systems biology of mammalian circadian clocks · systems biology of mammalian circadian clocks...

ANRV404-PH72-28 ARI 18 January 2010 10:16

Systems Biology ofMammalian Circadian ClocksHideki Ukai1 and Hiroki R. Ueda1,21Laboratory for Systems Biology and 2Functional Genomics Unit, RIKEN Center forDevelopmental Biology, Hyogo 650-0047, Japan; email: [email protected]

Annu. Rev. Physiol. 2010. 72:579–603

First published online as a Review in Advance onNovember 13, 2009

The Annual Review of Physiology is online atphysiol.annualreviews.org

This article’s doi:10.1146/annurev-physiol-073109-130051

Copyright c© 2010 by Annual Reviews.All rights reserved

0066-4278/10/0315-0579$20.00

Key Wordstranscriptional circuit, temperature compensation, singularity,artificial circuit

AbstractSystems biology is a natural extension of molecular biology; it can bedefined as biology after identification of key gene(s). Systems-biologicalresearch is a multistage process beginning with (a) the comprehensiveidentification and (b) quantitative analysis of individual system com-ponents and their networked interactions, which lead to the ability to(c) control existing systems toward the desired state and (d ) design newones based on an understanding of the underlying structure and dynam-ical principles. In this review, we use the mammalian circadian clock asa model system and describe the application of systems-biological ap-proaches to fundamental problems in this model. This application hasallowed the identification of transcriptional/posttranscriptional circuits,the discovery of a temperature-insensitive period-determining process,and the discovery of desynchronization of individual clock cells under-lying the singularity behavior of mammalian clocks.

579

Ann

u. R

ev. P

hysio

l. 20

10.7

2:57

9-60

3. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by U

nive

rsity

of T

okyo

on

02/1

4/10

. For

per

sona

l use

onl

y.


INTRODUCTION: SYSTEMSBIOLOGY AS BIOLOGY AFTERIDENTIFICATIONRecent large-scale efforts in genome se-quencing, expression profiling, and functionalscreening have produced an embarrassmentof riches for life science researchers, and bio-logical data can now be accessed in quantitiesthat are orders of magnitude greater than wereavailable even a few years ago. The growingneed for interpretation of data sets, as well asthe accelerating demand for their integration toa higher-level understanding of life, has set thestage for the advent of systems biology (1, 2),in which biological processes and phenomena

Mammalian circadian clock

Genetic networkof clock



Operatedcircadian rhythm

Networkcomponents

BMAL2DEC2

DEC1

E/E'-box D-box RRE

RevErbAb

RORa

NPAS2

CLOCK

BMAL1

CRY1

CRY2 RORb

RORc

DBP

TEF

HLF

RevErbAa

PER2

PER1

PER3

E4BP4

Step A6 h Step B4 h

Step C4 h

Step D6 h

Step E4 h

Oscillator

Naturalcircadian rhythm

−0.75−0.5

−0.25

0.250.5

0.751

Time (h)

Naturalcircadian rhythm

System controlOperate intentionally

System designSynthesize from scratch

System identi!cationDivide into components

System analysisAccurately measure

25 50 75 100 125 150 175PP1

FBXL3

PP2A

GSK3b

CKIe

CKId

E/E'box

RevErbAb

RORa

NPAS2

CLOCK

BMAL2

BMAL1

DEC2

DEC1

CRY1

CRY2

RORb

RORc

D-box

DBPTEF

HLF

RevErbAa

PER2

PER1

PER3

E4BP4

RRE

PP1

FBXL3

PP2A

GSK3b

CKIe

CKId

E/E'box

RevErbAb

RORa

NPAS2

CLOCK

BMAL2

BMAL1

DEC2

DEC1

CRY1

CRY2

RORb

RORc

D-box

DBPTEF

HLF

RevErbAa

PER2

PER1

PER3 RRE

E4BP4

PP1

FBXL3

PP2A

GSK3b

CKIe

CKId

E/E'box

RevErbAb

RORa

NPAS2

CLOCK

BMAL2

BMAL1

DEC2

DEC1

CRY1

CRY2

RORb

RORc

D-box

DBPTEF

HLF

RevErbAa

PER2

PER1

PER3 RRE

E4BP4

IdentifyPredictionVeri"cation

ControlRepair

DesignReconstructionStimuli

0

18 6

12

12

3

4

5

6

7

8

9

10111213

14

15

16

17

18

19

20

21

2223 24(0)

a b

c d

Systemsbiology

Figure 1Systems biology. (a) Systems-biological research begins with comprehensive identification. In this step, individual system componentsand their networked interactions are comprehensively identified. (b) In the second step, to derive the design principle of a target system,the behavior of the system is predicted and validated through an accurate measurement with perturbations. (c) An understanding of thedesign principle of the system is needed to derive the method of controlling the system toward the desired state. (d ) Finally, the level ofunderstanding is confirmed by reconstruction of the system.

