Dysregulated lipolysis and lipophagy in lipid droplets of macrophages from high fat diet‐fed obese mice

Abstract Obesity is associated with lipid droplet (LD) accumulation, dysregulated lipolysis and chronic inflammation. Previously, the caspase recruitment domain‐containing protein 9 (CARD9) has been identified as a potential contributor to obesity‐associated abnormalities including cardiac dysfunction. In the current study, we explored a positive feedback signalling cycle of dysregulated lipolysis, CARD9‐associated inflammation, impaired lipophagy and excessive LD accumulation in sustaining the chronic inflammation associated with obesity. C57BL/6 WT and CARD9−/− mice were fed with normal diet (ND, 12% fat) or a high fat diet (HFD, 45% fat) for 5 months. Staining of LDs from peritoneal macrophages (PMs) revealed a significant increase in the number of cells with LD and the number of LD per cell in the HFD‐fed WT but not CARD9−/− obese mice. Rather, CARD9 KO significantly increased the mean LD size. WT obese mice showed down regulation of lipolytic proteins with increased diacylglycerol (DAG) content, and CARD9 KO normalized DAG with restored lipolytic protein expression. The build‐up of DAG in the WT obese mice is further associated with activation of PKCδ, NF‐κB and p38 MAPK inflammatory signalling in a CARDD9‐dependent manner. Inhibition of adipose triglyceride lipase (ATGL) by Atglistatin (Atg) resulted in similar effects as in CARD9−/− mice. Interestingly, CARD9 KO and Atg treatment enhanced lipophagy. In conclusion, HFD feeding likely initiated a positive feedback signalling loop from dysregulated lipolysis, CARD9‐dependent inflammation, impaired lipophagy, to excessive LD accumulation and sustained inflammation. CARD9 KO and Atg treatment protected against the chronic inflammation by interrupting this feedforward cycle.


