Exercise training (EX) normalizes sympathetic outflow and plasma ANG II in chronic heart failure (CHF). The central mechanisms by which EX reduces this sympathoexcitatory state are unclear, but EX may alter components of the brain renin-angiotensin system (RAS). Angiotensin-converting enzyme (ACE) may mediate an increase in sympathetic nerve activity (SNA). ACE2 metabolizes ANG II to ANG-(1–7), which may have antagonistic effects to ANG II. Little is known concerning the regulation of ACE and ACE2 in the brain and the effect of EX on these enzymes, especially in the CHF state. This study aimed to investigate the effects of EX on the regulation of ACE and ACE2 in the brain in an animal model of CHF. We hypothesized that the ratio of ACE to ACE2 would increase in CHF and would be reduced by EX. Experiments were performed on New Zealand White rabbits divided into the following groups: sham, sham + EX, CHF, and CHF + EX (n = 5 rabbits/group). The cortex, cerebellum, medulla, hypothalamus, paraventricular nucleus (PVN), nucleus tractus solitarii (NTS), and rostral ventrolateral medulla (RVLM) were analyzed. ACE protein and mRNA expression in the cerebellum, medulla, hypothalamus, PVN, NTS, and RVLM were significantly upregulated in CHF rabbits (ratio of ACE to GAPDH: 0.3 ± 0.03 to 0.8 ± 0.10 in the RVLM, P < 0.05). EX normalized this upregulation compared with CHF (0.8 ± 0.1 to 0.4 ± 0.1 in the RVLM). ACE2 protein and mRNA expression decreased in CHF (ratio of ACE2 to GAPDH: 0.3 ± 0.02 to 0.1 ± 0.01 in the RVLM). EX increased ACE2 expression compared with CHF (0.1 ± 0.01 to 0.8 ± 0.1 in the RVLM). ACE2 was present in the cytoplasm of neurons and ACE in endothelial cells. These data suggest that the activation of the central RAS in animals with CHF involves an imbalance of ACE and ACE2 in regions of the brain that regulate autonomic function and that EX can reverse this imbalance.
- sympathetic nerve activity
- angiotensin II
- central nervous system
- angiotensin-converting enzyme
chronic heart failure (CHF) is a leading cause of death among developed nations, affecting nearly 5 million Americans. CHF is characterized by heightened sympathetic tone in compensation for reduced cardiac function (20, 21). This compensation is controlled, in part, by the activation of the renin-angiotensin system (RAS). Specifically, the brain RAS is hyperactive in CHF and contributes to an increase in sympathetic nerve activity (SNA) and vasoconstriction, which exacerbate CHF (36, 48, 58).
ANG II is a prime candidate for the regulation of SNA in CHF. Our laboratory has shown that rabbits with pacing-induced CHF exhibit a clear elevation in plasma and central ANG II and a corresponding increase in SNA (25, 34). Blockade of ANG II can decrease sympathetic outflow and improve baroreflex function in CHF (40). A metabolite of ANG II, ANG-(1–7), has been shown to exhibit cardiovascular effects that are in opposition to those of ANG II. An infusion of ANG-(1–7) improves cardiac function by reducing cardiac hypertrophy induced by ANG II infusion (5, 29). A recent study (11) has also shown that injections of the selective ANG-(1–7) antagonist d-Ala7-ANG-(1–7) (A-779) into the brain attenuates baroreflex sensitivity. In addition, it is now well documented that ANG-(1–7) stimulates the production of the sympathoinhibitory substance nitric oxide (49).
Increased central angiotensin-converting enzyme (ACE) may sustain SNA by promoting the generation of ANG II. There have been limited studies focusing on the mechanisms of ACE regulation in the brain during CHF. Several studies (10, 13, 31) have observed increases in cardiac ACE in heart failure, but few have focused on the central nervous system. Studies have found increases in ACE binding density in the paraventricular nucleus (PVN) after myocardial infarction (52) and increases in ACE protein and mRNA in the hypothalamus in CHF (56). These findings suggest that ACE may also play a significant role in RAS activation in CHF in more regions of the brain than have previously been recognized.
