Journal of Applied Physiology

Atrial natriuretic peptide levels and plasma volume contraction in acute alveolar hypoxia

T. S. E. Albert, V. L. Tucker, E. M. Renkin


Albert, T. S. E., V. L. Tucker, and E. M. Renkin.Atrial natriuretic peptide levels and plasma volume contraction in acute alveolar hypoxia. J. Appl. Physiol. 82(1): 102–110, 1997.—Arterial oxygen tensions ( PaO2 ), atrial natriuretic peptide (ANP) concentrations, and circulating plasma volumes (PV) were measured in anesthetized rats ventilated with room air or 15, 10, or 8% O2(n = 5–7). After 10 min of ventilation, PaO2 values were 80 ± 3, 46 ± 1, 32 ± 1, and 35 ± 1 Torr and plasma immunoreactive ANP (irANP) levels were 211 ± 29, 229 ± 28, 911 ± 205, and 4,374 ± 961 pg/ml, respectively. At PaO2 ≤40 Torr, irANP responses were more closely related to inspired O2(P = 0.014) than to PaO2 (P= 0.168). PV was 36.3 ± 0.5 μl/g in controls but 8.5 and 9.9% lower (P ≤ 0.05) for 10 and 8% O2, respectively. Proportional increases in hematocrit were observed in animals with reduced PV; however, plasma protein concentrations were not different from control. Between 10 and 50 min of hypoxia, small increases (+40%) in irANP occurred in 15% O2; however, there was no further change in PV, hematocrit, plasma protein, or irANP levels in the lower O2groups. Urine output tended to fall during hypoxia but was not significantly different among groups. These findings are compatible with a role for ANP in mediating PV contraction during acute alveolar hypoxia.

  • fluid balance
  • cardiac output
  • arterial hypoxemia
  • blood gases
  • plasma protein concentration

there is evidence that atrial natriuretic peptide (ANP) plays a role in systemic and pulmonary responses to hypoxia. Hypoxia is a known secretagogue for ANP (see Ref. 28 for review). Circulating levels of this cardiac peptide are rapidly increased in animals ventilated with hypoxic gas (2, 8, 16, 29) and are elevated in hypoxemic states such as sleep apnea (21), chronic obstructive pulmonary disease (5), and altitude sickness (9). High-affinity binding sites for ANP have been identified in the vasculature of numerous tissues, with particularly large numbers being expressed in the lung (see Ref. 6 for review). Hypoxia-induced ANP release is currently believed to mitigate the development of hypoxic pulmonary hypertension. ANP has been shown to have a direct vasodilatory effect on pulmonary arterial rings (15), and intravenous administration of ANP blunts pulmonary vasoconstriction during acute hypoxia (2, 16).

In addition to its vasodilatory properties, exogenous ANP has been shown to rapidly increase plasma protein extravasation (31, 34) and induce extrarenal fluid shifts from blood to interstitium (4, 32). Reductions in circulating volume subsequent to ANP-induced fluid shifts could provide an additional mechanism for limiting the pressure load on the right heart during hypoxia, thus minimizing right heart failure and pulmonary edema. Diminution of plasma volume (PV) is a common occurrence in hypoxia (11-13); however, a specific role for endogenous ANP in mediating this response has not been adequately tested.

The purpose of the present study was to assess the potential role of ANP in mediating PV reductions during acute hypoxia. The magnitude and time course of changes in circulating ANP concentrations and PV were determined in anesthetized rats ventilated with graded levels of O2 (8–21%) for 50 min. ANP levels were measured by radioimmunoassay, and circulating PV were assessed by using radiolabeled albumin. Blood pressures and arterial blood gases, hematocrits (Hct), pH, lactate (La), and protein concentrations were analyzed as well to further evaluate potential mechanisms of volume regulation during acute hypoxia. Albumin transport and fluid balance at the tissue level were also addressed in these experiments; the data are presented in a companion paper (3).


Surgical preparation.

Male Wistar rats (Charles River, Kingston, NY) weighing 260–320 g were fasted 14–18 h before experiments with water provided ad libitum. To induce anesthesia, pentobarbital sodium (12 mg/100 g body wt) was administered subcutaneously in two depots over a 1-h period. The right jugular vein and left carotid artery were cannulated with 50-gauge polyethylene tubing containing heparin in lactated Ringer solution (LR) for measurement of central venous and arterial pressures, respectively. The left jugular vein was similarly cannulated for infusion of fluids and test substances. A tracheotomy was performed, and a tube was inserted to facilitate spontaneous respiration during surgery and for mechanical ventilation during experiments. Body temperature was monitored by a rectal thermistor (Yellow Springs Instruments, Yellow Springs, OH). After surgery, 1 ml of arterial blood was collected into a syringe containing heparin and used to replace blood samples during the experimental period. The rat was left supine, and a 2-ml intravenous infusion of 5% (wt/vol) bovine serum albumin (BSA) was given over 10 min to compensate for fluid and protein loss due to anesthesia and surgery (27). The BSA infusion was followed by a slow infusion of LR (12 μl/min) for the remainder of a 15- to 20-min stabilization period. Supplementary doses of anesthetic were given intravenously as needed to maintain anesthesia.

