- Email: [email protected]
Factors affecting the circadian rhythm in plasma cortisol concentrations in the horse
DOMESTIC ANIMAL ENDOCRINOLOGY
Vol. 11 (2):227-238,1994
FACTORS AFFECTING THE CIRCADIAN RHYTHM IN PLASMA CORTISOL CONCENTRATIONS IN THE HORSE C.H.G. Irvine and S.L. Alexander Animal & Veterinary Sciences Group, Lincoln University, New Zealand ReceivedAugust 16, 1993 ABSTRACT In horses, a circadian rhythm in plasma cortisol concentrations has been reported in some but not all studies. When a rhythm occurred, horses were accustomed to a management routine, comprising stabling, feeding and sometimes exercise, which may entrain a circadian pattern. In this work, we monitored plasma cortisol by collecting jugular blood through indwelling cannulae from four groups: 1): 10 untrained, unperturbed maresgrazing excesspasture, bled hourly for 26 hr; 2) 4 mares housed in a barn for 48 hr before sampling every 15 rain for 20-24 hr; 3) 5 mares placed in an outdoor yard for sampling every 30 rain from 0930-2]00 hr; and 4) 4 stabled racehorsesin training, bled every 30 rain from 0730-2000 hr and once the following morning at 0830 hr. Plasma cortisol showed a similarly-timed circadian rhythm (P<0.0001) in all Group 1 horses,with a peak at 0000-0900 hr, and a nadir at 1800-2100 hr. By contrast, cortisol concentrations did not vary with time in either Group 2 or 3. Neither daily mean nor peak cortisol values differed in Group 1 and 2 (i.e. bled for > 20 hr); however nadir values were higher (P<0.05) in Group 2. In Group 4, cortisol declined (P=0.004) during the sampling period but had returned to initial concentrations the next morning. Values did not differ from those for Group 1, except between 1000 and 1300 hr when cortisol in Group 4 was lower (P<0.05). We conclude that a circadian cortisol rhythm exists in horses in the absence of any known cues imposed by humans. However, this rhythm can be obliterated by the minor perturbation of removing the horse from its accustomed environment. By contrast, the rhythm occurs in trained racehorses,
suggesting either that they have adapted to their environment thereby allowing an endogenous rhythm to emerge, or that the rhythm is entrained by their daily routine. These observations highlight the difficulties in determining the cortisol status of a horse, since measurements will be affected by time of day, the occurrence of short-term fluctuations, and how accustomed the horse is to its environment. INTRODUCTION A circadian rhythm in plasma glucocorticoid concentrations has been observed in many species, including humans (1) monkeys (2) and rats (3). In horses, some studies have found a circadian rhythm, with a peak between 0600 and 0900 hr and a trough between 1900 and 2300 hr (4-7). However, others have found that circadian changes do not (8) or only occasionally (9-11) occur in horses. In previous studies in which circadian rhythms were observed, horses were housed or yarded at universities, were presumably fed and cleaned to a regular routine, and in some cases, were exercised (7,9). These perturbations may have served to entrain or create a circadian rhythm. It is also possible that some horses found the experimental protocol to be stressful and the resultant cortisol secretion masked the normal daily pattern. This is suggested by the work of Hoffsis et al (9) in which a circadian cortisol rhythm was observed when horses were bled every 28 hr by venipuncture, but not when samples were collected more frequently through indwelling cannulae. The aim of the present study was to monitor plasma cortisol concentrations across 24 hr in untrained horses g a z i n g on excess pasture to determine whether a circadian rhythm exists in the absence of external cues imposed by humans. We then investigated the effect of housing or training on daily cortisol rhythms. We considered that the accuracy of clinical evaluations of adrenal status in horses could be improved by a better understanding of the Copyright© 1994Butterworth-Heinemann
227
0739-7240/94/$10.00
228
IRVlNE AND ALEXANDER
factors affecting plasma cortisol concentrations. Sampling or testing could then be carried out under more appropriate circumstances and interpreted more logically. M A T E R I A L S AND M E T H O D S Animals and Experimental Protocol. Experiment 1: Untrained horses in home paddock. During late summer in the morning, ten normal mares (nine Standardbreds, one Thoroughbred), aged 5-16 years, who had been grazing continually on excess pasture, were fitted with jugular cannulae (Angiocath; 16g, 5.25 inch; Deseret Co., Sandy, UT). Commencing either at 1600 h (n=6) or at 0800 h the next morning (n=4) and at hourly intervals thereafter for the next 26 hr 4-ml blood samples were collected. Horses were completely relaxed in their own two hectare paddock during this procedure and often remained in sternal recumbency while being bled. Night samples were collected with the aid of a small flashlight directed at the cannula and away from the mare's eyes. For five of the mares, a record was kept of behavior immediately before sampling. Levels of activity were graded as follows: 0=recumbent, l=standing quietly, 2=grazing, 3=active. Experiment 2: Housed, untrained horses with pituitary venous cannulae. During early autumn, four standardbred mares from the same group as those in Experiment 1 were brought into a large barn (40 m x 40 m), in which they could move freely with constant access to hay and water. On the following morning, pituitary venous and jugular cannulae were inserted. Placement of a pituitary venous cannula is a nonsurgical procedure which involves cannulation of a venous pathway unique to equids (12). Commencing at 1200 h the next day and at 15 min intervals for the next 20-24 hr 4-ml jugular blood samples were collected with minimal restraint and treated as above. Concurrently pituitary venous blood (3 ml each 5 min) was collected for another experiment. The photoperiod in the barn was ambient except that on the night of sampling it was dimly lit by three 150 watt bulbs approximately 2 m above the horses' heads. This lighting was similar to moonlight, and just allowed cannulae to be seen. Experiment 3: Untrained horses in a novel environment. During mid-winter, five mares from the same group used in the previous experiments were fitted with jugular cannulae. On the next day at 0900 hr, the mares were brought into a small (20 m x 50 m) outdoor yard and 4-ml blood samples were collected with minimal restraint every 30 min from 0930 hr to 2100 hr. Hay was available continually. Experiment 4: Housed, trained horses. Four Standardbred racehorses in training were used. The horses were castrated males, aged 3 to 6 years, and were performing well at approximately the same level of fitness. They underwent a regular daily routine which comprised overnight confinement in a stable, a large meal of oats at 0730 hr and vigorous, competitive exercise for 45-60 min by 1000 hr. After this, horses grazed and slept in paddocks until 1500 hr, then returned to their stable for a large meal at 1530 hr, followed by overnight confinement under ambient photoperiod. Jugular cannulae were inserted in the afternoon and the next day blood was collected every 30 rain from 0730hr to 2000hr. The horses followed their usual routine on this day except that they were not exercised. One blood sample was also collected at 0830 hr on the following day. The experiment was done in early winter. Experiment 5: Effect of excitement during the morning or evening on cortisol concentrations. A jugular blood sample was collected by venipuncture from 188 Standardbred racehorses of mixed sex, age and ability at one of four race meetings. On raceday, horses had been taken by truck from their home stables to the track for either a day (n=54) or night (n=134) meeting and were bled 90 min before racing. Regular drug testing was carried out at these tracks and it was unlikely that medication affecting adrenal hormones had been given.
