Adenosine and acetylcholine (ACh) trigger preconditioning through different signaling pathways. We tested whether either could activate myocardial phosphatidylinositol 3-kinase (PI3-kinase), a putative signaling protein in ischemic preconditioning. We used phosphorylation of Akt, a downstream target of PI3-kinase, as a reporter. Exposure of isolated rabbit hearts to ACh increased Akt phosphorylation 2.62 ± 0.33 fold (P = 0.001), whereas adenosine caused a significantly smaller increase (1.52 ± 0.08 fold). ACh-induced activation of Akt was abolished by the tyrosine kinase blocker genistein indicating at least one tyrosine kinase between the muscarinic receptor and Akt. ACh-induced Akt activation was blocked by the Src tyrosine kinase inhibitor 4-amino-5-(4-chlorophenyl)-7-(t-butyl)pyrazolo[3,4-d]pyrimidine (PP2) and by 4-(3-chloroanilino)-6,7-dimethoxyquinazoline (AG-1478), an epidermal growth factor receptor (EGFR) inhibitor, suggesting phosphorylation of a receptor tyrosine kinase in an Src tyrosine kinase-dependent manner. ACh caused tyrosine phosphorylation of the EGFR, which could be blocked by PP2, thus supporting this receptor hypothesis. AG-1478 failed to block the cardioprotection of ACh, however, suggesting that other receptor tyrosine kinases might be involved. Therefore, Gi protein-coupled receptors can activate PI3-kinase/Akt through transactivation of receptor tyrosine kinases in an Src tyrosine kinase-dependent manner.
- epidermal growth factor receptor
- Src tyrosine kinase
- phosphatidylinositol 3-kinase
ischemic preconditioning (IPC), first described by Murry et al. (25), refers to the remarkable reduction of infarct size that is observed when the heart has been exposed to a short period of ischemia before a prolonged ischemic insult. In the rabbit heart protection is triggered by occupation of Gi-coupled surface receptors, including those for adenosine, bradykinin, and opioids (32). An important step in the signaling pathway for preconditioning is protein kinase C (PKC) activation (45). Tong et al. (37) and more recently Mocanu et al. (20) demonstrated that activation of phosphatidylinositol 3-kinase (PI3-kinase) is also involved in the cardioprotective effects of ischemic preconditioning and that it is upstream of PKC. We recently noted that the various Gi-coupled receptors that trigger the protection of preconditioning do not necessarily use the same signaling pathway (7). Whereas most Gi-coupled receptors (i.e., muscarinic receptor) that have been tested triggered protection through a pathway dependent on the opening of ATP-sensitive K+channels (KATP) and generation of free radicals, the A1 adenosine receptor behaved quite differently and did not require either to protect the heart. We recently presented evidence that PI3-kinase may link the Gi-coupled muscarinic receptor to KATP (28). Wortmannin (Wort) blocked the acetylcholine (ACh)-induced generation of reactive oxygen species (ROS) in a vascular smooth muscle cell model, and ROS production was dependent on opening of KATP. If the above arrangement proved to be the case in the heart, then a muscarinic agonist such as acetylcholine would be expected to activate PI3-kinase, whereas adenosine might not.
In this study, we investigated the ability of ACh and adenosine to activate PI3-kinase in isolated rabbit hearts. The 57-kDa serine/threonine protein kinase Akt (also known as protein kinase B) is known to be a primary downstream target of PI3-kinase (4). So far, three isoforms of Akt have been identified (Akt1, -2, and -3) but it is still not fully known which isoforms are activated by specific ligand stimulation under physiological conditions (27). Akt1 is expressed at high levels in the heart and is activated through the hierarchical phosphorylation of both Thr308 and Ser473 by 3′-phosphoinositide-dependent kinase-1 (PDK1) and PDK2, respectively, and these latter kinases are both activated by the PI3-kinase lipid products phosphatidylinositol 3,4,5-trisphosphate [PtdIns(3,4,5)P 3] and PtdIns(3,4)P 2. In this study, we used phosphorylation of Akt1 at Ser473 as a reporter of PI3-kinase activation (5).
