• Open Access

Reduced oxygen tension attenuates differentiation capacity of human mesenchymal stem cells and prolongs their lifespan


  • Christine Fehrer,

    1. Extracellular Matrix Research, Institute for Biomedical Aging Research, Austrian Academy of Sciences, Rennweg 10, A-6020 Innsbruck, Austria
    Search for more papers by this author
  • Regina Brunauer,

    1. Extracellular Matrix Research, Institute for Biomedical Aging Research, Austrian Academy of Sciences, Rennweg 10, A-6020 Innsbruck, Austria
    Search for more papers by this author
  • Gerhard Laschober,

    1. Extracellular Matrix Research, Institute for Biomedical Aging Research, Austrian Academy of Sciences, Rennweg 10, A-6020 Innsbruck, Austria
    Search for more papers by this author
  • Hermann Unterluggauer,

    1. Department of Molecular and Cell Biology, Institute for Biomedical Aging Research, Austrian Academy of Sciences, Rennweg 10, A-6020 Innsbruck, Austria
    Search for more papers by this author
  • Stephan Reitinger,

    1. Extracellular Matrix Research, Institute for Biomedical Aging Research, Austrian Academy of Sciences, Rennweg 10, A-6020 Innsbruck, Austria
    Search for more papers by this author
  • Frank Kloss,

    1. Department of Cranio-Maxillofacial and Oral Surgery, University Hospital Innsbruck, Maximilianstrasse 10, A-6020 Innsbruck, Austria
    Search for more papers by this author
  • Christian Gülly,

    1. Center for Medical Research, Medical University of Graz, Stiftingtalstrasse 24, A-8010 Graz, Austria
    Search for more papers by this author
  • Robert Gaßner,

    1. Department of Cranio-Maxillofacial and Oral Surgery, University Hospital Innsbruck, Maximilianstrasse 10, A-6020 Innsbruck, Austria
    Search for more papers by this author
  • Günter Lepperdinger

    1. Extracellular Matrix Research, Institute for Biomedical Aging Research, Austrian Academy of Sciences, Rennweg 10, A-6020 Innsbruck, Austria
    Search for more papers by this author

  • Christine Fehrer and Regina Brunauer contributed equally to this article.

Günter Lepperdinger, Extracellular Matrix Research, Institute for Biomedical Aging Research, Austrian Academy of Sciences, Rennweg 10, A-6020 Innsbruck, Austria. Tel. 0043 5125839 1940; fax: 0043 5125839 198; e-mail: guenter.lepperdinger@oeaw.ac.at


Mesenchymal stem cells (MSC) are capable of differentiating into bone, fat, cartilage, tendon and other organ progenitor cells. Despite the abundance of MSC within the organism, little is known about their in vivo properties or about their corresponding in vivo niches. We therefore isolated MSC from spongy (cancellous) bone biopsies of healthy adults. When compared with the surrounding marrow, a fourfold higher number of colony-forming units was found within the tight meshwork of trabecular bone surface. At these sites, oxygen concentrations range from 1% to 7%. In MSC cultured at oxygen as low as 3%, rates for cell death and hypoxia-induced gene transcription remained unchanged, while in vitro proliferative lifespan was significantly increased, with about 10 additional population doublings before reaching terminal growth arrest. However, differentiation capacity into adipogenic progeny was diminished and no osteogenic differentiation was detectable at 3% oxygen. In turn, MSC that had previously been cultured at 3% oxygen could subsequently be stimulated to successfully differentiate at 20% oxygen. These data support our preliminary finding that primary MSC are enriched at the surface of spongy bone. Low oxygen levels in this location provide a milieu that extends cellular lifespan and furthermore is instructive for the stemness of MSC allowing proliferation upon stimulation while suppressing differentiation.


Spongy bone harbors hematopoietic and mesenchymal cells. The latter contain a rare subset of uncommitted progenitors that have been further defined ex vivo and termed mesenchymal stem cells (MSC) (Caplan, 1991). This cell type has been characterized by a stable undifferentiated phenotype, as well as by the ability to proliferate extensively while retaining the potential to differentiate along osteoblastic, adipocytic and chondrocytic lineages in vitro (Sakaguchi et al., 2004). Uncommitted MSC are not only located in the marrow but also in other tissues, where they appear to be functionally involved in the maintenance of characteristics of the resident tissue, as they also exhibit extensive renewal potential (Barry & Murphy, 2004).

It is important to note that the developmental fate of MSC within a particular tissue and their progeny depends on environmental influences occurring in the biological niche (Ankeny et al., 2004; Muguruma et al., 2005; Cui et al., 2006). The microenvironment and/or niche for the adult hematopoietic (Calvi et al., 2003; Zhang et al., 2003), neural (Shen et al., 2004) and epithelial stem cell (Cotsarelis et al., 1990) have previously been described. In contrast, no comprehensive data are available that describe characteristics and properties of the microenvironment for mesenchymal progenitors, neither within the bone marrow nor in other mesenchymal tissues. We therefore wondered whether MSC, which are involved in bone remodeling and bone repair after injury, can also be found at sites juxtaposed to bone. In a first set of experiments, it became evident that this type of stem cell is present at the surface of bone trabeculae at higher numbers than within the surrounding bone marrow.

