Author contributions: H.S. and C.C.: conception and design, carrying out experiments, data analysis and interpretation, and manuscript writing; W.Y.: design and supervision of the study, data interpretation, and manuscript revision. S.-H.H. and C.-T.C. contributed equally to this article.
Department of Biochemistry and Molecular Biology, National Yang Ming University, Taipei, Taiwan
Department of Biochemistry and Molecular Biology, National Yang Ming University, Taipei, Taiwan
Department of Medicine, Mackay Medical College, Taipei, Taiwan
Correspondence: Yau-Huei Wei, Ph.D., Department of Biochemistry and Molecular Biology, National Yang Ming University, 155 Li-Nong St., Sec. 2, Peitou, Taipei 112, Taiwan. Telephone: 886-2-28267118; Fax: 886-2-28264843; e-mail: email@example.com
We previously demonstrated that metabolic switch and mitochondrial activation are required for osteogenic differentiation of human mesenchymal stem cells (hMSCs). However, stem cells in niches or transplanted into injured tissues constantly encounter hypoxic stress that hinders aerobic metabolism. Therefore, we investigated the effects of oxygen tension (1% vs. 21%) on metabolism and osteogenic differentiation of hMSCs. We found that hypoxia impaired osteogenic differentiation as indicated by attenuation of alkaline phosphatase activity and expression of osteogenic markers core binding factor a-1 and osteopontin. In addition, differentiation-induced mitochondrial activation was compromised as shown by the decrease in the expression of respiratory enzymes and oxygen consumption rate. On the contrary, anaerobic metabolism was augmented as revealed by the upregulation of glycolytic enzymes and increase of lactate production, rendering the cells to rely more on anaerobic glycolysis for energy supply. Moreover, administration of 2-deoxyglucose (a glycolytic inhibitor) but not antimycin A (a respiratory inhibitor) significantly decreased intracellular ATP levels of hMSCs differentiating under hypoxia. Treatment with cobalt chloride, a hypoxia-inducible factor-1α (HIF-1α) stabilizer, recapitulated the inhibitory effects of hypoxia, suggesting that HIF-1α is involved in the compromise of hMSCs differentiation. These results suggest that hypoxia inhibits metabolic switch and mitochondrial function and therefore suppresses osteogenic differentiation of hMSCs. Stem Cells Stem Cells2013;31:2779–2788
Stem cells hold great promise in the application of cell therapy and tissue engineering due to their self-renewal ability and potential of differentiation into various cell lineages. The former allows stem cells to proliferate indefinitely in vitro, and the latter enables them to give rise to progenies of different types of tissue cells. The stem cell competence is governed by not only intrinsic signals such as transcriptional and epigenetic control of stemness genes but also extrinsic factors such as the microenvironment and oxygen tension in the niches they reside . Recently, the metabolic properties of stem cells have gained increasing interest in terms of the regulation of their differentiation, self-renewal, and even reprogramming at the molecular and cellular levels [1-3]. For example, it has been found that embryonic stem cells (ESCs) tend to keep a low rate of aerobic metabolism when they are maintained undifferentiated, but an upregulation of mitochondrial function is essential for their successful differentiation [4, 5]. In addition, hematopoietic stem cells were reported to have a reduced amount of mitochondria and low mitochondrial respiration rate even under normoxic environment, which indicates that they rely mainly on glycolysis for energy supply . Furthermore, recent studies demonstrated that reprogramming of fibroblasts into induced pluripotent stem cells can result in mitochondrial resetting and regain of the proglycolytic phenotype that resembles the metabolic features of ESCs [7, 8]. Together, these findings have implied the importance of regulation of energy metabolism, such as mitochondrial biogenesis and glycolytic flux, in the homeostasis and differentiation of stem cells, although the detailed mechanisms have remained unclear.
