Mitochondria complex III–generated superoxide is essential for IL-10 secretion in macrophages

Abstract

Mitochondrial electron transport chain (ETC) function modulates macrophage biology; however, mechanisms underlying mitochondria ETC control of macrophage immune responses are not fully understood. Here, we report that mutant mice with mitochondria ETC complex III (CIII)–deficient macrophages exhibit increased susceptibility to influenza A virus (IAV) and LPS-induced endotoxic shock. Cultured bone marrow–derived macrophages (BMDMs) isolated from these mitochondria CIII–deficient mice released less IL-10 than controls following TLR3 or TLR4 stimulation. Unexpectedly, restoring mitochondrial respiration without generating superoxide using alternative oxidase (AOX) was not sufficient to reverse LPS-induced endotoxic shock susceptibility or restore IL-10 release. However, activation of protein kinase A (PKA) rescued IL-10 release in mitochondria CIII-deficient BMDMs following LPS stimulation. In addition, mitochondria CIII deficiency did not affect BMDM responses to interleukin-4 (IL-4) stimulation. Thus, our results highlight the essential role of mitochondria CIII–generated superoxide in the release of anti-inflammatory IL-10 in response to TLR stimulation.

Mitochondrial ROS is necessary for the release of a critical anti-inflammatory cytokine.

INTRODUCTION

Macrophages play a crucial role in response to infection or tissue injury (1). Upon activation, these cells undergo significant metabolic changes to sustain their immune functions effectively. In their naïve state, monocytes and macrophages use mitochondrial respiration to generate adenosine 5′-triphosphate (ATP) through oxidative phosphorylation (OXPHOS) (2, 3). Monocyte/macrophage stimulation by various stimuli such as damage- or pathogen-associated molecular patterns (DAMPs or PAMPs) as well as cytokines induces an increase in glycolysis and its subsidiary pathways including the pentose phosphate pathway and the de novo serine synthesis pathway to maintain anabolic functions including synthesis of glutathione and a reduced form of nicotinamide adenine dinucleotide phosphate as well as cytokine production (3–5). In addition, mitochondria function as signaling hubs releasing reactive oxygen species (ROS), metabolites, and mitochondrial DNA or RNA to induce cytokine production and enhance both efferocytosis and phagocytosis (6–9).

Much of our understanding about the metabolism of macrophages has come from in vitro polarization studies using responses to lipopolysaccharide (LPS) or responses elicited by T helper 2 cytokines interleukin-4 (IL-4) or IL-13 (3). Historically, LPS- and IL-4–stimulated macrophage functions are linked to glycolysis and mitochondrial metabolism, respectively (10). Genetic manipulations of glycolytic enzymes (11) and mitochondrial electron transport chain (ETC) complexes (12, 13) indicate that both metabolic processes control macrophage production of cytokines in response to LPS. However, genetic in vivo evidence linking ETC function to macrophage function in inflammatory responses and the underlying mechanisms by which ETC controls macrophage cytokine production are not fully deciphered.

The ETC uses oxygen for respiration, a process intimately connected to ATP production through OXPHOS. This respiratory activity supports the tricarboxylic acid (TCA) cycle, crucial for creating metabolites needed for macromolecule synthesis. It also influences various biological functions in macrophages, including type I interferons (14, 15), epigenetic modifications, and hypoxia-inducible factors (16). A by-product of respiration is the formation of superoxide anions (O₂^·−) through reverse electron transport (RET) at mitochondrial complex I (CI) and forward electron transport at CIII (17). O₂^·− is then transformed into hydrogen peroxide (H₂O₂), which acts as a signaling molecule.

In the mitochondria, ETC CI and CII (CI and CII) transfer electrons to ubiquinone, converting it into ubiquinol. Subsequently, mitochondria CIII oxidizes ubiquinol (CoQH2) to ubiquinone (CoQ), simultaneously reducing cytochrome c (18). An increase in the CoQH2/CoQ ratio is a critical factor for increased RET and superoxide production at CI. One method to decrease the CoQH2/CoQ ratio involves the use of Ciona intestinalis alternative oxidase (AOX), which efficiently oxidizes CoQH2 back to CoQ (19). Recent studies in macrophages have shown that the absence of mitochondria CIII, combined with AOX expression, effectively reduces O₂^·− production by both mitochondria CI and CIII (12). AOX expression has been used as a research tool in various cell types to decrease mitochondrial superoxide production (20, 21). In the present study, we used AOX expression in conjunction with mitochondria CIII deficiency in macrophages to examine the necessity of ETC-generated O₂^·− production in both in vitro and in vivo responses to influenza A virus infection as well as LPS and IL-4 stimulation.

RESULTS

Macrophage mitochondrial CIII function is required for recovery following influenza A virus infection

