Twenty Novel Disease Group-Specific and 12 New Shared Macrophage Pathways in Eight Groups of 34 Diseases Including 24 Inflammatory Organ Diseases and 10 Types of Tumors

The mechanisms underlying pathophysiological regulation of tissue macrophage (Mφ) subsets remain poorly understood. From the expression of 207 Mφ genes comprising 31 markers for 10 subsets, 45 transcription factors (TFs), 56 immunometabolism enzymes, 23 trained immunity (innate immune memory) enzymes, and 52 other genes in microarray data, we made the following findings. (1) When 34 inflammation diseases and tumor types were grouped into eight categories, there was differential expression of the 31 Mφ markers and 45 Mφ TFs, highlighted by 12 shared and 20 group-specific disease pathways. (2) Mφ in lung, liver, spleen, and intestine (LLSI-Mφ) express higher M1 Mφ markers than lean adipose tissue Mφ (ATMφ) physiologically. (3) Pro-adipogenic TFs C/EBPα and PPARγ and proinflammatory adipokine leptin upregulate the expression of M1 Mφ markers. (4) Among 10 immune checkpoint receptors (ICRs), LLSI-Mφ and bone marrow (BM) Mφ express higher levels of CD274 (PDL-1) than ATMφ, presumably to counteract the M1 dominant status via its reverse signaling behavior. (5) Among 24 intercellular communication exosome mediators, LLSI- and BM- Mφ prefer to use RAB27A and STX3 than RAB31 and YKT6, suggesting new inflammatory exosome mediators for propagating inflammation. (6) Mφ in peritoneal tissue and LLSI-Mφ upregulate higher levels of immunometabolism enzymes than does ATMφ. (7) Mφ from peritoneum and LLSI-Mφ upregulate more trained immunity enzyme genes than does ATMφ. Our results suggest that multiple new mechanisms including the cell surface, intracellular immunometabolism, trained immunity, and TFs may be responsible for disease group-specific and shared pathways. Our findings have provided novel insights on the pathophysiological regulation of tissue Mφ, the disease group-specific and shared pathways of Mφ, and novel therapeutic targets for cancers and inflammations.


INTRODUCTION
As we reported previously (1)(2)(3)(4)(5), monocytes and macrophages (Mϕ) play significant roles in driving cardiovascular inflammations induced by various metabolic cardiovascular disease-related danger-associated molecular patterns (DAMPs) such as hyperlipidemia, hyperglycemia, hyperhomocysteinemia, and chronic kidney disease. Also, monocyte and Mϕ differentiation during various metabolic cardiovascular diseases has been characterized (5). In recent years, a complicated relationship between the bone marrow, monocytes/Mϕ, and the development of atherosclerotic plaques has begun to be revealed (6). The roles of Mϕ in modulating foam cell formation (7) and inflammation resolution (8) have also been reported. Moreover, several additional developments have been made, including in tissue Mϕ characterization, Mϕ polarization, subset characterization (9), clonal production, and trained immunity (trained immunity) (10). A recent success on the CANTOS trials with anti-interleukin-1β (IL-1β) monoclonal antibody Canakinumab (11) further emphasized the significant roles of inflammatory cytokines in the pathogenesis of metabolic cardiovascular diseases, in which monocytes and Mϕs secrete cytokines in large numbers and amounts in response to the stimulation of DAMPs or conditional DAMPs that we had reported (12). However, the molecular mechanisms underlying several vital aspects of Mϕ remain poorly determined: (1) the expression of Mϕ markers and Mϕ transcription factors, (2) pathways in the regulating roles of Mϕ in various diseases, and (3) the differentiation and transdifferentiation of tissue Mϕ subsets.
Macrophages play significant roles in the pathogenesis of various diseases including cardiovascular (13), metabolic (14), infectious (15), respiratory (16), digestive (17), autoimmune (18), and many types of cancers (19,20). However, three important questions remain: whether Mϕ use the same pathways and play the same roles or whether they use disease-specific pathways and play disease-specific roles in addition to the shared roles and pathways; whether 10 Mϕ subset markers and newly identified 27 TFs (21) and other 18 Mϕ subset TFs are differentially expressed in tissues; and whether all these newly reported proinflammatory features of Mϕ are differentially expressed in various tissues. Addressing these issues will improve our understanding of the disease-specific and shared roles and pathways of Mϕ in the pathogenesis of various diseases and cancers and will lead to the identification of novel therapeutic targets specific to those diseases and cancers.
Determining novel mechanisms underlying macrophage disease-specific and shared pathways first requires an understanding of how macrophages respond to environmental and tissue functional cues from several aspects such as cell surface receptor signaling, cell-cell interaction receptor signaling, cell-cell communication signaling, intracellular immunometabolic pathways, and transcription factors. Macrophages are present in almost all tissues of the body, displaying distinct location-specific phenotypes and gene expression profiles (22). In addition to central roles in innate immunity and as modifiers of the adaptive immune responses, tissue Mϕ play supportive functions to the tissues they reside in Hoeksema and Glass (13).
Several cell-surface-specific mechanisms could promote macrophage heterogeneity. First, by varying stimuli such as different cytokines or DAMPs to act on Mϕ cell surface receptors, Mϕ can be "polarized" into as many as total 10 macrophage subsets including the typical proinflammatory M1 Mϕ and anti-inflammatory M2 Mϕ (9). Also, as many as 28 T cell co-stimulation receptors and co-inhibition/immune checkpoint receptors as cell-cell contact signaling receptors may serve as a second cell surface mechanism to shape the antigen-presenting functions of Mϕ (23). Finally, recent reports showed that exosomes are local and distal cell-cell communication vehicles (24), which may serve as the third cell-surface mechanism.
