- Email: [email protected]
Effect of long-term feed restriction on seasonal endocrine changes in Soay sheep
Physiology & Behavior 71 (2000) 343 ± 351
Effect of long-term feed restriction on seasonal endocrine changes in Soay sheep S.M. Rhinda,*, S.R. McMillena, E. Duff b, C.E. Kylea, S. Wrighta a
b
Macaulay Land Use Research Institute, Craigiebuckler, Aberdeen AB15 8QH, UK Biomathematics and Statistics Scotland, Macaulay Land Use Research Institute, Craigiebuckler, Aberdeen AB15 8QH, UK Received 26 April 2000; accepted 7 July 2000
Abstract Groups of 15 adult, castrated, male Soay sheep were housed under natural daylength conditions at 57°N and fed a complete diet ad libitum (AL) or at a restricted rate (R) of 35 g dry matter (DM)/kg0.75 initial liveweight per day. The diet was based on barley and dried grass pellets and contained an estimated 11.6 MJ of metabolisable energy, 83% DM and 140 g crude protein/kg DM. In the AL animals, higher levels of feed intake during the periods of long daylength were associated with shorter intermeal intervals ( p < 0.001), a greater meal frequency ( p < 0.001), and a greater proportion of time spent eating ( p < 0.001) together with a greater rate of feed ingestion ( p < 0.001) and an increased meal size ( p < 0.001). Mean plasma concentrations of insulin, prolactin, insulin-like growth factor 1 (IGF-1), triiodothyronine (T3), and thyroxine (T4) were higher ( p < 0.001) in the spring or summer than in the autumn. Mean plasma GH concentrations did not differ with month. Compared with R animals, AL animals had higher mean plasma concentrations of insulin ( p < 0.001), prolactin ( p < 0.01), T3 ( p < 0.01), and T4 ( p < 0.01). Plasma GH and IGF-1 concentrations did not differ significantly with treatment. There was a greater increase in plasma insulin concentrations following feeding in R than AL animals ( p < 0.001) owing to higher pre-feeding concentrations in AL animals and the ingestion of larger amounts of feed by R than AL animals in the period after fresh feed was introduced. There were significant differences between months in this response, in R animals ( p < 0.01). Mean CSF insulin concentrations were significantly higher in AL than R animals ( p < 0.05) but were not affected by month. Neither was there a difference between pre-feeding concentrations and concentrations at approximately 12 h after feeding. It is concluded that the differences in the response of plasma insulin concentrations to feeding at different times of year, which were detected in R animals, were attributable, primarily, to differences in the vagally-induced insulin response to feeding and that these differences may provide important feedback signals to the appetite centre. D 2000 Elsevier Science Inc. All rights reserved. Keywords: Season; Food intake; Hormones; Sheep
1. Introduction Seasonal variation in voluntary food intake (VFI) in ruminants is probably controlled by both neural and endocrine signals [7]. These signals can be modified by both photoperiodic and nutritional influences Ð effects which are frequently confounded Ð which makes it difficult to assess the roles of hormones in the seasonal variation in VFI. The separate effects on hormone profiles of photoperiod and of seasonal changes in level of VFI by red deer have been described previously [16]. However, in that study, no clear links were found
* Corresponding author. Tel.: +44-1224-318611; fax: +44-1224311556. E-mail address: [email protected] (S.M. Rhind).
between seasonal changes in VFI and hormone profiles. The absence of clear links suggests that even if endocrine signals do control VFI in ruminants, they cannot be adequately described by single samples collected at intervals of several days. It is postulated that changes with season may also occur in the postprandial responses of metabolic hormones and that differences in these profiles may provide a differential feedback signal to the appetite centre of the brain. Previous studies of sheep [14,15] have shown that differences in such responses occur with changes in physiological state. Insulin is considered to be a likely signal since it exhibits marked postprandial changes in circulating concentrations [18], it is known to be transported across the blood ±brain barrier [17], and has been implicated previously in appetite regulation [17].
0031-9384/00/$ ± see front matter D 2000 Elsevier Science Inc. All rights reserved. PII: S 0 0 3 1 - 9 3 8 4 ( 0 0 ) 0 0 3 4 3 - 7
344
S.M. Rhind et al. / Physiology & Behavior 71 (2000) 343±351
The first aim of the present experiment was to determine whether or not variations in postprandial insulin profiles occur which could have a role in the control of seasonal variation in components of shortterm patterns of feed ingestion (rate of ingestion, meal size, etc.). The second objective of the present study was to determine whether or not there were seasonal differences in the insulin response to a glucose challenge. This measurement served two purposes. Firstly, it was designed to show whether or not the plasma insulin response to a standard challenge with an energy-providing substrate was modified by photoperiod per se. Secondly, it was designed to separate the later, nutrient-driven response [18] from the initial, vagally driven insulin response to feeding. The supply of many protein hormones to the brain is generally limited by the blood ± brain barrier [5], but since there is evidence of active transfer of insulin into the brain, this hormone could have a role in signalling the metabolic state of the animal and in the regulation of appetite [17]. The final objective of the study was to determine concentrations of insulin in the cerebrospinal fluid (CSF) at selected seasons and times of day in order to assess the potential significance of this signal to the brain in the control of patterns of food intake. 2. Materials and methods 2.1. Animals and treatments Two groups of 15 adult, castrated, male Soay sheep which had previously been maintained on pasture were housed in individual pens under natural daylength conditions at 57°N throughout a 15-month study beginning in January. All animals were fed a complete diet, based on barley and dried grass pellets, which contained an estimated 11.6 MJ of metabolisable energy, 83% dry matter (DM) and 140 g crude protein/kg DM. In order to separate the effects on hormone profiles of photoperiod and level of feed intake, animals of one group were fed ad libitum (AL) to a refusal rate of 0.2 (refusals were recorded daily and mean daily food intakes calculated) while animals of the second group were fed once daily between 0800 and 1100 h, at a restricted rate (R) of 35 g DM/kg0.75 initial liveweight, throughout the study. This was equivalent to the level of intake recorded previously in Soay sheep during the winter [10]. This level of intake was designed to ensure that the intake of animals was constant throughout the study while meeting the requirements for the maintenance of liveweight. Liveweights were recorded every 2 weeks during the study.
