Partitioning of Body Fluids and Cardiovascular Responses to Circulatory Hypovolaemia in the Turtle, Pseudemys Scripta Elegans

Vertebrate animals typically maintain a relatively constant blood volume and pressure through regulatory mechanisms that are both numerous and inter-related (see review by Little, 1981), resulting in a circulatory homeostasis upon which many bodily functions depend. The efficacy of these regulatory mechanisms has been largely demonstrated in experiments that induce changes in circulation volume (through haemorrhage or transfusion) in mammalian models; comparatively little is known concerning similar regulatory mechanisms operating in the cardiovascular systems of other vertebrate taxa.

A unique tolerance to circulatory hypovolaemia has been demonstrated in snakes in which cardiovascular and respiratory homeostasis are maintained despite haemorrhages exceeding 80–100% of the snake’s initial blood volume (Lillywhite & Smith, 1981; Lillywhite, Ackerman & Palacios, 1983). The primary mechanisms that counteract hypovolaemia in these reptiles are a rapid and extensive mobilization of interstitial fluid to the intravascular space (Lillywhite & Smith, 1981; Lillywhite & Pough, 1983; Smits & Lillywhite, 1985) and reflexive adjustments of cardiac performance and vascular tone (Lillywhite & Seymour, 1978; Lillywhite & Smith, 1981). The volume of interstitial fluid is also proportionately greater in snakes than in mammals, and essentially all the fluid absorbed into the vascular space during haemorrhage comes from the interstitium (Smits & Lillywhite, 1985).

Turtles represent an evolutionarily distinct taxon of reptiles and have served as model animals in cardiovascular research for decades. However, comparatively few data exist that describe the compartmentation of body fluids between the vascular and non-vascular locations in these reptiles, or the capacity for fluid movement between these locations to maintain circulatory homeostasis.

In this paper we report how the intracellular and extracellular body fluid is partitioned within the pond turtle, Pseudemys scripta. Further, the capacity for blood volume regulation in these reptiles has been tested by employing tracer dilution techniques to measure the extent and source of transcapillary fluid transfer during hypovolaemia induced by haemorrhage.

Pond turtles (Pseudemys scripta elegans) were obtained through a commercial supplier (Lemberger Associates, Wisconsin) and were kept in fresh water (24–28°C) on a diet of vegetables, fish and meat. The 14 adult turtles (body mass, x̄ ± S.D. = 1466 ± 173 g) used in this study were in apparent good health and normal state of hydration prior to experimentation.

Surgical anaesthesia was induced by lowering the turtles’ body temperature in an ice bath for 1 h; cold narcosis was maintained during surgery by covering the animals with chipped ice. A hole (4 cm diameter) was drilled through the plastron to allow access to the major blood vessels and the heart. The left subclavian artery was occlusively cannulated (PE-90, i.d. = 0·860 mm, o.d. = 1·270 mm) and the catheter was advanced to the most proximal portion of the artery. The catheter passed through the body cavity via a stab wound through the loose skin between the neck and left forelimb. The patency of the catheter was maintained by occasional flushing with small volumes (<0·2ml) of heparinized (100 units ml ⁻¹) turtle Ringer’s solution (Rogers, 1938).

The compartmentation of body fluid in each turtle prior to haemorrhage was assessed by tracer dilution analysis. Deuterium oxide (D₂O), ⁵¹Cr-labelled erythrocytes and thiocyanate ion (SCN⁻) were used to measure total body water (TBW), whole blood volume (BV) and extracellular fluid volume (ECFV), respectively.

The D₂O (approx. 2·5 ml kg⁻¹ body mass) was injected intraperitoneally at 16.00 h the day prior to the experiment. Turtles were kept overnight in dry but covered boxes to reduce both cutaneous D₂O exchange and evaporative water loss. A blood sample taken immediately preceding experimental haemorrhage was vacuum distilled to remove organic components, and the D₂O concentration of this body fluid was determined by comparison with D₂O standards (l·0–5·0ml D₂O1⁻¹) using infrared spectrometry (Zweens, Frankena, Reicher & Zijlstra, 1980).

