CD36’s effects on macrophages are manifold. As a receptor for lipoproteins, apoptotic cells, PAMPs and DAMPs, and a facilitator of fatty acid transport, CD36 plays an integral operative role, while also influencing differentiation and ultimate phenotype. Macrophages are functionally heterogeneous; they develop from bone marrow-derived monocytes or tissue-resident macrophages along a spectrum in response to situational factors and retain plasticity (34). They include those that prioritize pathogen killing, have immunosuppressive functions, or are involved in tissue homeostasis and wound healing. In the classic M1/M2 paradigm, CD36 is more highly expressed in macrophages associated with tissue homeostasis, remodeling and repair after injury (35–38). These homeostatic and tissue restoration macrophages store and oxidize fatty acids for the intermittent or extended energy needs involved in efferocytosis, phagocytosis of tissue debris and synthesis of degradative enzymes. Indeed, phagocytosis of cells couples with fatty acid oxidation by providing substrate that would otherwise contribute to excess lipid accumulation. Reparative macrophages show greater uptake of apoptotic cells than those involved in pathogen killing, and in this way decrease tissue exposure to necrosis and mitigate inflammatory responses (39). In a model of ischemia-induced brain injury, infiltrating blood monocyte-derived macrophages showed greater phagocytic activity, with concomitant increase in cell membrane CD36 (37). This was accompanied by elevated expression of lysosomal acid lipase, a critical mediator of the M2 phenotype through modulation of fatty acid supply for oxidation, through breakdown of triglycerides and lipoproteins (36). Inhibition of CD36 with antibody or drug significantly decreased phagocytosis, and altered the development of tissue-resolution macrophages as a result (37). In a more recent study, pathogenic polymorphisms in CD36 that reduced expression by ~50%, predisposed to myocarditis in patients immunized with the Pfizer-BioNTech BNT162b2 vaccine against COVID-19 (15). This was mechanistically linked to impaired efferocytosis. The authors noted that the effect was predominantly in M1 macrophages, but this is confounded by the fact that CD36 is necessary for expression of M2 markers (35, 36, 38). In both these cases, the functional role of CD36 was aligned with tissue repair and homeostasis; loss of CD36 function altered phenotype.

Reparative macrophages synthesize and secrete collagen, transforming growth factor-β, fibroblast growth factor, platelet-derived growth factor, and vascular endothelial growth factor (40–42). They are the conductors of the orchestrated remodeling process, attracting fibroblasts, stem cells, endothelial cells and others necessary in a temporal and coordinated manner. M2 macrophages also maintain a homeostatic environment through the synthesis and secretion of the anti-inflammatory cytokine IL10, and arginase-1, which converts arginine to ornithine and in this way thwarts the production of nitric oxide used in cytotoxic bursts (43). Classically activated (by interferon-γ or lipopolysaccharide) M1 and alternatively activated (by IL4 and IL13) M2 phenotypes demonstrate starkly different metabolic profiles: the overall ratio of oxidative phosphorylation to aerobic glycolysis has been shown to be 10-fold higher in M2 macrophages compared to M1 macrophages (36). Inhibition of fatty acid oxidation using the drug etomoxir decreased M2 signature gene expression, suggesting that the M2 phenotype is tightly associated with cellular metabolism (36). In line with this, delivery of triglycerides to the cell through CD36 was found to be necessary for M2 differentiation, and cells from mice and humans deficient in CD36 showed decreased expression of M2 markers (36). More recently, increased fatty acid oxidation as a necessary contributor to the immunosuppressive phenotype has been questioned (44). Differences in arginine and glutamine metabolism, and the channeling of fatty acids into anabolic pathways may be more essential in the determination of M1 vs M2 fate (44). Notwithstanding, uptake of lipids by CD36 remains a significant factor. A summary of metabolism-driving function for classical and alternatively activated macrophages is shown in Figure 1 .