2.2. Measurement of meal patterns At intervals of approximately 1 month, the meal patterns of animals fed AL were determined by automated recording, to the nearest 5 g, of the fresh weight of feed present in their feed bins at 5-min intervals for 2 days. Using these records, mean meal durations and frequencies and rates of feed ingestion were determined for AL animals for each month. An inter-meal interval was defined as the time during which there was no reduction in the weight of feed in the bin. The interval of 5 min between records was chosen because it was considered to be long enough to reflect a response to a physiological signal while intervals of, say, 1- to 2-min duration could be a reflection of transient distractions. Each feed bin was filled each morning and refusals were removed before refilling. For R animals, only the length of time taken to consume their ration was recorded. 2.3. Blood sample collection and hormone assays Single blood samples (7 ml) were collected by jugular venipuncture every 2 weeks for the purpose of determination of concentrations of insulin, growth hormone, thyroxine (T4), triiodothyronine (T3), prolactin, and insulin-like growth factor 1 (IGF-1). Samples were collected at a time between 0800 and 1100 h, before feeding. On four occasions, each within 2 weeks of the times of the summer or winter solstices or the spring or autumn equinoxes, jugular catheters were inserted under local anaesthesia. Four blood samples were collected from all animals at intervals of 15 min beginning at 1100 h. Each animal was then injected with 40 ml of 50% (w/v) glucose and samples were collected at intervals of 15 min for a further 3 h for determination of the profiles of glucose and insulin in the peripheral circulation. From 1600 to 0800 h, blood samples were collected from all animals at intervals of 1 h. The sheep were then given their feed and further samples were collected at intervals of 30 min until 1100 h and then hourly until 1700 h. This sample collection regime was designed to characterise plasma insulin and GH profiles throughout the day and, in particular, to determine the pattern of response to food ingestion. (Comprehensive description of plasma GH profiles would have required more frequent sample collection to take account of pulsatile secretion, but this could have disrupted normal feeding patterns and was therefore not undertaken.) Profiles of the thyroid hormones, IGF-1, and prolactin were not determined using these samples because preliminary measurements indicated that changes in concentration over that timescale were small. On three occasions, in October (intermediate daylength), December (short daylength), and May (long daylength), samples of CSF were collected before feeding from approximately half of the animals in each treatment
S.M. Rhind et al. / Physiology & Behavior 71 (2000) 343±351
group. This was done under mild sedation by insertion of a needle into the epidural space in the lumbo-sacral region of the spine. Samples contaminated with blood were discarded. Samples were collected from the remaining animals approximately 12 h later (and approximately 12 h after feeding) on each of these occasions. Previously described radioimmunoassay methods were used to determine circulating concentrations of insulin [13], growth hormone [16], T4 (Amerlex M RIA kit, Diagnostic Products, Los Angeles, CA, USA), T3 (Amerlex M RIA kit, Diagnostic Products), prolactin [12], and IGF-1 [4]. All methods were validated for sheep plasma by demonstrating that measures of hormone concentrations in serially diluted samples, selected randomly, ran parallel with the standard curve. Insulin concentrations in CSF were determined using a modification of an assay described previously [13]. An amount of 500 ml of CSF was used with a higher dilution of first antibody than in the original assay (1:75,000) and with standards ranging from 0.65 to 21 mU/l. 2.4. Statistical analyses 2.4.1. Meal patterns Serial records of weights of food in the food bins, for 48 h in each month, were used to determine, for each AL animal in the study, the average meal durations and frequencies and rates of ingestion at monthly intervals throughout the study. Data were analysed using Genstat 5 software [11]. Since the pattern of food intake differed between periods of daylight and darkness, days were divided into two recording periods, daylight and darkness, as described previously [16] except that the time before refilling of the feed bin was disregarded in the present analysis because for 9 months of the 15-month study, the bins were refilled very soon after sunrise and so this recording period was very brief. Preliminary analysis of the food bin weights revealed a degree of noise in the data from the food weighing system. To reduce the effect of this variability, the data were smoothed by applying a decreasing monotonic regression to the weights recorded, using the procedures described previously [16]. From both the entire 48-h interval and from daylight and darkness periods within the 48-h interval, frequency of meals (eating bouts), meal lengths, meal sizes (weight of feed ingested), inter-meal intervals, the proportion of time spent eating, and rate of food ingestion were calculated for each animal. These variables, describing meal patterns, were analysed using hierarchically structured analysis of variance (ANOVA) to assess (a) the effects of month (season) and (b) the effects of daylight/ darkness and the interaction of daylight/darkness with month. In the ANOVA to assess the seasonal effect only, the main effect of animal and the interaction of animal
345
with month were fitted as random effects. In analyses where daylight/darkness was included, a deeper hierarchical structure of ANOVA was required and, in addition to the above random effects, the interaction of animals with period of day and the interaction of this term with month were also fitted as random effects. The distribution of mean intermeal intervals, when determined for periods of daylight and darkness, was positively skewed (probably owing to the low frequency of long inter-meal intervals recorded during hours of darkness); these data were therefore log10 transformed before analysis. The mean times required by R animals to consume all the feed offered, at each month, were compared using ANOVA. 2.4.2. Hormone profiles Data were analysed using Genstat 5 software [11]. The effects of month (season) and feeding treatment on hormone concentrations were assessed using a hierarchical ANOVA, the structure being a nested design (animal and month (season) within animal). When necessary, hormone concentration data were log-transformed prior to ANOVA to satisfy the requirement for constant variance in ANOVA. Effects of feeding on plasma insulin concentrations, and effects of the glucose challenge on plasma glucose and insulin concentrations were compared with respect to treatment and season by comparing (a) the times (relative to feeding or glucose injection) at which maximal values were recorded and (b) the differences between mean concentrations for the two samples immediately before feeding and mean concentrations for the samples collected at 90 and 120 min after feeding (peak concentrations). When there was no increase, or a small decrease, in plasma insulin concentrations after feeding, the change was deemed to be zero. Plasma insulin values were transformed to log (value + 1) before analysis because the data exhibited a skewed distribution. Effects of treatment and month (season) were determined using ANOVA with mean basal plasma glucose and insulin concentrations included as covariates when comparing peak heights. Effects of feeding on plasma GH concentrations were estimated by comparing the mean interval between the introduction of food and the first time at which the GH concentration was below the mean baseline concentration (mean of the five samples collected immediately before feeding). The duration of the period during which concentrations were depressed below baseline concentrations (trough) and the mean concentration during that period were also compared. When concentrations remained below the baseline for more than 5 h after feeding, the duration of the trough was arbitrarily deemed to be 5 h. Effects of treatment and month (season) were determined using ANOVA with mean basal concentrations included as covariates.
346
S.M. Rhind et al. / Physiology & Behavior 71 (2000) 343±351
Correlation coefficients between mean daily intakes, averaged over 2-week periods, and plasma concentrations of hormones were calculated for each of the AL animals. Both average daily intake and hormone concentrations were adjusted to remove the effect of time (season) before calculating the correlations. These correlations were tested for departure from a mean correlation of zero using a t-test. Since the numbers of measurements taken varied between treatment groups, concentrations of insulin in the CSF were compared using restricted maximum likelihood (REML) analysis within Genstat 5 [11]. The model comprised fixed effects for treatment and season with a random effect for individual animals. 3. Results Three animals (two from the AL treatment and one from the R treatment) died during the study from causes unrelated to the experiment. Two were excluded from the analyses while the third (AL) died very late in the study and so the available data were included in the analysis. 3.1. Food intake and liveweight changes Animals of the R group consumed almost all of their food at all times of the year so that they had, as required by the experimental design, a constant level of feed intake throughout the study (Fig. 1b). However, the actual level of intake was higher than the 35 g DM/kg liveweight initially prescribed because the feeding level was set in relation to the initial liveweight when gut fill was probably relatively high. This level of intake was sufficient to maintain liveweight at a fairly constant level throughout the study (Fig. 1a). The time taken to consume all of the food (Fig. 1c) was significantly greater in January than in July ( p < 0.001), which in turn was lower than the peak in the following December ( p < 0.001). (The longer times taken during the first than second winter probably reflect ongoing adaptation to the diet.) Animals fed AL exhibited large differences with month in mean daily intakes ( p < 0.001; Fig. 1b), with mean intakes being approximately twofold higher in May than December. Daily intakes increased, from a level similar to that of the R animals in January, until about May. The higher rates of intake in AL animals during long-day than short-day photoperiod were associated with significant
differences in liveweight between months ( p < 0.001; Fig. 1a), with mean liveweights exhibiting an increase between March and July. This reflected an increase in tissue mass as well as an increase in gut fill associated with the higher levels of feed intake. The higher VFI during the periods of long daylength was associated with slightly longer meals, but the difference was not statistically significant (maximum (May): 10.4 min, minimum (December): 8.5 min; SED = 0.535; p = 0.099). The inter-meal intervals were shorter in long daylength (minimum (April): 37.2 min, maximum (December): 58.9 min; SED = 3.78; p < 0.001). Collectively, these differences resulted in a greater percentage of time spent eating (maximum (May): 21.1, minimum (December): 13.3; SED = 1.41; p < 0.001) during long compared with short days. Large seasonal changes were observed in the mean rate (g/min) of food ingestion (Fig. 1d) which was higher ( p < 0.001) during long days and this, combined with the greater proportion of the time spent eating, resulted in an increased ( p < 0.001) mean meal size (Fig. 1e). While there were significant ( p < 0.001) differences between months in the mean number of meals per hour (Fig. 1f), there was no clear seasonal effect. There were significant differences between periods of the day in the pattern of feeding behaviour. While the mean meal duration was similar (daylight: 9.33, darkness: 8.72; SED = 0.378; NS) during the daylight period, animals had a faster mean rate of feed ingestion (3.93 vs. 3.37 g/min; SED = 0.241; p < 0.05) and a larger mean meal size (36.2 vs. 29.6 g; SED = 2.82; p < 0.05) than during the period of darkness which, when combined with a shorter inter-meal interval (27.5 vs. 66.1 min; SED on log scale = 0.032; p < 0.001), resulted in a larger number of meals per hour (1.66 vs. 0.79; SED = 0.061; p < 0.001) and was associated with a greater proportion of the time spent eating (0.25 vs. 0.11; SED = 0.012; p < 0.001). 3.2. Hormone profiles In both treatments, significant seasonal variations in hormone profiles (comparing highest and lowest monthly values) were recorded ( p < 0.001) with mean plasma concentrations of prolactin and IGF-1 being higher during the long-day than short-day photoperiod while the overall mean changes in plasma insulin concentration were less marked (Fig. 2a ± c). Higher plasma concentrations of T3 and T4 were recorded during periods of increasing daylength than during
Fig. 1. (a) Mean daily food intakes (fresh weight) and (b) mean liveweights of castrated male Soay sheep fed ad libitum (AL) or at a constant rate of 35 g DM/ kg0.75 initial liveweight per day (restricted, R) during the study. (Mean intakes during weeks of serial blood sample collection are excluded because normal patterns of intake were temporarily disrupted by the sample collection procedure.) (c) Mean times taken by sheep fed a restricted ration (R) to consume all of the food offered, at selected times during the study. (d) Mean rates of feed ingestion (g/min); (e) mean meal sizes (g); and (f) mean number of meals per hour, at intervals of 4 weeks, in sheep fed AL. Results are shown separately for each of two periods of the day: daylight (the period between refilling of the feed bins and sunset), darkness (the period between sunset and sunrise), and for the whole day.
