Macrophage metabolic reprogramming during chronic lung disease

Abstract

Airway macrophages (AMs) play key roles in the maintenance of lung immune tolerance. Tissue tailored, highly specialised and strategically positioned, AMs are critical sentinels of lung homoeostasis. In the last decade, there has been a revolution in our understanding of how metabolism underlies key macrophage functions. While these initial observations were made during steady state or using in vitro polarised macrophages, recent studies have indicated that during many chronic lung diseases (CLDs), AMs adapt their metabolic profile to fit their local niche. By generating reactive oxygen species (ROS) for pathogen defence, utilising aerobic glycolysis to rapidly generate cytokines, and employing mitochondrial respiration to fuel inflammatory responses, AMs utilise metabolic reprogramming for host defence, although these changes may also support chronic pathology. This review focuses on how metabolic alterations underlie AM phenotype and function during CLDs. Particular emphasis is given to how our new understanding of AM metabolic plasticity may be exploited to develop AM-focused therapies.

Introduction

The respiratory mucosa is a unique site, as our airways are continually exposed to particulates, viruses, bacteria, and fungi, which challenge the pulmonary immune system¹. To maintain pulmonary homoeostasis and ensure efficient gas exchange, a complex regulatory system is in place, of which airway macrophages (AMs) are a core component. AMs are the most numerous immune cell type present in healthy lungs, are strategically positioned at the interface of airways and environment², and critical sentinels of barrier immunity. AMs form the first line of defence against inhaled particles, pathogens and antigens³. Although inherently suppressive, AMs exhibit significant functional and phenotypical specialisation, allowing efficient responses to environmental signals and rapid alterations in phenotype. Increasing evidence suggests that metabolic alterations provide an additional layer of functional plasticity to AM populations. Activation of macrophages in vitro with a range of inflammatory stimuli, induce profound metabolic adaptations, such as the switch from oxidative phosphorylation (OXPHOS) to glycolysis in oxygen-sufficient conditions, similar to the “Warburg effect” seen in some cancers⁴. It is now clear that how macrophages utilise energy dictates immune responses, and that manipulating cellular metabolism can alter the inflammatory response⁵. However in vivo, the unique oxygen rich environment of the airways coupled with specific local nutrient availabilities, shapes AM phenotype and function. Indeed, many recent studies have indicated that in chronic lung diseases (CLDs), such as asthma, chronic obstructive pulmonary disease (COPD), cystic fibrosis (CF), idiopathic pulmonary fibrosis (IPF), and during infection such as with Mycobacterium Tuberculosis (Mtb), there are significant alterations in AM metabolic processes and that targeting these pathways could represent an exciting therapeutic approach^6,7. This review focuses on how metabolic adaptations underlie AM phenotype and function during CLDs. Particular emphasis is given to how our new understanding of AM metabolic plasticity may be exploited to develop AM-focused therapies.

Airway macrophages: guardians of the lung environment

To maintain gas exchange, it is critical that AMs sustain a naturally hyporesponsive state whilst also reserving the ability to rapidly mount effective inflammatory responses. This balance is achieved through complex AM-airway epithelial cell (AEC) interactions via cell surface-expressed receptors and secreted products. AMs express transforming growth factor beta receptor (TGF-βR), interleukin (IL)-10 receptor (IL-10R), CD200 receptor (CD200R) and signal regulatory protein-α, key components mediating AM:AEC crosstalk and in turn, regulating AM activation⁸. For example, AM-AEC contact decreases AM phagocytosis and cytokine production in a TGF-β-dependent manner⁹. Conversely, loss of the integrin αvβ6 such as through loss of contact of AMs with AEC upon toll-like receptor (TLR) activation leads to initiation of the AM pro-inflammatory phenotype and inflammatory response¹⁰.

AMs are characterised by a distinct cellular phenotype. Human AMs highly express the lectin-binding transmembrane glycoprotein CD68, the adhesion molecule CD169, the integrin CD11c and mannose receptor CD206¹¹ (Fig. 1). In mice, expression of the CD68⁺CD206⁺CD11c^hiCD11b^lo cell surface phenotype is conserved at steady state^12–14, while murine AMs also express the Mer tyrosine kinase (MerTK), sialic acid dependent adhesion molecule SiglecF, hormone receptor F4/8, glycoprotein CD64 and the CD200 receptor¹⁵ (Fig. 1). Recent work in mice has indicated that many tissue resident macrophages, including those in the airways, are foetally derived and self-maintain locally with minimal contribution from circulating monocytes, during steady state conditions^16–20. During murine prenatal development, foetal liver or yolk sac macrophages are the major contributing pool to AM populations²¹ and AM colonization of the lung occurs in sequential waves in the first week of life²². Furthermore, post birth and during maturation, circulating monocytes do not significantly contribute to lung macrophage populations at homoeostasis¹⁸. During pulmonary inflammatory responses however, monocytes are recruited to the lung^23,24 and subsequently develop into AM-like cells^18,25. Thus, post-injury murine airways contain at least two ontologically distinct AM populations, tissue resident AMs (TR-AM) which are prenatally derived and monocyte-derived AMs (Mo-AMs). Several groups have studied samples from lung transplant patients to investigate the origins of AMs in the human lung^26–30. Utilizing bronchoalveolar lavage (BAL) from sex-mismatched lung transplant patients our group recently demonstrated that the majority of AMs in human lung post-transplant are derived from peripheral classical monocytes³¹. Thus, the unique airway niche combined with distinct ontological origins, age and environmental exposures results in remarkable AM plasticity and adaptability^32–34. While the recruitment of monocytes to the lungs to replenish the TR-AM pool is relatively well understood in mice, the origins of AMs in the human lung during healthy aging or CLDs requires further investigation; in particular clear markers which distinguish Mo-AMs and TR-AM are not well established.

Fig. 1. Human and murine airway macrophage surface receptors.

Murine AMs express the lectin-binding transmembrane glycoprotein CD68, the Mer tyrosine kinase (MERTK), the integrin alpha X chain protein CD11c, the type I membrane glycoprotein CD200 receptor, the mannose receptor CD206, the EGF-like module-containing mucin-like hormone receptor-like 1 (F4/80), the sialic acid binding lectin Siglec-F and the Fc receptor CD64. Human AMs express CD68, the adhesion molecule CD169, CD11c, CD206 as well as MHC class II receptor HLA-DR.