Single-cell RNA sequencing has greatly enhanced characterization of macrophage subtypes, and revealed key influencers in their phenotype. Triggering receptor expressed on myeloid cells (TREM) 2 has emerged as an important driver of subgroups of macrophages in obesity, metabolic dysfunction-associated steatohepatitis (formerly non-alcoholic steatohepatitis) and atherosclerosis (45–47). Lipid-associated macrophages (LAMs), a distinct population that expand and accumulate in the stromal-vascular fraction from obese human visceral adipose tissue and adipose tissue from diet-induced obese mice, have a TREM2-directed transcriptional signature that includes CD36 and other genes responsible for lipid uptake and storage (46). Kyoto encyclopedia of genes and genomes pathway analysis revealed enrichment of those related to phagocytosis, endocytosis and lipid metabolism. Other pathways upregulated in LAMs were peroxisome proliferator-activated receptor (PPAR)- γ signaling and oxidative phosphorylation (46). Deficiency in TREM2 did not alter the expansion of the monocyte/macrophage compartment in adipose tissue, but resulted in loss of the associated biomarkers, indicative of its obligatory role in specialization of this population (46). In studies in TREM2-knockout (KO) mice, loss of this macrophage subset had profound ramifications: “massive” adipocyte hypertrophy, rapid weight gain, hypercholesterolemia, increased body fat and serum insulin, and insulin-resistance (46). These data suggest that in the obese setting, TREM2 LAMs attempt to maintain homeostasis through adipose tissue remodeling with an important role for CD36. Unlike homeostatic macrophages, LAMs occur in diseased states, including obesity, metabolic syndrome, fatty liver and cancer. They may differentiate from monocytes recruited to the tissue or from resident macrophages. Homeostatic resident macrophages are self-renewing. LAMs differ from homeostatic resident macrophages by expression of a gene program that includes TREM2, FABPs 4 and 5, lipase A, apolipoprotein E (apoE) and CD36, and a focus on lipid uptake and metabolism (48). Morphologically, LAMs are distinguished by the presence of lipid droplets.

In single-cell transcriptomic analyses of human atherosclerotic plaques, studies have consistently identified a distinct myeloid cluster containing “foamy macrophages” (49–51). These macrophages are characterized by abundant lipid, upregulation of fatty acid utilization pathways and a less inflammatory cytokine profile. Similar analyses done in mouse atherosclerotic lesions revealed a conserved subtype (51, 52). In both species, expression of TREM2 was a determining characteristic. Analogous to LAMs, these macrophages showed enrichment for lipid and fatty acid uptake and metabolic pathways, activation of liver X receptor/retinoid X receptor, which increase cellular lipid efflux mechanisms, and deceased cholesterol synthesis (45, 52). These cells were further characterized by expression of smooth muscle cell actin, but in the context of myeloid-determining transcription factors, and anti-inflammatory pathways driven by signal transducer and activator of transcription (STAT) 6, a downstream effector of IL4 (49). As was the case for TREM2 LAMs in the obese setting, CD36 was upregulated (45, 49–53). As a receptor for oxidized and native lipoproteins and fatty acids, it likely plays a major role in the development of the foamy macrophage morphology.

In both the setting of obesity and atherosclerosis, these TREM2 macrophage subpopulations function in uptake and disposal of excess lipid. They utilize various pathways: fatty acid oxidation to burn lipids for power, efflux mechanisms to remove excess, and finally storage routes. In sustained conditions, however, lipid overload leads to endoplasmic reticulum (ER) stress and autophagy (54). Cells deficient in TREM2 were more sensitive to ER stress; this suggests that the upregulation of TREM2 and associated genes in early lesions is compensatory and pro-survival (52, 53, 55). Additionally, macrophage deficiency of TREM2 led to a greater number of apoptotic cells in atherosclerotic lesions (52). In mouse atherosclerosis studies in which TREM2 was deficient in monocytes/macrophages, hematopoietic cells or globally, there was no consistent change in necrotic core size, however (45, 52, 53). The discrepancies may have to do with the timing of analysis, the cell types deficient in TREM2 and the model itself. Increased numbers of apoptotic cells and increased necrotic core size functionally track to a defect in efferocytosis. Although not addressed, this may be a consequence of CD36 downregulation in TREM2 KO cells, as uptake of apoptotic cells is a scavenger receptor function. This would suggest that as lesions progress, loss of TREM2 can have negative consequences. Relevant to this point, in a database analysis of carotid endarterectomy samples from recent stroke patients and non-stroke controls, TREM2 was enriched in the non-stroke patients, suggesting that TREM2 promoted a stable plaque phenotype (52).

