Skeletal Muscle-specific PGC-1α Overexpression Suppresses Atherosclerosis in Apolipoprotein E-Knockout Mice

Endurance exercise training prevents atherosclerosis. Peroxisome proliferator-activated receptor γ coactivator-1α (PGC-1α) increases myokine secretion from the skeletal muscle, and these myokines have been shown to affect the function of multiple organs. Since endurance exercise training increases PGC-1α expression in skeletal muscles, we investigated whether skeletal muscle-specific PGC-1α overexpression suppresses atherosclerosis. Apolipoprotein E-knockout (ApoE-KO)/PGC-1α mice, which overexpress PGC-1α in the skeletal muscle of ApoE-KO mice, were sacrificed, and the atherosclerotic plaque area, spontaneous activity, plasma lipid profile, and aortic gene expression were measured. Immunohistochemical analyses were also performed. The atherosclerotic lesions in ApoE-KO/PGC-1α mice were 40% smaller than those in ApoE-KO mice, concomitant with the reduction in vascular cell adhesion molecule-1 (VCAM-1) and monocyte chemoattractant protein-1 (MCP-1) mRNA and protein levels in the aorta. Spontaneous activity and plasma lipid profiles were not changed by the overexpression of PGC-1α in the skeletal muscle. In human umbilical vein endothelial cells, Irisin and β-aminoisobutyric acid (BAIBA), PGC-1α-dependent myokines, inhibited the tumor necrosis factor α-induced VCAM-1 gene and protein expression. BAIBA also inhibited TNFα-induced MCP-1 gene expression. These results showed that the skeletal muscle-specific overexpression of PGC-1α suppresses atherosclerosis and that PGC-1α-dependent myokines may be involved in the preventive effects observed.


Decrease in atherosclerosis lesions by overexpression of PGC-1α in skeletal muscle.
We generated ApoE-KO/PGC-1α mice, overexpressing PGC-1α-b in the skeletal muscle of ApoE-KO mice. Since plasma triglyceride (TG) and total cholesterol (TC) levels are increased greatly and atherosclerosis is initiated in ApoE-KO mice fed with normal chow 24 , we fed ApoE-KO mice normal chow. The mean body weight of the ApoE-KO mice was 31.1 ± 0.7 g and that of ApoE-KO/PGC-1α mice was 27.9 ± 1.8 g, indicating that PGC-1α overexpression in skeletal muscle did not affect body weight. Representative photomicrographs of hematoxylin and eosin (H&E)-stained aorta sections from ApoE-KO and ApoE-KO/PGC-1α mice are presented in Fig. 1A. Atherosclerotic plaque areas were smaller in the aortic sections of ApoE-KO/PGC-1α mice than in ApoE-KO mice. Quantification of the aortic plaque areas showed that these were significantly lower in ApoE-KO/ PGC-1α mice, approximately 40% on average compared with those in ApoE-KO mice (3.1 ± 0.2 × 10 5 μm 2 versus 5.5 ± 0.6 × 10 5 μm 2 , p < 0.01; Fig. 1B). Such plaques were not formed in wild-type C57BL6/J (WT) mice (data not shown). Spontaneous activity of ApoE-KO mice was not significantly changed by the overexpression of PGC-1α (Fig. 2). Thus, skeletal muscle-specific overexpression of PGC-1α suppressed atherosclerosis progression in ApoE-KO mice without increasing spontaneous activity. The plaque areas of aged ApoE-KO/PGC-1α mice (37)(38)(39)(40)(41) weeks old) were also significantly lower in comparison to those in aged ApoE-KO mice (Supplemental Fig. S1). www.nature.com/scientificreports www.nature.com/scientificreports/ Effects of overexpression of PGC-1α on plasma lipid profiles. In order to reveal how skeletal muscle-specific PGC-1α overexpression suppressed the progression of atherosclerosis, we measured the plasma lipid profiles of mice because dyslipidemia is one of the leading high-risk factors for atherosclerotic disease 25 . Plasma lipid profiles are shown in Table 1. Only plasma chylomicron cholesterol concentrations were higher in ApoE-KO/PGC-1α mice than in ApoE-KO mice. However, there were no differences between the two groups with respect to other lipoproteins. Since it is reported that chylomicron cholesterol does not affect the progress of