Characterization of Glucose Transporter 6 in Lipopolysaccharide-Induced Bone Marrow–Derived Macrophage Function

Macrophages have important roles in host defense and tissue repair. In general, naive monocytes respond to extracellular signals and polarize to either a classically activated M1 state for host defense or an alternatively activated M2 state for tissue repair (1, 2). M1 polarization is triggered by bacterial infections and proinflammatory stimuli, which initiate genetic and metabolic reprogramming, including the Gram-negative bacterial cell wall component LPS (1). M1 cells are highly glycolytic and have phagocytic and bactericidal properties (1). In contrast, M2 polarization is triggered by signals associated with wound repair and parasite infection, including IL-4. M2 cells have low glycolytic activity and high oxidative metabolism that is more suited for producing substrates important for tissue repair and parasite defense (3). The differences in metabolism between M1 and M2 macrophages are discriminating phenotypes of these cells (1, 2).

The increase in glycolytic metabolism from naive macrophages (M0) to classically activated M1 macrophages is facilitated in part by increased expression of glucose transporter (GLUT) GLUT1 (4). GLUT1 is functionally important for M1 polarization, as knockdown of GLUT1 in bone marrow–derived macrophages (BMDMs) was sufficient to decrease glucose consumption, lactate production, and IL-6 secretion (5), whereas overexpression of GLUT1 increased glucose uptake, glycolytic metabolism, and secretion of IL-6 and TNF-α (4). GLUT1 clearly has an important role in M1 macrophage metabolism and function, but it is unclear whether other GLUT proteins also contribute to macrophage polarization or function.

There are 14 members of the GLUT family (belonging to the gene family Slc2a); however, most literature in the context of macrophage function focuses on the role of GLUT1. The best characterized glucose transporters are GLUTs 1-4, and of these, GLUT1 and GLUT3 are expressed in macrophages (4–7), whereas GLUT2 and GLUT4 are not expressed (4). GLUT6 has not been functionally investigated in macrophages.

GLUT6 mRNA is expressed primarily in the brain, spleen, and blood cells (8) of healthy humans and mice but becomes upregulated in some forms of cancer, including endometrial cancer (9). Importantly, knockdown of GLUT6 in endometrial cancer cells is sufficient to cause cell death (9), demonstrating a functional role for GLUT6 in cancer cell survival. To investigate potential adverse consequences of GLUT6 inhibition on normal tissues, we developed a GLUT6 knockout (KO) mouse and found that loss of GLUT6 had a negligible effect on whole body physiology (10). Together, these data identify GLUT6 as an anticancer drug target that is expected to have minimal adverse effects.

Next, we searched for a normal physiological process where GLUT6 is upregulated. Examining publicly available gene expression databases, including Functional Annotation of the Mammalian Genome and the Gene Expression Omnibus (GEO), we found that GLUT6 was highly upregulated following LPS stimulation of macrophages (11–15). In this study, we show that GLUT6 protein is abundantly expressed in LPS-induced M1 BMDMs but not M0 or M2 macrophages. Genetic deletion of GLUT6 had minimal impact on M1 polarization and cytokine production, thereby indicating that anticancer therapies targeting GLUT6 are unlikely to interfere with normal macrophage function.

GLUT6 KO (Slc2a6^−/−) mice were generated as described previously (10). Mice were bred at Australian BioResources (Moss Vale, NSW, Australia) until 5 wk of age and then transferred to the University of New South Wales (UNSW) Biological Resources Centre (Sydney, Australia). GLUT6 KO and their wild-type (WT) littermates were maintained in cages (maximum five per cage) in a specific pathogen–free facility with 12 h light/dark cycles. Mice were fed chow diet (Gordon’s Specialty Feeds) ad libitum. All mouse experiments were approved by the UNSW Animal Care and Ethics Committee.

WT and GLUT6 KO C57BL/6 littermates were euthanized between 6 and 10 wk of age by exposure to carbon dioxide followed by cervical dislocation. Both femurs were isolated, and bone marrow was flushed with ice-cold PBS using a 25-gauge needle. Cells were spun for 10 min at 10,000 × g at room temperature then resuspended in 80% maintenance medium (DMEM/F-12 [1:1], 10% [v/v] FBS, and 1% penicillin/streptomycin) with 20% conditioned medium (obtained from confluent L929 mouse fibroblast cells grown in maintenance medium) at a density of 2–10 × 10⁶ cells/ml in 10-cm dishes. Every 3 d, an additional 20% conditional medium was added to cells for 7–10 d prior to experimentation. For polarization, BMDMs were reseeded in maintenance medium (free from conditioned medium), and 3 h later, they were treated with LPS or IL-4 with doses and durations indicated in figure legends.

