Physiology&Behavior,Vol. 52, pp. 997-1008, 1992
0031-9384/92 $5.00 + .00 Copyright© 1992PergamonPressLtd.
Printedin the USA.
Resetting of a Circadian Clock by Food Pulses F R I E D R I C H K. S T E P H A N
Department of Psychology, Psychobiology-Neuroscience Program, Florida State University, Tallahassee, F L 32306-1051 Received 7 April 1992 STEPHAN, F. K. Resetting of a circadian clock byfood pulses. PHYSIOL BEHAV 52(5) 997-1008, 1992.--Rats with lesions of the suprachiasmatic nuclei were exposed to daily feeding until anticipatory activity (AA) developed. Meals were then phase advanced or delayed and presented for 1--4consecutive days. The phase of the circadian pacemaker was assessed during probes of total food deprivation before or after 8 days of intervening ad lib feeding. One or two food pulses caused phase delays for one cycle but were insufficientto reset the feedingentrainable pacemaker. Complete or partial resettingto 6- or 9-h advances or delays was observed in some rats after three food pulses and in all rats after four food pulses. In some rats, phase shifts of meals appeared to induce both advancing and delaying transients, and two bouts of AA appeared during food-deprivation probes. This suggests that the feeding entrainable pacemaker consists of two or more oscillators which became uncoupled after phase shifts. The persistence of AA at the preshift phase observed after initial phase delays, concomitantly with AA to the phase shifted meals, also suggests the presence of a second oscillator. Circadian rhythms
Suprachiasmaticnuclei
Phase shifts
RATS maintained on a single meal per day display an increase in activity beginning several hours prior to mealtime (9). This anticipatory activity (AA) persists in rats with lesions of the suprachiasmatic nuclei (SCN) which abolish many circadian rhythms in ad lib-fed rats, and considerable evidence indicates that AA is mediated by a circadian pacemaker that is anatomically distinct from the light-entrainable circadian pacemaker (presumably the SCN) [for review, see (10)]. The formal circadian properties of the feeding-entrainable pacemaker are best studied in rats with SCN lesions because the two circadian systems interact, albeit only weakly (18-20). Several important circadian properties of AA have been established in rats with SCN lesions. First, AA develops only when the period of meals is between 22 and 31 h (2,15,24). The upper limit of entrainment is greater if the period is increased in several steps, implying history dependence of the pacemaker (15). Second, the onset of AA relative to mealtime systematically increases with period (1), suggesting that this measure represents the phase angle of entrainment. Third, although AA rarely persists (i.e., free runs) in ad lib feeding conditions, it persists for up to 5 days during food deprivation (2). Furthermore, AA reappears during food deprivation even after weeks of intervening ad lib feeding at a phase near that of preceding entrainment (4,5,7,11). This suggests that the pacemaker free runs but that activity becomes decoupled in ad lib feeding. Fourth, following phase shifts of mealtime, several days are required until AA reentrains to the new mealtime (17,23). Transients between two steady states of entrainment are characteristic of oscillators, and can provide considerable insight into the dynamics of circadian clocks. A formal description of the effects ofa Zeitgeber on circadian oscillators can be obtained by establishing a phase-response curve (PRC) which displays the phase displacement resulting from a Zeitgeber delivered only
997
Food pulses
once at various circadian times. This technique has long been used to study light-entrainable circadian rhythms (8) and has also been used to assess the effects of dark pulses (3), as well as nonphotic signals [e.g., (6)]. As yet, no PRC for meals has been established. One of the major reasons is that AA fails to free run in ad lib feeding conditions and a state of food deprivation cannot be imposed for a sufficient number of days to use a standard protocol which assesses the effect of a single presentation of the Zeitgeber on the free-running rhythm. The present study was an attempt to study resetting behavior by a method which, while not equivalent to a standard PRC, would nevertheless provide some insight into the resetting dynamics of the feeding-entrainable circadian pacemaker. The method selected for the first experiment was to assess the phase displacement resulting from exposure to one phaseshifted signal after steady-state entrainment. Since AA is expressed during total food deprivation, the phase-shifted signal was followed by 3 days of food deprivation. Because the phase of AA might not yet have attained a steady state in 3 days, this was followed by 1 week of ad lib feeding and 3 more days of food deprivation [c.f., (4)]. The results of the first experiment indicated that a single food pulse was insufficient to reset the circadian pacemaker. Consequently, in a second experiment, the phase displacement in response to two, three, or four consecutive food pulses was assessed. METHOD
