12. Summary

28.11 Immortalization and transformation

Most tumors arise as the result of multiple events. It is likely that some of these events involve the activation of oncogenes, while others take the form of inactivation of tumor suppressors. The requirement for multiple events reflects the fact that normal cells have multiple mechanisms to regulate their growth and differentiation, and several separate changes may be required to bypass these controls. Indeed, the existence of many genes in which single mutations were tumorigenic would no doubt be deleterious to the organism, and has been selected against. Nonetheless, oncogenes and tumor suppressors define genes in which mutations create a predisposition to tumors, that is, they represent one of the necessary events. It is an open question as to whether the oncogenes and tumor suppressor genes identified in available assays are together sufficient to account entirely for the occurrence of cancers, but it is clear that their properties explain at least many of the relevant events.

The need for multiple functions fits with the pattern established by some DNA tumor viruses, in which (at least) two functions are needed to transform the usual target cells. In the same way, expression of two or more oncogenes in the cellular transfection assay is usually needed to convert a primary cell (one taken directly from the organism) into a tumor cell. The need for multiple functions of different types is sometimes described as the requirement for cooperativity. The division of functions may loosely be viewed as being concerned with immortalization or transformation (for review see Hanahan, 1988; Hunter, 1991).

Adenovirus carries the E1A region, which allows primary cells to grow indefinitely in culture, and the E1B region, which causes the morphological changes characteristic of the transformed state.

Polyoma produces three T antigens; large T elicits indefinite growth, middle T is responsible for morphological transformation, and small T is without known function. Large T and middle T together can transform primary cells.

Consistent with the classification of oncogenic functions, adenovirus E1A together with polyoma middle T can transform primary cells. This suggests that one function of each type is needed.

Several cellular oncogenes have been identified by transforming ability in the 3T3 transfection assay; 10 V20% of spontaneous human tumors have DNA with detectable transforming activity in this assay. Of course, 3T3 cells have been adapted to indefinite growth in culture over many years, and have passed through some of the changes characteristic of tumor cells. The exact nature of these changes is not clear, but generally they can be classified as involving functions concerned with immortalization. Oncogenic activity in this assay therefore depends on the ability to induce further changes in an established cell line.

The principal products of 3T3 transfection assays are mutated c-ras genes. They do not have the ability to transform primary cells in vitro, and this supports the implication that their functions are concerned with the act of transforming cells that have previously been immortalized. ras oncogenes clearly provide one major pathway for transforming immortalized cells; we do not know how many other transforming pathways may exist that are independent of ras.

Whatever functions are required for immortalization can be provided (or circumvented) by other oncogenes. Although ras oncogenes alone cannot transform primary fibroblasts, dual transfection with ras and another oncogene can do so. The ability to transform primary cells in conjunction with ras provides a general assay for oncogenes that have an immortalization-like function. This group includes several retroviral oncogenes, v-myc, v-jun, and v-fos. It also includes adenovirus E1A and polyoma large T. Mutant p53 genes have the same effect, suggesting that loss of p53 constitutes one route to immortalization. However, in many cases the distinction between immortalizing and transforming proteins is blurred. For example, although E1A is classified as having an immortalizing function, it has (some) of the functions usually attributed to transforming proteins, and loss of p53 confers some properties that are usually considered transforming.

One way to investigate the oncogenic potential of individual oncogenes independently of the constraints that usually are involved in their expression is to create transgenic animals in which the oncogene is placed under control of a tissue-specific promoter. A general pattern is that increased proliferation often occurs in the tissue in which the oncogene is expressed. Oncogenes whose expression have this effect with a variety of tissues include SV40 T antigen, v-ras, and c-myc (Brinster et al., 1984).

Increased proliferation (hyperplasia) is often damaging and sometimes fatal to the animal (usually because the proportion of one cell type is increased at the expense of another). However, the expression of a single oncogene does not usually cause malignant transformation (neoplasia), with the production of tumors that kill the animal. Tumors resulting from the introduction of an oncogene (for example, in transgenic mice) are probably due to the occurrence of a second event.

The need for two types of event in malignancy is indicated by the difference between transgenic mice that carry either the v-ras or activated c-myc oncogene, and mice that carry both oncogenes. Mice carrying either oncogene develop malignancies at rates of 10% for c-myc and 40% for v-ras; mice carrying both oncogenes develop 100% malignancies over the same period. These results with transgenic mice are even more striking than the comparable results on cooperation between oncogenes in cultured cells (Stewart et al., 1984; Sinn et al., 1987).