Because tyrosine kinase is well known to be involved in preconditioning (3, 23, 39), we also tested whether a tyrosine kinase might be required for Akt activation by receptor agonists. Finally, Gi-coupled receptors in other cell types are known to activate PI3-kinase through a non-ligand-dependent transactivation of receptor tyrosine kinases, including epidermal (EGFR), platelet-derived, and insulin-like growth factor receptors (18). We investigated this possibility by examining tyrosine phosphorylation of one of these receptor tyrosine kinases, the EGFR, and whether Src is down- or upstream of the EGFR in this signaling pathway.
MATERIALS AND METHODS
All experiments were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals (Revised 1996).
Isolated rabbit heart model.
New Zealand White rabbits of either sex were used. As previously described (41), hearts were removed from animals anesthetized with pentobarbital sodium, mounted on a Langendorff apparatus, and perfused with modified Krebs-Henseleit bicarbonate buffer containing (in mM) 118.5 NaCl, 25 NaHCO3, 4.8 KCl, 1.2 MgSO4, 1.2 KH2PO4, 1.7 CaCl2, and 10 glucose. As shown in Fig.1, seven groups were studied. All hearts were subjected to 30 min of global ischemia by arresting retrograde aortic perfusion. Control hearts were subjected to only 30 min of global ischemia. Hearts that were treated with either ACh or adenosine were electrically paced until the onset of ischemia to prevent bradycardia. In the drug-treated hearts, either 0.55 mM ACh or 100 μM adenosine was added to the perfusate for 5 min, followed by 10 min of washout before the onset of ischemia. In four groups of hearts exposed to ACh, an infusion of either 100 nM Wort, a blocker of PI3-kinase, 50 μM genistein (Gen), an inhibitor of tyrosine kinase, 1 μM 4-amino-5-(4-chlorophenyl)-7-(t-butyl)pyrazolo[3,4-d]pyrimidine (PP2), a Src tyrosine kinase inhibitor, or 300 nM 4-(3-chloroanilino)-6,7-dimethoxyquinazoline (AG-1478), an EGFR kinase inhibitor, was started 5 min before ACh and continued for either 15 min (Gen, PP2, and AG-1478) or 20 min (Wort). Serial transmural biopsies of the left ventricle, each weighing ∼25 mg, were obtained with a motorized biopsy tool. The tissue sample was ejected from the tool into liquid nitrogen within 1 s of excision. Serial biopsies were taken at the times indicated by arrows in Fig. 1.
Tissue preparation and immunoblotting.
Biopsies were homogenized with a Polytron homogenizer in ice-cold lysis buffer containing 20 mM Tris · HCl (pH 7.5), 150 mM NaCl, 1 mM Na2EDTA, 1 mM EGTA, 1% Triton, 2.5 mM sodium pyrophosphate, 1 mM β-glycerophosphate, 1 mM Na3VO4, and 1 μg/ml leupeptin. The samples were centrifuged for 15 min at 13,000 g, and the protein content of the supernatant was determined by the Bradford technique. After homogenization, the samples were immediately analyzed by standard SDS-PAGE methods. Samples were electrophoresed on a 10% polyacrylamide gel, followed by transfer to a nitrocellulose membrane. For phospho-EGFR separation, 6% polyacrylamide gel was used. Equal amounts of total protein were loaded in each lane. After being blocked with 5% milk, the membranes were treated with the primary antibody (1:1,000), followed by the secondary antibody (1:5,000) conjugated to horseradish peroxidase. Immunoreactive proteins were detected by enhanced chemiluminescence with LumiGLO (Cell Signaling Technology; Beverly, MA). Films were developed and the blots quantified by using a computer scanner, and the density of each band was calculated with Sigmagel software.
To correct for differences in protein loading and errors introduced during the transfer process, each membrane was stained with Ponceau S for 15 min at the end of each experiment as described previously (30). The stained membranes were scanned, and the density of at least two major bands at ∼45–50 kDa were calculated with Sigmagel software. The Ponceau stain density of each lane was expressed as a percentage of the average Ponceau stain density of each membrane. The final densities of the phosphorylated Akt lanes were then corrected using the corresponding result of the Ponceau stain and finally expressed as a percentage of that in the baseline sample.
Infarct size measurement.
