INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

SOME MAMMALS CAN MAINTAIN the brain, or at least parts of it, at a temperature cooler than that of the arterial blood supply from the trunk (1, 20). This phenomenon has been termed selective brain cooling (SBC). It is most conspicuous in mammals possessing a carotid rete. In these mammals, arterial blood on its way to the brain is cooled in the carotid rete via heat exchange with cool venous blood returning from the respiratory surfaces of the nasal cavity (1). It is thus postrete arterial blood temperature that is affected by SBC with the cooled arterial blood lowering the temperature of perfused regions of the brain. Quantification of SBC therefore requires knowledge of prerete blood temperature and brain temperature. Hayward and Baker (18) found that in monkeys, no difference in temperature existed between arterial blood in the aorta and at other sites as distant as the anterior cerebral artery, basilar artery, or abdominal aorta. On that basis, Baker and Hayward (3) proposed that carotid blood temperature provides a reliable measure of prerete arterial blood temperature.

Measurement of carotid blood temperature is not a simple task, and many investigators have relied on other measures of trunk temperature as a surrogate for prerete arterial blood temperature. If these surrogate temperatures were the same as arterial blood temperature, then using them to decide the degree to which an animal was using SBC would entail no error. However, temperatures measured at different trunk sites are not identical (5, 6, 19, 30, 40), so using surrogate temperatures may confound the determination of SBC. For example, there is a definite threshold of arterial blood temperature for onset of SBC in goats (27), but using rectal and brain temperature to estimate SBC, Mitchell et al. (33) concluded that a related artiodactyl, the sheep, always used SBC even during cold exposure. Rectal temperature (T_re) often exceeds arterial blood temperature in sheep (19, 40, personal observation), so the results of studies using T_re to estimate SBC require reinterpretation. Most of the studies that report apparent SBC in mammals that do not possess a rete, and in birds, have used some surrogate trunk temperature rather than arterial blood temperature to estimate SBC and so may be similarly confounded.

The main aim of our study was to evaluate the utility of T_re in estimating whether an animal is using SBC. We have used the adult sheep as the model for this study, because all published studies reporting SBC in sheep have relied on T_re as a surrogate for arterial blood temperature. We have measured brain temperature in the hypothalamus (T_hyp), arterial blood temperature in a common carotid artery (T_car), and T_re in sheep exposed to 40, 22, and 5°C. We also report observations on some nonthermal factors that influence SBC in the laboratory.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Animals. Four Dorper-cross ewes (average mass ± SD = 54.9 ± 8.4 kg) were used. The sheep were housed individually in connecting indoor pens where ambient temperature varied between 21 and 25°C and a natural light-dark cycle was maintained. Water was provided ad libitum, and commercial feed (Epol, Johannesburg, South Africa) and lucerne chaff were provided in the morning. Fresh straw was provided daily. The procedures were approved by the Animal Ethics Screening Committee of the University of the Witwatersrand (protocol 95/14/5).

Temperature measurements. Under halothane general anesthesia (8% for induction, 2% for maintenance) and in sterile conditions, each animal was fitted with blind-ended thermocouple guide tubes for measurement of carotid blood and brain temperature. The carotid guide tube consisting of a polythene core ensheathed in silicone rubber (1.4 mm internal diameter, 1.9 mm outer diameter) was inserted into the left common carotid artery through a puncture made with a hypodermic needle. The puncture was sealed, and the guide tube was secured in place, with a purse string suture around the tube that did not occlude the artery. Outside the artery, the guide tube was enclosed in a thicker silicon rubber tube that was secured to muscle near the site of insertion, exited the skin on the neck, and was attached to a cloth collar. The brain guide tube consisted of a length of acetate tubing (1.6 mm internal diameter, 3 mm outer diameter) through a nylon head plate. Anatomic markers were used to direct the guide tube toward the hypothalamus via a hole drilled in the skull 3 mm left of the midline. The nylon head plate was secured to the skull with bone screws. Correct positioning in the hypothalamus was confirmed post mortem.

