The Effect of Altering Pulmonary Blood Flow on Pulmonary Gas Exchange in the Turtle Trachemys (Pseudemys) Scripta

Many ectothermic vertebrates exhibit an intermittent breathing pattern characterized by long non-ventilatory periods interrupted by brief ventilatory periods consisting of one to several breaths (Shelton et al. 1986). During lung ventilation, pulmonary blood flow increases up to fivefold and thereby provides a temporal matching of lung perfusion and ventilation (Shelton and Boutilier, 1982; Shelton and Burggren, 1976; White and Ross, 1966). The increase in pulmonary blood flow during lung ventilation will affect both O₂ and CO₂ exchange (Burggren, 1987; Hicks and Wang, 1995); however, the physiological consequences of this cardio-respiratory coupling are unclear. One possibility is that the increase in pulmonary blood flow, per se, improves the spatial matching of ventilation and perfusion and thus improves the overall gas exchange efficiency of the lung (West et al. 1992; Wood, 1984). This possibility, although never tested experimentally, is appealing for reptiles, where the mean lung gas–end capillary blood difference may be greater than 20 mmHg, a value that is considerably larger that the equivalent difference observed in resting birds and mammals (Burggren and Shelton, 1979; Glass, 1991; Hicks and White, 1992). The large differences measured in reptiles could result from ventilation–perfusion heterogeneity, pulmonary diffusion limitation and intrapulmonary shunting, although the contribution of each of these factors is virtually unknown. The purpose of this study was to test the hypothesis that increases in pulmonary blood flow, per se, reduce heterogeneity in the lung of the turtle Trachemys (Pseudemys) scripta. We used the multiple inert gas elimination technique to determine the spatial distribution of ventilation and perfusion and to quantify the degree of pulmonary diffusion limitation and intrapulmonary shunting at different pulmonary blood flows in the turtle.

This study was approved by the Animal Subjects Committees of the University of California, Irvine. Six turtles Trachemys (Pseudemys) scripta (Gray) (mass 1.95±0.18 kg, mean ± S.E.M.) were obtained from a commercial dealer, housed in large tanks in a temperature-regulated room (28 °C) and fed goldfish. On the morning of the experiment, anaesthesia was induced by injection with Nembutal (35 mg kg⁻¹, intramuscularly), and the animal was placed on its back and ventilated (SAR-830 ventilator, CWE Inc., Ardmore, PA) with 21 % O₂, 3 % CO₂ (balance N₂) at 140 ml kg⁻¹ min⁻¹. The heart and central arteries were exposed by removing a 4 cm×4 cm piece of the plastron in order to implant flow probes and catheters. Ultrasonic flow probes (2R Transonic Systems Inc., Ithaca, NY) were placed around the left pulmonary artery and around the left aortic arch. Pulmonary flow was calculated by doubling the flow from the left pulmonary arch, and systemic flow was calculated by multiplying the flow from the left aortic arch by a factor of 2.8 (Comeau and Hicks, 1994; Shelton and Burggren, 1976). The common pulmonary artery and the left atrium were non-occlusively cannulated with PE 50 tubing (Ishimatsu et al. 1988) for direct sampling of blood flowing to and returning from the lung. The right jugular vein was occlusively cannulated for infusion of inert gases. A vascular occluder (model OC5A, In Vitro Metric, Healdsburg, CA) was placed around the common pulmonary artery. Ventilation was measured in triplicate at the beginning and end of the experiment by collecting expired gases in a Mylar gas-impermeable bag. Mixed expired and O₂ and CO₂ concentrations were also determined (Beckman OM-11, Sensormedics LB2, respectively).

