Bone marrow-derived mesenchymal stem cells express the pericyte marker 3G5 in culture and show enhanced chondrogenesis in hypoxic conditions


  • Wasim S. Khan,

    Corresponding author
    1. United Kingdom Centre for Tissue Engineering and Wellcome Trust Centre for Cell Matrix Research, Faculty of Life Sciences, Michael Smith Building, University of Manchester, Oxford Road, Manchester M13 9PT, United Kingdom
    • United Kingdom Centre for Tissue Engineering and Wellcome Trust Centre for Cell Matrix Research, Faculty of Life Sciences, Michael Smith Building, University of Manchester, Oxford Road, Manchester M13 9PT, United Kingdom. T: +44 (0) 7971 190720; F: +44 (0) 161 275 5752.
    Search for more papers by this author
  • Adetola B. Adesida,

    1. United Kingdom Centre for Tissue Engineering and Wellcome Trust Centre for Cell Matrix Research, Faculty of Life Sciences, Michael Smith Building, University of Manchester, Oxford Road, Manchester M13 9PT, United Kingdom
    Search for more papers by this author
  • Simon R. Tew,

    1. United Kingdom Centre for Tissue Engineering and Wellcome Trust Centre for Cell Matrix Research, Faculty of Life Sciences, Michael Smith Building, University of Manchester, Oxford Road, Manchester M13 9PT, United Kingdom
    Search for more papers by this author
  • Emma T. Lowe,

    1. United Kingdom Centre for Tissue Engineering and Wellcome Trust Centre for Cell Matrix Research, Faculty of Life Sciences, Michael Smith Building, University of Manchester, Oxford Road, Manchester M13 9PT, United Kingdom
    Search for more papers by this author
  • Timothy E. Hardingham

    1. United Kingdom Centre for Tissue Engineering and Wellcome Trust Centre for Cell Matrix Research, Faculty of Life Sciences, Michael Smith Building, University of Manchester, Oxford Road, Manchester M13 9PT, United Kingdom
    Search for more papers by this author


Bone marrow-derived mesenchymal stem cells are a potential source of cells for the repair of articular cartilage defects. Hypoxia has been shown to improve chondrogenesis in some cells. In this study, bone marrow-derived stem cells were characterized and the effects of hypoxia on chondrogenesis investigated. Adherent bone marrow colony-forming cells were characterized for stem cell surface epitopes, and then cultured as cell aggregates in chondrogenic medium under normoxic (20% oxygen) or hypoxic (5% oxygen) conditions. The cells stained strongly for markers of adult mesenchymal stem cells, and a high number of cells were also positive for the pericyte marker 3G5. The cells showed a chondrogenic response in cell aggregate cultures and, in lowered oxygen, there was increased matrix accumulation of proteoglycan, but less cell proliferation. In hypoxia, there was increased expression of key transcription factor SOX6, and of collagens II and XI, and aggrecan. Pericytes are a candidate stem cell in many tissue, and our results show that bone marrow-derived mesenchymal stem cells express the pericyte marker 3G5. The response to chondrogenic culture in these cells was enhanced by lowered oxygen tension. This has important implications for tissue engineering applications of bone marrow-derived stem cells. © 2010 Orthopaedic Research Society. Published by Wiley Periodicals, Inc. J Orthop Res 28:834–840, 2010

Articular cartilage has very limited intrinsic capacity for repair after injury. Most focal cartilage lesions, left untreated, progress to more extensive lesions and, in the long term, these require joint arthroplasty. Autologous chondrocytes harvested from low weight-bearing areas of articular cartilage are being used for the repair of focal hyaline cartilage defects.1 Although short-term clinical results have been good, evidence suggests some incidence of progressive degenerative changes in the joint.2 This procedure is also accompanied by donor site morbidity, and the limited amount of tissue available necessitates prolonged cell expansion. There is, therefore, interest in alternative sources of adult stem cells for cell-based tissue engineering approaches for cartilage repair.

