The INK4A/ARF locus codes for two, overlapping transcripts, INK4A and ARF, both of which are frequently mutated in human cancer. Knocking out either Arf or both gene products predisposes mice to tumorigenesis, so what role does Ink4a (also known as p16) play? Two studies in the 7 September issue of Nature now address this problem, as Sharpless et al. and Krimpenfort et al. show that Ink4a is a bona fide tumour suppressor.

The two proteins are created by the use of different first exons and reading frames. INK4A is a cyclin-dependent kinase inhibitor, which targets cyclin-D–CDK4/6 and prevents phosphorylation of RB; ARF, a known tumour suppressor, stabilizes p53 through its interaction with MDM2. To investigate whether Ink4a also participates in tumorigenesis, both groups mutated the gene in mice, independently of Arf. Krimpenfort et al. introduced a stop codon in exon 2 to mimic a mutation that is frequently found in human tumours (Ink4a*/*). This resulted in an unstable protein that could not be detected by immunoblotting. Sharpless et al. generated a knockout by deleting exon 1α (Ink4a−/−). Interestingly, mouse embryo fibroblasts (MEFs) harbouring either of these mutations have the same growth rate as wild-type MEFs, show normal Rb phosphorylation patterns and can still respond to γ-irradiation and serum starvation by arresting in G1.

Ink4a accumulates with the onset of senescence in a process that depends on telomere shortening. However, both Ink4a−/− and Ink4a*/* MEFs undergo growth arrest in culture or when RAS, a potent oncogene, is overexpressed. In contrast, Arf−/− MEFs continue to grow at the same rate under these conditions, which indicates that Arf is the principal mediator of senescence. Despite this, Ink4a−/− cells still immortalize at a greater frequency than Ink4a+/+ cells, and this does not always accompany loss of Arf or p53. Ink4a may therefore be able to facilitate escape from growth arrest.

When susceptibility to tumorigenesis was examined, some differences between the two groups' approaches emerged. Krimpenfort et al. showed no significant increase in the number of tumours when wild-type and Ink4a*/* mice were compared. Also, introduction of the Ink4a*/* mutation to Eμ- Myc mice — a well-established model of B-cell lymphoma — did not increase B-cell lymphomagenesis. However, when exons 2 and 3 of the second Ink4a allele were deleted — to generate a genotype frequently found in human tumours — there was a marked increase in tumour number. The remaining Arf allele was present in most of the tumours and gene silencing by methylation was not detected, so the increase in number was not due to loss of the second Arf allele.

One of the key uses for this mutant will be to model metastatic melanoma — the most predominant tumour type for humans with germ-line mutations at this locus. Application of DMBA, a known carcinogen, increased both the frequency of melanoma and the extent of metastasis.

Sharpless et al. showed that Ink4a−/− mice developed more tumours than wild-type and heterozygous mice; treatment with a variety of carcinogens further increased both malignancy and tumour type. As carcinogen-treated Ink4a+/− mice were more prone to tumours than wild-type mice, the authors investigated the status of the functional Ink4a allele in those tumours. Ink4a protein was not detected in any tumour; the Ink4a/Arf locus was not rearranged or deleted, but the gene could be epigentically silenced by promoter methylation — a mechanism of Ink4a loss also noted in many human tumours and an obvious target for therapy.

These papers provide a long-awaited answer to the conundrum of which protein is important in tumorigenesis — the answer being both. Questions for the future include which tumour suppressor is inactivated in which tumour type, and how this loss is brought about — through gene silencing or gene deletion.