Single Cell Clones Derived from a Patient's AML Xenograft Display Genetic and Functional Heterogeneity

Introduction:

Acute myeloid leukemia (AML) shows substantial genetic and epigenetic heterogeneity, even within an individual patient. Due to treatment resistance and ability to induce relapse, adverse subclones present a major clinical challenge in determining the patient's prognosis.

Here, we aimed at characterizing the genetic and functional heterogeneity within a single AML patient, and at identifying adverse subclones that result in therapy failure or give rise to relapse.

Methods:

Leukemic cells from an AML patient at first and second relapse were transplanted into immuno-compromised mice to generate patient-derived xenografts (PDX). PDX AML cells allowed serial transplantation and genetic engineering by lentiviruses. To distinguish single cells and generate PDX AML clones derived from a single cell (single cell clones, SCC), cells were transduced with a genetic barcode and transplanted into recipient mice near leukemia initiating cell frequency. Resulting SCC were genetically marked to express recombinant fluorochromes to enable flow cytometry analysis. All SCC were characterized for known subclonal mutations of the AML patient by targeted sequencing. Additionally, transcriptome and methylome analysis were performed by SCRB-seq and methylation array, respectively.

Results:

We successfully generated thirteen serially transplantable PDX SCC from a single AML patient, expressing combinations of up to four fluorochromes to enable competitive in vitro and in vivo experiments.

In targeted sequencing, we found that SCC originated from at least four genetically distinct AML subclones and were distinguished by mutations in KRAS (4/13), NRAS (5/13), EZH2 (2/13) or EZH2 and NRAS (2/13). While the NRAS mutation was detected in a minority of bulk cells over serial passages (<10%) from both the first and second relapse, 50% of SCC carried the NRAS mutation. This indicates that NRAS mutated AML cells have an increased stem cell capacity upon transplantation of low cell numbers.

Transcriptome analysis revealed 442 genes as differentially expressed between SCC and up to four biological replicates after adjustment for multiple hypothesis testing. Unsupervised clustering demonstrated a strong correlation of gene expression profiles with the respective genotype.

In competitive in vivo experiments, homing capability was comparable between the four genetically distinct subclones. However, 2/2 EZH2-mutated SCC overgrew all other clones showing a clear growth advantage within two weeks of in vivo growth. The EZH2-mutated SCC (2/2) were resistant towards in vivo treatment with Cytarabine, whereas the KRAS (4/4), NRAS (5/5) mutated as well as the EZH2 and NRAS double-mutated SCC (2/2) responded to treatment.

Conclusion:

Taken together, we experimentally prove the existence of genetically and functionally diverse subclones within an individual AML sample. Our approach allows not only genetic, but also functional in vitro and in vivo characterization of adverse subclones. Our approach can be used to identify novel therapeutic approaches in order to specifically target the most adverse cells within patients' AML sample.