Identification of phenotypically, functionally, and anatomically distinct stromal niche populations in human bone marrow based on single-cell RNA sequencing
- Division of Molecular Hematology and Stem Cell Center, Lund University, Sweden
- Department of Hematology, Skåne University Hospital, Sweden
Figures
Figure 1 with 1 supplement
Single-cell atlas of human bone marrow CD45low/-CD235a- cells.
(A) Overview of the study design including gating strategies for isolation of human bone marrow CD45low/-CD235a- and CD45low/-CD235a-CD271+ cells. (B) Uniform Manifold Approximation and Projection (UMAP) display of single-cell transcriptomic data of human bone marrow CD45low/-CD235a- cells containing enriched CD45low/-CD235a-CD271+ cells to allow for a detailed analysis of rare stromal cell subpopulations. Data are shown for a total of n=25067 cells [5704 CD45low/-CD235a- cells (22.76%) and 19363 CD45low/-CD235a-CD271+ cells (77.24%)] from a total of nine healthy donors. Color legend indicates cluster numbers and annotations. Basal, basal cell-like cluster; B prog., B cell progenitors; SC, stromal cells; PC, plasma cells; HSPC, hematopoietic stem and progenitor cells; Mk, megakaryocytes; DC, dendritic cells; GC, granulocytes; Ery., erythroid cells; NK, natural killer cells; EC, endothelial cells; Neuronal, neuronal cell-containing cluster. (C) UMAP display of non-hematopoietic clusters (n=9686 cells). The circled area in the overview (top left corner) indicates the non-hematopoietic clusters selected from (B). (D) Heatmap of representative differentially expressed genes for each of the non-hematopoietic clusters in (C). For stromal marker identification, the top 100 significant differentially expressed genes from each stromal cluster were selected to identify overlapping genes. Three genes shared by all nine clusters were identified in this comparison (NNMT, IFITM3, and DCN). Nine more genes were identified when OC clusters 23 and 29 (identified as OCs in Figure 2A) were removed from the comparison, as OCs showed considerably different gene expression profiles as compared with other stromal clusters. Cluster numbers and corresponding cell types are indicated under the heatmap. The scale bar indicates gene expression levels. EC, endothelial cells. A blow-up heatmap for clusters 38, 29, 23, 6, and 37 is shown under the main heatmap for better visualization.
Figure 1—figure supplement 1
Cluster annotation of human bone marrow CD45low/-CD235a cells.
UMAP display of normalized CXCL12 expression in sorted CD45low/-CD235a- cells (5704 cells) (A) and CD45low/-CD235a-CD271+ cells (19,363 cells) (B). Dashed lines outline shows the UMAP overview of the entire dataset. (C) scRNAseq quality measures expressed as gene numbers (left panel) and unique molecular identifier (UMI, right panel) versus cell number (y-axis). Cells with less than 1000 UMIs were removed. (D) UMAP display of single-cell transcriptomic data of human bone marrow CD45low/-CD235a- cells (5704 cells) from four healthy donors. Cluster number and color legend are consistent with Figure 1B. (E) UMAP display of single-cell transcriptomic data of human bone marrow CD45low/-CD235a-CD271+ cells (19,363 cells) from five healthy donors. Cluster number and color legend are consistent with Figure 1B. (F) UMAP display (as in Figure 1B) of human bone marrow CD45low/-CD235a- cells and CD45low/-CD235a-CD271+ cells from nine healthy donors with samples color coded (25,067 cells). BM indicates CD45low/-CD235a- cell donors (n=4), CD271 indicates CD45low/-CD235a-CD271+ cell donors (n=5). O, old; Y, young (see also Materials and Methods). (G) UMAP displays (as in Figure 1B) highlighting the expression of selected hematopoietic signature genes for major cell types. PF4-expressing cells were blown-up for better visualization. (H) UMAP display (as in Figure 1C) of normalized expression of selected non-hematopoietic signature genes for major cell types. Expression was normalized to 1000 UMIs. Scale bars are adjusted for optimal visualization in (G) and (H).
Figure 2 with 1 supplement
scRNAseq reveals distinct gene expression patterns in different bone marrow stromal cell populations.