are approached as complex and dynamicsystems. Systems biology is a natural extensionof molecular biology and can be defined asbiology after identification of key gene(s).

We consider systems-biological research asa multistage process (Figure 1) that comprisesfour steps: (a) identification of the whole net-work structure through functional genomicsand comparative genomics (system identifica-tion), (b) prediction and validation of the sys-tem’s behavior to derive the design principlethrough the accurate measurement and (static)perturbation of network dynamics (system anal-ysis), (c) control and repair of the network statetoward the desired state through the precise and

580 Ukai · Ueda

Ann

u. R

ev. P

hysio

l. 20

10.7

2:57

9-60

3. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by U

nive

rsity

of T

okyo

on

02/1

4/10

. For

per

sona

l use

onl

y.


dynamic perturbation of its components (sys-tem control), and (d ) reconstruction and designof systems based on the design principles de-rived from the identified structure and observeddynamics (system design). These processes arerequired to elucidate the design principles ofcomplex and dynamic biological systems, suchas the mammalian circadian clock.

MAMMALIAN CIRCADIANCLOCK AS A MODEL SYSTEMThe mammalian circadian clock is an idealmodel system for the study of complex and dy-namic biological systems. The mammalian cir-cadian clock consists of complexly integratedfeedback and feed-forward loops (3) and ex-hibits well-defined dynamical properties (4),including (a) endogenous oscillations with anapproximately 24-h period, (b) entrainment toexternal environmental changes (temperatureand light cycle), (c) temperature compensationover a wide range of temperatures, and (d ) syn-chronization of multiple cellular clocks despitethe inevitable molecular noise. All of these dy-namical properties would be difficult to elu-cidate without utilizing such system-level ap-proaches. In addition to its advantages as abasic model system for systems-biological re-search, the circadian clock is intimately in-volved in the control of metabolic and physi-ological processes (3, 5), and its dysregulationis associated with the onset and development ofnumerous human diseases, including sleep dis-orders, depression, and dementia. An improvedsystem-level understanding promises to pro-vide biomedical and clinical investigators with apowerful new arsenal with which to attack theseconditions.

Some systems-biological efforts to elucidatethe design principles of complex and dynamicbiological systems such as the mammalian circa-dian clock have been reported. In the followingsections, we describe in detail the strategies andtechnologies developed for systems-biologicalresearch, in addition to their applicability tothe specific case of the mammalian circadianclock.

Temperaturecompensation:phenomenon thatcauses the period of anoscillation to be stabledespite temperaturechange

SCN: suprachiasmaticnucleus

IDENTIFICATION OF CLOCKSFollowing the completion of genome projectsfor species such as mouse and human,genome-wide resources such as small interfer-ing RNA (siRNA) and complementary DNA(cDNA) libraries have undergone considerableexpansion. Development of high-throughputtechnologies also allows efficient use of theseresources. These genome-wide resources andtechnologies, as well as genome-associated in-formation, allow us to comprehensively identifycomplex systems such as the mammalian circa-dian clock (system identification).