Turtle BV was measured by the dilution of autologous, ⁵¹Cr-labelled erythrocytes infused into the arterial catheter approximately 3h prior to experimental haemorrhage. Details of the labelling procedure are described by Lillywhite & Smits (1985). Briefly, 1 ml of each turtle’s erythrocytes was incubated with 10 μCi of isotopic sodium chromate at room temperature for 1 h. Following incubation the erythrocytes were repeatedly washed with a plasma-saline solution to eliminate the unbound isotope. The erythrocytes were then diluted in normal saline to the volume originally withdrawn from the turtle, and a known volume and specific activity of the labelled blood was infused into the arterial catheter. The blood volume was obtained by dividing the total activity infused (counts per minute) by the specific activity of the turtle blood (c.p.m. ml⁻¹) sampled 3 h following tracer infusion. Radioactivity in all blood samples drawn for BV determination was assayed by counting each sample for 5 min (approx. 1% sigma) on a Packard Auto-Gamma 800C scintillation counter. The specific gravity of whole blood (1·05 gml⁻¹) was used for conversions of mass to volume (Thorson, 1968).

Dilution of the thiocyanate ion was used to estimate the ECFV of turtles. NaSCN was infused into the arterial catheter (0·025 g NaSCN kg⁻ body mass) 4h prior to experimental haemorrhage. Concentrations of the thiocyanate ion in turtle plasma were measured colourimetrically (Bradshaw & Shoemaker, 1967) using a Zeiss PM QII spectrophotometer. The pre-haemorrhage ECFV was calculated from concentrations of the thiocyanate ion in four plasma samples (0·2 ml) collected at 15-min intervals in the hour preceding haemorrhage. The exact quantity of each of the three tracers injected was determined gravimetrically.

Tracer equilibration times were determined on one turtle (1290 g), in which the tracer concentrations were measured in the turtle’s blood sampled at intervals up to 3 h (⁵¹Cr) and6h (NaSCN) following tracer infusion (Fig. 1). These data confirmed that adequate time was given (3 h for ⁵¹Cr; 4 h for NaSCN) for these tracers to equilibrate with their respective fluid spaces. Deuterium oxide was found to equilibrate with the TBW of turtles in 5–6 h at 25 °C; thus, adequate equilibration time was given for this tracer by administering it the day previous to the TBW determination.

Immediately following measurements of steady-state fluid volumes, turtles were subjected to a graded haemorrhage to assess the extent of inter-compartment fluid transfer in response to hypovolemia. Turtles were conscious and at room temperature (22–26°C) during haemorrhage, and rested in a horizontal, ventral up position. Arterial blood pressures and heart rates were measured from the subclavian arterial catheter using a Narco P1000A Linear Core pressure transducer coupled to a DMP-4A physiograph (Narco Biosystems, Inc.). Haemorrhage was begun only after arterial blood pressure and heart rate had appeared to stabilize.

Graded haemorrhage was accomplished by withdrawing 4% of the turtle’s estimated B V at 10-min intervals until a total of 48% of the pre-haemorrhage BV had been removed. Initial blood volumes were assumed to equal 7% of the turtle’s body mass based on previous BV measurements in the same species (Hutton, 1961; A. W. Smits & M. M. Kozubowski, unpublished observations). Each interval blood sample was withdrawn from the subclavian arterial catheter into a clean syringe and partitioned approximately equally for the separate analyses of whole blood (⁵¹Cr) and plasma (NaSCN). Samples of whole blood were delivered directly from the withdrawal syringe into pre-weighed counting tubes. These tubes were reweighed, sealed with Parafilm, and counted the following day for gamma radioactivity. Plasma fractions from each withdrawal were obtained by immediate centrifugation (5 min in IEC Clinical centrifuge), and were pipetted into sealable plastic vials for later analysis of NaSCN concentration. The total osmolality of this plasma was also determined using vapour pressure osmometry (Wescor, Model 5100C). Haematocrits were measured on every blood sample by spinning blood-filled microcapillary tubes for 5 min in an IEC MB microhaematocrit centrifuge (13 000g). Arterial blood pressures and heart rates were measured continuously between consecutive withdrawals from the subclavian arterial catheter; mean pressures and heart rates reported were those measured during the last minute of each 10-min sampling interval. Analysis of tracer concentrations in 1-ml blood samples taken 0·5, 1·0 and 2·0h after haemorrhage ceased (−48%) were used to assess body fluid distribution during post-haemorrhage recovery.

