Transcriptional profile of native CD271+ multipotential stromal cells: Evidence for multiple fates, with prominent osteogenic and Wnt pathway signaling activity


  • Sarah M. Churchman,

    1. NIHR Leeds Musculoskeletal Biomedical Research Unit and University of Leeds, Leeds, UK
    Search for more papers by this author
    • Dr. Churchman's work was supported by the NIHR Leeds Musculoskeletal Biomedical Research Unit.

  • Frederique Ponchel,

    1. NIHR Leeds Musculoskeletal Biomedical Research Unit and University of Leeds, Leeds, UK
    Search for more papers by this author
  • Sally A. Boxall,

    1. NIHR Leeds Musculoskeletal Biomedical Research Unit and University of Leeds, Leeds, UK
    Search for more papers by this author
    • Dr. Boxall's work was supported by the European Union Seventh Framework Programme (PurStem project; 223298).

  • Richard Cuthbert,

    1. NIHR Leeds Musculoskeletal Biomedical Research Unit and University of Leeds, Leeds, UK
    Search for more papers by this author
  • Dimitrios Kouroupis,

    1. NIHR Leeds Musculoskeletal Biomedical Research Unit and University of Leeds, Leeds, UK
    Search for more papers by this author
  • Tarek Roshdy,

    1. NIHR Leeds Musculoskeletal Biomedical Research Unit and University of Leeds, Leeds, UK
    Search for more papers by this author
  • Peter V. Giannoudis,

    1. NIHR Leeds Musculoskeletal Biomedical Research Unit and University of Leeds, Leeds, UK
    Search for more papers by this author
  • Paul Emery,

    1. NIHR Leeds Musculoskeletal Biomedical Research Unit and University of Leeds, Leeds, UK
    Search for more papers by this author
  • Dennis McGonagle,

    Corresponding author
    1. NIHR Leeds Musculoskeletal Biomedical Research Unit and University of Leeds, Leeds, UK
    • NIHR Leeds Musculoskeletal Biomedical Research Unit, LIMM, University of Leeds, Leeds LS9 7TF, UK
    Search for more papers by this author
    • Drs. McGonagle and Jones' work was supported by the Wellcome Trust/Engineering and Physical Sciences Research Council through WELMEC, a Centre of Excellence in Medical Engineering (grant WT088908/Z/09/Z), and in part by the the NIHR Leeds Musculoskeletal Biomedical Research Unit.

  • Elena A. Jones

    1. NIHR Leeds Musculoskeletal Biomedical Research Unit and University of Leeds, Leeds, UK
    Search for more papers by this author
    • Drs. McGonagle and Jones' work was supported by the Wellcome Trust/Engineering and Physical Sciences Research Council through WELMEC, a Centre of Excellence in Medical Engineering (grant WT088908/Z/09/Z), and in part by the the NIHR Leeds Musculoskeletal Biomedical Research Unit.



Controversy surrounds the identity and functionality of rare bone marrow–derived multipotential stromal cells (BM-MSCs), including their differentiation capabilities, their relationship to pericytes and hematopoiesis-supporting stromal cells, and the relevance of their culture-expanded progeny in studies of skeletal biology and development of cell-based therapies. The aim of this study was to clarify the nature of candidate BM-MSCs by profiling transcripts that reflect different aspects of their putative functions in vivo.


Rare, sorted BM-derived CD45−/low CD271bright (CD271) cells were analyzed using 96-gene expression arrays focused on transcripts relevant to mesenchymal-lineage differentiation (toward bone, cartilage, fat, or muscle), hematopoietic and stromal support, and molecules critical to skeletal homeostasis. These cells were compared to matched CD45+ CD271− hematopoietic-lineage cells, culture-expanded MSCs, and skin fibroblasts. When feasible, transcription was validated using flow cytometry.


CD271 cells had a transcriptional profile consistent with the multiple fates of in vivo MSCs, evident from the observed simultaneous expression of osteogenic, adipogenic, pericytic, and hematopoiesis-supporting genes (e.g., SP7 [osterix], FABP4 [fatty acid binding protein 4], ANGPT1 [angiopoietin 1], and CXCL12 [stromal cell–derived factor 1], respectively). Compared to culture-expanded MSCs and fibroblasts, CD271 cells exhibited greater transcriptional activity, particularly with respect to Wnt-related genes (>1,000-fold increased expression of FRZB [secreted frizzled-related protein 3] and WIF1 [Wnt inhibitory factor 1]). A number of transcripts were identified as novel markers of MSCs.


The native, BM-derived in vivo MSC population is endowed with a gene signature that is compatible with multiple functions, reflecting the topographic bone niche of these cells, and their signature is significantly different from that of culture-expanded MSCs. This indicates that studies of the biologic functions of MSCs in musculoskeletal diseases, including osteoporosis and osteoarthritis, should focus on in vivo MSCs, rather than their culture-adapted progeny.

Bone marrow (BM)–derived multipotential stromal cells (also variously known as mesenchymal stem cells, multipotential mesenchymal stromal cells, or marrow stromal cells; all designated MSCs) are highly proliferative cells capable of generating all skeletal tissues, including bone and cartilage (1, 2). As a result of their perceived rarity in vivo, it was thought that MSCs could only be used following their expansion in culture, for the development of novel therapies in regenerative medicine or for functional studies. Indeed, culture-expanded MSCs may be especially useful for regeneration of bone tissue (2), and they have also provided the theoretical foundation for the study of MSC-related biologic processes. However, the in vivo identity and molecular signature of MSCs in the native state, prior to culture expansion remains incompletely defined (1, 2).

