dissociation of per1 and bmal1 circadian rhythms in · primarily regulated by the circadian...
TRANSCRIPT
Dissociation of Per1 and Bmal1 circadian rhythms inthe suprachiasmatic nucleus in parallel withbehavioral outputsDaisuke Onoa,1,2, Sato Honmab,1,3, Yoshihiro Nakajimac, Shigeru Kurodad, Ryosuke Enokia,b,e, and Ken-ichi Honmab
aPhotonic Bioimaging Section, Research Center for Cooperative Projects, Hokkaido University Graduate School of Medicine, Sapporo, 060-8638, Japan;bDepartment of Chronomedicine, Hokkaido University Graduate School of Medicine, Sapporo, 060-8638, Japan; cHealth Research Institute, NationalInstitute of Advanced Industrial Science and Technology (AIST), Takamatsu, Kagawa 761-0395, Japan; dResearch Institute for Electronic Science, HokkaidoUniversity, Sapporo, 001-0020, Japan; and ePrecursory Research for Embryonic Science and Technology, Japan Science and Technology Agency, Saitama332-0012, Japan
Edited by Joseph S. Takahashi, Howard Hughes Medical Institute, University of Texas Southwestern Medical Center, Dallas, TX, and approved March 28, 2017(received for review August 11, 2016)
The temporal order of physiology and behavior in mammals isprimarily regulated by the circadian pacemaker located in thehypothalamic suprachiasmatic nucleus (SCN). Taking advantage ofbioluminescence reporters, we monitored the circadian rhythms ofthe expression of clock genes Per1 and Bmal1 in the SCN of freelymoving mice and found that the rate of phase shifts induced by asingle light pulse was different in the two rhythms. The Per1-lucrhythm was phase-delayed instantaneously by the light presentedat the subjective evening in parallel with the activity onset of be-havioral rhythm, whereas the Bmal1-ELuc rhythm was phase-delayed gradually, similar to the activity offset. The dissociationwas confirmed in cultured SCN slices of mice carrying both Per1-lucand Bmal1-ELuc reporters. The two rhythms in a single SCN sliceshowed significantly different periods in a long-term (3 wk) cul-ture and were internally desynchronized. Regional specificity inthe SCN was not detected for the period of Per1-luc and Bmal1-ELuc rhythms. Furthermore, neither is synchronized with circadianintracellular Ca2+ rhythms monitored by a calcium indicator,GCaMP6s, or with firing rhythms monitored on a multielectrodearray dish, although the coupling between the circadian firing andCa2+ rhythms persisted during culture. These findings indicate thatthe expressions of two key clock genes, Per1 and Bmal1, in the SCNare regulated in such a way that they may adopt different phasesand free-running periods relative to each other and are respec-tively associated with the expression of activity onset and offset.
clock gene | in vivo recording | suprachiasmatic nucleus |photic phase resetting | E and M oscillators
In mammals, the circadian pacemaker in the hypothalamicsuprachiasmatic nucleus (SCN) entrains to a light–dark (LD)
cycle and regulates circadian rhythms of behavior and physiology(1, 2). The circadian oscillation in the SCN is autonomous, andthe clock genes Per1, Per2, Cry1, Cry2, Clock, and Bmal1 playcrucial roles (3). A heterodimer of Clock and Bmal1 proteins(CLOCK/BMAL1) activates the transcription of Per and Crygenes; in turn, the protein products of these genes suppress theirown transactivation by CLOCK/BMAL1, closing a feedbackloop. One turn of the auto-feedback loop (core loop) takes∼24 h. On the other hand, Bmal1 expression is enhanced byRAR-related orphan nuclear receptor (ROR) and is repressedby an orphan nuclear receptor in the RevErb family (α, β)through ROR response element (4, 5). The expressions of RORand the RevErb family are enhanced in turn by a BMAL1/CLOCKheterodimer via an upstream E-box. Thus, the Bmal1 circadianrhythm is auto-regulated by a feedback loop (the Bmal1 loop)which is interlocked with the core loop, maintaining an antiphasicphase relationship with Per1 rhythm. This interlocked Bmal1 loophas been considered to contribute to stabilization and fine tuningof the core loop (4, 6) in addition to the regulation of downstreampathways (7).
The expression of Per genes in the SCN is activated by a timedexposure to light, which phase shifts the circadian pacemaker (8,9). The phase-dependent phase shifts of clock gene expressionare regarded as a key mechanism by which the circadian pace-maker is entrained to a LD cycle. Light signals from the retinastimulate the expression of Per genes, perturbing the core loopdynamics to produce a phase-dependent phase shift (8). How-ever, the mechanisms by which light-induced phase-shift signalsfrom the core loop are transduced to circadian rhythms in phys-iology and behavior are not well understood.On the behavioral level, the onset and offset of an activity
band (activity onset and offset) of circadian behavioral rhythmare known to respond differentially to a phase-shifted LD cycle(10) and to a single light pulse under continuous darkness (DD)in nocturnal rodents (11). In addition, the phase relation betweenthe activity onset and offset is known to change under differentphotoperiods (12). Furthermore, the two phases of behavioralrhythm occasionally split under the constant light condition (12).From these findings, the two oscillator hypothesis was advanced toexplain behavioral circadian rhythm in nocturnal rodents (13).
Significance
The circadian clock in the suprachiasmatic nucleus (SCN) regu-lates seasonality in physiology and behavior, which is bestcharacterized by the change in the activity time of behavioralrhythms. In nocturnal rodents, the activity time was shortenedin long summer days and lengthened in short winter days be-cause of the change in the phase relationship of activity onsetand offset, for which different circadian oscillators are predicted.Taking advantage of in vivo monitoring of clock gene expres-sion in freely moving mice, we demonstrated that the circadianrhythms of Per1 and Bmal1 in the SCN are associated differen-tially with the phase shifts of activity onset and offset, respec-tively, suggesting the existence of two oscillations with differentmolecular mechanisms in timing of circadian behavior.
Author contributions: D.O., S.H., and K.H. designed research; D.O. performed research;D.O., S.H., Y.N., S.K., and R.E. contributed new reagents/analytic tools; D.O. and Y.N.constructed a method of simultaneous measurement of bioluminescence; D.O., S.H., andR.E. constructed a method of fluorescence imaging; S.K. made an analytical program forimaging; D.O., S.H., and K.H. analyzed data; and D.O., S.H., and K.H. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.1To whom correspondence may be addressed. Email: [email protected] [email protected].
2Present address: Department of Neuroscience II, Research Institute of EnvironmentalMedicine, Nagoya University, Furo-cho, Chikusa-ku, Nagoya 464-8601, Japan.
3Present address: Research and Education Center for Brain Science, Hokkaido University,Sapporo 060-8638, Japan.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1613374114/-/DCSupplemental.
www.pnas.org/cgi/doi/10.1073/pnas.1613374114 PNAS | Published online April 17, 2017 | E3699–E3708
NEU
ROSC
IENCE
PNASPL
US
Dow
nloa
ded
by g
uest
on
Dec
embe
r 9,
202
0
Namely, one oscillator, designated an evening (E) oscillator, reg-ulates the activity onset, and the other oscillator, designated amorning (M) oscillator, controls the activity offset. Previously, re-gional differences were reported in the circadian rhythms of clockgene expression in the SCN (14–17), and the phase relation of thetwo rhythms changed under different photoperiods in parallel withthe change in activity time of behavioral circadian rhythm. Thesefindings suggested the existence of the E and M oscillators in theSCN. However, the molecular mechanisms underlying the E andM oscillators are totally unknown. Recently, we developed an invivo method for monitoring clock gene expression in the SCN of afreely moving mouse (18, 19), enabling us to compare the SCNcircadian rhythms with physiological and behavioral rhythms.In the present study, to obtain insight into the molecular
mechanism of light entrainment, we continuously monitoredcircadian rhythms in clock gene Per1 and Bmal1 expression in theSCN together with behavioral rhythms. Surprisingly, the rhythmsof Per1 and Bmal1 expression responded differentially to thelight pulse. The dissociation between the two circadian rhythmswas confirmed in the cultured SCN slices from double-transgenicmice carrying luciferase reporters for Per1 and Bmal1 expression.
These findings indicate that there are at least two circadianpacemakers in the SCN, which have different molecular mech-anisms and differentially regulate behavioral outputs.
