TM4SF19-mediated brake of macrophage efferocytosis contributes to obesity-induced inflammation of adipose tissue

doi:10.21203/rs.3.rs-2757116/v1

Download PDF

Article

TM4SF19-mediated brake of macrophage efferocytosis contributes to obesity-induced inflammation of adipose tissue

https://doi.org/10.21203/rs.3.rs-2757116/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Adipose tissue macrophages are critical players in obesity-induced inflammation and metabolic dysfunction. Here, we identify transmembrane 4 L six family member 19 (TM4SF19) as a gene upregulated in adipose tissue macrophages in obese states. We demonstrate that TM4SF19 is a lysosomal membrane protein that inhibits vacuolar H+-ATPase activity by interacting with its subunit ATP6V0B. In vitro gain- and loss-of-function assays in macrophages indicate that TM4SF19 reduces lysosomal acidification and impairs the clearance of dead fat cells. Single-nucleus RNA sequencing analysis indicates that TM4SF19 is specifically expressed in a subpopulation of Trem2 + lipid-associated macrophages, and global knockout reduces diet-induced obesity and recruitment of pro-inflammatory macrophages to adipose tissue. Moreover, global or macrophage-specific TM4SF19 knockout decreases the number of lipid-stressed adipocytes induced by high-fat feeding and restores tissue insulin sensitivity. Overall, these results suggest that TM4SF19-mediated control of lysosomal activity is a potential therapeutic target to resolve obesity-induced adipose tissue inflammation.

Biological sciences/Physiology/Metabolism/Metabolic diseases/Obesity

Health sciences/Diseases/Endocrine system and metabolic diseases/Obesity

Adipose tissue expansion during overnutrition is associated with the recruitment of specific immune cell subtypes, the functions of which are poorly understood¹. Mounting evidence indicates that immune cell recruitment is a double-edged sword. On one hand, immune cells support healthy tissue remodeling and expansion; on the other hand, chronic overnutrition often results in persistent recruitment of proinflammatory immune cells that exhibit stalled efferocytosis and contribute to local and systemic insulin resistance ². The mechanisms controlling the balance between healthy tissue expansion and persistent inflammation are incompletely understood.

Macrophage efferocytosis is the process by which macrophages phagocytose and clear apoptotic cells, which is an important process in maintaining tissue homeostasis and preventing/resolving inflammation ³. Growing evidence links defective efferocytosis to chronic inflammatory diseases, including obesity, autoimmune and neurological disorders, type 2 diabetes mellitus (T2DM), and atherosclerosis ⁴. In obesity, dysfunctional macrophages fail to efficiently clear apoptotic cells ^5,6, leading to the accumulation of dead adipocytes surrounded by proinflammatory macrophages, which contribute to insulin resistance and other metabolic abnormalities ^7,8.

In addition, lipid accumulation in macrophages is a characteristic of pathological remodeling of adipose tissue ⁹ and foam cells that appeared in atherosclerotic plaques ¹⁰. These lipid-associated macrophages have been molecularly defined by triggering receptor expressed on myeloid cells 2 (Trem2) expression in obese adipose tissue and various pathologic conditions ^9,11, although physiological and pathological functions of TREM2 in adipose tissue macrophages are controversial ¹².

Lysosomal catabolic activity of lipid-associated macrophages is required for lipid metabolic homeostasis ¹³ and clearance of lipid remnant from dead adipocytes in hypertrophied adipose tissue ^13–15. In particular, vacuolar H+-ATPase (V-ATPase) is critical for lysosomal acidification and catabolic activity. V-ATPase is a large protein complex that is composed of two multi-subunit domains: peripheral V1 domain involved in ATP hydrolysis and integral (membrane-embedded) V0 domain that transports protons ¹⁶;¹⁷. The regulated V1-V0 assembly is one of the major molecular mechanisms that control V-ATPase activity ¹⁶.

TM4SF19 is a protein of unknown function that is highly expressed in adipose tissue macrophages and is upregulated during diet-induced adipose tissue expansion. Here we show that Tm4sf19 is selectively expressed in a subpopulation of Trem2 + lipid-associated macrophages that are recruited to adipose tissue during high fat feeding. Global or macrophage-selective inactivation of Tm4sf19 greatly improves systemic metabolism and insulin action in adipose tissues. Mechanistically, we demonstrate that the loss of TM4SF19 in macrophages upregulates V-ATPase activities by enhancing V0-V1 assembly and increasing phagocytic activity and lipid clearance. As TREM2 + macrophages have been implicated in pathophysiological tissue remodeling involving aberrant lipid metabolism, our results indicate that Tm4sf19 might be targeted for treating a large class of obesity-associated metabolic diseases.

TM4SF19 is upregulated in adipose tissue of obese mice and humans

To identify genes upregulated in obese adipose tissue, we analyzed three publicly available bulk RNA-seq datasets (Fig. 1A): gonadal white adipose tissue (GWAT) of normal chow diet (NCD) and high-fat diet (HFD) fed mice (GSE182930)¹⁸, GWAT of wild type and leptin-deficient mice (ob/ob) (GSE150102)¹⁹, and abdominal subcutaneous WAT of lean subjects and patients with obesity (GSE59034)²⁰. Gene ontology (GO) analysis showed that the union of the top 100 genes upregulated by obesity from each dataset had significant enrichment of genes related to the immune system and inflammatory response (Extended Data Fig. 1A, Supplemental table 1). We identified 17 genes that were upregulated by obesity in all data sets, including genes involved in inflammatory responses [secreted phosphoprotein 1(Spp1), regulator of G protein signaling 1 (Rgs1), dehydrogenase/reductase 9 (Dhrs9), C-C motif chemokine ligand 2 (Ccl2), macrophage scavenger receptor 1 (Msr1), interleukin 1 receptor antagonist (Il1rn), Cd84] and markers of lipid-associated macrophages (triggering receptor expressed on myeloid cells 2 (Trem2), neutral cholesterol ester hydrolase 1 (Nceh1)) ^9,21 (Fig. 1B). We focused on Tm4sf19 (GSE182930: 3.64-fold, GSE150102: 6.98-fold, GSE59034: 2.38-fold), a gene with unknown function that might be associated with the inflammatory response of adipose tissue ²² (Fig. 1B). TM4SF19 belongs to the transmembrane 4 L six family ²³, and the monocyte/macrophage-specific expression of TM4SF19 in human adipose tissue was previously noted ^24–26. Consistent with bulk transcriptomic analyses, protein and mRNA levels of Tm4sf19 were significantly upregulated in GWAT of HFD fed mice, while basal expression levels of Tm4sf19 were detectable in bone marrow-derived macrophages (BMDM) obtained from NCD fed mice, further supporting its macrophage-specific expression (Fig. 1C and 1D). Furthermore, to determine the cellular distribution of TM4SF19, we isolated F4/80 + cells and floating adipocytes from the GWAT of HFD-fed mice by magnetic activated cell sorting (MACS) and found that Tm4sf19 was highly enriched in F4/80 + cells (Fig. 1E).

Next, we examined the potential upstream regulators of TM4SF19 expression using the transcription factor binding site prediction tool (Ciiider) ²⁷. We found that the promoter regions of both mouse Tm4sf19 and human TM4SF19 contain nuclear factor-kappa B (NF-ĸB) binding sites (Extended Data Fig. 1B), a well-known transcription factor involved in pro-inflammatory responses ²⁸. Binding sites for sterol regulatory element-binding transcription factors 1 and 2 (SREBF1 and SREBF2) were also predicted (Extended Data Fig. 1B), indicating that Tm4sf19 expression is potentially regulated by lipogenic stimuli ²⁹. ChIP analysis demonstrated that NF-kB was recruited to the putative NF-kB binding site of the Tm4sf19 promoter in response to lipopolysaccharide (LPS) treatment (Fig. 1F and 1G). Protein levels of TM4SF19 and NF-kB (p105 and p50) in RAW 264.7 macrophages were upregulated by LPS treatment (Fig. 1H), consistent with a previous report ³⁰. Additionally, TM4SF19 expression levels were positively correlated with body mass index (BMI) as well as NFKB1 expression in human subcutaneous WAT (Fig. 1I).

TM4SF19 is a lysosomal membrane protein that controls V-ATPase activity by interacting with ATP6V0B

Given the macrophage-specific expression of TM4SF19 in human adipose tissue and mouse GWAT, we established TM4SF19 overexpressing RAW264.7 macrophage cell lines to investigate the biological function of TM4SF19 in macrophages. We examined the subcellular localization of GFP-tagged TM4SF19 and found that TM4SF19 was specifically located at the lysosomal membrane (Pearson correlation coefficient = 0.732), indicated by LysoTracker staining (Fig. 2A). Next, we focused on lysosomal proteins that potentially interact with TM4SF19. TM4SF19 was previously suggested to interact with ATP6V0B in yeast two-hybrid arrays ³¹. Among several candidates, we tested TM4SF19 interaction with ATP6V0B, a membrane embedded component of V-ATPase. Immunoprecipitation analysis confirmed that TM4SF19 physically interacts with ATP6V0B (Fig. 2B). Based on this interaction, we hypothesized that TM4SF19-ATP6V0B interaction alters V0-V1 domain assembly of V-ATPase, which is critical for lysosomal acidification and catabolic activity ^32,33. To test this hypothesis, levels of assembled V0-V1 domains (relative abundance of ATP6V1B2 in membrane vs. cytosolic fractions) were quantified in BMDM of WT and TM4SF19 KO mice. Membrane-bound vs. cytosolic ATP6V1B2 levels were increased in BMDM by TM4SF19 KO, indicating that TM4SF19 KO suppresses V1-V0 assembly (Fig. 2C). Consistently, TM4SF19 overexpression (OE) increased cytosolic levels of ATP6V1B2, confirming that TM4SF19 inhibits V1-V0 assembly (Fig. 2D).

