Transcriptome Analysis of Severe Congenital Neutropenia Associated Leukemia with Different Types of RUNX1 Mutations in an iPSC-Based Model

Co-acquisition of CSF3R and RUNX1 mutations is the most frequent combination of acquired genetic abnormalities found in AML and MDS cases associated with severe congenital neutropenia (CN). Using the iPSC-based model for CN-associated leukemia, we compared the outcomes of RUNX1 missense mutations and RUNX1 haploinsufficiency by means of hematopoietic differentiation and transcriptome analysis.

We generated iPSC lines from 2 CN/AML patients harboring ELANE mutations p.C151Y or p.G214R (CN1 and CN2, accordingly). CN/AML iPSC clones of CN patient 1 were generated from leukaemia cells with CSF3R p.Q741X, missense RUNX1 p.R139G mutation, and trisomy 21 (CN/AML1.1 and 1.2). CN/AML iPSC clones of CN patient 2 (CN/AML2) have CSF3R p.Q743X mutation and truncated RUNX1 mutation p.E175fs introduced by CRISPR/Cas9 gene-editing.

The iPSC clones from CN (n = 4) and CN/AML stage (n = 6) were subjected for embryoid body (EB)-based differentiation into hematopoietic stem and mature myeloid cells. We observed dramatically increased proliferation of CN/AML iPSC-derived hematopoietic stem cell progenitors (HSPCs) collected at day 14 of EB differentiation and further cultured on the FLT3-L-secreting feeder cells, as compared to CN - iPSC derived HSPCs. Moreover, myeloid differentiation was severely diminished in CN/AML-iPSC group compared with CN iPSC group. To elucidate the mechanism of leukemogenic transformation in CN, we performed RNA-sequencing of CD34⁺ hematopoietic stem and progenitor cells (HSPCs) derived from CN and CN/AML iPSC clones.

Differential gene expression analyses of RNA-sequencing data of iPSC-derived HSPCs using DESeq2 R package identified 132 up- and 570 down-regulated (log2FC > 1 or < -1, adj. P-value < 0.05), as well as 579 up- and 1422 down-regulated (log2FC > 1 or < -1, adj. P-value < 0.05) genes between CN/AML and CN stages for CN patient 1 and CN patient 2, respectively. There were 189 differentially expressed genes shared between CN/AML1 vs CN1 and CN/AML2 vs CN2 comparisons.

Among the top significantly upregulated pathways in the gene set enrichment analysis (GSEA) of CN/AML1 iPSCs with missense RUNX1 mutation were "Hallmark E2F Targets", "Hallmark Oxidative Phosphorylation", "Hallmark MYC Targets V1", whereas genes corresponding to the gene set "Platelet Specific Genes" were significantly downregulated. In CN/AML2 iPSCs with truncated RUNX1 mutation, we detected enrichment of the gene sets "Hallmark G2M Checkpoint", "Hallmark E2F Targets" as well as "Hallmark TGFb signaling". In contrast, the gene set "GO:Structural Constituent of Ribosomes" was enriched in CN2 derived HSPCs.

Intriguingly, transcription factor enrichment analysis (TFEA) using the lists of differentially expressed genes (log2FC > 1 or < -1, adj. P-value < 0.05) demonstrated similarities between significantly enriched TFs motifs sets (P-value < 0.05) from both CN/AML vs CN comparisons. Among the transcription factor binding motifs significantly enriched in CN/AML cells (P-value < 0.05) with missense RUNX1 mutation and RUNX1 haploinsufficiency, we found motifs for GATA1, GATA2, and RUNX1, as well as for AML-associated factors such as SUZ12 and EZH2. The vast majority of enriched kinases (HIPK2, MAPK1/3/14, CSNK2A1, ERK1, AKT1, etc.) at the selected threshold (P-value <10^-8) were shared in kinase enrichment analysis (KEA) between both CN/AML vs CN comparisons.

Taken together, we found a dramatic difference in the gene expression signature between CN- and CN/AML-iPSCs derived HSPCs. The vast majority of significantly enriched transcription factors and kinases were overlapped in HSPCs carrying different types of RUNX1 mutations. These data will help to better understand the mechanistic outcomes of different types of the endogenously expressed mutated RUNX1 proteins in leukemogenesis. This info is crucial for the identification of specific drug targets for CN/AML and de novo AML with mutant RUNX1.