FGF21 induces autophagy‐mediated cholesterol efflux to inhibit atherogenesis via RACK1 up‐regulation

Abstract Fibroblast growth factor 21 (FGF21) acts as an anti‐atherosclerotic agent. However, the specific mechanisms governing this regulatory activity are unclear. Autophagy is a highly conserved cell stress response which regulates atherosclerosis (AS) by reducing lipid droplet degradation in foam cells. We sought to assess whether FGF21 could inhibit AS by regulating cholesterol metabolism in foam cells via autophagy and to elucidate the underlying molecular mechanisms. In this study, ApoE−/− mice were fed a high‐fat diet (HFD) with or without FGF21 and FGF21 + 3‐Methyladenine (3MA) for 12 weeks. Our results showed that FGF21 inhibited AS in HFD‐fed ApoE−/− mice, which was reversed by 3MA treatment. Moreover, FGF21 increased plaque RACK1 and autophagy‐related protein (LC3 and beclin‐1) expression in ApoE−/− mice, thus preventing AS. However, these proteins were inhibited by LV‐RACK1 shRNA injection. Foam cell development is a crucial determinant of AS, and cholesterol efflux from foam cells represents an important defensive measure of AS. In this study, foam cells were treated with FGF21 for 24 hours after a pre‐treatment with 3MA, ATG5 siRNA or RACK1 siRNA. Our results indicated that FGF21‐induced autophagy promoted cholesterol efflux to reduce cholesterol accumulation in foam cells by up‐regulating RACK1 expression. Interestingly, immunoprecipitation results showed that RACK1 was able to activate AMPK and interact with ATG5. Taken together, our results indicated that FGF21 induces autophagy to promote cholesterol efflux and reduce cholesterol accumulation in foam cells through RACK1‐mediated AMPK activation and ATG5 interaction. These results provided new insights into the molecular mechanisms of FGF21 in the treatment of AS.


| INTRODUC TI ON
Atherosclerosis (AS) is a chronic and degenerative disease in which lipids accumulate in the walls of large arteries. Macrophages promote foam cell development and plaque formation through the accumulation of cholesterol esters, which is the hallmark of AS. 1,2 Therefore, enhancing the clearance of cholesterol from macrophage foam cells, especially by increasing the cholesterol efflux from macrophages, is important to reduce plaque lipid build-up. Cholesterol efflux entails clearing cholesterol from peripheral macrophages, and its transformation into bile in the liver and ultimately into faeces, making it a promising anti-atherogenic strategy. 3 Recently, our group found that allicin and curcumin can induce foam cell cholesterol efflux. 4,5 FGF21 is also reported to be involved in cholesterol efflux. 6 However, the mechanisms underlying the effects of FGF21 on lipid metabolism in foam cells remain largely unclear.
As a metabolic regulator, FGF21 exhibits outstanding fat-regulating ability and anti-AS potential by increasing high-density lipoprotein (HDL) levels and reducing triglycerides, total cholesterol (TC) and low-density lipoprotein (LDL) levels in the serum. 7,8 Recent studies suggested that FGF21 can attenuate the oxidative stress protect against hydrogen peroxide-induced cytotoxicity by preventing mitogen-activated protein kinase (MAPK) activation within human umbilical vein endothelial cells (HUVECs). 9 FGF21 administration reverses oxidative stress induced by a high-fat diet (HFD) in rats and reduces circulating malondialdehyde, glutathione and superoxide dismutase. 10 Moreover, some studies demonstrated that FGF21 is anti-inflammatory. For instance, FGF21 inhibits lipopolysaccharide-induced NF-κB activation by promoting I-κB degradation and impairing p65 translocation to the nucleus. 11 FGF21 enhances AMPK activation and expression of Sirt1 in alcohol-treated HepG2 cells 12 and reduces the expression of C-reactive protein, tumour necrosis factor α and another inflammatory factor in macrophages. 7 Emerging evidence indicates that FGF21 reduces AS risk factors and inhibits certain AS-associated pathogenic mechanisms. [9][10][11]13 However, the mechanism behind the effect of FGF21 on macrophage foam cells, which are important for AS development, is poorly understood. Our previous study has reported that FGF21 can act on macrophage foam cells to reduce cholesterol accumulation. 6 Specifically, how FGF21 affects lipid metabolism in macrophage foam cells remain largely unknown. Therefore, an in-depth exploration of the underlying mechanisms of FGF21 is of great importance for the prevention of AS.
Autophagy is an essential process whereby cells are able to breakdown large portions of the cytoplasm and cellular organelles and recycle them via a catabolic and a highly conserved process. 14 Lipids are degraded to free cholesterol and fatty acids through autophagy. 15 Besides, Wang et al 16 have reported that inhibiting the activation of mTOR activates autophagy and drives cholesterol efflux, thereby decreasing foam cell lipid accumulation, while another study suggested that lysosomal acid lipase mediated autophagy-mediated regulation of cholesterol efflux. 17 Furthermore, mTOR inhibitor everolimus selectively depletes macrophages in atherosclerotic plaques by autophagy. 18 These findings suggest that induction of autophagy is an essential strategy to reduce foam cell lipid accumulation and the prevention of AS.
Activated kinase C receptor 1 (RACK1), a highly conserved intracellular adaptor protein, which was initially discovered to be integrated with Protein kinase C (PKC), is encoded by the GNB2L1 gene. 19 As an adaptor protein, RACK1 can provide a platform for the interaction with other proteins and regulate their activities. Studies reported that RACK1 can induce the formation of Atg14L-Beclin-1-Vps34-Vps15 20 or interact with ATG5 to promote RACK1-ATG5 complexes formation, 21 promote autophagy and inhibit hepatic lipid accumulation. In addition, RACK1 also induces the activation of AMPK to protect the liver from ischaemia-reperfusion (I/R) injury. 22 Therefore, we suggested that FGF21 may induce autophagy via RACK1 to enhance lipid degradation, increase cholesterol efflux and reduce lipid concentration within foam cells, thus inhibiting AS.
We observed that FGF21 inhibited AS significantly and that there was involvement of autophagy during this process. Moreover, RACK1 was highly expressed in the atherosclerotic plaque in ApoE −/− mice. Furthermore, the expression of RACK1 was up-regulated in the foam cells by FGF21. Consequently, autophagy was induced to drive cholesterol efflux and decrease cholesterol accumulation, thus preventing AS. Our findings provide novel information regarding the mechanisms of FGF21 and their role in the prevention and treatment of AS.