Cells with stem cell characteristics have been reported in many tissues including the bone marrow.3–6 Bone marrow-derived mesenchymal stem cells (BMSCs) represent a small portion of the cells in the stromal compartment, and conditions for the differentiation of these cells into chondrocytes, osteoblasts, and adipocytes have been used to show that they are multipotent.7 Because of their multipotency and practical access, cells from the bone marrow are of interest as a potential source of cells for the repair of focal cartilage defects in the knee.8 There have been limited reports of human autologous bone marrow stromal cell implantation for cartilage repair.6,9 The procedure involved the harvest of bone marrow stroma from the iliac crest and required culture expansion of cells.

A better understanding of the nature of these mesenchymal stem cells is essential to harness their optimal potential. In previous studies, a minor population of BMSCs has also been found to be positive for the pericyte marker 3G5.10 Pericytes have been shown to have multidifferentiation potential11, 12 and it has been suggested that, if distributed widely with vascular capillaries, pericytes may account for stem cells in other tissues.13 This theory is supported by the observation that many of the tissues from which stem cells have been isolated have good vascularization. In further support of this theory, a subendothelial network of pericyte-like cells has been identified using 3G5 in the vascular bed in many human tissues.14

Mammalian cells are normally cultured in air (containing 20% oxygen) with added 5% carbon dioxide, but some cells, including human bone marrow-derived hematopoetic stem cells, have been reported to proliferate more rapidly in lower oxygen concentrations.15–17 This is not surprising as the bone marrow oxygen tension in vivo is in the range of 4%–7%.18 Grayson et al.19 have shown that hypoxia in human BMSCs results in the ability to maintain a significantly higher number of stem cells, higher levels of stem cell genes, and higher levels of osteoblastic and adipocytic differentiation markers on differentiation. They did not look at the effects of hypoxia on the chondrogenic differentiation of human BMSCs. In chondrocyte culture systems, it has been shown that hypoxia results in increased synthesis of extracellular matrix by chondrocytes,20, 21 and this has been extended to stem cells from adipose tissue22 and infrapatellar fat pad23 undergoing chondrogenesis. Again, this relative hypoxia is normal in vivo where articular cartilage is avascular and exists at reduced oxygen tension of 7% at the surface to 1% in the deep layers.20, 21 There is evidence that hyperoxia may disturb the oxidation-reduction status of the cell.24 Thus, oxygen tension appears to be an important regulatory factor in the proliferation, differentiation, and matrix production of chondrocytes, but few studies have characterized gene expression changes. In our investigation of the potential of BMSCs, we characterized these cells using a number of markers and investigated the gene expression changes that characterized their response to hypoxic conditions in chondrogenic cultures.


Cell Isolation and Culture

Human bone marrow mononuclear cells (accompanied by certificate of analyses for the donor) and mesenchymal stem cell (MSC) media were obtained from a commercial source (Cambrex, Wokingham, UK). The bone marrow was extracted following fully informed consent of three 18–40-year-old patients. The cells were plated in MSC media supplemented with 5 ng/mL rh-FGF-2 at a density of 166,000 cells per cm2 in a T25 cell culture flask (Corning Inc., supplied through Fisher Scientific, Loughborough, UK). Nonadherant cells were removed after 24 h by washing twice in Dulbecco's phosphate buffered saline (DPBS) (Cambrex) and changing the medium. Colonies of BMSCs were observed after 10 days in culture, and confluence was reached after a further 20 days. The media was changed every 3 days. On confluence, the cells were split 1:3. Passage 2 cells were used for cell surface staining. Cultures were maintained at 37°C with 5% CO2 and normal oxygen (20%). Cultured cells from passage 2 were used for cell surface epitope characterization and cell aggregate culture.

Cell Surface Epitope Characterization

Confluent passage 2 cells were stained with a panel of antibodies for cell surface epitopes. This included antibodies to CD13 (aminopeptidase N), CD44 (hyaluronan receptor), CD90 (Thy-1), LNGFR (low affinity nerve growth factor receptor), STRO1 (marker for BMSCs), and CD56 (neural cell adhesion molecule, NCAM) from BD Biosciences (Oxford, UK); CD105 (SH2 or endoglin) and CD34 (marker for hematopoetic cells) from Dako (Ely, UK); and 3G5 (marker for vascular pericytes) courtesy of Dr. Ann Canfield (University of Manchester, UK). The cells were incubated for 1 h with the primary mouse antibodies (undiluted 3G5 and 1:100 dilution for others) followed by FITC-conjugated anti-mouse IgM secondary antibody (1:40 dilution) (Dako). For controls, nonspecific monoclonal mouse IgG antibody (Santa Cruz Biotechnology, Santa Cruz, CA) was substituted for the primary antibody. The cells were incubated with 1:100, 4′, 6-diamidino-2-phenylindole stain for 5 min, and images were captured with an Axioplan 2 microscope using an Axiocam HRc camera and AxioVision 4.3 software (all from Carl Zeiss Ltd, Welwyn Garden City, UK).