Ventilation–perfusion distributions were obtained using the multiple inert gas elimination technique (Hlastala, 1984; Wagner et al. 1974a,b) modified for reptiles (Hopkins et al. 1995). A mixture of six inert gases (sulphur hexafluoride SF₆, ethane, cyclopropane, enflurane, diethyl ether and acetone) was dissolved in normal saline and infused via the jugular vein (rate 0.4–0.6 ml min⁻¹). Mixed expired gases were collected with Mylar gas-impermeable bags and transferred into gas-tight glass syringes. Duplicate 3 ml blood samples for the inert gas analysis were taken from the pulmonary artery and left atrium. Solubilities, retentions (R equal to the ratio of left atrial to mixed venous partial pressure) and excretions (E equal to the ratio of mixed expired to mixed venous partial pressure) for the inert gases were determined using gas chromatography (Hewlett-Packard 5890) (Wagner et al. 1974a,b), and distributions were calculated from the inert gas data assuming an alveolar lung model. Using the multiple inert gas elimination technique, compartmental ventilation or blood flow to areas of different ratios can be calculated and represented graphically. Conventionally, a logarithmic scale is used for these data, and the standard deviation of the perfusion distribution (logSD Q̇) is used as an indicator of the degree of heterogeneity (i.e. the greater the logSDQ̇, the greater the heterogeneity).

Left atrial and pulmonary mixed venous and pH (Radiometer BMS3 MK2) were measured at 22 °C immediately after sampling. Correction factors for O₂ to allow for the difference between the electrode reading of the humidified calibration gas and the same gas tonometered in blood were determined by tonometry (DL213) of the blood of each animal on the day of the experiment.

Pulmonary blood flow was decreased slowly by inflating the pulmonary vascular occluder until flow was reduced to approximately one-sixth of the initial non-manipulated value. Pulmonary blood flow was increased by bolus injection of adrenaline (dose 0.1 μg kg–1; Comeau and Hicks, 1994) into the left atrium. The animals were studied under anaesthesia, at 22 °C, during initial non-manipulated conditions, after 10 min of low pulmonary blood flow and after 5 min of increased pulmonary blood flow. The order of the low-flow and high-flow conditions was randomized.

Data were analyzed for differences between conditions using analysis of variance (ANOVA) for repeated measures using a commercially available software package (SuperANOVA 1.11, Abacus Concepts Inc., Berkeley, CA). Since pulmonary blood flow varied between animals under the control and high-flow conditions, additional regression analyses (StatView 4.1, Abacus Concepts Inc., Berkeley, CA) were also performed to examine the relationship between pulmonary blood flow, and the inert gas data. Seventeen observations from the six animals were used in the regression analyses. The effects of between-subject differences on the dependent and independent variables were controlled for by assigning dummy variables to encode the different subjects and performing a multiple regression (Glantz and Slinker, 1990). Significance was accepted at P<0.05 (two-tailed, unless specified one-tailed in text). Results are reported as means ± S.E.M.

Complete data were obtained on six animals for control and high-flow conditions and are summarized in Tables 1 and 3. In one animal, we were not able to position the vascular occluder and therefore report data for five animals in the low-flow condition.

Pulmonary and systemic blood flows are given in Fig. 1. Initial (control) pulmonary blood flow varied markedly from animal to animal as shown by the high standard error. From control conditions, we were able to decrease pulmonary flow to one-sixth of the initial value and to increase it by a factor of almost two (P<0.001). Systemic blood flow did not change significantly throughout the experiment. Ventilation and metabolic data are given in Table 1. Ventilation was held constant throughout the experiment, and there were no significant changes in oxygen consumption rate ⁠, CO₂ production rate or respiratory exchange ratio during the study.

Blood gas data obtained from the pulmonary artery and left atrium are given in Table 1. There was a significant increase (one-tailed, P<0.05) in with increasing pulmonary blood flow, associated with a significant increase in (P<0.001). There were no other significant changes in blood gas levels. was significantly correlated with log(pulmonary blood flow) (Fig. 2, r=0.68, P<0.05) and changes in were reflected in (r=0.72, P<0.0001). Effective lung was calculated from the mixed expired gas concentrations and the inert gas dead space (Table 1) and increased significantly (P<0.05) with increasing pulmonary blood flow, although the changes were small. The difference between calculated effective and was significantly reduced with increasing pulmonary blood flow (one-tailed, P<0.05), and there was a significant negative correlation between log(pulmonary blood flow) and the effective difference (r=–0.67, P<0.05).