Cell fractions enriched for nonexpanded BM-MSCs have demonstrated excellent efficacy when used for bone repair in animal models (3, 4) and in clinical studies (5). However, the molecular mechanisms of such activity, with respect to direct differentiation and host cell interaction, remain to be elucidated.

Possibly of even greater relevance to the pathophysiology and study of bone and joint diseases is the fact that culture-expanded MSCs could be poor surrogates of native, bone-resident MSCs. Indeed, despite the recognized genetic and functional differences that underlie various bone phenotypes in osteoarthritis and osteoporosis (6, 7), the vast majority of the studies exploring MSC fitness in musculoskeletal diseases have often failed to reveal any significant differences (8–10). Given that key pathways known to be central to bone phenotypes, including the Wnt and bone morphogenetic protein (BMP) signaling pathways, may be aberrantly activated or silenced during culture adaptation of the cells as a result of the supraphysiologic proliferation that accompanies in vitro expansion, it may be no surprise to find that use of cultured MSCs in different disease states has been relatively uninformative. Furthermore, in vitro expansion of MSCs is associated with gradual accumulation of senescent cells (11), telomere erosion (12), and changing phenotypes (13, 14).

In addition to being endowed with remarkable tissue-repair capabilities, BM-MSCs are also capable of providing hematopoietic support (15), which is thought to be under the guise of adventitial reticular cells (16). Indeed, studies have suggested that MSCs occupy a pericyte niche and are, in fact, part of this stromal lineage (1, 17).

Based on previously reported strong phenotypic evidence indicating that the CD45−/lowCD271bright (CD271) cell population represents the candidate BM-MSC population in vivo (18–21), this study was designed to investigate the transcriptional signature of CD271 cells in order to address controversies relating to their in vivo identity and the relevance of their culture-expanded progeny to MSC-related biologic processes. As highlighted above, this candidate MSC population has the topography of hematopoietic-supportive adventitial reticular cells (22), which have been likened to “bona fide specialized pericytes of venous sinusoids in the marrow” (16). Similar to colony-forming unit–fibroblasts (CFU-Fs), CD45−/lowCD271bright cells are rare in BM aspirates (average frequency 0.02%) and express numerous MSC/osteoprogenitor markers, including CD73, CD105, and CD146 (13, 18, 19, 21, 23).

Bearing in mind the uncertainties pertaining to the identity of MSCs in vivo, we studied a selected panel of gene transcripts in order to address the following issues: 1) whether a gene profile compatible with multiple stromal- and skeletal-lineage differentiation potentials is present in the CD271 cell population; 2) whether BM-MSCs in vivo have a signature that defaults to their native bone residency; 3) to what degree the molecular signature of culture-expanded MSCs differs from that of in vivo MSCs; and 4) whether these differences in MSCs, with respect to key gene family members, are potentially pivotal in skeletal biologic processes. To this end, a 96-gene expression array was designed to investigate sorted CD271 cells for the expression of mesenchymal-lineage–related genes, pericyte–related genes, and stroma-specific molecules, with particular emphasis on markers of osteogenesis, and to compare these cells with culture-expanded MSCs, osteogenically differentiated MSCs, and negative control skin fibroblasts. Our findings support the notion of a multipotential CD271+ MSC population that has broad differentiation potentials in vivo, bearing a profile that is fundamentally different from that of its culture-adapted progeny. These findings may thus have further implications for the study of MSCs in rheumatic diseases and orthopedic settings.


Isolation of CD271+ cells for gene expression analysis.

Ethics permission for the collection of BM specimens was obtained from the Leeds NHS Trust Ethics Committee. Iliac crest BM aspirates were obtained from 8 healthy donors (median age 33.5 years, range 2–60 years). Mononuclear cells were isolated from the aspirates using a Lymphoprep density gradient (Axis-Shield), and cells positive for CD271 were selected with Anti-Fibroblast Microbeads (Miltenyi Biotec), according to the manufacturer's protocol. The CD45−/low CD271bright (CD271) cell population (mean of 4,903 cells) was then purified by fluorescence-activated cell sorting (FACS) in a MoFlo cell sorter (Dako), using mouse anti-human fluorescein isothiocyanate (FITC)–conjugated CD45 antibodies (Dako) and phycoerythrin (PE)–conjugated CD271 antibodies (BD Biosciences), along with 7-aminoactinomycin D (Sigma) to eliminate dead/dying cells, as previously described (13, 18). Donor-matched hematopoietic-lineage cells (HLCs) were sorted by identifying cells with the CD45+CD271− phenotype.

Generation of standard culture-expanded MSCs and control skin fibroblasts.

