All-trans-retinoic acid ameliorates atherosclerosis, promotes perivascular adipose tissue browning, and increases adiponectin production in Apo-E mice

All-trans-retinoic acid (atRA), an active metabolite of vitamin A, exerts a potential role in the prevention of cardiovascular diseases. It has been shown that atRA ameliorates atherosclerosis while the exact mechanism underlying this protection remains unknown. This study investigated the influence of atRA on insulin resistance (IR), atherosclerosis, and the process of perivascular adipose tissue (PVAT) browning. Moreover, syntheses of adiponectin, adipokine with anti-atherogenic effects, and tumor necrosis factor-alpha (TNF-α), a pro-inflammatory cytokine, were determined in PVAT. Apolipoprotein E-deficient mice (Apo-E) and control C57BL/6J wild-type mice were treated with atRA (5 mg/kg/day) or vehicle (corn oil) by plastic feeding tubes for 8 weeks. Long-term atRA treatment in Apo-E mice did not affect insulin resistance. AtRa administration ameliorated atherosclerosis, induced PVAT browning, and increased adiponectin production in PVAT in Apo-E mice. Furthermore, atRA increased nitric oxide (NO) level but did not affect adiponectin concentration in the aorta of Apo-E mice. These results indicate that atRA ameliorates atherosclerosis in Apo-E mice. We also observed the browning of PVAT. Besides, atRA increased the synthesis of adiponectin in PVAT and augmented NO level in the aorta in ApoE mice.

receptors (RXRs), which are activated specifically by the 9-cis isomer 7 . Previous studies have revealed the protective effects of atRA on the cardiovascular system [8][9][10] . It has been demonstrated that atRA inhibits experimental atherosclerosis by antioxidant and anti-inflammatory action and inhibition of platelet function 8 . Moreover, atRA enhances reverse cholesterol transport from macrophages 11 .
RA has been shown to promote the WAT browning 12 and its administration to mice upregulates the expression of uncoupling protein 1 (UCP1) both in vivo and in vitro 13 . BAT is increasingly recognized as a potential target to reduce atherosclerosis development and has an important role in lipoprotein metabolism as thermogenesis consumes large amounts of fatty acids 14 . Experiments with the use of a mouse as a model demonstrated that activation of BAT by cold temperature enhances the clearance of plasma lipids and prevents the development of atherosclerosis 15,16 . PVAT is a unique adipose tissue depot that has the potential to profoundly perturb the vasculature under certain conditions 17 . Atherosclerotic lesions were significantly increased in mice with impaired PVAT development, thus indicating that the lack of PVAT is sufficient to increase atherosclerosis 18 . In the PVAT surrounding critical plaques, macrophages and lymphocytes numbers were higher than in segments without atherosclerosis 19 . Recently, it has been shown that angiotensin 1-7 mimetic AVE0991 exhibited anti-inflammatory properties affecting monocyte/macrophage differentiation and recruitment to perivascular space in Apo-E mice 20 . Stimulation of PVAT browning seems to be a promising new therapeutic target in obesity and related complications like atherosclerosis.
In the current study, we used the Apo-E mouse model that develops lesions with morphological similarity to human atherosclerotic plaque 21 . We investigated the role of atRA in atherosclerosis and the process of PVAT browning. Further, we explored the atRA influence on insulin resistance (IR), and pro-inflammatory and antiinflammatory adipokines synthesis in PVAT. We also measured adiponectin and NO levels in the aorta of the atRA-or vehicle-treated mice.
