NEUROREPORT
RHYTHMS
Adenosine and caffeine modulate circadian
rhythms in the Syrian hamster
M. C. Antle, N. M. Steen and R. E. MistlbergerCA
Department of Psychology, Simon Fraser University, 8888 University Drive, Burnaby, BC V5A 1S6, Canada
CA
Corresponding Author
Received 13 June 2001; accepted 12 July 2001
Extracellular adenosine accumulates in some brain areas during
sleep deprivation. In Syrian hamsters, both sleep deprivation
and adenosine A1 agonists can inhibit phase shifts of circadian
rhythms to light at night. Sleep deprivation in the day (sleep
period) can shift circadian phase. We examined whether the
A1 agonist N-CHA mimics this effect. N-CHA (i.p. or i.c.) in
the mid-sleep period induced dose-dependent shifts similar to
those induced by 3 h sleep deprivation. The adenosine antagonist caffeine administered systemically at the mid-sleep period
induced arousal without shifts, and dose-dependently attenuated shifts to a 3 h sleep deprivation procedure (running in a
novel wheel). Adenosine may participate in resetting of the
circadian clock by manipulations of behavioral state. NeuroReport 12:2901±2905 & 2001 Lippincott Williams & Wilkins.
Key words: Adenosine; Caffeine; Circadian rhythms; N 6 -Cyclohexyladenosine; Sleep; Sleep deprivation
INTRODUCTION
Adenosine, a byproduct of cellular metabolism, functions
in the nervous system as a neuromodulator through
actions at four major receptor subtypes [1]. Adenosine has
been hypothesized to play a role in sleep regulation, based
on evidence that adenosine agonists increase slow-wave
sleep and decrease arousal, that adenosine antagonists,
such as caffeine and theophylline, increase arousal, and
that extracellular adenosine accumulates during spontaneous and extended wakefulness and dissipates during
subsequent sleep [1±4]. Short-term sleep deprivation (e.g.
3 h in the middle of the usual sleep period), by gentle
handling [5] or continuous locomotion (reviewed in [6]),
can induce large phase advance shifts of circadian rhythms
in the Syrian hamster. Sleep deprivation can also attenuate
phase shifts to brief light exposure at night [7]. Adenosine
A1 receptor agonists mimic this inhibitory effect [8,9],
raising the possibility that adenosine mediates one or more
effects of acute sleep deprivation on the circadian clock.
Here we report that an adenosine A1 agonist also mimics
the phase shifting effects of sleep deprivation, and that the
A1 receptor antagonist caffeine induces arousal without
phase shifts, and attenuates the phase shifting effect of
locomotor activity stimulated during the usual sleep period.
MATERIALS AND METHODS
Animals and housing: Young adult male hamsters (total
81; Charles River, PQ) were housed singly in standard
polypropylene cages equipped with 17 cm running wheels
monitored continuously by computer. Cages were maintained in ventilated isolation cabinets providing a 14:10 h
light:dark cycle.
0959-4965 & Lippincott Williams & Wilkins
Surgical procedures: Hamsters (n 31) were prepared
with indwelling cannulas (22 gauge, Plastics One, Roanoke,
VA) aimed at the suprachiasmatic nucleus (SCN), the site
of the light-entrainable circadian pacemaker. Stereotaxic
co-ordinates were 0.3 mm lateral from midline and 7.0 mm
below the skull surface at bregma, with the incisor bar at
2 mm below the interaural line.
Behavioral procedures: Animals received i.p. or intracerebral injections of the adenosine A1 agonist N 6 -cyclohexyladenosine (N-CHA, RBI), the adenosine antagonist
caffeine (RBI) or vehicle (50% DMSO/PBS) 6 h before the
time of usual dark onset (zeitgeber time (ZT) 6, where dark
onset is designated ZT12, by convention). Injection volumes were set at 1 ml/kg for i.p. injections, and at 0.5 ìl
delivered over 30 s for cannula injections. Lights were then
turned off for 3±4 days. Phase shifts were quanti®ed by
comparing the time of spontaneous activity onset (designated circadian time (CT) 12, by convention) on the second
day after drug treatment with the average time of activity
onset during the 5 days before drug treatment, as described elsewhere [5,7]. Phase changes were then compared
between drug and vehicle conditions.