DISTURBANCES IN CORTISOL RHYTHMS IN HORSES
229
Sample handling. Blood samples were placed into glass tubes containing 10 mg dried EDTA and kept in an ice bath until centrifuged at the end of the experiment. Plasma was then decanted and stored at -20 ° C until assayed. Assays. Cortisol was measured by enzyme immunoassay (13) in all jugular samples from Experiments 2-4 hr and in pituitary venous samples from one Expt. 2 mare in which sufficient plasma remained. The cortisol antibody (AbZ) was a gift from C.J. Munro (University of California, Davis) and has been described previously (14). The assay's detection limit was 1.7 nmol/L of plasma. Within and between assay coefficients of variation (CV) were 4.7% and 12.3%, respectively, as assessed by assay on each plate of quality control standards containing high, medium or low concentrations of cortisol (Lypocheck Control Sera, Biorad, Anaheim, CA). However, when these standards deviated from usual values by more than 20% the assay was repeated. For Experiments 1 and 5 cortisol was measured by radioimmunoassay using AbZ at a final dilution of 1:180,000 as described previously (15). Samples (20 ~tl) and standard amounts of cortisol in charcoal-adsorbed dexamethasone-suppressed equine jugular plasma were extracted in 1 ml reagent grade ethanol. The tracer was iodo-histamine-cortisol: Cortisol3-(O-carboxymethyl)-oxime (Sigma Chemical Co., St Louis, MO) was conjugated by the carbodiimide reaction to histamine which had been iodinated by the chloramine T method. The detection limit of the RIA was 2.5 nmol/L of plasma. Within and between assay CV's were 3.4% and 4.7%, respectively. The two assay systems gave very similar results as shown by linear regression of potency estimates (r= 0.94; [RIA] = 0.96 [EIA] + 3.8; n=50). Statistical analysis. Data from each horse were grouped into 4-hourly means. Analysis of variance (ANOVA) was used to assess the effect of time of day on cortisol concentrations taking into account the fact that repeated measurements had been made. The activity indices from Experiment 1 mares, which were ranked data, were analyzed using Friedman's non-parametric two-way analysis of variance ("Statistix" Analytical Software, St.Paul, MN). Data from Experiments 1 and 2 and from Experiments 1-4 were also combined and the effect on cortisol of sampling environment (e.g. paddock, barn or yard) and the interaction between environment and time over the periods covered by all experiments were tested by ANOVA. Pre-race data from Experiment 5 were assigned to the appropriate 4-hr period and the effect of time on cortisol concentration was determined by one-way ANOVA. When F was significant, means were compared by Fisher's lsd test (16). Mean activity indices were compared as described by Conover (17). In Experiment 2 in which samples were collected every 15 min, peaks in cortisol concentrations were detected using the Cluster programme (18). The programme was set to test 2x2 clusters, using the mean intra-assay CV%, and symmetrical t-statistics (3.34/3.34) that yielded a false positive rate of 2.5% with the test data supplied with the programme. We considered that the sampling frequency used in the other experiments was inadequate to define ultradian fluctuations. Data are given as means _+SE. RESULTS Experiment 1: Untrained horses in home paddock. Plasma cortisol concentrations showed a similarly-timed circadian rhythm in all 10 mares (e.g. Figure 1) and overall there was a significant effect of time on cortisol (F9,45=10.48; P<0.0001). Group mean peak levels occurred between 0600-0900 hr, while the nadir fell between 1800-2100 hr (Figure 2.a). The daily mean cortisol concentration was 163 -,- 14 nmol/L. For the group, the maximum percentage excursion (i.e. [(peak-nadir period)/nadir period]* 100) in cortisol concentrations was 129% -+ 35; however there was marked individual variation with values ranging from 30%
IRVINE AND ALEXANDER
230
to 380%. During nadir periods, cortisol levels appeared to be more stable than during peaks (Figure 1), although the hourly sampling frequency did not permit the definition of ultradian episodes. The activity of the five horses for which records were kept also varied with time of day (P=0.0003), with peak levels occurring between 1000-1300 hr, and the nadir between 22000100 (Figure2b). Experiment 2: Housed, untrained horses with pituitary venous cannulae. No circadian rhythm in cortisol concentrations was apparent either in individual mares (Figure 3) or in the group mean (F5,15 = 0.98; P=0.5; Figure 4). The daily mean cortisol concentration was 214 _+23 nmol/L, which was not different to that in pasture bled horses (F1,12 = 2.61; P=0.13). However there was interaction between environment and time (F5,60 = 2.48; P=0.04) with values spanning the circadian maximum (0200-1300 hr) being similar in the two groups, while trough values (1400-0100) were higher (P<0.05) in the housed, cannulated horses. Short-term fluctuations in cortisol concentrations were clearly defined (Figure 3) by the increased sampling frequency used in this experiment (every 15 min) and occurred with a mean frequency of 0.56 _+0.03 peaks per hr (range= 0.50-0.61). The frequency of peaks did not differ between day and night (t=0.54; NS). In the one mare in which cortisol was measured at 5-min intervals in pituitary blood, peak frequency was almost twice that determined from 15-min jugular measurements (1.09 vs. 0.59 peaks per hr; Figure 3a).
~
a
1
1
8
11
14
17
20 23 TIMEOF DAY
2
5
8
50
0
.
8
.
.
.
11
.
.
.
.
14
.
.
.
17
.
.
.
.
.
.
20 23 TIME OF DAY
.
.
.
2
.
.
.
5
.
.
.
8
Figure 1. Jugular cortisol concentrations in two representative untrained mares grazing excess pasture and bled hourly through indwelling cannulae.
DISTURBANCES IN CORTISOL RHYTHMS IN HORSES
(I-ll
b
1 0 " 1 3 14"17 18-21 11MEOF DAY
•
22-1
231
2"6
•
13
1
t.o. Ic t a
6-9
21
ccl
1 0 - 1 3 1 4 - 1 7 18'21 TIMEOF DAY
22-1
I" ° ~ ~
2-6
I
Figure 2. a) Mean (+ SE) cortisol concentrations during 4-hr periods in 10 untrained mares grazing excess pasture and bled hourly through indwelling cannulae. Means are different at the 0.05 level if they do not share a common letter, b) Mean (+ SE) cortisol concentrations and activity scores in 5 of these 10 mares whose behavior just before sampling was recorded. Levels of activity were scored as follows: 0=recumbent, lfstanding quietly, 2=grazing, 3=active. Letters indicate comparisons between times of either cortisol concentrations (lower case) or activity scores (upper case). Means are different at the 0.05 level if they do not share a common letter. a
,iw~--
•
|
'
--r
i
i-~
i
'
I i
z
Ti - - 1
i 12
b 3~,
16
~'~
---4
-I
%
24
,
I 1 1 i
;, "riMEOFDay
-
ti
I 16
Figure 3. Cortisol concentrations in two typical mares fitted with pituitary and jugular venous cannulae, and housed in a barn for 48 hr before sampling. The horizontal bars mark cortisol peaks detected by the Cluster program, a) The upper panel shows cortisol concentrations in samples collected every 5 rain (pituitary blood), while the lower panel shows concurrent jugular data collected every 15 rain from the same mare. b) Jugular data collected every 15 min from another mare.
232
IRVINE AND ALEXANDER
Experiment 3: Untrained horses in a novel environment; and Experiment 4: Housed, trained horses. There was no effect of time of day on cortisol (F2,8 = 0.05; NS) in the five horses bled between 0930 and 2100 hr in a novel environment (Figure 5). By contrast, cortisol in the four trained horses did change between 0730 and 2000 hr (F3,9 = 11.52; P<0.01), being highest in the early morning and declining steadily thereafter (Figures 6 & 8). However, the cortisol concentration in the single sample collected at 0830 hr the following day (224 _ 21.6 nmol/L) did not differ (paired t-test) from the first 0830 hr value (246 _+29.9 nmol/L). A comparison of cortisol values between 1000 and 2100 in Experiments 1-4 is shown in Figure 7. Environment affected cortisol concentrations during this period (F3,19 = 6.08; P=0.004), with mean levels being higher (P<0.05) in cannulated, housed mares than in trained horses or in untrained mares in their home paddock. Mean cortisol levels were also greater (P<0.05) in untrained horses in a novel environment than in trained horses.
Experiment 5: Effect of excitement during the morning or evening on cortisol concentrations. The cortisol concentration in pre-race blood samples was not affected by the time of collection (Figure 8; F3,184 = 1.35; NS).
00'
~
200"
8
100"
6-9
10-13
14-17 18-21 TIME OF DAY
22-1
2-5
Figure 4. Mean (+ SE) cortisol concentrations during 4-hr periods in 4 mares fitted with pituitary and jugular venous cannulae, and housed in a barn for 48 hr before sampling. Means did not vary with time.
DISTURBANCES IN CORTISOL RHYTHMS IN HORSES
233
400 a
i 9
10.5
12
13.5
15
' i2"
'13.s'
r iS'
16.5
TIMEOF DAY
18
19.5
' /a'
"~i.S'
21
2OO"
•
" "1~'
'1~'
' 21
"rIMEOF DAY
Figure 5. Jugular cortisol concentrations in two typical untrained mares brought from their paddock to an outside yard and bled there through indwelling cannulae. Both of these mares participated in another of the experiments described in this paper, a) Data from this mare are also shown in Figure 3a. b) Data from this mare are shown in Figure la. 2GO
5O
i
1'o
1"2
1"4
l"e
m
2o
TIIMEOF DAY 250
GO
~t
1'0
1'2 1'4 TtMEOF DAY
1'6
1'8
20
Figure 6. Jugular ¢ortisoi concentrations in two representative racehorses in training bled through indwelling cannulae. The star shows the cortisol concentration in a single sample collected the following day.
234
IRVINE AND ALEXANDER ;300" a
~
200a 17a
~
a
100-
10-13
I ~ GRASS ~
!
14-17 TIME OF DAY
BARN
~
i
YARD
18-21
~
TRAINED I
Figure 7. Summary of the mean cortisol concentrations during 4-hr periods in the 4 groups of horses described in other figures. Grass=Figures 1 & 2; Barn=Figures 3 & 4; Yard=Figure 5; Trained=Figures 6 & 8. Letters indicate comparisons between groups within time periods. Means are different at the 0.05 level if they do not share a common letter.
300
m !
