Integrated OMICs unveil the bone-marrow microenvironment in human leukemia

Abstract

The bone-marrow (BM) niche is the spatial environment composed by a network of multiple stromal components regulating adult hematopoiesis. We use multi-omics and computational tools to analyze multiple BM environmental compartments and decipher their mutual interactions in the context of acute myeloid leukemia (AML) xenografts. Under homeostatic conditions, we find a considerable overlap between niche populations identified using current markers. Our analysis defines eight functional clusters of genes informing on the cellular identity and function of the different subpopulations and pointing at specific stromal interrelationships. We describe how these transcriptomic profiles change during human AML development and, by using a proximity-based molecular approach, we identify early disease onset deregulated genes in the mesenchymal compartment. Finally, we analyze the BM proteomic secretome in the presence of AML and integrate it with the transcriptome to predict signaling nodes involved in niche alteration in AML.

Graphical abstract

Figure 1

BM niche populations identified with current markers (A) Schematic figure of the experimental design depicting stromal components isolated from the BM of NSG mice carrying specific fluorescent reporters. FACS populations were analyzed by RNA-seq. CD31 n = 7; Nes^high n = 4; Nes^low n = 5; Ng2 n = 3; Osx n = 5; Col^low n = 4; Col^high n = 5. (B) 3D PCA analysis showing similarities in expression profiles between different stromal components in healthy mice. Each dot represents an experimental replicate. (C) Flow-cytometry analysis of CD45^–Ter119^– BM cells derived from depicted mice showing overlap between markers associated to niche components. (D) Correlation matrix showing similarities between different stromal components in healthy mice. See also Figure S1.

Figure 2

Differential clusters between BM niche populations in homeostasis (A) Cluster behaviors among niche components. Left: within-cluster averages of Z-scored expression profiles across stromal types from control samples (means ± SEM). Right: top enriched processes associated to each cluster. (B) Process network map showing the relatedness of the different stromal types (size of each node proportional to the number of connections). See also Figures S1 and S2 and Tables S1 and S2.

Figure 3

Human AML impact on the functional identity of the BM niche (A) Schematic representation of the experimental setup. Depicted stromal components were isolated from the BM of NSG mice carrying specific fluorescent reporters and non-transplanted (ctrl) or xenografted with healthy hematopoietic cells (CB) or patient-derived AML samples (AML; see Table 1 for details). FACS-sorted populations were analyzed by RNA-seq. CD31: ctrl n = 7, CB = 4, AML = 15; Col^high: ctrl n = 5, CB = 2, AML = 9; Col^low: ctrl n = 4, CB = 3, AML = 13; Nes^high: ctrl n = 4, CB = 2, AML = 11; Nes^low: ctrl n = 5, CB = 2, AML = 11; Ng2: ctrl n = 3, CB = 3, AML = 12; Osx: ctrl n = 5, CB = 4, AML = 10. (B) Comparison of within-cluster averages of Z-scored expression profiles between control samples (black) and leukemic samples (red). (C–G) Deregulation (mean log2FC) of genes in depicted clusters in the presence of AML (^∗means FDR <0.1). See also Figures S3 and S4 and Table S3.

Figure 4

Progressive detriment of the BM mesenchymal niche in AML (A) Schematic representation of the experimental setup. Nes^low mesenchymal stromal cells were isolated from the BM of Nestin^GFP-NSG mice xenografted with human cherry⁺-AML cell lines at early and late time points of leukemic engraftment. FACS-sorted populations were analyzed by RNA-seq. Ctrl n = 5; AML early ch⁺ n = 6; AML early ch^– n = 6; AML late n = 6. (B) Flow cytometry profile of Nes^low cells from control mice or mice engrafted with human cherry⁺-AML cell lines at an early time point. (C) Pie chart showing the proportion of late deregulated genes also deregulated at early time points. (D) Heatmap showing the top transcripts deregulated in both the early and late stage of leukemic development in the mesenchymal niche. (E) Top upregulated (red) and downregulated (blue) processes in the early stage of leukemic engraftment in the mesenchymal niche. See also Figure S5.

Figure 5

The BM secretome in AML patient xenografts (A) Schematic representation of the experimental design. BM extracellular proteins were isolated from NSG mice non-transplanted (ctrl) or xenografted with healthy hematopoietic cells (CB) or human AML (AML; see Table 1 for details), and BM secretome was analyzed on a SOMAscan array. (B) 3D PCA showing similarities in protein expression profiles between samples. Each dot represents an experimental replicate. (C) Pie chart showing the number of deregulated proteins validated with comparison with depicted human datasets. (D) Top processes specifically upregulated (red) and downregulated (blue) in AML xenografts BM secretome compared to ctrl and validated on human data. See also Tables S4 and S5.

Figure 6

Integration of niche transcriptome and BM secretome (A) Graphic map of the ligand-receptor interactome between distinct niche components and the BM secretome. The thickness of the connecting lines is proportional to the number of interactions. (B–E) Sketch graphs of upregulated (red) and downregulated (blue) predicted ligand-receptor interactions between BM secretome and niche transcriptome. See also Figure S6 and dynamic figure https://giovannidiana.github.io/integratedomics/