| INTRODUC TI ON
Lipid droplets (LDs) are dynamic cytoplasmic organelles present in almost all cell types as a storage of neutral lipids. 1 Besides its versatile roles in many biological processes, LD hypertrophy and hyperplasia are known key pathological features of obesity. [2][3][4] Obesity is characterised by chronic low-grade inflammation that leads to pro-inflammatory conditions and metabolic diseases, such as insulin resistance, type II diabetes, atherosclerosis and cardiovascular disorders. [5][6][7] In the progression of this chronic inflammatory condition, macrophages are the predominant cell type infiltrated into target tissues to release pro-inflammatory mediators. 5,8,9 In addition to adipocytes, macrophages also are known to accumulate excessive LDs in obesity, due to an imbalance between lipid synthesis and catabolism. 9,10 Lipid catabolism is mediated through both the canonical lipolytic pathway and lipophagy. 2,11 The classical lipolysis is a complex process that generates free fatty acids (FFA) from LD storage. 4 Under basal conditions, adipose triglyceride lipase (ATGL) associates with LDs and hydrolyzes triacylglycerol (TAG) to diacylglycerol (DAG) and FFA, the so called basal lipolysis. [12][13][14][15] Activated by hormones, cAMP-dependent protein kinase A (PKA) phosphorylates hormone-sensitive lipase (HSL), which degrades DAG to monoacylglycerol (MAG) and FFA in a process known as stimulated lipolysis. 2,14 Monoacylglycerol lipase (MAGL) then converts MAG to glycerol and FFA. 16 Both basal and stimulated lipolysis are highly regulated by LD-associated proteins such as the perilipin (PLIN) family proteins. 17 Among them, PLIN1 is an essential LD coating protein that prevents the basal lipolysis by restricting the access of ATGL to LDs. 17,18 The phosphorylated PLIN1, on the contrary, is required for the stimulated lipolysis, that phosphorylation of HSL is critically dependent on p-PLIN1 for activity. 19 Studies have shown that high fat diet (HFD)-induced obesity is linked with DAG accumulation in cells such as adipocytes and hepatocytes. 4,20,21 However, whether the dysregulated lipolysis with associated DAG build-up plays a mechanistic role in the development of chronic inflammation in obesity is not fully understood. To determine if DAG accumulation drives the sustained inflammation and its downstream pathological effects, we tested Atglistatin (Atg), an ATGL inhibitor, to deplete the accumulated DAG and suppress its downstream inflammatory responses.
Autophagy is a cellular homeostasis process where cytosolic components such as proteins and damaged organelles are sequestered in double membrane-bound autophagosomes and fused with lysosome for degradation. 22 Recently, a macro-autophagy (hereafter referred to as autophagy) has been demonstrated as a novel mechanism in lipid metabolism, named lipophagy. 23 Lipophagy regulates LDs turnover and cells with defect in this process accumulate LDs. 24,25 Impaired autophagy in macrophages was associated with chronic inflammation in HFD-induced obesity, 26,27 suggesting a correlation between LD accumulation and chronic inflammation.
Further studies have shown that lipophagy functions in synergy with classical lipolysis in LD catabolism where lipolysis breaks down large LDs into small ones that are manageable for degradation by lipophagy. 11 This homeostatic LD metabolism process is perturbed in obesity, leading potentially to a cycle of dysregulated lipolysis, impaired lipophagy and exacerbated LD accumulation, which may result in sustained chronic inflammation. To test this hypothesis, it is critical to probe the specific signalling pathways involved in this cycle of LD accumulation and inflammation.
Caspase recruitment domain-containing protein 9 (CARD9) is an adaptor protein predominantly expressed in myeloid cells with an essential role in regulating innate and adaptive immune response. [28][29][30] Activation of CARD9 signalling complex by PKCδ leads to up-regulation of transcriptional factors p38 MAPK and NF-κB and production of pro-inflammatory cytokines such as IL-6, TNFα and IL-1β. 8,30,31 Previously, we have shown that CARD9 was up-regulated in peritoneal macrophages (PMs) of HFD-fed obese mice. 8 Moreover, fasting blood glucose level and insulin concentration were markedly elevated accompanied by impaired glucose disposal and insulin sensitivity in HFD-fed WT mice. 8 CARD9 KO significantly improved insulin sensitivity and glucose tolerance, suppressed plasma-and PM-derived cytokines and protected against myocardial dysfunction. In the current study, we hypothesize that HFD feeding induces dysregulated lipolysis in PMs that in turn activates DAG-PKCδ and CARD9-dependent inflammation. This further impairs lipophagy leading to a positive signalling cycle of LD accumulation and sustained chronic inflammation. We will test that CARD9 KO interrupts this positive feedback loop and suppresses inflammation associated with HFD-induced obesity.

| Experimental animals and HFD feeding regimen
C57BL/6 wild-type (WT) and CARD9 −/− mice were fed with either normal rodent chow having 12% kcal as fat (normal diet, ND) or a HFD having 45% kcal as fat (D12451, Research Diets, NJ), starting from 4 to 6 weeks of age. Body weight was monitored over the

| Atglistatin treatment
Atglistatin (Atg, Sigma-Aldrich) was suspended in 0.5% DMSO in PBS as described previously. 32 After 5-months ND and HFD feeding, mice were randomly assigned into: (1) Atg treatment group with i.p. injection at a dose of 100 μmol/kg body weight; (2) vehicle (Veh) control group with i.p. injection of 0.5% DMSO in PBS using the same volume; (3) control group without any treatment. Both Atg and veh treatment were performed 8 h before PMs harvesting. 32

| Macrophage isolation and quantification
Peritoneal macrophages were harvested according to the standard protocols described previously. 8,33,34 Briefly, 2 ml of 4% (wt/vol) thioglycollate (TG) broth (Sigma-Aldrich), an eliciting agent, was injected i.p. to each mouse at the end of 5-months HFD feeding. After 4 days of injection and 24 h of fasting, mice were sacrificed and peritoneal lavage performed using 10 ml ice-cold DMEM/F-12 media with 10% (vol/vol) heat-inactivated foetal bovine serum and 1% (vol/vol) penicillin-streptomycin (Life Technologies). Ammonium-chloridepotassium lysing buffer (ACK) (Quality Biological) was used for lysing any red blood cell residuals present in the samples. To quantitate the number of macrophages in the peritoneal fluids, the total leukocyte count from the hemacytometer was multiplied by the percent of macrophages determined from differential cell count. 33

