Renal blood flow (RBF) can be reduced in rats and rabbits by up to 40% without significant changes in renal tissue Po2. We determined whether this occurs because renal oxygen consumption changes with RBF or due to some other mechanism. The relationships between RBF and renal cortical and medullary tissue Po2 and renal oxygen metabolism were determined in the denervated kidneys of anesthetized rabbits under hypoxic, normoxic, and hyperoxic conditions. During artificial ventilation with 21% oxygen (normoxia), RBF increased 32 ± 8% during renal arterial infusion of acetylcholine and reduced 31 ± 5% during ANG II infusion. Neither infusion significantly altered arterial pressure, tissue Po2 in the renal cortex or medulla, nor renal oxygen consumption. However, fractional oxygen extraction fell as RBF increased and the ratio of oxygen consumption to sodium reabsorption increased during ANG II infusion. Ventilation with 10% oxygen (hypoxia) significantly reduced both cortical and medullary Po2 (60–70%), whereas ventilation with 50% and 100% oxygen (hyperoxia) increased cortical and medullary Po2 (by 62–298 and 30–56%, respectively). However, responses to altered RBF under hypoxic and hyperoxic conditions were similar to those under normoxic conditions. Thus renal tissue Po2 was relatively independent of RBF within a physiological range (±30%). This was not due to RBF-dependent changes in renal oxygen consumption. The observation that fractional extraction of oxygen fell with increased RBF, yet renal parenchymal Po2 remained unchanged, supports the hypothesis that preglomerular diffusional shunting of oxygen from arteries to veins increases with increasing RBF, and so contributes to dynamic regulation of intrarenal oxygenation.
- arteriovenous shunt
- diffusional shunt
both hypoxia and hyperoxia can compromise tissue function and integrity, so tissue oxygenation must be tightly regulated (29). Functional imperatives dictate that the mechanisms regulating oxygenation of the kidney differ from those in other organs (3, 29, 30). For example, to drive glomerular filtration, renal blood flow (RBF) must greatly exceed that required to meet renal metabolic demand (3, 29, 30). Renal vasoconstriction reduces RBF and thereby O2 delivery to the kidney. It might therefore be expected that renal tissue Po2 would vary with RBF, provided renal O2 consumption remains relatively stable. This is the case in the brain (38, 47), retina (36, 37), and skeletal muscle (4, 19). Indeed, tissue and intravascular Po2 are considered useful surrogate markers of blood flow in the retina (49), brain (32), and kidney (18). However, we recently reported two situations in which renal tissue Po2 remained stable during moderate renal vasoconstriction (31). When RBF was reduced by ∼15% by renal nerve stimulation in rabbits, or by ∼40% by ANG II infusion in rats, we did not detect changes in either cortical or medullary tissue Po2 (31).
The two possible explanations for these observations are the hypotheses: 1) that the changes in O2 delivery induced by moderate changes in RBF are matched by changes in O2 consumption or 2) that some other mechanism acts to control and maintain O2 delivery to renal tissue in the face of changes in RBF, such as diffusional shunting of O2 from arterial blood to venous blood. The first hypothesis is supported by observations, made over 40 years ago, showing a direct relationship between RBF and O2 consumption (13, 20, 23–26, 45). However, the maneuvers that were used to change RBF directly altered O2 consumption (e.g., hypotension, cooling the kidney). The interpretation of these seminal and influential studies may therefore have been confounded by effects independent of RBF. Thus these hypotheses remain to be directly tested.
To determine which hypothesis is correct, we altered renal O2 delivery in anesthetized and artificially ventilated rabbits by directly changing RBF through renal arterial infusion of vasoactive agents. We also varied inspired O2 concentration across the range from hypoxia to hyperoxia, at each level of RBF. We found that reducing or increasing RBF by ∼30% had little effect on renal O2 consumption and renal tissue Po2, despite the total flux of O2 across the kidney changing in proportion with RBF. This phenomenon was observed during normoxemia, hypoxemia, and hyperoxemia. Calculations based on the assumption that capillary blood O2 content remained stable in the face of changes in RBF indicate that arterial-venous O2 shunting increases with RBF and so makes a major contribution to the dynamic regulation of renal parenchymal oxygenation.
MATERIALS AND METHODS
Eight male New Zealand White rabbits were used (3.23 ± 0.13 kg). Procedures accorded with the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes and were approved by the Animal Ethics Committee of the Department of Physiology, Monash University.
