Studies were performed to determine the significance of temporal variation in vascular smooth muscle Ca2+ signaling during acute arteriolar myogenic constriction and, in particular, the importance of the stretch-induced intracellular Ca2+concentration ([Ca2+]i) transient in attaining a steady-state mechanical response. Rat cremaster arterioles (diameter ∼100 μm) were dissected from surrounding tissues, and vessel segments were pressurized in the absence of intraluminal flow. For [Ca2+]imeasurements, vessels were loaded with fura 2 and fluorescence emitted by excitation at 340 and 380 nm was measured using video-based image analysis. Ca2+ and diameter responses were examined after increases in intravascular pressure were applied as an acute step increase or a ramp function. Additional studies examined the effect of longitudinal vessel stretch on [Ca2+]i and arteriolar diameter. Step increase in intraluminal pressure (from 50 to 120 mmHg) caused biphasic change in [Ca2+]i and diameter. [Ca2+]i transiently increased to 114.0 ± 2.0% of basal levels and subsequently declined to 106.7 ± 4.4% at steady state. Diameter initially distended to 125.4 ± 2.1% of basal levels before constricting to 71.1 ± 1.2%. In contrast, when the same pressure increase was applied as a ramp function (over 5 min) transient vessel distension and transient increase in [Ca2+]i were prevented, yet at steady state vessels constricted to 71.3 ± 2.5%. Longitudinal stretch resulted in a large [Ca2+]i transient (158 ± 19% of basal) that returned to baseline despite maintenance of the stretch stimulus. The data demonstrate that the initial vessel distension (reflecting myocyte stretch) and associated global [Ca2+]i transient are not obligatory for myogenic contraction. Thus, although arteriolar smooth muscle cells are responsive to acute stretch, the resulting changes in myogenic tone may be more closely related to other mechanical variables such as wall tension.
- myogenic response
- stretch activation
- vascular smooth muscle
arterial resistance vessels typically exist in a state of partial contraction that is, in part, a direct response to transmural forces exerted by the intraluminal pressure. Although the existence of myogenic tone has been appreciated for almost a century (2), the exact cellular mechanisms that underlie this mode of smooth muscle activation remain uncertain (for review, see Ref. 7). Current evidence suggests that an acute increase in intravascular pressure results in increased wall tension or cell stretch that alters smooth muscle cell membrane ion conductance (6). This change in ion gating, either directly or indirectly via membrane depolarization, results in increased levels of myocyte intracellular Ca2+concentration ([Ca2+]i) and activation of the contractile process (6, 23, 24, 29, 30, 41).
In previous studies, it was shown that there is temporal variation in the [Ca2+]i response to an acute increase in arteriolar intraluminal pressure (41). Typically after pressure is acutely elevated, [Ca2+]i rises to a peak and then declines to a steady-state level, which remains increased relative to baseline, as a stable myogenic contraction is achieved. Although the magnitude of the initial increase in [Ca2+]i is related to the degree of vessel distension (5, 41), it is unclear whether the components of the overall [Ca2+]i response are dependent on cell stretch and/or a variable such as wall tension. Furthermore, the requirement of the initial [Ca2+]i transient for the development of the steady-state mechanical response is also uncertain. It is conceivable, for example, that this initial elevation in [Ca2+]i represents a generalized response to cell membrane deformation, because stretch-induced increases in [Ca2+]i have been demonstrated in many cell types and are not specifically related to expression of myogenic behavior (11, 25, 33, 35, 38, 39).
The present study compared the [Ca2+]i and diameter responses of isolated arterioles to equivalent increases in intravascular pressure applied as either an acute step (time < 1 s) increase or as a ramp (time = 5 min) function. Additional studies were performed to examine the effect of longitudinal vessel stretch on [Ca2+]i. The rationale for this mode of stimulation was that this form of vessel wall distension results in smooth muscle membrane deformation without an overt increase in circumferential wall tension as occurs during an increase in transmural pressure. These approaches were used to test the hypothesis that acute cell stretch and the resultant rapid transient increase in myocyte [Ca2+]i are obligatory signals to initiate or maintain the myogenic response.
MATERIALS AND METHODS
The studies used male Sprague-Dawley rats weighing between 200 and 350 g. Before experiments, rats were housed in pairs in a dedicated animal facility with a 12:12-h light-dark cycle. During this period rats were allowed free access to a standard rat chow and drinking water. All procedures were approved by the Animal Care and Use Committees at RMIT University and Eastern Virginia Medical School.