ResultsDifferential Responses of Per1-luc and Bmal1-ELuc Circadian Rhythmsin the SCN in Vivo. We continuously measured the expression ofthe clock genes Per1 and Bmal1 in the SCN of freely movingtransgenic mice carrying a bioluminescence reporter (Per1-luc orBmal1-ELuc) under DD. Spontaneous locomotor activity wasmonitored simultaneously by an infrared thermal sensor. Bio-luminescence emitted from the SCN was collected with animplanted plastic optical fiber connected to a cooled photo-multiplier tube (PMT), as described previously (18, 19).Phase responses of Per1-luc and Bmal1-ELuc circadian
rhythms in the SCN were examined when a single light pulse of9 h duration was given at circadian time (CT) 11.5 to make thelargest phase-delay shift according to a previous study (20), inwhich the activity onset of behavioral circadian rhythm was de-fined as CT12 (Fig. 1 and Fig. S1A). We also examined the effectof a light pulse of the same duration given at CT21.5, when
0
7
14
0 0 012 12Local time
Day
sDays
Behavior (counts/m
in)0
80
40
20
60
1 8 152 3 4 5 6 7 9 10 11 12 13 14
Per1-luc (SCN)
0
7
14
0 0 012 12Local time
Day
s
DaysR
LU (c
ount
s/m
in)
120
0
30
Behavior (counts/m
in)0
80
40
20
60
1 8 152 3 4 5 6 7 9 10 11 12 13 14
Bmal1-ELuc (SCN)
18126 0 6 12 181
7
14
Day
s
18126 0 6 12 181
7
14D
ays
Local time Local time
Per1-luc Bmal1-ELuc
Pha
se s
hifts
(h)
-6
-4
-2
0
Days after light pulse-1 0 1 2 3 4 5 6 7 8 9
Activity onsetActivity offset
*
60
90
RLU
(cou
nts/
min
)
120
0
30
60
90
Per1-luc
Pha
se s
hifts
(h)
-6
-4
-2
0
Days after light pulse-1 0 1 2 3 4 5 6 7 8 9
*
Bmal1-ELuc
Pha
se s
hifts
(h)
-6
-4
-2
0
Days after light pulse-1 0 1 2 3 4 5 6 7 8 9
*
Bmal1-ELucPer1-luc
Light pulse
A
B
C
D
Fig. 1. Light pulse-induced phase-delay shifts of circadian rhythms in the SCN and behavioral rhythms in freely moving adult mice. (A and B) Typical examplesof phase response at CT11.5 are illustrated for the Per1-luc (A) and Bmal1-ELuc (B) rhythms in the SCN with a behavioral rhythm (black histogram) (Left) indouble-plotting, in which the colored area indicates bioluminescence larger than the minimum value of a series, and in sequential plotting (Right), in whichbroken lines indicate raw data and solid lines indicate 4-h moving-averaged values (Per1-luc, blue; Bmal1-ELuc, green). The number of behavioral activities in1-min intervals is indicated by black vertical bars. A yellow vertical bar indicates the time of the light pulse. (C) Mean acrophases ± SEM (horizontal bar) areillustrated for Per1-luc (blue circles) (Left) and Bmal1-ELuc (green circles) (Right) together with the mean activity onsets (gray triangles) and offsets (blacksquares). A yellow horizontal bar indicates the time of a light pulse. A horizontal gray and black bar at the top of each panel indicates the LD cycle to whichmice had been entrained. (D) Daily phase shifts are reported by mean ± SEM (n = 4) for Per1-luc (blue circles) and Bmal1-ELuc (green circles) (Left) rhythms, forPer1-luc and two phase markers (onset and offset) of behavioral rhythm (Center), and for Bmal1-ELuc and the two phase markers (Right). The abscissa indicatesthe number of days after a light pulse. An asterisk indicates statistically significant difference (P < 0.05, two-way ANOVA with a post hoc t test) between Per1-lucand Bmal1-ELuc (Left), between Per1-luc and activity offset (Center), and between Bmal1-ELuc and activity onset (Right). RLU, relative luminescence units.
E3700 | www.pnas.org/cgi/doi/10.1073/pnas.1613374114 Ono et al.
Dow
nloa
ded
by g
uest
on
Dec
embe
r 9,
202
0
phase-advance shifts were expected (Fig. 2 and Fig. S1B). Dataof the in vivo experiments before a phase-delaying light pulse(n = 4 for Per1 and n = 4 for Bmal1) were used in a previousstudy (18) to calculate the free-running period and to estimatethe circadian and ultradian phases of behavioral and clock geneexpression rhythms.In response to a light pulse at CT11.5, the activity onset of the
behavioral rhythm was phase delayed immediately by 4.0 ± 0.4 hon average in Per1-luc mice (mean ± SD, n = 4) and by 3.5 ±0.7 h on average in Bmal1-ELuc mice (n = 4). In contrast, theactivity offset was gradually phase delayed over four or five cyclesby 4.4 ± 0.3 h in Per1-luc mice and by 3.6 ± 0.8 h in Bmal1-ELucmice before reaching a free-running steady state (Fig. 1 C andD). The amount of phase shift in either behavioral marker wasnot significantly different in the two reporter mice. Thus, theactivity band was temporarily compressed after the light pulse(Fig. 1C). On the other hand, the circadian peak of Per1-lucrhythm was immediately phase delayed in parallel with the ac-tivity onset (Fig. 1 C, Left and D), and the circadian peak ofBmal1-ELuc rhythm was gradually phase delayed in parallel withthe activity offset (Fig. 1 C, Right and D). The amount of phaseshift in the two bioluminescent rhythms on the second day afterthe light pulse was significantly different (two-way ANOVA, posthoc t test, P < 0.05) (Fig. 1D, Left). When each bioluminescent
rhythm was compared with behavioral rhythms, the amount ofphase shift in the Per1-luc rhythm was not different from thephase shift of activity onset but was significantly different fromthe phase shift of activity offset (two-way ANOVA, post hoct test, P < 0.05) (Fig. 1D, Center). The amount of phase shift inthe Bmal1-ELuc rhythm was not different from the amount ofphase shift in activity offset but was significantly different fromthe amount of phase shift in the activity onset (two-way ANOVA,post hoc t test, P < 0.01) (Fig. 1D, Right).When a single light pulse was given at CT21.5, the amount of
phase shift was small, and the dissociation between the Per1-lucand Bmal1-ELuc circadian rhythms was not detected at statisti-cally significant levels (Fig. 2 and Fig. S1B). Nevertheless, anassociation of the circadian peak in clock gene expression withthe behavioral marker was observed, similar to that observedwith a light pulse at CT11.5. These results indicate that the cir-cadian rhythms of Per1 and Bmal1 expression are dissociablewhen an external perturbation produces a large phase-delay shiftin the SCN circadian rhythm in vivo.
Dissociation of Per1 and Bmal1 Oscillations in the SCN Slice. To testwhether the dissociation of Per1 and Bmal1 circadian oscillationreally occurs, we made SCN slices from the double-transgenicmice carrying reporters for both Per1 and Bmal1 expression and
0
7
14
0 0 012 12Local time
Day
s
Days
Behavior (counts/m
in)0
80
40
20
60
1 8 152 3 4 5 6 7 9 10 11 12 13 14
Per1-luc (SCN)
0
7
14
0 0 012 12Local time
Day
s
Days
RLU
(cou
nts/
min
)
120
0
30
Behavior (counts/m
in)0
80
40
20
60
1 8 152 3 4 5 6 7 9 10 11 12 13 14
Bmal1-ELuc (SCN)
18126 0 6 12 181
7
14
Day
s
18126 0 6 12 181
7
14
Day
s
Local time Local time
Per1-luc Bmal1-ELuc
60
90
RLU
(cou
nts/
min
)
120
0
30
60
90
Light pulse
A
B
C
150
180
150
180
Pha
se s
hifts
(h)
Days after light pulse-1 0 1 2 3 4 5 6 7 8 9
Activity onsetActivity offset
Per1-luc
Pha
se s
hifts
(h)
01
2
3
Days after light pulse-1 0 1 2 3 4 5 6 7 8 9
*
Bmal1-ELuc
Pha
se s
hifts
(h)
Days after light pulse-1 0 1 2 3 4 5 6 7 8 9
Bmal1-ELucPer1-luc
D4
5
01
2
34
5
01
2
34
5
Fig. 2. Light pulse-induced phase-advance shifts of circadian rhythms in the SCN and behavioral rhythm in freely moving adult mice. (A and B) Typicalexamples of a light-induced phase response at CT21.5 are illustrated for the Per1-luc (A) and Bmal1-ELuc (B) rhythms in the SCN together with a behavioralrhythm in double plotting (Left) and sequential plot (Right) as in Fig. 1 A and B. (C) Mean acrophases are illustrated for Per1-luc (Left) and Bmal1-ELuc (Right)rhythms together with the mean activity onsets and offsets of behavioral rhythms. Also see the legend of Fig. 1C. (D) The amount of phase shifts after a lightpulse are shown as the mean and SEM (n = 3) for Per1-luc (blue circles) and Bmal1-ELuc (green circles) (Left), for Per1-luc and two phase markers (activity onsetand offset) of behavioral rhythm (Center), and for Bmal1-ELuc and the phase markers (Right). Also see the legend of Fig. 1D.