In vivo, the oxygen tension (pO2) depends on the level of vascularization and the type of microenvironment within the respective organ. Blood in human bone marrow contains about 7% O2 (Ishikawa & Ito, 1988), however, O2 gradients in surrounding tissues can be further extrapolated only by mathematical models, because direct in vivo measurements of pO2 spatial variations within the bone marrow is practically impossible. Applying detailed modeling, O2 levels in bone marrow have been calculated to be highest (~5%) near the sinuses and lowest (~1%) at the inner surface of cortical bone (Chow et al., 2001; Antoniou et al., 2004). Particularly in the case of stem cells, high pO2 may result in accelerated cellular aging, which in due course may lead to the loss of corresponding functional properties (Csete, 2005). In view of the fact that the importance and influence of low O2 concentration has been independently described for various other cell types, for example, animal cells were found to proliferate with elevated rates under reduced O2 conditions (Cooper et al., 1958; Zwartouw & Westwood, 1958; Packer & Fuehr, 1977), we strived to determine whether varying O2 levels alter the behavior of human MSC in culture. Zhou and colleagues (2005) have previously reported that human MSCs exhibit a diminished osteogenic potential when exposed for 12–24 h to reduced O2 levels. Similarly, a 2-day culture in conditions of reduced pO2 attenuated adipogenic differentiation (Zhou et al., 2005). However, telomerase-immortalized human MSCs attain a preadipocyte-like appearance when cultured at low pO2, although up-regulation of major adipogenic factors could not be detected under these conditions (Fink et al., 2004).

For human embryonic stem cells comparable results have been obtained. Culturing in 3% O2 helps to maintain this type of stem cell in an undifferentiated state (Ezashi et al., 2005). It is also well documented that growth of mammalian embryos in culture is optimal when the O2 concentration is decreased to about 5% (Okazaki & Maltpe, 2006). Furthermore, different levels of varying O2 concentrations are considered instructive for regenerative processes, for example, anoxic conditions occur after injury and during wound healing (Jauniaux et al., 1999). In particular after bone fracture, O2 levels are known to transiently drop to 0–2% at the center of the wound (Heppenstall et al., 1975; Knighton et al., 1981). In this context it is also interesting to note that hypoxic tumor cells appear to dedifferentiate and acquire stem cell-like properties, as demonstrated in neuroblastoma (Jogi et al., 2002), breast carcinoma (Helczynska et al., 2003) and prostate carcinoma cells (Ghafar et al., 2003). In summary, this incomplete listing of examples demonstrates that O2 content in the local environment of stem cells controls the fate and commitment of undifferentiated progenitor cells and tissue primordia.

Routinely, MSCs are cultured under ambient atmospheric conditions of ~20% O2. We, however, regard MSC, harvested from the surfaces of spongy bone, as being naturally adapted to low pO2. We therefore examined the impact of increasing O2 on this type of human stem cell, in particular with respect to cellular aging, which we presume to be the major cause for the reduction of stem cell self-renewal capacity and the potential to regenerate various connective tissues.


Cellular preparations of MSC from trabecular bone culture at low pO2

Primary colonies are considered the ex vivo source of MSCs. In the first step, spongy bone was separated into trabecular material and marrow. The remaining osseous material was further treated with collagenase to free cells from the tight meshwork located at the bone surface. The colony-forming unit (CFU) fibroblast obtained from the latter cellular fraction was compared with that derived from the explanted bone marrow stroma (Fig. 1A). In a second set of experiments the number of primary CFU was estimated from collagenase-treated bone and found elevated when grown at 3% O2 (Fig. 1B). The elevated number of viable cells demonstrated that these cells appear to be well adapted to a low oxygen concentration. We thus considered sites of low pO2 as those sites in vivo where uncommitted mesenchymal stromal cells preferentially reside. Therefore besides normal atmospheric conditions, we cultured MSC under decreased (3%) pO2. Hypoxia-induced or hypoxia-associated genes such as HIF-1α, PH-4, HIF1AN, VHL, HYOU1, HIG1 and HIG2 exhibited unaltered transcription when comparing both O2 concentrations (Fig. 2B). These results confirmed the previous assumption that 3% pO2 is sensed by primary MSC as normoxic. In order to gain detailed information about changes in gene expression, whole genome oligonucleotide arrays were hybridized. We found 4009 probe sets differentially expressed in these analyses, which referred to 2313 unique genes. 2244 of these genes were then further characterized with regard to their molecular function or involvement in particular biochemical pathways by querying the PANTHER database at http://www.pantherdb.org/ (Fig. 2A). Three hundred and twenty genes fell into signalling pathways such as Wnt, inflammation mediated by cytokines, angiogenesis, integrins, p53, insulin/insulin-like growth factor (IGF) and transforming growth factor β (TGF-β) signalling, several of which are believed to regulate cellular aging processes. Further quantitative assessment of differential expression levels of a variety of genes, which are functionally relevant to properties of uncommitted precursor cells as well as to determining progenitor cell type, was confirmed by quantitative reverse transcriptase–polymerase chain reaction (qRT-PCR). We probed cDNA prepared from primary MSCs that had been isolated from six healthy individuals of different ages (male: age 23, 50, 51 and 78; female: age 47 and 70) followed by cultivation under both oxygen conditions. As shown in Fig. 2B, for a subset of genes, differential expression with varying oxygen levels could be demonstrated, for example, for gene products, which are known to be involved in signal transduction (noggin, IGF-1, inhibin-β, leukemia inhibitory factor, Dickkopf-3, leptin receptor and follistatin) or which are involved in the control of transcription such as stem cell-specific transcription factors [NANOG and REX-1 (ZFP42)]. Furthermore, the expression levels of bone marrow stromal antigen 1, osteoprotegerin and hyaluronan synthase 1 changed significantly, suggesting that that MSCs actively restructure their microenvironment under varying oxygen levels.