Since enhanced V1-V0 association is required for V-ATPase activity ³⁴, we directly measured V-ATPase activity in lysosomal fractions (Extended Data Fig. 2). As expected, TM4SF19 OE decreased, whereas TM4SF19 KO increased, lysosomal V-ATPase activity in RAW264.7 (Fig. 2E). LysoSensor Yellow/Blue dextran staining also demonstrated that lysosomal pH was more acidic in TM4SF19 KO BMDM than in WT control (Fig. 2F). Next, we investigated the effects of TM4SF19 KO on the clearance of dying/dead adipocytes by macrophages. Inactivation of TM4SF19 increased the capacity of BMDM to clear dying/dead C3H10T1/2 adipocytes (Fig. 2G). Furthermore, high magnification images of co-cultures at the initial phase (90 mins after co-culture) showed that TM4SF19 KO accelerated phagocytosis by BMDM, as indicated by increased number of LipidTOX + BMDM (Fig. 2H).

Single-nucleus RNA sequencing analysis of GWAT dissects cell type-specific effects of Tm4sf19 deficiency

Next, we investigated the physiological roles of TM4SF19 using TM4SF19 KO mice. TM4SF19 KO was confirmed in WAT and BMDM by qPCR analysis (Extended Data Fig. 3A). TM4SF19 protein was not detectable by immunoblot analysis of adipose tissue of chow diet fed control mice. As anticipated, in mice fed normal chow diet, global TM4SF19 did not affect body weight, fat tissue mass, adipocyte size, or body composition (Extended Data Fig. 3B-3D). Similarly, TM4SF19 KO did not alter energy expenditure, food intake, physical activity, or thermogenesis induced by β3-adrenergic receptor agonist CL316,243 treatment (Extended Data Fig. 3E) in chow-fed mice.

We next focused on the effects of Tm4sf19 deficiency in mice fed high fat diet, in which TM4SF19 is greatly upregulated. Adipose tissue is a complex and dynamic mixture of immune, stromal, and parenchymal cell types {Rondini, 2020 #122}. To examine cell-type specific effects of Tm4sf19 deficiency in an unbiased manner, we performed single nuclei RNA-seq (snRNA-seq) on whole GWAT of WT and TM4SF19 global KO mice fed with NCD or HFD for 12 weeks. After filtering out low-quality cells, we profiled the transcriptomes of 54,960 cells from two multiplexed replicates for each condition. Cells were clustered by applying a graph-based clustering algorithm, annotated into 13 cell types based on the expression of curated cell-type marker genes for GWAT, and visualized in the uniform manifold approximation and projection (UMAP) representation (Fig. 3A, Extended Data Fig. 4A and 4B). Cell-type composition analysis revealed that high fat feeding expanded the subset of macrophages expressing Trem2 (Mac.Trem2) and inactivation of Tm4sf19 partially prevented its expansion (Fig. 3B, 3C, and Extended Data Fig. 4A). Tm4sf19 was selectively expressed in Mac.Trem2 subtype, (Fig. 3D and Extended Data Fig. 4C), suggesting that the observed phenotypes of TM4SF19 KO mice fed with HFD are associated with this macrophage subtype.

To compare the differentiation trajectories from monocytes to macrophages for each condition, we performed pseudotime analysis for monocytes and macrophage subsets. The constructed trajectory showed three distinct pathways from monocytes to each macrophage subset (Fig. 3E), indicating that the three subtypes have different characteristics. Lyve1 + macrophage was denoted as a perivascular-like macrophage (PVM), which might have a role in angiogenesis and M2-like features, and Trem2 + macrophages are known as a lipid-associated macrophage (LAM) which expand with HFD feeding ^35–37. Comparing the distribution of cells along the differentiation trajectories, monocytes primarily differentiated into Lyve1 + macrophages (Mac.Lyve1) rather than Trem2 + macrophages (Mac.Trem2) with NCD (Fig. 3F, Extended Data Fig. 4F and 4G). However, with HFD the majority of monocytes underwent differentiation into Mac.Trem2 and TM4SF19 KO attenuated this effect (Fig. 3F, 3G, and Extended Data Fig. 4G), which resulted in decreased Mac.Trem2 population (Fig. 3C and Extended Data Fig. 4D). To examine whether TM4SF19 KO also affects the characteristic of Mac.Trem2, we calculated the signature scores of Mac.Trem2 and Mac.Lyve1 subsets of wild-type mice. TM4SF19 KO decreased the Mac.Trem2 signature score and increased Mac.Lyve1 signature score of Mac.Trem2 in HFD-fed mice, while there was no significant change in NCD-fed mice (Fig. 3H). To confirm this result, we also calculated the signature scores for the Mac1 and Mac3 populations, considered to be Lyve1 + and Trem2 + macrophages in a previous study ⁹, respectively, and similar results were obtained (Extended Data Fig. 4E). Consequently, TM4SF19 KO decreases the diet-induced expansion of Mac.Trem2 and alters their expression profile.

TM4SF19 KO prevents HFD-induced inflammation of adipose tissue.

To further understand the function of Tm4sf19 in macrophages, we conducted co-expression network analysis on monocyte/macrophage populations from WT HFD mice and identified eight co-expressed gene modules (Mod1 ~ 8) (Extended Data Fig. 5A-C). Tm4sf19 was positively correlated with Mod4 and negatively correlated with Mod2 (Fig. 4A, 4B, and Extended Data Fig. 5B). Genes in Mod4 were enriched in Mac.Trem2, whereas genes in Mod2 were mainly expressed in Mac.Lyve1 (Fig. 4A and 4B). Genes positively correlated with Tm4sf19 expression (Mod4) were associated with lipid handling markers (Plin2, Pparg, Lipa), vacuolar acidification (Atp6v1h, Atp6v0d2, Atp6v1b2, Atp6v0a1, Atp6v1a), and immune system process (Cd300a, Cd300lb, Malt1, Alcam, Trim25), and were significantly reduced with TM4SF19 KO. Gene Ontology (GO) enrichment analysis of genes in Mod4 indicated that molecular functions of vacuolar-type ATPase (V-ATPase) and phagocytic cup were highly enriched, while cellular components of receptor-mediated endocytosis and complement and coagulation cascades, the mediators of immunity, were associated in genes negatively correlated with TM4SF19 expression (Mod2) (Extended Data Fig. 5D). Genes in Mod2 included anti-inflammatory markers (Mrc1, Clec10a, Rbpj) and phagocytosis markers (C1qa, C1qb, C1qc) (Fig. 4C). Interestingly, genes that correlated with markers of insulin sensitivity (Ly86, Pde3b, Igf1) and cholesterol efflux (Abca1, Apoe) in adipose tissue were upregulated by TM4SF19 KO (Fig. 4C). Gene set enrichment analysis (GSEA) showed that TM4SF19 KO downregulated genes involved in the inflammatory response in GWAT of HFD mice (Fig. 4D, 4E, 4F, and Extended Data Fig. 5E). Additionally, HFD-induced F4/80 + macrophage recruitment in GWAT and liver were significantly decreased in TM4SF19 KO mice (Fig. 4G and Extended Data Fig. 5F). Consistent with qPCR analysis of the macrophage marker, Adgre1 (Fig. 4F), flow cytometry analysis showed TM4SF19 KO decreased the proportion of total macrophages (CD45 + CD11b+) in HFD-fed mice (Fig. 4H and Extended Data Fig. 5E). However, the proportion of leukocyte (CD45+) did not differ between WT and TM4SF19 global KO after feeding HFD (Extended Data Fig. 5F). As expected from the anti-inflammatory gene signature of adipose tissue from TM4SF19 KO mice, the M2/M1 ratio (M2-like macrophages (CD11b+/CD11C-/CD206+) over M1-like macrophages (CD11b+/CD11C+/CD206-)) was significantly higher in GWAT of TM4SF19 global KO (Fig. 4H).

TM4SF19 KO restores insulin-sensitive adipocytes in adipose tissue of mice fed HFD for 12 weeks.

Next, from our snRNA-seq data, we further characterized adipocytes and adipocyte progenitor cells (APCs) which are major cell types that respond to nutritional stimuli in adipose tissue. Sub-clustering identified two populations of APCs (Pdgfrα, Dcn, Col1a1) and three populations of mature adipocytes (Adipoq, Cidec, Pparg, Plin4, Lep) (Fig. 5A, 5B). Three adipocyte populations were characterized by graduated expression of adipocyte marker genes. Adipoq was expressed in Adipocyte.1, and Plin4 and Lep were highly expressed in Adipocyte.3 compared to Adipocyte.1. Adipocyte.2 showed intermediate expression of these genes (Fig. 5B and Extended Data Fig. 4H). APCs could be divided into APC.Dpp4 (Dpp4, Pi16, Cd55) and APC.Icam1 (Icam1, Col4a1, F3), characterized with the known APC subtype marker genes (Fig. 5B). The pseudotime analysis of adipocytes and APCs showed the differentiation trajectories from APCs to adipocytes, divided into two pathways, which were enriched with adipocytes of WT NCD-fed and WT HFD-fed conditions, respectively (Fig. 5A, 5C, and Extended Data Fig. 4I). In WT NCD-fed condition, Adipocyte.1 occupied the largest population, with 73% of total adipocytes and Adipocyte.3 were rarely detected (less than 3%). In contrast, in WT HFD-fed mice, Adipocyte.3 composed about 59% of total adipocytes and became the major population among adipocytes (Fig. 5D). Interestingly, TM4SF19 KO partially recovered the Adipocyte.1 population in HFD-fed mice (Odd ratio = 3.71, P-value = 0.0002) while there was no difference in NCD-fed mice (Fig. 5C and 5D). This observation suggested that TM4SF19 KO potentially increases de novo adipogenesis from progenitors.

