Dogma would have it that both copies of a tumour-suppressor gene must be inactivated to promote tumour formation. But a new breed of tumour suppressor is now emerging — for which inactivation of one allele is enough. In 1998, James Roberts, Christopher Kemp and their collaborators established that CDKN1B, which encodes the cyclin-dependent kinase inhibitor Kip1, is haplo-insufficient for tumour suppression. Loss of just one copy of the Pten tumour suppressor is also sufficient to promote tumour progression in a transgenic model of prostate cancer. Now, Kazushi Inoue and colleagues define a new one-hit tumour suppressor in the 15 November issue of Genes & Development.

Most tumours downregulate the p53 pathway by mutating or deleting both TP53 alleles, expediting the disposal of p53 by upregulating MDM2, or inactivating ARF, which inhibits MDM2. These proteins are kept in check by a complex network of controls: ARF, for example, is regulated by several transcriptional activators and repressors, including the transcriptional activator DMP1.

Might DMP1 be a tumour suppressor in its own right? Inoue and colleagues previously reported that Dmp1−/− mice didn't spontaneously develop tumours in their first year of life, but now the mice are older and we're all a little wiser. In their second year, Dmp1−/− mice spontaneously developed a variety of tumour types but, intriguingly, so did Dmp1+/− mice. This was not due to loss of heterozygosity or epigenetic silencing of the wild-type Dmp1 allele because, in all tumours tested, the second allele was retained and mRNA and protein were produced.

To explore whether Dmp1 mutant tumours select for loss of other genes in the p53 pathway, the authors crossed Dmp1 mutant mice with Eμ-Myc mice, which spontaneously develop Burkitt's lymphomas. Around half of these tumours usually contain either p53 mutations or Arf mutations. Both Eμ-Myc/Dmp1+/− mice and Eμ-Myc/Dmp−/− mice developed tumours with a much shorter latency than Eμ-Myc/Dmp+/+ mice — 12 weeks rather than 6 months. Most of the tumours in the Eμ-Myc/Dmp+/− mice produced detectable wild-type Dmp1 protein, supporting the notion that losing just one Dmp1 allele promotes tumour formation. Furthermore, only a small percentage of these mice (9% for Dmp1−/− mice; 14% for Dmp1+/− mice) sustained mutations in p53 or Arf.

Dmp1, then, is a bona fide tumour suppressor and its downregulation, even by 50%, reduces the selection pressure for loss of p53 or Arf function. One of the central tenets of cancer biology — Knudson's two-hit hypothesis — might just have sustained another hefty blow.