The hearts were mounted on a Langendorff apparatus and perfused with modified Krebs-Henseleit bicarbonate buffer as described above. A 2-0 silk suture was passed around a branch of the left coronary artery with a taper needle and a snare formed by passing the ends of the thread through a small vinyl tube. All hearts were subjected to 30 min of regional ischemia by tightening the snare, followed by 120 min of reperfusion. Control hearts were subjected to regional ischemia-reperfusion only. In a second group, ACh (0.55 mM) was added to the perfusate for 5 min, followed by 10 min of washout before onset of ischemia, and in the third group the EGFR kinase blocker AG-1478 (300 nM) was present in the buffer from 5 min before onset of ACh exposure until 5 min after ACh was discontinued.
At the end of the experiment, the coronary artery was reoccluded, and 2–9 μm fluorescent polymer microspheres (Duke Scientific; Palo Alto, CA) were infused into the perfusate to demarcate the ischemic zone (region at risk) as the area of tissue without fluorescence. The heart was weighed, frozen, and then cut into 2-mm-thick slices. The slices were incubated in 1% triphenyltetrazolium chloride in sodium phosphate buffer (pH 7.4) at 37°C for 20 min. Triphenyltetrazolium chloride stains noninfarcted myocardium brick red. The slices were then immersed in 10% formalin to preserve the stained (viable) and unstained (necrotic) tissue. The risk zone was identified by illumination of the slices with ultraviolet light. The areas of infarct and risk zone were determined by planimetry of each slice and the volumes calculated by multiplying each area by the slice thickness and summing them for each heart. Infarct size is expressed as a percentage of the risk zone.
Phospho-Akt antibody to Ser473 (no. 9271), HRP-linked anti-rabbit IgG antibody, and cell lysis buffer were obtained from Cell Signaling Technology. Phospho-EGFR antibody to Y1173 (no. 05-483) and HRP-conjugated goat anti-mouse IgG antibody were from Upstate Biotechnology (Waltham, MA). Wort, Gen, ACh, adenosine, and Ponceau S solution were from Sigma (St. Louis, MO), whereas PP2 and AG-1478 were obtained from Calbiochem (La Jolla, CA). Wort, Gen, adenosine, PP2, and AG-1478 were dissolved in DMSO before being diluted in Krebs-Henseleit buffer, resulting in a DMSO concentration <0.01%. ACh was diluted directly in Krebs-Henseleit buffer. The stock solutions for ACh, adenosine, and AG-1478 were made fresh before every experiment.
Results are given as means ± SE. To analyze effects of drugs on hemodynamics, differences between baseline and 5 min of drug infusion were examined with the paired t-test. To further determine the effects of interventions on PI3-kinase activation, mean normalized levels of Akt phosphorylation over time were reduced to single points by calculating the area under the curve for each heart. The areas for different groups were then compared with one-way ANOVA and Tukey's post hoc test. A P value of <0.05 was considered to be significant.
Table 1 summarizes the body weight, heart weight, and baseline hemodynamic data (heart rate, coronary flow, and peak systolic pressure) obtained for all groups undergoing serial biopsies. Table 2 presents infarct and serial hemodynamic data for hearts used in the infarct size experiments. There were no differences at baseline between any of the groups. ACh depressed heart rate, peak systolic pressure, and coronary flow. Systolic pressure and coronary flow fell further during coronary occlusion with a rebound after reperfusion. The infusion of AG-1478 before ACh had no effect on hemodynamic variables (data not shown).
Akt phosphorylation after ACh and adenosine.
In untreated control hearts Akt phosphorylation was unchanged during the first 15 min of ischemia and fell below the baseline value after 30 min of ischemia (Figs. 2 A and3). After treatment for 5 min with ACh, the amount of phosphorylated Akt (phospho-Akt) increased >2.5-fold (262 ± 33% of baseline, P = 0.001 vs. pre-ACh) and remained elevated even after ACh was washed out for 5 min (Figs.2 B and 3). Phosphorylation gradually decreased toward baseline during the following 30 min of global ischemia. Figure2 C reveals that phosphorylation increased as well in the group treated with adenosine (152 ± 8% of baseline,P = 0.001 vs. preadenosine) (Fig.3). When the response over time was analyzed with ANOVA, ACh caused a significant increase in phosphorylation over that in untreated hearts (P = 0.006). This response was also greater than that seen in hearts treated with adenosine (P = 0.027). However, the difference between the adenosine-treated hearts and the untreated ones did not reach statistical significance (P = 0.48).
