AUTHORS
Alexandria P. Eiken, PhD, Harry J. Ashbaugh, MS2, Wenjie Sun, PhD3, Emma Baird4, Danielle C. Kirkey, MD5, Joanna Yi, MD6, Soheil Meshinchi, MD, PhD7, Tim Triche Jr, PhD8, Derek H. Janssens, PhD9
AFFILIATIONS
1 Van Andel Research Institute, Grand Rapids, Michigan, United States
2 Van Andel Research Institute, Grand Rapids, Michigan, United States
3 Van Andel Research Institute, Grand Rapids, Michigan, United States
4 Van Andel Research Institute, Grand Rapids, Michigan, United States
5 Fred Hutchinson Cancer Research Center, Seattle, Washington, United States, Seattle Children's Hospital, Seattle, Washington, United States
6 Baylor College of Medicine and Texas Children’s Hospital, Houston, Texas, United States
7 Fred Hutchinson Cancer Research Center, Seattle, Washington, United States, Seattle Children's Hospital, Seattle, Washington, United States
8 Keck School of Medicine of USC - Department of Surgery, Los Angeles, California, United States, MSU College of Human Medicine, Lansing, Michigan, United States, Van Andel Research Institute, Grand Rapids, Michigan, United States
9 Van Andel Research Institute, Grand Rapids, Michigan, United States
ABSTRACT
Large-scale sequencing of hematologic malignancies has revealed recurrent mutations in chromatin-modifying genes, underscoring the central role of chromatin dysregulation in disease pathogenesis and tumor heterogeneity. Although the genetic basis of intratumoral heterogeneity has been well studied, the contribution of chromatin dysregulation to cellular diversity within tumors remains largely unknown. Moreover, protein-DNA interactions in leukemia have been examined primarily using bulk assays, which mask the epigenetic heterogeneity present within individual patient cells. Mapping cell-type-specific binding sites of chromatin-modifying proteins in heterogeneous leukemia samples requires a method capable of simultaneously resolving distinct cell populations and detecting sparse signal in single cells. To meet this need, we developed CoCUT&Tag, a novel single-cell multifactorial chromatin profiling method that leverages antigen repeat scaffolds and nanobodies to amplify the genomic signal from two DNA-bound proteins or chromatin modifications simultaneously within individual cells.
We applied this approach to characterize mixed-lineage leukemia-rearranged (MLLr) leukemia samples. Although these leukemias are associated with substantial clinical and biological heterogeneity, they have a relatively silent mutational landscape, suggesting that epigenetic mechanisms may serve as the primary source of this heterogeneity. MLLr mutations generate potent oncoproteins that enforce aberrant transcriptional programs to drive leukemia progression. To evaluate MLL oncoprotein binding using CoCUT&Tag, we used a cocktail of rabbit antibodies targeting the MLL oncoprotein cofactors Menin, Dot1L, and ENL, along with a mouse antibody against H3K27me3 to identify cell populations. Using this approach, we were able to resolve the expected cell-line-specific MLL oncoprotein binding sites from a mixture of MLLr cell lines and observed absent signal over the oncoprotein target sites in a negative control cell line lacking the MLLr mutation. Furthermore, the recovered signal was enriched at MLL oncoprotein target sites rather than loci occupied by wild-type MLL demonstrating the specificity of CoCUT&Tag for oncoprotein-driven chromatin interactions.
Next, we used CoCUT&Tag to profile a mixture of 11 MLLr leukemia patient samples with two healthy bone marrow controls. Unsupervised clustering of H3K27me3 profiles from ~230,000 single cells identified biologically distinct cellular populations, including patient-specific clusters of blasts, as well as a group of patient cells that clustered with healthy bone marrow. Within the mixed-phenotype acute leukemia (MPAL) and B-ALL patient samples that went on to lineage switch, we identified substantial intratumoral heterogeneity, including a recurrent module enriched for B-cell commitment associated genes (e.g., PAX5, EBF1, and IKZF1). In the B-ALL samples, the subpopulation with highest H3K27me3 repression of this module had the lowest MLL oncoprotein binding over these genes. In the post-lineage switched samples, the population of cells with elevated H3K27me3 repression and reduced oncoprotein signal over the B-cell gene module had significantly expanded. This indicates that MLL oncoprotein binding sites change within leukemic blasts in individual patients and that this plasticity in the MLL oncoprotein binding may underlie the apparent lineage plasticity associated with these leukemias. Moreover, MLL oncoprotein binding sites with higher oncoprotein signal have lower H3K27me3 signal, indicating that H3K27me3 and the MLL oncoprotein have an inverse binding relationship in general. Lastly, we detected a rare hematopoietic stem and progenitor cell (HSPC)-like population with MLL oncoprotein signal that may represent preleukemic cells harboring the MLLr mutation present in the patient samples, suggesting that preleukemic HSPCs may be more prevalent in driving disease resistance than initially thought.
Altogether, CoCUT&Tag enables high-sensitivity mapping of MLL oncoprotein binding in single cells, allowing us to simultaneously resolve cellular identity and intratumoral heterogeneity. This provides new precision for identifying MLL-driven biomarkers and developmental states vulnerable to epigenetic and/or lineage plasticity and therapeutic resistance. We anticipate that CoCUT&Tag will enable similar studies of oncoprotein-regulated heterogeneity across hematologic malignancies.
DISCLOSURES
Alexandria P. Eiken, PhD: No relevant conflicts of interest to declare.