Author contributions: M.D.: conception and design, collection and/or assembly of data, data analysis and interpretation, and manuscript writing; M.M. and K.O.: collection and/or assembly of data; J.P. and L.E.M.: conception and design and data analysis and interpretation; M.B.: provision of study material or patients; D.L-A.: data analysis and interpretation and financial support; H.P.: conception and design, data analysis and interpretation, financial support, manuscript writing, and final approval of manuscript.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS December 7, 2012.
A major limitation in the development of cellular therapies using human mesenchymal stem cells (hMSCs) is cell survival post-transplantation. In this study, we challenged the current paradigm of hMSC survival, which assigned a pivotal role to oxygen, by testing the hypothesis that exogenous glucose may be key to hMSC survival. We demonstrated that hMSCs could endure sustained near-anoxia conditions only in the presence of glucose. In this in vitro cell model, the protein expressions of Hif-1α and angiogenic factors were upregulated by the presence of glucose. Ectopically implanted tissue constructs supplemented with glucose exhibited four- to fivefold higher viability and were more vascularized compared to those without glucose at day 14. These findings provided the first direct in vitro and in vivo demonstration of the proangiogenic and prosurvival functions of glucose in hMSC upon transplantation and identified glucose as an essential component of the ideal scaffold for transplanting stem cells. STEM CELLS2013;31:526–535
Mesenchymal stem cells (MSCs) hold considerable promise in bioengineering because of their ability to differentiate into various phenotypes. To date, MSCs have not met this promise, in part due to their high death rate upon transplantation [1–5]. In fact, cell transplantation strategies to replace damaged lost tissues are limited by the inability of MSCs to resist cell death in engineered construct after implantation.
A likely explanation for this limited cell survival is that, upon implantation, MSCs encounter an ischemic environment composed of both low oxygen tension, nutrient deprivation . Although either one of these insults has the potential to affect cell survival significantly, most studies have focused on the role of oxygen, because it modulates several critical cellular processes (e.g., cell adhesion [7, 8], metabolism, proliferation, and differentiation [9, 10]). In addition, oxygen is a poorly diffusive molecule whose passive diffusion from capillaries is limited to 100–200 μm [11, 12]. In fact, this limited diffusibility of oxygen poses a crucial challenge to tissue engineering, and it is held responsible for the difficulty in obtaining functional tissue construct of pertinent volume for clinical application [13–15].
In this study, we challenged the current paradigm that assigns a pivotal role to oxygen alone. We hypothesized that exogenous glucose (the main “metabolic fuel” source for MSCs) plays a key role in the survival of human MSCs (hMSCs) under transplantation conditions. We exposed hMSCs in vitro to a sustained, near-anoxic environment and assessed the influence of exogenous glucose on (a) cell viability, (b) expression of the hypoxia inducible factor Hif-1α, (c) secretion of angiogenesis proteins, and (d) functions pertinent to new tissue formation. We further investigated the critical contribution of glucose to in vivo survival of implanted hMSCs and to peri-implant vascularization in an ectopic mouse model.
MATERIALS AND METHODS
Chemicals and Molecular Assay
Details regarding chemicals and respective suppliers were as follows: Alpha minimum essential medium without glucose (αMEM 0glc), Dominique Dutscher (Brumath, France, http://www.dutscher.com). Glucose, Mannitol, deferoxamine, and hyaluronic acid, Sigma-Aldrich (St. Louis, MO, http://www.sigmaaldrich.com); antibiotics, trypsin, and fetal bovine serum (FBS), PAA (Pasching, Austria, http://www.paa.at); Guava reagents for flow cytometry and angiogenesis protein assay, Millipore (Bedford, MA, http://www.millipore.com); CellTiter-Glo Luminescent cell assay, Promega (Mannheim, Germany, http://www.promega.com); enzyme-linked immunosorbent assay (ELISA) assay, Invitrogen (Saint Aubin, France, http://www.invitrogen.com), and Hif-1α Activation Kit, Thermo Scientific (Quebec, Canada).
Near-anoxia was established, and maintained, in a sterile, 37°C, humidified environment using a hypoxic incubator (BINDER CO2 incubators CB-210; Binder Scientific, France, http://www.binder-world.com).
Pericellular pO2 was measured using an oxygen electrode connected to an Oxylab pO2 meter (Oxford Optronix; Oxford, U.K., http://www.oxford-optronix.com). To ensure sustained near-anoxia as well as cell-driven nutrient depletion, the culture plates containing the cells were not disturbed and the supernatant medium was not changed for the duration of the study.