Animals, Housing, and Data Collection Twenty male Sprague-Dawley rats (body weight approximately 300 g) with SCN lesions were housed in individual soundattenuated chambers that enclosed an activity wheel with a small adjacent cage. An exhaust fan provided fresh air flow and mask-
998 ing noise. The chambers and the experimental room were maintained in constant darkness. Water was available ad lib throughout the study and bottles were refilled at 2- to 3-day intervals without opening the chambers. Food hoppers were filled with pellets (Purina Laboratory Rat Chow, No. 5001) at 3- to 5-day intervals at irregular times between 0900 and 1700 h. During replacement of bedding (weekly) and restocking of food hoppers, the door of the experimental room was left ajar. Although the brief exposure (_81 responses (9 pins). For selected data segments, the amount of activity was assessed by averaging 30-min counts over a number of consecutive days. The expression of AA during food deprivation probes was determined from event records and printouts of 10-min data counts. A bout of AA was defined by sustained activity lasting 4-8 h with pauses less than 1 h, representing a tenfold increase over activity in the preceding and succeeding 4 h. In a few cases, bouts falling short of these criteria were accepted on the basis of consistency across rats in the same group. The effect of SCN lesions on circadian rhythms was assessed with the chi-square periodogram (14).
STEPHAN PROCEDURE
Experiment 1 Four rats were placed in the activity wheels on ad lib food for 10 days. One rat died during this phase of the experiment. For the remaining three rats, food access was then limited to 2 h/day (1200-1400 h). After stable entrainment of AA, food was presented with a 4-h delay. This was followed by 3 days of food deprivation, 8 days of ad lib feeding, and 3 additional days of food deprivation. Two h/day feeding was then resumed at the new phase. When AA appeared stable, phase shifts were repeated using delays of 6, 8, and l0 h. Three, or 8 additional days ofad lib feeding, were added prior to the resumption of restricted feeding to facilitate recovery from food deprivation (see Fig. l ). A 1-h phase advance was inadvertently imposed when the computer clock was reset to standard time.
Experiment 2 Sixteen rats were placed in the activity wheels with ad lib food for 20 days. Food access was then reduced to 3 h/day for 29 days. Four rats each were then exposed to phase delays or to phase advances of mealtime of 6 or 9 h. To reduce the potential effects of noise cues from running wheels among rats, two rats exposed to phase delays were housed alternately with two rats exposed to phase advances. First, the phase shifted meal was presented for 2 consecutive days. This was followed by 8 days of ad lib feeding and 3 days of food deprivation. Restricted feeding was then resumed at the new phase position. This procedure was repeated except that the phase shifted meals were presented for three and then for four consecutive meals. During the last phase of the experiment, rats were exposed three or four times to a single day of food deprivation interspersed with ad lib feeding (see Figs. 2, 4-6). RESULTS AND DISCUSSION
Histology and Periodicity Analysis Light microscopic evaluation of the lesions indicated complete destruction oftbe SCN in the three rats used in Experiment 1. Periodicity analysis of drinking over 9 days of ad lib feeding showed no significant peaks between 4 and 30 h. Of the 16 rats used in Experiment 2, 15 brains were available for light microscopic examination. The SCN appeared totally destroyed in 13 rats. In these rats, periodicity analysis of drinking over 10 days of ad lib feeding showed no significant peaks in the circadian range, although one rat (#60-4) had uitradian peaks at 8 and 12 h and another (#60-8) at 15 h. In two rats (#61-3, #61-7) a caudal unilateral SCN remnant remained intact. Both rats had low peaks at 25 h in the periodogram. Drinking data were used for periodicity analysis because low levels of activity in some rats would have made it difficult to detect residual rhythms. The absence of free-running activity rhythms is readily apparent in the event records (Figs. 1, 2, 4-6).
