Nicotinamide phosphoribosyltransferase (Nampt) affects the lineage fate determination of mesenchymal stem cells: A possible cause for reduced osteogenesis and increased adipogenesis in older individuals
Human aging is associated with a progressive decline in bone mass and an accumulation of marrow fat.1, 2 The marrow adipocytes and osteoblasts share common progenitors, mesenchymal stem cells (MSCs), which are capable of multilineage differentiation by selected activation of certain lineage-specific transcription factors.3 In osteogenesis, core binding factor 1 (Cbfa1/Runx2) is highly expressed. It binds in the promoters and stimulates the expression of other osteoblast-specific genes such as collagen α1,1 osteocalcin, bone sialoprotein, and osteopontin, which are associated with bone matrix production.4, 5 In adipogenesis, the expression of peroxisome proliferators–activated receptor γ (PPARγ) is upregulated. It binds to the promoters and enhances the expression of genes involved in lipid storage, such as the fatty acid binding protein aP2 and the cholesterol and fatty acid transporter FAFP/CD36.6 The lineage fate determination between osteoblasts and adipocytes has been shown to be mutually exclusive; for example, PPARγ insufficiency enhanced osteogenesis,7 whereas deficiency of Runx2 transdifferentiated preosteoblasts to adipocytes.8 In addition, PPARγ-deficient mice developed osteopetrosis spontaneously.9 Therefore, an unbalanced MSC differentiation scheme in aging could result in accelerated marrow adipogenesis. In support of this hypothesis, bone marrow MSCs derived from aged mice have been shown to develop fewer osteoblastic but more adipocytic colonies than MSCs from adult control mice.10 However, how aging makes the preferential shift from osteogenesis to adipogenesis is largely unknown.
Sirt1 is the mamamalian homologue of sirtuins, a group of nicotinamide adenine dinucleotide (NAD+)–dependent protein deacetylases that can regulate lifespan in lower organisms and affect age-related diseases in mammals.11 These enzymes act by transferring acetyl groups from their substrate proteins, including histones and transcriptional factors, to the ADP-ribose moiety of NAD+.12 In each reaction cycle, one molecule of NAD+ is consumed and cleaved into nicotinamide (NAM) and O-acetyl-ADP-ribose.13 NAM, functioning as a potent endogenous inhibitor,14–16 in turn noncompetitively inhibits the sirt1-catalyzed deacetylation.17, 18 It is noteworthy that the dual effects on NAD+ deprivation and NAM accumulation occur with several other enzymatic reactions in eukaryotic cells, such as those catalyzed by the poly(ADP-ribose) polymerases (PARPs), which regulate cell death programs,19 and mono(ADP-ribose) transferase (MART)20 and CD38/CD157,21 whose physiologic functions remain unclear. To maintain the NAD homeostasis and the constitutive activity of NAD+-consuming enzymes, a salvaging pathway for NAD+ resynthesis from NAM evolved in eukaryotic cells. The biochemical route of this so-called NAD+ salvaging is composed of two enzymatic steps, the first operated by nicotinamide phosphoribosyltransferase (Nampt), forming nicotinamide mononucleotide (NMN) from NAM and phosphoribosyl pyrophosphate, and the second driven by nicotinamide mononucleotide adenylyltransferases (Namnt), leading to NAD+ formation from NMN and ATP.22 These processes are currently believed to be as important as NAD+ neosynthesis from tryptophan or nicotinic acid for maintaining the cell energetic homeostasis.23
We have shown previously that Sirt1 activity affected the lineage fate determination between osteogenesis and adipogenesis of MSCs.24 Since Sirt1 activity is largely dependent on the availability of NAD+ and influenced by the intracellular concentration of NAM, we supposed that MSC differentiation also could be affected by enzymes involved in the NAD+ salvaging pathway. Here we report that the expression and activity of Nampt, the enzyme catalyzing NMN formation from NAM and phosphoribosyl pyrophosphate, affected the lineage fate determination of MSCs.
Materials and Methods
Modified Eagle's medium alpha (α-MEM), fetal bovine serum (FBS), L-glutamine, and gentamicin were purchased from Gibco (Invitrogen, Carlsbad, CA, USA). Rabbit anti-Nampt antibody was from BETHYL (Montgomery, TX, USA), rabbit anti-Sirt1 antibody was from Upstate (Stockholm, Sweden), and rabbit anti-Gapdh was purchased from Abcam (Cambridge, UK). KF866 and 1,5-isoquinolinediol were purchased from Sigma (St Louis, MO, USA).