Experimental animals. 8-week-old males of C57BL/6J Wild-type and Apo-E mice were obtained from Charles River Laboratories (Wilmington, MA, USA). The animals were housed under controlled conditions with 10/14-h light/dark cycles and temperature (22 °C). Mice were fed standard chow providing 25% calories from protein, 8% from fat, and 67% from carbohydrates (Morawski, Poland). Animals were fed ad libitum with free access to water. The animals were divided into four groups: C57BL/6J wild-type (n = 5), C57BL/6J wild-type treated with atRA (n = 6), Apo-E mice (n = 5) and Apo-E mice treated with atRA; (n = 5). AtRA (5 mg/kg/day) or vehicle (corn oil) were administered orally by plastic feeding tubes for 8 weeks. Blood samples were collected after 6 h of fasting. The animals were anesthetized with ketamine and xylazine and then euthanized. Plasma and tissues were collected and stored at − 80 °C until analysis. For quantification of the atherosclerotic lesions in the aortic root, the entire aorta, between the heart and iliac arteries was dissected. The heart with an attached 2-3 mm long aortic arch was dissected and the lower ~ 70% of the ventricular mass was cut away. The upper cardiac portion was frozen in the O.C.T. tissue-freezing medium (Sigma-Aldrich Chemical Co., St. Louis, MO, USA) and stored at − 80 °C with the aortic arch positioned upwards using a metal mold. The abdominal aorta with PVAT between the left renal vein and the iliac bifurcation were isolated under sterile conditions and cut into two portions. In one-part PVAT was isolated from the aorta under a dissecting microscope and both tissues were frozen and stored at − 80 °C until analysis. The second part of aorta and PVAT was fixed in 4% paraformaldehyde (Sigma-Aldrich Chemical Co., St. Louis, MO, USA) until further use.
All animal procedures were performed according to the Polish Guide for the Care and Use of Animals and were approved by the Local Ethics Committee of the Warsaw University of Life Sciences, Poland. All experiments and methods were carried out in compliance with relevant regulations and Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines.
Evaluation of insulin resistance. Plasma samples were collected from the tail vein before and after 8 weeks of treatment with atRA or vehicle. After 6 h of fasting blood samples were collected in EDTA/aprotinin tubes. Glucose concentration was measured using a glucometer (Accu-chek Performa, Roche, Mannheim, Germany). Insulin was determined using a commercially available ELISA kit (Mercodia AB, Uppsala, Sweden), according to the manufacturer's protocol. The detection limit was ≤ 0.025 μg/L. Intra-assay and inter-assay coefficient of variation were < 3% and < 5%, respectively. The HOMA-IR index was calculated as (fasting serum glucose × fasting serum insulin/22.5) to assess IR 22   (CCC), USA) and AKT mouse monoclonal antibody (Santa Cruz Biotechnologies, USA) overnight followed by rabbit anti-IgG-horseradish peroxidase secondary antibodies (Abcam, Cambridge, UK). Protein bands were detected with a light-emitting nonradioactive method using ECL reagent prepared in the laboratory 23 . The membranes were then subjected to autoluminography for 1 to 5 min.
Immunohistochemistry. The abdominal aortas from sacrificed animals were fixed in 4% paraformaldehyde for 12 h at 4 °C, then paraffin-embedded and sectioned at 5 µm. Sections of aortic tissues were deparaffinized, rehydrated, and washed with distilled water, 3% hydrogen peroxide to inactivate endogenous staining. Subsequently, sections were heated in 10 mM of citrate buffer (pH, 6.0), blocked with goat serum, and incubated sequentially with the primary antibody against UCP-1 (Cloud-Clone Corp. (CCC), USA) overnight at 4 °C and secondary antibody anti-rabbit (Abcam, Cambridge, UK) for 1 h at RT, respectively. Finally, the samples were incubated with 3,3′ diaminobenzidine (DAB) for coloration and counterstained with hematoxylin to stain the nuclei bluish. The positively stained cells showed a brown color. Photographs were taken using the Image-Pro Plus 5.1. Table 1. Gene primer sequences. Multiple comparisons were performed using a one-way analysis of variance (ANOVA) with appropriate post hoc analyses. Testing the significance of atRA on glucose and insulin levels before and after experiments were performed with two-way ANOVA. Values of p < 0.05 were considered statistically significant.

Results
The influence of the all-trans-retinoic acid treatment on insulin resistance. There was no difference in body weight between the vehicle-treated and the atRA-treated groups of Apo-E mice (Table 2). Similarly, there were no differences in body weight between subgroups of C57BL/6J mice the vehicle vs. atRA treated groups, respectively ( Table 2). Body weight was higher in the vehicle and atRA-treated Apo-E mice when compared to C57BL/6J mice (Table 2).
To define the influence of atRA on IR, glucose and insulin levels were measured in Apo-E and C57BL/6J mice before and after 8 weeks of treatment with atRA or vehicle and HOMA-IR (homeostasis model assessment of insulin resistance) was calculated ( Table 2). There was no significant difference in HOMA-IR between the vehicletreated and the atRA-treated group of Apo-E mice, as well as between the vehicle group and the atRA group of C57BL/6J mice ( Table 2). No significant differences were observed in Apo-E and C57BL/6J mice HOMA-IR before and after atRA administration ( Table 2).
