Macrophage polarization refers to the activation of macrophages under the stimulation of a variety of factors and their differentiation into different phenotypes according to the state and changes in the microenvironment (14, 36). AMs have two main macrophage phenotypes: classically activated or inflammatory (M1) macrophages and alternatively activated or anti-inflammatory (M2) macrophages (37, 38). The regulatory mechanisms and functional characteristics of M1/M2 macrophages are shown in Figure 1 . M1 macrophages constitute the first line of defense against intracellular pathogens (36). Currently, it is believed that M1 macrophages are mainly induced by lipopolysaccharide (LPS), interferon-γ (IFN-γ), and granulocyte–macrophage colony-stimulating factor (GM-CSF) (38, 39). M1 macrophages can guide acute inflammatory responses and produce a large number of pro-inflammatory cytokines such as IL-1β, inducible nitric oxide synthase (iNOS), tumor necrosis factor-α (TNF-α), IL-1, IL-6, IL-12, CCL8, IL-23, CXCL1-3, CXCL-5, CXCL8-10, monocyte chemotactic protein-1 (MCP-1), macrophage inflammatory protein 2 (MIP-2), reactive oxygen species (ROS), and cyclooxygenase 2 (COX-2) (16, 38, 40–42). M1 macrophages mainly induce Th1 response activation, exert antigen-presenting functions, and engage in pro-inflammatory responses, chemotaxis, radical formation, elimination of pathogenic microorganisms, and antitumoral activities (37, 43).

Conversely, M2 macrophages are induced in response to Th2 cytokines such as macrophage colony-stimulating factor (M-CSF), IL-4, IL-13, and IL-10 (43–45). M2 macrophages mainly express CD64 and CD209 and produce anti-inflammatory cytokines such as IL-8, IL-10, IL-13, CCL1, CCL2, CCL3, CCL4, CCL13, CCL14, CCL17, CCL18, CCL22, CCL23, CCL24, and CCL26 to exert anti-inflammatory effects, promote tissue remodeling, facilitate tumor development, and remove parasites (14, 43, 46–48). M2 macrophages can be further divided into four subsets, M2a, M2b, M2c, and M2d, according to the different activating stimuli received (41, 49). The M2a subset of macrophages can be stimulated by IL-4 or IL-13 to produce IL-10, CCL13, CCL17, and CCL22 (36, 49). These chemokines are associated with Th2 cell activation and can promote eosinophil recruitment to the lungs (50, 51). The M2b subset can be stimulated by LPS or IL-1β and produce pro-inflammatory cytokines (36, 49). The M2c subset is induced by IL-10, TGF-β, and glucocorticoids and releases high amounts of IL-10, CCL18, and CCL16 to exhibit anti-inflammatory activities (49, 52, 53). The M2d subset is induced by IL-6 and M-CSF and secretes high IL-10 and low IL-12 and TGF-β cytokine production and CXCL10, CXCL16, and CCL5 chemokines to promote angiogenesis and tumor metastasis (54, 55).

These polarized macrophages exhibit plasticity, as they can depolarize to M0 macrophages or exhibit the opposite phenotype through repolarization, which depends on the specific microenvironment (4, 43). For instance, a high αKG/succinate ratio further promotes the anti-inflammatory phenotype in M2 macrophages, whereas a low ratio enhances the pro-inflammatory phenotype in M1 macrophages (56). Specific microRNAs induced by different microenvironmental signals can regulate different patterns of macrophage polarization states by regulating transcriptional output. For instance, miR-155 promotes M1 polarization by directly inhibiting expression of BCL6, and overexpression of miR-155 can reprogram M2 macrophages into M1 macrophages (57, 58). The plasticity of epigenetic modification is an essential factor in macrophage identity and heterogeneity, which is remodeled in response to acute and polarizing stimulation (59). Histone deacetylases (HDACs) are strongly involved in M1 activation and play a prominent role in inflammatory responses (60). HDAC6 and HDAC7 are involved in the expression of pro-inflammatory genes in macrophages stimulated by LPS, and inhibition of their activity significantly limits M1 activation and the production of pro-inflammatory cytokines (61, 62). Overexpression of DNA methyltransferase 3B (DNMT3B) or loss of HDAC3 renders macrophages hyperresponsive to IL-4, skewing differentiation toward the M2 phenotype (59).