200-
~100"
0.~
6-9
10-13 14-17 TIME OF DAY
18-21
Figure 8. Mean (÷ SE) cortisol concentrations during 4-hr periods in 4 racehorses in training sampled in their home stable (bars) and in racehorses sampled at the racetrack during the indicated period which was 90 min before racing (symbols). Each symbol represents the mean of all horses in a given race; the total number of racehorses bled was 188. The letter ' m ' shows the mean value for the pre-race samples collected during the last period. Bars that do not share common lower case letters are different at the 0.05 level.
DISTURBANCES IN CORTISOL RHYTHMS IN HORSES
235
DISCUSSION When horses were bled in their home paddock, plasma cortisol concentrations showed a distinct circadian rhythm, with a peak early in the morning and a trough in the late afternoon-early evening. This is similar to the pattern reported in those earlier studies in which a circadian variation in cortisol was observed in horses (4-7,9,10). The amplitude of the rhythm, with approximately a doubling of concentrations between nadir and peak, was also comparable to that observed by others (4,6,7). In our horses, this circadian rhythm was fragile and could be obliterated by placing them in a different environment, such as a barn (Experiment 2) or yard (Experiment 3). In these cases, cortisol concentrations during the circadian trough were raised, while peak concentrations were unaffected. This resulted in a slight, but not significant elevation in daily mean cortisol levels as assessed by comparing Exps 1 and 2 in which sampling duration exceeded 20 hr. Although mean cortisol concentrations did not differ between horses in the barn and those in the yard, levels during the time that the circadian trough (i.e. 1800-2100 hr) occurred in unperturbed Experiment 1 horses were higher in the housed horses than in any other group. It is possible that several days in an enclosed barn were more 'stressful' than 12 hr in an outside yard, or that the more intensive sampling regimen (every 5 min) used in Experiment 2 caused additional perturbation of the horses. This latter possibility seems less likely because the horses were quiet and well acquainted with the samplers and usually seemed to welcome being handled. A blunted circadian glucocorticoid rhythm with increased concentrations during the nadir has also been reported in rats undergoing chronic or repeated stressots (19) and in humans suffering from depression (20,21). It is noteworthy that in rats, small elevations in nadir glucocorticoid levels can have a profound impact on the body, being associated with increased adrenal weight, decreased thymus weight and general ill-thrift (22). Five and 15-min blood sampling clearly showed that cortisol secretion was episodic. In horses in a novel environment, mean peak frequency was 0.56 ± 0.03 peaks per hr, with no difference in frequency between day and night. Comparison of 5- and 15-min sampling in one horse suggested that the less intensive regimen resulted in underestimation of peak frequency. Similar observations have been made in humans for cortisol (21) and ACTH (23). Other workers have bled horses frequently enough to observe ultradian cortisol fluctuations: Evans et al (10; 20-rain sampling with windows of 5-rain sampling) reported highly variable episodes, whereas Toutain et al (7; 10-rain sampling) found 10 ± 1.4 pulses per 24 hr which yielded 17 secretory bursts upon deconvolution analysis (24). Toutain's horses retained their circadian cortisol rhythm during sampling, and ultradian episodes were absent during the trough from early afternoon to 2200 hr (7,24). Likewise, in our horses bled in their own paddock, cortisol concentrations appeared to be more stable during the circadian trough than the peak. We suggest that the low-level perturbation of a novel environment maintains ultradian secretory episodes during the period when a trough would occur in unstressed horses. Our data also indicate that the circadian minimum is not due to lack of responsiveness of the adrenal or corticotrope to signals from the brain, but rather suggest that the central drive to the axis is diminished. Similarly, in the rat, much indirect evidence points to a reduction in or cessation of the secretion of corticotropin releasing factors during the circadian trough (22). The highly artificial and structured environment of the racehorse in training did not disrupt the occurrence of a circadian cortisol rhythm. Likewise, all horses in previous studies in which circadian rhythms were reported were housed at universities and were fed, cleaned (4-6,10,11) and in some instances exercised (7,9) on a daily routine. It is possible that these horses had become totally accustomed to and relaxed in their environment, allowing an endogenous circadian rhythm to emerge. Alternatively, regular large meals and/or exercise may act as cues to entrain a cortisol rhythm. In our trained horses, cortisol maxima and times of
236
IRVINE AND ALEXANDER
morning feeding and peak activity approximately coincided. Exercise does raise cortisol concentrations in horses (4,13,25) and we have shown that in some horses the anticipation of exercise activates the adrenal axis (13). On the other hand, Toutain et al (7) claimed that cortisol concentrations were falling to minimum values during their horses' period of maximum activity, which was presumably the 2-3 hr daily exercise (the usual time and type of exercise were not described). A single large morning feed produces a spike increase in cortisol in horses (26) while in rats restricting feeding to limited periods in the dark or light phase shifts the circadian maximum to coincide with eating (27). However, in horses, circadian cortisol patterns are roughly similar whether they are given a single daily feed or six small meals at 4-hr intervals (26). Our horses on grass also showed a circadian rhythm in activity which was similar in timing to that imposed on the racehorses in training. Peak activity occurred at 1000-1300 hr and both activity and plasma cortisol concentrations declined together during the afternoon and evening. In the horses on grass, the cortisol nadir was reached before that in activity, so that the rising ann of the cortisol rhythm tended to precede that in activity by one 4-hr period. Likewise, in humans, cortisol concentrations increase several hours before waking (28). In humans, controlled experiments have shown that the sleep-wake cycle is a major cue for entraining the daily cortisol rhythm (28). This seems less likely to be the case in horses, since sleep, either slow-wave or paradoxical, occupies only 12% of the 24 hr cycle, although it does only occur during the night (29). The subjects of this study were three stabled stallions and the author suggested that the percentages of drowsiness and perhaps sleep observed in them could be far in excess of those they would display in the field (29). Another possibility is that the cortisol rhythm is entrained to the light-dark cycle as has been shown in the horse for other hormones, e.g. melatonin (30). Supporting this theory is the observation that artificially extending the photophase by 4 hr into the evening for 11 to 13 weeks shifts the mean time of the horses'cortisol nadir to 4 hr later; however, in that study, the effect was not statistically significant (6). Our four experiments, although performed under different ambient photoperiods, do not give insight into this question since sampling regimens were designed to study the effect of environment and not of photoperiod on circadian cortisol rhythms. Certainly, the presence or absence of a circadian rhythm cannot be due to the time of year, since Experiments 1 and 2 (rhythm and no rhythm, respectively) were performed just weeks apart in the late summer-early autumn, while Experiments 3 and 4 (no rhythm and rhythm, respectively) were done together in the winter. It is remotely possible that the lack of circadian rhythm in Experiment 2 horses resulted from the horses perceiving the dim night-lighting in the barn as an extended photophase, since prolonged exposure to night-lighting intensities as low as 107 lux (i.e. 10% of the intensity used in ref #6 without significant effect on cortisol rhythms) can stimulate reproductive cyclicity in anestrous mares (31). However, it is questionable whether a single day of altered photoperiod would be sufficient to alter cortisol secretion, since neither increasing the photophase from 8 to 16 hr nor initiating melatonin treatment affects the 15.00 hr cortisol concentration in horses for at least 5 d (32). In any case, an altered photoperiod cannot explain the lack of a circadian rhythm in Experiment 3 horses who were sampled in an outside yard under ambient photoperiod. Cortisol concentrations in pre-race samples were not affected by the time of collection, which varied from 0930 to 2030 hr. If it can be assumed that resting cortisol values in these horses would have followed a pattern similar to that observed in our four trained horses, then the increment during the evening trough must have far exceeded that during the morning. This is, in fact, what occurs when rats (33) or humans (34,35) are stressed during the circadian peak or trough: the glucocorticoid increment is greater during the trough so that similar maxima are attained. This occurrence has been attributed to the diurnal variation in resting glucocorticoid concentrations and the consequent fluctuation in the level of the negative
DISTURBANCES IN CORTISOL RHYTHMS IN HORSES
237
feedback signal (3,36). However, this cannot be the sole explanation, since the circadian variation in stress-induced ACTH secretion also occurs in adrenalectomized rats (33). Therefore, it has been proposed that stress induces greater release of corticotrophin releasing factors during the circadian trough (33). Mean cortisol concentrations were similar in fit and unperturbed unfit horses. Values were also comparable in each 4-hr time period, except between 1000 and 1300 hr when cortisol in fit horses was lower. This finding conflicts with clinical observations of elevated cortisol concentrations in trained racehorses (25,37). However, such observations must be accepted cautiously since the present data show that it is impossible to determine the cortisol status of any horse from a single daily sample which has been the common method of diagnosis. We conclude that horses, in the absence of any known cues imposed by humans, have a circadian rhythm in cortisol concentrations, with a peak early in the morning and a trough in the late afternoon-early evening. However, this rhythm can be disrupted by the minor perturbation of removing the horse from its accustomed environment. This results in elevated cortisol levels during the normal trough and a consequent damping of the daily rhythm. These observations may explain the conflicting reports on the existence of a circadian cortisol rhythm in horses (See Introduction). They also highlight the difficulties in determining the cortisol status of a horse, since measurements will be affected by the time of day, the occurrence of ultradian fluctuations, and how relaxed the horse is in the sampling environment.