Blood gas partition coefficients for the six inert gases at 22 °C are presented in Table 2. The lower solubilities for SF₆, ethane and cyclopropane and the higher solubilities for the remaining gases, compared with mammalian values, are probably the result of the low haematocrits in these animals (18.1±1.6) (Young and Wagner, 1979). Inert gas data are summarized in Table 3. The residual sum of squares (RSS) is a measure of the goodness of fit of the inert gas data and is also an indicator of the technical adequacy of the data. The RSS follows a 𝒳² distribution and, with random experimental error, should be ⩽16.8 in 99 % of occurrences with six gases, although the RSS can also be increased in the presence of incomplete intrapulmonary gas mixing. However, there was no evidence for this, as all RSS data were within acceptable limits (Powell and Wagner, 1982). Fig. 3 shows a typical measured distribution during both low and high pulmonary blood flow conditions. The mean of the perfusion distribution decreased significantly (P<0.05) with increasing pulmonary blood flow. There was a significant decrease in logSD_Q̇ with increasing blood flow (one-tailed, P<0.05) and there was a significant relationship between logSD_Q̇ and log(pulmonary blood flow) (Fig. 4, r=–0.69, P<0.05), suggesting that matching is improved when pulmonary blood flow is increased.

Perfusion to areas of the lung with a low ratio (⁠ ratio less than 0.1) was 1 % of pulmonary blood flow during the low pulmonary flow condition and averaged 12 % during the high pulmonary flow condition. The difference between experimental conditions was not significant (Table 3). Intrapulmonary shunting varied from 11 % under the low pulmonary flow conditions to almost 20 % during the high pulmonary flow condition, but these differences were not significant (Table 3). Note that the catheter placement in the pulmonary artery and the left atrium excludes any contribution from cardiac shunts. Inert gas dead space decreased significantly with increasing pulmonary blood flow (P<0.01) (Table 3, VD). Inert gas dead space includes anatomical dead space and instrument dead space, unlike physiological dead space, which also includes ventilation to some areas of high ⁠.

Left atrial was predicted for each animal on the basis of the observed distribution and intrapulmonary shunt, and assuming diffusion equilibrium (Table 3) (Torre-Bueno, 1985). We used published data (Maginniss et al. 1980) on the oxygen dissociation curve of the turtle to modify the inert gas subroutine to allow for the difference in shape between the turtle and human oxygen dissociation curves. The turtle, unlike most mammals, has multiple haemoglobins and, therefore, the following data are subject to the limits of the approximation of the oxygen dissociation curve by the inert gas subroutine and must be interpreted with some caution. The predicted from the derived distribution was significantly greater than the observed at all levels of pulmonary blood flow (P<0.001), suggesting pulmonary diffusion limitation. The predicted did not change across conditions and was not significantly correlated with pulmonary blood flow.

In anaesthetized pump-ventilated turtles, we found that increasing pulmonary blood flow resulted in a decrease in ventilation–perfusion heterogeneity and an increase in left atrial ⁠. This is the first study to investigate the effect of pulmonary blood flow on relationships in an animal that markedly varies its blood flow under resting conditions.

Many reptiles, particularly turtles, are intermittent lung breathers. In these animals, brief ventilatory periods are interspersed among apnoeas of variable duration (Shelton et al. 1986). During apnoea, a bradycardia develops and there is a significant increase in the pulmonary vascular resistance, which leads to a reduction in Q̇l. For example, in the freely diving turtle Trachemys (Pseudemys) scripta, Q̇l may be as low as 10–15 ml kg⁻¹ min⁻¹ (Shelton and Burggren, 1976). In contrast, during brief ventilatory periods, the cardiovascular changes are the reciprocals of those occurring during apnoea. In Trachemys (Pseudemys) scripta, the heart rate increases, the pulmonary vascular resistance decreases and Q̇l can be as large as 50–70 ml kg⁻¹ min⁻¹ (Heisler et al. 1983; Shelton and Burggren, 1976). In the present study, we were able to alter pulmonary blood flow over a wide range (from 6 to 60 ml kg⁻¹ min⁻¹) and to measure the direct effects of these changes on pulmonary gas exchange.