(A) UMAP display of the nine stromal subsets (n=7069 cells). Color legends indicate stromal cluster numbers and annotations. MSSC, multipotent stromal stem cells; OC, osteochondrogenic progenitors. (B) Single cell heatmap of representative differentially expressed genes in each cluster shown in (A). A blow-up heatmap for clusters 38, 29, 23, 6, and 37 is shown in Figure 2—figure supplement 1A for better visualization. The scale bar indicates gene expression levels. (C–F) Stacked violin plots of adipogenic- (C), osteogenic- (D), osteochondrogenic- (E), and fibroblastic (F) markers in different stromal clusters. Dashed lines separate MSSCs, HAGEPs, balanced progenitors, and pre-osteoblast (left; cluster 3, 5, 16, and 38), OCs (middle; cluster 29, 23), and pre-fibroblasts (right; cluster 6, 37, 8). The boxes in the violin plots indicate the lower, median and upper quartiles.
Figure 2—figure supplement 1
Differential expression of markers in stromal clusters.
(A) Single-cell heatmap of representative differentially expressed genes in clusters 38, 29, 23, 6, and 37. Stacked violin plots of stromal markers (B), stress-related transcription factors (C) and hematopoietic markers expressed by stromal clusters (D). Cluster annotation as in Figure 2A.
Figure 3 with 1 supplement
RNA velocity analysis reconstructs the temporal sequence of transcriptomic events of stromal cells.
(A and A’) Single cell velocities of the nine stromal clusters visualized as streamlines in a UMAP. Black arrows indicate direction and thickness indicates speed along the stromal cell development trajectory. (A’) Colored thick arrows indicate the main directions of stromal cell developmental paths. Colors correspond with the differentiation destination. Red: HAGEPs; yellow: balanced progenitors; light green: OCs; dark green: pre-osteoblasts; blue: pre-fibroblasts. (B) UMAP display of stromal cells colored by inferred latent time. Inferred latent time is represented by a color scale from 0 (the earliest latent time) to 1 (the latest latent time). (C) Left: UMAP display of stromal cell clusters that form a continuum of different cellular states (n=6376 cells). Colors correspond to different stromal clusters as in Figure 2A. Right: Heatmap constructed by the top 300 likelihood-ranked genes demonstrates gene expression dynamics along latent time. Colors on top of the heatmap correspond with cluster colors in UMAP (left). Key genes are highlighted by different colors on the right. Gene name colors correspond to different developmental stage transitions. Red: genes responsible for MSSC intra-cluster transition; orange: genes responsible for the transition to HAGEPs; light green: genes responsible for the transition to OCs; dark green: genes responsible for the transition to balanced progenitors and pre-osteoblasts.
Figure 3—figure supplement 1
Latent time, pseudotime, velocity confidence and expression dynamics of selected genes inferred by RNA velocity analysis.
(A–B) UMAP display of latent time and pseudotime analyses for the selected stromal clusters that formed a continuum of different cellular states. Inferred latent time and pseudotime are represented by a color scale from 0 (the earliest latent time/pseudotime) to 1 (the latest latent time/pseudotime). (C) UMAP display of velocity confidence for the selected stromal cells. Scale bar indicates the average correlation of the velocity vector of a certain cell and those of its neighbors. (D) Different cellular states of the selected stromal cells were analyzed with scVelo and single cell velocities visualized as streamlines in a UMAP. Black arrows indicate direction and thickness indicates speed along the stromal cell development trajectory. Colors correspond to cluster colors in Figure 3C. (E–G) Expression dynamics of putative driver genes identified by likelihood model analysis. Upper panel: Phase portraits of selected putative driver genes (indicated on top) characterize their splicing kinetics. The dashed purple line corresponds to the estimated ‘steady-state’ ratio, i.e. the ratio of unspliced to spliced mRNA abundance. Positive (cells over the dashed purple line) and negative velocities (cells under the dashed purple line) correlate with up- and down-regulation of genes, respectively. The solid purple lines indicate the learned kinetics for each gene calculated with a likelihood-based model by scVelo. The colors in the phase portraits correspond with the colors in UMAP (as in Fig. S3D). Lower panel: UMAP display (as in Figure 3C) of the expression dynamics of putative driver genes indicated on top of the phase portraits in the upper panel. Colors correlate with expression levels as shown in the color bar.