Interestingly, in these studies of global, hematopoietic and monocyte/macrophage-specific TREM2 KO mice, atherosclerosis decreased in both the apoE KO and low-density lipoprotein receptor (LDLR) KO, classic mouse models of atherosclerosis (45, 52, 53). TREM2 KO macrophages were shown to have decreased expression of CD36 in vitro and in plaques, and decreased oxidized LDL (oxLDL) uptake and lipid accumulation, respectively (45). In experimental models where TREM2 was targeted with agonistic antibodies, there was consensus that plaques showed greater signs of stability and reduced necrotic cores (53, 56). In one model, however, there was no change in overall lesion size, while in the other, overall lesion size was greater with increased number of macrophages, due to better survival and increased proliferation in plaques (53, 56). Complementary in vitro studies showed that antibody agonism led to increased oxLDL uptake, increased cholesterol efflux and cell survival (56). Thus, functionally, TREM2 induced higher expression of CD36 that then significantly impacted cellular metabolism. CD36-accrued lipid was channeled to be metabolized or effluxed as a protective macrophage survival mechanism, but the accumulation of foamy macrophages increased plaque size. Deficiency of TREM2 in macrophages led to smaller lesions, dependent upon loss of CD36 function, but paradoxically, the cells in those lesions were more vulnerable to death. A summary of the effects of TREM2 on metabolism and macrophage function and effects of deficiency are shown in Figure 2 .

The significant impact of CD36-dependent lipid uptake on macrophage phenotype is further illustrated in mice deficient for caspase recruitment-domain containing protein 9 (CARD9). This adapter protein is essential for activation of mitogen-activated protein kinases by Toll-like receptors (TLRs) (57, 58). OxLDL has been shown by multiple groups to be an endogenous ligand for TLRs; thus, CARD9 may play a role in amplification of pro-inflammatory signaling through TLR activation of p38 and Jun N-terminal kinase (59, 60). There have been several reports on the role of CARD9 in atherosclerosis with conflicting results (61, 62). Recent mechanistic investigation, using multiple mouse models (apoE KO, LDLR KO, recombination activation gene (RAG) 2 KO, CD36 KO, CARD9 KO and combinations) to rule out effects of the adaptive immune system and gut microbiota, has shown that CARD9 deficiency increased atherosclerosis lesion burden, with marked enhancement in macrophage and necrotic core lesion areas (63). These mice had a less pro-inflammatory cytokine profile and no significant differences in serum cholesterol levels (63). Further investigation showed that cells deficient in CARD9 had greater oxLDL uptake and lipid accumulation, and oxLDL-mediated apoptosis (63). CD36 cell membrane expression was greatly increased and shown to be essential to the increase in atherosclerosis, macrophage and necrotic core areas in vivo (63). Mechanistically, upregulation of CD36 not only contributed to increased macrophage foam cell formation, but also resulted in lipid overload that impaired autophagy pathways, leading to enhanced and progressed lesion development (63). Autophagy was impaired as a result of inhibition of adenosine monophosphate-activated protein kinase (AMPK) and downstream activation of mammalian target of rapamycin (mTOR) C1/mTORC2 (63).

Other studies have also revealed CD36-mediated lipid uptake as an essential regulator of AMPK activation. AMPK is a cellular sensor of nutritional status, and through phosphorylation of key effector proteins, directly or indirectly regulates metabolism with far-reaching consequences (64). As a general statement, AMPK activation leads to an increase in catabolic pathways and an inhibition of anabolic pathways (64, 65). AMPK is phosphorylated as a result of an increase in the AMP to adenosine triphosphate (ATP) ratio (64, 65). AMPK increases the activities of glycolysis, fatty acid oxidation and autophagy, and induces mitochondrial biogenesis, while inhibiting lipid, protein, glucose, glycogen, sterol and ribosomal RNA (rRNA) synthesis, to bring the cell back to homeostasis (64, 65). In order to increase cellular levels of ATP, AMPK enhances fatty acid uptake, to couple with catabolic pathways (66–68). Hence, AMPK activation results in increased CD36 expression and localization to the plasma membrane.