ACE2 is a carbopeptidase that cleaves the terminal phenylalanine from ANG II to form ANG-(1–7). A recent study (12) has found ACE2 to be active in areas of the rat and mouse brain that express components of the RAS and to be localized in neurons. Overexpression of ACE2 in the rostral ventrolateral medulla (RVLM) of spontaneously hypertensive rats led to a decrease in blood pressure (54). Recently, neuron-specific overexpression of ACE2 in the subfornical organ was shown to reduce many ANG II-driven responses, such as blood pressure and water intake (16). More significantly, this study showed that ACE2 overexpression downregulated the ANG II type 1 (AT1) receptor.
Exercise training (EX) has recently been shown to be beneficial for patients with CHF (37). In humans with CHF, EX has been shown to enhance functional capacity by peripheral adaptations (4). Previous studies (27, 34, 35, 39) from this laboratory have shown that EX reduced SNA, plasma ANG II, and central AT1 receptor expression and enhanced baroreflex sensitivity in rabbits with CHF. Another study (18) has shown that EX increases tissue ANG-(1–7) levels and decreases plasma ANG II in hypertensive rats. The central mechanisms causing these beneficial effects are unknown, but the data suggest that they are mediated by changes in the upstream components of the RAS. For instance, EX decreased angiotensinogen mRNA in the nucleus tractus solitarii (NTS) in hypertensive rats (15) and plasma renin in normal humans (28). Changes in the balance between central ANG II and ANG-(1–7) during EX may be driven by similar alterations in ACE and ACE2.
Therefore, the purpose of the present study was to investigate the balance between ACE and ACE2 in CHF and to determine if EX alters the levels of ACE and ACE2 in the brain. We hypothesized that EX reduces ACE and increases ACE2 expression in the central nervous system of rabbits with pacing-induced CHF.
MATERIALS AND METHODS
Experiments were carried out on 55 male New Zealand White rabbits weighing between 3.0 and 4.5 kg (Charles River Laboratories, Wilmington, MA). These experiments were reviewed and approved by the Institutional Animal Care and Use Committee of University of Nebraska Medical Center and conformed with the National Institutes of Health Guiding Principles for the Use and Care of Laboratory Animals and guidelines of the American Physiological Society. Rabbits were randomly assigned to one of the following four groups: a normal sham-operated group (sham; n = 17), a sham + EX group (n = 10), a CHF group (n = 14), and a CHF + EX group (n = 14).
Surgical instrumentation and the induction of CHF.
CHF was induced by chronic ventricular pacing as previously described (26). A radiotelemetry catheter (Data Sciences, St. Paul, MN) was inserted into the descending aorta via a branch of the right femoral artery under general anesthesia to directly record arterial pressure and heart rate (HR). Left ventricular (LV) pacing (360–380 beats/min) was performed for 3–4 wk. Cardiac function was measured by echocardiography (Acuson Sequoia 512 C, Siemens Medical Solutions, Malvern, PA). Rabbits were hand held in the conscious state, and a two-dimensional, short-axis view of the LV was taken to measure LV internal dimensions and ejection fraction (EF). Heart failure was characterized by a reduction in EF to ∼45%, a 2-mm dilation of the LV in systole and diastole, and clinical signs of CHF, such as ascites, pulmonary congestion, and cachexia. Sham rabbits underwent the same surgical procedures but were not paced. At the end of the experiment, rabbits were anesthetized with ketamine (100 mg/kg ip). A common carotid artery was exposed by a midline incision in the neck and catheterized with a Millar transducer (model SPR-524, Millar Instruments) to measure LV pressure.
Animals in the sham + EX and CHF + EX groups were trained to run on a motor-driven treadmill. EX was performed for a total of 30 min/day for 6 days/wk. Rabbits were run for a warmup period of 5 min at 5 m/min, peak EX of 20 min at 15–18 m/min, and a cooldown period of 5 min at 5 m/min. The EX protocol was performed from the first day of rapid pacing for 3 wk.
At the end of the experiment, rabbits were euthanized with pentobarbital sodium. Their brains were removed, immediately frozen on dry ice, and blocked in the coronal plane. The cortex, cerebellum, medulla, and hypothalamus were then sectioned at −20°C in a cryostat. Additional brains were sectioned at 100 μm thickness, and the PVN, RVLM, and NTS were punched from each brain according to the methods described by Palkovits and Brownstein (26, 43). Tissues were frozen at −80°C until used for mRNA and protein analyses.
Western blot analysis.