Immunocytochemistry: To determine whether the adenosine agonist N-CHA mimics the effects of sleep deprivation
on basal fos expression in the SCN, hamsters received
injections of N-CHA (2 mg/kg, i.p.) at ZT6, and were
perfused transcardially with 50 ml PBS, followed by 4%
paraformaldehyde at ZT11.5. Brains were post®xed overnight, and then cryoprotected in 20% sucrose for 24 h.
Sections (50 ìm) were collected into PBS ®lled wells and
then received the following treatments, separated by PBSx
Vol 12 No 13 17 September 2001
2901
NEUROREPORT
Data analysis: The results were evaluated by repeatedmeasures ANOVA with Student±Newman±Keul's post hoc
analyses or paired t-tests. Pearson correlation coef®cients
were calculated to quantify relationships between phase
shifts and physiological variables. Values are reported as
mean s.d.
RESULTS
Pilot experiments established that systemic injections of NCHA at ZT6 induced phase advance shifts at doses of
2.5 mg/kg (mean shift 113 27 min, n 2) and 2.0 mg/kg
(50 24 min vs 25 22 min in response to vehicle alone;
paired t-test, t(11) 2.35, p , 0.05), but not at 1.5 mg/kg
(27 26 min). Following injections, hamsters were hypoactive for at least 24 h and hypothermic for 4 h. To minimize these acute side effects, N-CHA was microinjected
directly into the SCN area. These injections induced signi®cant, dose-dependent phase advance shifts (one-way RMANOVA, F(3,25) 25.9, p , 0.0001, Fig. 1a,b,e) at ZT6, and
signi®cant phase delay shifts at ZT18 (i.e. 6 h after the
usual time of lights off; paired t-test, t(9) 2.868, p , 0.01,
Fig. 1c±e). All cannula placements were within 400 ìm of
the SCN (measured by visual inspection of post-mortem,
Nissl stained brain sections), and there was no relation
between the position of the cannula and the size of phase
shifts to N-CHA (r 0.03 at ZT6, 0.25 ìg dose). Activity
was reduced to 8 12% baseline during the 24 h following
N-CHA (0.25 ìg) injections. Activity was also strongly
suppressed for a day or more after some vehicle injections
that were not associated with phase shifts (e.g. Fig. 1d).
Sleep deprivation by gentle handling (ZT6±9; [5]) or
wheel running (ZT4.5±7.5; [10]) signi®cantly suppresses
basal Fos levels within the SCN. If endogenous adenosine
participates in the phase resetting effect of sleep deprivation, then adenosine agonists may also suppress basal SCN
Fos expression. As predicted, N-CHA (2 mg/kg, i.p., n 4)
at ZT6 did signi®cantly suppress basal SCN Fos by comparison with the vehicle control (n 4; one-tailed t-test,
t(7) 2.28, p , 0.05; Fig. 2), although not to the same extent
observed following sleep deprivation [5]. Sleep deprivation
procedures induce larger shifts on average than those
observed here in response to N-CHA, suggesting that
phase shift magnitude is related to the degree of Fos
suppression.
