Genes VII

Whether activated by quantitative or qualitative changes, oncogenes may be presumed to influence (directly or indirectly) functions connected with cell growth. Transformed cells lack restrictions imposed on normal cells, such as dependence on serum or inhibition by cell-cell contact. They may acquire new properties, such as the ability to metastasize. Many phenotypic properties are changed when we compare a normal cell with a tumorigenic counterpart, and it is striking indeed that individual genes can be identified that trigger many of the changes associated with this transformation.

We assume that oncogenes, individually or in concert, set in train a series of phenotypic changes that involve the products of many genes. In this description, we see at once a similarity with genes that regulate developmental pathways: they do not themselves necessarily code for the products that characterize the differentiated cells, but they may direct a cell and its progeny to enter a particular pathway. The same analogy suggests itself for oncogenes and developmental regulators: they provide switches responsible for causing transitions between one discrete phenotypic state and another.

Taking this relationship further, we may ask what activities the products of proto-oncogenes play in the normal cell, and how are they changed in the transformed cell? Could some proto-oncogenes be regulators of normal development whose malfunction results in aberrations of growth that are manifested as tumors? We have stumbled across some examples of such relationships, but do not yet have any systematic understanding of the connection.

Oncoproteins are organized according to their types of functions in Figure 28.14. The left part of the figure groups the oncogenes according to the locations of their products. The boxes on the right give details of the corresponding proto-oncogenes. The functions of many oncogenes remain unknown, and further groups will no doubt be identified:

• Growth factors are proteins secreted by one cell that act on another. The oncoprotein counterparts can only transform cells bearing the appropriate receptor. (This is called autocrine transformation.; for review see Heldin and Westermark, 1984; Cross and Dexter, 1991)
• The growth factor receptors are transmembrane proteins that are activated by binding an extracellular ligand (usually a polypeptide). Most often the receptor is a protein tyrosine kinase. Oncogenicity may result from constitutive (that is, ligand-independent) activation of the kinase activity (Waterfield et al., 1983).
• One group of intracellular protein kinases phosphorylate tyrosine residues in target proteins. c-Src, which associates with the cytoskeleton as well as with the membrane, is the prototype of a family of kinases with similar catalytic activities (including c-Yes, c-Fgr, Lck, c-Fps, Fyn). We understand the effects of oncogenic mutations on the Src kinase activity in some detail, although we have yet to explain why the altered kinase activity is oncogenic. Other protein tyrosine kinases in the intracellular group are cytosolic; c-Abl is found in both cytosol and nucleus (Collet and Erikson, 1978; Hunter and Sefton, 1980; for review see Jove and Hanafusa, 1987).

Signal transduction pathways are often involved in oncogenesis The best characterized example is c-Ras, which plays a central role in transmitting the signal from receptor tyrosine kinases (see 26 Signal transduction). Oncogenic mutations change the regulation of Ras activity. Other stages in signal transduction are identified by Gsp and Gip, which are mutant forms of the α subunits of the G_s and G_i trimeric G proteins. Crk and Vav are proteins associated with later stages of signaling.

• A group of cytosolic enzymes are protein serine/threonine kinases, that is, they phosphorylate target proteins on serine or threonine. Little is known about the effects of oncogenic mutations beyond the fact they probably increase or constitutively activate the kinase activities. Mos is an example which can activate ERK MAPK.
• Nuclear proteins include transcription factors of several types. The functions of these proto-oncoproteins are rather well described (see 21 Regulation of transcription). Generally we understand what effects the oncogenic mutations have on the factors, but we cannot yet relate these changes to the activation or repression of a set of target genes that defines the oncogenic state (Bohmann et al., 1987; Angel et al., 1988; Bos et al., 1988).

The common feature is that each type of protein is in a position to trigger general changes in cell phenotypes, either by initiating or responding to changes associated with cell growth, or by changing gene expression directly. Before we consider in detail the potential of each group for initiating a series of events that has an oncogenic outcome, we need to consider how many independent pathways are identified by these factors.

Recall the example of the best characterized mitogenic pathway, the MAPK pathway which consists of the following stages:

growth factor

↓

growth factor receptor (tyrosine kinase)

↓

Ras

↓

kinase cascade (serine/threonine kinases)

↓

transcription factor(s)

When a growth factor interacts with its receptor, it activates the tyrosine kinase activity. The signal is passed (via an adaptor) to Ras. At this point, the pathway switches to a series of serine/threonine kinases. The targets at the end of the pathway may be controlled directly or indirectly by phosphorylation, and include transcription factors, which are in a position to make widespread changes in the pattern of gene expression.

If a pathway functions in a linear manner, in which the signal passes directly from one component to the next, the same results should be achieved by constitutive activation of any component (so that it no longer needs to be activated by a signal from an earlier component).

A signal transduction pathway, of course, is likely to branch at several stages, so that an initial stimulus may trigger a variety of responses. The activation of components that are downstream will therefore activate a smaller number of end-functions than the activation of components at the start of the pathway. But we can analyze any individual part of the pathway by tracing it back to the beginning as though it were strictly linear.

In the example of the Ras pathway, we know that it is activated by many growth factors to generate a mitogenic response. Mutations in the early part of this pathway, including the ras and raf genes, may be oncogenic. But oncogenic mutations are not found in the following components of the cascade, the MEK and MAP kinases. This suggests that there may be a branch in the pathway at the stage of ras or raf, and that activation of this branch is also necessary for oncogenicity. Ras activates a cytoskeletal GTPase called Rac, which may identify this branch. Of course, the ERK MAPK pathway terminates in the activation of several "immediate early" genes, including fos and jun, which themselves have oncogenic counterparts, suggesting that the targets of the MAPK pathway can be sufficient for oncogenicity.

The central role of this pathway is indicated by the number of its components that are coded by proto-oncogenes. One explanation of the discrepancies between the susceptibilities of MAP kinases and other components to oncogenic mutation may be that the level or duration of expression is important. It could be the case that mutations in MEK or MAP kinases do not activate the enzymes sufficiently to be oncogenic. Alternatively, the oncogenic mutations (which, after all, represent gain-of-function) may cause new targets to be activated in addition to the usual pathway. The general principle is clear: that aberrant activation of mitogenic pathways can contribute to oncogenicity, but we cannot yet relate the activation of these pathways to individual responses in terms of immortalization or transformation (for review see Cantley et al., 1991).