To analyze the role mitochondria CIII plays in monocyte/macrophage responses to infection, we crossed mice with floxed alleles of the mitochondria CIII subunit Uqcrq (QPC) (22) with CX3CR1^Cre-ert2-yfp mice (JAX 021160) (23) and tdTomato reporter mice Rosa26^Ai14/Ai14 (JAX 007914). Beginning 3 weeks before intratracheal infection with 15 plaque-forming units (PFU) influenza A virus (IAV) [A/WSN/H1N1(1933)], mice were fed tamoxifen chow to generate monocyte-/macrophage-specific mutants, i.e., QPC–knockout (KO) (CX3CR1^Cre-ERT2/WT; QPC^Fl/Fl; Rosa26^Ai14/Ai14), QPC-Het (CX3CR1^Cre-ERT2/WT; QPC^WT/Fl; Rosa26^Ai14/Ai14), and QPC–wild type (WT) (CX3CR1^Cre-ERT2/WT; QPC^WT/WT; Rosa26^Ai14/Ai14). In this model, immunopathology is in substantial part mediated by monocyte-derived alveolar macrophages (MoAMs) (24, 25). QPC-KO mice showed a significant decrease in survival and notable delay in weight recovery (Fig. 1, A and C) compared to QPC-WT and QPC-HET mice. As we saw no difference in response to influenza virus infection in QPC-WT and QPC-HET mice, QPC-HET mice were used for all subsequent comparisons in this study. Examination of whether mitochondrial CIII function in monocyte/macrophage populations is required for control of viral load revealed minimal differences in (PFU) per lung in QPC-HET and QPC-KO mice (Fig. 1B and fig. S1, C and D). This was consistent with similar kinetics in initial weight loss after influenza A virus infection (Fig. 1C). We also confirmed that classical monocytes circulating in the blood (Fig. 1D and fig. S1A) and residing in the lung (Fig. 1E and fig. S1B) as well as MoAMs (Fig. 1F) from QPC-HET and QPC-KO mice 7 days post-injection (DPI) have similar tomato⁺ frequency, indicating that recruited monocytes/macrophages continue to have active Cre recombination and that loss of QPC functionality does not affect monocyte transition to MoAMs in the lung. Frequencies and numbers of classical monocytes in the blood (fig. S1E and Fig. 1G) and lung (fig. S1F and Fig. 1H) as well as MoAMs (fig. S1G and Fig 1I) were also similar in QPC-HET and QPC-KO animals. Collectively, our data indicate that mitochondria CIII in different monocyte/macrophage populations is dispensable for control of influenza A virus infection as well as classical monocyte recruitment and development of MoAMs in the lung but is required for recovery and survival from influenza A virus infection.

Fig. 1. Macrophage mitochondrial CIII function is required for recovery following influenza A virus infection.

(A) Survival curves following intratracheal instillation of 15 PFU IAV [A/WSN/H1N1/(1933)] into mice expressing CX₃CR1^{Cre-ERT2-YFP/WT} while on tamoxifen chow. (B) IAV PFU from lung homogenates taken at 7 DPI. QPC-HET (n = 7) and QPC-KO (n = 6). (C) Weight was tracked for 23 days post-instillation in mice QPC-WT (QPC^WT/WT) [n = 6 (M = 3, F = 3)], QPC-HET (QPC^WT/Fl) [n = 14 (M = 4, F = 10)], and QPC- KO (QPC^Fl/Fl) [n = 14 (M = 9, F = 5)]. Cross indicates dates of mouse death events. Flow cytometric analysis 7 DPI of % Tomato⁺ of (D) classical monocytes in blood (E) classical monocytes in the lungs and (F) lung MoAMs. Seven DPI numerical calculations based on flow and cell counts for (G) classical monocytes in blood (H) classical monocytes in the lungs and (I) lung MoAMs. For (A) Kaplan-Meier tests with Mantle-Cox log rank and Gehan-Breslow-Wilcoxon test, adding extra weight for early time points was used. In (C), unpaired t tests were used. Data represent means ± SEM.

Mitochondrial CIII function is required for release of IL-10 protein in macrophages in response to LPS

Next, we examined whether mitochondria CIII is required for survival in an animal model of sub-lethal endotoxic shock induced by LPS. Myeloid-specific, constitutive QPC-KO (LysM^Cre/Cre; QPC^Fl/Fl; Rosa26^Ai14/Ai14) mice were significantly more susceptible to endotoxemia than QPC-HET (LysM^Cre/Cre; QPC^WT/Fl; Rosa26^Ai14/Ai14) controls (Fig. 2A). QPC-KO mice showed decreased levels of circulating plasma IL-10 (Fig. 2B) and increased plasma levels of tumor necrosis factor–α (TNF-α) (fig. S2A) 2 hours post-LPS administration. As IL-10 from LysM-expressing cells is required for controlling the inflammatory response to LPS-induced endotoxemia (26–29), we next examined whether mitochondrial CIII was required for IL-10 production in bone marrow–derived macrophages (BMDMs) following pattern recognition receptor stimulation. We confirmed that QPC-KO (LysM^Cre/Cre; QPC^Fl/Fl; Rosa26^Ai14/Ai14) BMDMs showed significantly decreased levels of basal respiration (Fig. 2C) and mitochondrial H₂O₂ compared to QPC-HET (LysM^Cre/Cre; QPC^WT/Fl; Rosa26^Ai14/Ai14) controls (Fig. 2D). Metabolomic labeling using U-¹³C₆-Glucose showed significant increase in lactate labeling 2 hours after glucose addition in untreated (Fig. 2E) and LPS-stimulated (Fig. 2F) BMDMs. After 2 hours of LPS stimulation, QPC-KO BMDMs released similar levels of TNF-α (fig. S2B) but showed diminished IL-10 release (Fig. 2G). Furthermore, stimulation with polyinosinic:polycytidylic acid [poly (I:C)], a double-stranded RNA mimetic and TLR3 agonist, for 2 hours in vitro induced similar levels of interferon-β (IFN-β) and TNF-α (fig. S2, C and D) release from BMDMs; however, IL-10 release in QPC-KO BMDMs (Fig. 2H) was significantly decreased compared to QPC-HET controls. Together, these data indicate that macrophage mitochondria CIII is required for the initial IL-10 response to PRR stimulation with PAMPs in vitro and in vivo.

Fig. 2. Mitochondrial CIII function is required for LPS induction of IL-10 protein in macrophages.