Blood pressures were measured by using Gould-Statham transducers (P23 ID, Oxnard, CA) referenced at midchest level and were continuously recorded on a Gilson polygraph (model ICT-2H, Middletown, WI). Mean arterial (MAP) and central venous pressures were obtained at regular intervals by electrical dampening of the pressure signals. Arterial and venous pressure records were used to calculate heart rate (HR) and respiration rate, respectively.

Experimental protocols.

After stabilization, animals were positive pressure ventilated with room air by using a Harvard rodent ventilator (model 683, Millis, MA) set at a respiratory rate of 60 breaths/min. Tidal volume was adjusted according to animal weight by using a rodent ventilatory chart (Harvard Bioscience). An example of the ventilatory parameters for a 300-g rat are respiration rate of 60 breaths/min and tidal volume of 2.0 ml/min. After 15 min of stabilization on the ventilator, baseline parameters and arterial blood samples were collected. Baseline blood sampling consisted of 1 ml taken for radioimmunoassay of ANP, two 70-μl Hct tubes for Hct, total protein (TP), and La measurements; and a 100-μl sample for blood gas and pH analysis. After blood sampling, 500 μl of replacement blood followed by 400–500 μl of LR were infused. The bladder was then emptied by application of gentle pressure to the abdomen. Ten additional minutes of stabilization were allowed before the start of control (air) or hypoxic gas treatment.

Animals were divided into a normoxic (room air) control group and three hypoxic (15, 10, and 8% O2-balance N2) groups. Room air and gas mixtures were administered from a Douglas bag attached to the inlet port of the ventilator via Tygon tubing. To humidify the gas, 2 ml of distilled water were added to the bag. Just before the start of the treatment period, the Tygon tubing connecting the Douglas bag to the ventilator was flushed with treatment gas. Arterial blood samples were collected ∼10 min into the treatment period, followed by intravenous injection of 131I-labeled rat serum albumin (131I-RSA; 6 μCi) at time (t) = 15 min. Small blood samples (100 μl) were drawn at 20 and 30 min, respectively, from the start of O2 treatment for measurement of plasma 131I activity, Hct, TP, and La. These samples were replaced with similar volumes of LR. At 45 min, a bolus of 125I-labeled RSA (6 μCi) was injected for a final measurement of PV. Terminal blood samples were drawn at 50 min, after which the rats were killed with intravenous KCl.

All blood samples taken during the experiment were centrifuged, and Hct was measured by the microtube method. Aliquots of plasma were saved for determination of TP and Laanalysis by colorimetric assay. Arterial pH and arterial PO2 ( PaO2 ) and PCO2 ( PaCO2 ) were immediately obtained from whole blood by using a Ciba-Corning pH/blood gas analyzer (model 238, Medfield, MA); corrections were made for animal body temperature. Urine and abdominal cavity fluids were collected postmortem by syringe.

PV measurements.

Whole animal PV was obtained from the 5-min distribution volumes of131I- and125I-RSA (t = 20 and 50 min, respectively). PV was calculated by dividing total counts injected by counts per milliliter of plasma after 5 min of circulation. The total counts injected were determined by counting a standard aliquot of each tracer. Plasma samples and tracer standards were counted in131I and125I channels of a gamma scintillation counter (Searle Analytical, Des Plaines, IL). Over 100,000 counts/sample were collected and corrected for background, crossover from 131I to125I, and decay of131I. Crossover of131I to the125I channel ranged from 10.9 to 11.3%.

Radioimmunoassay of plasma ANP.

Arterial blood was immediately placed in vials containing EDTA (1.6 mg/ml) and centrifuged at 4°C (12,000g for 5 min). Plasma was stored at −70°C. Samples (200–500 μl) were extracted one day before assay by reverse-phase chromatography by using C18 columns (Bond Elut, Varian, Harbor City, CA). Recovery of 100, 200, and 500 pg of peptide added to 500 μl of rat plasma was 72 ± 7.5, 71 ± 10, and 72 ± 10% (mean ± SD), respectively. Triplicate extracted samples and peptide standards were preincubated with antiserum for 24 h at 0–4°C. After incubation,125I-ANP was added (∼14,000 counts/tube) and the tubes were incubated for an additional 24 h at 0–4°C. Bound tracer was immunoprecipitated by sequential addition of pretitered goat anti-rabbit gamma globulin and rabbit serum, followed by a 2-h incubation period at room temperature. Bound tracer was measured in the gamma scintillation counter described above. A Log-logit analysis was used to determine the amount of hormone in the samples. Sensitivity (2 SD from 0) was 4.2 pg/tube. Values were corrected for incomplete recovery.