Figure 1.

Primary colony formation of plastic adherent mesenchymal progenitor cells. Colony-forming units (CFU) of mononuclear cells (MNC) derived from biopsies of healthy individuals were assessed. (A) MNC were isolated by centrifugation of spongy bone. Afterwards, the remaining trabecular material was treated with collagenase in order to free additional MNC. Subsequently, isolated cells were cultured in triplicates under standard conditions at ambient air for 2 weeks and colonies were visualized by crystal violet staining. Bone was obtained from eight donors (male: age 7, 10, 51, 68 and 78; female: age 40, 50 and 70) (B) Primary cells after isolation by collagenase from trabecular bone were cultured in triplicate at 3% or at 20% O2. Cells were obtained from 11 donors (male: age 7, 10, 23, 51, 63, 68 and 78; female: age 40, 50, 51 and 70).

Figure 2.

Molecular characterization of passage 1 MSC at 3% and 20% pO2. (A) Computational analysis of genes that were differentially expressed at high pO2 and included in pathways associated with curator-defined PANTHER families. Only groups of 10 or more genes, which have been annotated to participate in signaling (sign) pathways, biological processes and diseases are shown. (B) Differential expression of hypoxia-associated genes (hypoxia), stem cell markers/transcription factors (stemness), proteins involved in bone morphogenetic proteins signaling (osteogenesis), cytokines and receptors as well as membrane-bound marker or enzymes (pericellular), and other signaling molecules (signaling) were specifically quantified by qRT-PCR probing cDNA derived from six individual MSCs, passage 1 (male: age 78, 51, 50 and 23; female, age: 70 adn 47). Standard deviations (SD) and respective P values (P) for each set of measurements are depicted at the right side. For further information on gene identities please see Table 1.

Replicative senescence of MSC during long-term culture

Subsequent experiments involved MSCs from nine healthy individuals (female: age 7, 47, 51, 56 and 70; male: age 10, 50, 51 and 78). The explanted cells were plated at low density (50 cells cm−2), and were incubated either at 3% or 20% pO2 for long-term culture up to about 100 days. Figure 3A shows representative examples of growth curves and their respective cumulative population doublings (PD) for three individual MSC lines. Cells ceased growth after several passages under ambient O2 concentration, while at the same time, cells exposed to low O2 concentration maintained the ability to proliferate. Figure 3A shows three representative examples of MSCs derived from a 56-year-old and a 70-year-old female as well as from a 78-year-old male donor showing this phenomenon. The pooled results from all nine lines gave mean PDs of 28.5 ± 3.8 PD at 20% O2 and 37.5 ± 3.4 PD at 3% O2, respectively (Fig. 3B).

Figure 3.

Characterization of MSCs during long-term culture at 3% or 20% O2. (A) Three of nine growth curves are shown. MSCs were obtained from differently aged donors of both sexes and were grown at 3% (open circles) and at 20% O2 (filled circles). (B) Comparison of cumulative population doublings (PD) reached by MSC cultures derived from nine individual donors at 20% or 3% O2 showing a significant increase at 3% O2. (C) Cell viability of four independent MSC cultures at passages 1–5 was determined for cells cultured under normal conditions as well as cells treated with 200 µm staurosporine (SSP) for 20 h. (D) Photomicrographs of MSCs cultured for 16, 31 and 89 days. Bar is 100 µm.

This difference could be due to a higher susceptibility to cell death at elevated O2 concentration. Therefore, death rates of MSCs from early up to late passages at 3% and 20% pO2 were assessed by staining with propidium iodide (PI) and fluorescein isothiocyanate-conjugated annexinV (FITC-annexinV) (Fig. 3C). Combining both PI+ and annexinV+ labelling, cell viability for 20% O2 was 86 ± 9.25% and for 3% O2 was 90 ± 5.16%. When treating MSCs with 200 µm staurosporine for 20 h, no obvious differences both with regard to necrosis or apoptosis were detected. We concluded that elevated levels of cell death do not account for the decreased proliferative capacity of MSC at high O2.

During early passages, the morphological appearance of differently cultured MSCs was almost indistinguishable. In subsequent passages, cells cultured at 3% pO2 maintained their characteristic spindle-shaped. In turn during long-term culture, individual cells changed their morphology and displayed enlarged and flattened phenotypic appearances. The emergence of morphologically altered cells occurred at elevated rates in 20% O2 cultures as seen in Fig. 3D. The delayed onset of senescent phenotype in 3% O2 culture was also reflected by a decreased amount of advanced glycation end-products (AGE) (Fig. 4A). By the time MSCs ceased growth in 3% O2, high numbers of senescent cells were also identified, resembling late passage MSCs cultured at 20% O2. In addition to morphological evaluation, senescent phenotypes were also confirmed by the increase of transcripts for the cell cycle regulators (cyclin-dependent kinase inhibitors) and senescence markers p15, p16 and p21 (Fig. 4B) and the expression of senescence-associated β-galactosidase (SA-β-gal) (Fig. 4C). The latter was further quantified by assessing the number of SA-β-gal-positive cells in proliferating and senescent cultures both grown in long-term culture at 3% and 20% pO2 (Fig. 4D). Elevated numbers of SA-β-gal-positive cells were encountered at 20% pO2 when approaching the presenescent state, whereas at senescence the number and appearance of SA-β-gal-positive cells at 3% and 20% O2 were comparable.

Figure 4.

