Metabolic Reprogramming of Macrophages

Background: GLUT1 is the main glucose transporter in certain immune cells. Results: Overexpressing GLUT1 in macrophages results in increased glucose uptake and glucose utilization. Conclusion: Driving glucose uptake and metabolism through GLUT1 induces a proinflammatory response that is dependent upon glycolysis and reactive oxygen species. Significance: Understanding how macrophage substrate metabolism impacts inflammation is crucial to develop novel therapeutics for obesity and diabetes. Glucose is a critical component in the proinflammatory response of macrophages (MΦs). However, the contribution of glucose transporters (GLUTs) and the mechanisms regulating subsequent glucose metabolism in the inflammatory response are not well understood. Because MΦs contribute to obesity-induced inflammation, it is important to understand how substrate metabolism may alter inflammatory function. We report that GLUT1 (SLC2A1) is the primary rate-limiting glucose transporter on proinflammatory-polarized MΦs. Furthermore, in high fat diet-fed rodents, MΦs in crown-like structures and inflammatory loci in adipose and liver, respectively, stain positively for GLUT1. We hypothesized that metabolic reprogramming via increased glucose availability could modulate the MΦ inflammatory response. To increase glucose uptake, we stably overexpressed the GLUT1 transporter in RAW264.7 MΦs (GLUT1-OE MΦs). Cellular bioenergetics analysis, metabolomics, and radiotracer studies demonstrated that GLUT1 overexpression resulted in elevated glucose uptake and metabolism, increased pentose phosphate pathway intermediates, with a complimentary reduction in cellular oxygen consumption rates. Gene expression and proteome profiling analysis revealed that GLUT1-OE MΦs demonstrated a hyperinflammatory state characterized by elevated secretion of inflammatory mediators and that this effect could be blunted by pharmacologic inhibition of glycolysis. Finally, reactive oxygen species production and evidence of oxidative stress were significantly enhanced in GLUT1-OE MΦs; antioxidant treatment blunted the expression of inflammatory mediators such as PAI-1 (plasminogen activator inhibitor 1), suggesting that glucose-mediated oxidative stress was driving the proinflammatory response. Our results indicate that increased utilization of glucose induced a ROS-driven proinflammatory phenotype in MΦs, which may play an integral role in the promotion of obesity-associated insulin resistance.

Glucose is a critical component in the proinflammatory response of macrophages (M⌽s). However, the contribution of glucose transporters (GLUTs) and the mechanisms regulating subsequent glucose metabolism in the inflammatory response are not well understood. Because M⌽s contribute to obesityinduced inflammation, it is important to understand how substrate metabolism may alter inflammatory function. We report that GLUT1 (SLC2A1) is the primary rate-limiting glucose transporter on proinflammatory-polarized M⌽s. Furthermore, in high fat diet-fed rodents, M⌽s in crown-like structures and inflammatory loci in adipose and liver, respectively, stain positively for GLUT1. We hypothesized that metabolic reprogramming via increased glucose availability could modulate the M⌽ inflammatory response. To increase glucose uptake, we stably overexpressed the GLUT1 transporter in RAW264.7 M⌽s (GLUT1-OE M⌽s). Cellular bioenergetics analysis, metabolomics, and radiotracer studies demonstrated that GLUT1 overexpression resulted in elevated glucose uptake and metabolism, increased pentose phosphate pathway intermediates, with a complimentary reduction in cellular oxygen consumption rates. Gene expression and proteome profiling analysis revealed that GLUT1-OE M⌽s demonstrated a hyperinflammatory state characterized by elevated secretion of inflammatory mediators and that this effect could be blunted by pharmacologic inhibition of glycolysis. Finally, reactive oxygen species production and evidence of oxidative stress were significantly enhanced in GLUT1-OE M⌽s; antioxidant treatment blunted the expression of inflammatory mediators such as PAI-1 (plasminogen activator inhibitor 1), suggesting that glucose-mediated oxidative stress was driving the proinflammatory response. Our results indicate that increased utilization of glucose induced a ROSdriven proinflammatory phenotype in M⌽s, which may play an integral role in the promotion of obesity-associated insulin resistance.
Immune cells infiltrate adipose and other tissues at the onset of weight gain; these cells directly contribute to continued weight gain, adipose inflammation, and systemic insulin resistance associated with obesity (1)(2)(3). In this microenvironment, it is likely that particular substrate availability and utilization can alter M⌽ 2 phenotype. Indeed, fatty acids, lipid metabolism, and fatty acid-binding proteins play an important role in regulating M⌽-mediated inflammation in obesity and diabetes (4 -7). However, little work has been completed to date on the metabolic fate of other substrates, such as glucose, in metabolism-dependent reprogramming of M⌽ phenotypes.