Blood gas and cardiac output trials.

Animals in the main study group did not have blood gases analyzed after 15 min to avoid radioactive contamination of the blood gas machine. To assess blood gases throughout the hypoxia period, a small series of experiments were run by using a similar protocol without radioisotopes. Blood gases from both arterial and central venous samples were measured in three groups subjected to different O2 gas treatments: room air (n = 3), 15% O2(n = 2), and 8% O2(n = 3). Cardiac outputs were measured at approximately the same times by using a Columbus Instruments thermodilution machine (Cardiomax-II, model 85, Columbus, OH). Surgical preparation was the same as in the main study except for the insertion of a thermistor into the right carotid artery. Vital signs were measured as previously described. Blood gas and cardiac output measurements were taken at baseline (t= −10), after the first 10 min of O2 treatment (t = 10), after 30 min of O2 treatment (t = 30), and at the end of the experimental period (t = 50 min). Both arterial and venous blood samples were taken simultaneously. The volumes of fluids withdrawn and infused during the course of the experiment were identical to those in the main study. Saline (100 μl at room temperature) was used as the injectate for thermodilution measurement of cardiac output. The saline was injected through a right jugular vein cannula with the tip placed either at the entrance to the right atrium or inside the right atrium; the accompanying blood temperature changes were measured by a thermistor located just proximal to the aortic valve. The locations of both of these were verified by postmortem visualization. To minimize error in the cardiac output calculation due to warming of the saline in the jugular line, the line was flushed before each measurement. Duplicate measurements of cardiac output were taken at each point, and the values were averaged.


The 131I and125I radionuclides were purchased from New England Nuclear-DuPont (Wilmington, DE). BSA (5% solution) and RSA fraction V powder were purchased from Sigma Chemical (St. Louis, MO). ANP antibodies were obtained from Research and Diagnostic Antibodies (Berkeley, CA). ANP (rat, 28 amino acids) standard was from Bachem (Torrance, CA); (3-[125I]iodotyrosyl28) ANP was from Amersham (Arlington Heights, IL). Goat anti-rabbit gamma globulin and rabbit serum were purchased from Peninsula Laboratories (Belmont, CA). The La assay kit was from Sigma Chemical.

RSA was iodinated by using chloramine-T and purified by anion exchange and repeated concentration-dilution with Minicon filters (Amicon, Danvers, MA). Labeled RSA (specific activity = 10–50 μCi/mg) was repurified on the day of use to reduce free I levels to ≤1%.


Data are expressed as means ± SE unless otherwise indicated. Differences among treatment means were tested by using one-way analysis of variance followed by Duncan’s new multiple-range test for post hoc comparisons among individual means. Within-animal effects (paired comparisons) and relationships among variables were tested by using analysis of variance repeated measures and regression models, respectively. For the ANP radioimmunoassay results, a logarithmic transformation of the data was performed before analysis to reduce variance heterogeneity among treatments. Differences are reported as statistically significant when P ≤ 0.05.



Data from 24 experiments divided into room air control (n = 7), 15% O2(n = 6), 10% O2(n = 5), and 8% O2(n = 6) groups, respectively, are reported. Baseline values for all measured parameters were similar among groups. Table 1 summarizes group means for PaO2 , PaCO2 , and pH obtained ∼10 min before and 10 min after the onset of hypoxic (or normoxic) ventilation. All treatment groups that were ventilated with low O2 showed evidence of arterial hypoxemia. The 10 and 8% O2groups were the most severely hypoxemic ( PaO2 = 32 ± 1 and 35 ± 1 Torr, respectively); however, there was no significant difference in PaO2 between these two groups. Progressive reduction of PaCO2 was observed with decreasing PaO2 , even though minute ventilation was constant and similar for all groups. Treatment PaCO2 for both the 10 and 8% O2 groups were significantly reduced compared with paired baseline values. Arterial pH did not change significantly within the first 10 min of hypoxia. Body temperatures (not shown in Table 1) gradually dropped below the control group mean (38 ± 0.1°C) in the two most hypoxic groups and reached a significant reduction of ∼1°C by the end of the experimental period.

View this table:
Table 1.

