Proceedings of the National Academy of Sciences Vol. 68, No. 3, pp. 595-599, March 1971

Hour-Glass Behavior of the Circadian Clock Controlling Eclosion of the Silkmoth pernyi

JAMES W. TRUMAN The Biological Laboratories, Harvard University, Cambridge, Massachusetts 02138 Communicated by Carroll M. Williams, December 18, 1970

ABSTRACT The emergence of the Pernyi silkmoth thereby established (3). The phase assumed by the new from the pupal exuviae is dictated by a brain-centered, rhythm proves to be a function of the time at which the flash photosensitive clock. In continuous darkness the clock displays a persistent free-running rhythm. In photoperiod is administered during the free-running cycle. The experimen- regimens the interaction of the clock with the daily light- tally determined function permits one to plot a "phase-re- dark cycle produces a characteristic time of eclosion. sponse curve." For our present purposes, the phase-response But, in the majority of regimens (from 23L:ID to 4L:20D), relationship is of interest because it has been used success- the eclosion clock undergoes a discontinuous "hour- fully by Pittendrigh to document the operation .of the "eclo- glass" behavior. Thus, during each daily cycle, the onset of darkness initiates a free-running cycle of the clock. sion clock" throughout the days preceding eclosion (5). The next "lights-on" interrupts this cycle and the clock In the case of Drosophila, exposure to continuous light comes to a stop late in the photophase. The moment causes a dissolution of the eclosion rhythm over the course of when the Pernyi clock stops signals the release of an a few cycles. Pittendrigh demonstrated that this "damping" eclosion-stimulating hormone and is demonstrated to be a function of the time when the free-running cycle is effect of light occurs, not because the clock uncouples from the interrupted by lights-on. Moreover, the width (duration) overt rhythm, but because the clock stops (5)! Moreover, this of the eclosion peak in a photoperiod is shown to be de- effect occurs after a surprisingly short exposure to light. As a pendent upon the length of the dark phase, and, conse- result, the Drosophila eclosion clock stops during the final part quently, upon the amount of the free-running cycle that that a 12 is completed. This relationship demonstrates that the of each photophase has duration of or more hours (5). free-running cycle may be divided into two parts. The After the eclosion clock stops, it can be promply restarted attainment of maximal accuracy (and thus the narrowest by placing Drosophila in darkness. The resultant free-running eclosion peak) is dependent upon the completion of only rhythm then routinely shows a characteristic phase relation- the first 2 hr of the free-running cycle. The completion of ship to the "lights off" signal (5). It is also important to note succeeding portions of the cycle, while having an effect upon the time of eclosion, no longer affects the accuracy that, after the onset of the free-running rhythm, the first of the clock. A mechanistic model of the eclosion clock is cycle traversed by the clock is identical to that observed in presented. succeeding cycles (5). Thus, in continuous darkness the free- running cycle or free-running behavior of the eclosion clock is an invariant that becomes established immediately after dark- appears to be among The capacity to "keep-time" ubiquitous ness restarts the clock. organisms (1). Among the most striking examples of this As in other rhythmic phenomena the timing of Drosophila are circadian which temporal organization rhythms-events eclosion varies with the photoperiod (3, 4). A central problem in the absence of external cues occur at intervals of approxi- is the role of the "circadian" clock in this timing. As men- 24 The emergence ecdysis) of the fruit mately hr. (eclosion, tioned above, the Drosophila eclosion clock stops during pseudoobscura) from the puparium is one such fly (Drosophila longer than 12 hr and is subsequently restarted event that has received considerable attention during the past photophases in a free-running manner at lights-off. Near the end of the 15 years (2-5). Eclosion in this case terminates an intricate next photophase the clock again stops-and so on at daily in- series of developmental processes that transform the maggot tervals. Under this kind of discontinuous motion the timing into the fly. Although the fly may complete development at mechanism resembles not so much a clock as an "hour-glass" any time of the day, there is only a certain period during which it may emerge. Thus, eclosion is said to be "gated" (5). (5). In the case regimens having less than Studies of the role of photoperiod in the gating of eclosion of daily photoperiod 12 hr the Drosophila clock appears as a continuous have pinpointed several important properties of the control- of light, oscillation and a new cycle is started immediately upon the ling clock. Thus, when populations of flies are transformed termination of the preceding one. Thus, the situation is sim- from a photoperiod into continuous darkness, they display a photo- persistent "free-running" rhythm of eclosion. When this free- ilar to that seen in continuous darkness except that the period forces the clock into an exact 24 hr periodicity (4). The running population is exposed to a flash of light, the time of physiological mechanism for this entrainment is unknown. eclosion is either advanced or delayed and a new rhythm is. The analysis of the Drosophila eclosion rhythm by Pitten- drigh and his coworkers has provided a substantial basis for Abbreviation: #L: #D, a light-dark regimen that alternates the further studies of the mechanism of the eclosion clock. How- specified number of hours of light with the specified hours of ever, because of its small size, Drosophila is an unsatisfactory darkness. for investigating the physiological and biochemical as- 595 Downloaded by guest on September 28, 2021 596 Zoology: J. W. Truman Proc. Nat. Acad. Sci. USA 68 (1971)