A range of M⌽ phenotypes, or subtypes, exist in which "M1," or classically activated, proinflammatory M⌽s, occupy one extreme of the polarization spectrum, whereas "M2," or alternatively activated anti-inflammatory M⌽s, exist at the opposite end (2). M1 M⌽s are activated through the action of classical type 1 T-helper (Th1) inflammatory cytokines interferon-␥ (IFN␥) and tumor necrosis factor ␣ (TNF␣); these cytokines are produced in response to agents such as saturated fatty acids, cytokines, and Gram-negative bacterial endotoxin (lipopolysaccharide (LPS)). Upon activation, M1 M⌽s secrete inflammatory cytokines including TNF␣, IL-1␤, IL-6, and others. In contrast, M2 M⌽s are activated by prototypical type 2 T-helper (Th2) cytokines, IL-4, and IL-13 (2). Importantly, M1 M⌽s preferentially metabolize glucose as an energy substrate (8,9). Similarly, dendritic cells (10) and T-cells (11)(12)(13) exhibit a Warburg-like shift in cellular metabolism upon activation. However, alternatively activated M2 M⌽s primarily utilize fatty acids to fuel cellular behavior and activity (8,9). Taken together, glucose metabolism is central to the function of classically activated M⌽s and thus is a potential target for modifying inflammatory responses.
M⌽ subtypes modulate inflammation-associated insulin resistance (reviewed in detail in Ref. 2). Broadly, M2 M⌽s are immunoregulatory and maintain insulin sensitivity in adipose and liver, whereas M1 M⌽s disrupt insulin sensitivity in these tissues. In lean individuals, M2 M⌽s reside in adipose tissue and are involved in tissue remodeling and repair (14). At the onset of obesity and as weight gain continues, M1 M⌽s infiltrate the adipose tissue contributing to the low grade, chronic inflammation associated with excess adiposity (14). Based upon existing evidence for fuel preferences of M1 versus M2 M⌽s, we posited that manipulating M⌽ substrate metabolism may serve as a novel approach for controlling macrophage polarization and inflammatory capacity.
We sought to manipulate the M⌽ inflammatory response through the modulation of the primary glucose transporter in M⌽s, GLUT1 (15)(16)(17)(18). To this end, we overexpressed GLUT1 in the RAW264.7 murine M⌽ cell line to test the hypothesis that increased intracellular glucose availability and subsequently enhanced glucose metabolism can independently drive a hyperinflammatory response. Through radiotracer, metabolomics, bioenergetics, and expression analysis studies, we have demonstrated that elevated GLUT1-driven glucose metabolism drives reactive oxygen species (ROS) production and expression of proinflammatory mediators. Inhibiting glycolysis or treating cells with an antioxidant reversed GLUT1-mediated proinflammatory elevations. Finally, in vivo detection of GLUT1 in M⌽-laden crown-like structures in adipose tissue and inflammatory loci in livers of obese animals demonstrated important evidence for GLUT1-mediated glucose metabolism in tissue inflammation. Through understanding mechanisms of metabolic reprogramming driven by substrate availability, we provide important insights into the control of inflammation.

EXPERIMENTAL PROCEDURES
Cell Culture and Chemicals-For the GLUT1 construct, Rat Glut1 with a tandem FLAG epitope (DYKDDDDKDYKD-DDDK, inserted between amino acids 66 and 67) was cloned into the pEF6 vector (Invitrogen) (11,19,20). RAW264.7 (RAW) mouse M⌽s were transfected with either the empty vector or Glut1 construct using the AMAXA nucleofector V kit (Lonza, Cologne, Germany). Stable cell lines were established by selecting in 10 g/ml Blasticidin S (Invitrogen) for 2 weeks. Empty vector M⌽s (GLUT1-EV) were diluted serially to obtain clonal isolates. Glut1-transfected M⌽s were labeled with a fluorescent anti-FLAG antibody, and Glut1-overexpressing cells (GLUT1-OE) were purified by sterile cell sorting. This population of cells was diluted serially to obtain clonal isolates. RAW cells were maintained in 10 g/ml Blasticidin S. Studies reported herein were repeated in a second set of clones and the parent heterologous population of GLUT1-EV and GLUT1-OE from which the clones were isolated (data not shown). Lipopolysaccharide (Sigma) was diluted in sterile PBS at 1 mg/ml. Stock reagents were stored at Ϫ80°C under light-free conditions and diluted to appropriate final concentrations in the media. For ROS assays, N-acetylcysteine (Enzo Life Sciences, Inc., Farmingdale, NY) and 2-deoxyglucose (Sigma) were prepared in sterile PBS just before use.