We also reported that exosomes secreted from immune cells such as macrophages might propagate inflammation from the first inflamed cells to the secondary inflammatory cells (25); exosomes regulate inflammation and immune responses via intercellular exosome communications (26). In addition to cell surface mechanisms, the M1 Mϕ proinflammatory metabolic pathways and M2 metabolic pathways have been identified (27) as the intracellular mechanisms. Another development is the recognition of innate immune memory (trained immunity) pathways such as increased glycolysis pathway, enhanced acetyl-CoA (activated acetate, cellular acetyl donor) generation, and increased expression of mevalonate pathway enzymes (28,29), which were "zoomed in" to through extensive metabolic remodeling of 2,722 experimentally elucidated pathways (https://metacyc.org). Recent reports also identified numerous transcription factors involved in Mϕ differentiation and Mϕ subset polarization.
Regardless of significant progress in the field, several important questions remain. The first is whether, under various disease conditions, Mϕ uses both disease-specific signaling pathways and shared pathways. To determine the molecular mechanisms underlying disease-specific and shared pathways of macrophages, we examined macrophage features from tissue-specific differential expression of Mϕ cell surface markers, transcription factors, Mϕ cell-cell contact signaling receptors (T cell co-stimulation receptors and co-inhibition/immune checkpoint receptors), Mϕ cell-cell communication vesicle-exosome biogenesis and docking machinery, and Mϕ intracellular metabolism pathways such as bioenergy metabolism pathways and trained immunity (innate immune memory) pathways. We then narrowed in on the following questions: whether tissues have differential expression of Mϕ subset markers and transcription factors, whether tissue Mϕ have different inflammatory and trained immunity (innate immune memory) potentials, and whether tissue Mϕ have different bioenergy metabolism pathways and trained immunity pathways. To address these issues, we determined the expression of 207 Mϕ genes in several tissues such as lung, liver, intestine, spleen, and bone marrow-derived, including 10 subset markers, 45 transcription factors (TFs), and 127 other regulatory genes by analyzing the microarray experimental data sets that other investigators deposited in the NIH-NCBI GEO DataSets database, as shown in Figure 1. Of note, we pioneered this type of novel experimental data mining analysis in 2004 (30), which has allowed us to generate original findings and novel hypotheses for our experimental projects. The significant differences between our experimental database mining approaches and traditional literature reviews are detailed in Table 1. Based on the expression changes of 31 ten-Mϕ-subset markers and 45 TFs in eight groups of a total of 34 diseases, including 10 types of cancers, we have identified 20 novel disease group-specific and 12 new shared macrophage pathways. In addition, we also found new signaling and metabolic pathways underlying tissue Mϕ subset regulation in pathophysiological conditions as novel mechanisms for Mϕ heterogeneity, which serve as novel therapeutic targets specific to cancers and inflammations. Eight Groups of a Total  of 34 Diseases, Including 24 Inflammatory  Organ Diseases and 10 Types of Cancers; and Both Shared and Disease-Specific Pathways for Each Group of Disease/Tumor Have Been Identified Mϕ play a key role in the pathogenesis of various diseases. However, two critical questions remain: whether Mϕ use the same pathways and play the same roles or whether they use disease-specific pathways and play disease-specific roles in addition to the shared roles and pathways. To improve our understanding of the roles of Mϕ in various diseases, we examined the expression of 31 Mϕ subset markers and 45 Mϕ transcription factors (Table 2) in eight groupings of a total of 34 diseases, including four types of autoimmune diseases, four types of cardiovascular diseases, four types of digestive diseases, four types of infectious diseases, four types of metabolic diseases, four types of respiratory diseases, five types of digestive cancers, and five types of other cancers. As shown in Table 3A, some Mϕ markers were upregulated in more than 30% of the 34 diseases, including three M1 markers, CXCL11, CXCL10, and CXCL9, 2 M2 markers, CCL18 and IL1RN, and one M4 marker, MMP7, suggesting that these markers may play significant roles in the pathogenesis of the diseases. In addition, the diseases with Mϕ markers upregulated in more than 30% of the 34 diseases were of eight types, including #5 myocardial infarction, #6 coronary artery disease, #10 gastritis, #11 Crohn's ileitis, #12 Crohn's colitis, #29 esophageal cancer, #32 ovarian carcinoma, and #34 renal carcinoma, suggesting that these diseases may have significant Mϕ marker activities with the pathogenic processes. Moreover, as shown in Table 3B, some Mϕ transcription factors (TFs) were upregulated in more than 30% of the 34 diseases, including M1 TF STAT1 and three other Mϕ TFs such as HMGA1, E2F3, and NME1, suggesting that these TFs play significant roles in the pathogenesis of the diseases. Furthermore, the diseases having Mϕ TFs upregulated in more than 30% among the 34 diseases were of six types, including #6 coronary artery disease, #12 Crohn's colitis, #28 hepatocellular cancer, #29 esophageal cancer, #32 ovarian carcinoma, and #33 lung cancer, suggesting that these diseases have significant Mϕ TF activities with the pathogenic processes.

