Induction of senescence by chemotherapy was initially characterized as a suppressive response that prevents tumor cell proliferation. However, in response to treatment, it is not really known how cells can survive senescence and how irreversible this pathway is. In this study, we analyzed cell escape in response to irinotecan, a first line treatment used in colorectal cancer that induced senescence. We detected subpopulations of cells that adapted to chemotherapy and resumed proliferation. Survival led to the emergence of more transformed cells that induced tumor formation in mice and grew in low adhesion conditions. A significant amount of viable polyploid cells was also generated following irinotecan failure. Markers such as lgr5, CD44, CD133 and ALDH were downregulated in persistent clones, indicating that survival was not associated with an increase in cancer initiating cells. Importantly, malignant cells which resisted senescence relied on survival pathways induced by Mcl-1 signaling and to a lesser extent by Bcl-xL. Depletion of Mcl-1 increased irinotecan efficiency, induced the death of polyploid cells, prevented cell emergence and inhibited growth in low-adhesion conditions. We therefore propose that Mcl-1 targeting should be considered in the future to reduce senescence escape and to improve the treatment of irinotecan-refractory colorectal cancers.

Activated in response to chemotherapy, senescence is a tumor suppressive mechanism that induces a permanent loss of proliferation. However, in response to treatment, it is not really known how cells can escape senescence and how irreversible or incomplete this pathway is. We have recently described that cells that escape senescence are more transformed than non-treated parental cells, they resist anoikis and rely on Mcl-1. In this study, we further characterize this emergence in response to irinotecan, a first line treatment used in colorectal cancer. Our results indicate that Akt was activated as a feedback pathway during the early step of senescence. The inhibition of the kinase prevented cell emergence and improved treatment efficacy, both in vitro and in vivo. This improvement was correlated with senescence inhibition, p21waf1 downregulation and a concomitant activation of apoptosis due to Noxa upregulation and Mcl-1 inactivation. The inactivation of Noxa prevented apoptosis and increased the number of emergent cells. Using either RNA interference or p21waf1-deficient cells, we further confirmed that an intact p53-p21-senescence pathway favored cell emergence and that its downregulation improved treatment efficacy through apoptosis induction. Therefore, although senescence is an efficient suppressive mechanism, it also generates more aggressive cells as a consequence of apoptosis inhibition. We therefore propose that senescence-inducing therapies should be used sequentially with drugs favoring cell death such as Akt inhibitors. This should reduce cell emergence and tumor relapse through a combined induction of senescence and apoptosis.