NANCY L. WAYNE: BENOIT MALPAUX: and FRED J. KARSCH* ABSTRACT

Transcription

1 BIOLOGY OF REPRODUCTION 39,66-75 (1988) How Does Melatonin Code for Day Length in the Ewe: Duration of Nocturnal Melatonin Release or Coincidence of Melatonin with a Light-Entrained Sensitive Period?' NANCY L. WAYNE: BENOIT MALPAUX: and FRED J. KARSCH* Consortium for Research in Developmental and Reproductive Biology Department of Physiology The University of Michigan Ann Arbor, Michigan ABSTRACT The main objective of the study was to test the hypothesis that the phase of melatonin release with respect to the light-dark cycle mediates the effects of photoperiod on the reproductive response of the ewe. To test the phase hypothesis, we eliminated endogenous melatonin secretion by pinealectomy and then restored physiological levels of serum melatonin with rises of the same duration but at different phases of the light-dark cycle (either at night or in the middle of the day). Serum melatonin patterns were determined by radioirnmunoassay in samples taken hourly for 24 h. The reproductive state was monitored by measuring serum luteinizing hormone (LH) in ovariectomized ewes treated with constant-release estradiol implants. Infusion of a long-day pattern of melatonin was equally effective in maintaining reproductive suppression when given during the night or the middle of the day. LH remained low for approximately 150 days and then rose as ewes became refractory to the inhibitory melatonin signal. These results do not support the phase hypothesis. Rather, they are consistent with the hypothesis that the duration of the nocturnal secretion of melatonin codes for day length. INTRODUCTION For many seasonal breeders, photoperiod is the primary environmental factor that synchronizes reproductive activity to a time of year that promotes survival of the young. In a wide range of mammals, including both long- and short-day breeders, the reproductive response to photoperiod is mediated by the pineal gland (long-day breeders: hamsters, Hoffman and Reiter, 1965; voles, Farrar and Clarke, 1976; ferrets, Herbert et al., 1978; white-footed mice, Glass and Lynch, 1981; short-day breeders: sheep, Lincoln, 1979; Bittman et al., 1983a; deer, Plotka et al., 1984). Two major hypotheses have been advanced to explain how melatonin codes for day length: the Accepted February 23, Received November 27, Presented in preliminary form at the 20th Annual Meeting of the Society for the Study of Reproduction, 1987, Biol Reprod 36 (Suppl. 1):102 (Abstract 167). This work was funded by NIH-HD and NIH-HD 18258, T32-HD-07048, and the lnstitur National de la Recherche Agronomique, France. * Reprint requests: Fred J. Karsch, Developmental and Reproductive Biology, The University of Michigan, 300 N. Ingalls, Room 1101, Ann Arbor, MI 'Present address: Department of Biology. University of Virginia, Charlottesvik, VA Present address: INRA Station de Physiologie de la Reproduction, Nouzilly, France. 66 duration of nocturnal melatonin secretion and the phase of melatonin release with respect to the lightdark cycle (Rollag et al., 1978; Tamarkin et al., 1979; Goldman et al., 1982). Inclusive in the phase hypothesis is the concept of a rhythm of sensitivity to melatonin that is entrained by the light-dark cycle: when the periods of melatonin release and sensitivity coincide, a photoperiodic response occurs. There is experimental support for both hypotheses, stemming largely from work on rodents (phase: Stetson and Tay, 1983; Watson-Whitmyre and Stetson, 1983; Stetson et al., 1986; duration: Carter and Goldman, 1983a,b; Goldman et al., 1984; Dowel1 and Lynch, 1987). Which hypothesis most accurately describes how melatonin mediates the photoperiodic response remains an issue of considerable controversy. The primary objective of the present study was to test the phase hypothesis in the female sheep. In this species, a model has been developed in which circulating melatonin patterns can be artificially controlled and closely monitored for their physiological accuracy (Bittman et al., 1983b). Although definitive studies have not yet been performed in the sheep, earlier observations have been interpreted as being consistent with both hypotheses (phase: Rollag et al., 1978; Almeida and Lincoln, 1982; duration: Bittman

2 and Karsch, 1984; Yellon et al., 1985). Our present approach was to eliminate endogenous melatonin secretion by pinealectomy and then to restore physiological levels of serum melatonin with rises of the same duration but at different phases of the lightdark cycle. In the course of gathering information relative to the phase hypothesis, we have gained insight into other issues critical to the involvement of melatonin in timing seasonal breeding in the ewe. There is a growing body of evidence that seasonal reproductive transitions in the ewe are not directly driven by changes in photoperiod but are generated by an endogenous rhythm. This rhythm is entrained by photoperiod and, under natural conditions, reproductive transitions occur when animals become insensitive (or refractory) to the prevailing photoperiod (Worthy and Haresign, 1983 ; Robinson and Karsch, 1984; Robinson et al., 1985; Worthy et al., 1985). In the present study, we were able to gain information concerning the mechanisms of this photorefractoriness and the expression of endogenous reproductive changes. These issues, as well as the distinction between the phase and duration hypotheses, are central to our conceptual models for photoperiodic entrainment of the seasonal reproductive rhythm. General MATERIALS AND METHODS Adult ewes, either purebred Suffolk or predominately Suffolk breed, were maintained at the Sheep Research Facility near Ann Arbor, MI (42" 18' north latitude). They were fed a diet of hay and had free access to mineral licks