Iliac crest BM aspirates were obtained from 7 age-matched healthy donors (median donor age 35 years, range 2–72 years); of these, 4 samples were also used in CD271 cell selection. MSCs in 100 μl of BM aspirate were grown in culture, consisting of clinical-grade defined medium (α-minimum essential medium containing 10% fetal bovine serum [PAA Laboratories]) and 1 ng/ml fibroblast growth factor 2 (PeproTech) (24), followed by seeding into 98-mm dishes and expansion for 21 days. This medium has been compared to a research-grade nonhematopoietic culture medium that was previously optimized for MSC growth (Miltenyi Biotec), and no differences in cell growth rate or transcriptional profile were found (Supplementary Figure 1, available at The mean in vitro age of the culture-expanded MSCs was 15 population doublings (PDs) (range 14–16 PDs), calculated as follows: PD = log2(cell count on day 21/CFU-Fs on day 0), where CFU-Fs were determined from duplicate dishes similarly seeded on day 0. Visualization was by crystal violet staining on day 14 (25).

The culture-expanded MSCs were used directly for quantitative real-time polymerase chain reaction (qPCR). In addition, some cultures of MSCs (n = 7 donors; median donor age 28.5 years, range 2–72 years) were placed under osteogenic differentiation conditions, as described below. Skin fibroblast lines CRL-2068 and HFF1 were purchased from ATCC, and line NHDF was purchased from Lonza.

Generation of culture-expanded MSCs from CD271-selected cells and comparison with standard culture-expanded MSCs.

A CD45−/lowCD271+ cell population was FACS-purified from mononuclear cells in the same manner as described above (n = 6 donors; median age 44.5 years, range 15–64 years). Sorted cells were seeded into fibronectin-coated plates (BD Biosciences) in the defined medium, at a seeding density of 180 cells/cm2, and expanded for 21 days. Gene expression in these cultures was compared to that in standard culture-expanded MSCs from the same donors (Supplementary Figure 2, available at

Osteogenic and adipogenic differentiation of MSCs.

Osteoblast and adipocyte colony formation assays were performed with purified CD271 cells, as previously described, using standard differentiation media (13). Osteogenically differentiated MSCs were used as positive controls to study the involvement of newly discovered transcripts in MSC osteogenesis, and were generated in clonogenic osteogenic differentiation assays (13, 26). Briefly, 500 culture-expanded MSCs were seeded into 4 replicate 30-mm dishes in defined medium. On day 8, colonies were washed in phosphate buffered saline (PBS) and the medium was changed to standard osteogenic medium (13); one-half of the medium was changed twice weekly. After 21 days, calcium production assays (n = 3 dishes) were carried out, and calcium levels were determined following HCl extraction using calcium liquid (Sentinel Diagnostics), as previously described (13, 18).

Osteogenically differentiated MSCs from the fourth dish were used for qPCR. The fold increase in transcript levels in MSCs from different donors (osteogenically differentiated MSCs relative to culture-expanded MSCs) was assessed for correlations with the levels of calcium production by osteogenically differentiated MSCs. The adipogenic differentiation assays were performed as described previously (18).

Quantitative real-time PCR.

Sorted cells, trypsinized culture-expanded MSCs, osteogenically differentiated MSCs, and skin fibroblasts were washed in PBS prior to RNA isolation using the Norgen Biotek RNA/DNA Protein Kit (Geneflow). Complementary DNA (cDNA) was reverse transcribed using a High Capacity cDNA Reverse Transcription Kit for use on a Custom TaqMan Array (format 96a; both from Applied Biosystems). Exon-spanning, ‘3′ most’ TaqMan assays were selected for the array when possible (Supplementary Table 1, available at Based on the manufacturer's recommendation, 200 ng cDNA was used per port (2 ports/sample), when possible; alternatively, all available cDNA were used.

Quantitative analysis was carried out with the 2math image method, using Data Assist version 2.0 (Applied Biosystems), with results normalized to the values for the reference gene, HPRT. Cluster analysis was performed using open-source Cluster software (, utilizing clustering methods as described in the study by Eisen et al (27), with results visualized using the TreeView program (

Flow cytometry characterization of CD271 cells.

BM mononuclear cells (n = 8 donors; median donor age 31.5 years, range 9–44 years) were isolated using Lymphoprep, and CD271+ cells were preenriched using Anti-Fibroblast Microbeads. The cells were resuspended at 1 × 107 cells/ml in FACS buffer (PBS/0.5% bovine serum albumin [BSA]/2 mM EDTA) with 10% Fc receptor block (Miltenyi Biotec), and incubated at room temperature for 10 minutes. Antibodies were added at the recommended concentrations, and the cells were incubated for 40 minutes at 4°C. The antibodies used were as follows: PE–Cy7–conjugated CD45 (BD Biosciences), allophycocyanin (APC)–conjugated CD271 (Miltenyi Biotec), FITC-conjugated CD235a (Dako), PE-conjugated transforming growth factor β receptor 3 (TGFβR3) (R&D Systems), PE-conjugated CD73, PE-conjugated CD90, PE-conjugated CD19, FITC-conjugated CD33 (all from BD Biosciences), PE-conjugated CD105, and FITC-conjugated CD31 (both from Serotec). The cells were washed and resuspended in FACS buffer containing 100 ng/ml DAPI.

Live CD45−/lowCD271bright cells and donor-matched HLCs were gated using DAPI/PE–Cy7–conjugated CD45/APC-conjugated CD271. Isotype-specific negative control antibodies were purchased from Serotec and used to measure nonspecific antibody binding. All flow cytometry data were acquired and analyzed on an LSRII flow cytometer (BD Biosciences) using FACSDiva software. Overlay histograms were generated using FlowJo software.

