Uncoupling p53 from an Embryonic Regulome Exhausts Quiescent CML Stem Cells through Inhibition of a HIF1alpha Molecular Program

Although it has been recognized for many years that cancer stem cells and embryonic stem cells (ESC) share molecular features, identifying ways to exploit this therapeutically has proved challenging. To date, these shared features have not been examined in the leukemic stem cells (LSC) found in patients with chronic myeloid leukemia (CML).

By integrating known ES regulatory circuitry with transcriptomics datasets, including deep single-cell RNA-seq profiling of 15,670 LSC from five patients with CML, we identified a core ESC regulome in the LSC containing 1243 genes. The significant majority of this regulome (1102 genes) was up-regulated in cycling LSC, whilst quiescent LSC showed up-regulation of a characteristic set of 101 genes, unique to cells with high ESC identity and with regulatory circuitry enriched for c-Myc and Nanog modules. Membership of the ESC regulome included the TP53 gene which was transcriptionally repressed and detected at a lower frequency in quiescent LSC compared to cycling ones (11.8% vs 43.6%). We also demonstrated that tyrosine kinase inhibitors (TKI) repress the ESC regulome and TP53 expression in LSC, suggesting that the regulome safeguards against high levels of TP53 expression, thus promoting survival of quiescent LSC in the presence of TKI. We hypothesized that overcoming the influence of the regulome on TP53 expression would provide an opportunity to eradicate quiescent LSC.

To this end, we used an MDM2 inhibitor (MDM2i), RG7388 (idasanutlin) or RG7112, to stabilize the p53 protein, examining its potential in combination with nilotinib (NIL) to eradicate CML LSC in vitro and in vivo, with RG7388 being the most optimized and furthest in development. The combination of NIL plus MDM2i in vitro was more effective at targeted LSC from primary patient samples than NIL treatment alone, as evidenced by reduced CFC and LTC-IC outputs (p<0.05, 0.01 respectively). Intriguingly, the combination of NIL plus MDM2i did not result in significant reductions in the number of LSC compared to NIL only, when we quantified them at the end of drug treatments in pre-clinical mouse models. Instead, we observed a functional decline of the LSC as evidenced by diminished engraftment potential in 2 ^o recipient mice (p<0.05; SCLtTA x BCR-ABL1 transgenic model) or diminished colony-plating potential (p<0.05). This was followed by near complete depletion of the LSC population (p<0.05) 28 days after cessation of combination drug treatment (patient-derived xenografts/PDX in immunocompromised mice).

In order to understand the molecular events underpinning these drug effects on LSC, we performed RNA-seq analysis of drug-treated CD34 ⁺ cells in vitro (bulk cells), or of human CD34 ⁺ cells obtained from PDX (single cell RNA-seq). CD34 ⁺ cells treated with NIL plus MDM2i in vitro showed evidence of increased p53 stabilization and activation of p53 target genes, and this was accompanied by repression of the ESC regulome beyond that normally observed with NIL only. Similarly, in PDX we observed increased repression of the ESC regulome in human CD34 ⁺ cells exposed to the combination of NIL plus MDM2i that included repression of HIF1alpha and a signature of genes required for cellular adaptations to hypoxia, and growth factor-mediated resistance to TKI therapy. Further, single cell analysis of differentiated human CD45 ⁺ cells from our PDX model, provided compelling evidence that acquisition of this repressive signature in the LSC, through combined NIL plus MDM2i treatment, re-wires them towards a basophilic fate, consistent with functional exhaustion of the LSC compartment.

In conclusion, we have identified an ESC regulome in CML LSC and demonstrate that a combination of a TKI plus an MDM2i leads to p53 upregulation which antagonizes this regulome, providing a highly effective strategy to target near complete loss of functional LSC in pre-clinical models. Our study has revealed a new therapeutic paradigm to examine in other cancer stem cell populations that utilize ESC regulatory programs.