S.M. Rhind et al. / Physiology & Behavior 71 (2000) 343±351
periods of declining daylength ( p < 0.001; Fig. 2d ± e). There were no consistent seasonal trends in mean plasma concen-
347
trations (ng/ml) of GH (Mean SE on log (concentration + 1) scale: 0.439 0.047; back-transformed mean = 2.75; Fig. 2f).
348
S.M. Rhind et al. / Physiology & Behavior 71 (2000) 343±351
Fig. 2. Mean concentrations of (a) insulin, (b) prolactin, (c) IGF-1, (d) T3, (e) T4, (f) GH in sheep fed AL or at a constant rate of 35 g DM/kg0.75 initial liveweight per day (restricted, R) during the study.
S.M. Rhind et al. / Physiology & Behavior 71 (2000) 343±351
349
Compared with R animals, AL animals had higher mean plasma concentrations of insulin ( p < 0.001), prolactin ( p < 0.01), T3 ( p < 0.01), and T4 ( p < 0.01) (Fig. 2a, b, d, e). The decline in mean plasma thyroid hormone concentrations, which occurred in R animals during the late summer/early autumn, was less marked in the AL animals, which had a higher mean daily intake at this time; this resulted in significant ( p < 0.001) treatment time interactions for each of these hormones. Mean plasma GH and IGF-1 concentrations did not differ significantly with treatment. Mean correlation coefficients between hormone profiles and mean daily intakes in the AL animals were statistically significant only for insulin (0.162; SE = 0.074; p < 0.05) and IGF-1 (0.237; SE = 0.053; p < 0.001). 3.3. Diurnal insulin and growth hormone profiles The time between feeding and maximal plasma insulin concentrations was variable, but there was no significant difference in the mean time (82.4 min; SE = 3.94) with treatment or season. However, the mean magnitude of the increase (expressed as log10(values + 1)) was greater in R than AL animals (1.44 vs. 0.97; SED = 0.100; p < 0.001). This difference reflected the higher pre-feeding concentrations in AL animals which in turn were associated with a higher level of food intake (Fig. 3a). (The apparent absence of an obvious peak in concentrations in AL animals (Fig. 3a) is a function of the relatively small size of the increases and variation between individuals in the timing of the peak concentrations.) Comparing the combined treatments, there were significant differences between months in the size of the increase (September: 1.41 vs. March: 0.914; SED = 0.110; p < 0.001) and in the maximum concentrations achieved (June: 1.91 vs. March: 1.69; SED = 0.037; p < 0.001). However, AL animals exhibited relatively little seasonal variation in either of these variables while in R animals, mean peak concentrations in March were less than half of those recorded in June and September; this difference in pattern resulted in a significant treatment season interaction with respect to the magnitude of the increase in insulin concentrations ( p < 0.05) and peak concentrations ( p < 0.01). There were no significant differences with either treatment or season in the time from feeding to the decline in GH concentrations and no difference in the duration of the trough (Fig. 3b). When analysed using the baseline concentrations as a covariate, there was no difference with either treatment or season in the mean trough concentrations. 3.4. Glucose challenges Mean basal plasma glucose concentrations (mmol/l), expressed on a log scale, were slightly but significantly higher in AL than R animals (0.71 vs. 0.62; SED = 0.018;
Fig. 3. Mean (a) insulin and (b) GH concentrations before and after the introduction of feed (0 min) and (c) mean insulin concentrations before and after the injection (0 min) of glucose (i.v.; 40 ml of 50% w/v) in sheep fed (AL; ± ± ± ) or at a restricted rate of 35 g DM/kg0.75 initial liveweight per day (R; - - - - - ) in March (&, 5), June (8, *), September (., 6), and December (~, 4).
p < 0.001). Injection of glucose resulted in a significant increase in circulating concentrations within 15 min of injection ( p < 0.001). The magnitude of the increase (maximum value recorded ÿ mean of two pre-injection values) was slightly, but not significantly, lower in AL than R animals (log means: 0.88 vs. 0.95; SED = 0.034; p = 0.06) and this was reflected in the mean peak concentrations achieved which were also similar (1.08 vs. 1.14; SED = 0.033; p = 0.11). There was a small seasonal difference in the magnitude of the increase (maximum (December): 0.961, minimum (September): 0.860; SED = 0.033; p < 0.05) and in the peak concentrations
350
S.M. Rhind et al. / Physiology & Behavior 71 (2000) 343±351
(maximum (December): 1.14, minimum (September): 1.08; SED = 0.026; p < 0.05). However, these differences represented changes of just 24% and 14%, respectively, between the extremes. Injection of glucose resulted in increases ( p < 0 .001) in mean log plasma insulin concentrations (Fig. 3c). The magnitude of the increase in insulin concentrations (maximum value recorded ÿ mean of two pre-injection values) was similar in AL and R animals (2.12 vs. 2.01; SED = 0.084; NS) and this was reflected in similar mean peak concentrations (2.18 vs. 2.18; SED = 0.074; NS). While there were significant differences between months in the magnitude of the increase (maximum (September): 2.20, minimum (March): 1.97; SED = 0.059; p < 0.001) which were reflected in the peak values (maximum (September): 2.27, minimum (March): 2.12; SED = 0.043; p < 0.05), the pattern of change over season was inconsistent and differed for R and AL animals and so there was a significant interaction ( p < 0.05). 3.5. CSF insulin concentrations Overall mean CSF insulin concentrations (mU/l) were significantly higher in AL than R animals (3.14 vs. 2.53; SED = 0.264; p < 0.05). However, there was no significant effect of season (October: 2.74, December: 2.64, May: 3.11; SED = 0.268) or time of day relative to feeding (prefeeding: 2.76, approximately 12 h postfeeding: 2.91; SED = 0.26). There were no significant interactions between any of the factors. 4. Discussion The observations of patterns of feed intake confirm and extend previous reports of Soay sheep [10]. As found in a previous study of red deer [16], the correlations between concentrations of each of the hormones and level of feed intake were generally low, suggesting that the seasonal changes in hormones do not directly control intake. However, hormone measurements at infrequent intervals do not take account of diurnal changes in profiles associated with the ingestion of feed [2]. Changes with season in diurnal patterns of metabolic hormones, particularly insulin and GH which have been implicated previously in the regulation of metabolism [3,17], may result in a modified signal to the appetite centre. The higher mean plasma concentrations of insulin in AL compared with R animals were consistent with previous reports of sheep fed AL [2] and probably reflect the effects of both higher level of intake and frequent small meals, in contrast to the much shorter periods of ingestion of R animals which result in larger but transient increases in insulin concentrations followed by relatively low levels during the period of fasting [2]. It may also reflect greater body fat reserves in the AL animals [19].