Cellular aging of MSCs. (A) Determination of advanced glycation endproducts (AGE) in passage 1 (day 31) MSCs cultured at indicated O2 levels showing one representative blot (left panel); α-tubulin was used as a loading control. (B) mRNA levels of cyclin-dependent kinase inhibitors p15, p16 and p21 at passage 1 (early) and shortly before growth arrest (late) at 3% and 20% (grey) pO2. Blue line indicates the respective normalized relative expression levels at 3% O2. (C) Senescence marker, SA-β-gal at terminal growth arrest in 3 and 20% pO2 culture. Bar is 100 µm. (D) The number of SA-β-gal positive cells in 3 and 20% pO2 were assessed in four proliferating cultures (presenescent) as well as in their corresponding descendents, which eventually turned senescent at subsequent passages (senescent).

Evaluation of in vitro differentiation capacity of MSC during long-term culture

We next investigated in vitro differentiation capacities using primary cell isolates from both female and male donors of different ages. Regardless of 3% or 20% pO2 in the cultures, primary cells could be successfully differentiated into adipogenic and osteogenic progenitors as determined by histochemical analysis of lipid droplet-containing cells in the case of adipogenesis, or calcium-containing extracellular matrix deposited in the cultures when osteogenesis had taken place. However, when applying a graded evaluation scheme for this type of histochemical analysis at early passages (Fig. 5A), it became apparent that the primary cultures at 3% pO2 exhibited a significant decrease in differentiation capacity with respect to both lineages (Fig. 5B–D). In a subsequent set of experiments, we tested whether the respective changes are reversible by changing O2 content in later passages. Cells were induced to differentiate both at low and high O2 concentration following preceding passages at 3% or 20% pO2. After histochemical staining as well as by qPCR, which we applied to measure adipogenic [molecular markers: FABP4 (fatty acid binding protein 4, adipocyte), LPL (lipoprotein lipase)] and osteogenic [molecular markers: ALPL (alkaline phosphatase), IBSP (integrin-binding sialoprotein)] development, it became apparent that MSC, which had initially been grown at 3% pO2, regained full differentiation capacity when subsequently induced and cultivated at 20% pO2 (Figs 6 and 7). In contrast to cells which had been cultured at 20% pO2 during induction, those cultured at 3% pO2 displayed little or no capability for differentiation. During continuous long-term cell cultures at both 3% and 20% pO2, MSCs gradually lost their capacity to differentiate into osteogenic precursors (Fig. 5B–D). In sharp contrast to this finding, cell cultures grown at 3% O2 were still able to differentiate into preadipocytes. The capacity to differentiate into preadipocytes was further maintained even after reaching the point when cells had stopped to proliferate and were thus considered to be senescent. At 20% pO2 and with increasing passages, however, adipogenic differentiation capacity decreased continuously and was lost before MSCs reached terminal growth arrest.

Figure 5.

Adipogenic and osteogenic differentiation of MSCs at 3 or 20% O2. (A) Graded classification (0–3) for in vitro adipogenic differentiation (a) as revealed by Oil red staining (bar indicates 100 µm) or osteogenic (o) differentiation demonstrated by means of Alizarin Red staining (bar indicates 1 cm). (B–D) Differentiation capacity of primary cells and MSCs from a female donor, age 56 (B); a female donor, age 70 (C); and a male donor, age 78 (D) during long-term culture was scored as defined and shown in (A). Population doublings of cultured cells as well as the respective adipogenic and osteogenic differentiation potential of MSCs were examined in cultures either at 3% (triangles) or 20% O2 (squares/dashed line) and assessed at indicated passages.

Figure 6.

Influence of O2 on osteogenic and adipogenic differentiation capacity of MSCs in vitro. (A) Schematic drawing of cultivation and differentiation at different levels of O2: MSC that had been grown at 3% or 20% O2 were subsequently stimulated to differentiate at either O2 level. Osteogenic (B) or adipogenic (C) differentiation was assessed by histological staining and scored as defined in Fig. 5A. The two representative examples shown here were performed with MSCs at passage 1 derived from two donors (male: age 78; female: age 56).

Figure 7.

Assessment of molecular markers for adipogenic (A) and osteogenic (B) differentiation. Lineage development was monitored using qRT-PCR for fatty acid-binding protein-4 (FABP4) and lipoprotein lipase (LPL) in the case of adipogenic lineage differentiation, and for osteogenic differentiation using alkaline phosphatase (ALPL) and bone sialoprotein-2 (IBSP) as markers. MSCs used for these analyses were derived from three donors (male: age 33, 68; female: age 40) at passage 1. The cells were cultivated according to the scheme depicted in Fig. 6A. qPCR measurements were performed in duplicates.


Variations in O2 levels have long been known to play a crucial role in organ survival (Hicks et al., 2006), healing (LaVan & Hunt, 1990; Tandara & Mustoe, 2004) and aging (Harman, 1956; Stadtman & Berlett, 1998). High O2 concentrations could result in noxious oxidation of biomolecules through the generation of reactive oxygen species (Schoneich, 2006; Stadtman, 2006; Terman & Brunk, 2006; Chandel & Budinger, 2007). Traditional cell culture methods for stem cells as well as virtually all other ex vivo cell cultures are conducted in ambient air, which does not reflect physiological conditions. Standard tissue culture incubator conditions are set at 5% CO2 and 95% air (~20% O2), exposing cells to a nonphysiologically high amount of O2. However, the established O2 concentrations in mammalian tissues range from alveolar air with 14% O2, arterial blood with 12%, venous blood with 5.3% and interstitial tissue with ~5 to 1%. Consequently, the O2 tension used in the present study was set at 3%, approximating the mean tissue O2 concentration.

