Tissue-Resident Alveolar Macrophages Do Not Rely on Glycolysis for LPS-induced Inflammation

Abstract

Macrophage effector function is dynamic in nature and largely dependent on not only the type of immunological challenge but also the tissue-specific environment and developmental origin of a given macrophage population. Recent research has highlighted the importance of glycolytic metabolism in the regulation of effector function as a common feature associated with macrophage activation. Yet, most research has used macrophage cell lines and bone marrow–derived macrophages, which do not account for the diversity of macrophage populations and the role of tissue specificity in macrophage immunometabolism. Tissue-resident alveolar macrophages (TR-AMs) reside in an environment characterized by remarkably low glucose concentrations, making glycolysis-linked immunometabolism an inefficient and unlikely means of immune activation. In this study, we show that TR-AMs rely on oxidative phosphorylation to meet their energy demands and maintain extremely low levels of glycolysis under steady-state conditions. Unlike bone marrow–derived macrophages, TR-AMs did not experience enhanced glycolysis in response to LPS, and glycolytic inhibition had no effect on their proinflammatory cytokine production. Hypoxia-inducible factor 1α stabilization promoted glycolysis in TR-AMs and shifted energy production away from oxidative metabolism at baseline, but it was not sufficient for TR-AMs to mount further increases in glycolysis or enhance immune function in response to LPS. Importantly, we confirmed these findings in an in vivo influenza model in which infiltrating macrophages had significantly higher glycolytic and proinflammatory gene expression than TR-AMs. These findings demonstrate that glycolysis is dispensable for macrophage effector function in TR-AM and highlight the importance of macrophage tissue origin (tissue resident vs. recruited) in immunometabolism.

Glycolytic metabolism has been identified as an important driver of proinflammatory immune responses in macrophages (1–5); however, the current understanding of macrophage metabolism relies heavily on studies using macrophage cell lines (i.e., THP-1 and RAW 264.7) and bone marrow–derived macrophages (BMDMs). In vivo, macrophages exhibit marked heterogeneity based on the tissue of origin, and little is known about how the metabolic requirements of tissue-resident macrophages differ from those of BMDMs. Tissue-resident alveolar macrophages (TR-AMs) reside in the alveolar lumen, where the glucose concentration is less than 10% of blood glucose concentrations (6), suggesting that TR-AMs may use alternative metabolic programs to promote effector function. Here, we demonstrate that TR-AMs exhibit extremely low levels of glycolysis and, in contrast to BMDMs, do not undergo glycolytic reprogramming in response to LPS. Inhibition of glycolysis decreased proinflammatory cytokine production in BMDMs but had no effect on TR-AMs. TR-AMs predominantly rely on oxidative phosphorylation to meet their energy needs and are unable to upregulate glycolysis in response to electron transport chain (ETC) inhibitors, which cause cytotoxicity in these cells, but not in BMDMs. This cytotoxic effect of ETC inhibition can be overcome in TR-AMs by stabilization of HIF-1α (hypoxia-inducible factor 1α), which increases glycolytic enzyme expression and establishes glycolysis. However, HIF-1α stabilization is not sufficient for TR-AMs to mount further increases in glycolytic rate or enhance immune function in response to LPS. These findings reveal metabolic heterogeneity among macrophage populations and thus caution against making generalized conclusions regarding macrophage immunometabolism, which is increasingly being considered as a therapeutic target.

Methods

Primary Culture of BMDMs and TR-AMs

All animal experiments and procedures were performed according to protocols approved by the Institutional Animal Care and Use Committee of The University of Chicago. BMDMs were generated by isolating BM cells from the femur and tibia bones of 6- to 8-week-old C57BL/6 mice. BM cells were differentiated in RPMI 1640 (catalog number A1049101; ThermoFisher) supplemented with 10% FBS (catalog number 100-106; Gemini), 1% penicillin-streptomycin (catalog number 400-109; Gemini), and 40 ng/ml Recombinant Mouse M-CSF (catalog number 576406; BioLegend) for 1 week. On Day 7, the BMDMs were replated and allowed to adhere to tissue culture plates for 2 hours before experiments were initiated. During trypsinization, BMDMs were exposed to EDTA, which was washed away and removed. TR-AMs were isolated via standard BAL (intratracheal instillation) using PBS + 0.5 mM EDTA as we previously described (7, 8). After the TR-AMs were isolated, they were counted, plated under the same conditions as BMDMs, and allowed to adhere to tissue culture plates for 1 hour before experimentation. The media was removed after adherence of the BMDMs and AMs, and replaced with media containing 20 ng/ml LPS. For glucose-free experiments, macrophages were plated in glucose-free RPMI 1640 (catalog number 11879020; ThermoFisher) or glucose-free media supplemented with 25 mM D-glucose (catalog number BP350; ThermoFisher) for 2 hours. Cells were then stimulated with LPS under the same media conditions. Sodium oxamate (catalog number O2751; Sigma) was used at a concentration of 10 mM in experiments to inhibit glycolysis.