Rat weights, arterial blood gases, and pH measurements during gas treatment


All hypoxia groups showed similar and dramatic (≈50%) reductions in MAP immediately after the start of treatment, with no apparent relationship to the percentage of inspired O2 (Fig.1 A). The depression in MAP was maintained throughout the experimental period. The room air group demonstrated a much smaller, yet significant, reduction in MAP from its baseline over the course of the experiment. Unlike the consistent reduction in MAP, HR changes were more variable and clearly related to the percentage of inspired O2 (Fig.1 B). Within 5 min of the start of 15% O2 ventilation, HR fell significantly to 84% of its baseline. HR remained depressed for at least 20 min and then tended to rise, reaching control levels by 50 min of hypoxia. Exposure to 10% O2resulted in a smaller initial fall in HR followed by a gradual rise to average values slightly higher than, but not different from, controls. The 8% O2 group did not exhibit an early drop in HR, but it rose above control and 15% O2 after 10 min and remained significantly elevated for the remainder of the treatment period. Control HR tended to fall slightly during the course of the treatment period, but the changes were not statistically significant. Central venous pressures decreased to levels below paired baseline values in all groups (Fig. 1 C). Although there were no significant differences between group means at any time point, repeated-measures analysis indicated a greater decline in central venous pressure with time for both 8 and 10% O2 groups compared with either room air or 15% O2 treatment.

Fig. 1.

Hemodynamic responses to ventilation with hypoxic gas mixtures. Baseline measurements were taken 10 min before switch from air to low O2 gas, which is indicated by arrow. A: mean arterial pressure (MAP). B: heart rate (HR).C: central venous pressure (CVP). Values are means ± SE. TRT, onset of treatment. * Significantly different from control (room air),P ≤ 0.05.

Blood gases and cardiac output trials.

Because blood gases were not measured during the last 35 min of the experimental period during which radioisotopes were given, eight additional rats were subjected to the same experimental protocol except for omission of the radioisotopes. A control group (room air) and hypoxic groups at 15 and 8% O2were studied. Arterial and venous blood samples were taken for PO2 , PCO2 , and pH measurements 10 min before gas treatment (baseline) and at 10, 30, and 50 min, respectively, after the onset of gas treatment. Cardiac output was also measured at approximately the same times by using the thermodilution technique. These data are summarized in Table2. Animals in this series had similar baseline values and experienced comparable changes in arterial pressure, HR, and Hct as rats in the main study. Additional observations include 1) graded reduction in mixed venous PO2 with decreasing fraction of inspired O2,2) continued decline of PaCO2 in the face of constant minute ventilation, 3) increase in PaO2 (compared with paired baseline) during the later phase of 8% O2 treatment, and4) decrease in both arterial and venous pH after 50 min of 8% O2ventilation.

View this table:
Table 2.

Additional blood gas, pH, and CO measurements in nonradioactive animals

Baseline cardiac output (mean for all animals) was 276 ± 22 ml ⋅ min−1 ⋅ kg−1.1In the room air and 15% O2groups, cardiac output did not change significantly over the experimental period. In the 8% O2animals, cardiac output at 10 min was unchanged but subsequently fell to ≈50% of the baseline value at 50 min.

Fluid balance.

Figure 2 illustrates absolute changes in Hct (ΔHct) from baseline over the course of the gas treatment period. Control Hct (41 ± 0.8%) fell slightly but not significantly with time. After 10 min of hypoxia, rats ventilated with 15% O2 exhibited a significant decrease in Hct from 40.9 ± 0.7 to 38.3 ± 0.8%, and the level remained depressed (paralleling the controls) for the rest of the treatment period. The 10 and 8% O2 groups (baseline Hct = 42.9 ± 1.2 and 41.7 ± 0.65%, respectively) showed little or no early decline; however, Hct had increased significantly by 20 min of hypoxia, with the greatest increase occurring in the group given 8% O2. For all groups, Hct levels were sustained after 20 min of gas treatment.

Fig. 2.

Absolute changes in hematocrit (ΔHct) from individual baseline values. Arrow, onset of treatment (room air or hypoxia). * Significantly different from room air control,P ≤ 0.05. † Significantly different from 15% O2,P ≤ 0.05.

Table 3 shows values of PV, TP, and Hct obtained at 20 and 50 min after start of gas treatment and calculated values of total protein mass (TPM; TPM = PV × TP) at these times. Compared with controls, PV values were significantly lower in 10 and 8% groups after 20 min of hypoxia. These results indicate a loss of fluid from the circulation early in the treatment period. This is further supported by concomitant increases in arterial Hct that were linearly and inversely related to changes in PV (P = 0.004 by regression analysis,n = 24 points). There were no subsequent changes in either PV or Hct between 20 and 50 min of gas treatment.