Procedures. Beginning 10 days after surgery, several sets of measurements were made on each animal. Sheep were placed individually into cage trolleys, transported to a temperature-controlled room, and had thermocouples inserted by 0730 local time. There were always two animals in separate trolleys in the climate chamber during an experiment. Initial ambient temperature was maintained at 22-23°C. On different days, the animals were exposed in random order to different ambient temperatures after 0900. The animals either remained in the same climate room where ambient temperature was maintained at 22°C until 1500, remained in the same climate room where ambient temperature was increased to 40°C over ~1 h and then maintained at 40°C until 1500, or wheeled into a cool room maintained at 5°C until 1500. Every 30 min, an investigator entered the climate chamber to measure dry and wet-bulb temperatures with a whirling psychrometer and to time 30 respiratory movements.

Statistical analysis. Two-way, repeated-measures ANOVA, followed by Student-Newman-Keuls tests where appropriate, was used to compare hourly means of T_hyp, T_car, and T_re during 40, 22, and 5°C exposures. Trunk  brain temperature differences (T_car  T_hyp and T_re  T_hyp) were compared in the same manner. One-tailed t-tests were performed on the hourly means of T_car  T_hyp and T_re  T_hyp to test the null hypothesis that they were greater than zero. Breath frequency during heat exposure was tested with a repeated-measures ANOVA. Probabilities <0.05 were taken to indicate that effects were greater than could be accounted for by random variation. All data reported are means ± SE or means ± SD, as stated.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

During 40°C exposure (Fig. 1), temperatures varied among the three sites of measurement [F(2,6) = 13.78, P = 0.006] with T_re always being significantly higher than T_car and T_hyp. The temperature at all three sites increased with duration of the exposure [F(6,18) = 29.5, P < 10⁶]. There was a significant interaction between duration of exposure and measurement site [F(12,36) = 6.1, P < 10⁴], but only in the case of T_car and T_hyp, which started not different, rose at different rates, and ended with T_hyp significantly lower than T_car when T_car exceeded 39.5°C after 1300. 

View larger version (24K): [in this window] [in a new window]   Fig. 1.   Thirty-minute averages of carotid blood (Tcar), hypothalamic (Thyp), and rectal (Tre) temperatures of sheep exposed to 40, 5, and 22°C ambient temperatures. Ambient temperature was 22°C before 0900. Points show means ± SE, n = 4.

Breath frequency during heat exposure increased significantly at 1000 when all three body temperatures began to rise, and then again at 1030, when body temperatures had risen by ~0.4°C, wherein it remained stable until 1500 (Fig. 2).

View larger version (12K): [in this window] [in a new window]   Fig. 2.   Breath frequency of 4 sheep as ambient temperature increased from 22°C at 0900 to 40°C at 1000 and was then held at 40°C. Points show means ± SE. Means with different letters (a, b, and c) were significantly different from each other (P < 0.05).

During 5°C exposure (Fig. 1), temperatures also varied between measurement sites [F(2,6) = 15.6, P = 0.004] with T_re always being significantly greater than T_car and T_hyp, whereas the latter two temperatures were not significantly different at any time. The temperature at all three sites decreased with duration of the exposure [F(6,18) = 10.5, P < 10⁴]. There was no significant interaction [F(12,36) = 0.8].

During 22°C exposure (Fig. 1), there were no changes in temperatures over time [F(6,18) = 0.14, P = 0.98], and the difference between measurement sites approached significance [F(2,6) = 3.5, P = 0.09]. The interaction was not significant [F(12,36) = 0.88, P = 0.57]. Three of the animals maintained T_re on average 0.30°C greater than T_car, whereas in the other animal, the two temperatures were similar.

During 40°C exposure (Fig. 3), T_re  T_hyp was always significantly greater than T_car  T_hyp [F(1,3) = 40.0, P = 0.008]. Time had a significant statistical effect, with T_re  T_hyp and T_car  T_hyp increasing during the last 2 h of the exposure when T_car exceeded 39.5°C [F(6,18) = 7.9, P < 0.001]. The interaction was not significant [F(6,18) = 1.6, P = 0.19]. During 5°C exposure (Fig. 3), T_re  T_hyp was always significantly greater than T_car  T_hyp [F(1,3) = 553.7, P = 0.0001]. The effect of time was not significant [F(6,18) = 0.22, P = 0.96] nor was the interaction [F(6,18) = 1.5, P = 0.22]. During 22°C exposure (Fig. 3), the effect of time and the interaction were not significant [time F(6,18) = 0.9, P = 0.51; interaction F(6,18) = 0.86, P = 0.54]. Mean T_re  T_hyp was always greater than T_car  T_hyp, but the difference did not quite reach significance [F(1,3) = 7.5, P = 0.07].