In a previous study, we demonstrated that during osteogenic differentiation of human mesenchymal stem cells (hMSCs), mitochondrial oxidative metabolism was significantly activated as revealed by the upregulation of respiratory enzymes and increase of the oxygen consumption rate (OCR) of mitochondria . Conversely, the glycolic activity was dramatically reduced as shown by the decrease of lactate production, which led us to propose that there is a metabolic switch in stem cell differentiation . In addition, we observed a coordinated augmentation of antioxidant defense system during hMSCs differentiation, which was thought to be a response to cope with the reactive oxygen species (ROS) derived from the electron transport chain. Furthermore, inhibition of mitochondrial function by oligomycin A (an inhibitor of respiratory enzyme Complex V) or increase of intracellular ROS by exogenous hydrogen peroxide (H2O2) was found to disturb the osteogenic differentiation of hMSCs. These findings suggest that upregulation of mitochondrial function and ROS scavenging enzymes are important for proper differentiation of hMSCs in vitro. However, under physiological circumstances, stem cells reside in their niches with low oxygen tension . Besides, transplantation of stem cells after tissue injuries usually exposes the cells to low oxygen concentration due to the lack of vascular supply from damaged blood vessels. In addition, in tissue engineering, stem cells cultured on three-dimensional scaffolds are also subjected to a microenvironment of low oxygen concentration . The above-mentioned observations indicate that stem cells to be applied in regenerative medicine will inevitably be exposed to hypoxic challenges. Whether hypoxia will cause an inhibitory effect on the activation of aerobic metabolism and subsequently lead to compromised differentiation and functional maturation of stem cells awaits further investigation.
When cells are exposed to a hypoxic environment, hypoxia-inducible factor-1α (HIF-1α) plays an important role in directing cells from aerobic metabolism to anaerobic glycolysis [13, 14]. Under normal oxygen tension, glucose is metabolized to pyruvate and then converted to acetyl-CoA by pyruvate dehydrogenase (PDH) to enter TCA cycle where NADH and FADH2 are produced as reducing equivalents to drive the respiratory chain, which results in the production of ATP using oxygen as the terminal electron acceptor. However, when oxygen is in paucity, cells tend to fuel themselves through anaerobic glycolysis to meet the energy demand and the metabolic switch is mediated by HIF-1α. It has been documented that HIF-1α can activate gene expression of an array of glycolytic enzymes and pyruvate dehydrogenase kinase 1 (PDK1) . PDK1 phosphorylates the E1α subunit of PDH and inactivates the PDH enzyme complex, thereby preventing pyruvate from entering the TCA cycle and results in the attenuation of mitochondrial respiration [15, 16]. However, whether this regulatory mechanism elicited by HIF-1α is also involved in the metabolic switch of stem cells during the differentiation process is not clear.
Hypoxia has been shown to enhance stemness and repress differentiation of stem cells , but the role of hypoxia-induced metabolic changes in this regulation is not fully understood. In this study, we tested the hypothesis that mitochondrial function is attenuated under hypoxia and consequently the osteogenic differentiation capability of hMSCs is compromised. We found that osteogenic differentiation of hMSCs was repressed under hypoxia (1% oxygen), and the expression levels of the subunits of respiratory enzyme complexes and mitochondrial OCR were significantly decreased. Conversely, anaerobic glycolysis was drastically enhanced as revealed by the upregulation of glycolytic enzymes and increased lactate production. Treatment with metabolic inhibitors revealed that osteoblasts under hypoxia failed to activate mitochondrial respiration and relied extensively on glycolysis for energy supply. Similar findings were obtained when cells were treated with cobalt chloride (CoCl2) as the hypoxic stressor, which indicates that the attenuation of metabolic switch could lead to a decrease in the osteogenic differentiation capability of hMSCs under hypoxia. Understanding of the metabolic regulation may facilitate the development of effective transplantation of stem cells to damaged tissues exposed to a hypoxic environment.
Materials and Methods
Culture of Bone Marrow hMSCs
Human MSCs were isolated from the bone marrow aspirates of three healthy donors who received fracture fixation surgery after approval of the Institutional Review Board and with the informed consent of the donors. The hMSCs were cultured in the MesenPro medium (Gibco BRL, Invitrogen, Grand Island, NY, http://www.lifetechnologies.com/) at 37°C in a humidified chamber containing 5% CO2 according to the procedure described previously .
Induction of Osteogenic Differentiation Under Normoxia and Hypoxia
Osteogenic differentiation of hMSCs was performed according to the procedure described previously . Briefly, hMSCs were seeded at a density of 5,000 cells per square centimeter, grown to more than 90% confluence, and were then induced to differentiate in an osteogenic induction medium composed of serum-free Iscove's Modified Dulbecco's Media (IMDM) (Gibco BRL, Invitrogen), 0.1 µM dexamethasone, 10 mM β-glycerol phosphate, 0.2 mM ascorbic acid, 100 units/ml penicillin, and 100 µg/ml streptomycin sulfate (Sigma-Aldrich, St. Louis, MO, http://www.sigmaaldrich.com/taiwan.html). Upon induction, cells were maintained in either normoxic (21% O2) or hypoxic (1% O2) incubators as indicated in each experiment. The hypoxic incubator (Thermo Fisher Scientific, Waltham, MA, http://www.thermofisher.com/) maintained the oxygen tension through an adjustable gauge to control the rate of injecting N2. The induction medium was refreshed twice a week until the harvest of differentiated cells.