To characterize the difference between Adipocyte.1 and Adipocyte.3, we performed GO enrichment analysis on differentially expressed genes (DEGs) between Adipocyte.1 and Adipocyte.3. This analysis revealed that genes involved in insulin response and lipolysis were enriched in Adipocyte.1, while genes associated with lipid biosynthesis and apoptotic process were enriched in Adipocyte.3 (Fig. 5E). Genes up-regulated in Adipocyte.1 contained adipokines such as Adipoq and Cfd, which have beneficial effects on insulin sensitivity ^38,39 and included insulin signaling genes (Fig. 5F). Adipocyte.1 showed increased lipolysis and decreased lipid accumulation (Fig. 5F). In contrast, Adipocyte.3 showed the opposite gene expression patterns compared to Adipocyte.1, including decreased insulin signaling and lipolysis, but increased lipid accumulation (Fig. 5F). These results indicate that the Adipocyte.1 population, which was the major population in NCD-fed mice adipocytes, showed transcriptional signatures associated with insulin sensitivity, and the expansion of Adipocyte.1 in TM4SF19 KO restored the insulin sensitivity of adipocytes during high fat feeding

Additionally, we confirmed that Adipocyte.1-specific markers (IRS1, INSR, PDE3B) in human adipose tissue tended to be negatively correlated with BMI, whereas Adipocyte.3-specific markers (LEP, FASN) were positively correlated with BMI (Fig. 5G), as previously reported ⁴⁰. In addition, we identified that PDE4D, which affects the development of insulin resistance in obesity ^41,42, was positively correlated with BMI (Fig. 5G).

TM4SF19 KO prevents HFD-induced obesity and insulin resistance of adipose tissue.

We next addressed the impact of TM4SF19 KO on metabolic function during high fat feeding. Global TM4SF19 KO reduced the weight gain of mice over the 12-week course of HFD feeding (Extended Data Fig. 6A-C). Furthermore, glucose and insulin tolerance were improved in HFD-fed TM4SF19 global KO mice (Extended Data Fig. 6D-F). Global TM4SF19 KO also corrected HFD-induced hyperinsulinemia and hyperlipidemia (TG and cholesterol) (Extended Data Fig. 6G and 6H). As anticipated by snRNA-seq data (Fig. 5), TM4SF19 KO greatly improved in vivo insulin action as assessed by insulin-dependent phosphorylation of AKT and IRS1/2 (Extended Data Fig. 6I).

Hematoxylin and eosin (H&E) staining indicated that TM4SF19 KO reduces the size of inguinal and gonadal white adipocytes of mice fed HFD (Extended Data Fig. 7A). Furthermore, we found that TM4SF19 KO increased energy expenditure during HFD without affecting food intake or physical activity (Extended Data Fig. 7B and 7C). Consistent with elevated energy expenditure, GSEA analysis indicated that TM4SF19 KO prevents diet-induced reduction of mitochondrial biogenesis and oxidative phosphorylation gene expression (Extended Data Fig. 7D and 7E), which was confirmed by western blot analysis (Extended Data Fig. 7F).

Macrophage-specific TM4SF19 KO protects obesity-related metabolic dysfunction.

Finally, we established a tamoxifen-inducible macrophage-specific TM4SF19 KO mouse model (TM4SF19 MKO) by crossing Tm4sf19^flox/flox mice with Csf1r-CreER mice. After tamoxifen induction, macrophage-specific Tm4sf19 deletion was confirmed in GWAT of TM4SF19 MKO mice (Fig. 6A). The HFD-induced upregulation of TM4SF19 expression was not detectable in GWAT of TM4SF19 MKO mice (Fig. 6B). Consistent with global TM4SF19 KO mice, the HFD-induced increase in body weight, adipose tissue mass, and body fat percentage were suppressed by TM4SF19 MKO (Fig. 6C-E). TM4SF19 MKO also improved glucose tolerance and insulin sensitivity (Fig. 6F and 6G) and enhanced oxygen consumption and energy expenditure in HFD-fed mice without affecting locomotor activity or food intake (Fig. 6H and Extended Data Fig. 8). Furthermore, TM4SF19 KO lowered the expression of genes associated with pro-inflammatory responses [C-C motif chemokine ligand 2 (Ccl2)] and F4/80 + macrophage markers (Adgre1) during HFD feeding (Fig. 6I-J). Consistent with TM4SF19 global KO, macrophage-specific TM4SF19 KO reduced LAM in GWAT and decreased the proportion of LAM (BODIPY + F4/80 + CD45+) in the stromal vascular fraction (SVF) isolated from GWAT (Fig. 6K). Furthermore, TM4SF19 MKO prevented the recruitment of total macrophages (CD45 + F4/80+) by HFD feeding (Fig. 6K). The lipid accumulation in macrophages (CD45 + F4/80 + BODIPY+) and total CD45 + cells (CD45 + BODIPY+) were also reduced by TM4SF19 MKO (Fig. 6L).

As an alternative approach, we performed bone marrow transplantation (BMT) of TM4SF19 KO immune cells to WT mice to experimentally replace TM4SF19 selectively in macrophages and fed an HFD for 12 weeks (Extended Data Fig. 9A). We confirmed TM4SF19 was successfully removed (> 90%) in GWAT of BMT KO HFD mice after irradiation (Extended Data Fig. 9B). Consistent with HFD-fed TM4SF19 KO mice (Extended Data Figs. 6 and 7), BMT KO mice gained less weight from HFD feeding than BMT WT mice (Extended Data Fig. 9C). BMT KO significantly diminished the weights of WAT of HFD-fed mice compared to BMT WT mice (Extended Data Fig. 9D). Fasting glucose and insulin levels, as well as glucose and insulin tolerance, were improved (Extended Data Fig. 9E-G) in HFD-fed BMT KO mice (Extended Data Fig. 9H). Consistent with snRNA-seq analysis (Fig. 3C), the proportion of TREM2 + macrophages (CD45 + TREM2 + F4/80+) (Extended Data Fig. 9I) and LAMs (CD45 + F4/80 + BODIPY+) were reduced in GWAT of HFD-fed BMT KO mice (Extended Data Fig. 9J).

This study investigated the pathophysiological roles of TM4SF19, which is upregulated in macrophages in obesity. We discovered that TM4SF19 is a macrophage-enriched lysosomal membrane protein that represses lysosomal V-ATPase activity by interacting with ATP6V0B. We used snRNAseq analysis to examine cell-type specific responses to global genetic deficiency of TM4SF19. TM4SF19 KO reduced HFD-induced TREM2 + macrophage recruitment and pro-inflammatory activation in adipose tissue. Potentially due to the diminished inflammatory response of adipose tissue, TM4SF19 KO increased systemic energy expenditure, and improved glucose tolerance and insulin sensitivity of HFD-fed mice.

TREM2 + LAMs are known to surround stressed and apoptotic adipocytes during diet-induced obesity ⁹. One of the important findings in our study is that TM4SF19 ablation remodels the molecular phenotype of TREM2 + macrophages by increasing the expression of anti-inflammatory genes and reducing levels of lipid metabolism-related genes (Plin2, Cd36, Pparg). We speculate that TM4SF19 KO-induced lysosomal catabolic activity of TREM2 + macrophages increase the efficiency of efferocytosis/removal of apoptotic/dead adipocytes and consequently reduces lipotoxicity that recruits pro-inflammatory macrophage phenotypes. LAMs have been investigated in various disease conditions including obesity ⁹, nonalcoholic steatohepatitis (NASH) ⁴³, other hepatic disorders ^44,45, and Alzheimer’s disease ⁴⁶. TM4SF19 belongs to the transmembrane 4 L six family ²³, and its upregulation in human umbilical vein endothelial cells by lipopolysaccharide (LPS) stimulation and in atherosclerotic plaques of human patients with coronary heart disease have been reported previously ³⁰. Thus, we suggest that TM4SF19 is likely to play important roles in disease progression by modifying the activities of specialized TREM2 + immune cells, including vascular foam cells, lipid-loaded hepatic macrophages, and microglia.

Although the current study did not investigate the transcriptional regulators of Tm4sf19 gene expression, we identified NF-kb and SREBP1 and 2 as potential upstream regulators. In line with previous studies showing the roles of SREBP1a in inflammasome activation in conjunction with lipogenesis ⁴⁷, we speculate that Tm4sf19 is one of the downstream effector genes of SREBP1 that contribute to lipid accumulation and pro-inflammatory responses of macrophages. Further study is needed to understand the involvement of lipid signaling in the regulation of Tm4sf19 expression and activity.

In our snRNA-seq analysis, we demonstrated that Tm4sf19 rescued insulin-sensitive adipocytes, suggesting potential therapeutic targeting of Tm4sf19 in obesity-related metabolic diseases. Recently, single-cell transcriptomic analysis of WAT from human and mouse models enables us to characterize the molecular phenotypes of adipocytes. For instance, recent studies characterized human and murine white adipocyte clusters, distinguishing the thermogenic and T2DM-associated subpopulation in visceral WAT ⁴⁰. Consistently, in our study, gene expression profiles of lipid-stressed and insulin-sensitive adipocyte subpopulations correlated positively and negatively, respectively, with BMI in human subjects.

We demonstrated that TM4SF19 interacts with ATP6V0B in macrophages, interfering V1/V0 association. In addition to the critical roles of V-ATPase activity in lysosomal acidification, it has been shown that inhibition of V-ATPase activity induces macrophage-proinflammatory activation ⁴⁸. Similarly, a recent investigation demonstrated that V-ATPase regulates the biosynthesis of lipid mediators of M2-like macrophages, which are required for inflammation resolution ⁴⁹. Likewise, pharmacological inhibition of V-ATPase leads to proinflammatory cytokine secretion from M1 macrophages ⁵⁰. Further investigation is required to fully understand the molecular mechanisms of Tm4sf19 KO-mediated beneficial effects on obesity-induced inflammation and metabolic dysfunction.