Cells and Cell Culture
hMSCs were isolated from bone marrow obtained as discarded tissue during routine bone surgery from five donors by the Lariboisiere Hospital (Paris, France). hMSCs were isolated using a procedure adapted from literature reports . Cell passages 2–3 were used for the experiments that are described in the sections that follow. For the in vitro part of the study, MSCs obtained from five donors were used. Each test was conducted in triplicate using cells from each one of the five donors. For the in vivo part of the study, each test was conducted in sextuplicate using a pool of MSCs from the five donors who were also used in the in vitro part of this study.
hMSC Viability Under Sustained Near-Anoxia in Either the Presence or Absence of Glucose
hMSCs (1.25 × 104 cells per centimeter square) from each donor were seeded into individual wells of a 24-well plate, cultured overnight, washed twice with phosphate buffer solution (PBS), and maintained in near-anoxia under αMEM supernatant medium in either the absence or presence (1 or 5 g/l) of glucose. Cell viability and glucose and lactate levels were determined at days 3, 7, 14, and 21 of culture unless otherwise stated.
Determination of Glucose, lactate, and ATP
Glucose and lactate levels were monitored using a biomedical ARCHITECT C8000 (Abbott Diagnostic) robot. Intracellular ATP content was quantified using the CellTiter-Glo Luminescent cell assay (Promega), according to manufacturer's instructions. Data were expressed as “fold increase” compared to data obtained on day 0.
Viable and apoptotic cells were identified using the Viacount assay and the annexinV-phycoerythrin (PE)/7 amino-actinomycin (7-AAD) assay, respectively, according to the manufacturer's instructions. Briefly, the Viacount assay is based on incorporation of fluorescent propidium iodide (PI) after loss of cell membrane integrity, and the annexinV-PE/7-AAD assay is based on the fact that, in early apoptosis, annexin V binds to phosphatidylserine residues on the outer leaflet of the cell membrane, and that 7-AAD is excluded by viable cells. The cell cycle phase of hMSCs was identified using the Guava Cell Cycle Assay according to the manufacturer's instructions. Briefly, the Cell Cycle Assay uses PI, a nuclear DNA stain, to identify cell-cycle phase. Resting cells (G0/G1) contain two copies of each chromosome. Cycling cells synthesize chromosomal DNA (S-phase), which results in increased fluorescence intensity. When all chromosomal DNA has doubled (G2/M phase), the cells fluoresce with twice the intensity of the initial population. Data were collected using a Guava Easy-CyteTM PCA-96 System and were analyzed assessed using Viacount and Nexin and Cell cycle software, respectively, to quantify cell viability, apoptosis, and cell cycle.
Adenosine monophosphate-activated protein kinase (AMPK) activity, extracellular angiopoietin-2, and vascular endothelial growth factor (VEGF)-C concentration were determined by ELISA analysis. hMSCs were exposed to 0% O2 in the absence or in the presence of glucose (0.1, 1, and 5 g/L) for 3 and 7 days. At the prescribed time, hMSCs were lysed using cell lysis buffer containing phenylmethanesulfonylfluoride and a protease inhibitor. Each cell extract was centrifuged and treated according to manufacturer's instructions.
hMSCs Viability upon Reperfusion
After 21 days of exposure to sustained near-anoxia, hMSCs were transferred to normoxic (21% O2) conditions with αMEM containing 1 g/l of glucose and 10% FBS. Controls were hMSCs cultured in normoxic conditions at all times.
The phenotype of the isolated hMSCs was determined using flow cytometry. For this purpose, hMSCs were pretreated with the appropriate monoclonal antibody in the dark, at room temperature, for 30 minutes. Details regarding antibodies used and suppliers were as follows: PE-conjugated antibodies against CD31, CD73 (BD Pharmingen, San Diego, http://www.bdbiosciences.com/index_us.shtml), CD45, CD105 (Beckman Coulter, Fullerton, CA, http://www.beckmancoulter.com); fluorescein isothiocyanate (FITC)-conjugated antibodies against CD90 (Beckman Coulter) and CD34 (Beckman Dickinson). Nonspecific fluorescence was detected using goat anti-mouse immunoglobulin (Beckman Coulter).
