Macrophage-Specific Connexin 43 Knockout Protects Mice from Obesity-Induced Inflammation and Metabolic Dysfunction

Adipose tissue macrophages are a major immune cell type contributing to homeostatic maintenance and pathological adipose tissue remodeling. However, the mechanisms underlying macrophage recruitment and polarization in adipose tissue during obesity remain poorly understood. Previous studies have suggested that the gap junctional protein, connexin 43 (Cx43), plays a critical role in macrophage activation and phagocytosis. Herein, we investigated the macrophage-specific roles of Cx43 in high fat diet (HFD)-induced pathological remodeling of adipose tissue. Expression levels of Cx43 were upregulated in macrophages co-cultured with dying adipocytes in vitro, as well as in macrophages associated with dying adipocytes in the adipose tissue of HFD-fed mice. Cx43 knockdown reduced lipopolysaccharide (LPS)-induced ATP release from macrophages and decreased inflammatory responses of macrophages co-cultured with dying adipocytes. Based on global gene expression profiling, macrophage-specific Cx43-knockout (Cx43-MKO) mice were resistant to HFD-induced inflammatory responses in adipose tissue, potentially via P2X7-mediated signaling pathways. Cx43-MKO mice exhibited reduced HFD-induced macrophage recruitment in adipose tissue. Moreover, Cx43-MKO mice showed reduced inflammasome activation in adipose tissues and improved glucose tolerance. Collectively, these findings demonstrate that Cx43 expression in macrophages facilitates inflammasome activation, which, in turn, contributes to HFD-induced metabolic dysfunction.


INTRODUCTION
Obesity is characterized by abnormal fat accumulation, and its growing prevalence is closely associated with a high incidence of chronic metabolic diseases, including type 2 diabetes and cardiovascular diseases (Srivastava and Apovian, 2017). Adipose tissue dysfunction is one of the major contributing factors to the pathogenesis of obesity-related metabolic diseases (Srivastava and Apovian, 2017). Typically, the abnormal expansion of adipose tissue appears to be accompanied by the recruitment of pro-inflammatory immune cells, and these chronic inflammatory responses can lead to insulin resistance in multiple metabolic organs (Reilly and Saltiel, 2017).
Macrophages are a major immune cell type present in adipose tissues (Epelman et al., 2014). Notably, changes in macrophage phenotype and polarization status can contribute to the development of obesity-related diseases (Lee et al., 2010). For example, adipose tissue of individuals with obesity recruits proinflammatory macrophages (Reilly and Saltiel, 2017), whereas anti-inflammatory macrophages participate in tissue remodeling and homeostasis of adipose tissue (White and Ravussin, 2019). However, precise mechanisms underlying macrophage recruitment to adipose tissue during obesity need to be comprehensively elucidated.
Connexin, a gap junctional protein, plays a key role in gap junctional communication between cells, allowing the conduction of small molecules. For example, gap junctionmediated coupling of electrical stimuli between cells, such as cardiomyocytes, is critical for propagating electrical action potential (Evans and Martin, 2002). In addition to gap junctional channels, connexins can form hemichannels. Previous studies have indicated that connexin 43 (Cx43) hemichannels can induce ATP release from inflammatory cells and consequently regulate the autocrine activation of macrophages via ATP signaling (Dosch et al., 2019).
In the present study, we investigated the pathophysiological role of Cx43 using a co-culture system of dying adipocytes with adipose tissue macrophages and a macrophage-specific Cx43 knockout (KO; Cx43-MKO) mouse model. Interestingly, Cx43 expression was upregulated in macrophages co-cultured with dying adipocytes, as well as in adipose tissue of mice fed a high-fat diet (HFD). Cx43 knockdown reduced extracellular ATP levels and inflammatory responses in macrophages in vitro. In vivo, Cx43-MKO mice were protected against HFD-induced inflammation and glucose intolerance.

