Disruption of the normal pattern of imprinting has been implicated in cancer; for example, loss of imprinting (LOI) at the insulin-like growth factor 2 ( IGF2 ) locus is seen in several common cancers, and the 10% of the population with LOI at IGF2 have a higher than normal risk of developing colorectal cancer. A recent report by Andrew Feinberg and colleagues shows that epigenetic alterations to IGF2 promote intestinal tumorigenesis and that a shift towards an undifferentiated cell fate might contribute to an increased rate of tumour initiation.

Wanting to understand how epigenetic changes at the IGF2 locus cause intestinal tumorigenesis, the authors created a mouse model using a deletion of the differentially methylated region (DMR) that regulates Igf2 imprinting in mice. Female mice lacking the DMR inherit a normal active copy of Igf2 from their father, but also an abnormally active (imprint-free) copy of Igf2 from their mother. Crossing these females with adenomatous polyposis coli (Apc)-mutant males predisposes the resulting offspring to multiple intestinal neoplasia. The authors compared Apc mutants carrying a silenced maternal copy of Igf2 (they called these LOI(−) mice) to their Apc mutant littermates (LOI(+) mice) that had two active copies of Igf2.

Confirming the role of Igf2 in colon cancer, LOI(+) mice developed twice as many tumours in the small intestine and colon. Curiously however, LOI(+) mice also had longer intestinal crypts — the invaginations in the intestinal epithelium, at the bottom of which stem-cell proliferation occurs. Feinberg and co-workers postulated that this increase in crypt length could be due to a shift in the ratio of differentiated to undifferentiated epithelial cells. Sure enough, LOI(+) mice had higher levels of undifferentiated epithelial-cell markers in their small intestines than LOI(−) mice. A similar increase was seen in the intestinal epithelium of patients with IFG2 LOI.

The authors suggest that this shift increases tumour initiation, rather than the rate of tumour progression, and, supporting this, they found that LOI(+) and LOI(−) mice have the same ratio of small to large tumours. It is likely that the shift in normal tissue to a less differentiated state increases the number of cells that are vulnerable to subsequent genetic alterations. Future work will determine whether epithelial differentiation or LOI itself will be the better predictor of cancer risk.