Blockade of ACh-induced Akt phosphorylation.
As shown in Fig. 4 A, pretreatment with Wort greatly depressed phosphorylation of Akt and there was no increase in Akt phosphorylation after ACh. Pretreatment with Gen (Fig. 4 B) had no effect on Akt phosphorylation itself, but completely blocked the ability of ACh to increase it, suggesting that the pathway between the muscarinic receptor and phosphorylation of Akt contains at least one tyrosine kinase. PDK is a serine/threonine kinase and should be unaffected by Gen. Because muscarinic receptor activation has been reported to cause transactivation of receptor tyrosine kinases, including the EGFR (18), we tested for possible involvement of the latter by adding the EGFR kinase inhibitor AG-1478 to the perfusate of ACh-treated hearts. AG-1478 had no influence on hemodynamics during infusion (data not shown), but it completely blocked the ability of ACh to increase Akt phosphorylation (Fig. 4 C). Some reports suggest that Src tyrosine kinase plays an upstream role in receptor tyrosine kinase transactivation (19). The Src family tyrosine kinase inhibitor PP2 caused no significant increase of Akt phosphorylation before ACh and completely blocked the increased phosphorylation after ACh (Fig. 4 D). Figure5 is a summary of the phosphorylation data from all groups. The dramatic increase in phosphorylation of Akt induced by ACh is greatly attenuated by all four antagonists.
Phosphorylation of EGFR.
When the EGFR is activated by a ligand, autophosphorylation of its tyrosine residues occurs. To confirm that ACh also activated the EGFR, we used a phosphospecific antibody against tyrosine phosphorylation on site Y1173 of the EGFR. The EGFR became more phosphorylated after 5 min of treatment with ACh (increase of 2.3 ± 0.52-fold vs. baseline,P = 0.04, n = 6), thus confirming transactivation. Figure 6 Ashows a representative blot. This increase could be totally abolished by the Src family tyrosine kinase inhibitor PP2 as seen in Fig.6 B, suggesting that transactivation of EGFR is dependent on Src.
EGFR blockade failed to prevent the cardioprotection of ACh.
We also tested whether blocking the EGFR with AG-1478 could abort the protective effect of ACh. Whereas control hearts had infarct sizes of 37.8 ± 6.2% of the risk zone, the group pretreated with 5 min of ACh demonstrated protection (infarct size 9.3 ± 3.5%,P < 0.001 vs. untreated), which could not be abolished by the EGFR kinase blocker AG-1478 (infarct size 6.2 ± 2.4%,P < 0.001 vs. untreated and not significant vs. ACh) (Fig. 7).
The signal transduction pathways that mediate ischemic preconditioning in the heart remain to be defined in detail. The present study demonstrates in the Langendorff-perfused rabbit heart that ACh, and, to a lesser extent, adenosine, are both activators of the PI3-kinase/Akt signaling pathway. Furthermore, Src tyrosine kinase and transactivation of the EGFR are important steps in ACh-induced Akt activation.
It is well documented that Akt is a downstream target of PI3-kinase and mediates various physiological processes, including cell survival and insulin actions (4, 5). Akt contains an amino-terminal pleckstrin homology domain, which responds to PI3-kinase lipid metabolites PtdIns(3,4,5)P 3 and PtdIns(3,4)P 2, causing Akt to translocate to the plasma membrane. After translocation Akt becomes activated by phosphorylation within its kinase domain at Thr308 and near the carboxyl terminus at Ser473 (2). Phosphorylation of both sites is required for maximum activation (1). PDK1 is responsible for phosphorylation of Thr308, whereas PDK2 phosphorylates Ser473 (27, 33). Because both of the PDKs are dependent on the PI3-kinase lipid products PtdIns(3,4,5)P 3 and PtdIns(3,4)P 2, phosphorylation of Akt at Ser473 should serve as a good reporter of PI3-kinase activation (5).