While an association between TREM2 and AMPK in macrophages was not noted in studies of TREM2 KO mice, TREM2, SIRT1 and PPAR-γ, all necessary for shifts in energy metabolism in macrophages, have been shown to be tightly linked to AMPK and required for anti-inflammatory polarization of microglia, the resident myeloid cells of the brain (69). In microglia, TREM2 upregulates CD36 to promote phagocytosis of amyloid-β (70).

Mechanistically, AMPK-mediated increased expression of CD36 has been studied in cells other than myeloid cells. In intestinal cells, increased CD36 expression by AMPK was due to inhibition of polyubiquitination by Parkin (68). In myocytes, CD36 was found in a complex with the Src kinase, Fyn and liver kinase B1 (LKB1) (67). Increased levels of circulating fatty acids led to displacement of Fyn from the complex, leading to LKB1 phosphorylation of AMPK, and downstream effects (67). Interestingly, in myocytes, CD36 not only regulated AMPK activation, but then was itself regulated by AMPK. Stimulated recruitment of CD36 to the membrane by insulin, contraction or AMPK was dependent upon AK strain transforming (AKT)substrate 160 and Rab-GTPase to control vesicular trafficking and increased plasma membrane localization (66). Whether similar mechanisms account for AMPK-induced CD36 increased expression and localization in myeloid cells remains unknown. The mechanism of increased atherosclerosis in CARD9-deficient macrophages is shown in Figure 3 .

A major therapeutic initiative in cancer is targeting lipid metabolism as a way to reprogram immune cells in the tumor microenvironment (TME) (71, 72). This niche is unique due to an often immature and disorganized vasculature that limits supplies of oxygen and nutrients, and the removal of debris and waste products, while supporting tumor cell proliferation and immune cell emigration (73). Given the highly proliferative capacity of tumor cells, there is a significant increase in anabolic pathways, including lipid, cholesterol, rRNA and protein (74). Tumor cells develop strategies for sustained cellular biosynthesis via aerobic glycolysis (the Warburg effect) and glutamine-dependent anaplerosis, which provides substrates for a modified tricarboxylic acid cycle (74–76). While inefficient in generating ATP, it has been estimated that lactate production from aerobic glycolysis occurs at a faster rate (10-100x) compared to mitochondrial oxidation; thus, the net amount of ATP produced would be comparable (77). Additionally, increased glucose uptake and breakdown are hypothesized to feed into necessary biosynthetic pathways by providing carbon for the synthesis of proteins, lipids and nucleotides, nicotinamide adenine dinucleotide phosphate equivalents necessary for de novo fatty acid synthesis, and regeneration of nicotinamide adenine dinucleotide to keep glycolysis going (75, 76).

Evidence shows that high-fat diets and obesity contribute to more than a dozen cancers (78, 79). Tumor-driven modifications of adipose tissue enhance tumorigenesis of the breast, ovary, prostate, lung and bone marrow, amongst others (80–83). While fatty acid oxidation is disfavored in tumor cells, fatty acids can be utilized and stored for the biosynthesis of membranes and lipid signaling molecules (72). Thus, dietary and tumor cell-synthesized fatty acids are used to produce arachidonic acid, docosahexaenoic acid and eicosapentaenoic acid, eicosanoids, prostaglandins and lysophosphatidic acid (72). These can act autonomously or on other cells in the TME to influence phenotype. Prostaglandin E2 (PGE2), the product of the enzyme cyclooxygenase 2, which is preferentially expressed in cancer cells (e.g., colon, prostate, melanoma, breast and lung), has emerged as an important player in influencing TME cellular immunosuppressive phenotypes (84, 85).

Tumor-associated macrophages (TAMs) are found abundantly in solid tumors and evidence suggests that they play a significant role in modulation of other immune cells in the microenvironment through the expression of cytokines, growth factors and other products (44). As is true of macrophages in general, the phenotypes of TAMs exist along a spectrum, with pro- and anti-tumorigenic properties (86). Tumor-derived mediators and products activate signaling pathways that define both the metabolism and behavior of TAMs (86). Although numerous and with the capacity to mount an anti-tumor immune response, TAMs are often pro-tumorigenic.

