Metformin reduces saturated fatty acid-induced lipid accumulation and inflammatory response by restoration of autophagic flux in endothelial cells

Autophagy, an integral part of the waste recycling process, plays an important role in cellular physiology and pathophysiology. Impaired autophagic flux causes ectopic lipid deposition, which is defined as the accumulation of lipids in non-adipose tissue. Ectopic lipid accumulation is observed in patients with cardiometabolic syndrome, including obesity, diabetes, insulin resistance, and cardiovascular complications. Metformin is the first line of treatment for type 2 diabetes, and one of the underlying mechanisms for the anti-diabetic effect of metformin is mediated by the stimulation of AMP-activated protein kinase (AMPK). Because the activation of AMPK is crucial for the initiation of autophagy, we hypothesize that metformin reduces the accumulation of lipid droplets by increasing autophagic flux in vascular endothelial cells. Incubation of vascular endothelial cells with saturated fatty acid (SFA) increased the accumulation of lipid droplets and impaired autophagic flux. We observed that the accumulation of lipid droplets was reduced, and the autophagic flux was enhanced by treatment with metformin. The knock-down of AMPKα by using siRNA blunted the effect of metformin. Furthermore, treatment with SFA or inhibition of autophagy increased leukocyte adhesion, whereas treatment with metformin decreased the SFA-induced leukocyte adhesion. The results suggest a novel mechanism by which metformin protects vascular endothelium from SFA-induced ectopic lipid accumulation and pro-inflammatory responses. In conclusion, improving autophagic flux may be a therapeutic strategy to protect endothelial function from dyslipidemia and diabetic complications.


Results
Saturated fatty acids, but not unsaturated fatty acids affect autophagy. SFA causes stressful conditions, including inflammation and endoplasmic reticulum (ER) stress 25,26 . Since autophagy is one of the stress-response processes, we examined whether SFA regulates autophagy in human endothelial cells. Palmitic acid (PA), the most abundant SFA in the serum, increased the accumulation of microtubule-associated protein light chain 3 (LC3-II, a membrane component of autophagosome) in a time dependent manner (Fig. 1A). The LC3-II protein abundance was increased approximately 5-fold (p < 0.001) within 8 h treatment with 200 µM PA. The accumulation of LC3-II reached the maximum at the concentration of 100 μM PA (Fig. 1B). Because the lipidation of LC3 (e.g. LC3-II formation) indicates the steady-state level of autophagosome 33 , the result indicates that the amount of autophagosome was increased by palmitate. To examine whether the accumulation of LC3-II is a SFA specific response, the cells were treated with lauric acid (C12), another SFA, or oleic acid (C18), a monounsaturated fatty acid. Accumulation of LC3-II was observed when the cells were treated with lauric acid, but not with oleic acid (Fig. 1C). This result suggests that SFAs (e.g. lauric acid and palmitic acid) but not unsaturated fatty acid (e.g. oleic acid) cause the accumulation of autophagosome.