ACKNOWLEDGEMENTS/FOOTNOTES We thank Julie Turner, Sally Marks, Kate Hines and Mark Taylor for assistance with field work, and Natalie Shand for performing conisol enzyme immunoassays. We are grateful to Derek Jones for giving us access to racehorses in training. This work was supported by grants from the New Zealand Health Research Council, New Zealand Equine Research Foundation, and the United States NIH (DK-38322). Corresponding author: Dr S.L. Alexander, Animal & Veterinary Sciences Group, PO Box 84, Lincoln University, Canterbury, NEW ZEALAND
REFERENCES 1. Weitzman ED, Fukushima D, Nogeire C, Roffwarg G, Gallagher TF, Hellman L. Twenty-four hour pattern
of the episodic secretion of cortisol in normal subjects. Endocrinology 33:14-22, 1971. 2. Smith CJ, Norman RE Influence of gonads on cortisol secretion in female rhesus macaques. Endocrinology 121:2192-2198, 1987. 3. Keller-Wood ME, Dallman MF. 1984 Corticosteroid inhibition of ACTH secretion. Endocr Rev 5:1-24, 1984. 4. James VHR, Horner MW, Moss MS, Rippon AE. Adrenocortical function in the horse. J Endocrinol 48:319-335, 1970. 5. Bottoms GD, Rocsel OF, Rausch FD, Akins EL. Circadian variation in plasma cortisol and corticosterone in pigs and mares. Am J Vet Res 33:785-790, 1972. 6. Johnson AL, Malinowski K. Daily rhythm of cortisol, and evidence for a photo-inducible phase for prolactin secretion in nonpregnant mares housed under non-interrupted and skeleton photoperiods. J Anim Sci 63:169-175, 1986. 7. Toutain PL, Oukessou M, Autefage A, Alvinerie M. Diurnal and episodic variations of plasma hydrocortisone concentrations in horses. Domest Anim Endocrinol 5:55-59, 1988. 8. Eiler H, Oliver J, Goble D. Adrenal gland function in the horse: effect of dexamethasone on hydrocortisone secretion and blood cellularity and plasma electrolyte concentrations. Am J Vet Res 40:727-729, 1979. 9. Hoffsis GF, Murdik PW, Tharp VL, Ault K. Plasma concentrations of cortisol and corticosterone in the normal horse. Am J Vet Res 31:1379-1387, 1979. 10. Evans JW, Winget CM, Pollak El. Rhythmic cortisol secretion in the equine: Analysis and physiological mechanisms. J lnterdiscipl Cycle Res 8:111-121, 1977. 11. Sojka JE, Johnson MA, Bottoms GD. The effect of starting time on dexamethasone suppression test results in horses. Domest Anita Endocrinol 10:1-5, 1993. 12. lrvine CHG, Alexander SL. A novel technique for measuring hypothalamic and pituitary hormone secretion rates from collection of pituitary venous blood in the normal horse. J Endocrinol 113;183-192, 1987.
238
IRVINE AND ALEXANDER
13. Alexander SL, Irvine CHG, Ellis MJ, Donald RA. The effect of acute exercise on the secretion of corticotropin-releasing factor, arginine vasopressin, and adrenocorticotropin as measured in pituitary venous blood from the horse. Endocrinology 128:65-72, 1991. 14. Munro C, Stabenfetdt G. Development of a cortisol enzyme immunoassay in plasma. Clin Chem 31:956 (Abstract 281), 1985. 15. Alexander SL, Irvine CHG, Livesey JH, Donald RA. The acute effect of lowering plasma cottisol on the secretion of corticotropin-releasing hormone (CRH), arginine vasopressin (AVP) and ACTH as revealed by intensive sampling of pituitary venous blood in the normal horse. Endocrinology 133:860-866, 1993. 16. Steel RGD, Torrie JH. Principles and Procedures of Statistics: A Biometrical Approach; 2nd edition. McGraw-Hill Kogakusha, Ltd, Tokyo, p 191, 1980. 17. Conover WJ. Practical Non-Parametric Statistics. John Wiley & Sons, New York, pp 299-305, 1980. 18. Veldhuis JD, Johnson ML. Cluster analysis: a simple, versatile and robust algorithm for endocrine pulse detection. Am J Physiol 250:E486-E493, 1986. 19. Akana SF, Scribner KA, Bradbury MJ, Strack AM, Walker C-D, Dallman MF. Feedback sensitivity of the rat hypothalamo-pituitary-adrenal axis and its capacity to adjust to exogenous corticosterone. Endocrinology 131:585-594, 1992. 20. Linkowski P, Mendlewicz J, Kerkhofs M, LeClercq R, Golstein J, Brasseur M, Copinschi G, Van Cauter E. 24-hour profiles of adrenocorticotropin, cortisol, and growth hormone in major depressive illness: effect of antidepressant treatment. J Clin Endocrinol Metab 65:141-152, 1987. 21. Sachar EJ, Hellman L, Roffwarg HP, Halpern FS, Fukushima DK, Gallagher TF. Disrupted twenty-four hour patterns of cortisol secretion in psychotic depression. Arch Gen Psychiat 28:19-24, 1973. 22. Scribner KA, Walker C-D, Cascio CS, Dallman MF. Chronic streptozotocin diabetes in rats facilitates the acute stress response without altering pituitary or adrenal responsiveness to secretagogues. Endocrinology 129:99-108, 1991. 23. Iranmanesh A, Lizarralde G, Short D, Veldhuis JD. Intensive venous sampling paradigms disclose high frequency adrenocorticotropin release episodes in normal men. J Clin Endocrinol Metab 71:1276-1283, 1990. 24. Toutain PL, Laurentie M, Autefage A, Alvinerie M. Hydrocortisone secretion: production rate and pulse characterization by numerical deconvolution. Am J Physiol 255:E688-E695, 1988. 25. Snow DH, MacKenzie G. Some metabolic effects of maximal exercise in the horse and adaptations with training. Equine Vet J 9:134-140, 1977. 26. Clarke LL, Ganjam VK, Fichtenbaum B, Hatfield D, Garner HE. Effect of feeding on renin-angiotensinaldosterone system of the horse. Am J Physiol 254:R524-R530, 1988. 27. Wilkinson CW, Shinsako J, Dallman MF. Daily rhythms in adrenal responsiveness to adrenocorticotropin are determined primarily by the time of feeding in the rat. Endocrinology 104:350-359, 1979. 28. Weitzman ED, Zimmerman JC, Czeisler CA, Ronda J. Cortisol secretion is inhibited during sleep in normal man. J Clin Endocrinol Metab 56:352-358, 1983. 29. Ruckebusch Y. The relevance of drowsiness in the circadian cycle of farm animals. Anim Behav 20:637643, 1972. 30. Guillaume D, Palmer E. Effect of oral melatonin on the date of the first ovulation after ovarian inactivity in mares under artificial photoperiod. J Reprod Fertil [Suppl] 44:249-257, 1991. 31. Sharp DC, Cleaver BD, Davis SD. Photoperiod. In: Equine Reproduction. Ed, AO McKinnon, JL Voss. Lea & Febiger, Philadelphia, pp 179-185, 1993. 32. Argo CM, Cox JE, Gray, JL. Effect of oral melatonin treatment on the seasonal physiology of pony stallions. J Reprod Fertil [Suppl] 44:115-125, 1991. 33. Bradbury MJ, Cascio CS, Scribner KA, Dallman MF. Stress-induced adrenocorticotropin secretion: diurnal responses and decreases during stress in the evening are not dependent on corticosterone. Endocrinology 128:680---688, 1991. 34. Nathan RS, Sachar EJ, Langer G, Tabriz MA, Halpern FS. Diurnal variation in the response of plasma prolactin, cortisol, and growth hormone to insulin-induced hypoglycemia in normal men. J Clin Endocrinol Metab 49:231-235, 1979. 35. Ichikawa Y, Nishikai M, Kawagoe M, Yoshida K, Homma M. Plasma corticotropin, cortisol and growth hormone responses to hypoglycemia in the morning and evening. J Clin Endocrinol Metab 34:895-898, 1972. 36. DeCherney GS, Debold CR, Jackson RV, Sheldon Jr WR, Island DP, Orth DN. Diurnal variation in the response of plasma adrenocorticotropin and cortisol to intravenous ovine corticotropin-releasing hormone. J Clin Endocrinol Metab 61:273-279, 1985. 37. Persson SGB, Larsson M, Lindholm A. Effects of training on adreno-cortical function and red-cell volume in trotters. Zbl Vet Med A 27:261-268, 1980.