View this table:
Table 3.

Whole animal PV, arterial Hct, TP, and TPM during gas treatment

Changes in TP were small and irregular in direction; none were statistically significant relative to controls. This finding suggests that in addition to fluid, protein was likewise removed from the circulation in animals treated with 8 or 10% O2. TPM tended to be lower in both the 10 and 8% O2 groups; however, these changes did not reach statistical significance.

Volumes and protein concentrations of abdominal fluid (data not shown) were not different among groups. Total urine outputs (measured over the entire gas treatment period) tended to decrease in the more hypoxic groups (21% O2, 2.6 ± 0.2 μl/g body wt; 15% O2, 2.7 ± 0.6 μl/g body wt; 10% O2, 1.7 ± 0.3 μl/g body wt; and 8% O2, 1.5 ± 0.4 μl/g body wt), but these differences were not statistically significant.

Hormonal and metabolite changes.

Figure 3 shows plasma immunoreactive ANP concentrations (irANP) in individual rats plotted as a function of time. Mean values for the three sample periods are given in Table 4. Baseline levels were low and did not differ significantly among experimental groups. Plasma irANP increased in a graded fashion as the fraction of inspired O2 decreased; however, the pattern of response differed among hypoxia groups. For the 15% O2 group, the collective change in irANP after 10 min of gas treatment was not statistically significant. By the end of the treatment period, irANP was slightly but distinctly increased in five rats and substantially increased in one remaining animal. The latter was considered an outlier (irANP <3 SD from the average of all six rats) and was not included in the group mean reported in Table 4. Rats exposed to 10% O2 exhibited elevations of plasma irANP at 10 min after the start of gas treatment, with the group average being significantly greater than paired baseline values. Rats exposed to 8% O2 exhibited the largest increases in plasma irANP at 10 min. For both 8 and 10% O2 groups, irANP levels after 50 min of hypoxia were statistically indistinguishable from levels measured at 20 min of hypoxia.

Fig. 3.

Plasma immunoreactive atrial natriuretic peptide concentrations (irANP) plotted as a function of time in individual normoxic and hypoxic rats. Absolute values are indicated on right y-axis and corresponding logarithms onleft y-axis. Each graph represents separate treatment group. Individual rats are represented by a different symbol within each group. * Mean response significantly different from paired baseline (P ≤ 0.05).

View this table:
Table 4.

Plasma concentrations of irANP and La during gas treatment

Figure 4 illustrates the relationship between hypoxia and the initial increase in irANP levels. The logarithm of irANP is plotted as a function of systemic PaO2 (A) or inspired O2 fraction (B) measured after 10 min of gas treatment. Regression analysis of the data indicated a significant inverse relationship between PaO2 and irANP (P = 0.0014,n = 24). However, the response was not uniformly graded, as indicated by the abrupt increase in irANP when PaO2 decreased to ≤40 Torr. Among these “responders,” there was no correlation of irANP with PaO2 ; inspired O2 level was the best predictor of irANP response magnitude (P = 0.0001,n = 24) .

Fig. 4.

Logarithm of plasma irANP concentrations as a function of paired arterial Embedded Image ( Embedded Image ) values (A) and inspired O2 fraction percentage (B) in rats exposed to 10 min of gas treatment.

Table 4 also shows baseline, 20-, and 50-min plasma La concentrations. Baseline La did not differ among groups and remained unchanged over the treatment period in the control group. However, there was a progressive increase in plasma La with hypoxia, reaching significance compared with controls and increasing with time in both the 8 and 10% O2 groups.


Cardiovascular and metabolic state of the hypoxic animals.

All hypoxic groups showed directional changes in MAP that were similar to those reported by others for hypoxic pentobarbital sodium-anesthetized rats (14, 33). At all three levels of hypoxia, MAP was reduced by 50% as early as 5 min after the start of low-O2 ventilation. Given that cardiac output was maintained over the whole treatment period in 15% O2 animals and declined only in the later phase of 8% O2treatment (see Table 2), it is likely that the early hypotension in all hypoxic animals was due to reductions in total peripheral resistance (TPR). Decreased TPR in hypoxia has been attributed to the local vasodilatory effects of tissue hypoxia (25), but the early fall of HR in the 15% O2 group suggests an initial withdrawal of sympathetic nerve activity. If this is true, the subsequent rise in HR observed in hypoxic animals after the first few minutes of the treatment period could indicate a resurgence of sympathetic or sympathoadrenal activity (17). In the most severely hypoxic animals (8% O2), cardiac output fell progressively to at least one-half of its initial value between 10 and 50 min, whereas MAP remained constant. Thus TPR must have increased to levels equal to or greater than initial levels to maintain arterial pressure in the face of the lower cardiac output.