Cells were rinsed with PBS, and RNA was extracted using TRI Reagent (Sigma-Aldrich), according to manufacturer’s instructions. RNA was treated with DNase I (Sigma-Aldrich), and cDNA was synthesized using ≥250 ng of RNA using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems). Quantitivative real-time PCR was performed using iTaq Universal SYBR Green (Bio-Rad Laboratories) using the ViiA 7 Real-Time Machine (Applied Biosciences). The murine gene primer sequences were as follows: Chi3l3 forward, 5′-AGAAGGGAGTTTCAAACCTGGT-3′ and Chi3l3 reverse, 5′-GTCTTGCTCATGTGTGTAAGTGA-3′; Fizz1 forward, 5′-TACTTGCAACTGCCTGTGCTTACT-3′ and Fizz1 reverse, 5′-TATCAAAGCTGGGTTCTCCACCTC-3′; Ppia forward, 5′-CGATGACGAGCCCTTGG-3′ and Ppia reverse, 5′-TCTGCTGTCTTTGGAACTTTGTC-3′; Slc2a1 forward, 5′-AGCCCTGCTACAGTGTAT-3′ and Slc2a1 reverse, 5′-AGGTCTCGGGTCACATC-3′; Slc2a3 forward, 5′-TTCTGGTCGGAATGCTCTTC-3′ and Slc2a3 reverse, 5′-AATGTCCTCGAAAGTCCTGC-3′; Slc2a6 forward, 5′-GCGACTCCTGGAGAGAGAGA-3′ and Slc2a6 reverse, 5′-CAGGATGCCTGGATTTTGTC-3′; Slc2a8 forward, 5′-ACATCTCGGAAATCGCCT-3′ and Slc2a8 reverse, 5′-ACACAGCCCAGCACG-3′; IL-6 forward, 5′-AAAGCCAGAGTCCTTCAGAGAGATAC-3′ and IL-6 reverse, 5′-CTGTTAGGAGAGCATTGGAAATTG-3′; MCP-1 forward, 5′-TGCCCTAAGGTCTTCAGCAC-3′ and MCP-1 reverse, 5′-AAGGCATCACAGTCCGAGTC-3′; TNF-α forward, 5′-GCCACCACGCTCTTCTGTCT-3 and TNF-α reverse, 5′-GCCATAGAACTGATGA-3′. mRNA expression was determined relative to the housekeeping gene (Ppia) using the Pfaffl method (16).

Protein lysates were prepared in HES–SDS lysis buffer (250 mM sucrose, 20 mM HEPES [pH 7.4], 1 mM EDTA, and 2% [v/v] SDS) and diluted in loading dye (250 mM Tris [pH 6.8], 25% [v/v] glycerol, 10% [w/v] SDS, 5% [v/v] 2-ME, and 0.2% [w/v] bromophenol blue) prior to heating at 65°C for 5 min. Proteins (15 μg) were resolved on polyacrylamide Any kD gels (Bio-Rad Laboratories) and electrotransferred to a nitrocellulose membrane (Bio-Rad Laboratories). Protein expression was detected using rabbit anti-mouse GLUT6 (sc-134538) and mouse 14-3-3 (sc-1657) Abs from Santa Cruz Biotechnology. Primary Abs were detected with Donkey anti-mouse IgG (Alexa Fluor 790) or anti-rabbit IgG (Alexa Fluor 680) from Abcam, and membranes were scanned on the Odyssey CLx System (LI-COR).

BMDMs were seeded in Seahorse XF96 tissue culture plates at a density of 50,000 cells per well, incubated in BMDM medium (without conditioned medium) for 3 h, then polarized as indicated in figure legends. Prior to the Seahorse assay, cells were rinsed twice with PBS and then 180 μl of Seahorse medium (unbuffered Phenol red–free DMEM [pH 7.4] supplemented with 25 mM glucose, 4 mM glutamine, and 1 mM sodium pyruvate) was added to each well. The oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) were then measured in naive and polarized BMDMs using a Seahorse XF96 Flux Analyzer (Seahorse Bioscience) following treatment with oligomycin (1 μM), BAM15 (5 μM) and antimycin A/rotenone (1 μM) at the times indicated in figure legends. BAM15 was used as the mitochondrial uncoupler in this assay instead of FCCP because it does not depolarize the plasma membrane and drives maximal respiration over a broad dosing range (17). Each measurement involved a 2 min mixing step, a 2 min rest period, and a 3 min measurement of OCR and ECAR. The first three OCR and ECAR measures were averaged to give basal readings. For baseline data, data are expressed as a percentage of the third basal reading. OCR and ECAR measurements were normalized to protein content per well, where stated as determined by BCA assay (Thermo Fisher Scientific). For measurements of OCR and ECAR with acute LPS treatment, BMDMs were seeded as described above in BMDM medium (without conditioned medium). The next day, medium was removed, cells were rinsed twice with PBS, and 180 μl of Seahorse medium was added per well. OCR and ECAR were then measured as described above with vehicle or LPS (20 μl of 10× LPS, final concentration 100 ng/ml) injected at the indicated time point.

