Proximity-Based Differential Single-Cell Analysis of the Niche to Identify Stem/Progenitor Cell Regulators

Abstract

Physiological stem cell function is regulated by secreted factors produced by niche cells. In this study, we describe an unbiased approach based on the differential single-cell gene expression analysis of mesenchymal osteolineage cells close to, and further removed from, hematopoietic stem/progenitor cells (HSPCs) to identify candidate niche factors. Mesenchymal cells displayed distinct molecular profiles based on their relative location. We functionally examined, among the genes that were preferentially expressed in proximal cells, three secreted or cell-surface molecules not previously connected to HSPC biology-the secreted RNase angiogenin, the cytokine IL18, and the adhesion molecule Embigin-and discovered that all of these factors are HSPC quiescence regulators. Therefore, our proximity-based differential single-cell approach reveals molecular heterogeneity within niche cells and can be used to identify novel extrinsic stem/progenitor cell regulators. Similar approaches could also be applied to other stem cell/niche pairs to advance the understanding of microenvironmental regulation of stem cell function.

Figure 1. Proximity-based single cell analysis of the bone marrow niche

(A) Fluorescent microscopy image of femoral bone section from newborn col2.3GFP+ animal injected with DiI-labeled adult bone marrow LT-HSCs 48 hours after transplantation. Proximal OLC (circled) is defined as the nearest GFP+ cell within two cell diameters (red square) from HSPC (arrowhead). Distal cells are GFP+ cells located outside of this area at least five cell diameters away from transplanted HSPC (white arrows). Scale bar 10 microne. (B) Experimental workflow. (C) Example of micropipette aspiration of proximal OLC. Shown are overlaid single color (GFP and DiI) images before and after retrieval of proximal OLC (i) The proximal GFP⁺ OLC (green) was identified based on proximity to the DiI-labeled HSPC (red). (ii, iii) Following in-situ enzymatic dissociation, HSPC was dislodged from its original location and proximal OLC was aspirated into a micropipette. Scale bar 10 microne.

Figure 2. Transcriptional profiling of proximal and distal OLCs by single cell RNA-Seq

(A) An unbiased genome-wide classification of proximal and distal OLCs. The receiver-operator characteristic curve for the Support Vector Machine classification where all successive pairs of cells (one proximal and one distal) were classified based on the training data provided by other cells (P<0.005). (B) The use of SCDE to estimate the posterior distribution of expression levels based on the observations from each cell (colored lines, top and bottom panels). Analysis of the Vcam-1 gene is shown as an example. The joint posteriors (black lines) describe the overall estimation of likely expression levels within the proximal (top) and distal (bottom) OLCs, and are used to estimate the posterior of the expression fold difference (middle plot). The shaded area under the fold-difference posterior shows 95% confidence region.(C) Classification of individual OLCs based on the top 200 differentially expressed genes. Each row represents a gene, with the most likely gene expression levels indicated by color (blue – high, white – low/absent). (D,E) Expression analysis of known niche-derived HSPC regulators and OLC maturation genes. The violin plots show the posterior distribution of the expression fold-difference (y-axis, log2 scale) for each gene, with the shaded area marking the 95% confidence region (equivalent to the middle plot in B). The horizontal solid red lines show the most likely fold-change value. (F) Quantification of IL18 expression in neonatal post-transplant bone marrow niche by fluorescent in-situ hybridization (RNA Scope). Newborn col2.3GFP+ animals were transplanted with DiI-labeled long-term HSCs and sacrificed at 48 hours, as for cell harvesting experiments. Frozen sections of fixed undecalcified femoral bones were hybridized with fluorescently-labeled IL18 probe. Maximal intensity projections over 12 micrones are displayed. Top panel: Lower magnification images illustrating spatial definition of “proximal” (within two cell diameters from transplanted HSPC, arrowhead) and “distal” (greater than five cell diameters from transplanted HSPC) areas. Bottom panel: higher magnification images displaying DAPI stain (blue) and the RNA scope signal (red dots). The nuclei of GFP+ cells are circled. IL18 mRNA was quantified by comparing the intensity of the RNA scope signal (expressed as the average number of dots per cell) between GFP+ cells within proximal and distal areas, and the results are displayed in (E) (p<2.2×10⁻⁴, 95% confidence intervals are shown, based on the Poisson model of the dot occurrences per cell. P-value was estimated using Poisson rate ratio test).