Airway macrophage metabolic phenotype at homoeostasis

Beyond delineation of macrophage populations based on anatomical location or ontological origin, macrophage populations are also classified according to their activation status. Analogous to the Th1/Th2 paradigm, in vitro cultured human monocyte derived macrophages (MDMs) or murine bone marrow derived macrophages (BMDMs) are categorised as M1 or M2 macrophages, respectively. Seminal studies have demonstrated that pro-wound healing M2 (IL-4 stimulated) macrophages in vitro rely on fatty acid oxidation (FAO), an intact tricarboxylic acid cycle (TCA) cycle, high rates of OXPHOS and increased expression of arginase 1 (Arg1), which catalyses the production of ornithine from arginine as precursor for collagen to facilitate wound healing^5,35–37. Conversely, pro-inflammatory M1-macrophages rely on glycolysis and breaks in the TCA cycle lead to accumulation of metabolites, many of which have signalling functions such as citrate, succinate, fumarate and α-ketoglutarate^35,38,39. However, although useful in defining the range of potential macrophage responses, in vitro derived cells do not recapitulate the core aspects of AM phenotypes which are shaped by the local niche¹¹. As AMs are highly adapted to the unique environment of the airway lumen, it is perhaps unsurprising that the metabolic state of AMs is also distinct. Glucose concentrations in the alveolar lumen are less than 10% of blood glucose concentrations and AMs exhibit extremely low levels of glycolysis²⁰; in stark contrast to BMDMs, AMs do not undergo glycolytic reprogramming in response to LPS⁴⁰. Consequently, AMs readily engage OXPHOS and highly express the peroxisome proliferator-activated receptor gamma (PPARγ)⁴¹, which regulates lipid accumulation and promotes FAO to sustain OXPHOS.

AMs also play a major role in the catabolism of pulmonary surfactant, a monolayer composed mainly of phosphocholine-based lipids, phospholipids and cholesterol which lines the alveoli, lowers surface tension and prevents alveolar collapse during expiration⁴². Mice lacking GM-CSF and thus the AM compartment, develop pulmonary alveolar proteinosis (PAP), an inflammatory lung syndrome caused by the defective clearance of surfactant^43–45. In humans, mutations in genes encoding for GM-CSF receptors, result in hereditary PAP as a result of progressive alveolar surfactant accumulation^46–49. AM phenotype and behaviour are influenced by surfactant exposure, which has major implications for AM-mediated immune responses in pulmonary tissue. There are four principle surfactant proteins (SP-A-D) and SP-A and SP-D have been shown to directly influence AM functions such as cell migration, phagocytosis and activation phenotypes⁴². Both SP-A and SP-D bind carbohydrates, lipids, and nucleic acids and initiate phagocytosis of inhaled pathogens and apoptotic cells⁵⁰. Furthermore, SP-A blocks the binding of TLR ligands to TLR2, TLR4 and TLR co-receptors and furthermore inhibits complement activation^51,52.

Whilst the alveoli are covered with a monolayer of surfactant, a thin layer of mucus produced by goblet cells and ciliated epithelium protects the airways. Mucus serves as a barrier and facilitates clearance of microbes and pollutants. A major component of mucus are mucin glycoproteins, which may be categorized as polymerizing, nonpolymerizing and cell-surface associated. Of the cell-surface associated mucins, MUC1 is expressed in AMs and contributes to the resolution of inflammation by decreasing phagocytic potential and pro-inflammatory cytokine production⁵³. The polymerizing mucins include MUC5AC and MUC5B; in particular, MUC5B deficiency has been linked to particle accumulation in the lung, mucus obstruction and impaired macrophage phagocytosis⁵⁴. Pro-inflammatory macrophages induce MUC5B expression to aid mucociliary clearance⁵⁵. Furthermore, a single nucleotide polymorphism in the MUC5B promotor has been strongly associated with the risk of developing IPF, highlighting the importance of mucins for the pulmonary environment⁵⁶.

In addition to low glucose and a lipid rich environment, the airways also have a unique distribution of amino acids and central carbon metabolites. Surowiec et al. showed that whilst several glucogenic and ketogenic amino acids were present in the bronchial wash, only alanine is present in BAL⁵⁷ (Fig. 2). In addition, the central airways contained key glycolytic and OXPHOS metabolites such as fructose, glucose-6-phosphate, fumarate and malate as well as oxidised gluthathione (GSSG, indicating oxidative stress, Fig. 2); interestingly, these could not be detected in the periphery, suggesting either minimal secretion, high utilisation or as a result of anatomical location (i.e. close proximity to nutrient rich pulmonary capillaries)⁵⁷. Recently the lung microbiome has gained attention as a factor which modifies the pulmonary environment and directs immune responses by producing short chain fatty acids (SCFA). Whilst the airways and alveoli are colonised mainly by proteobacteria, bacteroidetes and firmicutes^58,59, the nasal mucosa additionally hosts actinobacteria^60,61 (Fig. 2). Proteobacteria, bacteroidetes and firmicutes produce large amounts of SCFA, including acetate, propionate and butyrate, which influence barrier function by regulating epithelial tight junctions⁶² and anti-inflammatory immune responses⁶³. Recent advances in understanding the pulmonary microbiome during homoeostasis and CLDs are described in detail elsewhere^58,59,64. Thus, at homoeostasis AMs are exposed to a unique environment, with minimal glucose availability and a distinct distribution of nutrients and SCFA, which depend on anatomical location (Fig. 2). However, despite the profound influence that local substrate availabilities may exert on macrophage development, activation and function, this is an understudied area. New knowledge, which further defines especially human AM substrate dependencies at homoeostasis is required in order to fully understand how local metabolic perturbations during CLDs may contribute to pathology. This is particularly relevant as already slight changes in nutrient availability during inflammation, such as succinate or citrate, can alter macrophage phenotypes through stabilization of Hif1α, post-translational modification of proteins and production of NO and ROS⁶⁵, thereby contributing to a pathological development.

Fig. 2. Nutrient environment at respiratory mucosal sites.

The respiratory mucosa stretches from the nasal cavity to the alveoli and its pseudostratified epithelium in the upper respiratory tract consists of mucus producing goblet cells, ciliated cells and progenitor basal cells on top of a basement membrane, covered with a layer of mucus. The alveoli are lined with type I alveolar epithelial cells (AEC) interspersed with surfactant producing type II AEC. Distinct metabolites of the key metabolic pathways have been detected in nasal secretions, bronchial washes and bronchoalveolar lavage (BAL) at the different respiratory mucosal sites. These key metabolic pathways are schematically represented here. Varying reactive oxygen species (ROS) and commensal bacteria have been detected in the nasal cavity, conducting airway and parenchyma, contributing to the unique environment in each compartment. The principal microbial phyla colonising the airway and alveoli are proteobacteria, bacteroidetes and firmicutes, while the nasal mucosa additionally hosts actinobacteria.