An interesting convergence is seen in the properties of the tumor antigens of the DNA tumor viruses: the antigens can bind to the cellular tumor suppressor products RB and p53. The two cellular proteins are recognized independently. Either different T antigens of the virus bind separately to RB and to p53, or different domains of the same antigen do so. So adenovirus E1A binds RB, while E1B binds p53; HPV E7 binds RB, while E6 binds p53. SV40 T antigen can bind both RB and p53. The consequences for function of the tumor suppressor are especially clear in the case of HPV E6, which targets p53 for degradation. In effect, HPV converts a target cell into a p53^- state. It seems likely that the loss of p53 (and/or RB) is a major step in the transforming action of DNA tumor viruses, and explains some significant part of the action of the T antigens. The critical events are inhibition of p53’s ability to activate transcription, and loss of RB’s ability to bind substrates such as E2F. Loss of the tumor suppressors may be one major route in the immortalization pathway.

The changes required for "immortalization" remain to be characterized in terms of changes in cellular properties. It is important to remember that immortalization is required for cells to be perpetuated indefinitely in tissue culture, and we do not know how the relevant events relate to the formation of a tumor in vivo. It is clear that some of the same tumor suppressors and oncogenes that are associated with immortalization in culture play roles in tumorigenesis. However, it is not necessarily the case that the events involved in passing through "crisis" to become immortalized in culture have any exact parallel in the formation of a tumor.

With this caveat in mind, it is interesting that p53 provides an important function in immortalization. Established cell lines have usually lost p53 function, which suggests that the role of p53 is connected with the acquisition of ability to support prolonged growth. However, loss of the known functions of p53 is not enough by itself to explain immortalization, since, for example, a p53 ^V mouse is viable, and therefore is able to undergo the usual pattern of cell cycle arrest and differentiation. Primary cells from a p53 ^V mouse can pass into the established state more readily than cells that have p53 function, which suggests that loss of p53 activity facilitates or is required for immortalization (Donehower et al., 1992).

In this context, we might view p53 as an immortalizing function, since the inability to trigger either growth arrest or apoptosis clearly must lead to continued growth. We do not know whether only one or both of these activities are required for immortalization in vitro. Also, it is important to realize that these activities of p53 have been described largely in terms of the response to irradiation. It is certainly consistent that cells that are unable to arrest or apoptose when damaged by DNA should develop the chromosomal abnormalities that are often associated with tumor cells. However, we should not assume that the failure in the response to irradiation is responsible for all the deficiencies seen in p53 ^V cells. For example, growth of normal cells in the body is controlled by a variety of signals, involving diffusible factors (such as TGFβ ), inhibition by cell-cell contacts, etc. Failure to respond to such signals may contribute to tumorigenesis, but we do not know how p53 is involved. We know that more than the growth arrest pathway is needed for p53’s contribution to tumorigenesis, because a p21 ^V mouse shows deficiencies in the G1 checkpoint (as would be expected) but does not develop tumors. The contrast with the increased susceptibility of a p53 ^V mouse to tumors shows that other functions of p53 are involved besides its control of p21.

In some systems, immortalization may be connected with an inability of the cells to differentiate. Growth and differentiation are often mutually exclusive, because a cell must stop dividing in order to differentiate. An oncoprotein that blocks differentiation may allow a cell to continue proliferating (in a sense resembling the immortalization of cultured cells); continued proliferation in turn may provide an opportunity for other oncogenic mutations to occur. This may explain the occurrence among the oncoproteins of products that usually regulate differentiation.

A connection between differentiation and tumorigenesis is shown by avian erythroblastosis virus (AEV). The AEV-H strain carries only v-erbB, but the AEV-E54 strain carries two oncogenes, v-erbB and v-erbA. The major transforming activity of AEV is associated with v-erbB, a truncated form of the EGF receptor, which is equivalent to the single oncogene carried by other tumor retroviruses: it can transform erythroblasts and fibroblasts. The other gene, v-erbA, cannot transform target cells alone, but it increases the transforming efficiency of v-erbB. Expression of v-erbA itself has two phenotypic effects upon target cells: it prevents the spontaneous differentiation (into erythrocytes) of erythroblasts that have been transformed by v-erbB; and it expands the range of conditions under which transformed erythroblasts can propagate. v-erbA may therefore contribute to tumorigenicity by a combination of inhibiting differentiation and stimulating proliferation. In fact, v-erbA has a similar effect in extending the efficacy of transformation by other oncogenes that induce sarcomas, notably v-src, v-fps, and v-ras.

Correlations between the activation of oncogenes and the successful growth of tumors are strong in some cases, but by and large the nature of the initiating event remains open. It seems clear that oncogene activity assists tumor growth, but activation could occur (and be selected for) after the initiation event and during early growth of the tumor. We hope that the functions of c-onc genes will provide insights into the regulation of cell growth in normal as well as aberrant cells, so that it will become possible to define the events needed to initiate and establish tumors.