Experiment 1: Effects of Single Food Pulses Several difficulties in interpreting the results must be noted at the outset. First, since the phase shifts of the Zeitgeber follow an entrained condition, the pacemaker is presumably not at its endogenous period and it is difficult to predict a priori what its period will be following a phase perturbation, i.e., the pacemaker could simply start to free run, or, because of history dependence, return to a period close to 24 h. Second, since AA rarely persists in ad lib conditions (see the Introduction section), the actual trajectory of the pacemaker to the phase of AA during food
RESETTING BY FOOD PULSES
999
M51-3 0 .......... .-
-
~
..-
~2 . . . . . . . . . ,
-
AL, . . . . . . .. - ,
.
.
.
.
-
24 ......... •
,...
-',
,
.~2 . . . . . . . . . ;
24
:"
" . . . . . .
:
-4HR
-6H~
-8HR
+1 HR -IOHR
FC AL . _-" -~._"
F~ ~
.....
: . -~: - . - - - = = ~
........
-,~
"
__L,
FIG. 1. Double-plotted event records o f wheel nlnning for one rat exposed to single food pulses. The left margin corresponds to midnight. Hatched rectangles indicate time o f food availability. Upright and inverted triangles
show the beginning and end of ad lib feeding. Phase shifts in hours are given in the left margin. AL = ad lib food, FD = food deprivation. The vertical line marks the initial phase of AA and interrupted lines mark the estimated phase during ad lib feeding.
deprivation probes must be inferred in most cases. Over 8 days ofad lib feeding, ambiguities may arise because, for example, a period of 23.5 h would lead to the same phase as a period of 26.5 h. Third, although there is considerable evidence that the pacemaker consists of two (or more) oscillators ( 16-19,23), few rats express two separate bouts of AA during food deprivation after forced dissociation (16,22). Consequently, presumed trajectories of a second oscillator are highly subjective. Finally, it is not clear whether transients indicate the phase of the pacemaker or of a driven process, i.e., the pacemaker may be fully reset on the first cycle while the driven process may require a number of cycles to catch up, generating transients. Because the phase of the pacemaker cannot be assessed directly, the phase of AA was taken to coincide with that of the pacemaker or one of its oscillators, although this may not actually be the case. A double-plotted event record for one of the three rats in Experiment 1 is shown in Fig. 1. Prior to the first phase shift, all rats had developed AA and the onset of AA was between 0600 and 0800 h. The effect of the phase-shifted food pulse was assessed to the nearest 0.5 h by comparing the onset of AA on the day of the shift with that on the subsequent day. The onset of AA was delayed by 1.5 h in two rats and by 0.5 h in one rat (Table 1). During the subsequent 2 days of food deprivation, the onset of AA occurred earlier and preceded the preshift phase of AA by about 2 h (Fig. 1). During the second food deprivation probe, distinct bouts of activity began between 0600 and 0800 h, slightly later than on the last day of the previous food deprivation probe. This suggests that an oscillator free ran with a period >24 h during ad lib feeding. The onset of AA clearly shows that the phase of the pacemaker had not been delayed in response to the phase shifted food pulse. When restricted feeding was resumed at the new phase position, activity initially increased between 0800 and 0900 h, i.e., near the phase of AA at the preceding food deprivation probe. All three rats continued to show this activity on some days but not on others. Since maintenance was performed during the day and some rats ran in wheels while food hoppers were restocked, this induced activity contributed to the increase in activity on some days. However, maintenance-induced activity typically lasted only 10 to 60 rain, rather than 6 to 8 h. The potential influence of maintenance will be more fully discussed in Experiment 2. The presence of the early onset of AA prevented a clear assessment of the effects of the 6-h delay of mealtime. On the day after the phase shift, AA consisted of multiple dispersed bouts of activity without a distinct onset. On days 2 and 3 of food deprivation, distinct bouts of AA appeared between 0600 and 0730 h, a time comparable to that during the preceding food deprivation probe. During the second food deprivation probe after ad lib feeding, AA had advanced even more in all three rats, suggesting an oscillator period of
0031-9384/92 $5.00 + .00 Copyright© 1992PergamonPressLtd.