| Animal model
All animals were killed using carbon dioxide. No anti-aesthetic agent was used. Ethical guidelines of the Chinese Association for Laboratory Animal Sciences from Directive GB14925-2010 regarding animal use in scientific studies were followed. proved these animal studies. ApoE −/− mice were housed in standard cages with a 12-hour light/dark cycle in climate-controlled conditions. The ApoE −/− mice were randomized into 6 groups (n = 10/ group). Control, ApoE −/− mice fed with a normal diet; HFD, ApoE −/− mice fed with a HFD for 12 weeks; HFD + NS, ApoE −/− mice fed a HFD with an intraperitoneal (IP) saline for 12 weeks; FGF21 + HFD, ApoE −/− mice fed a HFD with 10 mg/kg/d of FGF21 for 12 weeks; HFD + 3MA, ApoE −/− mice fed a HFD with an IP injection of 3MA (30 mg/kg) for 12 weeks; FGF21 + HFD + 3MA or LV-RACK neg, ApoE −/− mice fed a HFD with IP injection of 3MA (30 mg/kg) or LV-RACK shRNA + FGF21 (10 mg/kg/d) for 12 weeks. Animal body weights were recorded weekly, and after 12 weeks of daily injections, animals were killed using carbon dioxide, and blood samples were collected from the ophthalmic venous plexus. The concentrations of serum lipids were determined using instructions provided with the commercial detection kit. The mouse aorta was examined for the presence of atherosclerotic lesions.

| Cell culture
THP-1 cells were grown in RPMI 1640 containing 10% FBS and penicillin/streptomycin in an incubator as described above. Macrophage differentiation was achieved by adding 160 nm/mL PMA to the culture and incubating it for 24 hours. Subsequently, cells were washed with PBS and incubated with a serum-free medium supplemented using 50 µg/mL ac-LDL for 48 hours, in order to generate AS-associated macrophage foam cell model. To elucidate the mechanisms underlying the effects of FGF21, this cell model was treated with ATG5 siRNA, 3MA (5 mmol/L) or RACK1 siRNA prior to FGF21 treatment.

| Evaluation of foam cell formation
After removing the supernatants, cells were washed thrice using PBS followed by a 10-minute fixation step using 4% paraformaldehyde. After an additional wash, cells were placed in 60% isopropanol for 4 minutes. Cells were then stained using Oil Red O solution for 15 minutes that had just been filtered. After washing with 60% isopropanol, counterstaining was done for 40 seconds using haematoxylin. Finally, the stained cells were observed under a Leica phase-contrast microscope (DMI4000B, Wetzlar, Germany).