An important influence on the ability of a cell to grow is provided by telomerase, the ribonucleoprotein enzyme that is responsible for extending telomeres. The function of telomerase is to compensate for the shortening of telomeres that occurs at each replication cycle. Telomerase is turned off in many somatic cells, typically when differentiation occurs. However, its activity usually is reactivated in tumors. The limiting step in production of telomerase is the transcription of the catalytic subunit, which is repressed in differentiated somatic cells and restored in tumor cells. Similarly, telomerase activity is reactivated in cultured cells that have emerged from crisis. The critical question is whether telomerase activity is essential for tumor formation and at what stage it might be necessary.

Continued division in cells that lack telomerase activity (for example, when primary cells are placed into culture) will cause the telomeres to shorten in each generation. One possible cause of crisis in cultured cells is that telomere lengths become too short to ensure stability at the ends of the chromosomes. This could explain the frequent occurrence of chromosomal abnormalities in cultured cells. When cells enter senescence as the result of telomere shortening, p53 is activated, leading to growth arrest or apoptosis. The trigger that activates p53 is the loss of the telomere-binding protein TRF2 from the chromosome ends. The loss of p53 allows the cells to survive and divide, although of course chromosomal abnormalities result from the lack of proper telomeres.

This suggests that a critical parameter for immortalization might be the reactivation of telomerase. Telomerase activity can be restored by transfecting the gene for the catalytic subunit into target cells, and this allows them to be perpetuated in culture without passing through crisis. We might view the finite replicative capacity of primary human cells in general, or their inability to continue propagation once telomere lengths have become too short in particular, as a tumor suppression mechanism that in effect prevents cells from undertaking the indefinite replication that is needed to make a tumor (Meyerson et al., 1997).

Mice in which the gene coding for the telomerase RNA has been inactivated can survive for 6 generations. Mouse telomeres are exceptionally long, in the range from 10 V60 kb. In the absence of telomerase, telomeres shorten at 50 V100 bp per cell division. There are ~60 divisions in sperm cell production, and ~25 divisions in oocyte production, which fits with the observed rate of shortening of ~4.8 kb per male mouse generation. By the 6th generation, chromosomal abnormalities become more frequent, the ability to form tumors declines, and the mice become infertile (due to the inability to produce sperm). The effects of lack of telomerase are first seen in tissues consisting of highly proliferative cells (as might be expected). All of these observations demonstrate the importance of telomerase for continued cell division. However, cells from the telomerase-negative mice can pass through crisis and can be transformed to give tumorigenic cells, so the presence of telomerase is not essential, or at least is not the only means, of supporting an immortal state (although reactivation of telomerase is by far the most common mechanism) (Blasco et al., 1997).

There is a curious inconsistency between the results obtained with cultured cells and the survival of telomerase-negative mice. Crisis of mouse cells occurs typically after 10 V20 divisions in culture, but one would not expect the telomeres to have reached a limiting length at this point. Mice of the first telomerase-negative generation have passed a greater number of cell divisions without telomerase, and without suffering any ill effects.

Lack of telomerase is clearly associated with inability to continue growth, and reactivation of telomerase is one means by which cells can behave as immortal. It is not clear whether telomerase is the only relevant factor in driving cells into crisis and to what extent other mechanisms might be able to compensate for lack of telomerase. We do not know what pathway is responsible for controlling telomerase production in vivo, and how it is connected to pathways that control cell growth.

Reviews

Hanahan, D. (1988). Dissecting multistep tumorigenesis in transgenic mice. Ann. Rev. Genet. 22, 479-519.

Hunter, T. (1991). Cooperation between oncogenes. Cell 64, 249-270.

Research

Blasco, M. A. et al. (1997). Telomere shortening and tumor formation by mouse cells lacking telomerase RNA. Cell 91, 25-34.

Brinster, R. L. et al. (1984). Transgenic mice harboring SV40 T-antigen genes develop characteristic brain tumors. Cell 37, 367-379.

Donehower, L. A. et al. (1992). Mice deficient for p53 are developmentally normal but susceptible for spontaneous tumors. Nature 356, 215-221.

Meyerson, M. et al. (1997). hEST2, the putative human telomerase catalytic subunit gene, is up-regulated in tumor cells and during immortalization. Cell 90, 785-795.

Sinn, E., Muller, W., Pattengale, P., Tepler, I., Wallace, R., and Leder, P. (1987). Coexpression of MMTV/v-Ha-ras and MMTV/c-myc genes in transgenic mice: synergistic action of oncogenes in vivo. Cell 49, 465-4.

Stewart, T. A., Pattengale, P. K., and Leder, P. (1984). Spontaneous mammary adenocarcinomas in transgenic mice that carry and express MTV/myc fusion genes. Cell 38, 627-637.