The immunophenotype of all three populations

The immunophenotype of all three populations was comprehensively compared using a panel of 81 PE-conjugated TAPI-1 Supplier (Supplementary Table 1). A heat map showed extensive congruence in the percentage of cells positive for a given CD epitope between the three populations, although hierarchical clustering revealed that bmMSC and kCFU-F were more similar to each other than to cCFU-F (Figure 2A). An examination of all epitopes detected on >20% of cells in at least one of the three populations (Figure 2B) highlighted a number of matrix and adhesion molecules, potentially reflecting niche molecules expressed in common. These included CD29, CD44, CD49e, CD51, CD61, CD81 and CD24. CD105 (endoglin), a marker of primitive haematopoietic stem cells (HSCs), was also expressed on a high proportion of all cells. The expression of CD markers associated with immune regulation, including CD24, CD80, CD81 and CD90.2, implies a common immune regulatory function for MSC in multiple organs. CD49e, CD51, CD71 and CD105 have been previously reported on human MSC (da Silva Meirelles et al., 2008). CD73, generally accepted as a diagnostic human MSC marker, (Horwitz et al., 2005; Dominici et al., 2006), was not present on a significant percentage of any murine population analysed in this study (Supplementary Table 2). Conversely, CD80, negative on human MSC, was found on all 3 populations of murine MSC. Of note, whilst cCFU-F were isolated as GFP+ cells from a PDGFRα-GFP mouse strain, the mean percentage of cCFU-F cells positive for CD140a (detects PDGFRα) was 12%, comparable to the levels seen in both other populations. Analysis of the GFP positivity on FITC channel during immunophenotyping showed 23.7% GFP+ cells. This variation from the antibody may result from perdurance of the GFP (Data not shown).
Gene expression profiling reveals considerable congruence between the three populations To further compare phenotypes, Illumina microarray expression profiling was performed on RNA isolated from the three populations. Analysis of expression of >45,000 transcripts showed that 81.1% of genes were expressed to some degree across the combined set of samples. Pearson correlations of pair-wise comparisons revealed r2 values ranging from 0.878 to 0.924 (Figure 3A). This represents a very high level of similarity between the three cell types, although bmMSC were again more similar to kCFU-F than to cCFU-F. We ranked all genes in order of level of expression for all three populations. Ninety-five of the top 100 bmMSC-expressed genes were also ranked within the top 250-ranked genes listed for the other populations (Supplementary Table 4). As expected, most of the top 100 genes encoded house-keeping proteins. However, this list also included regulatory molecules and secreted proteins involved in tissue remodelling, repair and inflammatory modulation. Amongst the most highly expressed genes in all lines was Zyxin, Thymosin β10 and Macrophage inhibitory factor (MIF) (Supplementary Table 4). Zyxin (ranked 1 for cCFU-F, 10 for kCFU-F and 35 for bmMSC) encodes a member of the LIM domain family of focal adhesion adaptor proteins that is likely to also serve as a component of transcription factor complexes. Thymosin β10, a monomeric actin-sequestering protein, is secreted and has paracrine functions in wound repair (Huang et al., 2006), vasculogenesis (Lee et al., 2005), inflammation and cancer (Lapteva et al., 2001; Califano et al., 1998). In contrast to its paralogue thymosin β4, which is pro-angiogenic, thymosin β10 inhibits angiogenesis. MIF protein has emerged as a major inflammatory mediator through its chemokine-like functions, acting both upstream and downstream of inflammatory inducers. It has been proposed as a “master regulator” for leukocyte chemotaxis and arrest. MIF is also pro-angiogenic and inhibits migration and division of smooth muscle (SM) cells. Other commonly expressed secreted products included biglycan, a proteoglycan that modulates cellular proliferation and migration; serpinh1/HSP47, a procollagen chaperone involved in collagen remodelling and formation of scar tissue; serpinf1/PEDF, an inhibitor of endothelial cell proliferation and migration whilst also acting as a fibroblast chemoattractant; Sparc/Osteonectin, an MMP-activated, collagen-binding protein implicated in collagen fibril and basal lamina formation; and CTGF/CCN2, a known mediator of tissue fibrosis. These all strongly reflect the fibroblastic phenotype of these cells. Pathway analysis of genes expressed in common between these MSC populations identified genes involved in immunity, cell adhesion, BMP signalling, chemokine signalling, homeodomain transcription factors, metalloproteinases and their TIMP inhibitors (Table 1).