Leukemia Stem Cells in Acute Lymphoblastic Leukemia: Unveiling Hierarchical Structure at Single Cell Resolution

Leukemia stem cells (LSCs) are the root of leukemia, and are responsible for drug resistance and disease relapse. However, LSCs have not yet been identified for acute lymphoblastic leukemia (ALL). Among many challenges, lack of phenotypic markers is one of the major problems in identifying ALL LSCs. In this study, we demonstrated a novel method to isolate LSCs from both T- and B- cell ALLs and further characterized their transcriptome profile at the single cell level.

We have recently identified a novel method to isolate ALL LSCs based on cellular metabolic activities. We demonstrated that these isolated LSCs had in vivo leukemia-initiating capability (LIC). We have developed a series of primary ALL xenograft mouse models using patient samples and NOD/SCID/IL2Rg-/- (NSG) mice. Leukemia cells harvested from several generations of these mice were used in this study. We isolated LSCs and non-LSCs from 4 different B-cell type ALL samples and transplanted them separately into healthy NSG mice. Cell numbers used varied between 5, 10, and 50,000 per mouse, and the number of the animals varied between three and eight per group. All the animals transplanted with LSCs developed leukemia between 5-14 weeks, whereas those transplanted with non-LSCs did not develop the disease within the same timeframe or by the end of the study, which was more than 4 months after leukemia development in the LSC group.

In order to characterize and identify potential therapeutic targets in the LSCs, we investigated the transcriptome profile of these cells. First, we performed genomewide microarray gene expression profiling of RNA isolated from the LSCs and non-LSCs using 4 ALL cell lines (Reh, JM1, Jurkat, and Molt4). There were 173 genes which showed at least 2-fold difference in gene expression between the LSCs and non-LSCs. Using a panel of primer sets for the 100 genes exhibiting the highest difference in expression, we performed qRT-PCR for these genes in the isolated LSCs and non-LSCs from 11 primary ALL samples (10 B-cell and 1 T-cell type) transplanted and harvested from our NSG xenograft mouse models at different generations. There was a distinct difference in the transcriptome profile between the LSCs and non-LSCs in these primary ALL samples. Overall gene expression of 93 LSC signature genes was much lower in the LSCs than in the non-LSCs.

Recent advances in microfluidic technologies allowed us to investigate cells at single cell resolution. Growing evidence suggests that cancer stem cells consist of heterogeneous cell populations (subclones). Therefore, we further investigated whether these isolated LSCs have subclones using the Fluidigm C1 and Biomark system. Preliminary results using a primary ALL sample harvested from our xenograft mouse model, indicate that there are at least two distinct subclones in the LSCs based on principal component analysis of the single cell data.

In summary, we 1) developed a novel method to isolate ALL LSCs which have in vivo LIC, 2) demonstrated that isolated LSCs have a distinct transcriptome profile, and 3) discovered that the LSCs seem to consist of subclones. Currently we are in the process of performing detailed comprehensive transcriptome analyses and additional single cell transcriptome assays.