New-age tumour suppressors

The discovery of a gene that is inactivated in stomach cancers illustrates the value of mouse models for finding tumour-suppressor genes that are switched off by mechanisms other than mutation.

Tumour-suppressor genes form a crucial part of our natural anticancer defences. When these genes become inactive, tumours often develop. Most cells have two copies (alleles) of every gene, and the classical view of a tumour suppressor is that loss of function only occurs when both copies are inactivated genetically. First, a mutation takes place in one allele in a developing tumour cell; next, the remaining normal allele is knocked out by 'loss of heterozygosity' (LOH) — the cell loses a large part of the chromosome on which the normal allele resides1.

These ideas led to the discovery of a whole generation of tumour suppressors, including the well-known retinoblastoma protein and p53 (see ref. 2 for a review). However, although LOH is often seen in certain chromosomal regions in human and mouse cancers, the rate of discovery of the tumour suppressors in those regions seems to have slowed substantially. This may be because many tumour-suppressor genes take alternative routes to inactivation. Writing in Cell, Li et al.3 describe how they identified RUNX3 — which they find to be switched off in human gastric cancers — as just such a gene.

Why have the new-generation tumour suppressors been so difficult to identify? One reason may be the strategy that is typically used: find the characteristic region of LOH in a particular tumour; narrow down that region as much as possible by 'deletion mapping'; and hunt for inactivating mutations within genes in the corresponding region on the matching, undamaged chromosome. This allows the discovery of classical tumour suppressors, which have genetic alterations in both alleles. But tumours are more resourceful than we once thought, and as well as genetic mechanisms, they use an armoury of epigenetic mechanisms (which do not involve irreversible changes in DNA) to silence genes that impede rapid cell growth. For example, we now know that tumour-suppressor genes can be silenced by mechanisms such as the reversible modification of the regulatory promoters with methyl groups, either on both alleles, or on just one when the other has been deleted4.

In addition, there may be not one but several genes with overlapping functions within the LOH region of interest. One example of this might be provided by a region on human chromosome 3 that is frequently lost in lung tumours5; pinpointing the critical gene in this section has proved extremely difficult. A strong candidate is infrequently mutated but is often silenced by promoter methylation6. To add to this already complicated scenario, 'haploinsufficient' tumour-suppressor genes show loss of function after only one of the two alleles is altered7. This raises the spectre of large regions of LOH containing many genes, with one or more key genes being haploinsufficient and therefore hard to find by traditional means, because the corresponding alleles on the undamaged chromosome are not necessarily mutated. Clearly we need to find other approaches.

Functional studies in mice are proving extremely useful here, and lie behind the clever detective work of Li et al.3. Runx1, Runx2 and Runx3 are related mouse genes that encode gene-transcription factors with an impressive list of alternative names and functions8; for example, their human counterparts are known to be involved in leukaemias and certain congenital abnormalities. Li et al. studied the precise role of Runx3 by knocking out the gene in mice.

They found that the Runx3-deficient animals died shortly after birth, probably because they couldn't feed properly — the epithelial cells that made up their stomach lining had multiplied excessively. In other words, Runx3 is involved in keeping cell numbers under control, a typical feature of a tumour suppressor.

A similar effect is seen in mice that lack the transforming growth factor-β1 (TGF-β1) protein, which controls cell numbers by inhibiting cell proliferation and inducing cell death. This fact — together with the finding that Smad proteins, which transmit cellular signals coming from TGF-βs, interact directly with the Runx proteins9 — led Li et al. to investigate the TGF-β1 signalling pathway in cells from the Runx3-deficient mice. The TGFβ1-induced inhibition of proliferation was only modestly affected in Runx3-deficient gastric epithelial cells. But the ability of TGF-β1 to induce cell death was strongly impaired. Moreover, when the authors reintroduced a functional copy of the Runx3 gene into gastric tumour cells that lacked the protein, tumour growth was inhibited — a strong sign that the authors had identified a gene that suppresses gastric cancers in mice.

Li et al. also looked at the expression of the human RUNX3 gene in patients with gastric cancer. They found that the loss of function of this gene — as a result of deletion of one allele followed by methylation-induced silencing of the other — correlated with the progression of cancer, with all of the eight most advanced tumours studied showing deletions. But the authors' search for the mutational hallmarks of a classical tumour-suppressor gene was unsuccessful. They detected just one RUNX3 mutation, a 'missense' mutation that alters the encoded protein's structure but may leave some of its functions intact, in 119 tumours. So RUNX3 is by no means a classical tumour suppressor.