and water. Fifteen ewes were pinealectomized 1% yr prior to the studies; an additional 2 ewes were pinealectomized 1-2 mo before the start, and 6 ewes were pineal-intact. Pinealectomy was performed by the method of Roche and Dziuk (1969) and modified slightly as described by Dempsey et al. (1982). Completeness of pinealectomy was verified initially by visual inspection of the excised pineal gland and subsequently by documentation that the nocturnal rise in melatonin was eliminated (see below). Before pinealectomy, all ewes were ovariectomized and implanted S.C. with one Silastic capsule packed with 3.0 cm of estradiol (Karsch et al., 1973). Prior to the study, one ewe was found to be less sensitive to estradiol, and insertion of 2 capsules was necessary HOW MELATONIN CODES FOR DAY LENGTH 67 to produce an effect on luteinizing hormone (LH) secretion that was comparable to that in the other ewes. The estradiol capsules are known to maintain a physiological level of serum estradiol for a period exceeding 1 yr (3-5 pg/ml, Legan et al., 1977). Implants were replaced with new ones after 1 yr of the study. Reproductive activity was assessed from mean serum LH levels in blood samples (5 ml) taken twice weekly by jugular venipuncture. Changes in LH secretion reflect changes in the hypothalamic response to estradiol negative-feedback, and this constitutes a major neuroendocrine determinant of seasonal reproduction (Legan et al., 1977). Rising and falling levels of serum LH are coincident with the onset and termination, respectively, of breeding activity in intact ewes. These large swings in serum LH (>30- fold) provide a highly robust and photoperiodically sensitive marker of the seasonal reproductive state (Legan et al., 1977). All ewes were maintained in light-controlled rooms under artificial fluorescent illumination (350 lux at sheep head level, as measured at beginning of study). Dim red light (<3 lux) was supplied continuously to facilitate nighttime blood collection. Lights-on was at 0600 h EST during both long (16L:gD) and short (8L: 16D) day lengths. Rams were excluded from the buildings to minimize social and olfactory influences from the male on reproductive activity. Temperature was not controlled in this study. Melatonin Treatment Pinealectomized ewes were treated with melatonin using a modification of the procedure developed by Bittman and Karsch (1984). Each ewe was chronically implanted with a jugular-cardiac cannula that was attached to a portable, battery-operated infusion pump that could be programmed to turn on and off to the nearest minute (ASdMP, Auto-Syringe, Hooksett, NH). Pumps were contained in backpacks that consisted of a stainless steel box attached to a harness. This arrangement allowed complete mobility and normal behavior during melatonin treatment, which continued without interruption for the 1% yr of this study. Aliquots from a stock solution of melatonin (40 mg/ml of 95% ethanol) were diluted 3 00-fold in physiological saline containing sodium heparin (100 U/ml). This solution was delivered at a rate (38.5 pg melatonidh) calculated to achieve physiological nighttime concentrations of serum melatonin. On 2 occasions during the study, the

3 68 concentrations of serum melatonin were spot-checked in blood samples taken intermittently during a 24-h period. At the end of the study, hourly blood samples were taken from each ewe for 24 h to characterize the circulating melatonin pattern. All infusion lines were flushed with heparinized saline each week to verify that they were patent and had not become dislodged. WAYNE ET AL I 1 I 1 Nov FOb July FOb LD MEL SD MEL LD MEL L m O".YJEL Assays Serum was obtained after blood was allowed to clot overnight at 4 C and stored at -20 C until assayed. LH was determined in duplicate 25- to aliquots of serum and is expressed as ng NIH-LH-S12 per ml, as previously described (Niswender et al., 1968, 1969; Hauger et al., 1977). The intraassay coefficient of variation (CV) for standard sera containing 0.97 f 0.04 (mean k SEM), 2.28 f 0.05 and ng/ml averaged 7.9%. The interassay CV for the 3 serum pools averaged 15.2%. The limit of detection (2 standard deviations from the buffer controls) for the 21 assays in this study averaged 0.28 f 0.02 ng/ml (ranging from 0.13 to 0.48 ng/ml), for of serum. To standardize the baseline for statistical purposes, LH data below 0.48 ng/ml (upper value from the range of limit of detections in this study) were assigned this value. To verify the completeness of pinealectomy, melatonin was initially determined in duplicate extracts of serum (5 day- and 5 nighttime blood samples from each ewe) using the method of Rollag and Niswender (1976), modified as described by Bittman et al. (1983a). By the time the experiment was completed (about 3 yr after pinealectomy), a different radioimmunoassay for serum melatonin had been set up in the laboratory (English et al., 1986; modified as described by Malpaux et al., 1987, 1988a). This assay was used to characterize endogenous melatonin secretion in pineal-intact ewes and the infused melatonin patterns in pinealectomized ewes. For this purpose, melatonin was assayed in duplicate aliquots of unextracted serum. Intraassay CV for standard sera containing 81 and 402 pg/ml averaged 7% (26 assays). The interassay CV for the 2 serum pools averaged 21%. The limit of detection was 16 pg/ml (for serum). Experimental Design Our basic assumption was that, if a light-entrained sensitive period exists, it would coincide with at least I DAY 0 I 1 I Prlmlng p.rlod Exprlmontol Porlod 8 months 6.5 months FIG. 1. Experimental design. Light-dark cycles and schematic patterns of serum melatonin during the 8-mo priming period and the 6.5-mo experimental period. Hatched areas indicate the relative duration of the dark period. LD MEL indicates