Alkaline Phosphatase Activity Assay
Alkaline phosphatase (ALP) activity was measured by an ALP fluorimetric assay kit (BioVision, Inc., San Francisco, CA, http://www.biovision.com/) according to the manufacturer's instructions. Briefly, approximately 1 × 105 cells were homogenized in 100 µl assay buffer and centrifuged to remove insoluble cell debris. The supernatant was mixed with 20 µl 0.5 mM MUP substrate and incubated for 30 minutes at room temperature in the dark. The reaction was terminated by adding 20 µl stop solution and the fluorescent signal was measured by the Victor2 1420 Multilabel Counter (Perkin-Elmer Life Sciences, Boston, MA, http://www.perkinelmer.com/) with the excitation wavelength at 360 nm and emission wavelength at 440 nm, and normalized by the cell number.
Cells grown in a 35-mm dish were washed twice with phosphate buffered saline (PBS) and then fixed with 4% paraformaldehyde in PBS for 20 minutes. After fixation, cells were washed twice with PBS and permeabilized with 0.1% Triton X-100 (Bio-Rad Laboratories, Hercules, CA, http://www.bio-rad.com/) in PBS for 10 minutes. Intracellular ALP activity was revealed by staining with 5-bromo-4-chloro-3-indolyl phosphate/nitroblue tetrazolium (Sigma-Aldrich Chemical Co.) for 2 hours, and then washed with PBS. All procedures were carried out at room temperature.
Total RNA of cultured cells was extracted with the TRI reagent (Sigma-Aldrich), and 2 µg RNA was reverse-transcribed to cDNA with the Illustra Ready-to-Go RT-PCR kit (GE Healthcare, Buckinghamshire, U.K., http://www3.gehealthcare.co.uk/) according to the manufacturer's instructions. QPCR was performed using the LightCycler Taqman Master kit (Roche Applied Sciences, Mannheim, Germany, http://www.roche-applied-science.com), and the primers and probes (Supporting Information Table S1) used in the experiments were designed on the basis of the Universal Probe Library website (http://www.roche-applied-science.com). Gene expression levels in the cells were normalized by the expression level of 18S rRNA.
Western Blot Analysis
An aliquot of 25 µg protein was separated on a 10% SDS-PAGE gel and transferred to a piece of the PVDF membrane (Pall Corporation, Port Washington, NY, http://www.pall.com). Nonspecific bindings were blocked by 5% skim milk in Tris-buffered saline Tween 20 buffer (50 mM Tris-HCl, 150 mM NaCl, 0.1% Tween 20, pH 7.4), and the membrane was hybridized with indicated primary antibodies (Supporting Information Table S2). After incubation with a horseradish peroxidase-conjugated secondary antibody, the protein intensity was detected using an enhanced chemiluminescence reagent (Perkin-Elmer Life Sciences) and analyzed by a luminescent image analyzer Fujifilm LAS-4000 (Fujifilm Corporation, Tokyo, Japan, http://www.fujifilm.com/).
Measurement of OCR
OCR was measured on a set of the 782 Oxygen Meter (Strathkelvin Instruments, Scotland, U.K., http://www.strathkelvin.com/) at 37°C with a water circulation system. An aliquot of 5 × 105 cells was resuspended in 330 µl assay buffer (125 mM sucrose, 65 mM KCl, 2 mM MgCl2, and 20 mM Na+-K+-phosphate buffer, pH 7.2) in a closed chamber of the oxygen meter, and 0.0003% digitonin (Sigma-Aldrich) was added to permeabilize the mitochondrial outer membrane. An aliquot of 6 µl of 1 M glutamate plus 1 M malate or 6 µl of 1 M succinate alone (at a final concentration of 20 mM) was injected into the chamber as the electron donor. The efficiency of ATP synthesis and the integrity of mitochondrial respiratory chain were assessed by the response of the cells to the addition of 1 mM ADP.
Measurement of the Reserve Capacity of Mitochondrial Respiration
We used a Seahorse Extracellular Flux Analyzer (Seahorse Bioscience, North Billerica, MA, http://www.seahorsebio.com/) to measure the reserve capacity of mitochondrial respiration according to the manufacturer's instruction . An aliquot of 2 × 105 cells was seeded on custom-made plates and induced to differentiate for 3 weeks under either normoxia or hypoxia before bioenergetic function measurements. The basal level and reserved capacity of respiration were measured before and after the injection of 2 µM CCCP (carbonyl cyanide-3-chlorophenyl hydrazone, a conventional uncoupler of mitochondrial respiration).
