Development of the circadian oscillator duringdifferentiation of mouse embryonic stem cells in vitroKazuhiro Yagitaa,b,1, Kyoji Horiec, Satoshi Koinumad, Wataru Nakamurae, Iori Yamanakab, Akihiro Urasakif,Yasufumi Shigeyoshid, Koichi Kawakamif, Shoichi Shimadaa, Junji Takedac, and Yasuo Uchiyamag

aDepartment of Neuroscience and Cell Biology and cDepartment of Social and Environmental Medicine, Osaka University Graduate School of Medicine, Osaka565-0871, Japan; bCenter of Exellence Unit of Circadian Systems, Nagoya University Graduate School of Science, Nagoya 464-8602, Japan; dDepartment ofAnatomy, Kinki University School of Medicine, Osaka 589-8511, Japan; eLaboratory of Oral Chronobiology, Osaka University Graduate School of Dentistry,Osaka 565-0871, Japan; fDepartment of Molecular and Developmental Biology, National Institute of Genetics, and Department of Genetics, GraduateUniversity for Advanced Studies, Shizuoka 411-8540, Japan; and gDepartment of Cell Biology and Neuroscience, University of Juntendo, Tokyo 113-8421,Japan

Edited* by Joseph S. Takahashi, University of Texas Southwestern Medical Center, Dallas, TX, and approved January 5, 2010 (received for review November18, 2009)

The molecular oscillations underlying the generation of circadianrhythmicity in mammals develop gradually during ontogenesis.However, the developmental process of mammalian cellular circa-dianoscillator formation remainsunknown. Indifferentiatedsomaticcells, the transcriptional–translational feedback loops (TTFL) consist-ing of clock genes elicit the molecular circadian oscillation. Using abioluminescence imaging system to monitor clock gene expression,we show here that the circadian bioluminescence rhythm is notdetected in themouse embryonic stem (ES) cells, and that the ES cellslikely lack TTFL regulation for clock gene expression. The circadianclock oscillation was induced during the differentiation culture ofmouse ES cellswithoutmaternal factors. In addition, reprogrammingof the differentiated cells by expression of Sox2, Klf4, Oct3/4, andc-Myc genes, which were factors to generate induced pluripotentstem (iPS) cells, resulted in the re-disappearance of circadian oscilla-tion.These resultsdemonstrate thatan intrinsic programcontrols theformation of the circadian oscillator during the differentiation proc-essofEScells invitro.Thecellulardifferentiationand reprogrammingsystemusing culturedES cells allowsus to observe the circadian clockformationprocessandmayhelpdesignnewstrategies tounderstandthe key mechanisms responsible for the organization of the molec-ular oscillator in mammals.

circadian clock | induced pluripotent stem cells | real-time monitor

The circadian rhythm is a fundamental biological system inmammals involved in the regulation of various physiologicalfunctions such as the sleep–wake cycle, energy metabolism, andthe endocrine system (1, 2). These physiological rhythms developgradually in the first year of life in humans (3). It is well known thatthe human sleep–wake rhythm is generated within a few monthsafter birth. However, a weak circadian rhythm of core body tem-perature is present immediately after birth, suggesting that thedevelopment of the human circadian rhythms starts during fetallife. In fact, recent studies in rodents have suggested the appear-ance of circadian molecular rhythms in the suprachiasmaticnucleus (SCN) a few days before birth (4). However, little infor-mation is available on the development of themammalian cellularcircadian oscillator.In mammals, molecular oscillation of the circadian clock consists

of interlocked positive and negative transcription/translation feed-back loops (TTFL) involvinga set of clockgenes andclock-controlledoutput genes that link the oscillator to the clock-controlled processes(5). CLOCK and BMAL1 are basic-helix–loop–helix (bHLH) PAStranscription factors that heterodimerize and transactivate the coreclock genes such as Period (Per1, -2, and -3),Cryptochrome (Cry1 andCry2), and Rev-Erbα (2, 5, 6). PER and CRY proteins suppress theactivity of the CLOCK/BMAL1, whereas REV-ERBα suppressesBmal1 gene expression.In this study, we focused on the development of the mammalian

circadian oscillator during the differentiation culture of mouse

embryonic stem (ES) cells. Because the mouse ES cells are self-renewing pluripotent cells that have the potential to differentiateinto nearly all cell types of the mouse body, we investigated in thisstudy the formationprocess of the circadianoscillator in a cell culturesystem of mouse ES cells and differentiated cells derived thereof.The main results of the study were (i) the circadian feedback

loops did not seem to oscillate in ES cells; (ii) differentiationculture of ES cells in vitro induced molecular oscillation of thecircadian clock; and (iii) the developed oscillator disappearedagain when the differentiated cells were reprogrammed byintroducing four factors, Oct3/4, Sox2, Klf4, and c-Myc. Theresults strongly suggest that the generation of the circadianoscillator depends on an intrinsic program that correlates withthe cellular differentiation status of mammalian cells.

