Aurora Kinase a/MDM2-Mediated SETD2 Loss of Function in Chronic Myeloid Leukemia Patients in Blast Crisis Can be Therapeutically Targeted Inducing Apoptotic Cell Death in a Caspase-Dependent Way

One of the hallmarks of chronic myeloid leukemia (CML) is genomic instability, that fosters the acquisition of tyrosine kinase inhibitor (TKI)-resistant BCR-ABL1 mutations and/or of additional chromosomal aberrations leading to progression to blast crisis (BC).

Inactivating mutations in the SETD2 tumor suppressor occur in solid tumors and acute leukemias. SETD2 trimethylates histone H3 Lysine 36 (H3K36Me3) playing a key role in maintaining DNA integrity. We have recently demonstrated that, in CML, SETD2 loss of function may occur at the post-translational level. Reduced or null SETD2 and H3K36Me3 was detected in 83/96 (86%) patients (pts) with BC CML as compared to a pool of healthy donors and to chronic phase (CP) pts at diagnosis. Proteasome inhibition in primary cells from pts with undetectable SETD2 restored H3K36Me3 and led to accumulation of hyper-ubiquitinated SETD2. In K562 cells (SETD2/H3K36Me3^low), we observed that after proteasome inhibition hyper-ubiquitinated SETD2 co-immunoprecipitates with MDM2. MDM2 inhibition rescued SETD2 expression and activity, suggesting that MDM2 is implicated in SETD2 reduced stability. Co-IP also showed that SETD2 interacts with Aurora Kinase A (AKA) a S/T kinase frequently overexpressed in CML. We found that AKA phosphorylates SETD2, and its inhibition rescued SETD2 expression and activity.

To investigate whether SETD2/H3K36Me3 loss may be a druggable lesion, we performed clonogenic assays in LAMA84 (SETD2/H3K36Me3^high) cells before and after SETD2 silencing, in imatinib-sensitive K562 (SETD2/H3K36Me3^low) cells and in IM-resistant K562 cells, that are characterized by complete SETD2 loss. The extent of reduction of clonogenic growth after proteasomal, AKA or MDM2 inhibition was found to be inversely correlated to SETD2 residual expression. These observations were confirmed in cells from both CP (n=2) and BC (n=4) CML pts showing different levels of SETD2 expression and activity.

Further experiments were performed in the aforementioned cell lines to verify if reduced clonogenic potential was due to cytostatic or cytotoxic effects. Apoptotic cell death was quantified by annexin V/propidium iodide staining and flow cytometry. Proteasomal inhibition by bortezomib, carfilzomib and ixazomib and AKA de-phosphorylation by Danusertib caused a time-dependent increase of annexin-V-positive cells by activating the mitochondrial apoptotic pathway as reflected by an increase in Bax expression and induction of the cleavage of caspase-3,-9 and PARP. Moreover, all drug treatments as single agent, at nanomolar doses (Bortezomib: 10 nM, Carfilzomib: 5 nM, Ixazomib: 10 nM and Danusertib: 500 nM) induced a significant increase of the DNA double-strand break marker γH2AX, suggesting that in a SETD2 knock-down context, proteasomal and AKA inhibition propagates genomic instability by forcing the cells through successive replication cycles, ultimately resulting in apoptosis from mitotic catastrophe.

Reduced SETD2/H3K36Me3 levels, in association with MDM2 and AKA hyper-activation, were also detected when the CD34+ cell fraction of 10 CML-CP pts, was compared to the total mononuclear cell fraction or to the CD34+ compartment obtained from a pool of healthy donors. We thus hypothesized that leukemia progenitor cells, showing higher MDM2 and AKA activity and consequent SETD2 loss, accumulate genetic aberrations despite inhibition of BCR-ABL1 kinase. Studies are ongoing to verify if MDM2 or AKA inhibition may restore SETD2 expression and function and induce cell death.

Finally, it has already been shown that alterations of epigenetic regulators such as the KDM4 family members control tumor cell proliferation in a variety of cancers including acute myeloid leukemia. Recent findings have identified KDM4 demethylases as putative therapeutic targets in a SETD2 mutated context and illustrated the efficacy of KDM4 inhibitors in AML therapy. Starting from these evidences, we will test the same approach in BC CML models.

In conclusion, phosphorylation by AKA and ubiquitination by MDM2 contribute to SETD2 non-genomic loss of function in BC CML and in CD34+ leukemic progenitors. Restoring physiological H3K36Me3 may help to improve the outcome of this critical subset of pts.

Acknowledgments: Study supported by AIRC (project code 16996), AIL (Associazione Italiana contro le Leucemia, Linfomi e Mieloma), Italian Ministry of Health, project GR-2016-02364880.