The murine MSC line C3H10T1/2 was obtained from the American Type Culture Collection (ATCC, Boras, Sweden). The cells were cultured in α-MEM supplemented with 10% FBS, 2 mM L-glutamine, and 50 µg/mL of gentamicin at 37 °C in a humidified 5% CO2 atmosphere. Mouse bone marrow stromal cells were obtained from 4-week-old and 15-month-old male C57BL/6 mice. Briefly, mice were euthanized using 4% isofluorane in CO2, and the bones were excised aseptically from the hind limbs. External soft tissue was discarded, and the bones were place in α-MEM supplemented with 400 µg/mL of gentamicin. Both ends of the femur and tibia were clipped. An 18-gauge needle was inserted into the diaphysis at one end, and bone marrow was flushed out from the other end to a 50-mL Falcon tube by culture medium. After centrifugation at 1000 rpm for 5 minutes, the cell pellet was collected and diluted in 15 mL of culture medium and cultured in a 75-cm flask. Nonadherent cells were removed after 24 hours, and the remaining cells were passaged after reaching 80% confluence. For osteoblast differentiation, cells were cultured in α-MEM supplemented with 10% FBS, 10 mM β-glycerophosphate, 50 µg/mL of α-ascorbic acid, and 0.1 µM dexamethasone (osteoblast medium [OBM]) for 2 weeks, with medium changes twice per week.
Gene transfection of murine mesenchymal stem cell line C3H10T1/2
MISSION Nampt shRNA and non–target scramble shRNA lentiviral transduction particles were obtained from Sigma. Cells were plated at 5 × 104/well in 24-well plates and incubated at 37 °C for 18 hours. Then cells were exposed to lentiviral particles in the presence of 8 µg/mL of hexadimethrine bromide (Sigma) for 20 hours. Following transduction, cells were selected with 800 ng/mL of puromycin.
Western blot analysis
Cells were lysed in M-PER mammalian protein extraction reagent supplemented with Halt protease inhibitor cocktail (Thermoscientific, Stockholm, Sweden). Protein concentrations were measured with BioRad protein assay (BioRad, Hercules, CA, USA) using bovine serum albumin (BSA) as a standard. Then 30 µg of protein was mixed with Laemmli buffer and boiled at 97 °C for 10 minutes before loading onto SDS-PAGE gels. Proteins were transferred to polyvinylidene fluoride (PVDF) membranes by electroblotting, after which the membrane was blocked with 5% milk. The membranes were probed with anti-Nampt, anti-Sirt1, and anti-Gapdh antibodies diluted in blocking solution and then were horseradish peroxidase–conjugated with anti-rabbit IgG secondary antibody. The blots were visualized using the Immun-Star HRP Chemiluminescent Kit (BioRad) and exposed to X-ray films.
Measurement of Sirt1 deacetylase activity
Cells were lysed in M-PER mammalian protein extraction reagent. Protein concentrations were measured with a BioRad protein assay (BioRad) as described previously. Sirt1 deacetylase activity was measured by fluorometric SIRT1 Assay Kit (CS1040; Sigma) following the manufacturer's instructions. The assays were performed by incubating 20 µL of protein extract with SIRT1 substrate solution at 37 °C for 30 minutes. After addition of the developing solution and incubation at 37 °C for 10 minutes, fluorescent intensity was measured at 460 nm (excitation 355 nm) using a fluorescence plate reader and normalized by protein content. A standard curve was performed with the standard solution included in the kit.
Measurement of NAD+ amount
Intracellular NAD+ was measured with an NAD+/NADH quantification kit (K337-100; BioVision, San Francisco, CA, USA) following the manufacturer's instructions. Briefly, NAD+/NADH were extracted from cell samples with 500 µL of extraction buffer and filtered through a 10-kDa Micon filter (Biovision, Stockholm, Sweden). The protein concentration was measured with a BioRad protein assay (BioRad) as described previously. For each sample, 200 µL of the extract was heated to 60 °C for 30 minutes to decompose NAD+ but keeping NADH intact. During measurement, both heated and unheated extracts from each sample, together with the NADH standard solutions, were transferred into 96-well plates at 50 µL/well in triplicate. Then 100 µL of NAD cycling mix was added to each well and incubated at room temperature for 5 minutes to convert NAD+ to NADH. Then 10 µL of NADH developer was added, mixed, and incubated at room temperature for 2 hours. The optical density was read at OD 450 nm by a kinetic ELISA reader (Spectra MAX 250, Molecular Devices, Sunnyvale, CA, USA). The amount of NAD+ from each sample was calculated as total NAD (values from the unheated extracts) minus NADH (values from the heated extracts) and then divided by the protein concentration.
