Search

Cancer cells frequently exhibit chromosomal abnormalities. Specific cytogenetic aberrations often are predictors of outcome, especially in hematologic neoplasms, such as monosomy 7 in myeloid malignancies. The functional consequences of aneuploidy at the cellular level are difficult to assess because of a lack of convenient markers to distinguish abnormal from diploid cells. We performed single-cell RNA sequencing (scRNA-seq) to study hematopoietic stem and progenitor cells from the bone marrow of 4 healthy donors and 5 patients with bone marrow failure and chromosome gain or loss. In total, transcriptome sequences were obtained from 391 control cells and 588 cells from patients. We characterized normal hematopoiesis as binary differentiation from stem cells to erythroid and myeloid-lymphoid pathways. Aneuploid cells were distinguished from diploid cells in patient samples by computational analyses of read fractions and gene expression of individual chromosomes. We confirmed assignment of aneuploidy to individual cells quantitatively, by copy-number variation, and qualitatively, by loss of heterozygosity. When we projected patients' single cells onto the map of normal hematopoiesis, diverse patterns were observed, broadly reflecting clinical phenotypes. Patients' monosomy 7 cells showed downregulation of genes involved in immune response and DNA damage checkpoint and apoptosis pathways, which may contribute to the clonal expansion of monosomy 7 cells with accumulated gene mutations. scRNA-seq is a powerful technique through which to infer the functional consequences of chromosome gain and loss and explore gene targets for directed therapy.

Hematopoiesis, especially the early events of blood cell formation, has been mainly studied in bulk populations of cells and using progenitor colony formation assays; the familiar hierarchy of cell lineage differentiation and maturation, and associated regulatory factors have been inferred from these methods. However, these techniques often require extensive manipulation of cells, the exposure of cells to unphysiological conditions, aggregation of heterogeneous populations, and prior assumptions concerning cell function and gene expression. New single cell methodology avoids many of these potential experimental deficiencies. Here we have applied single-cell RNA-sequencing(scRNA-seq)to fresh human bone marrow CD34+cells: we profiled 391 single hematopoietic stem/progenitor cells (HSPCs) from four healthy donors by deep sequencing of individual cell transcriptomes. An average of 4560 protein-coding genes were detected per cell. Cells clustered into six distinct groups, which could be assigned to known HSPC subpopulations (Fig 1A), based on expression of lineage-specific genes. Lin-CD34+CD38+cells emerged as locally clustered cell populations (Clusters 2-6, including MEP, GMP, ETP and ProB), while Lin-CD34+CD38-cells formed a single cluster (HSC/MLP). Reconstruction of differentiation trajectories by transcription in single cells revealed four committed lineages derived from stem cell compartment. The earliest fate split separates MEPs from MLPs, which partition further into lymphoid, and granulocyte-monocyte progenitors (Fig 1B). The overall pattern differs from the classical hematopoietic model describing a single binary split between myeloid and lymphoid differentiation immediately downstream of multipotent cells. However, our data align well to recently published scRNA-seq data showing sequential commitment of stem cells to the lymphoid, erythroid/megakaryocytic, and finally myeloid lineages (Setty M, Nat Biotechnol2019; Pellin D, Nat Commun2019). We further examined trends in gene expression in each of the branches and found dynamic expression changes underlying cell fate during early lineage differentiation (Fig 1C). As confirmation, PCA plot of published single-cell assay for transposase-accessible chromatin (scATAC-seq) shows similar differentiation pattern. After projecting scATAC-seq data to our transcriptomic clusters' specific genes, MEP-dependent and myeloid/lymphoid-dependent genes were located on opposing sides of the PC1 with same direction (Fig 1D), indicating transcriptome and epigenome work on differentiation in concerted effort. scRNA-seq provides opportunities for discovery and characterization at the molecular levels of early HSC differentiation and developmental intermediates, retrospectively, without the need to isolate purified populations. However, information inferred from scRNA-seq may be obscured due to missing reads and limited cell numbers. More cells would provide greater detail and higher resolution mapping.Given the low frequency of megakaryocyte progenitors within the CD34+cells as well as the neglected Lin-CD34-BM compartment, we could not fully resolve the separation and maturation of all lineages. Nonetheless, we found good coverage of cell types and a similar HSPC Atlas as other published studies (Velten L, Nat Cell Biol2017; Pellin D, Nat Commun2019)despite our limited numbers of starting cells. Our data accurately reflect the pattern of normal hematopoiesis, which may help to revise and refine characterization of hematopoiesis and provide a general reference framework to investigate the complexities of blood cell production at single-cell resolution - especially when cell numbers are limited, as from patient samples and in marrow failure syndromes. Fig. 1scRNA-seq of human hematopoietic stem and progenitor cells. (A) Unsupervised hierarchical clustering of gene expression data for all cells. C1, HSC/MLP; C2, MEP; C3, GMP; C4, ProB; C5-C6, ETP. (B)Visualization of the HSPC continuum. Each ball represents one cell.(C) Large-scale shifts in gene expression during development of hematopoietic cells.Bars on top indicate locations of individual cells, colored by stages of development, along this developmental trajectory. (D) Projections of five transcriptomic gene modules onto PCA of scATAC-seq data (Buenrostro JD,Cell 2018). Each dot represents a transcriptional factor. Figure 1 Disclosures No relevant conflicts of interest to declare.