Figure 4 with 16 supplements
Stromal cell isolation based on data-driven gating strategies distinguishes cell subsets with different colony-forming capacities and differentiation capacities.
(A) Dot plot of surface marker gene expression in different stromal cell clusters. Cluster numbers and corresponding stromal cell groups are indicated on the y-axis legend. Dot sizes represent the percentage of cells expressing a certain gene in each cluster and dot colors represent the scaled average expression of that gene. (B) FACS plots illustrating the gating strategy for the isolation of different stromal subsets. The displayed cell populations are indicated on top of the plot. Following exclusion of doublets, dead cells, CD45- and CD235a-expressing cells, CD45low/-CD235a-CD71-CD271+ cells (left panel) were gated based on CD52 and NCAM1 expression (middle panel). The resulting three populations were labelled as A-C (middle panel), corresponding to the stromal cell groups in (A) (A, CD52-NCAM1-; B, CD52-NCAM1+; C, CD52+NCAM1-). CD45-CD235a-CD71-CD271+CD52-NCAM1- (group A) cells were further divided based on CD81 expression and four populations were identified (A1–A4) (right panel). A1, CD81++; A2, CD81+; A3; CD81+/-; A4, CD81-. (C) CFU-F frequencies of sorted stromal cell populations as shown in (B). Data are presented as individual data (dots) and median (horizontal lines) from independent experiments (n=3–6). Symbol colors and x-axis labels correspond to the cell population colors in (B). *: p<0.05; ****: p<0.0001 (Kruskal-Wallis test). (D) In vitro differentiation capacity of sorted stromal cell populations (as indicated in B) towards the adipogenic, osteoblastic, and chondrogenic lineage. Non-induction controls are shown in the left panel. Scale bars represent 200 µm. A representative set of pictures from a total of three independent experiments is shown. (F) Formalin-fixed, paraffin-embedded (FFPE) human BM slides were sequentially stained for DAPI (blue), CD45 (yellow), CD81 (green), CD271 (pink), and NCAM1 (cyan) and scanned with the OlympusVS120 slide scanner. Different staining combinations are shown as indicated under each picture to provide better visualization of individual staining obtained from the same FFPE slide. Scale bars represent 50 µm. Red arrows: CD271+CD81++ cells; white arrows: arteriolar walls; white lines: bone lining regions. Bone (b), adipocytes (a), and capillaries (*) are indicated.
Figure 4—figure supplement 1
Characterization of stromal cells based on gene expression, colony formation and in situ staining.
(A) Stacked violin plots of stromal, osteochondrogenic, and fibroblastic gene expression in different stromal clusters. Corresponding stromal cell groups are indicated on the x-axis legend. The color bar indicates gene expression in each cluster. (B) mRNA fold change of CD81 in FACS soreted A1-A4 subsets (n=3). Results are shown as mRNA fold change after standardization with GAPDH levels. For comparison, expression level of CD81 in A4 is set as one fold. *: p<0.05 (Kruskal-Wallis test). (C) The relative contribution of each sorted stromal cell populations in (B) to total CFU-F. Data are given as a normalized mean percentage as indicated in the figure (n=3). The sum of CFU-F of all populations is defined as 100%. (D) Formalin-fixed, paraffin-embedded (FFPE) human BM slides were sequentially stained for CD271 (pink), CD81 (green), NCAM1 (cyan) and CD45 (yellow) and scanned with the OlympusVS120 slide scanner. Single staining for each marker is shown as indicated under each picture. Scale bars represent 50 µm. Red arrows: CD271+CD81++ cells; white arrows: arteriolar walls; white lines: bone lining regions. Bone (b), adipocytes (a), and capillaries (*) are indicated. (E) FFPE human BM slides were sequentially stained for DAPI (blue), CD45 (yellow), CD81 (green), CD271 (pink), and NCAM1 (cyan) and scanned with the OlympusVS120 slide scanner. Scale bars represent 20 µm.
Figure 4—figure supplement 2
Visualization of CD271 and CD45 expression in bone marrow adipocyte regions by confocal microscopy.