For intracellular staining of FABP4, preenriched cells were initially stained using a Live/Dead Fixable Violet Dead Cell Stain kit (Invitrogen), according to the manufacturer's protocol. Cells were then washed in PBS and resuspended at 2 × 107 cells/ml in PBS/0.5% BSA before surface staining with APC-conjugated CD271 and PE–Cy7–conjugated CD45 antibodies at the recommended concentrations for 15 minutes at room temperature. The cells were then fixed and permeabilized using IntraStain (Dako). For detecting expression of FABP4, cells were sequentially stained with goat anti-human FABP4 antibodies (1:20 dilution; R&D Systems), rabbit anti-goat biotin (1:200; Dako), and streptavidin–FITC (1:50; BD Biosciences), for 15 minutes per antibody at room temperature; PBS washes were performed at each step. FABP4 expression was analyzed on CD45−/lowCD271bright cells and HLCs (CD45+CD271−). A minimum of 50,000 events were collected in each file.

Flow cytometry characterization of cultured MSCs.

Culture-expanded MSCs were lifted by trypsinization, and single-cell suspensions were stained with antibodies/isotype controls as previously described (13, 18). Briefly, cells were resuspended in FACS buffer and used for flow cytometry at 105 cells/test. For the analysis of the specificity of FABP4 staining, culture-expanded MSCs were induced to differentiate toward adipogenesis and osteogenesis according to standard protocols (18). FABP4 expression was analyzed on days 3, 7, 14, and 21 postinduction. Dead cells/debris were gated out based on exclusion by DAPI staining.

Statistical analysis.

Intergroup differences were tested using Kruskal-Wallis analysis, corrected with the Bonferroni-Dunn multiple-group comparison test, with results expressed as the median and interquartile range (presented as box and whisker plots, with boxes representing the upper and lower quartiles of the median, and whiskers representing the maximum and minimum values), calculated using GraphPad Prism version 5 software. Correlations were determined using Spearman's rank correlation analyses, with data analyzed using SPSS version 16.0.


Cell populations.

Consistent with the findings from previous studies (23, 28), sorted CD45−/low CD271bright (CD271) cells (purity >95%), when compared with CD45+CD271− HLCs, were larger and morphologically homogeneous (Figure 1A). As previously reported (13, 21, 29), CD271 cells were highly proliferative and contained clonogenic, adipogenic, and osteogenic progenitors (Figure 1B, left panel). A double-negative population (CD45−CD271− cells) was enriched for erythroblasts and lacked clonogenicity (results not shown), as has been previously reported (21, 30). CD271 cells were positive for CD73, CD105, and CD90, whereas expression of these markers on HLCs was negligible (Figure 1B, middle panel).

Figure 1.

Purification and characterization of CD271 cells as well as pericyte marker expression in CD271 cells, as compared to hematopoietic-lineage cells (HLCs) and culture-expanded multipotential stromal cells (cMSCs). A, Purification of CD271 cells (CD45−/lowCD271bright) and HLCs (CD45+CD271−) from bone marrow (BM) aspirates. Photomicrographs illustrate the homogeneous morphology of the sorted CD271 cells and heterogeneous morphology of the HLCs. B, Confirmation of the MSC-specific nature and purity of the sorted CD271 cells. Left panel, The clonogenicity and multipotentiality of the clones derived from 500 sorted CD271 cells were visualized by staining with crystal violet (top left), oil red (top right), alkaline phosphatase (bottom left), and alizarin red (bottom right). Middle panel, The MSC-specific phenotype of CD271 cells, when compared to HLCs, was confirmed by specific marker expression. Bars show the mean ± SD of 3 samples per group. Right panel, The purity of the sorted fractions was confirmed by gene expression analysis for NGFR (CD271); results are expressed relative to the values for HPRT. C, Pericyte-specific transcript expression. Results in right panel of B and in C are shown as box plots. Each box represents the upper and lower quartiles. Lines inside the boxes represent the median. Whiskers represent the minimum and maximum values. Triangles indicate data detected in a single donor, while values for the remaining donors were below detection, thus precluding full statistical analysis of the respective data set. ∗ = P < 0.05; ∗∗ = P < 0.01; ∗∗∗ = P < 0.001.

Culture-expanded MSCs generated from FACS-purified CD271 cells had a molecular profile identical to that of culture-expanded MSCs generated from standard plastic adherence (n = 6 donors; Supplementary Figure 2, available at Therefore, the latter cells were used for all subsequent analyses, representing the current method used in most laboratories. In previous analyses, culture-expanded MSCs were found to express CD73, CD105, and CD90 (13, 24) but lacked expression of CD271 (18, 23, 29). In the present study, this expression profile was evident at the transcriptional level, as shown by the higher expression of NGFR (CD271) in CD271 cells, compared to the down-regulated expression of NGFR in culture-expanded MSCs and low expression of NGFR in HLCs (Figure 1B, right panel), confirming the purity of the sorted fractions.

Pericyte transcript profile of the CD271 cells.