Rodents and Bone Marrow-derived M⌽s (BMDM)-All animals were housed in a temperature controlled facility with a 12-h light/dark cycle in compliance with the Duke and University of North Carolina at Chapel Hill Institutional Animal Care and Use Committee protocols. C57BL/6 male mice were purchased from The Jackson Laboratory (Bar Harbor, ME) and bred at the University of North Carolina. To generate C57BL/ 6-derived BMDM, bone marrow was collected from femur and tibia bones (5,6). BMDMs were generated by culturing marrow cells in 10-cm non-tissue culture dishes for 6 days with 30% L929 fibroblast-conditioned media and 70% RPMI supplemented with 10% FBS, 1ϫ L-glutamine, and 1ϫ penicillin/ streptomycin antibiotic mix. BMDMs were left untreated as naïve unstimulated M⌽s (M0) or polarized using 5 ng/ml LPS and 10 ng/ml IFN␥ (M1) or 10 ng/ml IL-4 (M2) for 24 h as in Vats et al. (8). To examine adipose and liver in lean and obese rodents, rats were given ad libitum access to either a purified diet containing 10% kcal from fat (low fat diet, Research Diets D07010502, New Brunswick, NJ) or a diet containing 45% kcal from fat (high fat diet, Research Diets, D06011802) as reported in Sampey et al. (4,5). For tissue collection, animals were anesthetized with tribromoethanol (0.02 ml/g of a 1.25% solution) and euthanized by cervical dislocation. Epididymal white adipose tissue and liver were fixed in 10% formalin and paraffinembedded for immunohistochemical analysis. To examine in vivo GLUT1 activation and inflammatory gene expression, Myc-epitope-tagged Glut1 knock-in mice (Glut1 myc) mice that express exofacially Myc-tagged GLUT1 at endogenous levels (12) were used (n ϭ 3, 12-week-old females). Mice were sacrificed, and an adipose tissue single cell suspension was generated using a GentleMACS tissue dissociator (Miltenyi Biotec Inc., Cambridge, MA) according to the manufacturer's recommended protocol. Cells were stained with anti-F4/80-APC (eBioscience, San Diego, CA), mouse-anti-Myc tag (Millipore clone 4A6), and V450-anti mouse IgG, then fixed in 1% paraformaldehyde and permeabilized with methanol. After permeabilization, cells were stained with anti-IL-6-phycoerythrin (PE) and anti-TNF␣-FITC antibodies (eBioscience) and analyzed using a MacsQuant flow cytometer (Miltenyi Biotec). Data show the mean Ϯ S.E. of the mean fluorescence intensity.
Immunocytochemistry and Immunohistochemistry-GLUT1 localization was determined in RAW264.7 GLUT1-OE cells. Cells were washed in PBS, 2% FBS and blocked with Fc block (1:100 in 5% BSA for 10 min in 5% rat serum). Primary rabbit anti-FLAG (Sigma #F7425), R-PE donkey anti-rabbit (Jackson Immuno catalog 711-116-152), and DAPI were used, and images were captured under a Nikon Microscope (Melville, NY). 5-m sections of liver and epididymal white adipose from lean and obese rats were used to examine GLUT1 expression in vivo by immunohistochemistry (4).
qPCR and qPCR Arrays-mRNA was isolated (Qiagen RNeasy mini kit, Qiagen, Valencia, CA) and reverse-transcribed into cDNA using the iScript cDNA synthesis kit (Bio-Rad). The expression of specific genes was quantitated by qPCR using Assay-On-Demands (ABI) on an ABI 7900HT machine (Invitrogen) (21). Mouse Innate and Adaptive Immune Responses Array qPCR arrays were purchased from SABiosciences (Qiagen). n ϭ 3 for GLUT1-OE and GLUT1-EV M⌽s.
Microarray Expression Analysis-Gene expression analysis was performed as previously reported (22) using Agilent microarrays (G4846A) as per the manufacturer instructions. A two-color labeling was performed using Low Input Quick Amp labeling kits (Agilent, Santa Clara, CA) with GLUT1 overexpressing cells labeled with Cy3 and a reference channel (Universal Mouse Reference, Stratagene, Santa Clara, CA) labeled with Cy5. Data analyses were performed using R (Version 2.9.2). Genes were filtered with only genes where Ͼ70% of microarrays had a signal Ͼ10 dots per inch in both channels included. Data were Lowess-normalized, and missing data were imputed using k-nearest neighbor imputation (with k ϭ 6). Data were analyzed by 2-class significance analysis of microarrays (SAM) to identify genes that significantly changed by GLUT1 overexpression (i.e. GLUT1-EV versus GLUT1-OE). Significant probes were collapsed by gene and visualized using Java Treeview. Significant genes were evaluated for ontological enrichment using Ingenuity Pathway Analysis (Ingenuity Systems), with Benjamini-Hochberg (B-H) multiple testing correction. Significant functions and pathways were defined as those with B-H p values less than 0.05.
Proteome Profiler Cytokine Array-GLUT1-EV and GLUT1-OE RAW264.7 M⌽s were plated into 24-well plates and either treated with media alone or 100 ng/ml LPS for 24 h. Conditioned media was assayed using Proteome Profiler Array Mouse Cytokine array panel A (ARY006, R and D Systems, Inc., Minneapolis, MN) following the manufacturer's protocol. Digital images were quantified using the Versadoc 4000, and relative abundance with background subtracted was normalized to internal standards, as in Sampey et al. (5).