Measurement of Lactate Release Rate
Lactate release rate was measured by a Lactate Reagent kit (Trinity Biotech Plc, Bray, Ireland, http://www.trinitybiotech.com/) according to the manufacturer's instructions. Cells in a six-well plate were replenished with the fresh medium and incubated at 37°C for 8 hours. An aliquot of 10 µl medium was transferred to a 96-well plate and reacted with 200 µl of the Lactate Reagent. The absorbance at 540 nm was measured by an ELISA reader (BioTek Instruments, Inc, Winooski, VT, http://www.biotek.com/), and the lactate release rate was calculated by dividing the amount of lactate by the cell number and the duration of incubation.
Measurement of Intracellular ATP Level
The intracellular level of ATP was measured by the Bioluminescent Somatic Cell Assay Kit (Sigma-Aldrich) according to the manufacturer's instructions. An aliquot of 50 µl suspension of viable cells was mixed with 150 µl Somatic Cell Releasing Reagent, and 100 µl mixture was then transferred to a black OptiPlate-96F 96-well plate (Perkin-Elmer Life Sciences) containing 100 µl ATP assay mixture in each well. The luminescence intensity was measured by the Victor2 1420 Multilabel Counter (Perkin-Elmer Life Sciences). The actual ATP content per cell was calculated using a standard curve and normalized by the cell number.
Measurement of Intracellular Superoxide Anions
Cells in 6-cm dishes were washed with PBS twice and incubated in a medium containing 5 µg/ml hydroethidine (Molecular Probes, Eugene, OR, http://www.lifetechnologies.com/tw/zt/home/brands/molecular-probes.html) at 37°C in the dark for 15 minutes. Cells were then trypsinized, centrifuged at 600g for 5 minutes, and resuspended in 0.4 ml of PBS. The fluorescence intensity, which reflected the intracellular superoxide anions ( ), was recorded for a total of 10,000 cells on the FL2 channel of a flow cytometer (Model EPICS XL-MCL, Beckman-CoulterBrea, CA, https://www.beckmancoulter.com/) with the excitation wavelength at 535 nm and emission wavelength at 590 nm, respectively.
Measurement of Intracellular Hydrogen Peroxide
Cells in 6-cm dishes were washed with PBS twice and incubated in a medium containing 40 µM 2′,7′-dichlorodihydrofluorescein diacetate (Molecular Probes) at 37°C in the dark for 15 minutes. Cells were then trypsinized, centrifuged at 600g for 5 minutes and resuspended in 0.4 ml of PBS. The fluorescence intensity, which reflected the intracellular hydrogen peroxide (H2O2), was recorded for a total of 10,000 cells on the FL1 channel of a flow cytometer (Model EPICS XL-MCL, Beckman-Coulter) with the excitation wavelength at 488 nm and emission wavelength at 535 nm, respectively.
Statistical analysis was performed by the Microsoft Office Excel 2007 statistical package. The data are presented as mean ± SD of the results obtained from three independent experiments. The significance level of the difference in the measured values between the hMSCs differentiated under normoxia and those under hypoxia was determined by the paired t test. A difference is considered statistically significant when p value <.05.
Hypoxia Impairs Osteogenic Differentiation of hMSCs
To study the effect of hypoxia on osteogenic differentiation of hMSCs, we cultured hMSCs under normal oxygen tension and then induced differentiation under either 1% or 21% O2 and the stabilization of HIF-1α and HIF-2α could be observed under hypoxia (Fig. 1A). After osteogenic induction, the activity of ALP, a differentiation biomarker, was significantly increased under normoxia, but the response was greatly attenuated under hypoxia (Fig. 1B). Moreover, the trend was also revealed by ALP staining (Fig. 1C). Similar results were also observed in hMSCs of another two donors (Supporting Information Fig. S1A, S1B). In addition, the mRNA levels of osteoblast-specific transcription factor core binding factor a-1 (Cbfa-1) and osteogenic markers osteopontin and osteocalcin were significantly increased after osteogenic induction under normoxia but not under hypoxia (Fig. 1D-1F). These findings indicate that low oxygen tension is unfavorable for osteogenic differentiation of hMSCs in vitro.
