Regulatory, Integrative and Comparative Physiology

Melatonin and activity rhythm responses to light pulses in mice with the Clock mutation

David J. Kennaway, Athena Voultsios, Tamara J. Varcoe, Robert W. Moyer


Melatonin and wheel-running rhythmicity and the effects of acute and chronic light pulses on these rhythms were studied in Clock Δ19 mutant mice selectively bred to synthesize melatonin. Homozygous melatonin-proficientClock Δ19 mutant mice (Clock Δ19/Δ19 -MEL) produced melatonin rhythmically, with peak production 2 h later than the wild-type controls (i.e., just before lights on). By contrast, the time of onset of wheel-running activity occurred within a 20-min period around lights off, irrespective of the genotype. Melatonin production in the mutants spontaneously decreased within 1 h of the expected time of lights on. On placement of the mice in continuous darkness, the melatonin rhythm persisted, and the peak occurred 2 h later in each cycle over the first two cycles, consistent with the endogenous period of the mutant. This contrasted with the onset of wheel-running activity, which did not shift for several days in constant darkness. A light pulse around the time of expected lights on followed by constant darkness reduced the expected 2-h delay of the melatonin peak of the mutants to ∼1 h and advanced the time of the melatonin peak in the wild-type mice. When theClock Δ19/Δ19 -MEL mice were maintained in a skeleton photoperiod of daily 15-min light pulses, a higher proportion entrained to the schedule (57%) than melatonin-deficient mutants (9%). These results provide compelling evidence that mice with the Clock Δ19 mutation express essentially normal rhythmicity, albeit with an underlying endogenous period of 26–27 h, and they can be entrained by brief exposure to light. They also raise important questions about the role of Clock in rhythmicity and the usefulness of monitoring behavioral rhythms compared with hormonal rhythms.

  • circadian
  • pineal gland
  • Clock genes
  • entrainment

in the suprachiasmatic nucleus (SCN), Clock is considered an essential transcription factor for cellular rhythmic processes. In theClock Δ19 mutant mouse (32), an A → T transversion in the splice donor site downstream from exon 19 leads to the skipping of this exon in the Clock mRNA and elimination of 51 amino acids in the COOH-terminal glutamine-rich region of the CLOCK protein (17). The BMAL1/CLOCKΔ19 heterodimer lacks the ability to promoteper transcription (9); consequently,per1, (13) BMAL1 (26),cry1 and cry2 (19), andarginine vasopressin (AVP) genes (13,29) are arrhythmic in the SCN. Nevertheless, homozygousClock Δ19 mutants can express daily behavioral rhythms, and the rhythms may persist for ≥1 wk in constant darkness, with a period of ∼27 h, before they become arrhythmic (20). Behavioral rhythms, however, can be masked by exogenous influences, and it can be quite challenging to distinguish between a masked and a genuine circadian rhythm.

The robust rhythm of pineal gland melatonin secretion is driven by the SCN in the anterior hypothalamus. The presence of melatonin receptors in the SCN (31) and their effects on SCN rhythmicity (21) suggest that the hormone may play an important role in the maintenance and entrainment of rhythms. The monitoring of melatonin rhythmicity is considered the “gold standard” for human studies, because it provides noise-free insight into SCN rhythmicity and is particularly useful in studies of rhythm disorders. In animals, it has proved useful in monitoring alterations in the melatonin rhythm in studies of the role of neurotransmitters in the control of rhythmicity by light (14).

Mice are widely used in circadian rhythm research because of the growth in our understanding of mouse genetics, particularly the ability to disrupt genes. They exhibit a robust rhythm of wheel-running activity, which allows simple, noninvasive quantitation. Unfortunately, the strains of laboratory mice that are so valuable for gene knockout studies, including the growing number of clock gene knockouts, are melatonin deficient because of mutations in arylalkylamine-N-acetyltransferase (AA-NAT) and hydroxyindole-O-methyltransferase (HIOMT) (8). The Clock Δ19 mutant mouse was developed in melatonin-deficient C57Bl mice (32). The use of this background strain has seriously restricted the usefulness of the mutants for understanding the control of rhythmicity, because the SCN is a major target site for melatonin as well as the generator of rhythmicity. The CBA strain has a robust melatonin rhythm (12), which can be manipulated by light pulses (15).

We therefore set out to produce homozygousClock Δ19/Δ19 mutant animals with normalAA-NAT and HIOMT genes using selective crossing with CBA mice. The study addressed whether mice with aClock Δ19 mutation can synthesize melatonin. On the affirmative answer to this question, we then asked whether melatonin production and secretion are rhythmic in a 12:12-h light-dark photoperiod, whether the rhythm is endogenously generated, and whether light can suppress melatonin production and entrain the melatonin rhythm. Finally, we examined whether the wheel-running rhythms ofClock Δ19 mutants that were melatonin proficient or deficient could be entrained with equal efficiency under conditions that minimized the effects of behavioral masking by light (skeleton photoperiods).