A then transferred to their respective photoperiod regimens at 260C. The methods of recording eclosion have been described (6). K////-///I * I Adult eclosion of saturniid is triggered by a neuro- secretory hormone released from the brain; the release in the case of the Pernyi takes place about 1.5 hr before eclo- sion (10). Since hormonal release, rather than eclosion per se, is the event controlled by the clock, the average time of re- lease in each photoperiod is assumed to occur 1.5 hr prior to the corresponding average eclosion time. RESULTS The Relationship of Eclosion to Photoperiod. Chilled Pernyi pupae were removed from their cocoons and placed at 260C under a 17L: 7D regimen. After the initiation of adult develop- ment, the pharate moths were placed in one of eight photo- period regimens, which ranged from 1L: 23D to 23L: ID. In all cases the developing moths were exposed to at least 10 cycles of the photoperiod prior to their eclosion. The moths usually emerged over a 4- to 5-day period. The cumulative data for each photoperiod is summarized in Fig. 1. In all of these regimens, the time of eclosion, and thus the timing of eclosion-hormone release, was characteristic of the particular photoperiod. Moreover, in most regimens, the majority of the moths emerged within a specific 3.5-hr period G of the day. Finally, we may note that the eclosion of males preceded that of females by about 1 hr. For reasons discussed *0 z z' zi I muI~ m j1.0/Z..Z I Pi below, the 1L:23D regimen disobeyed this generalization. The effects of continuous light or darkness were assessed on I I flulin * m pharate moths exposed to at least 10 cycles of a particular photoperiod. When moths were transferred from the 17L: 7D I regimen into continuous light, eclosion became arrhythmic N.- I 0 =I a E after one transient cycle (Fig. 2C). By contrast, pharate 0 6 12 Is 24 TIME (HOURS AFTER LIGHTS-OFF) moths transferred to continuous darkness developed a dis- tinct free-running rhythm that showed a period of 22 hr. As FIG. 1. The ecdysis of Pernyi moths under various photo- with Drosophila, identical rhythms were established when period conditions: A., first day of continuous darkness after moths were placed in darkness from photoperiods having prior exposure to a 12L:12D regimen; B., 1L:23D; C., 4L:20D; 8L: 12L: 22L:2D; moderate to long photophases (12L: 12D and 17L: 7D, respec- D., 16D; E., 12D; F., 17L:7D; G., 20L:4D; H., 2A first I., 23L: 1D. The white bar gives the estimated range of eclosion tively; Fig. and B). During the free-running cycle, hormone release for each regimen. The filled circles are the hormone release by the males and females occurred 21 and 22 average release times. The release time for males and for females hr, respectively, after the rhythm was started by lights-off. are presented separately in the iL: 23D regimen (see Discussion). DISCUSSION AND INTERPRETATION The hatched and white areas refer to dark and light, respectively. Termination of the clock cycle under photoperiod conditions pects of eclosion. By contrast, the giant silkmoths of the In photoperiods that have a day-length of 12 hr or more, the family have proven to be well suited for such behavior of the Drosophila eclosion clock is discontinuous in studies. Indeed, the photoreceptor, the clock, and a neuro- that the clock stops at some point late in the photophase. endocrine trigger necessary for eclosion have been localized to Pittendrigh demonstrated that in the 12L:12D regimen. the the cerebral-lobe area of the brain of these moths (6, 7). The clock stopped after 12 hr of light (5). Since this determination present communication deals with the response of the brain- was made for only the 12L: 12D photoperiod, it is unclear as centered clock of the Chinese oak silkmoth (Antheraea pernyi) to whether the 12-hr value can be applied to other regimens. to photoperiod conditions. The Pernyi moth proved to be well suited for determining when the eclosion clock stops. Thus, when this is MATERIALS AND METHODS placed in continuous darkness, we see that the eclosion hor- Cocoons containing diapausing pupae of Antheraea pernyi mone is released at the end-point of the free-running cycle. I were obtained from Japanese dealers and stored at 5YC until have assumed that the same is true in the presence of a photo- needed. The pupal diapause of Pernyi can be terminated by period. On that assumption, for each photoperiod regimen the exposing the to long-day photoperiod conditions median time of hormone release given in Fig. 1A and C-I (17L: 7D). This response is greatly accelerated after several identifies the time at which the clock cycle ends and, conse- months of preliminary chilling (8, 9). Routinely, pupae chilled quently, the time when the clock stops. for at least 4 weeks were placed in the 17L: 7D regimen at The data in Fig. 1 show that in photoperiods ranging from 260C until adult development began. Developing moths were 23L: ID to 4L: 20D the Pernyi eclosion clock reached the end- Downloaded by guest on September 28, 2021 Proc. Nat. Acad. Sci. USA 68 (1971) Silkmoth Eclosion and a Circadian Clock 597 point of its cycle when the lights were on-i.e., during the 1 1 photophase. This clock therefore differs from the Drosophila a 20 . clock in that its mechanism resembles an "hour-glass", even I_6 when the photophases are as brief as 4 hr. Since in these regi- mens the Pernyi clock stops during the photophase of each 12 daily cycle, lights-off of each cycle must restart the clock by