Repression of Mitochondrial Biogenesis and Aerobic Metabolism by Hypoxia During Osteogenic Differentiation of hMSCs
In the examination of hypoxic effects on mitochondrial function, we observed significant decreases in the expression of mitochondrial respiratory enzyme complexes including ND6 subunit of Complex I, core I subunit (UQCRC1) of Complex III, and cytochrome c oxidase subunit IV (COX4) in a time-dependent manner after differentiation of hMSCs under hypoxia (Fig. 2A). Similar decreases were found in multiple subunits of several respiratory enzyme complexes in osteoblasts after 3 weeks of differentiation (Fig. 2B) and the decrease was not due to the activation of autophagy (Supporting Information Fig. S2). In addition, the OCR driven by either glutamate plus malate or by succinate alone was lower in osteoblasts differentiated under hypoxia (H) than those differentiated under normoxia (N) (Fig. 2C; glutamate plus malate, N: 2.90 ± 0.57 vs. H: 1.86 ± 0.43 nmole O2/minute/106 cells; succinate, N: 4.58 ± 0.28 vs. H: 1.92 ± 0.81 nmole O2/minute/106 cells). The difference was even more pronounced when ADP was added to activate the mitochondrial respiration. We also examined the reserve capacity of mitochondrial respiration by challenging cells with a mitochondrial uncoupler CCCP on a Seahorse Extracellular Flux Analyzer. The results showed that osteoblasts differentiated under normoxia had a higher reserve capacity than did those differentiated under hypoxia or undifferentiated hMSCs (Fig. 2D). These findings indicate that hypoxia attenuated the differentiation-elicited activation of mitochondrial respiration and metabolic switch to aerobic metabolism that have been observed to occur under normoxia.
Enhancement of Anaerobic Glycolysis in Osteoblasts Differentiated Under Hypoxia
Hypoxia is known to activate glycolysis, we therefore examined whether glycolytic enzymes were altered during osteogenic differentiation under hypoxia. By Western blot analysis, we found that the protein expression levels of glycolytic enzymes such as hexokinase (HK), glucophosphate isomerase (GPI), glyceraldehyde 3-phosphate dehydrogenase (GAPDH), and lactate dehydrogenase (LDH) were upregulated in osteoblasts differentiated under hypoxia but this effect was attenuated under normoxia (Fig. 3A). In addition, PDK, which can phosphorylate PDH to inhibit its activity, was also dramatically upregulated throughout the differentiation process under hypoxia, whereas the expression level of PDH was significantly decreased (Fig. 3A). This inhibitory effect on the aerobic metabolism was completely reversed under normoxic conditions. We also measured the release rate of lactate, a principal indicator of glycolytic activity of cells, and found that the rate was significantly higher in the osteoblasts differentiated under hypoxia than in those differentiated under normoxia (Fig. 3B). The above findings indicate that hMSCs differentiated under hypoxia exhibit higher glycolytic activities as compared to those differentiated under normoxia. This observation was confirmed by treating hMSCs with inhibitors of either glycolysis (2-deoxyglucose, 2-DG) or mitochondrial respiration (antimycin A) to dissect the contribution of energy production from aerobic and anaerobic pathways, respectively. During osteogenic differentiation under hypoxia, continuous treatment of cells with 2.5 mM 2-DG resulted in a dramatic decrease of intracellular ATP levels, whereas treatment with 2 µg/ml antimycin A showed little effects (Fig. 3C). On the contrary, antimycin A reduced ATP level in osteoblasts differentiated under normoxia to a greater extent compared with the inhibitory effect on osteoblasts differentiated under hypoxia (Fig. 3C). Besides, a significant decrease of lactate release rate was observed only after 2-DG treatment of the osteoblasts differentiated under hypoxia but not those differentiated under normoxia (Fig. 3D). Moreover, 2-DG treatment under hypoxia and antimycin A treatment under normoxia dramatically reduced the ALP activity of differentiating osteoblasts (Fig. 3E). This suggests that disturbance of energy metabolism could impair osteogenic differentiation of hMSCs. These results indicate that osteoblasts differentiated under hypoxia tend to activate glycolysis instead of triggering the metabolic switch to aerobic metabolism that occurs under normoxia.
