Lack of progress in curing AML is likely, in part, to be due to the genetic and functional heterogeneity of AML. ~13 Tier 1 mutations occur per patient sample arrayed in 2-5 clones (CGARN, N Engl J Med, 2013). Not all AML cell populations may be equally chemosensitive; for example, leukemia-propagating leukemic stem cells (LSC) are more chemoresistant. (Ishikawa et al., Nat Biotechnol, 2007). Additionally, the impact of genetic heterogeneity on LSC function is unclear.

In ~ 70% of primary human AML with >2% CD34+ cells, LSCs exist within both CD34+CD38- and CD34+CD38+ compartments (Taussig et al., Blood, 2008), and have progenitor-like transcriptional programmes (Goardon et al., Cancer Cell, 2011). ~30% of AML with <2% CD34+ cells (CD34- AML) are genetically distinct, enriched for mutations in NPM1 and co-associated mutations (FLT3, IDH1/2, TET2, and DNMT3A). Here, LSC activity has been detected in both CD34+ and CD34- compartments (Taussig et al., Blood, 2010, Martelli et al., Blood, 2010, Sarry et al., J Clin Invest, 2011). Questions remain about CD34- AML LSC populations: (i) What is the relationship between CD34- and CD34+ LSCs? (ii) What are the nearest counterpart normal haemopoietic cells to LSCs at a global transcriptional level? (iii) What is the impact of genetic heterogeneity on LSC function? Do all clones have equal LSC potential?

Of a sequential cohort of 49 CD34- samples, 55% of 38 samples karyotyped had normal karyotype. 29/49 (59%) had mutated NPM1. Co-occurring mutations were FLT3 (54%), IDH1/2 (54%) and DNMT3A (26%). 11/28 samples with sufficient cells engrafted AML confirmed on mutation analysis. In 8/11 samples (7 were NPM1-mutated) sufficient available cells were available for detailed studies. LSC populations in serial transplant assays were present in both minor CD34+ and CD34-; CD34- populations were CD117+ and CD244+ or CD244-. Limit dilution analysis showed similar LSC frequencies in CD34+ and CD34- fractions from the same patient. Unexpectedly, there was no hierarchy with respect to CD34 expression and CD34 expression did not mark functionally distinct LSC populations. RNA-sequencing of 5 CD34+ and 14 CD34- LSC populations from 8 patient samples showed only 42 differentially protein coding genes out of 15539 expressed genes.

We used ANOVA analysis to identify 300 top-ranking differentially expressed genes between normal stem/progenitor and myeloid and erythroid precursor populations. Using this signature, principal component analysis showed CD34- AML LSCs (CD34+ and CD34-) are closest to CD34-CD117+CD244+ populations that are promyelocytes have no D14 progenitor function in CFC assays and express late granulocytic macrophage (GM) genes. CD244 separates normal CD34-CD117+ populations into GM (CD244+) from erythroid (CD244-) precursors so allowing greater precision in mapping CD34-AML LSC to normal GM counterparts. CD34-AML LCS were not only enriched for a GM precursor signature, but also for a transcriptional signature seen in normal HSC. CD34- AML LSC expressed 63/100 transcription factor (TF) genes expressed in HSC including HOX and HOX co-factors. Interestingly, LSCs in CD34- AML had a distinct RNA-Seq profile from CD34+ progenitor-like LSCs populations.

To explore genetic and functional heterogeneity of LSC populations, bulk and single cell genotyping of LSC populations revealed: (i) branching subclonal structures; with up to 5 genetically distinct LSC clones/patient; (ii) intermediate genotypes where the order of mutation acquisition was identified; (iii) some but not all patient LSC clones could be propagated in mice suggesting that current immunodeficient murine strains do not accurately model human AML. Thus, studies of LSC function have to combine both studies in mice and of primary human samples.

In summary, within CD34- AML there are multiple, non-hierarchically arranged LSC populations with transcriptional programmes most closely related to normal CD34- GM precursors. Unlike these normal mature cells, LSCs also express HSC transcriptional signatures. Functional and genetic analysis of single cells and populations from patient LSC, non-LSC compartments and of xenografts reveals clonal structure, order of acquisition of mutations, how subclones are distributed in different immunophenotypic populations, some with different functional properties and, differences in subclonal representation between patients and xenografts.

Disclosures

No relevant conflicts of interest to declare.

Author notes

*

Asterisk with author names denotes non-ASH members.

Sign in via your Institution