Isolated Arteriole Preparation
Rats were anesthetized with thiopental sodium (100 mg/kg), after which the right and left cremaster muscle were exteriorized, excised from the animal, and placed in a cooled (4°C) chamber containing dissection buffer (in mM: 3 MOPS, 145 NaCl, 5 KCl, 2.5 CaCl2, 1 MgSO4, 1 NaH2PO4, 0.02 EDTA, 2 pyruvate, and 5 glucose and 1% albumin) (13). Segments of the main intramuscular arteriole (1A) were dissected from the muscle as previously described (29, 40). Individual vessel segments were then cannulated with glass micropipettes, secured using 10-0 monofilament silk suture, and mounted in a custom-designed tissue chamber (vol 5 ml). The cannulated arterioles were continually superfused (2–4 ml/min) with a physiological salt solution containing (in mM) 111 NaCl, 25.7 NaHCO3, 4.9 KCl, 2.5 CaCl2, 1.2 MgSO4, 1.2 KH2PO4, 11.5 glucose, and 10 HEPES. Vessel segments were gradually pressurized to 70 mmHg (measured at level of inlet cannulation pipette) and warmed to 34°C during a 60-min equilibration period. During this period, vessels were checked for pressure leaks and allowed to develop spontaneous basal tone. Vessel length was adjusted before the development of spontaneous tone by increasing segment length such that pressure steps to 120 mmHg did not cause a lateral bowing of the vessel. This approach allowed consistency between preparations with respect to this parameter.
The vessel preparation was positioned on the stage of an inverted microscope equipped with a video-based imaging system (Universal Imaging, West Chester, PA). Measurements of vessel diameter (in μm) were made using an electronic video caliper (18).
In experiments requiring measurement of changes in [Ca2+]i, vessels were incubated (60 min, room temperature) with 1–5 μM fura 2-AM (Molecular Probes, Eugene, OR) in buffer containing 0.5% DMSO and 0.01% Pluronic. Only the abluminal surface of the vessel segment was exposed to the fura 2-AM solution, to restrict dye loading to the vascular smooth muscle layer (14, 31). The dye loading procedure was followed by a 30-min washout period with physiological salt solution. Fura 2-AM-loaded vessels were exposed to epi-illumination (75-W xenon source) with light of alternating excitation wavelengths (340 and 380 nm) using a computer-controlled filter wheel. Images of fluorescence emission at 510 nm were acquired using an image intensifier (Videoscope International, Washington, DC) and a charge-coupled device (CCD; Hamematsu, Bridgewater, NJ). Simultaneous transillumination with wavelengths >610 nm provided a nonfluorescent image that enabled measurement of internal arteriolar diameter while fluorescence images were collected. The high-wavelength image was directed, by a beam splitter, to a second CCD camera (Javelin Electronics, Los Angeles, CA). This procedure did not interfere with measurements of Ca2+-related fluorescence. Fluorescent image intensities were expressed as the 340- to 380-nm ratio to allow quantitative estimates of changes in arteriolar wall [Ca2+]i. Details of these procedures were presented in previous publications (30, 41).
[Ca2+]i measurements from the imaging system were verified using a second photometer-based instrument that enabled data collection at much higher sampling rates (30 Hz). The high illumination rates were made possible by the use of a spinning filter (divided into 340- and 380-nm sections) wheel. This system also utilized dual trans- and epi-illumination such that and internal diameter data could be simultaneously collected. Vessels were loaded with fura 2 as described above.
Comparison of acute step and ramp luminal pressure increases.
On the basis that a rapid increase in intraluminal pressure results in greater vessel distension than a slowly applied increase of equivalent magnitude, [Ca2+]i and diameter responses were compared for an acute step increase (<1 s) from 50 to 120 mmHg and a ramp increase applied slowly over a 5-min period (rate of change = 14 mmHg/min). To ensure that steady-state responses had been reached in both protocols, responses were followed for 10 min after the pressure change was initiated. To control for any possible buffering effects of fura 2-AM loading, mechanical responses were examined in both the presence and the absence of the fluorescent dye.
Effect of stretch applied along longitudinal axis of isolated arterioles.