Bioenergetic Measurements

Glycolytic and mitochondrial respiration rates were measured using the XFe24 Extracellular Flux Analyzer (Agilent). Both TR-AMs and BMDMs were also exposed to EDTA briefly during isolation and trypsinization, respectively; however, the EDTA was washed away and removed before bioenergetic measurements were obtained. BMDMs and TR-AMs were seeded at 4.0 × 10⁴/well onto Seahorse XF24 cell culture microplates. Cells were equilibrated with XF base media (catalog number 103334-100; Agilent) at 37°C for 1 hour in the absence of CO₂. The glycolytic rate was assessed using the manufacturer’s protocol for the Seahorse XF Glycolysis Stress Test, followed by sequential injections with glucose (10 mM), oligomycin (1.0 μM), and 2-deoxy-D-glucose (2-DG; 100 mM). The mitochondrial respiration rate was measured using the Seahorse XF Mito Stress Test according to the manufacturer’s protocol, followed by sequential injections with oligomycin (1.0 μM), carbonyl cyanide-4-(trifluoromethoxy)phenylhydrazone (FCCP) (1.0 μM for BMDMs and 4.0 μM for TR-AMs), and rotenone/antimycin A (1.0 μM). Quantification of the cellular ATP production rate was done using sequential injections of oligomycin (1.5 μM) and rotenone/antimycin A (0.5 μM), and performed according to Agilent’s protocol (9). Real-time metabolic responses to LPS were assessed using the protocol detailed in an application note provided by Agilent (10). In brief, after plating, cells were equilibrated in XF base media supplemented with 10 mM glucose, 2 mM L-glutamine, 1 mM sodium pyruvate (catalog number 11360070; Sigma), and 5 mM HEPES (pH 7.4, catalog number 15630080; Sigma), and incubated at 37°C without CO₂ for 1 hour before the XF assay was performed. Baseline metabolic rates were measured, followed by direct injection of LPS. Bioenergetic rates were subsequently measured every 3 minutes for ∼6 hours.

Metabolomics

BMDMs and TR-AMs were plated at 2.75 × 10⁶ on 60-mm plates for metabolite extraction. After treatment, the cells were washed twice with a 5% mannitol solution and metabolites were extracted using 400 μl 100% methanol. Then, 275 μl of aqueous internal standard solution was mixed in with the methanol and the extract solution was transferred to a microcentrifuge. The extracts underwent centrifugation at 2,300 × g at 4°C for 5 minutes to precipitate insoluble material, and the resulting supernatant was transferred to centrifugal filter units (Human Metabolome Technologies). The supernatant was filtered at 9,100 × g at 4°C for 2 hours. The filtrate was sent to Human Metabolome Technologies and analyzed by capillary electrophoresis–mass spectrometry.

Cell Lysis and IB

Cell lysis and IB were performed as we previously described (11–13). Details of the methodology are provided in the data supplement.

Quantitative PCR

RNA was isolated from cells using the Direct-zol RNA MiniPrep kit (catalog number R2052; Zymo Research) and reverse transcribed using iScript Reverse Transcription Supermix (catalog number 1,708,841; Bio-Rad). Quantitative mRNA expression was determined by real-time quantitative PCR using iTaq Universal SYBR Green Supermix (catalog number 172-5121; Bio-Rad). Rpl19 (Ribosomal protein L19) served as a housekeeping gene, and gene expression was quantified using the ΔΔCt method to determine the relative fold change. The mouse-specific primer sequences used for quantitative PCR are included in the data supplement.

Cytokine Analysis

Secreted IL-6 and TNF-α levels were evaluated in macrophage media using a standard sandwich ELISA (catalog numbers DY406 and DY410; DuoSet ELISA Development System, R&D Systems).

Lactate Dehydrogenase Assay

Extracellular lactate dehydrogenase (LDH) release was measured using the Pierce LDH Cytotoxicity Assay Kit (catalog number 88964; ThermoFisher). Data were represented as the ratio of extracellular LDH over intracellular LDH. The no-treatment group was normalized to one and is indicative of no cellular damage. The ETC inhibitor concentrations in BMDMs were as follows: 1 μM oligomycin, 1 μM rotenone, and 1 μM antimycin A. The ETC inhibitor concentrations in TR-AMs were as follows: 50 nM oligomycin, 500 nM rotenone, and 100 nM antimycin A.

Lactate Assay

Secreted lactate was measured using the Lactate Colorimetric Assay Kit (catalog number MAK064-1KT; Sigma). Cells were plated in complete Dulbecco’s modified Eagle medium (RPMI interferes with the assay), allowed to adhere, washed with PBS, and left untreated or treated with LPS. Samples were collected at 1, 3, 6, and 24 hours, and the manufacturer’s protocol was followed to measure lactate.

Glucose Uptake

Glucose uptake was measured using the Glucose Uptake-Glo Assay (catalog number J1342; Promega). Cells were plated in glucose-free RPMI media, allowed to adhere, washed with PBS, and left untreated or treated with LPS for 3 hours. The manufacturer’s protocol was followed to measure glucose uptake.

Cell Sorting of Resident and Nonresident Macrophages

C57BL/6 mice (6–8 weeks old) were anesthetized with isoflurane and underwent retro-orbital injection with 100 μl PKH26 Red Fluorescent Cell Linker Dye for Phagocytic Cell Labeling (catalog number PKH26PCL-1KT; Millipore Sigma) 1 day before lung challenge. The mice were then challenged intratracheally with either LPS (0.7 mg/kg) or mouse-adapted influenza (A/PR8/34; 100 plaque-forming units [pfu]) for 24 and 96 hours, respectively. After challenge, the mice were killed and immune cells were collected via BAL. BAL cells were first treated with Fc Block (clone 2.4G2, catalog number 553141; BD Biosciences) and stained with fluorochrome-conjugated antibodies. The antibodies used were APC/Cy7 anti-mouse Ly-6G (Clone 1A8, catalog number 127623, 1:250; BioLegend), Alexa Fluor 647 anti-mouse F4/80 (clone BM8, catalog number 123121, 1:500; BioLegend). Immediately before sorting, cells were resuspended in sorting buffer (0.2% BSA in PBS) containing 5 nM SYTOX Green Nucleic Acid Stain (catalog number S7020; ThermoFisher) to distinguish between live and dead cells. Cell sorting was performed on a FACS Aria II instrument and data were acquired using BDFACS Diva software and analyzed with FCS Express 6 software. Cell populations were identified and selected using the gating strategy described in the Results section. Sorted cells were immediately used for downstream experiments.