| Assessment of cellular cholesterol efflux
Cholesterol efflux was assessed based on methods reported previously. 6 Briefly, 0.2 µCi/mL of [ 3 H]-cholesterol was used for cell labelling in a black microtitre plate and incubated at 37°C overnight.
After 24 hours of incubation, cells were washed and resuspended using RPMI supplemented with 0.1% bovine serum albumin (BSA) overnight to equilibrate the cholesterol content. We then measured [ 3 H]-cholesterol in both the media and in the cells via liquid scintillation counting, with efflux being measured based on the formula: 100% × [total media counts/(total cell counts + total media counts)].

| Western blotting
RIPA lysis buffer was used to isolate cellular proteins at 4°C.
The concentration of total protein was detected using a BCA kit (Beyotime Biotech). Protein samples were then separated using 10% or 12% SDS-PAGE gels, followed by transfer to PVDF membranes (Millipore). Next, TBST (Tris-buffered saline, 0.1% Tween 20) containing 5% non-fat milk powder was used for blocking for 2 hours, and then, membranes were incubated overnight at 4°C with anti-

| HPLC assay
HPLC analysis was carried out as described previously. 6 PerkinElmer TotalChrom software was used for data analysis.

| Morphological assessments
The aortae were removed from killed animals and were stained for 30 minutes with Oil Red O. Samples were then differentiated for 15 minutes in 70% alcohol, washed and then observed. In addition, haematoxylin and eosin (H&E) staining was performed using aortic tissue samples that had been fixed using 10% (w/v) neutral formalin for 24 hours or more. Aortic roots were then processed using standard protocols prior to H&E staining. After that, intimal and medial thicknesses were assessed via an image processing system with a 0.1 mm lens and 400× amplification.
The aorta was placed in 10% neutral buffered formalin overnight. The aortic arch was then opened lengthwise through the lesser curvature and pinned flat en face in a wax-bottomed dissecting pan. The tissue was stained for 15 minutes using 0.5% Sudan IV solution (ProSciTech, China) in acetone and 70% ethanol (1:1). The tissue was then placed in 80% ethanol for 5 minutes and washed gently with water for several minutes. Digital images of the stained samples were acquired and staining was quantified as a percentage of the total tissue area.

| Transmission electron microscopy (TEM)
Transmission electron microscopy was conducted as described in a previous study. 23 Briefly, 0.1 mol/L sodium cacodylate-buffered (pH 7.4) 2.5% glutaraldehyde solution was used to fix cells at 4°C for 2 hours, after which the cells were washed thrice using 0.1 mol/L sodium cacodylate-buffered (pH 7.4) 7.5% sucrose. Next, 1% OsO4 solution was added and the cells were kept for 1.5 hours. Samples were then dehydrated using an ethanol gradient, and the cells were embedded using EMbed 812 (EMS) and cut into ultrathin sections.
Next, 2% uranyl acetate was used to stain the cells for 12 minutes, followed by treatment with Reynolds solution (pH 12.4) for 8 minutes. Finally, the stained ultrathin sections were assessed using a FEI Tacna microscope at 120 kV.

| Plasma lipid profile evaluation
We used 3% isoflurane to anaesthetize the mice, and the plasma samples were collected from the retro-orbital plexus. Afterwards, the levels of blood lipids were measured enzymatically (Boster Biological Technology Co., Ltd.), as described previously. 9

| Construction of lentivirus vector
shRNA (GGATGAGACCAACTATGGA) against RACK1 was ligated into the lentivirus (GenePharma, Shanghai, China). The lentivirus with the negative sequence (GTCACTCACCCTTCGGTTATT), which did not target any gene, was used as a negative control. LV-RACK1 shRNA was diluted to a total volume of 300 μL containing 4 × 10 7 pfu and was injected into the tail vein of 8-week-old male ApoE −/− mice (once a week for 4 weeks). Control mice were injected with the same dose of empty vector of lentivirus (the efficiency of LV-RACK1 shRNA is shown in Figure S1C).

| FGF21 altered blood cholesterol levels and inhibited AS in ApoE −/− mice
Fibroblast growth factor 21 is known to alter lipid profiles as well as to inhibit AS in a few animal models. 7,24 To verify the reported data, the effects of FGF21 were examined on the lipid profiles of ApoE −/− mice fed with a HFD-the levels of TC, TG, HDL-C and LDL-C were measured in these animals (Table 1). Relative to controls, serum TG, TC and LDL-C remarkably increased, whereas HDL-C levels markedly decreased (P < .05) in the mice with atherosclerosis (HFD group) ( Table 1). In addition, body weight of mice in HFD group significantly increased relative to the control group. These results indicated that the mice with atherosclerotic lesions are often accompanied by disturbances in lipid metabolism.
Total cholesterol, LDL-C and TG levels markedly reduced in the FGF21 + HFD group, while HDL-C levels markedly increased (P < .05), relative to HFD group (Table 1). These findings suggested that FGF21 treatment improves lipid metabolism in the ApoE −/− mice fed with a HFD. Moreover, the atherosclerotic lesions in mice with HFD were found to be significantly ameliorated by FGF21 treatment, as observed using aortic H & E and Oil Red O staining ( Figure 1A,B).
Taken together, FGF21 can alter lipid profiles and inhibit AS development in ApoE −/− mice.