The influence of the all-trans-retinoic acid treatment on cholesterol level. AtRA decreases the total and LDL/VLDL cholesterol level in Apo-E mice compared with vehicle-treated animals (Fig. 1). There was no significant difference in HDL fraction between those two groups. No significant differences were observed in total cholesterol and its fractions between AtRA-and vehicle-treated C57BL/6 J mice (Fig. 1).

Assessment of atherosclerotic lesion area.
To analyze atRA influence on the size of atherosclerotic plaques, we performed the Oil Red O staining for measurement of atherosclerotic lesions in the aortic root of mice. C57BL/6J mice did not develop atherosclerosis (Fig. 2). The total area of atherosclerotic lesions of the atRA group was 53% lower than those of the controls of Apo-E mice (Fig. 2).  In PVAT of Apo-E mice after 8 weeks of atRA administration, we observed increased expression of the key brown adipose genes: Ucp1, Cidea, and Prdm16 (Fig. 3A). We found a significantly higher mRNA level of Ucp1 in the atRA-treated than in the vehicle-treated C57BL/6J mice (Fig. 3A). Interestingly, there was no significant upregulation of pro-brown adipose tissue gene expression including Cidea, and Prdm16 in the atRAtreated C57BL/6J mice compared to the vehicle-treated C57BL/6J mice (Fig. 3A). In both Apo-E and C57BL/6J mice treated with atRA and in their counterparts, the vehicle-treated Apo-E and C57BL/6J mice, reduced size of adipocytes and induction of formation of beige adipocytes were demonstrated (Fig. 3B). Additionally, western blot analysis was used to determine the effect of the atRA treatment on the UCP1 protein level in PVAT. As shown, the UCP1 level was significantly higher in Apo-E and C57BL/6J mice treated with atRA than in their counterparts, the vehicle-treated Apo-E and C57BL/6J mice, respectively (Fig. 3C).
The all-trans-retinoic acid treatment increases adiponectin synthesis and secretion by perivascular adipose tissue in Apo-E mice. We tested if the atRA treatment results in altered production of adiponectin, an adipokine with anti-atherosclerotic properties, in PVAT in comparison with adiponectin concentration in visceral adipose tissue. Adiponectin concentration was significantly increased in PVAT of Apo-E group treated with atRA when compared to the vehicle-treated group of Apo-E mice (Fig. 4A). Moreover, there were no significant differences in adiponectin concentrations in visceral adipose tissue in all tested groups (Fig. 4B). After the atRA treatment, adiponectin mRNA level increased in PVAT of Apo-E mice when compared to Apo-E-mice treated with vehicle (Fig. 4C). Additionally, we measured adiponectin secretion to the medium after 24 h incubation from the same PVAT explants. The adiponectin concentration in medium of the atRA-treated Apo-E mice was increased as compared with medium of vehicle-treated Apo-E mice (Fig. 4D). No significant differences in adiponectin mRNA level, protein concentrations, and adiponectin secretion by PVAT were found between the atRA-and the vehicle-treated C57BL/6J mice (Fig. 4). www.nature.com/scientificreports/ TNF-α is a multifunctional inflammatory cytokine that may play a part in the pathogenesis of atherosclerosis. Therefore, we examined the effect of atRA on the TNF-α concentrations. There was no significant difference in TNF-α mRNA level (Fig. 5A) and protein concentration in PVAT (Fig. 5B) between the atRA-and the vehicletreated C57BL/6J and Apo-E mice. Moreover, no significant differences of TNF-α were observed in the medium concentrations after 24 h incubation of PVAT explants in all groups of mice (Fig. 5C).

All-trans-retinoic acid increase NO production in the abdominal aorta of Apo-E mice. Decreased
NO bioavailability, increased O 2 − production, and formation of peroxynitrite have been implicated in the pathogenesis of several cardiovascular diseases including atherosclerosis 24 . We measured NO and adiponectin levels in mice abdominal aorta to test whether atRA can increase their concentrations. We observed increased NO production in the abdominal aorta of the atRA-treated compared to the vehicle-treated Apo-E mice (Fig. 6A). There was no significant difference in NO level in the abdominal aorta between the atRA-and the vehicle-treated C57BL/6J groups (Fig. 6A). Moreover, there was no significant difference in adiponectin concentrations in all groups of mice (Fig. 6B).