In conclusion, our study demonstrates that TM4SF19 deletion in macrophages protects mice from obesity-induced metabolic dysfunction and suggests that TM4SF19-mediated control of lysosomal activity is a potential therapeutic target to facilitate the resolution of obesity-induced adipose tissue inflammation and related disorders.

Acknowledgments

This research was supported by the National Research Foundation of Korea (NRF) grants (NRF-2019R1C1C1002014 and NRF-2018R1A5A2024425 to Y.-H.L., NRF-2021M3H9A1030158 and 2019M3A9D5A01102794 to J.K.K) and Korea University Guro Hospital (KOREA RESEARCH-DRIVEN HOSPITAL) and grant funded by Korea University Medicine (No. K2210441).

Author contributions

Y.-H.L and J.K.K conceived and designed the study. C.C, M.K and Y-S.J conducted the animal experiments. Y.L.J, Y.H.C, and J.K.K conducted snRNA-seq analysis. C.C and M.K performed in vitro experiments. S.N. provided human samples. S.N. and C.C. analyzed human samples. C.C., M.K., G.C., and Y.C. conducted bone marrow transplantation experiments. Y.-H.L, C.C, Y.L.J, J.K.K and J.G.G. wrote the manuscript. All authors reviewed the manuscript.

Data and code availability

The accession number for publicly available database presented in Figure 1A and 1B is GSE59034, GSE150102, GSE182930. The raw single cell RNA-seq data presented in Figure 3,4,5 have been deposited in a public database under accession number PRJNA942977. (https://dataview.ncbi.nlm.nih.gov/object/PRJNA942977?reviewer=i97n7108kb7r9se0j3r2hcqo9g).

Declaration of interests

The authors declare that they have no competing interests.

Rosen, E. D. & Spiegelman, B. M. What we talk about when we talk about fat. Cell 156, 20-44, doi:10.1016/j.cell.2013.12.012 (2014).
Nance, S. A., Muir, L. & Lumeng, C. Adipose tissue macrophages: Regulators of adipose tissue immunometabolism during obesity. Mol Metab 66, 101642, doi:10.1016/j.molmet.2022.101642 (2022).
Doran, A. C., Yurdagul, A. & Tabas, I. Efferocytosis in health and disease. Nature Reviews Immunology 20, 254-267, doi:10.1038/s41577-019-0240-6 (2020).
Mehrotra, P. & Ravichandran, K. S. Drugging the efferocytosis process: concepts and opportunities. Nature Reviews Drug Discovery 21, 601-620, doi:10.1038/s41573-022-00470-y (2022).
Wang, X. et al. Prolonged hypernutrition impairs TREM2-dependent efferocytosis to license chronic liver inflammation and NASH development. Immunity 56, 58-77.e11, doi:10.1016/j.immuni.2022.11.013 (2023).
Li, S. et al. Defective Phagocytosis of Apoptotic Cells by Macrophages in Atherosclerotic Lesions of ob/ob Mice and Reversal by a Fish Oil Diet. Circulation Research 105, 1072-1082, doi:doi:10.1161/CIRCRESAHA.109.199570 (2009).
Cinti, S. et al. Adipocyte death defines macrophage localization and function in adipose tissue of obese mice and humans. Journal of Lipid Research 46, 2347-2355, doi:https://doi.org/10.1194/jlr.M500294-JLR200 (2005).
Lindhorst, A. et al. Adipocyte death triggers a pro-inflammatory response and induces metabolic activation of resident macrophages. Cell Death & Disease 12, 579, doi:10.1038/s41419-021-03872-9 (2021).
Jaitin, D. A. et al. Lipid-Associated Macrophages Control Metabolic Homeostasis in a Trem2-Dependent Manner. Cell 178, 686-698.e614, doi:10.1016/j.cell.2019.05.054 (2019).
Cochain, C. et al. Single-Cell RNA-Seq Reveals the Transcriptional Landscape and Heterogeneity of Aortic Macrophages in Murine Atherosclerosis. Circ Res 122, 1661-1674, doi:10.1161/circresaha.117.312509 (2018).
Katzenelenbogen, Y. et al. Coupled scRNA-Seq and Intracellular Protein Activity Reveal an Immunosuppressive Role of TREM2 in Cancer. Cell 182, 872-885.e819, doi:10.1016/j.cell.2020.06.032 (2020).
Winn, N. C., Wolf, E. M., Garcia, J. N. & Hasty, A. H. Exon 2-mediated deletion of Trem2 does not worsen metabolic function in diet-induced obese mice. J Physiol 600, 4485-4501, doi:10.1113/JP283684 (2022).
Xu, X. et al. Obesity activates a program of lysosomal-dependent lipid metabolism in adipose tissue macrophages independently of classic activation. Cell Metab 18, 816-830, doi:10.1016/j.cmet.2013.11.001 (2013).
Haka, A. S. et al. Exocytosis of macrophage lysosomes leads to digestion of apoptotic adipocytes and foam cell formation. J Lipid Res 57, 980-992, doi:10.1194/jlr.M064089 (2016).
Coats, B. R. et al. Metabolically Activated Adipose Tissue Macrophages Perform Detrimental and Beneficial Functions during Diet-Induced Obesity. Cell Rep 20, 3149-3161, doi:10.1016/j.celrep.2017.08.096 (2017).
Cotter, K., Stransky, L., McGuire, C. & Forgac, M. Recent Insights into the Structure, Regulation, and Function of the V-ATPases. Trends Biochem Sci 40, 611-622, doi:10.1016/j.tibs.2015.08.005 (2015).
Luzio, J. P., Pryor, P. R. & Bright, N. A. Lysosomes: fusion and function. Nat Rev Mol Cell Biol 8, 622-632, doi:10.1038/nrm2217 (2007).
Cho, Y. K. et al. Adipocyte lysoplasmalogenase TMEM86A regulates plasmalogen homeostasis and protein kinase A-dependent energy metabolism. Nat Commun 13, 4084, doi:10.1038/s41467-022-31805-3 (2022).
Jouffe, C. et al. Disruption of the circadian clock component BMAL1 elicits an endocrine adaption impacting on insulin sensitivity and liver disease. Proc Natl Acad Sci U S A 119, e2200083119, doi:10.1073/pnas.2200083119 (2022).
Petrus, P. et al. Transforming Growth Factor-beta3 Regulates Adipocyte Number in Subcutaneous White Adipose Tissue. Cell Rep 25, 551-560 e555, doi:10.1016/j.celrep.2018.09.069 (2018).
Hill, D. A. et al. Distinct macrophage populations direct inflammatory versus physiological changes in adipose tissue. Proc Natl Acad Sci U S A 115, E5096-E5105, doi:10.1073/pnas.1802611115 (2018).
Christensen, S. M. et al. Host and parasite responses in human diffuse cutaneous leishmaniasis caused by L. amazonensis. PLoS Negl Trop Dis 13, e0007152, doi:10.1371/journal.pntd.0007152 (2019).
Zhang, X. A., Bontrager, A. L. & Hemler, M. E. Transmembrane-4 superfamily proteins associate with activated protein kinase C (PKC) and link PKC to specific beta(1) integrins. J Biol Chem 276, 25005-25013, doi:10.1074/jbc.M102156200 (2001).
Uhlen, M. et al. Proteomics. Tissue-based map of the human proteome. Science 347, 1260419, doi:10.1126/science.1260419 (2015).
Karlsson, M. et al. A single-cell type transcriptomics map of human tissues. Sci Adv 7, doi:10.1126/sciadv.abh2169 (2021).
Ehrlund, A. et al. The cell-type specific transcriptome in human adipose tissue and influence of obesity on adipocyte progenitors. Sci Data 4, 170164, doi:10.1038/sdata.2017.164 (2017).
Gearing, L. J. et al. CiiiDER: A tool for predicting and analysing transcription factor binding sites. PLoS One 14, e0215495, doi:10.1371/journal.pone.0215495 (2019).
Zatterale, F. et al. Chronic Adipose Tissue Inflammation Linking Obesity to Insulin Resistance and Type 2 Diabetes. Front Physiol 10, 1607, doi:10.3389/fphys.2019.01607 (2019).
Griffiths, B. et al. Sterol regulatory element binding protein-dependent regulation of lipid synthesis supports cell survival and tumor growth. Cancer Metab 1, 3, doi:10.1186/2049-3002-1-3 (2013).
Ding, L. et al. TM4SF19 aggravates LPS-induced attenuation of vascular endothelial cell adherens junctions by suppressing VE-cadherin expression. Biochem Biophys Res Commun 533, 1204-1211, doi:10.1016/j.bbrc.2020.08.078 (2020).
Luck, K. et al. A reference map of the human binary protein interactome. Nature 580, 402-408, doi:10.1038/s41586-020-2188-x (2020).
Hooper, K. M. et al. V-ATPase is a universal regulator of LC3-associated phagocytosis and non-canonical autophagy. J Cell Biol 221, doi:10.1083/jcb.202105112 (2022).
Gao, Y. et al. Golgi-associated LC3 lipidation requires V-ATPase in noncanonical autophagy. Cell Death Dis 7, e2330, doi:10.1038/cddis.2016.236 (2016).
Parra, K. J. & Kane, P. M. Wild-type and mutant vacuolar membranes support pH-dependent reassembly of the yeast vacuolar H+-ATPase in vitro. J Biol Chem 271, 19592-19598, doi:10.1074/jbc.271.32.19592 (1996).
Chakarov, S. et al. Two distinct interstitial macrophage populations coexist across tissues in specific subtissular niches. Science 363, doi:10.1126/science.aau0964 (2019).
Sarvari, A. K. et al. Plasticity of Epididymal Adipose Tissue in Response to Diet-Induced Obesity at Single-Nucleus Resolution. Cell Metab 33, 437-453 e435, doi:10.1016/j.cmet.2020.12.004 (2021).
Cho, C. H. et al. Angiogenic role of LYVE-1-positive macrophages in adipose tissue. Circ Res 100, e47-57, doi:10.1161/01.RES.0000259564.92792.93 (2007).
Yamauchi, T. et al. The fat-derived hormone adiponectin reverses insulin resistance associated with both lipoatrophy and obesity. Nat Med 7, 941-946, doi:10.1038/90984 (2001).
Tafere, G. G., Wondafrash, D. Z., Zewdie, K. A., Assefa, B. T. & Ayza, M. A. Plasma Adipsin as a Biomarker and Its Implication in Type 2 Diabetes Mellitus. Diabetes Metab Syndr Obes 13, 1855-1861, doi:10.2147/DMSO.S253967 (2020).
Emont, M. P. et al. A single-cell atlas of human and mouse white adipose tissue. Nature 603, 926-933, doi:10.1038/s41586-022-04518-2 (2022).
Tao, X. et al. Overexpression of PDE4D in mouse liver is sufficient to trigger NAFLD and hypertension in a CD36-TGF-beta1 pathway: therapeutic role of roflumilast. Pharmacol Res 175, 106004, doi:10.1016/j.phrs.2021.106004 (2022).
Omar, B., Banke, E., Ekelund, M., Frederiksen, S. & Degerman, E. Alterations in cyclic nucleotide phosphodiesterase activities in omental and subcutaneous adipose tissues in human obesity. Nutr Diabetes 1, e13, doi:10.1038/nutd.2011.9 (2011).
Xiong, X. et al. Landscape of Intercellular Crosstalk in Healthy and NASH Liver Revealed by Single-Cell Secretome Gene Analysis. Mol Cell 75, 644-660 e645, doi:10.1016/j.molcel.2019.07.028 (2019).
Ramachandran, P. et al. Resolving the fibrotic niche of human liver cirrhosis at single-cell level. Nature 575, 512-518, doi:10.1038/s41586-019-1631-3 (2019).
Seidman, J. S. et al. Niche-Specific Reprogramming of Epigenetic Landscapes Drives Myeloid Cell Diversity in Nonalcoholic Steatohepatitis. Immunity 52, 1057-1074 e1057, doi:10.1016/j.immuni.2020.04.001 (2020).
Langston, P. K. et al. Glycerol phosphate shuttle enzyme GPD2 regulates macrophage inflammatory responses. Nature immunology 20, 1186-1195, doi:10.1038/s41590-019-0453-7 (2019).
Im, S.-S. et al. Linking Lipid Metabolism to the Innate Immune Response in Macrophages through Sterol Regulatory Element Binding Protein-1a. Cell Metabolism 13, 540-549, doi:https://doi.org/10.1016/j.cmet.2011.04.001 (2011).
Conboy, I. M., Manoli, D., Mhaiskar, V. & Jones, P. P. Calcineurin and vacuolar-type H<sup>+</sup>-ATPase modulate macrophage effector functions. Proceedings of the National Academy of Sciences 96, 6324-6329, doi:doi:10.1073/pnas.96.11.6324 (1999).
Rao, Z. et al. Vacuolar (H(+))-ATPase Critically Regulates Specialized Proresolving Mediator Pathways in Human M2-like Monocyte-Derived Macrophages and Has a Crucial Role in Resolution of Inflammation. Journal of immunology (Baltimore, Md. : 1950) 203, 1031-1043, doi:10.4049/jimmunol.1900236 (2019).
Thomas, L. et al. Selective upregulation of TNFα expression in classically-activated human monocyte-derived macrophages (M1) through pharmacological interference with V-ATPase. Biochemical pharmacology 130, 71-82, doi:10.1016/j.bcp.2017.02.004 (2017).
Toda, G., Yamauchi, T., Kadowaki, T. & Ueki, K. Preparation and culture of bone marrow-derived macrophages from mice for functional analysis. STAR Protoc 2, 100246, doi:10.1016/j.xpro.2020.100246 (2021).
Lee, S. et al. Real-time in vivo two-photon imaging study reveals decreased cerebro-vascular volume and increased blood-brain barrier permeability in chronically stressed mice. Sci Rep 8, 13064, doi:10.1038/s41598-018-30875-y (2018).
Lee, S. H. et al. Real-Time Monitoring of Cancer Cells in Live Mouse Bone Marrow. Front Immunol 9, 1681, doi:10.3389/fimmu.2018.01681 (2018).
Stransky, L. A. & Forgac, M. Amino Acid Availability Modulates Vacuolar H+-ATPase Assembly. J Biol Chem 290, 27360-27369, doi:10.1074/jbc.M115.659128 (2015).
Schmidt, H. et al. Enrichment and analysis of secretory lysosomes from lymphocyte populations. BMC Immunol 10, 41, doi:10.1186/1471-2172-10-41 (2009).
McCarthy, D. J., Campbell, K. R., Lun, A. T. & Wills, Q. F. Scater: pre-processing, quality control, normalization and visualization of single-cell RNA-seq data in R. Bioinformatics 33, 1179-1186, doi:10.1093/bioinformatics/btw777 (2017).
Lun, A. T., McCarthy, D. J. & Marioni, J. C. A step-by-step workflow for low-level analysis of single-cell RNA-seq data with Bioconductor. F1000Res 5, 2122, doi:10.12688/f1000research.9501.2 (2016).
Butler, A., Hoffman, P., Smibert, P., Papalexi, E. & Satija, R. Integrating single-cell transcriptomic data across different conditions, technologies, and species. Nat Biotechnol 36, 411-420, doi:10.1038/nbt.4096 (2018).
Germain, P. L., Lun, A., Garcia Meixide, C., Macnair, W. & Robinson, M. D. Doublet identification in single-cell sequencing data using scDblFinder. F1000Res 10, 979, doi:10.12688/f1000research.73600.2 (2021).
Korsunsky, I. et al. Fast, sensitive and accurate integration of single-cell data with Harmony. Nat Methods 16, 1289-1296, doi:10.1038/s41592-019-0619-0 (2019).
Setty, M. et al. Characterization of cell fate probabilities in single-cell data with Palantir. Nat Biotechnol 37, 451-460, doi:10.1038/s41587-019-0068-4 (2019).
Trapnell, C. et al. The dynamics and regulators of cell fate decisions are revealed by pseudotemporal ordering of single cells. Nat Biotechnol 32, 381-386, doi:10.1038/nbt.2859 (2014).
Qiu, X. et al. Reversed graph embedding resolves complex single-cell trajectories. Nat Methods 14, 979-982, doi:10.1038/nmeth.4402 (2017).
Ge, S. X., Son, E. W. & Yao, R. iDEP: an integrated web application for differential expression and pathway analysis of RNA-Seq data. BMC Bioinformatics 19, doi:10.1186/s12859-018-2486-6 (2018).
Huang da, W., Sherman, B. T. & Lempicki, R. A. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc 4, 44-57, doi:10.1038/nprot.2008.211 (2009).
Phipson, B., Lee, S., Majewski, I. J., Alexander, W. S. & Smyth, G. K. Robust Hyperparameter Estimation Protects against Hypervariable Genes and Improves Power to Detect Differential Expression. Ann Appl Stat 10, 946-963, doi:10.1214/16-AOAS920 (2016).
Fonseka, C. Y. et al. Mixed-effects association of single cells identifies an expanded effector CD4(+) T cell subset in rheumatoid arthritis. Sci Transl Med 10, doi:10.1126/scitranslmed.aaq0305 (2018).