Catheters were placed in central ear arteries and veins (31). Rabbits were then anesthetized with pentobarbital sodium (90–150 mg plus 30–50 mg/h), intubated, and artificially ventilated. Throughout the surgery and experiment, a maintenance solution containing compound sodium lactate and a 10% polygeline/electrolyte solution in a 4:1 ratio was infused at 10.8 ml·kg−1·h−1 (22). The left kidney was exposed via a flank incision and a catheter was placed in the left ureter (31). The kidney was then denervated to avoid the potentially confounding effects of changes in renal sympathetic nerve activity during experimental maneuvers (22). Catheters were placed in the renal artery (35) and renal vein (10). RBF was measured by transit time ultrasound flowmetry (31). To measure medullary tissue oxygenation (MPo2) and temperature, a Po2 optode/thermocouple (BF/OT, tip diameter = 230 μm Oxford Optronix, Oxford, UK) was advanced into the kidney using a micromanipulator so that its tip lay ∼10 mm below the midregion of the lateral surface of the kidney [i.e., within the inner medulla (15)]. For measurement of cortical tissue oxygenation (CPo2) and temperature, both a second Po2 optode/thermocouple and a Clark electrode (10-μm tip, Unisense, Aarhus, Denmark) were inserted 2 mm into the kidney using micromanipulators. Ventilation rate and tidal volume were then adjusted so that arterial Po2, Pco2, and pH were 90–110 mmHg, 13–25 mmHg, and 7.3–7.5, respectively. Experimental manipulations commenced 90 min later.
The protocol comprised three major experimental periods where RBF was manipulated by renal arterial infusion of ANG II [3.6–53.4 ng·kg−1·min−1, titrated to decrease RBF by ∼30% without altering mean arterial pressure (MAP)], acetylcholine (148–590 ng·kg−1·min−1, titrated to increase RBF by ∼30% without altering MAP), or the saline vehicle (20 μl·kg−1·min−1; 154 mM NaCl). The order of these treatments was randomized. Once the doses of ANG II or acetylcholine were titrated to the target effect, they remained fixed at this level throughout the major experimental period. At least 15 min was allowed after commencing the infusions, before responses to altered inspired O2 were tested. Within each of the major periods, there were four 15-min periods during which the rabbit was ventilated with 10, 21, 50, or 100% O2 in random order. During the final 10 min of each of the 12 ventilation periods, urine produced by the left kidney was collected. At the midpoint of the urine collection period, blood samples (0.3 ml) were obtained from the ear artery and renal vein for blood gas analysis and a 1-ml sample of arterial blood was collected for renal clearance measurements. Blood samples were replaced with washed red blood cells from previous samples and/or blood from a donor rabbit.
Measurement of hemodynamic variables and determination of renal function.
Glomerular filtration rate (GFR) was measured as [3H]inulin clearance (22). Sodium concentrations in plasma and urine were determined by atomic absorption spectrophotometry (31). Blood gas analysis was performed using an ABL510 oximeter (Radiometer, Copenhagen, Denmark). Renal O2 consumption was calculated as the product of RBF and the arterial/venous difference in O2 content. Arterial pressure (mmHg) was measured via an ear artery catheter (22). The Clark electrode was connected to a picoammeter (PA-2000, Unisense) and calibrated as previously described (31). The precalibrated fluorescence optodes were connected to a tissue oximetry system (Oxylite, Oxford Optronix) (31). These optodes operate across a range from 0 to 100 mmHg. In two rabbits, this upper limit was reached in either the cortex or medulla during hyperoxia. In these cases, Po2 values during hyperoxia were excluded from subsequent analyses. Before commencing the experiments, we tested the calibration of the fluorescence optodes against values obtained using a Clark electrode, in 154 mM NaCl at 37°C. We found close agreement between the two methods at Po2 levels of 2 and 32 mmHg. However, at higher levels of Po2, values provided by Clark electrode (77 mmHg) exceeded those provided by fluorescence optode (54 mmHg).