Discussion
Experimental studies point out the beneficial effects of atRA on metabolic health, prevention of atherosclerotic cardiovascular disease (CVD), and browning of WAT 8,12,13 . Results of the present study showed that atRA ameliorates atherosclerosis in Apo-E mice. Moreover, we demonstrated that the atRA treatment stimulated browning of PVAT of the same mouse model. We also observed an increased synthesis and secretion of adiponectin by PVAT. Furthermore, we demonstrated that atRA increased NO concentration in the aorta of Apo-E mice compared to mice treated with vehicle. Animals are unable to synthesize vitamin A, the major sources for these compounds are formed by forty carbons carotenoids 25 . The key step in the formation of vitamin A (retinol) and its derivatives (retinal, retinoic acid) are oxidative cleavage of b-carotene 25 . The role of vitamin A in glucose and insulin metabolism has not been clearly defined. On one hand, the atRA treatment improved glucose tolerance 26 and suppressed the expression of leptin 5 and resistin 20 , adipokines linked to insulin resistance. Moreover, short-term high-dose administration of atRA in high-fat diet mice was associated with a reduction of body weight and an increased UCP-1 expression in WAT 13 . In contrast, in another study, it has been demonstrated that the administration of the RXR agonist did not affect glucose and insulin metabolism 27 . Similarly, in our study, the body weight and HOMA-IR were not significantly different among the groups. The finding of no significant differences in HOMA-IR among the groups suggests unalert insulin sensitivity. We observed an increase in body weight in Apo-E mice compared All-trans-retinoic acid (5 mg/kg/day) or vehicle (corn oil) were administered orally by plastic feeding tubes to Apo-E and C57BL/6J mice for 8 weeks. Data are shown as mean ± SEM of 5-6 mice per group. The results were analyzed by one-way ANOVA with Tukey's Multiple Comparison Test. *p < 0.05 ApoE atRa vs. ApoE vehicle. www.nature.com/scientificreports/ to C57BL/6J treated with both atRA and vehicle. A recent study has demonstrated a similar effect, Apo-E mice presented an increasing tendency towards weight gain and were heavier compared to their counterparts 28 . Epidemiological and experimental studies indicate that the antioxidant properties of both β-carotene and vitamin A have the potential to prevent lipoprotein oxidation, which is an initial factor in the development of atherosclerotic CVD 29 . It has been also suggested that AtRA have a potential the same properties 8,[29][30][31] . The protective effects of atRA against atherosclerosis induced by a high-fat diet were demonstrated in the rabbit model, and included inhibited platelet activation and inflammation 8 . This beneficial action was associated with decreased caveolin-1 expression and endothelin-1 secretion, enhanced eNOS activity, and production of NO 8,30,31 . However, the rabbit model does not mimic the human development of aortic lesions because of the different lipoprotein metabolism. Apo-E deficient mice are excellent alternatives to rabbits as models of atherosclerosis in developing lesions with morphological similarity to human atherosclerotic lesions 32,33 . According to data from the literature, the deletion of ApoE is sufficient to drive massive hypercholesterolemia and develop spontaneous lesions under a standard diet, not only on a high-fat diet 34 . Because we used the descending aorta of chow-fed Apo-E mice before significant atherosclerotic plaque develops, we gained insight into the early phase of the  www.nature.com/scientificreports/ atherosclerosis process. Interestingly, besides dyslipidemia, these mice develop widespread, fibrous plaque-like lesions at the vascular sites. Previous studies using genetically modified mice revealed novel information on sex dimorphism in the development of cardiovascular phenotype. Mice with Apo-E or low-density lipoprotein receptor deficiency and disruption of estrogen receptors (ER), aromatase or follitropin receptor (both lead to estrogen deficiency) have convincingly shown that estrogen and ER-mediated signaling are protective against vascular dysfunction and atherosclerotic lesions [35][36][37] . To avoid these hormonal differences, only male mice were used in our experiments.
With the use of the Apo-E model, it has been demonstrated that feeding with β-carotene can inhibit atherosclerosis 38 . In the current study, performed in the same animal model that consumption of a natural β-carotene enriched foods, composed of all-trans and 9-cis isomers, inhibited atherogenesis compared to the vitamin A deficiency diet, while vitamin A did not 39 . Recently, it has been shown that β-carotene and retinoic acid administration reduces the development of atherosclerosis, presumably by a reduction of the secretion of newly synthesized triglyceride and cholesteryl ester 40 . It has also been shown that 9-cis-RA stimulated macrophage cholesterol efflux and inhibited atherosclerotic plaque formation in Apo-E mice 41 . Similarly, in our study, we demonstrated that atRA significantly ameliorated atherosclerosis in the aortic root of Apo-E mice.