Printedin the USA.
Resetting of a Circadian Clock by Food Pulses F R I E D R I C H K. S T E P H A N
Department of Psychology, Psychobiology-Neuroscience Program, Florida State University, Tallahassee, F L 32306-1051 Received 7 April 1992 STEPHAN, F. K. Resetting of a circadian clock byfood pulses. PHYSIOL BEHAV 52(5) 997-1008, 1992.--Rats with lesions of the suprachiasmatic nuclei were exposed to daily feeding until anticipatory activity (AA) developed. Meals were then phase advanced or delayed and presented for 1--4consecutive days. The phase of the circadian pacemaker was assessed during probes of total food deprivation before or after 8 days of intervening ad lib feeding. One or two food pulses caused phase delays for one cycle but were insufficientto reset the feedingentrainable pacemaker. Complete or partial resettingto 6- or 9-h advances or delays was observed in some rats after three food pulses and in all rats after four food pulses. In some rats, phase shifts of meals appeared to induce both advancing and delaying transients, and two bouts of AA appeared during food-deprivation probes. This suggests that the feeding entrainable pacemaker consists of two or more oscillators which became uncoupled after phase shifts. The persistence of AA at the preshift phase observed after initial phase delays, concomitantly with AA to the phase shifted meals, also suggests the presence of a second oscillator. Circadian rhythms
Suprachiasmaticnuclei
Phase shifts
RATS maintained on a single meal per day display an increase in activity beginning several hours prior to mealtime (9). This anticipatory activity (AA) persists in rats with lesions of the suprachiasmatic nuclei (SCN) which abolish many circadian rhythms in ad lib-fed rats, and considerable evidence indicates that AA is mediated by a circadian pacemaker that is anatomically distinct from the light-entrainable circadian pacemaker (presumably the SCN) [for review, see (10)]. The formal circadian properties of the feeding-entrainable pacemaker are best studied in rats with SCN lesions because the two circadian systems interact, albeit only weakly (18-20). Several important circadian properties of AA have been established in rats with SCN lesions. First, AA develops only when the period of meals is between 22 and 31 h (2,15,24). The upper limit of entrainment is greater if the period is increased in several steps, implying history dependence of the pacemaker (15). Second, the onset of AA relative to mealtime systematically increases with period (1), suggesting that this measure represents the phase angle of entrainment. Third, although AA rarely persists (i.e., free runs) in ad lib feeding conditions, it persists for up to 5 days during food deprivation (2). Furthermore, AA reappears during food deprivation even after weeks of intervening ad lib feeding at a phase near that of preceding entrainment (4,5,7,11). This suggests that the pacemaker free runs but that activity becomes decoupled in ad lib feeding. Fourth, following phase shifts of mealtime, several days are required until AA reentrains to the new mealtime (17,23). Transients between two steady states of entrainment are characteristic of oscillators, and can provide considerable insight into the dynamics of circadian clocks. A formal description of the effects ofa Zeitgeber on circadian oscillators can be obtained by establishing a phase-response curve (PRC) which displays the phase displacement resulting from a Zeitgeber delivered only
997
Food pulses
once at various circadian times. This technique has long been used to study light-entrainable circadian rhythms (8) and has also been used to assess the effects of dark pulses (3), as well as nonphotic signals [e.g., (6)]. As yet, no PRC for meals has been established. One of the major reasons is that AA fails to free run in ad lib feeding conditions and a state of food deprivation cannot be imposed for a sufficient number of days to use a standard protocol which assesses the effect of a single presentation of the Zeitgeber on the free-running rhythm. The present study was an attempt to study resetting behavior by a method which, while not equivalent to a standard PRC, would nevertheless provide some insight into the resetting dynamics of the feeding-entrainable circadian pacemaker. The method selected for the first experiment was to assess the phase displacement resulting from exposure to one phaseshifted signal after steady-state entrainment. Since AA is expressed during total food deprivation, the phase-shifted signal was followed by 3 days of food deprivation. Because the phase of AA might not yet have attained a steady state in 3 days, this was followed by 1 week of ad lib feeding and 3 more days of food deprivation [c.f., (4)]. The results of the first experiment indicated that a single food pulse was insufficient to reset the circadian pacemaker. Consequently, in a second experiment, the phase displacement in response to two, three, or four consecutive food pulses was assessed. METHOD