Statistics

The data were analyzed in Prism 8 (GraphPad Software Inc.). All data are shown as the mean ± SD. ANOVA was used for statistical analyses of data sets containing more than two groups, and Bonferroni’s post hoc test was used to explore individual differences. Statistical significance was defined as P < 0.05.

Results

TR-AMs Have a Low Glycolytic Rate Compared with BMDMs and No Glycolytic Reserve

Because the alveolar lumen represents a unique niche with particularly low glucose concentrations, which have been reported to be less than 10% of the plasma levels (6, 14), we sought to determine whether TR-AMs phenocopy the glycolytic capabilities of BMDMs. Freshly isolated TR-AMs and terminally differentiated BMDMs were plated in equal numbers, and the extracellular acidification rate (ECAR) was measured as a surrogate for glycolytic rate using the Seahorse XFe24 Extracellular Flux Analyzer. Compared with BMDMs, TR-AMs exhibited remarkably low levels of glycolysis after injection of glucose (Figures 1A and 1B, and Figure E1 in the data supplement). Moreover, after inhibition of mitochondrial ATP synthesis with oligomycin, TR-AMs were unable to increase their glycolytic rate, consistent with having no glycolytic reserve capacity (Figures 1A and 1B). This low rate of glycolysis observed in TR-AMs was not due to reduced expression of glucose transporters (1, 15) (Figure E2).

Figure 1.

Tissue-resident alveolar macrophages (TR-AMs) have a lower level of glycolysis than bone marrow–derived macrophages (BMDMs) and have no glycolytic reserve. (A) Glycolysis stress test in TR-AMs and BMDMs using Seahorse XFe24. Glycolysis was measured as the extracellular acidification rate (ECAR). Macrophages were sequentially treated with glucose, oligomycin (ATP synthase inhibitor), and 2-deoxyglucose (2-DG) (inhibitor of hexokinase 2, or glycolysis). (B) Stacked bar graphs quantify glycolytic parameters. Data represent at least three independent experiments (n = 8 separate wells per group). Glycolytic parameters were compared across macrophage types and significance was determined by two-tailed Student’s t test (*P < 0.05). (C) Extracellular (media) lactate levels were measured in TR-AMs and BMDMs at different time points over a 24-hour period. Data represent at least three independent experiments; n = 3 per group. Significance was determined by two-way ANOVA (*P < 0.05) with Bonferroni’s post hoc test. All error bars denote mean ± SD.

Because lactate levels correlate with acid production during glycolysis (16), we measured lactate levels in media from cultures of TR-AMs and BMDMs. Supporting the finding that TR-AMs have lower ECAR, we found that extracellular (media) levels of lactate were significantly lower in TR-AMs than in BMDMs, which exhibited a progressive increase in extracellular lactate levels overtime (Figure 1C). Collectively, these results support our findings that in contrast to BMDMs, TR-AMs have very low levels of basal glycolysis and no glycolytic reserve capacity.

TR-AMs Rely on Mitochondrial Respiration and Cannot Upregulate Glycolysis in Response to Mitochondrial Inhibition

We next investigated the mitochondrial function of these macrophage populations using the mitochondrial stress test (Seahorse XFe24). Compared with BMDMs, TR-AMs displayed significantly higher oxygen consumption rate (OCR) as well as a significantly higher spare mitochondrial capacity after administration of FCCP, which uncouples the ETC from ATP synthesis (Figures 2A and 2B). Assessment of ECAR during the mitochondrial stress test confirmed the increase in glycolysis in response to oligomycin in BMDMs and lack of response in TR-AMs (Figure 2C) that we observed earlier (Figure 1A). The sharp increase in ECAR in BMDMs after oligomycin injection suggests that BMDMs can compensate for the loss of mitochondrial ATP production by upregulating glycolysis. In contrast, the ECAR in TR-AMs decreased after oligomycin injection. Furthermore, although the ECAR was maintained in BMDMs after mitochondrial inhibition with Complex I and III inhibitors (rotenone and antimycin A, respectively), mitochondrial inhibition almost completely abolished the ECAR in TR-AMs (Figure 2C). Removal of glucose did not affect the pattern of changes in the ECAR during the mitochondrial stress test in TR-AMs, further supporting our findings about their bioenergetic profile characterized by low glycolysis and reliance on mitochondrial respiration (Figure E3).

Figure 2.

TR-AMs rely on mitochondrial respiration and cannot upregulate glycolysis in response to mitochondrial inhibition. (A) Mitochondrial stress test to measure the oxygen consumption rate (OCR) using Seahorse XF24. Macrophages were treated sequentially with oligomycin (ATP synthase inhibitor), carbonyl cyanide-4-(trifluoromethoxy)phenylhydrazone (FCCP) (uncoupler), and rotenone/antimycin A (Complex I and III inhibitor, respectively). (B) Stacked bar graphs quantify mitochondrial respiration parameters. Data represent at least three experiments (n = 8 separate wells per group). Mitochondrial parameters were compared across macrophage types and significance was determined by two-tailed Student’s t test (*P < 0.05). (C) ECAR measurements were obtained during a mitochondrial stress test to visualize the inability of TR-AMs to upregulate glycolysis in response to mitochondrial inhibition. Data represent at least three independent experiments; n = 8 separate wells per group. (D) Real-time ATP rate assay in TR-AMs and BMDMs using Seahorse XF24. Macrophages were treated sequentially with oligomycin and rotenone/antimycin A to determine the rate of ATP production derived from glycolysis and mitochondrial respiration. The bar graph depicts the percentage of ATP production. Data represent at least three independent experiments (n = 10 separate wells per group). Significance was determined by two-way ANOVA (*P < 0.05) with Bonferroni’s post hoc test. (E) Macrophages were treated with mitochondrial inhibitors (oligomycin, antimycin, and rotenone) overnight (16 h) and extracellular lactate dehydrogenase (LDH) was used as a surrogate for cytotoxicity. Bar graphs represent the ratio of extracellular LDH over intracellular LDH. Data represent at least three independent experiments (n = 3 per group). Significance was determined by two-way ANOVA (*P < 0.05) with Bonferroni’s post hoc test. All error bars denote mean ± SD. Mito = mitochondrial; OXPHOS = oxidative phosphorylation; Rot/Ant = rotenone/antimycin A.