infusion of a long-day pattern of melatonin. SD MEL indicates infusion of a short-day pattern of melatonin. Dates at top indicate onset of successive stages of the study. Number of animals is denoted by n. a portion of the night (when melatonin is normally secreted). In this case, melatonin administered during the middle of the day would miss any sensitive period and ewes should respond as if they did not receive melatonin. Therefore, we treated pinealectomized ewes with the same duration of melatonin but at different times of the light-dark cycle. Our present approach was to test the ability of ewes treated with a suppressive, long-day pattern of melatonin during the middle of the day to maintain a previously established inhibition of serum LH. The experimental design is illustrated in Figure 1. The study began with an 8-mo priming period that had 3 purposes: 1) to synchronize reproductive activity among sheep, 2) to demonstrate that the ewes were capable of exhibiting rises and falls in LH levels as a function of melatonin signals, and 3) to establish a state of reproductive suppression within a time frame in which ewes would still be responsive to the inhibitory melatonin signal during the experimental period. Beginning 27 November 1985, 17 pinealectomized ewes were subjected to 90 days of infusion of a long-day pattern of melatonin (Stage l), followed by 90 days of infusion of a short-day pattern of melatonin (Stage 2), and finally 65 days of a long-day pattern of melatonin (Stage 3). During these 3 stages of the priming period, melatonin patterns matched the photoperiod. The long-day pattern of melatonin consisted of a 8-h infusion of melatonin during an 8-h night, and the short-day pattern of melatonin consisted of a 16-h infusion of melatonin during a 16-h night. All infusions were begun at the time of lights-off.

4 HOW MELATONIN CODES FOR DAY LENGTH 69 On Day 0 of the experimental period (31 July 1986), the pinealectomized ewes were allocated to 3 treatment groups (Fig. 1). In one control group, the 8-h melatonin infusion was terminated and replaced with an 8-h infusion of saline during the night (STOP MEL, n = 5; one ewe not given saline due to technical problems). This determined the response to removal of an inhibitory signal. A second control group continued receiving 8-h melatonin during the 8-h dark period to maintain reproductive suppression (NIGHT MEL; n = 6). The experimental group had its 8-h melatonin shifted to the middle of the 16-h light period, thus testing whether the daytime infusions could maintain reproductive suppression (DAY MEL; n = 6). As an additional control, 6 pineal-intact ewes were treated during the priming period with the same photoperiod shifts experienced by the pinealectomized, melatonin-infused ewes (beginning 27 November 1985). On Day 0 of the experimental period, these ewes were transferred from 16L:8D to 8L:16D. The purpose of this group was to demonstrate a normal inductive response to short days and to compare this pattern to that which followed termination of the inhibitory melatonin signal in the STOP MEL group. Analysis of Data Rises and declines in serum LH were identified by slight modification of an algorithm that uses regressionlike statistics with a moving variable-length window to detect changes in hormone levels (Malpaux et al., 1988a). The modification consisted of centering the window at each time point using windows with lengths 7, 9, and 11 points. Therefore, the resolution into phases by the algorithm applied to the original time series is similar to that obtained after reversing the series with respect to time. A variable-length window is used because there is no unique window width that is optimal. A single width would reduce the flexibility of the algorithm to smooth across data values and to fulfill the criterion for identifying phases of a cycle. This algorithm can identify 4 phases of a cycle: baseline, rise, plateau, and decline. The series can start in any phase. The mean value during the plateau phase is described as the peak level. Times of rises and declines of LH were further analyzed by the Kruskall-Wallis test for analysis of variance and the Mann-Whitney U test (Siegal, 1956). Since the standard deviations for the peak levels of LH were similar between groups, differences in mean peak levels were analyzed by a one-way ANOVA (Brown and Hollander, 1977). Values were considered significantly different if p< Melatonin Profiles RESULTS The melatonin patterns at the end of the experimental period are shown in Figure 2. The STOP MEL group had a low to undetectable serum melatonin level throughout the 24-h period (Fig. 2A). The infusions produced elevations in serum melatonin in the NIGHT MEL and DAY MEL groups that were 12 h out of phase, yet identical in both amplitude and duration (Fig. 2B and C). These infused patterns closely replicated the endogenous melatonin pattern from a separate group of 6 pineal-intact ewes held on the same photoperiod of 16L:8D (Fig. 2D). M)o 400 Y UEL TIME OF DAY C FIG. 2. Diurnal patterns of serum melatonin (mean f SEM) in ewes exposed to 16L:8D. Hourly samples were taken for 24 h at the end of the experimental period. Panels A, B, and C show serum melatonin patterns in pinealectomized ewes that had their long-day melatonin infusion at night stopped (STOP MEL, 5 ewes), continued (NIGHT MEL, 6 ewes), or switched to rhe middle of the day (DAY MEL, 6 ewes), respectively. Panel D shows the serum melatonin pattern of a separate group of 6 pineal-intact ewes, maintained in 16L:8D, and sampled on the same day as the others (Malpaux et al., 1988b). Absence of SEM bar indicates this value is within the area of data point. Shaded burs represent the time and duration of the infusion period. Hatched bur beneath the panels indicates the dark period.

