Central metabolic interactions of immune cells and microbes: prospects for defeating infections

Abstract

Antimicrobial drug resistance is threatening to take us to the “pre‐antibiotic era”, where people are dying from preventable and treatable diseases and the risk of hospital‐associated infections compromises the success of surgery and cancer treatments. Development of new antibiotics is slow, and alternative approaches to control infections have emerged based on insights into metabolic pathways in host–microbe interactions. Central carbon metabolism of immune cells is pivotal in mounting an effective response to invading pathogens, not only to meet energy requirements, but to directly activate antimicrobial responses. Microbes are not passive players here—they remodel their metabolism to survive and grow in host environments. Sometimes, microbes might even benefit from the metabolic reprogramming of immune cells, and pathogens such as Candida albicans, Salmonella Typhimurium and Staphylococcus aureus can compete with activated host cells for sugars that are needed for essential metabolic pathways linked to inflammatory processes. Here, we discuss how metabolic interactions between innate immune cells and microbes determine their survival during infection, and ways in which metabolism could be manipulated as a therapeutic strategy.

Glossary

AA

amino acids

ADP

adenosine diphosphate

ATP

adenosine triphosphate

BCG

bacillus Calmette–Guérin (vaccine)

cGAS

GMP‐AMP synthase

CoA

coenzyme A

ER

endoplasmatic reticulum

ETC

electron transport chain

FA

fatty acids

GABA

γ‐aminobutyric acid

GAL4

galactose‐induced gene 4

GFP

green fluorescent protein

GlcN

glucosamine

HDACi

histone deacetylase inhibitors

HIF

hypoxia‐inducible factor

ICD

isocitrate dehydrogenase

ICL

isocitrate lyase

IDH

isocitrate dehydrogenase

IDO

indoleamine 2,3‐dioxygenase

IFN

interferon

IL

interleukin

IRF

interferon regulatory factor

IRG

immune‐responsive gene

KDM

lysine‐specific demethylase

LPS

lipopolysaccharide

MPLA

monophosphoryl lipid A

mROS

mitochondrial reactive oxygen species

MS

malate synthase

mTOR

mechanistic target of rapamycin

NAD

nicotinamide adenine dinucleotide

NADPH

nicotinamide adenine dinucleotide phosphate

NLRP

NACHT, LRR and PYD domain‐containing protein

NMN

nicotinamide mononucleotide

NO

nitric oxide

NOS

nitric oxide synthase

Nrf2

nuclear factor erythroid 2‐related factor 2

OXPHOS

oxidative phosphorylation

PARP

poly(ADP‐ribose) polymerase

PBMCs

peripheral blood mononuclear cells

PFK

phosphofructokinase

PPAR

peroxisome proliferator‐activated receptors

PYK

pyruvate kinase

QPRT

quinolinate phosphoribosyltransferase

ROS

reactive oxygen species

SDH

succinate dehydrogenase

SIRT

sirtuin (silent mating type information regulation 2 homolog)

STING

stimulator of interferon genes

TCA cycle

tricarboxylic acid cycle

TF

transcription factor

TLR

Toll‐like receptor

UDP‐GlcNAc

uridine diphosphate N‐acetyl glucosamine

Challenges in treating infectious diseases

The discovery of antibiotics in the 20^th century is often heralded as one of the major, if not the biggest medical breakthrough in history. Antibiotics save lives from common infections and support medical interventions that require invasive and immunosuppressive approaches, such as surgery, organ and stem cell transplantation and cancer chemotherapy. Unfortunately, less than a 100 years later, we are again in a crisis situation 1. We have overused and misused antibiotics and antifungals not only to treat humans, but also to support food production of animals and plants for our ever‐increasing population. Microbes can rapidly increase cell populations and possess adaptive mechanisms, such as genome rearrangements that promote drug resistance 2 and the appearance of so‐called “mutator” strains 3, 4. This means that the evolution of antimicrobial drug resistance can be fast in all classes of pathogens (bacteria, fungi and parasites) 1, 5, 6, 7. Development of new antimicrobial drugs has not been particularly successful, and pharmaceutical companies are pulling out from developing new antimicrobial agents because of reduced profit margins. A 2016 review sponsored by the UK Department of Health and the Wellcome Trust suggested that in the next three decades, deaths from drug‐resistant infections could be 10 million per year and thereby surpass deaths from cancer 8. While this assertion has been disputed 9, it is nevertheless clear that the threat from infectious diseases is not decreasing, but rather it is increasing. This is not just a problem of the developing world, but concerns all countries where opportunistic infections threaten many procedures of modern medicine, which we take for granted, and which make us feel that we can conquer any disease. Given that developing new antibiotics is challenging, we need to consider diverse facets of the host–microbe interface, to hopefully develop a range of complementary strategies to manipulate this interaction to benefit the host at the expense of the pathogen.

Developing alternative strategies to manage infections critically depends on understanding both sides of the host–microbe interaction. Antimicrobial immunity on the one hand, and immune evasion by microbes on the other, rely on appropriate regulation of transcriptional and signalling networks. These are in turn driven by chemical and physical principles and interactions. Therefore, not surprisingly, it has transpired that host–pathogen interactions are fundamentally regulated by an interplay between host and microbial metabolic pathways and the levels of metabolites in infection microenvironments. Humans face microbes not only in infection but, as is true for any other multicellular organism, they have evolved to coexist commensally with bacteria, fungi or parasites. This is likely to have shaped the metabolic adaptations of host and microbes alike. Commensal microbes generate nutrients essential to maintain human health, but also deplete nutrients to suppress growth of pathogenic microbes. Microbial metabolites also control immunity, by being sensed and metabolised by immune cells. Despite these beneficial interactions with microbiota, commensals and also bona fide pathogens can cause bacteraemia, fungaemia, sepsis, organ failure and mortality. Therefore, a functional epithelial layer and immunity are critically important for maintaining the commensal state of microbiota and defending us from infections. As we will review here, recent advances gave birth to a field of research commonly referred to as “immunometabolism”, which aims to understand how responses of innate immune phagocytes are shaped by cellular metabolism, metabolites and available nutrients. The metabolic adaptations are particularly evident in innate immune cells as they rapidly respond to changing environments, but metabolism also controls responses to infections in T cells and epithelial cells 10. As a counterpart to immunometabolism, microbes reshape their metabolic pathways to match host environments, and could also be employing metabolic tactics to counter immune attack by competing with immune cells for nutrients that are essential for effective antimicrobial responses. Here, we discuss what is known about the reprogramming of host and pathogen metabolic pathways in infection, the plasticity (or otherwise) of host and pathogen metabolism, and how the metabolic interplay controls the host–pathogen interaction. We will focus predominantly on macrophages, as they have been studied extensively in the context of microbial interactions, but several of the mechanisms are also relevant to other immune and non‐immune cell types. Finally, we will assess the potential of nutritional approaches and metabolic manipulations in aiding anti‐infective therapy.