Carbonic acid is a major source of nonglycolytic (mitochondrial CO₂–derived) acid in the cell and can be detected as an increase in ECAR (17). The reduction in ECAR after administration of oligomycin and mitochondrial inhibitors (rotenone and antimycin), which diminish the tricarboxylic acid (TCA) cycle, along with the increase in ECAR after administration of FCCP, which leads to an increase in the TCA cycle and consequently increased CO₂ production, support the notion that mitochondria-derived carbonic acid is responsible for the ECAR in TR-AMs. In summary, these results suggest that ECAR measurements in TR-AMs are more closely linked to mitochondrial respiration and the generation of carbonic acid from the TCA cycle than to glycolysis-derived protons, as mitochondrial inhibition inhibited not only the OCR but also the ECAR.

The inability of TR-AMs to induce glycolysis after inhibition of mitochondrial ATP production suggested that these cells depend completely on mitochondria for cellular energy homeostasis. In agreement with this assessment, we found that ∼92% of ATP production in TR-AMs resulted from mitochondrial respiration (Seahorse XF Real-Time ATP Rate assay) (Figure 2D). Conversely, only 70% of ATP production in BMDMs was derived from mitochondria (Figure 2D). Overnight (16 h) treatment of TR-AMs with ETC inhibitors resulted in significant cytotoxicity, whereas we did not observe any cytotoxicity after treatment with ETC inhibitors in BMDMs (Figure 2E). Together, these data suggest that TR-AMs rely almost solely on mitochondrial metabolism to meet their energy requirements, and lack the metabolic plasticity observed in BMDMs. Furthermore, the lack of metabolic plasticity makes TR-AMs vulnerable to drugs or conditions that affect mitochondrial function.

TR-AMs Do Not Upregulate Glycolysis in Response to LPS

BMDMs upregulate glycolysis upon exposure to LPS and other proinflammatory stimuli to support increased energy and biosynthetic demands required for effector function (1–5). To determine whether TR-AMs undergo a similar upregulation of glycolysis in response to LPS, we monitored changes in ECAR over time in TR-AMs and BMDMs after injection of LPS (20 ng/ml). In agreement with previous reports (18), the ECAR rose in BMDMs immediately after LPS injection and the elevated ECAR was maintained over an extended period of time (Figure 3A). In contrast, TR-AMs exhibited no change in ECAR after LPS injection (Figure 3A). Higher doses of LPS (1.0 μg/ml and 100 ng/ml) led to more rapid increases in ECAR in BMDMs than what was observed with 20 ng/ml, but still had no impact on ECAR in TR-AMs (Figure E4). Glycolytic reprogramming of BMDMs was associated with increased mRNA expression of key glycolytic enzymes, including Hk2 (Hexokinase 2) and Pfkfb3 (6-Phosphofructo-2-Kinase/Fructose-2,6-Bisphosphatase 3) (Figure 3B), as was previously reported (19–21). In contrast, although a modest increase in Hk2 expression was observed in TR-AMs, Pfkfb3 expression was not induced by LPS in these cells, and other glycolytic enzymes trended toward reduced expression (Figure 3B). Similar to the differences observed in mRNA expression, Western blot analysis revealed that TR-AMs exhibited significantly lower protein expression of HK2, PFKFB3, PKM2 (pyruvate kinase M2), and LDHA both at baseline and after LPS treatment compared with BMDMs (Figure 3C).

Figure 3.

TR-AMs do not upregulate glycolysis in response to LPS. (A) ECAR was measured in TR-AMs and BMDMs after LPS injection (final concentration: 20 ng/ml) using Seahorse XF24. (B) Quantitative PCR for glycolytic enzyme gene expression at 6 hours and 24 hours. Gene expression is represented as the fold change, determined using the ∆∆Ct method normalizing to unstimulated controls (represented by dotted line). Data represent at least three independent experiments (n = 3 per group). Significance was determined by two-way ANOVA (*P < 0.05) with Bonferroni’s post hoc test. (C) Western blot analysis for key regulatory enzymes in glycolysis. (D) Metabolomic heatmap for glycolytic intermediates after the LPS time course (n = 4 per group). Data are represented as relative normalized expression derived from the z-score. (E) Intracellular lactate levels in TR-AMs and BMDMs at 0, 6, and 24 hours after LPS treatment. Significance was determined by two-way ANOVA (*P < 0.05) with Bonferroni’s post hoc test. (F) Glucose uptake in TR-AMs and BMDMs after 3 hours of LPS stimulation. Significance was determined by two-way ANOVA (*P < 0.05) with Bonferroni’s post hoc test. All error bars denote mean ± SD. RLU = relative light units.

