Proc. Natl. Acad. Sci. USAVol. 84, pp. 1694-1698, March 1987Neurobiology

The suprachiasmatic nuclei contain a tetrodotoxin-resistantcircadian pacemaker

(circadian rhythms/hypothalamus/mini-osmotic pumps/Nat channel/action potentials)

WILLIAM J. SCHWARTZ*, ROBERT A. GROSSt, AND MATTHEW T. MORTONtNeuroendocrine Research Laboratory, Neurology Service, Massachusetts General Hospital and Harvard Medical School, Boston, MA 02114

Communicated by Colin S. Pittendrigh, November 17, 1986

ABSTRACT Tetrodotoxin was infused into the suprachi-asmatic nuclei of unanesthetized and unrestrained rats contin-uously for 14 days. The internal timekeeping mechanism of thecircadian pacemaker in the nuclei continued to oscillate unaf-fected by this treatment, although the toxin reversibly blockedfunction of both the input pathway for pacemaker entrainmentand an output pathway for expression of the circadian drinkingrhythm. Thus, Na+-dependent action potentials appear neces-sary for entrainment and expression of overt circadianrhythms, but they do not seem necessary for the pacemaker tokeep accurate time. The experimental approach presented inthis paper is useful because it allows systematic assessment anddistinction of the input, pacemaker, and output components ofa mammalian circadian timekeeping system in vivo.

The suprachiasmatic nuclei (SCN) in the anterior hypothal-amus appear to be the site of an endogenous circadianpacemaker in mammals (1). The nuclei receive retinal inputsfor entrainment to the environmental light-dark cycle andgenerate neural outputs for expression of overt, measurablerhythms. Two complementary measures of SCN activityhave helped to establish that the nuclei contain a functioningcircadian pacemaker. These two properties, in vivo glucoseutilization (2, 3) and unit discharge rates (4, 5), exhibitcircadian rhythmicity. SCN energy metabolism and electricalactivity are both elevated during the day and depressedduring the night in nocturnal and diurnal mammals (6, 7).However, the most recent investigations using these two

assays have generated some unexpectedly discordant data.On one hand, the rhythm of SCN metabolic activity appearsin fetal rats 72 hr before birth (8, 9). Such prenatal pacemakerfunction antedates the postnatal maturation of input andoutput pathways for photic entrainment and expression ofovert circadian rhythms (10). On the other hand, when SCNaction potentials are recorded in hypothalamic slices ob-tained from 7-, 11-, 14-, and 21-day-old rat pups, a circadianrhythm is observed only in those slices from the 14- and21-day-old animals (11). Although other interpretations arepossible, it seems that the circadian rhythm ofSCN unit firingrates appears weeks after the rhythm of SCN energy metab-olism first begins in utero.To resolve this discrepancy, we propose that the action

potentials recorded in the SCN are not a part of the internaltimekeeping mechanism of the circadian pacemaker; rather,the electrical impulses function to couple the pacemaker to itsinput and output pathways. We have tested this idea bychronically infusing tetrodotoxin (TTX) into the SCN ofunanesthetized, unrestrained rats. TTX selectively and re-versibly blocks voltage-dependent Na+ channels in axons,inhibiting the generation of action potentials without affectingresting membrane potential, K+ currents, Na+ pump mech-

anism, or local depolarization of postsynaptic membranes(12, 13). Importantly, in vitro recordings in hypothalamicslices demonstrate that SCN action potentials are abolishedwhen TTX is added to the bath (14).

MATERIALS AND METHODS

Animals. Adult male Sprague-Dawley rats were housedindividually in clear plastic cages. Twelve cages were con-tained within a well-ventilated, light-proof environmentalcompartment located in an animal facility with temperaturethermostatically controlled (24 ± 10C). Light was provided by15 W cool white fluorescent tubes delivering an intensity of600 lux at the mid-cage level. During darkness, 15 W safelights with dark red (series 2) filters remained on in the facilityto allow for routine care. Purina rat chow and water werefreely available and replenished once every 6 to 7 days atirregular hours.Some rats were blinded by bilateral orbital enucleation.