Cell Aggregate Culture

Three-dimensional cell aggregates (500,000 cells)25 were cultured at 37°C in 1 mL chondrogenic media for 14 days (medium changed every 2 days) in either normal oxygen (95% air containing 20% oxygen, and 5% carbon dioxide) or low oxygen (90% nitrogen, 5% carbon dioxide and 5% oxygen). The chondrogenic culture media contained basic media (as above, but without serum) with 1X insulin-transferrin-selenium supplement, ITS+1 (final concentration, 10 µg/mL bovine insulin, 5.5 µg/mL transferrin, 5 ng/mL sodium selenite, 4.7 µg/mL linoleic acid, and 0.5 mg/mL bovine serum albumin), 37.5 µg/mL ascorbate 2 phosphate, 100 nM dexamethasone, 10 ng/mL transforming growth factor beta-3 (TGFβ3), and 100 ng/mL insulin-like growth factor-1 (IGF-1) (all from Sigma, Poole, UK).

Glycosaminoglycan and DNA Assays

Cell aggregate wet mass was recorded at 14 days and the aggregates were digested overnight in 20 µL of 10 units/mL papain (Sigma), 0.1 M sodium acetate, 2.4 mM EDTA, 5 mM L-cysteine, pH 5.8 at 60°C. Sulphated glycosoaminoglycan (GAG) was assayed using 1,9-dimethylmethylene blue (Aldrich, Poole, UK) with shark chondroitin sulphate (Sigma) as standard, and DNA in the papain digest was measured using PicoGreen (Invitrogen, Paisley, UK) with standard dsDNA (Invitrogen).25, 26

Gene Expression Analysis

Quantitative real-time gene expression analysis was carried out for hypoxia-inducible transcription factor 1 alpha (HIF1α), hypoxia-inducible transcription factor 2 alpha (HIF2α), aggrecan, versican, perlecan, collagen type I (COL1A2), collagen type II (COL2A1), collagen type IX (COL9A1), collagen type X (COL10A1), collagen type XI (COL11A2), L-SOX5, SOX6, and SOX9. Total RNA was extracted with Tri Reagent (Sigma) from passage 2 cells in monolayer and from cell aggregates at 14 days ground up with Molecular Grinding Resin (Geno Technology Inc., St. Louis, MO). cDNA was generated from 10 to 100 ng of total RNA using reverse transcription followed by poly(A) PCR global amplification.27 Globally amplified cDNAs were diluted 1:1,000 and 1 µL aliquot of the diluted cDNA was amplified by quantitative real-time PCR in a 25 µL final reaction volume performed on a MJ Research Opticon using a SYBR Green Core Kit (Eugentec, Seraing, Belgium). Gene-specific primers were designed within 300 base pairs of the 3′ region of the relevant gene using ABI Primer Express software (Applied Biosystems, Foster City, CA). Gene expression analyses were performed relative to β-actin and calculated using the 2−ΔΔCt method.28 All primers (Invitrogen) were based on human sequences and are shown in Table 1.

Table 1. Forward 5′-3′ and Reverse 5′-3′ Sequences for Primers Used in the Gene Expression Studies
PrimerForward 5′-3′ SequenceReverse 5′-3′ Sequence

Safranin-O Staining

The cell aggregates were fixed in 4% formaldehyde (BDH Ltd, Poole, UK)/ DPBS for 2 h. The samples were then washed in 70% industrial methylated spirit (IMS) (BDH Ltd) and placed in a Shandon Citadel 2000 tissue processor (Thermo Electron Corporation, Runcorn, UK). Paraffin-embedded sections (5 µm) were taken and mounted on Superfrost Plus precoated slides (Menzel Glaser GMBH, Braunschweig, Germany), air dried, and left at 37°C overnight.