AM metabolism during CLDs

Chronic lung diseases affect a significant proportion of the world’s population, killing more than 100,000 people in the UK alone, each year⁶⁶. Persistent inflammation, impaired repair processes and pulmonary remodelling are cardinal features of CLDs^67–69. There are multiple overlaps in environmental exposures driving CLDs, such as smoking, pollution and environmental exposures; viral infection can also exacerbate symptoms in each disease^2,70,71. Interestingly, AM metabolic adaptation may play a central role in dictating pathology during CLDs and present novel therapeutic opportunities (Fig. 3).

Fig. 3. Altered metabolic pathways in AMs drive key features of chronic lung disease.

Several metabolic pathways are rewired during chronic lung disease. While this response exists to clear invading pathogens and launch an inflammatory response, long-term activation of these pathways has negative implications. The glycolysis pathway supports inflammatory responses of AM, while iron and metabolites produced in the TCA cycle can function as bacterial substrates and contribute to pathogen survival. While fatty acid synthesis and oxidation is useful as a way of storing energy and alternative energy source during times of macrophage activation, fatty acid synthesis can also contribute to mucus production. Leukotrienes contribute to the AM pro-inflammatory phenotype but also cause bronchial constriction and contribute to airway remodelling in asthmatics by causing smooth muscle thickening. The amino acid arginine is a proliferator for collagen via ornithine and proline and can thereby contribute to extracellular matrix deposition.

Asthma

Asthma is a heterogeneous disease of the airways characterized by airway remodelling, mucus production, airway hyperresponsiveness (AHR), and inflammation⁷². Although most patients have good control with standard medication, a proportion experience life-threatening, severe disease⁷³. AMs are central to mediating type-2 inflammation against allergens and parasitic worms². In vitro, macrophages respond to type-2 cytokines such as IL-4 that drive an ‘alternative’ M2 activation phenotype, linked to wound repair and type-2 pathology^74,75. Manipulation of AM phenotype via genetic deletion of the transcription factor interferon regulatory factor 5 (Irf5), a master regulator of macrophage activation⁷⁴, promoted pulmonary remodelling, AHR and mucus secretion in mice, in an IL-13 dependent manner⁷⁶. Indeed, a recent study has shown that both CD206⁺ AM (activated by IL-4 and IL-13) and pro-inflammatory Irf5⁺ AM are increased in asthmatic patients⁷⁷, highlighting the plasticity of macrophages and heterogeneity of human asthma. Roles of AMs during antigen induced airway inflammation include phagocytosis of apoptotic cells and eosinophils as well as triggering anti-inflammatory pathways to regulate airway hyper responsiveness, mucus secretion and matrix deposition⁷⁶. In severe asthma however, this protective function is impaired, resulting in the loss of phagocytic ability and anti-inflammatory programme⁷⁸, which can contribute to airway remodelling⁷⁹. Thus, AMs are uniquely involved in responses to allergen and type-2 inflammation, and aberrant AM-phenotypes can directly influence respiratory pathology.

Numerous lines of evidence suggest that metabolic stress leading to the production of reactive oxygen species (ROS) plays a role in asthma. Increased ROS have been detected in AMs of asthmatic patients⁸⁰, and contributes to lung injury⁸¹ and pro-inflammatory tumour necrosis factor-alpha (TNF-α) and IL-1β secretion by macrophages⁸². Furthermore, heme-oxygenase-1 (HO-1), which mediates ROS production in response to chemical and physical agents, is increased in AMs in asthmatics⁸³. In addition to these pro-inflammatory characteristics, AMs show key features of a more anti-inflammatory phenotype in studies using ovalbumin to model allergic asthma. Using this model, Al-Khami et al. show that expression of carnitine palmitoyltransferase (CPT) is increased in AMs, shuttling fatty acids into the mitochondria, as well as increased gene expression of FAO related genes⁸⁴.

Another functional pathway that is altered in asthmatic AMs and links to the underlying metabolic phenotype is the eicosanoid pathway, which is induced by Th2 cytokines IL-4 and IL-13. Eicosanoids, including prostaglandins and leukotrienes, are produced from the poly-unsaturated fatty acid arachidonic acid, which is released during asthma⁸⁵. Increased production of the eicosanoid 5-HETE and leukotrienes B₄ (LTB₄) and E₄ (LTE₄) has been detected in AMs from asthmatic patients stimulated ex vivo⁸⁶. This contributed to bronchial constriction and pro-inflammatory phenotype and failure to generate the anti-inflammatory eicosanoid 15-HETE and prostaglandin E₂ (PGE₂), which is associated with reduced AM phagocytosis⁷⁸. LTE₄ has been shown to cause AHR in subjects with aspirin-induced asthma⁸⁷ and can be produced in AMs by γ-glutamyl transpeptidase^85,88. IL-13 furthermore induces Arg1, which may further contribute to the asthmatic phenotype via metabolism of collagen precursors ornithine and proline to collagen and airway remodelling^89,90.

Several of these observed alterations have been targeted therapeutically, attempting to rewire AM phenotype. These include the eicosanoid pathway, ROS, glycolysis and FAO. Administration of the anti-inflammatory eicosanoid 15-HETE inhibited leukotriene synthesis and reduced AHR in asthmatics⁹¹. Ex vivo, the corticosteroid dexamethasone decreased levels of thromboxane B2 and LTB₄ in macrophages and asthmatic AMs⁹², while prednisone decreased LTB₄ production in AMs⁹³. Treatment with the antioxidant AD4 improved AHR and airway inflammation by decreasing ROS in the OVA-sensitised mouse model of allergic airway disease (AAD)⁹⁴. Inhibiting glycolysis with 2-DG, Zhao et al. show altered AMs phenotype ameliorated AAD, while Al-Khami et al. reported improvements in AHR after treatment with FAO inhibitor etomoxir^84,95.