Animals were fully anesthetized with ether, periorbital tissuewas dissected carefully with iris scissors, optic nerves werecut, and eyes were removed. The entire procedure wasaccomplished in <20 sec.Cannula Construction and Implantation. Guide cannulae

were constructed from 20-gauge stainless steel intravenousneedles ground flat at each end (total cannula length of 15mm). Stylets were manufactured from 24-gauge stainlesssteel tubing. Each was ground flat at the inserted end and bentat the other end to form a 5-mm-long handle. Infusioncannulae were made from 25-mm lengths of 24-gauge stain-less steel tubing. The inserted end was beveled at a 450 angle,whereas the opposite end was bent to form a smooth curveso that the tip of the infusion cannula projected 2 mm beyondthe end of the guide cannula when fully inserted therein. Theexposed, bent portion of the infusion cannula was capped bysoldering it to a 3- to 4-mm length of 20-gauge stainless steeltubing.The day before surgery, mini-osmotic pumps (model 2002,

Alza, Palo Alto, CA) were filled with infusate diluted inartificial cerebrospinal fluid (CSF) containing 140.0 mMNaCl, 3.5 mM KCl, 1.3 mM CaCl2, 1.0 mM MgCl2, and 5.0mM Hepes buffer at pH 7.2-7.3. Osmolality of the artificialCSF was 285 mosmol per liter ofH20, determined by freezingpoint depression. The pumps were primed by overnightimmersion in 0.9% NaCl.

Rats weighing -300 g were anesthetized with sodiumpentobarbital (50 mg/kg i.p.), treated with atropine sulfate

Abbreviations: SCN, suprachiasmatic nuclei; TTX, tetrodotoxin;CSF, cerebrospinal fluid.*Present address: Department of Neurology, University of Massa-chusetts Medical School, 55 Lake Avenue North, Worcester, MA01605.tPresent address: Department of Neurology, University of MichiganMedical School, Ann Arbor, MI 48109.iPresent address: Department of Neurology, Veterans Administra-tion Medical Center, 4150 Clement St., San Francisco, CA 94121.

1694

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

15, 2

020

Proc. Natl. Acad. Sci. USA 84 (1987) 1695

(0.4 mg/kg i.m.) and long-acting penicillin (150,000 unitsi.m.), and placed in a stereotaxic apparatus. The dorsalaspect of the parietal bones was exposed and cleaned. Fourholes were drilled in the parietal bones using a dental burr,two holes on each side of the sagittal suture, for placement ofsmall jewelers' screws that helped provide a foundation foranchoring the guide cannula to the skull with acrylic dentalcement. The guide cannula-stylet assembly was inserted intothe brain, aimed at the SCN (AP +7.6 mm anterior to earbars, L 0.0 mm in the midline, V -7.0 mm from the skull,incisor bar at the interaural line), and anchored with dentalcement. A 4.5-cm length of polyethylene tubing (PE 60) wassleeved over the 20-gauge cap of the infusion cannula, and thecatheter-cannula assembly was tilled with artificial CSFbefore it was attached to the flow modulator of a mini-osmotic pump. The stylet of the guide cannula was removed,and the infusion cannula was placed inside the open guide andcemented. The pump was implanted subcutaneously betweenthe scapulae. Artificial CSF in the cannula and catheter waspumped for =48 hr before the contents of the mini-osmoticpump were infused into the brain at the supplier's specifiedrate of 0.5 ,ul per hr for 14 days.At the conclusion of the experiment, each animal was

anesthetized with pentobarbital and perfused through theheart with ice-cold saline followed by 4% paraformaldehydein 0.1 M phosphate buffer at pH 7.3. After overnightimmersion in 20% sucrose as a cryoprotectant, the brain wascut on a cryostat, and 20-gm serial coronal sections werestained with cresyl echt violet to determine cannula locationand the extent of any infusion-induced damage.Measurement of Rat Drinking Activity. Leads from stain-

less steel grids on the cage floors and from the metal spoutsof drinking tubes were wired to drinkometer relays (LafayetteInstruments, Lafayette, IN) so that each lick of water by theanimal triggered a pen deflection on an Esterline-Angusevent recorder. The raw data were scored as number ofdrinking bouts per 30-min interval (values ranged from 0 to 3)and plotted as actograms-that is, drinking activity onsuccessive days was plotted vertically from top to bottom andthen "double plotted" to facilitate visual inspection: each24-hr record was photocopied and shifted up 1 day so that dayn was followed by day n + 1 horizontally from left to right.The free-running circadian period ('r) was calculated from theslope of a visually fitted line through successive daily activityonsets, as described previously (15). Averages for groups ofanimals are given as mean ± SEM.

