Pursuing Leukemia Stem Cells Definition in Pediatric Acute Myeloid Leukemia

BACKGROUND:

Standard chemotherapy targets actively proliferating cells and induces complete remission in most of children with acute myeloid leukemia (AML); however, relapse still represents the main cause of treatment failure. Leukemia stem cells (LSCs) have been proposed as the therapy-resistant reservoir of cells responsible for disease recurrence; therefore, dissecting peculiar LSCs biological properties and mechanisms of chemoresistance is key to identify and validate more effective treatment options. However, the gold standard CD34+/CD38- gating strategy to identify LSCs is nowadays debated, and no distinctive marker has been detected yet.

METHODS AND RESULTS:

By cell sorting, we isolated from primary pediatric AML bone marrow (BM) samples at diagnosis the 20% of blasts with the lowest (ROS low, RL) and the highest (ROS high, RH) levels of reactive oxygen species (ROS) by CellROX probe staining and studied peculiar features of the two cell fractions. We first tested key stemness features and demonstrated that in contrast to cells in RH fraction where colony forming capacity was lost at 2^nd passage, RL cells displayed a significantly higher clonogenic capacity, with continued colony growth thru 3 consecutive passages (p<0.05). Moreover, RL cells were more quiescent, predominantly in G0/G1 cell cycle phase (85% vs 52% in RH, p<0.0001) and with lower proliferation capacity by CFSE staining (at day 3, 66% undivided cells vs 20% in RH, p<0.05). Most significantly, transplantation of RL vs RH cells from the same donor source into NSG models provided significantly different engraftment dynamics. Xenografts transplanted with RL cells demonstrated rapid engraftment and supported serial transplantation into secondary and tertiary mice (at day 155 and 114 respectively). In contrast, mice transplanted with RH cells had a more prolonged engraftment (at day 240 in 2^nd and 258 in 3^rd recipients, p<0.001). We used our three-dimensional (3D) model of BM niche to further study the dynamics of RL vs RH engraftment. RL cells seeded in the 3D model of BM niche implanted in NSG mice for 3 months were found more distant from blood vessels than RH by histological analyses (p<0.0001), supporting their preferential survival in the endosteal niche. Additionally, a 72-hour treatment with the chemotherapy agent cytarabine in 3D showed that RL cells were insensitive to this treatment, supporting the RL fraction being chemo-resistant. All RL and RH cells underwent transcriptome sequencing to interrogate transcriptome make-up of the two fractions. Gene-set enrichment analysis on RNAseq data performed on RL and RH fractions provided significant insight into the underlying biologic distinction of the two fractions and revealed differences in mitochondrial processes. Then, by using differentially expressed genes identified above, we generated a specific RL gene signature and projected it on available single cell RNAseq (scRNAseq) (n=28). Results showed that RL cells clustered like HSCs in FLT3-ITD and NUP98-rearranged AML, whereas in KMT2A-r leukemias RL cells clustered on GMPs and monocytes populations, suggesting that RL cells represented the stem cell fraction of each AML irrespective of their differentiation status. Furthermore, to support RL cells initiating leukemia, we identified pseudotime developmental trajectories based on transcriptional signatures starting from RL cells and giving rise to all the hematopoietic cells composing the AML bulk.

Then, to deepen into mitochondrial features, we measured lower levels of mitochondrial membrane potential (TMRE, p<0.0001), mitochondrial ROS (MitoSOX, p<0.01) and mitochondrial mass (MitoTracker, p<0.05) in RL cells with respect to RH, together with a reduced ATP production (p<0.0001). Looking at cell metabolism by applying SCENITH technology, we showed a trend towards a higher mitochondrial dependence and a lower glycolytic capacity in RL cells. This result was corroborated by treating 3D scaffolds seeded with RL or RH cells with the complex I inhibitor IACS-010759, which was only effective in reducing RL viability (p<0.05).

CONCLUSIONS:

In conclusion, these results support RL cell fraction being enriched for LSCs. This novel strategy of RL cell recognition allows to identify a LSCs population that needs to be targeted to reduce relapse events. Novel mitochondrial vulnerabilities and nutrient dependences are under evaluation to improve AML eradication.