CHARACTERIZATION OF THE MOLECULAR MECHANISM RESPONSIBLE FOR THE LOSS OF THE TUMOR SUPPRESSOR NUMB IN BREAST CANCER

The tumor suppressor protein, Numb, safeguards against the emergence of cancer stem cells (CSCs) in the mammary gland by ensuring homeostasis of the stem cell (SC) compartment and proper progenitor maturation. Upon its downregulation, expansion of the mammary SC compartment accompanied by the acquisition of tumorigenic potential ensues, highlighting the potent tumor suppressor function of Numb in breast cancer. Indeed, Numb loss occurs in ~30% of human breast cancers and correlates with biological aggressiveness and poor prognosis. Notably, restoration of Numb expression curbs tumorigenic potential of Numb-deficient breast cancers through the selective targeting of CSCs. Thus, the development of therapeutic strategies capable of restoring Numb expression in breast cancer could present new treatment opportunities. A well-established mechanism responsible for Numb loss in BC is its excessive polyubiquitination and consequent proteasomal degradation. Thus, it has been proposed that deregulation of the ubiquitin-proteasome system (UPS) could underlie Numb loss. The identification of the UPS machinery responsible for Numb loss would not only provide clues as to the underlying lesions in Numb-deficient cancers, but could pave the way to the development of novel strategies to restore Numb expression. In our lab, through a siRNA-based high-throughput screening of the ubiquitination machinery, we previously identified the RING-type E3 ligase, RBX1, and the F-BOX protein, FBXW8, as determinants of Numb degradation. The involvement of these proteins was validated in established BC cell lines through high-resolution studies. Notably, RBX1 and FBXW8 are components of multiprotein E3-ligase complexes, Cullin RING Ligases (CRLs). Based on these observations, the founding hypothesis of this thesis is that deregulation of an RBX1-FBXW8 CRL complex could be responsible for Numb loss in BC, and, thus, represent a potential target for therapeutic intervention in Numb-deficient tumors. Here, we initially generated Numb-deficient (MDA-MB-361) and -proficient (MDA-MB-231) cell lines stably expressing doxycycline-inducible shRNA against candidate Numb regulators. This genetic tool enabled us to achieve the in vitro and in vivo conditional ablation of RBX1 or FBXW8, thus mimicking the administration of putative drugs inhibiting specifically CRL component activity. We exploited this tool to determine the impact of RBX1 and FBXW8 silencing on the formation of outgrowths in in vitro 3D Matrigel organotypic culture: a widely used proxy of in vivo tumorigenesis. We demonstrated that the downmodulation of CRL components led to Numb restoration and impaired tumorigenic potential in vitro, selectively in Numb-deficient 3D outgrowths. We next used pre-clinical in vivo models based on orthotopic xenografts of Numb-deficient and -proficient cells to elucidate the therapeutic value of targeting the components of the UPS system. We showed that doxycycline-induced RBX1 and FBXW8 ablation curbed tumor growth selectively in Numb-deficient models, recapitulating the effects achieved upon proteasome inhibition in vivo (e.g., with Bortezomib). To investigate whether this inhibition of in vivo tumor growth was strictly dependent on Numb protein restoration, Numb-deficient cells expressing doxycycline-inducible shRNA against Numb were used as genetic tool to prevent Numb rescue upon proteasome inhibitor treatment. We showed that Numb conditional ablation induced resistance to Bortezomib treatment. We further demonstrated in vitro that the more specific Cullin ligase inhibitor, MLN4924, successfully recapitulated the effects of Bortezomib on Numb rescue in Numb-deficient model cells. Overall, the results of this thesis argue for the involvement of a CRL complex, composed of RBX1 and FBXW8, in the excessive ubiquitination of Numb in Numb-deficient breast cancers. Through pre-clinical in vivo models we have also provided proof-of-principle of the therapeutic value of targeting the UPS machinery directly involved in Numb hyperdegradation. In particular, we have identified two potential drugs, already in clinical use or under clinical development (Bortezomib and MLN4924, respectively), which could eventually be repositioned towards the treatment of Numb-deficient breast cancers.