atherosclerosis 26 , these results showed that overexpression of PGC-1α in the skeletal muscle did not affect atherosclerosis-related plasma lipid concentrations. We measured the levels of plasma thiobarbituric acid reactive substance (TBARS), a biological marker of oxidative stress 27 . The reduction in plasma TBARS levels was not observed by the overexpression of PGC-1α in ApoE-KO mice (Supplemental Fig. S2). Blood glucose levels under fasting, one of the major risk factors for atherosclerosis 28 , were also measured in ApoE-KO/PGC-1α mice. Fasting blood glucose levels were not significantly different between ApoE-KO/PGC-1α mice and ApoE-KO mice (Supplemental Fig. S3). Although previous studies showed that plasma sphingomyelin (SM) correlated with the development of coronary artery diseases 29 , plasma SM concentrations were not lowered by the overexpression of PGC-1α in ApoE-KO mice (Supplemental Fig. S4).

Changes in mRNA expression in the aorta by overexpression of PGC-1α in skeletal muscle.
Inflammatory mediators have been shown to be involved in the progression of atherosclerosis 30 . Vascular cell adhesion molecule-1 (VCAM-1) and intercellular adhesion molecule-1 (ICAM-1) mediate monocyte adhesion on endothelial cells 31 . Monocyte chemoattractant protein-1 (MCP-1) mediates monocyte migration underneath  www.nature.com/scientificreports www.nature.com/scientificreports/ endothelial cells 32 . Nuclear factor kappa B (NFκB), which consists of two subunits (p50, p65), regulates the expression of VCAM-1, ICAM-1, and MCP-1 33,34 . Interleukin-6 and tumor necrosis factor α (TNFα), are inflammatory cytokines and are expressed in plaques 35,36 . The changes in mRNA levels of these inflammatory mediators in the abdominal aorta are shown in Fig. 3. In ApoE-KO mice, VCAM-1 mRNA expression was significantly higher compared to WT mice as reported previously 37 . However, VCAM-1 and MCP-1 mRNA expression levels in the aorta from ApoE-KO/PGC-1α mice were significantly lower compared to that in ApoE-KO mice. On the other hand, although mRNA expression of inflammatory mediators in peripheral blood cells was reported as an indicator for inflammatory status and the disease state 38 , no significant differences were observed in the mRNA expression of inflammatory mediators in blood cells among three groups of mice (Supplemental Fig. S5). These data suggested that PGC-1α overexpression in skeletal muscle suppressed VCAM-1 and MCP-1 mRNA expression in the aorta, and that VCAM-1 and MCP-1 suppression might be involved in prevention of plaque formation, particularly because inflammatory processes play key roles in the progression of atherosclerosis 39 .

Immunohistochemical analyses of aorta cryosections from ApoE-KO/PGC-1α mice.
Since the suppression of VCAM-1 and MCP-1 mRNA expression was identified in the abdominal aorta from ApoE-KO/ PGC-1α mice, immunohistochemical analyses were performed to determine whether VCAM-1 and MCP-1 protein abundance were also suppressed in atherosclerotic plaque areas around aortic valves. Multiple photomicrographs obtained after immunofluorescent staining are shown in Fig. 4. The positive staining areas for VCAM-1 and MCP-1 in ApoE-KO/PGC-1α mice were lower compared with those in ApoE-KO mice. Thus, skeletal muscle-specific PGC-1α overexpression suppressed VCAM-1 and MCP-1 protein expression in atherosclerotic plaques around aortic valves. Vascular smooth muscle cells are regarded as exerting a plaque-stabilization effect 40 . Macrophages are differentiated from monocytes and involved in atherosclerosis progression 39 . Therefore, we also analyzed vascular smooth muscle proliferation and macrophage localization using immunofluorescent staining with anti-α-smooth muscle actin (α-SMA) and anti-Galectin 3 (Mac-2) antibodies. However, α-SMA positive smooth muscle cells and Mac-2 positive macrophages were not significantly altered by skeletal muscle-specific overexpression of PGC-1α in ApoE-KO mice.