5 70 WAYNE ET AL. LH during Priming Period The LH patterns during the three stages of the priming period are shown in Figure 3. At the beginning of Stage 1, pinealectomized ewes were reproductively asynchronous as indicated by their variable serum LH concentrations. By the end of this period, LH was suppressed in all ewes in response to the long-day melatonin pattern. All pineal-intact ewes had elevated serum LH at the start of Stage 1; LH subsequently plummeted to low levels in response to long days, reaching baseline after 30 k 4 days (mean -+ SEM). During Stage 2, serum LH began to rise in the pinealectomized ewes 49 f 4 days after the switch to the inductive short-day melatonin pattern, reaching peak levels of 9.7 f 1.0 ng/ml. LH also rose in the pineal-intact ewes 64 f 3 days after the switch to short days, reaching levels of 9.4 f 1.1 ng/ml. This rise was significantly later than that in the infused ewes (p<0.025). During Stage 3, serum LH fell in the pinealectomized ewes, reaching baseline 39 f 4 days after the switch to the long-day melatonin pattern. LH levels fell in the intact ewes, reaching baseline 21 * 1 days after the switch to long days. This decline was significantly earlier than that in the infused ewes (p<o.ool). At the end of Stage 3, pinealectomized ewes were assigned to one of three experimental groups; there was no difference in the LH patterns of these three groups during the priming period. LH during Experimental Period Results are summarized in Figure 4 and Table 1. In the STOP MEL group, serum LH began to rise 92 f 1 LD MEL I SD MEL I LD MEL,n 1 Pinx + Me1 I _- I"." 5.0 c' E 1.0 c 0.5 Y 3 STAGE 1 c L 3 tf LD so LD w I DAY OF PRIMING FIG. 3. Serum luteinizing hormone (LH) levels (mean f SEM) in samples taken twice weekly during the 3 stages of the priming period. The upper panel shows LH values in pinealectomized ewes infused with melatonin (Pinx + Mel); lower panel shows LH values in pinealintact ewes. During Stage 1, Pinx ewes were infused with a long-day pattern of melatonin (LD MEL) and pineal intact ewes were maintained on long days (LD); Stage 2, Pinx ewes were infused with a short-day pattern of melatonin fsd MEL) and pineal-intact ewes were switched to short days (SD); Stage 3, Pinx ewes were infused with a long-day pattern of melatonin and pineal-intact ewes were transfered to long days. LH data were calculated using logarithmic transformations and are plotted on a logarithmic scale. Number of animals is indicated by n. = E o 0.5 \ - F [I 1.0 w 0.5 cn o 0.5 STOP MEL NIGHT MEL (n = 5) DAY MEL DAY OF EXPERIMENT FIG. 4. Serum luteinizing hormone (LH) levels (mean t SEM) during the experimental period. Panels A, B, and C show LH values in ewes that had their long-day melatonin infusion at night stopped (STOP MEL), continued (NIGHT MEL), or switched to the middle of the day (DAY MEL) on Day 0, respectively. In Panel B, ewe #015 is plotted separately and not included in the mean. LH data plotted as in Figure 3. Number of animals is indicated by n. A B C 1

6 HOW MELATONIN CODES FOR DAY LENGTH 71 Table 1. Characteristics of luteinizing hormone (LH) response during experimental period. Onset LH rise (days)a LH return to baseline (days)a Duration of peak Peak LH level (ng/ml)a STOP MEL NIGHT MEL DAY MEL PINEAL INTACT 92 f f 8C 152t 8b 64t 3b 179 f 7 n.d.d n.d. 144 t 6b 47 f 12 n.d. n.d. 45t t f f f 1.1 amean f SEM. b p < ~ relative. ~ ~ to STOP MEL group. 'p<0.02 relative to STOP MEL group. dn.d., No data for determination. 12 days following termination of the inhibitory long-day melatonin pattern, remained elevated at peak levels for approximately 50 days, and then fell to baseline (Fig. 4A). The one ewe that did not receive saline showed an LH response indistinguishable from that in the saline-treated ewes; therefore, her LH values were averaged with the others. In the NIGHT MEL group, serum LH remained suppressed for 152 * 8 days (Fig. 4B). All but one of the ewes then showed an increase in LH levels despite continued nightly infusion of the long-day melatonin pattern. This rise in LH was significantly delayed relative to that in the STOP MEL group. Ewe #4015 remained reproductively suppressed throughout the experiment (values plotted separately). The LH response in the DAY MEL group was virtually identi- L \ I cal to that in ewes infused with melatonin during the night (Fig 4C). The reproductive response in the STOP MEL group following termination of the long-day melatonin pattern is compared in Figure 5 to that in pinealintact ewes following transfer to a short photoperiod. Pineal-intact ewes switched from long to short days showed a rise in LH 64 k 3 days after transfer to the short photoperiod, remained elevated at peak levels for approximately 50 days, and then fell to baseline. The overall patterns of LH in the two groups were the same, but the timing of the LH rise and subsequent fall was significantly delayed in the STOP MEL group. In addition, these same pinealectomized ewes showed a rise in LH 46 * 7 days after transfer from the long- to short-day pattern of melatonin during Stage 2 of the priming period (gray line in Fig. 5). This rise in LH after the transfer to the short-day melatonin pattern was significantly earlier than that following termination of the long-day pattern of melatonin (p< 0.0 1). DISCUSSION DAY FIG. 5. Serum luteinizing hormone (LH) levelsfor 3 sets of responses. Solid points and black line indicate values from 6 pineal-intact ewes transfered from long to short days on Day 0 of experimental period (INTACT; mean