On arrival in Adelaide, mice heterozygous for theClock Δ19 mutation (Clock +/Δ19) on a BALB/c background were paired to produce Clock +/+,Clock +/Δ19, andClock Δ19/Δ19 lines. CBA/6CaH mice, a strain derived from stock originally provided by the Animal Resources Centre of the University of Western Australia (∼arcwa/), were obtained from the University of Adelaide Central Animal House. Animals were housed in our specific pathogen-free facility in a 12:12-h light-dark photoperiod and fed standard laboratory chow ad libitum.


Inheritance of the Clock Δ19 mutation was monitored in the breeding program by PCR of tail DNA (27). Briefly, mouse genomic DNA was extracted from tail biopsies and subjected to PCR analysis using primers reported previously (27). The products were digested with HincII and electrophoresed on a 1.5% agarose gel. The mutant allele produced a 460-bp product; the wild-type allele gave a 398-bp product.

Melatonin assays.

Plasma was assayed by RIA using reagents obtained from Buhlmann Laboratories (Allschwil, Switzerland), as previously described (15). Briefly, plasma (100 μl) was added to the prewashed columns and sequentially washed with 10% methanol; hexane and melatonin were finally eluted with pure methanol. After evaporation of the solvent, the residue was reconstituted in 1 ml of buffer, and two 400-μl aliquots were subjected to RIA. Sensitivity of the assay was 10 pM. Intra- and interassay coefficients of variation were <10% and <15%, respectively, across the range of the standard curve.

Pineal glands were stored frozen in 1 ml of RIA buffer. Before assay, the buffer was thawed, and the pineal gland was homogenized. Aliquots (100 μl) were assayed in duplicate by direct RIA with a sensitivity of 21 fmol/gland. Intra- and interassay coefficients of variation were <10% and <15%, respectively, across the range of the standard curve.

Phenotyping for melatonin production.

Because the mutations that disable AA-NAT vary between strains (23) or have not been characterized (HIOMT), we developed an in vivo phenotyping procedure for melatonin production based on the observation that early-morning administration of the β-adrenergic agonist isoproterenol increased plasma melatonin levels in CBA, but not C57Bl or BALB/c, mice (15). Briefly, the test involves administration of isoproterenol (20 mg/kg sc) 2 and 4 h after lights on. Under light halothane anesthesia, blood (∼300 μl) was collected from the retroorbital sinus 30 min after the last injection, and plasma was assayed for melatonin by RIA. Preliminary evidence indicated that mice carrying one mutant set of AA-NAT and HIOMTalleles produced half as much melatonin as those carrying two functional alleles (8). Thus we selected the highest melatonin producers at all stages of the breeding program to produce our target genotype, which we have designatedClock Δ19/Δ19 -MEL (seeresults).

Melatonin rhythmicity.

To monitor the endogenous production and rhythmicity of melatonin in mice with a disabled Clock gene,Clock +/+ -MEL andClock Δ19/Δ19 -MEL mice previously maintained in a 12:12-h light-dark photoperiod were released into continuous darkness, and blood and pineal glands were collected 4–64 h later from groups of 4–11 animals. Sampling times were concentrated around the time of subjective dawn during the first 2 days of continuous darkness. Tissue and blood were collected after brief exposure to dim red light.

In these and subsequent experiments, the time of day is reported as zeitgeber time (ZT) or circadian time (CT). ZT12 is the time of lights off in a 12:12-h light-dark photoperiod; CT12 is the time of expected lights off in continuous darkness (usually also coinciding with the onset of wheel-running activity).

Effects of a light pulse on melatonin.

Exposure of animals to an unexpected light pulse during the night causes an immediate cessation of melatonin synthesis through inhibition of AA-NAT activity (18). To test whether melatonin synthesis in Clock Δ19/Δ19 -MELmice could be suppressed by light, groups of five animals were killed in the dark around the expected time of peak melatonin production, specifically at ZT23.5, at ZT24, or 15 min after 15 min of exposure to a 100-lux light pulse that started at ZT23.5. As a control,Clock +/+ -MEL mice were killed around the time of their peak melatonin production, specifically at ZT21.5, at ZT22, or 15 min after a 15-min light pulse that started at ZT21.5.

Phase shifting of melatonin rhythms by light.

To address whetherClock Δ19/Δ19 -MEL mice could be phase advanced or prevented from phase delaying (i.e., entrained) by exposure to light, groups of five mice were exposed to light (100 lux for 15 min) around the time of peak melatonin production, ZT21.5 and ZT23.5 for Clock +/+ -MEL andClock Δ19/Δ19 -MEL mice, respectively, and kept in darkness until they were killed 21–28 h later. As a control, groups of mice from each line (n = 3–5 per time point) were not exposed to light. Potential phase shifting of the rhythm was analyzed by ANOVA from CT20 to CT23 and from CT2 to CT5 for Clock +/+ - MELand Clock Δ19/Δ19 -MEL mice, respectively, with significance set at P < 0.05. The degree of shift was determined by changes in the time of the peak.

Wheel running.