Although the data suggest that TPR returned toward control levels in the most severely hypoxic animals, it is likely that vascular resistance and blood flow in individual tissues and organs were abnormal during hypoxia. Plasma La concentrations in the 8% O2 animals were significantly higher than in 10% O2 animals, even though their PaO2 were similar. Decreased tissue perfusion due to progressive vasoconstriction over the course of the hypoxic period could be responsible for elevated levels of tissue hypoxia as well as the observed decrement in venous PO2 in the 8% group (see Table2). Given the large elevations in plasma irANP concentrations, one might suspect the vasodilatory actions of this peptide to modify vasoconstrictor responses. This possibility is difficult to assess in the present study because ANP measurements were not obtained in animals where CO was measured. If such a mechanism was operative, it apparently was unable to overcome systemic vasoconstriction during ventilation with 8% O2.

PaO2 in animals ventilated with 8% O2 for 10 min were equal to or greater than those in 10% O2animals. Because ventilation rate and volume were maintained constant, this cannot be due to increased stimulation of respiration by the lower inspired O2, nor can it be due to decreased O2 extraction by the tissues because venous PO2 was lower at 8% O2. Control group PaO2 values were on the low side of normal relative to their PaCO2 (which would indicate alveolar PO2 values >100 Torr), suggesting that there was substantial venous-to-arterial shunting in the lungs and/or considerable ventilation-perfusion inequality. It is possible that increased sympathetic nerve activity to bronchioles or pulmonary arterioles in the 8% O2group might have decreased shunting, thus preventing a further fall in PaO2 . Alternatively, the higher levels of ANP observed during initial treatment with 8% O2 could have antagonized the direct action of alveolar hypoxia on the lungs, reducing pulmonary vasoconstriction (16) (see below), thus diminishing ventilation-perfusion inequality and improving blood oxygenation. The continued rise in PaO2 during the later phase of 8% O2 ventilation could be attributed to an overall increase in ventilation-to-perfusion ratio because cardiac output decreased progressively in these animals in the face of a constant ventilation.

Hypoxia-induced changes in Hct and PV.

LaForte et al. (22) have provided evidence that reductions in circulating volume or pressure result in a shift of low Hct blood from the microcirculation into the large blood vessels. This could explain the decrease in arterial Hct observed in the 15% O2 group in the face of a negligible increase in PV, as well as the early tendency for Hct to fall in the lower O2 groups. By 20 min of gas treatment, animals ventilated with 10 and 8% O2 exhibited reciprocal changes in Hct and PV, indicating that the rise in Hct was primarily due to the loss of fluid from the circulation rather than an increase in erythrocyte volume. Additional support is provided by quantitative analysis of the relative changes in PV and Hct compared with the control group. Such analysis assumes that initial PV values were identical among groups, which is supported by similar weight, age, plasma protein concentrations, and starting Hct. For example, because 10 and 8% O2 groups had respective Hct values that were 3.5 and 5.2 Hct percentage points greater than control (at 20 min; Table 3), the calculated PV would be ∼8.1 and 11.6% lower than control, respectively. These figures are close to the observed 8.5 and 9.9% reductions in125I-RSA distribution volume (i.e., PV). Because urine flow was not increased during hypoxia in these anesthetized rats, plasma fluid must have entered extravascular compartments within the body.

Role of ANP in hypoxia-induced PV reduction.

What are the mechanisms responsible for hypoxia-induced reductions of PV? Fluid movement across capillary walls is a function of the prevailing hydrostatic and colloid osmotic pressure gradients, the permeability of the capillary walls to fluid and protein, and the exchange surface area. Given the early fall in arterial pressure and subsequent reductions in cardiac output, it is unlikely that increases in capillary hydrostatic pressures or surface area could have accounted for the elevated filtration. Furthermore, changes in MAP were similar for all levels of hypoxia, whereas significant reductions in PV were observed only in rats ventilated with 10 and 8% O2. The elevated plasma La concentrations suggest the possibility of a direct effect of tissue hypoxia on capillary permeability. Increases in capillary hydraulic conductivity and protein permeability in response to hypoxia have been observed (30, 33); however, there is presently no general consensus regarding this response. Alternatively, the rapid and large elevations in irANP could have contributed to the early reductions in PV. It is well established that intravenous administration of ANP enhances diuresis and natriuresis; however, ANP also reduces PV and increases Hct in nephrectomized rats (4, 32). The present experiments are analogous to the nephrectomized-rat model in the respect that plasma ANP levels were elevated in the absence of enhanced diuresis.