Effects of Glucose on Hif-1α Expression and Bioactivity
Expression and bioactivity of Hif-1α were determined by three different methods, all with the following procedures in common: hMSCs were exposed to either 21% or 0% O2 in the absence or in the presence of glucose (0.1, 1, and 5 g/l) for 3 days. The positive control for Hif-1α expression was obtained by adding 500 μM of deferoxamine (Sigma) to the supernatant medium of hMSCS cultured under normoxic (i.e., 21% oxygen) condition. Method-specific details follow.
Immunochemistry of Hif-1α
Hif-1α expression was assessed using the Hif-1α Activation Kit (Thermo Scientific), following the manufacturer's instructions. hMSCs were examined using confocal microscopy (LSM ZEISS 510).
Western Blot and Hif-1α Expression Analysis
Hif-1α expression was determined using a previously published technique . Antibodies against Hif-1α (1:1,000; Novus Biological) were used for immunoblotting. Specific bioluminescence signals was detected by bioluminescence (BLI) using an IVIS Lumina Bioluminescent imaging system (Xenogen, Caliper Life Science, Tremblay-en-France, France, http://www.caliperls.com) and were quantified by the IVIS Lumina imaging software (Living Image Software, version 3.1).
hMSCs were transfected with a pGL3/5HRE-CMV-Luc plasmid kindly donated by Dr. Masahiro Hiraoka containing five hypoxia responsive elements (HRE) sequences. Transfection was accomplished using the Amaxa Nucleofector II Device nucleoporation system (Lonza Walkersville, Walkersville, MD, http://www.lonza.com) following the manufacturer's instructions. Bioactivity of Hif-1α was assessed by BLI.
Assessment of MSCs Survival in 3D Constructs
The in vivo effect of glucose supplementation on cell viability was assessed using a mouse transplantation model (8-week-old female nih/nu/ xid/bg mice; Harlan-Sprague Dawley, Indianapolis, IN, http://www.harlan.com). All animal procedures were performed in compliance with institutional published guidelines (Directive du Conseil 24.11.1986. 86/609/CEE).
Tissue Construct Preparation
Cylindrical polyacrylonitrile-sodium methallyl sulfonate (PASM) scaffolds (diameter: 7 mm; height: 5 mm) were donated by Dr. Jiri Honiger (Saint-Antoine Hospital, Paris, France). The scaffolds were prepared by phase inversion of a polymer solution containing 6% acrylonitrile and sodium methallylsulfonate copolymer, 91% dimethylsulfoxide, and 3% physiological saline (0.9% NaCl) solution (w/w/w) . The scaffolds had pores with a mean diameter of 500–1,000 mm and a porosity of 77%. They were sterilized via immersion in 10% Dialox in physiological saline (v/v). After a thorough rinse of the PASM scaffolds using sterile physiological saline, hMSCs (3 × 105 cells) genetically modified by rMLV-LTR-eGFP-luc retroviral vector were seeded by injection and were cultured in αMEM (containing 10% FBS) at 37°C overnight.
In Vivo Experiments
At the time of implantation, 100 μl of either hyaluronic acid (2%) or fibrin gel (9 mg/ml) containing either 0 or 10 g/l of glucose was gently injected inside each cell-containing construct. Two cell constructs (loaded vs. not loaded with glucose) were implanted subcutaneously per animal on six mice and were imaged by BLI at days 1, 4, 7, and 14 postimplantation.
Assessment of HRE Expression in Vivo
The effect of glucose on HRE expression was assessed in vivo. Briefly, 3 × 105 hMSCs transfected with pGL3/5HRE-CMV-Luc were seeded by injection into cylindrical PASM scaffolds and were cultured in αMEM (containing 10% FBS) at 37°C overnight. At the time of implantation, these scaffolds were embedded into fibrin gel (9 mg/ml) containing either 0 or 10 g/l glucose. Two cell constructs per animal (loaded vs. not loaded with glucose) were implanted subcutaneously on six mice and were imaged by BLI at days 1, 4, 7, 10, and 14 postimplantation.