Data are expressed as means ± SE. Statistical tests were performed using SYSTAT (Version 9, SPSS, Chicago, IL). Two-sided P ≤ 0.05 was considered statistically significant. Our experiment was designed in a factorial manner. We therefore used ANOVA to test the global hypotheses that the variables we measured were dependent on the levels of renal blood flow (Pflow) and inspired O2 content (Pgas) and an interaction between these factors. Because hypoxia (10% O2) reduced MAP, analyses were performed both with and without these data (Table 1). Lines of best fit were determined by the least-products method (27).
Effects of altered inspired O2 concentration and renal arterial infusion of acetylcholine and ANG II.
Varying inspired O2 from 21 to 100% did not significantly alter RBF or MAP, but hypoxia (10% O2) reduced MAP (by 5 ± 2 mmHg during saline infusion). During normoxia (21% O2), RBF was 31 ± 5% less during infusion of ANG II compared with saline, and 32 ± 8% greater during infusion of acetylcholine compared with saline. Similar effects of these vasoactive infusions were observed under hypoxic (10% O2) and hyperoxic (50 and 100% O2) conditions. The vasoactive infusions did not significantly affect MAP (Fig. 1, Table 1).
Because our major interest was to relate renal oxygenation to RBF, in subsequent analyses we assigned RBF as the independent variable (Figs. 2, 3, 4, 5). Alterations in inspired O2 concentration produced the expected changes in Po2 of arterial blood, which were not influenced by the vasoactive infusions. Renal venous Po2 also increased with increasing inspired O2. Renal venous Po2 also increased as RBF was increased, at least under normoxic and hyperoxic conditions. Both (total) renal O2 delivery and renal venous O2 efflux increased with increasing RBF. The magnitudes of these changes were similar under normoxic, hyperoxic, and hypoxic conditions, but the absolute levels of renal O2 delivery and renal venous O2 efflux were markedly reduced under hypoxic conditions (Fig. 2, Table 1). Because arterial blood hemoglobin was nearly fully saturated under normoxic conditions, increasing inspired O2 from 21 to 100% during renal arterial infusion of saline increased arterial O2 content and renal O2 delivery by only 7.2 ± 2.5 and 5.0 ± 2.8%, respectively. In contrast, arterial blood hemoglobin was desaturated under hypoxic conditions, so decreasing inspired O2 from 21 to 10% decreased arterial O2 content and renal O2 delivery by 42.6 ± 9.1 and 37.9 ± 8.7%, respectively, during renal arterial infusion of saline. The vasoactive infusions profoundly altered RBF but not arterial blood O2 content. Consequently, renal O2 delivery changed in proportion to the changes in RBF.
Under normoxic conditions during renal arterial infusion of saline, the Pco2 and pH of arterial blood averaged 16.5 ± 1.9 mmHg and 7.40 ± 0.04 pH units, respectively. Corresponding values for renal venous blood were 17.9 ± 1.5 mmHg and 7.40 ± 0.04 pH units, respectively. These parameters were little affected by renal arterial infusions of acetylcholine and ANG II, and by hypoxia or hyperoxia.
CPo2 measured both by Clark electrode and fluorescence optode, and MPo2 measured by fluorescence optode, increased progressively as inspired O2 concentration was varied from 10 to 100%. However, these variables did not vary in a systematic manner with RBF (Fig. 3, Table 1).
Renal O2 consumption changed little as RBF changed. Both renal arterial-venous O2 concentration difference and fractional O2 extraction fell as RBF increased, similarly under normoxic and hyperoxic conditions. Hypoxia reduced the absolute level of renal O2 consumption, presumably due chiefly to reduced GFR secondary to reduced MAP. Hypoxia also reduced renal arterial-venous O2 concentration difference and increased fractional O2 extraction. Nevertheless, these variables varied with RBF in a similar manner under hypoxic, normoxic, and hyperoxic conditions (Fig. 4, Table 1).
Changes in RBF induced by renal arterial infusion of vasoactive agents were positively associated with GFR and sodium reabsorption (Fig. 5, Table 1). Filtration fraction was less under hypoxic conditions than normoxic or hyperoxic conditions. It also tended to fall as RBF increased, at least under normoxic and hypoxic conditions. However, this apparent effect was not statistically significant across all levels of inspired O2 (Fig. 5, Table 1). The ratio of renal O2 consumption to sodium reabsorption was similar under hypoxic, normoxic, and hyperoxic conditions, but was inversely related to RBF (Fig. 5, Table 1). This effect was mostly attributable to the actions of the ANG II infusion. When averaged over all inspired gases, the ratio of renal O2 consumption to sodium reabsorption averaged 0.51 ± 0.08 ml O2/mmol during renal arterial infusion of saline. Infusion of ANG II increased this value to 0.78 ± 0.16 ml O2/mmol (P = 0.02), but it was not significantly altered by infusion of acetylcholine (0.45 ± 0.10 ml O2/mmol; P = 0.51).