According to the literature, the RA administration increased multilocularity, expression of mitochondrial, thermogenic and fatty acid oxidation-related genes as well as retinoblastoma protein (pRb) phosphorylation in mice 13 . Vitamin A deficient diets led to increased body fat and decreased BAT UCP-1 content in mice, and both abnormalities were reversed by the atRA treatment 42 . RA-induced UCP-1 expression in BAT of mice was also demonstrated in other studies 42,43 . UCP-1 expression is regulated by binding specific elements in the UCP-1 enhancer like thyroid hormone receptor (THR), RA receptors (RAR and RXR), and the peroxisome proliferatoractivated receptor (PPAR) [43][44][45][46] . Our findings confirmed that atRA induced UCP-1 expression in both C57BL/6J and Apo-E mice treated with atRA. However, unexpectedly, we observed increased expression of key brown adipose genes like Cidea, and Prdm-16 only in Apo-E mice not in C57BL/6J. We hypothesized that these differences could be associated with the differences in gene regulation between C57BL/6J and Apo-E mice. PRDM16 is in down-regulated by a miRNA,miRNA-133 47 . Both the corepressor RIP140 and the PPAR coactivator 1α (PGC-1α) play key roles in the regulation of transcription of genes involved in energy homeostasis like CIDEA 48 . To our knowledge, we showed for the first time that the atRA administration induced PVAT browning in Apo-E mice.
It has been suggested that PVAT plays an important role in CVD through its direct local action on the vessels 17 . Herein, we investigated the influence of the atRA administration on browning of abdominal PVAT as it contains primarily WAT 1 . We also examined atRA influence on synthesis and secretion of anti-atherosclerotic and anti-inflammatory adipocytokine adiponectin, and pro-inflammatory TNF-α. The vasodilating effect of adiponectin secreted from PVAT could be achieved by multiple mechanisms. Adiponectin stimulates NO production from endothelial cells and adjacent adipocytes 3,49-51 . Adipose tissue-derived NO may activate large-conductance Ca 2+ -activated K + channels opening in VSMC and leads to vasodilation 52 . The change in NO bioavailability was suggested to be a major mechanism of endothelial dysfunction 53 . NO is an important protective molecule preventing vascular diseases, such as atherosclerosis [53][54][55] . We demonstrated in Apo-E mice that atRA increased adiponectin synthesis in PVAT. Additionally, high adiponectin concentration was secreted to medium after 24 h incubation of PVAT. We also measured adiponectin concentrations in visceral adipose tissue, but in this type of fat depots, we did not observe any significant differences among all studied groups of mice. Contrary to our results, Zhang et al. observed a decrease of adiponectin mRNA levels in WAT of mice treated with a single dose of atRA 56 . This effect was also confirmed in 3T3-L1 adipocytes 57 . These discrepancies may results from investigation of different depots of adipose tissue as well as different doses and administration time of atRA treatment.
Similarly, no differences were observed in adiponectin concentration in the aorta in all groups of mice. Furthermore, NO level increased in the aorta of the atRA-treated compared to the vehicle-treated Apo-E mice.
It has been demonstrated that adiponectin inhibited TNF-α-induced inflammatory response 4 . Additionally, atRA has been shown to exert immunomodulatory and anti-inflammatory functions in various cell types 58,59 . To our knowledge, so far, the effects of atRA on TNF-α have been investigated in vitro only 60 . It has been revealed that atRA downregulated NO synthase 2 and TNF-α expression, targeting the LPS/TLR4/NF-κB pathway in Colonic Mucosa 60 . In our present study, the atRA treatment did not affect the inflammatory marker, TNF-α, in Apo-E mice.
To sum up, the present study demonstrated that atRA ameliorates atherosclerosis in Apo-E mice. The AtRA treatment in ApoE mice also stimulates browning of PVAT and increases the synthesis of adiponectin. Increased NO concentration in the aorta suggests an important role of PVAT-derived adiponectin in the endothelial function of Apo-E mice.