Paraffin sections were deparaffinized with xylene and solutions of decreasing ethanol percentage. Sections were stained with safranin-O to highlight the sulfated glycosaminoglycan in the form of proteoglycan, representing active cartilage growth. Sections were stained for 6 min in 0.1% safranin-O (BDH Ltd) containing 1% acetic acid and counterstained for 2 min in 1% Fast Green containing 7% acetic acid.

Immunohistochemical Staining of Cell Aggregate Sections

Sections were also preincubated with 0.1 U/mL chondroitinase ABC (Sigma) at 37°C for 1 h and then immunostained with goat anti-human collagen type I (C-18 polyclonal) or collagen type II (N-19 polyclonal) (both from Santa Cruz Biotechnology), or with rabbit anti-human aggrecan (BR1) (all at 1:100 dilution) for 16 h at 4°C, followed by washing and incubating for 30 min at room temperature in donkey anti-goat IgG for collagen type I and collagen type II, and donkey anti-rabbit IgG for aggrecan (all at 1:250 dilution) biotin conjugated secondary antibodies (both from Santa Cruz Biotechnology). Goat IgG antibody was used as control for collagen, and rabbit IgG was used as a control for aggrecan (both from Santa Cruz Biotechnology).

Endogenous peroxidase activity was quenched with 3% hydrogen peroxide (Sigma) in methanol (BDH Ltd) for 5 min. Nonspecific binding was blocked with 10% normal donkey serum diluted in 1% bovine serum albumin (BSA) (both from Sigma) in DPBS for 1 h at room temperature. For visualization, sections were incubated for 30 min at room temperature in streptavidin/peroxidase complex (1:500 in DPBS) (Dako), rinsed in distilled water, and incubated in fast-DAB (3, 3′-diaminobenzidine) peroxidase substrate (Sigma) for 5 min and counterstained in diluted filtered hematoxylin (Sigma) for 15 s. Images were then taken with an Axioplan 2 microscope using an Axiocam HRc camera and AxioVision 4.3 software.

Statistical Analysis

Experiments were performed separately with cells from three sources, and all experiments were in triplicate. Gene expression data, wet mass, and DNA and GAG assay results are presented as a mean and standard error of the mean (SEM). Student's paired t-test was used to analyze the results from the two culture conditions and determine the level of significance. Statistical analyses were conducted with Sigma Stat software (SPSS Inc., Chicago, IL). Significance was set at p < 0.05.


Cell Surface Epitope Characterization of BMSCs

The BMSCs proliferated in culture and reached confluence by day 14. Figure 1 shows that cells at passage 2 stained strongly for CD13, CD44, CD90, and CD105 (markers for MSCs), and for 3G5 (marker for vascular pericytes). The cells stained poorly for LNGFR and STRO-1 (markers on freshly isolated BMSCs). Staining for CD34 (marker for hematopoetic cells) and CD56 (NCAM) was negative.

Figure 1.

Cell surface epitope characterization of BMSCs. Cell surface staining on BMSCs was performed using a panel of antibodies and FITC conjugated secondary antibody (green), and DAPI (blue). Results showed strong staining for CD13, CD29, CD44, CD90, CD105, and 3G5, and negative staining for LNGFR, STRO1, CD34, and CD56. No staining was observed for the IgG control. [Color scheme can be viewed in the online issue, which is available at]

Chondrogenic Culture of BMSCs and the Effect of Low Oxygen Tension

The BMSC aggregates cultured in chondrogenic medium showed evidence of induction of chondrogenesis under normal culture conditions (20% oxygen), and this was greatly enhanced in lower oxygen (5%) (Fig. 2). Cell aggregates cultured under hypoxic conditions at 14 days had 2.4-fold higher wet mass than those cultured under normoxic conditions (p < 0.05; Fig. 2a). The hypoxic conditions resulted in less cell proliferation as the aggregates contained 16% less total DNA (Fig. 2b). There was, however, a large increase (2.7-fold) in the GAG accumulation (p < 0.05; Fig. 2c), such that the proteoglycan content per cell at 14 days was much higher under hypoxic conditions (3.2-fold, p < 0.05; Fig. 2d).