The circadian clock is signi®cantly reset within 1 h
following a 3 h sleep deprivation from ZT6±9, as indicated
by a phase advance of the circadian rhythm of lightinduced c-fos expression in the SCN [5]. To determine if
clock resetting by N-CHA is similarly rapid, hamsters
2902
Vol 12 No 13 17 September 2001
Hours
12
0
24
(a)
(b)
(c)
(d)
(e) 300
*
200
Phase shift (min)
(0.3% Triton-X-100 in PBS) rinses; 0.3% H2 O2 in PBSx to
inactivate endogenous peroxidase, 10% normal goat serum
(NGS, Vector), 48 h at 48C in the primary fos antibody
(Oncogene Science, 1:80000 in 1% NGS-PBSx), 1 h in
secondary goat-anti-rabbit antibody (Vectastain Elite Kit,
Vector), 1 h in avidin±biotin complex (ABC, Vectastain
Elite Kit, Vector), diaminobenzidine (DAB) reaction for
5 min (0.04% DAB in Tris buffer 60 ìl 8% NiCl and 10 ìl
30% H2 O2 ). Tissue was mounted on slides, dehydrated and
coverslipped. Digital images were captured and analyzed
using MCID (Imaging Research Inc., St. Catherine's, ON,
Canada).
M. C. ANTLE, N. M. STEEN AND R. E. MISTLBERGER
*
100
0
19
20
5
4
12
10
2100
2200
0
*
0.05 0.25 0.5
0 0.25
CT6
ct18
Dose of N-CHA
Fig. 1. Actograms from a representative hamster that received either
N-CHA (diamonds, 0.25 ìg in 0.5 ìl 50%DMSO/saline) or vehicle control
(circles) at either ZT6 or ZT18. Each horizontal row represents 24 h,
with time in 10 min bins plotted from left to right. Vertical defections
from the horizontal line represent bins during which wheel revolutions
occurred. Shading represents lights-off period. (a) N-CHA at ZT6, (b)
Vehicle at ZT6, (c) N-CHA at ZT18, (d) Vehicle at ZT18. (e) Bar graphs
of mean ( s.e.m.) phase advance shifts to intra-SCN injections of vehicle
or N-CHA. Signi®cant at p , 0.01.
(n 5) received N-CHA (0.25 or 0.50 ìg in 0.5 ìl 50%
DMSO/saline, intra-SCN) or vehicle (n 3) at ZT6, and a
light pulse (15 min, 25 lux) at ZT11 (1 h before the
hamster SCN normally expresses Fos in response to light;
[11]). If N-CHA phase advances the clock by at least 1 h by
this time, then the light pulse should induce Fos in drug
but not vehicle-treated animals, and Fos-positive cells
should be concentrated ventrolaterally in the SCN, in the
spatial pattern characteristic of photically induced Fos
early in the active period. Both groups exhibited many Fospositive cells but these were concentrated (although not
NEUROREPORT
ADENOSINE A1 LIGANDS AFFECT CIRCADIAN CLOCK
ANOVA, F(3,27) 3.75, p , 0.05, Fig. 3a±c). The high dose
and vehicle control tests were repeated in an additional 30
animals, again yielding a signi®cant difference (mean shifts
23 7 min vs 80 13 min, respectively; paired t-test,
t(29) 5.348, p , 0.0001). There was no effect of caffeine
treatment on activity level during wheel con®nement
(paired t-test for the highest dose vs vehicle, t(41) 0.29,
p . 0.1).
con®ned) ipsilaterally along the cannula track, and did not
show the bilateral spatial distribution characteristic of
photic induction. Cell counts taken from representative
sections in the mid-caudal ventrolateral SCN, contralateral
to the injection site, showed no difference between groups
(56 10 vs 42 12 cell counts in vehicle and N-CHA
groups, respectively; t 1.5, p . 0.1).
The absence of Fos expression in the photic pattern in
the N-CHA group could indicate that the clock was not
shifted or that N-CHA has an extended inhibitory effect on
photic activation of SCN neurons. To test the latter alternative, two additional groups were processed for Fos
following N-CHA (0.25 ìg, n 3) or vehicle (n 3) at CT8,
and a light pulse at CT13, when the Fos response to light is
robust. Both groups again exhibited Fos associated with
the injection site, but only the vehicle-treated hamsters also
expressed a concentration of Fos-positive cells ventrolaterally in the SCN, in the pattern typical of photic stimulation
(75 42 vs 17 7 cell counts in vehicle and N-CHA groups,
respectively; t 2.35, p , 0.05) This result is consistent with
recent reports that N-CHA can inhibit photic input to the
clock [8,9], and further demonstrates that inhibition can be
prolonged (at least 5 h).