Protein was extracted from tissue samples and homogenized using a sonicator (Sonis Vibra Cell). Protein samples (30 μg) were loaded onto 7.5% polyacrylamide gels and separated by gel electrophoresis at 40 mA/gel for 40 min. The fractionized proteins were then electrophoretically transferred at 300 mA for 90 min. Nonspecific binding was blocked on the membranes with 5% nonfat dry milk in Tris-buffered saline-0.2% Tween 20 for 1 h, and membranes were probed with rabbit polyclonal IgG primary antibodies against ACE or ACE2 protein (1:500, Santa Cruz Biotechnology, Santa Cruz, CA) overnight at 4°C. After being washed, membranes were incubated with peroxidase-conjugated goat anti-rabbit secondary antibody (1:5,000, Santa Cruz Biotechnology) for 30 min at room temperature. The signal was detected by incubating membranes with Supersignal Femto Maximum Sensitivity reagents (Pierce, Rockford, IL). The bands were imaged, and optical density was analyzed using UVP BioImaging Systems (UVP, Upland, CA). GAPDH was measured to control for protein loading.
RNA extraction, cDNA synthesis, and PCR.
Additional brain punches of the PVN, NTS, and RVLM and sections of the cerebellum and cortex (n = 4 sections/group) were used to analyze ACE and ACE2 mRNA expression. Total RNA was isolated with the RNeasy minikit (Qiagen) and treated with DNase to eliminate genomic DNA in samples. The concentration of isolated RNA was measured by spectroscopy at 260 nm. Only RNA samples exhibiting an absorbance ratio (260- to 280-nm wavelength) of >1.6 and showing integrity of the RNA by electrophoresis were used in further experiments. cDNA was generated from 3 μg RNA by RT-PCR. Gene-specific primers for ACE (forward: 5′-GGATCCAACAAGACTGCCACCTGCTGG-3′ and reverse: 5′-AACATGTGGGCCCAGAGCTGGGTCGAC-3′), ACE2 (forward: 5′-ATGCCTCCCTGCTCATTTGCTT-3′ and reverse: 5′-AGCAGGAAAGGTTGCTTGGCAT-3′), and β-actin (forward: 5′-CTACAGCTTCACCACCAC-3′ and reverse: 5′-GCAGCTCGTAGCTCTTCTC-3′) were generated at the Eppley DNA synthesis Core Facility of the University of Nebraska Medical Center. The cDNA obtained after reverse transcription was then amplified with Hotstart Taq DNA polymerase (Qiagen) for 30 cycles as follows: 1 min at 92°C to denature the double-stranded DNA, 1 min at 50°C to allow reaction of the annealing primers, and 5 min at 69°C for primer extension. Samples were also amplified for β-actin as an internal control to insure that equal amounts of RNA were loaded. Aliquots of 5 μl were then separated by electrophoresis in a 1.5% agarose gel in Tris-borate-EDTA buffer. Gels were then visualized by ethidium bromide staining, and the optical density ratio of ACE or ACE2 to β-actin was quantified.
To determine whether ACE and ACE2 protein expression were present in neuronal or endothelial cells, a double-immunofluorescence protocol was used. Coronal sections (30 μm) of rabbit brains were made in a cryostat. Sections were fixed with 4% paraformaldehyde in 0.1 M PBS for 10 min and washed in PBS. Sections were permeabilized in chilled acetone for 10 min followed by antigen retrieval in 1 mM EDTA and 0.05% Tween 20 buffer for 20 min. Nonspecific staining was blocked by 10% normal goat serum (Sigma), 0.5% Triton X-100, and 1.0% BSA for 1 h at room temperature.
Sections were then simultaneously incubated in primary antibodies of goat polyclonal ACE or ACE2 (1:250, Santa Cruz Biotechnology) along with mouse anti-NeuN (a neuronal marker; 1:250, Millipore) or mouse CD31 (an endothelial marker; 1:250) overnight at 4°C. Sections were then stained in Alexa fluor 594-labeled donkey anti-mouse (1:500, Invitrogen) and Alexa fluor 488-labeled donkey anti-goat (1:500) secondary antibodies. The nuclei were then stained with 4′,6-diamidino-2-phenylindole, and coverslips were mounted with Fluoromount G (Southern Biotechnology Associates, Birmingham, AL). Immunostaining was imaged using confocal microscopy (Zeiss Confocal LSM 510), and specific lasers and channels for the emission of fluorescence were used according to each fluorescent compound. Images were edited and analyzed using Zeiss LS Image Browser software.