There is recent evidence that the activation of PI3-kinase is upstream of PKC in the signaling pathway of IPC (20, 37). IPC is triggered by occupation of several Gi-coupled receptors on the cardiomyocyte including the opioid δ, bradykinin B1, and adenosine A1 receptors. Several other Gi-coupled receptors exist on the heart, e.g., muscarinic and endothelin, which do not contribute to IPC because ligands for those receptors are not released during ischemia. Nevertheless, their occupation will also put the heart into a preconditioned state (32). The ability of the muscarinic receptor to mimic IPC has been extensively studied (36, 42-44). We (7) recently reported that in rabbit hearts either 5-hydroxydecanoate, an antagonist of the mitochondrial KATPchannel, or N-(2-mercaptopropionyl)glycine, a free radical scavenger, could block the cardioprotective effect of agonists to all Gi-coupled receptors tested except for the adenosine A1 receptor. Apparently adenosine and ACh trigger the preconditioning mechanism through different but parallel signaling pathways. To explain these observations we proposed that receptors in the group typified by the muscarinic receptor cause KATPchannels on the inner mitochondrial membrane to open. The channel opening, in turn, causes mitochondria to produce ROS, which then activate downstream protective pathways, including PKC (29). We recently explored this pathway in vascular smooth muscle cells (28). We found that activation of the muscarinic receptor led to the release of ROS from mitochondria and that its production could be abrogated by KATP channel blockers. More importantly, ROS release could also be prevented by the potent PI3-kinase blocker Wort. This suggests, at least in smooth muscle cells, that PI3-kinase is an important step in the pathway between the receptor and the KATP channels. The present study confirms that ACh can activate PI3-kinase in rabbit heart.
We were interested whether adenosine, which does not require KATP opening or ROS generation to trigger protection, could also activate PI3-kinase. Exposure to 100 μM adenosine, which would have saturated A1 receptors, caused less phosphorylation of Akt than did ACh but still increased it. We would conclude that failure of either free radical scavengers or KATP blockers to prevent the protection of adenosine is not because adenosine fails to activate the PI3-kinase pathway but rather because it probably has an additional coupling that can activate the downstream protective pathways even when the mitochondrial pathway is blocked.
How then does the muscarinic receptor actually activate PI3-kinase in the heart? The most obvious scenario would be that Gidirectly activates PI3-kinase-γ, an isoform that is known to be Gi dependent in other cell types (18). As shown in Fig. 4, however, the ACh-induced Akt activation could be blocked by the addition of Gen, indicating that at least one tyrosine kinase is part of that pathway. If a direct coupling between PI3-kinase γ and Gi were involved, then the only kinase between the receptor and Akt would be the PDKs, which are serine/threonine kinases. Therefore, this pathway could not explain our data.
The usual scheme for activation of PI3-kinase involves a receptor tyrosine kinase. Occupation of the receptor by its ligand causes autophosphorylation of tyrosine groups within the receptor, which in turn allows it to activate PI3-kinase. There are several reports in various cell types that several receptor tyrosine kinases, including EGFR, are transactivated when Gi subunits are liberated in response to muscarinic receptor occupation (8, 14, 38). This ligand-independent transactivation of the EGFR leads to activation of PI3-kinase. The mechanism seems to involve ErbB3, a member of the ErbB receptor protein tyrosine kinase subfamily, which is phosphorylated by the activated EGFR. ErbB3 then activates PI3-kinase through its p85 regulatory subunit (16, 34).
Several studies report that the protection of IPC can be blocked by tyrosine kinase antagonists, either by themselves (3, 9,22) or in combination with PKC inhibitors (10, 35,39). Could EGFR be the tyrosine kinase in question? Events in IPC can be divided into those that trigger protection and act before the ischemic insult and those that mediate it and act during the insult. Receptor occupancy and mitochondrial KATPopening can be shown to act as triggers (7, 29, 40). Fryer et al. (9) also showed that a tyrosine kinase acts as a trigger the protection of IPC.
Whereas three receptor tyrosine kinases have been identified to participate in Gi-dependent transactivation, an inhibitor is available for only one of them, the EGFR. We used the potent and selective EGFR kinase inhibitor AG-1478 to test whether ACh-induced Akt activation was dependent on EGFR activation in our model. Indeed, as shown in Fig. 4 C, the increase in Akt phosphorylation by ACh could be completely blocked by AG-1478. To confirm this novel finding we measured phosphorylation of the EGFR itself by using a phosphospecific antibody. ACh caused brisk phosphorylation of the cardiac EGFR (Fig. 6 A) confirming that the latter had been transactivated.