Expression of 31 Mϕ Markers and 45 Mϕ TFs Is Modulated in
We then determined whether there are disease-specific signaling pathways and shared pathways based on the expression changes of Mϕ subset markers and Mϕ TFs in eight groups of 34 diseases and tumors. After analyzing the Ingenuity Pathway Analysis results of the top 10 pathways in both upregulated and downregulated Mϕ subset markers and Mϕ TFs, respectively, we compared all the upregulated pathways, downregulated pathways, and the pathways either upregulated or downregulated in some diseases (upper panel, middle panel, and lower panel of Tables 3C,D). As shown in Table 3C, we found three disease-specific pathways upregulated and 14 FIGURE 1 | Flow chart of database mining strategy and two parts of data organization. (I) database mining strategy was used to analyze the macrophage (Mϕ) marker tissue expression profiles in physiological conditions. (II) Mϕ marker and Mϕ transcription factor expression changes were analyzed on experimental data from the microarray datasets in different diseases.
NIH  disease-specific pathways downregulated. As shown in Table 3D, we found 16 disease-specific pathways upregulated and 16 disease-specific pathways downregulated. We also compiled a list of pathways that are shared in several groups of diseases and tumors. As shown in Table 3A, we found that among 21 diseaseupregulated pathways, one pathway communication between innate and adaptive immune cells is shared among eight groups of diseases. We also found that three pathways, namely the role of hypercytokinemia/hyperchemokinemia in the pathogenesis of influenza, agranulocyte adhesion and diapedesis, and granulocyte adhesion and diapedesis, are shared by seven groups of diseases; four pathways, namely differential regulation of cytokine production in Mϕ and T helper cells by IL-17A and IL-17F, IL-10 signaling, the role of cytokines in mediating communication between immune cells, and pathogenesis of multiple sclerosis, are shared by 7 groups of diseases; and one pathway, altered T cell and B cell signaling in rheumatoid arthritis, is shared by five groups of diseases. In contrast, as shown in Table 3B, among 42 disease-downregulated pathways, 22 (52%) pathways are shared by two or more groups of disease, and 20 diseasespecific downregulated pathways may be important for the pathogenesis of the diseases. Furthermore, as shown in Table 3C, 13 Mϕ pathways are upregulated and/or downregulated in some disease groups in two different directions, suggesting that some Mϕ functional pathways are modulated in diseasespecific manners.
These results suggest that the expression changes of Mϕ TFs in eight groups of 34 diseases are more diseasespecific than that of Mϕ subset markers, allowing the identification of 20 disease-specific and 12 shared (more than 4 groups of diseases) modulations of Mϕ TFs pathways in eight groups of 34 diseases. As shown in Table 3E, in detail, we found five upregulated disease-specific pathways in autoimmune diseases, three upregulated disease-specific pathways in cardiovascular diseases, two upregulated diseasespecific pathways in digestive inflammatory diseases, four upregulated disease-specific pathways in infectious diseases, one disease-specific pathway in metabolic disease, one diseasespecific pathway in respiratory disease, one upregulated pathway (shared with autoimmune disease) and one downregulated specific pathway in digestive tumors, and two upregulated disease-specific pathways in other tumors. In addition, we found 12 pathways that are shared by more than four groups of diseases and tumors. These results demonstrate for the first time that the expressions of Mϕ TFs are modulated in both disease-specific, and shared signaling pathways; these results provide insights on the roles of Mϕ in various diseases and novel therapeutic targets for modulating Mϕ TFs and Mϕ functions for those diseases and tumors. These results have also demonstrated for the first time that certain "high hierarchical" functional pathways in pathological Mϕ are more important in the pathogenesis of various diseases than other pathways, making them novel pathological Mϕ-specific therapeutic pathways; disease-specific 2 | A total of 207 macrophage (Mϕ)-related regulator genes in seven representative groups were studied in this paper, including 31 Mϕ subset marker genes, 18 Mϕ subset transcription factor genes (TF), 27 Mϕ general transcription factor genes, 28 T cell co-stimulation and co-inhibition receptor genes, 56 bioenergetics pathway enzyme genes, 23 trained immunity (innate immune memory) pathway genes, and 24 exosome biogenesis/docking mediator genes. Detailed information see Figure S4 Pentose phosphate pathway   Mϕ pathways are also important for the pathogenesis of the diseases and are disease-specific therapeutic targets.

Macrophages (Mϕ) in Lung, Liver, Spleen, and Intestine Express Higher M1 Mϕ Markers Than Lean Adipose Tissue in Physiological Conditions
To determine the novel mechanisms underlying disease-specific and shared macrophage pathways, we and others previously reported that metabolic disease risk factors serve as conditional danger-associated molecular patterns (conditional DAMPs) (12,31,32) and induce monocyte/Mϕ differentiation into Ly6Chigh-(1-3) and CD40+ proinflammatory monocytes (4), and accelerate vascular inflammation. Other studies also reported that, under certain experimental conditions such as stimulation with lipopolysaccharide (LPS) and interferon-γ (IFN-γ) for M1 polarization or interleukin-4 (IL-4) for M2 polarization (33), Mϕ can be polarized into multiple subsets including proinflammatory M1 and M4 Mϕ, anti-inflammatory M2, M(Hb), and Mhem Mϕ (34) (Figure 2A). However, whether the different physiological environments present in tissues affect the expression of residential Mϕ subsets markers and other regulators has not been studied (35). We hypothesized that various tissue environments with tissue differentiation potentials, DAMPs/conditional DAMPs, cytokines, and cellcell contacts induce differential expression patterns of Mϕ subset markers.