Mice

All animal protocols were approved by the Institutional Animal Care and Use Committee at Seoul National University (SNU-211209-4-1, SNU-230130-5, SNU-210403-2) and conducted in strict compliance with the guidelines for humane care and use of laboratory animals specified by the Ministry of Food and Drug Safety. Mice were housed at 22 ± 1°C and maintained on a 12-h light/12-h dark cycle with free access to food and water. Only male mice were used for experiments. C57BL/6N-Tm4sf19^em1cyagen(TM4SF19 KO; Cyagen, KOCMP-20925-Tm4sf19) mice were established by the CRISPR/Cas9 system with the use of single guide RNA (5′-CTGACTCGTCTAGTGG-3′). The primers used for genotyping to confirm TM4SF19 KO were as follow: forward: 5’-TTAGAGGAAGTCCTTGAGACCCC-3’ and Reverse: 5’- ATCTGTATGACCGTTCACTTTGGA-3’.

Inducible macrophage-specific TM4SF19 KO (Tm4sf19^flox/flox/Csf1r-CreER: TM4SF19 MKO) mice were generated by crossing Tm4sf19^flox/flox mice (C57BL/6JSmoc-Tm4sf19^{em1(flox)Smoc}, Shanghai model organism, stock#: NM-CKO-200286) with Csf1r-CreER (FVB-Tg (Csf1r-cre/Esr1*)1Jwp/J, Jackson laboratory, stock# 019098) mice. Tm4sf19^flox/flox mice without Csf1r-CreER were used for wild-type (WT) control. For macrophage-specific TM4SF19 KO induction, tamoxifen (75mg/kg, Cayman, 13258) dissolved in sunflower oil was treated to Tm4sf19^flox/flox/Csf1r-CreER mice and WT controls at 6 weeks of age by oral gavage on each of 5 consecutive days. Ten days after the last dose of tamoxifen treatment, mice were used for each experiment. To prolong the duration of Cre recombinase activity in mice, tamoxifen treatment was performed at the age of 6, 10, 14 weeks and mice were sacrificed after 12 weeks of HFD feeding. The genotyping primers to detect the flox or WT were as follows: forward: 5′- AGGGCAAAGAAGGAAGTGGCTAAT -3′, reverse: 5′-CAGGAAGGGGGCAGACAAGGAGT-3′. The genotyping primers to detect CreER were as follows: forward: 5′-CTTCCAAAGCATGGTCCAGT-3′, reverse: 5′-TGAACCAGCTCCCTATCTGC-3′.