Ono et al. PNAS | Published online April 17, 2017 | E3701
NEU
ROSC
IENCE
PNASPL
US
Dow
nloa
ded
by g
uest
on
Dec
embe
r 9,
202
0
monitored Per1-luc and Bmal1-ELuc simultaneously from thesame SCN slice. Circadian Per1-luc and Bmal1-ELuc rhythms inthe cultured SCN were separated successfully, as previouslyreported (21, 22), and persisted at least for 3 wk (Fig. S2).In the neonatal SCN (Fig. 3 A–C), double-plotted circadian
rhythms in Per1-luc and Bmal1-ELuc showed internal dissocia-tion, with a pattern similar to relative coordination in which thephase relation of two rhythms changed continuously in thecourse of free running (23). Similar patterns also were detectedin other neonatal SCN slices (Fig. S3A). The mean circadianperiod of Bmal1-ELuc mice was significantly shorter than that ofPer1-luc mice. The circadian period determined by χ2 periodo-gram was 22.7 ± 0.4 h for Bmal1-ELuc and 23.1 ± 0.4 h for Per1-
luc; these periods were significantly different (n = 9, paired t test,P < 0.01) (Fig. 3D). The periods determined by other methods,such as a linear regression line fitted to the consecutive cyclepeaks, confirmed the difference (Fig. S3B). As a result, the phaserelation between the Per1-luc and Bmal1-ELuc circadian rhythmschanged gradually and significantly during culturing (Fig. 3C andFig. S3A), regardless of the circadian phase marker used [i.e., theacrophase of a fitted cosine curve (Fig. 3I and Fig. S3C, Left) orthe peak of a detrended circadian rhythm (Fig. S3C, Right)]. Toknow whether the changes in the phase relationship between thetwo circadian rhythms reflected gradual and systematic alterationsof the rhythm shape, we compared the shape of rhythmicity on day3 and day 15 of culture and did not find any difference (Fig. S3D).
Per1-luc
Bmal1-ELuc
SCN slice
PMT
Per1-lucBmal1-ELuc
Perio
d (h
)22.0
22.5
23.5
23.0
24.0**
A B
DLong pass
X
filter
RLU
(1,0
00 c
ount
s/15
sec) R
LU (1,000 counts/15sec)0 5 10 15 20
Days in culture
0
5
-5
0
20
-20
00 0 012 12
Day
s in
cul
ture
5
10
15
20
Local timeC
Per1-luc Bmal1-ELuc
Per1-luc Bmal1-ELuc
Per1-luc
Bmal1-ELuc
SCN slice
PMT
Per1-lucBmal1-ELuc
Perio
d (h
)
22.0
22.5
23.5
23.0
24.0
E F
HLong pass
X
filter
RLU
(1,0
00 c
ount
s/15
sec) RLU
(1,000 counts/15sec)0 5 10 15 20Days in culture
0
5
-5
0
20
-20
00 0 012 12
Day
s in
cul
ture
5
10
15
20
Local timeG
Pha
se d
iffer
ence
(h)
(Per
1-lu
c - B
mal
1-E
Luc)
8101214161820
0 2 4 6 8 10 12 14 16 18Days in culture
I
Per1-luc Bmal1-ELuc
Per1-luc Bmal1-ELuc
**
Neonate SCNAdult SCN
*********
Fig. 3. Simultaneous measurement of Per1-luc and Bmal1-ELuc rhythms in the cultured SCN slice. (A and E) Experimental schemes of simultaneous mea-surement of Per1-luc and Bmal1-ELuc expression in neonatal (A) and adult (E) SCN slices. (B, C, F, and G) Sequential plots and double plots of Per1-luc (blue)and Bmal1-ELuc (green) circadian rhythms in a neonatal (B and C) and an adult (F and G) SCN slice in culture. In double-plotting, the time zone in which thebioluminescence is higher than the mean value of detrended data in a series is indicated by colored horizontal bars. Colored circles in a double plot (C and G)are acrophases. (D and H) Mean circadian periods calculated by χ2 periodogram (± SD) (21 d) of Per1-luc and Bmal1-ELuc in the neonatal (n = 9) (D) and adult(n = 5) (H) SCN slices are indicated by colored bars. Double asterisks (**) indicate a statistically significant difference between Per1-luc and Bmal1-ELuc (P <0.01, paired t test). (I) Mean daily phase differences in terms of acrophase between Per1-luc and Bmal1-ELuc circadian rhythms in the neonatal (red, n = 9) andadult (black, n = 5) SCN slices in culture for 17 d. Two-way repeated-measure ANOVA revealed significant differences between the neonatal and adult SCN;**P < 0.01, *P < 0.05 (post hoc t test).
E3702 | www.pnas.org/cgi/doi/10.1073/pnas.1613374114 Ono et al.
Dow
nloa
ded
by g
uest
on
Dec
embe
r 9,
202
0
In addition, the skewness of circadian rhythm was not significantlydifferent between day 3 and day 15 for either Per1-luc or Bmal1-ELuc. Furthermore, to exclude a possible effect of rhythm dampingduring culture on the determination of rhythm phase, we en-hanced the rhythm amplitude by replacing the culture mediumwith fresh medium; this replacement is known to increase therobustness of rhythmicity (Fig. S4). However, because the mediumexchange also is known to shift the circadian rhythm of the neo-natal SCN depending on the time of replacement (24), we se-lected the time to minimize the phase shifts. The phase relationbetween the two rhythms was kept essentially as it had been aftermedium exchange, and there was no systematic relation betweenthe rhythm amplitude and phase difference.A difference in the circadian period of the two clock gene
rhythms also was detected in the adult SCN slice (Fig. 3 E–H andFig. S3). The circadian period of Per1-luc rhythm was 23.4 ±0.1 h, and that of Bmal1-ELuc rhythm was 23.2 ± 0.1 h (n = 5), asignificant difference (paired t test, P < 0.01) (Fig. 3H). How-ever, the amount of dissociation in the adult SCN was less robust,and the phase difference between the two rhythms after day 11 ofculture was significantly smaller in the adult than in the neonatalSCN (P < 0.01, two-way repeated-measure ANOVA) (Fig. 3I).
Spontaneous Firing Rhythms in the Cultured SCN Are Dissociatedfrom Per1 and Bmal1 Oscillations. Spontaneous firing is regardedas an important output signal from a core loop in the SCN (25,26) and is well correlated with behavioral circadian rhythms (27).However, the mechanism by which spontaneous firings in theSCN regulate behavioral circadian rhythms remains unknown.To understand the relationship between the circadian firing rhythmand activity onset or offset, or between spontaneous firing andclock gene expression in the SCN, we measured Per1-luc or Bmal1-ELuc expression simultaneously with spontaneous firing in theneonatal SCN slices using a multielectrode array dish (MED) anda CCD camera (Fig. 4 A and B and Fig. S5 A and B). For thisexperiment, we used single-transgenic mice carrying a bio-luminescence reporter for Per1-luc or Bmal1-ELuc expression.The characteristics of the bioluminescent circadian rhythms inthese single-transgenic mice were similar to those in double-transgenic mice (Per1-luc, 23.3 ± 0.1 h, n = 7; Bmal1-ELuc,22.7 ± 0.3 h, n = 7; Student’s t test, P < 0.01) (Fig. S6).The circadian peak of Per1-luc rhythm in the neonatal SCN
slice on the MED was gradually phase delayed relative to that ofthe spontaneous firing rhythms (Fig. 4 C–E). On the other hand,the circadian peak of Bmal1-ELuc rhythm was gradually phaseadvanced relative to that of the firing rhythms (Fig. 4 C–E). Thephase difference between the firing rhythms and Per1 or Bmal1expression rhythms was significantly larger at day 15 of culturethan that at the beginning of culture (Student’s t test, P < 0.01)(Fig. 4F). The circadian periods of Per1-luc and Bmal1-ELucrhythms also were significantly different (Student’s t test, P <0.05) (Fig. 4G). These results indicate that the firing rhythms andPer1 or Bmal1 rhythms in the cultured SCN are dissociable andthat the firing rhythms are not a direct consequence of the cir-cadian oscillation of either the core feedback loop or the inter-locked Bmal1 loop.