To examine this hypothesis in a very comprehensive manner, we collected 31 Mϕ markers of 10 Mϕ subsets, as reported in a recent publication (9) ( Table 2). As summarized in Figure 2A, the 10 Mϕ subsets perform various functions in regulating inflammation, immune responses, anti-oxidant, and tumor promotion (9). We included 10 Mϕ markers in our analysis, and these markers are differentially expressed in different Mϕ subsets. Furthermore, we also made a list of the transcription factors (TFs) of 18 Mϕ subsets, which are critical for the development and maintenance of seven out of 10 Mϕ subsets ( Table 2). The data for TFs that are critical for development of M2d, M4, and HA-Mac were not available at the time we conducted the analysis. Moreover, we included an additional 27 Mϕ TFs that were identified in Mϕ by RNA-Seq analysis (21).
By examining the expression of these Mϕ regulators in Mϕ microarray datasets deposited in the NI-NCBI GEO DataSets database (https://www.ncbi.nlm.nih.gov/gds/), we found that the expression of M1 Mϕ markers is higher in Mϕ that reside in tissues such as lung, liver, spleen, and intestine (LLSI) compared to lean adipose tissue Mϕ (ATMϕ), physiologically ( Figure 2B). This suggests that the majority of lung, liver, spleen, and intestine (LLSI) residential Mϕ are M1 Mϕ ( Figure 2C); therefore, these tissues have more potential to produce inflammatory responses than adipose tissue. In addition, as shown in Figures 2B,D, we found that: (1) Retnla, CD163, and MRC1 are relatively ATMϕspecific markers; (2) STAB1, NFE2L2, and SRXN1 are relatively bone marrow (BM)-specific Mϕ markers; (3) ARG1 is a relatively specific Mϕ marker for peritoneum, M2a, and M2c; (4) Chil4 is a relatively specific Mϕ marker for lung and M2a; (5) IL1B is a relatively specific Mϕ marker for liver and M1; (6) PDE4DIP and HMOX1 are relatively specific Mϕ markers for spleen, M2a, and MOX; and (7) CXCL9 is a relatively specific Mϕ marker for small intestine and M1. Our results suggest that these Mϕ subset markers modulated in the tissues may play important roles in tissue-specific Mϕ functions and subset compositions.    First, some Mφ markers were upregulated in more than 30% of the 34      First, some Mϕ transcription factors (TFs) were upregulated in more than 30% of the 34     Frontiers in Immunology | www.frontiersin.org

Pro-adipogenic Transcription Factors C/EBPα and PPARγ, and Proinflammatory Adipokine Leptin Upregulate the Expression of M1 Mϕ Markers
Adipose tissue releases more than 50 hormones, cytokines, and chemokines, collectively called adipokines, which regulate several physiological processes concerning energy, glucose metabolism, and immunity in an autocrine, paracrine, or systemic manner as well as several pathological processes including proinflammatory or anti-inflammatory processes, thereby contributing to insulin resistance and other inflammations (36). Adipose tissue from lean individuals releases anti-inflammatory adipokines such as adiponectin, transforming growth factor-β (TGF-β), IL-10, IL-4, IL-13, IL-1 receptor antagonist (IL-1Ra), and apelin. In contrast, obese adipose tissue secretes proinflammatory cytokines such as tumor necrosis factor-α (TNF-α), IL-6, leptin, visfatin, resistin, angiotensin II, and plasminogen activator inhibitor 1 (37). About one-third of obese adults and 10% of non-obese adults are metabolically healthy obese (MHO) (38,39). A series of reports suggest that patients with metabolically healthy obesity (MHO) have significantly higher rates of type II diabetes (40), metabolic syndrome (41), and chronic kidney disease (42) than metabolically healthy lean individuals. The molecular mechanisms underlying the pathogenesis of MHO remained poorly determined. In the search for master regulators responsible for MHO with the features of being pro-inflammatory/proatherogenic but anti-adipogenic, we reported that microRNA-155 (miR155) and, potentially, microRNA-221 are such master regulators for MHO (44). Deficiencies in those master regulators such as miR155 in an atherogenic apolipoprotein E (ApoE) −/− background led to the establishment of MHO in mice, significantly improving our understanding of the molecular mechanisms underlying MHO (43). Our recent findings further suggest that elevated adipokine resistin and leptin in a miR155 −/− /ApoE −/− MHO model fed a high-fat diet for 12 weeks may serve as a driver for the newly termed "second wave of inflammation status" in the MHO model (44). Along the same line, the issue of whether proinflammatory adipokines secreted by obese adipose tissues promote the expression of M1 Mϕ markers and other proinflammatory regulators remained poorly defined. We hypothesize that proinflammatory cytokine interferon-γ (IFN-γ) and lipopolysaccharide (LPS) upregulate M1 markers and regulators but not M2 (45). To test this hypothesis, we examined the expression of Mϕ markers and Mϕ TFs involved in M1 and M2 Mϕ polarization. As shown in Table 4A, when we examined the TF expression in the Mϕ polarization from human CD14+ monocytes, we made the following important findings: (a) M1 polarization promotes the expression of the Mϕ TFs for M1 and surprisingly also for M2c (IL-10 polarization); (b) the Krüppel-like family of transcription factor 4 (KLF4) was upregulated explicitly during M2a polarization (IL-4 polarization); and (c) four proinflammatory TFs (STAT1, STAT2, STAT3, and NF-kB) are more upregulated in M1 than in M2 polarization, suggesting their importance in promoting M1 Mϕ polarization (46).