Adrenaline has many effects and it is possible that the injection of the drug in the present study might have effects on the pulmonary vascular bed other than vasodilation. Adrenaline was chosen to increase pulmonary blood flow for the following reasons. In the rat snake Elaphe obseleta, recent studies have shown that electrical stimulation of vagal nerves produces a pulmonary vasoconstriction, followed by a pulmonary vasodilation during the post-stimulatory period (Donald et al. 1990). The post-stimulatory vasodilation is unaffected by atropine, but is eliminated following administration of propranolol (Donald et al. 1990). In turtles, stimulation of vagal afferents results in a twofold increase in pulmonary blood flow (Comeau and Hicks, 1994), which is attenuated following infusion of bretylium. These results suggest that the increase in pulmonary blood flow associated with intermittent lung ventilation may result from adrenergic vasodilation (Comeau and Hicks, 1994; Donald et al. 1990; Lillywhite and Donald, 1989).

The reduction in pulmonary blood flow during apnoea is under cholinergic control (Burggren, 1985; Comeau and Hicks, 1994; Hicks, 1994; White, 1976). In chelonians, approximately 50 % of the variable vascular resistance lies proximal to the lung parenchyma (Burggren, 1985). This has been demonstrated by in situ perfusion of distal pulmonary arteries, which show a strong cholinergic vasoconstriction and adrenergic vasodilation (Burggren, 1985). An additional important site for the regulation of pulmonary resistance is the bulbus cordis, a ‘sphincter-like’ internal wrapping of smooth muscle and an external covering of cardiac muscle at the base of the pulmonary artery (Burggren, 1977; March, 1961). The underlying smooth muscle is constricted by cholinergic agonists and dilated by adrenergic agonists (Burggren, 1985).

In the simplest approach, a single-compartment model, the mean alveolar ventilation divided by the mean blood flow can be used to determine the ratio. Although this approach has been used in previous studies in reptiles (Wood et al. 1977), it is of limited value, since it assumes that the lung is perfectly homogeneous with respect to and neglects the effect of the distribution of ratios within the many thousands of subunits within a lung. For example, a lung could have an overall ratio centred at 1; however, if there were marked heterogeneity about this mean ratio, pulmonary gas exchange would be substantially impaired. Conversely, if the mean ratio were low, yet blood flow was well matched to the available ventilation, with little heterogeneity, the overall effect on gas exchange would be minimized. The effect of increasing heterogeneity is to decrease both pulmonary oxygen uptake and carbon dioxide excretion, resulting in a decrease in and an increase in in the blood returning to the left atrium. Thus, most modern research focuses on the spatial distributions of ratios rather than the overall ratio of the lung, and several review articles explore this subject (Hlastala and Robertson, 1978; West, 1969; West and Wagner, 1991). Using the multiple inert gas technique, the distribution of ventilation and perfusion to areas of differing ratio can be represented graphically, and the dispersion or heterogeneity of blood flow and ventilation about the mean is given by the standard deviation of the distribution on a logarithmic scale (logSD_Q̇ or ⁠, respectively).

In normal mammalian lungs, logSD_Q̇ is almost always less than 0.5 (Gale et al. 1985); it may reach as high as approximately 0.7 during exercise in normal athletic humans (Hopkins et al. 1994) and approach 2.0 in patients with chronic obstructive pulmonary disease (West and Wagner, 1991). Ventilation–perfusion heterogeneity has been measured in a number of reptile species with values for logSD_Q̇ ranging from less than 0.5 in the alligator (Powell and Gray, 1989) and monitor lizard (Hopkins et al. 1995) to over 0.9 in the tegu lizard (Hlastala et al. 1985). In the turtle under control and low-flow conditions, there was moderate to marked heterogeneity; however, this improved with increasing pulmonary flow to values approaching that of the alligator and monitor lizard.