Spatiotemporal Features of Per1 and Bmal1 Expression in the CulturedSCN. To know whether the dissociation between the Per1 andBmal1 circadian rhythms is caused by region-specific gene ex-pression in the SCN, we analyzed the spatiotemporal features ofPer1-luc and Bmal1-ELuc expression in the SCN during culturingusing an automated rhythm analysis with time-series images onthe pixel level (28). In both Per1-luc and Bmal1-ELuc rhythms,the mean circadian phase on day 2 or 3 of culture was signifi-cantly delayed in the ventral region relative to the dorsal region,by ca. 1 h for Per1-luc (paired t test, P < 0.05) and by ca. 3 h forBmal1-ELuc (paired t test, P < 0.05) (Fig. 4 J and L). The phase
relation between the dorsal and ventral regions was not changedon day 13 or 14 of culture in either the Per1-luc or the Bmal1-ELuc rhythm (paired t test, Per1-luc, P = 0.841; Bmal1-ELuc, P =0.932) (Fig. 4 J and L). We also analyzed the distribution ofcircadian peaks in the dorsal and ventral regions of the SCNusing Rayleigh plots (Fig. 4 K and M and Fig. S5 C and D). Themean vector of the circadian rhythm was not changed in thedorsal and ventral regions of the SCN during culturing for eitherPer1-luc (Fig. 4K) or Bmal1-ELuc (Fig. 4M), suggesting that theinternal synchrony of cellular circadian rhythms remained un-changed during culture. These results indicate that the dissoci-ation between Per1-luc and Bmal1-ELuc rhythms is not causedby a regional difference in the SCN neural network.
Simultaneous Measurement of Circadian Per1, Bmal1, Calcium, andSpontaneous Firing in the Cultured SCN. To understand furtherthe relationships among the circadian firing rhythm, the corefeedback loop, and the interlocked Bmal1 loop, we simulta-neously measured Per1 and Bmal1 gene expression, spontaneousfiring, and the intracellular calcium level from single culturedSCN slices. The SCN of double-transgenic mice (Per1-luc andBmal1-ELuc) expressing the calcium indicator GCaMP6s wascultured on a MED probe, and the bioluminescence of firefly(F)-luc for Per1-luc and of E-Luc for Bmal1-ELuc, the fluores-cence of GCaMP6s, and spontaneous firing were monitored (Fig.5 A and B, Fig. S7, and Movie S1). In addition to an endogenouscircadian oscillation as an output of the core feedback loop,intracellular calcium in the SCN is known to be controlled by theinput signals from the SCN neural network (29, 30). Plotting ofthe acrophase or of the circadian peak in each cycle revealed thatthe circadian rhythms in Bmal1-ELuc, spontaneous firing, andintracellular calcium were gradually phase advanced relative tothe Per1-luc circadian rhythms in the course of culturing (one-way repeated-measure ANOVA, P < 0.05 and post hoc Tukey–Kramer test) (Fig. 5 C–H). Interestingly, the calcium circadianrhythm was essentially phase-locked to the circadian firingrhythms, keeping a stable phase difference of about 1 h for morethan 10 d (Fig. 5 D and H). Thus, the phase relations of circadianrhythms in Per1-luc, Bmal1-ELuc, and firing changed graduallyduring culturing. Although the phase relations among these cir-cadian rhythms changed substantially during culture, the phaserelation of the Per1-luc and Bmal1-ELuc circadian rhythms(Δphase) in the dorsal and ventral SCN regions did not change(day 2–3, 1.7 ± 0.3 h; day13–14, 2.1 ± 0.9 h; paired t test, P =0.283) (Fig. 5 F and G and Fig. S7C), suggesting that the changein the phase relationship of the two rhythms occurred at thesame rate in both regions. The mean vector in these circadianrhythms was not different between day 2–3 and day 13–14,again suggesting that the internal synchrony of cellular circadianrhythms was maintained during culture (Fig. S7D). On the otherhand, the mean vector of the Bmal1-ELuc circadian rhythm wasslightly but significantly shorter than those of the other circadianrhythms, suggesting differential internal synchrony and/or slop-piness of the Bmal1-ELuc circadian rhythms.
DiscussionIn the present study we found that the Per1-luc and Bmal1-ELuccircadian rhythms were transiently dissociated when freely movingmice under DD experienced a phase-shifting light pulse. Thegene-expression rhythms were differentially associated with theactivity onset and offset of behavioral circadian rhythm. Similardissociation of Per1 and Bmal1 oscillation was observed in thecultured SCN slices of mice carrying a dual reporter system. Sucha transient dissociation between the circadian rhythms was de-tected only by continuous and simultaneous measurement ofmultiple rhythms from a single SCN. The dual reporter systemused in the present study enabled us to identify the change in thephase relation of two different rhythms. However, because the
Ono et al. PNAS | Published online April 17, 2017 | E3703
NEU
ROSC
IENCE
PNASPL
US
Dow
nloa
ded
by g
uest
on
Dec
embe
r 9,
202
0
method is based on luciferin–luciferase reaction and the sepa-ration of emitted light by filtering, these findings should beconfirmed by more direct measurement of the transcription ortranslation of clock genes in future. Most of the current directmethods are still difficult to apply because of insufficient timeresolution resulting from individual differences and use of alarge number of animals.Interestingly, the circadian rhythm in spontaneous firing was
dissociated from the Per1 and Bmal1 circadian rhythms but not
from the circadian rhythm of the intracellular calcium level.These findings indicate that the Bmal1 loop has an oscillatorynature similar to that of the core loop and behaves as a consti-tutional subunit of the circadian system in the SCN.We demonstrated the dissociation of circadian rhythms in
clock gene expression of freely moving mice when the circadianrhythms were on the way to steady-state free running after thelight pulse (Fig. 1). A similar dissociation of the circadian rhythmsof clock gene expression has been observed previously with shifts
Per1-lucSCN slice
A
CCD
Bmal1-ELuc
lens
Days
(x10
0 co
unts
/h)
(x10
0 co
unts
/h)
RLU
,Bmal1-ELuc
Per1-luc
RLU
,
4
-4
0
3
-3
0Firing rate (H
z)
03
12
1821
Per1-luc Bmal1-ELuc
0 5 10 15 20
69
15
03
12
18
69
15
00 0 012 12
Day
s
5
10
15
20
Local time
Neuronal activity Bmal1-ELuc
Firing rate (Hz)
00 0 012 12
Day
s
5
10
15
20
Local time
Neuronal activity Per1-luc
MED
Days
B
C
D0 3 6 9 12 15 18 21
Phas
e di
ffere
nce
(h)
0
(vs.