To complement bioenergetic and glycolytic enzyme gene and protein expression, we also performed targeted quantitative metabolomics in TR-AMs and BMDMs both at baseline and after LPS treatment. Compared with BMDMs, all detectable glycolytic metabolites were significantly lower at baseline in TR-AMs (Figure 3D). TR-AMs exhibited small changes in glucose-6-phosphate and fructose-6-phosphate after LPS treatment, although BMDMs showed significant decreases in these metabolites. Fructose-1,6-bisphosphate, 3-phosphoglycerate, and phosphoenolpyruvate significantly increased in both TR-AMs and BMDMs after LPS treatment. The glycolytic metabolite profile in BMDMs corresponds with PFKFB3 activation at 6 hours after LPS treatment. PFKFB3 catalyzes the interconversion of fructose-6-phosphate and fructose 2,6-bisphosphate (F2,6BP). F2,6BP serves as an allosteric activator of phosphofructokinase-1, which catalyzes the irreversible commitment to glycolytic processes downstream of fructose-1,6-bisphosphate (22).

Consistent with the LPS-induced ECAR, intracellular lactate levels increased significantly in BMDMs compared with TR-AMs after LPS treatment (Figure 3E). In agreement with previous reports (1), glucose uptake was significantly increased in BMDMs treated with LPS (Figure 3F). In contrast, TR-AMs did not have a significant increase in glucose uptake when treated with LPS, and maintained substantially lower levels of glucose uptake than the BMDMs both at baseline and under LPS stimulation (Figure 3F). Collectively, these data demonstrate that LPS does not induce glycolytic reprogramming in TR-AMs.

Inhibition of Glycolysis Attenuates Proinflammatory Cytokine Production in BMDMs, but Not in TR-AMs

Because we observed contrasting metabolic phenotypes between BMDMs and TR-AMs, we next asked whether the inhibition of glycolysis had differential effects on proinflammatory cytokine production in BMDMs and TR-AMs in response to LPS. To inhibit glycolysis, we used glucose-free media or the LDHA inhibitor sodium oxamate, which inhibits LDH-mediated oxidation of reduced form of nicotinamide adenine dinucleotide (NADH), thus inhibiting the NAD⁺-dependent GAPDH reaction (23, 24). We chose oxamate instead of 2-DG to inhibit glycolysis because 2-DG is also known to significantly suppress mitochondrial function and induce endoplasmic reticulum stress (25–27). Confirming these off-target effects reported in the literature, our data also show that 2-DG led to substantial decreases in TR-AM procaspase 3 expression, with concomitant increases in cleaved caspase 3 (Figure E5), which signifies a commitment to apoptotic processes (28).

Consistent with increased glycolytic lactate production after LPS, oxamate inhibited the LPS-induced increase in ECAR in BMDMs (Figure 4A). In contrast, oxamate had no impact on ECAR in TR-AMs (Figure 4B). A glycolytic stress test verified that reductions in ECAR were directly linked to oxamate-mediated inhibition of glycolysis (Figures E6A–E6C). In contrast to the cytotoxic effects of inhibitors of mitochondrial respiration, we did not observe any cytotoxicity in TR-AMs or BMDMs cultured overnight with oxamate, or in glucose-free media (Figure E6D).

Figure 4.

Inhibition of glycolysis impairs proinflammatory cytokine production in BMDMs but not in TR-AMs. (A and B) ECAR measurements over time in BMDMs (A) and TR-AMs (B) using Seahorse XF24 to demonstrate that oxamate can inhibit LPS-induced glycolysis. The first arrow indicates oxamate injection (final concentration: 10 mM) and the second arrow indicates LPS injection (final concentration: 20 ng/ml). Data represent at least three independent experiments; n = 10 separate wells per group. Significance was determined by two-way ANOVA (*P < 0.05) with Bonferroni’s post hoc test. (C) BMDMs and (D) TR-AMs were pretreated for 1 hour (with or without glucose) and then stimulated with LPS (20 ng/ml) for 6 and 24 hours while maintaining pretreatment conditions. (E) BMDMs and (F) TR-AMs were pretreated for 1 hour (with or without 10 mM of oxamate) and then stimulated with LPS (20 ng/ml) for 6 and 24 hours while maintaining pretreatment conditions. (C–F) ELISA was used to measure secreted cytokine (IL-6 and TNF-α) in BMDMs (C and E) and TR-AMs (D and F). Secreted cytokines were not detectable under no LPS stimulation. Data represent at least three independent experiments; n = 3 per group. Significance was determined by two-way ANOVA (*P < 0.05) with Bonferroni’s post hoc test. All error bars denote mean ± SD.

To determine how inhibition of glycolysis affects macrophage effector function, BMDMs and TR-AMs were cultured in either glucose-free media or treated with oxamate and then subsequently treated with LPS. When BMDMs, which upregulate glycolysis in response to LPS, were cultured in glucose-free media or treated with oxamate, LPS-induced IL-6 and TNF-α secretion was significantly decreased (Figures 4C and 4E). In contrast, in TR-AMs, IL-6 and TNF-α secretion was unperturbed in the absence of glucose or treatment with oxamate (Figures 4D and 4F). In accordance with cytokine protein expression, proinflammatory gene expression was unaffected in TR-AMs in the absence of glucose or treatment with oxamate (Figures E7A and E7C), whereas Il6 and Il1b gene expression was considerably reduced in BMDMs cultured in glucose-free media or treated with oxamate (Figures E7B and E7D). Expression of Tnfa, Hk2, and Pfkfb3 was not affected under glucose-free conditions or in the presence of oxamate (Figure E7). However, PFKFB3 protein expression was reduced in BMDMs cultured in glucose-free media or oxamate (Figure E7E). Taken together, these data suggest that glycolysis is required for LPS-induced inflammatory response in BMDMs, but TR-AMs do not rely on glycolysis to generate an efficient inflammatory response and are consequently insensitive to the lack of glucose or LDH inhibition.