Renal O2 consumption was positively correlated with RBF, GFR, and sodium reabsorption (Fig. 6). On their own, RBF, GFR, and sodium reabsorption only accounted for 7–17% of the variance observed in renal O2 consumption (Table 2). Adding the categorical variable “Rabbit” to the analyses increased the proportion of the variance explained by the models to 50–53%. In contrast, adding the categorical variables “Gas” and “Flow” only increased the proportion of the variance explained by the models to 16–32 and 7–21%, respectively. This analysis suggests that renal O2 consumption varies between different rabbits because RBF, GFR, and sodium reabsorption vary between rabbits. The small increases in r2 achieved by the addition of Gas to the models are likely attributable to the fact that hypoxia reduced renal O2 consumption (Fig. 4). The fact that addition of Flow to the model had little impact on the values of r2 likely reflects the fact that changes in RBF induced by the vasoactive agents had little impact on renal O2 consumption (Fig. 4).
Tissue temperature in both the cortex and medulla varied slightly with RBF. Under normoxic conditions, infusion of acetylcholine increased cortical temperature by 0.3 ± 0.2°C (from 35.9 ± 0.6°C during saline infusion) and medullary temperature by 0.2 ± 0.2°C (from 37.8 ± 0.5°C during saline infusion). Infusion of ANG II reduced cortical temperature by 0.5 ± 0.2°C but had little effect on medullary temperature (−0.1 ± 0.3°C change).
We found that RBF could be reduced or increased by ∼30%, without detectable changes in tissue Po2 in the cortex or medulla. This was observed under normoxic, hypoxic, and hyperoxic conditions. Changes in RBF induced by renal arterial infusion of ANG II and acetylcholine were accompanied by changes in renal O2 delivery and efflux but not renal O2 consumption. Thus renal tissue oxygenation is maintained in the face of relatively large changes in RBF, and so renal O2 delivery. Under the conditions of our current experiment, this phenomenon appears not to be mediated by changes in renal O2 consumption.
The conventional understanding of regulation of intrarenal oxygenation might predict relatively stable renal tissue Po2 in the face of changes in RBF, because of concomitant changes in renal O2 consumption. This view is based on studies performed over 40 years ago, showing little change in fractional O2 extraction with altered RBF (23–26). Because renal O2 consumption is the product of RBF and O2 extraction, renal O2 consumption was found to vary directly with RBF. In contrast, we found that fractional O2 extraction fell progressively as RBF increased, so that renal O2 consumption remained relatively constant. This likely reflects an important methodological difference between our current approach and those of the classic studies in this field (23–26, 45), in which RBF was altered by maneuvers that alter O2 consumption independently of changes in RBF. For example, RBF was altered by chronic uninephrectomy (25, 45), by changes in renal perfusion pressure that would greatly alter GFR and tubular load (13, 20, 23, 24, 26), or by cooling the kidney, which would reduce tissue metabolic rate (23). In contrast, we acutely altered RBF by infusion of vasoactive agents, which did not significantly alter MAP (and so renal perfusion pressure) and only slightly altered renal tissue temperature. Thus our findings call for revision of the dogma that blood flow and O2 consumption are necessarily tightly linked in the kidney.
Nevertheless, tubular sodium reabsorption and renal O2 consumption are tightly linked (21). Our data are consistent with this concept, since renal O2 consumption was positively correlated with RBF, GFR, and sodium reabsorption. Nevertheless, under normoxic conditions, ANG II infusion reduced sodium reabsorption by 27 ± 15% and acetylcholine infusion increased sodium reabsorption by 18 ± 18%, yet neither treatment significantly altered O2 consumption. In contrast, we detected decreased renal O2 consumption during hypoxia (24 ± 11% during saline infusion), reflecting reduced GFR secondary to a small (5 ± 2 mmHg) fall in MAP. Because we could detect this physiologically relevant change in renal O2 consumption, we can be confident of our measurements of renal O2 consumption.