(A) Confocal scan of adipocytic region in BM biopsies with 3D orthographic cross-section view, co-stained with mouse anti-CD271, rabbit anti-CD45, and DAPI. (B) Single channel data for the florescent markers in A. (C) Intensity profile for all channels in A across a cell of interest in a representative z plane. Scale bars represent 50 μm. Yellow dashed lines indicate adipocyte surface.
Figure 4—figure supplement 3
Visualization of NCAM1 and CD45 expression in bone marrow adipocyte regions by confocal microscopy.
(A) Confocal scan of adipocytic region in BM biopsies with 3D orthographic cross-section view, co-stained with mouse anti-NCAM1, rabbit anti-CD45, and DAPI. (B) Single channel data for the florescent markers in A. (C) Intensity profile for all channels in A across a cell of interest in a representative z plane. Scale bars represent 50 μm. Yellow dashed lines indicate adipocyte surface.
Figure 4—figure supplement 4
Visualization of NCAM1, CD271, and SPP1 expression in bone marrow adipocyte regions by confocal microscopy.
(A) Confocal scan of adipocytic region in BM biopsies with 3D orthographic cross-section view, co-stained with rabbit anti-NCAM1, mouse anti-CD271, goat anti-SPP1, and DAPI. (B) Intensity profile for all channels in A across a cell of interest in a representative z plane. (C) Single channel data for the florescent markers in A. Scale bars represent 50 μm. Yellow dashed lines indicate adipocyte surface.
Figure 4—figure supplement 5
Visualization of CD81 and CD45 expression in bone marrow adipocyte regions by confocal microscopy.
(A) Confocal scan of adipocytic region in BM biopsies with 3D orthographic cross-section view, co-stained with mouse anti-CD81, rabbit anti-CD45, and DAPI. (B) Single channel data for the florescent markers in A. (C) Intensity profile for all channels in A across a cell of interest in a representative z plane. Scale bars represent 50 μm. Yellow dashed lines indicate adipocyte surface.
Figure 4—figure supplement 6
Visualization of CD271 and CD81 expression in bone marrow adipocyte regions by confocal microscopy.
(A) Confocal scan of adipocytic region in BM biopsies with 3D orthographic cross-section view, co-stained with mouse anti-CD271, rabbit anti-CD81, and DAPI. (B) Single channel data for the florescent markers in A. (C) Intensity profile for all channels in A across a cell of interest in a representative z plane. Scale bars represent 50 μm. Yellow dashed lines indicate adipocyte surface.
Figure 4—figure supplement 7
Visualization of CD271 and CD45 expression in bone marrow endosteal regions by confocal microscopy.
(A) Confocal scan of endosteal region in BM biopsies with 3D orthographic cross-section view, co-stained with mouse anti-CD271, rabbit anti-CD45, and DAPI. (B) Single channel data for the florescent markers in A. (C) Intensity profile for all channels in A across a cell of interest in a representative z plane. Scale bars represent 50 μm. White dashed lines indicate trabecular bone surface.
Figure 4—figure supplement 8
Visualization of NCAM1 and CD45 expression in bone marrow endosteal regions by confocal microscopy.
(A) Confocal scan of endosteal region in BM biopsies with 3D orthographic cross-section view, co-stained with mouse anti-NCAM1, rabbit anti-CD45, and DAPI. (B) Single channel data for the florescent markers in A. (C) Maximum intensity projection of the whole z-stack shown in A. (D) Intensity profile for all channels in A across a cell of interest in a representative z plane. Scale bars represent 50 μm. White dashed lines indicate trabecular bone surface.
Figure 4—figure supplement 9
Visualization of SPP1, NCAM1, and CD271 expression in bone marrow endosteal regions by confocal microscopy.
(A) Confocal scan of endosteal region in BM biopsies with 3D orthographic cross-section view, co-stained with rabbit anti-NCAM1, mouse anti-CD271, goat anti-SPP1, and DAPI. (B) Intensity profile for all channels in A across a cell of interest in a representative z plane. (C) Single channel data for the florescent markers in A. Scale bars represent 50 μm. White dashed lines indicate trabecular bone surface.