Resetting of behavioral rhythms and suppression of
basal Fos within the SCN by an adenosine agonist is
consistent with a role for adenosine in the induction of
phase shifts by acute sleep deprivation. We further tested
this hypothesis by determining whether an adenosine
receptor antagonist can inhibit phase shifting to sleep
deprivation procedures. The adenosine A1 receptor antagonist caffeine (75 mg/kg, i.p., delivered at ZT5.5) induced
substantial arousal (assessed at 1 min intervals by trained
observer), with behavioral sleep reduced to 6 2, 14 5
and 31 5 min, respectively, during each of the ®rst 3 h
following treatment, compared to hourly averages of
50 2, 56 1 and 55 1 min sleep following vehicle injections. Despite stimulating arousal, caffeine did not induce
phase shifts signi®cantly different from saline (1 19 min
vs ÿ2 23 min, paired t-test, t(10) 0.390, p . 0.10).
In a second test, hamsters were con®ned to novel
running wheels for 3 h beginning at ZT6, to stimulate
arousal by running. This procedure reliably induces large
(2±3 h) phase advance shifts in hamsters that run continuously [12]. Caffeine dose-dependently attenuated phase
shifts to this sleep deprivation procedure (one-way RM-
DISCUSSION
Previous work in Syrian hamsters has demonstrated that
phase shifts to light exposure early in the night can be
attenuated by the adenosine A1 agonist N-CHA [8,9] or by
short-term (6 h) sleep deprivation [7]. Given that sleep
deprivation is associated with increased levels of extracellular adenosine in at least some areas of the CNS [2],
adenosine may serve as a neurochemical signal by which
manipulations of behavioral state alter functional properties of the circadian clock. The results reported herein
provide additional support for this hypothesis by demonstrating that N-CHA mimics the clock resetting effect of
sleep deprivation, and has a prolonged inhibitory effect on
an immediate-early gene response of SCN cells to a photic
stimulus. Moreover, the A1 antagonist caffeine dose-dependently inhibited phase shifts to arousal with or without
continuous locomotor activity during the usual sleep
period.
Whether the effects of N-CHA and caffeine on clock
Hours
12
0
24
(a)
(b)
(c) 100
Phase advance (min)
Fig. 2. Coronal photomicrographs of fos-IR at the mid-caudal level of
the SCN (medial and ventral borders indicated by dashed lines). (a)
Vehicle control injections at ZT6. (b) N-CHA (2 mg/kg, i.p.) at ZT6.
Animals were perfused at ZT11.5. Scattered Fos-IR cells are evident in
N-CHA treated animal, but the number of cells and background Fos are
lower than in the control animal. Bar 100 ìm.
*
75
50
25
0
0
7.5
37.5
75
Dose of caffeine (mg/kg)
Fig. 3. Actograms from a representative hamster illustrating phase
shifts to 3 h wheel con®nement procedure at ZT6, following pretreatment at ZT5.5 with (a) PBS (circles) or (b) caffeine (diamonds, 75 mg/kg
i.p. in PBS). (c) Bar graph of mean ( s.e.m.) phase advance shifts to 3 h
of wheel con®nement following caffeine or vehicle injections, from eight
hamsters providing data at all doses. Signi®cant at p , 0.01.
Vol 12 No 13 17 September 2001
2903
NEUROREPORT
function are mediated by receptors within the SCN or at
sites afferent to the clock is a complex issue to be resolved.
Neuronal activity and metabolism within the SCN are
maximal during the subjective day, which corresponds to
the daily sleep period in the nocturnal Syrian hamster.
Neurotransmitters that phase advance the circadian clock
when administered to the SCN in the subjective day have a
predominantly inhibitory action on SCN neurons (e.g.