initiating a free-(Funning rhythm. In Fig. 3 is plotted the time _8 at which the clock stops as a function of the time that the free- 0'~

running cycle was interrupted by light. It is obvious that the -0 4- time from the onset of light until the termination of the cycle is not a constant [as it was thought to be in the case of Drc- 0[ 0 4 8 12 16 20 sophila (5) ], but is strikingly dependent upon when the prior TIME OF INTERRUPTION OF FREE-RUNNING free-running cycle was interrupted by light. CYCLE (HOURS AFTER ONSET OF CYCLE) FIG. 3. A plot showing the time that the Pernyi eclosion Gate width as a function of the photoperiod clock stops (or "damps out") as a function of when the free- As is evident in Fig. 1, the width of the eclosion gate varied running cycle is interrupted by the photophase. with the photoperiod. Under regimens with a very short scoto- phase (e.g., 23L: iD), the gate was very wide. As the amount of darkness increased, the gate width rapidly decreased, until running clock. This cycle occurs only in darkness, and in a width of 3.5 hr was attained. Further increases in the dura- Pernyi has a duration of 22 hr. The initiation of the scotonon tion of the scotophase had little effect on the gate width (with is defined as the moment when a stopped clock starts after the the exception of the special case in the 1L :23D regimen dealt onset of darkness. At the conclusion of the scotonon, secretion with below). The value of 3.5 hr, therefore, represents the of the eclosion hormone takes place. "minimal gate". The number of moths that emerged during Since the maximal accuracy is dependent upon the com- this 3.5-hr gate divided by the total number of moths was pletion of a process early in the free-running cycle, I have sub- taken as a measure of the relative accuracy of the clock divided the scotonon into two periods: the synchronization under the particular regimen. Accuracy was then expressed period and the dark decay period. The synchronization period on a scale from 0 to 1. is comprised of the events that lead to maximal accuracy of Under the photoperiod conditions listed in Fig. IC-I, the the clock; it encompasses the first 2 hr of the scotonon (i.e., eclosion clock initiates a free-running cycle at lights-off. In the approximately the time necessary for 50% of the population 23L: iD regimen, the clock free-ran for only 1 hr before the to become maximally accurate). The dark decay period in- onset of light; its accuracy, computed as described, was there- cludes the events that occur during the remaining 20 hr of the fore 0.38. At the other extreme, during the first cycle of con- scotonon. It may be noted (Fig. 3) that there is a sudden shift tinuous darkness (Fig. 1A), the clock free-ran for the entire in the response of the clock to lights-on at the outset of the 22-hr period with an accuracy of 0.96. Similarly, the relation- dark decay period. ship of accuracy to the completion of successive portions of The photonon the free-running cycle is shown in Fig. 4. It is obvious that the I have used the term photonon to describe the kinetics of the completion of an event that occurs early in the free-running clock after the onset of light. Fig. 3 defines the cycle is necessary for the attainment of maximal accuracy. relationship of the photonon to the scotonon. Thus, in photoperiod regimens, The scotonon the clock cycle begins with the initiation of the scotonon. The To facilitate a discussion of the mechanism of the eclosion onset of light then switches the clock to the photonon. The clock, I have used the term scotonon to refer to the sum total cycle terminates at the end of the photonon, whereupon the of processes that comprise one complete cycle of the free- clock stops. It restarts at the next lights-off signal, etc. Un-