Mice were housed individually in cages equipped with 11.5-cm-diameter running wheels. A data acquisition system (LabPro, Data Sciences, St. Paul, MN) was used to record the number of wheel rotations in 10-min bins. After the data were downloaded, the Actiview software package (MiniMitter, Bend, OR) was used to display the actograms and to perform the χ2 method of period analysis.

To investigate the pattern of wheel running in a 12:12-h light-dark photoperiod, 77 Clock +/+, 61Clock Δ19/Δ19, 71Clock +/+-MEL, and 83Clock Δ19/Δ19-MEL mice were studied on the 5th–10th days of access to wheels. The onset of running was evaluated in 12:12-h light-darkness by recording the time of day the wheel revolutions exceed 50 revolutions per 10 min for at least three successive 10-min periods within a 1-h period before and after lights off. Over the 5 days, the within- and between-animal mean onset times and the standard deviations were calculated. A subset of these animals was released into constant darkness for a further 10 days, and periodogram analysis was performed to estimate period length. The free-running period was calculated for 17Clock +/+, 17Clock Δ19/Δ19, 11Clock +/+-MEL, and 5Clock Δ19/Δ19- MEL mice.

To address the question of entrainment of wheel-running activity, 8Clock +/+, 22Clock Δ19/Δ19, 13Clock +/+-MEL, and 26Clock Δ19/Δ19-MEL mice were acclimated to the wheels and then exposed to skeleton photoperiods. In the first experiments, mice were exposed to a 0.25:23.75-h light-dark photoperiod (lights on at previous ZT0) or a 0.25:12:0.25:11.5-h light-dark-light-dark photoperiod (lights off at previous ZT0 and ZT12). Periodogram analysis was performed from the time the pulses commenced. In the second experiment, two groups ofClock Δ19/Δ19-MEL mice (n = 5) and their wild-type controls were exposed as described above to a 0.25:23.75-h light-dark photoperiod but with the light pulses timed for the previous ZT3 and ZT6. In all wheel-running experiments conducted in constant darkness, mice were considered to be entrained if the calculated period was between 23.8 and 24.2 h.


Breeding program for ClockΔ19/Δ19-MEL mice.

Male Clock Δ19/Δ19 mice were crossed with five female CBA mice to produce heterozygotes for theClock Δ19, AA-NAT, andHIOMT mutations (3 litters of 7, 5, and 6 pups). The female offspring from each of the litters (n = 3, 3, and 2) were then backcrossed to Clock Δ19/Δ19 male mice (not fathers) to produce the first backcross (BC1) line. This cross produced a total of 71 pups from 7 pregnancies, of which 37 were heterozygous and 32 were homozygous for theClock Δ19 mutation (2 died before genotyping). A second independent BC1 line produced 35Clock +/Δ19 and 54Clock Δ19/Δ19 mice. Figure1 shows the frequency distributions for the plasma melatonin levels after the isoproterenol phenotyping for the BC1 mice. On the basis of the expectation that the AA-NATand HIOMT genes are not linked (8), we expected that 25% of the BC1 animals would be heterozygous for the enzyme. Thus animals with melatonin levels in the 75th–100th percentile could be assigned this genotype (Fig. 1).Clock +/Δ19 andClock Δ19/Δ19 genotypes had similar distributions. Indeed, the top 25% secretors had melatonin levels of 29.5 ± 4.4 pM (n = 24) and 28.4 ± 3 pM (n = 30), respectively, after isoproterenol administration.

Fig. 1.

Distribution of plasma melatonin levels of 1st backcross (BC1) mice after isoproterenol administration. Data are shown as frequency and percentage of mice with <10–140 pM melatonin in 10-pM bins. A: all BC1 mice. Horizontal line is drawn at 75th percentile to indicate undetectable levels in 75% of the mice, as expected. B: mice heterozygous or homozygous forClock Δ19 mutation.

The four highest melatonin-producing male and female mice of the first backcross carrying the Clock Δ19 mutation were then intercrossed to provide the BC1 intercross line. This mating produced 40 pups from 4 pregnancies (18 female and 22 male mice). The second BC1 line was also intercrossed (3 pairs) to produce 11 male and 7 female mice that survived to phenotyping. Figure2 shows the frequency distribution of the plasma melatonin. Assuming again that all 3 genes segregated independently, we expected 1 in 16 mice to be homozygous for theClock Δ19 mutation and to carry 2 functional alleles of AA-NAT and HIOMT.

Fig. 2.

Distribution of plasma melatonin levels in intercross mice after isoproterenol administration. Data are shown as frequency and percentage of mice with <10–140 pM melatonin in 10-pM bins. Horizontal lines are drawn at 44th and 94th percentiles to identify mice presumed to have 0–1 functional arylaklylamine-N-acetyltransferase (AA-NAT) or hydroxyindole-O-methyltransferase (HIOMT) alleles (bottom 44%), ≥1 AA-NAT and 1 HIOMT allele (44–94%), or 2 functional AA-NAT and 2 functionalHIOMT alleles (>94%). Mice with 78 and 135 pM melatonin were males and were used to establish theClock Δ19/Δ19 -MEL mouse line.

