Leukaemia cell of origin identified by chromatin landscape of bulk tumour cells

Abstract

The precise identity of a tumour's cell of origin can influence disease prognosis and outcome. Methods to reliably define tumour cell of origin from primary, bulk tumour cell samples has been a challenge. Here we use a well-defined model of MLL-rearranged acute myeloid leukaemia (AML) to demonstrate that transforming haematopoietic stem cells (HSCs) and multipotent progenitors results in more aggressive AML than transforming committed progenitor cells. Transcriptome profiling reveals a gene expression signature broadly distinguishing stem cell-derived versus progenitor cell-derived AML, including genes involved in immune escape, extravasation and small GTPase signal transduction. However, whole-genome profiling of open chromatin reveals precise and robust biomarkers reflecting each cell of origin tested, from bulk AML tumour cell sampling. We find that bulk AML tumour cells exhibit distinct open chromatin loci that reflect the transformed cell of origin and suggest that open chromatin patterns may be leveraged as prognostic signatures in human AML.

Figure 1. Cell of origin determines potency of in vitro-derived MA9-driven AML.

(a) Schematic diagram of the primitive haematopoietic hierarchy. Colour code: purple for long-term HSCs (LT-HSC), blue for short-term (ST)-HSC, orange for multipotent progenitors (MPP), red for common myeloid progenitors (CMP), black for granulocyte macrophage progenitors (GMP). (b) Schematic diagram of experimental design to test in vitro transformation of distinct cells of origin by MA9. (c) Overall survival of mice transplanted with 100 K MA9-transformed cells from distinct cells of origin (n=5 per group). Technical replicates of 5 mice were each transplanted with 100 K cells from one set of cell lines. Trend and significance were replicated once using a set of independently derived cell lines. Log-rank test; P<0.0001. (d) Correlation between mean fluorescence intensity (MFI) of GFP in leukaemia cell lines and median survival (days) of mice transplanted with these lines (n=5 per group). Centre values indicate mean. Error bars indicate s.e.m. Pearson's r=−0.2653, not significant. (e) Limiting dilution analysis of leukaemia cell lines from distinct cells of origin (n=5 per group, per cell dose). Technical replicates of five mice were each transplanted with 100 K, 10 K or 1 K cells from one set of cell lines. Trend and significance were replicated once using a set of independently derived cell lines. Confidence intervals (1 per initiating cell frequency) were calculated with ELDA software. Pearson's χ²-test; P=0.00145.

Figure 2. Cell of origin alters tumour aggressiveness of in vivo-derived AML.

(a) Schematic diagram of experimental design to test in vivo transformation of distinct cells of origin by MA9. (b) Overall survival of mice transplanted with 500 MA9-transduced cells from distinct cells of origin (STHSC:MA9; n=5, MPP:MA9; n=11, CMP:MA9; n=15, GMP:MA9; n=14). Data were collected over three biological replicate experiments. Log-rank test; P=0.0004. (c) Peripheral blood leukocyte count in terminal mice transplanted with MA9-transduced cells (STHSC:MA9; n=5, MPP:MA9; n=11, CMP:MA9; n=15, GMP:MA9; n=14). Data were collected over three biological replicate experiments. Centre bars indicate mean. Error bars indicate s.e.m. Kruskal-Wallis test; P=0.5622. (d) Average frequency of LSCs (L-GMP20) in the bone marrow of terminal mice (STHSC:MA9; n=3, MPP:MA9; n=6, CMP:MA9; n=13, GMP:MA9; n=9). Data were collected over three biological replicate experiments. Centre bars indicate mean. Error bars indicate s.e.m. Kruskal–Wallis test; P=0.7031. (e) Relative expression of MA9 assessed by real-time PCR in splenic leukaemia cells in terminal mice (STHSC:MA9; n=3, MPP:MA9; n=6, CMP:MA9; n=4, GMP:MA9; n=6). Data were collected over three biological replicate experiments. Centre bars indicate mean. Error bars indicate s.e.m. Kruskal–Wallis test; P=0.0055. Dunn's multiple comparisons test; *P<0.05.

Figure 3. ST-HSC-derived AML has a unique and prognostically relevant gene expression signature.