For the diet-induced obesity model, mice at 6 weeks of age were fed a 60% fat diet (HFD; Research Diets, D12492) for 12 weeks. A standardized rodent pellet diet (NCD; Purina, 38057) was used for a control chow diet. Energy expenditure, locomotor activities, and food intake were estimated by indirect calorimetry system (PhenoMaster; TSE Systems). Body composition was measured by nuclear magnetic resonance scanning EchoMRI-700 (Echo Medical Systems).

For the glucose tolerance test, mice were fasted for 12 hours before measuring blood glucose levels. Mice were injected with d-glucose (2 g/kg body weights, Sigma, G7021) by intraperitoneal injection. For the insulin tolerance test, mice were treated with insulin by intraperitoneal injection (0.75 units/kg, Sigma, 91077C). Blood glucose levels were measured from tail vein blood samples at an indicated time using GlucoDoctor Top meter (Allmedicus, AGM-4100) and appropriate glucose indicator strips. For insulin tolerance test, mice were fasted for 6 hours and insulin was injected by intraperitoneal injection. To examine insulin signaling, insulin (0.75 units/kg) was injected into anesthetized mice by 2,2,2-tribromoethyl alcohol (avertin; Sigma, T48402) with the inferior vena cava and adipose tissues were harvested 10 min after the injection. To keep mice temperature at 37℃, we used the heating pad system (Toyotech, DP30-05A). For serum insulin, total cholesterol, and triglyceride levels, mouse insulin ELISA kit (Fujifilm, AKRIN-011T) and chemical blood analyzer (Fujifilm, DRI-CHEM 3500s) were used.

Bone marrow transplantation

Bone marrow transplantation (BMT) was performed with whole bone marrow cells from WT C57BL/6N and TM4SF19 KO mice. Bone marrow cells were prepared by flushing the tibia and femur with DMEM ⁵¹. Tail vein injection (2x10⁶ cells) was performed on lethally irradiated 6-week-old recipient WT mice under anesthesia. After a 4 weeks recovery period, mice were initiated to be fed an HFD for 12 weeks.

Human

Human subcutaneous fat tissues were collected from Korea University Guro Hospital (IRB no.:2022GR0095), and tissues were stored at -80℃. The characteristic of human fat tissues from 24 individuals is available in Supplementary Table 2.

Histology

BAT, IWAT, and GWAT were harvested and fixed in 10% formalin (Sigma, HT50128) for 24 hours at 4°C and then embedded in paraffin. 5 µm paraffin sections were prepared from the paraffin-embedded tissue blocks. For hematoxylin and eosin (H&E) staining, deparaffinized sections were stained with ClearView Staining Hematoxylin (BBC Biochemical, MA010081) and Eosin Y Alcoholic solution (BBC Biochemical, 3610). The images were obtained with Nikon Elements (NIS BR Analysis ver 5.10.00). For F4/80 staining in liver and GWAT, 5 µm paraffin sections were deparaffinized with 2 x 10 min in 100% xylene (Daejung, 1330-20-7), 10 min in 50:50 xylene:ethanol, 5 min in 100% ethanol (Daejung, 4022–4410), 5 min in 95% ethanol, 5 min in 70% ethanol. After deparaffinization step, the slides were rinsed with distilled water, boiled for 10 min in citrate buffer (pH 6.0) for antigen retrieval, and cooled to room temperature for 10 min. Blocking was performed with PBS mixed with 3% BSA for 1hours, and F4/80 was stained with antibody. DAPI (1000x) was stained for 5min after antibody staining. Images were acquired with Zeiss confocal microscope (LSM800) and analyzed with Zen software (version 3.0).

Western blot and quantitative PCR

To perform western blot analysis, BAT, IWAT, and GWAT depots from WT and TM4SF19 KO mice were homogenized in PRO-PREP Protein Extraction Solution (iNtRON Biotechnology, 17081) containing SIGMAFAST Protease Inhibitor Cocktail (Sigma, S8820) and PhoSTOP phosphatase inhibitors (Roche, 4906845001) using tacoPrep Bead Beater homogenization system. The protein concentration of protein samples was quantified by BCA assay and measured by spectrophotometry (MultiSkan GO, 51119000, Thermo Fisher Scientific) at 562 nm using SkanIt software ver 5.0. 10µg protein samples were denatured in 5x sample buffer (Elpis Biotech, EBA 1052) in 95 ℃ for 5 mins and separated on 8% or 12% of SDS-PAGE gel and transferred to PVDF membrane (Bio-Rad). The membrane was blocked with 5% non-fat dry milk or BSA in TBST (Tris-buffered saline with 0.1% Tween 20) and incubated with primary antibody at 4°C overnight followed by horseradish peroxidase (HRP)-conjugated secondary antibody incubation at room temperature for 1 hour. Anti-rabbit TM4SF19 antibody (YT6290, WB 1:1000) were purchased from Immunoway. Anti-rabbit α/β-Tubulin (2148 ,WB 1:2000), anti-rabbit NF-ĸB p105/50 (13586 ,WB 1:1000), anti-rabbit ATP6V1B2 (14617, WB 1:1000), anti-rabbit LAMP1 (3243, WB 1:1000), anti-rabbit F4/80 (30325, IHC 1:200), anti-rabbit p-IRS1 (S612) (3203, WB 1:1000), anti-rabbit IRS1 (3407, WB 1:1000), anti-rabbit COXIV (4850, WB 1:1000), anti-rabbit Myc-Tag (2278, IP 1:200), and normal rabbit IgG (2729, IP 1:100) were purchased in Cell Signaling. Anti-rabbit ATP6V0B (NBP2-83943 WB 1:1000) was purchased in Novus Biological. Anti-rabbit ATP6V1A (GTX110815, WB 1:1000) was purchased in GeneTEX. Anti-mouse p-AKT (S473) (Sc-514032, WB (1:1000)) and anti-mouse AKT (Sc-81434, WB 1:1000) were purchased from Santa Cruz. Anti-mouse total oxphos cocktail (Ab110413,WB 1:1000) was purchased from Abcam.

For quantitative PCR (qPCR) analysis, total RNA from fat tissues from mice and humans were isolated by TRIzol reagent (Thermo Fisher Scientific, 15596018) according to the manufacturer’s protocols. High-Capacity cDNA Reverse Transcription kit (Applied Biosystems, 4368814) was used to synthesize cDNA and qPCR reaction was performed with iQ SYBR Green Supermix (Bio-Rad, 170–8884) in Bio-Rad CFX Connect Real Time PCR Detection system. β-Actin and Peptidylprolyl isomerase A (PPIA) were used as a housekeeping gene and primers used for qPCR are listed in Supplementary table 3. 2-ΔCt method was used to calculate the relative expression levels of each gene.

Flow cytometric analysis

GWAT from WT and TM4SF19 KO mice after feeding NCD or HFD were used for flow cytometric analysis. Adipose tissues were isolated and digested with filtered 2 mg/mL collagenase type I (Gibco, 17100-017) in KRBB buffer containing 3% BSA at 37°C. Fully dissociated preparations were then passed through a 100 µm cell strainer and centrifuged at 500 × g for 5 min. Pellets were then washed with DMEM containing 20% FBS two times. Stroma vascular fraction (SVF) stained with fluorescence labeled appropriate primary antibodies at room temperature. BV421 Rat Anti-Mouse F4/80 (565411, 1:50) were purchased from BD Biosciences. Human/Mouse TREM2 Allophycocyanin Mab (FAB17291A, 1:50) were purchased from R&D systems. APC/Cyanine7 anti-mouse CD45 (103116, 1:100), FITC anti-mouse/human CD11b (101206, 1:100), Brilliant Violet 421™ anti-human CD206 (141717, 1:100), PE/Cyanine7 anti-mouse CD11c (117317, 1:100) were purchased from Biolegend.

Unstained and single stained controls were used for compensation. LSRFortessa X-20 Flow Cytometer (BD Biosciences) was used to analyze the samples and the data was acquired by BD FACSDiva 8.0 software which then was further analyzed by Flowjo software (version 10.5.3; TreeStar). The gating strategies are provided in Supplementary Fig. 2.

Adipocyte fractionation and magnetically activated cell sorting (MACS)

To collect floating adipocytes, fully dissociated GWAT as described above were centrifuged at 300 x g for 5min. Red blood cells were lysed with RBC lysis buffer (15.5 mM NH4Cl, 1.2 mM NaHCO3 and 50 mM EDTA, pH 7.4) and pellets (SVF) were then washed with PBS with 3% BSA (500 x g for 5min). SVF were incubated with anti-F4/80-FITC antibody (Biolegend) for 2 hours at 4℃ followed by anti-FITC-microbeads (Miltenyi Biotec, 130-048-701, 1:10) incubation for 1 hour at 4°C. After incubation, F4/80 + cells were isolated by column (Miltenyi Biotec, 130-122-727) according to the manufacturer’s instruction (Miltenyi Biotec, MACS technique).

Cell culture

Dulbecco’s Modified Eagle Medium (DMEM; Welgene, LM001-07) supplemented with 10% fetal bovine serum (FBS; Gibco, 16000044) and 1% penicillin/streptomycin (PS; Welgene, LS202-02) were used for growth medium. RAW264.7 macrophages (ATCC, TIB-71) were maintained in growth medium until experiments were initiated. To check the localization of TM4SF19, we used lysotracker following manufacturer’s protocols. For colocalization analysis, the pearson correlation coefficient was measured by JACOP in ImageJ software (version 1.52a).