ResultsEstablishment of Luminescent ES Cell Lines. To observe the internalcircadian clockoscillator in living cells, a real-timebioluminescenceassay was established using firefly luciferase as a reporter, which isdriven by the promoter of clock genes (7–11). In addition tomPer2and Bmal1, we reported previously that a clock-controlled Dbppromoter-driven luciferase reporter is also available to read out thecircadianmolecular oscillator in living cells (12). Thus in this study,we used Bmal1-promoter- and Dbp-promoter-driven luciferasereporters to visualize the intrinsic cellular circadian clock. Wecotransfected ES cells with Tol2 transposase (TP) expressionplasmid and Tol2 transposon-based Bmal1:luc or Dbp:luc reportervectors (Bmal1:luc-pT2A, Dbp:luc-pT2A) (13, 14). The Tol2transposon was originally discovered inMedaka fish, and the Tol2-based vector is considered a highly efficient gene transfer system inmouse ES cells (13, 15). All picked Bmal1:luc-pT2A ES cell clones(23 clones) were bioluminescent (Fig. 1A).

ES Cells Do Not Exhibit Circadian Bioluminescence Oscillation. Usingthese Bmal1:luc stably transfected ES cell lines, we investigatedthemBmal1 promoter-driven bioluminescence after changing themedium to luciferin-containing ES medium (ESM) by real-timephotomultiplier-tube (PMT)-based bioluminescence assay. Westimulated the ES cell culture with two known clock-synchronizingagents, forskolin and dexamethasone. The PMT-based analysisshowed no circadian bioluminescence oscillation in both syn-chronizing stimulations (Fig. 1B and Fig. S1A). These results

Author contributions: K.Y., designed research; K.Y., K.H., S.K., I.Y., and Y.S. performedresearch; A.U. and K.K. contributed new reagents/analytic tools; K.Y., K.H., S.K., W.N.,S.S., J.T., and Y.U. analyzed data; and K.Y. wrote the paper.

The authors declare no conflict of interest.

*This Direct Submission article had a prearranged editor.1Towhom correspondence should be addressed. E-mail: kyagita@anat1.med.osaka-u.ac.jp.

This article contains supporting information online at www.pnas.org/cgi/content/full/0913256107/DCSupplemental.

3846–3851 | PNAS | February 23, 2010 | vol. 107 | no. 8 www.pnas.org/cgi/doi/10.1073/pnas.0913256107

Dow

nloa

ded

by g

uest

on

July

2, 2

021

http://www.pnas.org/cgi/data/0913256107/DCSupplemental/Supplemental_PDF#nameddest=sfig01mailto:kyagita@anat1.med.osaka-u.ac.jphttp://www.pnas.org/cgi/content/full/0913256107/DCSupplementalhttp://www.pnas.org/cgi/content/full/0913256107/DCSupplementalwww.pnas.org/cgi/doi/10.1073/pnas.0913256107

suggest two possibilities. One is that the ES cells exhibit circadianrhythms but they are de-synchronized. Another is that the ES cellsdo not have a functional circadian clock oscillator. Although theyare desynchronized, or do not have a functional circadian clockoscillator. To answer this question, we used a microscope-basedhigh-sensitivity CCD camera system (LV200, Olympus) for real-time monitoring of Bmal1:luc bioluminescence of ES cells.Healthy undifferentiated ES cells form colonies of densely packedcells with smooth outlines, and ES cells move dynamically in thecolony. Thus, it is quite difficult to trace individual cells for timesas long as 72 h. To solve this problem, single or doubling cells werechosen at the first time point and traced as single cell-derivedcolonies. However, when colonies merged, we omitted them frombioluminescence analysis. The results demonstrated that coloniesobserved from single cells did not exhibit circadian oscillation ofbioluminescence intensity (Fig. 1C and Movie S1).To eliminate the possibility that these results are specific to the

KY1.1 ES cell line (F1 hybrid of C57BL/6J and 129S6/SvEvTac),we examined the circadian clock oscillation using other ES celllines such as E14Tg2a and EB5 (derived from 129P2/OlaHsd).These ES cells were stably transfected with Bmal1:luc-pT2A orDbp:luc-pT2A reporter vectors through a Tol2 transposon strat-egy. Using Bmal1:luc-pT2A or Dbp:luc-pT2A stably transfectedE14Tg2a and EB5 ES cells, we observed bioluminescence activ-

ities by a PMT-based real-time circadian clock monitoring system.For both ES cell cultures and both reporters, we did not detectapparent circadian fluctuation in bioluminescence from ES cellcultures (Fig. S1B). Microscopic analysis of Bmal1:luc EB5 cellsand Dbp:luc EB5 cells also showed that circadian fluctuation ofbioluminescence was not observed in EB5 ES cells at the singlecell/colony level (Fig. S1C). These series of results indicate thatES cells likely lack the capability of expressing the functionalcircadian clock oscillation. This conclusion is compatible withthose of recent ontogenic studies showing noncycling circadianmolecular clocks in early mammalian embryos (4, 16).

Development of Circadian Oscillation During the All-Trans RetinoicAcid Induced Differentiation Culture of ES Cells. Next, we moni-tored the circadian molecular oscillator during the cellular differ-entiation process of ES cells in the culture system. Previous studiesindicated that the self-sustaining circadian oscillator resides notonly in the central clock of SCN but also in the majority ofperipheral cells inmammals and even in cultured cell lines (8, 9, 17,18). In this regard, all-trans retinoic acid (RA) is used for ES celldifferentiation, because theRA treatment has been established as asimple procedure for differentiation of ES cells, mimicking thesequential Hox gene expression profiles seen in early embryos (19).KY1.1ES cells stably expressing theBmal1:luc reporter genewere