Infusion of ANG II reduced the efficiency of renal O2 consumption, as shown by a significant increase in the ratio of renal O2 consumption to sodium reabsorption. Presently, we can only speculate about the mechanisms mediating this effect. Potential mechanisms include 1) effects on mitochondrial O2 utilization, mediated by potential actions of ANG II on nitric oxide bioavailability (3); 2) differential effects of ANG II and/or nitric oxide on tubular transport mechanisms along the nephron that differ in their efficiency of O2 utilization (3); and even potentially 3) effects of ANG II on O2 consumption by renal vascular smooth muscle (16, 41, 42, 44, 48).
O2 transport to tissue occurs not just in capillaries, but also from arteries, arterioles, and sometimes venules (34, 44). Blood Po2 falls progressively along the arterial tree (7, 44). Indeed, approximately two-thirds of total O2 extraction occurs in the precapillary network in resting skeletal muscle (34). O2 losses from arterial blood occur from transport to parenchymal tissue, vessel wall O2 consumption (41), and where arteries and veins are in a countercurrent arrangement (e.g., kidney, skeletal muscle, and gut), arterial-venous O2 shunting (29, 30, 39, 46). The relative contributions of these three O2 sinks likely differ in different tissues (30, 34). Mathematical models predict that the progressive fall in blood Po2 along the arterial tree is inversely related to blood flow (40). Thus increased blood flow should lead to increased blood Po2 in downstream vascular elements and so an increased driving force for diffusion of O2 to tissue. In the absence of changes in O2 consumption, this should increase tissue Po2. Neither tissue Po2 nor O2 consumption varied with RBF in the current study, but fractional O2 extraction fell as RBF increased, suggesting that a mechanism operates to maintain longitudinal Po2 gradients in the face of changes in RBF.
What is the nature of this mechanism? One possibility is that changes in precapillary arterial-venous O2 shunting contribute to dynamic regulation of renal parenchymal oxygenation. O2 in renal venous effluent comes from only two sources: 1) O2 from blood within the renal capillaries and 2) O2 shunted directly from precapillary renal arteries to adjacent veins. If renal arterial O2 delivery and renal venous O2 concentration increase while renal O2 consumption remains stable, kidney tissue Po2 should increase unless some of the increased O2 delivered in the renal artery never actually gets to kidney tissue. This most plausible explanation for the “missing oxygen” is increased arterial-to-venous O2 shunting. Our observation of stable renal parenchymal Po2 therefore indicates that changes in the amount of arterial-venous O2 shunting may offset changes in renal O2 delivery induced by changes in RBF and so maintain homeostasis of renal parenchymal O2 delivery. Three caveats must be applied to this conclusion. First, at present we must limit our conclusions to the context of the current experimental conditions; since changes in RBF induced by different vasoactive factors from those used in the current study, or through changes in renal perfusion pressure, may have very different effects on arterial-to-venous O2 shunting than changes in RBF induced by renal arterial infusion of ANG II and acetylcholine. Second, we cannot exclude the possibility that changes in the pH of renal tissue and capillary blood, induced by changes in RBF, may have altered the Po2/hemoglobin saturation relationship. However, this seems unlikely since although we did not measure renal tissue pH in the current study, the Pco2 and pH of arterial and renal venous blood were not altered by changes in RBF or inspired O2 content. Third, our experiment was performed under conditions of hypocapnia. This arose because we needed to set arterial blood Po2 at ∼100 mmHg during ventilation with room air. Although the pH of arterial and renal venous blood was normal (i.e., 7.4) in the current study, our experiment must to be replicated under conditions of normocapnia before we can confidently generalize the concept that arterial-to-venous O2 shunting contributes to dynamic regulation of intrarenal oxygenation.
These caveats aside, our data indicate that the O2 content of blood within renal capillaries remained relatively stable across the range of RBF examined. This allows us to estimate the contribution of arterial-venous O2 shunting to dynamic regulation of renal oxygenation using the equation:
That is, if we assume the contribution of O2 within renal capillaries, to renal venous O2 efflux, changes in direct proportion with RBF, then changes in arterial-venous O2 concentration difference must be due to changes in arterial-venous O2 shunting. Based on this formula, we calculate that 0.17 ± 0.06 ml/min more O2 was shunted during saline infusion than during ANG II infusion, and 0.37 ± 0.09 ml/min more O2 was shunted during acetylcholine infusion than during ANG II infusion (across all gas mixtures). This equates to ∼15% of the difference in renal O2 delivery in each case. Thus arterial-venous O2 shunting may make an important contribution to maintenance of homeostasis of intrarenal oxygenation.
