Summary

Graphical abstract

Open in a separate window

Introduction

Adipose tissue is commonly categorized into three types with distinct functions, phenotypes, and anatomical localizations.1^,2^,3 White adipose tissue (WAT) is primary energy storage; brown adipose tissue (BAT) is responsible for energy dissipation during cold exposure; beige or brite (brown-in-white) adipose tissue, found interspersed in WAT, is characterized by a brown-like phenotype.4 These three types of adipose tissues also have endocrine functions and play important roles in systemic metabolic processes, especially in obesity and its comorbidities, such as cardiovascular diseases.4^,5^,6 The role of adipocytes, especially perivascular adipose tissue (PVAT), was first demonstrated in 1991.7 Their well-established non-uniform morphological structure surrounds blood vessels to support the cardiovascular system.6^,7 The morphology of PVAT is not always BAT in mice and humans,1^,4^,8 and its features depend on the regional location and heterogeneity of the vascular bed.1^,4 Unlike white fat, thoracic aortic PVAT has similar mitochondrial and metabolic-related gene expression and exerts thermogenic activity during cold adaptation, thereby regulating vascular tone, relaxation, or constriction.7^,9 As for the relationship between PVAT and metabolic disease and vasoconstriction, the anti-constriction ability of PVAT is reportedly attenuated when cardiovascular diseases, such as obesity, diabetes, or early atherosclerosis, are dysfunctional.5^,10^,11 PVAT is also an active endocrine organ that not only produces adipokine-derived relaxing factor (ADRF) but also induces the secretion of adipocyte-derived contracting factor (ADCF) to maintain homeostasis and contribute to the microenvironment of PVAT.6^,11 When the thoracic aortic PVAT releases ADRF or ADCF, ADRF functions like nitric oxide (NO), angiotensin 1–7 (Ang1–7), and adiponectin to dilate blood vessels and protect the vascular walls in hyperlipidemia, local hypoxia, and lipid deposition.12 Although ADCFs function similarly, such as AngII, prostaglandins, and norepinephrine (NE), they mediate vasoconstriction and maintain smooth muscle cell elasticity and efficiency.13

The activating transcription factor (ATF)-3 gene belongs to the ATF/cyclic adenosine monophosphate response element-binding (CREB) group and is a stress-response transcription factor.14^,15 ATF3 regulates the function of pancreatic β-cells and reduces the occurrence of diabetes,14^,15 promotes lipolysis, and attenuates metabolic dyshomeostasis caused by a high-fat diet (HFD).16 Previous studies have clarified that the pathogenic mechanism of diabetes is related to the abnormal expression of ATF3.15^,17 High-fat-induced diabetes and obesity models in mice with ATF3 deletion or deficiency have exhibited significantly reduced pancreatic β-cell function, decreased insulin secretion, and exacerbated metabolic dysfunction.15^,16 When the loss of ATF3 was restored by the adenovirus atf3 vector, glucagon expression in αTC-1.6 cells was significantly increased through a negative feedback regulation mechanism to improve blood glucose stability.15^,18 When ATF3 is activated, it can enter the nucleus and bind to DNA to inhibit CCAAT/enhancer-binding protein α expression and adipocyte differentiation.16 ATF3 overexpression inhibits not only the expression of adipogenesis-related genes but also carbohydrate-responsive element-binding protein (ChREBP) and lipid biosynthesis enzymes such as stearoyl-CoA desaturase (Scd-1), which promote lipolysis and fat browning.16 Based on gene polymorphisms19 and human gene expression databases to analyze clinically moderate to morbidly obese patients, ATF3 expression in the liver, adipose tissue, and muscle tissue has been found to be significantly lower than that in healthy subjects.16 It has been proven that ATF3 is related to obesity in humans.16^,19^,20

Although the biological function of ATF3 increases the resistance to obesity by regulating the metabolism of adipocytes and improving the metabolic abnormalities caused by an HFD,16^,17 the specific role of ATF3 in obesity-induced hypertension and vascular tone regulation remains to be thoroughly investigated. A critical role of the PVAT structure is that it not only modulates vascular homeostasis but also protects against the variation in blood pressure caused by cardiovascular dysfunction.6^,21 Previous reports have demonstrated that an increase in ATF3 activity significantly accelerates adipocyte tissue browning and lipolysis in HFD-induced obese mice.16^,17 Although accumulating evidence has led to our understanding of the role of PVAT in modulating vascular tone, its impact on vascular disease, especially on how ATF3 affects blood pressure control in obesity-mediated complications, remains to be clarified. To the best of our knowledge, this study is the first to demonstrate a protective function of ATF3 expressed in the thoracic aorta and to explore whether ATF3 in PVAT regulates the release of ADRF and ADCF to relax vascular smooth muscle cells (VSMCs).

Results

Specific heterogeneity of PVAT is essential for regulating vascular function

To reveal the importance of adipose tissue in different regions and phenotypes in vascular protection in response to pathophysiological conditions and vascular homeostasis, we first performed adipose tissue extraction experiments combined with blood vessel myography in wild-type (WT) obese mice. In strategy 1, inguinal WAT (iWAT) extract was administered at resting tone. Slight vasodilation was noted but with no significance compared with the control solvent (Figures 1A and 1D). Similar results were found for the suprascapular BAT extract, although the extract from suprascapular BAT did not cause vasodilation (Figures 1B and 1E). In strategy 2, inguinal WAT and suprascapular BAT extracts remained unchanged during PhE-induced vasoconstriction (Figures 1A, 1B, 1G, and 1H). In particular, PVAT extracts were given 1, 2.5, 5, and 10 μg/mL of adipose tissue concentration, which efficiently relaxed thoracic aortic rings from WT mice in a concentration-dependent manner (Figures S2). Thoracic aortic PVAT extract significantly induced resting tone vasodilation (Figures 1C and 1F) and further attenuated PhE-induced vasoconstriction in WT control mouse aortae (Figures 1C and 1I). We performed time course experiments to clarify further that thoracic aortic PVAT, but not inguinal WAT and suprascapular BAT, extracts exert vasodilatory effects. The results from Figure 1J demonstrated that thoracic aortic PVAT extract increased vasodilation over time, and the results are consistent with those shown in Figure 1F. Consequently, we draw a preliminary conclusion that only substances extracted from thoracic aortic PVAT, but not from inguinal WAT or suprascapular BAT, could modulate vasoconstriction or dilation.