The underlying differences between different times of year in postprandial changes in plasma insulin concentrations, which were masked in animals fed AL, mean that there is likely to be a difference with month in the insulin signal to the appetite centre of the brain. However, it does not indicate whether the difference is a function of a difference in the reflex, vagally mediated response, or in the later nutrient-induced response [8]. Measurements of the response in plasma insulin concentrations to the glucose challenge provide an index of the nutrient-induced component of the insulin response which is independent of the neurally controlled response associated with feeding. In R animals, there was no evidence of a consistent difference between months in the insulin response to the glucose challenge and so it is concluded that the nutrient-induced component of the insulin response was largely independent of month. Thus, the observed variation in insulin response to feeding is likely to be attributable primarily to differences in the CNS and vagal signals rather than to differences in the pancreatic response to changes in systemic hormone profiles or receptor activity. Variation in pancreatic response to the neural input could be a reflection of variation in either the neural signal or the responsiveness of the organ to the signal; this remains to be determined. Insulin secretion is known to depend on the systemic profiles of other hormones, such as the thyroid hormones [1]; thus, effects of seasonal changes in thyroid hormone concentrations could be expressed through differences in the insulin response to vagal stimuli. The effects of insulin on metabolism are modified by the actions of other hormones and, in particular, GH [3]. However, the absence of any difference with treatment or month in the postprandial GH profiles suggests that this hormone is not directly involved in the expression of differences between months in food intake. Recently, there has also been considerable interest in the role of leptin in the regulation of feed intake and fat depots. However, it has been shown that leptin concentrations are not significantly altered by increased insulin or glucose concentrations per se [9], suggesting that changes in the concentrations of this hormone in the immediate postprandial period are not likely to be a significant physiological signal in sheep. The insulin signal to the brain may be regulated partly through changes in the rate of transport across the blood ± brain barrier [17]. This differs with season in some species [6] and could represent an additional mechanism through which seasonal changes in appetite may be regulated. While higher mean concentrations of insulin in the CSF were recorded in AL than R animals, and there was a trend towards higher concentrations during long-day photoperiod in these animals, there were no significant differences between months of sample collection in R animals. This suggests that in sheep, there was little variation in the rate of insulin transfer into the CSF and so seasonal variation in VFI is unlikely to be mediated through differences in the rate of transfer of insulin into the appetite centre of the brain.
S.M. Rhind et al. / Physiology & Behavior 71 (2000) 343±351
However, systemic concentrations were positively related to level of food intake, as reported previously [2], and this relationship was reflected in CSF insulin concentrations. These observations, together with the fact that insulin concentrations in the CSF are known to be an index of the signal previously received by the appetite centre [17], are consistent with the hypothesis of Schwartz et al. [17] that insulin could be a medium- to long-term, intake-related signal to the appetite centre. On the other hand, the absence of consistent differences in insulin concentrations in the CSF with time of day, even though there were large differences in pattern of feed intake with time of day in both R and AL animals, suggests that changes in insulin signals to the appetite centre per se do not directly regulate food intake over a period of hours. However, conclusive evidence of this would require serial sample collection from the CSF to confirm that short-term changes in circulating insulin concentrations are not associated with parallel changes in CSF concentrations. Since the differences with month in postprandial insulin profiles did not result in equivalent changes in CSF concentrations, it is unlikely that the postprandial profiles directly influence the appetite centre to affect VFI. However, differences in postprandial insulin profiles could also act through changes in liver metabolism and therefore in the pattern of neural feedback to the appetite centre [20]. Clearly, the physiological response to postprandial hormonal changes will depend on both the nutritional and endocrine status of the animal and their interactions Ð factors which may modify hormone receptor populations and tissue responsiveness. While these remain to be investigated, the present results indicate that seasonal changes in VFI in sheep could be regulated, in part, through differences in the neurally induced postprandial increase in insulin secretion.
Acknowledgments The work was funded by the Scottish Executive Rural Affairs Department. The assistance of the staff of the Macaulay Land Use Research Institute's Glensaugh Research Station in the care of the animals is gratefully acknowledged. The provision of ovine prolactin (PRL-I-3 IOD; AFP- 10789B), prolactin antiserum (anti-ovine PRL2; AFP-C35810691), ovine GH (AFP-8758C), anti-ovine GH-2 antisera (AFP-C0123080), and anti-somatomedin-C antisera (UB2-495) by the NHPP, NIDDKD, NICHHD, and the USDA is gratefully acknowledged.
351
References [1] Achmadi J, Terashima Y. The effect of propylthiouracil-induced low thyroid function on secretion response and action of insulin in sheep. Domest Anim Endocrinol 1995;12:157 ± 66. [2] Bassett JM. Diurnal patterns of plasma insulin, growth hormone, corticosteroid and metabolite concentrations in fed and fasted sheep. Aust J Biol Sci 1974;27:167 ± 81. [3] Bauman DE, McCutcheon SN. The effects of growth hormone and prolactin on metabolism. In: Milligan LP, Grovum WL, Dobson A, editors. Control of digestion and metabolism in ruminants. Proceedings of the 6th International Symposium on Ruminant Physiology Banff, Canada, 1986. pp. 436 ± 55. [4] Bruce LA, Atkinson T, Hutchinson JSM, Shakespear RA, MacRae JC. The measurement of insulin-like growth factor 1 in sheep plasma. J Endocrinol 1991;128:R1 ± 4. [5] Feldman BF. In: Kaneko JJ, editor. Clinical biochemistry of domestic animals. London: Academic Press, 1989. pp. 836 ± 65. [6] Florant GL, Richardson RD, Mahan S, Singer L, Woods SC. Seasonal changes in CSF insulin levels in marmots: insulin may not be a satiety signal for fasting in winter. Am J Physiol 1991;260:R712 ± 6. [7] Forbes JM. Voluntary food intake and diet selection in farm animals. Wallingford CAB International: Oxon, 1997. [8] Godden PMM, Weekes TEC. Insulin, prolactin and thyroxine responses to feeding, and to arginine and insulin injections during growth in lambs. J Agric Sci (Cambridge) 1981;96:353 ± 62. [9] Kauter K, Ball M, Kearney P, Tellam R, McFarlane JR. Adrenaline, insulin, and glucagon do not have acute effects on plasma leptin levels in sheep: development and characterisation of an ovine leptin assay. J Endocrinol 2000;166:127 ± 35. [10] Kay RNB. Seasonal changes of appetite in deer and sheep. Agric Res Counc Res Rev 1979;5:13 ± 5. [11] Lawes Agricultural Trust. Genstat 5 Committee, Genstat 5 Release 3, Reference Manual. Oxford: Clarendon Press, 1994. [12] McNeilly AS, Andrews P. Purification and characterisation of caprine prolactin. J Endocrinol 1974;60:359 ± 67. [13] MacRae JC, Bruce LA, Hovell FD de B, Hart IC, Inkster J, Atkinson T. Influence of protein nutrition on the response of growing lambs to exogenous bovine growth hormone. J Endocrinol 1991;130:53 ± 61. [14] Rhind SM, Bass J, Doney JM, Hunter EA. Effect of litter size on the milk production, blood metabolite profiles and endocrine status of ewes lambing in January and April. Anim Prod 1991;53:71 ± 80. [15] Rhind SM, Bass J, Doney JM. Pattern of milk production of East Friesland and Scottish Blackface ewes and associated blood metabolite and hormone profiles. Anim Prod 1992;54:265 ± 73. [16] Rhind SM, McMillen SR, Duff E, Hirst D, Wright S. Seasonality of meal patterns and hormonal correlates in red deer. Physiol Behav 1998;65:295 ± 302. [17] Schwartz MW, Figlewicz DP, Baskin DG, Woods SC, Porte D. Insulin in the brain: a hormonal regulator of energy balance. Endocr Rev 1992;13:387 ± 414. [18] Weekes TEC, Godden PMM. Nutrition and metabolic hormones. Hormones and metabolism in ruminants. Proceedings of an Agricultural Research Council Workshop, 1981. pp. 99 ± 111. [19] Vandermeeschen-Doize F, Buchat JC, Bouckoms-Vandemeir M-A, Paquay R. Effects of long term feeding on plasma lipid components and blood glucose, b-hydroxybutyrate and insulin concentrations in lean adult sheep. Reprod Nutr Dev 1983;23:51 ± 63. [20] Vanderweele DA. Insulin is a prandial satiety hormone. Physiol Behav 1994;56:619 ± 22.