The volume and source of fluid mobilized into the vascular space were also determined by tracer dilution analysis, based on the assumption that the dilution of a particular tracer within a fluid space was proportional to the amount of unlabelled fluid entering that space. Decreases in the specific activity (c.p.m. ml⁻¹) of the ⁵¹Cr within blood samples withdrawn during haemorrhage were used to calculate the volume of extravascular (unlabelled) fluid transferred into the vascular space. Because the calculation of BV by ⁵¹Cr-labelled erythrocytes requires a knowledge of the total radioactivity infused (see above), all calculations of BV subsequent to the initial BV measurement were corrected to account for ⁵¹Cr activity withdrawn from the turtles. The concentration of the thiocyanate ion was used to determine the source of fluid entering the vascular space during haemorrhage. Because both the plasma and the interstitial fluid contained the same thiocyanate concentration at equilibrium, the thiocyanate concentration was assumed to remain unchanged if the source of absorbed fluid was solely from the interstitium. Reductions in the plasma thiocyanate concentration during haemorrhage were considered to indicate movement of intracellular (unlabelled) fluid into the extracellular space. The thiocyanate tracer is not a permanent marker within the ECF, but is normally eliminated from the extracellular space of reptiles at a very slow rate (A. W. Smits, unpublished observations). The time constant of NaSCN elimination was sufficiently large (Fig. 1) to support the assumption that natural decay of this tracer during experimental haemorrhage (2 h) was negligible.

In addition to these experiments where a discrete volume of blood (48% B V) was removed, five turtles were subjected to an extended haemorrhage. Four percent of the pre-haemorrhage BV was removed every 10 min until arterial pressure could no longer be regulated, or until blood could no longer be withdrawn through the arterial catheter. All procedures (surgery, tracers, haemorrhage rate, fluid sampling and analyses) except the extent of haemorrhage were the same as described above. Albumin and globulin fractions of protein within the blood plasma (sampled throughout haemorrhage) were assayed using Sigma Chemical Diagnostic Kits Nos. 630 and 560, respectively. Following the experiment turtles were held overnight in dry, plastic containers after their plastrons had been sealed with a wooden plug and medical grade silicone rubber (Elastomer Silastic, Dow Coming).

The compartmentation of body fluid measured within turtles by tracer dilution analysis is schematically illustrated in Fig. 2. The whole blood volume (PV+RCV) averaged 6·89 ± 0·33% of the body mass (x̄ ± S.E.), and ranged from 5·74 to 8·15%.

The maximum net decrease in circulating blood volume (BV) demonstrated by turtles experiencing a graded removal of whole blood averaged only 14·4% of their pre-haemorrhage BV (Fig. 3), despite a cumulative withdrawal of 48%. The dramatic departure of the lines in Fig. 3 depicting changes in BV (solid circles) and the predicted BV changes in the absence of autotransfusion (dashed line) indicates that turtles mobilized substantial quantities of fluid from extravascular to vascular locations. The rate at which this fluid was transferred into the vascular space accelerated as haemorrhage progressed, until the later stages of the experiment (−40 to −48% BV deficit) when rates of autotransfusion equalled the rates of experimental haemorrhage (Fig. 3). Restoration of the BV to pre-haemorrhage levels was accomplished within 1–2h after bleeding was discontinued.

Transcapillary fluid transfer determined by ⁵¹Cr dilution (Fig. 3) was consistent with the degree of haemodilution observed in the haematocrit (Fig. 4). Haematocrits decreased significantly (paired comparisons t-test; P< 0·001) from (x̄±S.E.) 22·1 ± 1·89% to 13·6±1·41% during the course of haemorrhage, representing a mean reduction of 38·5%. Following 2h of post-haemorrhage recovery, the haematocrit decreased further to 11·8 ± 1·22%. Thus, the cumulative decrease in haematocrit represented a haemodilution of 46·6%, a volume essentially identical to the haemorrhage deficit (−48%).

There were no measurable changes in plasma NaSCN concentration in any of the turtles during the course of haemorrhage, indicating that the source of fluid mobilized into the vascular space was entirely extracellular. The total osmolality of the plasma also appeared unaffected by the experimental haemorrhage, as average osmolalities before (298 ± 16mosmolkg⁻¹) and after haemorrhage (298 ±21 mosmolkg⁻¹) were identical.