One of our explanations is that pro-adipogenic TFs C/EBPα and PPARγ may tend to promote M2 in lean adipose tissues but that, in hypertrophic obese adipose tissues, these TFs may promote polarization of proinflammatory M1. Further detailed transcriptomic studies will be required to address this discrepancy.
(2) C/EBPα suppresses the expression of the M2a markers stabilin 1 (STAB1), coagulation factor XIII A chain (F13A1), chitinase-like 4 (Chil4), and Chil3. Since Mϕs are prototypic professional antigen-presenting cells (APCs) that modulate CD4+ T cell activation by providing T cell activation signal #1 and co-stimulation/co-inhibition-based signal #2(50), we also examined the expression of 28 T cell costimulation and co-inhibition (immune checkpoint) receptors (23), including 14 co-stimulation receptors, 10 co-inhibition receptors, and 4 dual-functional (both co-stimulation and coinhibition) receptors in tissue Mϕ (Figure 3A), as we reported previously (49). As shown in Figure 3A, we found that: (i) the Mϕ from LLSI express co-inhibition receptor CD274 (programmed death-ligand 1, PDL1) in much higher levels than ATMϕ and (ii) the Mϕ from peritoneum and ATMϕ express lower levels of CD274 than BM Mϕ. It has been reported that CD274 has significant reverse signaling activities (50). Antitumor immune response-enhancing transcription factor Forkhead box O (FoxO) inhibits CD274 expression (51), suggesting that CD274 expression may be responsible via reverse signaling for hiding immune response-enhancing features of tumor cells. Also, CD274 signals via conserved intracellular sequence motif "RMLDVEKC" inhibit JAK1-induced STAT3 activation and overcome interferon-mediated cytotoxicity (50). To correlate with the reported findings, our results suggest that: (i) peripheral tissue Mϕ, including LLSI Mϕ, express higher levels of T cell co-inhibition receptor CD274 than ATMϕ, to contribute to the establishment of immune tolerance at physiological conditions; and (ii) since our data suggested that LLSI tissue Mϕ are more proinflammatory than other Mϕ, higher expression of CD274 in LLSI Mϕ suggests that the high homeostatic and antiinflammatory functions of CD274 (programmed death-ligand 1, PD-L1) via its reverse signaling in Mϕ (52) may counteract the tissue Mϕ proinflammatory status (Figure 3B) in addition to CD274 inhibition of T cell activation via PD-1 (programmed  (30) are differentially expressed in macrophages from various tissues, and lung, liver, spleen, and small intestine upregulate more M1 Mϕ markers than M2 Mϕ markers in the physiologic condition in comparison to lean ATMϕ. (1) Retnla, CD163, and MRC1 are relatively adipose tissue-specific Mϕ (ATMϕ) markers; (2) STAB1, NFE2L2, and SRXN1 are relatively bone marrow-specific Mϕ markers; (3) ARG1 is a relatively specific Mϕ marker for peritoneum, Chil4 is a relatively specific Mϕ marker for lung, IL1B is a relatively specific Mϕ marker for liver, PDE4DIP and HMOX1 are relatively specific Mϕ markers for spleen, and CXCL9 is a relatively specific Mϕ marker for small intestine.  In addition to the above-discussed cell surface mechanisms such as Mϕ markers and cell-cell interaction (co-stimulation and co-inhibition/immune checkpoint) receptors, as cell-cell communication mechanisms of Mϕ and other cell types, exosomes can transport and deliver a large cargo of proteins, lipids, and nucleic acids and can modify cell and organ function. In addition to their key role as vehicles of intercellular communication, exosomes are increasingly recognized as biomarkers and prognosticators of disease (55). We reported that exosomes might modulate inflammation and immune responses (26) and propagate inflammation (25). We also examined the expression levels of 12 exosome biogenesis mediators and 12 exosome docking mediators in the tissue Mϕs ( Figure 4A). The results in Figure 4A showed that Mϕ from peritoneum, lung, liver, spleen, and small intestine prefer to use RAB27A and syntaxin 3 (STX3) than RAB31 and YKT6 in mediating exosome biogenesis and docking and that adipose tissue Mϕs use more Rab31, YKT6, and LGALS1 in mediating exosome biogenesis and docking. Of note, it has been reported that RAB27A-dependent exosome production inhibits chronic inflammation and enables acute response to inflammatory stimuli (56) and that microRNA-30c-2-3p regulates RAB31 and functions as an oncogene in gastric cancer tumorigenesis and development by interacting with glioma-associated oncogene homolog 1 (57). The results suggest that the differences in exosome biogenesis and docking in tissue Mϕs may be related to their proinflammatory functional status ( Figure 4B) as we reported previously (26), that Rab GTPases not only regulate the pathogenesis of cancer and neurodegenerative diseases (58) but may also regulate inflammation functions of Mϕ exosomes, and that syntaxin 3-identified homozygous likely deleterious variant (59) may regulate inflammatory Mϕ exosomes.