Inhibition of Osteogenic Differentiation by CoCl2-Induced Hypoxia
To confirm the effect of hypoxia on osteogenic differentiation of hMSCs, we used CoCl2, a HIF stabilizer, to mimic the hypoxic condition instead of using the hypoxic chamber. The protein amount of both HIF-1α and HIF-2α were accumulated upon treatment of hMSCs with 50 or 100 µM CoCl2 for 48-hour (Fig. 4A). Consequently, the expression of PDK was upregulated and PDH was downregulated. Then we treated osteoblasts differentiated under normoxia with 100 µM CoCl2. Similar to what we had observed under hypoxic condition, the 100 µM CoCl2-treated osteoblasts differentiated under normoxia showed a significant increase of the lactate release rate (Fig. 4B) and a dramatic decrease of the OCR (Fig. 4C). Furthermore, the ALP activity (Fig. 4D) and Cbfa-1 mRNA expression (Fig. 4E) were decreased to the levels similar to those of osteoblasts differentiated under hypoxia. These findings suggest that hypoxia attenuated metabolic switch and osteogenic differentiation of hMSCs through the effect of HIF-1α and/or HIF-2α. To further confirm the involvement of HIF-1α, we treated hMSCs with echinomycin, a small-molecule inhibitor of HIF-1α transcription activity , during differentiation under hypoxia. The results showed that treatment with echinomycin could partially restore the ALP activity of osteoblasts under hypoxia (Fig. 4F and Supporting Information Fig. S3), suggesting that HIF-1α is involved in the regulation of hMSCs differentiation under hypoxia.
Upregulation of Antioxidant Enzymes and Dynamic Changes of Intracellular ROS During Osteogenic Differentiation of hMSCs
In one of our previous studies, we observed a coordinated upregulation of antioxidant defense systems during osteogenic differentiation of hMSCs under normoxia to prevent the overproduction of ROS to interfere with the differentiation process . Whether this phenomenon also occurs under hypoxia, in which only mild increase of mitochondrial function was observed (Fig. 2), deserves further exploration. In this part of study, we found that the intracellular level of superoxide anion ( ) was slightly decreased in the early phase of differentiation under both normoxia and hypoxia, but the decrease was more pronounced under hypoxia (Fig. 5A). After the third day of differentiation, the levels began to increase with a higher rate in normoxia and reached approximately 150% of that of undifferentiated control on day 21. Conversely, the intracellular H2O2 levels were dramatically dropped down to approximately 60% of the control on the first day of differentiation with a similar trend under normoxia and hypoxia (Fig. 5B). The H2O2 levels continued to decrease gradually in both cases, but the osteoblasts differentiated under hypoxia had a lower H2O2 content as compared to those differentiated under normoxia. The protein expression level of manganese-dependent superoxide dismutase (MnSOD), which converts into H2O2, showed a temporal and dramatic increase at a greater extent in osteoblasts differentiated under hypoxia compared with that found under normoxia (Fig. 5C). The protein expression level of catalase, which turns H2O2 into water and oxygen, was also upregulated in a time-dependent manner during differentiation, but the osteoblasts differentiated under normoxia displayed much stronger induction than did those differentiated under hypoxia (Fig. 4C). These results indicate that the change of intracellular ROS levels of osteoblasts during differentiation shares similar trends regardless of oxygen tension, but the ROS levels in osteoblasts differentiated under hypoxia were lower than those differentiated under normoxia. Moreover, the expression of MnSOD and catalase was regulated under hypoxia in a manner different from the regulation of the two enzymes under normoxia.
Accumulating evidence supports the notion that the metabolic signatures are distinct between stem cells and their differentiated progenies . Our previous studies showed that hMSCs rely on anaerobic glycolysis while their differentiated osteoblasts depend mainly on oxidative phosphorylation for energy supply , which we termed the “metabolic switch” during stem cell differentiation. We also found that blockade of the metabolic switch by respiratory inhibitors resulted in compromised osteogenesis, which indicates that aerobic metabolism is crucial for osteogenic differentiation. Similar results were also observed in the differentiation of other types of stem cells including human ESCs , goat MSCs , and monkey ESCs . However, the application of stem cells in regenerative medicine usually encounters a hypoxic environment such as the transplantation sites with damaged blood vessels in bone fracture or three-dimensional-culture scaffold with limited oxygen perfusion. Under these conditions, the shortage of oxygen to serve as the terminal electron acceptor in the respiratory chain of mitochondria will inevitably compromise aerobic metabolism. Therefore, a better understanding of whether and how hypoxia affects the metabolic switch and the differentiation capability of hMSCs is essential for their clinical applications.