Figure 4—figure supplement 10
Visualization of CD81 and CD45 expression in bone marrow endosteal regions by confocal microscopy.
(A): Confocal scan of endosteal region in BM biopsies with 3D orthographic cross-section view for the z-axis, co-stained with mouse anti-CD81, rabbit anti-CD45, and DAPI. (B) Single channel data for the florescent markers in A. (C) Intensity profile for all channels in A across a cell of interest in a representative z plane. Scale bars represent 50 μm. White dashed lines indicate trabecular bone surface.
Figure 4—figure supplement 11
Visualization of CD271 and CD81 expression in bone marrow endosteal regions by confocal microscopy.
(A) Confocal scan of endosteal region in BM biopsies with 3D orthographic cross-section view, co-stained with mouse anti-CD271, rabbit anti-CD81, and DAPI. (B) Single channel data for the florescent markers in A. (C) Intensity profile for all channels in A across a cell of interest in a representative z plane. Scale bars represent 50 μm. White dashed lines indicate trabecular bone surface.
Figure 4—figure supplement 12
Visualization of NCAM1 and CD45 expression in bone marrow vascular regions by confocal microscopy.
(A) Confocal scan of vascular region in BM biopsies with 3D orthographic cross-section view, co-stained with mouse anti-CD271, rabbit anti-CD45, and DAPI. (B) Single channel data for the florescent markers in A. (C) Intensity profile for all channels in A across a cell of interest in a representative z plane. Scale bars represent 50 μm. Cyan dashed lines indicate vessel surface.
Figure 4—figure supplement 13
Visualization of NCAM1 and CD45 expression in bone marrow vascular regions by confocal microscopy.
(A) Confocal scan of vascular region in BM biopsies with 3D orthographic cross-section view, co-stained with mouse anti-NCAM1, rabbit anti-CD45, and DAPI. (B) Single channel data for the florescent markers in A. (C) Intensity profile for all channels in A across a cell of interest in a representative z plane. Scale bars represent 50 μm. Cyan dashed lines indicate vessel surface.
Figure 4—figure supplement 14
Visualization of SPP1, NCAM1, and CD271 expression in bone marrow vascular regions by confocal microscopy.
(A) Confocal scan of vascular region in BM biopsies with 3D orthographic cross-section view, co-stained with rabbit anti-NCAM1, mouse anti-CD271, goat anti-SPP1, and DAPI. (B) Intensity profile for all channels in A across a cell of interest in a representative z plane. (C) Single channel data for the florescent markers in A. Scale bars represent 50 μm. Cyan dashed lines indicate vessel surface.
Figure 4—figure supplement 15
Visualization of CD81 and CD45 expression in bone marrow vascular regions by confocal microscopy.
(A) Confocal scan of vascular region in BM biopsies with 3D orthographic cross-section view, co-stained with mouse anti-CD81, rabbit anti-CD45, and DAPI. (B) Single channel data for the florescent markers in A. (C) Intensity profile for all channels in A across a cell of interest in a representative z plane. Scale bars represent 50 μm. Cyan dashed lines indicate vessel surface.
Figure 4—figure supplement 16
Visualization of CD271 and CD81 expression in bone marrow vascular regions by confocal microscopy.
(A) Confocal scan of vascular region in BM biopsies with 3D orthographic cross-section view, co-stained with mouse anti-CD271, rabbit anti-CD81, and DAPI. (B) Single channel data for the florescent markers in A. (C) Intensity profile for all channels in A across a cell of interest in a representative z plane. Scale bars represent 50 μm. Yellow dashed lines indicated vessel surface.
Figure 5 with 3 supplements
Cell-cell interaction between stromal cells and hematopoietic cells in human bone marrow.