Isolated TAMs from various human and mouse cancers have been shown to be lipid-rich and to functionally have enhanced lipid uptake via CD36 (87). Metabolism of fatty acids is an essential determinant of TAM phenotype (87, 88). Interfering with fatty acid oxidation, for example, using the drug etomoxir, blocked the development of immunosuppressive TAMs (87, 88). Concomitant with increased suppressive phenotypes was increased expression of CD36 (87). Significantly, etomoxir-treated TAMs or those isolated from CD36 KO tumors did not have a suppressive phenotype and blocked the proliferation of multiple tumor cell types in in vitro co-culture experiments (87).

In the analysis of human tumors, single-cell sequencing data from breast and lung cancers revealed that CD36 is predominantly expressed on TAMs, and that these TAMs have significantly higher expression of CD36 compared with control tissues (87). Elevated expression of CD36 in TAMs correlated with genes involved in TAM differentiation and fatty acid oxidation (87). Published gene array datasets from renal cancer patients again showed that CD36 expression correlated with a TAM gene expression signature (87). In gene array datasets from glioblastoma patients, there was a strong correlation between the genes involved in fatty acid oxidation and CD36 (87). This supports work in mice as relevant to human cancers, and that metabolic programming of TAMs is CD36-dependent. Taken together, these data provide support for immunomodulation through metabolic manipulation as a potential therapeutic approach in cancer.

The liver is a major site of tumor cell metastasis and patients with liver metastases have an overall poor prognosis. Similar to TAMs, metastasis-associated macrophages (MAMs) have been phenotypically characterized in mice and humans and the data compellingly suggest an essential role for CD36 in an immunosuppressive, pro-tumorigenic role (89, 90). Metastatic Lewis lung carcinoma cells injected intraperitoneally into the spleen resulted in fewer metastatic nodules, overall decreased tumor area, reduced tumor proliferation and blood vessel number in CD36 KO mice compared to controls. Contribution from hepatocyte CD36 was ruled out using liver-specific CD36 KO mice (90). Depletion of macrophages with clodronate liposomes suppressed liver metastasis; liver metastases were also markedly reduced in macrophage-specific CD36 KO mice (90).

MAMs isolated from tumors had an increased number of lipid droplets compared to macrophages isolated from non-diseased liver (90). MAMs had higher expression of CD36 and increased uptake of fatty acids. Notably, the expression of other fatty acid uptake proteins (Fatty Acid Binding Proteins 1-6) was unchanged or decreased (90). In this model, the investigators uncovered a novel delivery mechanism for fatty acids derived from tumor cells: extracellular vesicles (90). These vesicles were shown to transfer fatty acids to MAMs, and they were subsequently trafficked to the mitochondria. Uptake of these tumor-derived vesicles correlated with CD36 expression and was significantly decreased in CD36 KO MAMs (90).

Like TAMs, MAMs were predominantly immunosuppressive. This suggests that the TME in some way promotes such a phenotype. Analysis of lipids in the liver metastatic TME showed enrichment for unsaturated and monounsaturated fatty acids; in vitro experiments demonstrated that CD36 expression was required for the immunosuppressive phenotype mediated by these lipids (90). In particular, oleic acid was shown to have an anti-inflammatory effect on bone marrow-derived myeloid cells and promoted the development of immunosuppressive dendritic cells and macrophages (88, 91). The mechanism was linked to metabolism: upregulation of pathways related to mitochondrial respiration, fatty acid oxidation, lipid droplet formation and downregulation of fatty acid synthesis pathways and desaturases. Oleate decreased the expression of mature macrophage markers, increased expression of signature TAM markers, including those associated with an inhibitory phenotype, and induced high levels of arginase activity (88, 91). Remarkably, these experiments revealed that oleate alone was sufficient to induce the differentiation of TAMs at phenotypical and functional levels. In vivo, short-term feeding of mice a diet supplemented with oleic acid increased macrophages with a suppressive phenotype in mesenteric adipose tissue (92).