Neu
rona
l act
ivity
)
-2-4-6-8
-10-12-14
E
Days0 3 6 9 12 15 18 21
79
11131517
Per1-luc
Bmal1-ELuc
camera
F
Per1-luc
Bmal1-ELuc
0
2
4
6
-2
-4
**
G
Perio
d (h
)
*
23.0
24.5
Pha
se d
iffer
ence
(h)
vent
ral-d
orsa
l
Day2-3
Day13
-140
2.0
4.0
6.0
8.0
Mea
n ve
ctor
(r)
Day2-3
Day13
-140
0.2
0.4
0.6
0.8
1.0dorsal
Day2-3
Day13
-140
0.2
0.4
0.6
0.8
1.0ventral
Day
2-3
Day
13-1
4
Day
2-3
Day
13-1
4
Per1-luc Bmal1-ELucSlice A Slice ASlice B Slice BSlice C Slice CSlice D Slice D
Pha
se d
iffer
ence
(h)
vent
ral-d
orsa
l
Day2-3
Day13
-140
2.0
4.0
6.0
8.0
Day2-3
Day13
-140
0.2
0.4
0.6
0.8
1.0dorsal
Day2-3
Day13
-140
0.2
0.4
0.6
0.8
1.0ventral
-12h
+12h
-12h
+12h
H I
J K L M
Mea
n ve
ctor
(r)
Mea
n ve
ctor
(r)
Mea
n ve
ctor
(r)
Phas
e di
ffere
nce
(h)
(vs.
Neu
rona
l act
ivity
)
d
v
d d dd
dd d
vv v
vv v v
n.s.
n.s. n.s. n.s. n.s. n.s.
dd
vv
Phas
e di
ffere
nce
(h)
Fig. 4. Simultaneous measurement of Per1-luc and Bmal1-ELuc together with spontaneous firing in the cultured SCN. (A) Experimental scheme of simul-taneous measurement of Per1-luc, Bmal1-ELuc, and spontaneous firing in the neonatal SCN. (B) Bioluminescence images of Per1-luc (Left) and Bmal1-ELuc(Right) on a MED probe. (Scale bars: 100 μm.) (C) Sequential plots of Per1-luc (Upper, blue) and Bmal1-ELuc (Lower, green) rhythms with firing rhythm (black)from the entire area of the SCN. (D) Double-plotted circadian rhythms of Per1-luc (Upper, blue) and Bmal1-ELuc (Lower, green) with respective acrophases(closed circles) are illustrated together with double-plotted circadian rhythms of spontaneous firing (gray bars) with acrophase (black circles). The coloredzones indicate the time when bioluminescence was higher than the mean value of detrended data in a series. Also see the legend of Fig. 3. (E) Daily acrophasedifferences in Per1-luc (Upper) or Bmal1-ELuc (Lower) rhythms and firing rhythms in four SCNs; SCNs are shown in different colors. One SCN partially lacksfiring data. (F) Phase difference (mean ± SD) between days 13–14 and days 2–3 in culture in the Per1-luc (Left) and Bmal1-ELuc (Right) circadian rhythms.Negative values indicate phase delay, and positive values indicate phase advance (**P < 0.01, Student’s t test). (G) Circadian period determined by χ2
periodogram (mean ± SD) using records of Per1-luc and Bmal1-ELuc at day 14 in culture (*P < 0.05, Student’s t test). (H and I) Acrophase maps of Per1-luc (H)and Bmal1-ELuc (I) at days 2–3 (Upper) and 13–14 (Lower) in culture (n = 4). The mean acrophase was adjusted to 0 h. The color key indicates phase dis-tribution in hours. (J–M) The phase difference (mean ± SD) in Per1 (J) or Bmal1 (L) circadian rhythms on the pixel level between the ventral and dorsal regionsand the length of the mean vector of the respective rhythm (K and M) in the dorsal and ventral region at days 2–3 and 13–14 in culture.
E3704 | www.pnas.org/cgi/doi/10.1073/pnas.1613374114 Ono et al.
Dow
nloa
ded
by g
uest
on
Dec
embe
r 9,
202
0
in LD schedules (31, 32), but the relevance to behavioral outputswas not clear. Importantly, the phase shifts of Per1 and Bmal1rhythms in our study were closely associated with the phase shiftsof either the activity onset or offset of behavioral rhythm. Pre-viously, Vansteensel et al. (33) reported the dissociation betweenthe Per1-luc circadian rhythm in the SCN and behavioral rhythmsthat occurred following a shift in the LD cycle. The finding raisedthe possibility that the Per1-luc circadian rhythm reports only asubset of SCN neurons. In agreement with this report, the presentfindings suggest not only the existence of two coupled circadianoscillations with different molecular mechanisms but also theirdifferential association with the activity onset and offset of be-havioral rhythms. According to Pittendrigh’s hypothesis (13), ac-tivity onset and offset of nocturnal rodents are regulated by twodifferent oscillators, the E and M oscillator, respectively. Thehypothesis was based on the findings of rhythm splitting underconstant light and differential responses to light (12). In agree-ment with the above hypothesis, Honma et al. (11) demonstratedin rats that activity onset and offset showed different phase re-sponses to a single light pulse. Thus, the Per1-luc circadian rhythmin the SCN seems to associate with the E oscillator, and theBmal1-ELuc circadian rhythm seems to associate with the M os-cillator. Previous reports demonstrated separate regional pace-makers which behaved differently under a long and a short
photoperiod in the SCN (14, 15). The pacemaker correspondingto the E oscillator is likely located in the rostral part of SCN, andthat corresponding to the M oscillator is likely located in thecaudal part. The present findings further suggest that the twocircadian oscillations have different molecular mechanisms.Dissociation between the Per1 and Bmal1 oscillation also was
observed in the SCN ex vivo (Figs. 3–5). Because of a significantdifference in the circadian period, the phase relation betweenthe Per1 and Bmal1 oscillation changed gradually during cul-turing; this change was not the result of systematic changes in theshapes of circadian rhythms (Fig. S3 and Fig. S7). The core andinterlocked Bmal1 loop have been regarded as the origin of thecircadian rhythms of Per1 and Bmal1 expression (3–5). Thepresent results indicate that the two loops are coupled to eachother but exhibit ongoing differences in period and relativephase that suggest they may oscillate independently. They aredissociable by a light pulse in vivo and by prolonged free runningex vivo. The theoretical backgrounds of the present findings are acoupling of two oscillating feedback loops and a dependency ofthe circadian period on the relative strength of interlocked loopsas formulated in another associated loop (34). In addition, therelative coordination between of Per1-luc and Bmal1-ELucrhythms suggests that the core loop and Bmal1 loop are not in-dependent but are mutually interactive even under dissociation.
Per1-luc
Bmal1-ELuc
A
X
SCN slice
CCD camera
lens
no-filter
DichroicExcitation
Emission
LED
+hSyn-GCaMP6s
Days0 2 4 6 8 10 12 14 16
RLU
(x10
0 co
unts
/h)
Firing (Hz)
0
0
6
12
18
-6
-12
-18
36
91215
18
21
MED
CPer1-luc Bmal1-ELuc hSyn-GCaMP6s Neuronal activity
D E
Long passfilter
Per1-luc Bmal1-ELuc
GCaMP6s
B
⊿ p
hase
(h)
vs P
er1-
luc
0
2
4
6
8
10
12
14
*
******
Days0 3 6 9 12 15
00 0 012 12
Day
s 5
10
15
Local time
Per1-lucBmal1-ELuc
hSyn-GCaMP6sNeuronal activity
H
Day2-3
Day13-14
0
12
18 6
0
12
18 6
F
0h
12h
24h
0h
12h
24h
0
12
18 6
⊿Phase (h) Bmal1 vs Per1
0
12
18 6
Whole SCN dorsal SCN ventral SCN
0
12
18 6
0
12
18 6
dorsal SCN ventral SCNWhole SCN
Neuronal activityBmal1-ELuc
hSyn-GCaMP6s
G
16 *****
****** * ** * *
Day
2-3
Day
13-1
4
Phase difference map(Bmal1 vs Per1)
⊿Phase (h) Bmal1 vs Per1
Local time0 12 244 8 16 20
0
Day
s
4
8
12
16
Acrophase
Peak phase
d
d
v
v
Fig. 5. Simultaneous measurement of Per1-luc, Bmal1-ELuc, intracellular calcium, and spontaneous firing in the cultured SCN slice. (A) Experimental schemeof simultaneous multifunctional measurement in the neonate SCN slice. (B) Bioluminescence images of Per1-luc (Upper Left) and Bmal1-ELuc (Upper Right),fluorescence images of GCaMP6s (Lower Left), and a bright field image (Lower Right) of a cultured SCN slice on a MED probe. (Scale bars: 200 μm.)(C) Sequential plots of circadian Per1-luc, Bmal1-ELuc, GCaMP6s, and firing rhythms from the entire area of the SCN. Spontaneous firing was expressed as themean firing rate from electrodes covered by bilateral SCN. (D) Double plots of circadian rhythms of four measures. Colored circles are acrophases of circadianPer1-luc, Bmal1-ELuc, GCaMP6s, and firing rhythms. Yellow bars indicate the time zone in which GCaMP6s fluorescence is higher than the mean value ofdetrended data in a series. Also see the legend of Fig. 4D. (E) Longitudinal plotting of daily acrophases and of cycle peaks of the four circadian rhythmsdemonstrated in C. The features of free running are almost identical regardless of the phase marker. (F) Phase-difference maps (Per1-luc vs. Bmal1-ELuc) atdays 2–3 (Upper) and 13–14 (Lower) in culture. The color key indicates the phase difference in hours. (G) Rayleigh plots of phase difference on the pixel levelbetween Per1-luc and Bmal1-ELuc in the whole (Left), dorsal (Center), and ventral (Right) SCN are shown for days 2–3 (Upper) and 13–14 (Lower) in culture.A red line in a Rayleigh circle indicates the mean phase. (H) The mean daily phase difference, in terms of acrophase, between the Per1-luc and three othercircadian rhythms (n = 3). (*P < 0.05, vs. day 2, one-way repeated-measure ANOVA with post hoc Tukey–Kramer test).