RESULTSThe polyethylene tubing joining the implanted pump to theinfusion cannula remained intact for the entire 14-day infu-sion period in over 80% of the animals. The mortality rate forcontrol rats receiving infusions of artificial CSF alone was18%. Animals tended to die early (within 2 to 3 days aftersurgery) or weeks later, when autopsy often disclosed ab-scess formation at the infusion site. When TTX was infusedat a concentration of 10 AM, all animals died (n = 5).Mortality fell to 17% at a concentration of 1 AM; this was thedose used in all studies described below.An actual implantation is shown in Fig. 1. In other rats

receiving either artificial CSF or TTX, the SCN were par-tially damaged (usually the dorsal aspect) by the infusions.Lesions had volumes of -0.3 mm3 each.TTX Uncouples an Output Pathway for Expression of Overt

Rhythms. As a first test of our hypothesis, we predicted thatthe circadian pacemaker in the SCN would continue itsaccurate timekeeping even when action potentials werepharmacologically suppressed by TTX. However, since theSCN drives overt rhythms via neural connections (4, 16),TT7X treatment ought to uncouple the pacemaker's output

Rat 8.11

1MM

FIG. 1. Coronal rat brain section showing infusion cannula trackinto the SCN.

pathway; rhythms normally driven by the SCN should berendered arrhythmic.

Rats were entrained to a 12 hr: 12 hr light-dark schedule forat least 2 weeks before they were blinded. Blinded rodents donot entrain to environmental illumination cycles, and there isno evidence for extraocular photoreception (17, 18). Free-running circadian rhythms of drinking activity were nextrecorded for at least 10 days in constant environmentalillumination. Cannula-pump assemblies were then inserted,and free-running circadian rhythms of drinking activity wererecorded during the infusions and for :=3 to 4 weeks after thepumps were exhausted.As expected, drinking activity remained rhythmic (Fig. 2A)

in control rats receiving infusions of artificial CSF alone (n =5). For the population of animals, r was 24.07 ± 0.01 hr be-fore infusion, and 24.18 ± 0.02 hr after infusion. When TTXwas infused into the SCN (n = 6), behavioral arrhythmicityoccurred in all rats by visual inspection of the records (Fig.2B) or by statistical analysis of the data (no circadianperiodicity by a linear-nonlinear least-squares multiple peri-odic regression analysis program) (19). After the infusionsended in five of the six animals, overt drinking rhythmsresumed with rs unaltered from those expressed preinfusion(24.10 ± 0.03 hrpreinfusion vs. 24.09 ± 0.03 hrpostinfusion).In some animals, the total number of licks over 24 hrappeared diminished after TTX treatment. In one of the sixTTX-treated rats, the SCN were completely lesioned, anddrinking activity remained arrhythmic. In each of the otherfive rats, the phase of the restored rhythm was that predictedby extrapolation of the is of the original (preinfusion)rhythms. That is, no measurable phase shift occurred in anyof the treated animals, whether they had been previouslyentrained to a light-dark or to a reversed, dark-light illumi-nation schedule. Thus, the underlying circadian pacemaker inthe SCN continued to keep accurate time, its oscillatormechanism unperturbed by the chronic infusion of TTX orthe resulting behavioral arrhythmicity. However, the treat-ment reversibly inactivated an output pathway for overtrhythm expression.Behavioral arrhythmicity did not occur when TTX was

chronically infused outside the SCN. This is consistent withthe fact that no single, discrete knife cut outside the SCNresults in drinking arrhythmicity; simultaneous interruptionof lateral, dorsal, and caudal SCN efferents is required (20).The cannula was lateral to the nuclei in one rat, anteroventraland in the subarachnoid space in another animal (Fig. 3A),and anterodorsal in the preoptic area in four rats (Fig. 3B).Some of the rats with implantations close to the SCN in thepreoptic area appeared to become arrhythmic near the end ofthe infusion (i. 3B). Thus, abolition of rhythmic drinkingactivity probably required that TTX covered the entire SCN,not just a portion of the nuclei. This is similar to theobservation that partial destruction of the SCN has only