Metabolism of macrophages

Macrophages come in different forms reflecting their diverse roles during infections, inflammation, wound repair and tissue homeostasis. While each macrophage is likely able to fulfil many, if not all, of these roles, tissue‐resident and recruited macrophages, as well as blood monocytes and neutrophils that accumulate at sites of infection and sterile inflammation, employ distinct immune responses. Macrophage responses may also differ within the same cell population due to concentration gradients of cytokines and pathogens, as well as between different innate immune cells, as all tissue‐resident macrophages likely originate from yolk sac and foetal liver progenitor cells, and are thus distinct from monocyte differentiated macrophages that originate from bone haematopoietic stem cells 11, 12, 13, 14. The former population is believed to be long‐lived and self‐renewing, suggesting different metabolic capabilities. The surrounding tissue cells are also important for macrophage development and function 15. The inflammatory milieu and microbial products further drive the differentiation of macrophages, which impacts their metabolism. For instance, the cytokines IL‐4 and IFN‐γ have long been used to characterise macrophage populations with distinct functions, which are best characterised by their metabolic state. In general, IL‐4 promotes amino acid metabolism via mitochondrial respiration and fatty acid oxidation and upregulates arginase‐1 to generate ornithine and urea for polyamine biosynthesis 16. In contrast, IFN‐γ‐treated mouse macrophages utilise exogenous arginine to generate nitric oxide (NO) radicals via NOS2, thereby inhibiting arginase‐1 activity 17. The bacterial cell surface component lipopolysaccharide (LPS) and IFN‐γ cause a general reduction in amino acid metabolism due to reduced mitochondrial respiration and the upregulation of catabolic enzymes, such as IDO1 that depletes tryptophan levels. Under these conditions, increased glucose import enhances glycolysis and the pentose phosphate pathway to generate energy (ATP) and oxygen radicals (ROS) via the NADPH oxidase. These metabolic responses promote antimicrobial activities: depletion of tryptophan and the generation of ROS prevent the growth of intracellular microbes, which are dependent on host‐derived amino acids and susceptible to ROS. Based on these observations, changes in metabolism were thought to support already committed macrophages to fulfil their diverse functions during infections and tissue homeostasis 17.

More recent studies, however, have demonstrated that the metabolic reprogramming that occurs after LPS exposure not only supports, but directly controls and enables immune responses (Fig 1), as inhibition of glycolysis prevents the expression of the pro‐inflammatory cytokine IL‐1β in primary mouse macrophages 18. This result with LPS was validated with bacterial infection of macrophages with Bordetella pertussis 18. Upregulation of glycolysis in LPS‐activated macrophages leads to the secretion of lactate rather than fuelling mitochondrial respiration despite the presence of oxygen 19. This resembles metabolic change in cancer cells and is commonly referred to as Warburg metabolism, which is utilised by rapidly dividing cells or immune cells that need to quickly respond to environmental cues 20. This is in contrast to naïve macrophages, which utilise both glycolysis and mitochondrial metabolism, whereby the latter generates important metabolites for amino acid and fatty acid synthesis, as well as ATP via oxidative phosphorylation (OXPHOS) via the electron transport chain (ETC) 21, 22.

Figure 1. Metabolic reprogramming of microbes and macrophages in infection.

When challenged with stimuli such as the bacterial ligand LPS, IFN‐γ, the bacterium M. tuberculosis and the fungus C. albicans, macrophages remodel their metabolism so that glucose import and glycolysis are enhanced, and the TCA cycle is compromised. Increased levels of TCA intermediates citrate and succinate play direct roles in promoting antimicrobial responses. Citrate is a precursor of itaconate, an inhibitor of microbial glyoxylate cycle enzyme isocitrate lyase (ICL) and host succinate dehydrogenase (SDH), while succinate stabilises the transcriptional activator of glucose utilisation genes and antimicrobial cytokines, HIF‐1α. Similarly, microbes adapt their metabolism by transcriptionally upregulating the glyoxylate shunt pathway, glycolysis and gluconeogenesis at distinct stages of infection. Reprogramming of microbial metabolism depends on available nutrients, including host‐derived lactate and itaconate. Mitochondria and nucleus (labelled orange and blue, respectively) are present in fungi, but not in bacteria. In fungi, the glyoxylate cycle reactions are further compartmentalised in the peroxisome (indicated by yellow shading). Metabolites and enzymes up‐ and down‐regulated during macrophage–microbe interaction are shown in red and purple, respectively. HIF‐1α, hypoxia‐inducible factor‐1α; ICL, isocitrate lyase; ICT/CCL, succinyl‐CoA:itaconate CoA transferase/(S)‐citramalyl‐CoA lyase; IDH, isocitrate dehydrogenase; IRG1, immune‐responsive gene 1; MS, malate synthase; SDH, succinate dehydrogenase; TF, transcription factor.

After LPS exposure of macrophages, the carbon flux through the TCA cycle is disrupted due to downregulation of isocitrate dehydrogenase (IDH1) and higher levels of the citrate carrier in mitochondria resulting in the accumulation of citrate 21, 23, 24. Increased levels of citrate fuel fatty acid synthesis to generate inflammatory prostaglandins and membrane lipids to promote cytokine secretion 25, 26. In addition, LPS‐activated macrophages upregulate immune‐responsive gene IRG1, which converts citrate into itaconate 21. Itaconate has antimicrobial effects by inhibiting the microbial glyoxylate shunt that generates essential metabolites for intracellular survival of bacteria and fungi 27. Itaconate also regulates host metabolic and immune responses by inhibiting the TCA enzyme succinate dehydrogenase (SDH), and activating transcription factor Nrf2 to control inflammation 28, 29, 30, 31, 32. The increased level of succinate stabilises transcription factor HIF‐1α resulting in the upregulation of IL‐1β 18, 33, 34. The metabolic changes in inflammatory macrophages, and their consequences for antimicrobial responses, are summarised in Fig 1.

Oxidation of succinate via SDH further triggers inflammatory responses via the generation of mitochondrial ROS (mROS), likely due to reverse electron transport. mROS has multiple functions. Firstly, mROS is antimicrobial, delivered by mitochondria that traffic to bacteria‐containing phagosomes 35, 36. Secondly, mROS further stabilises HIF‐1α, promoting inflammation 33. Thirdly, mROS may activate the NLRP3 inflammasome 37. Finally, excessive mROS may lead to irreversible mitochondria damage resulting in the release of DNA, which can lead to NLRP3 and cGAS/STING‐mediated inflammation 38, 39, 40, 41. LPS treatment has also recently been shown to cause depletion of NAD⁺ due to increased activity of NAD‐consuming pathways, such poly(ADP‐ribose) biosynthesis by PARPs and sirtuins, and downregulation of the NAD biosynthesis gene QPRT, despite upregulation of the initial enzyme IDO1 42. Loss of cellular NAD⁺ and mitochondrial deacetylase activity of SIRT3 results in the acetylation and inhibition of complex I, thereby reducing oxidative phosphorylation, and triggering increased pro‐inflammatory responses in macrophages and mice challenged with LPS 42. We note that many of the studies that we discussed used LPS stimulation as a proxy for bacterial infection, with validation with whole bacteria rarely performed. Nevertheless, challenge of primary mouse macrophages and human peripheral blood mononuclear cells (PBMCs) with the yeast pathogen Candida albicans resulted in similar metabolic changes to those seen with LPS 43, 44. However, differences compared to LPS have been reported in other cases. Specifically, challenge of human monocytes with the synthetic TLR2 receptor ligand Pam3CSK4, or whole‐cell lysates prepared from the bacterium Mycobacterium tuberculosis, Staphylococcus aureus or Escherichia coli resulted in upregulation of both glycolysis and mitochondrial oxidative phosphorylation 45. Similarly, both glycolysis and oxidative phosphorylation were increased (while fatty oxidation genes were mostly repressed) in human blood samples from people suffering from bacterial or fungal infection 43. Future experiments using live infections of immune cells with diverse microbial pathogens should broaden our understanding of which metabolic changes in innate immune cells are common, and which are specific to distinct infection agents.