Data are expressed as means ± SE. All statistical analysis was performed with Prism 3.0 software (GraphPad Software, San Diego, CA). Hemodynamic data between groups and for Western blot density and mRNA expression were analyzed with one-way ANOVA followed by the Newman Keuls post hoc test. Significance between pre- and postexperiment HRs within groups was analyzed by a paired t-test. P < 0.05 was considered as statistically significant.
Body weight, organ weight, hemodynamics, and echocardiographic data.
Table 1 shows the values for body weight, organ weight-to-body weight ratios, hemodynamics, and echocardiographic data from rabbits in the four groups studied. The CHF group exhibited a significantly higher wet lung weight-to-body weight ratio, suggesting pulmonary edema. Further clinical signs of CHF were observed, including pleural effusion, ascites, and subcutaneous edema. The CHF group exhibited a significantly lower EF and fractional shortening along with higher resting HR, LV end-diastolic diameter and pressure, and LV systolic diameter compared with the sham group, indicating a decrease in cardiac function characterizing CHF. EX significantly reduced HR in the sham + EX group compared with their HR before the beginning of the experiment, suggesting a training effect. In addition, CHF + EX rabbits had a slightly, although not significant, lower postexperiment HR compared with the beginning of the experiment and a significantly lower postexperiment HR compared with CHF rabbits. EX had no significant effects on the other indexes of CHF, suggesting that EX may exhibit its beneficial effects in the peripheral circulation rather than on the myocardium per se.
Protein expression of brain ACE and ACE2 in CHF.
In this experiment, we measured the protein expression of both ACE and ACE2. As shown in Fig. 1, the protein expression of ACE increased significantly in the CHF group in the cerebellum, medulla, and hypothalamus and in RVLM, NTS, and PVN nuclei. ACE expression did not increase in CHF in the cortex (data not shown). Conversely, the protein expression of ACE2 significantly decreased in the cerebellum, hypothalamus, RVLM, and NTS (see Fig. 3). There was a trend for ACE2 protein expression to decrease in the medulla and PVN, but this did not reach statistical significance. These data suggest an inverse relationship between ACE and ACE2 in CHF.
Gene expression of ACE and ACE2 in CHF.
We also measured the mRNA expression of ACE and ACE2 in CHF. ACE mRNA expression was significantly upregulated in the cerebellum, RVLM, NTS, and PVN in the CHF state (Fig. 2). ACE mRNA expression did not change in the cortex. Again, ACE2 showed an inverse relationship as ACE2 mRNA expression was downregulated in cerebellum, RVLM, NTS, and PVN nuclei compared with CHF (see Fig. 4). ACE2 mRNA expression was also not altered in the cortex (data not shown).
Effect of EX on ACE and ACE2.
The sham + EX group exhibited no change in ACE protein expression. However, ACE2 protein tended to be upregulated in the PVN, NTS, and hypothalamus in the sham + EX group but only reached statistical significance in the PVN (Fig. 3). There was a normalization of ACE protein and mRNA expression in CHF + EX rabbits in all areas measured (Figs. 1 and 2). EX also restored the protein and mRNA levels of ACE2 to near the levels seen in sham rabbits (Figs. 3 and 4). In addition, ACE2 protein and mRNA expression were upregulated in the RVLM in CHF + EX animals compared with sham animals.
Localization of ACE and ACE2.
To determine the type of brain cells expressing and mediating changes in ACE and ACE2, double immunofluorescence was performed in the RVLM of tissues from sham rabbits with cell-specific antibodies. ACE staining overlapped with CD31 (red; an endothelial marker), as shown in Fig. 5. On the other hand, ACE2 and NeuN colocalize in the cytoplasm of neurons, as shown in Fig. 6. These results suggest that, at least in the RVLM, ACE2 is expressed primarily in the cytoplasm of neurons and ACE is localized primarily in vascular endothelial cells.
The present study was carried out to determine the effects of EX on the regulation of RAS components ACE and ACE2 in the brains of normal and CHF rabbits subjected to a moderate EX protocol. This study demonstrates that ACE is upregulated in the brain of rabbits with CHF, which is accompanied by a downregulation of ACE2. EX normalized the expression of both ACE and ACE2. These data suggest that EX in CHF activates the ANG-(1–7) axis of the brain RAS and may involve a normalization of ACE and ACE2 to regulate the balance between ANG II and ANG-(1–7) and ostensibly sympathetic outflow.