Effect of long-term feed restriction on seasonal endocrine changes in Soay sheep S.M. Rhinda,*, S.R. McMillena, E. Duff b, C.E. Kylea, S. Wrighta a
b
Macaulay Land Use Research Institute, Craigiebuckler, Aberdeen AB15 8QH, UK Biomathematics and Statistics Scotland, Macaulay Land Use Research Institute, Craigiebuckler, Aberdeen AB15 8QH, UK Received 26 April 2000; accepted 7 July 2000
Abstract Groups of 15 adult, castrated, male Soay sheep were housed under natural daylength conditions at 57°N and fed a complete diet ad libitum (AL) or at a restricted rate (R) of 35 g dry matter (DM)/kg0.75 initial liveweight per day. The diet was based on barley and dried grass pellets and contained an estimated 11.6 MJ of metabolisable energy, 83% DM and 140 g crude protein/kg DM. In the AL animals, higher levels of feed intake during the periods of long daylength were associated with shorter intermeal intervals ( p < 0.001), a greater meal frequency ( p < 0.001), and a greater proportion of time spent eating ( p < 0.001) together with a greater rate of feed ingestion ( p < 0.001) and an increased meal size ( p < 0.001). Mean plasma concentrations of insulin, prolactin, insulin-like growth factor 1 (IGF-1), triiodothyronine (T3), and thyroxine (T4) were higher ( p < 0.001) in the spring or summer than in the autumn. Mean plasma GH concentrations did not differ with month. Compared with R animals, AL animals had higher mean plasma concentrations of insulin ( p < 0.001), prolactin ( p < 0.01), T3 ( p < 0.01), and T4 ( p < 0.01). Plasma GH and IGF-1 concentrations did not differ significantly with treatment. There was a greater increase in plasma insulin concentrations following feeding in R than AL animals ( p < 0.001) owing to higher pre-feeding concentrations in AL animals and the ingestion of larger amounts of feed by R than AL animals in the period after fresh feed was introduced. There were significant differences between months in this response, in R animals ( p < 0.01). Mean CSF insulin concentrations were significantly higher in AL than R animals ( p < 0.05) but were not affected by month. Neither was there a difference between pre-feeding concentrations and concentrations at approximately 12 h after feeding. It is concluded that the differences in the response of plasma insulin concentrations to feeding at different times of year, which were detected in R animals, were attributable, primarily, to differences in the vagally-induced insulin response to feeding and that these differences may provide important feedback signals to the appetite centre. D 2000 Elsevier Science Inc. All rights reserved. Keywords: Season; Food intake; Hormones; Sheep
1. Introduction Seasonal variation in voluntary food intake (VFI) in ruminants is probably controlled by both neural and endocrine signals [7]. These signals can be modified by both photoperiodic and nutritional influences Ð effects which are frequently confounded Ð which makes it difficult to assess the roles of hormones in the seasonal variation in VFI. The separate effects on hormone profiles of photoperiod and of seasonal changes in level of VFI by red deer have been described previously [16]. However, in that study, no clear links were found
* Corresponding author. Tel.: +44-1224-318611; fax: +44-1224311556. E-mail address: [email protected] (S.M. Rhind).
between seasonal changes in VFI and hormone profiles. The absence of clear links suggests that even if endocrine signals do control VFI in ruminants, they cannot be adequately described by single samples collected at intervals of several days. It is postulated that changes with season may also occur in the postprandial responses of metabolic hormones and that differences in these profiles may provide a differential feedback signal to the appetite centre of the brain. Previous studies of sheep [14,15] have shown that differences in such responses occur with changes in physiological state. Insulin is considered to be a likely signal since it exhibits marked postprandial changes in circulating concentrations [18], it is known to be transported across the blood ±brain barrier [17], and has been implicated previously in appetite regulation [17].
0031-9384/00/$ ± see front matter D 2000 Elsevier Science Inc. All rights reserved. PII: S 0 0 3 1 - 9 3 8 4 ( 0 0 ) 0 0 3 4 3 - 7
344
S.M. Rhind et al. / Physiology & Behavior 71 (2000) 343±351
The first aim of the present experiment was to determine whether or not variations in postprandial insulin profiles occur which could have a role in the control of seasonal variation in components of shortterm patterns of feed ingestion (rate of ingestion, meal size, etc.). The second objective of the present study was to determine whether or not there were seasonal differences in the insulin response to a glucose challenge. This measurement served two purposes. Firstly, it was designed to show whether or not the plasma insulin response to a standard challenge with an energy-providing substrate was modified by photoperiod per se. Secondly, it was designed to separate the later, nutrient-driven response [18] from the initial, vagally driven insulin response to feeding. The supply of many protein hormones to the brain is generally limited by the blood ± brain barrier [5], but since there is evidence of active transfer of insulin into the brain, this hormone could have a role in signalling the metabolic state of the animal and in the regulation of appetite [17]. The final objective of the study was to determine concentrations of insulin in the cerebrospinal fluid (CSF) at selected seasons and times of day in order to assess the potential significance of this signal to the brain in the control of patterns of food intake. 2. Materials and methods 2.1. Animals and treatments Two groups of 15 adult, castrated, male Soay sheep which had previously been maintained on pasture were housed in individual pens under natural daylength conditions at 57°N throughout a 15-month study beginning in January. All animals were fed a complete diet, based on barley and dried grass pellets, which contained an estimated 11.6 MJ of metabolisable energy, 83% dry matter (DM) and 140 g crude protein/kg DM. In order to separate the effects on hormone profiles of photoperiod and level of feed intake, animals of one group were fed ad libitum (AL) to a refusal rate of 0.2 (refusals were recorded daily and mean daily food intakes calculated) while animals of the second group were fed once daily between 0800 and 1100 h, at a restricted rate (R) of 35 g DM/kg0.75 initial liveweight, throughout the study. This was equivalent to the level of intake recorded previously in Soay sheep during the winter [10]. This level of intake was designed to ensure that the intake of animals was constant throughout the study while meeting the requirements for the maintenance of liveweight. Liveweights were recorded every 2 weeks during the study.