cultured in 1 μM RA-containing medium without leukemia inhib-itory factor (LIF). Then, PMT-based real-time bioluminescenceassays were performed at days 3, 8, and 15 following the start of RAtreatment (Fig. 2A).For synchronization todetect the circadianclockoscillation of differentiating cells, we used forskolin because thisstimulationdirectly affects the intracellular cAMP level andhas beenestablished as a clock-resetting factor for various types of cells (20,21).As shownabove, undifferentiatedES cell cultures did not exhibitcircadianfluctuations in their bioluminescence (Fig. 1B). After 3-dayRA differentiation culture, circadian fluctuation of biolumines-cence was still not observed, whereas one-cycle up-and-down bio-luminescence was observed in the cells at day 8 after RA treatmentwith 10 μM of forskolin synchronization (Fig. 2A Left and Center).However, cyclic bioluminescence no longer persisted after one cycle.In contrast, at day 15, the cells exhibited circadian bioluminescenceoscillationat least four cycles after synchronizing stimulation (Fig. 2ARight). These results suggest that a self-sustaining circadian oscillatorwas generated during differentiation by day 15.To confirm the developing process of circadian oscillation

during ES cell differentiation, EB5 ES cells stably expressing theDbp:luc reporter gene were analyzed. RA-induced differentiationculture was prepared using the same protocol as for KY1.1 EScells. RA-induced differentiation of EB5 ES cells also resulted inthe development of circadian bioluminescence oscillation drivenbyDbp promoter at day 14 of the differentiation culture (Fig. 2B).These results suggest that the circadian clock oscillation seems todevelop gradually during the ES cell differentiation culture.For quantitative evaluation of the rhythmicity, the above data

were subjected to fast Fourier transform (FFT) analysis (22). Boththe relative power and the relative amplitude were significantlyhigher after 15 days compared with 3 days of differentiation cultureinKY1.1 ES cells (Fig. 2C; P< 0.001). Similar to those inKY1.1 EScells, both the relative power and the relative amplitude of EB5-derived differentiated cells after 14 days were significantly higherthan in ES cells or 5-day differentiating cells (Fig. 2D; P < 0.001).Period lengths of differentiated cells were 22.73 ± 0.39 h (KY1.1RA Day 15, n= 10) and 23.09 ± 0.24 h (EB5 RADay 14, n= 11),respectively (Fig. 2E). Importantly, the peak phases of Bmal1:luc-and Dbp:luc-driven bioluminescence were in near anti-phase (Fig.2F), suggesting functional canonical circadian feedback loops inthese differentiated cells. These results indicate the development ofcellular circadian oscillation after RA-induced differentiation cul-ture of ES cells.

BABF

Luc

Forskolin 10 µM

Bio

lum

ines

cenc

e ES cell

CES cell

40000

0

Fig. 1. ES cells lack circadian bioluminescence rhythms. (A) Tol2 transposon-based Bmal1:luc reporter vectors were cotransfected with Tol2 transposaseexpression vector to establish the Bmal1:luc-pT2A stably transfected mouseES cells. The obtained Bmal1:luc-pT2A stably transfected ES coloniesexhibited apparent bioluminescence (Luc). (B) Photomultiplier-tubes (PMT)-based photon-counting assays were performed using Bmal1:luc ES cells afterforskolin stimulation. Bioluminescence showed no circadian oscillation. (C)Microscopic bioluminescence image analysis. Single ES cells were plated ontoa feeder cell layer and growing colonies raised from single cells were markedto trace the Bmal1:luc-driven bioluminescence, using an LV200 microscopicbioluminescence image analyzer (Movie S1 and Materials and Methods).

Yagita et al. PNAS | February 23, 2010 | vol. 107 | no. 8 | 3847

NEU

ROSC

IENCE

Dow

nloa

ded

by g

uest

on

July

2, 2

021

http://www.pnas.org/content/vol0/issue2010/images/data/0913256107/DCSupplemental/sm01.avihttp://www.pnas.org/cgi/data/0913256107/DCSupplemental/Supplemental_PDF#nameddest=sfig01http://www.pnas.org/cgi/data/0913256107/DCSupplemental/Supplemental_PDF#nameddest=sfig01http://www.pnas.org/content/vol0/issue2010/images/data/0913256107/DCSupplemental/sm01.avi

Continuous monitoring by PMT-based bioluminescence fromthe cells synchronized at day 11 after RA-induced differentiationculture also indicated that only a weak oscillation appeared at day11 but much higher-amplitude oscillation was detected after thesecond synchronization at day 15 of differentiation culture (Fig.S2). These results suggest that the circadian molecular oscillator

develops gradually in differentiating cells in RA treatmentdifferentiation culture.We next performed microscopic analysis to observe the bio-

luminescence in each cell (Fig. S3). Microscopic analysis of thedifferentiating cells revealed lack of apparent molecular oscillationin the cells at day 8 of RA-induced differentiation culture (Fig. S3A