(A) Circos plot visualization of cell-cell interaction between bone marrow stromal cells and different hematopoietic cell types. Colored lines connect stromal clusters with hematopoietic clusters based on LR expression analysis. Line colors indicate cell type and line thicknesses correspond to the number of interacting LR pairs. Cluster numbers and corresponding cell types and stromal groups are indicated. Hematopoietic cell type abbreviations are listed in the table. (B and C) Overview of selected ligand-receptor pairs between MSSCs and OCs, respectively, and different hematopoietic cell types from this dataset (B) and more defined hematopoietic progenitors from another study (C). The y-axis label indicates the pair of interacting cell clusters (‘cluster X_cluster Y’ indicates cluster X interaction with cluster Y by ligand-receptor expression). The x-axis indicates the interacting receptor/ligand (R/L) or ligand/receptor pairs, respectively (‘molecule L_molecule R’, molecule L interacts with molecule R). The means of the average expression level of the interacting molecules are indicated by circle size (adjacent lower scale bar). P values are indicated by circle color and correspond to the upper scale bar. Cluster numbers of the hematopoietic cell types are listed in the table. The dashed line separates MSSC from OC interactions. DC-I, dendritic cell progenitor cluster I; GMP, granulocyte-macrophage progenitor cluster I; HSC-I, hematopoietic stem cell cluster I; HSC-II, hematopoietic stem cell cluster II; Ly-I, lymphoid progenitor cluster I; MEP-I, megakaryocyte–erythroid progenitor cluster I; MPP-I, multipotent progenitor cluster I.
Figure 5—figure supplement 1
Cell-cell interactions inferred by CellPhoneDB.
(A) Heatmap showing the total number of interactions between different clusters in bone marrow CD45low/-CD235a- cells. Scale bar colors represent numbers of interactions. (B) Dot plot of selected ligand-receptor pairs between bone marrow stromal cells and different hematopoietic cell types. The y-axis label indicates the pair of interacting cell clusters (‘cluster X_cluster Y’ indicates that cluster X interacts with cluster Y by ligand-receptor expression). The x-axis indicates the interacting receptor/ligand (R/L) or ligand/receptor pairs, respectively (‘molecule L_molecule R’ means molecule L interacts with molecule R). The means of the average expression level of the interacting molecules are indicated by circle size. The scale is shown on the right side. p Values are indicated by circle color and correspond to the adjacent upper scale bar. Dashed lines separate the interactions between hematopoietic cells and different stromal clusters. (C) Schematic overview of the layout of the interaction matrix in (B). HAGEP, highly adipocytic gene-expressing progenitors; Bal., balanced progenitors; pre-ob, pre-osteoblasts; hemato, hematopoietic clusters. Cluster numbers of the hematopoietic cell types selected for this analysis are listed in the table on the right side.
Figure 5—figure supplement 2
UMAP illustration of selected ligands and receptors involved in cell-cell interaction.
(A–E, G–H, J) UMAP (as in Figure 1B) illustration of the normalized expression of selected ligand and receptor gene pairs. (F) Flow cytometric analysis of SPP1 expression of primary CD45low/-CD235a-CD71-CD271+CD56+ cells (red histogram), CD45low/-CD235a-CD71-CD271+CD56-CD81++ cells (blue histogram) and corresponding isotype control (orange histogram). (I) CFU-F assay of sorted CD45low/-CD235a-CD71-CD271+ cells (100 cells/well) in the presence (upper panel) or absence (lower panel) of sorted bone marrow CD45+ cells (3x105 cells/well), respectively.
Figure 5—figure supplement 3
Cell-cell interaction between different clusters and functional assays.
(A) CFU-F frequency of single-cell and bulk sorted MSSCs, i.e. sorted A1 cells (CD45low/-CD235a-CD71-CD271+NCAM1-CD52-CD81++). Data are shown as individual data points and medians (horizontal lines), n=3. **: p<0.01 (Mann-Whitney test). (B) A dot plot overview of top-ranked ligand-receptor pairs between different bone marrow stromal clusters. P values are indicated by circle color. The means of the average expression levels are indicated by circle size. Scale bars are provided on the right side of the plot. Stromal cell cluster pairs are indicated by ‘cluster number_cluster number’ (y-axis labels). Dashed lines separate the different stromal clusters. HAGEP, highly adipocytic gene-expressing progenitors; Bal., balanced progenitors; Pre-ob, pre-osteoblasts; Pre-fb, pre-fibroblasts. Interacting molecule pairs are indicated in the x-axis labels. (C) UMAP (as in Figure 1B) illustration of the normalized expression of genes involved in alpha 1 beta 1 integrin complex (a1b1 complex). (D) UMAP (as in Figure 1B) illustration of the normalized expression of selected ligand and receptor genes. (E) Fold change of total number of CD34+ cells (left) and CD34+CD90+ cells (right) produced after seven days in culture. Results are shown as fold change relative to the cell number of standard CD34+ culture without stroma support (No stroma). Data are presented as individual data (dots) and median (horizontal lines) from independent experiments (n=2).