Cellular Bioenergetics-Seahorse Analyses-60,000 GLUT1-EV or GLUT1-OE M⌽s were seeded into each well of a Seahorse Bioscience (Billerica, MA) tissue culture plate and incubated overnight at 37°C in an atmosphere of 5% CO 2 . Before analysis, cells were washed twice with DMEM assay media (sodium bicarbonate-and glucose-free DMEM supplemented with glutamine and penicillin/streptomycin, pH 7.4) and incubated for 1 h at 37°C without CO 2. Measures of glycolytic rate and glycolytic capacity were determined by recording extracellular acidification rates (ECAR, milli pH/min) and oxygen consumption rates (OCR, pmol/min) on a Seahorse Bioscience XF24 -3 Extracellular Flux Analyzer. The injection of glucose was used to measure glycolytic rate (final concentration 25 mM; Sigma), the injection of oligomycin was used to measure glycolytic capacity (final concentration 2.5 M; Sigma), and the injection of 2-deoxyglucose was used to blunt glycolysis (final concentration 20 mM). Background from cell-free wells was subtracted. Cells were lysed, and protein concentrations were determined using BCA protein assay. n ϭ 6 replicates per treatment for ECAR and OCR. Data were normalized to protein in each well. Experiments were repeated n ϭ 3 times.
Radiotracer Glucose Uptake Assay-GLUT1-EV and GLUT1-OE RAW264.7 M⌽s were plated at 1.5 ϫ 10 5 cells/well into 24-well plates and allowed to attach overnight. Cells were treated with vehicle or 100 ng/ml LPS for 24 h. The cells were washed twice with 0.5 ml of 37°C 1ϫ PBS and then incubated for 10 min at 37°C in 0.5 ml/well DMEM (136 mM NaCl, 4.7 mM KCl, 1.25 mM CaCl 2 , 1.25 mM MgSO 4 , 10 mM HEPES, pH 7.4) with 0.5 Ci/ml deoxy-D-[2-3 H]glucose (PerkinElmer Life Sciences). Cells were washed twice with 0.5 ml of cold 1ϫ PBS with 20 mM D-glucose and then lysed in 0.5 ml of lysis buffer (0.025% SDS, 1% Triton X-100 in 1ϫ PBS). Lysates were centrifuged at 20,000 ϫ g for 5 min at 4°C, and the protein concentration was determined with a BCA assay. Remaining lysate was counted by scintillation. The data are expressed as dpm/mg of protein.
Radiotracer Glucose Oxidation and Glycogen Synthesis-GLUT1-EV and GLUT1-OE RAW264.7 M⌽s were plated as above. Glucose oxidation to CO 2 was measured by incubating cells with ubiquitously labeled [ 14 C]glucose for 2 h and then measuring [ 14 C]CO 2 in the media. Briefly, the cells were washed twice with 0.5 ml of 37°C 1ϫ PBS and then incubated for 120 min at 37°C in 0.5 ml/well DMEM with 2 Ci/ml [ 14 C]uniformly labeled D-glucose (PerkinElmer Life Sciences). [ 14 C]CO 2 released from acidified media was captured and subjected to scintillation counting. The data are expressed as dpm/mg of protein/minute. Cells were then washed twice with 0.5 ml of iced 1ϫ PBS with 20 mM D-glucose and lysed in 0.5 ml of lysis buffer (30% KOH saturated with Na 2 SO 4 ) to ethanolprecipitate glycogen, which was measured by scintillation counting. The data are expressed as dpm/mg of protein/min.
Lactate Assay-Media from experiments was saved in aliquots at Ϫ80°C and assayed for lactate content. Briefly, fresh NADϩ was added to assay buffer (175 mM hydrazine sulfate, 68 mM glycine, 2.9 mM EDTA, 11.3 mM NADϩ, pH 9.5) just before assay. In a 96-well plate, 200 l of assay buffer and 40 l of lactate standard or media were combined in each well. The plate was then pre-read at 340 nm followed by the addition of 1 unit of lactate dehydrogenase in a 10-l volume of water. The plate was mixed and read every minute for 30 min at 340 nm. The standard curve measured concentrations of 0, 1, 2, 3, 4, 5, and 10 M lactate. The lactate concentration in each sample was determined from the standard curve.
Metabolomics-GLUT1-EV and GLUT1-OE M⌽s were plated at 5 ϫ 10 5 cells/ml and allowed to attach overnight. Fresh medium was added, and cells were grown for an additional 24 h. PBS-washed cells were detached by scraping and pelleted for analysis by Metabolon, Inc. (Durham, NC) as in Bhatt et al. (23). Data are presented as relative measures of "scaled intensity" after normalization to protein and median scaling to 1. Missing values were imputed with the minimum.