The two male mice with plasma melatonin levels higher than the 94th percentile after isoproterenol administration (135 and 78 pM) were therefore tentatively assigned the genotypeClock Δ19/Δ19 -AA-NAT +/+ -HIOMT +/+(hereafter designatedClock Δ19/Δ19 -MEL). These male mice were mated with 14 CBA female mice to produce heterozygotes for the Clock Δ19 mutation and homozygotes for theAA-NAT and HIOMT genes (53 male, 66 female, and 11 dead pups). The offspring from these matings were intercrossed to give Clock +/+ -MEL,Clock +/Δ19 -MEL, andClock Δ19/Δ19 -MEL mice. To establish the Clock Δ19/Δ19 -MELline, we used 13 male and 17 female mice. To establish theClock +/+ -MEL line, we used 9 male and 13 female mice. We currently maintain the lines using at least seven breeding pairs per generation and have produced five generations with no evidence that the original putativeClock Δ19/Δ19 -MEL male mice were misclassified (i.e., no melatonin-deficient litters have been produced).

Melatonin rhythms in ClockΔ19/Δ19-MEL mice.

The animals produced after the backcross and intercrossing provided an opportunity to conduct preliminary investigations on the endogenous production of melatonin in Clock Δ19 mutants. Figure 3 shows the mean plasma melatonin levels after the isoproterenol test in those BC1 animals in which levels were higher than the 75th percentile; at these levels, the animals were likely to be carrying at least one functional allele of AA-NAT and HIOMT. Nighttime (ZT22) plasma melatonin concentration and pineal content are also shown in Fig. 3. The stimulated daytime levels were almost identical in the two genotypes, but 2 h before lights on (the time of peak melatonin production in the founder CBA strain), plasma and pineal melatonin levels were 50% lower in theClock Δ19/Δ19 mice than in the heterozygotes. Low, but detectable, plasma and pineal melatonin levels were recorded in some of the presumptive melatonin-deficient mice at ZT22, but even in this group, Clock Δ19/Δ19 mice had lower levels at ZT22. We presume that the measurement of endogenous melatonin production in some animals categorized as deficient reflects a low false-negative assignment for the isoproterenol test.

Fig. 3.

Plasma melatonin levels and pineal gland melatonin content of presumptive melatonin-proficient (A) and -deficient mice (B) from 1st backcross (BC1) at 22 h zeitgeber time (ZT22). Values are means ± SE. Filled bars,Clock +/Δ19 mice; open bars,Clock Δ19/Δ19 mice. Plasma melatonin levels measured after isoproterenol injection to phenotype the mice as melatonin-proficient or -deficient are also shown. All animals were killed 2 h before lights on (ZT22). Melatonin-proficient mice had identical isoproterenol responses regardless of Clockgenotype, but levels were lower inClock Δ19/Δ19 mice at ZT22 than in heterozygotes.

For the intercross animals, we expected three melatonin phenotypes: deficient mice (50%), those carrying at least one allele each of AA-NAT and HIOMT (44%), and those carrying two functional AA-NAT and HIOMT alleles (6.25%). Because only four of the latter group were available for analysis at ZT22, no firm conclusions could be drawn about melatonin production at this time of day in mice with this genotype.

When sufficient Clock Δ19/Δ19 -MELanimals became available, the endogenous melatonin rhythm in 12:12-h light-darkness and during the first two cycles of continuous darkness was studied. Figure 4 shows thatClock +/+ -MEL mice had maximal pineal gland melatonin content and plasma melatonin concentration 2 h before expected lights on (ZT22) and that, by dawn (ZT0), production and secretion had decreased to near basal levels.Clock Δ19/Δ19 -MEL mice also produced melatonin during darkness, with peak pineal melatonin content and plasma melatonin levels just before the lights would have come on (ZT0), which is 2 h later than mice with a functionalClock gene. When groups of fiveClock Δ19/Δ19 -MEL mice were killed after 1 h of light in the morning or after 1 h of unexpected additional darkness, pineal and plasma melatonin were at basal/undetectable levels (data not shown).

Fig. 4.

Plasma melatonin levels and pineal gland melatonin content in Clock +/+ -MEL andClock Δ19/Δ19 -MEL mice maintained in a 12:12-h light-dark photoperiod and released into continuous darkness. ●,Clock +/+ -MEL mice; ○,Clock Δ19/Δ19 -MEL mice. Values are means ± SE. CT, circadian time. A: plasma melatonin in 4–11 mice killed at each time point. Black horizontal bars, time of actual and expected darkness; gray bars, subjective day. ZT12 is the normal time of lights off, and CT12 is expected time of lights off in constant darkness. B: pineal gland melatonin content for animals in A. Melatonin rhythm peaks inClock Δ19/Δ19 -MEL mice 2 h later than in Clock +/+ -MEL mice on the 1st night and shifts a further 2 h each day on release into constant darkness.

When mice with functional Clock genes were killed after 1 and 2 days of constant darkness, pineal and plasma melatonin levels were highest 8–10 h after the expected time of dark onset and low around subjective dawn on both days (Fig. 4).Clock Δ19/Δ19-MEL mice had basal levels of melatonin at CT22, with a peak around CT2 on the 1st day and around CT4 on the 2nd day of continuous darkness (Fig. 4).