All components of the RAS, including ACE, ACE2, ANG II, ANG-(1–7), and their receptors, are known to be present in the brain, especially in areas that regulate sympathetic outflow (8, 12, 36), suggesting that the brain has a local RAS that may regulate sympathetic function that is much more complex than previously thought. ACE and ACE2 have been extensively studied in the heart. Both appear to be elevated in the heart during CHF and may participate in the cardiac remodeling process (29). However, the balance of ACE and ACE2 in the brain in CHF is not as clear.
Recent findings have suggested a role of EX to modify the RAS, including decreasing ANG II (34) and AT1 receptor expression (39) in rabbits with CHF and increasing ANG-(1–7) and its Mas receptor in hypertensive rats (18). The present study suggests that a decrease in ACE and an increase in ACE2 after EX may lead to a normalization of ANG II and ANG-(1–7). However, neither of these peptides were measured in the brain in the present study.
The results of this study support increasing evidence that indicates an imbalance between ACE and ACE2 in CHF favoring ANG II production. Several studies (31, 46, 51, 52) have indicated an increase ACE in autonomic regions of the brain and in cardiac tissue in CHF. This suggests that increased ANG II in CHF may be due to an increase in its converting enzyme in the brain. Increased brain ACE expression in CHF is not a universal finding. Yoshimura et al. (55) found no increase in ACE mRNA expression in a high-output CHF model. However, this may be explained by the use of an aortocaval shunt method to induce CHF as opposed to the low-output pacing model in the present study.
The regulation of ACE2 in the brain is not as clear. Studies (3, 17, 57) of ACE2 in the heart in CHF have found increases of both ACE2 and ANG-(1–7). Differences in ACE2 expression between the brain and heart may be due to the role of ANG-(1–7) in cardiac remodeling as opposed to sympathetic regulation.
EX significantly reduced HR in sham + EX and CHF + EX rabbits compared with CHF sedentary rabbits. This suggests an endurance effect. Other training protocols have also reported a decrease in HR in normal animals (6). However, we did not observe changes in other cardiac parameters with EX. This suggests that the benefits of EX in CHF may be due to a peripheral vasodilatory effect. In addition, the lack of effect on other cardiac parameters may be because our study used a rather low level of EX to maintain consistency between normal and CHF rabbits.
The normalization of the balance between ACE and ACE2 in the brain after EX is also consistent with other studies (18, 45) of EX in CHF suggesting an activation of the ACE2 and ANG-(1–7) axis of the RAS. The finding that EX reduces ACE expression and increases ACE2 in CHF in the brain may explain previously found benefits of EX. For example, a decrease in plasma ANG II after EX (30, 34) may be due to a decrease in ACE expression. Similarly, an increase in ANG-(1–7) after EX (18) may be due to an increase in ACE2 expression. Feng et al. (16) have shown that overexpression of ACE2 initiates a decrease in AT1 receptor expression in cell culture and in vivo. This suggests that EX may reduce AT1 receptor expression (39), in part, by an ACE2-dependent mechanism. Similar changes in ACE and ACE2 in most nuclei and regions studied suggest that the normalization of ACE and ACE2 by EX may be a global effect observed in many brain nuclei.
The precise molecular mechanisms underlying the imbalance of ACE and ACE2 in CHF and the normalization of this imbalance with EX were not investigated in this study. Recent data from a study by Koka et al. (33) have suggested that the p38 MAPK and ERK1/2 signaling pathway is involved in ACE upregulation and ACE2 downregulation. We (27) have also shown that SOD contributes to the normalization of SNA and baroreflex function in CHF after EX. This suggests an important contribution of oxidant stress in the sympathoexcitatory process in CHF. Enhancement of antioxidant pathways in the brain with EX may also lead to the normalization of ACE and ACE2. Since there are several redox-sensitive transcription factors (1, 19, 50), this may be a transduction mechanism by which EX alters protein transcription in the brain. Finally, sympathoexcitation in CHF is known to be, in part, dependent on central inflammatory mediators, such as IL-6 and TNF-α (22). Recent evidence has indicated that EX lowers inflammatory substances in the brain of rats (9). Therefore, it is highly likely that both ACE and ACE2 may be regulated by the existing humoral environment in CHF and after EX.