Assessment of Implanted Constructs Vascularization 21 Days After Implantation
The in vivo effect of glucose on vascularization was histologically assessed. Briefly, 21 days postimplantation, constructs were explanted, fixed in 0.4% Paraformaldehyde (PFA) for 12 hours, and embedded in paraffin; thin (8 μm) sections of each construct were stained with hematoxylin, eosin, and safran. Peripheral vascularization was quantified in the area within 1 mm around each construct. Vascularization was also immunohistochemically assessed using FITC-conjugated Griffonia Simplicifolia Isolectine B4 (Sigma) according to the manufacturer's instructions. Briefly, embedded-paraffin-sections were stained with Isolectine B4 (dilution of 1/250), in the dark, at room temperature for 1 hour. Sections were examined using a fluorescent microscope (Nikon Eclipse TE2000-U; Nikon, Champigny sur Marne, France) fitted with a digital camera (DXM1200F). The Nikon NIS element F 2.20 software was used for imaging and analysis.
Numerical data are expressed as the mean ± SD. Statistical analyses were performed with ANOVA for in vitro data and Mann–Whitney U tests for in vivo data (GraphPad Prism Software). Asterisks indicate significant differences, as follows: *, p < .05; **, p < .001. Significant differences among the data of the same group are indicated by the symbol “#” for p < .001 in comparison with day 3.
Establishment of Near-Anoxia and Ischemia
To validate our in vitro model, hMSCs in α-MEM without serum in the absence or presence (either 1 or 5 g/l) of glucose were cultured in a humidified, 37°C, 5% CO2, 95% N2 incubator set at 0% oxygen for 21 days. The time course of oxygen tension, lactate (a hallmark of anaerobic metabolism), and glucose were determined by testing samples of supernatant cell medium (Fig. 1A). To prevent reoxygenation, to ensure cell-driven nutrient depletion, the culture plates containing the cells were not disturbed, and the medium was not changed over the course of the experiments.
In this experimental setting, when the cells were transferred to the environment of 0% oxygen, the oxygen tension dropped from 21% to 2.8% within 2 hours (Fig. 1B). After 72 hours, the oxygen level reached 0.12% and remained constant at this level for the duration of the experiments. This result demonstrated that hMSCs were exposed to sustained, near-anoxia.
Based on the time course of glucose concentration changes in the medium, hMSCs cultured in the absence of glucose faced abrupt ischemia from day 0 (Fig. 1C). hMSCs cultured in the presence 1 g/l of glucose faced complete depletion of glucose by day 14, and hMSCs cultured in the presence 5 g/l of glucose did not face glucose exhaustion during the entire 21 days of the experiment. The kinetics of lactate revealed that hMSCs had shifted to anaerobic metabolism, and the lactate accumulated in the supernatant medium only in the presence of glucose (Fig. 1D). In these conditions, hMSCs culture with 0, 1, and 5 g/L of glucose, respectively, faced abrupt ischemic like, cell driven ischemic, and near-anoxic like conditions for the duration of the experiments.
Abrupt and Cell-Driven Ischemia, but not by Sustained Near-Anoxia Affected hMSC Viability
Cell death during ischemia might result from either a shortage of glucose or exposure to near-anoxia. To assess the respective roles of near-anoxia and glucose concentrations under the conditions tested, we compared the survival rates of hMSCs cultured with or without glucose.
Under abrupt ischemic conditions (0 g/l glucose), the number of hMSCs decreased over 21 days (Fig. 2A). In addition, the ATP content per cell drastically decreased from day 3 to day 21 (Fig. 2B). Under cell-driven ischemic conditions (1 g/l glucose), hMSCs survived for 7 days; then, the number of viable cells significantly decreased from day 7 to day 21. The ATP content per cell remained stable for the first 7 days but rapidly decreased from days 7 to 21 of culture. Under near-anoxic conditions (5 g/l glucose), the number of viable hMSCs and the ATP content per cell remained stable for the duration of the experiment. Interestingly, at day 3, although ATP content per cell was similar in all conditions tested, a downregulation of phosphorylated AMPK (a key sensor of fuel and energy status in cell) was already observed (Fig. 2C).
Quantification of Annexin V expression (a marker of apoptosis) in hMSCs revealed that, after 3 days of either near-anoxia or ischemia, only 15% of hMSCs were Annexin V-positive (Fig. 2D). These results were corroborated by the absence of induction of phosphatidylinositide 3-kinase (PI3K)/mammalian target of rapamycin (mTOR) pathway  (as measured by phosphorylated p70S6K data not shown). In contrast, after 21 days in near-anoxia, the number of Annexin V-positive cells was significantly lower than that observed in hMSCs exposed to abrupt and cell driven ischemia (19% vs. 45%, p < .001; 19% vs. 38%, p < .001, respectively).