However, not all macrophages fit the classical paradigm of M1 and M2 macrophages. Tumor-associated macrophages (TAMs), one of the main types of tumor-infiltrating immune cells, are generally characterized as M2 macrophages, which are found to promote angiogenesis and invasion and tumor progression (63, 64). TCR+ macrophages, expressing the CD3/T-cell receptor (TCR)αβ complex, exist in tuberculous granulomas, atherosclerosis, and several types of cancer, which enhance phagocytosis and secrete IFN-γ, TNF, MIP-1β, and CCL2 to exhibit a specific pro-inflammatory profile (65–68).

Macrophage polarization is a complex process modulated by multiple factors, such as microRNAs (miR), proteins, glucocorticoids, and immunometabolism, involving numerous signaling pathways and transcriptional and post-transcriptional regulatory networks (14, 15, 42, 69). The precise regulatory mechanisms of macrophage polarization are still not completely understood and require further investigation. The phenotypic and functional changes in macrophages are accompanied by dramatic shifts in cell metabolism, with unique metabolic signatures related to their functional state, which is termed metabolic reprogramming (70–72). Recent studies have shown that specific metabolic pathways in macrophages are closely related to their phenotype and function, including glycolysis, tricarboxylic acid (TCA) cycle, PPP, arginine metabolism, glutamine metabolism, and fatty acid metabolism. The metabolic pathways of macrophages are shown in Figure 2 . Some metabolic intermediates can regulate macrophage activation and effector function through various mechanisms (72).

Glycolysis is a metabolic pathway that converts glucose to pyruvate, which plays a key role in energy metabolism, especially in the production of ATP in cells under hypoxia and other conditions. Tumor cells preferentially utilize glycolysis to produce lactic acid under normoxic conditions, known as “aerobic glycolysis.” Subsequently, several studies have observed that aerobic glycolysis is increased in pro-inflammatory macrophages (73). Although glycolysis produces far less ATP than mitochondrial OXPHOS (2 ATP vs. 36 ATP), glycolysis can be activated faster to match the immune response of macrophages (74–76). Multiple studies have shown that classically activated M1 macrophages in mice and humans are highly dependent on glycolysis. Van den Bossche et al. used extracellular flux analysis to demonstrate that M1 macrophages display enhanced glycolytic metabolism and reduced mitochondrial OXPHOS; conversely, M2 macrophages display enhanced mitochondrial OXPHOS (77). Rodríguez-Prados et al. showed that activation of murine peritoneal macrophages via the Toll-like receptor (TLR) pathway results in a hyperglycolytic phenotype, and classic activation favors the upregulation of gene expression from the glycolytic pathway and the repression of genes encoding proteins that participate in OXPHOS (73). IL-10, an anti-inflammatory cytokine that induces M2 macrophages, inhibits LPS-induced glucose uptake and glycolysis and promotes OXPHOS to oppose the switch to the metabolic program induced by inflammatory stimuli in macrophages (78).

After LPS stimulation of macrophages, transcription of the glucose transporter (GLUT1) is induced, leading to increased glucose uptake and aerobic glycolysis (79, 80). LPS specifically induces dimeric pyruvate kinase M2 (PKM2) protein expression and phosphorylation, and dimeric PKM2 interacts with HIF-1α, which is a transcription factor that contributes to both glycolysis and the induction of inflammatory genes and is critical for macrophage activation (81). HIF-1α can directly bind to the IL-1β promoter, an event that is inhibited by the activation of tetrameric PKM2 (82, 83). Activation of tetrameric PKM2 inhibits LPS-induced IL-1β production while promoting IL-10 production, thereby attenuating LPS-induced pro-inflammatory M1 macrophages and promoting the typical characteristics of M2 macrophages (82). Thus, PKM2 is a key determinant of LPS-activated macrophages that promote inflammatory responses (82). Another study demonstrated that PKM2-mediated glycolysis promotes NLRP3 and AIM2 inflammasome activation by modulating EIF2AK2 phosphorylation in LPS-primed mouse bone marrow-derived macrophages (BMDMs), consequently promoting the release of IL-1β and IL-18 and high-mobility group box 1 (HMGB1) (84). PFKFB3 is another LPS-regulated target involved in glycolytic conversion (73). PFKFB3 encodes u-PFK2, a subtype of 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase (PFK-2), which increases flux through the glycolytic pathway. PFKFB3 is a target gene of HIF-1α in response to hypoxia in human glioblastoma cell lines and mouse embryonic fibroblasts, thereby providing another mechanism by which HIF-1α promotes glycolysis (85).

