Analysis of Cardiac Shunting in the Turtle Trachemys (Pseudemys) Scripta: Application of the Three Outflow Vessel Model

The anatomically undivided ventricle of noncrocodilian reptiles has two inflow openings and three outflow vessels. The ventricular lumen of these animals is also divided into two subcavities. Two atria open into the dorsally located cavum dorsale (CD), which is often further subdivided into the cavum venosum (CV) on the right and the cavum arteriosum (CA) on the left, although such differentiation has not always been described (see Van Mierop and Kutsche, 1985). The right and left aortae (RAo and LAo respectively) arise from the right side of the CD. The cavum pulmonale (CP) is a smaller chamber located ventrally and to the right of the CD. The CP has no direct communication with either of the atrial orifices and is incompletely separated from the CD by a horizontal muscular septum. The pulmonary artery (PA) emerges from the CP. This complex spatial arrangement indicates that blood may be translocated within the ventricle during each cardiac cycle, resulting in intracardiac shunting. However, little information is available regarding details of blood movement within reptilian ventricles.

Cardiac shunting has been conventionally defined as either right-to-left (R–L) or left-to-right (L–R). A R–L shunt represents a fraction of the systemic venous blood being recirculated into the systemic circulation. A L–R shunt represents a fraction of the pulmonary venous blood flowing back into the pulmonary circulation. The conventional analysis of cardiac shunting in reptiles assumes identical blood oxygen levels in the two aortae (e.g. Burggren and Shelton, 1979). Therefore, four blood pathways, shunted and unshunted flows from each atrium into the pulmonary and systemic circuits, were assumed to exist. However, several studies have reported that oxygen levels in the two aortae are different (Khalil and Zaki, 1964; Tucker, 1966; Baker and White, 1970; Burggren and Shelton, 1979; Ishimatsu et al. 1988). This fact led us to develop a model delineating the six possible blood flow pathways within the ventricle (the three outflow vessel model, Ishimatsu et al. 1988).

The three outflow vessel model can be tested using several approaches. Our previous study on the Nile monitor lizard Varanus niloticus was performed by combining blood gas analysis and the microsphere method for determination of relative flow rates (Ishimatsu et al. 1988). Although cardiac shunting patterns could be demonstrated for chronically cannulated, recovered animals, this approach did not allow the analysis of blood flow distribution in terms of absolute flow rates. In the present study, this gap has been filled by measuring flow rates and oxygen levels of blood in all central vascular sites in an acute preparation of a turtle, Trachemys (Pseudemys) scripta. To extend this study to cover the dynamic range of blood flow rates, stimulation of the cervical vagus was used as a tool for an extended analysis of shunting patterns.

Turtles, Trachemys (Pseudemys) scripta Gray, were purchased from a commercial animal dealer (Lemberger, Oshkosh, WI, USA) and airfreighted to Germany. They were kept in large aquaria with dry basking areas and infrared lamps to allow behavioural regulation of body temperature. Water temperature in the aquaria was set at 25±1 °C. Animals were regularly fed chopped beef liver and chicken meat. Food was withheld for 3 days before experimentation.

Anaesthesia was induced as described by Hicks and Malvin (1992). Animals were then tracheotomized and mechanically ventilated (12–20 min⁻¹) with a humidified gas mixture of 17 % O2, 3 % CO₂ and 80 % N₂ prepared using a gas-mixing pump (Wösthoff, Bochum, Germany). A rectangular opening (approximately 3 cm×4 cm) was cut in the plastron with a bone saw to expose the heart and central blood vessels. The plastron piece was carefully removed from the underlying musculature. Bleeding from the musculature was stopped by cauterization. Short sections of the central vessels were freed from surrounding connective tissues and blood flow probes (2R ultrasound, transit-time blood flow probes; Transonic Systems Inc., Ithaca, NY, USA) were placed around the left pulmonary artery (PA), the LAo, the common carotid of one side (CC) and the main trunk of the right aorta (RAo,t). Care was taken not to kink or distort the vessels. Subsequently, five central vascular sites [PA, LAo, RAo,t, the left atrium (LAt) and the right atrium (RAt)] were non-occlusively catheterized as described by White et al. (1989). A thin thermistor was inserted into the cloaca to measure body temperature.

The cervical vagus on each side was exposed and prepared for nerve stimulation as described by Hicks and Comeau (1994). A few animals showing vasoconstriction following probe or catheter placement were excluded from data analysis.