Levels of Immunometabolism Pathway
Enzymes Are Higher in Mϕ in Peritoneal, Lung, Liver, Spleen, and Intestine Than in Adipose Tissue Mϕ Since Mϕ bioenergetics metabolism, as an immunometabolism pathway (60), regulates their polarizations (61), we hypothesized that tissue Mϕs from different tissues would have various metabolic pathway enzyme genes expressed at different levels.
To test this hypothesis, we collected 59 metabolic enzymes involved in six metabolic pathways including the tricarboxylic acid (TCA) cycle (13 enzymes), pentose phosphate pathway (Warburg-Limpam-Dickens cycle and phosphogluconate shunt, 7 enzymes), glutamine pathway (7 enzymes) (62), fatty acid pathway (16 enzymes), fatty acid B-oxidation pathway (63) (13 enzymes), and fatty acid C pathway (9 enzymes), as shown in Table 1. Of note, six genes overlapped in different bioenergetics metabolic pathways. Comparing the Mϕ from peritoneal and LLSI tissues with that of adipose tissues (Figure 5A), we found that 2 out of 13 TCA cycle enzymes, 2 out of seven pentose phosphate pathway enzymes, one out of seven glutamine pathway enzymes, 7 out of 16 enzymes in the fatty acid pathway, 5 out of 13 regulators in the fatty acid β-oxidation pathway, and 3 out of 9 fatty acid C pathway enzymes were upregulated. We also found that 1 out of 13 TCA cycle enzymes, 1 out of seven pentose phosphate pathway enzymes, four out of seven glutamine pathway enzymes, and 2 out of 16 enzymes in the fatty acid pathway were downregulated. In addition, comparing the Mϕ from peritoneal, intestine, and adipose tissue with that of bone marrow, we found that 1 out of 13 TCA cycle enzymes, 1 out of seven pentose phosphate pathway (PPP) enzymes, 2 out of seven glutamine pathway enzymes, 2 out of 16 enzymes in the fatty acid pathway, and 2 out of 13 regulators in the fatty acid β-oxidation pathway were upregulated and that 5 out of 13 TCA cycle enzymes, 1 out of seven pentose phosphate pathway enzymes, three out of seven glutamine pathway enzymes, 4 out of 16 enzymes in the fatty acid pathway, 5 out of 13 enzymes in the fatty acid β-oxidation pathway and 3 out of 9 fatty acid C pathway enzymes were downregulated. These results suggest that Mϕ in peritoneal, lung, liver, spleen, and intestine may upregulate bioenergetics pathway enzyme expression more than in Mϕ in adipose tissue Mϕ, where expression of the enzymes in the PPP pathway and the three fatty acid pathways increased and expression of glutamine pathway enzymes decreased. Surprisingly, BMderived Mϕ expresses higher bioenergetics pathway enzymes than that of Mϕ in peritoneum, intestine, and adipose tissues. As shown in Figure 5B, our Ingenuity Pathway Analysis (IPA) showed that: (1) comparing all the differences among tissue Mϕ from peritoneal, lung liver, spleen, intestine, adipose tissue, and bone marrow, type II diabetes signaling is shared; (2) the fatty acid activation pathway is also shared among three groups: One of the major differences between the adaptive immune system and innate immune systems is that the cells in the   Tissue macrophages (Mϕ) have differences in the expression of several mediators for exosome biogenesis and docking. First, Mϕ from peritoneum, lung, liver (STX3), spleen, and small intestine prefer to use RAB27A and STX3 than RAB31 and YKT6 in mediating exosome biogenesis and CD44 for docking in comparison to lean ATMϕ. In addition, Mϕ from lung also upregulates STX6 (biogenesis), CAV1, and LGALS3 (docking) comparing to lean ATMϕ. Moreover, Mϕ from peritoneum and intestine upregulate ITGA6 for docking in comparison to lean ATMϕ. (B) Mϕ from peritoneum, lung, liver (STX3), spleen, and small intestine prefer to use RAB27A and STX3 rather than RAB31 and YKT6 in mediating exosome biogenesis and CD44 for docking, presumably to make exosomes more effective in propagating inflammation than adipose tissue macrophages. As cell-cell communication vehicles, exosomes propagate inflammation from first inflammatory cells to secondary inflammatory cells [see also Figure 5 of our previous report for more evidence and the experimental data of others (PMID: 27842563)].
adaptive immune system such as T cells have an antigenspecific memory function (64). However, recently it became clear that innate immune cells also have trained immunity (innate immune memory) functions in the form of increases in three key metabolic pathways: glycolysis, acetyl-CoA synthesis, and the mevalonate pathway (65). Thus, in addition to the FIGURE 5 | (A) Mechanism V: The macrophages from peritoneum, lung, liver, and spleen upregulate more bioenergetics pathway enzymes (immunometabolism pathway) than adipose tissues, with a significantly increased pentose phosphate pathway, three significantly increased fatty acid-related pathways, and a decreased glutamine pathway, and the macrophages from bone marrow express higher bioenergetics pathway enzymes than peritoneum, intestine, and adipose tissues (For detailed expression data, see Figure S7). (B) Ingenuity Pathway Analysis of bioenergetics/immunometabolism pathway markers.
