Functionally and phenotypically distinct subpopulations of marrow stromal cells are fibroblast in origin and induce different fates in peripheral blood monocytes

Abstract

Marrow stromal cells constitute a heterogeneous population of cells, typically isolated after expansion in culture. In vivo, stromal cells often exist in close proximity or in direct contact with monocyte-derived macrophages, yet their interaction with monocytes is largely unexplored. In this report, isolated CD146(+) and CD146(-) stromal cells, as well as immortalized cell lines representative of each (designated HS27a and HS5, respectively), were shown by global DNase I hypersensitive site mapping and principal coordinate analysis to have a lineage association with marrow fibroblasts. Gene expression profiles generated for the CD146(+) and CD146(-) cell lines indicate significant differences in their respective transcriptomes, which translates into differences in secreted factors. Consequently, the conditioned media (CM) from these two populations induce different fates in peripheral blood monocytes. Monocytes incubated in CD146(+) CM acquire a tissue macrophage phenotype, whereas monocytes incubated in CM from CD146(-) cells express markers associated with pre-dendritic cells. Importantly, when CD14(+) monocytes are cultured in contact with the CD146(+) cells, the combined cell populations, assayed as a unit, show increased levels of transcripts associated with organismal development and hematopoietic regulation. In contrast, the gene expression profile from cocultures of monocytes and CD146(-) cells does not differ from that obtained when monocytes are cultured with CD146(-) CM. These in vitro results show that the CD146(+) marrow stromal cells together with monocytes increase the expression of genes relevant to hematopoietic regulation. In vivo relevance of these data is suggested by immunohistochemistry of marrow biopsies showing juxtaposed CD146(+) cells and CD68(+) cells associated with these upregulated proteins.

FIG. 1.

Three 1-dimensional principal coordinate analyses using the DHS landscape (linear patterning of DHSs) for each lineage class combined with data of primary marrow stromal cells from healthy donors and marrow stromal cell lines (HS5, HS27a). The plot shows the relative positioning of the primary stromal cells and the cell lines with clusters of endothelial in panel (A), hematopoietic in (B), and fibroblast lineages in (C). P values of stromal cells and each cluster were indicated. A list of GEO accession numbers, a detailed description of the cells, the source for the cell type, and a weblink to the tissue culture protocol for the cell type are shown in Supplementary Table S3. Color images available online at www.liebertpub.com/scd

FIG. 2.

Expression of CD146 in stromal cells. (A) Surface protein expression of CD146 was determined by flow cytometry for freshly isolated CD14/CD45-negative marrow stromal cells from healthy donors (black line), and cloned stromal cell lines, HS27a (blue line) and HS5 (red line). Cells were selected for analysis by forward and side scatter properties. The gray area and dotted line indicate the fluorescence intensity of isotype-matched control antibody staining. (B) Expression of CD146 gene transcripts in HS27a (blue bar), HS5 (red bar), and primary marrow stromal (black bar) cells was determined by quantitative PCR after normalized to G3PDH. Color images available online at www.liebertpub.com/scd

FIG. 3.

Morphological and phenotypic changes of CD14⁺ monocytes after culture in conditioned media (CM) from primary and immortalized stromal cells. (A–C) Phase-contrast images of CD14⁺ cells from healthy donors cultured in CM from HS27a (A), HS5 (B), and primary marrow stromal cultures (C) are shown. Images were captured with an inverted microscope using the 10×objective. A scale bar (0.1 mm) is shown in (A). (D, E) Predendritic (DC)-like cells induced by HS5-CM, acquire a DC phenotype when IL-4, TNFα are added to the culture with HS5 CM (D). The cells in the gated population were further analyzed and found to be CD80⁺/CD86⁺/DR⁺, similar to the standard DC phenotype (E). These cells present a prototypic tumor-associated antigenic epitope to cognate cytotoxic lymphocytes (CTL), and increase T-cell proliferation in a mixed lymphocyte reaction (data not shown). Color images available online at www.liebertpub.com/scd

FIG. 4.

K-means clustering analysis of microarray data obtained from HS27a cells, freshly isolated peripheral blood CD14⁺ monocytes (Mo), CD14⁺ cells cultured in HS27a conditioned media (Mo+CM), and cocultures of HS27a and CD14⁺ cells (HS27a+Mo). Normalized log₂ intensities were probewise median centered across all conditions (see Materials and Methods section for details). (A) A heat map of 8 clusters generated from the 631 probe sets showing differential expression (defined as |log₂(ratio)| ≥1 and FDR of 0.1%) in the comparison between the cocultures and HS27a cells. (B) Shows a Venn diagram of the number of highly expressed probe sets. The positions of Cluster 3 and 8 are indicated by arrows. (C) Shows the consistent values obtained in Cluster 8 for two sample preparations for the 102 probe sets upregulated in monocytes after CM stimulation. (D) Shows consistent results for 122 probe sets in Cluster 3 that are upregulated in cocultures. The red lines indicate the cluster's centroid. The y-axis displays the log₂ signal intensity relative to the median probe signal for a given cluster. Color images available online at www.liebertpub.com/scd

FIG. 5.

Immunohistochemical detection of CD146⁺ cells and CD163+ or CD68+ macrophages in marrow biopsies. Marrow biopsies were subjected to dual immunohistochemistry. (A, C) CD146 (red) and CD163 (gray/black). (B) Isotype controls (red and gray/black). (D–F) CD146 (red) and CD68 (gray/black). Several areas of increased numbers of both macrophages and CD146⁺ cells, particularly in association with bone, are shown at low and high magnifications (E, F). (G) CD146 (blue) and CDK6 (brown). (H) DKK3 (brown) with nuclei counter stained with hematoxylin (blue). Marrow biopsies from healthy and arthritic donors (A–D, G, H) and a patient with myelodysplastic syndrome (E, F) were used. Original objective 40×for (A), (B), and (F); 10×for panel E; 100×for (C, D, G, H). Scale bars (0.02–0.1 mm) are shown in each panel. Color images available online at www.liebertpub.com/scd