Reactive Oxygen Species Production-1.5 ϫ 10 6 GLUT1-EV or GLUT1-OE M⌽s were seeded into 6-well plates. Cells were either untreated or incubated with 5 mM N-acetyl-cysteine (NAC) or 1 mM 2-deoxyglucose (2-DG) and incubated overnight in DMEM at 37°C in an atmosphere of 5% CO 2. Total ROS and superoxide levels in live, unfixed cells were measured by a total ROS/superoxide kit per manufacturer's instructions (Enzo Life Sciences) and analyzed using a Beckman Coulter CyAn ADP flow cytometer (Brea, CA). Flow cytometry data were analyzed using FlowJo software (TreeStar, Ashland, OR), and median fluorescence intensity was used to compare total ROS and superoxide produced by GLUT1-EV or GLUT1-OE M⌽s. At least 5 ϫ 10 3 cellular events were analyzed.
Statistics-Statistical differences between experimental groups were determined using statistics software within GraphPad Prism (GraphPad Software, Inc., La Jolla, CA). 1-Way analysis of variance was used to determine the existence of statistically significant differences between RAW, GLUT1-EV, and GLUT1-OE M⌽s. Non-paired Student's t test was then used to determine p values comparing GLUT1-OE M⌽s to the two control cell lines. For all experiments, representative results are reported for at least n ϭ 3. Experiments were conducted with n ϭ 3-6 replicates per assay. All data are shown as means Ϯ S.E. For metabolomics study, two-way analysis of variance with p Ͻ 0.05 was considered significant.

GLUT1 Was the Predominant Glucose Transporter in M⌽s-
BMDM from C57BL/6J mice were plated and left untreated as naïve unstimulated M⌽s (M0) or polarized to M1 or M2 as described under "Experimental Procedures." qPCR analysis of the glucose transporters GLUTs 1-4 revealed that GLUT1 mRNA was up-regulated 10-fold by M1 polarization compared with M0 BMDMs (Fig. 1A). M2-polarized M⌽s up-regulated GLUT1 expression by 2-fold compared with M0. GLUT3 was minimally detected in M0 and down-regulated with M1 polarization. GLUT2 and -4 were undetected (Fig. 1A). Similar to BMDMs, the most abundant glucose transporter in the RAW M⌽ cell line was GLUT1 followed by a low level of expression of GLUT3. Expression of GLUT2 or GLUT4 was not detected (Fig. 1B). RAW M⌽s up-regulated GLUT1 mRNA with no change in GLUTs 2-4 in response to 24-h LPS stimulation (Fig.  1B).
GLUT1 Was Up-regulated in Inflamed Obese Adipose and Liver-To determine the relevance of M⌽ GLUT1 in vivo, we stained for GLUT1 expression in adipose and liver of lean and diet-induced obese rats. GLUT1-positive staining was associated with inflammatory M⌽ infiltrates in epididymal white adipose tissue and liver from obese animals fed a high fat diet compared with low fat diet-fed lean controls. GLUT1-positive staining was co-localized with M⌽s in crown-like structure (CLS) and inflammatory loci (Fig. 1C) (4,5). In addition, to confirm in vivo expression of GLUT1 and cytokine expression, we isolated F4/80ϩ macrophages from adipose tissue of mice (Fig. 1D). Macrophages that expressed high levels of surface GLUT1 contained significantly greater levels of IL-6 (p ϭ 0.018) and TNF␣ (p ϭ 0.0062) compared with macrophages expressing low levels of GLUT1.
Generation of GLUT1 Overexpression M⌽ Model-To investigate the role of glucose metabolism in the inflammatory state of M⌽s, GLUT1 was stably transfected into RAW M⌽s. Clonal lines stably expressing either pEF6 vector alone (GLUT1-empty vector (GLUT1-EV) or pEF6 overexpressing GLUT1 (GLUT1-OE) were generated and grown according to ATCC recommendations in 4.5 g/liter glucose containing DMEM. Immunocytochemical staining of GLUT1-EV and GLUT1-OE clones demonstrated successful overexpression of GLUT1 and correct localization in the GLUT1-OE cell line, as seen in the red punctate and primarily membrane-localized areas ( Fig. 2A). qPCR of GLUT1 in untreated RAW, GLUT1-EV, and GLUT1-OE M⌽s revealed a nearly 12-fold overexpression of GLUT1 mRNA in the GLUT1-OE cells compared with the other M⌽ groups (Fig.  2B). GLUT1-OE M⌽s expressed 2.28-fold greater GLUT1 protein levels relative to GLUT1-EV and RAW M⌽s (Fig. 2, C and  D). LPS stimulation resulted in a 2-fold increase in GLUT1 protein in RAW and GLUT1-EV cells and a 1.5-fold increase in GLUT1-OE cells. The overexpression of GLUT1 at the protein level in this model was at physiologic levels and was comparable to a 1.64-fold induction in GLUT1 protein on adipose derived macrophages after a 3-h intraperitoneal LPS injection (data not shown) as well as 1.8-fold LPS-induced elevations in glucose uptake in BMDMs. The viability and growth characteristics of the M⌽s were not altered by the overexpression of GLUT1 as

GLUT-1-mediated Glucose Metabolism Drives M⌽ Inflammation