View larger version (25K): [in this window] [in a new window]   Fig. 3.   Thirty-minute averages of Tre  Thyp and Tcar  Thyp of sheep exposed to 40, 5, and 22°C. Ambient temperature was 22°C before 0900. Points show means ± SE, n = 4.

Comparison of the trunk and brain temperature differences with zero revealed that during 40°C exposure, T_re  T_hyp was always greater than zero, except between 1000 and 1100 when all temperatures were rising rapidly, whereas T_car  T_hyp was greater than zero only for the final 2 h of the heat exposure. During 5°C exposure, T_re  T_hyp was greater than zero for the last 5 h of exposure, whereas at no time was T_car  T_hyp greater than zero. During exposure to 22°C, neither T_re  T_hyp nor T_car  T_hyp was significantly greater than zero at any time.

Figure 4 shows for one animal the average T_hyp at each 0.1°C category of T_car during 40, 22, and 5°C exposure. The intersection of the regression line fitted to the 40°C data and the line of identity (i.e., the threshold for SBC) occurred at 39.0°C. The threshold for SBC varied between animals, with the values for the other three animals being 38.7, 39.4 (this animal depicted in Fig. 5), and 39.7°C, but the shape of the relationship was the same in all four animals. The four animals entered the chamber with similar body temperatures (T_car = 38.6  38.8°C), but T_car at the end of the 6-h exposure to 40°C varied from 39.6 to 40.5°C. There was a tendency for the animals with the lowest SBC threshold to experience the larger increases in T_car during 40°C exposure, but there was no significant relationship between SBC threshold and final T_car [F(1,2) = 3.7, P = 0.19], change in T_car [F(1,2) = 1.29, P = 0.37], or body mass [F(1,2) = 5.1, P = 0.15] during 40°C exposure.

View larger version (17K): [in this window] [in a new window]   Fig. 4.   Thyp (means ± SD) at each 0.1°C category of Tcar during 40, 5, and 22°C exposure. Dashed line shows line of identity. Minimum and maximum Thyp at each Tcar category during 40°C exposure are indicated as continuous solid lines. Selective brain cooling (SBC) threshold for this animal was 39.0°C.

View larger version (33K): [in this window] [in a new window]   Fig. 5.   Thirty-second measurements of Tcar, Thyp, and Tre of 1 sheep for the final 4 h of 40°C exposure (top) and SBC calculated as Tcar  Thyp for the same time period (bottom). Arrows on the abscissa indicate times when an investigator entered the climate chamber. An investigator was present in the chamber for ~5 min.

The plots showing 30-min averages for each body temperature (Fig. 1) mask variability at a fine time scale, especially the variability in T_hyp reflected in the SD error bars and range of T_hyp in Fig. 4. As an example of the short-term variability in T_hyp, Fig. 5 shows the 30-s records of T_hyp, T_car, and T_re in one animal during the final 4 h of heat exposure (top) and the calculated SBC over this time (T_car  T_hyp; bottom). The figure shows regular oscillations in T_hyp over a 0.5°C range resulting in periods where the animal was using SBC of up to 0.3°C, interspersed with periods where SBC was abolished and T_hyp exceeded T_car by 0.2°C. The other three animals showed similar variability. The arrows on the abscissa indicate the times when an investigator entered the climate chamber to measure psychometric data and breath frequency, which took on average 5 min. In every instance, the entry of the investigator into the climate chamber led to a decrease in SBC and a consequent increase in T_hyp.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Our results are the first reports of carotid blood and brain temperature in sheep and show that this species can use SBC as part of the response to thermal stress. The occurrence and magnitude of SBC depend on body temperature but also on nonthermal factors such as the presence of an investigator. The study has highlighted the consequences of using T_re as a surrogate for arterial blood temperature, and we concur with the advice originally offered by Baker and Hayward (3) that because it is postrete arterial blood temperature that is altered by the SBC effector mechanism, T_car is the temperature that should be used to calculate SBC. As judged by the difference between T_hyp and T_car, the sheep selectively brain-cooled only during the final few hours of 40°C exposure and not at all during exposure to 22 or 5°C. If we had used T_re to estimate SBC, we would have erroneously concluded that the animals always were using SBC even after 6 h exposed to 5°C.