E

A

**

B

biol

umin

esce

nce ES cells

5 6 7 8 9 Day after RA treatment

biol

umin

esce

nce RA Day5

14 15 16 17 18 Day after RA treatment

biol

umin

esce

nce

RA Day14

F

C D

Fig. 2. Circadian clock formation during ES cell differentiation in the cell culture system. (A) Bmal1:luc stably transfected KY1.1 ES cells were dif-ferentiated by 1 μM of retinoic acid (RA) and PMT-based bioluminescence was monitored on days 3, 8, and 15. Cells were synchronized by 10 μM offorskolin treatment (blue arrows). (B) A PMT-based bioluminescence monitor was performed using Dbp:luc stably transfected EB5 ES cells and RA-induced differentiated cells. Cells were synchronized by 10 μM of forskolin treatment (blue arrows). (C ) FFT spectral power analysis of RA Day 3 cells(n = 3) and RA Day 15 cells (n = 10) differentiated from Bmal1:luc stably transfected ES cells showed that both relative power (Left) and relativeamplitude (Right) were significantly induced after 15 days of differentiation culture. (D) FFT spectral power analysis of Dbp:luc stably transfected EScells (n = 9), RA Day 5 differentiated cells (n = 3), and RA Day 14 differentiated cells (n = 11). (Left) FFT-relative power. (Right) FFT-relative amplitude.(E ) Period length of bioluminescence oscillation obtained from RA Day 15 cells differentiated from Bmal1:luc stably transfected KY1.1 ES (22.73 ±0.39 h, n = 10) cells and RA Day 14 cells differentiated from Dbp:luc stably transfected EB5 ES cells (23.09 ± 0.24 h, n = 11). (F ) Peak phases of thebioluminescence oscillation obtained from Bmal1:luc RA Day 15 differentiated cells (first peak, 15.92 ± 0.24 h; second peak, 37.77 ± 0.31 h aftersynchronization) and Dbp:luc RA Day 14 differentiated cells (second peak, 26.35 ± 0.11 h; third peak, 49.53 ± 0.34 h after synchronization). Data in C–Eare mean ± SEM.

3848 | www.pnas.org/cgi/doi/10.1073/pnas.0913256107 Yagita et al.

Dow

nloa

ded

by g

uest

on

July

2, 2

021

http://www.pnas.org/cgi/data/0913256107/DCSupplemental/Supplemental_PDF#nameddest=sfig02http://www.pnas.org/cgi/data/0913256107/DCSupplemental/Supplemental_PDF#nameddest=sfig02http://www.pnas.org/cgi/data/0913256107/DCSupplemental/Supplemental_PDF#nameddest=sfig03http://www.pnas.org/cgi/data/0913256107/DCSupplemental/Supplemental_PDF#nameddest=sfig03www.pnas.org/cgi/doi/10.1073/pnas.0913256107

and Movie S2). Interestingly, although it was unstable and lowamplitude, someof the cells differentiatedbyRAfor12days showednear-circadian bioluminescence fluctuation (Fig. S3B and MovieS3). Then, the circadian bioluminescence oscillation with higheramplitudewaselicited in the cells atDay15 (Fig. S3CandMovieS4).The FFT relative power of RA-induced 15-day differentiated cellswas significantly higher than that ofES cells at the cellular level (Fig.S3D). FFT spectral power of cellular bioluminescence fluctuationincreased in RA Day 15 cells relative to RA Day 8 and RA Day 12cells (Fig. S3E), which supported the findings obtained through thePMT-based analysis. Because the data at days 12 and 15 wereobtained subsequently from the same sample, the circadian oscil-lator of each cell seems to develop gradually.

Development of CircadianOscillation inNeural StemCells Differentiatedfrom the ES Cell. To determine whether the formation of the circa-dian oscillator is an RA-induced differentiation-specific event ornot, we established neural stem (NS) cells from Bmal1:luc stablytransfected KY1.1 ES cells using the previously reported protocol(23). Fig. S4A presents a differentiation procedure used. Estab-lished adherentNS cells were spindle shaped and expressed nestin,a NS cell marker (Fig. S4B andC). The circadian bioluminescenceoscillation was clearly observed in these established NS cell cul-tures by PMT-based bioluminescence analysis after forskolinstimulation (Fig. S4D). This result suggests that generation of thecellular circadian clock is likely to correlate with cellular differ-entiation of ES cells rather than a specific procedure such asRA treatment.

Disappearance of Circadian Oscillation in the Reprogrammed Cells.To further investigate the relationship between circadian clockgeneration and cellular differentiation, we reprogrammed theestablished NS cells following the method used to establish theinduced pluripotent stem (iPS) cells using the recently reportedreprogramming factors (Oct3/4, Sox2, Klf4, and c-Myc) (24). Theprocedure used to establish the iPS cells from NS cells was basedon the recently reported protocols (25, 26) (Fig. 3A). Reprog-ramming factors-infected NS cells gradually formed colonies andgenerated ES cell-like colonies after a few weeks in culture (Fig.3B). Many of the round-shaped ES cell-like colonies (indicated byarrows in Fig. 3C) expressed theNanog protein, a pluripotent stemcell marker, suggesting the successful establishment of reprog-rammed cells from NS cells. In comparison, the flat and cobblestone-like cells (indicated by arrowheads in Fig. 3C) showed veryweak expression of Nanog, suggesting that these cells were notfully reprogrammed to stem cells. Microscopic bioluminescencemonitoring of these heterogeneous reprogrammed coloniesrevealed that themajority of the obtained ES cell-like colonies didnot exhibit circadian bioluminescence fluctuation (Fig. S5 andMovie S5). PMT-based bioluminescence assay of Nanog-positivecloned cells confirmed that the reprogrammed cell culture lost thecircadian bioluminescence oscillation (Fig. 3D). These resultssuggest that the reprogrammed cells lose the capacity to maintaincircadian clock oscillation. Interestingly, some of the reprog-rammed cell colonies still showed bioluminescence cycle (e.g.,indicated by a red arrow in Fig. S5 A and B). This oscillation-likebioluminescence may result from insufficient reprogramming,because these colonies exhibited cobble stone-like morphologyand were different from the ES cell-like round colonies.Then, reprogrammed cells were again differentiated in RA-

treated culture. Obvious circadian oscillation again appearedafter a 12-day RA-induced differentiation culture (Fig. 3E).These results indicate that the formation process of the mam-malian circadian clock correlates with cellular differentiationand that this process is reversible.