HIF-1α Activation Alters TR-AM Metabolism

HIF-1α is a master transcriptional regulator that is activated in response to low oxygen levels. As a result, HIF-1α induces transcription of genes related to angiogenesis, cell survival, and glycolysis to compensate for hypoxia (29). Interestingly, under normoxic conditions, myeloid cells are highly reliant on HIF-1α for efficient immune activation and infiltration (30). Recent studies have shown that terminal maturation of TR-AMs is dependent on the expression of Von Hippel-Lindau tumor suppressor (VHL), the ubiquitin ligase that is required for degradation of HIF-1α (31, 32). Thus, we asked whether stabilization of HIF-1α could alter TR-AM metabolism to promote glycolysis.

We assessed glycolytic enzyme expression in TR-AMs after 16 hours of treatment with dimethyloxalylglycine (DMOG), an analog of α-ketoglutarate that inhibits degradation of HIF-1α. DMOG induced robust HIF-1α accumulation within the nucleus of TR-AMs (Figure 5A) and led to increased protein (Figure 5B) and mRNA (Figure 5C) expression of HIF-1α target genes that are key regulators of glycolysis. We then assessed the effect of DMOG on glycolytic and mitochondrial metabolism in TR-AMs. Consistent with increased HIF-1α activity, DMOG-treated TR-AMs exhibited marked increases in glycolysis and glycolytic reserve capacity compared with controls under baseline conditions (Figure 5D). DMOG also led to significant reductions in mitochondrial oxidative phosphorylation and spare mitochondrial capacity (Figure 5E). TR-AMs treated with DMOG were able to upregulate glycolysis in response to ETC inhibition (Figure 5F), and glycolysis-derived ATP production increased from 5% in controls to 50% in the DMOG-treated group (Figure 5G). Most importantly, DMOG treatment could rescue TR-AM cytotoxicity induced by ETC inhibitors (Figure 5H). These results suggest that HIF-1α stabilization is sufficient to promote glycolytic metabolism in TR-AMs.

Figure 5.

Activation of HIF-1α (hypoxia-inducible factor 1α) alters metabolism by inducing glycolysis in TR-AMs. TR-AMs were treated with dimethyloxalylglycine (DMOG; 1 mM) overnight (16 h) to activate HIF-1α. (A) Western blot analysis of nuclear extracts (NE) and cytosolic extracts (CE) to verify HIF-1α activation. (B) Protein (Western blot) and (C) mRNA (quantitative PCR) expression of glycolytic enzymes in TR-AMs treated with DMOG. Gene expression is represented as fold change using the ∆∆Ct method. Data represent at least three independent experiments; n = 3 per group. Significance was determined by two-way ANOVA (*P < 0.05) with Bonferroni’s post hoc test. (D) A glycolytic stress test was performed to measure DMOG’s effect on glycolysis. (E) A mitochondrial stress test was performed to measure DMOG’s effect on mitochondrial function. In D and E, stacked bar graphs quantify glycolytic and mitochondrial parameters, respectively. Data represent at least three independent experiments (n = 8 separate wells per group). Glycolytic and mitochondrial parameters were compared across treatment groups and significance was determined by two-tailed Student’s t test; *P < 0.05. (F) ECAR measurements during mitochondrial stress test showed that TR-AMs were able to increase glycolysis after mitochondrial inhibition. Data represent at least three independent experiments; n = 8 separate wells per group. (G) Rate of ATP synthesis in TR-AMs after DMOG treatment (Seahorse XF24, real-time ATP rate assay). Data represent at least three independent experiments (n = 8 separate wells per group). Significance was determined by two-way ANOVA (*P < 0.05) with Bonferroni’s post hoc test. (H) DMOG-treated TR-AMs were exposed to mitochondrial inhibitors overnight (16 h) and extracellular LDH was used as a surrogate for cytotoxicity. Bar graphs represent the ratio of extracellular LDH over intracellular LDH. Data represent at least three independent experiments (n = 3 per group). Significance was determined by two-way ANOVA (*P < 0.05) with Bonferroni’s post hoc test. (I) ECAR measurements in DMOG-treated TR-AMs after LPS injection (final concentration: 20 ng/ml) using Seahorse XF24. (J) Quantitative PCR was used to measure cytokine mRNA expression. (K) ELISA was used to measure secreted cytokines. All error bars denote mean ± SD. Significance was determined by two-tailed Student’s t test; *P < 0.05.

Given that DMOG treatment increased glycolytic enzyme expression in TR-AMs, enabling them to perform glycolysis similarly to BMDMs, we next evaluated whether DMOG treatment enhanced the responsiveness of TR-AMs to LPS. Despite these changes in baseline glycolysis, DMOG treatment did not result in a further increase of ECAR after LPS injection (Figure 5I). Furthermore, DMOG-treated TR-AMs did not exhibit enhanced cytokine mRNA expression in response to LPS (Figure 5J); however, dramatic reductions in TNF-α and IL-6 cytokine secretion were observed (Figure 5K). These data indicate that HIF-1α stabilization can promote glycolysis in TR-AMs and prevent ETC-induced cytotoxicity at baseline without any stimuli, but it is insufficient for TR-AMs to mount further increases in glycolytic rate or enhance their effector function in response to LPS.