Together, these studies indicate that there is significant disruption of AM metabolism during asthma and AAD, notably via dysfunction in eicosanoid, glycolysis and fatty acid pathways. In order to evaluate candidate therapies, it is crucial that studies utilise relevant preclinical models and ex vivo patient samples to understand disease. Models which more closely recapitulate the complex immune response to allergens, are more likely to reveal viable targets for intervention; in particular the ovalbumin model of AAD, which requires an adjuvant and a sensitization phase⁹⁶, is a poor murine model of asthma. Furthermore, our new understanding of asthma heterogeneity has allowed the development of biologics which target “type-2 high” asthma⁹⁷; delineation of how metabolic changes underlie distinct asthma phenotypes could lead to new treatments for other phenotypes, such as neutrophilic and paucigranulocytic asthma.

COPD

COPD is the 5th leading cause of death in high income countries⁹⁸, affecting over 200 million people worldwide⁹⁹. COPD is a heterogeneous disease, characterised by destruction of the parenchyma and emphysema, narrowing of the airways, remodelling and chronic inflammation driven by chronic exposure to cigarette smoke and particular matter¹⁰⁰. AM numbers are increased in COPD BAL¹⁰¹ and contribute to COPD pathology through numerous pathways. During COPD, AMs are found in areas of lung destruction and produce pro-inflammatory cytokines¹⁰², chemokines¹⁰⁰ and matrix metalloproteases (MMPs) with elastolytic properties^103,104. At the same time, tissue inhibitor of metalloproteases (TIMP)-1 is decreased in AMs in COPD¹⁰⁵ and furthermore, decreased phagocytic capacity and impaired bacterial killing have been described in COPD-AMs^106–108.

AMs of COPD patients experience a high level of oxidant burden induced through cigarette smoke and subsequent increased ROS production is a key feature¹⁰⁹. Compared to controls, COPD-AMs secrete increased levels of mitochondrial ROS (mtROS)¹¹⁰, superoxide and hydrogen peroxide^81,108 whilst glutamyl cysteine ligase for GSH synthesis is downregulated¹¹¹. Cigarette smoking also alters iron homoeostasis¹¹² and AMs in COPD show increased sequestering of iron¹¹³, which can furthermore contribute to ROS production. Bewley et al. showed recently that increased generation of mtROS in COPD AMs results in impaired bacterial clearance¹⁰⁸. This study also reported a decrease in the mitochondrial membrane potential¹¹⁴, which has recently been linked to AM exposure to particulate matter¹¹⁰. This may explain the impaired phagocytic capacity of AMs in COPD as decreased mitochondrial membrane potential results in energy failure in the cell, proton leakage and increased mtROS¹¹⁵. Another study by O’Beirne et al. further investigated the metabolic profile of AMs from healthy smokers, non-smokers and COPD patients. While all groups had similar baseline glycolysis rates, there was a decrease in coupling efficiency, maximal respiration and spare respiratory capacity in COPD-AMs, while proton leak was significantly increased¹¹⁶. In addition, expression of genes related to glutathione metabolism, mitochondrial transport, pyruvate metabolism, TCA cycle and electron transport chain were altered in smokers and COPD patients, compared to non-smoking healthy controls¹¹⁶.

Other metabolic alterations in COPD AMs include increased expression of inducible nitric oxide synthase (iNOS) contributing to increased levels of nitric oxide (NO)¹¹⁷ and increased levels of the adenosine receptor A2BR¹¹⁸, suggesting increased adenosine metabolism, which might be linked to the increased levels of Hif1α in COPD AMs¹¹⁹. While excessive ROS production through oxidant burden and iron accumulation has been identified as an important regulator of AM phenotype in COPD, it has only recently been linked to mitochondrial dysfunction and metabolic reprogramming. It would be interesting to follow up on these transcriptomic and metabolic alterations to understand their underlying disease driving role and to identify ways to rewire AM metabolism.

As corticosteroids have been found to be particularly ineffective in COPD, more specific pathways involved in AM function and metabolism have been investigated recently, such as the ROS pathway and iron accumulation. A study by Harvey et al. showed that treatment with sulforaphane in COPD AMs ex vivo improved bacterial clearance by activating the antioxidant and anti-inflammatory NRF2 pathway¹²⁰, while Cloonan et al. found that treatment with an iron chelator or a low iron diet protected mice from cigarette smoke induced COPD¹²¹. Furthermore, procysteine, a precursor of GSH, increased AM efferocytosis in a mouse model of COPD¹²².

Overall, COPD is marked by distinct iron sequestration, ROS, NO production and energetic dysfunction in AMs; further delineation of how mitochondrial phenotype links to inflammatory processes and pathology in COPD will allow the identification of molecular targets for modulating mitochondria during the disease.

Cystic fibrosis

Cystic fibrosis (CF) is caused by mutation of the CF transmembrane conductance regulator (CFTR), a chloride channel, which regulates fluid homoeostasis in mucosal surfaces. In the lung, CFTR mutation and subsequent loss of function results in a reduced aqueous film covering the epithelium and mucus thickening, leading to impaired mucociliary clearance and frequent bacterial infection¹²³. CF is furthermore characterised by hyper-inflammation of the lungs, airway obstruction, structural damage and progressive reduction of lung function¹²⁴. During recurring airway inflammation, large numbers of neutrophils, macrophages and T lymphocytes infiltrate the lungs and secrete pro-inflammatory cytokines, while anti-inflammatory IL-10 is reduced^125,126. Although AM numbers are increased in CF patients^127,128, pathogen clearance is attenuated, leading to colonisation of the airways and chronic inflammation^106,128. Meyer et al. report a more pro-inflammatory phenotype of AMs in a murine model of CF, even in the absence of infection¹²⁹ and MDMs differentiated from CF patients show an increased inflammatory profile¹³⁰, while others have shown that monocytes from CF patients had an impairment in activation upon IL-13 stimulation¹³¹. CF-AM phenotype can be heterogenous, depending on infection status and local environment. While AMs from P. Aeruginosa infected CF patients showed increased expression of mannose receptor CD206 and augmented arginase activity¹³², in CF sputum AMs a decrease in expression of CD206 and scavenger receptor MARCO was detected¹³³. Furthermore, AMs are involved in the structural damage in CF airways by secreting serine- and metalloproteases, which subsequently degrade connective tissue components¹³⁴. The lower volume of airway surface liquid in CF airways activates AMs to increase their release of MMP12, resulting in the cleavage of elastin and degradation of the airway and parenchyma¹³⁵.