To exclude that the effect of glucose on MSC viability was driven by the glucose-mediated hypertonicity of the cell environment, hMSCs were cultured under near anoxic conditions in the presence of 5 g/l mannitol (Fig. 2A). Under these conditions, hMSCs viability was not statistically different from the one observed in the absence of glucose suggesting that hypertonicity is not cytoprotective. Moreover, to verify that only glucose, not glutamine (an alternative source for ATP production), is responsible of cell survival in ischemia, the role of glutamine was evaluated in the in vitro cell model tested in this study. We found that adding glutamine in the supernatant medium did not prevent cell death and did not maintain the ATP content (Supporting Information Fig. 1). Taken together, these results demonstrated that, in the near-absence of oxygen, glucose depletion (abrupt or cell driven) affected hMSC viability. In contrast, hMSCs remained viable in sustained near-anoxia conditions in the presence of glucose.
hMSCs Exposed to Near-Anoxia Remained Viable and Retained Proliferative Ability After Reperfusion
To assess the functional status of hMSCs after a 21-day exposure to either sustained near-anoxia, abrupt ischemia, or cell-driven ischemia, we simulated the conditions of blood reperfusion, and then assessed select hMSC stem cell function using observation of morphology, CD marker characterization, and cell cycle analysis.
We found that hMSCs cultured under either abrupt or cell-driven ischemia did not proliferate upon reperfusion (Fig. 3A). In contrast, hMSCs exposed to near-anoxia for 21 days exhibited normal fibroblast-like morphology after reperfusion (Fig. 3A). Cell cycle analyses showed that these hMSCs proliferated at a normal rate, that is, the percentage of mitotic cells was similar to that observed under standard cell culture conditions (Fig. 3B). Flow cytometry analysis revealed that, after 21 days under near-anoxic conditions and 3 days after reperfusion, the cells were positives for CD73, CD90, and CD105, and negatives for CD31, CD34, and CD45 (Fig. 3C). Taken together, these data confirmed that in sustained near-anoxic conditions the presence of a glucose supply was necessary to ensure cell viability and function.
Glucose Regulated Hif-1α, Angiopoietin-2, and VEGF-C Expression
The hypoxia-inducible factor Hif-1α is a major regulator of the cellular response to hypoxia because it activates the transcription of many genes, including those involved in cell survival, angiogenesis, oxygen delivery, and metabolic adaptation to hypoxia. To evaluate the effect of glucose on Hif-1α expression, immunohistochemistry was performed on hMSCs cultured for 3 days under various conditions including the following: standard cell culture conditions in the absence (negative control) or presence (positive control) of deferoxamine and near-anoxia in the absence or presence of glucose (1 and 5 g/l) (Fig. 4A). Under normoxia, no expression of Hif-1α was detected; however, in the presence of deferoxamine (positive control for Hif-1α expression), Hif-1α was localized in the cell nucleus. Under near-anoxia, Hif-1α was always localized in the nucleus, regardless of glucose concentration. Most importantly, analysis of images acquired with the same exposure settings showed that Hif-1α expression was positively correlated with increases in glucose concentration in the supernatant medium (Fig. 4A). This observation was also confirmed by Western blot analysis (Fig. 4B).
The effect of glucose concentration on Hif-1α bioactivity was also evaluated by transfecting hMSCs with the reporter plasmid, pGL3/5HRE.CMVmp-Luc, which contained five copies of the hypoxia responsive element (HRE), a DNA binding sequence for Hif-1α. The intensity of the Hif-1α-induced signal was directly correlated to the glucose concentration present in the supernatant media of hMSC cultures exposed to near-anoxia for 3 days. These results showed that the increases in Hif-1α expression resulted in increased bioactivity (Fig. 4C).
To further investigate the role of glucose concentration on angiogenesis, the expression of angiopoietin-2 and VEGF-C was quantified by ELISA assays. Results showed that glucose significantly enhanced the amount of angiopoietin in hMSC cultures exposed to near-anoxia at 3 and 7 days (a fourfold increase when compared to results obtained without glucose) (Fig. 4D). Moreover, in the presence of glucose, VEGF-C level increase in hMSC exposed to near-anoxia at 3 and 7 days. Taken together, these results provided evidence that the glucose concentration enhance Hif-1α expression and secretion of angiogenic factors (such as angiopoietin and VEGF-C).