Ono et al. PNAS | Published online April 17, 2017 | E3705
NEU
ROSC
IENCE
PNASPL
US
Dow
nloa
ded
by g
uest
on
Dec
embe
r 9,
202
0
The Bmal1 loop consists of a positive-feedback component ofROR and a negative-feedback component of RevErbα for Bmal1expression and potentially is capable of oscillating without os-cillation of the core loop, at least in theory (35).The circadian oscillation in the dorsomedial region of the SCN
was reported to be faster than that in the ventrolateral region(36). Regional differences also were reported in the intensity ofclock gene expression (37). Therefore the dissociation betweenthe two clock gene rhythms could be ascribed to regional dif-ferences in circadian oscillation. However, in the present study,the phase relation of the Per1 and Bmal1 expression rhythms wasnot different in the dorsal and ventral SCN during culturing,bringing into question the possibility that regional differences inrespective circadian oscillation might be a cause of dissociation.Recently, we found sharply differentiated clusters of cells thatexpressed the PER2 circadian rhythm with different periods inthe SCN of Cry1,2 double deficient mice (38). The cells in eachof these cell clusters showed no regional specificity but insteadwere distributed diffusely. The finding suggested that in thecircadian system of the SCN a specific type of oscillating cellbuilds up a constitutional oscillator that does not necessarilyshow regional specificity. The Per1- and Bmal1-specific oscillatorcells could be located diffusely throughout the SCN. Alterna-tively, the dissociation of two clock gene rhythms could becaused by the dissociation of the core loop and the Bmal1 loop insingle SCN cells. If so, within a single cell, the core molecularloop would be responsible for the activity onset and the Bmal1loop would be responsible for the activity offset of behavioralrhythm. In this case, we should assume different output signalsfrom a single cell. In any case, the coupling of two oscillationswas less strong in the neonatal SCN than in the adult SCN,suggesting a developmental change in the SCN circadian system.We had expected that the circadian firing rhythm in the SCN
slice might synchronize with either of the two rhythms of clockgene expression or show a splitting into two components. Con-trary to our expectation, the simultaneously determined circa-dian firing rhythms dissociated from both of the circadian gene-expression rhythms (Figs. 4 and 5). However, a coupling betweenthe circadian firing and intracellular calcium rhythms persistedfor at least for 2 wk in culture, suggesting that the two overtrhythms are regulated by the same oscillatory mechanism, whichcould be neither the core loop nor the interlocked Bmal1 loop.The cellular circadian rhythms in the SCN are likely regulated bytwo circadian oscillations of different origins: the intracellularmolecular feedback loop and the SCN network in which the cellis involved. The relative intensities of the two oscillations maydetermine the nature of the cellular circadian rhythms examined.Circadian firing and calcium rhythms would be affected byfeedback loops and the SCN network (39). A strong coherenceof the circadian firing rhythm to the intracellular calcium rhythmcould be interpreted in terms of causality rather than oscillatorycoupling. Neither Ca2+ nor firing rhythm in the SCN is associ-ated directly with the onset or offset of circadian behavioralrhythm. The present findings indicate that the links between theSCN circadian rhythms and behavior are more complicated thanpreviously thought (40).Taken together, the present findings indicate that external
perturbation, such as a single light pulse, transiently uncouplesinterlocked molecular loops that are separately associated withthe onset and offset of behavioral rhythms.
Materials and MethodsAnimals. Male and female Per1-luc (heterozygous or homozygous) (14) andBmal1-ELuc (heterozygous) (41) reporter mice of C57BL/6j background wereused. Per1-luc and Bmal1-ELuc mice were produced in the YS New Tech-nology Institute and the National Institute of Advanced Industrial Scienceand Technology, respectively, by injecting the reporter construct into fer-tilized eggs of C57BL/6j background. We back-crossed these mice with
C57BL/6j background mice for more than seven generations. Mice werereared in our animal quarters where environmental conditions were con-trolled (lights-on 6:00–18:00; light intensity ∼100 lx at the cage bottom;humidity 60 ± 10%). Mice had free access to food pellets and water. Ex-periments were conducted in compliance with the rules and regulationsestablished by the Animal Care and Use Committee of Hokkaido Universityunder the ethical permission of the Animal Research Committee of HokkaidoUniversity (approvals no. 13-0053 for in vivo experiments and no. 13-0064 forex vivo experiments).
Measurement of Behavioral Activity. Mice (nine males, four females) wereused in the in vivo study at age 2–6 mo. Mice were individually housed in apolycarbonate cage (115 mm wide × 215 mm long × 300 mm high), whichwas placed in a light-tight and air-conditioned box (40 × 50 × 50 cm). AnLED light source with an intensity of 150–250 lx on the ceiling of a box wasturned on when a light pulse was given after 5 d exposure to DD. Spon-taneous movements were measured by a passive infrared sensor, whichdetects a change in thermal radiation from an animal caused by bodymovements (42). The amount of behavioral activity was recorded auto-matically every minute by a computer (The Chronobiology Kit; StanfordSoftware System).
Surgery. Surgery was performed under isoflurane anesthesia as previouslydescribed (18, 19). To measure bioluminescence from the SCN in vivo, ahandmade guide cannula (i.d. 1.12 mm; o.d. 1.48 mm) was stereotaxicallyinserted into the brain 0.2 mm posterior to the bregma, 0.2 mm lateral fromthe midline, and 3.0 mm from the surface of the skull and was fixed to theskull with dental resin. After a recovery period of more than 4 d, a polymethylmethacrylate optical fiber (fiber diameter, 0.5 mm; surface cladding, 0.25 mmthick) was inserted into the guide cannula aimed at the SCN at a 5.8-mm depthfrom the surface of the skull and was fixed to the skull with dental resin.
More than 4 d after the insertion of the optical fiber, an osmotic pump(model 2002; Alzet; flow speed, 0.5 μL/h, pump volume, 200 μL) containingD-luciferin, a substrate of luciferase, was implanted in the peritoneal cavity.To deliver the substrate, the osmotic pump was filled with D-luciferinK (100 mM) dissolved in physiological saline. After each surgery, penicillin-G(40 units/g of body weight) was administered i.m. to prevent infection. As-pirin (120 mg/kg of body weight, administered orally) or buprenorphine(0.05–0.1 mg/kg of body weight, administered by s.c. injection) was used forpostoperative analgesia.
In Vivo Measurement of Bioluminescence. Three to five days after the im-plantation of the osmotic pump, bioluminescence from the SCN was mea-sured in freely moving mice under DD. The measurement was performedevery minute via an optical fiber connected to a photon-counting device (Invivo Kronos; Atto) (18, 19) equipped with a photomultiplier tube (Hama-matsu Photonics). Recorded data were fed into a computer and analyzed.
Histological Examination. Once the measurements were completed, mice wereanesthetized with ether and were intracardially perfused with physiologicalsaline followed by 4%paraformaldehyde in 0.1Mphosphate buffer (PB). Brainswere cryoprotected with 20% sucrose in 0.1 M PB and stored at −80 °C. Serial30-μm-thick coronal sections of the brain were made using a Leica cryostat andwere stained with cresyl violet to identify the placement of the tip of theoptical fiber.