1.0 * A 12L:12D DD ' i

0.8

X~~~~' 7/0'' 0i0,,, ,,, 0.6 0L

0.4

m 1~~2 24 36 48 60 72 84 7L:7D-'{LLHUSATRIGS-F 0.2 z

0.0

l S- 0 4 8 12 16 20 24 36 48 60 72 84 96 PORTION OF FREE-RUNNING CYCLE COMPLETED PRIOR TO LIGHTS- HOURS AFTER LIGHTS-ON ON (HOURS AFTER ONSET OF CYCLE)

FIG. 2. The eclosion of Pernyi moths exposed to a photo- FIG. 4. The curve describes the accuracy of the eclosion period regimen and subsequently transferred to constant con- clock as a function of the amount of the free-running cycle com- ditions: A., 12L: 12D regimen to continuous darkness; B., 17L: 7D pleted prior to the photophase. Maximal accuracy is attained in regimen to continuous darkness; C., 17L: 7D regimen to con- all moths essentially after the first 4 hr of the cycle have been tinuous light. completed. Downloaded by guest on September 28, 2021 598 Zoology: J. W. Truman Proc. Nat. Acad. Sci. USA 68 (1971)

A dictated by the scotonon at the time of interruption by light, and decays until [S] becomes equal to So. The photonon is de- void of a rising limb; therefore, if So is reached in the light, the clock stops. But, if So is reached in the dark, a new scotonon is initiated. Under most circumstances light is necessary to trigger, but not to sustain, the photonon. In all cases, as [SI approaches So, the eclosion hormone is released. Fig. 6 shows the application of the scheme to the last three days of adult development. In a given regimen, the clock IS] undergoes daily fluctuations characteristic of that photo- period. During the day of eclosion, the endocrine system be- comes "plugged-in" and the eclosion hormone is released. When a lights-on signal arrives during the decay of the scot- ..a. . a. . onon, I would expect little variation in [SI and, thus, a cor- '0 8 16 24 respondingly uniform gate width. By contrast, when lights-on H 0 UR S occurs during the rising limb, I would anticipate a much S C 0 T 0 N 0 N greater variation in the concentration of S and, consequently, a much broader peak. This scheme, therefore, provides a B simple description of the interaction of the eclosion clock with photoperiod regimens with respect to both the time and the H 0 U R S width of the eclosion gate. 0 8 16 Sd Interpretation of the IL:23D regimen In Fio. 6, we see that in photoperiods from 23L: iD to 4L: 20D the Pernyi eclosion clock initiates a cycle at lights-off, [s]

P H O T O N O N FIG. 5. A schematic representation of the components of the Pernyi eclosion clock. Both are represented as the fluctuation of a hypothetical substance (S) versus time. The scotonon occurs only during darkness. The photonon is triggered by interruption of the scotonon by light; Sd is the concentration of S reached by the scotonon at the time of the interruption. The clock stimu- lates the release of the eclosion hormone as S approaches So.

like the case with the scotonon, which, as we have seen, is greatly affected by the onset of light, there is a wealth of evi- dence from Drosophila that indicates that the photonon, under nearly all circumstances, is insensitive to interruption by darkness (4).