bioenergetics metabolic pathway analysis, we further examined whether tissue Mϕ have differences in the expression of trained immunity pathway enzymes. We hypothesize that the expression of trained immunity pathway enzyme genes in Mϕ from peripheral tissues such as lung, liver, spleen, and intestine is higher in than that of ATMϕ. As shown in Table 1, we found that 24 enzymes are involved in three pathways of trained immunity functions including 14 enzymes in glycolysis, three enzymes in acetyl-CoA generation, and 7 enzymes in the mevalonate pathway (28,66). As shown in Figure 6A, Mϕ in peritoneal, lung, liver, spleen, and intestine upregulate higher levels of 11 out of 24 enzymes in trained immunity pathways in comparison to ATMϕ. In addition, Mϕ in peritoneal and intestine and ATMϕ upregulate 2 out of 24 enzymes and downregulate 14 out of 24 enzymes in comparison to BM Mϕ. The Ingenuity Pathway Analysis showed, as seen in Figure 6B, that the top two pathways involved in upregulation of trained immunity enzymes in Mϕ are LPS/IL-1-mediated inhibition of RXR function and stearate biosynthesis I. The top pathway shared by peripheral Mϕ and ATMϕ upregulation enzymes is acetyl-CoA biosynthesis III. The two pathways among the top three pathways shared by peripheral Mϕ and BM Mϕ are the superpathway of cholesterol biosynthesis and mevalonate pathway I. The macrophages from peritoneum, lung, liver, and spleen upregulate more trained immunity (innate immune memory)-related metabolic genes than that of adipose tissues, and the macrophages from peritoneum, small intestine, and adipose tissues upregulate more trained immunity-related metabolic genes than bone marrow macrophages (For detailed expression data, see Figure S8). (B) Ingenuity pathway-Venn Diagram analyses show that the macrophages (Mϕ) from peritoneum, lung, liver, spleen, and intestine upregulate more shared trained immunity regulators such as mevalonate pathway regulators than Mϕ from bone marrow. (C) The expression changes of 19 new enzymes involved in intracellular immunometabolism pathways (IMPs) and trained immunity pathways (TIPs) of M1 Mϕ and 6 new enzymes involved in IMPs and TIPs of M2 Mϕ may be the mechanisms underlying the higher M1 Mϕ proinflammatory status of lung, liver, spleen, and intestine and the disease group-specific pathways and shared disease pathways (PMIDs:28381829; 27396447;26694790; 30679807; 28396078). The green boxes are bioenergetics pathways ( Figure 5A), and the blue boxes are trained immunity pathways (A). SLC1A5, GLUT1, LDH, PDH, ACLY, and CPT1A are mentioned in previous studies. However, IDH is found to decrease in M1, but not sure in M2. *25 others (bolded and underlined) are not found in the reviews listed above. Figure 6C summarizes the findings from Figure 5A (in the green boxes) and Figure 6A (in the blue boxes) into a new map related to the Mϕ metabolic pathways identified in M1 and M2. We found that the 19 new enzyme expression changes (with * , bolded, and underlined) involved in immunometabolism pathways and trained immunity pathways may be the mechanisms underlying the higher M1 proinflammatory status of lung, liver, spleen, and intestine. Also, we found that six new enzyme changes in M2-related pathways may also be the mechanisms underlying the higher M2 anti-inflammatory status of adipose tissue Mϕ as well as the disease group-specific pathways and shared disease pathways.

DISCUSSION
Macrophages play a key role in the pathogenesis of various diseases including cardiovascular diseases (13), metabolic diseases (14), infectious diseases (15), respiratory diseases (16), digestive diseases (17), autoimmune diseases (18), and many types of cancers (19,20). However, it remained unclear whether Mϕ use the same pathways and play the same roles or use diseasespecific pathways and play disease-specific roles in addition to the shared roles and pathways. To address this question and also to identify the potential mechanisms underlying this issue, we performed a novel type of big-omics database mining analysis, which we pioneered in 2004 (30). We have made the following significant findings. (1) The expression of 31 Mϕ markers and 45 Mϕ TFs are modulated in eight groups comprising a total of 34 diseases including 10 types of cancers, and both shared and disease-specific pathways for each group of disease/tumor have been identified. To identify the potential mechanisms underlying the Mϕ heterogeneity related to disease-group-specific pathways, we examined several novel aspects of Mϕ. (2) The expression of M1 Mϕ markers is higher in Mϕ in lung, liver, spleen, and intestine compared to in lean adipose tissue Mϕ in physiological conditions. (3) Pro-adipogenic transcription factors C/EBPα and PPARγ and proinflammatory adipokine leptin upregulate the expression of M1 Mϕ markers. Our results correlated well with a recent report implicating a pleiotropic protein prohibitin in regulating adipose-immunometabolism (67). (4) Immunologically peripheral tissue Mϕ from lung, liver, spleen, intestine, and bone marrow express higher levels of T cell coinhibition/immune checkpoint receptor CD274 (programmed death-ligand 1, PDL-1) among ten co-inhibition receptors than that of lean adipose tissues, presumably to counteract the M1 dominant status via its reverse signaling and high homeostatic and anti-inflammatory functions (52). Our results reveal a new mechanism underlying the toxicities of the anti-PD-1 and anti-PD-L1 immune checkpoint antibodies (68). (5) Tissue Mϕ from lung, liver, spleen, intestine, and bone marrow prefer to use RAB27A and STX3 than RAB31 and YKT6 in mediating exosome biogenesis and docking, suggesting new inflammatory exosome markers and a new inflammatory exosome status for propagating inflammation from inflamed cells to secondary inflammatory cells as we reported previously (25). Our results correlated well with recent findings that inflammation leads to distinct populations of extracellular vesicles (69). (6) To address why Mϕ in peripheral tissues have a higher M1 status than those in adipose tissues, we found that Mϕ in peritoneal, lung, liver, spleen, and intestine upregulate higher levels of immunometabolism pathway enzymes than adipose tissue Mϕ (ATMϕ). (7) To address the potential mechanism underlying the higher M1 proinflammatory status of Mϕ in peripheral tissues, we found that Mϕ from peritoneum, lung, liver, and spleen upregulate more trained immunity (innate immune memory)-related metabolic genes than that of adipose tissues and that the macrophages from peritoneum, small intestine, and adipose tissues upregulate more trained immunityrelated metabolic genes than bone marrow macrophages. Taken together, our results suggest that multiple mechanisms such as those at the cell surface including M1 Mϕ markers, cell-cell contact receptors, cell-cell communication exosomes, intracellular immunometabolism and trained immunity, and Mϕ nuclear transcription factors may be responsible for the disease group-specific pathways and shared pathways that we found in eight groups of 34 diseases. Since CD274 reverse signaling works via its interaction with CD279 (PD1) expressed in CD4+ T cells, CD8+ T cells, B cells, macrophages, and dendritic cells (70), these analyses emphasize the following. (1) CD279+ T cells (both CD4+ and CD8+) and B cells have significant innate immune functions in controlling CD274+ Mϕ proinflammatory status and maintaining tissue homeostasis in addition to having antigen-specific adaptive immune functions.