Preparation of SCN Slices for Culture. To measure the bioluminescence fromthe SCN slice, mice were killed to harvest the SCN between 8:00 and 16:00 h.To measure the level in tissue, 300-μm-thick coronal slices from the SCN ofneonatal (postnatal day 7) mice were made with a tissue chopper (Mcllwain),and slices from the SCN of adult (2- to 6-mo-old) mice were made with amicroslicer (DTK-1000; Dosaka EM). The SCN tissue was dissected at the mid-rostrocaudal region, and paired SCN slices were cultured on a Millicell-CMculture insert (Millipore Corporation) under culture conditions as describedpreviously (43). Briefly, the slices were cultured in air at 36.5 °C with 1.2 mLDMEM (Invitrogen) with 0.1 mM D-luciferin K and 5% supplement solution.
Measurement of Bioluminescence in SCN Tissue. Bioluminescence in SCN tissuewas measured using a Lumicycle (Actimetrics) or Kronos (Atto) PMT at 10-minintervals with an exposure time of 1 min. For the simultaneous measurementof Per1 and Bmal1 expression, the bioluminescence of Per1-luc and Bmal1-ELucfrom the same SCN slice was monitored alternatively by a dish-type luminometerwith a turntable for eight recording dishes (Kronos, Atto). Bioluminescence wasmeasured for 15 s in the presence of a 600-nm long-pass filter (R60 filter; Hoya),
E3706 | www.pnas.org/cgi/doi/10.1073/pnas.1613374114 Ono et al.
Dow
nloa
ded
by g
uest
on
Dec
embe
r 9,
202
0
then in the presence of a 560-nm long-pass filter (O56 filter; Hoya), and thenwithout any filter. The measurement was repeated at 10-min intervals. Theintensities of Per1-luc and Bmal1-ELuc bioluminescence were calculatedas described previously (21, 22). For bioluminescence calculation, we used a600-nm long-pass filter.
Simultaneous Recordings of Bioluminescence and Spontaneous Firing from anSCN Slice. An SCN slice from a 3- to 5-d-old pup was cultured on a MED with64 electrodes. The size of an electrode was 20 × 20 μm (MED-P210A).Spontaneous firings were recorded using a MED 64 system (Alpha MEDScientific). Spike discharges with a signal-to-noise ratio >2.0 were collectedby Spike Detector software (Alpha MED Scientific) as described previously(43). The number of spikes per minute was calculated for each electrodecovered by the SCN slice.
The MED was placed in a mini-incubator installed on the stage of a mi-croscope (ECLIPSE TE2000-U or ECLIPSE E1000; Nikon). The culture conditionswere as described previously (43). Bioluminescence was recorded with a CCDcamera (ORCA-ІІ; Hamamatsu Photonics) cooled at −60 °C. The pixel size was4.3 × 4.3 μm.
Simultaneous Recordings of Per1-luc, Bmal1-ELuc, Calcium, and SpontaneousFiring from an SCN Slice. An SCN slice from a 3- to 5-d-old pup of Per1-lucand Bmal1-ELuc pup was cultured on a Millicell-CM culture insert. Aliquotsof adeno-associated virus (AAV) serotype rh10 harboring GCaMP6s, a ge-netically encoded calcium sensor under the control of human synapsin-1 promotor (University of Pennsylvania Gene Therapy Program Vector Core),were inoculated onto the surface of the cultured SCN slice 3–5 d after slicepreparation. On the day after AAV infection, the SCN slice was transferredonto the MED probe. Simultaneous measurement of bioluminescence,fluorescence, and neuronal activity began 7–10 d after the SCN was cultured.
The MED was placed in a mini-incubator installed on the stage of a mi-croscope (ECLIPSE-80i; Nikon) equipped with an EM-CCD camera (ImagEM;Hamamatsu Photonics). A fluorescent calcium sensor (GCaMP6s) was ex-cited at cyan color (475/28 nm) with an LED light source (Retra Light Engine;Lumencor) and was visualized with a 495-nm dichroic mirror and 520/35-nmemission filters (SemLock). To measure F-luc and E-Luc bioluminescence(Per1-luc and Bmal1-ELuc) separately, a long-pass filter (AT610, ChromaTechnology Corporation) was used. Bioluminescence with or without thefilter was measured every hour (exposure time, 29 min for each condition).Per1-luc and Bmal1-ELuc bioluminescence were calculated on the pixellevel with the method used for PMT measurement.
Data Analysis. For the analyses of behavioral rhythms in vivo, spontaneousmovements obtained every minute were used. The number of phase shiftswas determined on the basis of a double-plotted actograph by visual in-spection. A regression line was fitted to the succeeding activity onsets oroffsets during a steady-state free-run before and after the light pulse. The
circadian period was determined by the slope of the regression line, and theamount of daily phase shift was calculated from the phase difference be-tween the phase on the day of light pulse and that on the following day.Three chronobiology experts (D.O., S.H., and K.H.) participated in the visualinspection.
For the analyses of circadian bioluminescent rhythms in vivo and ex vivo,data obtainedwith a PMTwere smoothed by a 4-hmoving average for in vivodata and a 50-min moving average for ex vivo data. The smoothed data thenwere detrended by a 24-h moving average subtraction method (18, 19).Neuronal activity rhythms also were detrended by a 24-h moving averagesubtraction method. To compare the peak phases of circadian rhythms invivo and ex vivo, we used an acrophase obtained by ClockLab software(Actimetrics) or simply the peak phase in a cycle. The circadian period in theSCN slice was calculated by a periodogram or from the slope of a regressionline fitted to consecutive peak phases by the least-squares method. Thedifference in phase angle between two circadian rhythms was calculatedusing the acrophase of a best-fitted cosine curve or the peak of a circadiancycle. For time-lapse images obtained by a CCD camera, the properties ofcircadian rhythm in bioluminescence and fluorescence signals were ana-lyzed on the pixel level using custom-made software based on a cosinecurve-fitting method as described previously (39, 44). The dorsal and ven-tral areas of the SCN were separated by a line drawn at the midpointbetween the dorsal and ventral edges so that it crossed the third ventricleat a right angle. A Rayleigh plot was made using the Oriana4 software(Kovach Computing Service). Data are expressed as mean ± SD unlessotherwise indicated.
Statistics. Student’s t test was used when two independent group meanswere compared. A paired t test was used when two dependent group meanswere compared. A one-way ANOVA with a post hoc Tukey–Kramer test wasused to analyze data of a single time series. A two-way ANOVA with a posthoc t test was used when the data of two independent time series werecompared (Statview or Statcel 3).
ACKNOWLEDGMENTS. We thank H. Tei for providing the Per1-luciferasereporter plasmid; M. Shimogawara and H. Kubota for the development ofan in vivo bioluminescence measurement system; K. Baba for technical ad-vice; T. Ueda for the analysis program; M. Mieda for advice on AAV purifi-cation; M. P. Butler for thorough discussions; and the Genetically-EncodedNeuronal Indicator and Effector (GENIE) Project and the Janelia Farm Re-search Campus of the Howard Hughes Medical Institute for sharing GCaMP6sconstructs. This work was supported in part by the Uehara Memorial Foun-dation; the Nakajima Foundation; the Project for Developing InnovationSystems of the Ministry of Education, Culture, Sports, Science and Technol-ogy (MEXT); the Creation of Innovation Centers for Advanced Interdisciplin-ary Research Areas Program, MEXT; and Japan Society for the Promotion ofScience Grants in Aid for Scientific Research (KAKENHI) Grants 15H04679,26860156, and 15K12763.
1. Moore RY, Eichler VB (1972) Loss of a circadian adrenal corticosterone rhythm fol-
lowing suprachiasmatic lesions in the rat. Brain Res 42:201–206.2. Stephan FK, Zucker I (1972) Circadian rhythms in drinking behavior and locomotor
activity of rats are eliminated by hypothalamic lesions. Proc Natl Acad Sci USA 69:
1583–1586.3. Reppert SM, Weaver DR (2002) Coordination of circadian timing in mammals. Nature
418:935–941.4. Preitner N, et al. (2002) The orphan nuclear receptor REV-ERBalpha controls circadian
transcription within the positive limb of the mammalian circadian oscillator. Cell 110:
251–260.5. Triqueneaux G, et al. (2004) The orphan receptor Rev-erbalpha gene is a target of the
circadian clock pacemaker. J Mol Endocrinol 33:585–608.6. Sato TK, et al. (2004) A functional genomics strategy reveals Rora as a component of
the mammalian circadian clock. Neuron 43:527–537.7. Liu AC, et al. (2008) Redundant function of REV-ERBalpha and beta and non-essential
role for Bmal1 cycling in transcriptional regulation of intracellular circadian rhythms.