GABA, benzodiazepines, serotonin, neuropeptide Y)
[13,14]. Spontaneous locomotor activity is also associated
with suppression of SCN neuronal activity [15,16]. Adenosine and N-CHA, which inhibit neuronal activity via
postsynaptic A1 receptors, presumably share with these
other stimuli the property of hyperpolarizing SCN neurons, and this common effect on membrane potential may
be the means by which phase advance shifts are induced
in the subjective day.
Although exogenous adenosine agonists may mimic the
phase shifting effect of sleep deprivation by actions at SCN
A1 receptors, it is not clear that endogenous extracellular
adenosine within the SCN plays this role in vivo. Given
that neural activity within the hamster SCN is high during
the usual sleep period and is attenuated during periods of
locomotor activity, extracellular adenosine in the hamster
SCN is more likely to accumulate during sleep, and to
dissipate when hamsters are aroused. Although direct
evidence for this is lacking, this raises the intriguing
possibility that the daily rhythm of SCN neural activity,
and extracellular accumulation of adenosine, might normally contribute to the relative insensitivity of the circadian clock to light during the day, possibly via presynaptic
actions at retinal terminals [17]. These considerations also
suggest that we should consider sites outside of the SCN
were adenosine accumulation, and exogenous agonists,
might exert clock resetting effects similar to those that
follow sleep deprivation.
A candidate area is the basal forebrain. N-CHA administered directly to the SCN could diffuse into the basal
forebrain directly, or via leakage into the third ventricle.
Actions of N-CHA in this region would be expected to
induce sleep [18] and alter activity of cholinergic and other
neurons afferent to the SCN [19]. The suppression of
locomotor activity that followed intra-SCN infusions of NCHA supports the possibility of actions at a distance from
the injection site, although it is also possible that behavioral
suppression is a local effect. Wheel running in hamsters is
typically decreased by SCN lesions, and N-CHA may
simulate a reversible SCN ablation by inhibiting SCN
output.
The behavioral suppression induced by peripheral and
central adenosine injections had an intriguing similarity to
torpor, particularly at the higher systemic doses. Wheel
running was greatly suppressed, and the animals were
noticeably cool. During most observations, the animals
appeared asleep in a prone position that was distinct from
the typical sleep posture. Although arousal threshold
appeared elevated, this state was rapidly reversed when
the animals were disturbed, which enabled us to ensure
that the animals received the photic stimulus. Also, the
hamsters usually exhibited at least some spontaneous
wheel running during the period of maximal behavioral
suppression. Whether this was a true state of torpor
2904
Vol 12 No 13 17 September 2001
M. C. ANTLE, N. M. STEEN AND R. E. MISTLBERGER
triggered by hypothalamic A1 receptor stimulation remains
to be examined further. However, clock resetting was likely
independent of behavioral and metabolic suppression, for
several reasons. First, suppression of locomotor activity at
night, by restraint, induces small phase delays of circadian
rhythms [20] (Antle and Mistlberger, unpublished observations), rather than the large phase advances evident in the
present study. Second, vehicle injections also suppressed
activity in some cases, but did not produce large phase
shifts. Third, cooling can induce large phase shifts in
hamsters, but only if body temperature is reduced below
288 deg for 12 h or longer, which is a much more severe
hypothermia than that observed in the present study.
Moreover, only phase delays have been observed in
response to severe cooling [21].
While systemic injections of caffeine presumably access
A1 receptors within the SCN, caffeine almost certainly also
modulates neuronal activity at sites afferent to the clock.
An area of special interest is the raphe, as caffeine is
known to promote terminal release of serotonin [22].
Serotonergic cell groups in the midbrain raphe can modulate SCN neuronal activity via a direct projection from the
median raphe, and via an indirect projection from the
dorsal raphe to the intergeniculate lea¯et (IGL) [23], an
area of the thalamus thought to convey non-photic input to
the clock [6,24]. Conceivably, caffeine-enhanced release of
serotonin in the IGL might inhibit activity of IGL neurons
that project to the SCN, thereby preventing the full phase
resetting effect of wheel running stimulated in the usual
sleep period [25]. Such indirect effects may combine with
direct effects of caffeine on A1 receptors in the SCN. At
high doses (. 30 mg/kg in rats), caffeine inhibits locomotor
activity, but we did not see signi®cant suppression of
running in hamsters at 75 mg/kg, although suppression
was evident at 150 mg/kg (unpublished observations).