2.2. Measurement of meal patterns At intervals of approximately 1 month, the meal patterns of animals fed AL were determined by automated recording, to the nearest 5 g, of the fresh weight of feed present in their feed bins at 5-min intervals for 2 days. Using these records, mean meal durations and frequencies and rates of feed ingestion were determined for AL animals for each month. An inter-meal interval was defined as the time during which there was no reduction in the weight of feed in the bin. The interval of 5 min between records was chosen because it was considered to be long enough to reflect a response to a physiological signal while intervals of, say, 1- to 2-min duration could be a reflection of transient distractions. Each feed bin was filled each morning and refusals were removed before refilling. For R animals, only the length of time taken to consume their ration was recorded. 2.3. Blood sample collection and hormone assays Single blood samples (7 ml) were collected by jugular venipuncture every 2 weeks for the purpose of determination of concentrations of insulin, growth hormone, thyroxine (T4), triiodothyronine (T3), prolactin, and insulin-like growth factor 1 (IGF-1). Samples were collected at a time between 0800 and 1100 h, before feeding. On four occasions, each within 2 weeks of the times of the summer or winter solstices or the spring or autumn equinoxes, jugular catheters were inserted under local anaesthesia. Four blood samples were collected from all animals at intervals of 15 min beginning at 1100 h. Each animal was then injected with 40 ml of 50% (w/v) glucose and samples were collected at intervals of 15 min for a further 3 h for determination of the profiles of glucose and insulin in the peripheral circulation. From 1600 to 0800 h, blood samples were collected from all animals at intervals of 1 h. The sheep were then given their feed and further samples were collected at intervals of 30 min until 1100 h and then hourly until 1700 h. This sample collection regime was designed to characterise plasma insulin and GH profiles throughout the day and, in particular, to determine the pattern of response to food ingestion. (Comprehensive description of plasma GH profiles would have required more frequent sample collection to take account of pulsatile secretion, but this could have disrupted normal feeding patterns and was therefore not undertaken.) Profiles of the thyroid hormones, IGF-1, and prolactin were not determined using these samples because preliminary measurements indicated that changes in concentration over that timescale were small. On three occasions, in October (intermediate daylength), December (short daylength), and May (long daylength), samples of CSF were collected before feeding from approximately half of the animals in each treatment
S.M. Rhind et al. / Physiology & Behavior 71 (2000) 343±351
group. This was done under mild sedation by insertion of a needle into the epidural space in the lumbo-sacral region of the spine. Samples contaminated with blood were discarded. Samples were collected from the remaining animals approximately 12 h later (and approximately 12 h after feeding) on each of these occasions. Previously described radioimmunoassay methods were used to determine circulating concentrations of insulin [13], growth hormone [16], T4 (Amerlex M RIA kit, Diagnostic Products, Los Angeles, CA, USA), T3 (Amerlex M RIA kit, Diagnostic Products), prolactin [12], and IGF-1 [4]. All methods were validated for sheep plasma by demonstrating that measures of hormone concentrations in serially diluted samples, selected randomly, ran parallel with the standard curve. Insulin concentrations in CSF were determined using a modification of an assay described previously [13]. An amount of 500 ml of CSF was used with a higher dilution of first antibody than in the original assay (1:75,000) and with standards ranging from 0.65 to 21 mU/l. 2.4. Statistical analyses 2.4.1. Meal patterns Serial records of weights of food in the food bins, for 48 h in each month, were used to determine, for each AL animal in the study, the average meal durations and frequencies and rates of ingestion at monthly intervals throughout the study. Data were analysed using Genstat 5 software [11]. Since the pattern of food intake differed between periods of daylight and darkness, days were divided into two recording periods, daylight and darkness, as described previously [16] except that the time before refilling of the feed bin was disregarded in the present analysis because for 9 months of the 15-month study, the bins were refilled very soon after sunrise and so this recording period was very brief. Preliminary analysis of the food bin weights revealed a degree of noise in the data from the food weighing system. To reduce the effect of this variability, the data were smoothed by applying a decreasing monotonic regression to the weights recorded, using the procedures described previously [16]. From both the entire 48-h interval and from daylight and darkness periods within the 48-h interval, frequency of meals (eating bouts), meal lengths, meal sizes (weight of feed ingested), inter-meal intervals, the proportion of time spent eating, and rate of food ingestion were calculated for each animal. These variables, describing meal patterns, were analysed using hierarchically structured analysis of variance (ANOVA) to assess (a) the effects of month (season) and (b) the effects of daylight/ darkness and the interaction of daylight/darkness with month. In the ANOVA to assess the seasonal effect only, the main effect of animal and the interaction of animal
345
with month were fitted as random effects. In analyses where daylight/darkness was included, a deeper hierarchical structure of ANOVA was required and, in addition to the above random effects, the interaction of animals with period of day and the interaction of this term with month were also fitted as random effects. The distribution of mean intermeal intervals, when determined for periods of daylight and darkness, was positively skewed (probably owing to the low frequency of long inter-meal intervals recorded during hours of darkness); these data were therefore log10 transformed before analysis. The mean times required by R animals to consume all the feed offered, at each month, were compared using ANOVA. 2.4.2. Hormone profiles Data were analysed using Genstat 5 software [11]. The effects of month (season) and feeding treatment on hormone concentrations were assessed using a hierarchical ANOVA, the structure being a nested design (animal and month (season) within animal). When necessary, hormone concentration data were log-transformed prior to ANOVA to satisfy the requirement for constant variance in ANOVA. Effects of feeding on plasma insulin concentrations, and effects of the glucose challenge on plasma glucose and insulin concentrations were compared with respect to treatment and season by comparing (a) the times (relative to feeding or glucose injection) at which maximal values were recorded and (b) the differences between mean concentrations for the two samples immediately before feeding and mean concentrations for the samples collected at 90 and 120 min after feeding (peak concentrations). When there was no increase, or a small decrease, in plasma insulin concentrations after feeding, the change was deemed to be zero. Plasma insulin values were transformed to log (value + 1) before analysis because the data exhibited a skewed distribution. Effects of treatment and month (season) were determined using ANOVA with mean basal plasma glucose and insulin concentrations included as covariates when comparing peak heights. Effects of feeding on plasma GH concentrations were estimated by comparing the mean interval between the introduction of food and the first time at which the GH concentration was below the mean baseline concentration (mean of the five samples collected immediately before feeding). The duration of the period during which concentrations were depressed below baseline concentrations (trough) and the mean concentration during that period were also compared. When concentrations remained below the baseline for more than 5 h after feeding, the duration of the trough was arbitrarily deemed to be 5 h. Effects of treatment and month (season) were determined using ANOVA with mean basal concentrations included as covariates.
346
S.M. Rhind et al. / Physiology & Behavior 71 (2000) 343±351
Correlation coefficients between mean daily intakes, averaged over 2-week periods, and plasma concentrations of hormones were calculated for each of the AL animals. Both average daily intake and hormone concentrations were adjusted to remove the effect of time (season) before calculating the correlations. These correlations were tested for departure from a mean correlation of zero using a t-test. Since the numbers of measurements taken varied between treatment groups, concentrations of insulin in the CSF were compared using restricted maximum likelihood (REML) analysis within Genstat 5 [11]. The model comprised fixed effects for treatment and season with a random effect for individual animals. 3. Results Three animals (two from the AL treatment and one from the R treatment) died during the study from causes unrelated to the experiment. Two were excluded from the analyses while the third (AL) died very late in the study and so the available data were included in the analysis. 3.1. Food intake and liveweight changes Animals of the R group consumed almost all of their food at all times of the year so that they had, as required by the experimental design, a constant level of feed intake throughout the study (Fig. 1b). However, the actual level of intake was higher than the 35 g DM/kg liveweight initially prescribed because the feeding level was set in relation to the initial liveweight when gut fill was probably relatively high. This level of intake was sufficient to maintain liveweight at a fairly constant level throughout the study (Fig. 1a). The time taken to consume all of the food (Fig. 1c) was significantly greater in January than in July ( p < 0.001), which in turn was lower than the peak in the following December ( p < 0.001). (The longer times taken during the first than second winter probably reflect ongoing adaptation to the diet.) Animals fed AL exhibited large differences with month in mean daily intakes ( p < 0.001; Fig. 1b), with mean intakes being approximately twofold higher in May than December. Daily intakes increased, from a level similar to that of the R animals in January, until about May. The higher rates of intake in AL animals during long-day than short-day photoperiod were associated with significant
differences in liveweight between months ( p < 0.001; Fig. 1a), with mean liveweights exhibiting an increase between March and July. This reflected an increase in tissue mass as well as an increase in gut fill associated with the higher levels of feed intake. The higher VFI during the periods of long daylength was associated with slightly longer meals, but the difference was not statistically significant (maximum (May): 10.4 min, minimum (December): 8.5 min; SED = 0.535; p = 0.099). The inter-meal intervals were shorter in long daylength (minimum (April): 37.2 min, maximum (December): 58.9 min; SED = 3.78; p < 0.001). Collectively, these differences resulted in a greater percentage of time spent eating (maximum (May): 21.1, minimum (December): 13.3; SED = 1.41; p < 0.001) during long compared with short days. Large seasonal changes were observed in the mean rate (g/min) of food ingestion (Fig. 1d) which was higher ( p < 0.001) during long days and this, combined with the greater proportion of the time spent eating, resulted in an increased ( p < 0.001) mean meal size (Fig. 1e). While there were significant ( p < 0.001) differences between months in the mean number of meals per hour (Fig. 1f), there was no clear seasonal effect. There were significant differences between periods of the day in the pattern of feeding behaviour. While the mean meal duration was similar (daylight: 9.33, darkness: 8.72; SED = 0.378; NS) during the daylight period, animals had a faster mean rate of feed ingestion (3.93 vs. 3.37 g/min; SED = 0.241; p < 0.05) and a larger mean meal size (36.2 vs. 29.6 g; SED = 2.82; p < 0.05) than during the period of darkness which, when combined with a shorter inter-meal interval (27.5 vs. 66.1 min; SED on log scale = 0.032; p < 0.001), resulted in a larger number of meals per hour (1.66 vs. 0.79; SED = 0.061; p < 0.001) and was associated with a greater proportion of the time spent eating (0.25 vs. 0.11; SED = 0.012; p < 0.001). 3.2. Hormone profiles In both treatments, significant seasonal variations in hormone profiles (comparing highest and lowest monthly values) were recorded ( p < 0.001) with mean plasma concentrations of prolactin and IGF-1 being higher during the long-day than short-day photoperiod while the overall mean changes in plasma insulin concentration were less marked (Fig. 2a ± c). Higher plasma concentrations of T3 and T4 were recorded during periods of increasing daylength than during
Fig. 1. (a) Mean daily food intakes (fresh weight) and (b) mean liveweights of castrated male Soay sheep fed ad libitum (AL) or at a constant rate of 35 g DM/ kg0.75 initial liveweight per day (restricted, R) during the study. (Mean intakes during weeks of serial blood sample collection are excluded because normal patterns of intake were temporarily disrupted by the sample collection procedure.) (c) Mean times taken by sheep fed a restricted ration (R) to consume all of the food offered, at selected times during the study. (d) Mean rates of feed ingestion (g/min); (e) mean meal sizes (g); and (f) mean number of meals per hour, at intervals of 4 weeks, in sheep fed AL. Results are shown separately for each of two periods of the day: daylight (the period between refilling of the feed bins and sunset), darkness (the period between sunset and sunrise), and for the whole day.