| Autophagy is involved in FGF21-mediated inhibition of AS
Previously, some studies have identified a role of autophagy in AS progression, and FGF21 induces autophagy to protect against myocardial injury in ischaemia-reperfusion 25 and diabetic cardiomyopathy. 26 However, whether autophagy is involved in FGF21-inhibited treatment. We found that beclin-1, p62 levels, LC3-II/LC3-I ratio and the AS region were apparently reversed in the FGF21 + HFD + 3MA group compared with FGF21 + HFD group ( Figure 2D-E). These findings revealed that FGF21 induced autophagy in order to inhibit AS.

| FGF21 up-regulated RACK1 expression to induce autophagy and inhibit AS
As a scaffold/adaptor protein, RACK1 can interact with ATG5 to induce autophagy. 21 RACK1 can induce the formation of Atg14L-Beclin-1-Vps34-Vps15 and RACK1-ATG5 complexes, promote autophagy and inhibit hepatic lipid accumulation. 20 In addition, RACK1 also induces the activation of AMPK to protect the liver from ischaemiareperfusion (I/R) injury. 22  To further determine whether RACK1-mediated FGF21 can affect autophagy-inhibited AS, Ad-RACK1 shRNA was transfected into ApoE −/− mice for the purpose of RACK1 silencing before FGF21 treatment. The results showed that the effects of FGF21 on LC3-I to LC3-II conversion, beclin-1, p62 expression ( Figure 4A-C) and AS ( Figure 4D,E) were reversed. Therefore, these findings suggested that FGF21 up-regulated RACK1 expression to promote autophagy-inhibited AS.

| FGF21 decreased foam cell lipid accumulation
The formation of foam cells is a major hallmark of early-stage AS and thus can be an appropriate target for therapy. 27 FGF21 is a metabolic modulator that regulates lipid and glucose metabolism in

| FGF21 induced autophagic flux to promote cholesterol efflux in foam cells
Recent studies have demonstrated that autophagy is critical for regulating lipid metabolism and is able to reduce lipid accumulation by promoting cholesterol efflux in the foam cells. 32,33 However, whether  and FGF21 + bafilomycin A groups ( Figure S2D). Furthermore, similar results were observed using monodansylcadaverine (MDC) staining and TEM analysis ( Figure S2E,F). Overall, these findings showed that FGF21 can induce autophagic flux in foam cells.

TC (mg/g) FC (mg/g) CE (mg/g) CE/TC (%)
Also, to investigate whether autophagy mediates FGF21-induced cholesterol efflux, the foam cells were treated using FGF21 + 3MA (autophagy inhibitor) or ATG5 siRNA (Table 4 and Figure 6). Pre-treatment with 3MA significantly inhibited the effects of FGF21 by increasing TC, FC and CE levels, cholesterol accumulation (Table 4; Figure 6A,B) and cholesterol efflux ( Figure 6C) in the foam cells. Transient ATG5 knockdown yielded similar results. Taken together, FGF21 promoted cholesterol efflux and reduced lipid accumulation in foam cells via autophagy. showing that FGF21 increased RACK1 expression levels ( Figure   S1D). Next, the foam cells were pre-treated with RACK1 siRNA.

| FGF21 induced autophagy to promote cholesterol efflux in foam cells via RACK1 upregulation
Our results demonstrated that RACK1 knockdown significantly suppressed FGF21-induced autophagy, decreased LC3-I to LC3-II conversion, up-regulated p62 expression ( Figure 7A-C) and inhibited autophagosome formation ( Figure 7D-F). Furthermore, FGF21 decreased the levels of TC, FC and CE, reduced cholesterol accumulation and increased cholesterol efflux in foam cells, but these effect also reversed by RACK1 knockdown (Table 5; Figure 7G,H). These data indicated that FGF21 induced autophagy to promote cholesterol efflux in foam cells via RACK1 up-regulation.
More recently, it was shown that AMPK pathway is activated by RACK1 to protect against hepatic I/R injury. 22 Therefore, we suggested that RACK1 may activate AMPK and interact with ATG5 to regulate FGF21-induced autophagy and promote foam cell cholesterol efflux. In the present study, RACK1 siRNA was used in foam cells, showing that knocking down RACK1 greatly reduced FGF21induced AMPK activation and ABCA1 expression ( Figure S3A-C).
Additionally, RACK1 knockdown disrupted the interaction between RACK1 and ATG5 ( Figure S3D,E). These data suggested that FGF21induced autophagy promoted cholesterol efflux via RACK1 up-regulation, in order to activate AMPK and interact with ATG5 in the foam cells. Furthermore, the activation of AMPK also up-regulated the expression levels of ABCA1.