Previous studies have suggested that M2 macrophages do not exhibit increased glycolysis and that aerobic glycolysis is predominantly associated with M1 macrophages. However, recent studies have shown that M2 macrophages also display an upregulated rate of glycolysis in addition to augmented mitochondrial metabolism (71). The glycolysis inhibitor 2-deoxyglucose (2-DG) (1 mM) attenuated enhanced mitochondrial respiration, significantly reduced 13C-labeled Krebs cycle metabolite levels and intracellular ATP levels, impaired IL-4-stimulated activation of early BMDMs, and reduced the expression of M2 phenotypic markers, such as Arg1, YM-1, FIZZ-1, and MRC1 (86–88). M-CSF associated with M2 polarization was found to induce similar glucose transporter expression, oxidative metabolism, mitochondrial biogenesis, and increased expression of glycolytic enzymes in macrophages compared with GM-CSF associated with the M1 phenotype (89). Glycolysis may play a more important role in M2 macrophages than previously thought, and further studies are needed.

PPP is an essential step in glucose metabolism, which is required for maintaining the cellular redox state and carbon homeostasis and provides precursors for nucleotide and amino acid biosynthesis. PPP can be divided into an initial oxidative phase that converts glucose 6-phosphate into carbon dioxide, ribulose 5-phosphate, and NADPH and a later non-oxidative phase that produces ribose 5-phosphate for nucleic acid synthesis and phosphate sugar precursors for amino acid synthesis (75, 90).

PPP is a major source of NADPH, which is a cofactor for NADPH-oxidase (Nox2)-dependent ROS production and is required for the generation of the antioxidant glutathione (75). The regulation of ROS levels was mediated in M1 macrophages by Cybb-encoded Nox2, which was transcriptionally upregulated in M1 macrophages and downregulated in M2 cells. Moreover, both the total pool of pentose-5-phosphates and their labeled fraction increased significantly in M1 macrophages (91). CARKL, a sedoheptulose kinase, also known as sedoheptulokinase (SHPK), catalyzes sedoheptulose to sedoheptulose-7-phosphate (S7P), which is a PPP intermediate and PPP flux restraint (92). The downregulated enzyme CARKL causes glycolysis to feed the PPP, thereby generating nucleotides, amino acids, and NADPH, leading to an increase in ROS. The CARKL expression level was rapidly downregulated in mice and humans in vitro and in vivo upon stimulation by LPS; conversely, it was upregulated in response to IL-4, leading to PPP inhibition in M2 macrophages (80). PPP was upregulated in M1 macrophages but downregulated in M2 macrophages.

The Krebs cycle, also known as the TCA cycle, is the final common pathway for the oxidation of carbohydrates, fatty acids, and amino acids and is also a source of precursors for many other biological molecules, such as non-essential amino acids, nucleotide bases, and porphyrin (93, 94). Therefore, the Krebs cycle is an important hub for cellular anabolism (gluconeogenesis and lipid synthesis) and catabolism (glycolysis).

In recent years, several studies have shown that the TCA cycle of M1 macrophages is disrupted at various points, and OXPHOS is inhibited, leading to the accumulation of citrate, itaconate, and succinate (91, 95, 96). Conversely, M2 macrophages maintain robust oxidative Krebs cycle activity, while increasing OXPHOS and ATP levels (88, 91).