Identification of theTranscriptional CircuitThe mammalian circadian master clock is pri-marily located in the suprachiasmatic nucleus(SCN) (3). Transcript analyses have indicatedthat circadian clocks are not restricted to SCNbut are found in various tissues (6), includingliver, and in cultured fibroblast cells such as Rat-1 (7, 8), NIH3T3 (9), and U2OS (10) cells. In-creasing numbers of so-called clock genes wereidentified in mammalian clocks; these genes in-clude three basic helix-loop-helix (bHLH)-PAS(PER-ARNT-SIM) transcription factors (Clock,Npas2, and Arntl; the latter is also known asBmal1 or Mop3) (11–14), two period genes (Per1and Per2) (15, 16), two cryptochrome genes(Cry1 and Cry2) (17, 18), casein kinase I epsilonand delta (Csnk1e and Csnk1d ) (19, 20), andtwo orphan nuclear hormone receptors (Nr1d1and Rora, also known as RevErbα and Rorα,respectively) (21, 22). A number of other tran-scriptional factors also thought to function inthe circadian regulation of gene expression weregradually clarified; these include four bZip-family genes (Dbp, Tef, Hfl, and Nfl3; the latter isalso termed E4bp4 ) (23, 24), three bHLH tran-scription factors (Arntl2, Bhlhb2, and Bhlhb3,also known as Bmal2, Dec1 or Stra13, and Dec2,respectively) (25), one period-related gene(Per3) (26), and three genes related to Nr1d1and Rora (Nr1d2, Rorb, and Rorc, also knownas RevErbβ, Rorβ, and Rorγ , respectively)(22, 27).

www.annualreviews.org • Systems Biology of Mammalian Clocks 581

Ann

u. R

ev. P

hysio

l. 20

10.7

2:57

9-60

3. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by U

nive

rsity

of T

okyo

on

02/1

4/10

. For

per

sona

l use

onl

y.


CCEs: clock-controlled elements

dLuc: destabilizedluciferase

In cellulo cyclingassay: means ofperforming real-timemonitoring oftranscriptionaldynamics of clock-controlled promotersin a cell via dLuc

Molecular interactions among these clockgenes and clock-related genes have also beenat least partly identified. For example, thetranscription factors CLOCK and BMAL1dimerize and directly and indirectly activatetranscription of the Per and Cry genes throughE-box elements (5′-CACGTG-3′) (28, 29).The PER and CRY proteins accumulate inthe cytosol and are then translocated followingphosphorylation into the nucleus, where theyinhibit the activity of CLOCK and BMAL1(3). The turnover of the inhibitory PER andCRY proteins leads to a new cycle of activationby CLOCK and BMAL1 via E-box elements.The transcriptional regulation network ofthese genes forms a circadian clock oscillator,which is known to control output genes andto affect physiological and metabolic processes(3, 5). This type of transcriptional feedbackmechanism underlying circadian rhythmsseems to be conserved across species (4).Despite reports of many transcriptional regu-lations of each gene, however, an overview ofthe circadian clock core network remains to beput forward.

Complicated networks cannot be elucidatedwithout both a comprehensive identificationof network circuits and an accurate measure-ment of system dynamics. In the first steptoward complete identification of the circadianclock core network, several groups utilizedDNA microarray to comprehensively andquantitatively measure genome-wide geneexpression of mammalian circadian clocks (27,30–54). For example, Ueda et al. (27) identifiedso-called clock-controlled genes and demon-strated circadian oscillation with characteristicexpression patterns in mouse SCN and liverthrough use of a high-density oligonucleotideprobe array (55) and biostatistics. In thesecond step, the authors comprehensivelydetermined the transcription start sites via theoligo-capping method (56) to construct thegenome-wide promoter/enhancer database.Using these data, the authors predicted arelationship between expression patterns ofidentified genes and DNA-regulatory ele-ments on their promoter/enhancer regions.

Three types of clock-controlled elements(CCEs)—E-boxes (5′-CACGTG-3′) (57)plus E′-boxes (5′-CACGTT-3′) (58, 59),D-boxes [5′-TTATG(C/T)AA-3′] (60), andRevErbA/ROR-binding elements (RREs) [5′-(A/T)A(A/T)NT(A/G)GGTCA-3′] (61)—aredistributed throughout the oscillatory genes.

To determine the role of these elements inthe circadian clock, Ueda et al. (27, 58) utilized acell-culture system in which circadian rhythmsin transcriptional dynamics are monitored via adestabilized luciferase (dLuc) reporter driven byclock-controlled promoters. In this cell-culturesystem—termed the in cellulo cycling assay—cultured Rat-1 cells were transiently transfectedby reporter constructs and stimulated with dex-amethasone, and their bioluminescences weremeasured. Dexamethasone was administratedto induce macroscopic circadian oscillationsin the cultured cells. Through the genome-wide searching process described above, theauthors found CCEs on 16 clock/clock-controlled gene promoters/enhancers. Via thecellulo cycling assay system, Ueda et al. re-vealed that functionally and evolutionarily con-served E/E′-boxes are located on the noncod-ing regions of nine genes (Per1, Per2, Cry1,Dbp, Rorγ , RevErbα/Nr1d1, RevErbβ/Nr1d2,Dec1/Bhlhb2, and Dec2/Bhlhb3); that D-boxesare located on those of seven genes (Per1, Per2,Per3, RevErbα, RevErbβ, Rorα, and Rorβ); andthat RREs are located on those of six genes(Bmal1/Arntl, Clock, Npas2, Cry1, E4bp4/Nfil3,and Rorγ ). On the basis of this functional andconserved transcriptional regulatory mecha-nism, investigators successfully drew transcrip-tional circuits underlying mammalian circadianrhythms (Figure 2a) (58). We note that furthercharacterization of these three CCEs was re-cently performed (62–65).