During experimentation, animals were placed supine and mechanically ventilated as described above. Experiments were performed on seven turtles at 25±1 °C.

The first set of measurements was conducted to obtain blood gas levels and flows with intact vagi (V_intact). Blood samples were drawn simultaneously from the five sites while blood flows were recorded continuously. Immediately after withdrawal, blood was analyzed for ⁠, and pH and was subsequently reinfused into the animal to minimize blood loss.

After completing the first set of measurements, the vagi were exposed and sectioned as described above. Five different conditions for vagal stimulation were tested. These included measurements without electrical stimulation (V_cut), efferent stimulation of either right (RVEF) or left (LVEF) vagus, and afferent stimulation of either right or left vagus (RVAF and LVAF, respectively). Electrical conditions of stimulation were as follows: 1–3 Hz, 4–8 V and 200 ms. Animals were allowed to stabilize between stimulation periods. In three animals, sodium nitroprusside (32 μg) was injected as a bolus into the right atrium (RAt) to evoke extreme deviations of cardiovascular conditions. Nitroprusside injection tended to give lower α (see below) values, while no apparent effects were seen in the other partition fractions.

Blood ⁠, and pH were determined using two sets of blood gas analyzers (Radiometer BMS3 Mk2) thermostatted to 25±0.1 °C. We used previously reported data on the oxygen dissociation curve for T. scripta (Burggren et al. 1977; Maginniss et al. 1980) to calculate oxygen content from and pH. Oxygen capacity was estimated from its relationship with haematocrit (Hct, %): O2 capacity = 0.788+0.284Hct (J. W. Hicks, unpublished observation). Plasma bicarbonate concentration was calculated from pH and using values for pK′1 (the apparent first dissociation constant of the bicarbonate buffer system in plasma) and CO2 solubility reported by Nicol et al. (1983) for Chrysemys picta bellii. Two dual-channel Transonic flowmeters (T206, Transonic Systems Inc.) were used to determine blood flow in the four central blood vessels. Between blood sampling, catheters were connected to pressure transducers (Statham P23B) to record blood pressure.

Flow partitioning of the blood derived from both atria into the PA, LAo and RAo was calculated by substituting and blood flow data into equations previously reported (see equations 19–21 in Ishimatsu et al. 1988). Fractions of the inflow from the left atrium (LAt) into the ventricle are designated as α (to the PA), β (to the RAo) and 1–α–β (to the LAo). Fractions of the RAt inflow are referred to as X (to the LAo), Y (to the RAo) and 1–X–Y (to the PA).

Analysis of variance (ANOVA) with randomized block design was used to detect differences in measured and calculated parameters between experimental conditions. Dunnett and Bonferroni tests were used for pairwise comparison, as appropriate. A paired t-test was used to detect any significant difference between two variables obtained under the same conditions. Curve-fitting was performed using SigmaPlot for Windows (Jandel Scientific, San Rafael, CA, USA). Data are expressed as means ± S.D. wherever possible.

Blood gas levels in the RAo prior to vagal sectioning are summarized in Table 1. Blood and pH were in good agreement with previously determined data for the same species (Hicks and Comeau, 1994), and the acid–base status agreed well with the values interpolated from the data for chronically cannulated Chrysemys picta bellii (Glass et al. 1985). The blood was found to be lower by 10 mmHg (1.3 kPa) than that of Chrysemys picta bellii (Glass et al. 1985). The fact that was 50 mmHg (6.7 kPa) lower than in mechanically ventilated P. scripta (Hicks and Comeau, 1994) is at least partially attributable to differences in the oxygen concentration of the ventilation gas mixture (17 % O₂ in this study versus 21 % O2 in Hicks and Comeau, 1994).

Pulmonary blood flow (Q̇_p) varied, averaging 42±14 ml min⁻¹ kg⁻¹ (mean ± S.D., N=7), which is 14 ml min⁻¹ kg⁻¹ lower than the value reported for similarly prepared P. scripta by Hicks and Comeau (1994). Systemic blood flow (Q̇_s) was 20±6 ml min₋₁ kg₋₁. Detailed data on the haemodynamic variables of the animals are given in Hicks et al. (1996). Since the aim of the present study was to analyse patterns of intraventricular blood distribution under various conditions, no attempt was made to normalize the physiological conditions of the animals at the beginning of the experiment.