Neurobiology: Schwartz et al.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

15, 2

020

1696 Neurobiology: Schwartz et al.

A BClock Time Rat 6.3

0600 0600

RatClock Time

0800 2000 0800

Days' i '-.e,;;

...re.a.TTX20 ; :.adit;;3

FIG. 2. Actograms of two blinded rats with cannulae located in the SCN. Artificial CSF (A) or TTX (B) was continuously infused for the14 days bracketed by arrows.

minor effects on drinking rhythms; complete lesions arerequired for arrhythmicity to occur (21-23).TTX Uncouples the Input Pathway for Entrainment to

Light-Dark Cycles. If our interpretation of the previousexperiments is correct, then chronic infusion ofTTX into theSCN should also prevent entrainment of the circadian pace-maker to the external light-dark cycle. In the hypothalamicslice, TTX eliminates SCN field potentials evoked by in vitroelectrical stimulation of the optic nerve (24). Thus, pharma-cological suppression of SCN action potentials ought todisconnect the input pathway from the circadian pacemaker,inhibiting re-entrainment of the free-running pacemaker to a

phase-shifted light-dark cycle.Rats were first entrained to a 12 hr: 12 hr light-dark

schedule for at least 2 weeks. Cannula-pump assemblies werethen inserted, and rodents were maintained in constantenvironmental darkness for the first 8 days of the 14-dayinfusion period. During the last 6 days ofthe infusion, animalswere exposed to a 12 hr: 12 hr light-dark cycle progressively