Ueda et al. (58) further suggested that reg-ulation of E/E′-boxes is the topological vulner-ability point in the mammalian circadian clock,and they functionally verified this concept us-ing in cellulo cycling assay systems. Overex-pression of repressors of E/E′-box regulation[CRY1 (29)], RRE regulation [REVERBα (22,27)], or D-box regulation [E4BP4 (24)] affected

582 Ukai · Ueda

Ann

u. R

ev. P

hysio

l. 20

10.7

2:57

9-60

3. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by U

nive

rsity

of T

okyo

on

02/1

4/10

. For

per

sona

l use

onl

y.


PP1

FBXL3

PP2A

GSK3b

CKIε

CKIδ

E/E'-box

RevErbAβ

RORα

NPAS2

CLOCK

BMAL2

BMAL1

DEC2

DEC1

CRY1

CRY2

RORβ

RORγ

D-box

DBPTEF

HLF

RevErbAα

PER2

PER1

PER3

E4BP4a

b

RRE

Period length(h real)

Peri

od le

ngth

(h C

T)

Concentration (µM)0 200 400

Concentration (µM)0 200 400

48

40

32

24

50

42

34

26

TG003 SP60012548

40

32

24

50

42

34

26

Figure 2System identification of clocks. (a) Schematic representation of the transcriptional network of themammalian circadian clock. Genes and clock-controlled elements are depicted as ovals and rectangles,respectively. Transcriptional/translational expression, activation, and repression are depicted as gray, green,and magenta lines, respectively. (b) Dose-dependent effects of SP600125 and TG003 on period length inU2OS-hPer2-Luc cells. The period length is indicated both in real time (right axis) and in circadian time(CT; left axis). For CT, the average period length in two independent control experiments was assigned as24 h. The two lines in each graph correspond to two independent experiments (99). Each value representsthe mean ± SEM (standard error of the mean). At the concentrations without data points, the cells behavedarrhythmically. High-dose pharmacological inhibition of the casein kinase I epsilon (CKIε) activitylengthened the circadian (∼24-h) period to an almost circabidian (∼48-h) period.


Ann

u. R

ev. P

hysio

l. 20

10.7

2:57

9-60

3. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by U

nive

rsity

of T

okyo

on

02/1

4/10

. For

per

sona

l use

onl

y.


circadian rhythmicity in Per2 or Bmal1 pro-moter activity. The effects differed, however,among repressors, and the severest effect wasobserved when the E/E′-box was attacked. Suchdifferent modes of effect cannot be explainedby mere quantitative differences in the strengthof these three repressors, indicating that thereis some qualitative difference among E/E′-box, D-box, and RRE regulation in circadianrhythmicity (58).

We also note that a novel strategy termedgene-dosage network analysis was recentlydeveloped by Baggs et al. (66). In this strategy,siRNA-induced dose-dependent changes ingene expression were used to build gene-association networks consistent with knownbiochemical constraints (66). Baggs et al.observed that several genes are upregulatedfollowing knockdown of their paralogs, whichsuggests that the clock network utilizes activecompensatory mechanisms rather than simpleredundancy to confer robustness and maintainfunction.