When exposed to light at the time of peak production, the mice responded by rapid and complete suppression of melatonin production and secretion, irrespective of the presence of a functionalClock gene (Fig. 5).

Fig. 5.

Plasma melatonin levels (A) and pineal gland melatonin content (B) inClock +/+ -MEL andClock Δ19/Δ19 -MEL mice before and after exposure to 15-min 100-lux light pulse around the time of peak melatonin production. Because the 2 lines differed in the time of the peak, Clock +/+ -MEL mice were pulsed or kept in darkness from ZT21.5 to ZT21.75 and killed at ZT22, whereasClock Δ19/Δ19 -MEL mice were pulsed from ZT23.5 to ZT23.75 and killed at ZT0. Values are means ± SE (n = 5) for each group. Exposure to light pulse decreased plasma melatonin levels and pineal melatonin content (open bars) to detection limit of assay and basal levels, respectively.

It was clear thatClock Δ19/Δ19 -MEL mice placed in constant darkness delayed the peak of melatonin production by 2 h each cycle. It was therefore of interest to test whether light exposure at the time of peak production would advance or at least prevent the 2-h delay in the melatonin rhythm. Figure6 shows peak melatonin production between CT0 and CT4 in Clock Δ19/Δ19 -MELmice after one cycle of continuous darkness, as observed in the first experiment. After brief exposure to light, the mutant mice had a different pattern of plasma and pineal melatonin, with significantly lower levels at CT2 and CT3 (P < 0.05 by ANOVA). TheClock +/+ -MEL mice also responded to the light pulse with significantly lower plasma melatonin levels and pineal melatonin contents at CT20, CT21, and CT22 (P < 0.05). From the previous experiment where we observed a 2-h delay in the time of the peak melatonin production in theClock Δ19/Δ19-MEL mice, we expected that light presentation at the time of peak melatonin production would prevent this delay. In the case of theClock +/+ -MEL mice, we would expect the rhythm to be advanced. The results shown in Fig. 6 are consistent with these hypotheses.

Fig. 6.

Plasma melatonin levels and pineal gland melatonin content inClock +/+ -MEL andClock Δ19/Δ19 -MEL mice after a 15-min 100-lux light pulse. Clock +/+ -MELmice received the light pulse (arrows) at ZT21.5, whereasClock Δ19/Δ19 -MEL mice were pulsed at ZT23.5, the times of peak melatonin production in the 2 lines. Stippled areas represent melatonin levels from the 1st experiment (see Fig. 5). Values are means ± SE. A and C:Clock +/+ -MEL mice. ●, No light exposure; ○, light pulse. B andD: Clock Δ19/Δ19 -MELmice. Alternating black and gray horizontal lines indicate subjective dark and light periods, respectively. Light pulse resulted in significantly reduced melatonin levels between CT20 and CT23 inClock +/+ -MEL mice and between CT2 and CT3 in Clock Δ19/Δ19 -MEL mice.

Wheel-running rhythms.

Wheel running in all four mouse lines in a 12:12-h light-dark photoperiod was characterized by intense running during the dark phase, with the greatest concentration of running early in the night and decreasing steadily toward lights on (Fig.7). In theClock +/+ andClock +/+ -MEL mouse lines, the onset of running occurred 6 ± 29 (SD) min after darkness and 11 ± 25 min before lights off, respectively. In theClock Δ19/Δ19 andClock Δ19/Δ19 -MEL lines, the mean onset times were 6 ± 42 and 8 ± 42 min before lights off. The intra- and interanimal variance was higher in the mutants than in the wild-type animals. When the average number of wheel revolutions during the last 6 h of light was compared between genotypes,Clock +/+ andClock +/+ -MEL mice accumulated 4% and 3.7%, respectively, of their total wheel revolutions per day during this time. By contrast, 13.7% of the total daily running activity ofClock Δ19/Δ19 mice and 14% of the activity of Clock Δ19/Δ19 -MEL mice occurred during this period. The melatonin-deficientClock +/+ andClock Δ19/Δ19 mice also accumulated 12.3% and 11.6%, respectively, of their total running during the first 6 h of light, in contrast to their melatonin-proficient genotypes (2.3% and 6.3%). In Fig. 7, the differences in the time of offset of running on light onset in the melatonin-deficient and -proficientClock +/+ lines are clearly evident.

Fig. 7.

Wheel-running activity rhythms ofClock Δ19/Δ19 andClock Δ19/Δ19 -MEL mice and their wild-type controls before and after release into constant darkness. Data are averages of the number of wheel revolutions per 10 min. A: Clock +/+ andClock Δ19/Δ19 mice (light and darkness,n = 77 and 61, respectively, with 17 of each genotype continuing into constant darkness). B:Clock +/+ -MEL andClock Δ19/Δ19 -MEL mice (light and darkness, n = 71 and 83, respectively, with 5 mutant and 11 wild-type mice continuing into constant darkness). Stippled areas, data from mutant lines; lines, data from wild-type controls. Vertical arrows, actual and expected times of lights off; square wave, light-dark transitions. Representative double-plotted actograms of 1 wild-type mouse, 2 free-running mutants, and 1 arrhythmic mutant for melatonin-deficient and -proficient lines are shown below Aand B. DD, constant darkness.