Expression of Clock Genes in ES Cells, iPS Cells, and DifferentiatedCells.Next, to determine the reasonswhyES cells (and iPS cells) do

not express the circadian molecular oscillation, the endogenousexpression levels of indispensable core clock genes were deter-mined in ES cells, reprogrammed cells, and differentiated cells.We first examined the expression levels of clock genes in ES celllines including KY1.1, E14Tg2a, and EB5 cells (Fig. S6). Inter-estingly, all lines of ES cells exhibited lower mPer2 and mBmal1transcription levels (P < 0.001) and higher expression levels ofmCry1 (P < 0.001) compared to nonsynchronized NIH 3T3 cells,and similar expression levels ofmPer1 andmCry2were noted inEScell lines and NIH 3T3 cells. These results suggest that certain ES-cell-specific mechanisms regulate the expression of clock genes.We also analyzed the expression of clock genes in reprogrammedcells. The reprogrammed cells also showed expression patterns ofclock genes similar to those of ES cells, with lower mPer2 andmBmal1 (P < 0.001) and higher mCry1 (P < 0.001) expressionlevels in ES and iPS cells than those in NIH 3T3 cells (Fig. S7A).On the other hand, similar expression levels of mPer1 and mCry2were observed among the ES, iPS, andNIH 3T3 cells. Because iPScells were established from NS cells, we checked the expressionlevels of clock genes in NS cells (Fig. S7B). Expression patternssuch as significantly higher expression ofmPer2 andmBmal1 (P <0.001) in NS cells than in ES cells suggest that the reprogrammingprocess may change the expression of clock genes in NS cells froman NIH 3T3-like to an ES-like pattern.To clarify the relation between the expression levels of core

clock genes and the cellular differentiation process, we examined

Day -1 0 1 2 3 7 14

infectionSeed on MMC-treated MEFs

NSMmedium ESM ESM/KSR

plating

Day 2 Day 284 factor Infection

A

B C

Biol

umin

esce

nce

E

Biol

umin

esce

nce

D

Colony pick-up, passage

BF Nanog

Nanog

Fig. 3. Reprogramming of differentiated NS cells results in the dis-appearance of the Bmal1-promoter-driven circadian oscillation. (A) Procedureused for generating inducedpluripotent stem (iPS) cells fromBmal1:luc ES cell-driven NS cells. NSM, ESM, and ESM/KSR indicate culture media (SI Materialsand Methods). (B) Morphological changes in NS cells to ES cell-like coloniesafter infectionwithOct3/4,Klf4, Sox2, and c-Myc expression retrovirus vectors.(C) Expression of Nanog protein in the reprogrammed cell colonies. A strongNanog-positive signal is observed in the ES cell-like smooth colony (arrow). Onthe other hand, only a faint Nanog signal is observed in the flat and cobblestone-like cell colony (arrowhead). (D) Real-time bioluminescence analysis ofcloned Nanog-positive reprogrammed cells (Inset) showed no circadian oscil-lation of Bmal1:luc driven bioluminescence. (E) Retinoic acid-induced differ-entiation culture of Nanog-positive reprogrammed cells shown in D results inthe reappearance of distinct circadian bioluminescence oscillation (red lines).Culture of the reprogrammed cells without RA and LIF does not exhibitapparent circadian oscillation (blue lines).

Yagita et al. PNAS | February 23, 2010 | vol. 107 | no. 8 | 3849

NEU

ROSC

IENCE

Dow

nloa

ded

by g

uest

on

July

2, 2

021

http://www.pnas.org/content/vol0/issue2010/images/data/0913256107/DCSupplemental/sm02.avihttp://www.pnas.org/cgi/data/0913256107/DCSupplemental/Supplemental_PDF#nameddest=sfig03http://www.pnas.org/content/vol0/issue2010/images/data/0913256107/DCSupplemental/sm03.avihttp://www.pnas.org/content/vol0/issue2010/images/data/0913256107/DCSupplemental/sm03.avihttp://www.pnas.org/cgi/data/0913256107/DCSupplemental/Supplemental_PDF#nameddest=sfig03http://www.pnas.org/content/vol0/issue2010/images/data/0913256107/DCSupplemental/sm04.avihttp://www.pnas.org/cgi/data/0913256107/DCSupplemental/Supplemental_PDF#nameddest=sfig03http://www.pnas.org/cgi/data/0913256107/DCSupplemental/Supplemental_PDF#nameddest=sfig03http://www.pnas.org/cgi/data/0913256107/DCSupplemental/Supplemental_PDF#nameddest=sfig03http://www.pnas.org/cgi/data/0913256107/DCSupplemental/Supplemental_PDF#nameddest=sfig04http://www.pnas.org/cgi/data/0913256107/DCSupplemental/Supplemental_PDF#nameddest=sfig04http://www.pnas.org/cgi/data/0913256107/DCSupplemental/Supplemental_PDF#nameddest=sfig04http://www.pnas.org/cgi/data/0913256107/DCSupplemental/Supplemental_PDF#nameddest=sfig05http://www.pnas.org/content/vol0/issue2010/images/data/0913256107/DCSupplemental/sm05.avihttp://www.pnas.org/cgi/data/0913256107/DCSupplemental/Supplemental_PDF#nameddest=sfig05http://www.pnas.org/cgi/data/0913256107/DCSupplemental/Supplemental_PDF#nameddest=sfig06http://www.pnas.org/cgi/data/0913256107/DCSupplemental/Supplemental_PDF#nameddest=sfig07http://www.pnas.org/cgi/data/0913256107/DCSupplemental/Supplemental_PDF#nameddest=sfig07http://www.pnas.org/cgi/data/0913256107/DCSupplemental/Supplemental_PDF#nameddest=STXThttp://www.pnas.org/cgi/data/0913256107/DCSupplemental/Supplemental_PDF#nameddest=STXT