In contrast to the situation described above for LPS and IFN‐γ, IL‐4‐mediated transcriptional programmes increase mitochondrial metabolism and respiration in macrophages to promote wound healing mechanisms 17. For instance, inhibiting the reprogramming of mitochondrial respiration prevents anti‐inflammatory phenotype and the upregulation of arginase‐1, which is required to increase proline levels for collagen synthesis to support the extracellular matrix in wound repair 46, 47. Increased rates of fatty acid oxidation and mitochondrial respiration may directly drive IL‐4 activation of macrophages 46, 48, 49. In addition, IL‐4 treatment results in increased hexosamine biosynthesis, generating UDP‐GlcNAc, which is essential for N‐glycosylation of cell surface receptors that are upregulated in these macrophages 21. Hexosamine biosynthesis depends on glucose and glutamine utilisation, and both metabolites are critical to induce the expression of IL‐4‐dependent genes 21, 48, 50. Besides hexosamine biosynthesis, glucose also feeds into glycolysis, which is upregulated in IL‐4‐activated macrophages depending on the transcription factor IRF4 48, 50. It is thought that increased glycolysis enables fatty acid biosynthesis, which is upregulated in IL‐4‐activated macrophages to promote oxidative metabolism. Inhibition of glycolysis and fatty acid biosynthesis prevents the expression of several IL‐4‐dependent markers 48. However, more recent genetic studies suggest that coenzyme A (CoA) homeostasis, rather than mitochondrial respiration, controls IL‐4 activation of macrophages 51. CoA is involved in many cellular processes, ranging from fatty acid synthesis to post translational modification of histones and other proteins. Treatment with IL‐4 causes accumulation of acetyl‐CoA partly due to citrate cleavage and increased uptake of glucose, glutamine and fatty acids, which are catabolised to acetyl‐CoA 50. Increased acetyl‐CoA levels promote histone acetylation, mediating IL‐4‐dependent immune responses and arginase‐1 expression 50.

IL‐4‐treated macrophages remain metabolically flexible, as they can utilise fatty acids or glucose to fulfil their roles, likely because IRF4‐dependent transcription increases mitochondrial metabolism and glycolysis. As such, IL‐4‐treated macrophages can be reprogrammed by LPS/IFN‐γ to express inflammatory markers 49. In contrast, LPS‐ and IFN‐γ‐treated macrophages primarily utilise glucose to fuel metabolic pathways and immune responses, are metabolically inflexible and fail to respond to IL‐4 48, 49.

Immune cell metabolism in sepsis patients

Blood‐derived immune cells of sepsis patients that are exposed to circulating LPS and other microbial products provide unique opportunities to study the role of metabolic reprogramming during and after human infections. Transcriptional profiling of blood‐derived lymphocytes isolated from sepsis patients with acute hyper‐inflammation or LPS‐induced experimental endotoxemia showed differential expression of glycolysis and mitochondrial respiration genes, with either up‐ or downregulation of these pathways detected relative to healthy people depending on the condition 43. Several metabolites are increased in the blood of sepsis patients, including lactate, raising the possibility that the metabolic reprogramming of leucocytes during sepsis contributes to the hyperlactaemia 52. Hyperglycaemia is also strongly associated with severe sepsis, due to insulin resistance and impaired degradation of the insulin receptor 53. Increased blood glucose levels trigger glucose transporter‐mediated transcriptional reprogramming of epithelial cells, which causes loss of gut barrier integrity and infections 54. Whether increased blood glucose levels are critical to support inflammatory responses of activated macrophages and monocytes remains unknown. The plasma concentrations of most amino acids are altered in sepsis patients compared to healthy controls and may indicate disease severity 52. For instance, glutamine levels are increased, and may fuel succinate synthesis via the GABA shunt and thus promote inflammatory responses 55. Also, sepsis patients show increased arginine levels that may support NO production via NOS2 in inflammatory macrophages. Compared to survivors, arginine levels were reduced in sepsis patients that failed to recover 55. Similarly, several plasma lipid levels, particularly carnitine esters, are increased in sepsis non‐survivors relative to controls, together with decreased expression of fatty acid transporters 56. This likely reflects defects in mitochondrial import of fatty acids, which requires carnitine palmitoyltransferase and reduced rates of β‐oxidation. Consistent with this, the expression of peroxisome proliferator‐activated receptors (PPAR) α, β and δ, which regulate several aspects of β‐oxidation, is reduced in severe sepsis patients 57, 58. Collectively, these data support in vitro observations in LPS‐treated macrophages showing that increased glycolysis fuels fatty acid synthesis. Conversely, uptake of fatty acid and their catabolism via oxidation into acetate may further enable the accumulation of citrate and succinate in the TCA cycle to drive inflammatory responses. Of note and as mentioned previously, the reduction of mitochondrial respiration, as occurs in LPS‐treated macrophages, is not always observed in sepsis patients 43, 52, possibly reflecting the infection site and/or microbes involved, as in vitro experiments showed that activation with ligands such as Pam3CSK4, E. coli and S. aureus does not induce a reduction of mitochondrial respiration in monocytes 45.

Current efforts are aimed at identifying the role of individual metabolites in promoting survival of sepsis patients, and how this knowledge could be applied to manipulate metabolism for treatment. The role of metabolism to control macrophage immune responses, however, has largely been studied in ex vivo immune cells or cell lines treated with defined stimuli in rich culture media. Determining the metabolism of tissue‐resident macrophages and monocyte‐derived macrophages within infection microenvironments remains technically challenging. It is equally difficult to quantify the levels of key immune‐related metabolite levels in these niches. While inflammatory responses are critically important to eliminate microbes during sepsis, excessive inflammation contributes to tissue damage, dissemination and increased mortality. As such, timing of nutrient supplementation during sepsis may be critical. Inflammatory macrophages also secrete cytokines to dampen inflammation, including IL‐4 59. Consequently, tissue macrophages may not strictly adhere to an IFN‐γ/IL‐4 dichotomy, but rather follow a broad spectrum of activation states 60. This warrants further studies to determine the metabolism of other activated macrophages and individual lymphocytes isolated from sepsis patients 45.