Glutamine, serving as a carbon and nitrogen source for metabolic reprogramming of M1 macrophages, can be broken down to produce glutamate and α-KG, the latter of which enters the TCA cycle to provide energy. Furthermore, glutamine metabolism represents an important metabolic module governing the alternative activation of macrophages in response to IL-4 (91). Palmieri et al. reported that inhibition of glutamine synthetase skewed M2-polarized macrophages toward the M1-like phenotype, characterized by decreased intracellular glutamine and increased succinate with enhanced glucose flux through glycolysis, showing an enhanced capacity to induce T-cell recruitment and reduced T-cell suppressive potential, which could be partly related to the activation of HIF-1α (117). Production of αKG by glutamine hydrolysis is important for the alternative activation of macrophages (56). αKG promotes M2 activation through Jmjd3-dependent metabolic and epigenetic reprogramming, and a high αKG/succinate ratio further promotes the anti-inflammatory phenotype in M2 macrophages, whereas a low ratio enhances the pro-inflammatory phenotype in M1 macrophages (56). Moreover, αKG inhibits the nuclear factor-κB (NF-κB) pathway through prolyl hydroxylase (PHD)-dependent proline hydroxylation of protein kinase IKKβ, thereby impairing the pro-inflammatory response of M1 macrophages (56). The peroxisome proliferator-activated receptor γ (PPARγ) is involved in the alternative activation of macrophages (56, 118, 119). PPARγ is required for IL-4-induced M2 macrophage respiration, and the absence of PPARγ dramatically affects glutamine oxidation (120). Unstimulated macrophages lacking PPARγ contain elevated levels of the inflammation-associated metabolite itaconate and express a pro-inflammatory transcriptome (120).

Arginine metabolism is a complex process characterized differently in M1 and M2 macrophages. M1 macrophages express NO synthase, which metabolizes arginine to NO and citrulline, and citrulline can be reused for efficient NO synthesis via the citrulline–NO cycle (121). M2 macrophages are characterized by the expression of arginase, which hydrolyzes arginine to generate ornithine and urea, a precursor of l-proline and polyamines involved in tissue repair and wound healing (122). There are two isozymes of arginase: arginase I and arginase II. The hepatic urea cycle arginase I is expressed as a cytosolic enzyme, while human granulocyte arginase I is found in the granular compartment and arginase II is found as a mitochondrial enzyme (123). Arg1 inhibits NO-mediated inflammatory pathways by competing with iNOS for l-arginine. A recent study demonstrated that renal tubular cells apically exposed to dead cell debris induce reparative macrophage activation, expressing Arg1, which is required for the S3 tubular cell proliferative response that promotes renal repair after ischemia–reperfusion injury (124). Moreover, Zhang et al. found that polarization of M2a macrophages promotes the expression of Arg1, which restores axonal regeneration and promotes the structural and functional recovery of the contused spinal cord (125). Hardbower et al. reported that Arg2 deletion leads to enhanced M1 macrophage activation, pro-inflammatory cytokine expression, and immune cell-derived chemokine production (126). Arg2 is essential for the IL-10-mediated increase in mitochondrial oxidative metabolism and succinate dehydrogenase/complex II activity and the decrease in the inflammatory mediators succinate, HIF-1α, and IL-1β (127). Therefore, arginine metabolism plays an important role in the anti-inflammatory effects and tissue repair of macrophages.

The TCA cycle of M1 macrophages is disrupted, causing the accumulation of citrate, which is exported from the mitochondria to the cytoplasm to generate acetyl-CoA via ACLY, which fuels the de-novo synthesis of cholesterol and FA (128). These steps involve ACLY, acetyl-CoA carboxylase (ACC), fatty acid synthase (FASN), desaturases, and elongation proteins. Fatty acids tightly couple glucose and lipid metabolism via the de-novo FA synthesis pathway, supporting cell adaptation to environmental changes. The FAO pathway produces acetyl-CoA, NADH, and FADH2, which are further used in the TCA cycle and ETC to generate large amounts of ATP, which is thought to promote M2 polarization. This process is coordinated by multiple enzymes such as fatty acyl CoA synthetase, carnitine palmitoyltransferase I (CPT1), and carnitine palmitoyltransferase II (CPT2).