The goal of the study was to investigate the effects of low vs. high O2 levels on the aging behavior of human MSC. The study builds on previous results obtained with rodent MSC (Lennon et al., 2001) that are known to be affected in vitro by variations of pO2. As an example, MSC aspirates from rat bone marrow cultured at 5% pO2 form more colonies and show increased proliferation rates compared to those grown at 20% pO2. Furthermore, their in vivo and in vitro differentiation potential has been shown to increase relative to control cultures at 20% pO2. Similar effects were also reported with monolayer cultures using plastic substrates and with three-dimensional cultures on ceramic substrate as well as with cells grown on hydrophilized polyethylene terephthalate fibrous matrices (Grayson et al., 2005). In the present work we investigated human MSCs cultured on plastic substrates in the presence of 3% pO2 and besides phenotypic evaluation, gene expression profiles under high and low pO2 content were also established. Under hypoxic conditions, the activities of the HIF-1α inhibitors PH-4 and HIF1AN are hampered; thus, stabilizing and rendering the transcription factor active (Maxwell et al., 1999), which in consequence regulates the transcription of downstream genes such as HIG1, HIG2 and HYOU1 (Schofield & Ratcliffe, 2004). To date, a battery of genes is known to be regulated by HIF-1 in a cell type-specific manner. Particular HIF-1 target genes directly affect physiologic responses such as erythropoiesis, angiogenesis and glycolysis or are associated with cell proliferation, cell survival as well as cell death via apoptosis (Semenza, 2003; Zhou et al., 2006). At 3% pO2 MSCs express mRNA for HIF-1α at sufficient levels. However, no distinct activation of a range of hypoxia-associated marker genes could be observed, which is indirect evidence for the undisturbed activity of PH-4 and HIF1AN under these conditions. In contrast to this, a further extended bioinformatic analysis of these microarray experiments clearly revealed that changing pO2 to 3% does in fact exert instructive stimuli on the transcriptional control of protein families. Molecular networks that are known to be associated with particular aging pathologies such as Alzheimer's or Parkinson's disease or with aging-related alterations in signalling systems such as insulin/IGF or TGF-β signaling pathways were found to be differentially regulated at 3% pO2, while at the same time the expression of hypoxia-associated genes was not significantly altered. Because pO2 gradients within individual connective tissues range from 1 to 5%, and MSCs freshly isolated from trabecular bone appeared to be well adapted to low pO2 environment, we therefore consider 3% pO2 as normoxic for this type of primary stem cell.

Exposure to elevated levels of O2 may induce cell death through the generation of reactive oxygen species resulting in activated proapoptotic Bcl-2 proteins, Bax or Bak, which eventually lead to mitochondrial membrane permeabilization and cell death (Chandel & Budinger, 2007). However, after the exposure of normoxic MSC cultures to elevated O2 levels, no change in the rate of cell death was detectable. In parallel, the proliferation rate of primary cells at early passages was found equal to those cultured under routine culture conditions. In this context, it is important to note that the expression of the stem cell-specific transcription factor, NANOG decreased at high pO2 which is indicative of a diminished self-renewal capacity. The latter assumption is in good accordance with the finding that the proliferative capacity was significantly higher at low pO2. Comparable results have been reported for bone marrow MIAMI cells (D’Ippolito et al., 2006). In contrast, adult stem cells derived from subcutaneous fat, which are considered closely related to MSC from bone marrow, exhibited decreased proliferation rates at low O2 (Wang et al., 2005). It has been further reported that MSCs successfully adapt to low pO2 by increasing their glucose consumption and thus maintain their viability partly depending on glycolysis (Grayson et al., 2006). Several research groups have reported that low O2 levels positively influence the viability of hematopoietic stem cells (HSC), which reside in close association to MSCs in bone (Li & Li, 2006), and it appears likely that the progenies of HSCs are spatially oriented within the bone marrow with respect to O2 concentrations, with naïve HSCs residing in regions having the lowest O2 content (Gong, 1978; Lambertsen & Weiss, 1984; Nilsson et al., 1997; Balduino et al., 2005). At low pO2, HSCs also exhibit enhanced proliferation rates (Krishnamurthy et al., 2004). In line with this, human embryonic stem cells, when cultured at low pO2, showed improved formation of embryoid bodies together with higher Oct-4 expression which is required to sustain stem cell renewal (Ezashi et al., 2005). These examples suggest that self-renewal capacities of various types of stem cells appear to be influenced by O2 content.

In accordance with this assumption, we also demonstrate that MSCs not only exhibit undisturbed proliferation rates at low pO2, but more than that, it prevents them from transforming into specific progenitor cell types. This view was further corroborated by the finding that factors that propagate progenitor cell development, such as bone morphogenetic protein 4, osteoprotegerin or leptin receptor, are down-regulated at low O2, whereas noggin and inhibin β B were strongly up-regulated. Noggin is a secreted protein that binds to and inactivates members of the TGF-β family, such as BMP2, BMP4 and BMP7 (Cohen, 2006). Inhibin is also thought to oppose TGF-β signaling (Harrison et al., 2005). Pronounced down-regulation of hyaluronan synthase 1 at high pO2 results in decreased hyaluronan production. Hyaluronan is a highly hydrated polysaccharide within the pericellular space. Reducing the amount of hydrated matrix results in the condensation of cells and this eventually may also promote cellular differentiation (Cohen, 2006).