The reason for a worsening of matching under conditions of low flow is uncertain. In the mammalian lung, inter-regional heterogeneity may occur, where changes in the pulmonary vascular–alveolar pressure relationships may result in increased heterogeneity by increasing the area of the lung where the alveolar pressure exceeds the pulmonary arterial pressure (zone 1). In the turtle, however, different lung zones are unlikely, since gravity-dependent regional differences in blood flow are small and lung air space pressures are expected to be equal throughout. A second possibility is intra-regional heterogeneity, where local changes in vascular impedance secondary to changes in pulmonary vascular recruitment may alter relationships. In perfused dog lung preparations, where care was taken to keep the lung under zone 2 conditions (zone 2 conditions occur when pulmonary arterial pressure exceeds alveolar/lung air space pressure, and is expected in the turtle lung under the conditions of the present study) (Domino et al. 1991; Ohlsson et al. 1989), decreasing blood flow produced marked heterogeneity (logSD_Q̇ =1.00), suggesting that the effect of lowering pulmonary blood flow on heterogeneity can occur in the absence of inter-regional changes in heterogeneity.

However, in the dog lung, in contrast to the turtle lung, increasing pulmonary flow resulted in increased heterogeneity (Domino et al. 1991), possibly as a result of increased perfusion of corner vessels, increased perfusion to areas of high ventilation–perfusion ratio, or subtle pulmonary oedema resulting from the experimental procedure. Similar increases in heterogeneity are seen in humans (Hammond et al. 1986) and varanid lizards (Hopkins et al. 1995) during exercise, where pulmonary blood flow and vascular pressures are increased, and may be the result of interstitial pulmonary oedema. The reason for the differences between mammals and turtles is unknown. However, turtles are one of the few vertebrates that have large changes in pulmonary blood flow associated with increasing ventilation, in the absence of exercise.

We measured an intrapulmonary shunt of 17 % under control conditions, which is similar to the value obtained by Seymour (1983) using 100 % oxygen breathing at a lung volume providing neutral buoyancy. Since the oxygen method for detecting shunt measures the contribution of desaturated blood to the output of the left heart, both the intrapulmonary shunt and the shunt from extrapulmonary sources will be measured. These extrapulmonary sources of shunting, such as the bronchial circulation, will extract oxygen from the blood and will therefore be measured as contributing to the oxygen shunt. However, since inert gases will not be extracted by the bronchial circulation, they will not measure these extrapulmonary contributors to the oxygen shunt. Therefore, the shunt measured by the oxygen method will be greater than the inert gas shunt, particularly at low levels of shunting where extrapulmonary shunting will contribute a greater proportion of the total shunt (Hlastala et al. 1975). In our experimental preparation, care was taken to exclude any contribution from cardiac shunts by the placement of the sampling catheters in the pulmonary artery and the left atrium, and we cannot comment on other sources of shunting since 100 % O₂ was not given. Unlike the varanid lizard, where increasing pulmonary blood flow with exercise resulted in a decreased intrapulmonary shunt (Hopkins et al. 1995), the intrapulmonary shunt in the present study tended to increase with increasing pulmonary blood flow. This may represent a difference between species or may result from the differing experimental means (exercise versus adrenaline) of increasing pulmonary blood flow.

Inert gas dead space includes anatomical dead space and instrument dead space. Dead space ventilation was significantly reduced when pulmonary blood flow was increased. This reduction is probably a result both of the increasing perfusion of areas of the lung with high ratio with increasing pulmonary blood flow and of the recruitment of unperfused areas of the pulmonary circulation with increasing pulmonary blood flow.

The inert gas method allows an estimate of pulmonary diffusion limitation by computing the arterial and expired that would be expected from the recovered distribution. For each of the 50 compartmental values, compartmental oxygen concentrations are calculated for the inspired and mixed venous gas concentrations. Assuming lung gas–end capillary diffusion equilibrium, the appropriately weighted mixed arterial and mixed expired concentrations of O₂ are calculated and can be compared with the measured values. When the observed value is less than that predicted from the inert gas exchange, this suggests pulmonary diffusion limitation. Another possible explanation for the discrepancy between observed and predicted O₂ concentration is extrapulmonary shunting (e.g. cardiac, bronchial arteries and thebesian veins), since this would also result in a lower than that predicted from the distribution. Note that the placement of the catheters in the pulmonary artery and left atrium would preclude any cardiac shunts, and the effect of other sources of shunting is likely to be very small (Hlastala et al. 1975).