PGC-1α-mediated changes of gene expression involved in the production of Irisin and BAIBA in
the skeletal muscle of Apoe-Ko mice. Irisin and BAIBA are known as PGC-1α-dependent myokines 20,21 .
To determine whether PGC-1α-mediated production of these myokines was observed in ApoE-KO mice, we measured the expression levels of genes involved in biosynthesis of Irisin and BAIBA. As shown in Fig. 5A, FNDC5, which is the gene symbol of Irisin, was increased 2.3-fold in the muscle of ApoE-KO/PGC-1α mice. mRNA required for BAIBA biosynthesis, such as Acyl-CoA dehydrogenase short chain (Acads), hydroxyacyl-CoA dehydrogenase trifunctional multienzyme complex subunit alpha (Hadha), and hydroxyacyl-Coenzyme A dehydrogenase (Hadh) were increased in ApoE-KO/PGC-1α mice compared to ApoE-KO mice (Fig. 5B). Therefore, skeletal muscle-specific PGC-1α overexpression increased the expression of genes involved in production of Irisin and BAIBA in the muscles of ApoE-KO mice.

Effect of Irisin and BAIBA, PGC-1α-dependent myokines, on VCAM-1 and MCP-1 expression in human umbilical vein endothelial cells (HUVECs). PGC-1α overexpression in the skeletal muscle
suppressed VCAM-1 and MCP-1 expression in the aorta; however, it is unclear how VCAM-1 and MCP-1 were suppressed. We hypothesized that PGC-1α-dependent production and secretion of myokines might suppress VCAM-1 and MCP-1 expression in endothelial cells. Therefore, we exposed TNFα-treated HUVECs to Irisin and BAIBA because TNFα is involved in atherosclerosis progression 41,42 . TNFα, for instance, is present in atherosclerotic arterial walls in humans 36 and administration of TNFα to ApoE-KO mice promotes the progression of atherosclerosis 42 . Irisin and BAIBA reduced TNFα-induced mRNA expression of VCAM-1 in HUVECs (Fig. 6A).   www.nature.com/scientificreports www.nature.com/scientificreports/ MCP-1 mRNA expression was also decreased by addition of BAIBA (Fig. 6B). The effect of the myokines on the expression of other genes related to atherogenesis is shown in Supplemental Fig. S6.
In order to confirm whether Irisin and BAIBA suppress VCAM-1 and MCP-1 protein expression as well as gene expression, we also measured protein expression levels directly. Incubation of the cells with TNFα increased VCAM-1 and MCP-1 protein expression significantly. Irisin and BAIBA suppressed TNFα-induced VCAM-1 protein expression in HUVECs (Fig. 7). Therefore, Irisin and BAIBA decreased the expression of VCAM-1 at the level of both mRNA and protein production.

Discussion
The present study shows that skeletal muscle-specific PGC-1α overexpression suppressed the progression of atherosclerosis in ApoE-KO mice without changing spontaneous activity and plasma lipid profiles. Overexpression of PGC-1α in skeletal muscle suppressed VCAM-1 and MCP-1 expression in the aorta. Furthermore, Irisin and BAIBA suppressed TNFα-induced VCAM-1 mRNA and protein expression in HUVECs.