| Oil Red O (ORO) staining of LD
Once harvested, a total of 1-2 × 10 6 cells in a complete DMEM/F12 medium were added to 35-mm glass bottom cell culture dishes to form a monolayer. 33

| Western immunoblotting analysis
The isolated macrophages were lysed in ice-cold RIPA buffer. The SC-374665). After overnight incubation, membranes were probed with horseradish peroxidase-conjugated secondary antibody (CST, #7074) for 1 h at room temperature. Signals were then detected with chemiluminescence reagent (LumiGlo; CST, #7003). Images were taken using ChemiDoc XRS imager (Bio-Rad). Band intensity was determined by densitometry analysis using Quantity One® software and normalised using GAPDH as the loading control.

| Measurement of diacylglycerol (DAG) and triacylglycerol (TAG) content
Concentrations of DAG and TAG were measured using DAG ELISA assay kit (#CEC038Ge, Cloud-Clone Corp.) and picoProbe TAG assay kit (#ab178780, Abcam) according to manufacturer's instruction. Briefly, isolated macrophages were re-suspended in fresh lysis buffer (# IS007, Cloud-Clone Corp.) for DAG extraction or in triglyceride assay buffer for TAG extraction followed by ultra-sonication until the solution was clear. DAG and TAG concentrations were determined from the supernatant of each sample after generating a standard curve using a microplate reader (SpectraMax 190, Molecular Devices). Results were normalised to the total protein concentration.

| Immunofluorescence staining
The isolated PMs were cultured to obtain monolayer of cells as described previously. 33,35 This was followed by fixation with 4% paraformaldehyde for 20 min and blocking with a buffer containing PBS, 5% normal goat serum (#5425, CST), and 0.3% Triton™ X-100 (#T8787, Sigma-Aldrich) for 1 h. Then, the cells were incubated with primary antibody, LC3B (1:200; #3868, CST) diluted in PBS containing 1% BSA and 0.3% Triton X-100 for overnight at 4°C. After washing three times in PBS, the samples were incubated with Alexa Flour 555 secondary antibody (#4413, CST) and BODIPY 493/503 (#D3922, Invitrogen) for 1 h followed by counterstaining with DAPI for 10 min. BODIPY 493/503 was used to stain LDs in green. Finally, Z-stack images were taken using Zeiss LSM 980 confocal microscope. Images were analysed for co-localization using ImageJ software (NIH) with at least 80 cells per group.

| Statistical analysis
Statistical analysis was performed using GraphPad Prism Software (Version 8.0) by One-Way ANOVA followed by a Tukey test for post hoc analysis where appropriate. When comparing the means of two groups, unpaired t-test was used. All measurements were repeated three times with at least three mice per group, and normality of data distribution was checked by Shapiro-Wilk test. For co-localization study, Pearson's correlation coefficient was applied. Data were presented as mean ± SEM. Significant difference was determined at p < 0.05.

| HFD feeding increased the numbers of LDs and PMs with LD in a CARD9-dependent manner
As shown in Figure

| HFD feeding dysregulated LD lipolysis in PMs in a CARD9-dependent manner
As lipolysis is essential to the dynamics of LD accumulation and CARD9 KO ameliorated the reduction of p-PLIN1 ( Figure 2B), but significantly increased p-HSL ( Figure 2D). HFD feeding also markedly reduced PLIN1 in both HFD-fed WT and CARD9 −/− mice compared with their respective ND-fed controls ( Figure 2C). There was no significant difference in ATGL level among all the experimental groups ( Figure 2E). Taken together, these results suggest that the basal lipolysis (depending on PLIN1) is enhanced in both HFD-fed WT and CARD9 −/− mice but the stimulated lipolysis (depending on p-PLIN1) is only down regulated in the HFD-fed WT mice. CARD9 KO restored the stimulated lipolysis.