Arterial blood pressures (Fig. 5) were significantly reduced (33%) as a result of experimental haemorrhage (paired comparisons t-test, P<0·01), from the initial mean (±s.E.) of 39·2 ± 3·27mmHg (range, 25·0 – 57·5 mmHg) to 26·1 ± 3·17mmHg (range, 16·5–45·6mmHg). Initially-measured pulse pressures (5·10mmHg) typically decreased to 3–5 mmHg in response to the bleeding. Turtles raised arterial blood pressures during the 2-h recovery period to a level (34·8 ± 3·14) that was not significantly different from that of the pre-haemorrhage value (P> 0·2). The average heart rate of turtles (Fig. 5) at the cessation of haemorrhage (41·9 ± 2·19 beats min⁻¹) was 120% higher than the initially-measured heart rate (19·1 ± 3·43 beats min⁻¹). Although heart rate slowed significantly (P<0·02) during the recovery period to 36·1 ± 2·49 beats min⁻¹, this rate was still significantly greater (86·5%, P<0·001) than that measured just prior to haemorrhage.

The changes in B V measured during the initial 48% haemorrhage deficit in turtles subjected to an extended haemorrhage (Fig. 6) were similar in pattern and magnitude (−10·6 ±3·5%; t-test following arcsin transformation, P> 0·35) to those observed in turtles in the previous experiment (Fig. 2). Despite continued haemorrhage, BV in these turtles increased dramatically during the later stages of the experiment to levels that exceeded pre-haemorrhage values by 30% (Fig. 6). The greatest increments in BV occurred following the removal of 80% of the initial BV and concomitant with an accelerated decline in arterial blood pressure. The maximum heart rate (42·3 ± 1·07 beats min⁻¹) was observed at an 80% B V deficit and declined from this point until the cessation of haemorrhage.

Concentrations of albumin (16·9 ± 0·51 gdl⁻¹) and globulin (28·5 ± l·78gdl⁻¹) within the plasma decreased gradually throughout extended haemorrhage to values that were 33% (11·3 ± 0·96gdl⁻¹) and 38% ( 17·6 ± l·38gdl⁻¹) of the pre-haemorrhage levels, respectively. All turtles survived the extended haemorrhage to the following day without access to water, and several were kept alive for 1–2 months in the laboratory in plastic tanks containing fresh water.

The mean BV of turtles at steady state (6·89% body mass) is similar to the BV assumed (7%) for the subsequent haemorrhage protocol, and is consistent with BV values reported for Pseudemys scripta and Chrysemys picta (combined; 8·1%) measured by the dilution of ⁵¹Cr-labelled erythrocytes (Wilson, Hansard & Cole, 1960). The small discrepancy in the BV of Pseudemys scripta reported here and that measured by Wilson et al. (1960) or Hutton (1961; 8·75% body mass) presumably reflects differences in body sizes of the experimental animals, as plasma volume appears to be inversely related to body mass inPseudemys (Hutton, 1961).

Apparently no marker substance used in the dilution analysis of the ECFV is universally acceptable (Law, 1982), thus, only studies utilizing similar techniques and tracers may be compared. The ECFV (thiocyanate space) of Pseudemys scripta (Fig. 2) exceeds the thiocyanate space measured in rats (33% body mass; Huang & Bondurant, 1956) and is intermediate in volume compared to the thiocyanate spaces of other reptiles (Coulson & Hernandez, 1953; Minnich, 1982; Smits & Lillywhite, 1985).

Total body water in Pseudemys (75·3% body mass; Fig. 2) is greater than that measured in other freshwater or marine turtles (64·0–72·9%; Thorson, 1968), but is very similar to values of TBW in other species of reptiles (review by Minnich, 1982).

Decreases in systemic arterial pressure induced by haemorrhagic hypovolaemia in higher vertebrates initiate a baroreceptor-controlled stimulation of the sympathetic nervous system that primarily increases peripheral resistance and venous return through vasoconstriction, and secondarily improves cardiac output (Guyton, 1981; p. 333). Compensation for hypovolaemia via fluid resorption is typically considered a more passive process resulting from a reduction in capillary hydrostatic pressure (Djojosugito, Folkow & Kovach, 1968) and from the osmotic gradients caused by a hyperglycaemic hyperosmolality of the plasma (Ware, Norberg & Nylander, 1980), although activation of β-adrenoreceptors (Hillman, Gustafsson & Lundvall, 1982) appears to contribute significantly to fluid resorption.