Plasma lipid profiles were not changed by muscular overexpression of PGC-1α in ApoE-KO mice. Previous studies showed that improvement of plasma lipid profiles was not necessary for the suppression of atherosclerosis by endurance exercise training in mouse models. In the report by Pellegrin, 24 weeks of swimming exercise decreased the progression of atherosclerotic lesions by 30% in ApoE-KO mice without changes to plasma TG, TC, HDL-cholesterol, non-HDL-cholesterol, and phospholipid levels 43 . Eight weeks of swimming also suppressed fatty streak plaque lesions developed in ApoE-KO mice fed a high-fat diet without changing plasma TG and TC levels 44,45 . These previous studies suggested that exercise-induced antioxidant effects via the vascular NO system suppressed atherosclerosis. However, plasma TBARS levels and mRNA expression of eNOS in the aorta were not improved in 16 to 20-week-old ApoE-KO mice by overexpression of PGC-1α in skeletal muscles, suggesting that plasma lipid profiles and antioxidant systems were not involved in the suppression of atherosclerosis observed in ApoE-KO/PGC-1α mice. Exercise is recommended to lower LDL-cholesterol and non-HDL-cholesterol 46 for the prevention of atherosclerotic disease. However, exercise therapy appears to suppress atherosclerosis even when there is no change in plasma lipid profiles.
In addition to lipid profiles and antioxidant systems, chronic arterial wall inflammation is a key component of atherosclerosis 47 . VCAM-1, ICAM-1, MCP-1, NFκB, IL-6 and TNFα are involved in atherogenesis 30 . Among these factors, VCAM-1 and MCP-1 have been reported as factors for the initiation of atherosclerosis 31,32 . VCAM-1 induces monocyte adhesion and accumulation on the vessel wall, and MCP-1 induces monocyte migration and infiltration under the vessel wall, respectively 31,32 . Requirement of VCAM-1 for the initiation of atherosclerosis was shown using a Vcam1 D4D /LDL receptor-knockout mouse model 48 . In addition, MCP-1 deficiency was reported to suppress the progression of atherosclerosis in ApoE-KO mice 49 . Therefore, decreased expression of VCAM-1 and MCP-1 suppresses the development of atherosclerosis. Furthermore, the expression of VCAM-1 www.nature.com/scientificreports www.nature.com/scientificreports/ and MCP-1 in the aorta was reduced by exercise training 50,51 , suggesting that endurance exercise training might decrease aortic VCAM-1 and MCP-1 expression in preventing atherosclerosis. In the present study, we found that VCAM-1 and MCP-1 expression was lower in the aorta from ApoE-KO/PGC-1α mice. It seems likely that decreases in the expression of these factors are implicated in the suppression of atherosclerosis by the overexpression of PGC-1α in the skeletal muscle.
In this study, Irisin and BAIBA reduced VCAM-1 protein expression in HUVECs, as observed in the aorta of ApoE-KO/PGC-1α mice. However, although MCP-1 protein expression was suppressed in the aorta by skeletal muscle PGC-1α overexpression, TNFα induced expression was not reduced in HUVECs treated with Irisin and BAIBA. In order to rationalize this difference, we considered that the suppression of MCP-1 expression may be occurring in macrophages. VCAM-1 is expressed by endothelium and mediates monocyte adhesion 48 . MCP-1, however, is expressed, not only in endothelial cells, but also in monocytes and macrophages and acts as a potent monocyte chemotactic factor 32 . Therefore, myokines may act on macrophages to suppress the expression of MCP-1. Furthermore, unidentified factor(s) other than Irisin and BAIBA might also involve in the suppression of MCP-1 expression in aorta, in vivo.
Recently, it was reported that myokines are secreted from skeletal muscle and affect not only the skeletal muscle itself but also other organs such as the liver and adipose tissue 15 . Epidemiological studies revealed a correlation between Irisin, and CVDs risk 22,23 . Protective effects of Irisin on atherosclerosis were also reported in two different ApoE-KO mouse models 37,52 . In these experiments, Irisin treatment significantly decreased atherosclerotic plaque area concomitantly with the reduction of inflammatory cytokines expression in aorta. Since overexpression of PGC-1α in skeletal muscle increased the production and secretion of Irisin from skeletal muscle 20 , Irisin secretion might represent one of the mechanisms for suppression of atherosclerosis. In the present study, VCAM-1 and MCP-1 expression were lower in the aorta as a result of PGC-1α overexpression in skeletal muscle. Moreover, we found that Irisin treatment of HUVECs suppressed TNFα-induced expression of VCAM-1 mRNA and protein. In addition to Irisin, BAIBA was reported as a myokine and secreted from cells with forced expression of PGC-1α, which results in the browning of white adipose tissue and increases fat oxidation in the liver 21 .