(A) Survival curve of mice following intraperitoneal injection of 30 mg/kg to LPS QPC-HET (LysM^Cre/CreQPC^WT/FlRosa26^Ai14/Ai14) [n = 22 (M = 13, F = 9)], QPC-KO (LysM^Cre/CreQPC^Fl/FlRosa26^Ai14/Ai14) [n = 16 (M = 8 F = 8)]. (B) Plasma IL-10 levels 2 hours post-LPS injection in QPC-HET (n = 6) and QPC-KO (n = 6). BMDMs generated from QPC-HET (LysM^Cre/Cre QPC^WT/Fl Rosa26^Ai14/Ai14) or QPC-KO (LysM^Cre/CreQPC^Fl/FlRosa26^Ai14/Ai14) mice were examined via seahorse for (C) Basal respiration (n = 4 for each group) and (D) Amplex red assay for mitochondrial ROS production (n = 6 for each group). BMDMs treated with heavy-labeled U-¹³C-Glucose for 0, 1, or 2 hours to examine lactate labeling in (E) untreated and (F) LPS-stimulated BMDMs QPC-HET (100 ng/ml; n = 5) and QPC-KO (n = 3). IL-10 secretion from QPC-HET (n = 4) and QPC-KO (n = 4) BMDMs following 2-hour stimulation with (G) LPS (100 ng/ml) or (H) poly (I:C) (20 μg/ml). For data in (A) Kaplan-Meier tests with Mantle-Cox log rank and Gehan-Breslow-Wilcoxon test, adding extra weight for early time points was used. In (B) to (H), unpaired Student’s t tests were used. Data represent means ± SEM.

Mitochondria CIII–generated superoxide is required for release of IL-10 protein from LPS-stimulated macrophages

To test whether mitochondria respiration is necessary for release of IL-10 protein in response to LPS, we restored respiration in QPC-KO mice by crossing our LysM^Cre/Cre to QPC^Fl/Fl mice and C. intestinalis AOX (ROSA26^SNAPf-AOX) mice (12, 30–32) to generate QPC-KO + AOX mice. AOX transfers electrons from ubiquinol (CoQH2) to O₂, allowing mitochondria CI and CII to pass electrons to ubiquinone (CoQ), thus allowing the regeneration of nicotinamide adenine dinucleotide (oxidized form) (NAD⁺) and flavin adenine dinucleotide, respectively (Fig. 3A) (21, 30). AOX reduces ETC generated superoxide but allows ATP generation from mitochondrial CI–dependent proton pumping. We have previously shown that AOX expression in QPC-deficient macrophages reduces superoxide production from the ETC (12). AOX expression in QPC-KO BMDMs was sufficient to rescue mitochondrial respiration, which was sensitive to AOX inhibitor salicylhydroxamic acid (SHAM) but not mitochondrial CIII inhibitor myxothiazol (Fig. 3, B to D). Critically, AOX expression was sufficient to rescue basal (Fig 3E) and phosphorylating respiration (Fig. 3F). However, the basal glycolytic rate remained elevated (Fig. 3G), while glycolytic capacity (Fig. 3H) was unchanged in KO + AOX BMDMs. This may occur as glyceraldehyde-3-phosphate dehydrogenase (GAPDH) is known to be oxidized by H₂O₂ to decrease glycolysis (33) and mitochondria CIII–generated superoxide, which rapidly becomes H₂O₂, may be required for GAPDH oxidation. AOX expression rescued NAD⁺ regeneration in QPC-KO macrophages (Fig. 4A) but was not sufficient to rescue IL-10 secretion from LPS-treated BMDMs in vitro (Fig. 4B). Il10 mRNA transcription in response to LPS was unaffected by mitochondria CIII loss or expression of AOX (fig. S3A). In addition, LPS stimulation of TNF-α release in vitro was slightly increased with AOX rescue compared to mitochondria CIII KO; however, it was not significantly elevated above QPC-HET controls (fig. S3B). Neither mitochondria CIII deletion, nor its functional rescue with AOX, altered the early response to LPS stimulation (Fig. 4, C and E, and fig. S3C); however, mitochondria CIII function was required for fulminant induction of the IL-10/ signal transducers and activators of transcription (STAT3) anti-inflammatory response (AIR) to LPS (Fig. 4D) (34). In addition, AOX was not sufficient to rescue survival following intraperitoneal injection of LPS (Fig. 4F), increase IL-10 levels (Fig. 4G), or decrease TNF-α levels in plasma (Fig. 4H) 2 hours post-intraperitoneal injection of LPS. Therefore, mitochondria CIII–dependent superoxide is required for release of IL-10 in response to LPS.

Fig. 3. AOX restores mitochondrial CIII–dependent respiration.

(A) Schematic of mitochondrial ETC and mitochondrial function in QPC-HET, QPC-KO, and QPC-KO + AOX mice was created in https://BioRender.com. (B) Seahorse analysis of oxygen consumption of QPC-HET (n = 7), QPC-KO (n = 5), and QPC-KO + AOX (n = 5) BMDMs with injections of mitochondrial CIII inhibitor myxothiazol (100 μM), AOX inhibitor SHAM (16 mm), and combination antimycin (1 μM)/piericidin (1 μM). Calculations of (C) mitochondrial CIII–dependent and (D) AOX-dependent respiration. Calculations of (E) basal respiration QPC-HET (n = 8), QPC-KO (n = 8), and QPC-KO + AOX (n = 7) (F) coupled respiration of QPC-HET (n = 6), QPC-KO (n = 6), and QPC-KO + AOX (n = 5) and (G) basal glycolytic rate QPC-HET (n = 8), QPC-KO (n = 8), and QPC-KO + AOX (n = 7) (H) glycolytic capacity QPC-HET (n = 4), QPC-KO (n = 4), and QPC-KO + AOX (n = 4) of BMDMs with injections of mitochondrial CIII inhibitor oligomycin (2 μM), AOX inhibitor SHAM (16 mm), and combination antimycin (1 μM)/piericidin (1 μM). In (C) to (H), ordinary one-way analyses of variance (ANOVAs) with Holm-Sidak’s multiple comparisons test with a single pooled variance were used to compare levels between the three groups. Data represent means ± SEM.

Fig. 4. Mitochondrial CIII, independent of enabling respiration, is required for LPS induction of IL-10 protein in macrophages.