t SEM). Shaded area indicates SEM envelope around the mean value from 5 ewes that had their long-day melatonin infusions stopped on Day 0 (STOP MEL; data repeated from Fig. 4). Gray line connects mean values from 5 ewes transfered from a long- to short-day pattern of melatonin at the start of Stage 2 of priming period (SD MEL; same ewes as in STOP MEL group). LH data plotted as in Figure 3. Rationale for Experimental Design We assumed that if a light-entrained sensitive period exists, it would coincide with at least a portion of the night when melatonin is normally secreted. We chose to use a long-day pattern of melatonin (short duration) to test for phase so that the daytime infusion would not overlap with the night, thereby avoiding the putative sensitive period. There are two ways to demonstrate a long-day reproductive effect in our experimental model: (1) drive an inhibition of high levels of circulating LH, or (2) maintain already suppressed serum LH concentrations. For the

7 72 WAYNE ET AL. first approach, serum LH must be elevated in response to an inductive short-day signal prior to driving inhibition with long days. Since refractoriness to a short-day signal develops soon after the onset of this rise (Karsch et al., 1986; Malpaux et al., 1987), it is difficult to distinguish between inhibition caused by long-day melatonin and inhibition due to refractoriness to short-day melatonin. In fact, on two previous occasions we attempted to test for phase using this approach; the results did not permit a definitive conclusion because we could not differentiate between long-day inhibition and short-day refractoriness (Wayne, Bittman and Karsch, unpublished observations). On the other hand, refractoriness to an inhibitory photoperiod takes much longer, approximately days from the onset of the long-day signal (Karsch et al., 1986; Malpaux et al., 1988b). This allows plenty of time to distinguish between an LH rise resulting from long-day refractoriness and one resulting from the termination of an inhibitory signal; this discrimination was of critical importance for definitive interpretation of our data. Therefore, we considered the maintenance of longday inhibition to be the most appropriate way to test for phase in the ewe. Test for Phase If there were a sensitive period to melatonin during the night, the day infusion should have missed it, and the ewes should have responded as if melatonin infusions had stopped. This however, was clearly not the case. Rather, ewes interpreted the 8-h block of melatonin given in the middle of the day as an inhibitory signal. These results do not support a role for phase, but are consistent with the hypothesis that duration of the melatonin rise mediates the photoperiodic response. The present data corroborate our earlier work in which the duration of the melatonin elevation was systematically manipulated in the pinealectomized ewe; the reproductive response always conformed to duration (Bittman and Karsch, 1984; Yellon et al., 1985). Although our data are in accordance with the duration hypothesis, we cannot completely rule out a role for phase. It could be argued, for example, that there is a broad band of melatonin sensitivity and that both of our infusion blocks overlapped with this period. In addition, we did not test the possibility that a period of sensitivity to melatonin might be entrained by factors other than the light-dark cycle, such as the melatonin rhythm itself, as recently postulated by Stetson and coworkers (1986). Indeed, there is evidence that the pineal gland and melatonin can influence entrainment of the circadian rhythms of wheel-running activity and drinking in rats, and it has been suggested that melatonin can affect the rhythms of melatonin secretion and body temperature in humans (Quay, 1970; Redman et al., 1983; Armstrong et al., 1986; Wright et al., 1986). Nevertheless, testing the hypothesis that melatonin can entrain its own rhythm of sensitivity is difficult, at best, since there is no direct assay for a melatonin sensitive period. The controversy over the phase and duration hypotheses has arisen largely from work in rodents. Studies in the juvenile Djungarian hamster treated with timed infusions of melatonin have led to the conclusion that the reproductive response depends on the duration of elevated melatonin, not its temporal relationship to the light-dark cycle (Carter and Goldman, 1983a,b; Goldman et al., 1984). Recent work in the white-footed mouse, in which melatonin was implanted intracranially for varying durations at different times of day, also supports the duration hypothesis (Dowel1 and Lynch, 1987). These observations are at variance with the findings in Syrian and Djungarian hamsters injected with melatonin at different times of the light-dark cycle. Injections given in the late afternoon or early morning produced gonadal regression; injections at other times of the day had no effect, thereby supporting the phase hypothesis (Stetson and Tay, 1983; Watson-Whitmyre and Stetson, 1983 ; Stetson et al., 1986). The inconsistencies of the findings in rodents are most likely due to differences in the mode of administration of melatonin. The dose of melatonin used in the infusion studies (low nanogram range) was calculated to approximate the rate of melatonin synthesis in the hamster pineal during the night (Carter and Goldman, 1983a), and it was similar to the infusion rate used in the present study (normalized for body weight). In marked contrast, the dose of melatonin used in the injection studies (microgram quantities) far exceeds the nanogram amounts synthesized by the pineal during the course of the night. In our view, this compromises the physiological relevance of the injection studies and weakens the experimental basis for the phase hypothesis. Further, it is important to stress that the characteristics of the melatonin signal provided in those