Monosomy 7 is a frequent cytogenetic abnormality in hematopoietic malignancies and a general indicator of poor prognosis. Due to lack of distinct cell surface markers between monosomy 7 cells and normal cells, it is not feasible to physically separate aneuploid from diploid cells. We performed single-cell RNA-seq (scRNA-seq), which allows the entire transcriptome of large numbers of single cells to be assayed in an unbiased way, to investigate hematopoietic differentiation of normal and aneuploid human hematopoietic cells. Bone marrow samples were collected from four patients (P1-P4) with myelodysplastic syndrome, and four healthy volunteers. Conventional cytogenetics showed -7/7q- in bone marrow cells from P1, P3 and P4, and dup(1)(q21q32) in cells from P2; retrospectively, P2 was found positive for monosomy 7 as well as trisomy 8 by fluorescence in situ hybridization. Fresh CD34+CD38- and CD34+CD38+cells were sorted by flow cytometry and then subjected to Fluidigm C1 Single-Cell Auto Prep System for scRNA-seq. After excluding cells with low transcriptome coverage, 326 cells from P1 and P2 (analysis is in progress for P3 and P4), and 391 cells from healthy subjects were analyzed by comparison of transcriptomes from 17,071 genes. Nonlinear dimension reduction and visualization were achieved using t-distributed Stochastic Neighbor Embedding (tSNE). Cells from healthy controls clustered into seven subgroups based on their gene expression pattern, and each group could be associated with a previously reported hematopoietic cell type by known marker genes (Laurenti E, Nat Immunol, 2013). These cell types included hematopoietic stem cell (HSC), multilymphoid progenitor (MLP), granulocyte-monocyte progenitor (GMP), Pro-B cell (ProB), earliest thymic progenitor (ETP), and megakaryocytic-erythroid progenitor (MEP) (Fig 1a). Individual cells from healthy controls were ordered by Monocle software based on their expression profile similarity to uncover a differentiation hierarchy. A two-branch trajectory of development from HSC was revealed, with one branch progressing towards erythroid cell and the other to lymphoid/myeloid cells (Fig 1b). This pattern differs from the classic hematopoietic model, but is consistent with reports claiming existence of early-lymphoid-biased progenitors that retain myeloid but not erythroid potential (Doulatov S, Nat Immunol, 2010), and of dominance of multipotent and unipotent progenitors over scarce oligopotent progenitors in the adult marrow hematopoietic hierarchy (Notta F, Science, 2016). We compared single cells from patients and healthy controls for regional and chromosomal copy number differences in gene expression. We identified subclonal populations of cells from patients that showed decreased expression of chromosome 7 genes (60% in P1, and 55% in P2; Fig 1c and 1d), and increased expression of chromosome 8 (77% in P2) and chromosome 1 long arm genes (P2), at FDR=0.05 estimated with cells from health donors. Gene Ontology enrichment analysis using topGO indicated that cells with low global expression of chromosome 7 genes had dysregulated expression of immune related genes, including B cell receptor signaling pathway, T cell activation and differentiation, antigen receptor-mediated signaling pathway, as well as signal transduction and Fc-γ Receptor signaling pathway. ScRNA-seq analysis reveals a simple pattern of normal human hematopoietic development and the molecular signature of aneuploid cells from patients with developing "clonal evolution". This powerful method should improve characterization of functional changes in human cells with chromosome abnormalities. Figure 1 a. Single-cell gene expression patterns assigned single cells from healthy controls to seven clusters. 38N: CD34+CD38- population; 38P: CD34+CD38+ population. Different shapes represent cells from different subjects. b. Pseudo-time ordering of cells using Monocle reveals a two-branch stepwise development from stem cells to erythroid or lymphoid/myeloid cells. c. Heatmap of the copy-number variation (CNV) signal normalized against healthy controls shows CNV changes by chromosome (columns) for patients' individual cells (rows). d. Genome-wide gene expression binned per chromosome in single cells from P1, P2 and healthy controls. Chromosomal mapping reads values were median centered. Figure 1. a. Single-cell gene expression patterns assigned single cells from healthy controls to seven clusters. 38N: CD34+CD38- population; 38P: CD34+CD38+ population. Different shapes represent cells from different subjects. b. Pseudo-time ordering of cells using Monocle reveals a two-branch stepwise development from stem cells to erythroid or lymphoid/myeloid cells. c. Heatmap of the copy-number variation (CNV) signal normalized against healthy controls shows CNV changes by chromosome (columns) for patients' individual cells (rows). d. Genome-wide gene expression binned per chromosome in single cells from P1, P2 and healthy controls. Chromosomal mapping reads values were median centered. Disclosures Desierto: GSK/Novartis: Research Funding. Townsley:GSK/Novartis: Research Funding. Young:Novartis: Research Funding.