CONCLUSION
Additional studies will be necessary to unravel the sites of
action of adenosinergic compounds within the circadian
system. Nonetheless, the present data provide the ®rst
evidence that adenosine, a putative sleep signal within the
CNS, may have signi®cance as an endogenous circadian
phase resetting stimulus, in addition to its recently hypothesized role as a modulator of photic inputs to the
clock. Furthermore, the failure of caffeine to induce phase
shifts at pharmacological doses that induce signi®cant
behavioral arousal is the ®rst formal demonstration of an
important principle that behavioral and pharmacological
means of stimulating arousal can have different, even
opposing effects on the state of the circadian clock. Given
the world-wide use of caffeine, particularly during jet
travel and shift-work rotations, when clock resetting is
often necessary for adaptation, the potential interaction
between this drug and phase resetting stimuli clearly
merits further attention.
REFERENCES
1. Dunwiddie TV, Masino SA. Annu Rev Neurosci 24, 31±55 (2001).
2. Porkka-Heiskanen T, Strecker RE, Thakkar M et al. Science 276,
1265±1268 (1997).
3. Porkka-Heiskanen T. Ann Med 31, 125±129 (1999).
4. Radulovacki M. Rev Clin Basic Pharm 5, 327±339 (1985).
ADENOSINE A1 LIGANDS AFFECT CIRCADIAN CLOCK
5. Antle MC, Mistlberger RE. J Neurosci 20, 9326±9332 (2000).
6. Mistlberger RE, Antle MC, Glass JD et al. Biol Rhythm Res 31, 240±283
(2000).
7. Mistlberger RE, Landry GJ, Marchant EG. Neurosci Lett 238, 5±8 (1997).
8. Elliott KJ, Weber ET, Rea MA. Eur J Pharmacol 414, 45±53 (2001).
9. Watanabe A, Moriya T, Nisikawa Y et al. Brain Res 740, 329±336 (1996).
10. Mikkelsen JD, Vrang N, Mrosovsky N. Brain Res Bull 47, 367±376 (1998).
11. Chambille I, Doyle S, Serviere J. Brain Res 612, 138±150 (1993).
12. Reebs SG, Mrosovsky N. J Biol Rhythms 4, 39±48 (1989).
13. Strecker GJ, Y. B, Dudek FE. Semin Neurosci 7, 43±51 (1995).
14. Cutler DJ, Piggins HD, Selbie LA et al. Eur J Pharmacol 345, 155±162
(1998).
15. Schaap J, Meijer JH, Eur J Neurosci 13, 1955±1962 (2001)
NEUROREPORT
16. Yamazaki S, Kerbeshian MC, Hocker CG et al. J Neurosci 18,
10709±10723 (1998).
17. Goodman RR, Kuhar MJ, Hester L et al. Science 220, 967±969 (1983).
18. Ticho SR, Radulovacki M. Pharmacol Biochem Behav 40, 33±40 (1991).
19. Bina KG, Rusak B, Semba K. J Comp Neurol 335, 295±307 (1993).
20. Van Reeth O, Hinch D, Tecco JM et al. Brain Res 545, 208±214 (1991).
21. Gibbs FP. Am J Physiol 244, R607±610 (1983).
22. Okada M, Nutt DJ, Murakami T et al. J Neurosci 21, 628±640 (2001).
23. Meyer-Bernstein EL, Morin LP. J Neurosci 16, 2097±2111 (1996).
24. Hastings MH, Duf®eld GE, Smith EJ et al. Chronobiol Int 15, 425±445
(1998).
25. Mintz EM, Gillespie CF, Marvel CL et al. Neuroscience 79, 563±569
(1997).
Vol 12 No 13 17 September 2001
2905