Glucose Enhanced In Vivo hMSCs Survival in 3D Constructs
To extend the results of the in vitro studies to the in vivo setting, we assessed hMSC survival in an ectopic transplantation mouse model. Briefly, eGFP-Luc hMSCs expressing constitutively the green fluorescent protein and the gene of luciferase (eGFP-Luc hMSCs) were seeded into PASM scaffold in the absence of glucose. The hMSC PSAM scaffolds were implanted ectopically in the back of mice, and the BLI of each implant (a noninvasive measurement of transplanted hMSC viability) was quantified on days 1, 4, 7, and 14 postimplantation. The number of viable cells decreased from days 1 to 14 (Fig. 5A). On day 14, only 15% viable cells were detected.
In order to limit glucose (which is a small, highly diffusive molecule) leakage outside the cell-containing construct, glucose was loaded into either fibrin or hyaluronic acid hydrogels with which the PASM scaffold were filled before implantation. The number of viable cells in PASM scaffolds loaded with fibrin alone significantly decreased from days 1 to 14, and, on day 14, only 15% viable cells were detected (Fig. 5B). In contrast, the number of viable cells in PASM scaffolds loaded with fibrin and glucose remained constant from days 1 to 7 but decreased thereafter. Importantly, the number of viable cells on day 14 in PASM scaffolds loaded with fibrin was five times higher in the presence of glucose than that observed in the absence of glucose. To confirm the positive influence of glucose on hMSC viability, we conducted a similar study with hyaluronic acid as the delivery system for glucose. Of interest, in these conditions, the number of viable cells increased from days 1 to 7 and remained constant between days 7 and 14 (Fig. 5C). Importantly, the percentage of viable cells on day 14 was 4.5-fold higher in the presence of glucose than that observed in the absence of glucose. Taken together, these results established, for the first time, that the presence of glucose in engineered scaffold could significantly limit and even prevent massive cell death upon implantation of cell containing constructs.
Glucose Enhanced HIF-1 Bioactivity and Vascularization of 3D Construct
To evaluate the effect of glucose on HIF-1 bioactivity, HRE-Luc hMSCs were seeded into PASM scaffolds, embedded in a fibrin gel, and implanted ectopically in the back of mice in either the presence or in the absence of glucose. The BLI of each implant was quantified on days 1, 4, 7, 10, and 14 (Fig. 6A). In the absence of glucose, the BLI signal remained constant from day 1 to 14. In contrast, in the presence of glucose, the BLI signal increased from day 1 to 10 but remained stable at day 14. At day 14, the BLI signal was seven times higher in the presence of glucose (2.12 vs. 15; p < .001). Taken together, these data suggest that the presence of glucose enhance Hif-1α bioactivity.
To evaluate the effect of glucose on vascularization of implanted 3D construct they were imaged (Fig. 6C), explanted, embedded, included, and stained by Hematoxylin-Eosin-Safran (Fig. 6B). The number of peripheral blood vessels was determined from histological sections (1 mm thin) from tissue surrounding each implant. In the absence of glucose, few peripheral blood vessels were observed. In contrast, when the constructs were supplemented with either hyaluronic acid and glucose or with fibrin gel and glucose, the number of peripheral blood vessels increased significantly (2 vs. 4; p < .01 and 2 vs. 9; p < .001, respectively) (Fig. 6E). The presence of blood vessels was confirmed immunochemically by staining endothelial cells with isolectine B4 (Fig. 6D).
In order to realize the full therapeutic potential of hMSCs, their survival rate upon implantation must be improved. To achieve this important goal, we require a better understanding of the underlying mechanisms of cell death upon implantation, and we must develop a niche that reduces hMSC sensitivity to ischemia.
In this study, we developed an in vitro model to investigate the deleterious effects of ischemia on hMSC survival and function under conditions of near-anoxia, that is, low pO2, Hif-1α expression, and anaerobic metabolism. In the absence of glucose, most hMSCs cultured under near-anoxia died within 14 days; this result established glucose as a key player in hMSC survival. An alternative explanation might be that a sudden metabolic shift induced by abrupt depletion of glucose had caused massive cell death. To exclude this alternative possibility, hMSCs were cultured in near-anoxia in the presence of 1 g/l glucose. Under these conditions, hMSCs faced a progressive cell-driven exhaustion of glucose. Thus, by day 14, near-anoxic (low pO2) conditions switched to ischemic (low pO2 and glucose depletion) conditions (Fig. 2). Ischemia caused an early AMPK upregulation, cell shrinking, reduced cell viability, and resulted in low ATP content. We also definitively ruled out a possible effect of glucose-mediated hypertonicity on cell viability (Fig. 2A) and the contribution of glutamine as an alternative nutrient source in this model (Supporting Information data 1). When exposed to glutamine alone under near-anoxic conditions, hMSCs died at day 7. These results provided evidence that both abrupt and cell-driven glucose depletion led to massive cell death and validated the present model.