Immune checkpoint therapy, particularly inhibition of the PD-1 axis, has shown remarkable efficacy in multiple solid tumor types and some hematological malignancies. Clinical results using anti-PD1 immunotherapy in AML have not yet been reported. Immune checkpoint expression levels and magnitude and/or repertoire of tumor-infiltrating lymphocytes have all been proposed as potential predictive biomarkers of response to such targeted immunotherapy. We therefore examined the marrow infiltrating T-lymphocyte compartment of Relapsed/Refractory Acute Myeloid Leukemia (RR-AML) patients undergoing salvage chemotherapy and that of healthy donors (HD) to determine baseline data for PD-1 expression and T-lymphocyte clonality. Nine adult RR-AML patients recruited to the clinical trial NCT02527447 (Myeloid Malignancies Section, NHLBI, NIH) were analyzed. Cryopreserved bone marrow aspirate mononuclear cells (BMMCs) from these patients prior to salvage chemotherapy (Day 0) and HD (n=9) were assessed using two custom ten-color flow cytometry panels. Frequencies of naive (TN), central memory (TCM), effector memory (TEM), terminal effector (TEMRA), stem cell memory (TSCM), and activated T-lymphocyte subsets, and the expression of PD-1, TIM-3, CTLA-4, 4-1BB on CD8+ T-lymphocyte subpopulations were calculated. DNA extracted from bone marrow of 5 of the RR-AML patients and 5 HD was used for sequencing of the CDR3 region of TCRB gene to evaluate TCR repertoire (ImmunoSEQ, Adaptive Biotechnologies). Immunohistochemistry of pre-treatment bone marrow biopsy sections revealed 10-40% CD3+ cells, typically scattered with focal small aggregations. Flow cytometry of marrow aspirate demonstrated more activated CD8+ CD69+ T-lymphocytes in the marrow of RR-AML patients compared with HD (19.8% vs. 9.4%, p=0.031). Significantly more CD8+ TEMRA T-lymphocytes (60.8% vs. 45.2%, p=0.040) and a trend towards fewer CD8+ TEM T-lymphocytes (19.0% vs. 32.1%, p=0.062) were observed in RR-AML versus HD. An increased proportion of total CD8+ T-lymphocytes in RR-AML had PD-1 expression compared to HD (22.0% vs. 9.2%, p=0.008). This pattern held true for every CD8+ sub-population in AML versus HD: TN (14.0% vs. 4.0%, p=0.014); TCM (24.2% vs. 7.3%, p=0.003); TEMRA (19.6% vs. 10.5%, p=0.060) and TEM (37.3% vs. 16.2%, p=0.004) (Figure 1a). There were no significant differences in frequencies of CD8+ T-lymphocyte subsets or PD-1 expression between Day 0 and Day 28 in AML patients (in 4 patients tested). Furthermore, there was negligible PD-1 co-expression with other immune checkpoint markers and no differences in 4-1BB or CTLA4 expression in AML versus HD. However, we noted significantly higher TIM-3 expression on TN and TCM CD8+ T-lymphocytes in AML (6.5% vs. 2.2%, p=0.001, and 4.5% vs. 0.7%, p=0.021, respectively). Sequencing of productive TCRB gene rearrangements revealed significantly higher average marrow T-cell clonality in AML versus HD (0.232 vs. 0.086, respectively, n=5 AML, n=5 HD). When considering the top 10 most frequent clonotypes for each AML patient and HD, frequencies were found to be much higher in AML than HD (Figure 1b), consistent with large clonal T-lymphocyte expansions in the marrow of RR-AML patients. Furthermore, we identified several TCR clones that were found in AML patients but never seen in HD. Sequencing of an additional cohort of 20 newly diagnosed adult AML patients however discovered that only a subset (25%) of those patients had T cell clonal expansions above the upper limit of that seen in HD. AML is already known to be an immunogenic malignancy, as demonstrated by the measurable efficacy of allogeneic transplantation and donor lymphocyte infusion. Our data demonstrate that the bone marrow tumor microenvironment in RR-AML is enriched for PD-1+ CD8+ marrow-infiltrating lymphocytes, and that large clonal expansions of T-lymphocytes can be found in the bone marrow in these patients. This is suggestive that RR-AML patients, for whom cytotoxic chemotherapy is often suboptimal, may benefit from immune checkpoint blockade therapies, particularly PD-1 axis inhibition, to enable improved immunologic control of AML by autologous T-lymphocytes already resident in the tumor microenvironment. Based in part on these data, a clinical trial of anti-PD-1 immunotherapy in combination with a hypomethylating agent for treatment of relapsed and refractory AML patients will open at our institution. Disclosures Hourigan: Sellas Life Sciences Group: Research Funding.