Figure 6
Expression of selected ligand and receptor pairs involved in cell-cell interaction between stromal cells and hematopoietic cells in human bone marrow.
(A–D, F–G) UMAP (as in Figure 1B) illustration of the normalized expression of selected ligand and receptor gene pairs. Red dashed lines in figures A, B and D mark the erythroid clusters. The blow-up shown in (G) is to better visualize PDGFA- and PDGFB-expressing cells. (E) Left panel: Formalin-fixed, paraffin-embedded (FFPE) human BM slides were sequentially stained for DAPI (blue), CD45 (yellow), SPP1 (red), CD271 (pink), and NCAM1 (cyan) and scanned with an OlympusVS120 slide scanner. Left upper image: all markers are shown; left lower image: just NCAM1 channel is shown. Middle panel: confocal laser scanning analysis of BM biopsies co-stained with CD271 (red), NCAM1 (green), SPP1 (white), and DAPI (blue) presented as 3D orthographic cross-section view. Right panel: Single staining channel data for CD271 (red), NCAM1 (green), SPP1 (white), and corresponding blow-ups for indicated areas. NCAM1, SPP1 and CD271(NGFR). White scale bars represent 50 μm and yellow scale bars represent 25 μm. White dashed lines indicate the trabecular bone surface lining regions.
Figure 7
Cell-cell interaction between the endothelial cluster and hematopoietic cells in human bone marrow.
(A) Overview of selected ligand-receptor pairs between the endothelial cluster (cluster 28) and different hematopoietic cell types. The y-axis label indicates the pair of interacting cell clusters (‘cluster X_cluster Y’ indicates cluster X interaction with cluster Y by ligand-receptor expression). The x-axis indicates the interacting receptor/ligand (R/L) or ligand/receptor pairs, respectively (‘molecule L_molecule R’, molecule L interacts with molecule R). The means of the average expression level of the interacting molecules are indicated by circle size (adjacent lower scale bar). P values are indicated by circle color and correspond to the upper scale bar. Cluster numbers of the hematopoietic cell types are listed in the table in Figure 5B.
Additional files
-
MDAR checklist
- https://cdn.elifesciences.org/articles/81656/elife-81656-mdarchecklist1-v3.pdf
-
Supplementary file 1
Donor information.
- https://cdn.elifesciences.org/articles/81656/elife-81656-supp1-v3.docx
-
Supplementary file 2
Cluster annotation.
- https://cdn.elifesciences.org/articles/81656/elife-81656-supp2-v3.docx
-
Supplementary file 3
Cell numbers and percentages of each cluster in each sample.
- https://cdn.elifesciences.org/articles/81656/elife-81656-supp3-v3.xlsx
-
Supplementary file 4
DE (differentially expressed) genes in each cluster.
- https://cdn.elifesciences.org/articles/81656/elife-81656-supp4-v3.xlsx
-
Supplementary file 5
List of the 300 top-ranked likelihood genes shown in Figure 3C.
- https://cdn.elifesciences.org/articles/81656/elife-81656-supp5-v3.xlsx
-
Supplementary file 6
List of ligand-receptor interactions in scRNAseq dataset as identified by CellPhoneDB analysis.
- https://cdn.elifesciences.org/articles/81656/elife-81656-supp6-v3.xlsx
-
Supplementary file 7
Stromal cluster summary.
- https://cdn.elifesciences.org/articles/81656/elife-81656-supp7-v3.docx
Download links
A two-part list of links to download the article, or parts of the article, in various formats.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
Identification of phenotypically, functionally, and anatomically distinct stromal niche populations in human bone marrow based on single-cell RNA sequencing
eLife 12:e81656.
https://doi.org/10.7554/eLife.81656