Previous reports have claimed that sheep used SBC of up to 0.8°C during heat exposure and 1.3°C during fever (29, 33, 35, 36), a much larger magnitude than we observed for T_car  T_hyp in this study. In all of these previous studies, SBC was estimated using T_re. Because T_re always exceeded T_car (under the conditions of our study), the use of T_re to estimate SBC will result in an overestimate of SBC. The magnitude of SBC we observed in sheep is consistent with that reported in other ungulates where T_car has been used to calculate SBC (11, 15, 21, 23, 27, 28, 34).

Not only will using T_re to estimate SBC lead to false conclusions regarding the magnitude of SBC, but it could also lead to the conclusion that the SBC effector mechanism was activated when it was not. Nijland et al. (35) concluded that because angularis oculi vein occlusion in heat-stressed sheep did not alter T_re  T_hyp, this vessel was not a major source of cool blood to the cavernous sinus. We later reported that in anesthetized sheep, blood flow in the angularis oculi vein accounted for over 80% of the cooling responsible for SBC (31). We can now explain the discrepancy. Figure 1 shows that after a 3-h exposure to 40°C, T_hyp and T_car were only beginning to diverge after becoming equal during the preceding hour. It was after 3 h of 40°C exposure that Nijland et al. (35) occluded the vein. Thus there was no SBC to be inhibited. But T_re was 0.4°C greater than T_hyp, leading to the false conclusion that SBC was present at the time of occlusion. A similar explanation presumably accounts for the conclusion reached by Mitchell et al. (33). With the use of T_re to estimate SBC in cold-exposed sheep, they concluded that sheep always selectively brain-cooled in the cold. Proper estimates using T_car show that SBC did not occur in our animals exposed to 5°C (Fig. 3). Indeed, SBC in the sheep has a threshold temperature in the upper normothermic range as is observed in other species.

Nijland et al. (36) showed that superior cervical sympathectomy in the sheep led to a decrease in T_hyp, which they attributed to an increase in SBC under conditions of heat exposure, cold exposure, and fever. Our present results raise doubts about their estimated magnitude of the SBC but do not affect the conclusion that sympathectomy led to a decrease in T_hyp. Sympathetic mechanisms could explain the phenomena illustrated in Fig. 5, a rapid increase in T_hyp independent of any change in T_car subsequent to an investigator entering the climate chamber. Sympathetic activity is an important nonthermal input to the SBC controller (4, 14, 21, 23, 34) with increased activity suppressing SBC even at high body temperatures. It appears that entry of an investigator caused activation of the sympathetic nervous system and suppression of SBC in our experimental animals. Sympathetically mediated attenuation of SBC may account for the observed increase in SBC threshold in some species during exercise (12, 28, 34). Two scenarios have been proposed to explain the effect of sympathetic stimulation on SBC: constriction of either the angularis oculi vessels or arteriovenous anastomotic shunts in the nasal mucosa (3, 24, 31).

An implication of suppression of SBC by the presence of an investigator is that the magnitude of SBC indicated by Fig. 4 underestimates the true nature of the thermally driven relationship between T_car and T_hyp in sheep. A similar implication may apply to other laboratory studies so that the "thresholds" for SBC we and others have calculated should be viewed with caution. The magnitude of SBC was periodically suppressed by the experimental protocol that influenced the mean T_hyp at each category of T_car. For example, the calculated SBC threshold for the animal depicted in Fig. 5 was 39.4°C, but the animal was selectively brain-cooling with a T_car of 39.25°C at the start of the trace.

The effect of sympathetic activity on SBC onset was proposed by Jessen et al. (23) to account for the absence of SBC in free-ranging black wildebeest during intense exercise. Such an effect may also underlie differences in SBC patterns between field and laboratory studies. The SBC threshold is ~38.8-38.9°C in tame goats in the laboratory (26, 27). However, when measurements were made on goats in outdoor enclosures, the threshold was less predictable and there was a range of T_car from 38.6 to 39.3°C over which the animals may or may not have been employing SBC (21, 22). Similarly, laboratory studies on reindeer reveal a precise threshold of 38.7°C (28), whereas free-ranging wildebeest and springbok have ranges for SBC onset between 38.7-39.3 and 39.0-39.3°C, respectively (23, 34). Presumably, this discrepancy reflects the additional stressors that free-ranging animals experience. As a corollary, determining the true nature of the thermally driven relationship between central blood and brain temperature requires elimination of the stress on experimental animals.