Immune cell tolerance and paralysis during severe sepsis

In severe sepsis patients, macrophages and monocytes often fail to respond appropriately to microbial and inflammatory signals, resulting in an immune tolerant state that can develop into immune paralysis. The absence of a functional innate immune system may lead to increased infection and mortality rates commonly observed in severe sepsis patients. The underlying mechanisms that lead to immune tolerance, however, remain poorly understood. In vitro studies have shown that LPS‐treated macrophages fail to respond to subsequent LPS exposure, and this is likely driven by epigenetic modifications of histones and changes to gene transcription. In particular, prolonged LPS exposure of macrophages increases cellular levels of NAD⁺, and NAD⁺‐dependent sirtuins (SIRT1 and SIRT6) downregulate glycolysis and increase β‐oxidation 61. SIRT6 targets HIF‐1α promoter sites to decrease acetylation of lysine 9 of histone H3 (H3K9Ac), causing reduced expression of genes encoding glycolytic enzymes 62.

NAD⁺ levels are also increased in immunotolerant monocytes from sepsis patients 43. Here, epigenetic changes at H3K9 similarly correlate with reduced rates of glycolysis in immunotolerant monocytes 43. Tolerant monocytes show reduced mitochondrial and fatty acid metabolism, further compounding their inability to respond to LPS. Nevertheless, immunotolerant monocytes secrete 10‐fold more IL‐10 compared to LPS‐treated control cells 43. IL‐10 is thought to contribute to immunotolerance in macrophages by inhibiting glycolysis and inducing the degradation of damaged mitochondria (mitophagy) via mTOR signalling 63. Loss of IL‐10 increases inflammation, partly due to activation of the NLRP3 inflammasome mediated by increased glycolysis and mitochondrial ROS 63. IL‐10 secretion, as well as defects in metabolism and immune response, is reversed in patients who have recovered from severe sepsis 43. The molecular pathways involved in the resolution of immunotolerance, however, remain to be described.

Metabolic aspects of trained immunity

Immunotolerance likely acts as a mechanism to dampen excessive inflammation and appears to be chiefly mediated by reducing leucocyte metabolism. In a contrasting scenario, innate immune cells can be reprogrammed to respond to a secondary signal more quickly and vigorously compared to the first time. This suggests that innate immune cells are capable of memory, which is referred to as “innate immune memory” or “trained immunity” 64. Trained immunity is not restricted to a particular stimulus, and it confers cross‐protection against diverse infections 65. Similar to macrophages activated by the bacterial cell surface component LPS, the cell wall component from fungal cells, β‐glucan, increases macrophage glycolysis and reduces mitochondrial respiration. Inhibiting glycolysis reduces macrophage immune responses to β‐glucan, as described for LPS 66. In contrast, however, β‐glucan‐mediated metabolic reprogramming lasts for weeks and months, whereas LPS activation leads to metabolic dysfunction and immune tolerance within 6 days 66, 67. It is thought that these different metabolic responses are due to changes in the epigenetic state of chromatin and therefore gene expression, including histone acetylation and methylation 66, 67, 68. Epigenetic modifications are triggered by cellular metabolites, which in turn enhance the expression of metabolic pathways. For instance, β‐glucan is sensed by the cell surface receptor dectin‐1 that signals via Akt/mTOR/HIF‐1α to induce metabolic reprogramming due changes in the TCA cycle and the accumulation of fumarate 66, 68. Fumarate inhibits the KDM5 demethylase and causes the accumulation of the H3K4 trimethylation mark 69. It also has a further effect on H3K27 acetylation and HIF‐1α levels and was associated with increased expression of glycolytic enzymes 69. Epigenetic modifications underpin the long‐lasting changes in the metabolism of innate immune cells, as activation of histone deacetylases reduces trained immunity and promotes immunotolerance 66. This is observed in animals and humans, whereby clearance of initial fungal infections or vaccinations induce permanent changes in the cellular metabolism of innate immune cells 64.

Activated monocytes are thought to be short‐lived whereas trained immunity can last for weeks and months. β‐glucan and other stimuli including the BCG vaccine and Western diets trigger reprogramming of haematopoietic stem cells within the bone marrow that give rise to circulating trained monocytes, explaining some aspects of the long‐lasting effect of trained immunity 70, 71. However, several tissue‐resident phagocytes, including alveolar macrophages, are of embryonic origin and self‐renew independently of the bone marrow 72. Despite this, viral lung infections trigger long‐lasting memory in alveolar macrophages depending on INF‐γ priming by T cells, which increases glycolysis and enables protection from bacterial pathogens 73. This suggests that many microbes encounter innate immune cells that already have a reprogrammed metabolism and display a heightened immune response compared to the naïve monocytes/macrophages commonly utilised in infection studies. This raises the prospect that microbes, such as commensals which regularly interact with host immunity, may have evolved mechanisms to combat reprogrammed innate immune cells to survive, proliferate and establish infections.

Microbial metabolism in infection

Depending on their mode of host interaction (commensal/pathogen) and/or route of infection, microbes encounter diverse host sites, including the bloodstream, skin, gut, mucosal surfaces, organs and tissues, and intracellular host environments when phagocytosed by innate immune cells. These environments are nutritionally heterogeneous, and microbial metabolism has therefore evolved to be adapted to their particular lifestyle, and respond rapidly when nutrient level change. Specifically, the immunometabolic shift in activated macrophages directly “talks” to the metabolic pathways of microbes (Fig 1), and this crosstalk modulates host–pathogen interactions. While immunometabolic shifts are meant to help the host to defeat the infection, in some instances microbes might benefit from the changes in metabolite levels that result from host metabolic reprogramming 74, 75, 76, 77. Furthermore, it is also possible for the microbial pathogen to seize the opportunity and turn the immunometabolic reprogramming of host cells into a nutritional disadvantage 44. Below we review the metabolic strategies of key microbial pathogens and discuss the concept of microbes turning the tables on immune cells during their metabolic interactions.

Extracellular pathogens

Pathogens that persist in the blood as extracellular microbes, such as Trypanosoma brucei, or within red blood cells, such as Plasmodium falciparum, will invariably utilise glucose as the primary carbon source 78. In these microbes, mitochondrial metabolism is thought to be reduced, altered or absent, although more recent genetic and metabolic profiling approaches have identified additional pathways and alternative carbon sources that promote survival of T. brucei and P. falciparum 79, 80. These additional pathways either replenish metabolites otherwise depleted by glycolysis or support microbial survival when glucose levels are limited.

Mycobacterium tuberculosis

In contrast to extracellular pathogens, intracellular microbes frequently encounter environmental niches that contain low glucose levels. This is particularly evident in microbes that reside in specialised vacuoles of macrophages, as these host cells rapidly alter their own metabolism during infections and thus available nutrients. As discussed above, macrophages infected with BCG/killed Mycobacteria tuberculosis upregulate glycolysis to trigger inflammatory responses and trained immunity. In contrast, macrophages harbouring virulent M. tuberculosis show reduced rates of glycolysis and mitochondrial respiration, but increased uptake of exogenous fatty acids 81. Perhaps as a consequence of this, M. tuberculosis fails to gain access to sufficient levels of glucose and uses fatty acids a primary carbon source within macrophages. While determining metabolite concentrations within organelles remains challenging, M. tuberculosis mutants that are no longer able to utilise glucose are still able to establish lung infections in mice, suggesting that the levels of glucose are low in infection microenvironments and that intracellular bacteria utilise alternative carbon sources 82. However, mutants that cannot utilise glucose fail to maintain viability during chronic infections, indicating that the bacteria have access to at least some pools of glucose and that intracellular glucose levels likely fluctuate during the course of infection 82.