In CF airways, GSH is depleted^136,137, while levels of iron, transferrin, haem and haemoglobin are increased¹³⁸, resulting in high oxidative stress and ROS production. ROS in turn can induce TGF-β1¹³⁹, which has recently been shown to be increased in CF-BAL and AMs and inhibits CFTR biogenesis and cellular trafficking to the surface of epithelial cells¹³⁴, while also contributing to airway remodelling by recruitment and differentiation of myofibroblasts¹⁴⁰. However, during infection with bacteria from the Burkholderia family, both MDMs and AMs from CF patients showed reduced superoxide production as well as decreased phosphorylation of NADPH oxidase (NOX) components p47^phox and p40^phox, suggesting an inherent deficit in CF-AMs generating oxidative bursts for pathogen defence¹⁴¹.

P. Aeruginosa is one of the most common pathogens to cause recurrent pulmonary infection in CF patients and exploits the host to maintain infection by inducing production of the TCA cycle metabolite itaconate in AMs. Itaconate exerts antimicrobial properties via inhibition of bacterial isocitrate lyase in the glyoxylate shunt¹⁴² and to evade this mechanism P. Aeruginosa has developed a way to use itaconate as an energy source¹⁴³. Similarly succinate, which is secreted in high levels during CF and especially during bacterial infection¹⁴⁴, can be utilised by P. Aeruginosa and S. Aureus as a substrate to generate oxidative stress.

Changes in lipid metabolism are a hallmark of CF and increased FAO, lipid turnover in cell membranes and eicosanoid production in AMs have been reported. Furthermore, sterol regulatory-element binding protein (SREBP), a regulator of lipid homoeostasis, has been linked to CFTR loss of function¹²⁴. This results in altered plasma and tissue fatty acid profiles, and while levels of the omega-3 fatty acid docosahexaenoic acid (DHA) were unchanged in AMs upon loss of CFTR¹⁴⁵, ex vivo treatment with DHA decreased TNF-α levels¹⁴⁶. Furthermore, in CF AMs the anti-inflammatory lipoxin A₄ (LXA₄), which is synthesised from the fatty acid arachidonic acid, is reduced and the LXA₄/LTB₄ ratio in CF BAL is decreased¹⁴⁷, while the fatty acid metabolite resolvin D1 (RvD1) has been suggested as a biomarker¹⁴⁸.

Increased energy demand by AMs in CF, either by manipulation through bacterial pathogens or to fight sustained infection, results in increased utilization of all available metabolic pathways. Recently, Lara-Reyna et al. reported increased glycolysis, mitochondrial function, and production of TNF-α in CF macrophages is due to an alteration in the serine/threonine-protein kinase/endoribonuclease IRE1α pathway and this supports exacerbated inflammation¹⁴⁹. While this study used PBMCs and monocyte-derived M1 macrophages from CF patients, it would be important to detect such a mechanism in AMs and to target this pathway specifically.

Several of the above described pathways have been identified as potential drug targets in CF, however yet there are no treatments targeting AMs. Delivery of GSH to the lower respiratory tract improves the antioxidant barrier of CF epithelium¹⁵⁰, while treatment with cysteamine and restoration of MicroRNA 17 (MiR17) and MiR20 expression improves disease by restoring autophagy^151,152. Several studies administered omega-3 fatty acids (DHA/EPA) to CF patients^153–156, although only one trial reported improved FEV1 after 8 months treatment with DHA¹⁵⁷. Treatment with DHA in a murine model of CF decreased liver inflammation but did not improve lung morphology¹⁵⁸.

In conclusion, AM metabolic phenotype during CF is marked by increased energy expenditure to support exacerbated inflammation and is readily exploited by bacterial pathogens, leaving AMs deficient of oxidative burst capability during infection. It will be important to clarify the role of fatty acids in CF and furthermore, to target metabolic changes in AMs such as increased glycolysis, OXPHOS and FAO specifically to rewire AM phenotype and prevent exploitation through bacterial pathogens.

Idiopathic pulmonary fibrosis (IPF)

IPF is a chronic interstitial lung disease characterised by excessive extracellular matrix deposition in the lung parenchyma and has a particularly poor prognosis¹⁵⁹. Repetitive alveolar injury in genetically susceptible individuals causes activation of mesenchymal cells, recruitment of fibroblasts and differentiation into myofibroblasts to replace damaged alveolar epithelial cells and provide a matrix for wound healing and tissue repair¹⁶⁰. During IPF, the wound healing process is dysregulated, leading to fibrotic plaque formation and excessive build-up of extracellular matrix, resulting in impaired gas exchange. AMs have been identified as key contributors to the dysregulated wound healing process, by secreting large amounts of ROS and TGF-β¹⁶¹. Furthermore, AMs can shape the extracellular matrix by secreting factors contributing to the matrix (proline, collagen) and breaking down the matrix (plasmin, MMPs)^162–164.

Several changes to the central carbon metabolism pathways have been identified recently in AMs of IPF patients, including dysmorphic mitochondria¹⁶⁵. In murine models of pulmonary fibrosis, increased glucose consumption, glycolysis and enhanced expression of key glycolytic mediators was detected¹⁶⁶, while in IPF AMs, expression of the pulmonary glucose transporter GLUT1 was increased¹⁶⁷, which enabled augmented glucose uptake¹⁶⁶. The increased glucose uptake via GLUT1 can furthermore sustain NADPH production in the pentose phosphate pathway and TCA cycle¹⁶⁸ and is therefore a key substrate for ROS production via NOX¹⁶⁹. Activation of macrophages results in the accumulation of endogenous metabolites capable of adopting immunomodulatory roles such as succinate¹⁷⁰ and itaconate^171–173. Recently, our laboratory identified itaconate as an endogenous anti-fibrotic in the human and murine lung. In patients with IPF, there were reduced levels of airway itaconate, and decreased expression of ACOD1 (which controls the synthesis of itaconate) in AMs compared to healthy controls. Acod1 deficiency in mice leads to more severe disease pathology and exogenous itaconate limits fibroblast activity¹⁷⁴. These data indicate that AM metabolites may play a key role in the pathogenesis of lung fibrosis and may be exploited for the development of anti-fibrotic therapies.