8 studies cannot be resolved because there were no descriptions of the melatonin patterns produced in hamsters treated with either infusions or bolus injections. One of the unique strengths of using the sheep as a model lies in our ability to control and characterize infused melatonin patterns. Using infusions to produce melatonin patterns that were documented to be physiological, we have gathered strong support for the duration hypothesis in this and earlier studies. Further support for this hypothesis has recently been obtained by Arendt and coworkers (1988) who exposed ewes to photoperiodic manipulations which altered the phase of the melatonin peak, but not its duration. The reproductive response of those ewes reflected the duration of melatonin elevation. It is important to note that there is no definitive support for the phase hypothesis in the sheep. Loss of Response to a Long-Day Melatonin Signal Refractoriness to long days has been associated with onset of the natural breeding season in the ewe, and thus plays an important role in timing the seasonal reproductive cycle (Robinson et al., 1985; Worthy et al., 1985). Two mechanisms have been proposed to account for the development of photorefractoriness, an alteration in the generation of the melatonin signal and a spontaneous change in the post-pineal processing of that signal. There is evidence in the Soay ram and Saanen goat that photorefractoriness is associated with altered melatonin patterns, in that elevated levels were observed during the day (Almeida and Lincoln, 1984; Maeda et al., 1986). In the course of distinguishing between phase and duration, we have gathered support for the hypothesis that the loss of response to an inhibitory photoperiod results, at least in part, from a post-pineal process. Specifically, pinealectomized ewes infused with a fixed long-day pattern of melatonin became refractory to the inhibitory signal after approximately 15 0 days, this was evident from the increase in LM despite the continued melatonin infusion. The time course for the loss of response to melatonin was similar to that which we repeatedly observe for development of photorefractoriness in pineal-intact ewes held on a fixed long day (Karsch et al., 1986; Malpaux et al., 1988b). One of the 6 ewes (#4015) did not exhibit refractoriness to melatonin during the course of the experiment; it has been our experience that approxi- HOW MELATONIN CODES FOR DAY LENGTH 73 mately 10% of the Suffolk ewes in our flock do not become refractory to an inhibitory photoperiod. Our results, therefore, provide initial support for the hypothesis that refractoriness to long days results from a spontaneous change in the post-pineal processing of the melatonin signal. Additional support for this hypothesis was gathered in a separate study in which the 24-h secretory patterns of melatonin were characterized during prolonged exposure of Suffolk ewes to a fixed long day (Malpaux et al., 1988b). There was no evidence for an alteration in phase or duration of the melatonin pattern, or for elevated daytime levels, during the development of long-day refractoriness. In complimentary studies, we have examined the mechanism for development of refractoriness to an inductive day length. Pinealectomized ewes that were infused with a fixed, short-day pattern of melatonin became refractory to the inductive signal over much the same time course a group of pineal-intact ewes that became refractory to short days (Karsch et al., 1986). In addition, we obtained no evidence for an alteration of the phase or duration of the melatonin rhythm in pineal-intact ewes during the development of refractoriness to a fixed short day (Malpaux et al., 1987). Our results suggest that refractoriness to an inductive photoperiod, like that for an inhibitory one, develops at a post-pineal level. It is of interest that the time course observed for the development of refractoriness to the long-day pattern of melatonin in the present study was identical in ewes infused with melatonin during the night or in the middle of the day. Whatever mechanisms underlie refractoriness to a melatonin signal, therefore, it appears to make no difference what time of day the melatonin signal is present. Termination of an Inhibitory Photoperiod Cue The response of the STOP MEL group shows that, following the termination of a long-day pattern of melatonin, both reproductive induction and its subsequent suppression occur. This response closely resembled that of the pineal-intact ewes transfered from long to short days. Specifically, following the removal of the long-day signal by either stopping the melatonin infusion or switching to short days, LH levels rose after a certain lag, remained maximal for about 50 days, and then fell to baseline. These results lead to the hypothesis that the primary role of short days is to remove inhibition produced by long