The replicative potential of MSCs in culture can be predicted by colony-forming assays (Digirolamo et al., 1999). We found CFU numbers of primary cultures to be higher at low O2. This initial finding was corroborated by counting the number of cumulative PDs in order to determine the proliferation capacity of MSC in long-term culture. At low pO2, this was significantly enhanced, although proliferation rates remained similar for both atmospheric conditions during the initial cultivation phase. The reasons for the observed effects of pO2 on proliferative capacity are not fully understood. Yet, similar results have been published recently. For example, cultures of satellite cells from skeletal muscle of elderly donors at decreased O2 levels showed continued cell cycle progression and in consequence enhanced in vitro replicative lifespan (Chakravarthy et al., 2001). Cells exposed to modest hyperoxia are likely to undergo senescence earlier, which could be explained by the postulation that normal aging can at least in part be attributed to the generation of reactive oxygen species (Wallace, 2005). The formation of reactive carbonyl compounds resulting from auto-oxidation or lipid peroxidation is also closely related to oxidative processes (Hein, 2006). Reactive carbonyl intermediates, products of auto-oxidized sugar moieties as well as AGEs are chemical modifications of proteins that accumulate during cellular aging. Mammalian skin accumulates AGEs at a rate that increases with age (Sell et al., 1996) and it is therefore believed that the emergence of AGEs in the course of the aging process is also paralleled by an elevation of the endogenous level of oxidative stress. AGE formation was enhanced in MSCs at elevated pO2, which also accords with the observed decrease of CFU, together with a significant loss in proliferative capacity. During the last phase of long-term culture, MSC up-regulate markers characteristic for terminal growth arrest such as the cyclin-dependent kinases p15, p16 and p21 and, in addition, senescence-associated β-galactosidase positive cells were found to increase in number. Taken together, all these features indicate that the replicative aging taking place in MSCs occurs more rapidly under mild oxidative stress.

The observed enhanced replicative capacity at low pO2 and the ability to differentiate into specific progenitor cells at high O2 levels suggested the following conclusions: MSC may remain quiescent in their respective primary niche; however, as soon as a daughter cell leaves the niche after cell division, it is likely to proliferate efficiently under low O2 conditions, while being sufficiently protected from further differentiation. In turn, expanded progeny can only gain full differentiation capacity at elevated pO2. To a large extent, O2 can only be supplied by the circulation, which further supports differentiation by distributing nutrients and growth factors. This model agrees with what is known for HSC: these cells remain quiescent under low O2 conditions, while proliferating and differentiating optimally at elevated pO2 (Cipolleschi et al., 1993). Whether O2 is a decisive signal for the differentiation of mesenchymal progenitor cells in vivo has to be tested in future experiments.

Experimental procedures

Isolation and culture of human MSC

Bone from the iliac crest of systemically healthy individuals was harvested for reconstructive bone surgery of defects within other areas of the body. A small biopsy of substantia spongiosa osseum, which would otherwise have been discarded was taken for further investigation under an Institutional Review Board-approved protocol, having obtained written consent. After surgery, the bone was transferred into growth medium [MEM (Gibco; no. 25300-054 Invitrogen GmbH, Lofer, Austria) supplemented with 20% fetal calf serum (FCS), 100 units mL−1 penicillin, 100 µg mL−1 streptomycin (Gibco; no. 15140-122)] at room temperature. The biopsies were fragmented and marrow cells were isolated from pieces (20–100 mm3) by centrifugation (400 g, 1 min) as described previously (Peister et al., 2004). After centrifugation, the remaining pieces were treated with collagenase (2.5 mg mL−1 in growth medium) (Sigma, St. Louis, MO, USA; no. C-9722) for 2 h at 37 °C, 20% O2, 5% CO2. Thereafter, the samples were again centrifuged (400 g, 1 min). Both cell fractions, whether derived from the bone marrow by centrifugation or from bone by collagenase treatment, were loaded seperately on a Ficoll-Paque Plus gradient (Amersham Biosciences, Piscataway, NJ, USA; no. 17-1440-02) and centrifuged at 2500 g for 30 min. Cells were harvested from the interphase (density < 1.075 g mL−1), washed and collected by centrifugation (1500 g, 15 min). Cells were cultured at a density of 0.2–0.5 × 106 cells cm−2 either at 20% O2, 5% CO2 and 37 °C (Heraeus, HeraCell 240, Keudro, Vienna, Austria) or at 3% O2, 5% CO2 and 37 °C (Thermo Electron Corporation 3110, Bartelt, Graz, Austria). After 24 h, the nonadherent cell fraction was removed by washing twice with phosphate buffered saline (PBS; Gibco; no. 14190-094, Invitrogen GmbH, Lofer, Austria). Since considerably more colonies were obtained from the cell fraction isolated from bone using collagenase, all subsequent procedures were carried out with cells obtained using this method. After the primary culture had reached approximately 30–50% confluence, cells were washed twice with PBS, and subsequently treated with 0.05% trypsin per 1 mm ethylenediaminetetraacetic acid (EDTA) (Gibco; no. 25300-05, Invitrogen GmbH, Lofer, Austria) for 3–5 min at 37 °C. Cells were harvested, washed once in growth medium and further expanded at a density of 50 cells cm−2. At a final density of 1 × 106 cells mL−1, cells were frozen in freezing medium (growth medium supplemented with 30% FCS, 0.5% dimethylsulfoxide (DMSO) (Sigma; no. D-841, Vienna, Austria) with the aid of Cryo Freezing Container (Mr Frosty; Nalgene, Rochester, NY, USA) and thereafter transferred to liquid nitrogen slowly reaching –70 °C over several hours.