After collection of baseline diameters and [Ca2+]i levels, arterioles were rapidly stretched in the longitudinal direction by adjusting a calibrated micromanipulator in the x direction (length). The extent of the applied stretch was adjusted to cause arteriolar diameter (y) to be reduced to 80% of baseline. This procedure is based on the assumption that wall surface area (x × y) remained constant when the vessel was acutely stretched; thus lengthening equated to a longitudinal increase of 125% of baseline (i.e., 1.25 × x). To determine whether this degree of stretch mechanically compromised the vessel preparation, several arterioles were subsequently exposed to norepinephrine to test vasoconstrictor capacity and to acetylcholine to test their ability to undergo relaxation.
In protocols not examining changes in intraluminal pressure, arteriolar diameters (d) were normalized for each experiment to the passive (zero calcium) diameter at an intraluminal pressure of 70 mmHg (d 70) and are expressed as the ratiod/d 70. This pressure approximates the in vivo intravascular pressure for arteriole 1A. Alternatively, for the ramp and step changes in pressure, vessel diameter was expressed as a percentage of the basal diameter at 50 mmHg (i.e., the starting pressure). Results were tested for statistical significance using Student's paired t-test and were considered different whenP < 0.05. Throughout this paper, n represents both the number of vessel segments studied and the number of animals utilized.
Comparison of Acute Step and Ramp Increases in Luminal Pressure
Figure 1 shows the average responses to acute pressure steps and ramped pressure increases. In nine arterioles examined in the absence of fura 2-AM loading, the acute pressure step from 50 to 120 mmHg resulted in distension to 125.4 ± 2.1% of baseline diameter whereas no distension was detected during the ramp increase. Baseline diameters (at 50 mmHg) were similar during the two protocols [115.6 ± 6.2 compared with 113.6 ± 5.4 μm,P = not significant (NS)]. After steady-state constriction was achieved at 120 mmHg, arteriolar diameters (as percentage of baseline) were 71.1 ± 1.2% for the step protocol and 71.3 ± 2.5% for the ramp increase (NS). Similarly, calculated wall tension rose abruptly to a transient peak (1,126 ± 65 dyn/cm) during the step protocol, whereas a slow increase (to a maximum of 812 ± 70 dyn/cm; P < 0.01 compared with step protocol) occurred in response to the ramp pressure increase. Steady-state wall tension levels were similar regardless of the pressure-increase protocol (649 ± 50 dyn/cm for step compared with 643 ± 33 dyn/cm for ramp, NS).
[Ca2+]i responses were examined in a separate group of arterioles (n = 6) loaded with fura 2-AM (1 μM). As with the diameter responses reported above, steady-state [Ca2+]i levels were similar regardless of whether intraluminal pressure was increased from 50 to 120 mmHg by a ramp or a step function. The mean [Ca2+]i responses shown in Fig.2 A illustrate that the ramp pressure increase was not associated with an initial transient increase in [Ca2+]i as occurred in response to the step pressure increase. Figure 2 B shows the mean peak increase in [Ca2+]i occurring within the first minute after an acute pressure step was initiated and the mean steady-state [Ca2+]ilevels for both pressure-increase protocols during the last 2.5 min of the observation period.
Because time-dependent variations (unrelated to the pressure step) in the signal could conceivably contribute to the peak response and artifactually contribute to the biphasic appearance of the signal, the mean of five fluorescence intensity measurements across the peak was compared with the mean fluorescence level from the final five measurements (i.e., during steady state). With this method of analysis, the peak fluorescence ratio was 110.2 ± 0.4 compared with a steady-state level of 106.7 ± 0.2 (P < 0.05; paired Student's t-test).
The results of the imaging experiments were confirmed using a photometer-based system (see materials and methods). A typical response is shown in Fig. 3. Figure3 A again illustrates the biphasic nature of the pressure step-induced response, whereas Fig.3 B demonstrates that the pressure step and ramp increase reach similar steady-state [Ca2+]ilevels.