Vol. 11 (2):227-238,1994
FACTORS AFFECTING THE CIRCADIAN RHYTHM IN PLASMA CORTISOL CONCENTRATIONS IN THE HORSE C.H.G. Irvine and S.L. Alexander Animal & Veterinary Sciences Group, Lincoln University, New Zealand ReceivedAugust 16, 1993 ABSTRACT In horses, a circadian rhythm in plasma cortisol concentrations has been reported in some but not all studies. When a rhythm occurred, horses were accustomed to a management routine, comprising stabling, feeding and sometimes exercise, which may entrain a circadian pattern. In this work, we monitored plasma cortisol by collecting jugular blood through indwelling cannulae from four groups: 1): 10 untrained, unperturbed maresgrazing excesspasture, bled hourly for 26 hr; 2) 4 mares housed in a barn for 48 hr before sampling every 15 rain for 20-24 hr; 3) 5 mares placed in an outdoor yard for sampling every 30 rain from 0930-2]00 hr; and 4) 4 stabled racehorsesin training, bled every 30 rain from 0730-2000 hr and once the following morning at 0830 hr. Plasma cortisol showed a similarly-timed circadian rhythm (P<0.0001) in all Group 1 horses,with a peak at 0000-0900 hr, and a nadir at 1800-2100 hr. By contrast, cortisol concentrations did not vary with time in either Group 2 or 3. Neither daily mean nor peak cortisol values differed in Group 1 and 2 (i.e. bled for > 20 hr); however nadir values were higher (P<0.05) in Group 2. In Group 4, cortisol declined (P=0.004) during the sampling period but had returned to initial concentrations the next morning. Values did not differ from those for Group 1, except between 1000 and 1300 hr when cortisol in Group 4 was lower (P<0.05). We conclude that a circadian cortisol rhythm exists in horses in the absence of any known cues imposed by humans. However, this rhythm can be obliterated by the minor perturbation of removing the horse from its accustomed environment. By contrast, the rhythm occurs in trained racehorses,
suggesting either that they have adapted to their environment thereby allowing an endogenous rhythm to emerge, or that the rhythm is entrained by their daily routine. These observations highlight the difficulties in determining the cortisol status of a horse, since measurements will be affected by time of day, the occurrence of short-term fluctuations, and how accustomed the horse is to its environment. INTRODUCTION A circadian rhythm in plasma glucocorticoid concentrations has been observed in many species, including humans (1) monkeys (2) and rats (3). In horses, some studies have found a circadian rhythm, with a peak between 0600 and 0900 hr and a trough between 1900 and 2300 hr (4-7). However, others have found that circadian changes do not (8) or only occasionally (9-11) occur in horses. In previous studies in which circadian rhythms were observed, horses were housed or yarded at universities, were presumably fed and cleaned to a regular routine, and in some cases, were exercised (7,9). These perturbations may have served to entrain or create a circadian rhythm. It is also possible that some horses found the experimental protocol to be stressful and the resultant cortisol secretion masked the normal daily pattern. This is suggested by the work of Hoffsis et al (9) in which a circadian cortisol rhythm was observed when horses were bled every 28 hr by venipuncture, but not when samples were collected more frequently through indwelling cannulae. The aim of the present study was to monitor plasma cortisol concentrations across 24 hr in untrained horses g a z i n g on excess pasture to determine whether a circadian rhythm exists in the absence of external cues imposed by humans. We then investigated the effect of housing or training on daily cortisol rhythms. We considered that the accuracy of clinical evaluations of adrenal status in horses could be improved by a better understanding of the Copyright© 1994Butterworth-Heinemann
227
0739-7240/94/$10.00
228
IRVlNE AND ALEXANDER
factors affecting plasma cortisol concentrations. Sampling or testing could then be carried out under more appropriate circumstances and interpreted more logically. M A T E R I A L S AND M E T H O D S Animals and Experimental Protocol. Experiment 1: Untrained horses in home paddock. During late summer in the morning, ten normal mares (nine Standardbreds, one Thoroughbred), aged 5-16 years, who had been grazing continually on excess pasture, were fitted with jugular cannulae (Angiocath; 16g, 5.25 inch; Deseret Co., Sandy, UT). Commencing either at 1600 h (n=6) or at 0800 h the next morning (n=4) and at hourly intervals thereafter for the next 26 hr 4-ml blood samples were collected. Horses were completely relaxed in their own two hectare paddock during this procedure and often remained in sternal recumbency while being bled. Night samples were collected with the aid of a small flashlight directed at the cannula and away from the mare's eyes. For five of the mares, a record was kept of behavior immediately before sampling. Levels of activity were graded as follows: 0=recumbent, l=standing quietly, 2=grazing, 3=active. Experiment 2: Housed, untrained horses with pituitary venous cannulae. During early autumn, four standardbred mares from the same group as those in Experiment 1 were brought into a large barn (40 m x 40 m), in which they could move freely with constant access to hay and water. On the following morning, pituitary venous and jugular cannulae were inserted. Placement of a pituitary venous cannula is a nonsurgical procedure which involves cannulation of a venous pathway unique to equids (12). Commencing at 1200 h the next day and at 15 min intervals for the next 20-24 hr 4-ml jugular blood samples were collected with minimal restraint and treated as above. Concurrently pituitary venous blood (3 ml each 5 min) was collected for another experiment. The photoperiod in the barn was ambient except that on the night of sampling it was dimly lit by three 150 watt bulbs approximately 2 m above the horses' heads. This lighting was similar to moonlight, and just allowed cannulae to be seen. Experiment 3: Untrained horses in a novel environment. During mid-winter, five mares from the same group used in the previous experiments were fitted with jugular cannulae. On the next day at 0900 hr, the mares were brought into a small (20 m x 50 m) outdoor yard and 4-ml blood samples were collected with minimal restraint every 30 min from 0930 hr to 2100 hr. Hay was available continually. Experiment 4: Housed, trained horses. Four Standardbred racehorses in training were used. The horses were castrated males, aged 3 to 6 years, and were performing well at approximately the same level of fitness. They underwent a regular daily routine which comprised overnight confinement in a stable, a large meal of oats at 0730 hr and vigorous, competitive exercise for 45-60 min by 1000 hr. After this, horses grazed and slept in paddocks until 1500 hr, then returned to their stable for a large meal at 1530 hr, followed by overnight confinement under ambient photoperiod. Jugular cannulae were inserted in the afternoon and the next day blood was collected every 30 rain from 0730hr to 2000hr. The horses followed their usual routine on this day except that they were not exercised. One blood sample was also collected at 0830 hr on the following day. The experiment was done in early winter. Experiment 5: Effect of excitement during the morning or evening on cortisol concentrations. A jugular blood sample was collected by venipuncture from 188 Standardbred racehorses of mixed sex, age and ability at one of four race meetings. On raceday, horses had been taken by truck from their home stables to the track for either a day (n=54) or night (n=134) meeting and were bled 90 min before racing. Regular drug testing was carried out at these tracks and it was unlikely that medication affecting adrenal hormones had been given.