Values for in blood from the RAo and LAo are presented in Fig. 1. A paired t-test revealed that the mean of the RAo (70±8 mmHg, 9.3±1.1 kPa) was significantly higher than that of the LAo (64±12 mmHg, 8.5±1.6 kPa; P<0.001, N=43). Thus, the three outflow vessel model described in Materials and methods is applicable to the present data set.

The relationship between calculated and measured Q̇_RAo is presented in Fig. 2. The correlation coefficient (r) between the two variables is 0.715 (P<0.01). The ratio of calculated to measured values was 1.10±0.77 (mean ± S.D.).

Q̇_s and Q̇_p were similar below a Q̇_tot value of approximately 30 ml min⁻¹ kg⁻¹ but progressively deviated at higher values of Q̇_tot (Fig. 3A). The data were fitted by the following curves shown in Fig. 3A (for S.D. of the parameters and dependencies of the curves, see legend to Fig. 3A):

The distribution of LAt blood into the three outflow vessels is plotted against Q̇_tot in terms of partition fractions (Fig. 4A–C) and volume flow partitioning (Fig. 4D–F). As Q̇_tot rose to approximately 40–50 ml min⁻¹ kg⁻¹, the fraction α, flowing to the PA, increased sharply with increasing Q̇_tot to values as high as 0.7–0.8, but then levelled off (Fig. 4A). The fraction β, flowing to the LAo, varied considerably at low Q̇_tot, ranging from below 0.1 to above 0.9, but at higher Q̇_tot the variation was greatly reduced, with values for β being close to 0.2 (Fig. 4B). A similar but less variable pattern was found for 1–α–β (Fig. 4C). Flow into the PA (Q̇_p×α, unidirectional L–R shunt) increased linearly with Q̇_tot (Fig. 4D). Flow into the RAo (Q̇_p×β) was similar to the flow into the LAo [Q̇_p(1–α–β)] when Q̇_tot was lower than 40 ml min⁻¹ kg ⁻¹, but rose at higher Q̇_tot values (Fig. 4E,F, see figure legend for equations for regression lines).

The distribution of RAt blood into the three outflow vessels is plotted against Q̇_tot in Fig. 5. The turtles never developed a large net R–L shunt as reported in earlier in vivo studies (e.g. White et al. 1989). In only one animal was a high X+Y value of 0.55 attributable to a high X value (Fig. 5A), while in all the others X+Y was approximately 0.2. At larger values of Q̇_tot, both X and Y were well below 0.1, resulting in a value of 1–X–Y approaching 1 (Fig. 5C). The PA blood flow originating from the RAt [Q̇_s(1–X–Y)] increased sharply as Q̇_tot became larger (Fig. 5F). The blood flows into the Lao (Q̇_s×X) and RAo (Q̇_s×Y) were low, as expected from the observed low partition fractions (Fig. 5D,E), and were not significantly dependent on Q̇_tot (see figure legend).

Of the six partition fraction values, 1–X–Y showed the largest change in response to vagal stimulation. The mean 1–X–Y value (0.81) during RVEF was significantly lower than the values measured under all the other conditions (P<0.05, Bonferroni test in a randomized block design ANOVA). The value of 1–X–Y was highly constant during the other periods and there were no significant differences between any pair of values (P>0.05). X and β were not significantly affected by vagal stimulation (P>0.05). The very large variability of X, particularly for RVEF, is due to an exceptionally high value from one animal. If this value is disregarded, the mean partition fraction value X during RVEF was 0.062±0.042. This is not significantly different from the values found during the other periods. Values of Y were significantly different only between RVEF and LVAF (P<0.05). Values of α were significantly different between RVEF and LVAF (P<0.01), as well as between RVEF and RVAF (P<0.05). Significant differences in 1–α–β were found in all three pairwise comparisons, between RVEF and LVAF (P<0.05), between RVAF and LVEF (P<0.01), and between LVEF and LVAF (P<0.01).

in the LAt was statistically independent of Q̇_tot at approximately 80 mmHg (10.7 kPa; Fig. 6D), while in the RAt changed significantly as a function of Q̇_tot (Fig. 6D, see figure legend for regression equation). Only at low Q̇_tot, where the partition fraction value Y indicated some increase (see Fig. 5B), did in the RAo differ from in the LAt (Fig. 6A). in the LAo was lower than that in the LAt at low Q̇_tot (Fig. 6B). The difference in between PA and LAt remained at approximately 30 mmHg (4 kPa) at Q̇_tot values of more than approximately 40 ml min⁻¹ kg⁻¹, but at lower Q̇_tot there was a tendency for this difference to increase (Fig. 6C).