Effect of Stretch Applied Along Longitudinal Axis of Isolated Arterioles
Acute longitudinal stretch resulted in a rapid increase in [Ca2+]i, reaching a peak (158 ± 19% of basal) within 2 s of application of the mechanical stimulus (Fig. 4). This peak in [Ca2+]i subsequently declined to a level not significantly different from baseline despite maintenance of the stretch stimulus. Although the longitudinal stretch passively reduced the vessel diameter, a small but transient vasoconstriction was subsequently observed (∼7%); however, a significant level of sustained active constriction was not apparent despite maintenance of the stretch stimulus. This was not a function of vessel damage or the vessel being stretched to a point where it was mechanically impaired, because subsequent application of 10−6 M norepinephrine (while the vessel remained longitudinally stretched) caused vasoconstriction comparable to that before stretch (Table 1). Further evidence of viability of the vessels after longitudinal stretch was provided by the observations that 1) vessels remained myogenically reactive to increases in intraluminal pressure (see Fig. 5) and that 2) on return of longitudinal length to that of the basal state, arteriolar diameters were not statistically different from those recorded before stretch (0.56 ± 0.02 compared with 0.56 ± 0.02d/d 70, n = 6). Maintenance of longitudinal stretch was associated with a reduced arteriolar dilator response to ACh (10−6 M), but this was largely a passive effect because the diameter in 0 mM Ca2+ buffer was also significantly less after longitudinal stretch (153 ± 5 μm before stretch compared with 116 ± 5 μm after stretch, P< 0.01 paired t-test, n = 5).
Figure 5 illustrates the diameter responses to a pressure step from 50 to 120 mmHg in a group of arterioles (in the absence of fura 2-AM loading; n = 6) before and during maintained longitudinal stretch. It should be noted that although this group of vessels was less reactive to an intraluminal pressure stimulus compared with those described above, 1) all responses were qualitatively similar and 2) all experiments were analyzed on a paired basis and thus each vessel acted as its own control. Before longitudinal stretch the step from 50 to 120 mmHg resulted in a transient distension to 121.8 ± 3.4% of diameter at 50 mmHg. After longitudinal stretch, the same pressure step resulted in distension to 113.8 ± 1.3% (P < 0.05; paired t-test). Despite the significant difference in distension, vessels reached a similar steady-state level of constriction (93.9 ± 3.9% before stretch vs. 90.1 ± 8.4% after longitudinal stretch, NS).
The results of these studies suggest that a global, stretch-induced [Ca2+]i transient is not an absolute requirement for the development of a steady-state level of myogenic tone. This suggestion is based on the observation that, regardless of whether arteriolar pressure was raised from 50 to 120 mmHg via a ramp function or an acute pressure step, a similar level of steady-state myogenic constriction was attained. Furthermore, in the case of the ramp increase in pressure there was no observable passive stretch of the arteriolar smooth muscle layer nor was there a transient increase in myocyte [Ca2+]i. Despite this, it was evident that myocyte stretch, per se, could lead to an increase in [Ca2+]i. Thus transient increases in [Ca2+]i were observable both after the cell stretch associated with an acute pressure step and when arterioles were stretched along their longitudinal axis.
The results obtained in the present study are possibly relevant to the earlier in vivo studies of Mellander and colleagues (19, 20), who reported that the myogenic response contains a rate-sensitive component. In particular, these authors suggested that a myogenic response consists of two components, the first of which correlates with the rate of application of a given pressure change (dP/dt) and the second, or steady-state phase, which occurs independently of the rate of pressure application. Thus in vivo, high dP/dt (5 mmHg/s) was reported to result in a marked transient increase in resistance that declined to a lower steady-state level. In the present study the actual constrictor response induced by a step increase in pressure (50 to 120 mmHg) appeared monophasic, that is, once it developed, the pressure-induced constriction was maintained. This constrictor pattern occurred despite a high dP/dt, estimated to be >50 mmHg/s (9). It is possible that had smaller vessels been studied, as reported by Davis and Sikes (9), biphasic constrictor responses would have been observed. In contrast, the [Ca2+]i response revealed in the present experiments was indeed biphasic, reaching a peak at a time when the vessel remained distended relative to baseline and declining to a steady-state level as the arteriole achieved a steady-state level of constriction. Rather than reflecting a phenomenon such as rate sensitivity, we have suggested that the change in [Ca2+]i is, at least in part, a function of transmural wall tension that declines as myogenic constriction develops.