(a) Principal component analysis of normal and MA9-transformed haematopoietic stem and progenitor cells (STHSC; n=5, MPP; n=6, CMP; n=5, GMP; n=5, MEP; n=4, STHSC:MA9; n=3, MPP:MA9; n=6, CMP:MA9; n=4, GMP:MA9; n=6). Each sample was obtained from an individual mouse. Data were collected from two biological replicate experiments. Clustering segregates STHSC:MA9 leukemias (blue cloud) from other leukaemias (red cloud). (b) K-medioids clustering (K=5) of 133 differentially expressed genes in leukaemias based on cell of origin (foldchange >2, FDR<0.05) (STHSC:MA9; n=3, MPP:MA9; n=6, CMP:MA9; n=4, GMP:MA9; n=6). Representative biological replicates are shown. (c) Overall survival of human AML patients based on expression of CBFA2T3 (n=200). Log-rank test; P=0.0113. (d) Number of somatic nucleotide variants (SNVs) per tumour derived from distinct cells of origin (STHSC:MA9; n=3, MPP:MA9; n=6, CMP:MA9; n=4, GMP:MA9; n=6). Each sample was obtained from an individual mouse. Data were collected from two biological replicate experiments. Kruskal–Wallis test; P=0.0019.

Figure 4. Global chromatin remodelling during MA9 transformation.

(a) Number of ATAC-seq peaks distributed across annotated features of the genome (n=4 biological replicates from individual mice). Centre bars indicate mean. (b) Heatmap of Pearson correlation analysis of open chromatin regions in in vivo-derived primary bulk leukaemias from distinct cells of origin and their normal cellular counterparts (LSK; n=3, CMP; n=3, GMP; n=2, STHSC:MA9; n=2, MPP:MA9; n=2, CMP:MA9; n=2, GMP:MA9; n=2). LSK; Lin⁻ Sca-1⁺ c-Kit⁺, includes ST-HSC and MPP cells. Each sample was obtained from an individual mouse. Data were collected from two biological replicate experiments. (c) Normalized ATAC-seq profiles of in vivo-derived primary bulk leukaemias from distinct cells of origin and their normal cellular counterparts, showing gain and loss of enhancer elements (dotted lines) around Mina and Crybg3 in a 150-kb region. Single samples are shown. Trends were replicated once with independent biological replicates. Shown at bottom are H3K4me1 and H3K27ac ChIP-seq peak regions in normal murine bone marrow (BM) and H3K27ac peak regions in a MA9/Nras^G12D murine AML cell line (RN2). RNA-seq FPKM of Mina and Crybg3 are shown (STHSC; n=5, MPP; n=6, CMP; n=5, GMP; n=5, STHSC:MA9; n=3, MPP:MA9; n=6, CMP:MA9; n=4, GMP:MA9; n=6). Centre bars indicate mean. Error bars indicate s.e.m. Kruskal–Wallis test; P=0.0012 and P<0.0001, respectively. Dunn's multiple comparisons test; *P<0.05, **P<0.01.

Figure 5. Unique open chromatin regions in bulk leukaemia cells identify AML cell of origin.

(a) Number of intersecting ATAC-seq peaks based on leukaemia cell of origin (STHSC:MA9; n=2, MPP:MA9; n=2, CMP:MA9; n=2, GMP:MA9; n=2). Calculated numbers exclude peaks found in non-selected leukaemia subtypes. Each sample was obtained from an individual mouse. Data were collected from two biological replicate experiments. (b) Normalized ATAC-seq profiles of in vivo-derived primary bulk leukaemias from distinct cells of origin (biological replicates are shown), highlighting unique open chromatin regions around Alcam, Hoxa5, Lrrc4c, Armc1, Suv39h2, Rac2, Bambi and Tap2 in 6 kb regions. Shown at bottom are H3K4me1 and H3K27ac ChIP-seq peak regions in normal murine bone marrow (BM) and H3K27ac peak regions in a MA9/Nras^G12D murine AML cell line (RN2). (c) Percentage of unique ATAC-seq peaks in each primary bulk leukaemia retained or gained from their respective normal cell of origin (STHSC:MA9; n=2, MPP:MA9; n=2, CMP:MA9; n=2, GMP:MA9; n=2). Each sample was obtained from an individual mouse. Data were collected from two biological replicate experiments. (d) Overall survival of human AML patients based on DNA methylation status of the CpG probe shown (n=200). Log-rank test; P=0.0170 and P=0.00137, respectively.