(2) T cell co-stimulation and co-inhibition receptors serve as prototypic cell surface receptor-mediated cell-cell contact signaling in addition to classical signaling pathways from cytokine receptors, growth factor receptors, pathogen-associated molecular pattern receptors (PAMP-Rs), danger-associated molecular pattern receptors (DAMP-Rs), and conditional DAMP-Rs, as we reported previously (12). (3) Our results suggest a potential molecular mechanism underlying the clinical finding that elevated immune-related adverse effects (irAEs) of systematically injected anti-PD-L1 monoclonal antibody (mAb) (Durvalumab) in patients with cancers (71). Blocking CD274 reverse signaling could activate all the CD274+ tissue macrophages and contribute to elevated immune-related adverse effects (irAEs). (4) Since CD279 is also expressed in tumor-associated macrophages (TAMs) (72), our results also suggest a possibility that CD279-CD274 interaction on TAMs may suppress the anti-tumor functions of TAMs via reverse signaling.
To summarize our findings, we have proposed a new working model with three connected parts. Figure 7A illustrates the first part: based on the expression levels of two groups of 31 Mϕ subset markers (9) and 45 Mϕ transcription factors in the eight groups of 34 diseases (also see Table 2), we have identified for the first time 20 novel Mϕ disease group-specific pathways and 12 disease-shared pathways (shared in more than four major disease groups). These results have demonstrated two aspects for the first time. First, the pathogenesis of various diseases and tumors significantly modulates Mϕ signaling pathways in disease group-specific and shared manners. Figure 7B illustrates the second: the potential tissue mechanisms underlying the above-mentioned Mϕ heterogeneity in diseased conditions. Based on the differential expression of regulators including M1 markers, M1 TFs, co-stimulation and co-inhibition/immune checkpoint receptors, cell-cell communication exosome biogenesis machinery, M1 bioenergetic enzymes, and trained immunity enzymes, we proposed a new concept of tissue M1 Mϕ status. We found that first, Mϕs in liver, small intestine, and bone marrow-derived Mϕ have the highest macrophage inflammation potential and, second, adipose tissue from lean animals and surprisingly spleen have low Mϕ inflammation potential. Of note, white adipose tissue hypertrophy recruits significant numbers of inflammatory cells including Mϕ (73). The new data have demonstrated that various tissue Mϕ have significant differences in M1 proinflammatory status, which could be controlled by high expression of a co-inhibition receptor such as CD274 (PDL1)-initiated anti-inflammatory reverse signaling. It is noteworthy that each tissue has its own composition of embryonically derived and adult-derived Mϕ, but it is unclear whether the Mϕ of distinct origins are functionally interchangeable or have unique roles at steady state (74). These issues can be examined in the future when new microarray/RNA-sequencing (RNA-Seq) data are available. Figure 7C illustrates the third part of the model: three novel major cell/molecular mechanisms underlying the above-mentioned Mϕ heterogeneity in diseased conditions. The six novel cell and molecular mechanisms include three cell surface mechanisms (Mϕ subset markers with potential of signaling, cell-cell interaction receptors (co-stimulation and co-inhibition/immune checkpoint receptors), and inflammationmodulating cell-cell communication exosomes), two new metabolism mechanisms (the immunometabolism/bioenergistic and trained immunity metabolic pathways), and, finally, nuclear transcription factors. Taken together, the new tissue, cell, and molecular mechanisms may contribute to the novel Mϕ signaling heterogeneity in diseased conditions that we have found.
We acknowledge that carefully designed in-vitro and invivo experimental models will be needed to verify all the results we report here. These experimental models will enable the consolidation of the Mϕ disease group-specific pathways in various pathological conditions. However, the big data mining analyses that we pioneered in 2004 (30) have provided significant insights into the Mϕ disease group-specific and shared pathways and heterogeneity, homeostasis, and functions of Mϕ in various diseases and cancers/tumors and have also identified novel therapeutic targets for treating cancers/tumors and inflammation, tissue regeneration, and tissue repair.