C3H10T1/2 cells (ATCC, CCL-226) were used to conduct in vitro adipocyte experiments. For adipogenic lineage commitment, C3H10T1/2 cells were exposed to 20 ng/mL bone morphogenetic protein 4 (BMP4) (R&D systems, 314-BP) in the growth medium with for 3 days. For adipogenic differentiation, the cells were incubated with the differentiation medium [growth medium containing 0.125 mmol/L indomethacin (Cayman Chemical, 70270), 2.5 mmol/L isobutyl methylxanthine (Cayman Chemical, I5879), 1 µmol/L dexamethasone (Cayman Chemical, D4902), 10 µg/mL insulin, and 1 nmol/L triiodothyronine (Cayman Chemical, 16028)] for 3 days. Then the cells were maintained with the growth medium containing 10 µg/mL insulin and 1 nmol/L triiodothyronine until the experiments performed.

Bone marrow-derived macrophages (BMDMs) were extracted from femur and tibia of 6 weeks WT and TM4SF19 KO mice ⁵¹. Briefly, the bone marrows from the femur and tibia were washed and resuspended by injecting 2mL of DMEM and centrifuge bone marrow suspension at 500 x g for 5 minutes at 4 ℃. Bone marrow was resuspended in 1mL of RBC lysis buffer and centrifuge at 500 x g for 5 minutes at 4℃. For BMDM differentiation, bone marrow cells were incubated with bone marrow culture medium [growth medium with 1ng/mL M-CSF1 (Peprotech Ltd, 315-02)] for 3days at 37℃, 5% CO₂. On the third day, media was changed with fresh bone marrow culture medium for an additional 3 day of differentiation. LysoSensor Yellow/Blue dextran (Invitrogen, L22460) staining was performed to measure lysosomal pH following the manufacturer’s protocols. 1mg/mL of LysoSensor Yellow/Blue dextran was incubated with 24 hours, and 4% PFA was incubated 10 min for fixation. Images were acquired with Zeiss (LSM800) and analyzed with Zen software (version 3.0).

Subcellular fractionation

To collect the membrane and cytosolic fractions, we performed subcellular fractionation using the method described previously ⁵⁴. A total of 2x10^8 cells were lysed and homogenized using fractionation buffer. The homogenates were then centrifuged at 500 x g for 10 minutes to clear the nuclei and intact cells. The resulting supernatant was collected and subjected to further centrifugation at 100,000 x g for 4 hours to isolate the cytosolic (supernatant) and membrane (pellet) fractions. The resulting pellet was lysed with RIPA buffer (Thermo Fisher Scientific, 89901). To confirm the purity of the isolated membrane and cytosol fractions, two marker proteins, LAMP1 and b-actin, were used for membrane and cytosolic fraction markers, respectively.

Lysosome isolation

Lysosome isolation was performed using a previously described method ⁵⁵. A total of 2x10⁸ cells were rinsed and collected with a fractionation buffer. The samples were then centrifuged at 1000 x g for 10 minutes, and collect the resulting supernatant. The supernatant was centrifuged at 20000 x g for 20 minutes and the resulting pellets were mixed with 19% diluted Optiprep density gradient medium (Biovision, M1248). The mixture was then loaded onto a low osmotic discontinuous density gradient with 27%, 22.5%, 19%, 16%, 12%, and 8% ⁵⁵ and centrifuged at 150,000 x g for 4 hours. Subcellular fractions were eluted with RIPA lysis buffers (Thermo Fisher Scientific, 89901). To verify the purity of the lysosome fractions, the lysosome marker LAMP1 was tested in individual fractions by immunoblot analysis. It was found that fraction 2 (22.5%) was enriched with the lysosome. High-purity lysosome fractions (fraction 2) were used to measure V-ATPase activity by ATPase activity assay (Abcam, ab234055) following the manufacturer’s protocols. Briefly, 50µg of fresh lysosomal proteins were incubated with 5mM sodium azide (mitochondrial ATPase inhibitor) in the presence or absence of concanamycin A (ConA: specific inhibitor of V-ATPase) for 30 minutes at 37℃. The V-ATPase activity was normalized with experimental values in the absence of ConA and was then subtracted with control values in the presence of ConA.

Gene overexpression

Lentiviral transfer plasmid Tm4sf19 ORF was generated by inserting Tm4sf19 (NM_001160402.1) ORF clone to the transfer vector pLenti-EF1a-C-mGFP-P2A-Puro lentiviral vector (Origene, PS100121). Myc-tagged transfer plasmid was generated through GFP-fused TM4SF19 by replacing mGFP with myc-DDK. Lentivirus was produced by transfection of HEK293T cells with transfer plasmids (5 µg), psPAX2 (3.75 µg, Addgene, 12260), and pMD2.G (1.25 µg, Addgene, 12259) in 4:3:1 ratio using jetPRIME transfection reagent (Polyplus, 114 − 15). The medium containing viral particles was collected after 48 h of transfection, centrifuged, and filtered through a 0.45-nm syringe filter (Sartorius, STR.16555K). RAW264.7 macrophages were transduced with the viral supernatants containing 8 µg/mL polybrene (Santa Cruz, sc-134220) for 72 hours and were replaced with growth medium containing 2 µg/mL puromycin (Sigma, P8833) for selection for 48 hours.

Immunoprecipitation of Myc-tagged proteins

For immunoprecipitation analysis, RAW264.7 macrophages overexpressing Myc-fused TM4SF19 were harvested by lysis buffer (50 mM Tris-HCl (pH 7.6), 150 mM NaCl, 2 mM EDTA, and protease inhibitor). After 10 minutes of incubation on ice, lysates were centrifuged at 13,000 rpm, at 4℃ for 15 minutes and the pellet was discarded. 50 µL of Pierce™ Protein A/G Magnetic Beads (Invitrogen, 88802) were incubated with 5 µL of Myc and IgG antibodies for 2 hours rotating at 4℃. Beads were washed once in lysis buffer before incubation with 300 µg of lysate for 4 hours, rotating at 4℃. Beads were then washed five times in a buffer containing Tris-buffered saline (TBS) containing 0.05% Tween-20 Detergent. Bound proteins were eluted by 0.1M glycine (pH 2.0).

Phagocytosis assay

For phagocytosis analysis, fully differentiated C3H10T1/2 adipocytes were treated 5µg/ml brefeldin A (BFA; Biolegend, 420601) for 24 hours and were stained with 4,4-difluoro-5-(2-thienyl)-4-bora-3a, 4a-diaza-s-indacene3-dodecanoic acid (BODIPY 558/568 C12) (Invitrogen, D3835), and macrophages were labeled with Vybrant DiO Cell-Labeling Solution (Invitrogen, V22886) overnight. BODIPY-labeled C3H10T1/2 cells were detached, added to the DiO-labeled macrophages, and then co-cultured for 16 hours. Live cell imaging was performed with Operetta CLS High-Content Analysis System (Perkin Elmer) and analyzed by Harmony High-Content Imaging and Analysis Software (Perkin Elmer). For detection of high-resolution images of phagocytosis of dying adipocytes by the macrophages, fully differentiated adipocytes were treated 5µg/ml BFA(Biolegend) for 24 hours and then co-cultured with BMDM stained by Alexa488-cholera toxin B (CtB, Invitrogen, C22841) for 90 min. After co-culturing, fixation was performed with 4% PFA for 15 min and then LipidTOX (Invitrogen, H34476) was stained for 1 hour according to the manufacturer’s protocols. Images were acquired with Zeiss (LSM800) or Leica (TCSSP8) confocal microscope and analyzed with Zen software (version 3.0) or Leica software (LASX).

Promoter analysis

CiiiDER, an integrated computational toolkit for transcription factor binding analysis, was performed to analyze the promoter reigon (2.5kb) of mouse Tm4sf19 and human TM4SF19 gene (ENSMUSG00000079625, ENSG00000145107). Only match scores of 0.85 or above were accepted to retrieve the accurate transcription factor binding sites. For the chromatin immunoprecipitation (ChIP) analysis, MAGnify Chromatin Immunoprecipitation System kit (Invitrogen, 49-2024) was used. RAW264.7 macrophages were treated with LPS for 4 hours prior to crosslink. The macrophages were crosslinked with 1% formaldehyde for 10 minutes at room temperature and the reaction was stopped by adding 1.25 M glycine. Immunoprecipitation was performed using the antibody against NF-κB1 p105/p50 (Cell Signaling, 13586) or a normal IgG (Cell Signaling, 2729) as a negative control. To perform ChIP-qPCR analysis and gel-electrophoresis, precipitated DNA fragments were analyzed by specific primers for the putative NF-κB binding site bearing promoter region of TM4SF19 (Forward: 5′- AGGGGTGCTCTCTCAGAAG-3′ and Reverse: 5′-GCTCAACGGTCTTAGT-3′).

Single-nucleus RNA sequencing

Tissues were incubated in TST lysis buffer (146 mM NaCl, 10mM Tris-Cl pH7.5, 1mM CaCl₂, 21 mM MgCl₂, 0.03% Tween-20, 0.01% BSA) for 10 minutes at 4℃ and minced using Noyes spring scissors. Samples were filtered through a 70µm strainer and centrifuged at 500 x g for 5 min at 4℃. Pellets were washed once with TST lysis buffer and twice with 1% BSA/PBS. Pelleted nuclei were resuspended in 1% BSA/PBS with 0.2U/µl RNase inhibitor and filtered through a 30µm strainer. Nuclei were stained with acridine orange/propidium iodide solution and counted using an automated counter (LUNA-FX7, Logos Biosystems, L70001).

For sample multiplexing, 1–2 x 10⁶ nuclei per a sample were stained with 100µl of CellPlex (10X Genomics, PN-1000261). Samples were washed three times with 2ml of cold 1% BSA/ PBS and centrifuged at 500 x g for 5 min at 4℃. Nuclei were resuspended in 1% BSA/PBS and equal numbers of cells from each biological replicate for a condition were mixed. Nuclei were counted and added to the reverse transcription mixture with the aim of capturing 20,000 nuclei of the combined sample.