9 74 days. This hypothesis is further supported by recent work in the adult and immature ewe in which longday signals were terminated by either pinealectomy or superior cervical ganglionectomy and the response was similar to that of ewes switched to short days (Wayne and Karsch, unpublished observations, Foster et al., 1987). Other observations in our study suggest that, in addition to removing long-day inhibition, short days can actively promote reproductive induction. The pineal-intact ewes receiving an inductive signal showed an earlier rise in serum LH than did the pinealectomized ewes that had their inhibitory melatonin signal terminated (STOP MEL group). Conclusions drawn from direct comparisons between the pineal-intact and pinealectomized ewes, however, should be tempered because of differences observed in their responses during the priming period. Nevertheless, a role for short days in advancing reproductive induction is also suggested by our finding that pinealectomized ewes switched from a long- to short-day pattern of melatonin during the priming period showed a significantly earlier rise in LH than when these same ewes had their long-day melatonin pattern terminated (SD MEL vs. STOP MEL in Fig. 5). Additional studies are therefore required to clarify whether or not short days can stimulate an active process that accelerates reproductive induction. Duration Hypothesis The central focus of this study was to investigate whether the phase of the melatonin rise, with respect to the light-dark cycle, or the duration of this elevation mediates the effects of photoperiod on the reproductive response of the ewe. Using a model in which circulating melatonin patterns could be artificially controlled and closely monitored for their physiological accuracy, we could not find evidence for a role of phase. Rather, our results are consistent with the hypothesis that the duration of nocturnal melatonin secretion codes for day length. In absolute terms, however, the duration hypothesis is likely an oversimplification of the operation of the photoperiodic timekeeping system in animals exposed to gradual changes in day length, such as during the natural photoperiodic cycle. Recent reports in both the sheep and hamster show that the same duration of elevated melatonin results in opposite reproductive responses depending on the previous melatonin pattern (Hoffman et al., 1986; Robinson WAYNE ET AL. and Karsch, 1987; Hastings et al., 1987). Therefore, it is not the absolute duration, but a change in duration of nocturnal melatonin secretion, that conveys photoperiodic information to the reproductive system. ACKNOWLEDGMENTS We thank Mr. Douglas Doop and Kirk VanNatter for their assistance with the animal experimentation; Drs. Lee E. Claypool, Douglas L. Foster, Alan H. Kaynard, Jane E. Robinson, and Ms. Celia Woodfill for the design and conduct of the experiment; Drs. Josephine H. Arendt, Gordon D. Niswender, and Leo E. Reichert, Jr. for supplying assay reagents; the Sheep Research Core Facility for maintaining rhe animals; Dr. Morton B. Brown and the Data Analysis Core Facility for processing the data; and the Standards and Reagents Core Faciility for preparing the reagents. REFERENCES Almeida OFX, Lincoln GA, Photoperiodic regulation of reproductive activity in the ram: evidence for the involvement of circadian rhythms in melatonin and prolactin secretion. Biol Reprod 27: Almeida OFX, Lincoln GA, Reproductive photorefractoriness in rams and accompanying changes in the patterns of melatonin and prolactin secretion. Biol Reprod 30: Arendt J, Symons AM, English J, Poulton AL, Tobler I, How does melatonin control seasonal reproductive cycles? Reprod Nutr Dev (in press) Armstrong SM, Cassone VM, Chesworth MJ, Redman JR, Short RV, Synchronization of mammalian circadian rhythms by melatonin. J Neural Transm 21 (Suppl): Bittman EL, Dempsey RJ. Karsch FJ, 1983b. Pineal melatonin secretion drives the reproductive response to daylength in the ewe. Endocrinology 113: Bitrman EL, Karsch FJ, Nightly duration of pineal melatonin secretion determines the reproductive response to inhibitory day length in the ewe. Biol Reprod 30: Bittman EL, Karsch FJ, Hopkins JW, 1983a. Role of the pineal gland in ovine photoperiodism: regulation of seasonal breeding and negative feedback effects of estradiol upon luteinizing hormone secretion. Endocrinology 113: Brown BW Jr, Hollander M, Statistics: A Biomedical Introduction. New York: John Wiley and Sons, Inc., pp Carter DS, Goldman BD, 1983a. Antigonadal effects of timed melatonin infusion in pinealectomized male Djungarian hamsters (Phodopus sungorus sungorus): duration is the critical parameter. Endocrinology 11 3 : Carter DS, Goldman BD, 1983b. Progonadal role of the pineal in the Djungarian hamster (Phodopus sungorus sungorus): mediation by melatonin. Endocrinology 113~ Dempsey RJ. Hopkins J. Bittman EL, Kindt GW, Total pinealectomy by an occipital parasaggital approach in sheep. Surg Neurol 18: Dowell SF, Lynch GR, Duration of the melatonin pulse in the hypothalamus controls testicular function in pinealectomized mice (Peromyscus leucopus). Biol Reprod 36: English J. Poulton AL, Arendr J. Symons AM, A comparison of the efficiency of melatonin treatments in advancing oestrus in ewes. J Reprod Fertil 77: Farrar GM, Clarke JR, Effect of chemical sympathectomy and pinealectomy upon gonads of voles (Microcus ugresris) exposed to short photoperiod. Neuroendocrinology 22: Foster DL, Ebling FJP, Claypool LE, Normal timing of puberty in a short-day breeder in the absence of ambient short days. Biol Reprod 36 (Suppl 1): 160 (Abstr. 340) Glass JD. Lynch GR, The effect of superficial pinealectomy on reproduction and brown fat in the adult white-footed mouse, Peromyscus leucopus. J Comp Physiol 144: Goldman BD, Carter DS, Hall VD, Roychoudhury P, Yellon SM, 1982.