[Ca2+]i responses to acute stretch have been examined in a number of cell types including vascular smooth muscle (6, 8), endothelial cells (11), epithelial cells (39), osteoclasts (38), cardiac myocytes (25), and astrocytes (35). Interestingly, when vascular smooth muscle cells cultured on deformable membranes are acutely stretched, [Ca2+]i transiently increases before returning to near-baseline levels despite maintenance of the stretch stimulus (12). Consistent with this, Kulik et al. (28) reported that acute stretch of cultured smooth muscle cells resulted in a rapid and short-lived increase in inositol trisphosphate. The time course of the [Ca2+]i change in these cultured cells therefore may be more similar to the effect of longitudinal arteriolar stretch. In the present study, the vascular smooth muscle [Ca2+]i response to longitudinal stretch was transient compared with that after an acute increase in transmural pressure, where, although there is a Ca2+ transient, [Ca2+]i remains significantly elevated after a steady-state level of constriction has been achieved.
In the case of vascular smooth muscle cells, there are several conceivable reasons why differing responses to cell stretch could occur relative to those in the intact vessel. For example, smooth muscle cells in culture tend to undergo phenotypic changes such that they exhibit the properties of a synthetic cell rather than those of a contractile cell. In addition, when grown under static culture conditions smooth muscle cells lose the directional orientation that is seen in the intact arteriole, whereas smooth muscle cells grown under cyclic stretch tend to orientate perpendicularly to the direction of stretch (10). In culture, these cells also tend to lose their in situ spindle shape and more closely resemble the spreading shape of a fibroblast (3). This is in contrast to the morphology of arteriolar smooth muscle cells in situ, which are arranged circumferentially within ±15° to perpendicular to the long axis of the vessel (4,31, 36). Thus, when a population of cultured cells is stretched, the variability in orientation may result in a heterogeneous application of the stimulus to individual cells. Furthermore, under conditions of cyclic strain, cultured vascular smooth muscle cells exhibit differences in the distribution of signaling molecules (32).
An additional reason for examining step versus ramp pressure increases relates to the possibility that cellular responses to mechanical stimuli may be affected independently by both the rate of change of a particular mechanical force and the absolute level of that force.
For example, Hamill and McBride (22) suggested that mechanosensitive channels (in Xenopus oocytes) can alter their gating characteristics to detect either phasic or tonic mechanical signals. Consistent with this, Kirber et al. (27) showed that a burst of channel activity occurs in smooth muscle cells as a mechanical stimulus is applied. Furthermore, Frangos and co-workers (16) showed that endothelial cells respond differently to the shear stress associated with initiation of flow compared with that evoked by steady-state flow. Similarly, the same group has shown in studies of cardiac fibroblasts (21) that the rate of application of mechanical strain is important for G protein-mediated signaling events. Fibroblasts were subjected to 6% maximal strain applied over either 10 or 60 s with GTP binding being detectable only in the cells subjected to the higher strain rate (i.e., shorter duration). Although this is analogous to the comparison of ramp and step pressure increases, it should be noted that the signaling molecule observed in the present study ( ) achieved similar steady-state levels regardless of the protocol used to increase pressure. However, equally important and perhaps indicative of the involvement of other events when pressure is increased rapidly is the observation that a stretch-dependent transient [Ca2+]iincrease occurs in response to step but not ramp increases in pressure (Fig. 2).
Because noncontractile cells respond to acute stretch with transient, and in some cases sustained, increases in [Ca2+]i, it is possible that this represents a generalized response to membrane deformation. Alternatively, it may represent a signal for a compensatory pathway leading to growth/remodeling rather than a specific mechanism coupled only to contraction. Thus it is conceivable that the mechanical forces associated with an increase in intraluminal pressure activate more than one Ca2+-dependent pathway in vascular smooth muscle. Consistent with this, an increase in intraluminal pressure in isolated small arteries not only stimulates contractions but within ∼30 min stimulates protooncogene mRNA production consistent with activation of a growth pathway (1).