DISTURBANCES IN CORTISOL RHYTHMS IN HORSES
229
Sample handling. Blood samples were placed into glass tubes containing 10 mg dried EDTA and kept in an ice bath until centrifuged at the end of the experiment. Plasma was then decanted and stored at -20 ° C until assayed. Assays. Cortisol was measured by enzyme immunoassay (13) in all jugular samples from Experiments 2-4 hr and in pituitary venous samples from one Expt. 2 mare in which sufficient plasma remained. The cortisol antibody (AbZ) was a gift from C.J. Munro (University of California, Davis) and has been described previously (14). The assay's detection limit was 1.7 nmol/L of plasma. Within and between assay coefficients of variation (CV) were 4.7% and 12.3%, respectively, as assessed by assay on each plate of quality control standards containing high, medium or low concentrations of cortisol (Lypocheck Control Sera, Biorad, Anaheim, CA). However, when these standards deviated from usual values by more than 20% the assay was repeated. For Experiments 1 and 5 cortisol was measured by radioimmunoassay using AbZ at a final dilution of 1:180,000 as described previously (15). Samples (20 ~tl) and standard amounts of cortisol in charcoal-adsorbed dexamethasone-suppressed equine jugular plasma were extracted in 1 ml reagent grade ethanol. The tracer was iodo-histamine-cortisol: Cortisol3-(O-carboxymethyl)-oxime (Sigma Chemical Co., St Louis, MO) was conjugated by the carbodiimide reaction to histamine which had been iodinated by the chloramine T method. The detection limit of the RIA was 2.5 nmol/L of plasma. Within and between assay CV's were 3.4% and 4.7%, respectively. The two assay systems gave very similar results as shown by linear regression of potency estimates (r= 0.94; [RIA] = 0.96 [EIA] + 3.8; n=50). Statistical analysis. Data from each horse were grouped into 4-hourly means. Analysis of variance (ANOVA) was used to assess the effect of time of day on cortisol concentrations taking into account the fact that repeated measurements had been made. The activity indices from Experiment 1 mares, which were ranked data, were analyzed using Friedman's non-parametric two-way analysis of variance ("Statistix" Analytical Software, St.Paul, MN). Data from Experiments 1 and 2 and from Experiments 1-4 were also combined and the effect on cortisol of sampling environment (e.g. paddock, barn or yard) and the interaction between environment and time over the periods covered by all experiments were tested by ANOVA. Pre-race data from Experiment 5 were assigned to the appropriate 4-hr period and the effect of time on cortisol concentration was determined by one-way ANOVA. When F was significant, means were compared by Fisher's lsd test (16). Mean activity indices were compared as described by Conover (17). In Experiment 2 in which samples were collected every 15 min, peaks in cortisol concentrations were detected using the Cluster programme (18). The programme was set to test 2x2 clusters, using the mean intra-assay CV%, and symmetrical t-statistics (3.34/3.34) that yielded a false positive rate of 2.5% with the test data supplied with the programme. We considered that the sampling frequency used in the other experiments was inadequate to define ultradian fluctuations. Data are given as means _+SE. RESULTS Experiment 1: Untrained horses in home paddock. Plasma cortisol concentrations showed a similarly-timed circadian rhythm in all 10 mares (e.g. Figure 1) and overall there was a significant effect of time on cortisol (F9,45=10.48; P<0.0001). Group mean peak levels occurred between 0600-0900 hr, while the nadir fell between 1800-2100 hr (Figure 2.a). The daily mean cortisol concentration was 163 -,- 14 nmol/L. For the group, the maximum percentage excursion (i.e. [(peak-nadir period)/nadir period]* 100) in cortisol concentrations was 129% -+ 35; however there was marked individual variation with values ranging from 30%
IRVINE AND ALEXANDER
230
to 380%. During nadir periods, cortisol levels appeared to be more stable than during peaks (Figure 1), although the hourly sampling frequency did not permit the definition of ultradian episodes. The activity of the five horses for which records were kept also varied with time of day (P=0.0003), with peak levels occurring between 1000-1300 hr, and the nadir between 22000100 (Figure2b). Experiment 2: Housed, untrained horses with pituitary venous cannulae. No circadian rhythm in cortisol concentrations was apparent either in individual mares (Figure 3) or in the group mean (F5,15 = 0.98; P=0.5; Figure 4). The daily mean cortisol concentration was 214 _+23 nmol/L, which was not different to that in pasture bled horses (F1,12 = 2.61; P=0.13). However there was interaction between environment and time (F5,60 = 2.48; P=0.04) with values spanning the circadian maximum (0200-1300 hr) being similar in the two groups, while trough values (1400-0100) were higher (P<0.05) in the housed, cannulated horses. Short-term fluctuations in cortisol concentrations were clearly defined (Figure 3) by the increased sampling frequency used in this experiment (every 15 min) and occurred with a mean frequency of 0.56 _+0.03 peaks per hr (range= 0.50-0.61). The frequency of peaks did not differ between day and night (t=0.54; NS). In the one mare in which cortisol was measured at 5-min intervals in pituitary blood, peak frequency was almost twice that determined from 15-min jugular measurements (1.09 vs. 0.59 peaks per hr; Figure 3a).
~
a
1
1
8
11
14
17
20 23 TIMEOF DAY
2
5
8
50
0
.
8
.
.
.
11
.
.
.
.
14
.
.
.
17
.
.
.
.
.
.
20 23 TIME OF DAY
.
.
.
2
.
.
.
5
.
.
.
8
Figure 1. Jugular cortisol concentrations in two representative untrained mares grazing excess pasture and bled hourly through indwelling cannulae.
DISTURBANCES IN CORTISOL RHYTHMS IN HORSES
(I-ll
b
1 0 " 1 3 14"17 18-21 11MEOF DAY
•
22-1
231
2"6
•
13
1
t.o. Ic t a
6-9
21
ccl
1 0 - 1 3 1 4 - 1 7 18'21 TIMEOF DAY
22-1
I" ° ~ ~
2-6
I
Figure 2. a) Mean (+ SE) cortisol concentrations during 4-hr periods in 10 untrained mares grazing excess pasture and bled hourly through indwelling cannulae. Means are different at the 0.05 level if they do not share a common letter, b) Mean (+ SE) cortisol concentrations and activity scores in 5 of these 10 mares whose behavior just before sampling was recorded. Levels of activity were scored as follows: 0=recumbent, lfstanding quietly, 2=grazing, 3=active. Letters indicate comparisons between times of either cortisol concentrations (lower case) or activity scores (upper case). Means are different at the 0.05 level if they do not share a common letter. a
,iw~--
•
|
'
--r
i
i-~
i
'
I i
z
Ti - - 1
i 12
b 3~,
16
~'~
---4
-I
%
24
,
I 1 1 i
;, "riMEOFDay
-
ti
I 16
Figure 3. Cortisol concentrations in two typical mares fitted with pituitary and jugular venous cannulae, and housed in a barn for 48 hr before sampling. The horizontal bars mark cortisol peaks detected by the Cluster program, a) The upper panel shows cortisol concentrations in samples collected every 5 rain (pituitary blood), while the lower panel shows concurrent jugular data collected every 15 rain from the same mare. b) Jugular data collected every 15 min from another mare.
232
IRVINE AND ALEXANDER
Experiment 3: Untrained horses in a novel environment; and Experiment 4: Housed, trained horses. There was no effect of time of day on cortisol (F2,8 = 0.05; NS) in the five horses bled between 0930 and 2100 hr in a novel environment (Figure 5). By contrast, cortisol in the four trained horses did change between 0730 and 2000 hr (F3,9 = 11.52; P<0.01), being highest in the early morning and declining steadily thereafter (Figures 6 & 8). However, the cortisol concentration in the single sample collected at 0830 hr the following day (224 _ 21.6 nmol/L) did not differ (paired t-test) from the first 0830 hr value (246 _+29.9 nmol/L). A comparison of cortisol values between 1000 and 2100 in Experiments 1-4 is shown in Figure 7. Environment affected cortisol concentrations during this period (F3,19 = 6.08; P=0.004), with mean levels being higher (P<0.05) in cannulated, housed mares than in trained horses or in untrained mares in their home paddock. Mean cortisol levels were also greater (P<0.05) in untrained horses in a novel environment than in trained horses.
Experiment 5: Effect of excitement during the morning or evening on cortisol concentrations. The cortisol concentration in pre-race blood samples was not affected by the time of collection (Figure 8; F3,184 = 1.35; NS).
00'
~
200"
8
100"
6-9
10-13
14-17 18-21 TIME OF DAY
22-1
2-5
Figure 4. Mean (+ SE) cortisol concentrations during 4-hr periods in 4 mares fitted with pituitary and jugular venous cannulae, and housed in a barn for 48 hr before sampling. Means did not vary with time.
DISTURBANCES IN CORTISOL RHYTHMS IN HORSES
233
400 a
i 9
10.5
12
13.5
15
' i2"
'13.s'
r iS'
16.5
TIMEOF DAY
18
19.5
' /a'
"~i.S'
21
2OO"
•
" "1~'
'1~'
' 21
"rIMEOF DAY
Figure 5. Jugular cortisol concentrations in two typical untrained mares brought from their paddock to an outside yard and bled there through indwelling cannulae. Both of these mares participated in another of the experiments described in this paper, a) Data from this mare are also shown in Figure 3a. b) Data from this mare are shown in Figure la. 2GO
5O
i
1'o
1"2
1"4
l"e
m
2o
TIIMEOF DAY 250
GO
~t
1'0
1'2 1'4 TtMEOF DAY
1'6
1'8
20
Figure 6. Jugular ¢ortisoi concentrations in two representative racehorses in training bled through indwelling cannulae. The star shows the cortisol concentration in a single sample collected the following day.
234
IRVINE AND ALEXANDER ;300" a
~
200a 17a
~
a
100-
10-13
I ~ GRASS ~
!
14-17 TIME OF DAY
BARN
~
i
YARD
18-21
~
TRAINED I
Figure 7. Summary of the mean cortisol concentrations during 4-hr periods in the 4 groups of horses described in other figures. Grass=Figures 1 & 2; Barn=Figures 3 & 4; Yard=Figure 5; Trained=Figures 6 & 8. Letters indicate comparisons between groups within time periods. Means are different at the 0.05 level if they do not share a common letter.
300
m !