Beta-oxidation regulates autophagic flux. Since lysosomal degradation is the final process of autophagy, we examined whether PA regulates autophagic flux (a net change between autophagosome formation and degradation). The autophagic flux was assessed by subtracting the ratio of LC3-II/LC3-I in the NH 4 Cl/ leupeptin (Leu)-untreated cells (e.g. steady-state LC3-II) from the ratio of LC3-II/LC3-I in the NH 4 Cl/Leutreated cells (the total amount formed during the given time). The amount of degraded LC3-II was much larger in the vehicle (BSA)-treated samples than in the PA-treated samples (22% of BSA-treated samples) ( Fig. 2A), indicating that the accumulated LC3-II was due to the impaired lysosomal degradation of LC3-II. This impairment of autophagic flux was blunted by triacsin C, an inhibitor of acyl-CoA synthetases (ACSLs, enzymes that convert FA to fatty acyl-CoA), suggesting that long chain acyl-CoA or its metabolic products may block the lysosomal function 34 . The data suggest that the abundance of intracellular acyl-CoA regulates autophagic flux (Fig. 2B). Next, we examined whether β-oxidation regulates autophagic flux because most of the intracellular acyl-CoA is metabolized by mitochondrial β-oxidation. Mitochondrial fatty acid oxidation was blocked by using etomoxir, an inhibitor of carnitine pamitoyltransferase-1 (CPT1). In this experiment, submaximal concentrations of PA (10 μM and 50 μM) instead of saturating dose (200 μM) were used to observe the effect of etomoxir on autophagic flux. Blocking β-oxidation by etomoxir exacerbated the impairment of autophagic flux (Fig. 2C). To further corroborate this result, we examined the effect of reduced Cpt1b on autophagic flux in the primary mouse cardiac endothelial cells (MHEC) isolated from Cpt1b (+/−) heterozygous mice. Consistent with Fig. 2C, the autophagic flux was only partially impaired by SFA in the MHECs isolated from wild type (WT) mice, Scientific RepoRtS | (2020) 10:13523 | https://doi.org/10.1038/s41598-020-70347-w www.nature.com/scientificreports/ whereas the autophagic flux was completely blocked by SFA in the MHECs isolated from Cpt1b (+/−) mice. Altogether, these results indicate that a reduction in mitochondrial fatty acid oxidation contributes to the accumulation of LC3-II and autophagosome degradation (Fig. 2D).
Metformin ameliorates the SFA-induced impairment of autophagic flux and reduces the accumulation of lipid droplet. Metformin promotes lipid metabolism in skeletal muscle, liver, and brown adipose tissue [35][36][37] . However, the effect of metformin in SFA-induced impairment of autophagy in endothelial cells is unknown. We examined whether metformin can improve autophagic flux. Pre-treatment with metformin decreased the accumulation of LC3-II and ameliorated the SFA-induced impairment of autophagic flux (Fig. 3A). The results indicate the ability of metformin to improve autophagic flux. Next, we examined whether AMPKα is involved in the effect of metformin on autophagic flux. Human endothelial cells were transfected with siRNA for AMPKα or scrambled, and then the cells were treated with or without SFA in the presence or absence of metformin. The knock-down of AMPKα blunted the effect of metformin on autophagic flux by 42% (from 72.2 to 28.2%) (Fig. 3B,C). The results suggest that metformin improves autophagic flux through an AMPKαmediated mechanism. Since the accumulation of intracellular lipid droplets (LD) is regulated by autophagy 3,21,38 , we investigated whether metformin regulates the accumulation of LD. Interestingly, pre-treatment with metformin decreased the abundance of LD (Fig. 3D,E). Moreover, the blocking β-oxidation by etomoxir significantly suppressed the effect of metformin on reducing LD accumulation (Fig. 3D,E). This suggests that the cytosolic amount of acyl-CoA may play an important role in not only autophagic flux but also LD accumulation.
Metformin reduces LD by enhancing the degradation of LD, but not by inhibiting the formation of LD. We examined whether metformin increases the degradation or reduces the formation of LD. Endothelial cells were treated with metformin during or after the LD was formed. The number of LD was reduced to a similar extent of ≃ 86% under both conditions (Fig. 4A,B). The result suggests that metformin enhances the degradation of LD rather than inhibition of LD formation. Thus, the reduction of LD by metformin seems to be mainly mediated by autophagic degradation.  www.nature.com/scientificreports/ lysosomal pH was assessed by a lysotracker. Treatment with PA elevated the lysosomal pH, which was restored by AICAR (an AMPK activator) or metformin (Fig. 5A). This suggests that metformin may ameliorate the SFAinduced lysosomal dysfunction by regulating lysosomal pH. Lysosomal acid lipase (LAL) is a lysosomal specific lipase, which contributes to lipophagy 40,41 . To examine whether LAL is involved in the reduction of LD by metformin, LAL was knocked down by using siRNA for LAL (Fig. 5A). We confirmed that LAL was reduced by more than 80% by siRNA transfection (Fig. 5B). The ability of metformin to reduce the number of LD was blunted by LAL knock-down (Fig. 5C). The co-localization of LD with LAMP-1, a lysosomal marker protein indicates that LD is degraded by autophagolysosome 21 . Interestingly, treatment with metformin decreased the co-localization of LD with LAMP-1 by 68.3%, which suggests that LD was degraded by metformin in cells transfected with scrambled siRNA. The knock-down of LAL increased the co-localization of LD with LAMP-1 by twofold, both www.nature.com/scientificreports/ in the presence and absence of metformin compared to the fatty acid-treated control cells (Fig. 5D). The data suggests that LD degradation by metformin is dependent on LAL activity.