| HFD feeding increased DAG, but CARD9 KO and Atglistatin treatment ameliorated the increase of DAG
Next, we examined if the reduction in PLIN1 (suggesting an increase from TAG to DAG), p-PLIN1 and p-HSL (suggesting a decrease from DAG to MAG) observed in the HFD-fed WT mice leads to DAG accumulation. As shown in Figure

| HFD feeding induced but CARD9 KO and Atg treatment ameliorated the activation of PKCδ , NF-kB and p38 MAPK
Next, we tested if the dysregulated lipolysis and the increased DAG accumulation in the HFD-fed WT mice activates PKCδ which further activates CARD9-dependent NF-κB and p38 MAPK inflammatory pathways as suggested. 28,30,36 Western immunoblotting analysis was performed on protein expression of PKCδ, p-PKCδ, p38 MAPK, p-p38 MAPK, p-p65 and p65 with representative blots shown in Figure 4A,F. HFD feeding markedly increased the ratio of p-PKCδ/ PKCδ by 1.4-fold in the WT mice compared with the ND-fed control group ( Figure 4B, p < 0.05), with no significant difference among the CARD9 −/− groups. HFD feeding also significantly increased the ratio of p-p65/p65 ( Figure 4C) and p-p38/p38 ( Figure 4D) by 1.5-(p < 0.05) and 1.6-fold (p < 0.01) respectively compared with their ND-fed groups. CARD9 KO ameliorated these changes. Further, Atg treatment suppressed the increase in p-PKCδ/ PKCδ, p-p65/ p65 and p-p38/p38 MAPK in the HFD-fed WT mice ( Figure 4G-I).
These data indicated that the dysregulated lipolysis in the HFD-fed WT mice activated the DAG-PKCδ pathway which further activated the CARD9 inflammatory signalling. CARD9 KO and Atg treatment ameliorated these inflammatory responses.

| HFD feeding reduced but CARD9 KO and Atg treatment restored p38IP
It was reported that p-p38 MAPK interacts with the p38-interacting protein (p38IP) resulting in suppression of p38IP-initiated autophagosome formation, an alternative autophagy pathway. 37 To test if the activated CARD9 signalling inhibits autophagy/lipophagy, Western immunoblotting analysis was performed on protein expression of p38IP. As shown in Figure 4E, p38IP was significantly down regulated by 0.8-fold (p < 0.05) in the HFD-fed WT mice compared with the ND-fed WT mice. CARD9 KO markedly increased the level of p38IP by 1.6-fold (p < 0.0001) while Atg treatment restored p38IP expression ( Figure 4J), suggesting protected autophagosome formation and potentially lipophagy.

| HFD feeding impaired, CARD9 KO and Atg treatment restored autophagy signalling
As defects in macrophage autophagy were associated with chronic inflammation in obesity, 26 Figure 5A,E. HFD feeding significantly reduced LC3BII/I ratio by 26% ( Figure 5B, p < 0.05) and Beclin-1 by 46% ( Figure 5D, p < 0.05) in the HFD-fed WT mice compared with the ND-fed WT mice with no significant change to p62 ( Figure 5C). CARD9 KO restored LC3BII/I ( Figure 5B), Beclin-1 ( Figure 5D) and reduced p62 ( Figure 5C). Interestingly, Atg treatment markedly increased LC3BII/I and Beclin-1 in the HFD-fed WT mice compared with the ND-fed WT mice without significant change to p62 ( Figure 5F-H). These results indicated that the activated DAG-PKCδ and CARD9 signalling resulted in altered autophagy signalling in the HFD-fed WT mice, and CARD9 or Atg treatment restored autophagy.