The Role of Negative FeedbackThe identification of transcriptional circuitsof mammalian circadian clock has revealedthe topological importance of E-box, the so-called morning element. This is consistentwith the prevailing transcriptional feedbackmodel, which is believed to be mediated in themammalian circadian clock by the CRY1 and-2 (17, 18, 29) and PER1 and -2 (16, 67,68) proteins. More specifically, the CRY andPER proteins are hypothesized to autoregu-late their own expression by repressing theheterodimeric complex of the bHLH-PAS-domain transcriptional activators CLOCK andBMAL1, which bind to E-box elements in theCRY (69) and PER (28, 58, 59) promoters.Similar autoregulatory loops, in which negativefeedback regulation of transcriptional and/ortranslational processes are involved, seem tobe evolutionarily conserved among a wide va-riety of species (3, 70, 71). Thus, transcrip-tional/translational feedback repressions areproposed to generate a 24-h periodicity. The

question of universal necessity for transcrip-tional/translational feedback repression, how-ever, arose primarily from recent studies ofcyanobacterial circadian rhythms, in which therepression was shown to be not necessary (72,73). Direct evidence for the requirement ofCRY-mediated repression of CLOCK/BMAL1transcriptional activity in the maintenance ofcircadian clock function thus remains to bepresented.

To determine the requirement of feedbackrepression in circadian clock function, Satoet al. (78) changed the molecular parameter forfeedback repression by functional genomics,then tested the cellular phenotype caused bythis parameter change via the in cellulo cyclingassay system. The authors sought to identifythe CLOCK alleles that were insensitiveto CRY1 repression but maintained normaltranscriptional activity. They generated alibrary of ∼6000 random point mutants ofhuman alleles for both CLOCK and BMAL1,then screened clones individually in cell-basedreporter assays with wild-type Bmal1 cDNAand a Per1 promoter–luciferase (Per1-Luc)construct (28) in the presence of cotransfectedCry1. Compared with wild-type alleles, of theCLOCK and BMAL1 clones screened, severalreproducibly maintained threefold or greaterreporter activity in the presence of CRY1.Notably, these clones demonstrated similartranscriptional activities as wild types in theabsence of cotransfected Cry1, suggesting thatthese mutations do not cause overt alterationsin the heterodimerization, nuclear localization,DNA-binding, and transactivation propertiesof the mutant CLOCK/BMAL1 complex.

The prevailing transcriptional feedbackmodel predicts that impairment of CRY-mediated repression should have marked ef-fects on circadian expression of the Per genes.This notion is supported by in vivo observa-tions that expression of Per1 and Per2 is consti-tutively elevated in Cry1/Cry2 double-knockoutmice (18, 74). To determine whether these mu-tations in CLOCK and BMAL1 cause pheno-typic changes in circadian gene expression, in-vestigators performed in cellulo cycling assays

584 Ukai · Ueda

Ann

u. R

ev. P

hysio

l. 20

10.7

2:57

9-60

3. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by U

nive

rsity

of T

okyo

on

02/1

4/10

. For

per

sona

l use

onl

y.


(27, 58). Mouse NIH3T3 cells were trans-fected with plasmids harboring dLuc drivenby the Per2 or SV40 basic promoters, alongwith the BMAL1 and CLOCK mutant alleles,and were monitored via the in cellulo cyclingassay. Compared with empty vector transfec-tion, cotransfection of wild-type CLOCK andBMAL1 did not substantially alter rhythmicity,and the transfected cells’ period lengths were21.4 ± 0.4 h. In contrast, transfection of ei-ther CLOCK or BMAL1 mutant alleles, whencompared with wild-type CLOCK/BMAL1,resulted in substantial impairment of circa-dian rhythmicity after one or two cycles ofoscillations. Notably, cotransfection of CRY-insensitive mutant CLOCK and BMAL1 to-gether resulted in the loss of circadian Per2promoter activity. Sato et al. (78) thereforedemonstrated that the transcriptional repres-sion of CLOCK/BMAL1 by CRY was requiredfor circadian E-box activity.

In addition to that of Per and Cry, the rhyth-mic expression of Bmal1 mRNA is also sub-ject to circadian clock regulation (75). TheBmal1 promoter used in this study, however,does not have E-box sites and instead containsRRE (22, 27), whose activities are reciprocallycontrolled by the rhythmically expressed tran-scriptional repressor REVERBα (22) and theactivator RORα (21). In an additional test forcircadian clock function, the effects of mutantCLOCK and BMAL1 on rhythmic RRE ac-tivity were examined via in cellulo cycling as-says with a Bmal1-dLuc reporter. Similar to theresults obtained from the Per2-dLuc reporter,transfection of single CLOCK or BMAL1mutants resulted in the decreased amplitudeof cycling of Bmal1-dLuc activity comparedwith wild-type CLOCK/BMAL1 transfection.Moreover, this decrease in cycling amplitudewas further exacerbated upon cotransfectionof the double-mutant heterodimer. These re-sults indicate that transcriptional repression ofCLOCK/BMAL1 by CRY is also required forcircadian BMAL1 expression through RRE,which in turn depends on transcriptional, trans-lational, and posttranslational actions of en-dogenous cellular factors.

