Tumour suppressors

Teaming up to restrain cancer

Article metrics

Foremost among the protective mechanisms that have evolved to reduce the likelihood of cancer are the tumour-suppressor genes. As long as they are fully active, these genes serve as potent buffers against tumour progression. Only when one (or more) of them becomes defunct is the gate opened for uncontrolled cell multiplication and cancer.

Perhaps the best known of all tumour-suppressor genes is p53 (ref. 1), which is altered in about half of all human tumours, making it the most frequent target for genetic alterations in cancer. Presumably, loss of the normal p53 protein facilitates the emergence of malignant traits. But, as Garkavtsev et al.2 report on page 295 of this issue, p53 does not restrain cancer single-handedly. Rather, it seems to require an intimate partner — a protein encoded by another tumour-suppressor gene, ING1.

The ING1 gene was identified through a cloning strategy aimed at genes whose expression is selectively reduced in cancer cells3. Overexpression of ING1 was found to block cell proliferation3 and to increase programmed cell death in the absence of survival factors4. When expression of ING1 is quenched using antisense RNA, cellular replicative senescence is delayed, whereas unrestrained proliferation and illegitimate cell survival are promoted. All of these properties are consistent with ING1 behaving like a tumour suppressor, underscoring a striking functional similarity between ING1 and p53.

Prompted by these parallels, Garkavtsev et al. took a closer look at the relationship between the two proteins, p33ING1 and p53. They now report that this relationship extends a long way — in fact, each of the two proteins seems to require the other in order to exert its inhibitory effects. More specifically, p33ING1 cannot block the proliferation of cells lacking p53, nor can it render p53-deficient cells sensitive to chemotherapy. Conversely, if production of p33ING1 is suppressed using antisense RNA, cells can escape p53-mediated growth inhibition. So p53 and p33ING1 seem to operate as a close team, presumably specializing in tumour suppression.

How is the team effort coordinated? The data indicate that p33ING1 and p53 physically interact to form a specific protein-protein complex (see Fig. 1, overleaf). p53 is mainly a sequence-specific transcription factor and, due to its interaction with p33ING1, it becomes a more effective transcriptional activator. For example, p33ING1 can increase p53-dependent activation of the p21/WAF1 promoter. The p21WAF1 protein potently blocks the cell-cycle machinery and mediates p53-dependent growth arrest1. Elevated production of p21WAF1 could, therefore, largely account for the ability of p33ING1 to block cell proliferation in concert with activated p53.

Figure 1: p33ING1 and p53 as mediators of cell fate.

p53 can arrest cell proliferation by activating a set of target genes, among which p21/WAF1 is pivotal. p53 can also induce cell death through apoptosis. This outcome relies, at least in part, on the transcriptional activation (Trcn) of death-promoting genes such as Bax and Fas/ApoI, and of genes involved in the production of reactive oxygen species (ROS). To exert its cellular effects p53 must first be biochemically activated, typically entailing an increase in cellular levels of p53. As Garkavtsev et al.2 now report, many of p53's activities are carried out in cooperation with p33ING1, presumably through a physical interaction between the two proteins. But it is not known which signals modulate the activity of p33ING1, and whether p53-mediated apoptosis also requires p33ING1.

So far, the cooperation between p53 and p33ING1 has been shown for only one gene (p21/WAF1) and, primarily, for one biological outcome — growth arrest. The pathway that leads to p53-mediated apoptosis is distinct from that leading to growth arrest, and probably involves the activation of a different subset of target genes. Some of these genes may promote apoptosis through the generation of reactive oxygen species5, and it remains to be seen whether p33ING1 cooperates with p53 here too. But the answer will probably be positive: p33ING1 also exerts pro-apoptotic effects4, and it cooperates with p53 to confer sensitivity to chemotherapy2, a treatment that works mainly by inducing apoptosis of cancer cells.

In the experimental systems used by Garkavtsev et al.2, the presence of p33ING1 seems to be obligatory for the inhibitory effects of p53. However, these systems are rather artificial and prone to exaggeration. So, within the human body, at least some of the physiological tumour-suppressor activities of p53 may occur independently of p33ING1. Nevertheless, if the loss of p33ING1 compromises the function of p53 only slightly, this may provide emerging cancer cells with a selective growth advantage. And even a small advantage might make the difference between cancer and health.