analyzer. ECAR was more than doubled in GLUT1-OE compared with GLUT1-EV M⌽s (p ϭ 0.0003, Fig. 3A). The addition of oligomycin (an inhibitor of ATP synthase) to test glycolytic capacity induced a greater increase in GLUT1-OE M⌽s ECAR compared with GLUT1- EV (p ϭ 0.01, Fig. 3B). Inhibiting gly-colysis with 2-DG reduced ECARs equivalently in GLUT1-OE and GLUT1-EV M⌽s, indicating no difference in glycolytic reserve (not shown). Interestingly, GLUT1-OE M⌽s demonstrated significantly reduced OCR compared with GLUT1-EV M⌽s throughout the experiment regardless of whether glucose FIGURE 3. GLUT1 up-regulated glycolytic rates and capacity and blunted respiratory capacity. ECAR and OCR were measured using the Seahorse Bioanalyzer. Glycolytic rate (A; *, p ϭ 0.0003) and glycolytic capacity (B; *, p ϭ 0.01) were measured. Glycolytic reserve was equal in GLUT1-EV and GLUT1-OE (not shown). Oxygen consumption was measured using the Seahorse before glucose addition to the media (C, Ϫ glucose, *, p ϭ 0.002) and after glucose injection into the assay (D, ϩ glucose, *, p ϭ 0.0005). n ϭ 3 experiments with n ϭ 6 replicates were conducted, and data were normalized to protein in each well.  MARCH 14, 2014 • VOLUME 289 • NUMBER 11

JOURNAL OF BIOLOGICAL CHEMISTRY 7889
was present in the assay media (Fig. 3, C and D, p Ͻ 0.002 and p Ͻ 0.0005, respectively).
Comprehensive metabolomic analysis was performed on GLUT1-EV and GLUT1-OE cell lysates to examine alterations in glycolytic and other pathway metabolites. 152 biochemicals (45 up-regulated and 107 down-regulated) were significantly different (p Ͻ 0.05) between unstimulated GLUT1-OE compared with GLUT1-EV supplemental Table 1). Several glycolytic intermediates were significantly up-regulated ( Fig. 5A and supplemental Table 1). GLUT1-OE M⌽s contained significantly less intracellular glucose (p ϭ 0.033), likely due to increased glycolysis, as evidenced by elevated intracellular lactate concentrations (p ϭ 0.049) (Fig. 5A and supplemental Table 1). In addition, greater amounts of metabolites in the PPP

GLUT1 Drives a Proinflammatory Phenotype and Alters Redox Status of M⌽s-Having established that GLUT1-OE
M⌽s take up and metabolize more glucose compared with the GLUT1-EV M⌽s, we next set out to test our central hypothesis that increasing glucose metabolism in M⌽s renders these cells more proinflammatory. Supervised two-class analysis comparing GLUT1-OE to GLUT1-EV revealed that 1547 genes were differentially regulated by GLUT1 (775 genes down-regulated, 772 genes up-regulated) using a false discovery rate of 3.2% (supplemental Table 2). Significantly regulated genes were visualized using hierarchical clustering (Fig. 6A). Using Ingenuity Pathway Analysis, the genes in cluster 1 (Fig. 6A) were defined by expression of genes involved in the cellular function and maintenance, protein synthesis, cellular development, cellular growth and proliferation, and cell morphology functional pathways. Down-regulated canonical pathways included cross-talk between dendritic cells and Natural Killer cells, antigen presentation pathway, communication between innate and adaptive immune cells, allograft rejection signaling, and role of pattern recognition signaling in recognition of bacteria and viruses (Table 1). Cluster 2 (Fig. 6A) was associated with a GLUT1-mediated increase in expression of genes involved in cell cycle, DNA replication, recombination, and repair, cellular assembly, and organization, cell-to-cell signaling and interaction, and cellular function and maintenance. Up-regulated canonical pathways in cluster 2 included cholesterol biosynthesis I, II, III and caveolar-mediated endocytosis signaling (Table  1).
Because GLUT1 overexpression and subsequent enhanced glucose metabolism regulated important signaling and metabolic pathways that could affect M⌽ inflammatory function, we next investigated the effect of GLUT1 overexpression on inflammatory activation using quantitative gene and protein expression analysis. First, using a targeted qPCR array for 84 genes involved in the innate and adaptive immune response, GLUT1-OE was found to significantly regulate 12 genes including a 5-fold increase in serpine1, also known as plasminogen FIGURE 6. GLUT1 gain of function drives increased expression of inflammatory mediators. A, microarray analysis of untreated GLUT1-EV and GLUT1-OE M⌽s was conducted. Supervised two-class significance analysis of microarray analysis identified 1,547 genes differentially regulated by GLUT1 at a false discovery rate of 3.2% (supplemental Table 2). 775 were down-regulated and 772 were up-regulated in a GLUT1-dependent manner. Cluster analysis revealed two clusters of genes regulated by GLUT1. B, media from GLUT1-EV and GLUT1-OE M⌽s were examined for secretion of inflammatory mediators using a proteome profiler array and normalized to protein lysates. Significantly (p Ͻ 0.05) regulated proteins are shown with the exception of eotaxin and IL-10. MARCH 14, 2014 • VOLUME 289 • NUMBER 11 activator inhibitor 1 (PAI-1, p Ͻ 0.05, Table 2). We next examined secreted cytokines using a proteome profiling array. GLUT1-OE cells significantly increased secretion of 11 cytokines and growth factors (Fig. 6B and Table 3). Notably, G-CSF, IL-6, CXCL2, and RANTES all show a Ͼ10-fold increase compared with GLUT1-EV M⌽s, whereas CXCL1, TNF-␣, IL-1 receptor antagonist, IL-5, CCL4, IL-17, and CCL1 were up-regulated by at least 2-fold (Fig. 6B). Three cytokines (IL-2, IL-27, and CXCL10) were significantly reduced by GLUT1-OE (Table  3).