A BClock Time Rat 6.8 Clock Time Rat 3.3

0600 1800 0600 0600 0800 2000 0800 0800

~~~~~~~Days

10 o -

20 M~~~~~T20~

FIG. 3. Actograms of two blinded rats with cannulae located

outside the SCN. TTX was continuously infused for the 14 days

bracketed by arrows.

delayed by three successive 4-hr delays or two successive6-hr delays (A1D) until a completely reversed cycle wasattained. At the end of the infusion, rats were again placed inconstant darkness, and free-running circadian rhythms ofdrinking activity were monitored for -1 month.

Overt drinking rhythms were successfully phase-shifted bythe imposed light-dark cycle (Fig. 4A) in control rats (n = 3)and in rats with TTX infusions outside the SCN (n = 5),although the phase shift obtained in these animals (6.6 0.3hr and 6.2 ± 0.8 hr, respectively) was less than that observedin three control animals whose polyethylene tubing brokemid-infusion (9.0 ± 1.0 hr). When TTX was infused into theSCN (n = 12), extensive (>50%) SCN damage in six animalsresulted in postinfusion arrhythmicity or drinking activitywith unstable or aberrant rs less than 24 hr. In the remainingsix animals, TTX prevented the phase-resetting action of light(Fig. 4B), with only a 1.9 + 0.7-hr phase shift for thepopulation of animals. That is, the phase of the restoredrhythm was nearly that expected had the rhythm continuedits free-run despite application of the reversed light-darkcycle. In many of the animals, light applied during the TTXinfusion acutely suppressed drinking activity (Fig. 4B), eventhough the underlying circadian pacemaker in the SCNcontinued to oscillate undisturbed (judged by the phase ofthepostinfusion free-running drinking rhythm). Previous evi-dence had suggested that ambient light may bypass thepacemaker to directly influence overt rhythms by such"masking" (25-27).To ensure that the TTX effect was fully reversible, we

prepared a final group of animals and subjected them to thesame paradigm described above. At the end of this trial,however, the rats were re-exposed to a 12 hr: 12 hr light-darkschedule progressively delayed in a manner identical to thatimposed earlier while TTX was being infused. In rats withTTX infusions outside the SCN (n = 4), overt drinkingrhythms were phase-shifted both during infusion and laterduring recovery (5.9 ± 0.7 hr and 6.2 ± 0.5 hr, respectively).When TTX was infused into the SCN (n = 9), five animalsexhibited postinfusion arrhythmicity or drinking activity withunstable or aberrant ms less than 24 hr. In the remaining four

3.6

0800

Proc. Natl. Acad. Sci. USA 84 (1987)

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

15, 2

020

Proc. Nati. Acad. Sci. USA 84 (1987) 1697

AClock Time

BRat 8.4

070c

10

0700 1900 0700 0700

ys *- Lht-Dark

-V.~~~~~~A LCycleU* *; b4* Ia ' ;' *' .' Constant

Darkness

20 Light-DarkiA-..-12 hr

:i .

| ' : . ....I.e *.30 I. Darkness. * i. . .. ..a3 :

d .*3.7 .: .||49 l

| .|; U as.*40

* ins d;l" rlzf;|@

60-' *t~ Sb lj ***4'*w5% 'a , e...7. J . I . .1..6a6. I| . l.

.4.*

I

60-; * Is>,@ o I.s

70LII : J.11 -| A

Clock Time Rat 8.110 1900 0700 0700

* * @Light-DarkCycle

d. . |

:J . I

Constant

_ Darkness

* |. ~ ..; \ ,::' ? Light-Dark* . l~. l . "@ I. d. .~A '-f i|, :<' ih|;;I 'jai:.. b P - 12 hr

i4'>I I C\X1 5 .+qW 1,\,\>,"Darkness

4sr IIdjEaa I.

IL A.LI I...

_ I a I l. U a

.81' as

L4 J@.

;z4'~-LIJLL I Isal

.; (Lr If"@*

Id~ L. d6te

70 -

FIG. 4. Actograms of two rats with cannulae located in the SCN and exposed to a phase-shifted light-dark cycle. Artificial CSF (A) or TTX(B) was continuously infused for the 14 days bracketed by arrows.

animals, TTX prevented the phase-resetting action of light(1.9 + 0.6-hr phase shift for the population of animals). Afterrecovery (Fig. 5), these animals could be re-entrained andtheir rhythms phase-shifted (5.7 ± 0.8 hr) by the light-darkcycle. Thus, the TTX-induced insensitivity to light was not a

permanent deficit but a reversible inactivation of the inputpathway. Parenthetically, circadian rhythmicity was notrestored to any of the arrhythmic animals upon re-exposure

