more sectors can be distinguished within a single polyp, the cells in one sector appearing relatively normal, those in the other appearing frankly cancerous, as though they have arisen as a mutant subclone within the original clone of adenomatous cells. At later stages in the disease, the tumor cells become invasive, first breaking through the epithelial basal lamina, then spreading through the layer of muscle that surrounds the gut (Figure 23-38B), and finally metastasizing to lymph nodes, liver, lung, and other tissues.

A Few Key Genetic Lesions Are Common to a Majority of Cases of Colorectal Cancer

What are the mutations that accumulate with time to produce this chain of events? Of those genes so far discovered to be involved in colorectal cancer, three—K-Ras (a member of the Ras gene family), p53, and a third gene, APC, to be discussed below (Table 23-3)—stand out as very frequently mutated. Others are involved in smaller numbers of colon cancers. Still other critical genes remain to be identified.

Table 23-3

Some Genetic Abnormalities Detected in Colorectal Cancer Cells.

One approach to discovery of the mutations responsible for colorectal cancer is to screen the cells for abnormalities in genes already known or suspected to be involved in cancers elsewhere. This type of genetic screening has revealed that about 40% of colorectal cancers have a specific point mutation in K-Ras, activating it as an oncogene, and about 60% have inactivating mutations or deletions of p53.

Another approach to finding cancer-critical genes is to track down the genetic defects in those rare families that show a hereditary predisposition to colorectal cancer. The first of these hereditary colorectal cancer syndromes to be elucidated was a condition known as familial adenomatous polyposis coli (FAP), in which hundreds or thousands of polyps develop along the length of the colon (Figure 23-39). These make their appearance in early adult life, and if they are not removed, it is almost inevitable that one or more of them will progress to become malignant; on average, 12 years elapse from the first detection of polyps to the diagnosis of cancer. The disease can be traced to deletion or inactivation of APC, a gene on the long arm of chromosome 5. Individuals with FAP have inactivating mutations or deletions of the APC gene in all the cells of the body. Of the patients with colorectal cancer who do not have the hereditary condition—the vast majority of cases of the disease—more than 60% have similar mutations in the cells of the cancer but not in their other tissues; in other words, both copies of the APC gene have been lost during their lifetime. Thus, by a route similar to that which we have discussed for retinoblastoma, mutation of APC is identified as one of the central ingredients of colorectal cancer.

Figure 23-39

Colon of familial adenomatous polyposis coli patient compared to normal colon. (A) The polyposis colon is completely covered by hundreds of projecting polyps (shown in section in Figure 23-38A), each resembling a tiny cauliflower when viewed with the (more...)

As explained earlier, tumor suppressor genes can also be tracked down, even where there is no known hereditary syndrome, by searching for genetic deletions in the tumor cells. A systematic scan of a large number of colorectal cancers reveals frequent losses of large sections of certain chromosomes, suggesting that those regions may harbor tumor suppressor genes. One of them is the region including APC. Another includes the SMAD4 gene, which is mutated in perhaps 30% of colon cancers. At least one other important tumor suppressor gene is thought to lie in the same neighborhood as SMAD4 on chromosome 18 but remains to be identified. Specific parts of other chromosomes also show frequent losses—or gains—in colorectal cancers and are now being searched for additional cancer-critical genes.

As our knowledge of genes and their functions has expanded, another fruitful approach has been to look for genes that interact with a known cancer-critical gene in the hope that these, too, may be targets for mutation. The APC protein is now known to be an inhibitory component of the Wnt signaling pathway (discussed in Chapter 15). It acts by binding to β-catenin, another component of the pathway, and thereby preventing activation of TCF4, a gene regulatory protein that stimulates growth of the colonic epithelium when it has β-catenin bound to it. As we saw in Chapter 22, loss of TCF4 causes a depletion of the gut stem-cell population, so that loss of the antagonist APC may cause overgrowth by the opposite effect. When the β-catenin gene was sequenced in a collection of colorectal tumors, it turned out that among the few that did not have APC mutations, a high proportion had activating mutations in β-catenin instead. Thus, it is the Wnt signaling pathway, rather than any single oncogene or tumor suppressor gene that it contains, that is critical for the cancer.