Given the increased levels of fatty acids within infected macrophages, it was thought that M. tuberculosis utilises lipids to maintain growth and viability. Unlike humans, bacteria are able to utilise 2‐carbon acetyl‐CoA derived from fatty acid oxidation, to generate 6‐carbon glucose via TCA intermediates and gluconeogenesis. This is only possible because of the glyoxylate shunt, and enzymes isocitrate lyase (ICL) and malate synthase (MS) that convert the TCA intermediate isocitrate to glyoxylate and then to malate. This metabolic pathway enables bypassing of the oxidative decarboxylation steps within the TCA cycle and replenishes 4‐carbon metabolites important for amino acid synthesis and generation of the gluconeogenic substrate oxaloacetate. Deletion of either enzyme, ICL or ML, reduces the ability of M. tuberculosis to establish and maintain lung infections in mice 83, 84. As other TCA intermediates are likely required to maintain energy bioenergetics and for biosynthesis of important metabolites, bacteria must ensure that isocitrate remains available for isocitrate dehydrogenase (ICD). In M. tuberculosis, this is regulated by glyoxylate itself, which activates ICD, thus establishing a rheostat so that both pathways, TCA cycle and the glyoxylate shunt, have access to sufficient levels of isocitrate 85. As such, the flux of metabolic pathways determines survival of intracellular microbes, but also enables metabolic responses to resist immune attack, because the glyoxylate shunt is primarily protective in INF‐γ‐activated macrophages 86.

Candida albicans

As a human commensal, the fungus C. albicans colonises the gut, oral and genital mucosa, while the transition to pathogenesis is associated with an increased load and, in systemic disease, bloodstream location and invasion of organs. Although C. albicans is phagocytosed by macrophages as a host defence mechanism, this yeast is not an intracellular pathogen. In vitro experiments have shown that C. albicans is equipped with effective mechanisms for escape from macrophages via a switch from yeast to hyphal cell morphology, which in turn activates pyroptosis, a programmed cell death of macrophages that depends on the NLRP3/caspase‐1 inflammasome 87, 88. Since it only occupies the macrophage environment transiently, and it therefore sees both intracellular and extracellular environments, C. albicans displays dynamic regulation of its metabolic pathways.

Analysis of C. albicans mutants is consistent with non‐glucose carbon sources being important during infection, as mutations in the glyoxylate cycle or gluconeogenesis modulate in host fitness in mouse infections 89, 90. The situation with fatty acid oxidation is less clear 91, 92. At least in part, utilisation of non‐glucose carbon sources by C. albicans stems from the necessity to survive and grow within macrophages, because phagocytosed C. albicans upregulates enzymes of the glyoxylate shunt, gluconeogenesis and fatty acid oxidation 44, 93. Transcription of enzymes needed to utilise non‐glucose carbon source was also detected in C. albicans facing human blood 94, 95, in a model of epithelial infection 96, in vivo in mouse infections 97, 98, 99 and in isolates obtained from AIDS patients suffering from mucosal candidiasis 96.

Glucose is the preferred carbon source for C. albicans in vitro, but is it important in vivo? Yes, it is, for pathogenesis 90, and also likely for commensal gut colonisation 100. A C. albicans deletion mutant in the glycolytic enzyme pyruvate kinase (PYK1) shows reduced virulence in mouse infections 90, and so does a mutant in which the two transcription factors needed to transcribe glycolytic and other carbohydrate metabolism genes, TYE7 and GAL4, are deleted 101. But, if alternative carbon source utilisation is “on” in host environments (as suggested by studies discussed above), shouldn't glycolysis be “off”? While fungal glycolysis and glucose utilisation genes are repressed when facing host conditions 93, 95, 98, 102, it is worth noting that several studies compared the transcriptome of C. albicans in the host to in vitro growth in glucose‐rich medium. This means that upregulation of alternative carbon utilisation in host environments is easily detected over the reference condition, but the expression of glycolysis and glucose utilisation genes appears repressed or not changed in this comparison. Indeed, imaging of GFP driven by promoters of phosphofructokinase PFK2 and pyruvate kinase PYK1 showed they were expressed when C. albicans was intracellular in macrophages 90, and the pyk1 mutant of C. albicans cannot form hyphae or kill macrophages in vitro 44. Glycolytic gene expression has also been reported in C. albicans in transcriptomic experiments in mouse infections 97, using GFP fusions to glycolytic gene promoters in C. albicans disseminated to the kidney 90, and during incubation with human blood 94. Glycolytic growth could be needed by C. albicans to proliferate in the host and could further represent an immune evasion mechanism, as activated inflammatory phagocytes have high needs for glucose 44.

Candida albicans has evolved regulatory mechanisms to have metabolic flexibility and consume multiple carbon sources at the same time 103. It is also possible that host–pathogen interactions change over the time frame of infection, and therefore, regulation of carbon source utilisation by C. albicans is dynamic. This is most clearly seen during infection of macrophages by C. albicans in vitro, where the initial repression of fungal glycolysis is reversed coinciding with C. albicans egressing from macrophages and finding glucose again 44, 93. This dynamic regulation is mediated by transcription factors Tye7 and Gal4 that activate glycolytic gene expression in a timely manner as C. albicans escapes macrophages 44. Whether these mechanisms of macrophage escape through hyphal formation and dynamic reprogramming of fungal metabolism also operate in vivo remains to be understood. Finally, lactate is thought to be an important carbon source for C. albicans in humans, where it is produced by other microbiota and by activated inflammatory phagocytes that have switched to Warburg metabolism. Again, utilisation of lactate might serve a few different functions for C. albicans, enabling the fungus not only to proliferate, but also to evade immunity by impacting on cell surface structure and immune cell interactions 104.

Leishmania parasites

Not all intracellular microbes express glyoxylate shunt enzymes, including Leishmania parasites, and may thus rely on sugar uptake within macrophages. Because glucose levels are limited, Leishmania also scavenges lipids and amino acids from its macrophage host to fuel mitochondrial respiration 105, 106, 107. This indicates that some microbes have the ability to utilise different carbon sources to promote growth and expression of cell surface virulence factors. Surprisingly, however, Leishmania parasites suppress sugar and amino acid uptake by inducing the degradation of cell surface transporters, resulting in reduced rates of glycolysis and mitochondrial metabolism 105. The minimal metabolism is thought to prevent depletion of important nutrients, but also reduces mitochondria‐mediated oxidative stress that is likely detrimental in an oxidative environment created by antimicrobial responses of macrophages. The stringent metabolism of Leishmania is also associated with reduced expression of cell surface carbohydrates, typically considered virulence factors, and a slow growth rate (doubling time: ~12 days) 16. This raises the prospect that Leishmania parasites evade immune attack by reducing metabolism and the expression of cell surface molecules that are sensed by macrophages. Alternatively, intracellular microbes that rely on host cell for survival may forfeit glucose to support host metabolism. This would also minimise the metabolic stress on the host, which can be sensed to mount potent immune responses 108. In support of this, Leishmania scavenges amino sugars that are nutritionally less important for macrophages to replenish its metabolic pathways during intracellular growth 109, 110. This suggests that microbes adapt their metabolic capabilities depending on the environmental niche, but also indicates that some microbes may have been driven into hostile territory, such as lysosomes, in search for essential nutrients and to evade immune attack 16.