Primary bone marrow stromal cell culture

Total bone marrow cells prepared either by centrifugation or by collagenase treatment were plated on plastic (105 per 58 cm2) (BD; no. 353003, BD Biosciences, Schurechat, Austria). After 24 h, the nonadherent cell fraction was rinsed off with PBS. The medium was changed every 3–4 days. After 14 days, the established cultures were fixed in acetone/methanol (1 : 1) and stained with 2% crystal violet. Excess stain was removed with tap water. Only those colonies consisting of more than 20 cells were considered.

Assessment of cell viability

Cell viability was examined by monitoring the reduction of nuclear PI fluorescence together with forward/sideways light scattering analysis (Nicoletti et al., 1991) and the annexinV method (Martin et al., 1996). MSCs were grown to near confluence at 3% and 20% pO2. After trypsination, cells were washed twice in PBS, resuspended in binding buffer (10 mm HEPES, 140 mm NaCl, 2.5 mm CaCl2, 7.4) to 1 × 106 cells mL−1 and stained with 5 µL annexinV-FITC (BD, AnnexinV-FITC Apoptosis Detection Kit I, # 556547) and 5 µL PI for 15 min at room temperature in the dark. Quantification of dead cells was performed with the aid of an argon laser-equipped FACScalibur (BD). Cell debris and small particles were excluded from analysis. Cell death of cultured cells was induced by the addition of 200 µm staurosporin (S-4400; Sigma) for 20 h.

In vitro differentiation

MSCs were plated in duplicates in six-well plates at 50 cells cm−2, grown to near confluence, and then induced to differentiate. In the case of primary cultures, 105 total bone marrow cells per 58 cm2 were grown for 7–10 days, before induction. The induction medium was changed twice per week for a period of 3 weeks. Osteogenesis was stimulated using growth medium containing 20 mmβ-glycerol phosphate, 1 nm dexamethasone and 0.5 µm ascorbate-2-phosphate. For evaluation, the cells were fixed with 4% paraformaldehyde in PBS for 10 min and stained with Alizarin Red, pH 4.1 for 20 min at ambient temperature. Excess stain was removed by several washes with PBS adjusted to pH 4. For adipogenic differentiation of MSC, growth medium was supplemented with 1 µm dexamethasone, 50 µm indomethacine, 0.5 µm 3-iso-butyl-1-methylxanthine and 0.5 µm hydrocortisone. After fixation, lipid vesicles in differentiated cells were stained with 0.7% Oil Red O in 85% propylene glycol for 20 min at room temperature. Excess stain was removed by several washes with PBS. The stained cells were scored with respect to their differentiation potential. Adipogenesis was scored as follows: undifferentiated cells were scored 0; small lipid vesicles around the nucleus, 1; small vesicles together with cells containing large lipid vacuoles, 2; and predominantly large lipid vacuoles, 3. For osteogenesis, no differentiation was scored as 0; < 40% of positive cells as 1; 40–60% as 2; and 60–100% as 3 (see also Fig. 4A).

Determination of AGEs

Cells were washed twice in ice-cold PBS, harvested in lysis buffer containing 50 mm sodium phosphate buffer (pH 7.8), 150 mm NaCl, 1% sodium deoxycholate, 0.1% sodium dodecyl sulfate (SDS), 2 mm EDTA and 1 mm Pefabloc SC (Roche, Basel, Switzerland). Lysates were centrifuged at 14 000 g for 15 min at 4 °C, and the protein concentration was determined by a detergent compatible assay (DC-assay; Bio-Rad, Munich, Germany). After electrophoresis on 4–15% gradient SDS/polyacrylamide gels (Bio-Rad), the proteins were transferred to polyvinylidene difluoride (PVDF) membranes by wet electroblotting and blocked in PBS supplemented with 5% dry milk powder and 0.1% Tween-20. Immunodetection was performed using a mouse monoclonal anti-AGE antibody (6D12; Trans Genic Inc., Kumamoto, Japan) or mouse monoclonal anti-α tubulin (Sigma, Vienna, Austria) followed by horseradish peroxidase-conjugated antimouse secondary antibody (DAKO, Glostrup, Denmark) and enhanced chemiluminescence (Amersham Life Science, Braunschweig, Germany).

SA-β-gal staining

Cells were seeded at a density of 100 per cm2 and grown to 50% confluency. Thereafter, cells were fixed in 4% paraformaldehyde in PBS, washed with PBS and stained with freshly prepared 1 mg mL−1 bromo-chloro-indoyl-galactopyranoside, 20 mm Na2HPO4, 150 mm NaCl, 2 mm MgCl2, 10 mm citric acid, 5 mm potassium ferricyanide and 5 mm potassium ferrocyanide, pH 5.8 for 1–2 days in a humid chamber at 37 °C in the dark.