Defects in DNA Mismatch Repair Provide an Alternative Route to Colorectal Cancer

In addition to the hereditary disease associated with APC mutations, there is a second, and actually commoner, kind of hereditary predisposition to colon carcinoma in which the course of events is quite different from the one we have described for FAP. In patients with this condition, called hereditary nonpolyposis colorectal cancer, or HNPCC, the probability of colon cancer is increased without any increase in the number of colorectal polyps (adenomas). Moreover, the cancer cells in the tumors that develop are unusual, inasmuch as examination of their chromosomes in a microscope reveals a normal (or almost normal) karyotype and a normal (or almost normal) number of chromosomes. In contrast, the vast majority of colorectal tumors in non-HNPCC patients have gross chromosomal abnormalities, with multiple translocations, deletions, and other aberrations, and a total of 55 to 70 or more chromosomes instead of the normal 46 (Figure 23-40).

Figure 23-40

Chromosome complements (karyotypes) of colon cancers showing different kinds of genetic instability. (A) The karyotype of a typical cancer shows many gross abnormalities in chromosome number and structure. Considerable variation can also exist from cell (more...)

The mutations that predispose an individual with HNPCC to colorectal cancer turn out to be in one of several genes that code for central components of the DNA mismatch repair system in humans, homologous in structure and function to the mutL and mutS genes in bacteria and yeast (see Figure 5-23). Only one of the two copies of the involved gene is defective, so the inevitable DNA replication errors are efficiently removed in most of the patient's cells. However, as discussed previously for other tumor suppressor genes, these individuals are at risk, because the accidental loss or alteration of the remaining good gene copy will immediately elevate the spontaneous mutation rate by a hundred-fold or more (discussed in Chapter 5). These genetically unstable cells presumably go speeding through the standard processes of mutation and natural selection that allow clones of cells to progress to a malignancy.

This type of genetic instability produces invisible changes in the chromosomes—most notably changes in individual nucleotides and short expansions and contractions of mono- and dinucleotide repeats such as AAAA… or CACACA…. Once the phenomenon was recognized, mutations in mismatch repair genes were found in about 15% of the colorectal cancers occurring in normal people, with no inherited mutation. Again, the chromosomes in these cancers were unusual in having nearly normal karyotypes.

Thus genetic instability is found in practically all colorectal cancers, but it can be acquired in at least two very different ways. The majority of colorectal cancers have become unstable by rearranging their chromosomes—some perhaps as the combined result of a p53 mutation and telomere shortening, as previously discussed. Others have been able to avoid this type of trauma; their genetic instability occurs on a much smaller scale, being caused by a defect in DNA mismatch repair. The fact that many carcinomas show either chromosomal instability or defective mismatch repair—but rarely both—clearly demonstrates that genetic instability is not an accidental byproduct of malignant behavior, but a contributory cause; it is a property that most cancer cells need in order to become malignant, but one that they can acquire in several alternative ways.

The Steps of Tumor Progression Can Be Correlated with Specific Mutations

In what sequence do K-Ras, p53, APC, and the other identified colorectal cancer genes undergo their mutations, and what contribution does each of them make to the eventual unruly behavior of the cancer cell? There cannot be a simple answer to this question, because colorectal cancer can arise by more than one route. Thus, as we have just seen, in some cases the first mutation leading toward the cancer may be in a DNA mismatch repair gene; in other cases, it may be in a gene regulating growth. A general feature, such as genetic instability, can arise in a variety of ways, through mutations in different genes.

Nevertheless, certain patterns of events are particularly common. Thus, mutations inactivating the APC gene appear to be the first, or at least a very early, step in most cases. They can be detected already in small benign polyps at the same high frequency as in large malignant tumors. Loss of APC seems to increase the rate of cell proliferation in the colonic epithelium relative to the rate of cell loss, without affecting the way the cells differentiate or the details of the histological pattern they form.