In humans, RUNX3 is located on the short arm of chromosome 1, in a region designated 1p36 that has long been of interest to cancer researchers. Is RUNX3 the critical tumour-suppressor gene in this region? No doubt studies of gene expression and promoter methylation, and mutation hunting, will soon provide the solution. In the meantime, there are other questions to answer, not least how RUNX3 works to keep cell numbers in the stomach under control.

The interaction between the RUNX and SMAD proteins, downstream targets of TGF-β signalling, may be important here. But the answer will not be straightforward. Although SMAD proteins can inhibit cell proliferation, they also induce tumour invasion and metastasis10,11. Moreover, in genetically engineered animal models, TGF-β can inhibit early stages of tumour development but stimulate later tumour progression and metastasis10. Perhaps the loss of RUNX3 blocks some of the inhibitory effects of TGF-β on cell proliferation and death but leaves intact the positive effects of SMADs on malignant progression. In fact, there is evidence that the three RUNX genes have multifunctional roles in different tumours and, perhaps, at different stages of cancer12,13.

To muddy the waters still further, although Li et al.3 provide convincing evidence that RUNX3 helps to suppress gastric cancer, it seems that a retrovirus activates the Runx3 gene in T-cell lymphomas in mice14. Clearly, much remains to be done to unravel the complexities of this interesting gene family and its roles in cancer.

The bad news about the new wave of tumour-suppressor genes is that they are hard to find. But, unlike the old-style tumour suppressors, they may be good drug targets because they are not mutated, and so could perhaps be reactivated using small molecules that prevent or reverse methylation. Some drugs of this sort are now under development or in clinical trials15. Such new-wave drugs may be what is needed to tackle the new generation of tumour suppressors.


  1. 1

    Cavenee, W. K. et al. Nature 305, 779–784 (1983).

    ADS  CAS  Article  PubMed  Google Scholar 

  2. 2

    Balmain, A. Nature Rev. Cancer 1, 77–82 (2001).

    CAS  Article  Google Scholar 

  3. 3

    Li, Q. L. et al. Cell 109, 113–124 (2002).

    CAS  Article  Google Scholar 

  4. 4

    Jones, P. A. & Laird, P. W. Nature Genet. 21, 163–167 (1999).

    CAS  Article  Google Scholar 

  5. 5

    Lerman, M. I. & Minna, J. D. Cancer Res. 60, 6116–6133 (2000).

    CAS  PubMed  Google Scholar 

  6. 6

    Dammann, R. et al. Nature Genet. 25, 315–319 (2000).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  7. 7

    Philipp-Staheli, J., Payne, S. R. & Kemp, C. J. Exp. Cell Res. 264, 148–168 (2001).

    CAS  Article  Google Scholar 

  8. 8

    Cohen, M. M. Am. J. Med. Genet. 104, 185–188 (2001).

    Article  PubMed  Google Scholar 

  9. 9

    Hanai, J. et al. J. Biol. Chem. 274, 31577–31582 (1999).

    CAS  Article  PubMed  Google Scholar 

  10. 10

    Derynck, R., Akhurst, R. & Balmain, A. Nature Genet. 29, 117–129 (2001).

    CAS  Article  Google Scholar 

  11. 11

    Oft, M., Akhurst, R. J. & Balmain, A. Nature Cell Biol. (in the press).

  12. 12

    Osato, M. et al. Blood 93, 1817–1824 (1999).

    CAS  PubMed  Google Scholar 

  13. 13

    Blyth, K. et al. Oncogene 20, 295–302 (2001).

    CAS  Article  PubMed  Google Scholar 

  14. 14

    Stewart, M., MacKay, N., Cameron, E. R. & Neil, J. C. J. Virol. 76, 4364–4369 (2002).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  15. 15

    Brown, R. & Strathdee, G. Trends Mol. Med. 8, S43–S48 (2002).

    CAS  Article  PubMed  Google Scholar 

Download references

Author information



Corresponding author

Correspondence to Allan Balmain.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Balmain, A. New-age tumour suppressors. Nature 417, 235–237 (2002).

Download citation

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.


Quick links

Sign up for the Nature Briefing newsletter for a daily update on COVID-19 science.
Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing