Life Sciences 125 (2015) 79–87

Contents lists available at ScienceDirect

Life Sciences journal homepage: www.elsevier.com/locate/lifescie

Perivascular adipose tissue and vascular responses in healthy trained rats Hygor N. Araujo a, Carmem P. Valgas da Silva a, Amanda C.S. Sponton b, Stefano P. Clerici b, Ana P.C. Davel b, Edson Antunes c, Angelina Zanesco a, Maria A. Delbin b,⁎ a b c

Department of Physical Education, Institute of Biosciences, Univ. Estadual Paulista (UNESP), Rio Claro, SP, Brazil Department of Structural and Functional Biology, Institute of Biology, University of Campinas (UNICAMP), Campinas, SP, Brazil Department of Pharmacology, School of Medical Sciences, University of Campinas (UNICAMP), Campinas, SP, Brazil

a r t i c l e

i n f o

Article history: Received 14 July 2014 Accepted 30 December 2014 Available online 29 January 2015 Keywords: Aerobic exercise training Vascular response Perivascular adipose tissue Adipokines

a b s t r a c t Aims: The importance of perivascular adipose tissue (PVAT) in vascular function has recently been recognized. The aim of the study was to investigate the effects of exercise training on anticontractile responses of periaortic adipose tissue. Main methods: Male Wistar rats were divided into sedentary (SD) and trained (TR). Running training was performed for 60 min/day, 5 days/week, for 8 weeks. Concentration–response curves to acetylcholine (ACh), sodium nitroprusside (SNP), phenylephrine (PHE) and serotonin (5-HT) were obtained in aortic rings without (PVAT–) or with (PVAT+) PVAT. The protein expressions of eNOS, AMPKα, pAMPKThr172 and mtTFA were determined in PVAT. The contents of adiponectin, leptin and TNF-α were evaluated systemically and locally. Key findings: The PVAT+ rings did not modify the relaxing responses to ACh and SNP whereas it showed anticontractile effects for both PHE and 5-HT agents in the SD and TR groups. The amount of PVAT was markedly reduced in TR (3.6 ± 0.3 mg/mm) compared with SD (6.8 ± 0.6 mg/mm). Increased protein expressions of eNOS, pAMPKThr172 and mtTFA were observed in PVAT from TR animals, without modifications in PVAT-derived adiponectin, leptin and TNF-α. Circulatory leptin levels were reduced in TR without changes in adiponectin. Significance: Our findings show that exercise training for 8 weeks did not alter the anticontractile effects induced by PVAT in rat-isolated aorta. Moreover, PVAT-derived adipokine, adiponectin and leptin levels were not different in trained healthy animals despite a significant metabolic adaptation and reduction in periaortic adipose tissue amount. © 2015 Elsevier Inc. All rights reserved.

Introduction It is well known that physical exercise promotes beneficial health effects in a variety of functional systems, preventing chronic disorders such as arterial hypertension, type 2 diabetes mellitus and obesity [5, 49]. Exercise improves health span, enhances cardiovascular function and delays age-associated frailty in healthy animals [14,23]. Different from the old concept that the periorgan adipose tissue (pericardial, perimuscular, and perivascular) plays a mechanical role by protecting organs against injury impact, more recent evidence has shown that, particularly, perivascular adipose tissue (PVAT) releases a wide range of biologically active molecules [8] including adipocytederived relaxing factor (ADRF) [7,30], adiponectin [31], leptin [12], and nitric oxide (NO) [16]. It has been demonstrated that, functionally, PVAT exhibits an anticontractile effect, which might involve ⁎ Corresponding author at: Department of Structural and Functional Biology, Institute of Biology University of Campinas (UNICAMP), P.O. Box 6109, Campinas, SP 13083-970, Brazil. Tel.: + 55 19 3521 6189. E-mail address: [email protected] (M.A. Delbin).

http://dx.doi.org/10.1016/j.lfs.2014.12.032 0024-3205/© 2015 Elsevier Inc. All rights reserved.

endothelial-dependent and/or endothelial-independent pathways [39]. In contrast, the anticontractile effect of PVAT is lost in experimental models of chronic obesity despite its higher amount. An imbalance between ADRF and vasocontractile agent production as well as a great releasing of proinflammatory substances might be the primary causes of the loss of the anticontractile effect of PVAT in obesity models [13,32]. On the other hand, increased contractile responses were observed in an experimental model of hypertension which was associated with a reduction in PVAT amount as well as a lower release of vasodilatory adipokines [11,12]. Besides their fundamental role in the regulation of the endocrine system, evidences show that leptin and adiponectin also participate in the regulation of the vascular tone [9,25]. Indeed, it has been demonstrated that leptin elicits an anticontractile effect that depends on an intact and functional endothelium [28,36,43]. Additionally, adiponectin derived from PVAT is considered a physiological modulator of local vascular tonus by increasing NO bioavailability and/or by inhibiting synthesis of inflammatory markers [26]. However, the number of studies that have evaluated the contribution of these adipocyte-derived adipokines on the vascular response is scarce.

80

H.N. Araujo et al. / Life Sciences 125 (2015) 79–87

Regarding exercise training and PVAT function, only two studies have examined this issue. However, the results were controversial [2, 35]. In addition, no one has investigated the anticontractile effects of PVAT in healthy trained animals which is crucial for a better understanding of the insight mechanisms by which exercise training promotes lower incidence of cardiometabolic diseases. Therefore, the aim of this work was to examine the effects of 8 weeks of aerobic exercise training on the vascular responses of isolated aortic rings and its relationship with the amount of the periaortic cushion fat. To further elucidate the paracrine control involved in the vascular responses, we focused on the mediators released from PVAT by measuring adiponectin, leptin and tumor necrosis factor-alpha (TNF-α). Given that adipose tissues play a key role in metabolism, we evaluated the protein expressions of endothelial nitric oxide synthase (eNOS), AMP-activated protein kinase α (AMPKα), phosphoAMPKThr172 (pAMPKThr172) and mitochondrial transcription factor A (mtTFA) in an attempt to detect the insight mechanisms by which exercise training might have a protective effect on cardiometabolic biomarkers. We also performed histology assessment in PVAT to analyze potential phenotype changes in adipose tissue. To exclude the contribution of systemic adipokines on the vascular responsiveness in reply to exercise training, we also evaluated circulating adiponectin, leptin, TNF-α and nitrite/nitrate (NO− x ) levels.

Material and methods Animals This study was approved by the Ethical Committee for Animal Research (CEUA 014/2012) at the University of São Paulo State (UNESP) established by the Brazilian College for Animal Experimentation (COBEA). Male Wistar rats (weighing 250–300 g) were obtained from the Animal Care Facility of the University of Campinas (UNICAMP) and were maintained in a room at 20–21 °C with a normal 12 h light/dark cycle. The animals were housed in groups of two/three and had free access to water and commercial chow (Nuvilab Radiated-CR1, Brazil). Animals were divided into two groups: sedentary (SD) and trained (TR). Body weight and food intake measurements were performed weekly during all periods of the study.

Aerobic exercise training Animals were trained on a treadmill designed for small animals with individual lanes (Gesan, São Paulo—SP, Brazil). One week before starting the training program, the animals were adapted to the treadmill to minimize potential stress; during this week the duration and speed begun at 5 meters/minute (m/min) for 15 min and were progressively increased to 10 m/min for 20 min. Only the animals adapted to the treadmill were used in the present study. After four days of adaptation, the animals performed an acute incremental exercise testing on the treadmill, where the intensity of exercise was increased by 5 m/min (5–40 m/min) every 3 min at 0% grade until exhaustion. The maximal speed was used to calculate the percentage corresponding to moderate intensity. At the beginning of the training program, the duration and speed started at 10 m/min for 30 min and were progressively increased to 60 min and at a speed of 60–80% of maximal capacity (15 m/min-17 m/min), 5 days/week, for 8 weeks and at 0% grade. All the animals were trained in the early morning, between 6:00 a.m. to 8:00 a.m. At the last week of the training program, the effectiveness of exercise was evaluated by acute incremental exercise testing on the treadmill for both groups, SD and TR. This test provided the total distance, total time and the maximal speed performed for each animal.

Blood sample and epididymal fat pad collection After 48 h of the last exercise session and 12 h of fasting, blood samples were collected from the tail vein and glycemia was measured using standard test strips (Accu-Chek Performa Roche Diagnostics, Indianapolis—IN, USA). Immediately after glycemia measurement, animals were anesthetized with sodium thiopental (40 mg/kg, i.p.) and arterial blood samples were collected from the abdominal aorta in different tubes (one for serum and one for plasma using EDTA as anticoagulant) and centrifuged (3000 rpm, for 15 min). Fresh serum was separated for lipid profile measurements and serum aliquots were also stored at − 80 °C; all plasma supernatants were stored at − 80 °C. After that, animals were euthanized and the epididymal fat pad was collected and weighted. Concentration–response curves Intact thoracic aorta was isolated carefully and placed in freshly prepared ice-cold Krebs solution containing (mM): NaCl, 118; NaHCO3, 25; glucose, 5.6; KCl, 4.7; KH2PO4, 1.2; MgSO4·7H2O, 1.1; and CaCl2·2H2O, 2.5. Further, thoracic aorta was cut into 3 mm rings using a calibrated eyepiece with a dissecting microscope (Nikon Instruments, Melville—NY, USA) and rings were isolated without (PVAT−) or with (PVAT +) the perivascular adipose tissue (PVAT). Each ring was suspended between two wire hooks and mounted in an organ chamber (Panlab Harvard Apparatus, Barcelona, Spain) with 10 ml Krebs solution at 37 °C, pH 7.4 and continuously gassed with 95% O2 and 5% CO2 under a resting tension of 10 milliNewton (mN). The rings were allowed to equilibrate for 60 min, during this period the tension was verified every 15 min and washed with Krebs solution. After the equilibration period, rings were precontracted with KCl 80 mM and washed with Krebs to verify their viability. Cumulative concentration–response curves to vasodilator agents: acetylcholine (ACh, 1 nM–30 μM) and sodium nitroprusside (SNP, 1 nM–100 μM) were obtained. Relaxing responses were plotted as percentage of the contraction induced by phenylephrine (in a concentration necessary to produce 50–70% of maximal response of KCl 80 mM). In accordance with standard in vitro analysis of vascular responses, concentration–response curves to phenylephrine (PHE, 1 nM–100 μM) were obtained in the presence of propranolol (100 nM; [42]). We also performed concentration–response curves to serotonin (5-HT, 10 nM–100 μM). Contractile responses were plotted according to the force and length from each ring measured as milliNewton/millimeter (mN/mm). Data acquisition was performed using PowerLab 8/30 (LabChart 7, ADInstruments, Sydney, Australia). After the concentration–response curves, PVAT of each ring was collected, the excess of Krebs solution was dried with a filter paper and the tissue was weighted wet and measured as milligram/ millimeter (mg/mm). All the concentration–response data were fit to a logistic function in the equation: E = EMAX /((1 + (10c/10x)n) + Φ), where E is the effect of

Table 1 Body weight, epididymal fat pad, food intake, glucose, total cholesterol and triglycerides in rats from the sedentary (SD) and trained (TR) groups. Groups

I-body weight (g) F-body Weight (g) Epididymal fat (g) Food intake (g/rat/day) Blood glucose (mg/dl) Total cholesterol (mg/dl) Triglycerides (mg/dl)

SD

TR

281 ± 15 (8) 445 ± 15 (8) 8.0 ± 0.7 (8) 29 ± 1 (8) 86.6 ± 3.5 (8) 49.8 ± 2.9 (5) 83.1 ± 6.7 (8)

276 ± 14 (9) 399 ± 11⁎ (9) 4.5 ± 0.3⁎ (9) 24 ± 1⁎ (9) 80.8 ± 1.9 (9) 55.1 ± 2.1 (5) 56.9 ± 4.4⁎ (9)

Initial (I) and Final (F). Data are mean ± SEM. The number of animals per group is indicated in parentheses. ⁎ p b 0.05 compared with SD.

H.N. Araujo et al. / Life Sciences 125 (2015) 79–87 Table 2 Potency values (pEC50) obtained from concentration–response curves to acetylcholine (ACh) and sodium nitroprusside (SNP) in aortic rings without (PVAT−) or with (PVAT+) the perivascular adipose tissue from sedentary (SD) and trained (TR) rats. Groups

ACh–PVAT− PVAT+ SNP–PVAT− PVAT+

SD

TR

7.3 ± 0.10 (6) 7.3 ± 0.10 (6) 7.5 ± 0.28 (7) 7.6 ± 0.17 (7)

7.2 ± 0.07 (5) 7.2 ± 0.10 (5) 7.9 ± 0.08 (6) 7.6 ± 0.15 (6)

Potency is represented as –log of molar concentration to produce 50% of the maximal responses. Data are mean ± SEM. The number of animals per group is indicated in parentheses.

above basal; EMAX is the maximum response produced by the agonist; c is the logarithm of the EC50, the concentration of agonist that produces half-maximal response; x is the logarithm of the concentration of agonist; the exponential term, n, is a curve-fitting parameter that defines the slope of the concentration response line, and Φ is the response observed in the absence of added agonist. Nonlinear regression analysis was used to determine the parameters EMAX, log EC50, and using GraphPad Prism (GraphPad Software, Prism 4, San Diego—CA, USA) with the constraint that Φ = 0. The responses for each agonist are shown as the mean ± SEM of potency (pEC50) and maximal responses (EMAX). Western blotting analyses In order to evaluate the contribution of endothelial nitric oxide synthase (eNOS), the expression of this protein was determined by

81

Western blotting in thoracic aorta as well as in the respective PVAT. We also evaluated the metabolic adaptations of PVAT by the protein expressions of AMP-activated protein kinase α (AMPKα), phosphoAMPKThr172 (pAMPKThr172) and mitochondrial transcription factor A (mtTFA). Frozen tissue was isolated and homogenized in a RIPA lysing buffer (Upstate, Temecula—CA, USA) with 1 mM Na 3 VO 4 , 1 mM phenylmethylsulfonyl fluoride and protease inhibitor cocktail (2 μl/ml) (Sigma-Aldrich Co., Saint Louis—MO, USA). The tissue lysate was centrifuged (2500 rpm, for 30 min at 4 °C) and the supernatant was collected. The protein concentration was determined by a BCA protein assay kit (Pierce, Rockford—IL, USA). Proteins from homogenized (50 μg for aorta and 100 μg for PVAT) were eletrophoretically (Mini-Protean II, Electrophoresis Cell, Bio-Rad, Hercules, CA, USA) separated by 7.5% or 12% SDS-PAGE. The proteins were subsequently transferred to polyvinylidene difluoride membranes, overnight at 4 °C, using a Mini Trans-blot Cell System (BioRad) containing 25 mM Tris, 190 mM glycine, 20% methanol and 0.05% SDS. After blockade of nonspecific sites in a Tris-buffered solution (10 mM Tris, 100 mM NaCl, and 0.1% Tween 20) with 5% albumin, membranes were incubated overnight at 4 °C with the primary antibody anti-eNOS (1:750, BD Bioscience, San Jose—CA, USA); anti-AMPKα (1:1000, Abcam, Cambridge—MA, USA), anti-pAMPKThr172 (1:1000, Cell Signaling, Danvers—MA, USA) and anti-mtTFA (1:500, Santa Cruz Biotechnology, Santa Cruz—CA, USA). After being washed (10 mM Tris, 100 mM NaCl, and 0.1% Tween 20), the membranes were incubated with secondary antibody (1:5000) IgG antibody conjugated to horseradish peroxidase. The membranes were thoroughly washed and immunocomplexes were detected using an enhanced horseradish peroxidase–luminol chemiluminescent system (ECL Plus Amersham,

Fig. 1. Concentration–response curves to vasodilator agents. Concentration–response curves to acetylcholine (ACh, panels A and B) and sodium nitroprusside (SNP, panels C and D) in aortic rings; without (PVAT−, open symbol) or with (PVAT+, close symbol) the perivascular adipose tissue from sedentary (○PVAT−, ●PVAT+: SD panels A and C) and trained (ΔPVAT−, ▲PVAT+: TR panels B and D) rats. The maximal response values are inserted in the figure. Data are mean ± SEM. The number of animals per group is indicated in parentheses.

82

H.N. Araujo et al. / Life Sciences 125 (2015) 79–87

Piscataway—NJ, USA). Scanning densitometry was used to quantify the immunoblot signals using specific software (Scion Image, Scion Corporation, Frederick—MD, USA). The same membrane was used to determine α-actin (for aorta) or α-tubulin (for PVAT) protein expressions as an internal control using anti-α-actin (1:5000, Abcam) or anti-αtubulin (1:1000, Santa Cruz), and its content was used to normalize eNOS, AMPKα and mtTFA protein expressions in each sample. The AMPKα was used to normalize pAMPKThr172 protein expression.

Adiponectin, leptin and tumor necrosis factor-alpha (TNF-α) in the perivascular adipose tissue Adiponectin and leptin concentrations were measured using a commercial ELISA kit (R&D Systems, Minneapolis—MN, USA and Merck Millipore, Billerica—MA, USA, respectively). For adiponectin, samples were prepared with assay diluent in a concentration of 10 ng of protein/μl (500 ng of protein in 50 μl of diluent). For leptin, samples were prepared with assay diluent in a concentration of 5 μg of protein/μl (50 μg of protein in 10 μl of diluent). The concentrations of adiponectin and leptin in the samples were captured using specific antibody, and the concentrations were determined by comparison with standard curve absorbance measured at 450 nm. For TNF-α, samples were prepared with matrix diluent in a concentration of 2 μg of protein/μl (100 μg of protein in 50 μl of matrix diluent). The concentrations were detected with specific antibody using an ELISA kit (BioLegend, San Diego—CA, USA) and absorbance measured at 450 nm.

Histology assessment Thoracic aorta was cut into rings of 3 mm with the surrounding perivascular adipose tissue. The rings were fixed in a 4% formaldehyde solution and embedded in paraffin and transverse sections (5 μm) were obtained. Slides were deparaffinized in xylene and histologically stained with hematoxylin and eosin for characterization. At the sequence, images were obtained with an inverted microscope (Eclipse Ti, Nikon, Japan) equipped with a digital camera (DS-U3, Nikon, Japan) and NIS Elements Basic Research (3.2 Software, Nikon, Japan), using a 40× objective. Plasma nitrate/nitrite (NO− x ), adiponectin and leptin Plasma NO− x (μM) concentrations were measured using a commercial kit (Cayman Chemical, Ann Arbor—MI, USA). Briefly, plasma samples were ultra-filtrated through microfilter cups (10,000 rpm, for 60 min at 4 °C) (Microcon Centrifugal Filter Units, 10 kDa; Millipore, Billerica—MA, USA). The NO− x concentration of the resulting filtrate was determined based on the enzymatic conversion of nitrate to nitrite by nitrate reductase. The addition of the Griess reagents converted nitrite into a deep purple azo compound and absorbance measured at 540 nm determined the nitrite concentration. The measurements of plasma adiponectin and leptin were performed using the commercial ELISA kits (R&D Systems and Merck Millipore, Billerica—MA, USA, respectively) and the absorbance measured at 450 nm.

Fig. 2. Concentration–response curves to contractile agents. Concentration–response curves to phenylephrine (PHE, panels A and B) and serotonin (5-HT, panels C and D) in aortic rings; without (PVAT−, open symbol) or with (PVAT+, close symbol) the perivascular adipose tissue from sedentary (○PVAT−, ●PVAT+: SD panels A and C) and trained (ΔPVAT−, ▲PVAT+: TR panels B and D) rats. Panel E: The amount of perivascular adipose tissue. The maximal response values are inserted in the figure. Data are mean ± SEM. The number of animals per group is indicated in parentheses. +p b 0.05 compared with PVAT− from the respective groups. *p b 0.05 compared with SD.

H.N. Araujo et al. / Life Sciences 125 (2015) 79–87 Table 3 Potency values (pEC50) obtained from concentration–response curves to phenylephrine (PHE) and serotonin (5-HT) in rat aortic rings without (PVAT−) or with (PVAT+) the perivascular adipose tissue from sedentary (SD) and trained (TR) groups. Groups

PHE–PVAT− PVAT+ 5-HT–PVAT− PVAT+

SD

TR

6.4 ± 0.08 (7) 6.2 ± 0.05 (7) 5.4 ± 0.10 (7) 5.2 ± 0.09 (7)

6.6 ± 0.07 (5) 6.2 ± 0.08 (5) 5.4 ± 0.03 (6) 5.2 ± 0.06 (6)

Potency is represented as −log of molar concentration to produce 50% of the maximal responses. Data are mean ± SEM. The number of animals per group is indicated in parentheses.

Serum total cholesterol, triglycerides and TNF-α In order to analyze the lipid profile, total cholesterol and triglyceride concentrations were determined using fresh serum samples and an available colorimetric kit (Katal Biotechnology, São Paulo—SP, Brazil) with the absorbance measured at 500 nm. The TNF-α concentration was detected in serum samples with specific antibody using the ELISA kit (BioLegend) and absorbance measured at 450 nm.

83

(SD: 13 ± 2 min). Similarly, the total distance performed was increased in the TR group (TR: 383 ± 62 m) as compared with the SD group (SD: 169 ± 35 m). The maximal speed was also significantly different in TR animals (TR: 34 ± 3.5 m/min) as compared with the SD group (SD: 21 ± 2.6 m/min). Body weight, epididymal fat pad and food intake The body weight values were similar in all groups at the initial time of the study. As expected, after 8 weeks, exercise training was effective in preventing weight gain in the TR group, by approximately 10% as compared with SD (Table 1). Epididymal fat pad was also reduced in the TR group, in approximately 44% as compared with that of SD (Table 1). Finally, the food intake was also decreased in the TR group, approximately 17% as compared with that of SD (Table 1). Blood glucose and lipid profiles The blood glucose and total cholesterol were similar in all groups (Table 1). On the other hand, exercise training was effective in reducing triglycerides in the TR group as compared with those of SD, approximately 30% (Table 1). Concentration–response curves

Statistical analysis All data are expressed as mean ± SEM of n experiments. Unpaired Student's t-test was performed using GraphPad Prism (GraphPad Software, Prism 4, San Diego—CA, USA) and values of p b 0.05 were considered statistically significant. Drugs Acetylcholine chloride, DL-propranolol hydrochloride, sodium nitroprusside dihydrate, phenylephrine hydrochloride and serotonin were purchased from Sigma-Aldrich CO. (Saint Louis—MO, USA). Results Exercise training performance TR animals had an increase in the total time performed in the incremental exercise test (TR: 21 ± 2 min) as compared with the SD group

The vasodilator agents acetylcholine (ACh: 1 nM–30 μM) and sodium nitroprusside (SNP: 100 pM–100 μM) produced concentrationdependent relaxation in aortic rings PVAT− and PVAT+. The PVAT+ did not modify the endothelium-dependent relaxation responses evoked by ACh. Similarly, aerobic exercise training did not affect this response (Table 2 and Fig. 1, panels A and B). Regarding the concentration–responses to NO donor, SNP, no alterations were observed at the pEC50 levels or in the EMAX values in both groups, PVAT − or PVAT + (Table 2 and Fig. 1, panels C and D). The contractile agents phenylephrine (PHE, 1 nM–100 μM) and serotonin (5-HT, 10 nM–100 μM) produced concentrationdependent contraction responses in aortic rings PVAT − and PVAT +. The PVAT + rings showed anticontractile effects for both agents PHE (Δ reduction: 1.8 mN/mm) and 5-HT (Δ reduction: 2.4 mN/mm) in the SD group (Fig. 2, panels A and C). The PVAT + also produced anticontractile effects in the TR group, for PHE (Δ reduction: 1.6 mN/mm) and 5-HT (Δ reduction: 2.1 mN/mm) (Fig. 2, panels B and D). No alterations were verified for pEC 50 values for both agonists PHE and 5-HT (Table 3). The amount of perivascular

Fig. 3. Protein expression of endothelial nitric oxide synthase (eNOS) and plasma nitrite/nitrate (NO− x ). Aorta protein expression of eNOS (panel A), perivascular adipose tissue protein expression of eNOS (panel B), and plasma NO− x (panel C); from sedentary (SD) and trained (TR) rats. The bottom panel is representative of Western blotting and the top panel is representative of quantitative analysis. Data are mean ± SEM. The number of animals per group is indicated in the bars. *p b 0.05 compared with SD.

84

H.N. Araujo et al. / Life Sciences 125 (2015) 79–87

Fig. 4. Protein expressions of AMP-activated protein kinase α (AMPKα), phosphoAMPKThr172 (pAMPKThr172) and mitochondrial transcription factor A (mtTFA). Perivascular adipose tissue protein expressions of AMPKα (panel A), pAMPKThr172 (panel B) and mtTFA (panel C); from sedentary (SD) and trained (TR) rats. The bottom panel is representative of Western blotting and the top panel is representative of quantitative analysis. Data are mean ± SEM. The number of animals per group is indicated in the bars. *p b 0.05 compared with SD.

Fig. 5. Adiponectin, leptin and tumor necrosis factor-alpha (TNF-α). Perivascular adipose tissue content of adiponectin (panel A) and plasma adiponectin (panel B), perivascular adipose tissue content of leptin (panel C) and plasma leptin (panel D) and perivascular adipose tissue content of TNF-α (panel E); from sedentary (SD) and trained (TR) rats. Data are mean ± SEM. The number of animals per group is indicated in the bars. *p b 0.05 compared with SD.

H.N. Araujo et al. / Life Sciences 125 (2015) 79–87

adipose tissue was markedly reduced after 8 weeks of exercise training weighted after concentration–response curves (3.6 ± 0.3 mg/mm) when compared with the SD group (6.8 ± 0.6 mg/mm), approximately 48% of reduction (Fig. 2, panel E). Protein expression of eNOS and plasma NO− x The quantification of aortic eNOS protein expression was not modified in the SD and TR groups. Nevertheless, in PVAT the eNOS protein expression was significantly increased in the TR group as compared with SD, by approximately 50%. No alteration was observed in plasma NO− x levels (Fig. 3). Protein expression of AMPKα, pAMPKThr172 and mtTFA The total of the protein expression of AMPKα was remarkably decreased in the TR group, approximately 70%, whereas the pAMPKThr172 was increased as compared with SD, approximately 75% (Fig. 4, panels A and B). We also detected an enhancement of approximately 110% on mtTFA protein expression in the TR group as compared to SD (Fig. 4 C). Adiponectin, leptin and TNF-α Interestingly, neither the content of adiponectin from aortic PVAT nor the plasma adiponectin levels was altered in both groups (Fig. 5, panels A and B). The aerobic exercise training significantly reduced the plasma leptin in the TR group, approximately 78%, as compared with the SD group. However, the content of leptin in aortic PVAT was similar in both groups (Fig. 5, panels C and D). Serum TNF-α was not detected in the present study in all groups (data not shown). No differences in the TNF-α content were found in PVAT from aortic rings in the SD and TR rats (Fig. 5 E). Histology The images obtained from the SD and TR groups demonstrated similarities related to architecture and structure of PVAT, with round nuclei and multilocular adipocytes (characteristic of brown adipose tissue) (Fig. 6, panels A and B). Interestingly, analysis of hematoxylin and eosin stained tissue showed visual differences in lipid droplet size between the SD and TR groups. The adipocytes from PVAT in the SD group exhibit enlargement and coalescence of lipid droplets in comparison to the TR group, wherein the lipid droplets are smaller (Fig. 6, panels A and B).

85

Discussion The main findings of this study were that exercise training for 8 weeks promotes an up-regulation of eNOS, pAMPKThr172 and mtTFA protein expressions in periaortic adipose tissue that was accompanied by a significant reduction in the PVAT amount and morphological alterations. On the other hand, these alterations did not affect the antincontractile effects of PVAT for either agonists, phenylephrine and serotonin, as well as the levels of adiponectin, leptin and TNF-α measured locally. It has been well established that aerobic exercise training promotes body weight loss in different species such as rats [4,6], horses [3] and mice [44]. The cellular mechanisms by which exercise training might regulate body weight are mainly related to increase in energy expenditure, increase in sympathetic activity as well as increase in the β-oxidation rate in mitochondria [29,37,40]. Confirming previous work from our laboratory, in the present study, the 8-week exercise training promotes a reduction in body weight gain, by approximately 10%, and in parallel, food intake was also reduced by 17%. A marked reduction in the content of the epididymal fat pad and PVAT from aortic rings was found, by approximately 50%. The reason for this higher sensitivity of aortic PVAT, particularly, in reducing perivascular fat mass content in response to exercise training might be due to its unique morpho-functional property. Indeed, recent evidence has shown that aortic PVAT presents a mix of white and brown adipose tissues with functional characteristics of brown adipose tissue [10, 39,48]; in accordance with these studies we also evidenced morphological similarity of rat thoracic PVAT to brown adipose tissue, with round nuclei, and multilocular lipid droplets. Specially, in the TR group, the adipocytes exhibit smaller lipid droplets in comparison to the SD group. Additionally, we found increased protein expressions of pAMPKThr172 and mtTFA in rat aortic PVAT. AMPK has been classically described as an intracellular energy sensor mainly in skeletal muscle, adipose tissues and liver [19]. Further, this protein kinase plays an important role in glucose transport, lipid metabolism, cellular growth and proliferation, mitochondrial biogenesis, inflammation, and oxidative stress [38]. Its activation is dependent on upstream kinases that phosphorylates at the threonine residue (Thr172) of the enzyme [20]. Thus, an up-regulation of protein expression of pAMPKThr172 indicates that exercise training has a profound effect in aortic PVAT metabolism. This is confirmed by increased protein expression of mtTFA in PVAT, an important transcription factor for mitochondrial biogenesis including replication and transcription of mitochondrial DNA [21,45]. The clinical relevance of these data is

Fig. 6. Perivascular adipose tissue histology. Representative histology of perivascular adipose tissue from thoracic aorta segments from sedentary (SD, n = 4) and trained (TR, n = 6) groups. Solid arrows indicate round nuclei and disrupted arrows indicate multilocular adipocytes with larger lipid droplets in the SD group and smaller lipid droplets in the TR group. Digital images were captured using the 40× objective. Scale bar = 50 μm.

86

H.N. Araujo et al. / Life Sciences 125 (2015) 79–87

not clear, however, given that AMPK regulates vascular smooth muscle proliferation as well as re-endothelialization. We can speculate that these alterations in the AMPK/mtTFA signaling pathway of the aortic PVAT from healthy animals may be related to the beneficial effects of exercise training in preventing vascular disease. Regarding vascular responses, we found no effects of exercise training on the relaxing responses to endothelium-dependent and independent agonists in rat aortic rings with or without PVAT. Accordingly, previous studies using swine coronary artery with or without PVAT also found no alterations in the concentration–response curves to vasodilators agents, bradykinin and acetylcholine, in sedentary or trained animals [2,35]. Corroborating with the functional assays, the protein expression of eNOS in aortic rings from trained rats was not modified. Intriguingly, an up-regulation of eNOS protein expression in aortic PVAT was seen after exercise training, even though no changes were found in the concentration– response curves to vasodilator agents and in the plasma NO− x levels. Considering the complexity of the adipose tissue as an endocrine organ as well as the existence of a variety of signaling pathways in this tissue, the mechanistic insight by which an upregulation of eNOS protein expression in aortic PVAT from trained animals did not affect relaxing response is not clear at moment. We can speculate that an up-regulation of eNOS protein expression in PVAT (50%) might be contributing to the lipolytic effects in the periorgan fat tissue rather than relaxing responses. Indeed, a study has shown that the eNOS/NO signaling pathway is associated with increased lipolysis rate ([22]) and it was recently demonstrated that exercise induces mitochondrial biogenesis in subcutaneous adipose tissue through eNOS-dependent mechanisms ([41]). Conversely, the contractile responses to phenylephrine and serotonin were affected in the presence of PVAT showing the classical anticontractile effects of PVAT on the vascular response. Interestingly, exercise training failed to modify the anticontractile effects of aortic PVAT even though the amount of perivascular fat pad was significantly reduced. Furthermore, no changes in adiponectin levels were found in aortic PVAT from trained rats. Previous studies have shown a central role of adiponectin on the anticontractile effects in mesenteric artery [9,47], aorta [9] and human small vessel [1,17] Likely, the lack of effect on anticontractile actions in response to aerobic exercise training in aortic rings might be due to the absence of any changes in adiponectin levels in PVAT as well as in the plasma of healthy trained animals. Accordingly, it has been demonstrated that systemic adiponectin concentrations do not change in response to moderate exercise training [27]. In accordance with this result, we found no changes in TNF-α content in periaortic adipose tissue from healthy trained rats indicating a balance between adiponectin and TNF-α content in PVAT in healthy trained rats. Indeed, it was demonstrated that increased TNF-α levels in obesity reduce NO availability and also inhibit the anticontractile effects of PVAT, especially through promoting superoxide generation [24,46]. Interestingly, studies have associated the cardiovascular protective role of adiponectin to its ability to reduce the production and activity of TNF-α in mice [15,33]. Leptin plays a fundamental role in energy expenditure regulation and its production is mainly derived from white adipocytes [50]. In addition, it has been demonstrated that leptin is also produced in PVAT, eliciting anticontractile effects and modulating vascular tone through different signaling pathways [28,36,43]. However, the exact role of PVAT-derived leptin on the vascular responses and on the downstream pathway are not clear yet. A recent study demonstrated a low level of leptin in periaortic adipose tissue from spontaneously hypertensive rats which was positively associated with loss of anticontractile effects [11,12]. On the other hand, a higher NO production in mesenteric PVAT was associated with hyperleptinemia on an obesity model [16]. Furthermore, it has been reported that leptin release from PVAT has deleterious effects on the underlying endothelium and vascular smooth muscle cells in obesity and other metabolic disorders [34]. In our

study, we clearly demonstrated that aerobic exercise training for 8 weeks had no effects on PVAT-derived leptin; however, a marked reduction in its systemic levels was found in trained animals, approximately 80%. Leptin mRNA expression and secretion are strongly correlated with the size of white adipocytes [18,51] and we could relate this systemic alteration of leptin to a significant reduction (of approximately 44%) in epididymal fat pad, essentially white adipose tissue, of the trained animals. Conclusions Collectively, our findings show that exercise training for 8 weeks promotes a profound effect on aortic PVAT morphology and amount which were associated with an up-regulation of the eNOS/ AMPKThr172/mtTFA signaling pathway without changes in local adipokines. On the other hand, no effects of exercise training were found upon relaxing or contractile agents of the aortic rings from healthy animals. Conflict of interest statement The authors declare that there are no conflicts of interest.

Acknowledgments Edson Antunes, Angelina Zanesco and Maria A. Delbin were supported by the Fundação de Amparo a Pesquisa do Estado de São Paulo (FAPESP). References [1] R. Aghamohammadzadeh, A.S. Greenstein, R. Yadav, M. Jeziorska, S. Hama, F. Soltani, P.W. Pemberton, B. Ammori, R.A. Malik, H. Soran, A.M. Heagerty, Effects of bariatric surgery on human small artery function: evidence for reduction in perivascular adipocyte inflammation, and the restoration of normal anticontractile activity despite persistent obesity, J. Am. Coll. Cardiol. 9 (2013) 128–135. [2] A.K. Bunker, A.A. Arce-Esquivel, R.S. Rector, F.W. Booth, J.A. Ibdah, M.H. Laughlin, Physical activity maintains aortic endothelium-dependent relaxation in the obese type 2 diabetic OLETF rat, Am. J. Physiol. Circ. Physiol. 298 (2010) 1889–1901. [3] R.A. Carter, L.J. McCutcheon, E. Valle, E.N. Meilahn, R.J. Geor, Effects of exercise training on adiposity, insulin sensitivity, and plasma hormone and lipid concentrations in overweight or obese, insulin-resistant horses, Am. J. Vet. Res. 71 (2010) 314–321. [4] G.G. de Araujo, M. Papoti, M.A. Delbin, A. Zanesco, C.A. Gobatto, Physiological adaptations during endurance training below anaerobic threshold in rats, Eur. J. Appl. Physiol. 113 (2013) 1859–1870. [5] M.A. Delbin, A.J. Trask, The diabetic vasculature: physiological mechanisms of dysfunction and influence of aerobic exercise training in animal models, Life Sci. 102 (2014) 1–9. [6] C. de Moraes, A.P. Davel, L.V. Rossoni, E. Antunes, A. Zanesco, Exercise training improves relaxation response and SOD-1 expression in aortic and mesenteric rings from high caloric diet-fed rats, BMC Physiol. 29 (2008) 8–12. [7] G. Dubrovska, S. Verlohren, F.C. Luft, M. Gollasch, Mechanisms of ADRF release from rat aortic adventitial adipose tissue, Am. J. Physiol. Heart Circ. 286 (2004) 1107–1113. [8] E.C. Eringa, W. Bakker, V.W.M. Van Hinsbergh, Paracrine regulation of vascular tone, inflammation and insulin sensitivity by perivascular adipose tissue, Vasc. Pharmacol. 56 (2012) 204–209. [9] G. Fesus, G. Dubrovska, K. Gorzelniak, R. Kluge, Y. Huang, F.C. Luft, M. Gollasch, Adiponectin is a novel humoral vasodilator, Cardiovasc. Res. 75 (2007) 719–727. [10] T.P. Fitzgibbons, S. Kogan, M. Aouadi, G.M. Hendricks, J. Straubhaar, M.P. Czech, Similarity of mouse perivascular and brown adipose tissue and their resistance to dietinduced inflammation, Am. J. Physiol. Heart Circ. Physiol. 301 (2011) 1425–1437. [11] B. Galvez, J. Castro, D. Herold, G. Dubrovska, S. Arribas, M.C. González, A. Aranguez, F.C. Luft, P. Ramos, M. Gollasch, M.S. Fernández-Alfonso, Perivascular adipose tissue and mesenteric vascular function in spontaneously hypertensive rats, Arterioscler. Thromb. Vasc. Biol. 26 (2006) 1297–1302. [12] B. Galvez-Prieto, B. Somoza, M. Gil-Ortega, C.F. García-Prieto, A.I. de Las Heras, M.C. González, S. Arribas, I. Aranguez, J. Bolbrinker, R. Kreutz, M. Ruiz-Gayo, M.S. Fernández-Alfonso, Anticontractile effects of perivascular adipose tissue and leptin are reduced in hypertension, Front. Pharmacol. 5 (2012) 1–8. [13] Y.J. Gao, Dual modulation of vascular function by perivascular adipose tissue and its potential correlation with adiposity/lipoatrophy-related vascular dysfunction, Curr. Pharm. Des. 13 (2007) 2189–2192. [14] R. Garcia-Valles, M.C. Gomez-Cabrera, L. Rodriguez-Mañas, F.J. Garcia-Garcia, A. Diaz, I. Nogueira, G. Olaso-Gonzalez, J. Viña, Life-long spontaneous exercise does not prolong lifespan but improves health span in mice, Longev. Healthspan 2 (2013) 14. http://dx.doi.org/10.1186/2046-2395-2-14.

H.N. Araujo et al. / Life Sciences 125 (2015) 79–87 [15] Q. Ge, L. Ryken, L. Noel, E. Maury, S.M. Brichard, Adipokines identified as new downstream targets for adiponectin: lessons from adiponectin-overexpressing or -deficient mice, Am. J. Physiol. Endocrinol. Metab. 301 (2011) 326–335. [16] M. Gil-Ortega, P. Stucchi, R. Guzmán-Ruiz, V. Cano, S. Arribas, M.C. González, M. Ruiz-Gayo, M.S. Fernández-Alfonso, B. Somoza, Adaptative nitric oxide overproduction in perivascular adipose tissue during early diet-induced obesity, Endocrinology 151 (2010) 3299–3306. [17] A.S. Greenstein, K. Khavandi, S.B. Withers, K. Sonoyama, O. Clancy, M. Jeziorska, I. Laing, A.P. Yates, P.W. Pemberton, R.A. Malik, A.M. Heagerty, Local inflammation and hypoxia abolish the protective anticontractile properties of perivascular fat in obese patients, Circulation 119 (2009) 1661–1670. [18] K.Y. Guo, P. Halo, R.L. Leibel, Y. Zhang, Effects of obesity on the regulation of leptin mRNA expression and adipocyte size in anatomically distinct fat depots in mice, Am. J. Physiol. Regul. Integr. Comp. Physiol. 287 (2004) 112–119. [19] D.G. Hardie, AMP-activated protein kinase: an energy sensor that regulates all aspects of cell function, Genes Dev. 18 (2011) 1895–1908. http://dx.doi.org/ 10.1101/gad.17420111. [20] S.A. Hawley, M. Davison, A. Woods, S.P. Davies, B.K. Beri, D. Carling, D.G. Hardie, Characterization of the AMP-activated protein kinase kinase from rat liver and identification of threonine 172 as the major site at which it phosphorylates AMPactivated protein kinase, J. Biol. Chem 271 (1996) 27879–27887. [21] D.A. Hood, Mechanisms of exercise-induced mitochondrial biogenesis in skeletal muscle, Appl. Physiol. Nutr. Metab. 34 (2009) 465–472. http://dx.doi.org/10.1139/ H09-045. [22] W.S. Jobgen, S.K. Fried, W.J. Fu, C.J. Meininger, G. Wu, Regulatory role for the arginine–nitric oxide pathway in metabolism of energy substrates, J. Nutr. Biochem. 17 (2006) 571–588. [23] S. Judge, Y.M. Jang, A. Smith, C. Selman, T. Phillips, J.R. Speakman, T. Hagen, C. Leeuwenburgh, Exercise by lifelong voluntary wheel running reduces subsarcolemmal and interfibrillar mitochondrial hydrogen peroxide production in the heart, Am. J. Physiol. Regul. Integr. Comp. Physiol. 289 (2005) 1564–1572. [24] J. Ketonen, J. Shi, E. Martonen, E. Mervaala, Periadventitial adipose tissue promotes endothelial dysfunction via oxidative stress in diet-induced obese C57Bl/6 mice, Circ. J. 74 (2010) 1479–1487. [25] J.D. Knudson, G.A. Payne, L. Borbouse, J.D. Tune, Leptin and mechanisms of endothelial dysfunction and cardiovascular disease, Curr. Hypertens. Rep 10 (2008) 434–439. [26] C. Kobashi, M. Urakaze, M. Kishida, E. Kibayashi, H. Kobayashi, S. Kihara, T. Funahashi, M. Takata, R. Temaru, A. Sato, K. Yamazaki, N. Nakamura, M. Kobayashi, Adiponectin inhibits endothelial synthesis of interleukin-8, Circ. Res. 97 (2005) 1245–1252. [27] R.R. Kraemer, V.D. Castracane, Exercise and humoral mediators of peripheral energy balance: ghrelin and adiponectin, Exp. Biol. Med. (Maywood) 232 (2007) 184–194. [28] G. Lembo, C. Vecchione, L. Frattal, G. Marino, V. Trimarco, G. D'amati, B. Trimarco, Leptin induces direct vasodilation through distinct endothelial mechanisms, Diabetes 49 (2000) 293–297. [29] V. Ljubicic, A.M. Joseph, A. Saleem, G. Uguccioni, M. Collu-Marchese, R.Y. Lai, L.M. Nguyen, D.A. Hood, Transcriptional and post-transcriptional regulation of mitochondrial biogenesis in skeletal muscle: effects of exercise and aging, Biochim. Biophys. Acta 1800 (2010) 223–234. [30] M. Lohn, G. Dubrovska, B. Lauterbach, F.C. Luft, M. Gollasch, A.M. Sharma, Periadventitial fat releases a vascular relaxing factor, FASEB 16 (2002) 1057–1063. [31] F.M. Lynch, S.B. Withers, Z. Yao, M.E. Werner, G. Edwards, A.H. Weston, A.M. Heagerty, Perivascular adipose tissue-derived adiponectin activates BKCa channels to induce anticontractile responses, Am. J. Physiol. Heart Circ. Physiol. 304 (2013) 786–795. [32] C. Marchesi, T. Ebrahimian, O. Angulo, P. Paradis, E.L. Schiffrin, Endothelial nitric oxide synthase uncoupling and perivascular adipose oxidative stress and inflammation contribute to vascular dysfunction in rodent model of metabolic syndrome, Hypertension 54 (2009) 1384–1392.

87

[33] T. Masaki, S. Chiba, H. Tatsukawa, T. Yasuda, H. Noguchi, M. Seike, H. Yoshimatsu, Adiponectin protects LPS-induced liver injury through modulation of TNF-alpha in KK-Ay obese mice, Hepatology 40 (2004) 177–184. [34] G.A. Payne, J.D. Tune, J.D. Knudson, Leptin-induced endothelial dysfunction: a target for therapeutic interventions, Curr. Pharm. Des. 20 (2014) 603–608. [35] M.S. Reifenberger, J.R. Turk, S.C. Newcomer, F.W. Booth, M.H. Laughlin, Perivascular fat alters reactivity of coronary artery: effects of diet and exercise, Med. Sci. Sports Exerc. 39 (2007) 2125–2134. [36] A.S. Sahin, H. Bariskaner, H. Gökbel, N. Okudan, The dual effects of leptin on aortic rings with and without endothelium isolated from streptozotocin-induced diabetic rats, Methods Find. Exp. Clin. Pharmacol. 31 (2009) 325–329. [37] T. Sakurai, J. Ogasawar, T. Kizaki, S. Sato, Y. Ishibashi, M. Takahashi, 0. Kobayashi, S. Oh-Ishi, J. Nagasawa, K. Takhashi, H. Ishida, T. Izawa, H. Ohno, The effects of exercise training on obesity-induced dysregulated expression of adipokines in white adipose tissue. Int, J. Endocrinol. 801743 (2013). [38] G.R. Steinberg, B.E. Kemp, AMPK in health and disease, Physiol. Rev. 89 (2009) 1025–1078. http://dx.doi.org/10.1152/physrev.00011.2008. [39] T. Szasz, G.F. Bomfim, R.C. Webb, The influence of perivascular adipose tissue on vascular homeostasis, Vasc. Health Manag 9 (2013) 105–116. [40] D. Thompson, F. Karpe, M. Lafontan, K. Frayn, Physical activity and exercise in the regulation of human adipose tissue physiology, Physiol. Rev. 92 (2012) 157–191. [41] E. Trevellim, M. Scorzeto, M. Olivieri, M. Granzotto, A. Valerio, L. Tedesco, R. Fabris, R. Serra, M. Quarta, C. Reggiani, E. Nisoli, R. Vettor, Exercise training induces mitochondrial biogenesis and glucose uptake in subcutaneous adipose tissue through eNOS-dependent mechanisms, Diabetes 12 (2014) pi: DB_1311234. Epub ahead of print. [42] J.M. Van Rossum, Cumulative dose–response curves. II. Techniques for the making of dose–response curves in isolated organs and the evaluation of drug parameters, Arch. Int. Pharmacodyn. Ther. 143 (1963) 299–330. [43] C. Vecchione, A. Maffei, S. Colella, A. Aretini, R. Poulet, G. Frati, M.T. Gentile, L. Fratta, V. Trimarco, B. Trimarco, G. Lembo, Leptin effect on endothelial nitric oxide is mediated through Akt–endothelial nitric oxide synthase phosphorylation pathway, Diabetes 51 (2002) 168–173. [44] V.J. Vieira, R.J. Valentine, K.R. Wilund, N. Antao, T. Baynard, J.A. Woods, Effects of exercise and low-fat diet on adipose tissue inflammation and metabolic complications in obese mice, Am. J. Physiol. Endocrinol. Metab 296 (2009) 1164–1171. [45] J. Viña, M.C. Gomez-Cabrera, C. Borras, T. Froio, F. Sanchis-Gomar, V.E. MartinezBello, F.V. Pallardo, Mitochondrial biogenesis in exercise and ageing, Adv. Drug Deliv. Rev. 61 (2009) 1369–1374. http://dx.doi.org/10.1016/j.addr.2009.06.006. [46] A. Virdis, F. Santini, R. Colucci, E. Duranti, G. Salvetti, I. Rugani, C. Segnani, M. Anselmino, N. Bernardini, C. Blandizzi, A. Salvetti, A. Pinchera, S. Taddei, Vascular generation of tumor necrosis factor-α reduces nitric oxide availability in small arteries from visceral fat obese patients, J. Am. Coll. Cardiol. 58 (2011) 238–247. [47] S.B. Withers, L. Simpson, S. Fattah, M.E. Werner, A.M. Heagerty, cGMP-dependent protein kinase (PKG) mediates the anticontractile capacity of perivascular adipose tissue, Cardiovasc. Res. 101 (2014) 130–137. [48] J. Wu, P. Bostrom, L.M. Sparks, L. Ye, J.H. Choi, A.H. Giang, M. Khandekar, K.A. Virtanen, P. Nuutila, G. Schaart, K. Huang, H. Tu, W.D. Van Marken Lichtenbelt, J. Hoeks, S. Enerback, P. Shrauwen, B.M. Spiegelman, Beige adipocytes are a distinct type of thermogenic fat cell in mouse and human, Cell 150 (2012) 366–376. [49] A. Zanesco, E. Antunes, Effects of exercise training on the cardiovascular system: pharmacological approaches, Pharmacol. Ther. 114 (2007) 307–317. [50] Y. Zhang, R. Proenca, M. Maffei, M. Barone, L. Leopold, J.M. Friedman, Positional cloning of the mouse obese gene and its human homologue, Nature 372 (1994) 425–432. [51] Y. Zhang, K.Y. Guo, P.A. Diaz, M. Heo, L. Leopold, R.L. Leibel, Determination of leptin gene expression in fat depots of lean mice, Am. J. Physiol. Regul. Integr. Comp. Physiol. 282 (2002) 226–234.