A schematic representation of the eclosion clock As seen in Fig. 5, the photoperiod-sensitive portion of the eclo- sion clock can be most simply represented as a fluctuation in concentration of a hypothetical substance (S) as a function of time. The kinetics of the scotonon is represented by a saw- -p I-..' tooth curve; the 2-hr rising limb and the 20-hr falling limb FIG. 6. The behavior of the eclosion clock during days 17, 18, correspond to the synchronization period and the dark decay and 19 of adult development under photoperiod conditions rang- period, respectively. When [S] reaches So, a new scotonon can ing from 23L: 1D to 4L:20D. On day 19 of development, the begin, but only in darkness. The shape of the photonon is eclosion clock becomes plugged into the endocrine system and then determined by the relationship given in Fig. 3. The the eclosion hormone is released. The mean behavior of the photonon is initiated at Sd, where Sd is the concentration of S population is represented by the curve. Downloaded by guest on September 28, 2021 Proc. Nat. Acad. Sci. USA 68 (1971) Silkmoth Eclosion and a Circadian Clock 599

switches to the photonon kinetics at lights-on, and then stops and comprised of two components. These parts, the scotonon during the final portion of the photophase. As the scotophase and photonon, respond to light and darkness in a defined is increased, the overall duration of the clock cycle (scotonon manner and can account for the variation of the time and the plus photonon) also increases, at the expense of the interval width of the eclosion peak with photoperiod. This scheme during which the clock is stopped. The extreme of this trend derived from the "hour-glass" behavior of the Pernyi clock should be seen in a certain critical photoperiod regimen in may also be applicable to regimens in which the clock runs which the clock is restarted as soon as it stops. In this reg- continuously. inen, and those having even longer scotophases, the clock would appear to oscillate continuously through successive I thank Profs. C. M. Williams and L. M. Riddiford for helpful photoperiod cycles. Thus, for the "hour-glass" to act as a con- discussions during this study and the preparation of the manu- script. This investigation was supported in part by an NSF tinuous oscillation would only require that the photonon end predoctoral fellowship to me and by NSF grants GB-7966 (LMR) in darkness. and GB-7963 (CMW). Consider, for example, the 1L: 23D photoperiod shown in Fig. 1B. Under these conditions, the male Pernyi release the 1. Bunning, E., The Physiological Clock, 2nd ed. rev. eclosion hormone prior to lights-on, whereas release by the (Springer-Verlag, New York, 1967). females occurs 2.5-hr later, in the early part of the succeeding 2. Pittendrigh, C., V. Bruce, and P. Kaus, Proc. Nat. Acad. Sci. USA, 44, 965 (1958). scotophase. Adapting this data to the "hour-glass" mech- 3. Pittendrigh, C. S., Cold Spring Harbor Symp. Quant. Biol., anism, I would assume that the scotonon began early in the 25, 159 (1960). scotophase (after the end of the preceding photonon). Late in 4. Pittendrigh, C. S., in Circadian Clocks, ed. J. Aschoff the course of the scotonon, the clock signaled the release of the (North-Holland Publishing Co., Amsterdam, 1965), p. 277. eclosion hormone in the males. But, before this event oc- 5. Pittendrigh, C. S., Z. Pflanzenphysiol., 54, 275 (1966). 6. Truman, J. W., and L. M. Riddiford, Science, 167, 1624 curred in the females, lights-on occurred and a photonon was (1970). initiated. This carried over into the succeeding dark period 7. Truman, J. W., in Biochronometry, ed. M. Menaker and, at its termination, the females released the eclosion hor- (National Academy of Sciences, Washington, D.C.), in press. mone. The termination of the photonon during darkness would 8. Williams, C. M., and P. L. Adkisson, Biol. Bull., 127, 511 (1964). then allow for the immediate initiation of a new clock cycle. 9. Williams, C. M., Symp. Soc. Exp. Biol., 23, 285 (1969). Thus, we see that in the majority of photoperiod regimens, 10. Truman, J. W., Ph.D. dissertation, Harvard University the daily cycle of the Pernyi eclosion clock is discontinuous (1970). Downloaded by guest on September 28, 2021