It is not clear why there was little effect of EX on ACE protein expression in sham rabbits. Low ACE in the normal state or the level and duration of the EX may have prevented changes in sham rabbits. This also suggests that the imbalance between ACE and ACE2 is particularly activated by EX in pathological conditions such as CHF. However, ACE and ACE2 mRNA may be altered by EX in the normal state, but this was not examined in the present study due to a lack of sufficient tissue in this group.
The imbalance between ACE and ACE2 expression in the cerebellum in CHF and its normalization after EX is of interest. This suggests that there may be high levels of ANG II and low levels of ANG-(1–7) in the cerebellum in CHF. The cerebellum is functionally related to movement and coordination, not autonomic function. However, the fastigial nucleus has been shown to be intimately involved in blood pressure and sympathetic regulation (38). Importantly, other studies of the cerebellum have also reported changes in ACE and ACE2. In a recent study by Gallagher et al. (24), rat astrocytes from the cerebellum showed a decrease in ACE2 after treatment with ANG II. In addition, ACE and ACE2 may have multiple roles in the brain (2, 23, 42, 53).
To localize ACE and ACE2 expression to specific cell types in the brain, we performed double immunofluorescence. In the RVLM, ACE appeared to be associated with the endothelium of blood vessels, colocalizing with the endothelial marker CD31. On the other hand, ACE2 was colocalized with neurons staining positive for the neuronal marker NeuN. These findings agree with other studies (7, 14, 41) that have localized ACE in vascular endothelial cells in several areas, including the lung, heart, liver, kidney, and pancreas. ACE2 has also been localized in the cytoplasm of neuronal cells in specific nuclei in the mouse brain, including the RVLM, NTS, and PVN (12). A study (24) performed in brain cell cultures from the cerebellum and medulla has reported that ACE2 is expressed in glial cells. However, this finding may be due to the use of cultured astrocytes from neonatal rats and developmental changes in ACE2 localization. The presence of intracellular ACE2 may seem to limit ANG-(1–7) to only intracellular actions. However, both ACE and ACE2 have been shown to be membrane bound and secreted from cells (23, 47, 53). Therefore, central ACE and ACE2 may have a significant role in the formation of plasma and tissue ANG II and ANG-(1–7). If ACE is confined to the endothelium in the RVLM or other structures, this begs the following question: what influence might the vascular production of ANG II have on cells within the RVLM given the assumption that a tight blood-brain barrier exists at this location? The answer to this is not apparent, but one might speculate that the local production of intravascular ANG II might act on vascular AT1 receptors to have indirect actions on neurons in this area (44).
Another potential limitation of these findings is that our immunofluorescence techniques were only able to detect ACE2 protein in the intracellular compartment. Therefore, it is not clear if the changes in ACE2 that we describe alter the balance between ANG II and ANG-(1–7) in the interstitial space.
In conclusion, the new findings in this study may have important ramifications for the treatment of CHF. The development of ACE inhibitors targeted to the brain may be even more effective for CHF treatment. Increasing ACE2 and potentially ANG-(1–7) in the brain may be used as a strategy for reducing central sympathetic outflow in CHF. EX in CHF is now an area of intense investigation. While many studies (4, 28) have shown beneficial effects of EX in CHF, the mechanism by which this occurs is largely unknown. The data provided here implicate a central mechanism in which the balance between ANG II and ANG-(1–7) may be normalized due to a change in their converting enzymes after EX. On the basis of the present work, it may also be possible to design an EX regimen to normalize ACE and ACE2 levels in CHF without pharmacological interventions.
This work was supported by National Heart, Lung, and Blood Institute Grant PO1-HL-62222.
No conflicts of interest are declared by the author(s).
The authors thank Janice A. Taylor and James R. Talaska (Confocal Laser Scanning Microscope Core Facility, University of Nebraska Medical Center) for providing assistance with the confocal microscopy and the Nebraska Research Initiative and Eppley Cancer Center for the support of the Confocal Laser Scanning Microscope Core Facility. The authors also acknowledge the expert technical assistance of Johnnie F. Hackley, Pamela Curry, Phyllis Anding, and Li Yu. The authors thank Dr. James Anderson for help with the statistical analysis of this study.
- Copyright © 2010 the American Physiological Society