(A) NAD⁺/NADH ratio in BMDMs ±2-hour LPS (100 ng/ml) stimulation (n = 4 all groups). (B) IL-10 secretion 2-hour post-LPS simulation QPC-HET (n = 8), QPC-KO (n = 10), and QPC-KO + AOX (n = 7) BMDMs. RNA-seq data for BMDMs ±2-hour LPS stimulation [QPC-HET (n = 5), QPC-KO (n = 5), and QPC-KO + AOX (n = 3)] examining summed gene module expression for (C) GO:0071222, (D) IL-10/STAT3 AIR (34), and a (E) heatmap of genes from gene module (GO:0071222). (F) Survival of QPC-KO and QPC-KO + AOX mice following injection of LPS (10 mg/kg) from three experiments QPC-KO (n = 32) and QPC-KO + AOX (n = 26). Plasma (G) IL-10 and (H) TNF-α levels 2-hour post-LPS injection (30 mg/kg) in QPC-HET (n = 11), QPC-KO (n = 13), and QPC-KO + AOX (n = 8) mice. (I) Intracellular IL-10 concentration in BMDM lysates following LPS stimulation ± Brefeldin A. (J) Secreted IL-10 in BMDMs following 2-hour stimulation with LPS (100 ng/ml) ± db-cAMP (100 μM) or H89 (1 μM). (QPC-HET, n = 7; QPC-KO, n = 11; db-cAMP, n = 4; LPS + db-cAMP+H89, n = 3). In (A) and (I), an ordinary two-way ANOVA was performed with Sidak’s multiple comparisons test. In (B), (G), and (H), an ANOVA with mixed effects analysis with Dunnett’s multiple comparisons test was performed. In (C) and (D), pairwise comparisons were performed using pairwise Mann-Whitney U tests with false discovery rate (FDR) correction. In (E), hierarchical clustering was performed using Ward’s method (D2) with a Euclidean distance metric. Gene expression values are z-normalized DESeq2 counts. In (F), Kaplan-Meier tests with Mantle-Cox log rank and Gehan-Breslow-Wilcoxon test were used. In (J), an ordinary one-way ANOVAs with Holm-Sidak’s multiple comparisons test were used to compare groups. Data represent means ± SEM.

Mitochondria CIII function is required for secretion of IL-10 through a cAMP/PKA-dependent pathway

Next, we tested whether mitochondrial CIII function is required for production and secretion of IL-10 protein. BMDMs were stimulated with LPS for 2-hour ±Brefeldin A to inhibit protein transport, and cellular lysates were examined for intracellular IL-10 levels by enzyme-linked immunosorbent assay (ELISA). QPC-HET, QPC-KO, and QPC-KO + AOX BMDMs all generated similar levels of intracellular IL-10 in response to LPS stimulation in the presence of Brefeldin A (Fig. 4I). Thus, mitochondria CIII function is not required for IL-10 production in response to LPS. Previous studies have shown that some IL-10 is secreted from the Golgi directly to the cell surface in a cAMP-dependent protein kinase (PKA)–dependent manner (35, 36). In addition, PKA activity is up-regulated in a mitochondrial ROS-dependent manner in hypoxic conditions (37). To test whether PKA activation was sufficient to rescue IL-10 release in our mitochondria CIII–deficient macrophages, we administered PKA activator dibutyryl–cyclic adenosine 3′,5′-monophosphate (db-cAMP) (38). The db-cAMP was sufficient to rescue IL-10 release from LPS-stimulated, mitochondria CIII–deficient BMDMs (Fig. 4J and fig. S3D). This was PKA-dependent as addition of PKA inhibitor, H89, abrogated this rescue. Together, these data show that mitochondrial CIII activation of PKA, likely through production of mitochondrial ROS, is required for secretion of IL-10 following LPS stimulation.

Mitochondria CIII function is not required for macrophage response to IL-4

Previous reports identified an association between mitochondrial ETC activity and the IL-4–induced polarization of macrophages (39, 40). To test the requirement of mitochondria CIII function in IL-4 polarization, we performed RNA sequencing (RNA-seq) analysis of QPC-HET and QPC-KO BMDMs 18 hours post–IL-4 stimulation (10 ng/ml). Analysis showed minimal differences in IL-4–mediated gene expression (Fig. 5, A and B); however, summed module expression of the IL-4–stimulated program (41) revealed a small but statistically significant decrease in program induction in QPC-KO macrophages (Fig. 5C) partially driven by Cre-mediated loss of Uqcrq gene (fig. S4A). To examine whether major markers of IL-4 activation in macrophages were altered we assessed standard protein markers of IL-4-stimulated macrophage function 18 hours post-stimulation (fig. S4B). The frequency of Arginase-1⁺ (Arg-1⁺) (Fig. 5D), RELM-α⁺ (Fig. 5E), PD-L2⁺ (Fig. 5F), CD301 (fig. S4C), and CD206 (fig. S4D) were not different between QPC-HET and QPC-KO BMDMs following 18-hour IL-4 stimulation. We also did not observe significant differences in staining intensity in Arg-1⁺ (fig. S4E) or PD-L2⁺ (fig. S4F) from IL-4–stimulated QPC-HET or QPC-KO BMDMs. As RELM-α is a critical mediator of in vivo responses to IL-4 (42), we activated peritoneal macrophages with an injection of thioglycolate followed by injections of IL-4 complex (IL-4c) on 0 and 2 days post-thioglycolate injection (fig. S5A). IL-4c is a 5:1 mixture, by weight, of IL-4 cytokine to monoclonal rat anti-IL4 (A11B11) to stabilize IL-4 in vivo and prolong its signaling, (25/5 μg per injection) (42). At day 4 after injection, peritoneal macrophages (CD45⁺CD11b^hiSSC^loTomato⁺; fig. S5B) from LysM^Cre-driven QPC-HET and QPC-KO mice showed no difference in frequency of Arg-1⁺ (Fig. 5G) or RELM-α⁺ (Fig. 5H) peritoneal macrophages. In addition, we examined cellular LPS response modules by RNA-seq upon stimulation with LPS or IL-4 in BMDMs from QPC-HET, QPC-KO, and QPC-KO + AOX mice. Our data show that the IL-4 response RNA-seq module is only observed with IL-4 stimulation (18 hours), which is independent of mitochondria CIII or AOX expression (fig. S6A). In addition, the cellular response to LPS is only observed with LPS stimulation (2 hours) and is also independent of mitochondria CIII or AOX expression (fig. S6B). These data reveal that mitochondria ETC activity is dispensable for the broad transcriptional responses to IL-4 and LPS in macrophages.

