NEW YORK (GenomeWeb) – A team from Massachusetts General Hospital, Harvard Medical School, and elsewhere has come up with an atlas of acute myeloid leukemia (AML) cell types using single-cell genomic and transcriptomic analyses — a collection that is revealing expression features in AML cells with specific mutations or sub-clone sources.
"It is now clearer that genetic mutations can drive different cancer cell types that need to be specifically targeted with precision therapies, so our findings can guide personalized therapies to eradicate AML cells," co-first author Peter van Galen, a pathology and cancer researcher affiliated with Massachusetts General Hospital, Harvard Medical School, and the Broad Institute, said in a statement.
The researchers used a single-cell RNA sequencing genotyping approach — along with machine learning-based cell type classification — to assess tens of thousands of cells from individuals with or without AML. As they reported online today in Cell, the results revealed clusters of normal and cancerous cell types, including differentiated AML cell clusters exhibiting T cell suppression features in the expression data and in their subsequent immunohistochemistry experiments.
"Immune therapies that harness T cells have been less successful in AML, and our findings suggest that their efficacy could be improved by overcoming the inhibitory signals from differentiated tumor cells," Galen said, noting that a "better understanding of the abnormal regulation of developmental genes in leukemia stem cells may provide fundamental insights into the origins of this disease."
For their analysis, he and his colleagues used a modified, nanowell-based approach to do single-cell RNA-seq on 38,410 individual cells, representing 35 cryopreserved bone marrow aspirate samples from 16 AML patients as well as frozen or progenitor-enriched bone marrow samples from five healthy individuals.
In parallel, they produced targeted, single-cell genotyping on a subset of the same cells, using a streptavidin-based bead capture, PCR amplification, and Illumina short read or Oxford Nanopore long-read sequencing to identify AML-related mutations, insertions, fusions, and cancer sub-clones.
The team analyzed the resulting data — single-cell transcriptomes for 30,712 AML cells and 7,698 healthy bone marrow cells, along with genotypes for 3,799 of the cells — with a machine learning classifier that highlighted half a dozen malignant AML cell types present to differing degrees in tumors with distinct genetic alterations.
Those cell types "project along the [hematopoetic stem cell]-to-myeloid differentiation axis," the authors explained, and provided a resource "to relate developmental hierarchies to genotypes, to evaluate properties and prognostic significance of primitive AML cells, and to identify differentiated AML cells with immunomodulatory properties."
At one end of the spectrum, for example, the researchers saw AML cells with features similar to hematopoetic stem cell that expressed genes associated with "stemness" and "myeloid priming genes." But other cells clustered into groups resembling granulocyte-macrophage progenitor-like cells and more differentiated "monocyte-like" cells that appeared to restrain the immune system and staunch T immune cell activity.
From these and other findings, the authors reported that the "[p]rimitive AML cells exhibited dysregulated transcriptional programs with co-expression of stemness and myeloid priming genes and had prognostic significance," while "[d]ifferentiated monocyte-like AML cells expressed diverse immunomodulatory genes and suppressed T cell activity in vitro."