the expression of clock genes in ES cells at 5 days (RADay 5) and14 days (RA Day 14) after RA-induced differentiation culture(Fig. 4). The expression levels ofmPer2,mCry2, andmClock geneswere significantly increased byRAday 5 from those inES cells (P<0.001), whereas mCry1 expression was significantly decreased inRA Day 5 cells from its expression in ES cells (P < 0.001). Inter-estingly, the expression levels of all examined clock genes weresimilar (P > 0.05) in RA Day 5 cells and RA Day 14 cells. Theseresults indicate that differentiation has already occurred in RADay 5 cells at least with regard to clock gene expression. Toevaluate the extent of cell differentiation, we performed Leish-man’s staining of ES cells, RA Day 5 cells, and RA Day 14 cells.Leishman’s solution stains ES cells strongly, whereas differ-entiated cells are stained weakly. EB5 ES cells stained strongly indark blue, whereasRADay 5 cells stainedmoreweakly and similarto the staining ofRADay 14 cells (Fig. S8A andB). The expressionlevels of ES cell-specific genes Oct3/4 and Nanog were alsoexamined in these ES and differentiating cells (27). Compatiblewith the results of Leishman’s staining, the expression levels ofbothOct3/4 andNanogweremarkedly decreased inRADay 5 cellscompared with ES cells (Fig. S8C). These results indicate that RAtreatment results in rapid differentiation of ES cells and that

undifferentiated ES cells almost no longer remain in the RA Day5 condition.

Dysfunctional Canonical Circadian Feedback Loops in the ES Cells. Arecent study reported that fertilized oocytes and very earlyembryos seem to escape from the regulation of circadian feedbackloops (16). Because ES cells are generally established from theinner cell mass of blastocysts, it is possible that the circadianfeedback loops do not function in ES cells. To examine this pos-sibility, we established the Cre-recombinase-dependent condi-tional dominant-negative BMAL1 (BMAL1-DN) expression EScell line (Fig. 5 A and B). The BMAL1-DN, which lacks the C-terminal region of BMAL1 protein, strongly suppresses theBMAL1/CLOCK-mediated transcription (28). To confirm theeffect of BMAL1-DN on the endogenous clock genes in somaticcells, we analyzed the expression levels of endogenous clock genesbefore and after the expression of BMAL1-DN in NIH 3T3 cells.

Rel

ativ

e ge

ne e

xpre

ssio

n le

vel

Rel

ativ

e ge

ne e

xpre

ssio

n le

vel

Rel

ativ

e ge

ne e

xpre

ssio

n le

vel

Rel

ativ

e ge

ne e

xpre

ssio

n le

vel

Rel

ativ

e ge

ne e

xpre

ssio

n le

vel

Rel

ativ

e ge

ne e

xpre

ssio

n le

vel

Fig. 4. Temporal expression profiles of endogenous core clock genes in RA-induced differentiation culture using EB5 ES cells. Dbp:luc stably transfectedEB5 ES cells were used for Taqman quantitative PCR analysis. All ES cells andDay 5 and Day 14 samples were cultured under the same conditions for bio-luminescence monitoring analysis. Day 5 and Day 14 samples were differ-entiated by 1 μMof RA for 5 days and 14 days, respectively. Compared with EScells, the expression levels of mPer2 and mClock genes were significantlyhigher (P < 0.001) andmCry1 expression was significantly lower (P < 0.001) onDay5. In contrast, the expression levels of all analyzed clock geneswere similaron Day 5 and Day 14. Data are mean ± SEM of three independently cultureddishes under each condition. *P < 0.05, **P < 0.01; n.s., not significant.

YFP-BMAL1-DN -

YFP-BMAL1-DN +

A B YFP BF

Cre +

Cre -

C

D

Fig. 5. Effects of dominant negative BMAL1 expression on NIH 3T3 and EScells. (A) Design of inducible dominant negative BMAL1 (BMAL1-DN)expression vector. A neomycin-resistant gene with poly(A) signals flanked byloxPwas placed under the control of the constitutively active CAG promoter.(B) Stable transfectants were selected by G418. The obtained ES cell coloniesdid not express YFP (Upper). To express the YFP fused BMAL1-DN protein,transient transfection was performed using cre recombinase expressionvector that also expressed a puromycin-resistant gene. YFP-positive ES cellcolonies were obtained after 2 days selection by puromycin (Lower). (C)Overexpression of YFP-BMAL1-DN expression in NIH 3T3 cells resulted inreduced expression levels of E-box regulated clock genes such as mPer1,mPer2, mCry1, mCry2, and mDbp. Data are mean ± SEM of three inde-pendent culture dishes. *P < 0.05, **P < 0.01; n.s., not significant. (D)Overexpression of YFP-BMAL1-DN did not suppress the expression levels ofany of the examined E-box regulated clock genes in ES cells. To detect theendogenous Bmal1, the 5′-UTR region of mBmal1 mRNA was targeted forqPCR assay. In EYFP, the average expression level of Cre+ cells was adjustedto the average level of Bmal1 expression in Cre+ cells. Each histogramindicates independent culture dishes.