Open in a separate window

Figure 1

PVAT possesses the potential for resting vasodilation and attenuating contractile responses induced by vasoconstrictors

This strategy was performed using inguinal WAT (iWAT) (A), suprascapular BAT (B), and thoracic aortic PVAT (C) extracts collected from WT mice fed a high-fat diet (HFD) for 22 weeks, removing PVAT and endothelial cells of the aortic arteries of control WT mice and placing arterial rings into the organ bath to assess vascular function. Representative tracing showing only the VSMCs of the aorta isolated from WT mice fed a normal diet (ND; A–C), the vascular response at the resting tone assessment for strategy 1 (D–F), and phenylephrine-induced vasocontraction (PhE; 1 μM) for strategy 2 (G–I). Determination of iWAT (A, D, and G), suprascapular BAT (B, E, and H), and thoracic aortic PVAT extracts (C, F, and I) of WT mice fed an HFD for recording the vasodilation effects at 22 weeks or the end of the experiments. PVAT extract-induced time-dependent vasorelaxation is summarized in J. Data are presented as mean ± SEM. Statistical comparisons between two groups were performed using unpaired t test (G‒I), one-way ANOVA in more than two groups (D‒F), and two-way ANOVA (J). p < 0.05, compared with the control solvent group in each experiment. ns: not significant. The dotted symbols in each column represent the number of mice examined. W: wash.

Hypertensive mice and patients both have reduced ATF3 levels in different organs

To explore the candidate genes that may mediate PVAT-induced vasodilation, we screened the human genomic database from the GEO. In the hypertensive disease model, the expression of ATF3 was significantly altered (Figures 2A and 2B). Compared with the control (normotension) group, the expression of ATF3 in the aorta and kidney segments (Figures 2A and 2B, respectively) was significantly lower in the hypertensive group, whereas ATF3 expression in the heart (Figure 2C) and liver tissues (Figure 2D) was not significantly different. We further evaluated the expression of ATF3 in human peripheral blood cells from healthy and hypertensive subjects (Figure 2E) and found that hypertensive patients had a marked decrease in ATF3 expression compared with healthy volunteer controls (Figure 2E). Based on the GEO results in the aorta and kidney, which plays a central role in regulating arterial blood pressure, ATF3 expression was associated with hypertension in both murine strains and human subjects.

Open in a separate window

Figure 2

Analysis of ATF3 expression level among aorta, kidney, heart, and liver from mouse models of hypertension and human subjects in the NCBI GEO datasets

(A–E) ATF3 expression levels in the aorta (A), kidney (B), heart (C), and liver (D) of normotensive and hypertensive mouse models. The expression of ATF3 in human peripheral blood cells was assessed, as shown in (E). The dotted symbol in each column represents the number of mice examined (A–D); the number of experiments in humans = 3. Statistical comparisons were performed using unpaired t test (A‒D). Data are mean ± SEM; p < 0.05 compared with the normotension (control) group in mice and the healthy group in humans. ns: not significant.

Abnormal blood pressure regulation and weight gain are associated with ATF3 deletion

To further examine whether ATF3 plays an essential role in regulating blood pressure in mice fed an ND or HFD, blood pressure was determined using a non-invasive tail-cuff system. Physiological parameters, pathological examination, and related traits in ATF3-null (ATF3^−/−) mice were determined for comparison. ATF3^−/− mice fed an HFD, but not a regular diet, showed a higher body area with moderate to severe obesity (Table 1). Food intake was not different between the diet-challenged groups (Table 1); however, the body weight changes were significantly increased in ATF3^−/− mice compared with their WT littermates (Figures 3A and 3B). In addition, the fasting glucose levels and hepatic steatosis significantly increased in ATF3^−/− mice when the HFD was administered for a long period of 22 weeks (Table 1). Surprisingly, both the systolic (Figure 3C) and diastolic (Figure 3E) blood pressures increased in ND-fed mice but were significantly exacerbated in HFD-fed mice (Figures 3D and 3F). The mean arterial pressure was significantly elevated in ATF3^−/− mice and was associated with weight gain (Figures 3G and 3B). Heart rates between the control WT and ATF3^−/− mice were not statistically different following the different diets (Figure 3H). The cohort, GEO analysis, and murine model results suggest that ATF3 is a pivotal regulator in hypertension disease.

Table 1

Physiological parameters and pathological examination of wild-type (WT) and ATF3−/− fed normal diet (ND) or high-fat diet (HFD) at the final gain

	ND		HFD
Groups	WT	ATF3−/−	WT	ATF3−/−
Body weight (g)	28.9 ± 0.6	28.4 ± 0.7	48.8 ± 0.7	52.4 ± 0.8a
Body width (cm)	3.0 ± 0.0	3.4 ± 0.1	4.1 ± 0.1	4.8 ± 0.1
Body length (cm)	9.0 ± 0.4	9.2 ± 0.1	11.2 ± 0.4	9.3 ± 0.8
Body area (mm)	2640.8 ± 12.9	2669.9 ± 9.9	3193.5 ± 15.9	3662.4 ± 19.6a
Food intake (g/wks)	21.4 ± 2.3	21.1 ± 2.1	28.6 ± 2.1	25.8 ± 2.4
Fasted glucose (mg/dL)	98.1 ± 4.6	131.0 ± 4.5a	127.3 ± 5.3	187.0 ± 10.2a
Hepatic steatosis	No observations	No observations	Yes (mild)	Yes (moderate)

Open in a separate window

Data are mean ± SEM.

Body weight, width, length, area, food intake, and fasting glucose in wild-type (WT) and ATF3^−/− mice fed a normal diet (ND) or high-fat diet (HFD) for 22 weeks. The hepatic steatosis score was estimated based on the histological examination. Data are mean ± SEM and p < 0.01 compared with the WT group fed ND or HFD in each experiment (n = 5 per group).

^ap < 0.01 indicates significant difference from the WT mice fed ND or HFD ingestion. n = 5 per each group.

Open in a separate window

Figure 3

Deletion of ATF3 aggravated high-fat diet (HFD)-induced body weight and blood pressure abnormality regulation in mice

ATF3-gene-deleted mice (ATF3^−/−) and their wild-type (WT) littermates were fed a normal diet (ND) or high-fat diet (HFD) for 22 weeks. Body weight measurements were performed after 22 weeks of ND (A) or HFD (B) feeding in both groups. Determination of systolic pressure (C–D) and diastolic pressure (E–F) of WT and ATF3^−/− mice fed ND (C–E) or HFD (D–F) for 3 days of tail-cuff recording at 22 weeks or the end of the experiment. G shows the mean arterial pressure (MAP) and heart rate (HR) represented in H. The number of experiments in body weight measurement = 14 (A–B); the systolic and diastolic pressure represented the dot symbol in each column representing the number of mice examined (G–H). Statistical comparisons were performed using one-way ANOVA (G and H) and two-way ANOVA (A‒F). Data are presented as mean ± SEM; p < 0.05 compared with the WT group in each experiment. ^#Indicates a statistically significant difference among the sets of curves by the generalized estimating equation (GEE) analysis (A and B). ns: not significant.

ATF3 deficiency alters the morphology and structure of PVAT in high-fat diet-induced obesity

Recent evidence has demonstrated that obesity causes inflammation and oxidative stress to promote increased PVAT mass and lipid droplet hypertrophy surrounding the aortic artery.11 WT mice fed an HFD showed a decreased ATF3 expression (Figure S3). We found that ATF3^−/− mice fed a long-term HFD, but not an ND, showed increased PVAT mass and density expansion, significant lipid deposition, and fat distribution in the surrounding PVAT compared with their WT littermates (Figure 4A). HFD-fed ATF3^−/− mice showed a markedly increased PVAT area (Figure 4B), width (Figure 4C), and all PVAT tissue weights (Figure 4D) compared with WT mice fed a regular diet and HFD, respectively (Figures 4A–4D). A histological examination further determined the morphological changes in the aortic specimens. The inner diameter (Figures 4E and 4F) and media area (Figures 4E and 4G) showed no differences between the WT and ATF3^−/− mice fed the ND or HFD for 22 weeks (Figures 4E–4G). The adventitia area (Figures 4E and 4H) and A/M ratio were significantly increased in HFD-fed ATF3^−/− mice (Figure 4I). The high A/M ratio indicated noticeable early fibrotic changes in the adventitia of vessels and PVAT in ATF3^−/− mice (Figure 4I). In contrast, fibrosis-like pathological changes were not observed in the WT littermates (Figure 4I). Despite excessive perivascular adipose accumulation and increased fat mass, the blood vessel diameter and media remained unchanged compared with the WT littermates fed the HFD (Figures 4E–4G). Next, we calculated the PVAT adipocyte cell size in each group (Figures 4E, 4J, and 4K) to demonstrate that ATF3 deletion in mice aggravated HFD-induced (Figure 4K), but not ND-induced, adipocyte hypertrophy (Figure 4J). The analysis showed enlarged lipid sizes and estimated fat to the hypertrophy stage (Figures 4E, 4J, and 4K). HFD-fed ATF3^−/− mice showed a marked phenotypic conversion of average to enlarged PVAT adipocyte cell size (Figure 4K). This implied that vascular diameters or media formation were unaltered in HFD-fed ATF3^−/−mice (Figures 4E–4G); however, the pathological changes in PVAT adipocyte phenotype may have contributed to the susceptibility of arteries to obesity (Figures 4E, 4H, and 4I). These data confirm that the presence or absence of ATF3 ablation in mice modulates PVAT genetic programming, in turn affecting WAT/BAT balance in HFD-fed mice, leading to pathological changes.

Open in a separate window

Figure 4

ATF3-deficient mice indicated PVAT pathological conversion after HFD feeding

The characterization of thoracic aortic PVAT morphology and phenotype in wild-type (WT) and ATF3^−/− mice subjected to a normal diet (ND) and high-fat diet (HFD). To compare all the pathological changes caused by HFD induction, mice fed an ND (4% kcal from fat) were used as controls. Representative image illustrating blood vessels with PVAT structure isolated from the thoracic artery of mice (A). Lower magnification shows the entire vascular system with a standard scale, indicating that HFD-fed mice caused the PVAT morphology to be significant in irregular and fat hypertrophy in ATF3-deficient mice (A). Quantification of the thoracic aortic PVAT area (B), width (C), and tissue weight (D) was performed after 22 weeks of ND or HFD feeding for both groups. Representative hematoxylin and eosin Y staining of blood vessels with PVAT structure (E), and further measured the inner diameter (blue dotted line, F), media area (yellow dotted line to blue dotted line, G), and adventitia area (red dotted line to yellow dotted line, H) in both WT and ATF3^−/− mice. The gray box shows the zoomed-in capture location of the image (E). The effects of ND and HFD feeding strategies were assessed by the pathological examination of the cross-sections of mouse aortas, as shown in the enlarged picture in E, and further quantified the A to M (A/M) ratio as collagen deposition summarized in I. Lipid droplets were classified into three groups according to diameters of <20 μm, 20–40 μm, and >40 μm. Average proportions were calculated from 100 lipid droplets in each microscopic field. The data were obtained from six field/mouse in both WT and ATF3^−/− mice fed ND (J) and HFD (K). A, adventitia; L, lumen; M, media. Scale bar: 50 μm for hematoxylin and eosin Y staining. Statistical comparisons were performed using one-way ANOVA. Data are presented as mean ± SEM. p < 0.05 compared with the WT group in each experiment. ns: not significant. The dotted symbols in each column represent the number of mice examined.

Global ATF3 deficiency is responsive and more sensitive to low concentrations of stimulated vasoconstrictors

To determine whether the deletion of ATF3 alters vasoconstricting agent-stimulated contractility, which is enhanced in the aortic arteries of obese mice, we first investigated the vascular reactivity of VSMCs in healthy WT and global ATF3^−/− mice fed an ND and a 60% kcal from fat to test the contractile response. Both PVAT and endothelium of the aortic rings were completely removed before myography. In both groups fed the ND, the contractile responses in isolated VSMCs from ATF3^−/− mice to a cumulative KCl concentration did not significantly differ compared with vessels from WT mouse VSMCs (Figures 5A and 5B). KCl-induced vasoconstriction, regardless of the concentration-induced tracing curve or tension induced by the maximal concentration of 120 mM, did not differ between the WT and ATF3^−/− mice (Figures 5B and 5C). In contrast, low concentrations of KCl in age-matched ATF3^−/− mice fed an HFD led to significantly increased vasoconstriction (Figures 5D and 5E). However, high concentrations of KCl did not substantially affect the tension of KCl-induced maximal contraction (Figure 5F), which was consistent with the results shown in Figure 5C. In parallel, we further investigated the concentration-response curve to the thromboxane A2 analog U46619 for isolated aortic arteries with VSMCs (Figures 5G–5L). There was no difference in U46619-induced concentration-dependent vasoconstriction between the WT and ATF3^−/− mice fed an ND (Figures 5G and 5H). In contrast, in ATF3^−/− mice fed an HFD, a low concentration of U46619 led to significantly increased vasoconstriction (Figures 5J and 5K). Similar results were found in absolute tension at 1 μM; the maximal concentration induced by U46619 showed no differences in either WT or ATF3^−/− mice fed an ND or HFD (Figures 5I and 5L). Furthermore, we tested PhE, an α1-adrenergic receptor agonist, to induce VSMC contraction in WT and ATF3^−/−mouse arteries. There was no significant difference in the PhE-induced vascular tone (Figure 5M). This implied that the deletion of ATF3 did not affect KCl-, U46619-, or PhE-induced vasoconstriction in mice fed a regular diet. However, these effects were more sensitive and observed in mice fed an HFD, especially in ATF3-deficient conditions followed by low concentration-induced contractions. In addition, the endothelium-independent relaxation of SNP led to no significant differences in the aortic rings from all groups fed the HFD (Figures 5N and 5O). The results showed that the defect of the ATF3 gene in smooth muscle cells did not affect the maximum vascular tone but increased the sensitivity of blood vessels to low-dose vasoconstrictors. Thus, our data suggest that the presence of ATF3 is important for VSMC-regulated vasoconstrictor-induced contractions during the initial phase of obesity.

Open in a separate window

Figure 5

The vascular reactivity in VSMCs is hyperresponsive to vasoconstrictor stimuli in aortic rings isolated from ATF3 deficiencies in HFD feeding

To determine vascular smooth muscle responses to vasoconstrictors, both PVAT and endothelium of the aortic rings were wholly removed before myography. Challenges in normal diet (ND) and high-fat diet (HFD) ingestion in wild-type (WT) and ATF3^−/− mice preserved only in vascular smooth muscle cells (VSMCs only; EC and PVAT were denudated) in response to vasoconstrictors potassium chloride (KCl; A‒F), U46619 (G‒L), and phenylephrine (PhE)-induced vasoconstriction (M). Representative tracing showing only the VSMCs of the aorta isolated from WT and ATF3^−/− mice fed normal chow (A) or HFD (D) was induced by a cumulative-dose KCl (7.5 mM–120 mM). In B and E, the percentage (%) of contraction was normalized to the changes observed in response to treatment with 120 mM KCl. The maximal contraction to KCl was analyzed and represented by the absolute tension (T) unit in grams (g) (C and F). Following similar procedures, representative tracing in G and J showed U46619-induced concentration-dependent curve responses in ND (H) and HFD (K)-fed mice. The maximal contraction induced by U46619 represented the absolute tension (T) unit as gram (g), shown in ND feeding (I) or HFD feeding (L). The α1-adrenergic receptor agonist phenylephrine (PhE) at 3 μM was administered to induce constriction for comparison between WT and ATF3^−/− mice in the HFD challenge (M). Compared with WT littermates, sodium nitroprusside (SNP) led to a significant reduction in PhE-induced contraction in the original tracing (N); however, there was no difference in either group (N and O). The dot symbol in each column represents the number of mice examined (C, F, I, L, and M). Number of experiments in SNP-induced relaxation = 5 (O). Statistical comparisons were performed using unpaired t test (C, F, I, L, and M). Figures B, E, H, K, and O were performed by two-way ANOVA. Data are mean ± SEM; p < 0.05 compared with the WT group in each experiment and single concentration application of the concentration curve. ^#Indicates a statistically significant difference among the sets of curves by generalized estimating equation (GEE) analysis. ns: not significant. W: wash.

PVAT from ATF3-deficient mice potentiates and aggravates the constriction ability of aortic rings in obese mice

As shown in Figure 5, we observed a marked increase in vasoconstrictors with KCl- or U46619-induced vascular contractility in ATF3^−/− mice fed an HFD compared with the WT controls. However, these effects did not change in the group with normal dietary consumption. We hypothesized that PVAT might regulate agonist-stimulated contractions, enhanced in HFD-induced obesity in mouse aortic rings. In the perivascular adipose-intact aortic rings (+PVAT/VSMC) of control WT littermates fed the HFD, cumulative concentrations of NE significantly caused vasoconstriction. This NE-induced contractility, however, was markedly enhanced in ATF3-deficient aortic rings with PVAT (Figures 6A and 6C). The removal of PVAT increased NE-induced contractions in both groups of aortic rings (Figures 6B and 6D). Furthermore, contractions in response to NE in ATF3^−/− VSMCs were significantly higher than that in WT mice at high concentrations (0.03 nM‒1 μM) of NE induction (Figures 6B and 6D). The NE-induced maximal contractions in rings with PVAT were slight in WT mice (Figures 6A and 6E); however, they were markedly increased in ATF3^−/− mice presenting with PVAT, which was indicative of the anticontractile effect of PVAT in the WT littermates, but not in the ATF3-deficient mice (Figures 6A and 6E). Our findings of potentiating contractions induced by NE and NE-induced maximal contractions, as represented by tension, were significantly enhanced in ATF3^−/− mice fed an HFD (Figures 6A and 6E). Pearson’s equation was used to correct the data and evaluate the ATF3^−/− and WT control mice (Figure 6F). We found that increased PVAT weight was associated with a significant increase in NE-induced vasoconstriction in the WT mice (Figure 6F). In ATF3 deficiency, PVAT weight was positively correlated with NE-induced contractions; however, the differences were not statistically significant (Figure 6F).

Open in a separate window

Figure 6

ATF3 deficiency in PVAT and VSMCs enhances norepinephrine-induced vascular contractility on HFD feeding

Representative tracing showing concentration-dependent contractions of norepinephrine (NE) in thoracic aortic arteries with PVAT (A) and removal of PVAT (B) from wild-type (WT) littermates and ATF3-deficient (ATF3^−/−) mice 22 weeks after HFD-induced obesity (A and B). The concentration-response curve shows the logarithm of NE-induced vasoconstriction in the presence of PVAT (C) and the absence of PVAT is referred as VSMC only (D). (E) The quantitative analysis of the constriction tension (T) of isolated aortic rings in response to maximal concentration of NE (1 μM). +PVAT represents preserved with PVAT structure; −PVAT represents the removal of the PVAT structure shown in panel E. Statistical analysis of correction in thoracic aortic PVAT weights and NE-induced contractility in both groups subjected to the HFD (F). The response to treatment with 120 mM KCl was considered 100% contraction. The percentage (%) of NE-induced contractions was normalized to that of KCl. The number of experiments in the NE-induced contraction was six (C and D). Dotted symbols in each column represent the number of mice examined (E). A statistical comparison was performed using one-way ANOVA in E and two-way ANOVA in C and D. Data are mean ± SEM; p < 0.05 compared with the WT group in each experiment and single concentration application of the concentration curve. ^#Indicates a statistically significant difference among the sets of curves by generalized estimating equation (GEE) analysis. SNP: sodium nitroprusside.

ATF3-deficient PVAT is less effective in buffering vasoconstrictor serotonin and attenuating the anticontractile ability in obese HFD-fed mice

To test the hypothesis that PVAT influences buffering activity against vascular contraction, we examined the dose-response curve of 5-HT for aortic arteries with PVAT (+PVAT) and without PVAT (−PVAT). As expected, WT and ATF3^−/− vessels with PVAT constricted significantly less than those without PVAT; however, 5-HT-induced vasoconstriction was markedly augmented in ATF3^−/− mice (Figures 7A and 7B). In addition, WT mice were fed an HFD for 22 weeks, and contraction in response to 5-HT at low concentrations was significantly lower compared with ATF3-deficient mice when the longitudinal removal of PVAT was performed (Figures 7A and 7C). In the VSMCs of ATF3^−/− mice, the sensitivity to low concentrations of 5-HT-induced contractions was substantially greater than that in WT mice (Figures 7A and 7C). These findings were consistent with previous observations in the arterial response to U46619 and KCl. However, the loss of ATF3 in VSMCs was substantially sensitive to vasoconstrictors when mice were fed an HFD. We further calculated the area under the curve (AUC) of the concentration-response curve as arbitrary units (AU). As shown in Figure 7D, the measured total area of the 5-HT-induced contractile pattern showed that the whole area level in WT mice was significantly lower than that in ATF3^−/− mice in the presence of PVAT (Figures 7D and 7E). In contrast, responses to 5-HT produced by arteries from WT and ATF3^−/− mice without PVAT were not significantly different in the contractile patterns and levels of the total area in VSMCs alone (Figures 7F and 7G). To further investigate the buffering activity of PVAT in the vasoconstriction response, we overlapped the 5-HT-induced contraction profiles to analyze the vascular buffering capacity in arteries with and without PVAT. The dotted line area represents the capacity of vascular buffering for vasoconstriction and the Δabc value obtained by subtracting the total area of VSMCs from PVAT (Figures 7H and 7I). In the WT control, the value of Δabc of the buffering capacity to 5-HT-induced contraction was significantly higher when ATF3 was present in PVAT, which was indicative of an anticontractile effect of PVAT in WT mice (Figures 7H and 7I). However, this anticontractile ability was markedly decreased in the absence of ATF3 in mouse PVAT (Figures 7H and 7I). These results indicated that ATF3 might play an epigenetic role in PVAT, resulting in anticontractile effects via VSMC under HFD conditions.

Open in a separate window

Figure 7

ATF3-deficient PVAT failed to elicit an anticontractile effect

Representative trace in (A) showing concentration-response curves to serotonin (5-HT) in intact(+) thoracic aortic PVAT preparations and aortic rings without EC and PVAT (VSMC only). Quantitative analysis of constriction percentage (%) of aortic rings in response to 5-HT on vessel rings with PVAT (B) or without PVAT (C). Aortic rings with PVAT (+PVAT/VSMC) in (D) and only preserved vascular smooth muscle cells (VSMCs) in F from WT and ATF3-deficient (ATF3^−/−) mice were exposed to an increasing concentration of 5-HT. Each bar graph inset in (D and F) represents the percentage of contraction when each concentration of 5-HT was applied to the organ chamber. The total area under the curve (AUC) of the concentration-response curve is expressed as an arbitrary unit (AU) in E and G. Overlapping the whole area profiles in arteries with PVAT and preserved VSMC in both groups for comparison (H), summarized data from H are expressed as Δabc in I. The number of experiments on 5-HT-induced contractions was 10 (B and C). The dot symbol in each column represents the number of mice examined (E, G, and I). Statistical comparisons were performed using unpaired t test in E, G, and I and two-way ANOVA in B and C. Data are mean ± SEM; p < 0.05 compared with the WT group in each experiment and single concentration application of the concentration curve. ^#Indicates a statistically significant difference among the sets of curves by generalized estimating equation (GEE) analysis. ns: not significant.

ATF3 regulates vascular reactivity via the PVAT-mediated release of ADRF and ADCF

To clarify the molecular mechanisms underlying the regulation of the PVAT anticontractile response and decreased sensitivity of vasoconstrictor-induced contraction in aortic rings by ATF3, we hypothesized that ATF3 might regulate ADCF or ADRF release in HFD-fed mice. Accordingly, adipokines secreted from adipose tissue may have systemic effects and shift to pro-inflammation as adipose tissue expands during the development of obesity.22 Based on our results, we tested whether significant obesity-induced changes in PVAT-derived relaxation and contraction factors balanced affected vessel contractility. We used a protein array analysis and ImageJ quantification to determine the factors involved in PVAT regulation.23 The protein array analysis of adipokines (Figure 8A) revealed higher plasma levels of monocyte chemoattractant protein-1 (MCP-1) (Figure 8B), C-reactive protein (Figure 8C), and interleukin (IL)-6 (Figure 8D) in ATF3^−/− mice than in WT mice. The plasma levels of resistin (Figure 8E), PAI-1 (Figure 8F), and adiponectin (Figure 8G) did not differ between groups. Parallel experiments with immunoblotting (Figures 8H, 8I, 8J, and S8) and RT-PCR (Figures 8K and 8L) were performed. Further evidence showed that adipokine ADCF and ADRF secretion, including ADCF MCP-1 protein (Figure 8H) and mRNA (Figure 8K) expression, was higher in the PVAT tissue of ATF3^−/− mice than in WT mice (Figures 8H and 8K). We found no changes in ADRF adiponectin protein production (Figure 8I), but there was a significant difference in the mRNA levels in PVAT (Figure 8L). We determined whether the downregulation of adipoR1 in VSMCs was worse than adiponectin-mediated vasorelaxation. Notably, protein levels of adipoR1 in VSMCs were markedly reduced in ATF3^−/− mice fed an HFD (Figure 8J). MCP-1 and adiponectin expression in PVAT were examined using double-immunofluorescence staining. Fluorescence intensity in the PVAT surrounding the vessel was quantified and indicated that the ADCF MCP-1 release and overexpression in ATF3^−/− aortic rings with PVAT were much stronger than in WT mice (Figures 8M and 8O). We found that ATF3^−/− mice fed an HFD showed significantly increased MCP-1 production and release from the PVAT. However, the lack of ATF3 did not affect adiponectin levels in PVAT from both groups fed the HFD (Figures 8M and 8P). AdipoR1 expression was co-stained with the smooth muscle cell marker SM22 in the medial layer of the vessel, indicating that the adipoR1 fluorescence intensity was significantly decreased in ATF3^−/− mice (Figures 8N, 8Q, and 8R). Adiponectin expression in both rings with PVAT was not significantly different (Figure 8P); the expressed adipoR1 in VSMCs was markedly reduced in ATF3^−/− mice (Figure 8Q). In contrast, the levels of circulating and tissue location with MCP-1 and decreased adipoR1 in VSMCs, which may be related to vascular tone abnormalities in diet-induced obesity, were significantly higher in ATF3^−/− mice (Figures 8O and 8Q). We, therefore, propose that the mechanisms of PVAT in the regulation of vascular homeostasis involve the ATF3-mediated regulation of ADCF MCP-1 but not ADRF adiponectin expression and secretion from adipocyte tissue. Our findings prove that ATF3 protects against high vasoconstrictor-induced sensitivity to contractions. Furthermore, ATF3 function in PVAT significantly improved vascular hyperreactivity and the loss of anticontractile ability in HFD-induced vascular dysfunction. Therefore, these effects of ATF3 regulation may be beneficial in reversing obesity-related vascular complications, such as hypertension (Graph abstract).

Open in a separate window

Figure 8

ATF3 influences vascular reactivity via obese PVAT-mediated ADCF MCP-1 release and downregulated adiponectin receptor expression in VSMCs

Protein levels of adipokine and inflammation-related expression in wild-type (WT) and ATF3^−/− mice of thoracic aortic PVAT after HFD feeding for 22 weeks by adipokine assays (A), and the ImageJ analyzer software was used for quantification of densitometry of blots in B‒G. Protein levels of MCP-1 (H) and adiponectin (I) in PVAT tissue and adipoR1 in VSMC (J) were examined by immunoblot assay and summarized in a bar graph presented as the relative densities (%) for each dependent experiment in both groups. Actin was used as the internal control. Bar graphs in K and L quantify RT-PCR analysis of mRNA levels of MCP-1 (K) and adiponectin (L) in PVAT of WT and ATF3^−/− mice. Representative double-immunofluorescence (IF) images of MCP-1 (red IF) and adiponectin (green IF) in WT and ATF3^−/− mice (M). In N, images of adipoR1 (red IF) and SM22 (green IF) are shown. The yellow line indicates the Z-plot measurement of fluorescence intensity. The fluorescence intensity changes are shown in O for MCP-1, P for adiponectin, Q for adipoR1, and R for the SM22 VSMC marker for comparison. The fluorescence intensity was determined from six field/mouse (n = 3) in both WT and ATF3^−/− mice fed an HFD (O‒R). Scale bar for image: 100 μm; enlarged picture illustrates scale bar: 50 μm. Arrow indicates MCP-1 expression of PVAT in panel M and adipoR1 expression of VSMC in N. The dotted symbols in each column represent the number of mice examined. Statistical comparisons were performed using unpaired t test. Data are presented as mean ± SEM; p < 0.05 compared with the WT group in each experiment. ns: not significant.

Discussion

To the best of our knowledge, this study is the first to provide substantial evidence that the functional regulator ATF3 in PVAT, but not BAT or WAT, is essential for physiological homeostasis in blood vessels, contributing to regulating blood pressure. We used traditional whole-body knockout mice and generated recombinant AAV carrying ATF3 to restore the original function of the ATF3 gene. Mice lacking ATF3 gene expression revealed hypertension, vascular tone regulation abnormalities, and potentiated KCl- and U46619-induced contractility in VSMCs.

ATF3 deletion resulted in a lack of anticontractile ability in PVAT, enhanced sensitivity to vasoconstrictors, and increased ADCF MCP-1 release. These results suggest that ATF3 contributes to vascular variations in HFD-induced obesity and related vascular tone regulation abnormalities. This study identified ATF3 as a promising modulator of obesity-induced cardiovascular complications.

ATF3 is an adaptive response transcription factor widely expressed in various organs, such as the aorta, kidney, liver, and heart.16^,20 This is in accordance with the NCBI data in Figure 2. However, the specific role of ATF3 in regulating blood pressure remained unclear. Thus, we propose that the global deletion of ATF3 may cause aggregate vasoconstriction and that mice have hypertensive features, even in the absence of an HFD. This is supported by data obtained from ATF3^−/− mice in our study, which displayed increased systolic and diastolic pressures with increased mean arterial pressure (Figure 3). Conversely, Chou et al. previously demonstrated that in ATF3-deficient mice with fructose-induced metabolic syndrome, systolic blood pressure was not significantly elevated in either WT or ATF3^−/− mice.24 Notably, different model-induced diseases, diet sources, and experimental periods may yield differences in blood pressure measurements.25 Another possibility is that ATF3 induces chronic adaptation to modulate stress-induced vascular dysfunction.14 The model of HFD-induced obesity or age-dependent vascular abnormality for studying ATF3 function depends on the duration of the experiment. We ruled out changes that may have occurred if the feeding duration was shorter than 22 weeks, as both the systolic and diastolic blood pressures showed trends toward an increase in both HFD groups and were age-dependent in the standard diet groups. However, it is clear if the knockout of ATF3 significantly affects systemic blood pressure regulation and the morphological appearance of body fat distribution. In a human study based on data from the NCBI, hypertensive patients showed decreased ATF3 expression (Figure 2).26 This result supports our finding of the correlation between ATF3 expression and hypertension.26 In addition, this is consistent with previous studies that used microarray analyses of other mouse strains to induce hypertensive disease, which indicated that ATF3 expression in the aorta and kidney is involved in the progression of hypertension.27 Our results support the hypothesis that ATF3 deletion exacerbates blood pressure variation and increases contraction responses to contractile stimuli. A previous study demonstrated that IL-18-deficient mice exhibited blood pressure abnormalities.28 However, there were no unusual morphological differences or apparent infiltrating cell accumulation in the tissues.28 The pathological mechanisms affecting blood pressure include an enlarged heart, thickened vascular wall with fibrotic changes, and endothelial dysfunction.28 We revealed that ATF3^−/− mice had impaired blood pressure regulation and vascular dysfunction of the aorta, including in VSMCs and PVAT. Changes in the regulation of blood pressure may be related to altered PVAT function and morphological features, but not changes in the heart. Other possible mechanisms of action in the aorta of ATF3^−/− mice include impaired EC function (verified by ACh-induced relaxation) and related molecular target AMPK-mediated eNOS enzyme expression. The possible role of the endothelial function remains to be investigated. Our results indicate that both systolic and diastolic blood pressure in ATF3^−/− mice were higher than in their WT littermates, even though there was no apparent PVAT abnormality or differences in PVAT anticontractile ability in ND-fed mice for 22 weeks. Therefore, future studies are needed to elucidate the detailed mechanisms of increased blood pressure in ATF3^−/− mice. The renin-angiotensin system (RAS) control of blood pressure regulation has been well established.29 A published result from Chang et al. demonstrated that brown adipocyte-selective Bmal1-deficient (BA-Bmal1-KO) mice were mediated by angiotensinogen expression and increased release of AngII from PVAT, which contributed to the regulation of vasoactivity and blood pressure during the resting phase.29 In our study, however, we did not examine the function of RAS in regulating blood pressure, nor did we propose that the level of AngII in ATF3^−/− mice when fed normal chow was higher than that in WT mice. We cannot rule out that the RAS system mediates angiotensinogen expression and that AngII levels in the PVAT of ATF3^−/− mice may be significantly increased. Therefore, the possibility that ATF3 also regulates the angiotensinogen, AngII, and AT2 receptor pathways, including Bmal1 gene function in PVAT, warrants further investigation.

The metabolic balance of adipose tissue in the lipolysis and lipogenesis pathways involved in regulating blood pressure and vascular homeostasis by ATF3 is poorly understood. Evidence has shown that increased visceral fat due to metabolic disorders is associated with increased cardiovascular diseases, such as hypertension.30^,31 Moreover, the accumulation of visceral white adipose fat due to obesity may result in a stressed and dysfunctional state that leads to the release of proinflammatory mediators, including tumor necrosis factor (TNF)-α and IL-6. These results are significant complications in the cardiovascular system.9^,10 Based on previous studies, ATF3 has anti-inflammatory and protective effects in various tissues, including the kidneys, heart, and blood vessels.16^,17^,32^,33^,34 According to a previous study, ATF3 inhibits obesity-induced inflammation by targeting adipocytes.16 Here, we found that PVAT lipid composition and droplet size showed significant changes in ATF3^−/− mice fed an HFD. Morphological analyses revealed no differences in PVAT between WT and ATF3^−/− mice fed an ND. The lack of ATF3 expression significantly promoted PVAT morphological abnormalities and increased its adipocyte size and adventitia fibrosis in HFD feeding. Our results are consistent with those of Cheng et al., suggesting that ATF3 may affect the function of adipocytes,16 especially the PVAT structure, to regulate vascular homeostasis. The possible involvement of the protection of vascular dysfunction by controlling PVAT buffering activity supports our hypothesis that may underlie the ATF3-mediated PVAT morphological alternation mechanisms.

One of the findings of this study was the functional effect of ATF3 on the regulation of vascular tone by PVAT in different diet challenges, especially the HFD. We further investigated whether the anticontractile ability of aortic PVAT is lost in ATF3-deleted mice, which is due to the enhanced generation of ADCF MCP-1 release by the PVAT. Our results demonstrated that the protective effect was impaired by the loss of ATF3 in HFD-induced obesity, suggesting that PVAT plays a protective regulator role in aortic contractions. Similar results were found for the PVAT extract switching. WT PVAT extracts induced significant vasodilation. However, the PVAT extract from ATF3^−/− mice caused slight vasodilation in WT aortic rings (Figures S4A and S4B). Thus, it was considered that the harmful effects on the ATF3^−/− mouse aorta were caused in an additive manner by the release of ADCF and imbalance of ADRF. We previously discussed that the altered PVAT function in ATF3 deficiency appeared to be related to morphological changes. The loss of PVAT function was associated with the increased arterial tone, with an essential role of ATF3 in regulating these processes. This is further supported by our observation that obese ATF3^−/− mice were exacerbated with vasoconstrictor agents. We found that WT PVAT significantly attenuated the vascular responsiveness of aortic arteries to contractile agents, including NE and 5-HT. To reach the direct conclusion that ATF3 plays a role in PVAT, a precise mechanism of action of ATF3 in PVAT will be delineated in future studies using a CRISPR knockout experiment model, siRNA knockdown assay in an ex vivo model, and blood-vessel myography assay.

