Second-site POLE mutations in the long-term adaptation of ultramutant cancers

Polymerase-associated ultramutagenesis is associated with the highest mutation rates in human cancers (up to 400-500 mutations/megabase) and distinctive clinical features (excellent prognosis, infrequent metastasis, and dramatic responses to anti-cancer therapies). Mutational burden correlates with cell surface neoepitopes, sensitizing tumors to 1) innate immune surveillance and 2) immune checkpoint blockade. However, elevated mutational burden might also decrease cell fitness independent of immune surveillance. Here we formulate and explore the POL automutation hypothesis, which postulates that ultramutating tumor cells escape relentless ultramutation by inactivating their mutant POLE alleles via second site mutations. We developed ex vivo models based on serial passaging mouse PoleP286R/+ endometrial cancer cells. Short-read sequencing permitted detection of second site mutations, with longread sequencing to discriminate those in cis vs. trans. While second site mutations were identified, these were as likely to be in cis as in trans. Diverse human cancer cell lines with POLE driver mutations did not have second site mutations. Taken together, our findings argue against intrinsic cell fitness as a common, physiologically relevant mechanism explaining the distinct clinical features of ultramutant cancers. Instead, our findings and prior studies together point to immune surveillance or other adaptive mechanisms as playing more important roles