3850 | www.pnas.org/cgi/doi/10.1073/pnas.0913256107 Yagita et al.

Dow

nloa

ded

by g

uest

on

July

2, 2

021

http://www.pnas.org/cgi/data/0913256107/DCSupplemental/Supplemental_PDF#nameddest=sfig08http://www.pnas.org/cgi/data/0913256107/DCSupplemental/Supplemental_PDF#nameddest=sfig08www.pnas.org/cgi/doi/10.1073/pnas.0913256107

The expression levels of all E-box-driven mPer1, mPer2, mCry1,mCry2, and mDbp genes were suppressed after the expression ofBMAL1-DN (Fig. 5C), indicating that the BMAL1-DN func-tionally inhibited the endogenous BMAL1/CLOCK activity. Onthe other hand, BMAL1-DN expression did not suppress anyexamined E-box-driven clock genes in ES cells (Fig. 5D). Espe-cially, it has been reported thatDbp gene expression level directlyreflects BMAL1/CLOCK-mediated transcriptional activity via E-box enhancer elements (29). Therefore, these results suggestdysfunction of the BMAL1/CLOCK-regulated canonical circa-dian feedback loops in ES cells. Although the exact mechanismthat controls the expression of clock genes in ES cells needs to bedetermined in future studies, dysfunction of the feedback loopsmay be one of the reasons why the circadian molecular oscillationis not observed in ES cells.

DiscussionThe main findings of the present study were (i) the mammaliancircadian molecular oscillator develops during cellular differ-entiation and (ii) this process can be reversed by expression ofthe reprogramming genes. These findings suggest some func-tional cross-talk between cellular differentiation and circadianclock formation in mammals. It was reported previously that thecircadian oscillator in zebrafish embryo appeared within 1 dayafter fertilization (30). This rapid generation of the circadianclock may correlate with the rapid development of zebrafish. Therelationship between the speed of development and circadianclock generation should be clarified in future studies for a betterunderstanding of the development process of the circadian clock.Many questions on the mechanisms of “chronogenesis” remain

unanswered at this stage. Our circadian oscillator generation assaydescribed in this study is a potentially useful model system forinvestigating the mechanisms of mammalian chronogenesis. More-over, the combination of a genetic modification strategy and dif-

ferentiation culture using ES cells may provide a powerful tool foridentification of key factors indispensable in the generation of thecircadian clock oscillator. Especially, one of the possible applica-tions may be ES-cell-based screening. As shown in Fig. 2 E and F,phases and period lengths of differentiated cells could be welldetermined, suggesting that the assay system presented in this studyhas a potential for the screening study (31). These lines of researchmay contribute to the design of new therapeutic or preventivestrategies for human developmental disorders such as autism thatare often complicated with circadian rhythm disorders (32, 33).In addition, it has been shown that reprogramming of somatic

cells can result in cells with properties of cancer stem cells (34).The assay system proposed in this study using cellular reprog-ramming may be also available for the investigation of cross-talkbetween the circadian clock and cancer, because the cellularcircadian clock has been implicated in carcinogenesis and tumorprogression (35, 36).

Materials and MethodsDifferentiation Culture. For differentiation, ≈1 × 105 ES cells/well were platedon gelatinized 35-mm dishes without feeder cells, using LIF-free ES mediumcontaining 1 μM all-trans retinoic acid (RA, Sigma). Differentiating cells werepassaged on days 2 and 6. Real-time bioluminescence analysis was per-formed using differentiating cells at the indicated days.

More materials and methods information is available in SI Text.

ACKNOWLEDGMENTS. We thank Dr. Y. Hatta-Ohashi and Dr. H. Suzuki(Olympus) for the technical support and analysis of microscopic cellular liveimages, Dr. J. Miyazaki (Osaka University) for technical advice about Leish-man’s staining, Dr. K. Yusa (Osaka University) and Dr. H. Niwa (RIKEN, Kobe)for providing the ES cells, and Dr. R. Matoba (Osaka University) for support-ing research. We also thank Dr. T. Kondo (Nagoya University) for equipmentsupport, technical advice, and discussion. This study was supported by aGrant-in-Aid (to K.Y.) from the Ministry of Education, Culture, Sports, Sci-ence, and Technology of Japan.

1. GachonF, Nagoshi E, BrownSA, Ripperger J, SchiblerU (2004) Themammalian circadiantiming system: From gene expression to physiology. Chromosoma 113:103–112.

2. Lowrey PL, Takahashi JS (2004) Mammalian circadian biology: Elucidating genome-wide levels of temporal organization. Annu. Rev. Genomics Hum. Genet. 5:407–441.

3. Waterhouse J, DeCoursey P (2004) Chronobiology: Biological Timekeeping, edsDunlap J, Loros J, DeCoursey P (Sinauer, Sunderland, MA).

4. Sumová A, et al. (2006) Setting the biological time in central and peripheral clocksduring ontogenesis. FEBS Lett 580:2836–2842.

5. Schibler U, Naef F (2005) Cellular oscillators: Rhythmic gene expression andmetabolism. Curr Opin Cell Biol 17:223–229.