Consistent with the hypothesis of an overlapping identity of MSCs and pericytes (1, 31), the reported pericyte-specific transcripts MCAM (CD146), ANGPT1 (angiopoietin 1), ACTA2 (smooth muscle actin), CSPG4 (chondroitin sulfate proteoglycan 4/neuron-glial antigen 2), and PDGFRA (platelet-derived growth factor receptor A) (17, 31, 32) were expressed at higher levels in CD271 cells compared to HLCs (Figure 1C). This provided molecular support for the common identity of CD271 cells and marrow sinusoidal pericytes that express CD271 in situ (21, 22).

Transcript categories based on differential expression among CD271 cells, HLCs, and culture-expanded MSCs.

Gene expression in the cells was analyzed in arrays using a panel of 96 genes (including 3 references genes). Fifty-one genes were significantly overexpressed in CD271 cells when compared to donor-matched HLCs. These mesenchymal-lineage genes were further subdivided into 2 groups: 1) genes whose expression was similar between CD271 cells and culture-expanded MSCs (n = 27 genes) (designated MSC markers) (Table 1), and 2) genes whose expression was significantly lower in culture-expanded MSCs (n = 24 genes) (designated CD271-specific markers (Table 2). A further 10 genes were identified as being additional CD271-specific markers; the expression of these genes was significantly lower in culture-expanded MSCs compared to CD271 cells, but was nondiscriminatory between CD271 cells and HLCs (Table 2).

Table 1. Multipotential stromal cell (MSC) markers*
GeneHLCsCulture-expanded MSCs
Fold decrease relative to CD271 MSCsPFold decrease relative to CD271 MCSsP
  • *

    Gene expression in hematopoietic-lineage cells (HLCs) was significantly decreased, or not present, compared to that in CD271 MSCs, whereas differences were not significant (NS) in culture-expanded MSCs relative to CD271 MSCs. NT = not tested where gene levels were either low (low detection [LD]) or below detection (BD).

  • Gene expression was correlated with age.

F208LD in HLCsNT3.0NS
Table 2. CD271-specific multipotential stromal cell (MSC) markers*
 HLCsCulture-expanded MSCs
Fold decrease relative to CD271 MSCsPFold decrease relative to CD271 MSCsP
  • *

    For CD271-specific MSC markers, gene expression was significantly decreased, or not present, both in hematopoietic-lineage cells (HLCs) and in culture-expanded MSCs relative to CD271 MSCs. For the additional CD271 markers, significant differences in gene expression were observed in culture-expanded MSCs relative to CD271 MSCs, whereas differences were not significant (NS) in HLCs relative to CD271 MSCs. NT = not tested where gene levels were either low (low detection [LD]) or below detection (BD).

  • Gene expression was correlated with age.

CD271-specific markers    
 BMP5LD in HLCsNTBD in culture-expanded MSCsNT
 PDGFRLLD in HLCsNT46.7<0.01
 OMDLD in HLCsNT34.5<0.05
 SP7BD in HLCsNTLD in culture-expanded MSCsNT
 CDH597.5<0.05BD in culture-expanded MSCsNT
Additional CD271 markers    

The remaining genes were, overall, considered nondiscriminatory (n = 14 genes) or were not expressed (levels below detection; n = 18 genes) (Supplementary Table 2, available at Among the genes not expressed, these included TERT (telomerase reverse transcriptase) and NEUROD1 (neurogenic differentiation 1) as well as myogenesis-related markers such as MYF5 (myogenic factor 5), MYOD1 (myogenic differentiation 1), and MYOG (myogenin). This lack of myogenic transcripts is noteworthy, given that this MSC lineage cannot be consistently and robustly generated in the laboratory, when compared to fat-, bone-, and cartilage-lineage markers. Hierarchical clustering highlighted these transcript categories and confirmed the similarity of gene expression profiles between CD271 cells and culture-expanded MSCs, but not HLCs (Figure 2).

Figure 2.

Hierarchical clustering analysis (single-linkage clustering) of gene expression data from quantitative polymerase chain reaction analyses of CD271 cells, culture-expanded multipotential stromal cells (cMSCs), and hematopoietic-lineage cells (HLCs) in bone marrow aspirates from age-matched donors. Log2 transformation and data filtering (filter = 67% present) were performed according to the methods described in the Cluster&TreeView Manual, resulting in identification of 54 of the 96 genes tested. Scores were assigned as follows: black = 1, red = >1, green = <1; grey = missing data (below detection).

Markers indicative of the simultaneous stromal function and multipotentiality of CD271 cells.

We next explored the stromal-supporting potential and multipotentiality of CD271 cells using a range of stroma-, adipogenesis-, chondrogenesis-, and osteogenesis-related genes (Figure 3A). High-level expression of genes encoding hematopoiesis-supporting cytokines (CXCL12 [stromal cell–derived factor 1], IL7 [interleukin-7]) was observed in CD271 cells, along with high-level expression of transcripts traditionally associated with adipogenesis (PPARG [peroxisome proliferator–activated receptor γ], LPL [lipoprotein lipase], FABP4 [fatty acid binding protein 4]) and with osteogenesis (SP7 [osterix], SPARC [osteonectin], SPP1 [osteopontin], BGLAP [osteocalcin]). The expression of stroma-, adipogenesis-, and osteogenesis-related markers was commonly lower in culture-expanded MSCs compared to CD271 cells (Figure 3A).