ROS production is a key feature of AMs in IPF¹⁷⁵ and can occur during OXPHOS, by the membrane bound NOX or by reaction of hydrogen peroxide with intracellular iron¹⁷⁶. NOX, and subsequent superoxide production, is activated by binding GTP-bound Rac1¹⁷⁷, which is secreted from AMs in IPF¹⁷⁸ and can also activate the mTOR signalling hub¹⁷⁹. Superoxide produced by NOX can further react with NO to form peroxynitrite (OONO^-), another type of ROS. At the expense of NADPH, NO is produced in the mitochondria by iNOS, which is upregulated in pro-inflammatory macrophages¹⁸⁰ and in IPF-AMs leading to increased levels of the cytotoxic OONO^- in IPF AMs¹⁸¹. In the bleomycin mouse model of pulmonary fibrosis, increased levels of superoxide, NO and OONO^- were measured in AMs¹⁸². MtROS is furthermore linked to expression of PPAR-γ coactivator 1-alpha (PGC-1α), which induces metabolic reprogramming to FAO and is regulated by the mitochondrial calcium uniporter (MCU), which is increased in IPF AMs¹⁸³. MCU has furthermore been shown to regulate expression of the fatty acid transporter CPT-1, which is increased in AMs from IPF patients and bleomycin exposed mice¹⁸³. While human IPF-AMs have increased levels of MCU, mitochondrial calcium and expression of PGC-1a, bleomycin exposed mice utilise increased FAO¹⁶⁶, which is reduced in mice expressing dominant-negative MCU¹⁸³. Furthermore, these mice were protected from bleomycin induced pulmonary fibrosis. These findings highlight calcium transport and FAO as pathways to target in IPF AMs; however, a better understanding of the linking mechanism will be necessary.

IPF AMs have also been shown to be iron laden¹⁸⁴, which further induces oxidative stress and ROS production¹⁸⁵. Using RNA-sequencing, Lee et al. show furthermore, that macrophage activation is increased in iron laden AMs in IPF, suggesting that iron accumulation plays a role in macrophage activation¹⁸⁵. The proportion of AMs expressing transferrin receptor (CD71), importing transferrin bound iron into the cell, are decreased in IPF AMs, leading to an extracellular accumulation of transferrin. Furthermore, numbers of CD71-negative macrophages are an independent predictor of survival in IPF¹⁸⁶. Iron metabolism is therefore likely a key pathway in IPF-AMs and targeting it would be a viable option to decrease ROS, oxidative stress and macrophage activation.

Recently, therapies targeting metabolic processes in IPF are of considerable interest. While antioxidant therapy in IPF was promising in vivo, the double-blind placebo controlled PANTHER trial, administering either N-acetylcysteine or placebo to IPF patients for 60 weeks did not show a change in lung function parameters¹⁸⁷. Another arm of this study investigated the combined potential of corticosteroid prednisone, immunosuppressant azathioprine and N-acetylcysteine but was stopped prematurely due to increased mortality and adverse effects without evidence of benefit¹⁸⁸. Another randomized, double-blind clinical trial assessed the safety and tolerability of N-acetylcysteine in patients already receiving pirfenidone anti-fibrotic therapy. While this trial showed that N-acetylcysteine in combination with pirfenidone was safe, no change in FVC, 6-minute walk test or occurrence of adverse effects was detected¹⁸⁹. Another promising therapeutic avenue was the use of metformin, a potent metabolic remodelling drug often prescribed for type II diabetes. While on a global level metformin lowers the amount of blood sugar in diabetic patients, on a cellular level metformin activates AMP-activated protein kinase (AMPK) leading to inhibition of TGF-β induced NOX activity¹⁹⁰. Sato et al., have shown that metformin inhibited TGF-β induced NOX activity via AMPK leading to inhibition of myo-fibroblast differentiation in vitro and reduced bleomycin induced collagen deposition in vivo¹⁹¹. Consistent with this, Rangarajan et al. showed that metformin treatment reversed bleomycin induced pulmonary fibrosis via AMPK activation, while in IPF patients AMPK phosphorylation was decreased¹⁹². A posthoc analysis study of the effect of metformin in IPF patients however showed no change in clinically outcomes¹⁹³, once again showing the difficulty of translating in vitro and in vivo findings into the clinic. Another study investigating the NOX-NRF2 imbalance as a therapeutic target showed that in vivo knockdown of NOX4 and NOX1/4 inhibition restored the capacity of fibrosis resolution in aged mice¹⁹⁴. Furthermore, treatment with nitrated fatty acids, reversed pulmonary fibrosis in a mouse model by promoting collagen uptake by AMs and dedifferentiating myofibroblasts¹⁹⁵.

While these treatment approaches targeted metabolic changes during pulmonary fibrosis, none was specific to AMs. Targeting macrophage specific metabolic reprogramming, which sustains ROS and TGF-β production and contributes to dysregulated wound healing in IPF would therefore be a promising approach.

Respiratory tract infections

During respiratory tract infections, activation of pattern recognition receptors expressed by AM can elicit a variety of pro-inflammatory host responses². For example, severe Coronavirus disease-19 (COVID-19) associated pneumonia patients may exhibit features of systemic hyper-inflammation also known as macrophage activation syndrome or “cytokine storm” which is associated with sustained elevation of macrophage/monocyte-derived pro-inflammatory cytokines (e.g., IL-6, IL-8, TNF-α, IL-1β) leading to acute respiratory distress syndrome (ARDS)^196–198. Using single cell approaches a recent study demonstrated that highly inflammatory, monocyte recruited AMs, rather than quiescent pulmonary resident AMs, predominate in the BAL in COVID-19 patients with severe pathology, implicating these cells in COVID-19-associated ARDS¹⁹⁹. Rather than direct infection of AMs, AM:AEC cross-talk has been identified as a major mechanism for control of many respiratory viral infections²⁰⁰ and AEC have been shown to be a key source of pro-inflammatory cytokines, modulating AM phenotype^198,201. For example, Rhinovirus (RV), the causative agent of the common cold, primarily infects the upper airways, however prior infection with RV attenuates subsequent AM antibacterial responses²⁰². Although AMs are susceptible to influenza A viral infection (IAV), replication within AMs has been shown to be minimal with the exception of several highly virulent strains^203–205. Here, we will focus on Mycobacterium tuberculosis (Mtb) infection, as AMs are the primary infected cell type and metabolic changes in response to Mtb infection are well studied.