Turning the tables on immune cells: how microbial pathogens might benefit from immunometabolic shifts

From everything that we discussed so far, the consensus is that the metabolic shifts in immune cells upon infection, i.e. heightened glycolysis, disruptions to the TCA cycle and changes to mitochondrial respiration, drive antimicrobial inflammation and the production of direct antimicrobial effectors, such as itaconate and ROS. To launch a productive infection, microbes must either evade immune cell activation or, alternatively, adapt to, or even benefit from, the metabolic shifts in immune cells.

Immunometabolic shifts cause increased concentrations of glucose and lactate in host cells that could promote intracellular survival of Listeria monocytogenes, Legionella pneumophila, Brucella abortus and M. tuberculosis 74, 75, 76, 77, 111. A glucose‐consuming microbe would potentially also interfere with inflammatory responses, similar to the glycolytic inhibitor 2‐deoxy‐glucose that is commonly used in studying Warburg metabolism in immune cells. Furthermore, LPS‐/IFN‐γ‐activated or C. albicans‐activated macrophages are metabolically inflexible and rely on glucose due to the downregulation of mitochondrial oxidative phosphorylation 44, 49. Not only that, activated macrophages rapidly lose viability in the absence of glucose, which means that glucose‐consuming microbes could perturb host glucose homeostasis with potentially catastrophic consequences, as we have recently described for macrophages infected ex vivo with C. albicans 44 (Fig 2).

Figure 2. Glucose competition between microbes and host cells determines host–pathogen responses and outcomes.

(A) Upon C. albicans infection, macrophage glycolysis is turned on rendering them metabolically inflexible and necessitating glucose. Once internalised by macrophages, C. albicans switches from yeast to filamentous hyphae supported by amino and fatty acid (AA, FA) metabolism. Initial hyphal escape is mediated in part by NLRP3 inflammasome‐dependent pyroptosis 87, 88. Fungal hyphae then reactivate glycolysis following egress from macrophages 44, 93, which leads to depletion of glucose by the fungus 44. Glucose starvation triggers death of macrophage, as these immune cells are metabolically inflexible and become dependent on glucose as the sole carbon source 44. Once glucose levels drop, C. albicans may utilise macrophage‐derived lactate and other carbon sources, such as amino and fatty acids, but this remains to be firmly shown. (B) S. Typhimurium competes for glucose with macrophages, disrupting host glucose homeostasis and leading to increased levels of ROS and decreased levels of NADH 112. This metabolic stress causes activation of the inflammasome, with consequent production of cytokine (IL‐1β) and pyroptotic lysis. Pyroptosis is a host inflammatory response, but also enables bacterial egress and dissemination. C) During skin infection, S. aureus consumes glucose, triggering glucose stress in keratinocytes 113. This contributes the signal for Warburg reprogramming in keratinocytes, and consequent activation of transcription factor HIF‐1α. HIF‐1α mediates pro‐inflammatory cytokine transcription, and also transcription of genes related to glycolysis and glucose consumption. D) While Leishmania preferentially uses glucose as carbon source, in intracellular stages it relies on alternative carbon sources including glucosamine (GlcN), amino acids (AA) and fatty acids (FA) 105, 106, 107, 110. Glucose uptake is reduced, likely promoting macrophage viability and parasite growth.

Glucose competition between host cells and microbes has also been suggested for Salmonella enterica serovar Typhimurium infecting macrophages and Staphylococcus aureus infecting skin keratinocytes 112, 113 (Fig 2). Glucose consumption by intracellular S. Typhimurium disrupts host glycolysis, which creates a metabolic signal in the form of lower levels of NADH and higher ROS, activating the NLRP3 inflammasome 112. In this scenario the consequence of competition for glucose is an inflammatory response, which includes maturation and secretion of IL‐1β cytokine and lysis of macrophages by pyroptotic cell death. Pyroptosis contributes to inflammation, but also releases intracellular bacteria to spread infection, potentially benefitting the pathogen. In the case of S. aureus, the glycolysis mutant pyk1 was less able to activate metabolic reprogramming in keratinocytes towards high glycolysis, and it triggered less IL‐1β cytokine and reduced lysis of keratinocytes and was less virulent in an animal model of skin infection 113. The authors proposed that glucose consumption by S. aureus causes nutritional stress in keratinocytes, leading to upregulation of glycolysis and driving inflammatory processes 113 (Fig 2). Staphylococcus aureus can replicate and grow extracellularly. Therefore, here also, glucose competition could be benefitting the pathogen by contributing to host cell lysis and microbial release. In both bacterial and fungal interactions with macrophages, it has been shown that the consequences of glucose competition can be reversed by restoring host metabolic homeostasis. Macrophage cell death following C. albicans infection can be delayed by increasing the availability of glucose 44, and inflammasome activation by S. Typhimurium can be suppressed by addition of pyruvate, the final metabolite produced by glycolysis 112.

Does infection disrupt host glucose homeostasis in vivo? While sepsis, as many other severe illnesses, is associated with increased blood glucose levels (so‐called “stress hyperglycaemia”), sepsis patients might be hyper‐susceptible to insulin interventions aimed at reducing glucose levels, which can cause life‐threatening hypoglycaemia 114. Interestingly, we have seen an equivalent mechanism in the mouse candidaemia model, whereby C. albicans‐infected mice developed severe hypoglycaemia following administration of the glucose‐lowering drug metformin, whereas in the absence of infection mice could maintain blood glucose levels when injected with the same dose of metformin 44. In another mouse study, injection of LPS or infection with S. Typhimurium caused hypoglycaemia when co‐administered with a low dose of insulin 53. It is worth noting that in both infection models, with C. albicans and with S. Typhimurium, the microbial infection alone was also causing some hypoglycaemia, which was dramatically increased with glucose‐lowering interventions (metformin or insulin) 44, 53. A transient reduction in host blood glucose was also reported in a mouse model of polymicrobial bacterial septic infection 115.

At present, it is not clear whether, during infection, microbes directly perturb host glucose homeostasis (by competing with host cells for glucose), or whether the observed hypoglycaemia results from dysfunction of physiological processes during sepsis. These two scenarios are not mutually exclusive. Infection can impact on insulin clearance 53 and reduce glucose biosynthesis by gluconeogenesis in the liver 115. This is consistent with broad physiological effects of infection on glucose metabolism and uptake. Supporting direct effects of microbial competition for glucose, glycolytic mutants of C. albicans have reduced ability to trigger hypoglycaemia in response to metformin 44. The interpretation of these data is however complicated: do the C. albicans glycolytic mutants maintain host glucose homeostasis because they consume less glucose? Or is it because they are hypovirulent, and therefore trigger distinct physiological responses by the host compared to fully virulent strains? Testing microbial mutants in a variety of pathways required for virulence (metabolic and non‐metabolic) should resolve this question.