| D ISCUSS I ON
Fibroblast growth factor 21 is a hormone-like growth factor that lowers the levels of glucose and lipids in non-human primate and rodent models. Administration of FGF21 has greatly improved the lipo-profiles, by decreasing LDL-C and elevating HDL-C in rodent models. 36 Recent in vitro and in vivo works suggest that FGF21 can inhibit AS by attenuating inflammation and inhibiting oxidative stress. 36,37 However, how FGF21 affects lipid metabolism in foam cells is still elusive. Our findings suggested that FGF21 may serve as an essential factor for inhibiting AS progression. More importantly, we found that induction of autophagy can be considered as a novel mechanism of FGF21 to reduce cholesterol accumulation in foam cells and prevent subsequent AS development.
Numerous studies have demonstrated that FGF21 is effective in ameliorating the metabolic syndrome, lowering LDL-C and improving other related diseases and conditions such as type 2 diabetes, non-alcoholic steatohepatitis, obesity and chronic inflammation. 36,38,39 We have found FGF21 to be able to cause a marked improvement in the lipo-profiles (reducing LDL-C and elevating HDL-C) and prevent AS in ApoE −/− mice. In this respect, the increased circulating FGF21 levels may serve as the body's defence mechanism to prevent vascular damage in patients and rodents Note: Foam cells were divided into five groups and cultured in medium containing 200 ng/mL FGF21 for indicated time. The levels of cellular total cholesterol (TC), free cholesterol (FC) and cholesterol ester (CE) were determined by HPLC. Data are expressed as the mean ± SD from three independent experiments, each of which was performed in triplicate.
*P < .05 vs control group.controlgroup:BSA group.  interaction to induce autophagy in macrophage-derived foam cells ( Figure S3). The involvement of AMPK in autophagy has been previously identified. 6 inhibiting AS progression ( Figure S4). The mechanism is as followson the one hand, RACK1 mediated FGF21-induced autophagic activity in macrophage foam cells via ATG5 interaction and, on the other hand, FGF21 increased RACK1 expression to induce AMPK activation and then promote ABCA1 expression ( Figure S4). Collectively, our results offered novel insights into the roles and mechanisms of FGF21 in the development of AS and suggested that RACK1 may be a therapeutic target for AS treatment.

| Inadequacies of the article
In the current study, we only explore the role of FGF21 in inhibiting AS induced by autophagy, but the mechanism of the FGF21regulated autophagy remains to be completely elucidated. On the other hand, FGF21 therapy is accompanied by lipid-lowering effects in non-human primates and reduced atherosclerotic plaque formation in mice. Due to its anti-oxidative, anti-inflammatory, lipidlowering and adiponectin-increasing effects, FGF21 directly or indirectly represses signalling pathways that lead to atherosclerosis. Note: The foam cells were transfected using scrambled (neg) or RACK1 siRNA for 12 h and then incubated with 200 ng/mL of FGF21 for 24 h, HPLC assay was subsequently performed. Data are expressed as the mean ± SD from three independent experiments, each of which was performed in triplicate.

TA B L E 5
The role of RACK1 in the effects of FGF21 on cholesterol content in foam cells However, a number of these studies were conducted in animals. As there is difference in atherosclerosis susceptibility between humans and rodent, further studies are needed to confirm the therapeutic role of FGF21 against atherosclerosis in humans or large humanoid animals such as pigs. Also, further prospective studies are needed to clarify whether FGF21 can be used as a predictive biomarker to identify individuals at high risk of atherosclerosis in atherosclerosisassociated diseases and whether FGF21 therapy can reduce the risk of atherosclerosis in these diseases.

ACK N OWLED G EM ENTS
The authors would like to express their gratitude towards EditSprings (https://www.edits prings.com/) for the expert linguistic services.

CO N FLI C T O F I NTE R E S T
The authors declare 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 that support the findings of this study are available from the corresponding author upon reasonable request.