WT and GLUT6 KO BMDMs were seeded in 24-well plates at a density of 150,000 cells per well and incubated overnight in BMDM maintenance medium with conditioned medium. Media were removed, cells were rinsed with PBS, and media were replaced with BMDM maintenance medium (without conditioned medium). Three hours later cells were treated with PBS (M0), 100 ng/ml LPS (M1), or 10 ng/ml IL-4 (M2) for 24 h. Medium was removed, cells were rinsed twice with PBS, then incubated with 5 mM 2-deoxyglucose (Cayman Chemical) and 0.5 μCi/ml [³H]2-deoxyglucose (PerkinElmer) for 15 min in Krebs-Ringer phosphate buffer (pH 7.4) (120 mM NaCl, 12.5 mM HEPES [pH 7.4], 6 mM KCl, 1.2 mM MgSO₄, 1 mM CaCl₂0.2H₂O, 0.6 mM Na₂HPO₄, and 0.4 mM NaH₂PO₄0.2H₂O). Cells were rinsed twice with ice-cold PBS before permeabilization with 1% (v/v) Triton X-100 for 1 h before counting using the TriCarb Scintillation counter (Perkin Elmer).

WT and GLUT6 KO BMDMs were seeded in 24-well plates at density of 100,000 cells per well in BMDM maintenance medium with conditioned medium. The next day, media were removed and cells were rinsed with PBS and incubated with BMDM maintenance medium (without conditioned medium) for 3 h before treating with PBS or LPS at concentrations and for durations indicated in figure legends. Conditioned medium was collected at the indicated times, and secreted TNF-α levels were measured by the mouse TNF-α ELISA (PeproTech) as per manufacturer’s instructions. TNF-α levels were normalized to protein content per well as determined by BCA (Thermo Fisher Scientific).

WT and GLUT6 KO BMDMs were seeded as per the TNF-α ELISA and treated with PBS (M0) or 100 ng/ml LPS (M1) for 24 h. Medium was harvested, centrifuged at 14,000 × g for 5 min to remove cellular debris, then secreted cytokines were detected using the mouse cytokine Proteome Profiler Ab Array (R&D Systems) as per manufacturer’s instructions.

Statistical significance was determined between polarization states by two-way ANOVA using the Sidak multiple comparison test, as indicated in figure legends. Multiple t tests using the Holm–Sidak method were used without assuming a consistent SD between groups (α = 0.05) for comparison between genotypes. These comparisons were performed in GraphPad Prism version 7.02. A linear mixed model with blocks as a random effect was performed using RStudio to account for a batch effect where stated.

To identify physiological conditions where GLUT6 may be functionally important, we first searched the Functional Annotation of Mammalian Genome 5 database using Semantic catalog of Samples, Transcription initiation And Regulators (18, 19). This analysis revealed that GLUT6 was highly expressed in activated immune cells, with the highest expression in macrophages (library identification CNhs11532). Additional searches of Affymetrix data in the GEO Profiles database identified that GLUT6 was highly upregulated in eight studies of macrophages treated with LPS (Table I). Furthermore, GLUT6 contains three consensus binding sites for RelA, a transcription factor well-documented to be activated by LPS signaling.

To validate these data, we treated BMDMs from WT and GLUT6 KO mice with 100 ng/ml of LPS or 10 ng/ml of IL-4 for 12 h to promote the M1 and M2 polarization states, respectively. We also investigated the expression of other glucose transporters (GLUTs 1, 3, 4, and 8) in these cells to indicate whether they are affected by polarization and if loss of GLUT6 results in compensation by other GLUTs. GLUT8 was investigated because it shares the greatest amino acid identity with GLUT6 (8), whereas GLUTs 1 and 3 are known to be expressed in macrophages. Expression of GLUT1 mRNA significantly increased by >20-fold in both M1 WT and GLUT6 KO BMDMs, whereas LPS increased GLUT6 mRNA by ∼17-fold in WT cells (Fig. 1A, 1C). In contrast, GLUT3 and GLUT8 expression decreased with LPS and IL-4 treatment (Fig. 1B, 1D). GLUT4 was not detected in naive, M1- or M2-polarized cells of either genotypes (data not shown). GLUT6 mRNA increased by ∼2.5-fold in IL-4–treated cells, although this effect was not statistically significant (Fig. 1C). GLUT6 protein was undetectable in naive WT macrophages but was highly abundant in LPS-treated WT BMDMs (Fig. 1E). As expected, GLUT6 KO BMDMs did not express GLUT6 mRNA or protein (Fig. 1C, 1E).