in the LAt was not affected by vagal stimulation and, accordingly, values are not presented (P>0.05, Bonferroni test in a randomized block design ANOVA). In only two cases could differences that were statistically significant be detected in values for RAt determined during different stimulation periods: V_intactversus LVEF and RVEF (P<0.05) (not shown in Fig. 7). Differences in LAo between RVEF and all other treatments were highly significant (P<0.01). As a result, the difference in between the LAt and LAo blood became significantly larger during RVEF compared with the values found in the other conditions (P<0.01). in the RAo was lowered by RVEF, but the resulting differences were only significant between V_intact and RVEF, between Vcut and RVEF and between LVAF and RVEF (P<0.05). The differences in PA measured during RVEF and under all other conditions were significant (P<0.05), except between LVEF and RVEF.

LAt was unaffected by vagal stimulation (overall mean 78±5 mmHg (10.4±0.7 kPa, N=40). RAt was also stable except for a small decline in response to LVEF and RVEF. Overall mean values for RAt were 28±5 mmHg (3.7±0.7 kPa, N=25) for all conditions except LVEF and RVEF and 22±5 mmHg (2.9±0.7 kPa, N=14) for these two conditions.

The oxygen content, calculated for blood from the LAt (LAt ), was not significantly different during RVEF from that during any other condition (P>0.05). The oxygen content of RAt blood (RAt ) during V_intact differed significantly from values during RVEF (P<0.05) and RVAF (P<0.01).

In several cases, large discrepancies were found between calculated and measured Q̇_RAo (Fig. 2). Checking these instances against data sets with better agreement revealed no apparent relationship with the absolute values of variables, except for indicating a correlation coefficient (r) of 0.61 for Q̇_p (P<0.01). Indirect estimation of oxygen content is subject to errors in the determination of and pH in addition to errors introduced by data from the oxygen dissociation curve. The effect of such errors on the estimation of partition fractions becomes larger at high Q̇_p because the differences in between two sites become smaller under these conditions (Fig. 6), introducing larger deviations in the equations for partition fractions in addition to Q̇_PA/Q̇_LAo (the partition fraction representing a R–L shunt, see Ishimatsu et al. 1988).

However, the differences found may also be related to slight violations of the boundary conditions of the model. There are two fundamental conditions for the validity of the vessel model: (1) conservation of mass (blood volume and amount of oxygen) during blood transit through the ventricle and (2) no net fluid movement through the peripheral capillary walls. If a significant fraction of the pulmonary circulation were filtered through the capillary walls, as claimed for the lungs of Trachemys scripta (Burggren, 1982), some degree of deviation would be expected.

Since LAt generally exceeds RAt , LAo should be lower, as the ratio of Q̇_s×X to Q̇_p(1–α–β) increases, and vice versa. There was a clear tendency for the ratio to increase when Q̇_tot fell below approximately 40 ml min⁻¹ kg⁻¹ (Fig. 8A). An inverse relationship between in the LAo and this ratio is clearly demonstrated in Fig. 8B. Similar relationships exist for the other two outflow vessels (RAo and PA, data not shown).

To our knowledge, this is the first complete description of blood flow partitioning within the turtle ventricle. The extensively invasive nature of the present technique makes it difficult to apply the method to recovered animals, but the information obtained from the present study forms the basis for a detailed analysis of the mechanisms involved in cardiac shunting in turtles.

The present analysis clarifies the complex relationship between the partitioning of blood from the two atrial inflow tracts towards the three cardiac output pathways. In the turtle, in most instances, blood shunting occurs simultaneously in both directions (R–L and L–R shunts) as previously found in Varanus niloticus (Ishimatsu et al. 1988). Where Q̇_tot is lower than 30 ml min⁻¹ kg⁻¹, α varies from below 0.1 to almost 0.6, and X and Y showed some positive values (Figs 4, 5) at which Q̇_p and Q̇_s were approximately the same (cf. Fig. 2). Patterns of blood mixing should therefore be carefully discriminated from net volume shunts. Net volume shunt relates to differences between Q̇_p and Q̇_s, and may be zero when there is substantial blood mixing in both directions.