Metformin suppressed leukocyte adhesion by enhancing lysosomal function. SFA stimulates
a pro-inflammatory response in vascular endothelial cells 25,26,42 . Thus, we examined whether autophagy plays a role in the SFA-induced inflammatory response. The co-culture of human monocyte with endothelial cells was performed and the adhesion of human monocyte was evaluated with or without inhibition of autophagy. Treatment with bafilomycin A1 increased monocyte adhesion by fivefold, which is comparable to the results from the treatment with PA (Fig. 6A). Furthermore, the treatment with metformin decreased PA-induced monocyte adhesion by 79.5% (Fig. 6B). The result suggests that metformin-stimulated autophagic flux may contribute to the anti-inflammatory action of metformin.

Discussion
Autophagy is an essential catabolic process for normal cellular physiology and is activated by stressful conditions including starvation and over-nutrition as a survival mechanism 7 . Human subjects with metabolic dysfunction, such as diabetes and obesity, have higher fatty acid levels with increased inflammatory responses 43,44 . These inflammatory responses are associated with cardiovascular diseases, neurodegeneration, and metabolic syndrome [44][45][46][47] . Although the involvement of autophagy in normal immune function and inflammation is manifold, a growing body of evidence indicates that autophagy is one of the crucial mechanisms that regulate inflammatory responses 34 . We previously demonstrated that SFA causes pro-inflammatory responses through a tolllike receptor-mediated mechanism in vascular endothelial cells 25,26 . However, the link between SFA-induced inflammatory response and autophagy in endothelial cells is largely unknown. Metformin is the first line of treatment for T2DM and is known to have additional beneficial effects in patients with cardiovascular disease or cancer 29,48,49 . In the present study, we demonstrate the anti-inflammatory effect of metformin which is mediated by the increased autophagic flux in endothelial cells. This presents a novel mechanism by which metformin protects cardiovascular function in patients with diabetes and dyslipidemia (Fig. 7). Subjects with obesity and/or diabetes have elevated circulating lipid levels that cause endothelial dysfunction 25,26,[50][51][52] . The mechanisms by which SFA causes endothelial dysfunction are increased reactive oxygen species inflammatory response, and endoplasmic reticulum stress 25,51-54 . We demonstrate that SFA but not monounsaturated fatty acid increases the accumulation of LC3-II (Fig. 1C). Although lauric acid is shorter than PA, lauric acid induces pro-inflammatory response as potent as palmitic acid in macrophage cell line 55,56 . Likewise, lauric acid may cause more stress than oleic acid. These cellular stress conditions may affect autophagy, which could be a novel mechanism for endothelial dysfunction and vascular inflammation [57][58][59] . Therefore, it is plausible that SFA may affect the autophagy process. In fact, SFA increases autophagosome formation, but the autophagic degradation (flux) is impaired due to the elevated lysosomal pH (Fig. 5). This impairment of autophagic flux may contribute to ectopic lipid accumulation, the fat deposit in non-adipose tissue. Since ectopic lipid accumulation is associated with metabolic syndrome, our findings suggest that SFA causes cardiovascular and metabolic dysfunction through impaired autophagic flux.