| HFD feeding impaired but CARD9 KO and Atg treatment restored lipophagy-mediated degradation of LDs
Next, we tested if the autophagic protein LC3B overlaps with LDs to serve as a lipophagic signalling. PMs were harvested for immunofluorescence staining where anti-LC3B antibody and BODIPY were used to stain autophagosome and LDs, respectively. Representative confocal microscopy images are presented in Figure 6A. As shown in Figure 6B,C, HFD feeding markedly increased the number of LD per cell without significant change to the number of LC3B + autophagy puncta per cell in the HFD-fed WT mice compared with the ND-fed WT mice. The HFD-fed CARD9 −/− mice also exhibited a slight but significant increase in the number of LD per cell compared with the respective ND-fed mice, similar to ORO staining results ( Figure 6B vs. Figure 1C). The number of LC3B + puncta per cell was markedly increased by about 2-fold (p < 0.01) in the HFDfed CARD9 −/− mice compared with the respective ND-fed mice ( Figure 6C). The total area of LC3B + staining per cell, corresponding to the total LC3B expression, was also markedly increased by about 2-fold in the HFD-fed CARD9 −/− mice compared with the respective ND-fed group ( Figure 6D, p < 0.05), with no significant difference among the WT groups. The incidence of LC3B co-localization with LDs was evaluated by Pearson's correlation analysis. As shown in Figure 6E, the HFD-fed CARD9 −/− mice exhibited a 1.7-fold increase in the correlation coefficient compared with the respective ND-fed group (p < 0.01), with no significant difference among the WT groups.
To assess if Atg treatment enhances lipophagy, WT mice were treated with Atglistatin as in the Material and Methods section. Representative confocal microscopy images are presented in Figure 7A. As shown in Figure 7B, Atg treatment did not affect the number of LD per cell. However, Atg treatment significantly increased  Figure 7D). The number of LDs per cell was still significantly higher in the treated HFD-fed WT mice compared with the ND-fed WT mice, but was significantly reduced from that of the untreated HFDfed WT mice as in Figure 6B. Altogether, these data indicated that CARD9 KO and Atg treatment increased the number of LC3B + autophagosome, enhanced lipophagy therefore lipophagy-mediated LD degradation.

| DISCUSS ION
Obesity is hallmarked with increased lipid accumulation in various tissues. 38,39 It is associated with a constellation of metabolic disorders with chronic inflammation. 25,40 Our previous study have shown that CARD9 deficiency attenuated HFD-induced insulin resistance, glucose intolerance and PM-derived inflammatory cytokines. 8 Here, we report that HFD feeding dysregulated lipolysis and activated DAG-PKCδ signalling which further activated CARD9-dependent There are several key findings in the current study. First, CARD9 KO reduced the number of PMs harbouring LDs and the number of LDs in PMs following HFD-feeding. Interestingly, these LDcontaining PM cells from the HFD-fed CARD9 −/− mice showed larger LD with mean droplet size up to 6 μm 2 . A previous report suggested that the size of LD is an essential physical characteristic that governs the process of LD metabolism following diet-induced obesity in hepatocytes. 11 The authors were able to show that while the lipolytic enzymes targeted large LDs, the lipophagic machinery could only encompass the small ones preferentially for degradation. Another study also suggested that the size of a cargo destined for catabolism through autophagy may be a crucial factor governing degradation by the autophagic machinery. 41 The size-dependent LD catabolism might explain our observation of larger LDs in the HFD-fed CARD9 −/− mice despite improved lipolysis and enhanced lipophagy, as also suggested previously. 42,43 Secondly, CARD9 KO and Atg treatment restored the dysregulated lipolysis following HFD-feeding. Previously, it was demonstrated that two-months HFD feeding increased basal lipolysis and reduced stimulated lipolysis in adipocytes. 4 The lipolytc proteins p-HSL and PLIN1 were found to be compromised, leading to DAG build-up. 44 47,48 but hepatocytes showed the opposite regulation. 23 It was reported that HFD-induced obesity is linked with reduced autophagy in PMs resulting in pro-inflammatory polarization. 27 Our data on the co-localization of LC3B + autophagosome with LD in CARD9 KO or Atg treatment mice suggested a potential mechanistic link between obesity-induced inflammation and defective autophagy. Previous studies have shown that the phosphorylated p38 MAPK interacts with p38IP, preventing its association with Atg9 and negatively impairing autophagy. Based on our observation on the decrease of p-p38/p38 and increase of p38IP in the HFDfed CARD9 −/− mice, we speculate that p38IP, a potential partner to Atg9 trafficking and autophagosome formation, as an alternative autophagy pathway, may have mediated the CARD9-dependent impairment of lipophagy. 37 Future mechanistic studies are warranted to fully understand how CARD9 mediates the interaction between p38IP and Atg9.
In conclusion, our study demonstrated that HFD-induced obe- writing -review and editing (lead).

ACK N OWLED G EM ENT
This study was supported by Grants from American Heart Association (18TPA34170232) and National Institute of Health (P20GM103432).

CO N FLI C T O F I NTE R E S T
The authors declared that they have no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data sets supporting the findings of the current study are available from the corresponding author upon reasonable request.