MicroRNA Regulated Macrophage Activation in Obesity

Obesity is a metabolic disease of ever-increasing prevalence characterized by excess accumulation of white adipose tissue resulting from a combination of overnutrition, energy imbalance, and genetics. In contrast to its original characterization as an inert tissue depot of triglycerides, adipose tissue has since been recognized as a dynamic organ orchestrating metabolic, endocrine, and immune responses.[1] Accumulating evidence from recent decades has linked obesity to chronic low-grade inflammation, which underlies obesity-associated insulin resistance, diabetes mellitus, metabolic syndrome, and cardiovascular diseases. It has become clear that the role of white adipose tissue has exceeded its original notion for energy storage. In addition to adipokines, adipose tissue also produces a variety of cytokines and chemokines and plays an important function as an endocrine organ for orchestrating systemic physiology. Chronic overnutrition-induced obesity is also characterized by two hallmark responses, insulin and metainflammation, which are the causal factors for obesity-associated metabolic syndromes and other health risks.

Adipose tissue macrophages (ATMs) represent the largest immune population in the adipose tissue stroma and contribute essential support for tissue remodeling, metabolic homeostasis, as well as inflammatory responses under obesity stress. In healthy lean adipose tissues, ATMs exert crucial functions for maintaining immunological and metabolic homeostasis, including immune regulation, efferocytosis, lipid buffering, and angiogenesis.[2, 3] Obesity is associated with a 10-fold increase in macrophages in adipose tissue due to recruitment of circulating monocytes and/or local proliferation of tissue macrophages.[4, 5] Within expanding adipose tissue, macrophage release pro-inflammatory cytokines such as TNF-α and form crown-like structures that surround dying adipocytes, the latter of which is a histologic hallmark of inflammation within adipose tissue.[6, 7] Such pro-inflammatory environments orchestrated by macrophages contribute significantly to long-term high-fat diet-induced insulin resistance in mice.[8] In addition, multiple studies support that pro-inflammatory macrophages inhibit adipogenitor cell proliferation and differentiation.[9, 10, 11, 12] Through these mechanisms, ATMs profoundly alter adipose tissue functions, impact local and systemic metabolism, and orchestrate pathological changes during obesity stress.

Tissue-residing macrophages including ATMs are highly plastic, allowing them to adopt diverse functions in response to various stimuli, such as cytokines, infections, and chemicals.[13] Researchers have broadly classified macrophages by their activation states as M1 “classically” activated or M2 “alternatively” activated, and also have defined macrophage responses with this classification. In the case of ATMs, obese conditions induce a switch from an anti-inflammatory M2 to a pro-inflammatory M1 activation state. M2 ATMs contribute to homeostasis maintenance and tissue remodeling, whereas M1 ATMs promote insulin resistance through the activation of pro-inflammatory pathways (such as JNK, ERK, p38, and NF-κB) that target insulin receptor signaling.[14] Macrophages regulate tissue functions mainly through the release of secreted products, including cytokines, reactive molecules, and extracellular RNAs. Cytokines released by M1 macrophages are usually pro-inflammatory, such as TNF-α, IL-1β, IL-12, and IL-23;[15, 16] M2 macrophages, on the other hand, are known to secrete anti-inflammatory cytokines and growth factors, such as IL-10 and TGFβ.[17] Interestingly, both M1 and M2 macrophages produce cytokine IL-6 that has both proinflammatory and anti-inflammatory functions[18] and is known to promote insulin secretion[19] or skeletal muscle tissue repair, depending on specific conditions.[20] M1 macrophages also release reactive molecules such as nitric oxide (NO) and reactive oxygen species (ROS),[21, 22] which impose profound impacts on molecular functions and are critically involved in obesity-associated pathology. [23, 24, 25, 26] In addition, recent studies suggest extracellular RNAs secreted by macrophages regulate tissue functions: cultured macrophages release microRNAs (miRNAs) in response to pro-inflammatory stimuli[27] and transfer miR-142 and miR-223 to co-cultured hepato-carcinoma cells;[28] and ATMs from obese mice release miR-155 and cause insulin resistance in insulin target cells.[29] However, the role of extracellular RNAs in regulating macrophage activation and tissue function is still largely unknown and requires further investigation.

Within the past decade, researchers have elucidated several key regulators that drive heterogeneous macrophage activation, including the signal transducer and activator of transcription (STAT) family, PPARγ/LXR, CREB-C/EBP, and interferon regulatory factors (IRFs). Interferons are one of the first cytokines identified as activators of the pro-inflammatory macrophage phenotype.[30, 31, 32] Interferon regulatory factors (IRFs) have long been known to control the activity of interferons (IFN), and increasing evidence supports their important role in regulating macrophage activation.[33] Of these, IRF1, 2, 5, and 6 drive macrophages toward the M1 pro-inflammatory type. Pro-inflammatory activation of murine macrophages elicited by LPS or IFNγ are inhibited by knockout of IRF1 or IRF2.[34] IRF5 promotes “M1-like” activation of human peripheral blood macrophages and inhibits expression of M2-associated genes.[35] IRF6 was recently found to suppress “M2-like” activation of murine bone marrow-derived macrophage (BMDM) by inhibiting PPARγ gene transcription.[36] In contrast, some other members of the IRF family, such as IRF3, 4, and 9, mediate anti-inflammatory signaling through type I Interferon responses. In human microglia, IRF3 mediates anti-inflammatory responses by activating the PI3K/Akt pathway and promoting “M2-type” cell activation.[37] Induction of IRF4 by IL4 contributes to “M2-like” alternative macrophage priming in murine BMDMs. In addition, during the “M2-like” activation of murine macrophages triggered by parasites or fungi, researchers observed significant Lysine Demethylase 6B (KDM6B, or Jmjd3)-mediated histone demethylation at the locus of the Irf4 gene, further implicating its involvement in the activation process.[38] Another member of this family, IRF9, is believed to mediate interferon tau-induced anti-inflammatory responses and M2 activation of murine BMDMs.[39]