Animals
All animal protocols were reviewed and approved by the Institutional Animal Care and Use Committee at Seoul National University (SNU-181120-3, SNU-191204-4-1). All animal experiments were conducted in strict compliance with the guidelines for humane care and use of laboratory animals specified by the Ministry of Food and Drug Safety. Male mice were used for experiments. Mice were housed at 22 ± 1°C and maintained on a 12-h light/12-h dark cycle with free access to FIGURE 1 | Cx43 is upregulated in gonadal white adipose tissue macrophages of mice by high-fat diet feeding (A) Immunoblot analysis of Cx43 and F4/80 expression in gonadal white adipose tissue (gWAT) of mice fed normal chow diet (NCD) or high-fat diet (HFD) for 8 weeks (n = 6) (B) Immuno-fluorescence images of paraffin sections of gWAT of mice fed NCD or HFD, stained for Cx43 and F4/80 (C) Flow cytometric analysis of stromovascular cells (SVCs) obtained from gWAT of WT mice fed with NCD or HFD for 8 weeks (n = 6). Unpaired, two-tailed t-tests (**p < 0.01, ***p < 0.001), each point represents biological replicate. Data are presented as mean ± S.E.M.

Western Blot Analysis
Western blot analysis was performed as described previously (Lee et al., 2015). Briefly, protein was extracted in RIPA buffer (Thermo Fisher) containing protease inhibitors (Sigma) and phosphatase inhibitors (Roche). Resolved proteins were transferred to polyvinylidene difluoride (PVDF) membranes, the membranes were incubated with blocking buffer (5% skim milk Tris Buffered Saline with Tween ® 20), primary and secondary antibodies. The antibodies used for western blot are listed in Supplementary Table S1.

Gene Expression Analysis
Quantitative PCR was performed as described previously (Lee et al., 2015). Briefly, RNA was extracted using TRIzol ® reagent (Invitrogen), and was reverse transcribed using a cDNA synthesis kit (Applied Biosystems). 100ng of cDNA was subjected to quantitative polymerase chain reaction (qPCR) in 20-μL reaction volumes (iQ SYBR Green Supermix, Bio-Rad) with 100 nM primers. qRT-PCR was performed using SYBR Green dye and CFX Connect Real-time system (Bio-Rad) for 45 cycles and fold change for all samples was calculated by using the 2−ΔΔCt method. The primers used for qPCR are listed in Supplementary Table S2. Peptidylprolyl Isomerase A (PPIA) was used as a housekeeping gene for mRNA expression analysis. Primers used for qRT-PCR were described previously Lee et al., 2016).

RNA Sequencing Analysis
RNA sequencing (RNA-seq) analyses were performed as previously described. Briefly, Trizol reagent (Invitrogen) was used for total RNA extraction of gWAT, according to the manufacturer's instruction. RNA integrity number (RIN), rRNA ratio, and concentration of samples were verified on an Agilent Technologies 2100 Bioanalyzer (Agilent Technology) using a DNA 1000 chip. For RNA-seq analysis, cDNA libraries were constructed with the TruSeq mRNA Library Kit using 1 mg of total RNA. The total RNA was sequenced by the NovaSeq 6000 System (Macrogen).
Raw sequenced reads were trimmed for adaptor sequence, and then HISAT2 v2.1.0 was used to map the trimmed reads to the reference genome. After read mapping transcript assembly was performed with Stringtie v1.3.4 days and calculated raw transcription profiles as a fragment per Kilobase of transcript per Million mapped reads (FPKM) for each gene and each sample (Fiechter et al., 2021).
The DESeq2 package (v.1.24.0) was used to normalize read counts and determine differentially expressed genes (DEGs) among samples (Love et al., 2014). DEGs were defined by a cutoff values of 1.5-fold change and p value <0.05. Principal component analysis (PCA) was performed for selection of variable genes and dimensionality reduction using prcomp function in R v.3.6.1 (R Core Team, Vienna, Austria). The Hallmark gene sets (h.all.v7.5.symbols) were used for the Gene Set Enrichment Analysis (GSEA, v.4.0.3) by using the list of genes pre-ranked by principal component (PC) one loadings (Subramanian et al., 2005). Module scores were obtained by calculating the average of Z-normalized expression levels of genes in each gene set. Using the lists of DEGs, gene ontology (Subramanian FIGURE 3 | Cx43 knockdown reduces inflammatory responses of macrophages co-cultured with dying adipocytes (A) Live imaging of RAW264.7 cells transfected with Gja1 siRNA or controls and co-cultured with dying adipocytes (dACs) for 12 h. Adipocytes were tagged with C12-BODIPY (red) and macrophages were stained with DiO (Green) (n = 3) (B) qPCR analysis of Gja1 and inflammatory cytokine markers in RAW264.7 cells transfected with Gja1 siRNA or controls and co-cultured with dACs for 24 h (n = 6) (C) Extracellular ATP level in lipopolysaccharide (LPS)-stimulated RAW 264.7 macrophages treated with Gja1 siRNA or control for 30 min (n = 3) (D) qPCR analysis of inflammatory cytokines in RAW264.7 cells transfected with Gja1 siRNA or controls, and co-cultured with dACs in the presence or absence of apyrase (2U/ml) (n = 3) (E) Schematic diagram illustrating the experimental method used for F. Conditioned media obtained from control RAW264.7 cells co-cultured with dACs were transferred into Gja1KD RAW264.7 cells (F) qPCR analysis of inflammatory cytokine markers in Gja1 KD RAW264.7 cells co-cultured with dACs (GjaKD + dAC) or exposed to the conditioned media obtained from the control co-cultures (Gja1KD + CM) (n = 3). Unpaired, two-tailed t-tests (*p < 0.1, **p < 0.01, ***p < 0.001). Each point represents biological replicate. Data are presented as mean ± S.E.M.