On release into continuous darkness, rhythmicity was maintained in the melatonin-deficient and -proficient Clock +/+lines for ≥14 days, with periods of 23.8 ± 0.04 h (n = 16) and 23.3 ± 0.07 h (n = 10), respectively. By contrast, rhythmicity was sustained in 76% (n = 25) and 44% (n= 9) of the Clock Δ19/Δ19 andClock Δ19/Δ19 -MEL mice, respectively. In those animals that showed rhythmicity, the free-running periods were 27.0 ± 0.17 and 26.9 ± 0.24 h, respectively. When the revolutions per 10 min were averaged in constant darkness, it was clear that the onset of running activity in theClock +/+ mice changed little over the first four cycles (Fig. 7), consistent with their free-running period of 23.8 h. Similarly, the Clock +/+ -MEL mice changed their onset times over the first four cycles, consistent with their shorter periodicity, most evident from the second onset in constant darkness. Despite having free-running periods of ∼27 h, it was not until the second and fourth cycles in constant darkness that the Clock Δ19/Δ19 andClock Δ19/Δ19 -MEL mice showed clear changes (delays) in onset.

Entrainment of running activity by light pulses.

When melatonin-deficient and -proficientClock +/+ mice were released into the skeleton photoperiod with the pulse occurring at the time of the previous lights on (ZT0), both lines free ran with a period of 23.5 ± 0.04 and 23.4 ± 0.25 h, respectively, over the duration of the experiment (Fig. 8). Over the duration of the experiment, the pulses never coincided with wheel-running activity in these mice. Therefore a double-pulse skeleton photoperiod (0.25:23.75:0.25:11.5-h light-darkness-light-darkness) was utilized (with the pulses occurring at the previous lights on and lights off). Under these conditions, it was hypothesized that theClock +/+ mice would be entrained to the “evening” pulse. As expected, the Clock +/+lines had periods of 24.0 ± 0.01 and 24.0 ± 0.03 h, showing that the pulses were sufficient to entrain the mice.

Fig. 8.

Wheel-running activity rhythm periods ofClock Δ19/Δ19 (A) andClock Δ19/Δ19 -MEL (B) mice and their wild-type controls before (open bars) and after (filled bars) release into a 0.25:23.75-h light-dark or 0.25:12:0.25:11.5-h light-dark-light-dark skeleton photoperiod. Values are means ± SE (hours). Results are grouped into mice that entrained (period 24 ± 0.5 h), entrained late (free ran and then had a period of 24 ± 0.5 h), free ran (period > 24.5 h), or became arrhythmic. Wild-type mice are shown as 2 groups, because all wild-type mice free ran in the 0.25:23.75-h light-dark photoperiod, where the pulse occurred at the subjective dawn, or all entrained to the double-pulse regimen, i.e., 0.25:12:0.25:11.5-h light-dark-light-dark photoperiod, where pulses occurred at subjective dusk and dawn. Double-plotted actograms show data from individual mice that are representative of the various categories in A andB. Vertical arrows, time of daily light pulses.

Under single- and double-pulse regimens, it was expected that only the morning pulse would influence entrainment of the mutants because of the long free-running period. Data from both regimens are shown in Table1. Because the results obtained were similar for single and double pulses, they were combined. In the case of the Clock Δ19/Δ19 mice, 2 of 22 entrained to the pulses (period 23.5–24.5 h), whereas 13 of 22 free ran with a period of 27.1 ± 0.32 h and 7 of 22 became arrhythmic immediately (Fig. 9). By contrast, 12 of 26 Clock Δ19/Δ19 -MEL mice were entrained and 3 of 26 were entrained after free running for ∼8 days (late entrainment), indicating that 58% of the mice entrained. Another 6 of 26 mice free ran with a period of 26.3 ± 0.35 h, and 5 of 26 mice became arrhythmic immediately on release into the pulse conditions.

View this table:
Table 1.

Animals that were entrained, arrhythmic, or free ran in single- and double-pulse regimens

Fig. 9.

Proportions of Clock Δ19/Δ19and Clock Δ19/Δ19 -MEL mice that entrained, free ran, or became arrhythmic in 0.25:23.75-h light-dark or 0.25:12:0.25:11.5-h light-dark-light-dark photoperiod. Values are percentages of total number of mice of each phenotype. Note high proportion of Clock Δ19/Δ19 -MELmice that were entrained compared with melatonin-deficient mutants.