Their study also revealed that, in humans, plasma BAIBA levels were increased with exercise and inversely associated with metabolic risk factors, such as fasting glucose, insulin, homeostasis model assessment of insulin resistance (HOMA-IR), TG, and TC levels 21 . We have previously reported that BAIBA was detected only in mice that overexpressed PGC-1α in the skeletal muscle, but not in WT mice 53 . However, the protective effects of BAIBA on the progression of atherosclerosis and inflammatory reactions on the vessel walls have not been uncovered. In the present study, we observed that BAIBA treatment of HUVECs suppressed the TNFα-induced expression of VCAM-1, as also observed for Irisin. The plasma BAIBA concentration was 6.5 ± 2.5 µM in mice overexpressing www.nature.com/scientificreports www.nature.com/scientificreports/ PGC-1α in skeletal muscle 21 and we also observed decreased mRNA and protein expression of VCAM-1 at 10 and 40 µM of BAIBA treatment, respectively, in in vitro experiments, suggesting that increased plasma BAIBA levels may play an important role in the anti-atherogenic effects of PGC-1α overexpression in skeletal muscle. On the other hand, adiponectin has beneficial effects in terms of atherosclerotic progression 54 . Okamoto et al. reported that adenovirus-mediated elevation of plasma adiponectin suppressed atherosclerosis progression in ApoE-KO mice concomitantly with suppression of aortic VCAM-1 mRNA expression 55 . Irisin enhanced adiponectin expression in lipopolysaccharide-treated adipocytes 56 . Therefore, adiponectin-mediated mechanisms might also be involved in the prevention of atherosclerosis by PGC-1α overexpression in the skeletal muscle.
In this study, we show that atherosclerosis was suppressed in the ApoE-KO/PGC-1α mice, and the mRNA expression levels of FNDC5 and the genes involved in BAIBA biosynthesis in skeletal muscle were increased about 2-fold in the ApoE-KO/PGC-1α mice. It has been reported that 3 weeks of exercise training could increase plasma myokines concentrations (2.5-fold for Irisin 20 and 1.2-fold for BAIBA 21 ), and exercise training for 20 weeks suppressed atherosclerosis in ApoE-KO mice 4 . Prolonged secretion of these myokines for 20 weeks may be involved in suppression of atherosclerosis in exercise-trained ApoE-KO mice. In humans, endurance training causes a 2-fold increase in the protein expression of PGC-1α in skeletal muscle 57 . Endurance training increases serum Irisin levels 1.2-fold 58 and plasma BAIBA concentrations by 17% 21 . Continuous training is expected to increase the expression of PGC-1α and the increased levels of these myokines in plasma. Therefore, Irisin and BAIBA might be involved in exercise-induced atherosclerosis suppression in humans in the same manner as in ApoE-KO/PGC-1α mice.
In conclusion, we have shown that PGC-1α overexpression in skeletal muscle suppressed VCAM-1 and MCP-1 expression in the arterial wall and inhibited the progression of atherosclerosis. Furthermore, we demonstrated that PGC-1α-dependent myokines, namely Irisin and BAIBA, reduced TNFα-induced VCAM-1 expression. These findings indicate that adaptive effects of endurance training on the skeletal muscle might be one of the reasons for exercise training-mediated anti-atherogenic effects. Irisin and BAIBA may be useful biomarkers to identify whether endurance exercise training is sufficient to prevent atherosclerotic diseases.