To confirm that glucose, not near-anoxia, was the key factor in the cell death process, hMSCs were exposed to severe, continuous (21 days) near-anoxic condition in the presence of glucose (5 g/l). Under these conditions, MSCs remained viable, expressed characteristic phenotypic markers and proliferated in vitro (after simulation of reperfusion) (Fig. 3). These findings challenged the traditional view that severe near-anoxia per se is responsible for the massive MSC death observed postimplantation [15, 19]. The results of this study provided evidence that hMSCs can withstand exposure to severe, continuous near-anoxia, provided that glucose is available during all the duration of the experiment.
When hMSCs were exposed to near-anoxia in the presence of various glucose concentrations, a dose effect on Hif-1α (a key regulator of the cellular response to hypoxia ) expression and bioactivity was observed. This result provided the first evidence that glucose, in addition to supplying fuel for ATP production, is required for the hMSC response to near-anoxia through Hif-1α activation (Fig. 4) and confirmed previous studies with carcinoma cells .
A key observation of this study is that glucose supplementation of cell-containing constructs in vivo resulted in a four- to fivefold increase of the bioluminescent signal directly correlated with the percentage of present viable cells [18, 22]. These results are the first direct demonstration that glucose per se significantly reinforced the ability of hMSCs to survive in vivo transplantation. They are important in the context of tissue engineering, because they identified glucose as an essential component of the “ideal niche” to reduce MSC sensitivity to ischemia. More interestingly, the presence of glucose strongly enhanced peripheral vascularization of the implanted tissue constructs (Fig. 6). This desirable result might be due to in vivo overexpression of Hif-1α, which activates a large panel of genes involved in angiogenesis  including angiopoietin and VEGF-C which were both overexpressed in vitro in the presence of glucose under near anoxic condition (Fig. 4D, 4E). In addition, exposure to high glucose concentrations may induce locally protein glycation, which affects the local immune surveillance. Although, this outcome is unlikely in this study due to the limited duration of the experiment, the occurrence of such a glycation process cannot be excluded during long-term experiments (for review ).
In the presence of glucose, the number of hMSCs in the constructs on day 14 was, at best, similar to that of MSCs in constructs at day 0. Although we cannot exclude the possibility that only a transient exposure to glucose is needed to reinforce hMSC survival in vivo, we speculate that, the full potential of glucose in fostering MSC survival and construct vascularization after implantation will be realized with advanced drug delivery systems that entrap glucose and release it at a rate that matches MSC demand for glucose over a long period of time. An alternative, but not mutually exclusive, explanation for this observation might be that cell death occurred after several insults, including the absence of glucose, metabolic waste accumulation with concomitant pH changes, and induction of inflammatory reaction [6, 25].
This study was the first to provide in vitro and in vivo demonstrations that glucose (but not glutamine) significantly enhanced the ability of hMSCs to survive in a near-anoxic environment. At last but not the least, in vitro and in vivo glucose supply significantly enhances Hif-1α expression and angiogenesis by the secretion of angiogenic factors such as angiopoietin and VEGF-C. This finding provides valuable insights to current understanding of the mechanisms underlying MSC death upon implantation. Our in vivo results provided evidence that glucose improved the viability of a tissue-engineered construct after implantation. These experiments demonstrated the feasibility of our strategy and have important implications for applications of major clinical impact.
We thank Dr. J. Honniger for donating the PASM scaffolds; Professor R. Bizios for valuable comments on the manuscript; Y. Calando for immunochemistry advice regarding Hif-1α and the IFR65 platform for the quantification of angiopoietin and VEGF-C. We also acknowledge the financial support from the Fonds d'amorçage Biothérapie BTH06003, the ANR 07-RIB-011-01 MYOCELLOS and ANR 08-TECS-004 GLASSBONE, and the Contrat d'Interface AP-HP/INSERM.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.