The regulation of systemic blood pressure in Pseudemys in response to hypovolaemia appears poor despite dramatic increases in heart rate (Figs 5, 6). Systemic pressure seems largely dependent on the BV deficit (Figs 3, 5). The dependence of pressure regulation on BV is perhaps best exemplified by the return of arterial blood pressure to pre-haemorrhage values during the recovery period, concomitant with a complete restoration of pre-haemorrhage BV (Figs 3, 5). This recovery of blood pressure indicates that turtles can withstand a gradual removal of at least 48% of their resting BV without entering severe and irreversible circulatory shock.

The experimental rate of haemorrhage used to demonstrate the tolerance of terrestrial snakes to hypovolaemia (Lillywhite & Smith, 1981; Smits & Lillywhite, 1985) was duplicated in the present study, thus the haemodynamic responses and changes in B V during hypovolaemic stress in turtles and snakes may be quantitatively compared (Table 1). Turtles and snakes are similar in their capacity to mobilize substantial volumes of extracellular fluid from extravascular to vascular locations during hypovolaemia; this supports the inference that the resistance to fluid movement across the capillary wall in reptiles is comparatively low (Lillywhite & Smits, 1984; Smits & Lillywhite, 1985). Unlike mammals, turtles and snakes do not demonstrate an increase in plasma osmolality during haemorrhage. Thus, the transfer of fluid between extravascular and vascular locations does not appear to be facilitated by a hyperglycaemic hyperosmolality in these reptiles. The observed stability of extracellular osmolality also indicates that turtles do not resorb significant volumes of dilute urine (< 100 mmol kg⁻¹) during acute hypovolaemia.

Turtles do not maintain systemic arterial pressure as effectively as snakes in response to hypovolaemia, despite a proportionately greater increment in heart rate (Table 1). Baroreceptors located at the base of the pulmonary artery in Pseudemys have been characterized by recording afferent vagal nerve activity synchronous with the heart rate (Faraci, Shirer, Orr & Trank, 1983), but the primary effector(s) of these mechanoreceptors remain in question. The haemodynamic responses of Pseudemys to hypovolaemia in the present study (Figs 5, 6) and to drug-induced changes in blood pressure (Millard & Moalli, 1980) suggest that the heart, rather than the peripheral vasculature, may be the primary effector of barostasis.

Because measurements of B V in turtles are based on changes in the concentration of ⁵¹Cr-labelled erythrocytes, a rapid change in haematocrit influenced by contraction of the spleen during haemorrhage may invalidate B V estimates. However, experimental turtles did not demonstrate transient increases in haematocrit during any stage of the haemorrhage (Fig. 4), thus we feel that the contribution of the spleen to erythrocyte recruitment and changes in BV during the experiment is negligible.

Turtles enduring an extended haemorrhage over-compensate for their initial BV deficits and become increasingly hypervolaemic (Fig. 6). The restoration of BV in these turtles does not augment the regulation of systemic blood pressure or significantly reduce the heart rate, in contrast to turtles bled 48% of their initial BV (Fig. 5). This dramatic shift of fluid into the vascular space appears more related to changes in the hydrostatic than osmotic pressures of the Starling equilibrium, as the proportional decreases in protein fractions that contribute to plasma oncotic pressure are constant throughout the experiment (Fig. 6). We have also measured dramatic increases in the concentrations of plasma epinephrine and norepinephrine at times concurrent with the accelerated vascular resorption (−84% to −96% deficit; A. W. Smits, M. M. Kozubowski & K. J. Renner, in preparation), and suggest that this dramatic change in circulating BV may be related to the humoral effects of these catecholamines.

In summary, the capacity of aquatic turtles to counteract hypovolaemic stress through fluid recruitment (resorption) into the vasculature is exceptional and equals (if not exceeds) rates of autotransfusion measured in terrestrial snakes (Smits & Lillywhite, 1985). As in snakes, the ECFV is comparatively large and is the primary source of fluid that is apparently mobilized across a low resistance between the interstitial and intravascular spaces. Although turtles tolerate and survive substantial losses of their blood, their haemodynamic responses to experimental hypovolaemia suggest that barostasis through reflexogenic adjustments in vascular tone in these reptiles is comparatively weak.

This study was supported by a Grant-in-Aid research award by the Sigma Xi Research Society to the authors, and by the U.S. Public Health Service, NIH Grant HL 24640 to H. B. Lillywhite. We thank H. W. Shirer and H. B. Lillywhite for the use of their laboratories, and Fran Williams for typing the manuscript.