Statistical Analysis
Statistical analyses were performed using GraphPad Prism 7 software (GraphPad Software). Data are presented as mean ± standard errors of the means (SEMs). Statistical significance between two groups was determined by unpaired t-test. Comparisons among multiple groups were performed using a two-way analysis of variance (ANOVA), with Bonferroni post hoc tests to determine p values.

Data and Resource Availability
All data generated or analyzed during this study are included in this article or are available from the corresponding authors on request. The raw RNA-seq data have been deposited in Gene Expression Omnibus (GEO) (GSE 204794).

Macrophage Expression of Cx43 Was Upregulated in Gonadal White Adipose Tissue of HFD-Fed Mice
We examined Cx43 expression levels in gonadal white adipose tissue (gWAT) of mice fed a HFD for 8 weeks. Western blot FIGURE 6 | Macrophage-specific Cx43 KO protected mice from HFD induced metabolic dysfunction (A) Immunoblot analysis of gonadal white adipose tissue (gWAT) of WT and Cx43 MKO mice fed with normal chow diet (NCD) or high-fat diet (HFD) for 8 weeks (n = 6). Significant effects of genotype (NLRP3: p < 0.0001, Caspase1 p10: p < 0.0001, ASC: p < 0.0001) and significant effects of diet (NLRP3: p < 0.0001, Caspase1 p10: p = 0.0001, ASC: p = 0.0006) were observed. Significant differences between WT HFD and KO HFD groups were determined by Bonferroni post hoc tests (B) Flow cytometric analysis of SVCs obtained from gWAT of WT and Cx43 MKO mice fed with NCD or HFD for 8 weeks (n = 3). Significant effects of genotype (p < 0.0001) and diet (p < 0.0001) were observed. Significant differences between WT HFD and KO HFD groups were determined by Bonferroni post hoc tests (C) Glucose tolerance and insulin tolerance tests of WT and Cx43 MKO mice fed with HFD for 8 weeks (n = 6). Two-way ANOVA with Bonferroni post hoc tests was used in A-B. Unpaired, two-tailed t-tests (**p < 0.01, ***p < 0.001) were used in C. Each point represents biological replicate. Data are presented as mean ± S.E.M.

Macrophages Co-cultured With Dying Adipocytes Upregulated Cx43 Expression
HFD-induced adipose tissue remodeling is characterized by adipocyte hypertrophy, adipocyte death, and macrophages surrounding damaged/dying adipocytes (Strissel et al., 2007). To identify the molecular features of adipose tissue macrophages involved in HFD-induced adipose tissue remodeling, we established in vitro system by co-culturing macrophages with dying adipocytes (dAC) (Cho et al., 2019). As previously described , dying adipocytes were obtained from prolonged cultures of adipocytes differentiated from C3H10T1/2 cells. MACS-isolated adipose tissue macrophages were cultured with the dying adipocytes (Anand et al., 2008). We confirmed the upregulation of Cx43 expression in macrophages using immunoblotting, quantitative PCR, and immunostaining analysis (Figures 2A-C).