S.M. Rhind et al. / Physiology & Behavior 71 (2000) 343±351
periods of declining daylength ( p < 0.001; Fig. 2d ± e). There were no consistent seasonal trends in mean plasma concen-
347
trations (ng/ml) of GH (Mean SE on log (concentration + 1) scale: 0.439 0.047; back-transformed mean = 2.75; Fig. 2f).
348
S.M. Rhind et al. / Physiology & Behavior 71 (2000) 343±351
Fig. 2. Mean concentrations of (a) insulin, (b) prolactin, (c) IGF-1, (d) T3, (e) T4, (f) GH in sheep fed AL or at a constant rate of 35 g DM/kg0.75 initial liveweight per day (restricted, R) during the study.
S.M. Rhind et al. / Physiology & Behavior 71 (2000) 343±351
349
Compared with R animals, AL animals had higher mean plasma concentrations of insulin ( p < 0.001), prolactin ( p < 0.01), T3 ( p < 0.01), and T4 ( p < 0.01) (Fig. 2a, b, d, e). The decline in mean plasma thyroid hormone concentrations, which occurred in R animals during the late summer/early autumn, was less marked in the AL animals, which had a higher mean daily intake at this time; this resulted in significant ( p < 0.001) treatment time interactions for each of these hormones. Mean plasma GH and IGF-1 concentrations did not differ significantly with treatment. Mean correlation coefficients between hormone profiles and mean daily intakes in the AL animals were statistically significant only for insulin (0.162; SE = 0.074; p < 0.05) and IGF-1 (0.237; SE = 0.053; p < 0.001). 3.3. Diurnal insulin and growth hormone profiles The time between feeding and maximal plasma insulin concentrations was variable, but there was no significant difference in the mean time (82.4 min; SE = 3.94) with treatment or season. However, the mean magnitude of the increase (expressed as log10(values + 1)) was greater in R than AL animals (1.44 vs. 0.97; SED = 0.100; p < 0.001). This difference reflected the higher pre-feeding concentrations in AL animals which in turn were associated with a higher level of food intake (Fig. 3a). (The apparent absence of an obvious peak in concentrations in AL animals (Fig. 3a) is a function of the relatively small size of the increases and variation between individuals in the timing of the peak concentrations.) Comparing the combined treatments, there were significant differences between months in the size of the increase (September: 1.41 vs. March: 0.914; SED = 0.110; p < 0.001) and in the maximum concentrations achieved (June: 1.91 vs. March: 1.69; SED = 0.037; p < 0.001). However, AL animals exhibited relatively little seasonal variation in either of these variables while in R animals, mean peak concentrations in March were less than half of those recorded in June and September; this difference in pattern resulted in a significant treatment season interaction with respect to the magnitude of the increase in insulin concentrations ( p < 0.05) and peak concentrations ( p < 0.01). There were no significant differences with either treatment or season in the time from feeding to the decline in GH concentrations and no difference in the duration of the trough (Fig. 3b). When analysed using the baseline concentrations as a covariate, there was no difference with either treatment or season in the mean trough concentrations. 3.4. Glucose challenges Mean basal plasma glucose concentrations (mmol/l), expressed on a log scale, were slightly but significantly higher in AL than R animals (0.71 vs. 0.62; SED = 0.018;
Fig. 3. Mean (a) insulin and (b) GH concentrations before and after the introduction of feed (0 min) and (c) mean insulin concentrations before and after the injection (0 min) of glucose (i.v.; 40 ml of 50% w/v) in sheep fed (AL; ± ± ± ) or at a restricted rate of 35 g DM/kg0.75 initial liveweight per day (R; - - - - - ) in March (&, 5), June (8, *), September (., 6), and December (~, 4).
p < 0.001). Injection of glucose resulted in a significant increase in circulating concentrations within 15 min of injection ( p < 0.001). The magnitude of the increase (maximum value recorded ÿ mean of two pre-injection values) was slightly, but not significantly, lower in AL than R animals (log means: 0.88 vs. 0.95; SED = 0.034; p = 0.06) and this was reflected in the mean peak concentrations achieved which were also similar (1.08 vs. 1.14; SED = 0.033; p = 0.11). There was a small seasonal difference in the magnitude of the increase (maximum (December): 0.961, minimum (September): 0.860; SED = 0.033; p < 0.05) and in the peak concentrations
350
S.M. Rhind et al. / Physiology & Behavior 71 (2000) 343±351
(maximum (December): 1.14, minimum (September): 1.08; SED = 0.026; p < 0.05). However, these differences represented changes of just 24% and 14%, respectively, between the extremes. Injection of glucose resulted in increases ( p < 0 .001) in mean log plasma insulin concentrations (Fig. 3c). The magnitude of the increase in insulin concentrations (maximum value recorded ÿ mean of two pre-injection values) was similar in AL and R animals (2.12 vs. 2.01; SED = 0.084; NS) and this was reflected in similar mean peak concentrations (2.18 vs. 2.18; SED = 0.074; NS). While there were significant differences between months in the magnitude of the increase (maximum (September): 2.20, minimum (March): 1.97; SED = 0.059; p < 0.001) which were reflected in the peak values (maximum (September): 2.27, minimum (March): 2.12; SED = 0.043; p < 0.05), the pattern of change over season was inconsistent and differed for R and AL animals and so there was a significant interaction ( p < 0.05). 3.5. CSF insulin concentrations Overall mean CSF insulin concentrations (mU/l) were significantly higher in AL than R animals (3.14 vs. 2.53; SED = 0.264; p < 0.05). However, there was no significant effect of season (October: 2.74, December: 2.64, May: 3.11; SED = 0.268) or time of day relative to feeding (prefeeding: 2.76, approximately 12 h postfeeding: 2.91; SED = 0.26). There were no significant interactions between any of the factors. 4. Discussion The observations of patterns of feed intake confirm and extend previous reports of Soay sheep [10]. As found in a previous study of red deer [16], the correlations between concentrations of each of the hormones and level of feed intake were generally low, suggesting that the seasonal changes in hormones do not directly control intake. However, hormone measurements at infrequent intervals do not take account of diurnal changes in profiles associated with the ingestion of feed [2]. Changes with season in diurnal patterns of metabolic hormones, particularly insulin and GH which have been implicated previously in the regulation of metabolism [3,17], may result in a modified signal to the appetite centre. The higher mean plasma concentrations of insulin in AL compared with R animals were consistent with previous reports of sheep fed AL [2] and probably reflect the effects of both higher level of intake and frequent small meals, in contrast to the much shorter periods of ingestion of R animals which result in larger but transient increases in insulin concentrations followed by relatively low levels during the period of fasting [2]. It may also reflect greater body fat reserves in the AL animals [19].