Tuberculosis

Tuberculosis (TB) is a contagious, chronic disease and one-third of the world’s population is infected with Mtb, the causative agent of TB, resulting in ~2 million deaths per year (2009 World Health Organization Report)²⁰⁶. During infection, Mtb colonises AMs intracellularly and disables innate intracellular defence mechanisms such as the phagolysosome and inflammasome and accesses macrophage intracellular nutrients²⁰⁷. AM host defence mechanisms against Mtb include production of ROS and reactive nitrogen species (RNS) for bacterial killing and fusing mycobacteria-containing phagosomes with lysosomes as well as autophagy and apoptosis²⁰⁸. However, virulent or multi-drug resistant strains can evade these host responses e.g. by preventing phagolysosome fusion and surviving ROS/RNS²⁰⁹.

During Mtb infection, AMs shift their metabolic programme from OXPHOS to aerobic glycolysis, which is regulated by HIF1α and interferon-gamma (IFN-γ). This metabolic shift and subsequent enhanced glycolytic flux in infected AMs is crucial to control infection. Mice lacking HIF1α in the myeloid lineage are more susceptible to infection and show decreased cytokine and antimicrobial effector production²¹⁰. To support this metabolic reprogramming, key glycolysis genes are upregulated in the early stages of granuloma formation in mice, supporting the shift towards aerobic glycolysis²¹¹. Blocking this shift on the other hand results in decreased levels of IL-1β, increased IL-10 and subsequent increased bacterial survival²¹². Gleeson et al. suggest furthermore that infection-induced glycolysis limits Mtb survival through the induction of IL-1β during infection with drug susceptible Mtb, as absence or inhibition of the IL-1 receptor (IL-1R) negated the effect of aerobic glycolysis²¹². In contrast, during infection with multi-drug resistant Mtb, AM metabolic reprogramming and induction of glycolysis is regulated through IFN-β as overexpression of Mtb cell wall lipids blocks the IL-1 receptor type 1 pathway²¹³. These findings suggest that infection-induced glycolysis is necessary for control of bacterial intracellular replication though it remains to be investigated whether glycolytic reprogramming could be further induced to support AM defence.

M. Tuberculosis has developed several mechanisms to evade host defence. Using mice deficient of cystathionine-gamma-lyase (CSE), which catalyses the synthesis of hydrogen sulphide (H₂S), H₂S has been identified as a regulator for central carbon metabolism during Mtb infection in AM. CSE^−/− mice had increased flux through the glycolysis and pentose phosphate pathway in AM, while Mtb infected WT mice produced increased levels of H₂S, reducing HIF1α levels, glycolysis and host defence²¹⁴. Howard et al. show that multi-drug resistant Mtb can drive augmented cell wall lipids synthesis, thereby bypassing the IL-1 receptor pathway and resulting in induction of IFN-β signalling, reprogramming host metabolism²¹³. Furthermore, increased expression of MiR-21 in BMDMs and human MDMs upon exposure with Mtb, resulted in decreased glycolytic response and facilitated bacterial survival by targeting phosphofructokinase isoform M (PFK-M) and limiting IL-1β production. IFN-γ however inhibits MiR-21, forcing an isozyme switch in the PFK-M complex and rescuing glycolysis and host defence²¹⁵.

Several signalling metabolites have been identified to be important for host defence in AMs during Mtb infection. Decreased levels of fumarase result in accumulation of bactericidal fumarate, which can modify metabolites and proteins through succination²¹⁶. Furthermore, the antimicrobial metabolite itaconate is increased during Mtb infection, although it can be disabled by β-hydroxylacyl CoA lyase in Mtb and used as a nutrient. Wang et al. show that deletion of this enzyme resulted in attenuated Mtb infection in mice²¹⁷. Similarly, Mtb exploit intracellular iron as a nutrient. This is highlighted by worse TB outcome with increased dietary ingestion of iron and inhibition of Mtb growth when iron is unavailable^218,219. Abreu et al. show that heparin reduced hepcidin expression in macrophages infected with Mtb while heparin-treated macrophages had increased expression of ferroportin and subsequent iron export, limiting iron availability for intracellular bacilli²⁰⁶. Mtb furthermore induce ferroptosis, associated with reduced levels of GSH, superoxide and increased free iron. The ferroptosis inhibitor ferrostatin-1 (Fer-1) as well as iron chelation decreased necrotic cell death of Mtb-infected macrophages in vitro, while in vivo treatment with Fer-1 reduced bacterial load²²⁰. Mtb can cope in low iron environments however by downregulating their non-essential protein content via specific sRNA²²¹.

Several changes in fatty acid metabolism of Mtb infected AMs were identified recently. Compared to interstitial macrophages during Mtb infection, which are reliant on glycolysis, AMs utilise FA, which is induced by PPAR-α²²² and have a lower burden of Mtb infection²²³. To escape host defence, Mtb has developed a mechanism inhibiting pathways related to autophagy, lysosomal function and FAO in support of replication by inducing microRNA-33 (MiR-33) in the host cell. Silencing of MiR-33 however induced AM lipid catabolism and autophagy and rescued host defence²²⁴. Furthermore, amino acid metabolism is altered during Mtb infection. In mice and macaque lungs, indoleamine 2,3-dioxygenase (IDO), which is involved in tryptophan catabolism, was increased during Mtb infection, while inhibition of IDO in a macaque model of TB decreased bacterial burden and pathology, as tryptophan metabolites suppress host immunity²²⁵.

While Mtb relies on host lipids as energy source, existing therapies such as targeting PPAR transcription factors or cholesterol synthesis have been successful mainly in animal models^226–229, whereas retrospective human studies, which investigated the effect of statins in diabetic TB patients did not show any results²³⁰. As Mtb can also utilise iron as a substrate, another approach is to prevent iron accumulation. Treatment of Mtb infected human MDMs and primary AM with iron chelator Desferrioxamine (DFX) ex vivo induced the expression of glycolytic enzymes and enhanced glycolysis, as well as IL-1βα, thereby supporting host defence²³¹ and offers a novel therapeutic approach, which will need to be investigated in clinical trials. Together, these findings highlight the distinct phenotype of AMs during Mtb infection, which counteracts intracellular infection through aerobic glycolysis, but is also heavily exploited by Mtb bacteria feeding on host lipids and iron.

