Casein Kinase 1 Alpha Maintains Normal and Leukemic Stem Cells by Regulating p53 Activity

Abstract 209

In order to identify novel approaches to the targeting of acute myeloid leukemia (AML), we performed a pooled in vivo shRNA screen on murine leukemic stem cells (LSCs) targeting factors related to Wnt-signaling. We found that silencing of casein kinase 1 alpha (Csnk1a1), a serine-threonine kinase and a critical negative regulator of beta catenin, dramatically depleted murine LSCs in vivo. This is a surprising result since beta catenin is essential for MLL-AF9 AML. Validation experiments with shRNA vectors co-expressing GFP recapitulated the result from the pooled screen and confirmed efficient knockdown of both the Csnk1a1 transcript and protein. To rule out off-target effects of the Csnk1a1 shRNAs, we co-expressed the shRNAs with a Csnk1a1 cDNA mutated at the shRNA binding sites, and observed a complete rescue of the proliferative defect. Additionally, we demonstrated that a kinase dead form of Csnk1a1(D136N) failed to rescue this proliferation defect. These results indicate the specific effect of these hairpins on Csnk1a1 function in leukemia cells.

The role of Csnk1a1 in normal hematopoietic stem and progenitor cells (HSPCs) is not known. We introduced the Csnk1a1 shRNA vectors into HSPCs and followed GFP over time in a bone marrow transplantation setting. Over a 24-week period, we observed a 3–4 fold depletion of GFP positive donor cells with two independent Csnk1a1 shRNAs compared to control. In contrast, the same shRNAs resulted in a 20–25 fold depletion of leukemia cells in vivo over a 2-week time period, suggesting that leukemia cells are selectively dependent on Csnk1a1.

To more rigorously study Csnk1a1 in hematopoiesis, we generated a Csnk1a1 conditional knockout mouse (loxP sites flanking critical exon 3) and crossed it with the Mx1-cre mouse, allowing for hematopoietic specific inducible Csnk1a1 excision. In competitive bone marrow transplantations, Csnk1a1(−/−) donor cells exhibited a severe competitive disadvantage resulting in a 20-fold depletion of donor cells over a 12-week period. Interestingly, Csnk1a1(−/−) donor cells were devoid of myeloid lineage cells, suggesting that Csnk1a1 is particularly important for the generation or survival of myeloid cells. Moreover, in line with our shRNA results, we found that Csnk1a1(−/−) cells were resistant to MLL-AF9 mediated transformation, demonstrating that Csnk1a1 is essential also for leukemia initiation.

To identify critical targets of Csnk1a1, we performed gene expression profiling of Csnk1a1 silenced cells. We identified enrichment of a p53 signature using Gene Set Enrichment Analysis (FDR= 0.001). Induction of p53 and its target p21 was confirmed by western blots in both Csnk1a1 silenced leukemia cells and in Csnk1a1(−/−) bone marrow cells. Furthermore, we demonstrated that p53(−/−) leukemia cells are resistant to the proliferative defect induced by Csnk1a1 silencing.

We next tested whether D4476, a small molecule casein kinase inhibitor, would exhibit selective anti-leukemic effects. Whereas treatment of LSCs with D4476 inhibited their proliferation (IC50: 7μM), concentrations up to 40μM had minimal effects on normal HSPCs. Confirming the specificity of the compound, we found that cells carrying Csnk1a1 shRNAs were sensitized to D4476 in a dose dependent manner. In contrast, overexpression of Csnk1a1 desensitized leukemia cells for D4476 treatment, suggesting that D4476 kills leukemia cells in a Csnk1a1 dependent manner. Finally, we mixed 10,000 HSPCs with 10,000 LSCs and treated them ex vivo with either D4476 or DMSO control for 48 hours followed by injection into lethally irradiated mice. Whereas exposure to the drug caused prolonged latency of disease with some recipients never developing leukemia, there was no significant effect on HSPC donor cell chimerism at 8 weeks post transplantation compared control, indicating limited toxicity from the drug.

In summary, these findings identify Csnk1a1 as critical for maintaining both normal HSCs and LSCs via modulation of p53 activity. Importantly, LSCs were significantly more sensitive to small molecule inhibition of Csnk1a1, suggesting that Csnk1a1 may be an attractive new drug target in AML.