Figure 2.

Chondrogenic cultures of BMSCs and the effects of hypoxia. Wet weight (a), GAG analysis (b), DNA analysis (c), and GAG per DNA measurements (d) of cell aggregates following chondrogenic differentiation for 14 days under normoxic and hypoxic conditions. Data is mean ± SEM (n = 3). *p < 0.05 (Student's paired t-test).

Gene Expression Analysis of Chondrogenic BMSC Aggregates

In the presence of lowered oxygen tension, there was a more enhanced chondrogenic response with changes in gene expression (Fig. 3). The expression of SOX6, collagen types II and XI, and aggrecan were significantly increased in hypoxia by 81-, 128-, 7-, and 6-fold, respectively (p < 0.05). Collagen type I was highly expressed in the chondrogenic cultures and there was no downregulation of its expression under hypoxic conditions.

Figure 3.

Gene expression in chondrogenic cultures of BMSCs. Relative gene expression for hypoxia induction factors (HIF) and SOX genes (a), collagens (b), and proteoglycans (c) in monolayer culture and following chondrogenic differentiation for 14 days under normoxic (black) and hypoxic (white) conditions. Data is mean ± SEM (n = 3). *p < 0.05 (Student's paired t-test).

Safranin-O Staining and Immunohistochemistry of Chondrogenic BMSC Aggregates

The cell aggregates cultured under both normoxic and hypoxic conditions showed evidence of chondrogenesis with safranin-O staining of proteoglycan matrix and immunolocalization of cartilage-associated matrix, including collagen type II and aggrecan (Fig. 4). Cell aggregates cultured under hypoxic conditions were larger and less cellular than aggregates cultured under normoxia. All cells had a rounded appearance and were surrounded by extracellular matrix. Cell aggregates under normoxic and hypoxic conditions both stained, albeit weakly, for collagen type I.

Figure 4.

Safranin-O staining and immunohistochemistry of chondrogenic cultures of BMSCs. Safranin-O staining for proteoglycan, and immunohistochemical staining for collagen type I and II, aggrecan, and control IgG in cell aggregates following chondrogenic differentiation for 14 days under normoxic and hypoxic culture conditions. [Color figure can be viewed in the online issue, which is available at]


The cell surface epitope characterization of BMSCs showed characteristic staining and, although they stained poorly for STRO-1 and for LNGFR, the expression of these markers in BMSCs is reported to decline with culture.29 The antigen recognized by 3G5 is a cell surface ganglioside, characterized originally on vascular pericytes from bovine retina, which have been shown to have multidifferentiation potential.12 The expression of 3G5 is specific for pericytes and is known to vary in culture.23, 30 Shi and Gronthos10 showed that a minor population of BMSCs positive for STRO1 was also positive for 3G5. In our experiments, none of the BMSCs were positive for STRO1, but a high number of cultured BMSCs stained positively for 3G5. Pericytes have been shown to have multidifferentiation potential,11, 12 and our study shows that pericytes may account for stem cells in bone marrow.

The BMSCs responded to chondrogenic culture in cell aggregates as demonstrated by the GAG and DNA assays, gene expression studies, and safranin-O and immunohistochemical staining, and this was much enhanced under hypoxic conditions. The wet mass of cell aggregate provides a simple measure of in vitro chondrogenesis in MSCs,31, 32 and cell aggregates cultured under hypoxic conditions had a 2.4-fold higher wet mass than those cultured under normoxia. The GAG content reflected proteoglycan biosynthesis and accumulation in the matrix and, under hypoxic conditions, there was a 2.7-fold increase in the total GAG per aggregate. In spite of the increased mass, there was a lower DNA content compared to normoxia, which reflected a lower cell proliferation rate, and this was balanced by a much greater production of GAG per cell. These results with BMSCs are comparable to a previous study on stem cells derived from other human liposuction-derived adipose tissue, where 5% oxygen was reported to increase collagen and GAG synthesis, and reduce cell proliferation.22

