By Sharon Kingman

BioWorld International Correspondent

LONDON ¿ Mouse models of human genetic diseases can be very useful, but only when disruption of a gene common to both species causes similar abnormalities. Scientists find it disconcerting if they mutate a mouse gene and end up with a very different clinical picture to the human genetic disorder.

Such was the case with the part of the genome known as CDKN2A. Mutations in this area are known to be common in human cancers, and are associated with more human cancers than any other except the p53 gene.

The trouble was that the human families that had germline mutations in CDKN2A were susceptible to melanomas, while mice which had mutations induced in CDKN2A tended to have lymphomas.

Those who have lost sleep over this particular conundrum can now rest easy in their beds, however. Two independent research groups have proved that loss of part of the CDKN2A section in mice can indeed cause the animals to develop melanomas.

The two groups, one in the Netherlands and one in the U.S., report their findings in separate letters to Nature. The title of the Dutch group¿s letter is ¿Loss of p16Ink4a confers susceptibility to metastatic melanoma in mice¿ while that of the American group is ¿Loss of p16Ink4a with retention of p19Arf predisposes mice to tumorigenesis.¿

Anton Berns, director of research at the Netherlands Cancer Institute in Amsterdam, told BioWorld International: ¿The results reported in these two papers are exciting from the point of view of the analysis of the CDNK2A locus, because they have explained something that, formerly, we didn¿t understand. We didn¿t know how to explain the discrepancy between what we saw in humans and what we saw in the mouse, but now we have, and everything has fallen into place. It is all very reassuring.¿

The CDNK2A region of the genome is highly unusual in that it encodes two completely different proteins, depending on where the cell¿s machinery for reading DNA begins. Some bacteria are known to do this, but CDNK2A is the only part of the genome in mammals that works in the same way.

The proteins thus encoded are called p16Ink4a and p19Arf. For this reason, CDNK2A also is known as Ink4a/Arf. Interestingly, both proteins play a role in turning off cell division.

Scientists wanted to model the loss of p16Ink4a in mice, because they knew that mutations in the equivalent gene in humans were associated with an increased risk of melanoma. Berns explained: ¿This was when researchers discovered that, although there was a predisposition to tumors in these mice, the types of tumors they developed were primarily lymphomas. Then it was realized that the CDNK2A locus contained two genes, which overlap and use the same coding sequence, but in different reading frames.¿

The early knockout experiment, it turned out, had deleted sections of both genes, inactivating both of them. A team in the U.S. went on to knock out p19Arf alone. To everyone¿s surprise, the phenotype of these animals appeared to be almost indistinguishable from the double knockout.

Berns said, ¿This left us with a situation where we knew that, in humans, p16Ink4a was proven to be a tumor suppressor gene, whereas in mice, p19Arf seemed to be the relevant tumor suppressor gene, because mice in which either p19Arf alone or p19Arf and p16Ink4a together had been inactivated showed nearly the same tumor phenotype. We all wanted to know what on earth p16Ink4a does in the mouse. Is it also a tumor suppressor, or does it do something different in the mouse?¿

This was the question that the Dutch and American teams endeavored to answer. Berns said, ¿What we found was that when p16Ink4a was made defective, mice were more prone to develop tumors, especially when treated with carcinogens. Melanomas were frequently found under these conditions and seemed more aggressive. Interestingly, when, in addition, one copy of p19Arf was inactivated, the animals¿ predisposition to tumors significantly increased.¿

Berns added that these observations suggest that there is some interplay between the two genes, which will need to be studied further.

¿We expect this work to lead eventually to models for testing drugs to treat melanoma,¿ he said. ¿An obvious group to test would be inhibitors of the cyclin-dependent kinases.¿

The group expects to develop the model further by adding in more mutations so that melanoma development is more reproducible and occurs more rapidly.