Most cytosolic fatty acids are ligated to coenzyme A (CoA) by acyl-CoA synthetases before the fatty acyl-CoAs undergo β-oxidation or further cellular utilization 60 . Inhibition of fatty acyl-CoA synthesis by triacsin C reversed the PA-induced impairment of autophagic flux (Fig. 2B). Because there are several isotypes expressed in endothelial cells, it would be interesting to identify a specific ACSL isotype that plays an important role in autophagic degradation. However, it is also possible that each isotype may compensate for the functions of each other. To clarify this issue, further studies are necessary. Inhibition of β-oxidation reduces the flux amount (Fig. 2C). This result suggests that the cytosolic fatty acyl-CoA may contribute to the impairment of autophagic flux due to the abundance of mitochondria (Fig. 7). Moreover, the effect of β-oxidation on autophagic flux may be dependent on cell types, because the amount of cytosolic acyl-CoA is dependent on the number of mitochondria and its function. Because endothelial cells are not a mitochondria-rich cell type compared to skeletal muscle or cardiac tissue, endothelial cells may be more sensitive to FA-induced impairment of autophagy.
Since fatty acyl-CoA is a component of triglycerides (TG), increased the cytosolic fatty acyl-CoA is associated with TG synthesis, which makes lipid droplets. When the cells are undergoing the anabolic cycle, the lysosomal degradation is likely to be inhibited. Deficiency of long chain fatty acyl-CoA synthetase 1 increases dysfunctional mitochondria in cardiac tissue 61 . Due to mitochondrial dysfunction, fatty acid utilization is impaired while glucose utilization is enhanced 61 . Therefore, the increased glucose utilization may stimulate the expression of ITM2A, which is a negative regulator of vacuolar ATPase (V-ATPase) 62 . Thus, V-ATPase may be inhibited by accumulated fatty acyl-CoA and the increased ITM2A, which causes the elevation of lysosomal pH. The detailed mechanisms by which SFA regulates V-ATPase are remained to be further investigated.
We observed that the treatment with etomoxir increased the accumulation of LD in the presence of metformin (Fig. 3D, E). This may be due to the effect of metformin on the LD accumulation is by increasing β-oxidation or by some other unknown degradation pathway. Furthermore, the knock-down of LAL abolished the effect of metformin on LD accumulation (Fig. 5C). This suggests that metformin reduces LD accumulation at least in part through autophagic degradation.
The pharmacological concentration of metformin reaches 40-70 μM in the portal vein, and 10-40 μM in circulation 63 . The primary action of metformin is inhibition of hepatic gluconeogenesis by both AMPK-dependent and -independent mechanisms 29,64 . Metformin suppressed gluconeogenesis through activation of AMPK at lower concentrations (< 250 μM), while decreasing ATP levels at higher concentrations (> 250 μM). High concentrations of metformin (2-5 mM) inhibit respiratory chain complex 1 in hepatocytes, which leads to an increased in the AMP/ATP ratio by an AMPK-independent mechanism 64,65 . We used a low concentration of www.nature.com/scientificreports/ metformin (100 μM) in the present study. Thus, the effect of metformin in our study seems to be through an AMPK-dependent mechanism rather than by increasing AMP/ATP ratio via inhibition of respiratory chain complex 1 (Fig. 3B,C). Also, a previous study demonstrates that metformin directly activates V-ATPase 66 , which is consistent with our result (Fig. 5). This suggests that the protective effect of metformin in fatty acid-induced lipotoxicity may be mediated by the stimulation of AMPK-dependent autophagic flux.
In the present study, we demonstrate the effects of metformin on autophagy and its role in LD accumulation in vascular endothelial cells (Figs. 3, 4, 5). These suggest a novel mechanism by which metformin protects cardiovascular functions from lipotoxicity and dyslipidemia (Fig. 7). In addition to anti-diabetic action, metformin has beneficial effects on cardiovascular functions, including improvement of endothelial function and prevention of heart failure and atherosclerosis 28,31,[67][68][69][70][71] . Activation of AMPK is one of the major mechanisms for the vasculoprotective effects of metformin 72 . Further study is warranted on the involvement of AMPK and eNOS in metformin-stimulated autophagic flux and reduced LD accumulation.