Animals, Housing, and Data Collection Twenty male Sprague-Dawley rats (body weight approximately 300 g) with SCN lesions were housed in individual soundattenuated chambers that enclosed an activity wheel with a small adjacent cage. An exhaust fan provided fresh air flow and mask-
998 ing noise. The chambers and the experimental room were maintained in constant darkness. Water was available ad lib throughout the study and bottles were refilled at 2- to 3-day intervals without opening the chambers. Food hoppers were filled with pellets (Purina Laboratory Rat Chow, No. 5001) at 3- to 5-day intervals at irregular times between 0900 and 1700 h. During replacement of bedding (weekly) and restocking of food hoppers, the door of the experimental room was left ajar. Although the brief exposure (_81 responses (9 pins). For selected data segments, the amount of activity was assessed by averaging 30-min counts over a number of consecutive days. The expression of AA during food deprivation probes was determined from event records and printouts of 10-min data counts. A bout of AA was defined by sustained activity lasting 4-8 h with pauses less than 1 h, representing a tenfold increase over activity in the preceding and succeeding 4 h. In a few cases, bouts falling short of these criteria were accepted on the basis of consistency across rats in the same group. The effect of SCN lesions on circadian rhythms was assessed with the chi-square periodogram (14).
STEPHAN PROCEDURE
Experiment 1 Four rats were placed in the activity wheels on ad lib food for 10 days. One rat died during this phase of the experiment. For the remaining three rats, food access was then limited to 2 h/day (1200-1400 h). After stable entrainment of AA, food was presented with a 4-h delay. This was followed by 3 days of food deprivation, 8 days of ad lib feeding, and 3 additional days of food deprivation. Two h/day feeding was then resumed at the new phase. When AA appeared stable, phase shifts were repeated using delays of 6, 8, and l0 h. Three, or 8 additional days ofad lib feeding, were added prior to the resumption of restricted feeding to facilitate recovery from food deprivation (see Fig. l ). A 1-h phase advance was inadvertently imposed when the computer clock was reset to standard time.
Experiment 2 Sixteen rats were placed in the activity wheels with ad lib food for 20 days. Food access was then reduced to 3 h/day for 29 days. Four rats each were then exposed to phase delays or to phase advances of mealtime of 6 or 9 h. To reduce the potential effects of noise cues from running wheels among rats, two rats exposed to phase delays were housed alternately with two rats exposed to phase advances. First, the phase shifted meal was presented for 2 consecutive days. This was followed by 8 days of ad lib feeding and 3 days of food deprivation. Restricted feeding was then resumed at the new phase position. This procedure was repeated except that the phase shifted meals were presented for three and then for four consecutive meals. During the last phase of the experiment, rats were exposed three or four times to a single day of food deprivation interspersed with ad lib feeding (see Figs. 2, 4-6). RESULTS AND DISCUSSION
Histology and Periodicity Analysis Light microscopic evaluation of the lesions indicated complete destruction oftbe SCN in the three rats used in Experiment 1. Periodicity analysis of drinking over 9 days of ad lib feeding showed no significant peaks between 4 and 30 h. Of the 16 rats used in Experiment 2, 15 brains were available for light microscopic examination. The SCN appeared totally destroyed in 13 rats. In these rats, periodicity analysis of drinking over 10 days of ad lib feeding showed no significant peaks in the circadian range, although one rat (#60-4) had uitradian peaks at 8 and 12 h and another (#60-8) at 15 h. In two rats (#61-3, #61-7) a caudal unilateral SCN remnant remained intact. Both rats had low peaks at 25 h in the periodogram. Drinking data were used for periodicity analysis because low levels of activity in some rats would have made it difficult to detect residual rhythms. The absence of free-running activity rhythms is readily apparent in the event records (Figs. 1, 2, 4-6).