To further test the ability ofClock Δ19/Δ19 -MEL mice to entrain to single light pulses, we maintained mice in a 12:12-h light-dark photoperiod and then released them into the skeleton (0.25:23.75-h light-dark) photoperiod with light pulses timed for 3 or 6 h after expected lights on for 13 days. Figure10 shows that two of fiveClock Δ19/Δ19 -MEL mice were entrained to the single pulses presented 3 h after expected lights on, one free ran, and two became arrhythmic. Four of five mice entrained to the single pulses presented 6 h after expected lights on, and one free ran. The Clock +/+ -MELmice free ran until the onset of wheel running coincided with the pulse, after which they all entrained (data not shown).

Fig. 10.

Double-plotted actograms of wheel running inClock Δ19/Δ19 -MEL mice exposed to a 0.25:23.75-h light-dark skeleton photoperiod, where daily pulses occurred at 3 h (A) or 6 h (B) after expected time of lights off. Top left actograms show timing of darkness (black) during experiments and time of daily single light pulses. e, Entrained; f, free running, a, arrhythmic.


A principal aim of the experiments was to investigate hormonal (melatonin) rhythmicity for the first time in mice carrying theClock Δ19 mutation. Unfortunately, the originalClock Δ19 mutants were produced in C57Bl/6J mice and maintained on a BALB/c background and, therefore, are melatonin deficient. The CBA mouse strain produces melatonin rhythmically (12, 33), and the rhythm shifts in response to light pulses (15); therefore, we chose to introduce functional AA-NAT and HIOMT genes intoClock Δ19 mice by selective breeding. It was possible that the mutants would not synthesize melatonin, because BMAL1/CLOCK heterodimer binding in the promoter region ofAA-NAT is required for induction of the enzyme in the chicken pineal gland (7) and rat retina (5). Although there appears to be no such requirement in the rat pineal gland in vitro (5), this had not been tested in vivo. Our program showed that Clock is not required for AA-NAT induction in vivo and thatClock Δ19/Δ19 -MEL mice produced and secreted melatonin in amounts comparable to wild-type mice.

The selective breeding program was based on the assumption that the three genes segregated independently. AA-NAT has been mapped to mouse chromosome 11 (11) and Clock to mouse chromosome 5 (16), and AA-NAT andHIOMT genes are known to segregate independently (8). To our knowledge, mouse HIOMT has not been mapped; therefore, we had to assume that it was not close toClock on chromosome 5. Therefore, a conservative selection procedure was used, with only top-ranking mice being used for breeding, especially after the intercross stage. The two putativeClock Δ19/Δ19 -MEL male mice were crossed again with CBA female mice to rapidly expand the colony size and reduce inbreeding and the chances of transmitting a mutant enzyme allele. After five generations, there has been no evidence of melatonin deficiency in the colony.

Melatonin-proficient mice carrying theClock Δ19 mutation showed a rhythm in melatonin production and secretion, with the peak occurring at the time of lights on, 2 h later than the wild-type strain. The melatonin production decreased to baseline within 1 h, even in the absence of light. This, together with the persistence of the melatonin rhythm in theClock Δ19/Δ19 -MEL mice in constant darkness and the delay in the timing of the peak each cycle by 2 h, confirmed the endogenous nature of the melatonin rhythm.

To test whether control of the melatonin rhythm in theClock Δ19/Δ19 -MEL mice was normal, these mice were subjected to unexpected exposure to light at night (10). When exposed to a brief light pulse at the time of the peak melatonin production,Clock +/+-MEL andClock Δ19/Δ19 -MEL mice responded by suppressing melatonin levels to baseline within 15 min. WhenClock +/+ - MEL mice were exposed to a light pulse at the time of the peak of melatonin production and followed over the next cycle in constant darkness, the melatonin production ceased significantly earlier than nonpulsed mice, as shown previously in CBA mice (15). Exposure ofClock Δ19/Δ19 -MEL mice to light at the time of the melatonin peak resulted in significantly decreased pineal melatonin content and plasma melatonin levels at CT3 and CT4 compared with the nonpulsed animals. The fact that the peak at CT1 was still 1 h later than the time of the peak in a light-dark environment and 1 h earlier than in the control mutant animals suggests that the response to the light pulse was weak or that only a portion of the population responded. Nevertheless, we conclude that a light pulse can shorten the period of the melatonin rhythm and potentially entrain it. These responses to light are particularly interesting in view of the reported poor responsiveness ofClock Δ19/Δ19 mutants to light pulses, at least with respect to induction of the immediate-early gene c-fos and per1 and per2(27).

Using the more traditional method of wheel running to monitor rhythmicity, we confirmed the entrainment ofClock Δ19/Δ19 mice (32) to a 12:12-h light-dark photoperiod and report similar results forClock Δ19/Δ19 -MEL mice. The precision of the entrainment in both mutant lines was poor compared with the wild-type animals, and running during the light period was common. Because Clock Δ19 mutants have a period of 26–27 h, the onset of running activity should be 2–3 h after lights off (1), but in theClock Δ19/Δ19 andClock Δ19/Δ19 -MEL mice it occurred within ±10 min of lights off, similar to the wild-type mice. By contrast, the melatonin rhythm peak occurred 2 h later than in wild-type animals, which is expected from Aschoff's predictions (1). A similar lack of correlation between phase angle difference and period of the wheel running rhythm is evident in otherClock gene knockouts: per1 (2, 4),per2 (2, 34), per3(25), per1/per2, per2/per3, andper1/per3 (2), cry1,cry2, and cry1/cry2 (30), andBMAL1 (3). Furthermore, whenClock Δ19 mutants were released into constant darkness, there was a lag of one to two cycles before the onset shifted (so-called transients), a phenomenon not observed with the melatonin rhythm. By contrast, Clock Δ19/Δ19 maintained on a Jcl:ICR genetic background have activity and body temperature rhythm acrophases in 12:12-h light-darkness consistent with the long endogenous period (24). This raises the question whether wheel running is a universally suitable output measure of the circadian clock in mice.