200-
~100"
0.~
6-9
10-13 14-17 TIME OF DAY
18-21
Figure 8. Mean (÷ SE) cortisol concentrations during 4-hr periods in 4 racehorses in training sampled in their home stable (bars) and in racehorses sampled at the racetrack during the indicated period which was 90 min before racing (symbols). Each symbol represents the mean of all horses in a given race; the total number of racehorses bled was 188. The letter ' m ' shows the mean value for the pre-race samples collected during the last period. Bars that do not share common lower case letters are different at the 0.05 level.
DISTURBANCES IN CORTISOL RHYTHMS IN HORSES
235
DISCUSSION When horses were bled in their home paddock, plasma cortisol concentrations showed a distinct circadian rhythm, with a peak early in the morning and a trough in the late afternoon-early evening. This is similar to the pattern reported in those earlier studies in which a circadian variation in cortisol was observed in horses (4-7,9,10). The amplitude of the rhythm, with approximately a doubling of concentrations between nadir and peak, was also comparable to that observed by others (4,6,7). In our horses, this circadian rhythm was fragile and could be obliterated by placing them in a different environment, such as a barn (Experiment 2) or yard (Experiment 3). In these cases, cortisol concentrations during the circadian trough were raised, while peak concentrations were unaffected. This resulted in a slight, but not significant elevation in daily mean cortisol levels as assessed by comparing Exps 1 and 2 in which sampling duration exceeded 20 hr. Although mean cortisol concentrations did not differ between horses in the barn and those in the yard, levels during the time that the circadian trough (i.e. 1800-2100 hr) occurred in unperturbed Experiment 1 horses were higher in the housed horses than in any other group. It is possible that several days in an enclosed barn were more 'stressful' than 12 hr in an outside yard, or that the more intensive sampling regimen (every 5 min) used in Experiment 2 caused additional perturbation of the horses. This latter possibility seems less likely because the horses were quiet and well acquainted with the samplers and usually seemed to welcome being handled. A blunted circadian glucocorticoid rhythm with increased concentrations during the nadir has also been reported in rats undergoing chronic or repeated stressots (19) and in humans suffering from depression (20,21). It is noteworthy that in rats, small elevations in nadir glucocorticoid levels can have a profound impact on the body, being associated with increased adrenal weight, decreased thymus weight and general ill-thrift (22). Five and 15-min blood sampling clearly showed that cortisol secretion was episodic. In horses in a novel environment, mean peak frequency was 0.56 ± 0.03 peaks per hr, with no difference in frequency between day and night. Comparison of 5- and 15-min sampling in one horse suggested that the less intensive regimen resulted in underestimation of peak frequency. Similar observations have been made in humans for cortisol (21) and ACTH (23). Other workers have bled horses frequently enough to observe ultradian cortisol fluctuations: Evans et al (10; 20-rain sampling with windows of 5-rain sampling) reported highly variable episodes, whereas Toutain et al (7; 10-rain sampling) found 10 ± 1.4 pulses per 24 hr which yielded 17 secretory bursts upon deconvolution analysis (24). Toutain's horses retained their circadian cortisol rhythm during sampling, and ultradian episodes were absent during the trough from early afternoon to 2200 hr (7,24). Likewise, in our horses bled in their own paddock, cortisol concentrations appeared to be more stable during the circadian trough than the peak. We suggest that the low-level perturbation of a novel environment maintains ultradian secretory episodes during the period when a trough would occur in unstressed horses. Our data also indicate that the circadian minimum is not due to lack of responsiveness of the adrenal or corticotrope to signals from the brain, but rather suggest that the central drive to the axis is diminished. Similarly, in the rat, much indirect evidence points to a reduction in or cessation of the secretion of corticotropin releasing factors during the circadian trough (22). The highly artificial and structured environment of the racehorse in training did not disrupt the occurrence of a circadian cortisol rhythm. Likewise, all horses in previous studies in which circadian rhythms were reported were housed at universities and were fed, cleaned (4-6,10,11) and in some instances exercised (7,9) on a daily routine. It is possible that these horses had become totally accustomed to and relaxed in their environment, allowing an endogenous circadian rhythm to emerge. Alternatively, regular large meals and/or exercise may act as cues to entrain a cortisol rhythm. In our trained horses, cortisol maxima and times of
236
IRVINE AND ALEXANDER
morning feeding and peak activity approximately coincided. Exercise does raise cortisol concentrations in horses (4,13,25) and we have shown that in some horses the anticipation of exercise activates the adrenal axis (13). On the other hand, Toutain et al (7) claimed that cortisol concentrations were falling to minimum values during their horses' period of maximum activity, which was presumably the 2-3 hr daily exercise (the usual time and type of exercise were not described). A single large morning feed produces a spike increase in cortisol in horses (26) while in rats restricting feeding to limited periods in the dark or light phase shifts the circadian maximum to coincide with eating (27). However, in horses, circadian cortisol patterns are roughly similar whether they are given a single daily feed or six small meals at 4-hr intervals (26). Our horses on grass also showed a circadian rhythm in activity which was similar in timing to that imposed on the racehorses in training. Peak activity occurred at 1000-1300 hr and both activity and plasma cortisol concentrations declined together during the afternoon and evening. In the horses on grass, the cortisol nadir was reached before that in activity, so that the rising ann of the cortisol rhythm tended to precede that in activity by one 4-hr period. Likewise, in humans, cortisol concentrations increase several hours before waking (28). In humans, controlled experiments have shown that the sleep-wake cycle is a major cue for entraining the daily cortisol rhythm (28). This seems less likely to be the case in horses, since sleep, either slow-wave or paradoxical, occupies only 12% of the 24 hr cycle, although it does only occur during the night (29). The subjects of this study were three stabled stallions and the author suggested that the percentages of drowsiness and perhaps sleep observed in them could be far in excess of those they would display in the field (29). Another possibility is that the cortisol rhythm is entrained to the light-dark cycle as has been shown in the horse for other hormones, e.g. melatonin (30). Supporting this theory is the observation that artificially extending the photophase by 4 hr into the evening for 11 to 13 weeks shifts the mean time of the horses'cortisol nadir to 4 hr later; however, in that study, the effect was not statistically significant (6). Our four experiments, although performed under different ambient photoperiods, do not give insight into this question since sampling regimens were designed to study the effect of environment and not of photoperiod on circadian cortisol rhythms. Certainly, the presence or absence of a circadian rhythm cannot be due to the time of year, since Experiments 1 and 2 (rhythm and no rhythm, respectively) were performed just weeks apart in the late summer-early autumn, while Experiments 3 and 4 (no rhythm and rhythm, respectively) were done together in the winter. It is remotely possible that the lack of circadian rhythm in Experiment 2 horses resulted from the horses perceiving the dim night-lighting in the barn as an extended photophase, since prolonged exposure to night-lighting intensities as low as 107 lux (i.e. 10% of the intensity used in ref #6 without significant effect on cortisol rhythms) can stimulate reproductive cyclicity in anestrous mares (31). However, it is questionable whether a single day of altered photoperiod would be sufficient to alter cortisol secretion, since neither increasing the photophase from 8 to 16 hr nor initiating melatonin treatment affects the 15.00 hr cortisol concentration in horses for at least 5 d (32). In any case, an altered photoperiod cannot explain the lack of a circadian rhythm in Experiment 3 horses who were sampled in an outside yard under ambient photoperiod. Cortisol concentrations in pre-race samples were not affected by the time of collection, which varied from 0930 to 2030 hr. If it can be assumed that resting cortisol values in these horses would have followed a pattern similar to that observed in our four trained horses, then the increment during the evening trough must have far exceeded that during the morning. This is, in fact, what occurs when rats (33) or humans (34,35) are stressed during the circadian peak or trough: the glucocorticoid increment is greater during the trough so that similar maxima are attained. This occurrence has been attributed to the diurnal variation in resting glucocorticoid concentrations and the consequent fluctuation in the level of the negative
DISTURBANCES IN CORTISOL RHYTHMS IN HORSES
237
feedback signal (3,36). However, this cannot be the sole explanation, since the circadian variation in stress-induced ACTH secretion also occurs in adrenalectomized rats (33). Therefore, it has been proposed that stress induces greater release of corticotrophin releasing factors during the circadian trough (33). Mean cortisol concentrations were similar in fit and unperturbed unfit horses. Values were also comparable in each 4-hr time period, except between 1000 and 1300 hr when cortisol in fit horses was lower. This finding conflicts with clinical observations of elevated cortisol concentrations in trained racehorses (25,37). However, such observations must be accepted cautiously since the present data show that it is impossible to determine the cortisol status of any horse from a single daily sample which has been the common method of diagnosis. We conclude that horses, in the absence of any known cues imposed by humans, have a circadian rhythm in cortisol concentrations, with a peak early in the morning and a trough in the late afternoon-early evening. However, this rhythm can be disrupted by the minor perturbation of removing the horse from its accustomed environment. This results in elevated cortisol levels during the normal trough and a consequent damping of the daily rhythm. These observations may explain the conflicting reports on the existence of a circadian cortisol rhythm in horses (See Introduction). They also highlight the difficulties in determining the cortisol status of a horse, since measurements will be affected by the time of day, the occurrence of ultradian fluctuations, and how relaxed the horse is in the sampling environment.