Cx43 Knockdown Reduced Inflammatory Response of Macrophages Co-cultured With Dying Adipocytes
Next, we investigated the effects of Cx43 knockdown in macrophages co-cultured with dying adipocytes. We performed Gja1 (encoding Cx43) siRNA knockdown (KD) in RAW264.7 cells. For visualization, we labeled adipocytes with BODIPY and macrophages with DIO dye . Using this co-culture study, we revealed that Gja1KD increased the clearance of dying adipocytes in macrophages ( Figure 3A). Interestingly, Gja1KD decreased the expression of pro-inflammatory markers and increased the expression of genes involved in anti-inflammatory responses (M2 macrophage markers) ( Figure 3B). Previous studies have suggested that Cx43 plays a critical role in the pro-inflammatory activation of macrophages by facilitating the release of ATP and regulating extracellular ATP signaling. Thus, we examined ATP levels in LPS-treated Gja1KD macrophages and found that Gja1KD reduced extracellular ATP levels ( Figure 3C). Treatment with apyrase decreased the expression levels of pro-inflammatory cytokines by degrading extracellular ATP ( Figure 3D). In addition, we tested the effects of conditioned media obtained from control RAW264.7 cells with dAC ( Figure 3E). Data indicated that the conditioned media were sufficient to increase pro-inflammatory cytokines in Gja1KD RAW264.7 cells ( Figure 3F), indicating that ATP released through Cx43 hemichannels in the media was a critical signal to induce inflammatory responses.

Global Transcriptomic Analysis Indicated That Macrophage-specific Cx43 KO Reduced Macrophage Recruitment and Inflammation in Adipose Tissue of HFD-Fed Mice
To investigate the role of Cx43 expression in vivo, we used Cx43-MKO mice. Cx43 floxed mice were crossed with Csf1r_CreER mice to delete Cx43 expression in a macrophage-specific manner. After tamoxifen induction, macrophage-specific KO was confirmed by qPCR analysis of MACS-isolated macrophages from the adipose tissue ( Figure 4A).
We next aimed to characterize the effects of Cx43-MKO on the molecular phenotype of adipose tissue in an unbiased manner. Accordingly, we performed RNA-seq analysis of gWAT from wild-type (WT) and Cx43-MKO mice fed a HFD or normal chow diet (NCD). Principal component analysis (PCA) based on the expression levels of 9,895 variable genes indicated that the WT-HFD group was characterized by a transcriptional pattern distinct from the other groups (WT-NCD, KO-NCD, and KO-HFD) ( Figure 4B). On listing genes ordered according to their contribution levels to PC1 was subjected to Gene Set Enrichment Analysis (GSEA), eight out of 24 hallmark gene sets were significantly affected (Supplementary Table S3). In HFD-fed WT mice, the hallmark inflammatory response was one of the most significantly enriched pathways (p < 0.01) in gWAT ( Figure 4C). The average expression of genes associated with the inflammatory response was positively regulated in the WT-HFD group; this was not observed in the other groups (Supplementary Figure S1). In addition, PC1 showed a high correlation with the average expression of genes associated with the inflammatory response, thereby confirming that it is a critical biological process differentiating the WT-HFD group from other groups (Supplementary Figure S1). The heatmap indicated that among these 106 genes, 83 genes, including Adgre1, P2rx7, and P2rx4, were highly expressed in the WT-HFD group, while 23 genes were downregulated ( Figure 4D). In the DEG analysis, HFD upregulated 821 genes and downregulated 985 genes in WT mice but significantly affected only 130 genes in KO mice, thus indicating that Cx43-MKO counteracted the effect of HFD (Supplementary Figures  S2A, S2D).
In addition, we analyzed differentially expressed genes between the WT-HFD and KO-HFD groups to determine the effect of Cx43 KO. We found that 475 genes were upregulated, whereas 426 genes were downregulated ( Figure 4E). In the KO-HFD group, pathways associated with lipid metabolic (catabolic) processes were enriched ( Figure 4F, Supplementary Figure S2C). Pathways associated with leukocyte activation and chemotaxis were enriched in the WT-HFD group ( Figure 4G, Supplementary Figure S2C). We further validated the RNA-seq data at transcript and protein levels using qPCR, immunoblotting, and immunohistochemical analyses. Based on qPCR analysis, HFD feeding upregulated P2RX7 and P2RX4 expression levels in the gWAT of WT mice, but not in the gWAT of Cx43-MKO mice ( Figure 5A). Histological analysis of H&Estained paraffin sections indicated that Cx43-MKO reduced the accumulation of crown-like structures in gWAT after HFD feeding ( Figure 5B). Western blot analysis further confirmed that Cx43-MKO reduced HFD-induced recruitment of F4/80+ macrophages ( Figure 5C). qPCR analysis confirmed the RNAseq data, presenting reduced mRNA expression of proinflammatory cytokines in Cx43-MKO mice under HFD when compared with WT-HFD-fed mice ( Figure 5A). It is well-established that P2RX7 signaling in macrophages facilitates the inflammasome signaling pathway. Consequently, inflammasome marker expression was decreased in the gWAT of MKO mice fed an HFD for 8 weeks ( Figure  6A). Moreover, FACS analysis indicated that pro-inflammatory macrophages (F4/ 80+CD11c+) were reduced in KO mice fed an HFD ( Figure 6B). Phenotypic analysis revealed that Cx43-MKO mice exhibit improved glucose tolerance and insulin sensitivity ( Figure 6C).