Fig. 5. Mitochondrial CIII function is not required for IL-4 stimulation of macrophages.

RNA-seq analysis of QPC-HET (n = 5) and QPC-KO (n = 5) BMDMs 18 hours post–IL-4 (10 ng/ml) stimulation or vehicle control by: (A) Principal components analysis (PCA), (B) heatmap of gene expression of the top 20 genes, and (C) summed module expression of the IL-4 response module as defined in (41). Frequency of (D) Arg-1⁺, (E) Relm-α⁺, and (F) PD-L2⁺ BMDMs 18 hours post–IL-4 (10 ng/ml) stimulation. Frequency of (G) Arg-1⁺ and (H) Relm-α⁺ peritoneal macrophages 4 days post-initial thioglycolate and IL-4c stimulation QPC-HET (unstimulated, n = 3; IL-4c stimulated, n = 10) and QPC-KO (unstimulated n = 3; IL-4c stimulated, n = 12). Data represent means ± SEM. In (A), PCA was performed using single value decomposition of the top 1000 genes by variance. In (B), hierarchical clustering was performed using Ward’s method (D2) with Euclidean distance as a distance metric. Gene expression values are z-normalized DESeq2 counts. In (C), pairwise comparisons were performed using pairwise Mann-Whitney U tests with FDR correction. In (D) to (H), ordinary one-way ANOVAs with Holm-Sidak’s multiple comparisons test with a single pooled variance were used to compare groups. Data represent means ± SEM.

DISCUSSION

IL-10 plays a pivotal role in mitigating inflammation by influencing both innate and adaptive immune cells (43), including macrophages (44). It exerts its anti-inflammatory effects by suppressing the synthesis of pro-inflammatory cytokines like TNF-α (45). IL-10 is crucial for maintaining the balance between host defense and disease tolerance (46). Perturbations in IL-10 levels have also been linked to various pathological conditions, including inflammatory bowel disease (47, 48), cancer (49), and viral infections (50).

Previous investigations have connected the functionality of the ETC with IL-10 transcription (51, 52). However, the genetic tools to examine the specific ETC functions, e.g., ATP production or ROS generation, essential for IL-10 generation and release were unavailable. Our current research demonstrates the necessity of mitochondria CIII in both in vitro and in vivo IL-10 response, independent of its role in ATP production. Unexpectedly, the loss of mitochondria CIII in our model is dispensable for the transcription and production of IL-10. However, it is required for IL-10 secretion following LPS or poly (I:C) stimulation. Our study demonstrates that this process is independent of mitochondria CIII’s ability to enable respiration and ATP production as the expression of AOX, which compensates for respiratory and ATP deficits due to mitochondria CIII impairment, does not rescue secreted IL-10 protein levels in vitro and in vivo.

Our findings uncover a mechanism of action and location of IL-10 regulation by the ETC. Previous studies have focused on transcriptional regulation of IL-10 by the ETC. One study found that mitochondria CIII function, in the context of myocardial infarction, was required for ETC-dependent NAD⁺ regeneration to induce IL-10 transcription during efferocytosis (51). Our models focus on early monocyte/macrophage responses to PRR stimulation with PAMPs and infection, rather than in the context of a sterile inflammatory response to tissue damage. These differences may be crucial in determining the dominant method of mitochondrial control of IL-10 production in macrophages.

AOX is known for its ability to reduce superoxide production originating from mitochondria CI and CIII while preserving crucial aspects of mitochondria metabolism, including respiratory-linked ATP generation, NAD⁺ regeneration, and oxidative TCA cycle activity (53). Our findings suggest that mitochondria CIII–derived superoxide, through PKA activation, serves as the critical mediator linking ETC functionality to IL-10 secretion. These findings synthesize two known PKA pathways as mitochondria ROS are known regulators of PKA in hypoxic conditions (37), and PKA modulates IL-10 secretion (35, 36). Administration of PKA activator db-cAMP was sufficient to rescue IL-10 secretion levels in mitochondria CIII–deficient BMDMs 2 hours following LPS stimulation. Notably, previous studies also indicate that IL-10 enhances mitochondrial quality control, preventing the accumulation of dysfunctional mitochondria (44). Therefore, mitochondria CIII–generated superoxide not only modulates macrophage IL-10 release but also contributes to enhancing mitochondrial health in these cells.

One unexpected finding of our study is that diminished mitochondria CIII function does not significantly alter the overall transcriptional response to LPS. A key distinction in our research, in contrast to previous studies, is that we limited our LPS stimulation to 2 hours, allowing us to examine the early-phase response to LPS (54). Many prior studies have focused on responses to LPS stimulation lasting 24 to 48 hours (16). Perhaps the most notable finding was that diminished mitochondria CIII function did not affect IL-4–dependent transcriptional responses, including the in vitro IL-4 polarization phenotype in macrophages. Previous reports have linked IL-4–dependent polarization and transcriptional responses in macrophages to fatty acid oxidation, mitochondrial respiration, and the oxidative TCA cycle (39, 40, 55). Note that disabling mitochondria CIII function or metabolic pathways through genetics may lead to a rewiring of metabolism and transcriptional responses that differ significantly from strategies involving short-term inhibition of ETC (56). However, AOX expression can restore metabolic impairments caused by the genetic loss of mitochondria CIII function without the generation of superoxide. Thus, our study highlights that mitochondria CIII–generated superoxide plays an essential role in IL-10 release.