Experiment 1: Effects of Single Food Pulses Several difficulties in interpreting the results must be noted at the outset. First, since the phase shifts of the Zeitgeber follow an entrained condition, the pacemaker is presumably not at its endogenous period and it is difficult to predict a priori what its period will be following a phase perturbation, i.e., the pacemaker could simply start to free run, or, because of history dependence, return to a period close to 24 h. Second, since AA rarely persists in ad lib conditions (see the Introduction section), the actual trajectory of the pacemaker to the phase of AA during food
RESETTING BY FOOD PULSES
999
M51-3 0 .......... .-
-
~
..-
~2 . . . . . . . . . ,
-
AL, . . . . . . .. - ,
.
.
.
.
-
24 ......... •
,...
-',
,
.~2 . . . . . . . . . ;
24
:"
" . . . . . .
:
-4HR
-6H~
-8HR
+1 HR -IOHR
FC AL . _-" -~._"
F~ ~
.....
: . -~: - . - - - = = ~
........
-,~
"
__L,
FIG. 1. Double-plotted event records o f wheel nlnning for one rat exposed to single food pulses. The left margin corresponds to midnight. Hatched rectangles indicate time o f food availability. Upright and inverted triangles
show the beginning and end of ad lib feeding. Phase shifts in hours are given in the left margin. AL = ad lib food, FD = food deprivation. The vertical line marks the initial phase of AA and interrupted lines mark the estimated phase during ad lib feeding.
deprivation probes must be inferred in most cases. Over 8 days ofad lib feeding, ambiguities may arise because, for example, a period of 23.5 h would lead to the same phase as a period of 26.5 h. Third, although there is considerable evidence that the pacemaker consists of two (or more) oscillators ( 16-19,23), few rats express two separate bouts of AA during food deprivation after forced dissociation (16,22). Consequently, presumed trajectories of a second oscillator are highly subjective. Finally, it is not clear whether transients indicate the phase of the pacemaker or of a driven process, i.e., the pacemaker may be fully reset on the first cycle while the driven process may require a number of cycles to catch up, generating transients. Because the phase of the pacemaker cannot be assessed directly, the phase of AA was taken to coincide with that of the pacemaker or one of its oscillators, although this may not actually be the case. A double-plotted event record for one of the three rats in Experiment 1 is shown in Fig. 1. Prior to the first phase shift, all rats had developed AA and the onset of AA was between 0600 and 0800 h. The effect of the phase-shifted food pulse was assessed to the nearest 0.5 h by comparing the onset of AA on the day of the shift with that on the subsequent day. The onset of AA was delayed by 1.5 h in two rats and by 0.5 h in one rat (Table 1). During the subsequent 2 days of food deprivation, the onset of AA occurred earlier and preceded the preshift phase of AA by about 2 h (Fig. 1). During the second food deprivation probe, distinct bouts of activity began between 0600 and 0800 h, slightly later than on the last day of the previous food deprivation probe. This suggests that an oscillator free ran with a period >24 h during ad lib feeding. The onset of AA clearly shows that the phase of the pacemaker had not been delayed in response to the phase shifted food pulse. When restricted feeding was resumed at the new phase position, activity initially increased between 0800 and 0900 h, i.e., near the phase of AA at the preceding food deprivation probe. All three rats continued to show this activity on some days but not on others. Since maintenance was performed during the day and some rats ran in wheels while food hoppers were restocked, this induced activity contributed to the increase in activity on some days. However, maintenance-induced activity typically lasted only 10 to 60 rain, rather than 6 to 8 h. The potential influence of maintenance will be more fully discussed in Experiment 2. The presence of the early onset of AA prevented a clear assessment of the effects of the 6-h delay of mealtime. On the day after the phase shift, AA consisted of multiple dispersed bouts of activity without a distinct onset. On days 2 and 3 of food deprivation, distinct bouts of AA appeared between 0600 and 0730 h, a time comparable to that during the preceding food deprivation probe. During the second food deprivation probe after ad lib feeding, AA had advanced even more in all three rats, suggesting an oscillator period of