Inhibition of autophagy increased monocyte adhesion which was suppressed by metformin (Fig. 6). We and others reported that SFA causes endothelial dysfunction and pro-inflammatory responses through toll-like receptor (TLR)-mediated mechanisms 25,26,42 . Since metformin has an anti-inflammatory effect and attenuates TLR signaling in liver, skeletal muscle, and heart, it would be of interest to investigate whether metformin-induced autophagic flux affects innate immune response [73][74][75][76] . The deficiency of autophagy in endothelial cells exacerbates atherosclerosis 77 . This suggests the impaired autophagic flux by SFA may contribute to the development of atherosclerosis.  www.nature.com/scientificreports/ In the current study, we demonstrate that SFA causes accumulation of LD and pro-inflammatory responses and that metformin ameliorates the adverse effects of SFA through an autophagy-dependent mechanism. In conclusion, autophagic flux is a novel mechanism by which metformin protects vascular endothelium from lipotoxic injuries.

Materials and methods
Materials. Anti-LC3 (#4599) and anti-β-actin (#3700) antibodies were obtained from Cell Signaling Technology (Danvers, MA, USA). NH 4 Cl, etomoxir, and leupeptin were purchased from Sigma-Aldrich (St. Louis, MO, USA). DsiRNA for AMPKα, LAL and scrambled dsiRNA were purchased from Integrated DNA Technologies (Coralville, IA, USA). Metformin, and bafilomycin A1 were obtained from Millipore Sigma (Burlington, MA, USA). Most of the compounds used in this study were dissolved in dimethyl sulfoxide, and we confirmed that the vehicle alone did not affect our results. isolation of mouse primary heart endothelial cells. The mouse heart endothelial cells (MHEC) were isolated using a modification of previously described methods 79 . Briefly, after 6 wk old mice were euthanized by isoflurane inhalation, the heart was dissected, and then minced heart tissue was incubated in 2 mg/ml collagenase I/PBS (Worthington Biochemical Co., Lakewood, NJ, USA) for 45 min at 37 °C. The collagen treated hearts were triturated with the syringe attached a cannula, filtered through a cell strainer and washed twice with DMEM. The cells were isolated by using microbeads (Dynal beads M-450, Invitrogen, Carlsbad, CA, USA) coated with anti-CD31 (BD Biosciences, San Jose, CA, USA) in PBS with 0.1% BSA for 10 min. The isolated endothelial cells were seeded onto gelatin-coated plates in DMEM supplemented with 20% FBS, 45 μg/ml endothelial cell growth supplement (ECGS), 100 U/ml penicillin, 100 μg/ml streptomycin, and 10 U/ml of heparin. MHEC were used between passages 3 and 6. endothelial cell culture and transfection. Maintenance and transfection of endothelial were performed as previously described 25,80 . Immortalized human endothelial cell line EA.hy926 cells (CRL-2922) were obtained from ATCC (Manassas, VA, USA), and maintained in DMEM low glucose media containing 10% FBS, penicillin (100 U/ml) and streptomycin (100 μg/ml). Cells were transiently transfected with 100 nM of siRNA duplex oligonucleotides (siRNA for LAL or scrambled) using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. Two days after transfection, cells were serum-starved for 2 h and then treated with BSA or PA as indicated in legends to figures. preparation of palmitic acid. Preparation of PA was carried out as described by Mott et al. 81 . Briefly, 10.5% BSA (Fitzgerald, MA, USA) was dissolved in 25 mM HEPES/DMEM, and filtered (0.22 μM, Millipore, Burlington, MA, USA). Sodium palmitate was heated to be dissolved in water and rapidly added to warmed BSA solution. Then this BSA-conjugated PA, oleic acid, or lauric acid (Nu-Chek, Elysian, MA, USA) were added to the concentrations as indicated in the figure legends. We used endotoxin-free reagents and we tested all the reagents