RNA isolation, array analysis and qPCR

RNA was isolated after homogenization in 4.2 m guanidinium thiocyanate, phenol extraction and ethanol precipitation (Chomczynski & Sacchi, 1987). The resulting total RNA was further purified by LiCl precipitation (final concentration 4.5 m). Forty µg of total RNA were reverse transcribed and labeled cDNA was subjected to microarray analysis using Applied Biosystems 1700 Expression Array, version 2 (Foster City, CA, USA) capable of quantifying the level of 57 967 individual transcripts. Those exhibiting a twofold higher or lower expression level were considered differentially expressed. For qPCR, cDNA was synthesized from total RNA by reverse transcription using RevertAid H Minus MMuLV reverse transcriptase (Fermentas, Vilnius, Lithuania) and oligo(dT) primer (MWG Biotech, Ebersberg, Germany). qRT-PCR assays were performed on a LightCycler instrument (Roche) with the LC-FastStart DNA Master SYBR Green I kit (Roche). The 15-µL reactions contained 2 µm forward and reverse primer and 3 mm MgCl2. After the activation of the enzyme at 95 °C for 8 min, 50 cycles at 95 °C for 15 s, 57 °C for 8 s and 72 °C for 15 s were performed. mRNA expression levels were calculated relative to eukaryotic translation elongation factor 1 α 1 (EEF1α1). Primers were obtained from MWG Biotech. For detailed information regarding primer sequences and transcript IDs of the corresponding genes see Table 1.

Table 1.  Gene lds and respective RT-PCR primer information
GeneNameForward/reverse (5’–3’)AMP (bp)Temperature (°C)Ensembl ID
ALPLAlkaline phosphatase, liver/bone/kidneyCTTCAAACCGAGATACAAGC / TCAGCTCGTACTGCATGTC12560ENST00000344573
BMP1Bone morphogenetic protein 1CGACTGCGGCTATGACTAC / ATCCGAGTGGAACTTCACC13960ENST00000306385
BMP4Bone morphogenetic protein 4AAGAGCAGATCCACAGCAC / GAGATCACCTCGTTCTCAGG17560ENST00000245451
BMPR1ABone morphogenetic protein receptor, type IAGCAATTGCTCATCGAGAC / TGGTATTCAAGGGCACATC13360ENST00000224764
BMPR2Bone morphogenetic protein receptor, type IICACAAATGTCCTGGATGG / CTTCACAGTCCAGCGATTC16460ENST00000374580
BST1Bone marrow stromal cell antigen 1AGCTCTTACAGTGCGTGGAC / GATAAGACCCGCCCTTTG11660ENST00000265016
DKK3Dickkopf homolog 3 (Xenopus laevis)GACAACCAGAGGGACTGC / CAGGTGATGAGGTCCAGAAG13760ENST00000326932
EEF1AEukaryotic translation elongation factor 1 α 1CACACGGCTCACATTGCAT / CACGAACAGCAAAGCGACC19360ENST00000311405
EEF1AEukaryotic translation elongation factor 1 α 1AAGATGGCCCTAAATTCTTG / TTCTTGTCCACTGCTTTGAT16956ENST00000311405
FABP4Fatty acid binding protein 4, adipocyte (aP2)GCTTTGCCACCAGGAAAGT / GGACACCCCCATCTAAGGTT19560ENST00000256104
HIF1AHypoxia-inducible factor 1, α subunitCGTTCCTTCGATCAGTTGTC / TCAGTGGTGGCAGTGGTAGT14360ENST00000323441
HIF1ANHypoxia-inducible factor 1, α subunit inhibitorTTCCCTCCGGATCAGTTC / GGAAATTAGGGAACCTCTCG11260ENST00000299163
IBSPIntegrin-binding sialoprotein (bone sialoprotein, bone sialoprotein II)CCGAAGAAAATGGAGATGACAG / CCATAGCCCAGTGTTGTAGCA14560ENST00000226284
IGF1Insulin-like growth factor 1 (somatomedin C)AGTCAGCTCGCTCTGTCC / TTGGCCAACCTTTCCTTC15060ENST00000337514
IGF1RInsulin-like growth factor 1 receptorCAAGTCCTTCGCTTCGTC / GAAGGAAGGCCTCATCTTG12060ENST00000268035
INHBBInhibin, β B (activin AB β polypeptide)CGCGTTTCCGAAATCATC / GGACGTAGGGCAGGAGTTTC15160ENST00000295228
LIFLeukaemia inhibitory factorCTGAGGTTTCCTCCAAGG / TGTTTCCAGTGCAGAACC9560ENST00000249075
p15Cyclin-dependent kinase inhibitor 2BCGGGGACTAGTGGAGAAGGT / GGTGAGAGTGGCAGGGTCT174, 29760ENST00000276925
p16Cyclin-dependent kinase inhibitor 2ACAACGCACCGAATAGTTACG / AGCACCACCAGCGTGTC177, 45160ENST00000304494
p21Cyclin-dependent kinase inhibitor 1AGGCGGCAGACCAGCATGACAGATT / GCAGGGGGCGGCCAGGGTAT22265ENST00000244741
PH-4Hypoxia-inducible factor prolyl 4-hydroxylaseGAGCATTCAGGAGATGTACG / AGGTATGGTGGCTGTTCC16160ENST00000318125
VHLVon Hippel-Lindau tumor suppressorTCAATGTTGACGGACAGC / GTAGAGCGACCTGACGATG13760ENST00000256474


Array raw data were normalized using CARMAweb (Rainer et al., 2006). Differentially expressed genes were grouped into protein families associated with biological processes or characterized pathways by querying the PANTHER database (Mi et al., 2005).


All values were expressed as means ± standard deviation of the mean. Statistical differences of experimental scores were evaluated using Student's t-tests. Differences were considered significant when the P value was less than 0.05.


This work was supported by the Austrian Science Fund (FWF) through a National Research Network grant (NRN 093 05) to Günter Lepperdinger. We are particularly indebted to Brigitte Greiderer and Evi Hütter for their help, the sharing of reagents and expertise. We would like to thank Dirk Strunk and Peter Ertl for fruitful discussions and careful reading of the manuscript.