The Src family of tyrosine kinases plays an important role in the transactivation process either by phosphorylating the receptor directly or by phosphorylating some important component of the receptor tyrosine kinase's signaling unit (19). Indeed, Mockridge et al. (24) found that PI3-kinase activation in a model of simulated ischemia-reperfusion in cardiomyocytes could be blocked by the Src kinase inhibitor PP2. PP2 also could block protection from IPC in rat heart (15). PP2 completely blocked Akt activation from ACh in the present study supporting a role for an Src kinase in this pathway. Because we found protection from the opioid receptor to be both KATP and ROS dependent just like that from the muscarinic receptor (7), it is interesting that Fryer at al. (11) could not block opioid-induced protection with PP2 in rat hearts. To further test whether Src plays a role in activation of EGFR, we tried to block the ACh-induced increase in EGFR phosphorylation with PP2. As shown in Fig. 6 B, the increase in EGFR phosphorylation is totally abolished in the presence of PP2 confirming participation of Src kinase in transactivation.
Taken together, we would propose that activation of cardiac muscarinic receptors causes a Src-dependent transactivation of receptor tyrosine kinases, including EGFR that in turn activate PI3-kinase leading to the eventual phosphorylation of Akt. Because blocking only the EGFR was so effective in interfering with the ability of ACh to activate Akt, we wondered whether using only the EGFR inhibitor AG-1478 could also block the ability of ACh to trigger preconditioning. To our surprise, AG-1478 failed to block the cardioprotective effects of ACh. A reasonable explanation would be that the EGFR is not the only receptor tyrosine kinase involved in this pathway and other unblocked receptors such as the insulin-like growth factor and the platelet-derived growth factor receptors (12, 18) could still carry the signal. The other explanation would be that the PI3-kinase/Akt pathway is not as central to ACh-induced preconditioning as we thought. We have tested Wort, PP2, and Gen in a similar setting and found that all could block the protection of ACh but not adenosine (31), suggesting that this pathway is indeed important.
The protective effect of IPC is thought to occur during the ischemic insult (6). For that reason, a popular strategy has been to examine biochemical parameters during the ischemic period to see whether a difference can be detected between protected and unprotected hearts (13, 41). The data in this study show that Akt is activated after 5 min of ACh infusion and remains activated, even during the following ischemic period. It should be noted, however, that the level of Akt phosphorylation declines back toward baseline during ischemia. It is not known whether this residual phosphorylation during ischemia is important.
Although PI3-kinase seems to carry the signal for protection (20,37), PI3-kinase has several downstream targets including Src-family tyrosine kinases as well as Akt. These experiments do not reveal whether Akt also might carry the protective signal or whether it is merely a reporter for PI3-kinase activation, thus making its phosphorylation an epiphenomenon. We are not aware of a suitable inhibitor of Akt, which would allow us to actually test whether Akt is involved in the protection of IPC. We could find two studies (12,21), in which hearts were transfected with either a dominant negative or a constitutively activated Akt. Both studies indicated that Akt activation per se was associated with protection.
In summary, we have presented new information that both the muscarinic and adenosine receptors can cause phosphorylation of Akt through PI3-kinase in the rabbit heart. Furthermore, the activation of this pathway was found to be the result of ligand-independent transactivation of at least one receptor tyrosine kinase, the EGFR. An important role for Src tyrosine kinase in the transactivation could also be shown. A schematic representation of the hypothesized pathway of ACh's cardioprotection is shown in Fig.8. Whereas blockade of only the EGFR abrogated the ability of ACh to activate Akt, it could not block protection in the same model suggesting that other receptor tyrosine kinases are involved. The importance of this pathway in ischemic preconditioning remains unclear.
This study was supported in part by National Heart, Lung, and Blood Institute Grants HL-20648 (to J. M. Downey) and HL-50688 (to M. V. Cohen). T. Krieg was supported by a grant from Nachwuchsförderung Community Medicine Greifswald of the Alfried Krupp von Bohlen und Halbach-Stiftung, Essen, Germany.
Address for reprint requests and other correspondence: J. M. Downey, Dept. of Physiology, MSB 3024, Univ. of South Alabama, College of Medicine, Mobile, AL 36688 (E-mail:).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
August 22, 2002;10.1152/ajpheart.00474.2002
- Copyright © 2002 the American Physiological Society