Mutations activating the K-Ras oncogene—a member of the Ras family—appear to take place a little later than those in APC; they are rare in small polyps but common in larger ones that show disturbances of cell differentiation and histological pattern. When malignant colorectal carcinoma cells containing such Ras mutations are grown in culture, they show typical features of transformed cells, such as the ability to proliferate without anchorage to a substratum. Loss of cancer-critical genes on chromosome 18 and mutations in p53 may come later still. They are rare in polyps but common in carcinomas, suggesting that they may often occur late in the sequence (Figure 23-41). As discussed earlier, loss of p53 function is thought to allow the abnormal cells not only to avoid apoptosis and to divide, but also to accumulate additional mutations at a rapid rate by progressing through the cell cycle when they are not fit to do so, creating many abnormal chromosomes.

Figure 23-41

Suggested typical sequence of genetic changes underlying the development of a colorectal carcinoma. This over-simplified diagram provides a general idea of the way mutation and tumor development can fit together. But there are certainly other mutant genes (more...)

Each Case of Cancer Is Characterized by Its Own Array of Genetic Lesions

As we have just seen in colorectal cancer, the traditional classification of cancers is simplistic: a single one of the conventional categories of tumor will turn out—on close scrutiny—to be a heterogeneous collection of disorders with some features in common, but each characterized by its own array of genetic lesions (Figure 23-42). Many types of cancers that have been analyzed genetically show a large variety of genetic lesions and a great deal of variation from one case of the disease to another. In the form of lung cancer known as small-cell lung cancer, for example, one finds mutations not only in Ras, p53, and APC, but also in Rb, in members of the Myc gene family (in the form of amplification of the number of Myc gene copies), and in at least five other known proto-oncogenes and tumor suppressor genes. Different combinations of mutations are encountered in different patients and correspond to cancers that react differently to treatment.

Figure 23-42

Each tumor will generally contain a different set of genetic lesions. In this schematic diagram, W, X, Y, and Z denote alterations in as yet undiscovered tumor suppressor genes or oncogenes. Tumors that arise from different tissues are generally more (more...)

In principle, molecular biology provides the tools to find out precisely which genes are amplified, which are deleted, and which are mutated in the tumor cells of any given patient. As we see in the next section, such information may soon prove to be as important for the diagnosis and treatment of cancer as is the identification of microorganisms in patients with infectious diseases.

Summary

Studies of developing embryos and transgenic mice have helped to reveal the functions of many cancer-critical genes. Most of the genes found to be mutated in cancer, both oncogenes and tumor suppressor genes, code for components of the pathways that regulate the social and proliferative behavior of cells in the body—in particular, the mechanisms by which signals from a cell's neighbors can impel it to divide, differentiate, or die. Other cancer-critical genes are involved in maintaining the integrity of the genome and guarding against damage. The molecular changes that allow cancers to metastasize, however, escaping the parent tumor and growing in foreign tissues, are still largely unknown.

DNA viruses such as papillomaviruses can promote the development of cancer by sequestering the products of tumor suppressor genes—in particular, the Rb protein, which regulates cell division, and the p53 protein, which is thought to act as an emergency brake on cell division in cells that have suffered genetic damage and to call a halt to cell division in senescent cells with shortened telomeres.

The p53 protein has a dual role, regulating both progression through the cell cycle and the initiation of apoptosis. So loss or inactivation of p53, which occurs in about half of all human cancers, is doubly dangerous: it allows genetically damaged and senescent cells to continue to replicate their DNA, increasing the damage, and it allows them to escape apoptosis. The loss of p53 function may contribute to the genetic instability of many full-blown metastasizing cancers.

Generally speaking, the steps of tumor progression can be correlated with mutations that activate specific oncogenes and inactivate specific tumor suppressor genes. But different combinations of mutations are found in different forms of cancer and even in patients that nominally have the same form of the disease, reflecting the random way in which mutations occur. Nevertheless, many of the same types of genetic lesions are encountered repeatedly, suggesting that there is only a limited number of ways in which our defenses against cancer can be breached.