To indicate whether GLUT6 has a role in macrophage metabolism, we assessed glycolytic and oxidative capacity of naive, M1, and M2 BMDMs from WT and GLUT6 KO mice using a Seahorse Extracellular Flux Analyzer. No significant difference in the OCR or ECAR were found between genotypes for M0 (Fig. 2A, 2B), M1 (Fig. 2C, 2D), or M2 (Fig. 2E, 2F) macrophages under basal conditions or when challenged with a mitochondrial stress test involving the ATP synthase inhibitor oligomycin, the mitochondrial uncoupler BAM15, and the electron transport chain I/III inhibitors rotenone and antimycin A. These data suggest that GLUT6 does not have a functional role in LPS-induced macrophage basal respiration, ATP-dependent respiration, maximal respiratory capacity, mitochondrial reserve capacity, or proton leak. As expected, M1 cells of both genotypes exhibited greater glycolytic metabolism than naive M0 cells as reflected by the higher basal ECAR and ECAR/OCR readings (Fig. 2H, 2I) and elevated glucose uptake (Fig. 2J). M2 BMDMs displayed a characteristic increase in spare respiratory capacity following treatment with the mitochondrial uncoupler BAM15 (Fig. 2E) and a trend for higher OCR (Fig. 2G) compared with M0 cells. To confirm that the M2 polarization state was reached, mRNA expression of FIZZ1 and CHI3L3 genes characteristically upregulated in M2 macrophages was quantified. As expected, mRNA levels of FIZZ1 and CHI3L3 were substantially higher than M0 BMDMs, validating that the M2 activation state was achieved (Supplemental Fig. 1A, 1B respectively). GLUT6 deletion had no effect on glucose uptake in M0, M1, or M2 macrophages (Fig. 2J). Overall, no genotype-specific differences in glucose uptake, glycolytic (ECAR), or oxidative (OCR) metabolism were detected for M0, M1, or M2 cells (Fig. 2A–J).

High concentrations of LPS might induce a plateau effect whereby GLUT1 expression dominates GLUT6; therefore, we next investigated whether GLUT6 might have a functional role at submaximal concentrations of LPS. WT and GLUT6 KO BMDMs were polarized with 0.1, 1, and 10 ng/ml of LPS for 24 h. Both 1 and 10 ng/ml of LPS but not 0.1 ng/ml of LPS increased basal ECAR in both genotypes, but no genotype-specific differences were observed for either OCR or ECAR (Supplemental Fig. 2A, 2B). Furthermore, no change in OCR or ECAR during a mitochondrial stress test was found between LPS-treated WT and GLUT6 KO cells (Supplemental Fig. 2C–J).

To determine whether GLUT6 could be involved in rapid changes in metabolism, M0 BMDMs from WT and GLUT6 KO mice were acutely stimulated with 100 ng/ml of LPS (via injection through the Seahorse XF Analyzer port). This experiment showed that LPS treatment rapidly increased ECAR but had no effect on OCR. No genotype-specific effect was observed between WT and GLUT6 KO BMDMs treated with submaximal concentrations of LPS (Supplemental Fig. 2K, 2L).

GLUT1 expression alters inflammatory cytokine production in M1-polarized macrophages (4); therefore, we investigated the role of GLUT6 in cytokine production. Expression of TNF-α and IL-6, two of the most functionally important cytokines produced by M1 macrophages, were examined in WT and GLUT6 KO BMDMs. Interestingly, basal TNF-α was decreased 2-fold in GLUT6 KO BMDMs; however, this effect was not seen following M1 polarization (Fig. 3A). The expression of IL-6 was not affected by GLUT6 deletion in M0- and M1-polarized BMDMs (Fig. 3B). To determine whether GLUT6 affects cytokine secretion, we measured TNF-α protein in conditioned medium collected from M1-polarized BMDMs. In a pilot experiment, we determined that TNF-α levels were saturating for all doses of LPS >1 ng/ml (data not shown); therefore, we measured TNF-α levels at LPS concentrations of 0.1, 0.5, and 1 ng/ml at 4, 8, 12, and 24 h (Fig. 3C–F). This experiment showed that TNF-α secretion was significantly lower in GLUT6 KO BMDMs treated with 0.5 ng/ml LPS for 12 h (Fig. 3E). However, this genotype-specific effect disappeared by 24 h, indicating that GLUT6 may have a relatively minor role in regulation of TNF-α production (Fig. 3F).