The Janus Kinase-STAT pathway can drive M1 activation. IFNγ is thought to induce expression of M1-associated genes by triggering dimerization of STAT1,[40] and mice bearing STAT1 deficiency fail to respond to IFNγ and IFNα.[41] In addition, LPS-induced IFNβ activation enhances the formation of STAT1-STAT2 heterodimers to mediate the induction of M1-associated genes by forming the IFN-stimulated gene factor 3 complex.[42] Another member of the STAT family, STAT6, is associated with M2 macrophage activation. IL-4 and IL-13 are considered two critical M2 inducers, as suggested by several in vitro and in vivo studies,[43, 44, 45] and STAT6 can mediate IL-4a signaling and regulate expression of M2 signature genes.[43, 46] STAT6 signaling is further mediated by monocyte chemoattractant protein-1-induced protein (MCPIP), which induces M2-promoting ROS, endoplasmic reticulum stress, and autophagy.[47]

Another important group of macrophage activation regulators are in the CCAAT-enhancer-binding proteins (C/EBP) family, in particular, C/EBPα, β, and σ. C/EBPβ mediates signaling of toll-like receptor (TLR) and cAMP-responsive element-binding protein (CREB) that induce expression of arginase 1 (ARG1), an M2 marker protein in mice. C/EBPβ also promotes the expression of another M2 signature gene, mannose receptor c-type 1 (Mrc1), upon induction by CREB. Deletion of CREB binding sites in the promoter region of C/EBPβ consistently abolishes muscle tissue repair (an M2 macrophage-mediated function in mice) and inhibits the expression of numerous M2 signature genes, including macrophage scavenger receptor 1 (Msr1), IL-10, IL-13 receptor subunit, receptor α1 (Il13ra1), and Arg1 within macrophages; deletion of these binding sites does not alter the levels of pro-inflammatory M1 signature genes.[48] Further, CREB exerts the negative feedback regulation of pro-inflammatory TLR signaling mediated by MSK1/2 kinases through the induction of IL-10 production and dual specificity protein phosphatase 1 to limit inflammation.[49, 50] In contrast to C/EBPα, C/EBPσ induces M1 pro-inflammatory responses in murine BMDMs.[51] Another member of the C/EBP family, C/EBPα, is believed to be necessary for both M1 and “M2 activation of murine macrophages.[52] All these studies have suggested a critical role for CREB-C/EBP signaling in regulating heterogeneous activation of macrophages.

The lipid metabolism regulator Peroxisome Proliferator-Activated Receptor gamma (PPARγ) can negatively regulate numerous pro-inflammatory genes.[53, 54] Knockout of PPARγ in murine myeloid cells can abolish their “M2-like” activation and enhance susceptibility to inflammation-associated health issues including obesity, insulin resistance, and glucose-intolerance.[55] In addition, in murine thioglycollate-elicited macrophages and human peripheral blood monocytes, PPARγ is thought to mediate the signaling of IL-4 and IL-13, which are well-known inducers of “M2-like” phenotypes.[56] In cultured murine macrophages, PPARγ interacts with STAT6 as a cofactor to facilitate the induction of genes under the regulation of PPARγ.[57] Similarly, liver X receptors (LXRs; nuclear transcription factors that heterodimerize with the retinoid X receptor RXRα) contribute to reduced pro-inflammatory signaling.[58, 59, 60] Indeed, LXRs have long been recognized for their role in the amelioration of autoimmune diseases and regression of atherosclerotic plaques, which are extensively orchestrated by M2 macrophages.[61, 62]

In recent years, increasing evidence supports the critical roles of miRNAs in regulating macrophage activation by targeting key regulators. PPARγ directly binds upstream of miR-223 and mediates anti-inflammatory signaling by inhibiting the expression of Nuclear Factor of Activated T-Cells 5 (NFATt5) and RAS p21 Protein Activator 1 (RASA1), thereby promoting “M2-like” activation. Deletion of miR-223 can abolish PPARγ-regulated M2 activation of murine macrophages in vivo and ex vivo.[63] In addition, miR-223 inhibits the expression of PBX/Knotted 1 Homeobox 1 (Pknox1), thus suppressing NFκB/JNK signaling and “M1-like” pro-inflammation activation.[64] miR-223 also inhibits the pro-inflammatory differentiation of murine intestinal macrophages by targeting C/EBPβ,[65] and blunts the transition of THP-1 cells and peripheral human blood monocytes toward inflammatory macrophages.[66]