Finally, ROS production is a hallmark of M1 M⌽ polarization (2). Based on the increased glycolytic rate of the GLUT1-OE M⌽s, we hypothesized that elevated ROS may be the mechanism linking increased glucose metabolism to proinflammatory cytokine production. Metabolomic analysis indicated a shift in intracellular redox balance with GLUT1 overexpression, such that that GLUT1-OE had a greater GSSG/GSH ratio (0.764) than GLUT1-EV (0.337) (supplemental Table 1). To examine ROS production directly, RAW, GLUT1-EV, and GLUT1-OE M⌽s were labeled with fluorescent probes to measure total ROS and superoxide. Indeed, mean fluorescent intensity of GLUT1-OE M⌽s was 3-fold greater than RAW cells and GLUT1-EV for both total ROS (p Ͻ 0.0005, Fig. 7C) and superoxide (p Ͻ 0.0005, Fig. 7D). To test the dependence of PAI-1 expression on ROS signaling, GLUT1-EV, and GLUT1-OE, M⌽s were incubated with 5 mM NAC for 24 h to quench ROS, and PAI-1 expression was again assessed by qPCR. NAC treatment resulted in a complete reversal in the up-regulation of PAI-1 in the GLUT1-OE M⌽s (p ϭ 0.0001, Fig. 7B).

DISCUSSION
Immune cell activation in response to infection or other stressors is a metabolically expensive event (2). To date there is evidence that M⌽ glucose metabolism plays a key role in mounting the robust, proinflammatory response to bacterial
Adipose tissue M⌽s displaying a proinflammatory M1-like phenotype interfere with insulin sensitivity and perpetuate obesity-induced inflammation (2). Strategies that modulate substrate metabolism in M⌽s would be innovative approaches to address insulin resistance and diabetes. To this end we investigated whether manipulating the amount of glucose substrate available to M⌽s for metabolism by modulating the expression of the GLUT1 transporter would alter the inflammatory characteristics of M⌽s. Herein, we report that GLUT1 is the primary glucose transporter in murine BMDM and RAW264.7 M⌽s and that proinflammatory activation of M⌽s either through M1 polarization of BMDM or LPS stimulation of RAW M⌽s increased GLUT1 expression. When GLUT1 was overexpressed in M⌽s, elevations in glycolytic rates and lactate production in conjunction with increased glucose oxidation and glycogen storage were detected, suggesting that there was an overall increased flux though glucose-dependent metabolic pathways within the cell. Metabolomic analysis also demonstrated activation of the PPP. The PPP utilizes glucose to generate NADPH for use in biosynthetic pathways, ROS production, and ribose sugars for nucleotide biosynthesis. Increased levels of PPP metabolic intermediates in GLUT1-OE indicate glucose availability in excess of energetic requirements of the cell. Bioenergetic analysis further demonstrated that GLUT1-overexpressing cells are highly glycolytic, with blunted overall oxidative capacity compared with controls. GLUT1-overexpressing cells also contained less CoA and B vitamin pantothenate (necessary for CoA generation). The lack of CoA in GLUT1-OE cells may limit excess pyruvate or fatty acids from entering the TCA cycle for oxidative metabolism. Taken together, GLUT1-mediated glucose metabolism drives metabolic programming of macrophages by switching fuel metabolism from oxidative to glycolytic.
M⌽s, which display elevated GLUT1-mediated glucose uptake and metabolism, are forced into a hyperinflammatory state with increased production of multiple inflammatory pathways and protein mediators. Our findings support those of Haschemi et al. (34) that demonstrated classical, M1 M⌽ activation FIGURE 7. GLUT1 overexpression inflammatory mediator production is dependent upon glycolysis and ROS production. A, qPCR of PAI-1 was measured in GLUT1-EV and GLUT1-OE M⌽s left unstimulated or treated with 100 ng/ml LPS for 24 h. Graphs are of combined data from three independent experiments Ϯ S.E. *, p Ͻ 0.05 versus unstimulated; , p Ͻ 0.05 versus GLUT1-EV). B, qPCR of PAI-1 was measured in GLUT1-EV and GLUT1-OE M⌽s left untreated or co-treated with 1 mM 2-DG to inhibit glycolysis or 5 mM NAC to quench ROS for 24 h. Graphs are of combined data from three independent experiments Ϯ S.E. GLUT1-EV is set to 1. *, p ϭ 0.0001. C, ROS production was measured in unstimulated RAW, GLUT1-EV, and GLUT1-OE cells. MFI, mean fluorescence intensity. Total ROS (C) and superoxide (D) in RAW, GLUT1-EV, and GLUT1-OE were detected. Graphs shown are representative of three separate experiments done in triplicate. Means Ϯ S.E. are shown. *, p Ͻ 0.0005 versus RAW and GLUT1-EV.
