Multi-Omic Approaches Reveal Leukemia Stem Cell Gene Regulatory Mechanisms

Background: Acute myeloid leukemia (AML) is initiated and maintained by self-renewing leukemic stem cells (LSCs) that can also differentiate into leukemic progenitor cells (LPCs) that comprise the bulk of leukemic blasts. Since LSCs represent chemoresistant cells and the reservoir of disease post-treatment, understanding the mechanisms that promote their function is critical to developing more effective therapies. We previously identified the cell surface antigen CD99 as overexpressed in LSCs in the vast majority of human AML. Differences in CD99 expression allow prospective separation of functional LSCs from LPCs, allowing us to conduct refined studies of the molecular features of LSCs.

Methods: To investigate the molecular differences between CD99hi LSCs, and CD99lo LPCs, we sorted the highest (top 20%), and lowest (bottom 20%) CD99 expressing blasts from 5 AML patient samples representing diverse genetic and cytogenetic subtypes of AML. We then generated molecular profiles for each of these samples by characterizing their chromatin accessibility landscapes (ATAC-seq), transcriptomes (total RNA-seq), translatomes (polysomal RNA-seq), and proteomes (LC/MS).

Results: LPCs showed slightly greater chromatin accessibility (174 open peaks, p<0.1) compared to LSCs (150 open peaks, p<0.1). Transcription factor motif enrichment analysis in proximal promoter regions did not identify significant differences between LSCs and LPCs, suggesting that changes in chromatin accessibility are not major drivers of the functional differences between LSCs and LPCs.

LPCs exhibited more differentially expressed genes (DEGs) than LSCs, based both on total RNA-seq data (610 vs 426 upregulated genes, log2FC >=1, p<0.05), and polysomal RNA-seq data (511 vs 265 upregulated genes, log2FC >=1, p<0.05). Pathway analysis of DEGs identified in total RNA-seq data showed that LSCs exhibit increased expression of cell cycle associated genes and MYC target genes. In contrast, LPCs showed significant enrichment of genes associated with inflammation and interferon alpha response. Intriguingly, the pattern of enrichment of these pathways differed when we evaluated the polysomal RNA-seq data, as LSCs exhibited enrichment of NOTCH signaling, interferon alpha response, and DNA repair genes, suggesting that LSCs and LPCs express unique translational programs due to post-transcriptional and translational mechanisms of gene regulation.

Comparison of transcriptional and proteomic profiles revealed minimal overlap, as only 4.6% of total RNA, and 5.9% of polysomal RNA changes were reflected in the proteome. In addition, although LPCs exhibited a larger number of genes with high translational efficiency than LSCs (218 genes in LPCs vs 172 genes in LSCs, log2FC >=1, p <0.05), LSCs showed more differentially expressed proteins (334 up-regulated proteins) than LPCs (282 up-regulated proteins) (p <0.1). Together, these data indicate that a significant number of differentially expressed proteins are due to post-translational mechanisms of regulation rather than transcriptional or translational mechanisms. Proteins upregulated in LSCs were associated with MTOR signaling, glycolysis, reactive oxygen species, and interferon alpha response, compatible with a model in which LSCs express genes associated with protein and carbohydrate metabolism to support cell cycling.

Conclusion: The ability to prospectively isolate LSCs and LPCs based on differential CD99 expression has allowed us to perform the first integrated genome-wide multi-omic investigation of human LSCs and LPCs. These studies have revealed the importance of post-transcriptional, translational, and post-translational regulatory mechanisms in shaping the molecular differences between these two populations. It is expected that these studies will lay the groundwork for the identification and selection of biomarkers and therapeutic targets for risk stratification, disease prevention, and cure.

No relevant conflicts of interest to declare.