Knockout and conditional mouse models allow investigators to compare phenotypic differences when gene expression is either 'on' or 'off', but mutations that occur in real life could generate expression levels that fall between these two extremes. In the March issue of Nature Genetics, Greg Hannon, Scott Lowe and colleagues use RNA interference (RNAi) to generate a series of p53 hypomorphs, and show that the level of p53 markedly affects the level of disease.

The authors generated three retroviral vectors that each contained a distinct short hairpin RNA to target the Trp53 gene. When transfected into cultured cells, these constructs — p53-A, -B and -C — had differential effects on p53 protein expression, which corresponded to differences in the ability of mouse embryonic fibroblasts (MEFs) to form colonies in a colony-formation assay. Following expression of p53-A, which had the least effect on protein levels, only a few colonies were able to grow, whereas p53-C significantly decreased protein expression and many more colonies grew.

So, do these RNAi constructs also affect the tumorigenic capacity of cells? Haematopoietic stem cells were infected with viruses produced from these constructs and were transplanted into Eμ- Myc mice, which develop B-cell lymphomas at 4–6 months. All mice developed palpable lymph nodes within 3–5 weeks, indicative of lymph-node hyperplasia, but only those with p53-B and p53-C developed B-cell lymphomas and showed a decrease in survival, compared with control Eμ-Myc mice.

Further analysis showed that p53-B recipients developed smaller and less-malignant tumours than p53-C recipients. The tumours of p53-B recipients still had a high rate of apoptosis, a low mitotic index and did not significantly infiltrate into the lung and liver, whereas the p53-C recipients had low levels of apoptosis, a high mitotic index and a large amount of infiltration. Interestingly, the tumours produced in both of these recipients were not genomically unstable, unlike those that are found in Trp53−/− mice. These results confirm how important the apoptotic function of p53 is for tumour suppression.

In Eμ-Myc mice that are heterozygous for Trp53, all tumours that develop show loss of heterozygosity at the Trp53 locus. Might expression of the RNAi constructs relieve this selection pressure? One of two tumours that developed in p53-B recipients and four of four tumours that developed in p53-C recipients had not lost the second wild-type Trp53 allele, indicating that loss of heterozygosity is not required for tumour development in the presence of the RNAi constructs.

So, as well as producing null phenotypes, RNAi can generate intermediate phenotypes that can help dissect protein function. This protocol also has interesting therapeutic implications, as it shows that stable RNAi can suppress the deleterious function of genes in stem cells ex vivo, and that these retain their function in vivo.