Figure 3.

Markers indicative of the multipotentiality of CD271 cells. A, Expression of stroma-, adipogenesis-, chondrogenesis-, and osteogenesis-related genes, relative to the values for HPRT, was assessed in CD271 cells, hematopoietic-lineage cells (HLCs), and culture-expanded multipotential stromal cells (cMSCs). B, Expression of FABP4 was assessed in each cell group. Left panel, Percentage of FABP4-positive cells, shown as the mean ± SD of 3 samples per group. Middle panels, Representative histograms, confirming FABP4 expression by flow cytometry. Right panel, Up-regulated expression of FABP4 during adipogenic differentiation of MSCs (solid diamonds), but not during osteogenic differentiation (open diamonds). C, Expression of TGFBR3 was assessed in each cell group. Left panels, Comparison of TGFBR3 gene expression between the groups, and correlation of TGFBR3 gene expression with the extent of Ca++ production in osteogenically differentiated MSCs. Middle panel, Percentage of TGFBR3-positive cells, shown as the mean ± SD of 3 samples per group. Right panels, Representative histograms, confirming TGFBR3 expression by flow cytometry. Empty histogram = marker; filled histogram = isotype control. Results in A and far left panel of C are shown as box plots. Each box represents the upper and lower quartiles. Lines inside the boxes represent the median. Whiskers represent the minimum and maximum values. Triangles indicate data from single samples, while the remaining values were below detection, thus precluding full statistical analysis of the respective data set. ∗ = P < 0.05; ∗∗ = P < 0.01; ∗∗∗ = P < 0.001.

The chondrogenesis-related transcription factor SOX9 was expressed in CD271 cells (Figure 3A), but the mature markers COL2A1 and EPYC were absent (Supplementary Table 2, available at, suggesting that CD271 cells were ready for, but not actively engaged in, chondrogenic differentiation. This is in contrast to the high-level expression of transcripts for late-stage adipogenic proteins (LPL, FABP4) and osteogenic proteins (SPARC, SPP1, BGLAP) in CD271 cells, indicating their steady-state engagement in adipogenesis and osteogenesis. Indeed, when cells were stained with antibodies against FABP4, a continuum of FABP4-positive cells was observed in CD271 cells, suggesting that there were differing levels of adipogenic commitment, whereas control cells (HLCs and culture-expanded MSCs) were uniformly negative for FABP4 (Figure 3B). The specificity of FABP4 staining was further confirmed using adipogenically differentiated MSCs, while staining was negligible in osteogenically differentiated MSCs (Figure 3B, far right panel).

Flow cytometry confirmed the pattern of gene expression of the osteogenesis-related surface receptor TGFBR3 (Figure 3C). The gene expression of TGFBR3 was elevated in osteogenically differentiated MSCs, and this was significantly correlated with the levels of calcium deposition by osteogenically differentiated MSCs from the same donors, thus confirming the value of TGFBR3 in identifying osteogenically committed cells.

An overlap in the expression of TGFBR3 was demonstrated by flow cytometry in CD271-positive and CD271-negative subpopulations, indicating differing levels of osteogenic commitment between individual CD271 cells. The observed coexistence of both adipogenically and osteogenically committed cells within the CD271 cell population suggests that cell sorting can be used for future study of clonogenicity and multipotentiality of these subpopulations. With our use of relevant commercially available antibodies in flow cytometry, the results confirmed the adipogenic and osteogenic gene expression in CD271 cells at the protein level.

Differences in expression of bone-related and Wnt pathway–related transcripts in CD271 cells and culture-expanded MSCs.

As shown in Table 2, there were some striking differences between CD271 cells and culture-expanded MSCs in the patterns of gene expression for key bone-related molecules, especially in the Wnt and BMP pathways, both of which are known to be intimately associated with bone disease phenotypes in humans and in experimental models. For example, the expression of the BMP2 gene was >20-fold lower in culture-expanded MSCs compared to CD271 cells, and BMP5 expression was reduced to undetectable levels. The expression of OMD (osteomodulin), an osteoblast maturation marker that is induced by osteoclast activity (33), was similarly lower (∼35-fold) in culture-expanded MSCs compared to CD271 cells. The other 2 most significantly down-regulated molecules in culture-expanded MSCs were the Wnt pathway inhibitors FRZB (secreted frizzled-related protein 3) (>1,200-fold) and WIF1 (Wnt inhibitory factor 1) (>4,200-fold) (Table 2). In contrast, expression of the osteoblast markers COL1A1 and COL1A2 (34, 35) was reduced only slightly (P not significant) (Table 1).

Given the high expression levels of FRZB, SFRP1 (secreted frizzled-related protein 1), and SFRP4 (secreted frizzled-related protein 4) transcripts on CD271 cells and the importance of the Wnt pathway in osteogenesis (36), we next investigated the expression of selected Wnt pathway receptors and molecules that have been previously reported to act as antagonists in CD271 cells, and compared their expression to that in culture-expanded MSCs and negative control skin fibroblasts (Figure 4A). With the exception of FZD7 (frizzled homolog 7) and DVL2 (dishevelled homolog 2), the transcripts for all other tested molecules were up-regulated in CD271 cells compared to culture-expanded MSCs. Excluding DVL2 and FZD7, Wnt transcript expression in fibroblasts was, on average, 50- and 10-fold lower compared to that in CD271 cells and culture-expanded MSCs, respectively. The expression of the Wnt target genes NANOG (Nanog homeobox) and POU5F1 (POU class 5 homeobox 1/Oct4) (37) was the highest in CD271 cells (Figure 4A).