is dependent upon increased flux through the glycolytic and pentose phosphate pathways and extend those findings to show that increasing glucose uptake through GLUT1 is, by itself, sufficient to impart an M1-like phenotype in M⌽s. Furthermore, Tannahill (9) elegantly demonstrated glycolytic dependence of LPS-induced IL-1␤ release in BMDMs; however, alterations in succinate or GABA were not detected in our cell system, suggesting an alternate mechanism linking glucose metabolism to inflammation. One mechanism by which GLUT1-mediated metabolism may drive inflammation is through ROS production, which we demonstrated was highly increased with GLUT1 overexpression. Upon activation, classically activated M1 M⌽s undergo a "respiratory burst," referred to as the "glycolytic burst," driven by significantly enhanced NADPH oxidase activity, the purpose of which is to generate large amounts of ROS to destroy invading bacteria (35). To support redox balance, the glycolysis-PPP axis is activated in response to M1 polarization, presumably to sustain enhanced production of NADPH for use by NADPH oxidase in addition to production of glutathione used by the M⌽s to protect from the excessive amounts of superoxide being produced (35). We report that overexpression of GLUT1 alone leads to increased overall ROS burden and increased intracellular oxidative stress as evidenced by an increased GSSG/GSH ratio. GLUT1-mediated metabolism ties increased cellular glucose to oxidative stress; the role of oxidative stress in obesity-induced inflammation is well established (2).
PAI-1 is an important proinflammatory mediator linked to the pathophysiology of obesity and vascular diseases such as atherosclerosis (2). It was robustly up-regulated in GLUT1-OE M⌽s regardless of LPS stimulation or not. Therefore, we tested the dependence of PAI-1 expression on GLUT1-mediated glycolysis. Pharmacologic inhibition of glycolysis using 2-DG blunted expression PAI-1. The fact that PAI-1 is an established NFB target gene and NFB is regulated by redox status (36) could be a mechanism linking GLUT1, enhanced glucose metabolism, and increased proinflammatory mediators, including ROS. We have reported that treatment of GLUT1-OE M⌽s with the antioxidant NAC also blunted GLUT1-driven PAI-1 expression. Taken together, because the M1 M⌽ subtype has been shown to modulate inflammation-associated insulin resistance in tissues such as adipose and liver (14,37,38), it is likely that GLUT1 may contribute to maintenance of the proinflammatory M1 phenotype present in obese tissue. Indeed, in vivo evidence for the role of GLUT1 in proinflammatory M⌽s was demonstrated; GLUT1 positive M⌽s co-localized with crown-like structures and inflammatory loci in white adipose and livers, respectively, of obese rats versus lean controls (4,5). In addition, adipose tissue macrophages displaying high surface GLUT1 contained greater cytokines, suggesting that GLUT1 correlates with in vivo inflammation. The dependence of these findings upon GLUT1 remains to be determined in future studies.
Elevated lipogenesis and cholesterol accumulation are other potential mechanisms by which glucose metabolism can lead to inflammation (39 -45). GLUT1-regulated pathways identified by Ingenuity Pathway Analysis included elevated cholesterol synthesis. Cholesterol is a known activator of inflammatory pathways (46,47). Importantly, we and others have shown that lipids such as free cholesterol can induce endoplasmic reticulum stress and promote M⌽-mediated inflammation via elevations in inflammatory kinase activity such as JNK and IK, which promote insulin resistance, obesity, and atherosclerosis (46). Hence, the link between GLUT1-mediated elevated glucose metabolism and elevated cholesterol production in M⌽s is currently under investigation.
Taken together our work demonstrates that M⌽ metabolic reprogramming, which is critical to M⌽ inflammatory capacity, is driven by GLUT1-mediated glucose uptake and subsequent enhanced glucose metabolism through the PPP. This phenomenon was detected in the immortalized RAW macrophage cell line, which are inherently highly glycolytic. The relevance of GLUT1 to in vivo inflammatory states is currently under investigation. It is likely that glucose metabolism may aid in finetuning the chronic, obesity-associated M⌽-mediated inflammation in insulin resistance. As obesity has reached globally epidemic proportions (48), the ability to better understand obesity-induced inflammation and identify putative targets that result in the manipulation of substrate metabolism in M⌽s will aid in the discovery of novel pathways that regulate M⌽-mediated inflammation and the promotion of obesity and diabetes.