In parallel, ATF3^−/− mice without PVAT were more hyperpolarized in the VSMC membrane by activating the potassium channel activator KCl. The mechanism has been described in Figure 5, where potassium channels are primarily involved only in HFD feeding. Based on previous reports, the potassium channels Kv and K_ATP could contribute to blood pressure elevations in obesity-related hypertension.35^,36 In addition, PVAT from BK_Ca knockout mice showed decreased anticontractile activity and attenuated adiponectin release.37 We cannot rule out that VSMCs and PVAT BK_Ca channels mediate the process of PVAT-derived relaxation abolished by ATF3 deletion. Therefore, the thromboxane A₂ analog U46619 induces vascular constriction, which depends on the resistance to opening the potassium channel of VSMCs.36 Low concentrations of U46619 significantly increased vasoconstriction in HFD ATF3^−/− mice. These results were similar to those for KCl, indicating that the difference in the contractile response to KCl and U46619 between VSMCs and PVAT is involved in potassium channel regulation.35^,38 The electrophysiological mechanisms underlying ATF3 involved in different types of potassium channel activation require further investigation.

PVAT releases NO, adiponectin, H₂S, and methyl palmitate as ADRFs to regulate blood pressure.6^,21^,39 Studies in animals and humans have also suggested that ADCFs, such as AngII, reactive oxygen species (ROS), and MCP-1 are associated with hypertension.6^,21 Currently, direct evidence from animal models to investigate the anticontractile or contractile roles of PVAT in blood pressure regulation is well investigated. Body fat and renal function are further associated with obesity-induced hypertension,3^,31^,40 and ATF3 might be involved in those organ regulations. The mechanisms of ATF3 deficiency in blood pressure regulation and the maintenance of vascular homeostasis in cells and tissues are also complicated and, thus far, unclear. We used the obese model of a mouse lacking ATF3 expression in the global system and reported that ATF3^−/− mice are more likely to have high blood pressure following a long-term ND and HFD feeding. Our present results further demonstrate that the loss of anticontractile ability to NE and buffering activity to vasoconstrictor 5-HT stimulation in ATF3-deficient mice is due to alterations in the structure and morphology of PVAT; moreover, the enhanced early fibrosis parameter A/M ratio to the absence of ATF3 regulation in the VSMCs when HFD diet challenged. Because ATF3 deletion in VSMCs augmented with low-concentration contractions to KCl and U46619, it is also likely due to an increased sensitivity of the medial VSMCs vasoconstrictor application. A comparison of adiponectin expression in the PVAT of 22-week-old WT and ATF3-deficient mice allowed us to conclude that the ADRF adiponectin levels in both groups had no significant difference but markedly declined in adipoR1 expression in ATF3^−/− mice on HFD feeding. The lack of change in gene or protein production does not rule out changes in ATF3-mediated adiponectin secretion.

Our results are consistent with those of previous studies reporting that the PVAT-mediated anticontractile effect is lost and impaired in obese models.6^,11 The mechanism seemingly involves the reduced release of the relaxing factors or the release of constricting factors that contribute to the loss of the anticontractile effect of PVAT. To further explore the mechanism of the altered function of ATF3 in PVAT, we measured critical adipokines, including adiponectin, NO, MCP-1, and IL-6, in WT and ATF3^−/− mouse PVAT extracts. We found that the protein array analysis significantly elevated ADCF MCP-1 (Figure 8B) release and the proinflammatory adipokines C-reactive protein (Figure 8C) and IL-6 (Figure 8D) in ATF3^−/− mice. However, unlike MCP-1, C-reactive protein and IL-6 did not induce significant vasoconstriction in our previous findings. We further explored proteomic and KEGG pathway map analysis, and our preliminary data show ATF3-conducted PVAT function and that the PPAR signaling pathway might be involved in blood pressure regulation (unpublished results). We did not find any significant differences in the release and expression of ADRF adiponectin. Interestingly, adipoR1 expression in VSMCs was significantly lower in ATF3^−/− mice. The critical molecules in the mechanism of impaired function of ATF3 deficiency may target the adipoR1 receptor but not adiponectin. In this study, we found that MCP-1 and IL-6 release can be enhanced by ATF3 deficiency, which may cause a marked increase in vasoconstriction in ATF3^−/− mice, but not in WT aortic rings. Indeed, reduced AMPK activity, leading to an upregulation of inflammatory cytokines IL-6 and TNF-α, is one of the reasons for exploring the possible role of ATF3 in vascular tone regulation.41^,42 It must be noted that there is another target of AMPK in addition to adiponectin, and these cannot be ruled out from local hypoxia and ROS generation in HFD-fed states involved in the loss of the anticontractile effect of PVAT.6^,33 In this study, PVAT inflammation, oxidative stress, and adipocyte hypertrophy could also lead to a harmful change in AMPK-mediated NOS production or perhaps act as a compensatory mechanism following adipoR1 expression in the PVAT or VSMCs. These pathological changes may cause severe vascular hyper-responsiveness to the inflamed and injured cardiovascular system. The homeostasis needed for maintaining blood pressure at physiological levels during vascular dysfunction associated with obesity requires ATF3 expression.

Adiponectin is a well-established vasodilator secreted by the adipocytes of the PVAT.39^,43 Published evidence suggests that the loss of the buffering activity of PVAT is accomplished by decreased adiponectin secretion.42 MCP-1, an ADCF induced by inflammation and local hypoxia, has been demonstrated to contribute to increased vascular hyperreactivity in obese animals and humans.10^,22 Attenuated MCP-1 expression and production have been shown to improve PVAT dysfunction, increasing vascular hyperreactivity. Based on a previous report, MCP-1 and IL-6 mRNA expression is modulated by ATF3 transcription,16^,32 which plays a role in the negative feedback regulation of inflammation. Our results showed that adiponectin expression in WT and ATF3^−/− mice showed no significant changes; however, adipoR1 expression was significantly decreased in HFD-fed ATF3^−/− mice. In addition, MCP-1 released from PVAT was greatly enhanced in obese mice but markedly higher in ATF3^−/− mice. Augmented MCP-1 and IL-6 overproduction due to ATF3 dysfunction in ATF3^−/− mice PVAT supports this notion. This further highlights the importance of ATF3 in maintaining vascular health by controlling ADCF and ADRF release.

Consent for publication

Not applicable.

Ethics approval and consent to participate

All animal protocols were approved by the Animal Care and Use Committee of Tzu Chi Hospital in Taiwan (approval no.:108–39 and 110–47).

STARMethods

Acknowledgments

Notes

Published: December 22, 2022

Footnotes

References