An interesting question relates to why the myocyte [Ca2+]i responses associated with longitudinal stretch were found to be markedly different (in both the magnitude of the [Ca2+]i transient and steady-state levels) than those induced by an increase in intraluminal pressure. In both cases the degree of stretch applied to the cells was on the order of 125% of baseline (see materials and methods) whether the deformation was perpendicular or parallel to the direction of the smooth muscle cells. The difference may be a reflection of differences in orientation of the contractile proteins, although it is generally considered that these proteins are arranged as a crisscrossing lattice within the cells (15, 26). Alternatively, vascular smooth muscle cells in situ may exhibit a degree of polarity such that structures detecting a change in wall tension or a related variable are preferentially stimulated by deformation along their long axis. The binding of extracellular matrix to integrin receptors with subsequent activation of focal adhesion kinase and intracellular signaling pathways may provide such a mechanism, particularly if the distribution of such molecules is polarized. Supporting the notion of polarity, Gabella (17) provided data suggesting that dense bodies are asymmetrically distributed in arteriolar smooth muscle, with a greater abundance of such structures being located on the adventitial surface compared with the medial surface of the vessel. Interestingly, endothelial cells subjected to physiological levels of strain show an alignment of β1-integrins parallel to the long axis of the cell (40). Similarly, smooth muscle cells are known to align perpendicular to flow (34); however, whether integrins are polarized under such conditions is not known.
An alternative explanation for the differing [Ca2+]i responses in pressurized versus stretched vessels is that longitudinal stretch results in cell damage. This would appear not to be the case, because 1) a similar degree of stretch applied in the transverse direction initiated clear myogenic constriction without apparent vessel damage and2) longitudinally stretched vessels remained responsive to receptor-mediated agonists and vasodilators and recovered to baseline diameters after removal of the stretch stimulus. Furthermore, longitudinally stretched vessels retained the ability to demonstrate myogenic constriction to an acute increase in intraluminal pressure despite passive elongation and reduced distensibility (Fig. 4).
A related question in the comparison of the effects of the different mechanical stimuli is, Why was the larger [Ca2+]i transient that occurred during longitudinal stretch not associated with sustained contraction? As outlined above, vessels remained reactive to agonist stimulation so the apparent decrease in responsiveness to the stretch-induced increase in [Ca2+]i is unlikely to be a result of vessel damage. A possible explanation relates to the fact that longitudinal vessel stretch reduces transmural wall tension as a result of the passive decrease in vessel radius (tension = pressure × radius). The decrease in transmural tension might be expected to result in a decreased Ca2+ entry caused by deactivation of mechanosensitive channels or removal of a membrane depolarizing stimulus (7). Thus the combination of a longitudinal stretch-induced [Ca2+]i mobilization and a simultaneous decrease in transmural wall tension could result in a net transient [Ca2+]i signal that does not lead to maintained contraction.
The finding that overt cell stretch is not required for the development of a steady-state myogenic constriction provides further evidence that a variable such as wall tension, rather than cell length, is sensed and/or regulated during a myogenic constriction. This is consistent with previous studies taking a more biochemical approach in which wall tension was found to be well correlated with levels of smooth muscle [Ca2+]i and myosin light chain phosphorylation (41). It should be acknowledged, however, that the present studies could not address the possibility that an element within the cell membrane or in the cell interior remains stretched during a pressure increase despite overall cell contraction and reduction of wall tension. Alternatively, an absence of a global increase in [Ca2+]i does not adequately reflect what is occurring within discrete cellular compartments. In this regard, a recent report by Wier et al. (37) indicates that Ca2+ sparks occur in arteries with myogenic tone despite the stable baseline global [Ca2+]i reported under similar conditions in a number of arteriolar preparations (5, 30, 41). Regardless of the possibility of Ca2+ modulation within intracellular compartments, it is clear that, in the present study, global Ca2+ measurements were differentially affected by mechanical stimuli applied along the circumferential and longitudinal axes of the arterioles.
In summary, the present data do not support the hypothesis that overt cell stretch and the resultant rapid transient increase in [Ca2+]i are obligatory signals to initiate or maintain a myogenic contraction. Although arteriolar vascular smooth muscle cells do exhibit changes in global [Ca2+]i in response to acute stretch, it appears that myogenic tone may be more closely related to other variables such as wall tension. The results further underscore the complexity of the [Ca2+]isignal that occurs when arteriolar intraluminal pressure is increased and simultaneous changes in cell length and wall tension occur.
The authors extend thanks to Dr. Gerald Meininger for constructive criticisms before submission.
Address for reprint requests and other correspondence: M. A. Hill, Dept of Human Biology and Movement Science, RMIT Univ., Plenty Rd, Bundoora, Victoria 3083, Australia (E-mail:).
The studies described were supported, in part, by grants from the National Health and Medical Research Council of Australia and the US National Institutes of Health.
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. §1734 solely to indicate this fact.
- Copyright © 2000 the American Physiological Society