DISCUSSION
Obesity and related metabolic diseases can be characterized by a chronic inflammatory status (Reilly and Saltiel, 2017). Changes in macrophage phenotypes are correlated with the pathological remodeling of adipose tissue, thereby contributing to the development of insulin resistance and obesity-related metabolic dysfunction (Reilly and Saltiel, 2017).
Cx43 plays a key role in gap junction communication between cells (Evans and Martin, 2002). In addition to functioning as a gap junction channel, Cx43 hemichannels can mediate ATP release from cells and consequently regulate the autocrine activation of macrophages via ATP signaling (Dosch et al., 2019). Consistently, our data indicated that macrophage expression of Cx43 is critical during extracellular ATP signaling for proinflammatory macrophages activation in adipose tissue during HFD feeding.
Interestingly, we observed that macrophage-specific deletion of Cx43 protected mice from HFD-induced inflammation, consequently ameliorating glucose tolerance and insulin sensitivity. It is well-known that inflammatory responses in adipose tissue are one of the primary factors that contribute to the development of over-nutrition-induced metabolic dysfunction. In this regard, adipose tissue macrophages have been characterized as the major cell types responsible for pro-inflammatory cytokine production and inflammation. For instance, in patients with obesity, hypertrophic adipose tissue is frequently associated with pro-inflammatory macrophages and crown-like structures. Previous studies using diet-induced obesity mouse models reported that gWAT manifests more prominent macrophage infiltration, compared to other depots at different anatomical locations, such as subcutaneous and mesenteric WAT (van Beek et al., 2015). Thus, in this study, we focused on gWAT to investigate macrophagespecific roles of Cx43 during the development of obesity in mice.
We hypothesized that Cx43-mediated ATP release and the purinergic receptor P2RX7 signaling pathway in macrophages are factors that facilitate over-nutrition-induced inflammatory responses. This observation further suggests that pharmacological inhibition of Cx43 could afford a novel therapeutic strategy for obesity-associated inflammation and resistance. However, although the current study focused on the roles of macrophage-expressed Cx43, it is crucial to note that Cx43 activation in adipocytes mediates beneficial effects by facilitating cAMP coupling between adipocytes to increase protein kinase A (PKA)-signaling-mediated lipid catabolism and energy expenditure (Zhu et al., 2016). Therefore, the cell type-specific delivery of Cx43 modulators is required to develop novel therapies to treat obesity-related metabolic diseases.
Herein, our findings suggest that Cx43 KO inhibits the P2RX7-mediated activation of macrophages. Although not examined in the current study, extracellular ATP signaling is also known to be involved in regulating adipocyte function. For example, P2X7 receptor downstream signaling regulates lipid metabolism and adipogenesis, and P2RX7 KO affects fat distribution in vivo (Beaucage et al., 2014). In the present study, we could not precisely determine the cell type-specific contribution of Cx43/ATP/P2RX7 purinergic signaling. Further studies using single-cell level analysis are necessary to comprehensively clarify the effects of macrophage-specific Cx43 KO on other cell types, including adipocytes, through paracrine mechanisms.
Collectively, our study demonstrates the critical role of Cx43 in the pro-inflammatory activation of macrophages during HFDinduced adipose tissue remodeling in mice. Mechanistically, Cx43-mediated ATP release could induce autocrine macrophage activation, potentially through P2RX7. Identifying molecular players in over-nutrition-induced macrophage activation is critical for understanding adipose tissue inflammation and insulin resistance.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: NCBI GEO, accession no: GSE204794.