The underlying differences between different times of year in postprandial changes in plasma insulin concentrations, which were masked in animals fed AL, mean that there is likely to be a difference with month in the insulin signal to the appetite centre of the brain. However, it does not indicate whether the difference is a function of a difference in the reflex, vagally mediated response, or in the later nutrient-induced response [8]. Measurements of the response in plasma insulin concentrations to the glucose challenge provide an index of the nutrient-induced component of the insulin response which is independent of the neurally controlled response associated with feeding. In R animals, there was no evidence of a consistent difference between months in the insulin response to the glucose challenge and so it is concluded that the nutrient-induced component of the insulin response was largely independent of month. Thus, the observed variation in insulin response to feeding is likely to be attributable primarily to differences in the CNS and vagal signals rather than to differences in the pancreatic response to changes in systemic hormone profiles or receptor activity. Variation in pancreatic response to the neural input could be a reflection of variation in either the neural signal or the responsiveness of the organ to the signal; this remains to be determined. Insulin secretion is known to depend on the systemic profiles of other hormones, such as the thyroid hormones [1]; thus, effects of seasonal changes in thyroid hormone concentrations could be expressed through differences in the insulin response to vagal stimuli. The effects of insulin on metabolism are modified by the actions of other hormones and, in particular, GH [3]. However, the absence of any difference with treatment or month in the postprandial GH profiles suggests that this hormone is not directly involved in the expression of differences between months in food intake. Recently, there has also been considerable interest in the role of leptin in the regulation of feed intake and fat depots. However, it has been shown that leptin concentrations are not significantly altered by increased insulin or glucose concentrations per se [9], suggesting that changes in the concentrations of this hormone in the immediate postprandial period are not likely to be a significant physiological signal in sheep. The insulin signal to the brain may be regulated partly through changes in the rate of transport across the blood ± brain barrier [17]. This differs with season in some species [6] and could represent an additional mechanism through which seasonal changes in appetite may be regulated. While higher mean concentrations of insulin in the CSF were recorded in AL than R animals, and there was a trend towards higher concentrations during long-day photoperiod in these animals, there were no significant differences between months of sample collection in R animals. This suggests that in sheep, there was little variation in the rate of insulin transfer into the CSF and so seasonal variation in VFI is unlikely to be mediated through differences in the rate of transfer of insulin into the appetite centre of the brain.
S.M. Rhind et al. / Physiology & Behavior 71 (2000) 343±351
However, systemic concentrations were positively related to level of food intake, as reported previously [2], and this relationship was reflected in CSF insulin concentrations. These observations, together with the fact that insulin concentrations in the CSF are known to be an index of the signal previously received by the appetite centre [17], are consistent with the hypothesis of Schwartz et al. [17] that insulin could be a medium- to long-term, intake-related signal to the appetite centre. On the other hand, the absence of consistent differences in insulin concentrations in the CSF with time of day, even though there were large differences in pattern of feed intake with time of day in both R and AL animals, suggests that changes in insulin signals to the appetite centre per se do not directly regulate food intake over a period of hours. However, conclusive evidence of this would require serial sample collection from the CSF to confirm that short-term changes in circulating insulin concentrations are not associated with parallel changes in CSF concentrations. Since the differences with month in postprandial insulin profiles did not result in equivalent changes in CSF concentrations, it is unlikely that the postprandial profiles directly influence the appetite centre to affect VFI. However, differences in postprandial insulin profiles could also act through changes in liver metabolism and therefore in the pattern of neural feedback to the appetite centre [20]. Clearly, the physiological response to postprandial hormonal changes will depend on both the nutritional and endocrine status of the animal and their interactions Ð factors which may modify hormone receptor populations and tissue responsiveness. While these remain to be investigated, the present results indicate that seasonal changes in VFI in sheep could be regulated, in part, through differences in the neurally induced postprandial increase in insulin secretion.
Acknowledgments The work was funded by the Scottish Executive Rural Affairs Department. The assistance of the staff of the Macaulay Land Use Research Institute's Glensaugh Research Station in the care of the animals is gratefully acknowledged. The provision of ovine prolactin (PRL-I-3 IOD; AFP- 10789B), prolactin antiserum (anti-ovine PRL2; AFP-C35810691), ovine GH (AFP-8758C), anti-ovine GH-2 antisera (AFP-C0123080), and anti-somatomedin-C antisera (UB2-495) by the NHPP, NIDDKD, NICHHD, and the USDA is gratefully acknowledged.
351
References [1] Achmadi J, Terashima Y. The effect of propylthiouracil-induced low thyroid function on secretion response and action of insulin in sheep. Domest Anim Endocrinol 1995;12:157 ± 66. [2] Bassett JM. Diurnal patterns of plasma insulin, growth hormone, corticosteroid and metabolite concentrations in fed and fasted sheep. Aust J Biol Sci 1974;27:167 ± 81. [3] Bauman DE, McCutcheon SN. The effects of growth hormone and prolactin on metabolism. In: Milligan LP, Grovum WL, Dobson A, editors. Control of digestion and metabolism in ruminants. Proceedings of the 6th International Symposium on Ruminant Physiology Banff, Canada, 1986. pp. 436 ± 55. [4] Bruce LA, Atkinson T, Hutchinson JSM, Shakespear RA, MacRae JC. The measurement of insulin-like growth factor 1 in sheep plasma. J Endocrinol 1991;128:R1 ± 4. [5] Feldman BF. In: Kaneko JJ, editor. Clinical biochemistry of domestic animals. London: Academic Press, 1989. pp. 836 ± 65. [6] Florant GL, Richardson RD, Mahan S, Singer L, Woods SC. Seasonal changes in CSF insulin levels in marmots: insulin may not be a satiety signal for fasting in winter. Am J Physiol 1991;260:R712 ± 6. [7] Forbes JM. Voluntary food intake and diet selection in farm animals. Wallingford CAB International: Oxon, 1997. [8] Godden PMM, Weekes TEC. Insulin, prolactin and thyroxine responses to feeding, and to arginine and insulin injections during growth in lambs. J Agric Sci (Cambridge) 1981;96:353 ± 62. [9] Kauter K, Ball M, Kearney P, Tellam R, McFarlane JR. Adrenaline, insulin, and glucagon do not have acute effects on plasma leptin levels in sheep: development and characterisation of an ovine leptin assay. J Endocrinol 2000;166:127 ± 35. [10] Kay RNB. Seasonal changes of appetite in deer and sheep. Agric Res Counc Res Rev 1979;5:13 ± 5. [11] Lawes Agricultural Trust. Genstat 5 Committee, Genstat 5 Release 3, Reference Manual. Oxford: Clarendon Press, 1994. [12] McNeilly AS, Andrews P. Purification and characterisation of caprine prolactin. J Endocrinol 1974;60:359 ± 67. [13] MacRae JC, Bruce LA, Hovell FD de B, Hart IC, Inkster J, Atkinson T. Influence of protein nutrition on the response of growing lambs to exogenous bovine growth hormone. J Endocrinol 1991;130:53 ± 61. [14] Rhind SM, Bass J, Doney JM, Hunter EA. Effect of litter size on the milk production, blood metabolite profiles and endocrine status of ewes lambing in January and April. Anim Prod 1991;53:71 ± 80. [15] Rhind SM, Bass J, Doney JM. Pattern of milk production of East Friesland and Scottish Blackface ewes and associated blood metabolite and hormone profiles. Anim Prod 1992;54:265 ± 73. [16] Rhind SM, McMillen SR, Duff E, Hirst D, Wright S. Seasonality of meal patterns and hormonal correlates in red deer. Physiol Behav 1998;65:295 ± 302. [17] Schwartz MW, Figlewicz DP, Baskin DG, Woods SC, Porte D. Insulin in the brain: a hormonal regulator of energy balance. Endocr Rev 1992;13:387 ± 414. [18] Weekes TEC, Godden PMM. Nutrition and metabolic hormones. Hormones and metabolism in ruminants. Proceedings of an Agricultural Research Council Workshop, 1981. pp. 99 ± 111. [19] Vandermeeschen-Doize F, Buchat JC, Bouckoms-Vandemeir M-A, Paquay R. Effects of long term feeding on plasma lipid components and blood glucose, b-hydroxybutyrate and insulin concentrations in lean adult sheep. Reprod Nutr Dev 1983;23:51 ± 63. [20] Vanderweele DA. Insulin is a prandial satiety hormone. Physiol Behav 1994;56:619 ± 22.