Infiltrating Macrophages Metabolically Resemble BMDMs during Acute Lung Injury

To ascertain whether our in vitro studies focusing on BMDMs are relevant for in vivo models of inflammation, we sought to determine the metabolic phenotype of TR-AMs and monocyte-derived macrophages recruited during acute lung injury. Although LPS is a commonly used immunostimulus for in vitro macrophage activation studies, intratracheal LPS instillation yields airway inflammation that is predominantly neutrophilic in cellular composition (33). We found that intratracheal LPS instillation failed to elicit a significant infiltrating macrophage response (Figure E8), which prevented us from studying the metabolism of BMDMs/monocyte-derived macrophages in vivo. Thus, we elected to use intratracheal influenza administration as an alternative to assess the metabolic phenotype of infiltrating macrophages (monocyte-derived macrophages) recruited to lungs. Influenza is a clinically relevant model for acute lung injury and results in significant macrophage recruitment into the airway (34, 35), allowing for the collection of adequate cell numbers for the metabolic analyses of resident (TR-AM) and nonresident macrophage populations (BMDMs/monocyte-derived macrophages).

To evaluate the cellular metabolism of macrophage subpopulations in the airway during inflammation, mice were injected with PKH26 Red Fluorescent Cell Linker dye (Sigma-Aldrich) 1 day before lung challenge. The PKH26 dye labels the lipid membrane of resident phagocytic cells (TR-AMs) but will not label the BM cells from which infiltrating macrophage populations arise (36, 37). This allows for the identification of resident (PKH26⁺) and nonresident (PKH26⁻) macrophages via flow cytometry based on the presence or absence of PKH26 fluorescence (Phycoerythrin [PE] fluorochrome filter). PKH26-treated mice were challenged with 100 pfu of influenza virus (A/PR8/34). Four days later, BAL cells were collected, stained with antibodies, and flow sorted. We devised a sorting strategy that excluded neutrophils (Ly6 g⁺) and selected for macrophages (F4/80⁺). Macrophages were then separated into resident (PKH26⁺) and nonresident (PKH26⁻) populations (Figure 6A). Siglec5, a TR-AM–specific marker, was used to validate the flow results, and was significantly higher in PKH26⁺ macrophages (Figure 6B).

Figure 6.

Infiltrating macrophages metabolically resemble BMDMs during acute lung injury. (A) FACS plots of BAL samples collected from C57BL/6 mice treated with PR8 (100 plaque-forming units; 4 dpi). Samples were first gated on single cells based on the side scatter (SSC)/forward scatter signal, and then live cells were selected (SYTOX Green⁻). Ly6G⁻ F4/80⁺ gating was used to exclude neutrophils and select for macrophages. Resident macrophages were identified as being PKH26⁺, and nonresident/infiltrating macrophages were PKH26⁻. (B) Expression of siglec5 in flow-sorted cells. (C) Mitochondrial stress test to measure the OCR using Seahorse XF24. Flow-sorted macrophages were treated sequentially with oligomycin (ATP synthase inhibitor), FCCP (uncoupler), and rotenone/antimycin A (inhibitors of Complexes I and III, respectively). (D) Stacked bar graphs quantify mitochondrial respiration parameters. Data represent at least three experiments (n = 8 separate wells per group). Mitochondrial parameters were compared across macrophage types and significance was determined by two-tailed Student’s t test; *P < 0.05. (E) ECAR measurements during a mitochondrial stress test showed that PKH26⁻ cells increased glycolysis after mitochondrial inhibition. (F) Graphical representation of ECAR reading after oligomycin injection. Data are represented as the percent change in baseline ECAR (measurement #3) from the ECAR after oligomycin injection (measurement #6). Significance was determined by two-tailed Student’s t test; *P < 0.05. (G and H) Expression of (G) glycolytic genes and (H) proinflammatory genes. Gene expression is represented as the fold change using the ∆∆Ct method. Data represent samples from three independent experiments; n = 4–5 per group. Significance was determined by two-tailed Student’s t test. *P < 0.05 for gene expression analysis. All error bars denote mean ± SD.

To assess how metabolic phenotypes may differ between TR-AMs (PKH26⁺) and infiltrating macrophages (PKH26⁻), freshly sorted cells were subjected to mitochondrial stress tests. Consistent with TR-AMs under steady-state conditions, PKH26⁺ resident macrophages from influenza-infected mice exhibited higher basal OCR as well as higher ATP-coupled respiration and spare mitochondrial capacity compared with PKH26⁻ nonresident macrophages (Figures 6C and 6D). Moreover, PKH26⁺ macrophages failed to compensate for loss of mitochondrial ATP production by upregulating glycolysis (ECAR) in response to oligomycin (Figures 6E and 6F). In contrast, PKH26⁻ macrophages exhibited a 40% increase in ECAR from baseline in response to oligomycin, suggesting that these cells more closely resemble the metabolism of in vitro–differentiated BMDMs. After rotenone/antimycin injection, the PKH26⁺ macrophage ECAR dropped substantially compared with that was observed in PKH26⁻ macrophages, signifying that PKH26⁺ macrophage acidification is predominantly tied to mitochondrial CO₂ production (Figure 6E). PKH26⁻ macrophages had higher glycolytic gene expression than PKH26⁺ macrophages, consistent with the differences observed between TR-AMs and BMDMs (Figure 6G). Moreover, PKH26⁻ macrophages had significantly higher proinflammatory gene expression (Figure 6H). These data suggest that tissue-resident and infiltrating macrophage populations exhibit differences in metabolism during lung challenge similar to those observed in BMDMs and TR-AMs during in vitro experiments.