PLoS Genet 4:e1000023.8. Shigeyoshi Y, et al. (1997) Light-induced resetting of a mammalian circadian clock is
associated with rapid induction of the mPer1 transcript. Cell 91:1043–1053.9. Albrecht U, Sun ZS, Eichele G, Lee CC (1997) A differential response of two putative
mammalian circadian regulators, mper1 and mper2, to light. Cell 91:1055–1064.10. Aschoff J, Hoffmann K, Pohl H, Wever R (1975) Re-entrainment of circadian rhythms
after phase-shifts of the Zeitgeber. Chronobiologia 2:23–78.11. Honma K, Honma S, Hiroshige T (1985) Response curve, free-running period, and
activity time in circadian locomotor rhythm of rats. Jpn J Physiol 35:643–658.12. Pittendrigh CS, Daan S (1976) A functional analysis of circadian pacemakers in noc-
turnal rodents. V. Pacemaker structure: A clock for all seasons. J Comp Physiol A
Neuroethol Sens Neural Behav Physiol 106:333–355.
13. Pittendrigh CS (1974) Circadian oscillation in cells and the circadian organization ofmulti-cellular systems. The Neurosciences, Third Study Program, eds Schmitt FO,Worden FG (The MIT Press, Cambridge, MA), pp 437–458.
14. Inagaki N, Honma S, Ono D, Tanahashi Y, Honma K (2007) Separate oscillating cellgroups in mouse suprachiasmatic nucleus couple photoperiodically to the onset andend of daily activity. Proc Natl Acad Sci USA 104:7664–7669.
15. Naito E, Watanabe T, Tei H, Yoshimura T, Ebihara S (2008) Reorganization of thesuprachiasmatic nucleus coding for day length. J Biol Rhythms 23:140–149.
16. Evans JA, Leise TL, Castanon-Cervantes O, Davidson AJ (2013) Dynamic interactionsmediated by nonredundant signaling mechanisms couple circadian clock neurons.Neuron 80:973–983.
17. Myung J, et al. (2015) GABA-mediated repulsive coupling between circadianclock neurons in the SCN encodes seasonal time. Proc Natl Acad Sci USA 112:E3920–E3929.
18. Ono D, Honma K, Honma S (2015) Circadian and ultradian rhythms of clock geneexpression in the suprachiasmatic nucleus of freely moving mice. Sci Rep 5:12310.
19. Ono D, Honma S, Honma K (2015) Circadian PER2:LUC rhythms in the olfactory bulbof freely moving mice depend on the suprachiasmatic nucleus but not on behaviourrhythms. Eur J Neurosci 42:3128–3137.
20. Comas M, Beersma DG, Spoelstra K, Daan S (2006) Phase and period responses of thecircadian system of mice (Mus musculus) to light stimuli of different duration. J BiolRhythms 21:362–372.
21. Noguchi T, Ikeda M, Ohmiya Y, Nakajima Y (2008) Simultaneous monitoring of in-dependent gene expression patterns in two types of cocultured fibroblasts withdifferent color-emitting luciferases. BMC Biotechnol 8:40.
22. Noguchi T, et al. (2010) Dual-color luciferase mouse directly demonstrates coupledexpression of two clock genes. Biochemistry 49:8053–8061.
23. Aschoff J (1981) Freerunning and entrained circadian rhythms. Handbook of BehavioralNeurobiology, ed Aschoff J (Plenum, New York), Vol 4, Biological Rhythms, pp 81–93.
Ono et al. PNAS | Published online April 17, 2017 | E3707
NEU
ROSC
IENCE
PNASPL
US
Dow
nloa
ded
by g
uest
on
Dec
embe
r 9,
202
0
24. Nishide SY, Honma S, Honma K (2008) The circadian pacemaker in the cultured su-
prachiasmatic nucleus from pup mice is highly sensitive to external perturbation. Eur J
Neurosci 27:2686–2690.25. Schwartz WJ, Gross RA, Morton MT (1987) The suprachiasmatic nuclei contain a
tetrodotoxin-resistant circadian pacemaker. Proc Natl Acad Sci USA 84:1694–1698.26. Jones JR, Tackenberg MC, McMahon DG (2015) Manipulating circadian clock neuron
firing rate resets molecular circadian rhythms and behavior. Nat Neurosci 18:373–375.27. Nakamura W, et al. (2008) In vivo monitoring of circadian timing in freely moving
mice. Curr Biol 18:381–385.28. Yoshikawa T, et al. (2015) Spatiotemporal profiles of arginine vasopressin transcrip-
tion in cultured suprachiasmatic nucleus. Eur J Neurosci 42:2678–2689.29. Ikeda M, et al. (2003) Circadian dynamics of cytosolic and nuclear Ca2+ in single su-
prachiasmatic nucleus neurons. Neuron 38:253–263.30. Hastings MH, Maywood ES, O’Neill JS (2008) Cellular circadian pacemaking and the
role of cytosolic rhythms. Curr Biol 18:R805–R815.31. Reddy AB, Field MD, Maywood ES, Hastings MH (2002) Differential resynchronisation
of circadian clock gene expression within the suprachiasmatic nuclei of mice subjected
to experimental jet lag. J Neurosci 22:7326–7330.32. Kiessling S, Eichele G, Oster H (2010) Adrenal glucocorticoids have a key role in cir-
cadian resynchronization in a mouse model of jet lag. J Clin Invest 120:2600–2609.33. Vansteensel MJ, et al. (2003) Dissociation between circadian Per1 and neuronal and
behavioral rhythms following a shifted environmental cycle. Curr Biol 13:1538–1542.34. Yan J, et al. (2014) An intensity ratio of interlocking loops determines circadian period
length. Nucleic Acids Res 42:10278–10287.
35. Uryu K, Tei H (2016) Feedback loops interlocked at competitive binding sites amplifymammalian circadian oscillations. J. Chronobiol 22:155.
36. Koinuma S, et al. (2013) Regional circadian period difference in the suprachiasmaticnucleus of the mammalian circadian center. Eur J Neurosci 38:2832–2841.
37. Evans JA, et al. (2015) Shell neurons of the master circadian clock coordinate thephase of tissue clocks throughout the brain and body. BMC Biol 13:43.
38. Ono D, Honma S, Honma K (2016) Differential roles of AVP and VIP signaling in thepostnatal changes of neural networks for coherent circadian rhythms in the SCN. SciAdv 2:e1600960.
39. Enoki R, et al. (2012) Topological specificity and hierarchical network of the circadiancalcium rhythm in the suprachiasmatic nucleus. Proc Natl Acad Sci USA 109:21498–21503.
40. Ramkisoensing A, Meijer JH (2015) Synchronization of biological clock neurons bylight and peripheral feedback systems promotes circadian rhythms and health. FrontNeurol 6:128.
41. Noguchi T, Ikeda M, Ohmiya Y, Nakajima Y (2012) A dual-color luciferase assay systemreveals circadian resetting of cultured fibroblasts by co-cultured adrenal glands. PLoSOne 7:e37093.
42. Abe H, et al. (2001) Clock gene expressions in the suprachiasmatic nucleus and otherareas of the brain during rhythm splitting in CS mice. Brain Res Mol Brain Res 87:92–99.
43. Ono D, Honma S, Honma K (2013) Cryptochromes are critical for the development ofcoherent circadian rhythms in the mouse suprachiasmatic nucleus. Nat Communs 4:1666.
44. Mieda M, et al. (2015) Cellular clocks in AVP neurons of the SCN are critical for in-terneuronal coupling regulating circadian behavior rhythm. Neuron 85:1103–1116.
E3708 | www.pnas.org/cgi/doi/10.1073/pnas.1613374114 Ono et al.
Dow
nloa
ded
by g
uest
on
Dec
embe
r 9,
202
0