Adipocyte staining and quantification
Cells were washed twice with PBS and fixed with 10% formaldehyde for 45 minutes at room temperature. After washing with distilled water twice and 60% isopropanol once, the cells were stained with filtered oil red O–60% isopropanol for 20 minutes at room temperature. Then the cells were washed twice with distilled water and photographed by an inverted fluorescence microscope system (Zeiss Axiovert S100, Spectral Solutions, Stockholm, Sweden). To quantify retention of oil red O, stained adipocytes were extracted with 4% Igepal CA630 (Sigma-Aldrich, St Louis, MO, USA) in isopropanol for 15 minutes, and absorbance was measured by spectrophotometry at 520 nm. For flowcytometric adipocyte quantification, trypsinized cells were washed with PBS once and fixed in 10% formaldehyde for 30 minutes at room temperature. The cells were stained with 10 µg/mL of Nile red for 20 minutes at room temperature. Fluorescent emission between 564 and 604 nm with a bandpass filter was measured using a FACSCalibur flow cytometer (BD Biosciences, San Jose, CA, USA).
Alkaline phosphatase staining and quantification
A TRACP & ALP Double-Stain Kit (Karara Bio, Inc., Otsu, Japan) was used for staining of alkaline phosphatase (ALP) in the cell cultures. Cells were plated at 3 × 104/well in 24-well plates and grown for 4 days in bone-inducing medium. Substrate for ALP was dissolved in 10 mL of distilled water before use. After washing twice with PBS, cells were fixed in fixation buffer for 5 minutes. Then cells were washed three times with distilled water, and 250 µL of ALP substrate was added to each well. After incubation at 37 °C for 30 minutes and then washing again with distilled water, cells were observed and photographed by an inverted fluorescence microscope system (Zeiss Axiovert S100, Spectral Solutions). Phosphatase Substrate Kit (Pierce, Rockford, IL, USA) containing p-nitrophenyl phosphate disodium salt (PNPP) was used to quantify the ALP activity in cell cultures. PNPP solution was prepared by dissolving two PNPP tablets in 8 mL of distilled water and 2 mL of diethanolamine substrate buffer. Cells were plated at 20,000/well in 6-well plates. After culturing in OBM for 4 days, the cells were lysed in 500 µL of M-PER mammalian protein extraction reagent without protease inhibitors. Protein concentrations were measured with BioRad protein assay (BioRad) using BSA as a standard. Then 20 µL cell lysate was incubated with 100 µL of PNPP solution in 96-well plate at room temperature for 30 minutes. Then 50 µL of 2N NaOH was added to each well to stop the reaction. The blank control was made of 20 µL of M-PER reagent and 100 µL of PNPP solution. The absorbance was measured at 405 nm in a kinetics ELISA reader (Spectra MAX 250, Molecular Devices). The results were normalized with protein concentration of the cell lysates.
RNA preparation and quantitative real-time PCR analysis
Total RNA was extracted with a RNeasy Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's instruction. cDNA synthesis was performed with 500 ng of RNA using a High Capacity cDNA Reverse Transcription Kit (Applied BioSystems, Foster City, CA, USA). Quantitative real-time PCR was performed with Maxima Probe/ROX qPCR Master Mix (Fermentas, Helsingborg, Sweden) and monitored with a LightCycler 480 (Roche Diagnostics, Mannheim, Germany). The gene primers and probes were used for mouse PPARγ (Mm00440945_m1; Applied BioSystems), Runx2 (Mm00501578_m1; Applied BioSystems), osteocalcin (Mm00649782_gH; Applied BioSystems), OPG (Mm00435452_m1; Applied BioSystems), RANKL (Mm00441908_m1; Applied BioSystems), and Gapdh (Mm99999915_g1; Applied BioSystems). The reaction protocol included preincubation at 95 °C for 10 minutes, amplification of 40 cycles that was set for 15 seconds at 95 °C, annealing for 30 seconds at 60 °C, and extending at 72 °C for 30 seconds. Levels of gene expression were shown relative to the internal standard (mouse Gapdh).