Figure 4.

Expression of transcripts for Wnt pathway genes and newly discovered genes, as well as gene expression related to aging, in CD271 cells compared to hematopoietic-lineage cells (HLCs), culture-expanded multipotential stromal cells (cMSCs), or skin fibroblasts. A, Gene expression of Wnt-related molecules, relative to the values for HPRT, was assessed in CD271 cells (diamonds), culture-expanded MSCs (squares), and fibroblasts (circles). B, Expression of SFRP1 and LEPR was age dependent in CD271 cells (diamonds), but not in culture-expanded MSCs (squares). C, Expression of novel MSC markers was assessed in each cell group (top panels), and levels were correlated with the extent of Ca++ production in osteogenically differentiated MSCs. Results in top panels of C are shown as box plots. Each box represents the upper and lower quartiles. Lines inside the boxes represent the median. Whiskers represent the minimum and maximum values. ∗ = P < 0.05; ∗∗ = P < 0.01; ∗∗∗ = P < 0.001.

The high-level expression of Wnt receptors and Wnt pathway inhibitors in CD271 cells provided evidence of their readiness for stimulation by Wnts and for potential autocrine regulation of the Wnt pathway in these cells. Furthermore, these findings indicate that Wnt signaling activity is higher in MSCs compared to fibroblasts, particularly in their native state (CD271 cells).

Age-related changes in transcript expression in CD271 MSCs.

The expression of SFRP1 (Figure 4B) and several other Wnt-related transcripts (Tables 1 and 2) in CD271 cells appeared to be age-related, suggesting that aging of MSCs in vivo may be associated with a gradual reduction in Wnt signaling activity. This finding is complementary to previous observations, on shorter telomeres, in BM-MSCs from older donors (12, 13). The expression of LEPR (leptin receptor) in CD271 cells was also age-related (Figure 4B); this is interesting, given the known role of leptin in controlling the bone–fat balance in the marrow (38), which changes with age (39). Age-related trends in selected transcript expression were lost in culture-expanded MSCs (Figure 4B). This may be because culture-expanded MSCs artificially “age” in vitro, by acquiring successive, cumulative changes as they move toward senescence during culture expansion (11).

Novel BM-MSC molecules in vivo and in culture.

We looked at additional molecules known to be important in fibroblast-, osteoblast-, and stromal-lineage cell biology in vivo to ascertain their expression on CD271 MSCs and on culture-adapted progeny. Expression of the following molecules was observed for the first time in CD271 cells: COL1A2, PCOLCE (procollagen C–endopeptidase enhancer), TNFRSF11B (tumor necrosis factor receptor superfamily, member 11b/osteoprotegrin), GJA1 (gap junction protein α1, 43 kd/connexin 43), IGF2 (insulin-like growth factor 2), IGFBP3 (insulin-like growth factor binding protein 3), CDH11 (cadherin 11), PDGFRL (platelet-derived growth factor receptor-like), DDR2 (discoidin domain receptor 2), and ANGPTL4 (angiopoietin-like 4) (Tables 1 and 2). Expression of CDH11, DDR2, IGFBP3, and IGF2 was increased in osteogenically differentiated MSCs, in close correlation with the levels of calcium production (Figure 4C), suggesting the role of these molecules in osteoblast differentiation; this was not evident for GJA1. Among the sorted fractions, the expression levels of all of these molecules were significantly lower in HLCs (Figure 4C), indicating their utility as novel in vivo MSC markers.


The purpose of this study was to explore the basis for the apparent simultaneous multipotentiality and stromal hematopoietic-supportive role of BM-MSCs and to further relate these in vivo patterns of gene expression to the transcriptional profile of culture-expanded MSCs, the latter of which is the current gold standard for molecular, functional, and putative therapeutic targeting. Our results show, for the first time, simultaneous osteogenic, adipogenic, pericytic, and hematopoiesis-supporting transcriptional activity in CD271 cells, thus confirming, at the molecular level, their multipotential stromal nature. The hierarchical clustering of CD271 cells with culture-expanded MSCs and the widespread expression of pericyte markers further support the notion that multipotentiality is inextricably linked with pericyte function, which could lead to better understanding of vessel wall calcification in autoimmune muscle disease and scleroderma.

The CD271 cell population had fundamental differences in the transcriptional profile of the BMP- and Wnt pathway–related molecules, when compared to that of culture-expanded MSCs. Collectively, these pathways are known to be crucial to skeletal homeostasis in health and disease. Given that it is now possible to isolate CD271 cells to high purity (21) and in large numbers (13), these findings have implications for biologic studies of MSCs in bone disease, including osteoarthritis and osteoporosis, to which polymorphisms in the Wnt pathway molecules have been linked (6, 7, 40).