6. Reppert SM, Weaver DR (2002) Coordination of circadian timing in mammals. Nature418:935–941.

7. Ueda HR, et al. (2002) A transcription factor response element for gene expressionduring circadian night. Nature 418:534–539.

8. Yamazaki S, et al. (2000) Resetting central and peripheral circadian oscillators intransgenic rats. Science 288:682–685.

9. Yoo S, et al. (2004) PERIOD2::LUCIFERASE real-time reporting of circadian dynamicsreveals persistent circadian oscillations in mouse peripheral tissues. Proc Natl Acad SciUSA 101:5339–5346.

10. Brown SA, et al. (2005) The period length of fibroblast circadian gene expressionvaries widely among human individuals. PLoS Biol 3:e338.

11. Nagoshi E, et al. (2004) Circadian gene expression in individual fibroblasts: Cell-autonomous and self-sustained oscillators pass time to daughter cells. Cell 119:693–705.

12. Kiyohara YB, et al. (2008) Detection of a circadian enhancer in the mDbp promoterusing prokaryotic transposon vector-mediated strategy. Nucleic Acids Res 36:e23.

13. Kawakami K (2005) Transposon tools and methods in zebrafish. Dev Dyn 234:244–254.

14. Urasaki A, Morvan G, Kawakami K (2006) Functional dissection of the Tol2transposable element identified the minimal cis-sequence and a highly repetitivesequence in the subterminal region essential for transposition. Genetics 174:639–649.

15. Koga A, Suzuki M, Inagaki H, Bessho Y, Hori H (1996) Transposable element in fish.Nature 383:30.

16. Amano T, et al. (2009) Expression and functional analyses of circadian genes in mouseoocytes and preimplantation embryos: Cry1 is involved in the meiotic processindependently of circadian clock regulation. Biol Reprod 80:473–483.

17. Balsalobre A, Damiola F, Schibler U (1998) A serum shock induces circadian geneexpression in mammalian tissue culture cells. Cell 93:929–937.

18. Yagita K, Tamanini F, van Der Horst GTJ, Okamura H (2001) Molecular mechanisms ofthe biological clock in cultured fibroblasts. Science 292:278–281.

19. Forrester LM, et al. (1996) An induction gene trap screen in embryonic stem cells:Identification of genes that respond to retinoic acid in vitro. Proc Natl Acad Sci USA93:1677–1682.

20. Balsalobre A, Marcacci L, Schibler U (2000) Multiple signaling pathways elicit circadiangene expression in cultured Rat-1 fibroblasts. Curr Biol 10:1291–1294.

21. Yagita K, Okamura H (2000) Forskolin induces circadian gene expression of rPer1,rPer2 and dbp in mammalian rat-1 fibroblasts. FEBS Lett 465:79–82.

22. Liu AC, et al. (2007) Intercellular coupling confers robustness against mutations in theSCN circadian clock network. Cell 129:605–616.

23. Conti L, et al. (2005) Niche-independent symmetrical self-renewal of a mammaliantissue stem cell. PLoS Biol 3:e283.

24. Takahashi K, Yamanaka S (2006) Induction of pluripotent stem cells from mouseembryonic and adult fibroblast cultures by defined factors. Cell 126:663–676.

25. Kim J-B, et al. (2008) Pluripotent stem cells induced from adult neural stem cells byreprogramming with two factors. Nature 454:646–650.

26. Shi Y, et al. (2008) A combined chemical and genetic approach for the generation ofinduced pluripotent stem cells. Cell Stem Cell 2:525–528.

27. Silva J, et al. (2009) Nanog is the gateway to the pluripotent ground state. Cell 138:722–737.

28. Kiyohara YB, et al. (2006) The BMAL1 C terminus regulates the circadian transcriptionfeedback loop. Proc Natl Acad Sci USA 103:10074–10079.

29. Zhang EE, et al. (2009) A genome-wide RNAi screen for modifiers of the circadianclock in human cells. Cell 139:199–210.

30. Dekens MP, Whitmore D (2008) Autonomous onset of the circadian clock in thezebrafish embryo. EMBO J 27:2757–2765.

31. Takahashi JS, Shimomura K, Kumar V (2008) Searching for genes underlying behavior:Lessons from circadian rhythms. Science 322:909–912.

32. Richdale AL, Prior MR (1992) Urinary cortisol circadian rhythm in a group of high-functioning children with autism. J Autism Dev Disord 22:433–447.

33. Corbett BA, Schupp CW, Levine S, Mendoza S (2009) Comparing cortisol, stress, andsensory sensitivity in children with autism. Autism Res 2:39–49.

34. Liu Y, et al. (2009) Mouse fibroblasts lacking RB1 function form spheres and undergoreprogramming to a cancer stem cell phenotype. Cell Stem Cell 4:336–347.

35. Fu L, Lee CC (2003) The circadian clock: Pacemaker and tumour suppressor. Nat RevCancer 3:350–361.

36. Filipski E, et al. (2004) Effects of chronic jet lag on tumor progression in mice.Cancer Res 64:7879–7885.

Yagita et al. PNAS | February 23, 2010 | vol. 107 | no. 8 | 3851

NEU

ROSC

IENCE

Dow

nloa

ded

by g

uest

on

July

2, 2

021

http://www.pnas.org/cgi/data/0913256107/DCSupplemental/Supplemental_PDF#nameddest=STXT