Given that the melatonin and wheel-running rhythms ofClock Δ19/Δ19 -MEL mice were strongly entrained to a 12:12-h light-dark photoperiod, we were interested in the extent to which rhythmicity could be entrained by short light pulses in a skeleton photoperiod. Vitaterna et al. (32) reported that a single 6-h light “pulse” could restore a long free-running period inClock Δ19/Δ19 mice that had become arrhythmic in constant darkness. We found that daily 15-min light pulses presented at the time of the previous lights on or 3 or 6 h later prevented animals from free running or eventually entrained them with a period close to 24 h. We observed that theClock Δ19/Δ19 -MEL mice were more likely to be entrained than the melatonin-deficient mutants, although ≥40% still failed to entrain or became arrhythmic. Because we have introduced genes other than AA-NAT and HIOMT from the CBA strain, we cannot attribute the increased proportion of entrained animals solely to melatonin.

How can these results of robust hormone rhythmicity and entrainment inClock Δ19/Δ19-MEL mutant mice be reconciled with what is known about the molecular biology of the “essential” Clock gene? TheClock Δ19 mutation is an antimorph that inhibits wild-type function (16). Indeed, not only is the BMAL/CLOCKΔ19 heterodimer incapable of driving transcription of an mper1-luciferase reporter, it is less active than BMAL1 alone (9). As a result,Clock Δ19 mutant mice should have decreased transcription of mper1 and other clock genes. This has been shown to be the case, with major disturbances in the rhythm ofper1, per2, and per3 (13),cry1 and cry2 (19), BMAL1(26), AVP (13, 29), and, more recently, prokineticin2 (6). One clock gene,mper2, appears to retain significant rhythmicity (13), and per1 and per2 have been shown to be induced, albeit at a reduced level, by a 15-min light pulse at ZT16 (27). The BMAL1/CLOCK heterodimer is the key driver of cellular rhythmicity; yet theClock Δ19/Δ19-MEL mutant mice show hormonal and behavioral circadian rhythmicity and entrainment to very brief light pulses. How can these hormonal and behavioral rhythms be sustained, especially at the unchanged amplitudes maintained by the mutants? One possibility is that the BMAL1/CLOCK heterodimer does have some intrinsic transcriptional activity at the E-boxes of some of the clock genes, but, as indicated earlier, at least for per1,there is no evidence for this (9). We are aware of no similar studies on per2. Another possibility is that another partner exists for BMAL1 that may drive transcription with decreased efficiency in the SCN or in centers that project to the SCN, resulting in the relatively poor precision of entrainment and long period length. NPAS2 would appear to be such a candidate protein, because it has been shown to be a functional analog of CLOCK and can partner with BMAL1 and drive transcription of appropriate reporter genes (22). However, npas2 is not expressed in the normal mouse SCN (28); instead, it is widely expressed in the forebrain. Although it is possible that npas2 is expressed in the SCN of Clock Δ19 mice, and not normal mice, it is more likely that other parts of the brain that maintain rhythmicity via BMAL1/NPAS2 drive, project to the SCN, and maintain essential sleep-wakefulness and hormonal rhythms.

In conclusion, we have produced Clock Δ19mutant mice that can synthesize melatonin. We have shown that, despite producing a protein that fails to initiate transcription ofClock and clock-controlled genes, the animals have a normal endogenous melatonin rhythm than can be altered by light pulses. The timing of the melatonin rhythm is consistent with the period expressed in constant darkness, unlike the case with locomotor activity. Finally, we have shown that theClock Δ19/Δ19 -MEL mice can be entrained to skeleton photoperiods, where the duration of the light period is 15 min.


We are grateful to Drs. J. Takahashi, M. Vitaterna, and M. Niekrasz (Northwestern University, Evanston, IL) for generously supplying the original Clock mutant mice. We thank Dr. David Weaver for helping us establish the PCR genotyping procedure in our laboratory and Drs. Wes Whitten and Bob Seamark for help in designing the breeding program.


  • These studies were supported in part by grants from the University of Adelaide.

  • Address for reprint requests and other correspondence: D. J. Kennaway, Dept. of Obstetrics and Gynaecology, University of Adelaide Medical School, Frome Rd., Adelaide, South Australia 5005, Australia (E-mail: david.kennaway{at}

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

  • First published January 9, 2003;10.1152/ajpregu.00697.2002


View Abstract