to diurnal lighting. This observation discounts the theoreticalpossibility that the arrhythmicity was due to an intactcircadian pacemaker held motionless because its limit cycleoscillation had been driven to the "singularity point" (28).

DISCUSSION

Our results indicate that TTX blocked the function of SCNinput and output pathways without affecting the actualoscillatory mechanism of the circadian pacemaker in thenuclei. Thus, although Na'-dependent action potentials doappear necessary for photic entrainment and overt expres-

sion of circadian rhythms, they do not seem to be required forthe pacemaker to keep accurate time. This hypothesizedorganization for a mammalian circadian pacemaker is com-patible with that proposed for the pacemaker in the eye ofAplysia. This invertebrate oscillator generates a circadianrhythm of optic nerve impulses, but the discharges per se arenot part of the pacemaker mechanism (29-31). Similarly, thecircadian pacemaker in silkworm pupae is resistant to sys-temic TTX at doses causing flaccid paralysis (32).

Presumably, the SCN pacemaker is not confined entirely toa single neuron within the nuclei. Therefore, some mecha-nism(s) for intercellular communication are required to syn-

thesize a precise circadian oscillation from the disparateactivities of a number of individual cells. Although Na'-dependent action potentials do not appear to be part of sucha mechanism, SCN dendrites might generate Ca2'-dependentspikes, as described elsewhere in vertebrate (33, 34) andinvertebrate (35) brain. Alternatively, SCN "local circuit"

neurons might interact by graded Ca2l-dependent release ofneurotransmitters (36, 37) rather than by firing all-or-nonespikes. Finally, it is possible that communication betweenpacemaker cells might be ephaptic (38) rather than synaptic;interestingly, the bulk of SCN synapses are not formed untilthe first few weeks of postnatal life (39).Two methodological difficulties in our experimental para-

digm need to be noted. First, chronic infusions of eitherartificial CSF or TTX into the SCN altered the amount ofdrinking activity in some animals. Perhaps the osmolality ofthe infusate affected the discharge rates of neighboring neuralfibers from the anteroventral third ventricle that regulatethirst (40). Future studies will determine whether otheranimal models (e.g., hamster locomotor activity) may provesuperior to rat drinking activity in this regard.

Second, partial damage to the SCN was frequently asso-ciated with the chronic infusions. We do not believe, how-ever, that our results can be explained solely on the basis ofincomplete SCN lesions. Of note, TTX infusion did not causepartial lesions in every case (cf. Fig. 1); on the other hand,incomplete lesions were found in control rats receivinginfusions of artificial CSF alone. Moreover, a reportedfeature of partial lesions of the SCN is a postoperative r ofless than 24 hr (41-44); this we did observe in some of ouranimals with extensive damage to the SCN, and we did notinclude them in the analysis. Finally, the TTX-inducedarrhythmicity reported here could be terminated by cuttingthe polyethylene tubing that joins the mini-osmotic pump toinfusion cannula (data not shown), suggesting that it is thecontinued flow of infusate (not the creation of a partial lesion)that is responsible for the TTX effect.An advantage of the experimental approach outlined here

is that it permits systematic assessment and distinction of theinput, pacemaker, and output components of a mammaliancircadian timekeeping system in vivo. We anticipate that thisstrategy will also be fruitful for testing reversible pharmaco-logical probes other than TTX. This paradigm and oui- dataclearly illustrate the principle that a treatment can abolish an

Neurobiology: Schwartz et al.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

15, 2

020

1698 Neurobiology: Schwartz et al.

Clock rime Rat 10.70700 1900 0700 0700

Days ] IlF.... LhD

...J1.-Cycle

10 Constant|TX 2.0U

t 13k 1 P- 2Darkness

Light-Dark

20~ ~ ~~ A4 -12 hr

FIG. 5. Actogram of a rat with cannula located in the SCN andexposed to a phaseL-shifted light-dark cycle both during and afterTTXinfusion. TTX was continuously infused for the 14 days bracketed byarrows.