MATERIALS AND METHODS

Mice

All mice in these studies were generated through in-house crosses of five strains. Three strains from the Jackson Laboratory (Bar Harbor, Maine)—Cre-expressing Cx3cr1^Cre-ERT2/YFP (strain# 021160) or Lys2^cre (Strain #:004781) and Cre reporter strain Ai14 (strain# 007914)—were crossed with QPC

^Fl/Fl strain (22) to generate experimental mice. Mice expressing AOX in myeloid cells were generated by first backcrossing C57Bl6/N Rosa26^SNAPf-AOX/wt mice given to us by M. Szibor (31, 32) onto a C57Bl6/J background for >6 generations. Then, we backcrossed Rosa26^{SNAPf-AOXl/WT} mice onto our LysM^Cre/CreQPC^Fl/FlROSA26^Ai14/Ai14 strain to generate three groups: control LysM^Cre/CreQPC^WT/FlROSA26^Ai14/Ai14, KO LysM^Cre/CreQPC^Fl/FlROSA26^Ai14/Ai14, and KO-AOX LysM^Cre/CreQPC^Fl/FlROSA26^{SNAPf-AOX/Ai14}. All experimental mice with the Cx3cr1^Cre-ERT2/YFP were used as heterozygous to have one functional Cx3cr1 allele. Experimental mice were homozygous for LysM^Cre to ensure complete cleavage in peripheral monocyte/macrophage populations in vitro and in vivo. Previous reports show that loss of LysM function has no effect on response to LPS administration (57). All mouse lines were maintained at Northwestern University under specific pathogen–free conditions in ventilated microisolator cages with automatic water access. Housing rooms had standard 12-hour light/12-hour dark cycles and an ambient temperature of 23°C. We complied with all relevant ethical regulations in accordance with Federal and University guidelines and protocols approved by the Institutional Animal Care and Use Committee and Northwestern University, protocol numbers IS00014481 and/or IS00001588.

Influenza infection

At least 3 weeks before IAV administration, mice were put on tamoxifen chow (Envigo, TD130860) to induce QPC deficiency in all CX₃CR1-expressing cells. Following recording mouse weight, mice were administered intratracheal dose of A/WSN/H1N1(1933) (15 PFU). Mouse weights were tracked for 23 days following infection.

BMDM isolation and cell culture

Bone marrow was isolated from mice and plated in 10 cm Primaria tissue culture plates (Thermo Fisher Scientific, 25382-701). To induce differentiation into macrophages, cells were cultured in RPMI medium containing 11 mM glucose, 10% fetal⁺ serum (Atlas Biologics, P16E19A1), 1 mM methyl pyruvate (Sigma-Aldrich, 371173), 400 μM uridine (Sigma-Aldrich, U3003), 1% antibiotic/mycotic (Thermo Fisher Scientific, 15-240-062), 1% Hepes (Thermo Fisher Scientific, catalog no. MT25060CI), and 4 mM glutamine (Gibco, catalog no.11965-126) supplemented with M-CSF (20 ng ml⁻¹; PeproTech, 315-02) at 37°C with 5% CO₂. The medium was changed every 3 days, and BMDMs were harvested by scraping on day 6 and plated in 12-well plates (2 million cells per well), 24-well plates (940,000 cells per well), 48-well plates (300,000 cells per well), or 96-well plates (150,000 cells per well).

BMDM in vitro stimulation

BMDMs were stimulated with ultrapure O5:B55 LPS (0.1 μg/ml; Invivogen, catalog no. tlrl-pb5lps) or poly (I:C) (20 μg/ml) for 2 hours or IL-4 (10 ng/ml) for 18 hours before collection. For stimulation of PKA activation in BMDMs, db-cAMP (100 μM; Sigma-Aldrich, 28745) was added at time of LPS stimulation. PKA inhibitor, H89 (1 μM; Cayman Chemical, A613329), was added 20 min before LPS stimulation.

Seahorse

Oxygen consumption rate (OCR) was performed on BMDMs using the XF96 extracellular flux analyzer seahorse (Seahorse biosciences). Following BMDM isolation and culture, we adhered 150,000 cells to 96-well seahorse plates (Seahorse Biosciences) using Cell-Tak (Corning, 354240) following the manufacturer’s instructions. Basal respiration was measured by subtracting the OCR values after treatment with 1 μM antimycin A (Cayman Chemical, 19433) and 1 μM piercidin A (Cayman chemical, 15379). Coupled respiration was determined by treatment with 2 μM oligomycin A (Sigma-Aldrich, 75351) and subtracting oligomycin A values from basal respiration. Injections of 100 nM myxothiazol (Sigma-Aldrich, T5580) and 16-mm SHAM (Sigma-Aldrich, S607) were used to inhibit mitochondrial CIII function and AOX function, respectively.

Cytokine analysis

For analysis of IL-10, TNF-α, and IFN-β cytokine levels, DuoSet ELISA (R&D Systems) or MILLIPLEX plates (Millipore) were used for supernatant cytokine release. In Fig. 2 (G and H) and fig. S2 (B to D), milliplex plates were used to analyze cytokine release. Duoset ELISA was used for analysis of plasma samples in Fig. 2B and fig S2A. In Fig 4B, MILLIPLEX or Duoset ELISA plates were used to analyze BMDM IL-10 release, and data were combined. In Fig. 4I, intracellular IL-10 on stimulated BMDM lysates was determined with DuoSet IL-10 ELISA. In Fig. 4 (G and H), MILLIPLEX plates were used for analysis of plasma cytokines.

Endotoxemia

Mice were injected with either Escherichia coli 055.B5 LPS (10 or 30 mg/kg; Sigma-Aldrich, L2880) diluted in 1x phosphate-buffered saline (PBS) for plasma collection at 2 hours post-injection or monitoring for endotoxemia survival. For monitoring, mice were carefully scored from 0 to 4 regarding their appearance, level of consciousness, activity, response to stimulus, eye health, respiration rate, and respiration quality (58). When mice reach an average score across all categories of 3.5, they are euthanized.