Therapeutic options for infections based on nutrients and metabolism

The prospect that inhibiting metabolic enzymes or boosting metabolic pathways may be beneficial in infectious diseases is gaining traction (Table 1). These anti‐infective approaches would target the host, compared to traditional antimicrobial therapy, which targets the pathogen. Given the difficulties in developing new antimicrobial therapeutics, host‐targeted approaches should be considered, while keeping in mind potential side effects of metabolic therapies with broad effects on multiple host cell types.

Table 1.

Possible metabolic strategies in infection

Strategy	Target	Compound	Effect
I. Target metabolic enzymes	Glycolysis	2‐Deoxy glucose (2‐DG)	Repressed inflammation via HIF‐1α
	GABA transaminase of the GABA shunt; production of succinate	Vigabatrin	Repressed inflammation via HIF‐1α
II. Target chromatin and transcription factors that drive metabolic reprogramming	Demethylase	Fumarate	Induction of trained phenotype in monocytes. Induction of HIF‐1α‐dependent metabolic genes
	SIRT1	EX‐527	Reactivation of immune response in tolerant immune cells. Inhibition of SIRT1 inhibits metabolism of fatty acids. Acts on transcription factors NF‐kappa Β and PGC‐1
	Histone lysine deacetylases (HDACs)	Panobinostat Entinostat RGFP966	Reduction of inflammatory and anti‐inflammatory cytokine production
III. Modulate signalling pathways that control immunometabolism	AMPK (mTOR indirectly) Mitochondrial respiration	Metformin	Repress inflammation by inhibiting metabolic reprogramming of immune cells.
	Dectin‐1 signalling	β‐Glucan	Promote trained immunity and reverse immunoparalysis by triggering epigenetic changes in metabolic gene expression.
	TLR4 signalling	Monophosphoryl lipid A (MPLA)	Triggers reversible reprogramming on macrophages to Warburg effect and the reactivation of mitochondrial functions
	IFN‐γ signalling	IFN‐γ	Induces glycolysis and reactivates immunotolerant monocytes.
	NAD+ levels	β‐Nicotinamide mononucleotide (NMN)	Reduce inflammation by promoting SIRT3‐dependent deacetylation of substrates in mitochondria, to increase mitochondrial respiration and metabolism.
IV. Supplement metabolites	Various cells	Glucose	Maintain homeostatic blood glucose levels Maintain immune cell function Maintain tissue homeostasis, reduce ER stress and protect brain function

Several clinical studies have focused on inhibiting inflammatory pathways during sepsis, as uncontrolled cytokine secretion is thought to lead to organ and tissue damage. One approach to this is to control metabolic enzymes. Given the importance of glycolysis and its enzymes for inflammatory responses in innate immune cells, glycolysis blockers, such as 2‐deoxy glucose, prevent secretion of IL‐1β during LPS administration in mice 18. Another possible target is GABA transaminase, which fuels the TCA cycle to promote accumulation of succinate. Its inhibition with vigabatrin reduced IL‐1β secretion after LPS administration or Salmonella infections in mice 18. Vigabatrin is currently used in the clinic to treat epilepsy and is associated with increased risks of infections, possibly due to its anti‐inflammatory properties 116.

Another approach is to control chromatin modification enzymes. Histone deacetylase inhibitors (HDACi), some of which are approved for cancer treatment, also reduce inflammation in fungal and bacterial infections 117. Whether this is solely due to changes in histone acetylation and gene expression, or also involves the many non‐histone targets of HDACs, awaits confirmation. It is worth keeping in mind that bacterial metabolites, such as the short‐chain fatty acid butyrate, are potent HDAC inhibitors. Moreover, the TCA metabolite fumarate or derivatives that are less likely to be further metabolised inhibit histone demethylation, and their administration may be beneficial in modulating inflammatory responses to infection 69.

Signalling pathways that control metabolism and inflammatory responses could also be targeted. For instance, metformin, which is commonly used to treat diabetes via activation of AMP kinase to lower blood glucose levels, dampens inflammation. This is mediated via reduced mTOR activity, which in turn dampens HIF‐1α‐dependent secretion of IL‐1β 43. Metformin treatment is associated with reduced risk of developing sepsis in humans 118; however, studies in mice suggest that it may increase opportunistic fungal infections by lowering immune responses to C. albicans and causing hypoglycaemia 43, 44. LPS triggers loss of NAD⁺, and restoring NAD levels by administration of the precursor metabolite, β‐nicotinamide mononucleotide (NMN), prevents inflammatory responses in macrophages after LPS challenge 42. Interestingly, macrophages show increased inflammatory responses in the elderly, which are known to have altered NAD homeostasis. Again, NMN administration reverted those inflammatory responses to levels typically associated with younger individuals 42. Monophosphoryl lipid A (MPLA), which activates the macrophage receptor TLR4 and is an FDA‐approved vaccine adjuvant, is also thought to have promise as an anti‐infective agent in bacterial (S. aureus) as well as fungal (C. albicans) infections 119. In vitro experiments with bone marrow‐derived mouse macrophages showed that while initially MPLA treatment triggered a Warburg effect, 3 days later macrophages reactivated mitochondrial oxidative phosphorylation 119. This result is interesting in the light of our work with C. albicans that showed that bone marrow‐derived macrophages cannot reactivate mitochondrial metabolism in time to counter glucose depletion by the fungus 44. Making macrophages able to reactivate mitochondrial functions could protect them under glucose starvation conditions. We note that under our experimental conditions (multiplicity of infection of 3 Candida to 1 macrophage and standard tissue culture medium RPMI that contains ≈ 10 mM glucose), the time frame for the glucose depletion and catastrophic consequence on host mitochondria was a matter of hours, rather than days. The precise conditions of infection in vivo, including multiplicity of infection, glucose concentrations and levels of other metabolites, will all impact on these processes. Future work will need to understand how we can translate the in vitro data to in vivo effects.

Given that sepsis can progress to an immune‐tolerance state, metabolic reprogramming of innate immune cells to restore inflammatory responses and prevent infections is also currently being explored experimentally. For instance, fungal β‐glucan induces trained immunity characterised by increased glycolysis and partially reverses LPS‐induced immunoparalysis of human monocytes after experimental endotoxemia 67. This was largely due to induced histone modifications that restored the expression of several inflammatory cytokines. Administration of glucans showed protection not only against C. albicans, but also against S. aureus infections, in keeping with the observation that trained immunity is not a pathogen‐specific response 120. While β‐glucan administration has not been validated in human sepsis patients, it is currently being trialled to enhance anti‐cancer immunotherapies 121. β‐glucan as well as the BCG vaccine also protect from unrelated infections by enhancing pro‐inflammatory responses, suggesting the possibility to stimulate protective mechanisms against infectious diseases and also cancers 121. Both β‐glucans and BCG cause epigenetic changes on key glycolytic genes to maintain their increased expression 66, 122. Targeting these epigenetic mechanisms directly has emerged as a potential mechanism to reactivate immunotolerant leucocytes, as specific histone marks have been associated with reduced expression of pro‐inflammatory cytokines. This is partly driven by the NAD⁺‐dependent deacetylase, SIRT1 61. The SIRT1 inhibitor EX‐527 was able to reverse the tolerance phenotype of septic mice, resulting in improved mouse survival rates, bacterial clearance and innate immune responses 123. Presumably this also reflects increased metabolic rates of leucocytes to promote inflammatory response, but this has not been investigated so far. It has also become apparent that cytokines themselves affect the metabolism of leucocytes, including autocrine signalling. For instance, IFN‐γ was able to restore the immunotolerant phenotype of human monocytes isolated from sepsis patients, which includes increased glycolytic flux as determined by lactate secretion 43. This suggests that reactivation of inflammatory metabolism is feasible and that this restores immunoparalysis in fungal sepsis patients 43, 124.