In contrast to miR-223, miR-155 regulates pro-inflammatory “M1-like” activation. TLR agonists (LPS, hypomethylated DNA, or Pam3CSK4) and pro-inflammatory cytokines (TNF-α, IFNβ, or IFNγ) increase miR-155 expression in cultured murine macrophages.[67, 68] In alcoholic liver disease, miR-155 mediates an NF-κB-regulated pro-inflammatory response by stabilizing TNF-α mRNA, thus promoting its synthesis.[69] In addition to directly boosting pro-inflammatory factors, miR-155 also contributes to inflammation by suppressing signaling on the anti-inflammatory side, targets of which include suppressor of cytokine signaling 1 (SOCS1), phosphatidylinositol-3,4,5-trisphosphate 5-phosphatase 1 (SHIP1), [68, 70] and IL13RA1/STAT6.[71] Interestingly, despite extensive support for a pro-inflammatory role of miR-155, it also reduces the production of pro-inflammatory cytokine in murine BMDMs by targeting TGF-β Activated Kinase 1/MAP3K7 Binding Protein 2 (TAB2),[72] suggesting dual roles, each of which are likely dependent on specific conditions.

Inhibiting the expression of key transcription factors is an important mechanism by which microRNAs manipulate macrophage functions, and is exemplified by the action of miR-125 and Let7c. miR-125 mediates pro-inflammatory signaling of IFNγ by suppressing the expression of the M2-promoting transcription factor IRF4, and thus contributes to an “M1-like” phenotype;[73] it can, however, also suppress pro-inflammatory signaling in other macrophage types.[74] In contrast to miR-125, Let-7c reduces the levels of inflammation-associated factors such as C/EBPσ, IL-12, and major histocompatibility complex (MHC) class II, and is required for “M2-like” phenotypes of macrophages.[51] In addition, other miRNAs have also been reported to manipulate macrophage activation, including the M1-promoting miR-9, miR-127, miR-124,[75, 76, 77] and the M2-promoting miR-132, and miR-146a.[78, 79] Interestingly, miR-21 contributes to both M1 and M2 activation, although each under different conditions in separate studies.[80, 81]

The profound impact of miRNAs on activation of macrophages that critically orchestrate obesity-associated adipose tissue pathogenesis suggests their potential as promising therapeutic targets. However, despite intensive research on miRNA involvement in diseases, most studies focused on cancer-related conditions, and therefore, less is known about their role in adipose tissue of healthy and obese subjects. Another challenge of addressing miRNA-regulated macrophage activation in obesity originates from the complex nature of macrophage heterogeneity: despite the widely adopted M1/M2 polarization system, increasing studies suggest that this model largely built using in vitro cell cultures is too simple to depict the multi-faceted and dynamic activation states of tissue-residing macrophages. This is not surprising considering the highly complex and diverse microenvironments in healthy or diseased tissues. In recent years, the accumulating evidence [82, 83, 84] suggests macrophage activation states display spectrum-like patterns across a variety of tissues and species (Figure 1); however, a systemic model that comprehensively annotates complex macrophage features under different conditions is still lacking.

Figure 1.

Schematic description of multi-faced macrophage activation. Upon monocyte differentiation into macrophages, dependent on specific antigens or stimuli they encounter and controlled by intracellular regulators, macrophages may adopt different phenotypes that 1) orchestrate immune defense and/or tissue inflammation; 2) conduct anti-inflammation and/or tissue remodeling functions; 3) maintain a resting/pre-activation state until further stimulation; or other functions dependent on the specific micro-environment of the tissue loci. Secretory factors, including cytokines, reactive chemicals, and exRNAs, are an important approach for macrophages to impact tissue functions.

Advances in high-throughput sequencing and single-cell technologies allow in-depth analyses of cell populations to identify distinct subsets and dissect regulatory mechanisms underlying cell function. However, currently available algorithms are not tailored to depict macrophage activation and often result in ambiguous characterization of dynamic activation state changes in vivo. To address this major knowledge gap, we generated single-cell transcriptome data of ATMs from healthy and obese mice and primary bone marrow-derived monocytes and macrophages to develop new high-resolution algorithms. The outcome was the creation of a two-index platform, MacSpectrum (https://macspectrum.uconn.edu) that enables comprehensive high-resolution mapping of monocyte/macrophage activation states from diverse mixed cell populations. The capability of MacSpectrum to dissect macrophage heterogeneity was well-supported by its performance on the samples from human and murine species, under in vitro and in vivo conditions, and in bulk and single-cell sequencing formats. Importantly, MacSpectrum revealed an unprecedented sequential activation pattern of monocytes/macrophages in obesity and unique cell programs under the regulation of miRNAs. The performance of MacSpectrum suggests that novel bioinformatic algorithms tailored to macrophage study could provide promising strategies to address the challenges of investigating miRNA-regulated activation in obesity and facilitate more focused therapeutic development through sub-population separation, functional annotation, and signature gene identification.