In conscious humans and animals, acute hypoxic exposures usually result in natriuresis and diuresis (10, 13). However, these conscious models did not exhibit the significant hypotension and cardiac output reduction observed in our anesthetized rats ventilated with 10 or 8% O2. For example, Colice et. al (10) found arterial pressure to be only slightly depressed (−10%) in chronically instrumented rats exposed to a chamber filled with 10.5% O2. Thus one explanation for a lack of diuresis in the present study is a reduction in renal perfusion pressure secondary to systemic hypotension. Along this same line, Karim and Al-Obaidi (18) observed that direct stimulation of carotid chemoreceptors with mixed venous blood resulted in diuresis and natriuresis in anesthetized dogs only when carotid sinus pressure was maintained at high levels. At low sinus pressure (72 ± 1.3 mmHg), chemoreceptor stimulation resulted in a reflex decrease in renal blood flow and urine output. These findings are also consistent with a reduction in renal perfusion as a mechanism of antidiuresis in the present study.

Another similarity between the action of severe hypoxia and exogenous ANP administration is that changes in TP concentrations were negligible compared with changes in Hct. ANP has been shown to elevate microvascular permeability to plasma proteins, increasing their extravasation (31, 34). The resulting decreases in transcapillary colloid osmotic gradient could have favored filtration into tissues. The role of ANP in hypoxia-induced protein extravasation is discussed in more detail in a companion study (3).

Mechanisms of increased irANP during hypoxia.

Elevation in the plasma levels of ANP under conditions of acute hypoxia has been reported in humans (19) and other mammals (2, 29) and is attributed to increased secretion from the heart rather than decreased metabolism of circulating ANP (8). In the present study, irANP increased dramatically after only 10 min of hypoxic treatment with the two lowest concentrations of O2. Such immediate ANP responses to acute hypoxia have also been described in conscious lambs (8) and in rats (29) and anesthetized pigs (2). Although the response time is similar between species, the magnitude of elevation is very different. The irANP response to hypoxia in our study was 11-fold greater than that measured in either lambs or pigs (2, 8). However, the 24-fold increase in irANP that we observed after 10 min of 8% O2 ventilation corresponds closely to the ∼22-fold increase seen by Shirakami et al. (29) in conscious rats breathing 5% O2. This suggests that even though arterial pressures were depressed during acute hypoxia, the ANP response was not blunted in our pentobarbital sodium-anesthetized rats. Sustained elevations of plasma irANP in rats ventilated with 8 or 10% O2 have also been reported in rats subjected to 90 min of hypoxia (16) and in chronically hypoxic rats (26, 35).

A number of mechanisms appear to be responsible for release of ANP in hypoxic animals. ANP is released directly from isolated hearts perfused with substantially lower PO2 s than those in our experiments (24). One additional effect of low inspired O2 in vivo is pulmonary vasoconstriction. Correlation of pulmonary arterial pressure and plasma ANP levels (1) and the presence of elevated plasma ANP in diseases in which pulmonary hypertension is present (5, 21) have led others to the conclusion that pulmonary hypertension is a major cause of ANP release. Jin et al. (16) demonstrated an association between hypoxic pulmonary arterial hypertension and elevated plasma irANP in rats after 90 min of exposure to 10% O2; however, the observed increases in plasma ANP were small (≈2-fold) compared with the present study. Moreover, these investigators have found that observed increases in pulmonary arterial pressure were unaffected by pretreatment with an ANP monoclonal antibody during the first 6 h of hypoxia, whereas pressures were significantly reduced after 7 h of hypoxia. Their results suggest that antihypertensive actions of ANP may be more important in the subacute rather than acute phase of hypoxia. It is not clear how this finding relates to the present study, in which larger increases in irANP were observed.

Whether ANP was released immediately after the development of acute hypoxic pulmonary hypertension in the present study is not known; the first measurements of irANP were obtained after 10 min of hypoxia. A pulmonary neural reflex active during hypoxic pulmonary hypertension has been shown to account for ∼50% of the ANP response to hypoxia in the pig (7). One way that a pulmonary neural reflex could act on the heart to release ANP is through the release of sympathetic agonists. A study by Lew and Baertschi (24) supports such a role for sympathetic control of hypoxia-induced ANP release, although the release mechanisms they considered were intrinsic to the heart itself. They demonstrated that about one-half of the hypoxia-induced ANP released from the isolated rat heart is mediated by adrenergic receptors.