Apart from the lower leptin-binding capacity of BM-MSCs in osteoporosis (41), early functional and molecular studies of culture-expanded MSCs in these diseases, compared to MSCs in the healthy state, revealed little abnormality (8–10). It is clear from our data that culture-expanded MSCs may only poorly resemble their in vivo counterparts in terms of the expression of these genes. Therefore, extracting and studying CD271 cells from different disease environments is likely to show more substantial differences in their molecular profiles, which may have value in defining different molecular cascades that lead to disease phenotypes. Since both osteoporosis and osteoarthritis are also considered diseases of aging, the observed decreases in some Wnt-related transcripts in CD271 cells from older individuals are additionally noteworthy.

Simultaneous expression of genes characteristic of multiple mesenchymal lineages and providing stromal support for hematopoiesis in CD271 cells can be explained by the coexistence of nonoverlapping subpopulations of osteo- and adipogeneically committed cells and cytokine-producing cells. Based on our FABP4 and TGFBR3 flow cytometry data, however, such a clear-cut subdivision appears rather unlikely. Consistent with our findings, the existence of CD271 cells that simultaneously express multiple lineage markers has been recently shown by single-cell multiplex reverse transcription–PCR for a limited number of transcripts (21). Therefore, both sets of data suggest the presence of a rather dominant cell population with “promiscuous” gene expression, a notion consistent with some stem cell definitions (42). Subdivision of CD271 cells into CD146+ and CD146− subpopulations was recently shown to be related to their topography and local oxygen levels, rather than any intrinsic functional difference between cells (21). In a similar manner, physiologic requirements for bone remodeling, and hence the production of osteoblasts, may dictate the precise balance between osteogenically and adipogenically driven CD271 cells. The lack of mature chondrogenic gene marker expression in CD271 cells is consistent with the fact that in adults, chondrogenesis occurs only following trauma, i.e., during endochondral ossification (43), whereas adipogenesis and bone remodeling occur physiologically under steady-state conditions (44).

In this study, we also identified several novel molecules whose expression sheds more light on the multiple functions of CD271 MSCs in vivo. Commonly referred to as osteoblastic cadherin, CDH11 has been shown to be expressed in both preosteoblast and preadipocyte cell lines, which declines markedly during adipogenesis and concordantly rises during osteogenesis (45). In our study, it was found to be a good MSC marker, potentially applicable in the investigation of MSCs in the synovium (46). GJA1 (gap junction protein α1) and JAG1 (jagged 1) regulate hematopoietic cell proliferation and maturation (47, 48). S1PR1 (sphingolipid G protein–coupled receptor 1) is the G protein–coupled receptor engaged in the egress of newly formed B cells from the BM into the blood (49) as well as trafficking of osteoclast precursors (50). TNFRSF11B (osteoprotegerin) (51) and EPHB4 (EPH receptor B4) (52) are involved in osteoblast–osteoclast interactions. The expression of these molecules in CD271 cells provides further insight into the role of native MSCs not only in osteoblast-mediated bone formation, but also in osteoclast-mediated bone breakdown. As far as we are aware, no known PDGFRL, DDR2, or ANGPT4 function has yet been described in relation to bone or stromal physiology. BMP2 is known to be produced by megakaryocytes and platelets, but it can also be strongly expressed by mesenchymal cells during fracture repair (53). IGF2 signaling has been recently shown to be an important mechanism in triggering osteogenic differentiation in human MSCs (54). To the best of our knowledge, no previous study has shown significantly higher expression of BMP2 and IGF2 in native MSCs compared to culture-expanded MSCs. Our results, showing strong BMP2 expression in CD271 cells, suggest that local injections of freshly harvested autologous CD271 cells may be an effective adjuvant therapy for fracture repair, instead of rather expensive pharmacologic BMP-2. Overall, this study provides a molecular basis for the superior osteogenic capacity of native MSCs compared to culture-expanded MSCs, as described in in vivo studies (3, 5). This further supports the concept of one-stage procedures for bone tissue regeneration, based on prospectively selected, uncultured MSCs.

In conclusion, this study evaluated gene expression in CD271 cells in the BM of healthy donors and confirmed their in vivo MSC identity. These findings highlight the value of using CD271 cells, rather than culture-expanded MSCs, in future studies focused on the effects of aging on MSCs in health and in abnormal conditions, such as osteoporosis, osteopenias, osteoarthritis, and other skeletal diseases. Beyond the rheumatic diseases, this work also provides a platform from which to address potential dysfunctions in resident MSCs in hematologic diseases, including myelodysplasias, myeloma, and marrow failure. Cell-tracking experiments in relevant animal models (55) would shed new light on the normal physiologic functions of MSCs in vivo and their responses to acute or chronic injury. Collectively, these results highlight the relevance of studying native multipotential stromal cells prior to culture manipulation, which could further our understanding of their biologic impact in humans.


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. McGonagle had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Churchman, Ponchel, Emery, McGonagle, Jones.

Acquisition of data. Churchman, Boxall, Cuthbert, Kouroupis, Roshdy, Giannoudis, Jones.

Analysis and interpretation of data. Churchman, Ponchel, Boxall, McGonagle, Jones.


We thank Mrs. Anne English for optimizing the flow cytometry staining for FABP4, and Drs. Sally Kinsey, Argiris Papathanassopoulos, and Geoff Shenton for providing clinical samples.