Materials and Methods
Animals. Homozygous ApoE-KO mice and heterozygotes PGC-1α mice were crossbred and backcrossed into a murine ApoE-KO background. ApoE-KO mice and homozygous ApoE-KO/heterozygotes PGC-1α-b mice (ApoE-KO/PGC-1α mice) were obtained, and 16-20-week-old male offspring were used in experiments. All mice were maintained on a C57BL6/J background. ApoE-KO mice were obtained from The Jackson Laboratory (Bar Harbor, ME, USA) 59 . The methods for generation of PGC-1α mice were described previously 14 . The promoter for human α-skeletal actin, provided by Drs E. C. Hardeman and K. L. Guven (Children's Medical Research Institute, Australia) was used to express PGC-1α-b in skeletal muscle. Animals were housed in groups of 5 mice per cage in a room with a 12-hour light/dark cycle at 22 °C and provided with standard mouse chow (CE-2, CREA Japan Inc., Tokyo, Japan) and drinking water ad libitum. Mice were cared for according to the National Institutes of Health Guide for the Care and Use of Laboratory Animals (https://www/ncbi.nlm.nih.gov/books/NBK54050/) and our institutional guidelines. All animal experiments were approved by the Institutional Animal Care and Use Committee of the University of Shizuoka (number 165123).

Spontaneous activity.
Nineteen-week-old mice were housed in a single cage equipped with infrared ray sensor (NS-AS01, NeuroScience, Tokyo, Japan) for 24 hours. The data were analyzed using ARCO-2000 (ARCO SYSTEM Inc., Chiba, Japan).
Fasting blood glucose levels. Mice (16-17 weeks old) were starved for 12 hours before blood sampling.
The blood samples were collected from the tail vain. Blood glucose levels were measured using Breeze 2 (Bayer, Leverkusen, Germany).
Plasma lipid analysis. Mice were starved for 12 hours before blood sampling. The blood samples were collected from the orbital sinus under isoflurane anesthesia. EDTA was used as an anti-coagulant. The plasma was separated and stored at −80 °C until analysis. Plasma lipid profiles were analyzed by LipoSEARCH service (Skylight Biotech, Inc., Akita, Japan) 60 . Thiobarbituric acid reactive substance concentrations in pooled plasma were measured using Calorimetric TBARS Microplate Assay Kit (FR40, Rochester Hills, Oxford Biochemical Research, MI, USA) according to the manufacturer's instructions. Plasma SM was measured as described previously 61 . In brief, SMs were analyzed by LCMS-8040 (Shimadzu Corp., Kyoto, Japan) under the positive-ion mode using precursor-ion mode scanning at m/z 184 to specifically detect substances containing choline. Obtained MS data were searched with a database of sphingolipids (http://www.lipidmaps.org/tools/ms/sphingo-lipids_batch.html). The relative peak area for each species was normalized by the peak area of internal standard (1,2-diheptadecanoyl-sn-glycero-3-phosphocholine, 850360 P, Avanti Polar Lipids, Alabaster, AL, USA).

Determination of aortic lesion area.
Hearts were fixed with 10% formalin neutral buffer solution for 72 hours, followed by replacement with phosphate buffered saline (PBS). Cryosections (5-μm-thick) were taken at the four levels of the aortic valves and stained with H&E. The area of the atherosclerotic lesions from 4 cross-sections of each mouse heart was measured using Image J software (http://imagej.nih.gov/ij/). Atherosclerotic lesions were calculated as the sum of lesion area across 4 cross-sections 62 . Immunohistochemistry. Cryosections of the aortic valve were fixed on ice (4 °C) with acetone for 5 minutes and washed with 1% Tween in PBS (PBS-T) for 5 minutes 3 times. Cryosections were incubated with Blocking Mouse IgG (MKB-2213, Vector Laboratories, CA, USA) in 1 mL PBS. After washing with PBS-T for 5 minutes 3 times, cryosections were incubated with 5% Normal Goat Serum (50062Z, Thermo Fisher Scientific,