Staining for mineralization
Mouse primary bone marrow stromal cells were cultured in 24-well plates for 2 weeks before analysis of mineralization. Both alizarin red and von Kossa staining were used. Briefly, for von Kossa, the cells were washed with PBS and fixed with 4% paraformaldehyde. The cells were rinsed with distilled water and thereafter incubated under ultraviolet (UV) light in the presence of 5% silver nitrate. After 30 minutes, the cells were washed with distilled water and incubated with 5% sodium thiosulfate. For alizarin red staining, the cells instead were incubated with 2% alizarin red (pH 4.2) for 10 minutes and subsequently washed with distilled water.
Measurement of nicotinamide (NAM) by high-performance liquid chromatography (HPLC)
Authentic NAM was obtained from Sigma and produced a single peak on a Waters laboratory HPLC system (Waters Corporation, Milford, MA, USA) equipped with a 717+ autosampler, 600 pump, 486 tunable UV detector (λ = 260 nm), and Empower software. Separation of samples was achieved isocratically at ambient temperatures and 1 mL/min on a Chrompack Microsper C18 column (Middelburg, the Netherlands) (100 × 4.6 mm, 3 µm particle size) in 40 mM phosphate buffer (pH 7.0). The amounts of NAM in each sample were calculated from calibration curves of NAM standards run simultaneously with every set of unknown samples. Results were normalized for sample volume and cell protein amount.
The increased adipocyte formation in bone marrow stromal cells from aged mice was associated with a relatively lower intracellular NAD+ concentration and a lower Sirt1 activity
Senile osteoporosis is signaled by a progressive decline in bone mass and an accumulation of marrow fat. Consistent with the previous report of Moerman and colleagues,10 we found that after 2 weeks of culture in OBM, the bone marrow stromal cells derived from 15-month-old mice developed fewer bone nodules (as shown by von Kossa and alizarin red staining) and more adipocytes (as shown by oil red O staining) than cells derived from 4-week-old mice (Fig. 1A). In addition, the cells derived from aged mice had significantly lower intracellular NAD+ concentration (Fig. 1B) and lower Sirt1 activity (Fig. 1C) than cells derived from young mice. However, the expression of Sirt1 protein was not reduced by aging (Fig. 1D).
The Sirt1 inhibitor EX-527 increased adipocyte formation in C3H10T1/2 cells growing in OBM
We showed previously that Sirt1 regulators resveratrol and NAM affected the formation of adipocytes when C3H10T1/2 cells were cultured in OBM.24 Since the specificity of resveratrol and NAM in regulating Sirt1 activity is still questionable,25 we now also studied the influence of the newly identified Sirt1 regulator EX-527 on MSC differentiation. EX-527 is currently considered to be the most specific and potent Sirt1 inhibitor.26 EX-527 dose-dependently increased the adipocyte number, which was confirmed by oil red O staining (Fig. 2A), followed by retention quantification (Fig. 2B), as well as a further quantification by Nile red staining and flow cytometry (FACS) analysis (Fig. 2C).
Nampt deficiency enhanced adipogenesis at the expense of osteogenesis in MSCs
In order to show whether Nampt activity also could influence the lineage fate determination of MSCs, we first tested the effects of a potent and specific Nampt inhibitor, FK866.27 We found that at concentrations higher than 10 nM, FK866 exerted an apparent cytotoxic effect on MSCs (data not shown), whereas at an apparently nontoxic concentration (1 nM), FK866 significantly increased adipocyte formation and reduced ALP activity in primary cultured bone marrow stromal cells (Fig. 3A, B). Consequently, bone nodule formation, as demonstrated by von Kossa and alizarin red staining, apparently was reduced by treatment with FK866 (Fig. 3C). To further confirm the effect of Nampt deficiency on MSC differentiation, we produced Nampt-deficient C3H10T1/2 cells by transfecting the cells with Nampt shRNA lentiviral transduction particles. As indicated in Fig. 4A, Nampt protein expression was reduced successfully compared with cells tranfected with non–target scramble shRNA lentiviral particles. Consistent with the effect of FK866, adipocyte formation was increased significantly in Nampt-deficient cells (Fig. 4B). In addition, the activity of the nonspecific osteoblast marker ALP was reduced, as demonstrated by ALP staining and PNPP semiquantification (Fig. 4C). These effects were further supported by quantitative PCR analysis, which showed that expression of the adipocyte-specific transcription factor PPARγ was significantly higher in Nampt-deficient cells, whereas the osteoblast key transcription factor Runx2, as well as the osteoblast marker genes, osteocalcin and OPG, were significantly downregulated (Fig. 4D).
