Abstract P34: Bayesian networks modeling identifies a reliance of TP53 mutant AML on NF kappa B signaling

Abstract

In acute myeloid leukemia (AML), TP53 mutations ( TP53Mut ) confer the worst prognosis. Leukemia stem cells (LSCs) are rare AML cells endowed with self-renewal capacity, allowing them to recapitulate leukemia after therapy and cause disease progression. In human AML, signaling pathways associated with inflammation have been implicated in the pathogenesis of TP53Mut AML, but the pathways that mediate self-renewal in human TP53Mut AML are not well-known. To define the signaling pathways that are activated in TP53Mut human LSCs, we used CyTOF, a form of flow cytometry that can measure up to 40 proteins simultaneously at single-cell resolution. We compared the levels of activated signaling molecules in the LSC-enriched (CD34+CD38−) subset of seven TP53Mut and seven TP53 wild-type ( TP53WT ) AML samples. Phosphorylated NF kappa B (pNFkB), pSTAT1, and pP38 are significantly higher in TP53Mut AMLs. In contrast, Ki67 and pMAPKAPKII are higher in TP53WT AML. These data demonstrate a unique signaling activation state in TP53Mut AML. The influence of signaling molecules on downstream targets can vary by cellular context. Therefore, we asked whether the influence and interaction of the activated signaling molecules vary between TP53Mut and TP53WT LSCs. To model the signaling architecture, we performed Bayesian networks modeling, a machine learning algorithm that has been validated as a method to discover statistical correlations and dependencies between signaling molecules. We found a similar global signaling network structure and hierarchy in TP53Mut and TP53WT AMLs. However, TP53Mut model showed NFkB have strong influences on phosphorylated Histone H3 (pHIS3) and Ki67, suggesting that proliferation depends on NFkB levels in TP53Mut LSCs. Furthermore, NFkB levels were highly dependent on pSTAT1 in TP53Mut LSCs. Next, we compared the transcriptional profiles of TP53Mut and TP53WT samples in the BEAT AML dataset and found that TP53Mut AML displayed significant enrichment of all the NFkB gene sets (n=19) in the molecular signatures database, providing more evidence for a unique reliance of TP53Mut human AML on NFkB signaling. Therefore, we used lentivirus transduced shRNA constructs to determine the influence of mutant p53 on signaling in human AML. We found that knockdown of TP53Mut reduced NFkB protein levels in Kasumi AML cells (which are TP53R248Q/ − ) but not in MOLM13 AML cells (which are TP53WT). RNA sequencing of transduced Kasumi cells showed that TP53Mut knockdown led to loss of NFkB and LSC self-renewal signatures. Knockdown of TP53Mut in primary human TP53Mut AMLs (S127F/- and S241Y/-) led to a similar loss of NFkB and LSC signatures in these primary samples as well. Colony formation in Kasumi cells and TP53Mut primary human AML were also abrogated after TP53Mut knockdown. These data show TP53Mut promotes NFkB activation in human AML and suggests that human TP53Mut LSCs may be dependent on NFkB for self-renewal. These data implicate NFkB as a possible therapeutic strategy to target TP53Mut human LSCs. Citation Format: Daniel Chang, Klara Noble-Orcutt, Marie Lue Antony, Yoonkyu Lee, Fiona He, Karen Sachs, Chad L Myers, Zohar Sachs. Bayesian networks modeling identifies a reliance of TP53 mutant AML on NF kappa B signaling [abstract]. In: Proceedings of the Blood Cancer Discovery Symposium; 2024 Mar 4-6; Boston, MA. Philadelphia (PA): AACR; Blood Cancer Discov 2024;5(2_Suppl):Abstract nr P34.