Discussion

Here, we show that TR-AMs have a distinct metabolic phenotype compared with BMDMs. Under steady-state conditions, TR-AMs have very low glycolytic capability and fail to upregulate glycolysis when treated with ETC inhibitors, which lead to cytotoxicity. We also show that, unlike BMDMs, TR-AMs do not exhibit classical glycolytic reprogramming in response to LPS, and that glycolytic inhibition has no effect on their ability to produce inflammatory cytokines. In vivo influenza studies demonstrated that nonresident, infiltrating monocyte-derived macrophages more closely resembled BMDMs with regard to metabolic phenotype, supporting the relevance of our in vitro experiments to in vivo models of acute lung injury.

Assessment of ECAR measurements during a mitochondrial stress test, particularly after mitochondrial inhibition, can provide insight into whether extracellular acidification arises from mitochondria-derived CO₂ (carbonic acid) or glycolytic flux. This is critical for an accurate interpretation of these data, to ensure that cells with unique metabolic phenotypes, such as TR-AMs, are not mischaracterized. During a mitochondrial stress test, ECAR increases after oligomycin injection due to enhanced glycolysis to compensate for the reduction in ATP synthesis (38). Although BMDMs exhibited the expected increase in ECAR, TR-AMs showed decreased ECAR after oligomycin injection. The reduction in ECAR after oligomycin injection suggests that the ECAR is mitochondrially derived and the cells may be exhibiting a glycolytic deficiency. Moreover, BMDMs maintained high ECAR after exposure to the mitochondrial complex inhibitors rotenone and antimycin, but the ECAR in TR-AMs decreased even below their baseline value. These results strongly suggest that mitochondria-derived CO₂ contributes to acidification in TR-AMs (38).

To our knowledge, no other cell type has been identified to have such limited glycolytic capabilities. Long-lived plasma cells import glucose primarily for glycosylation of antibodies, but will divert glucose to form pyruvate under metabolic stress (39). Sekine and colleagues found that glucose use and lactate production in pancreatic β-cells were significantly blunted in the presence of mitochondrial inhibitors (40). Like TR-AMs, pancreatic β-cells have low LDH expression and activity. The authors proposed that low LDH expression ensures that the glycolytic flux to pyruvate is tightly coupled to mitochondrial respiration. This could be the case with our observed TR-AM phenotype; however, the limited availability of glucose in the alveolar space makes it unlikely that TR-AMs rely on glucose to fuel mitochondrial respiration. Moreover, we observed no increase in oxygen consumption upon glucose injection in TR-AMs. Although more work needs to be conducted, the low-glucose and high-oxygen environment of the alveolar lumen may be the most plausible explanation for the unique metabolic phenotype observed in TR-AMs.

In agreement with our findings, Xie and colleagues found that unsorted AMs from saline-treated mice exhibited limited glycolytic capabilities compared with unsorted AMs from bleomycin-treated mice (41). AMs from bleomycin-treated mice had enhanced glycolytic gene and protein expression as well as enhanced maximal glycolysis. It is likely that the observed changes in AM glycolytic function were due to recruited macrophages that remained in the respiratory tree after bleomycin-induced lung injury. Based on our in vivo influenza data, we would expect the findings of Xie and colleagues to mirror ours if they were to segregate cell populations within the lungs based on the tissue of origin.

HIF-1α stabilization with DMOG was sufficient to enhance glycolytic function and enzyme expression in TR-AMs, but ultimately failed to induce glycolytic reprogramming in response to LPS or enhance inflammation. The reduced cytokine expression and secretion observed in DMOG-treated TR-AMs could be linked to depressed mitochondrial function. Our data show that TR-AMs are highly reliant on mitochondria to efficiently maintain cellular processes, and that DMOG greatly suppresses OCR in favor of enhanced glycolysis. Others have reported that DMOG directly inhibits mitochondrial function, and that this process precedes and does not depend on HIF-1α signaling (42). Moreover, DMOG treatment can interfere with other reactions that require dioxygenases, such as Ten-eleven Translocation enzymes, and therefore can result in changes in DNA methylation that can impact cellular function outside of HIF-1α signaling (43). Thus, impaired immune effector function could be linked to depressed mitochondrial function or epigenetic alterations. Future experiments will be required to determine whether HIF stabilization is required for the effects of DMOG on cytokine expression in TR-AMs; however, our findings are consistent with a recent study that showed that oxygen sensing leading to loss of HIF through VHL expression is required for terminal maturation of TR-AMs (31).

Unique macrophage subsets exist within different organ systems, yet little is known regarding how distinct environments shape macrophage metabolism and effector function. Unlike circulating monocytes that arise from the BM, TR-AMs originate from fetal yolk tissue, are long-lived, and have self-renewing capabilities (44–46). Interestingly, monocytes that infiltrate the alveolar space (i.e., recruited infiltrating monocyte-derived macrophages or BMDMs) have been shown to retain monocytic characteristics, which have been shown to play a role in the pathogenesis of chronic lung diseases (45, 47). Infiltrating macrophages derived from monocytes have been implicated not only in the pathogenesis of lung disorders, such as acute lung injury and pulmonary fibrosis, but also in models of multiple sclerosis and inflammatory bowel disease (48–50). In many of these disease states, infiltrating monocytes begin to acquire tissue-resident characteristics over time, suggesting that early blockade of these cells may be most beneficial in attenuating disease. These observations demonstrate that both developmental and environmental cues are critical in defining macrophage characteristics. Thus, to accurately link macrophage metabolism to effector function, future studies should account for tissue specificity. This is of major significance when considering targeting immunometabolism for therapeutic purposes, as decisions regarding treatment will need to take into account metabolic differences between populations of macrophages. It is also of note that many environmental pollutants target the ETC and inhibit mitochondrial function. The mitochondrial dependency of TR-AMs may make these cells vulnerable to environmental exposures and contribute to the exposure-related, complex biological responses.