overt circadian rhythm merely by uncoupling an outputpathway from the still oscillating pacemaker; arrhythmicityof a measured function may represent loss of the "hands" ofthe pacemaker rather than damage to its "gears."

We thank Dr. Martin Zatz for critical and helpful discussions andMs. Caroline Coletti for technical help. W.J.S. is supported byNational Institute ofNeurological and Communicative Disorders andStroke Teacher Investigator Award K07 NS00672, National InstituteofNeurological and Communicative Disorders and Stroke Grant R01NS23029, and March of Dimes Basil O'Connor Research Grant5-433.

1. Moore, R. Y. (1983) Fed. Proc. Fed. Am. Soc. Exp. Biol. 42,2783-2789.

2. Schwartz, W. J. & Gainer, H. (1977) Science 197, 1089-1091.3. Schwartz, W. J., Davidsen, L. C. & Smith, C. B. (1980) J.

Comp. Neurol. 189, 157-167.4. Inouye, S. T. & Kawamura, H. (1979) Proc. Natl. Acad. Sci.

USA 76, 5962-5966.5. Inouye, S. T. & Kawamura, H. (1982) J. Comp. Physiol. A

146, 153-160.

6. Schwartz, W. J., Reppert, S. M., Eagan, S. M. & Moore-Ede,M. C. (1983) Brain Res. 274, 184-187.

7. Sato, T. & Kawamura, H. (1984) Neurosci. Res. 1, 45-52.8. Reppert, S. M. & Schwartz, W. J. (1983) Science 220, 969-

971.9. Reppert, S. M. & Schwartz, W. J. (1984) J. Neurosci. 4,

1677-1682.10. Davis, F. C. (1981) in Handbook ofBehavioral Neurobiology,

ed. Aschoff, J. (Plenum, New York), Vol. 4, pp. 257-274.11. Shibata, S., Liou, S. & Ueki, S. (1983) Neurosci. Lett. 43,

231-234.12. Kao, C. Y. (1966) Pharmacol. Rev. 18, 997-1049.13. Narahashi, T. (1974) Physiol. Rev. 54, 813-889.14. Sugimori, M., Shibata, S. & Oomura, Y. (1984) Neurosci.

Abstr. 10, 294.15. Pittendrigh, C. S. & Daan, S. (1976) J. Comp. Physiol. A 106,

223-252.16. Stephan, F. K. & Nunez, A. A. (1977) Behav. Biol. 20, 1-16.17. Nelson, R. J. & Zucker, I. (1981) Comp. Biochem. Physiol. A

69, 145-148.18. Underwood, H. & Groos, G. (1982) Experientia 38, 1013-1021.19. Rummel, J., Lee, J. K. & Halberg, F. (1974) in Biorhythms &

Hormone Reproduction, eds. Ferin, M., Halberg, F., Richart,R. M. & Vande Wiele, R. L. (Wiley, New York), pp. 53-82.

20. Nunez, A. A. & Stephan, F. K. (1977) Behav. Biol. 20, 224-234.

21. Stephan, F. K. & Zucker, I. (1972) Proc. Natl. Acad. Sci.USA 69, 1583-1586.

22. Mosko, S. & Moore, R. Y. (1979) Brain Res. 164, 17-38.23. Robbins, A. & Stephan, F. K. (1983) Neurosci. Abstr. 9, 1070.24. Shibata, S., Oomura, Y., Hattori, K. & Kita, H. (1984) Brain

Res. 302, 83-89.25. Rusak, B. (1977) J. Comp. Physiol. A 118, 145-164.26. Dark, J. (1980) Physiol. Behav. 25, 863-873.27. Aschoff, J., Daan, S. & Honma, K. I. (1982) in Vertebrate

Circadian Systems, Structure and Physiology, eds. Aschoff,J., Daan, S. & Groos, G. (Springer, Berlin), pp. 13-24.

28. Winfree, A. T. (1980) The Geometry ofBiological Time (Spring-er, New York).

29. Jacklet, J. W. (1973) in Neurobiology of Invertebrates, ed.Salanki, J. (Hungarian Acad. Sci., Budapest), pp. 363-380.

30. Eskin, A. (1977) J. Neurobiol. 8, 273-299.31. Strumwasser, F., Alvarez, R. B., Viele, D. P. & Woolum,

J. C. (1979) in Biological Rhythms & Their Central Mecha-nism, eds. Suda, M., Hayaishi, 0. & Nakagawa, H. (Elsevier,Amsterdam), pp. 41-66.

32. Williams, C. M. (1968) Science 160, 444.33. Llinas, R. & Yarom, Y. (1981) J. Physiol. 315, 549-584.34. Mason, W. T. & Leng, G. (1984) Exp. Brain Res. 56, 135-143.35. Carrow, G. M., Calabrese, R. L. & Williams, C. M. (1984) J.

Neurosci. 4, 1034-1044.36. Shibata, S., Shiratsuchi, A., Liou, S. Y. & Ueki, S. (1984)

Neurosci. Lett. 52, 181-184.37. Thomson, A. M. (1984) Neuroscience 13, 761-767.38. van den Pol, A. N. (1980) J. Comp. Neurol. 191, 661-702.39. Lenn, N. J., Beebe, B. & Moore, R. Y. (1977) Cell Tissue Res.

178, 463-475.40. Andersson, B. & Rundgren, M. (1982) Annu. Rev. Med. 33,

231-239.41. Pickard, G. E. & Turek, F. W. (1983) Brain Res. 268, 201-210.42. Davis, F. C. & Gorski, R. A. (1984) J. Comp. Physiol. A 154,

221-232.43. Rietveld, W. J. (1984) in Biological Rhythms & Medications,

Annual Review of Chronopharmacology, eds. Reinberg, A.,Smolensky, M. & Labrecque, G. (Pergammon, Oxford), Vol.1, pp. 1-4.

44. Prosser, R. A. & Satinoff, E. (1985) Neurosci. Abstr. 11, 817.

Proc. Natl. Acad. Sci. USA 84 (1987)

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

15, 2

020