As is often the case with crucial bioregulatory processes, the story is probably far more complicated. For example, at least some anti-proliferative activities of p53 rely on another candidate tumour-suppressor gene, IRF1 (ref. 6). Cooperative activation of the p21/WAF1 gene is implicated in this case, too. However, the mechanism may be distinct from that described for p33ING1, and there may not be direct protein-protein association between p53 and IRF1. Furthermore, p53 can physically associate with yet another tumour-suppressor gene product — the WT1 protein7. Perhaps all associates belong to the same team or, maybe, p53 chooses different partners under different circumstances.

What are the practical implications of the new findings? For one, the concentration of p33ING1 might determine the extent to which a given tissue can mount an effective p53 response on exposure to stress. This could explain the puzzling observation that, unlike the promiscuous activation of p53 in cultured cells, only a limited number of tissues within a normal organism show strong p53 activation in response to DNA damage. An even smaller number show the expected biological effects8.

The findings of Garkavtsev et al.2 may also relate to the important question of whether p53 is fully functional in the 50 per cent of human tumours that retain a wild-type p53 gene. Certain types of tumour inactivate their p53 through increased expression of the p53-binding protein Mdm2 (ref. 9). So loss of p33ING1 function is another potential mechanism for inactivation of p53 in cancer cells. Moreover, in a limited analysis of three neuroblastoma cell lines, one line harboured a rearranged ING1 gene resulting in a truncated protein3. Many more lines must be analysed before firm conclusions can be drawn, but it is remarkable that neuroblastoma cells often contain high levels of wild-type p53 (ref. 10) — a counter-intuitive behaviour given the highly malignant nature of these cells.

Despite being highly abundant, the p53 in neuroblastoma cells is functionally defective11 (although this is somewhat controversial). The possibility that these cells carry a defect in p33ING1 might explain their unusual immunity to p53. Furthermore, the excess p53 accumulates in the cytoplasm of neuroblastoma cells, rather than in the nucleus, raising the possibility that p33ING1 might enable efficient nuclear translocation of p53.

A growing effort is now being invested in developing p53-based cancer gene therapies12. If p33ING1 is essential for the effective anti-tumour action of p53, we may need to determine the status of p33ING1 in every tumour before subjecting the patient to such experimental therapy. There is already reason to suspect that ING1 may be deleted in some head and neck tumours, which are prime targets for clinical trials in the near future12. What is badly lacking now is an extensive assessment of alterations in p33ING1 — at the level of the gene as well as the protein — in human cancer. If such alterations are encountered at a significant frequency, we can take it for granted that p33ING1 will soon follow in the footsteps of its better-known team-mate and, like p53, will become a hot item in biomedical research.


  1. 1

    Levine, A. J. Cell 88, 323–331 (1997).

  2. 2

    Garkavtsev, I.et al. Nature 391, 295–298 (1998).

  3. 3

    Garkavtsev, I., Kazarov, A., Gudkov, A. & Riabowol, K. Nature Genet. 14, 415–420 (1996).

  4. 4

    Helbing, C. C., Veillette, C., Riabowol, K., Johnston, R. N. & Garkavtsev, I. Cancer Res. 57, 1255–1258 (1997).

  5. 5

    Polyak, K., Xia, Y., Zweier, J. L., Kinzler, K. W. & Vogelstein, B. Nature 389, 300–305 (1997).

  6. 6

    Tanaka, N.et al. Nature 382, 816–818 (1996).

  7. 7

    Maheswaran, S., Englert, C., Bennett, P., Heinrich, G. & Haber, D. A. Genes Dev. 9, 2143–2156 (1995).

  8. 8

    MacCallum, D. E.et al. Oncogene 12, 2575–2587 (1996).

  9. 9

    Momand, J. & Zambetti, G. J. Cell. Biochem. 64, 343–352 (1997).

  10. 10

    Davidoff, A. M., Pence, J. C., Shorter, N. A., Iglehart, J. D. & Marks, J. R. Oncogene 7, 127–133 (1992).

  11. 11

    Moll, U. M.et al. Mol. Cell. Biol. 16, 1126–1137 (1996).

  12. 12

    Barinaga, M. Science 278, 1036–1039 (1997).

Download references

Author information

Rights and permissions

Reprints and Permissions

About this article

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.