Targeting metabolism during chronic lung disease

Many potential targets have been identified recently that could rewire macrophage metabolic and phenotypic changes driving chronic lung disease. Since all cells depend on oxidative phosphorylation or cytoplasmic glycolysis to synthesize ATP, there is the potential for unwanted side effects by targeting specific metabolic processes. However, it is becoming increasingly apparent that it is possible to safely target metabolic pathways in patients. For example, dimethyl fumarate, a known regulator of macrophage phenotype, is a first-line-treatment for relapsing-remitting multiple sclerosis²³². Indeed, metabolic processes are highly plastic with significant redundancy, modulation of these processes may have the added benefit of selectively targeting cells with high metabolic demands²³³. Targeted delivery to AMs may add another layer of selectivity, improving efficacy, sustained drug release and evading capture by mucus²³⁴. Systems for inhaled AM targeted drug delivery include the use of micro- and nanocarriers, including liposomes, which are phagocytosed by AMs. Rifampicin-loaded microspheres as a therapeutic approach for Mtb have been described²³⁵, and have been further refined to allow a one-step assembly for rifampicin containing microspheres²³⁶. Recently, aerosolised delivery of siRNA, which post-translationally downregulates gene expression, has been developed to target AMs specifically²³⁷, whilst mannose coated microspheres have been developed which exploit the phagocytotic activity of AMs²³⁸. Many of these delivery vehicles have been developed to transport antibiotics targeting intracellular AM bacterial infections, which are helpful for treating TB, however other drugs could be incorporated into aerosolised micro- or nano delivery systems. Specifically, treatment with iron chelators, antioxidants and nitrated fatty acids has shown to rewire AM phenotype and improve diverse chronic lung disease; these may be ideal candidates to develop novel, aerosolised vehicle-assisted drug delivery to AMs during chronic lung disease.

Conclusion

In the last decade enormous strides have been made regarding our understanding of how adaptations in metabolic pathways underlie macrophage phenotype and function. AMs are remarkably plastic cells, orchestrating not only pathogen defence and efferocytosis, but also pulmonary tolerance and resolution. It has become increasingly clear that AMs tailor their metabolic profile to fit their local niche generating ROS for pathogen defence, utilising aerobic glycolysis to rapidly generate cytokines, employing the TCA cycle to fuel inflammatory responses and generating metabolites with secondary signalling functions such as citrate, itaconate, succinate and fumarate. Work elucidating the complexities of AM metabolic alterations in the context of CLDs has highlighted many potential therapeutic targets (summarized in Table 1). Indeed, a lack of understanding of shared cellular mechanisms, which underlie CLDs has been a major obstacle in respiratory biology; identification of common AM-metabolic pathways/metabolites which directly influence core features of CLDs would be a significant advance on the route to devising new AM-directed strategies to treat pulmonary diseases which affect millions worldwide.

Table 1.

Specific metabolic alterations in AMs during chronic lung disease^a.

Disease	Metabolic change	Functional change	Ref.
Asthma	↑ HO-1	ROS production	83
	↑ ROS	Lung injury, TNF-α, IL-1β production	80,82
	↑ 15-LOX	5-HETE and leukotrienes	239
	↑ Leukotriene B4/E4	Bronchial constriction, AHR	85–87
	↑ Prostaglandin E2	↓ AM phagocytosis	78
	↑ FAO gene expression (CPT)	↑ FAO metabolism	84
	↑ Glycolysis function PKM2	HIF1a dependent gene expression
	↑ Arg1, ↓ NOS2	Ornithine and proline	89,90
COPD	↑ ROS, ↑ mtROS, ↑ Superoxide	Oxidative stress	108–110
	↓ mtROS after challenge	Impaired bacterial clearance	108
	↓ Glutamyl cysteine ligase	Loss of GSH synthesis	111
	↑ iNOS	NO production	117
	↓ Mt membrane potential	Impaired phagocytosis	114
	↑ Iron sequestration	ROS generation	113
	↑ HIF1α	↑ Glycolysis	119
	↓ Compensatory glycolysis	Unmet energetic demand	7
	↓ Coupling efficiency, ↓ OXPHOS ↑ Proton leak	Dysfunctional metabolism and macrophage phenotype	7
	↑ A2BR	↑ Adenosine metabolism	118
CF	↑ Arginase	Impaired phagocytosis & efferocytosis	132
	↑ ROS release, ↓ GSH	Oxidative stress	136,137
	↓ Superoxide, ↓ NOX phos.	↓ Oxidative burst	141
	↓ Lipoxin A4	↓ Anti-inflammatory potential	147
	↑ Acod11	Persistent P. Aeruginosa infection	143
	↓ CFTR-PTEN complex	ROS production, succinate release	144
	↑ Iron	↑ ROS production	138
	↑ IRE-a pathway	Increased glycolysis & Mt. function	149
IPF	↑ ROS	Oxidative stress	175
	↑ Secreted Rac1	NOX, superoxide, mTOR activation	178,179
	↑ iNOS	NO and OONO- production	181,182
	↓ HO-1	Loss of oxidative response	240
	↑ Iron uptake	Oxidative stress & ROS	184,185
	↓ % of CD71+ AM	Accumulation of transferrin	186
	↑ AKT	Activation of HIF1α	241
	↑ GLUT-1, ↑ Glucose uptake	NADPH production, superoxide	166–168
	↑ Glycolysis, glycolysis genes	M2-like AM profile	166
	↑ FAO		166
	↑ MCU, ↑ Calcium, ↑ PGC-1α	FAO reprogramming, mtROS
TB	↑ ROS/RNS production	Bacterial killing	208
	↑ HIF1a	Aerobic glycolysis, IFN-γ host defence	210
	Warburg shift, ↑ glycolysis genes	↑ IL-1β and bacterial killing	211,215
	↑ Host MiR-21	↓ Glycolytic response, ↑ bacterial survival	215
	↑ Host MiR-33	↓ autophagy, FAO, ↓ host defence	224
	↓ Hydrogen sulphide	↑ Glycolysis & PPP	214
	↑ fumarate	Bactericidal	216
	↑ itaconate	Antimicrobial, modulates host response	217
	Heparin ↓ Hepcidin	Decreased iron availability to bacilli	206
	↑ Iron, superoxide, lipid perox.	Ferroptosis, bacterial spread	220
	↑ FAO, ↑ Lipid accumulation	↑Host response, ↓ Bacterial burden	223,242
	PPAR-α activation	FAO, autophagy & host defence	222
	↑ IDO, ↑ tryptophan	↓ host defence	225
	↑ Glutaminolysis	↑ cytokine profile	243
	↑ Arg1	NO production	244
	↑ NAD+, creatine, GSH	Host defence	245

^aUpward arrow represents increased expression, downward arrow represents decreased expression.

Author contributions

P.P.O. and A.J.B. conceived of and wrote the manuscript.

Competing interests

The authors declare no competing interests.