Flux analysis

Glucose-free RPMI (Thermo Fisher Scientific, 11-879-020) supplemented with 10% FBS and 11 mM d-glucose (U-¹³C₆) (Cambridge Isotope Laboratories, CLM-1396-1). BMDMs were plated 2 × 10⁶ cells per well in 12-well plates and washed with blank glucose-free RPMI twice before treatment with or without LPS (100 ng/ml; Sigma-Aldrich) for times indicated in text. After treatment, cells were washed two times with ice-cold saline and scraped into 1 ml of ice-cold methanol (80% methanol/20% H₂O) and stored at −80°C overnight. Samples were lysed by 3× cycles of freeze thaw in liquid N₂ followed by 37°C water bath. Samples were then spun at 16,000g for 15 min. Supernatant was collected and analyzed as below.

Metabolomics and LC-MS/MS

All samples were prepared as previously described (59); briefly, a 15-μl aliquot of the sample was used for high-resolution high-performance liquid chromatography (HPLC) tandem mass spectrometry. High-resolution HPLC tandem mass spectrometry was performed on a Q Exactive (Thermo Fisher Scientific) in line with an electrospray source and an UltiMate 3000 (Thermo Fisher Scientific) series HPLC consisting of a binary pump, degasser, and autosampler outfitted with an XBridge Amide column (Waters; 4.6 mm–by–100 mm dimension and a 3.5-μm particle size). Mobile phase A contained water and acetonitrile (95/5, v/v), 10 mM ammonium hydroxide, and 10 mM ammonium acetate (pH 9.0). Mobile phase B contained 100% acetonitrile. The gradient was set to 0 min, 15% A; 2.5 min, 30% A; 7 min, 43% A; 16 min, 62% A; 16.1 to 18 min, 75% A; 18 to 25 min, 15% A, with a flow rate of 400 μl min^–1. The capillary of the electrospray ionization source was set to 275°C, with sheath gas at 45 arbitrary units, auxiliary gas at 5 arbitrary units, and the spray voltage at 4.0 kV. A mass/charge ratio scan ranging from 70 to 850 was used in positive/negative polarity switching mode. MS1 data were collected at a resolution of 70,000. The automatic gain control (AGC) target was set at 1 × 10⁶, with a maximum injection time of 200 ms. The top five precursor ions were fragmented using the higher-energy collisional dissociation cell with normalized collision energy of 30% in MS2 at a resolution of 17,500. Data were acquired with Xcalibur software (v.4.1, Thermo Fisher Scientific).

NAD/NADH ratio measurement

BMDM NAD/NADH ratio was measured using the NAD/NADH-Glo Assay (Promega, G9071) as per the manufacturer’s instructions.

Tissue preparation for flow cytometry

Lung and blood single-cell suspensions were prepared as previously described (60). Briefly, the blood was collected from facial vein bleeds into K₂ EDTA-coated collection vials (Thermo Fisher Scientific, 02-669-33), and the volume was recorded. Blood plasma was separated from cells via 15 min spin at 10,000g at 4°C. The plasma was placed in fresh tubes and stored at −80°C for further use. For lung digestion homogenates, the protocol from (61) was performed with some modifications. Mice were euthanized, and their lungs perfused through the right ventricle with Hanks’ balanced salt solution (HBSS). Following removal, the lungs were filled with 10 ml of digestion buffer [collagenase D (2 mg/ml; Roche, 11088858001) and deoxyribonuclease I (0.1 mg/ml; Roche, 10104159001) dissolved in HBSS + Ca²⁺ and Mg²⁺), minced with scissors, and placed in C Tubes (Miltenyi) for processing on gentle MACS (Miltenyi) using the program m_lung_01. The processed lungs were then incubated at 37°C with gente agitation for 30 min and then pressed through a 40-μm filter. Single-cell suspension was then incubated with mouse anti-CD45 microbeads (Miltenyi, 130-052-301). CD45⁺ cells were collected through the MultiMACS Cell25 separator (Miltenyi) according to the manufacturer’s instructions. For all cell suspensions, red blood cells were lysed using Ammonium-Chloride-Potassium (ACK) lysis buffer. Cells were counted on the Cellometer K2 Fluorescent Cell Counter (Nexelcom) using acridine orange/propidium iodide solution to enumerate live (AO) and dead (PI) cells before dilution for staining with fluorochome-cojugated antibodes in fluorescence-activated cell sorting (FACS) buffer [1x PBS + 10% NuSerum IV culture supplement (Thermo Fisher Scientific, CB-55004)].

Flow cytometry and antibodies

For flow cytometric analysis of arginase and relma expression in peritoneal macrophages, we initially incubated cells with ghost dye violet 510 (Cytek, 13-0870-T100) as per the manufacturer’s instructions. For surface antibody staining, we initially resuspended cells in 1:50 anti-mouse CD16/32 (BioLegend, 101302, clone 93) and, subsequently, stained with the following antibodies: 1:100 anti-mouse CD45.2-PacBlue (BV605, BioLegend, 109841, clone 104) and 1:200 anti-mouse CD11b (Mac-1)–allophycocyanin (APC) (BV785, BioLegend, 101243, clone M1/70) in 200-μl final volume of FACS buffer. Fixation and permeabilization were performed with BD Fix/Perm (BD, 51-2090KZ) and, subsequently, stained with 5.5 μl of anti-RELMα (APC, Invitrogen, 17-5441-82, clone: DS8RELM) and 5.5 μl of anti–Arg-1 [fluorescein isothiocyanate (FITC), Intivrogen, 53-3697-82, clone: A1exF5]. For flow cytometric analysis of Arg and Relmα expression in BMDMs, we initially incubated cells with ghost dye violet (Cytek, 13–0870-T100) as per the manufacturer’s instructions. For surface antibody staining, we initially resuspended cells in 1:50 anti-mouse CD16/32 (BioLegend, 101302, clone 93) and, subsequently, stained with the following antibodies: 1:100 anti-mouse CD45.2-PacBlue (BV605, BioLegend, 109841, clone 104) and 1:200 anti-mouse CD11b (Mac-1)–APC (BV785, BioLegend, 101243, clone M1/70) in 200-μl final volume of FACS buffer. Fixation and permeabilization were performed with BD Fix/Perm (BD, 51-2090KZ) and, subsequently, stained with 5.5 μl of anti-RELMα (PerCP-efluor710, Invitrogen, 53-3697-82, clone DS8RELM), 5.5 μl of anti–Arg-1 (FITC, Invitrogen, 53-3697-82, clone: A1exF5), and anti-CD206 (BV421, BioLegend, 141717, clone C068C2).