Nampt insufficiency inhibited Sirt1 activity in C3H10T1/2 cells by increasing intracellular NAM level
To know whether the increased adipogenesis with Nampt insufficiency was mediated by Sirt1, we tested the expression of Sirt1 protein. As shown in Fig. 5A, no significant change in Sirt1 expression was found between cells transfected with Nampt shRNA and scramble shRNA lentiviral particles. However, when Sirt1 activity was studied, a significant reduction was found in Nampt-deficient cells (Fig. 5B). In order to know by which mechanism Nampt deficiency affected Sirt1 activity, we analyzed the intracellular concentrations of NAD+ and NAM. As shown in Fig. 5C, Nampt-deficient cells had significantly lower NAD+ levels. In addition, HPLC analysis demonstrated that the intracellular concentration of NAM was higher in Nampt-deficient cells (Fig. 5D). We next tested the effects of different exogenous intermediates in the NAD+ salvaging pathway on the differentiation of MSCs. As shown in Fig. 6, adipocyte formation was not significantly affected by the addition of 100 µM NAD+ or 100 µM NMN, but the same concentration of NAM apparently increased the adipocyte formation in both C3H10T1/2 cells and primary cultured bone marrow stromal cells.
NAD+, as a coenzyme in redox reactions, plays essential roles in cell metabolism and is indispensible for the energy release from nutrients.28 Low intracellular NAD+ results in mitochondrial dysfunction, production of reactive oxygen species (ROS), and protein damage, which are associated with cell senescence and toxicity.29 NAD+ also serves as a substrate in protein modification processes, such as adenosine 5'-dephosphate (ADP)–ribosylation reactions catalyzed by poly(ADP-ribose) polymerases (PARPs) and Sirt1-mediated histone/protein deacetylation. The former is involved in cell death programs,19 and the latter is known to be a regulator in the aging process.11 In this study, we found that inhibition of Nampt, the key enzyme in the NAD+ salvaging pathway, enhanced adipogenesis at the expense of osteogenesis in MSC cultures. Intracellular Nampt expression has been shown to decline with aging.30 Our findings therefore indicate that this enzyme might play a role in the pathogenesis of senile osteoporosis.
We showed here that Sirt1 activity in MSCs is lower in aged mice. The strong connection between aging and sirtuins was revealed by a series of investigations on the mysteries of the relationship between lifespan extension and calorie restriction (CR).31–35 The common findings are that the enzymatic activity of sir2 in lower organisms is enhanced during CR and that lifespan is increased with overexpression of sir2.36 While it is not known whether the mammalian Sir2 orthologue Sirt1 similarly regulates aging in mammals, Sirt1 levels have been shown to be increased in tissues such as muscle, brain, liver, and white fat in rodents in response to CR.37–40 In addition, the transcription and expression levels of Sirt1 were shown to decline in senescent cells and in tissues such as lung, fat, and heart from senescent mice.41, 42 These facts, together with our previous finding that Sirt1 enhanced osteogenesis and inhibited adipogenesis,24 led us to investigate the correlation between Sirt1 activity and senile osteoporosis. In this study, we confirmed that the adiopogenesis potential of marrow MSCs increased with aging. However, although Sirt1 activity is significantly lower in MSCs of aged animals, we did not find any decrease in Sirt1 protein expression (Fig. 1), indicating that the enzymatic activity is more important than expression for Sirt1 functions. This notion may help to explain the previous findings that Sirt1 transgenic mice did not have extended lifespans43 and that tissue-specific Sirt1 overexpression failed to provide resistance to aging or metabolic disorders in experimental animals.44–47
Our hypothesis that the downregulation of Sirt1 activity in MSCs in aging is mediated by Nampt is based on the recent report of Koltai and colleagues,30 who showed that the age-related downregulation of intracellular NAD+ was correlated with a decline in Nampt expression. In vivo, Nampt catalyzes the conversion of NAM to NMN, which has been supposed to be the rate-limiting step for NAD+ biosynthesis. As expected, we found that after knocking down Nampt, the intracellular NAD+ concentration decreased and intracellular NAM increased in C3H10T1/2 cells. This finding is also consistent with the recent report of Pittelli and colleagues, who showed that intracellular NAD+ concentration decreased when the specific Nampt inhibitor FK866 was added to HeLa cells.48