Simultaneous measurement of Q̇_p and Q̇_s can provide information on net shunting, but will not detect unidirectional blood mixing. Conversely, oxygen analysis of the blood in the central vascular site can, on its own, provide partition fractions only when blood gas levels are identical in the two systemic arches. In other cases, additional information is required; for example, that obtained by application of the microsphere method to assess partition fraction values. Even then, an estimate of absolute volume flow cannot be made. Accordingly, a combination of blood gas analysis and flow measurement is required to obtain a full, quantitative picture of cardiac shunting as presented in this study.

With increasing Q̇_tot, 1–X–Y approached 1.0 while α reached a maximum value of approximately 0.8. Under these conditions, all RAt blood would be ejected into the PA whereas 20 % of the LAt blood would be directed into the systemic arches. Although theoretically possible, a total shutdown of systemic circulation and exclusive perfusion of the pulmonary circuit did not actually take place. In contrast, there is apparently an upper limit for Q̇_s, above which Q̇_p increases Q̇_tot linearly within the blood flow range covered by this study. The differences in blood oxygen levels (Fig. 1) and flow rates (Fig. 3B) between the two systemic arches enables a preferential supply of oxygen to the upper body mass, including the central nervous system.

199_11_2467C18With increasing Q̇_tot, the in PA blood rises above Rat , indicating recirculation of pulmonary venous blood back into the pulmonary circuit (Fig. 6). Hicks et al. (1996) have demonstrated that a net L–R shunt is a function of Q̇_tot. Fig. 9 plots the X, Y and α values against percentage volume flow in the pulmonary blood circuit ejected before the systemic blood flow starts (%Q̇_ppre). Nine data points, determined simultaneously in the same individual, were available. α apparently increased with %Q̇_ppre (Fig. 9C), while Y decreased with %Q̇_ppre (Fig. 9B). There was no apparent relationship between X and %Q̇_ppre (Fig. 9A). Regression analysis revealed a significant linear relationship between Q̇_tot and %Q̇_ppre (%Q̇_ppre=–4.51+0.128Q̇_tot, P<0.05; not shown). It should be noted that the flow from the LAt to the PA (Q̇_p×α) far exceeded the flows from the RAt to the RAo or the LAo (Q̇_s×X or Q̇_s×Y) when Q̇_tot was greater than 30–40 ml min⁻¹ kg⁻¹ (Figs 3A, 5). Thus, at high Q̇_tot, relatively small volumes of RAt blood residing in the CV at the end of diastole would be conveyed into the CP or even into the pulmonary circuit by movement of blood within the ventricle induced by the earlier ejection into the PA. This blood movement (counter-clockwise, viewed from the atrium) and the position of the RAo opening (lying further ‘upstream’ than that of the LAo) may be partly responsible for the higher oxygen levels in this arch. It has been pointed out that blood ejection into the pulmonary circuit precedes that into the systemic circuit by about 50–150 ms in Reptilia (Johansen et al. 1970; Shelton and Burggren, 1976; Ishimatsu et al. 1988; Hicks et al. 1996).

Angiographic observations (Hicks et al. 1996) have demonstrated that opaque material, injected into the CA just before pulmonary ejection during stimulation of vagal afferent nerves, flowed as a thin streak into the PA. This clearly indicates translocation of the material from the CA into the CP. During early systole, the muscular ridge evidently did not abut the opposing ventricular wall. At the end of systole, the CD is totally filled by LAt blood. During the next diastole, LAt blood lying in the CV will be washed into the CP (the washout mechanism, Heisler et al. 1983; Heisler and Glass, 1985). The extent of blood movement would be affected by (1) the ejection timing differences which, in turn, are influenced by the difference in diastolic pressures between the pulmonary and systemic circuits (Hicks et al. 1996), (2) the vascular resistance of the pulmonary circuit, (3) the contractility of the ventricular myocardium, and possibly (4) the direction of propagation of the excitation wave, as found by Burggren (1978).

In vivo studies have shown that, under certain conditions (e.g. during ventilation), Q̇_p greatly exceeds Q̇_s (Shelton and Burggren, 1976; West et al. 1992). Such physiological states may result in a blood distribution pattern similar to that seen in mammals and birds, where the systemic circuit is perfused exclusively with LAt blood and the whole volume of RAt blood is directed into the pulmonary circulation. A perfusion pattern such as this, which would be advantageous for gas exchange, would entail L–R shunting in turtles, as depicted by the present study.

We would like to thank Professor J. Piiper for invaluable discussion. This research was supported by Deutsche Forschungsgemeinschaft.