Because our experimental protocol evoked regular unplanned increases in brain temperature, the breath frequencies reported in Fig. 2 probably are overestimates. The rapid rise in T_hyp when the investigator entered the climate chamber would have enhanced heat-loss effectors including panting (26). So, in our study, as in any other study on artiodactyls, attenuation of SBC by experimenter presence would have resulted in an increase in panting rate compared with that which would have prevailed when an experimenter was not present.

If the relationship between T_re and T_car was predictable, then the state of the SBC effector could be deduced using measures of T_re. However, in our study, the relationship between the two temperatures was not consistent between treatments. The situation is complicated further when extended to other species. For example, T_re in the horse was similar to or higher than T_car under thermoneutral conditions but consistently lower than T_car during heat exposure and also during exercise (32). We conclude that surmising the state of the SBC effector without actually measuring arterial blood temperature is not possible, and studies that have used surrogate measures of arterial blood temperature are all likely to have reached erroneous conclusions regarding the occurrence and magnitude of SBC.

The process leading to the constant positive temperature difference between the rectum and arterial blood in sheep is not known. The difference could be due to either a heat source in the rectum or a heat sink in the carotid artery. Given the constancy of arterial blood temperature even at sites distant from the heart (18), a source of heat to the rectum is the likely explanation. Heat of microbial activity in the rectum may lead to an increase in rectal relative to other body temperatures (30). Although this hypothesis has been tested and discounted in humans (37), the protocol used can be criticized. The research compared the T_re of people before and after antibiotic treatment and showed no change in T_re. However, the relationship between rectal and other body sites was not measured. Another potential heat source to the rectum is venous blood draining from active leg muscles in the standing posture. To our knowledge, the influence of changes in external ileac vein temperature on T_re has not been addressed in any species.

The effect of a surrogate measure for arterial blood temperature is pertinent to the arguments concerning SBC in animals that do not possess a carotid rete. Most studies tendering support for SBC in nonrete mammals have used rectal, abdominal, or interscapular temperature as a surrogate for arterial blood (7-10, 17, 25, 38, 39). If a difference exists in these species between arterial blood and other trunk temperatures, as exists in sheep, then the SBC reported in these studies may be an artifact. The experimental problem of measuring arterial blood temperature in small animals presently precludes investigation of SBC in many nonrete species.

Only four studies have measured arterial blood and brain temperature concurrently in conscious nonrete mammals, two on equids, and one each on the rabbit and a primate. McConaghy et al. (32) found evidence of SBC in the horse during exercise and heat exposure in the laboratory, but Fuller et al. (13) found no evidence of SBC in free-ranging zebra, a close relative of the horse. Selective brain cooling in the horse either is a laboratory phenomenon or the two closely related equids exhibit a very different thermal physiology. In exercising rabbits, brain temperature was lower than T_re but not lower than aortic blood temperature (2). Hayward and Baker (18) measured carotid arterial blood and brain temperature in the rhesus monkey under thermoneutral and hot conditions and found no evidence of SBC. A later study compared rectal and brain temperature in squirrel monkeys and concluded the animals did use SBC in the heat, which was augmented by face fanning (16). That study also measured mixed venous and hypothalamic temperature in anesthetised squirrel monkeys exposed to 25°C, concluding that the animals were capable of SBC. It is not known what influence anesthesia had on these results, but T_hyp was more than a degree cooler than the calculated onset for SBC in conscious monkeys (Figs. 1 vs. 4 of Ref. 16). Thus the onset and control of SBC, if present, were distinctly different from those seen in conscious rete mammals. That study also showed that T_re was consistently higher than mixed venous blood temperature in monkeys. Given that this relationship is similar to that in sheep, the SBC reported in conscious monkeys may be an artifact.

In conclusion, we have shown that sheep in the laboratory exhibit SBC with characteristics very similar to that of laboratory goats. Previous studies describing SBC in sheep generated artifacts, because T_re was used as a surrogate for arterial blood temperature. Assessing the status of the SBC effector is not possible without a measure of prerete arterial blood temperature. We further show that investigator presence can influence the SBC effector mechanism in experimental animals and possibly lead to erroneous conclusions regarding the magnitude of SBC and even thermal effector responses employed during heat stress.