Known as NAD+-dependent protein deacetylase, Sirt1 activity is supposed to be regulated by intracellular NAD+ concentration. However, we found that neither exogenous NAD+ nor its direct precursor, NMN, could affect the adipogenesis of C3H10T1/2 cells. Instead, the same concentration of NAM apparently increased the adipocyte number. Therefore, it is likely that NAM plays a major role in regulation of intracellular Sirt1 activity. The mechanism of NAM-mediated Sirt1 inhibition has been clarified by Sauve and colleagues14 and Jackson and colleagues.15 Briefly, the enzymatic activity of sir2 can be considered as a combination of two reactions—base exchange and deacetylation. These two reactions are competitive processes emerging from a common intermediate, ADP-ribosyl-enzyme-acetyl peptide. Unlike deacetylation, the base-exchange process is reversible, and for this reason, NAM also can be considered as a base-exchange substrate. Therefore, an increase in NAM concentration will favor the base-exchange process, regenerating NAD+ from the ADP-ribosyl-enzyme-acetyl peptide intermediate, and sacrifice the other process, deacetylation. The physiologic importance of NAM in regulating Sirt1 activity in mammals may have been underestimated because most of the original results came from studies in lower organisms. However, the inhibitory effect of NAM on Sirt1 seems to be phylum-dependent. In bacteria and yeast, approximately 79% and 35% of the maximal sir2 activity remained following millimolar NAM treatment.14 In contrast, the IC50 of NAM on human Sirt1 is shown to be less than 50 µM,17 and over 95% inhibition of Sirt1 occurred at the presence of 160 µM NAM in mouse cells.14 Since levels of NAM in mammalian tissues are supposed to lie in the range of 10 to 400 µM,49–52 it is highly likely that under normal physiologic conditions, the intracellular Sirt1 activity is regulated mainly by the concentration of NAM.
With the renewed interest in NAD biology, pharmaceuticals interfering with NAD homeostasis have undergone a renaissance.53 For example, niacin (nicotinic acid) is regarded as the most effective agent for increasing high-density lipoprotein-cholesterol. Its potential application in reducing the cardiovascular risks in patients with metabolic syndrome has led to enormous investments in developing niacin analogues without the vasoactive side effects.54 NAM, which lacks the side effects of niacin,55 has been used in dermatology in the treatment for autoimmune blistering disorders for many years.56 Recent studies indicate that by increasing skin NAD+ levels, NAM inhibits photocarcinogenesis and therefore could be used as an inexpensive and readily available compound in sunscreens.57 Furthermore, FK866-dependent intracellular NAD+ depletion has been shown to effectively trigger cell death in tumors.58 This compound is now under scrutiny in phase 2 trials for the treatment of solid and hematologic malignancies.58, 59 Our finding that NAM and FK866 enhanced adipogenesis at the expense of osteogenesis in MSC cultures suggests that the side effects on bone should be considered when evaluating their safety for long-term use.
In conclusion, this study showed that the lineage fate determination of MSCs could be affected by the activity of Nampt, the enzyme catalyzing NAD+ resynthesis from NAM. The functional mechanism may lie in the deprivation of NAM by Nampt, which subsequently regulates Sirt1 activity. This finding indicates that the MSC differentiation could be influenced by cell energy metabolism. Further studies therefore are needed to investigate whether this mechanism is involved in the development of senile osteoporosis. Our findings also indicate that side effects on bone should be considered when evaluating the long-term safety of the NAD-interfering pharmaceuticals.
All the authors state that they have no conflicts of interest.
YL and XH contributed equally to this work.
We are extremely grateful to Dr Yunjian Xu for the technical instructions with some of the cell biologic experiments. We thank Åsa-Lena Dackland and Monica Eriksson for the technical support with the FACS scan and HPLC equipments, respectively. This work was supported in part by the Ulla and Gustaf af Ugglas Foundation and the Loo and Hans Ostermans Foundation for geriatric research.
Authors' roles: YL designed the study, conducted some of the experiments, and drafted the manuscript. XH conducted the experiments and collected and analyzed the data. JH participated in the experiments with primary cultured cells. BA designed and participated in the HPLC analysis. YL and GA participated in some of the research discussion. UL participated in the research discussion, data review, and manuscript modification.