Lastly, several studies have shown that administration of metabolites can be beneficial in diverse infections. Blood glucose levels are tightly controlled to maintain tissue homeostasis, but likely also contribute to maintaining immune function. During sepsis, blood glucose levels initially increase but can become hypoglycaemic in patients where glucose control is implemented, which is associated with poor clinical outcomes 114. In murine infection models with diverse pathogens (bacteria, fungi and viruses), glucose has been shown to have either positive or negative effects on host outcomes (Fig 3). These contrasting results highlight the need for deeper understanding of how glucose regulates host physiology depending on the nature of infection. It will also be essential to study how this research in experimental animals translates to humans. Glucose supplementation rescued mice that failed to maintain blood glucose levels after polymicrobial infection 115. Glucose administration is also protective in experimental influenza infections by promoting protein glycosylation and reducing ER stress to maintain tissue homeostasis in mice 125. Loss of blood glucose is also partly due the reduced nutritional intake, termed sickness behaviour, during infections. Reduced food intake, and thus reduced blood glucose levels, has been shown to be protective in some bacterial infections and LPS sepsis model, although food restrictions negatively affect host outcomes in many other infections 125, 126. Not surprisingly then, microbes themselves target host metabolism as Salmonella prevents reduced food intake of its host to increase survival of bacteria and host 126. Similarly, host metabolism affects microbial virulence, as decreased glucose absorption due to insulin resistance increases gut glucose levels, which then suppresses expression of a virulence factor in pathogenic E. coli 127. Consequently, oral glucose administration attenuates gut infections and increases mouse survival 127. Moreover, feeding infected mice glucose improved animal survival rates in systemic C. albicans infection (precisely why glucose was beneficial in this context remains to be determined) 44. Glucose also rescued animals from the severe hypoglycaemia that developed when metformin and C. albicans were administered in conjunction 44.

Figure 3. Diverse effects of glucose metabolism on host outcomes in animal infections.

Glucose administration prevents hypoglycaemia (loss of blood glucose) in Candida albicans‐infected mice treated with metformin and in polymicrobial infections in the absence of liver gluconeogenesis, thereby promoting mouse survival 44, 115. Glucose is also beneficial in mice during lung infections with influenza by preventing neuronal death due to increased glucose demand, although loss of blood glucose levels was not restored 125. In contrast, glucose administration accelerates accumulation of reactive oxygen species (ROS) in the brain after Listeria monocytogenes infection and decreases mouse survival 125.

Conclusions and outlook

In the era of raising concerns regarding antimicrobial drug resistance and lack of investment to develop new antibiotics and antifungals, it is paramount that we develop complementary strategies for the treatment of infectious diseases. As discussed in this review, metabolic interactions of innate immune cells and pathogens present a range of opportunities for interventions that will benefit the host by assisting immune cells in their antimicrobial activities, and by controlling inflammation so that it is helpful, but not toxic. However, several challenges and open questions remain (see Box 1). In particular, we need to understand “local versus global issues”, to borrow an expression from politics. For example, we need to be able to know how the levels of metabolites and nutrients fluctuate in the infection foci in local microenvironments in infected tissues and organs, which could be achieved by applying new imaging methods that are able to capture metabolites, such as NMR microscopy. Global changes in metabolites affect the physiology that accompanies infections, and understanding how metabolism affects sickness behaviour promises new therapeutic options. There is also a temporal dimension to this problem, in that different stages of infection (i.e. early acute phases and later septic phases) are likely different with respect to metabolism of immune cells and pathogens, as they adapt to each other and to changing nutrient levels in their environment. This means that different strategies might be needed at different stages of disease. Regardless of these challenges, we and many others believe that manipulating metabolism holds great promise for combatting microbial pathogenesis. It is also a strategy that might be broadly applicable to infections with diverse pathogens, as demonstrated by the discussed examples of protection in bacterial and fungal infections. In this context, it is worth noting that polymicrobial infections occur, including mixed bacterial–fungal infections that present distinct challenges 128. Immunometabolic interventions could potentially be very useful in the situation of multiple infecting pathogens.

Box 1.  In need of answers.

1. What nutrients are available to immune cells and microbes locally in infection microenvironments? How do the levels of nutrients change during infection?

   • During infection, the host–pathogen interaction “wars” occur in localised environments. These include intracellular locations in immune cells (phagosome, cytoplasm), as well as various tissues and organs. Tissue culture media and ex vivo infections might not faithfully reproduce the situation in these infection microenvironments. To fully understand the metabolism of both immune cells and microbes during their interaction, we need to define the local availability of nutrients in infection microenvironments, as well as understand how the levels of nutrients might dynamically change during infection.

2. What are the metabolic profiles of immune cells and microbes in infection microenvironments?

   • Metabolic profiles of host immune cells and microbial pathogens have mostly been analysed in ex vivo challenge scenarios, with defined nutrient composition of the media and the presence of only two cell types (the immune cell and the microbe under study). While these reductionist approaches are critical for mechanistic insight, infection microenvironments are complex, not only nutritionally as described above in Question 1, but also in relation to the presence of other cell types. To fully understand how we could manipulate metabolism to help the host fight infection, it will be important to profile the metabolic status of both host and pathogen from infection microenvironments.

3. How can metabolic interventions modulate inflammation to promote pathogen clearance, but prevent pathology driven by over inflammation?

   • The different stages of infection in patients might require distinct metabolic approaches to promote antimicrobial functions of immune cells, restart the immune system in immune paralysis or dampen inflammation if it has become pathological. Which metabolic interventions will be most useful in these diverse scenarios and how the timing of the intervention during diseases might impact on the outcome are open questions.

4. What is the nature of metabolic host–pathogen interactions in polymicrobial infections?

   • The majority of studies in the field of immunometabolism have been done using infections with a single species of microbe. However, polymicrobial infections are common. Microbes can influence immune cells in diverse ways 45. Moreover, supplementation of nutrients can have opposing effects (i.e. be beneficial or make things worse) depending on the nature of infection, as has been shown for glucose in viral versus bacterial infection 125. Therefore, we should study immunometabolic shifts and the interactions of host and pathogen metabolism not only in single‐microbe infections, but also in polymicrobial challenge of immune cells and in disease.

Conflict of interest

The authors declare that they have no conflict of interest.

EMBO Reports (2019) 20: e47995