Figure 3. Conditional deletion of Ang from niche cell subsets leads to the loss of quiescence in LT-HSCs and CLPs

(A) Comparison of Ang expression in proximal and distal OLCs. (B) LT-HSC number per femur and (C) LT-HSC cell cycle status following conditional deletion of Ang from distinct niche cell subsets, as per the color-coded legend (n=4-10). Non-shaded graphs: control animals, shaded graphs: Ang-deleted animals. (D) CLP number per femur and (E) CLP cell cycle status following conditional deletion of Ang from distinct niche cell subsets (n=4-10). (F) Long-term reconstitution following competitive (1:1) transplantation of bone marrow from control animals (solid lines) and animals with conditional deletion of Ang (broken lines) into WT congenic recipients (n=8). *P<0.05, **P<0.01, ***P<0.001. Data are presented as mean+/− SEM.

Figure 4. In vivo analysis of IL18 function in HSPC regulation

(A) Comparison of IL18 expression in proximal and distal OLC. (B) IL18 receptor expression in HSPC. Representative histograms are shown (n=3). A comparable cell population from IL18R KO mouse was used as a negative control (shaded histogram). (C) BrdU incorporation by HSPC in IL18KO mice (n=5). (D) Flow cytometric assessment of multi-lineage response to 5-FU in IL18KO mice. The statistical significance was assessed by ANOVA. Boxplots illustrating log ratios of cell numbers between 5- FU-treated and vehicle-treated animals in WT and IL18 groups are shown (n=7). (E) Enumeration of apoptotic LKS cells and lin-negative cells in WT animals pre-treated with rIL18 prior to 5-FU exposure (n=5). (F) Myeloid and lymphoid reconstitution in IL18KO mice following transplantation of (WT) LKS cells (n=7). (G) Myeloid and lymphoid reconstitution following transplantation of LKS cells from IL18R1KO or WT animals into WT hosts (n=8 per group). *P<0.05, **P<0.01. Data are presented as mean+/− SEM.

Figure 5. Identification of Embigin as a hematopoietic regulator

(A) Comparison of Embigin expression in proximal and distal OLC. (B) Quantification of primitive hematopoietic cells, (C) colony-forming cells and (D) cell cycle status following treatment with anti-Embigin or isotype control antibody (n=5). (E) The effect of anti-Embigin treatment on mobilization of lin⁻kit⁺ cells, lin⁻kit⁺Sca1⁺ cells and colony-forming cells into peripheral blood (n=4). (F,G) Quantification of homing and proliferation by intra-vital microscopy following transplantation of WT LKS cells into anti-Embigin-treated host or anti-Embigin-treated LKS cells into WT host (n=4). Cell number per calvarial marrow at 24 hours and calculated proliferation rate based on assessment of cell number by repeat imaging at 48 hours comparing are shown. Green – transplanted GFP+ cells, blue – second harmonic generation (bone signal). * p<0.05, ** p<0.01. Data are presented as mean+/− SEM.

Figure 6. Isolation and characterization of VCAM⁺Embigin^high subset of osteolineage cells (VE cells)

(A) Flow cytometric strategy for isolation of VE cells. (B) Expression of known niche factors and OLC maturation genes in VE cells from col2.3GFP mice (n=3). (C) Principal component analysis of transcriptomes from VE cells, nesitn-GFP^high and nestin-GFP^low cells (n=3). (D) Assessment of VE cell frequency after irradiation (n=4). (E) Changes in niche factor expression in VE cells following irradiation. (F) Results of GSEA analysis (cell-cell adhesion) in VE cells and non-VE cells from LT-HSC-versus saline-injected animals (n=3). Data are presented as mean+/− SEM.