we used including BSA, PA, media and reagent diluents. We checked the endotoxin level of all the reagents we used in this study by Chromogenic Endotoxin Quantitation assay kit (Thermo Fisher, Waltham, MA, USA). Lower than 25 pg/ml of endotoxin is undetectable. Figure 5. Impairment of lysosomal dysfunction causes the accumulation of LD. (A) Endothelial cells were pretreated with either AICAR (200 μM) or metformin (100 μM) 30 min prior to PA (200 μM). The acidic lysosome was stained with a lysotracker (red), and the nucleus was stained with Hoechst 33342 (blue). Treatment with PA elevated pH of the lysosome and pre-treatment with AICAR or metformin restored the acidic PH of the lysosome. (B) Endothelial cells were transfected with scrambled or siRNA for LAL. siRNA for LAL reduced the expression of LAL. (C) The knock-down of LAL blunted the effect of metformin in LD reduction. Human endothelial cells were transfected with siRNA for scrambled or LAL. Endothelial cells were pre-treated without or with metformin (100 μM) 30 min prior to PA (200 μM). LD was stained with BODIPY (green). The number of LD was quantified by using Image J and is shown in bar graphs (mean ± SEM). ****p < 0.0001, ***p < 0.001, **p < 0.01, and *p < 0.05 indicate that the samples are statistically different between the indicated samples. (D) The knock-down of LAL increased the co-localization of lipid droplet (BODIPY, green), with LAMP-1 (lysosome marker, red). Six microscopic fields per condition were obtained and quantified by using Image J. The quantification of colocalization of LD with LAMP-1 is quantified with Image J and is shown in bar graphs (mean ± SEM). **p < 0.01 and *p < 0.05, indicates that the samples are statistically different between the indicated samples. The full-length agarose gel is presented in Supplemental Figure S5A, and the inverted image is presented in Supplemental Figure S5B. The dotted lines in the Supplemental figures indicate the cropped area.
immunocytochemistry. Immunocytochemistry was performed as described previously 21 . To visualize lipid droplets, cells grown on coverslips were fixed with 4% formaldehyde in PBS, and then BODIPY 493/503 (Thermo Fisher Scientific, Waltham, MA, USA) was stained for 30 min at room temperature. For immunofluorescent staining of cells, cells were treated as described in the figure legends. After stimulation, cells were fixed in 4% paraformaldehyde/PBS for 10 min and washed the cells with PBS. Cells were then permeabilized with 0.1% Triton X-100/PBS for 5 min and washed with PBS. Cells were blocked with 5% BSA/PBS for 1 h and then incubated with anti-LAMP-1 antibody in 5% BSA/PBS at 4 °C overnight. Cells were washed with PBS 3 times (5 min each) and then incubated with Alexa 555 conjugated-goat anti-rabbit IgG (Thermo Fisher, Waltham, MA, USA) for 1 h at room temperature. Images were acquired with an Axiovert fluorescence microscope (Carl Zeiss Ltd., Thornwood, NY, USA). Ten to fifteen cells were randomly selected from each treatment to calculate the average number of lipid droplets and the percentage of co-localization per cell. Quantification was performed using Image J software (NIH, MD, USA) and the percentage of co-localization was calculated by JACoP plugin of Image J. The data shown are from one representative experiment of three independent repeats. The quantification of adherent cell numbers is shown in bar graphs (mean ± SEM). ***p < 0.001 indicates that the samples are statistically different between the indicated samples.

Statistical analysis.
Values are presented as mean ± standard error of the mean (SEM). Western blots were analyzed by one-way ANOVA.
Ethical approval for animal study. Ethical approval is taken for using 'mice' in the study by the Animal Use and Care Committee at The University of Alabama at Birmingham.