The present finding confirms previous reports that hypoxia-induced ANP release is related to the level of alveolar O2 rather than to the degree of arterial hypoxemia (8). This points to an airway or pulmonary origin for the difference in ANP release between the 10 and 8% O2 groups. This could be due to a higher sustained level of pulmonary arterial pressure in the 8% O2 group relative to the 10% O2 group or to a greater rate of development of pressure to the same level. There is experimental support that ANP release is more strongly correlated with the rate of increase in pulmonary arterial pressure rather than the average pressure (7). Another possibility is that airway or pulmonary chemoreceptors sensitive to the level of inspired O2 are involved in a reflex leading to ANP release. The presence of hypoxia-sensitive airway receptors called pulmonary neuroepithelial bodies has been described in the lungs of rabbits (23), and recent work has identified the presence of an O2-sensing mechanism on the plasma membranes of these cells (36).

Release of ANP due to hypoxic damage to atrial tissue cannot be ruled out, but the likelihood of this being the primary cause is remote. Elevations of irANP in the 8% O2group occurred immediately after the start of the hypoxia treatment, before any evidence of possible heart damage or failure was present (e.g., decreased CO). Studies on isolated perfused rat hearts have shown elevations in ANP release during hypoxia to be reversible, repeatable, and independent of tissue damage (24). HR changes were not associated with ANP release immediately after the start of hypoxia; however, a significant association was found between HR and ANP levels toward the end of the hypoxic period. This is consistent with the idea that increased HR may be involved in the maintenance of ANP elevation during hypoxia, as Lew and Baertschi (24) observed in isolated rat hearts.


Ventilation of anesthetized rats with 10 and 8% O2 resulted in PaO2 s <40 Torr and large increases in plasma irANP within 10 min. In rats ventilated with 15% O2( PaO2 ranging from 40 to 50 Torr), there was no initial increase of irANP, but after 50 min of hypoxia, irANP was slightly elevated. Although PaO 2s at 10 and 8% O2 did not differ significantly, irANP concentrations were substantially and significantly higher at 8% O2. Thus the main stimulus for ANP release in these animals is more closely related to inspired or alveolar PO2 than to PaO2 and may have arisen from receptors in the airways or lungs. Proportional increases in Hct and decreases in PV were observed in irANP-responsive rats. Plasma TP did not increase in proportion to Hct; thus plasma protein was lost from the circulation as well as fluid. Similar reductions in PV have been produced by infusion of exogenous ANP in normoxic anesthetized rats (4, 32, 34). Furthermore, our findings reveal that hypoxia-induced PV contraction can occur even under conditions in which vascular pressures and renal excretory function are suppressed. In a similar fashion, transvascular fluid and protein shifts induced by exogenous ANP occur in the absence of diuresis (4, 32) and independently of changes in capillary hydrostatic pressure (31). The results of the present study are consistent with the hypothesis that ANP contributes to PV contraction during acute hypoxia by favoring a shift of fluid and protein from the circulation into extravascular spaces.


We thank E. Bravo, T. Myers, and G. Johnson for their technical assistance.


  • Address for reprint requests: V. L. Tucker, Dept. Human Physiology, Univ. of California, Davis, Davis, CA 95616-.

  • This study was supported by National Heart, Lung, and Blood Institute Grant HL-18010 and University of California President’s Undergraduate Fellowship (to T. S. E. Albert).

  • Preliminary results of this study were published in abstract form: FASEB J. 8: A1032, 1994.

  • 1 Kissling et al. (20) have shown that measurements of cardiac output in the rat by the thermodilution technique overestimate the true value because of exchange of heat with other tissues than blood, with errors being greater as CO decreases. Because our baseline CO of 276 ± 22 ml ⋅ min−1 ⋅ kg−1is almost exactly the same as that reported in the study by Kissling et al., care must be used in interpreting our CO data. In general, thermal exchange would result in underestimation of the fall in CO produced by our treatments. Qualitative assessments of CO changes should be correct in our study; however, because the direction and relative magnitude of errors can be estimated. Briefly, this can be done by reliance on the results from the Kissling et al. study, which show that heat diffusion is directly related to the time that it takes for the bolus of sub-body temperature fluid to flow through the lungs and be registered by the thermocouple (i.e., passage time). In other words, an increased passage time for the injectate leads to greater overestimation of CO. Because passage times did not change among any groups over the first 10 min, we can assume errors to be similar among groups. Therefore, the observation that CO was not different among groups over the first 10 min of treatment is not in error. The potential meaning of the insignificantly lower CO for the 8% O2 group at 30 min is difficult to interpret because passage time did show a tendency to increase above that for the other groups. In a consideration of the later significant drop in CO, it appears likely that CO was decreasing in the 8% O2 group at this time. The measurement of a significantly lower CO in the 8% O2 group at 50 min is accurate, even though injectate passage time increased, because the value is overestimated to a greater extent than that for the control.


View Abstract