Lipid droplets, which are a prominent common characteristic of immunosuppressive macrophages, play a key role in maintenance of phenotype: they provide a reservoir of fatty acids to support the metabolism and functions of the immunosuppressive state (88). Interference with enzymes and associated proteins involved in mobilization of fatty acids from lipid droplets disrupted the development of the immunosuppressive phenotype (88, 91, 93).

When the immune cell population of the TME was examined in macrophage-specific CD36 KO mice compared with controls, there was a diminished number of MAMs and an increased number of B cells and T cells, including cytotoxic CD8+ T cells (90). Expression of interferon-γand granzyme B were also significantly increased in these cytotoxic CD8+ T cells, suggesting greater anti-tumor potential. In human hepatic metastases, CD36 expression correlated with an M2-like, immunosuppressive phenotype in two datasets examined (90). These studies bolster the hypothesis that CD36 expression is essential to support the TAM and MAM immunosuppressive phenotype by contributing to lipid metabolism. Figure 4 summarizes key metabolic and suppressor pathways in TAMs.

The mechanism of CD36 upregulation in TAMs and MAMs is likely multifactorial. IL4 is a differentiating cytokine for M2 macrophages. Due to the oxidative stress in the TME, macrophages are exposed to IL4 in the context of oxLDL, and this has been shown to have a strong influence on CD36 expression (94). IL13, important in the differentiation of TAMs, also upregulates CD36 (95). Secretion of long-chain fatty acids, especially saturated and monounsaturated classes, likely leads to a feed-forward response that increases CD36 expression (90). Similarly, as the macrophages increase fatty acid metabolism and oxidation, this also has a feed-forward effect, driving up the expression of fatty acid uptake receptors, including CD36. Some of these same molecules also activate PPARγ, which again increases CD36 expression (96).

In addition to TAMs, there are also myeloid-derived suppressor cells (MDSCs) in the TME (97–99). MDSCs are divided into polymorphonuclear (PMN-MDSCs) and monocytic (MO-MDSCs) subtypes (97, 99). They are recruited to the TME as a result of signals similar to TAMs. PMN-MDSCs stimulate T-cell tolerance in an antigen-specific manner, while MO-MDSCs are potently immunosuppressive and exert effects in an antigen-independent manner (97, 100). MDSCs are believed to arise in response to sustained pressure on myelopoiesis as a result of chronic infections, cancer and autoimmune diseases (100). The initial proliferation and training of MDSCs occurs in the bone marrow and spleen, but importantly, the phenotype is then refined in the peripheral sites of disease, where condition-specific growth factors, cytokines and derived products exert influence (101).

During differentiation, reports show that MDSCs have high rates of glycolysis, glutamine uptake and reliance on the tricarboxylic acid cycle and pentose phosphate pathway for energy (102). Investigations have shown that activation of MDSCs in the TME led to upregulation of CD36 via STAT3 and STAT5, and increased lipid uptake (103). Phenotyping of spleen vs. tumor-infiltrating MDSCs revealed significant differences in mitochondrial mass, oxygen consumption and fatty acid oxidation, all increased in those cells educated in the TME (97). As has been suggested for other immunosuppressive myeloid cells, fatty acid oxidation was the source of intermediates for the biosynthesis of fatty acids, amino acids, proteins and ribose for rRNA. These feed into biosynthetic pathways for anti-inflammatory molecules, including IL10, programmed cell death protein-1 and inducible nitric oxide in MO-MDSCs. The development of PMN-MDSCs has also been shown to be dependent upon lipid metabolism to produce reactive oxygen species and myeloid peroxidase, which limit dendritic cell antigen presentation and T-cell activation (104). Tumor-secreted PGE2 and fatty acid transport protein (FATP) 2 play essential roles in their determination (85, 105). Thus, strategies that interfere with FATP2 or CD36 curb MDSC suppressive effects on effector cells (105, 106). While many of these studies have been done in mouse models, recent single-cell RNA sequencing analyses showed greater than two-fold increases in CD36 expression in MDSCs from head and neck and breast cancer (107). A comparison between PMN-MDSCs and MO-MDSCs is shown in Figure 5 .