10 HOW MELATONIN CODES FOR DAY LENGTH 75 Physiology of pineal melatonin in three hamster species. In: Klein DC (ed), Melatonin Rhythm Generating System. Basel, Switzerland: S Karger, pp Goldman BD. Darrow JM, Yogev L, Effects of timed melatonin infusions on reproductive development in the Djungarian hamster (Phodopus sungorus). Endocrinolgoy 114: Hastings MH, Herbert J, Hutchison J, Powers JB, Stell EA, Walker AP, Photoperiodic history controls the neuroendocrine interpretation of the pineal melatonin signal in the male Syrian hamster. Society for Neuroscience Program, 17th Annual Meeting, New Orleans, LA: Abstr Hauger RL, Karsch FJ, Foster DL, A new concept for control of the estrous cycle of the ewe based on the temporal relationships between luteinizing hormone, estradiol and progesterone in peripheral serum and evidence that progesterone inhibits tonic LH secretion. Endocrinology 101 : Herbert J, Stacey PM, Thorpe DH, Recurrent breeding seasons in pinealectomized or optic-nerve-sectioned ferrets. J Endocrinol 78: Hoffman RA, Reiter RJ, Pineal gland: influence on gonads of male hamsters. Science 148: Hoffmann K, Illnerova H, Vanecek J, Change in duration of the nighttime melatonin peak may be a signal driving photoperiodic responses in the Djungarian hamster (Phodopus sungorus). Neurosci Lett 67:68-72 Karsch FJ, Bittman EL, Robinson JE, Yellon SM, Wayne NL, Olster DH, Kaynard AH, Melatonin and photorefractoriness: loss of response to the melatonin signal leads to seasonal reproductive transitions in the ewe. Biol Reprod 34: Karsch FJ, Dierschke DJ, Weick RF, Yamaji T, Hotchkiss J, Knobil E, Positive and negative feedback control by estrogen of luteinizing hormone secretion in the rhesus monkey. Endocrinology 92: Legan SJ, Karsch FJ, Foster DL, The endocrine control of seasonal reproductive function in the ewe: a marked change in response to the negative feedback action of estradiol on luteinizing hormone secretion. Endocrinology 101 : Lincoln GA, Photoperiodic control of seasonal breeding in the ram: participation of the cranial sympathetic nervous system. J Endocrinol82: Maeda K, Mori Y, Kano Y, Superior cervical ganglionectomy prevents gonadal regression and increased plasma prolactin concentrations induced by long days in goats. J Endocrinol 110: Malpaux B, Brown MB, Moenter SM, Wayne NL, Woodfill CJI, Karsch FJ, 1988b. Reproductive refractoriness of the ewe to inhibitory photoperiod is not caused by alteration in the circadian secretion of melatonin. Neuroendocrinology (in press) Malpaux B, Robinson JE, Brown MB, Karsch FJ, Reproductive refractoriness of the ewe to inductive photoperiod is not caused by inappropriate secretion of melatonin. Biol Reprod 36: Malpaux B, Robinson JE, Brown MB, Karsch FJ. 1988a. Importance of changing photoperiod and melatonin secretory pattern in determining the length of the breeding season in the Suffolk ewe. J Reprod Fertil (in press) Niswender GD, Midgley AR, Reichert LE, Radioimmunologic studies with murine, bovine, ovine, and porcine luteinizing hor- mone. In: Rosenberg E (ed.), Gonadotropins Los Altos, CA: Geron-x, pp Niswender GD, Reichert LE Jr, Midgley AR Jr, Nalbandov AV, Radioimmunassay for bovine and ovine luteinizing hormone. Endocrinology 84: Plotka ED, Seal US, Letellier MA, Verme LJ, Ozoga JJ, Early effects of pinealectomy on LH and testosterone secretion in white-tailed deer. J Endocrinol 103: 1-7 Quay WB, Physiological significance of the pineal during adaptation to shifts in photoperiod. Physiol Behav 5: Redman J, Armstrong S, Ng KT, Free-running activity rhythms in the rat: entrainment by melatonin. Science 219: Robinson JE, Karsch FJ, Refractoriness to inductive day lengths terminates the breeding season of the Suffolk ewe. Biol Reprod 31: Robinson JE, Karsch FJ, Photoperiodic history and a changing melatonin pattern can determine the neuroendocrine response of the ewe to daylength. J Reprod Fertil80: Robinson JE, Wayne NL, Karsch FJ, Refractoriness to inhibitory day lengths initiates the breeding season of the Suffolk ewe. Biol Reprod 32: Roche JF, Dziuk PJ, A technique for pinealectomy of the ewe. Am J Vet Res 30: Rollag MD. Niswender GD, Radioimmunoassay of serum concentrations of melatonin in sheep exposed to different lighting regimens. Endocrinology 98: Rollag MD, O Callaghan PL, Niswender GD, Serum melatonin concentrations during different stages of the annual reproductive cycle in ewes. Biol Reprod 18: Siegal S, Nonparametric Statistics for the Behavioral Sciences. New York: McGraw-Hill Book Co., Inc., pp , Stetson MH, Sarafidis E, Rollag MD, Sensitivity of adult male Djungarian hamsters (Phodopus sungorus sungorus) to melatonin injections throughout the day: effects on the reproductive system and the pineal. Biol Reprod 35: Stetson MH, Tay DE, Time course of sensitivity of golden hamsters to melatonin injections throughout the day. Biol Reprod 29: Tamarkin L, Reppert SM, Klein DC, Regulation of pineal melatonin in the Syrian hamster. Endocrinology 104: Watson-Whitmyre M, Stetson MH, Stimulation of peak pineal melatonin release restores sensitivity to evening melatonin injections in pinealectomized hamsters. Endocrinology 112: Worthy K, Haresign W, Evidence that the onset of seasonal anoestrus in the ewe may be independent of increasing prolactin concentrations and daylength. J Reprod Fertil 69:41-48 Worthy K, Haresign W, Dodson S, McLeod BJ, Foxcroft GR, Haynes NB, Evidence that the onset of the breeding season in the ewe may be independent of decreasing plasma prolactin concentrations. J Reprod Fertil 75: Wright J, Aldhous M, Franey C, English J, Arendt J, The effects of exogenous melatonin on endocrine function in man. Clin Endocrinol 24: Yellon SM, Bittman EL, Lehman MN, Olster DH, Robinson JE, Karsch FJ, Importance of duration of nocturnal melatonin secretion in determining the reproductive response to inductive photoperiod in the ewe. Biol Reprod 32:523-29