In this study, we maintained hMSCs at 21% O2 concentration and induced osteogenic differentiation under either 21% or 1% O2 to examine the effect of O2 tension on the metabolic switch and differentiation of hMSCs. A comprehensive downregulation of mitochondrial respiratory enzymes in osteoblasts differentiated under hypoxia was clearly shown (Fig. 2A, 2B). Complexes I, III, and IV were significantly downregulated whereas Complex II and V were decreased mildly, which indicate that mitochondrial respiration was suppressed under hypoxia. Interestingly, Fukuda et al.  showed that hypoxic signals could activate the expression of COX4-2 (cytochrome c oxidase subunit 4-2) and induce COX4-1 degradation in mammalian cells. The fluctuation between these two types of COX4 isoforms is critical for the optimal efficiency of mitochondrial respiration in response to the change of cellular oxygen concentration. However, we did not observe a significant change in the expression level of COX4-2 in osteoblasts differentiated under hypoxia (data not shown). This indicates that hMSCs differentiated under hypoxia were not able to tune up COX4-2 to compensate for the compromised mitochondrial respiration.
In addition to analyzing the protein amount of mitochondrial respiratory enzymes, we examined the mitochondrial OCR as a functional index. Under normoxic condition, the substrate (glutamate plus malate or succinate)-supported and ADP-stimulated OCRs in osteoblasts differentiated for 3 weeks (ost3w) were higher than those of the undifferentiated hMSCs (Fig. 2C), indicating an activation of aerobic metabolism after osteogenic induction. However, the OCR of osteoblasts differentiated under hypoxia was lower than that of their normoxic counterparts. Moreover, by the treatment of hMSCs with CCCP to induce uncoupled respiration, we found that the reserve capacity was increased in ost3w under normoxia but no such change in ost3w under hypoxia as compared to that of hMSCs (Fig. 2D). These findings together reveal that hypoxia impairs both mitochondrial integrity and respiratory function of the cells during osteogenic differentiation. The decline of mitochondrial OCR under hypoxia may be attributed to the paucity of oxygen to support the electron transport in mitochondrial inner membranes. Besides, there has been strong evidence to show that HIF-1α was able to repress mitochondrial respiration because knockdown or knockout of HIF-1α under hypoxia can restore the OCR to the level under normoxia . Therefore, downregulation of aerobic metabolism during osteogenic differentiation under hypoxia may be the result of both oxygen deprivation and HIF-1α-induced mitochondrial inactivation.
Since aerobic metabolism was attenuated in osteoblasts differentiated under hypoxia, we further investigated whether anaerobic glycolysis was altered to cope with the energy demand of differentiating cells. As expected, the protein expression levels of glycolytic enzymes including HK, GPI, GAPDH, and LDH were highly induced under hypoxia (Fig. 3A), indicating a profound activation of glycolysis. In addition, PDK, the critical regulator in the metabolic switch, was also significantly upregulated under hypoxia. By contrast, the protein expression level of PDH was dramatically decreased in hMSCs differentiating under hypoxia. Consequently, the lactate release rate, an indicator of glycolytic activity, of the osteoblasts differentiated under hypoxia was higher than that of the osteoblasts differentiated under normoxia (Fig. 3B). The above findings that mitochondrial respiration was attenuated and glycolysis was enhanced under hypoxia may be due to the fact that pyruvate, the product of glycolysis, could not be converted to acetyl-CoA by PDH to enter the TCA cycle because PDH was downregulated and inactivated by PDK. Instead, pyruvate was catalyzed to lactate and regenerate NAD+ to support the continuous execution of glycolysis under hypoxia.
To confirm the metabolic switch, we treated hMSCs with inhibitors of glycolysis or mitochondrial respiration and evaluated the changes in metabolic activities. When osteoblasts differentiated under hypoxia were treated with 2-DG, their ATP content significantly dropped to approximately 20% of the control, but antimycin A (a mitochondrial Complex III inhibitor) failed to affect the ATP level (Fig. 3C). This finding indicates that osteoblasts differentiated under hypoxia rely mainly on glycolysis for supply of energy. Similarly, lactate production of osteoblasts under hypoxia was dramatically reduced to approximately 15% by 2-DG but not influenced by antimycin A (Fig. 3D), which is consistent with the above finding. Moreover, the inhibition of glycolytic activity by 2-DG could further repress the ability of hMSCs to undergo osteogenic differentiation under hypoxia (Fig. 3E). These observations suggest that the metabolic switch to mitochondrial respiration during osteogenic differentiation of hMSCs was strongly compromised by hypoxia.