For flow cytometric analysis of IAV-infected lung single-cell suspensions, we initially incubated cells with ghost dye violet 780 (Cytek, 13-0865-T100) as per the manufacturer’s instructions. For surface antibody staining, we initially resuspended cells in 1:50 anti-mouse CD16/32 (BioLegend, 101302, clone 93) and, subsequently, stained with the following antibodies: 1:100 anti-mouse CD45.2-PacBlue (BV605, BioLegend, 109841, clone 104), 1:200 anti-mouse CD11b (Mac-1)–APC (BV785, BioLegend, 101243, clone M1/70), 1:250 anti-mouse CD64 (BV421, BioLegend, 139309, clone X54-5/7.1), 1:200 anti-mouse Ly6G (BV650, BioLegend, 127641, clone 1A8), anti-mouse Siglec-F (BB700, eBioscience, 46-1702-82, clone 1RNM44N), and anti-mouse Ly6C (PeCy7, Thermo Fisher Scientific, 25-5932-82, clone HK1.4) in a final volume of 200 μl for 20 min on ice. All samples were run on BD FACS Symphony A5-laser analyzer, and analysis was done using FlowJo 10 (10.9.0).

In vivo stimulation of peritoneal macrophages with IL-4

In vivo stimulation of IL-4 macrophage protocol was adapted from (42). Briefly, LysM^Cre/CreQPC^WT/Fl (QPC-HET) and LysM^Cre/CreQPC^Fl/Fl (QPC-KO) mice were injected with thioglycolate with or without IL-4c (IL-4c, a 5:1 mixture of IL-4 cytokine to monoclonal rat anti-IL-4 (A11B11) to stabilize IL-4 in vivo and prolong its signaling). IL-4c (25/5 μg per mouse per injection) was administered again on day 2. On day 4, peritoneal cells were lavaged, and single-cell suspension was stained with fluorescent-labeled antibodies.

Bulk RNA-seq

RNA and DNA from 1.5 × 10⁵ BMDMs were isolated using AllPrep DNA/RNA Micro kit (QIAGEN, 80284). RNA quality and quantity were assessed using TapeStation 4200 High Sensitivity RNA tapes (Agilent), and RNA-seq libraries were prepared from 100 ng of total RNA using the NEBNext Ultra DNA Library Prep Kit for Illumina (NEB E7370L). Library quality control was then performed using TapeStation 4200 High Sensitivity DNA tapes (Agilent). Dual-indexed libraries were pooled and sequenced on a NextSeq2000 instrument (Illumina) for 100 cycles, single-end, to an average sequencing depth of 14.99 million reads per sample. FASTQ files were generated using bcl-convert 4.0.3 using default parameters. To facilitate reproducible analysis, samples were processed using the publicly available nf-core/RNA-seq pipeline version 3.9 implemented in Nextflow 22.04.5.5708 using Singularity 3.8.1 with the minimal command nextflow run nf-core/rnaseq \ -r ‘3.9’ \ -profile nu_genomics \--additional_fasta ‘transgenes.fa’ \ --star_index false \ --genome ‘GRCm38’ for SMARTer-seq samples. For NEBnext samples, 3′ clipping was not performed. Briefly, lane-level reads were trimmed using trimGalore! 0.6.7 and aligned to the hybrid genome described above using STAR 2.6.1d. Gene-level assignment was then performed using salmon 1.5.2.

Bulk differential expression analysis

All analysis was performed using custom scripts in R version 4.1.1 using the DESeq2 version 1.34.0 framework. A “local” model of gene dispersion was used as this better fit dispersion trends without obvious overfitting, and pairwise comparisons were performed using Wald tests on a combined factor of treatment and genotype. Alpha was set at 0.05 for all DEA. Otherwise, default settings were used. High-level analysis was performed using custom scripts available in the NUPulmonary/utils/R GitHub repository.

k-means clustering of bulk samples

The k_means_figure function from NUPulmonary/utils was used for k-means clustering. Briefly, variable genes were identified using a likelihood-ratio test with local estimates of gene dispersion in DESeq2 with diagnosis as the full model and a reduced model corresponding to intercept alone (~1). Genes with q ≥ 0.05 were discarded. Extant genes were then clustered using the Hartigan-Wong method with 25 random sets and a maximum of 1000 iterations using the k-means function in R stats 4.1.1. Samples were then clustered using Ward’s method and plotted using pheatmap version 1.0.12. GO term enrichment was then determined using Fisher’s exact test (classic mode) in topGO version 2.46.0, with org.Mm.eg.db version 3.14.0 as a reference.

RNA-seq statistical analysis

For all analyses, normality was first determined using the Shapiro-Wilk test and visual inspection of histograms. In cases where distributions were non-normal, nonparametric statistics were used. All analysis was performed using custom scripts in R 4.1.1, all of which are publicly available on Github at NUPulmonary/utils. Plotting was performed using ggplot2 3.4.2 unless otherwise noted. Comparisons for these figures were added using ggsignif 0.6.4. Heatmaps were generated using ComplexHEatmap 2.10.0 using Euclidean distance as the distance metric and the Ward D2 clustering method. In all box plots, box limits represent the interquartile range (IQR) with a center line at the median. Whiskers represent the largest point within 1.5× IQR. All points are overlaid with jittering.

Statistical analysis

Statistical analysis and visualization of all figures, other than RNA-seq was performed using Graphpad Prism (v10.4.1). Specific statistical methods for each figure are located at the end of the associated figure legend for clarity.

Supplementary Materials

This PDF file includes:

Figs. S1 to S6