The gene expression analysis provided an assessment of the changes induced by hypoxia in the chondrogenic cultures. The transcription factors have been shown to be essential for chondrocyte differentiation and cartilage formation.33 One of their actions is to activate specific enhancer elements in cartilage matrix genes, such as collagen type II and aggrecan.34, 35 This action of SOX9 is further enhanced by SOX5 and SOX6. In the chondrogenic cultures of BMSCs cells, there was an increase in the expression of SOX6 in hypoxia. The expression of key cartilage collagens II and XI was increased under hypoxia, showing that hypoxia enhanced the potential for the assembly of a complete cartilage fibrillar template. Aggrecan, and to a lesser extent perlecan, were also increased in hypoxia. This is the first study that has characterized the chondrogenic gene expression in hypoxia in BMSCs.

Lennon et al.17 looked at the effects of hypoxia on the monolayer expansion of calvarial bone marrow-derived MSCs. We, on the other hand, have looked at the effects of hypoxia on human bone marrow-derived MSCs undergoing chondrogenesis, as this is more relevant and applicable to a future therapeutic procedure for the repair of articular cartilage. Malladi et al.36 showed that inguinal fat pad-derived cells from mice showed enhanced chondrogenesis in 2% oxygen. There are many differences between that study and our study. Malladi et al. used a different species, lower oxygen concentration, and different chondrogenic conditions; they used the micromass technique, and the chondrogenic medium contained 1% FCS and 10 ng/ml TGF-β1.

The hypoxic conditions used differed from the normoxic conditions not only in the lower oxygen tension, but also in elevated nitrogen tension and absence of trace gases including argon and neon. There is no data in the literature to suggest that these additional factors contributed significantly towards the results seen.

The response by cells to hypoxia is complex and mediated by several genes.37 HIF1α is one of the major regulators of hypoxic response in most cells and tissues,38 where it is frequently associated with angiogenesis and new blood vessel formation. Targets of its molecular signaling are reported to include a cluster of hydroxylases crucial for collagen fiber formation, such as prolyl-4 hydroxylase and procollagen lysyl-hydroxylase.39–41 Through these actions, HIF1α affects the rate of synthesis of procollagen chains in vivo and in vitro.42 HIF2α is closely related to HIF1α, with similarities in DNA binding and dimerization, but with differences in transactivation domains.43 In previous work, we have shown that HIF2α and not HIF1α was upregulated in infrapatellar fat pad-derived MSCs undergoing chondrogenesis in hypoxic culture conditions.23 In our experiments with BMSCs, although the levels of HIF1α and HIF2α were higher with hypoxia, the differences were not statistically significant. HIF genes are transcription factors, and hypoxia was associated with increases in gene expression of collagen type II, collagen type XI, and aggrecan, and immunohistochemical staining of collagen type II. We have previously shown that HIF2α and not HIF1α was upregulated in the fat pad-derived MSCs in response to 5% oxygen.21 It has previously been noted that in with lung epithelial cells, the upregulation of HIF1α was transient, whereas increases in HIF2α were sustained.44 A transient upregulation of the HIF genes cannot be ruled out in our study and warrants further investigation.


Our results show that BMSCs express the typical MSC markers and the pericyte marker 3G5. Cells from three BMSC samples showed the ability to undergo chondrogenic differentiation, and 5% oxygen selectively enhanced the expression of the cartilage matrix genes. The results showed that chondrogenesis was enhanced in an atmosphere of reduced oxygen tension and that this is mediated by significantly increased expression of key genes expressed by chondrocytes, notably the transcription factor SOX6. These findings show that oxygen tension has an important role in regulating the synthesis and assembly of matrix by BMSCs, as they undergo chondrogenesis by acting as a “metabolic switch,” and this has important implications for the use of BMSCs in cartilage tissue engineering.


The authors are grateful to the UK Medical Research Council (MRC) and the Royal College of Surgeons of Edinburgh (RCSEd) for funding a clinical research fellowship. The authors are also grateful to Dr. A. Canfield, University of Manchester, UK for the supply of 3G5 antibody. The Research Councils (BBSRC, MRC, EPSRC) are thanked for funding UKCTE and The Wellcome Trust for support for The Wellcome Trust Centre for Cell-Matrix Research.