Next, we used an unbiased approach to detect changes in other cytokines using a cytokine proteome profiler array with conditioned media from WT and GLUT6 KO BMDMs treated with vehicle or 100 ng/ml LPS. This experiment showed that naive BMDMs from GLUT6 KO mice appeared to have lower MCP1 levels (Supplemental Fig. 3A), but there was no obvious change in other cytokines (e.g., IL-6 and TNF-α) from LPS-treated WT and GLUT6 KO BMDMs (Supplemental Fig. 3B). The minor change in MCP1 expression was unexpected considering that M0 macrophages do not express abundant GLUT6; nevertheless, we measured MCP1 mRNA levels in BMDMs. In line with the proteome profiler array, expression of MCP1 was decreased in M0 GLUT6 KO BMDMs but was not different following M1 polarization (Fig. 3G).

Activating the classical M1 program is an energy-intensive process that is fueled by increased glycolytic metabolism. This process is facilitated by increased expression of glucose transporters and glycolytic enzymes (4, 20). GLUT1 is important for M1 polarization, but whether other glucose transporters also play a role in this metabolic switch has not previously been assessed. Examination of public gene expression datasets showed that GLUT6 was highly upregulated to nearly the same degree as GLUT1 in LPS-polarized macrophages; therefore, we aimed to determine the role of GLUT6 in this LPS-induced macrophage polarization.

In previous studies, we have not been able to detect endogenous GLUT6 protein expression in normal mouse tissues by Western blot, including brain, spleen, or muscle, when using GLUT6 KO mouse tissues as a negative control (10). However, in this study, we detected endogenous GLUT6 protein expression in BMDMs following stimulation with LPS. The GLUT6 protein band was confirmed by a lack of band in LPS-treated GLUT6 KO cells. The only previous study to directly measure GLUT6 gene expression by PCR in macrophages showed that GLUT6 was upregulated ∼4-fold by M-CSF and GM-CSF in peritoneal macrophages compared with naive cells (21). However, that study did not investigate GLUT6 protein or function. In this study, we explored the functional role of GLUT6 in BMDMs using primary macrophages derived from GLUT6 KO mice. Overall, GLUT6 deletion did not have an effect on glucose uptake in basal or activated states and did not affect macrophage polarization or cytokine production. These data show that GLUT6 does not have a major functional role in LPS-induced macrophage function, but it does not exclude that GLUT6 may play roles in macrophages of different origin other than bone marrow, in disease states, or in other models of macrophage activation, including IFN-γ or M-CSF.

Following close examination of the inflammatory factors characteristic of M1 activation, including TNF-α and MCP1, we found that the expression of these genes decreased in M0 GLUT6 KO BMDMs. However, these genes are not normally expressed at high levels in M0 macrophages, and stimulation with LPS increased their gene expression to similar levels in GLUT6 KO BMDMs and WT cells. Therefore, it is likely that GLUT1-mediated glucose uptake is sufficient to overcome the loss of GLUT6. Alternatively, GLUT6 may have another role, apart from extracellular glucose uptake, that has not yet been identified. Previous studies have shown that GLUT6 possesses a dileucine motif that is known to limit the time that proteins spend at the plasma membrane (22). For example, GLUT6 may transport other nonglucose sugars or metabolites, or it may be located in different cellular locations to direct the transport of incoming glucose.

Overall, this study supports previous work showing that GLUT1 is the dominant glucose transporter that drives M1 polarization and glycolytic metabolism in murine BMDMs. Although no outward phenotype caused by loss of GLUT6 was observed in these cells, this study is an important addition to the literature, as GLUT6 is commonly overexpressed in cancer and represents a possible anticancer drug target. Our data suggests that a GLUT6 inhibitor targeting cancer cells would not adversely affect macrophage activation or interfere with antitumor immunity.

We thank Ellen Olzomer (School of Biotechnology and Biomolecular Sciences, UNSW) for assistance with mouse genotyping and Dr. Peter Geelan-Small (Stats Central, UNSW) for guidance with statistics.

The authors have no financial conflicts of interest.