There was evidence of pulmonary diffusion limitation in the turtle, since the of left atrial blood was 16–21 mmHg lower than that predicted from the recovered distribution. Caution must be used in interpreting these data since the analyses are limited by our ability to approximate the oxygen dissociation curve of the turtle. We did not measure the oxygen dissociation curve in our animals, and therefore any deviation of these animals from published data (Maginniss et al. 1980) would be expected to affect the results. For example, if the P₅₀ of the animal were overestimated, it would increase the predicted and reduce the effective –predicted difference; conversely, the effect of overestimating P₅₀ would be to reduce the predicted ⁠. These uncertainties in P₅₀ are minimized in lungs with little inequality and increased in lungs with marked inequality. In humans (Hammond et al. 1986; Hopkins et al. 1994) and the varanid lizard (Hopkins et al. 1995), pulmonary diffusion limitation is seen only during very heavy exercise. Although the degree of diffusion limitation was constant across conditions, since the degree of heterogeneity and therefore the effective difference was reduced with increasing blood flow, the relative contribution of pulmonary diffusion limitation to the effective difference was larger at higher blood flow in the turtles.

In awake, spontaneously breathing varanid lizards (Hopkins et al. 1995), we found evidence of incomplete intrapulmonary gas mixing, and this has also been reported in the anaesthetized alligator (Powell and Gray, 1989). In the present study, our turtles were anaesthetized and ventilated at a high rate corresponding to the level of ventilation observed during a ventilatory episode. We did not find evidence of incomplete intrapulmonary gas mixing, as all residual sums of squares (RSS) were within the expected 𝒳² distribution for six gases (Powell and Wagner, 1982). Incomplete intrapulmonary gas mixing will increase the RSS since the retention of the heaviest gas (enflurane) will be increased in comparison with that of gases of low molecular mass (cyclopropane) (Downs and Wagner, 1983). It is likely that the high ventilation rate that was used in these animals improved convective gas mixing within the lung and prevented incomplete intrapulmonary gas mixing. Also, the turtle may have additional mechanisms which prevent limitations of gas exchange, resulting from incomplete gas mixing during periods of apnoea. Spragg et al. (1980) found that movement of limbs and contraction of smooth muscle within the lung increased the displacement of gas from the posterior to the anterior parts of the lung and facilitated gas mixing.

Increased pulmonary blood flow resulted in a decrease in heterogeneity, a decrease in inert gas dead space and an unchanged intrapulmonary shunt and pulmonary diffusion limitation. The net effect was a reduction in the effective difference, an increase in and improved pulmonary gas exchange. Although the intrapulmonary shunt was numerically greater in the high-flow than the low-flow conditions, the consequences of intrapulmonary shunting are minimized, since pulmonary mixed venous was significantly increased from 32 mmHg under the low pulmonary flow conditions to 80 mmHg under high pulmonary flow conditions. Thus, the shunted blood is at a higher and the effect on arterial of combining shunted blood with mixed venous blood is reduced. Arterial was also improved with increasing pulmonary blood flow because of reduced dead space ventilation and improved matching. The net effect of the improvement in heterogeneity was an increase in the predicted arterial from 127 mmHg to 135 mmHg.

We can conclude that, in the anaesthetized turtle, intrapulmonary shunting, pulmonary diffusion limitation and heterogeneity all contribute to the difference between lung gas and left atrial blood. This is in contrast to the situation in mammals and varanid lizards, which do not show appreciable pulmonary diffusion limitation at rest. In the turtle, heterogeneity is significantly reduced with increasing pulmonary blood flow and it appears that the increase in pulmonary blood flow observed with increasing ventilation in the awake animal probably results in reduced heterogeneity and improved pulmonary gas exchange.

Supported by NSF IBN-9218936 (J.H.). T.W. was supported by a postdoctoral fellowship from the Carlsberg Foundation (Copenhagen, Denmark) and S.H. was supported by NIH T32 HL 07212.