ACKNOWLEDGEMENTS/FOOTNOTES We thank Julie Turner, Sally Marks, Kate Hines and Mark Taylor for assistance with field work, and Natalie Shand for performing conisol enzyme immunoassays. We are grateful to Derek Jones for giving us access to racehorses in training. This work was supported by grants from the New Zealand Health Research Council, New Zealand Equine Research Foundation, and the United States NIH (DK-38322). Corresponding author: Dr S.L. Alexander, Animal & Veterinary Sciences Group, PO Box 84, Lincoln University, Canterbury, NEW ZEALAND
REFERENCES 1. Weitzman ED, Fukushima D, Nogeire C, Roffwarg G, Gallagher TF, Hellman L. Twenty-four hour pattern
of the episodic secretion of cortisol in normal subjects. Endocrinology 33:14-22, 1971. 2. Smith CJ, Norman RE Influence of gonads on cortisol secretion in female rhesus macaques. Endocrinology 121:2192-2198, 1987. 3. Keller-Wood ME, Dallman MF. 1984 Corticosteroid inhibition of ACTH secretion. Endocr Rev 5:1-24, 1984. 4. James VHR, Horner MW, Moss MS, Rippon AE. Adrenocortical function in the horse. J Endocrinol 48:319-335, 1970. 5. Bottoms GD, Rocsel OF, Rausch FD, Akins EL. Circadian variation in plasma cortisol and corticosterone in pigs and mares. Am J Vet Res 33:785-790, 1972. 6. Johnson AL, Malinowski K. Daily rhythm of cortisol, and evidence for a photo-inducible phase for prolactin secretion in nonpregnant mares housed under non-interrupted and skeleton photoperiods. J Anim Sci 63:169-175, 1986. 7. Toutain PL, Oukessou M, Autefage A, Alvinerie M. Diurnal and episodic variations of plasma hydrocortisone concentrations in horses. Domest Anim Endocrinol 5:55-59, 1988. 8. Eiler H, Oliver J, Goble D. Adrenal gland function in the horse: effect of dexamethasone on hydrocortisone secretion and blood cellularity and plasma electrolyte concentrations. Am J Vet Res 40:727-729, 1979. 9. Hoffsis GF, Murdik PW, Tharp VL, Ault K. Plasma concentrations of cortisol and corticosterone in the normal horse. Am J Vet Res 31:1379-1387, 1979. 10. Evans JW, Winget CM, Pollak El. Rhythmic cortisol secretion in the equine: Analysis and physiological mechanisms. J lnterdiscipl Cycle Res 8:111-121, 1977. 11. Sojka JE, Johnson MA, Bottoms GD. The effect of starting time on dexamethasone suppression test results in horses. Domest Anita Endocrinol 10:1-5, 1993. 12. lrvine CHG, Alexander SL. A novel technique for measuring hypothalamic and pituitary hormone secretion rates from collection of pituitary venous blood in the normal horse. J Endocrinol 113;183-192, 1987.
238
IRVINE AND ALEXANDER
13. Alexander SL, Irvine CHG, Ellis MJ, Donald RA. The effect of acute exercise on the secretion of corticotropin-releasing factor, arginine vasopressin, and adrenocorticotropin as measured in pituitary venous blood from the horse. Endocrinology 128:65-72, 1991. 14. Munro C, Stabenfetdt G. Development of a cortisol enzyme immunoassay in plasma. Clin Chem 31:956 (Abstract 281), 1985. 15. Alexander SL, Irvine CHG, Livesey JH, Donald RA. The acute effect of lowering plasma cottisol on the secretion of corticotropin-releasing hormone (CRH), arginine vasopressin (AVP) and ACTH as revealed by intensive sampling of pituitary venous blood in the normal horse. Endocrinology 133:860-866, 1993. 16. Steel RGD, Torrie JH. Principles and Procedures of Statistics: A Biometrical Approach; 2nd edition. McGraw-Hill Kogakusha, Ltd, Tokyo, p 191, 1980. 17. Conover WJ. Practical Non-Parametric Statistics. John Wiley & Sons, New York, pp 299-305, 1980. 18. Veldhuis JD, Johnson ML. Cluster analysis: a simple, versatile and robust algorithm for endocrine pulse detection. Am J Physiol 250:E486-E493, 1986. 19. Akana SF, Scribner KA, Bradbury MJ, Strack AM, Walker C-D, Dallman MF. Feedback sensitivity of the rat hypothalamo-pituitary-adrenal axis and its capacity to adjust to exogenous corticosterone. Endocrinology 131:585-594, 1992. 20. Linkowski P, Mendlewicz J, Kerkhofs M, LeClercq R, Golstein J, Brasseur M, Copinschi G, Van Cauter E. 24-hour profiles of adrenocorticotropin, cortisol, and growth hormone in major depressive illness: effect of antidepressant treatment. J Clin Endocrinol Metab 65:141-152, 1987. 21. Sachar EJ, Hellman L, Roffwarg HP, Halpern FS, Fukushima DK, Gallagher TF. Disrupted twenty-four hour patterns of cortisol secretion in psychotic depression. Arch Gen Psychiat 28:19-24, 1973. 22. Scribner KA, Walker C-D, Cascio CS, Dallman MF. Chronic streptozotocin diabetes in rats facilitates the acute stress response without altering pituitary or adrenal responsiveness to secretagogues. Endocrinology 129:99-108, 1991. 23. Iranmanesh A, Lizarralde G, Short D, Veldhuis JD. Intensive venous sampling paradigms disclose high frequency adrenocorticotropin release episodes in normal men. J Clin Endocrinol Metab 71:1276-1283, 1990. 24. Toutain PL, Laurentie M, Autefage A, Alvinerie M. Hydrocortisone secretion: production rate and pulse characterization by numerical deconvolution. Am J Physiol 255:E688-E695, 1988. 25. Snow DH, MacKenzie G. Some metabolic effects of maximal exercise in the horse and adaptations with training. Equine Vet J 9:134-140, 1977. 26. Clarke LL, Ganjam VK, Fichtenbaum B, Hatfield D, Garner HE. Effect of feeding on renin-angiotensinaldosterone system of the horse. Am J Physiol 254:R524-R530, 1988. 27. Wilkinson CW, Shinsako J, Dallman MF. Daily rhythms in adrenal responsiveness to adrenocorticotropin are determined primarily by the time of feeding in the rat. Endocrinology 104:350-359, 1979. 28. Weitzman ED, Zimmerman JC, Czeisler CA, Ronda J. Cortisol secretion is inhibited during sleep in normal man. J Clin Endocrinol Metab 56:352-358, 1983. 29. Ruckebusch Y. The relevance of drowsiness in the circadian cycle of farm animals. Anim Behav 20:637643, 1972. 30. Guillaume D, Palmer E. Effect of oral melatonin on the date of the first ovulation after ovarian inactivity in mares under artificial photoperiod. J Reprod Fertil [Suppl] 44:249-257, 1991. 31. Sharp DC, Cleaver BD, Davis SD. Photoperiod. In: Equine Reproduction. Ed, AO McKinnon, JL Voss. Lea & Febiger, Philadelphia, pp 179-185, 1993. 32. Argo CM, Cox JE, Gray, JL. Effect of oral melatonin treatment on the seasonal physiology of pony stallions. J Reprod Fertil [Suppl] 44:115-125, 1991. 33. Bradbury MJ, Cascio CS, Scribner KA, Dallman MF. Stress-induced adrenocorticotropin secretion: diurnal responses and decreases during stress in the evening are not dependent on corticosterone. Endocrinology 128:680---688, 1991. 34. Nathan RS, Sachar EJ, Langer G, Tabriz MA, Halpern FS. Diurnal variation in the response of plasma prolactin, cortisol, and growth hormone to insulin-induced hypoglycemia in normal men. J Clin Endocrinol Metab 49:231-235, 1979. 35. Ichikawa Y, Nishikai M, Kawagoe M, Yoshida K, Homma M. Plasma corticotropin, cortisol and growth hormone responses to hypoglycemia in the morning and evening. J Clin Endocrinol Metab 34:895-898, 1972. 36. DeCherney GS, Debold CR, Jackson RV, Sheldon Jr WR, Island DP, Orth DN. Diurnal variation in the response of plasma adrenocorticotropin and cortisol to intravenous ovine corticotropin-releasing hormone. J Clin Endocrinol Metab 61:273-279, 1985. 37. Persson SGB, Larsson M, Lindholm A. Effects of training on adreno-cortical function and red-cell volume in trotters. Zbl Vet Med A 27:261-268, 1980.