Libraries for single-nucleus RNA sequencing were generated using the Chromium Next GEM Single Cell 3’Kit (10X Genomics, PN-1000269), 3’Feature Barcode Kit (10X Genomics, PN-1000262), Dual Index Kit TT Set A (10X Genomics, PN-1000215), and Dual Index Kit NN Set A (10X Genomics, PN-1000243). To generate gel bead-in-emulsions (GEMs), samples were loaded onto the Chromium Next GEM Chip G (10X Genomics, PN-2000127) and the subsequent microfluidic process was performed on a Chromium controller (10X Genomics). Reverse transcription was performed by incubating GEMs in a thermal cycler (C1000 Touch Thermal Cycler with Deep Well, Bio-Rad). After incubation, GEMs were disrupted and pooled cDNA was purified using silane magnetic beads. Amplification of cDNA was performed using a thermal cycler. During cDNA size selection using SPRIselect beads, the supernatant and bead pellet portion were handled separately to obtain cDNA from poly-adenylated mRNA and DNA from cell multiplexing oligo feature barcode. To generate cell multiplexing libraries, 75µl of supernatant was retained and further purified. The cell multiplexing barcode DNA and indexing primers (10X Genomics, PN-3000482) were added to the PCR reaction mixture and PCR amplification was performed. For cDNA library generation, bead pellets were purified and cDNA was eluted. Subsequent library construction procedures, including enzymatic fragmentation, end repair, A-tailing, adaptor ligation, and sample index PCR, were performed according to the manufacturer’s protocol.

Size and concentration of cDNA and libraries were analyzed using a Bioanalyzer (Agilent). Libraries were pooled and sequenced on a HiSeq X Ten sequencer (Illumina) using a 100bp paired-end protocol to generate a minimum of 20,000 read pairs per cell for 3’gene expression library and 5,000 read pairs per cell for the cell multiplexing library.

Single-nucleus RNA sequencing data pre-processing

Raw reads from multiplexed single-nucleus RNA sequencing data were aligned to the mouse genome (mm10) and demultiplexed by sample-specific barcodes using cellranger (v6.1.1). The raw count matrix for each sample was generated using cellranger multi with min-assignment-confidence = 0.7, expect-cells = 20000 and include-introns = true options. R (v4.1.0) was used for analysis. Low-quality cells with detected UMIs lower than 1,000 or percentage of genes mapped to mitochondrial genes higher than 5% were filtered out for further analysis. The threshold was determined by visually checking the distribution of per-cell UMI count using addPerCellQC function of scater (v1.22.0) R package ⁵⁶. All count matrices were combined and cells were clustered using quickCluster function of scran (v1.22.1) R package ⁵⁷. Cell-specific size factors were computed using computeSumFactors function of the same package. The normalized count matrix was generated with logNormCounts function using the calculated size factors with pseudo_count = 1. To select highly variable genes (HVGs), variance of normalized expression for each gene was modeled using modelGeneVar and getTopHVGs function of scran R package with fdr.threshold = 0.05. The dataset was scaled and centered using ScaleData function of Seurat (v4.0.5) ⁵⁸. The top 25 principal components (PCs) are computed on HVGs and used for clustering and UMAP dimension reduction. To identify molecular characteristics of each cluster, marker genes of each cluster were obtained using FindAllMarkers function of Seurat. Cell types were manually annotated in resolution of clusters based on identified molecular characteristics. Putative doublets were confirmed using scDblFinder (v1.8.0) R package ⁵⁹. These cells were manually double-checked whether they expressed two or more cell type markers from different lineages and removed. Cells expressing erythroid cell markers such as Hba-a1, Hba-a2 and Alas were removed. There were some clusters presumed to be derived from epididymis, anatomically close to the GWAT. These cells were annotated as epididymal cells (Abcb5, Ces5a, Adam7), spermatozoas (Dnah12, Spef2, Hydin) and efferent duct cells (Adcy8, Dnah5, Slc9a3) and removed for further analysis. The remained cells after quality control were re-clustered using the same functions described above with the top 15 PCs. The batch effect caused by different samples was corrected by RunHarmony function of harmony (v0.1.0) R package ⁶⁰.

Single-nucleus RNA sequencing data analysis

Trajectory analysis

Pseudotime analysis was conducted using Palantir (v1.0.0) python package for monocyte/macrophage and APC/Adipocyte cell types ⁶¹. To construct pseudotime trajectory on tSNE plot, the variable features were selected using getTopHVGs function and renormalized count matrix of intended populations was extracted. For monocyte and macrophage lineage trajectory, the start_cell was chosen randomly in the monocyte cluster. The first 150 PCs were used to make a diffusion map using run_diffusion_maps with n_components = 10 and determine_multiscale_space function with the default options. Pseudotime and differentiation trajectory were obtained using run_palantir function with num_waypoints = 2000. The coordinates of tSNE on diffusion components were computed using run_tsne function with perplexity = 700. The constructed trajectory was visualized on tSNE plot using plot_palantir_results function. After that, to find the more reliable starting cell, the cell which has the lowest pseudotime was selected as a start_cell and the above process was repeated. For adipose lineage trajectory, the start_cell was designated as one of the cells expressing Dpp4 and Cd34 (Figure S2F). Trajectory was constructed following the same methods as mentioned above except for these options: the number of used PC = 150, num_waypoints = 5000 for run_palantir function and perplexity = 900 for run_tsne function. The start_cell was selected again as described above.

Monocle3 (v1.3.1) was used for confirmation of monocyte to macrophage trajectory ^62,63. We used same HVGs and normalized count matrix of monocyte and macrophage populations as used in Palantir. We preprocessed the data using preprocess_cds function with 20 PCs and reduced dimension using UMAP. Clustering was performed using cluster_cells function with resolution of 0.0001. The trajectory was constructed using learn_graph with default parameters.

Cell type composition

To calculate the cell type composition, the number of cells in each cell type for each sample or condition was divided by the total cell number of each sample. The total count of the cell in each sample was scaled as 100%.

DE (Differential Expression) and GO (Gene Ontology) analysis

For publicly available transcriptomic database analysis, differentially expressed genes (DEGs) between different body fat conditions (NCD vs HFD and lean vs obese) were demonstrated by iDEP 9.6⁶⁴ and GEO2R provided by NCBI. The significant DEGs of top 100 upregulated genes in each dataset were determined by adjusted P-value < 0.05 and fold change > 2. DAVID (https://david.ncifcrf.gov) was used for functional enrichment analysis ⁶⁵. DEGs were determined between Adipocyte.1 and Adipocyte.3 using limma (v3.50.0) R package ⁶⁶. Briefly, lmFit function was used to fit the model on the normalized count for each gene and batch effects were considered using duplicateCorrelation function by adding sample information to the block option. Statistical test was conducted using eBayes function with trend = T and robust = T options. The significant DEGs (adjusted P-value < 0.05) between Adipocyte.1 and Adipocyte.3 were used as input. The DEGs up-regulated in Adipocyte.1 and up-regulated in Adipocyte.3 were separately used for each cluster-enriched pathway analysis.

For gene set enrichment analysis (GSEA), DEGs between the WT HFD and TM4SF19 KO HFD conditions were obtained using FindMarkers function of Seurat R package and Molecular Signatures Database (MSigDB) was downloaded using msigdbr (v7.4.1) R package. DEGs were sorted by average log2 fold change and applied for GSEA using fgsea (v1.21.2) R package.

Calculating Signature score

Trem2 + macrophage and Lyve1 + macrophage signature scores were calculated using AddModuleScore function of Seurat. Up-regulated DEGs were used as signature genes. DEGs were determined between Lyve1 + macrophage and Trem2 + macrophage in wild-type mice (WT NCD, WT HFD), filtered with adjusted P-value < 0.05. Mac1 and Mac3 signature scores were computed using same function. The DEGs between Mac3 (Trem2-expressing macrophage) and Mac1 (Lyve1-expressing macrophage) was obtained from the previous study (Jatin et al., 2019) and filtered with p-value < 0.05 and log2fc > 1. The significance was determineded by MASC (v0.0.0.9000) R package ⁶⁷ and batch was considered as random effect.

Co-expression network analysis

The co-expression gene modules were obtained using hdWGCNA (v0.2.04) R package, which was applied after sub-clustering the monocyte and macrophage population of the WT HFD condition. 1837 metacells were constructed using the MetacellsByGroups function with k = 25 parameter and genes expressed in at least 5% of the cells used for downstream analysis. Co-expression gene network was constructed using the ConstructNetwork function with soft power threshold of 3, chosen using the TestSoftPowers function. DEGs between WT HFD and TM4SF19 KO HFD conditions of monocyte and macrophage population were obtained using FindMarkers function of Seurat R package, and down- or up-regulated genes were determined with adjusted P-value < 0.05 and |log2fc| > 0.2.

Statistics

GraphPad Prism 7 software (GraphPad Software, USA) was used for statistical analysis and correlation analysis. Data are presented as mean ± standard errors of the mean (SEM) or mean ± standard deviation (SD) as indicated in the Figure Legends. National Institutes of Health ImageJ software (version 1.52a) was used to quantify the intensity of immunoblot and immunostaining.

There is NO Competing Interest.

ExtendedDataFigures.docx
Tm4sf19SupplementalInformation0328.docx
Supplementary Figures and Tables

Download PDF

Version 1

posted

You are reading this latest preprint version

TM4SF19-mediated brake of macrophage efferocytosis contributes to obesity-induced inflammation of adipose tissue

Status:

Version 1

Abstract

Figures

Introduction

Results

Discussion

Declarations

References

Methods

Additional Declarations

Supplementary Files

Status:

Version 1