A number of studies focused on the effect of hypoxia on the differentiation of hMSCs have shown that hypoxia could affect osteogenesis, chondrogenesis, and adipogenesis, and this may be related to the oxygen requirement of the differentiated cells. For example, chondrocytes in avascular cartilage have low oxygen requirement so that hypoxia is beneficial to chondrogenesis. Conversely, osteoblasts in vascular bone require high oxygen levels for optimal differentiation so that hypoxia is thought to inhibit osteogenesis. However, there are still controversies about this important issue. In the case of osteogenesis, hypoxia was shown to inhibit the in vitro osteogenic differentiation potential of MIAMI cells  and downregulate osteoblastic genes in hMSCs in vitro , which are in agreement with our findings. Moreover, a recent study reported that hypoxia can inhibit osteogenesis of hMSCs through repression of Cbfa-1 transcription via activation of Twist (a HIF-1α downstream effector) . This has provided a novel molecular mechanism to explain the inhibitory effect of hypoxia. On the contrary, it has also been reported that hypoxia can enhance bone-forming potentials of MSCs in vitro and in vivo [26-28] and knockout of HIF-1α or HIF-2α resulted in decreased trabecular bone volume. Interestingly, Hung et al.  demonstrated that hMSCs cultivated under hypoxia for a short period of time could increase the expression levels of stemness genes, proliferation rate, as well as osteogenic differentiation potential, but reduced the adipogenic and chondrogenic differentiation abilities. These findings indicate that preconditioning of hMSCs under hypoxic environment can dramatically alter the physiological properties and differentiation outcome of stem cells. Therefore, the contradictory observations may result from differences in the cell types (primary cultures or cell lines), donor species (rat, mouse or human), culture conditions (maintained in normoxia or hypoxia), and oxygen concentration (from 0.02% to 5% oxygen) during differentiation of stem cells. Nevertheless, in this study, we demonstrated that hypoxia attenuated mitochondrial aerobic metabolism and suppress osteogenic differentiation of hMSCs in vitro. By the use of CoCl2 as a hypoxia mimetic and echinomycin (an inhibitor of the HIF-1α transcription activity) to treat hMSCs, we confirmed that stabilization of HIF-1α and/or HIF-2α is involved in the disturbance of metabolic switch and osteogenic differentiation of hMSCs under hypoxia (Fig. 4).
Alterations of cellular redox status as well as antioxidant enzymes have been documented in the differentiation of different types of stem cells [4, 9, 30] and in the reprogramming of somatic cells [1, 7, 8], which implies that ROS might play important roles in stem cell differentiation. In a previous study, we found that the intracellular ROS level was significantly decreased upon osteogenic induction and gradually increased as differentiation went on . We investigated this phenomenon under hypoxia and found that the ROS (both and H2O2) levels were decreased, and were maintained at a lower level under hypoxia compared with those under normoxia (Fig. 5A, 5B). This difference may come from the inactive mitochondria, which produce less ROS through mitochondrial respiration under hypoxia. Besides, an alteration of the expression of antioxidant enzymes may be also involved. Therefore, we examined the expression levels of several antioxidant enzymes and found that MnSOD and catalase were dramatically upregulated in osteoblasts differentiated under normoxia and hypoxia, respectively (Fig. 5C). Surprisingly, the amplitude of increase in MnSOD was significantly larger in hypoxia than in normoxia, but the expression of catalase showed the opposite trend. Taken together, the antioxidant defense enzymes were still upregulated during osteogenic differentiation of hMSCs under hypoxia even though the aerobic metabolism was not significantly activated. This finding suggests that the highly responsive antioxidant defense system and suppression of intracellular ROS are important for the osteogenic differentiation process under both normoxia and hypoxia. The mechanism that governs the dynamic expression of MnSOD and catalase in the differentiation of stem cells warrants further investigation.
In conclusion, our findings of this study revealed that osteoblasts differentiated under hypoxia fail to undergo metabolic switch and therefore rely on glycolysis rather than mitochondrial respiration for supply of most of the energy required for differentiation, which contributes to compromised osteogenesis. This is the first report to address the hypoxic effects on hMSCs differentiation from the perspectives of energy metabolism. Therefore, manipulation of metabolic properties of stem cells may be useful for the development of appropriate cell culture conditions or in vivo transplantation protocols for the preservation of the differentiation potentials of hMSCs.
This work was supported by grants from the National Science Council of Taiwan (NSC99–2811-B-010–054, NSC100–2320-B-010–024-MY3, and NSC 100–2811-B-010–023). The authors would like to express their appreciation to Dr. Oscar K. Lee for supporting the work on mesenchymal stem cells in the past few years.
Disclosure of Potential Conflicts of Interest
The authors declare no potential conflicts of interest.