Our current and previous (31) observations suggest that renal tissue Po2 remains remarkably stable when RBF is altered within the physiological range, although tissue hypoxia does occur when RBF is reduced by more than ∼30% (31). How can we reconcile these findings with those of previous studies showing changes in renal tissue Po2 in response to vasoactive factors? An important difference between our current study, and most previous studies of the relationship between RBF and renal tissue oxygenation (5, 6, 8, 12, 28), is our provision of detailed information on both RBF and renal O2 consumption. Thus it is equally possible that the changes in CPo2 observed in these previous experiments resulted from altered renal metabolism rather than the direct effects of altered RBF. Our data might also appear at odds with those of Juillard et al. (18) who observed progressive desaturation of hemoglobin (assessed by functional magnetic resonance imaging) in both the cortex and medulla when RBF was reduced by renal artery constriction. However, blood O2 levels do not necessarily reflect tissue Po2, as evidenced by the fact that renal venous Po2 varied with RBF in our study, yet renal tissue Po2 did not.
As we have found previously (31), cortical Po2 measured by fluorescence optode was always less than that measured by Clark electrode. Nevertheless, responses of cortical tissue Po2 to changes in RBF and arterial blood Po2, assessed using Clark electrodes and fluorescence optodes, were qualitatively similar. Furthermore, the relationship between simultaneous measurements of cortical Po2 made with the two methods in the current study could be fitted to a straight line with proportional bias (i.e., slope ≠ 1) but no fixed bias (i.e., no zero offset). Thus it seems very unlikely that the conclusions we draw from our current experiment are confounded by the techniques we used to measure tissue Po2.
Although changes in RBF had little effect on renal tissue Po2, changes in arterial Po2 profoundly altered renal tissue Po2. These observations reflect the importance of gradients in O2 partial pressure in tissue O2 delivery. Thus changes in atmospheric O2 content would likely profoundly alter renal tissue Po2, and thus the signaling cascades that mediate erythropoietin release (43). Our observations suggest that changes in RBF within the physiological range will likely have little impact on these signaling cascades. Dissociation of erythropoietin synthesis and RBF makes adaptive sense, since it allows independent regulation of extracellular fluid volume and blood O2-carrying capacity.
In conclusion, tissue Po2 must be regulated within tight limits to maintain cellular respiration, prevent hypoxic damage, prevent excessive production of reactive oxygen species, and regulate O2-dependent gene expression. This is no less true in the kidney than in other organs (1, 3, 9, 29, 30, 43). However, the mechanisms regulating tissue oxygenation in the kidney differ somewhat from those operating in other organs. In particular, unlike organs such as skeletal muscle and the heart (11), brain (2), and retina (36), changes in tissue Po2 in the kidney do not profoundly alter renal vascular tone (11, 22). This allows control of glomerular filtration, the primary function of the kidney, to dominate control of renal vascular tone. However, this could result in large fluctuations in renal tissue Po2, unless additional mechanisms operate to maintain homeostasis of renal tissue oxygenation. Our current results indicate that renal tissue Po2 changes little when RBF is altered within the physiological range (±∼30%). This occurs despite relatively stable renal O2 consumption and may be mediated by changes in the efficiency of preglomerular arterial-venous O2 shunting. We speculate that this mechanism could be adaptive, by allowing RBF and GFR to change in response to physiological requirements, without concomitant changes in tissue Po2. Diabetes (33) and hypertension (46) are both associated with renal tissue hypoxia, at least in part due to reduced efficiency of O2 utilization by the kidney. Future studies should examine whether malfunction of renal arterial-venous O2 shunting also contributes to renal hypoxia under these pathological conditions. Renal arterial-venous O2 shunting may also contribute to development of renal hypoxia during acute hemodilution (17).
This work was funded by National Health and Medical Research Council of Australia (143785, 143603, 384101).
We thank Dr. J. Ludbrook (Biomedical Statistical Consulting Service, Melbourne, Australia) for advice regarding the statistical analysis of the data.
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.
- Copyright © 2007 the American Physiological Society