Arrhythmic Per2 expression, observed ina population of cells expressing the double-mutant CLOCK/BMAL1 complex, may bedue to the disruption of oscillator functionor a lack of synchrony between individualrhythmic cells. To address these possibilities,quantitative imaging of Per2-dLuc reporteractivity from individual NIH3T3 cells wasmeasured by an approach similar to that used inanalyzing Bmal1 reporter rhythms from singlecells (76). As in the whole-well assays, themedian reporter activity for the population ofimaged individual cells coexpressing wild-typeCLOCK/BMAL1 oscillated rhythmically.In contrast, the population of individualClock/Bmal1 mutant cells was visibly arrhyth-mic. Individual reporter activities from singlewild-type cells were rhythmic, as expected,whereas individual Clock/Bmal1 double-mutantcells showed arrhythmic reporter activities.These differences in activity patterns were eval-uated by two independent statistical methodsthat score the circadian rhythmicity of exper-imental time-course data. These data providedirect evidence that CRY-mediated feedbackrepression of the CLOCK/BMAL1 complex isrequired for mammalian clock function.

We also note that functional genomics tech-nology utilized in Sato et al.’s study has suc-cessfully been applied to other clock proteins(CRY1 and CRY2) to further characterize theirmolecular properties (77). Thus, the applica-tion of cellular genetics technology will prob-ably have as significant an impact on mam-malian biology as similar approaches have hadon prokaryotic and yeast biology.

The Identification of theRate-Limiting ProcessThe E-box-mediated transcriptional/posttranscriptional loop is the critical core loopof the mammalian circadian clock (58, 78).Despite this finding, the specific processes inthe core loop that determine the period lengthremain elusive. In this section, we describe theidentification of the rate-limiting process ofthe mammalian circadian clock.


Ann

u. R

ev. P

hysio

l. 20

10.7

2:57

9-60

3. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by U

nive

rsity

of T

okyo

on

02/1

4/10

. For

per

sona

l use

onl

y.


Circadian time (CT):standardized 24-hnotation of the phasein a circadian cycle.CT0 indicates thebeginning of asubjective day, andCT12 indicates thebeginning of asubjective night

As described above, comprehensive gene-expression profiling is a basic approachto system identification (27, 30–54), andsuch gene-expression profiles are integratedwith comparative genomics to characterizethe clock-controlled DNA elements in themammalian genomes (62, 64, 65). However,mammalian circadian clocks cannot be un-derstood only through the identification oftranscriptional circuits; posttranscriptionalregulations need to be identified (79–82). Thislack of information about posttranscriptionalregulations will be able to be complemented byhigh-throughput measurement technologiessuch as proteomics (46, 83), metabolomics (84,85), and other functional genomics resourcessuch as RNA interference (RNAi) (10, 66,86, 87) and chemical or peptide libraries (88,89). Using these technologies and resources,Nakahata et al. (88) identified 15d-PGJ2 andHatcher et al. (84) identified little SAAS as en-trainment factors for NIH3T3 cells and SCN,respectively. In a high-throughput RNAi-basedgenetic screening, Maier et al. (86) identifiedCK2 as a PER2-phosphorylating kinase andcomponent of the mammalian circadian clock.

A chemical-biological approach is also ef-fective at elucidating the basic processes thatunderlie circadian clocks (90). Although per-formed on a relatively small scale, early stud-ies revealed that several protein kinase in-hibitors [such as IC261 (91, 92), CKI7 (93),lithium chloride (94, 95), SP600125 (92,96), SB203580 (97), DRB, LY294002, BML-297, and SB202190 (98)], adenylate cyclaseinhibitors [such as THFA (9-tetrahydro-2-furyl-denine), 2′5′-dideoxyadonosine, and 9-cyclopentyladenine (92)], and proteasome in-hibitors [such as MG132 and lactacystin (91)]can lengthen the period of mammalian circa-dian clocks by 2% to ∼40%. High-throughputscreening of a large chemical compound librarywas also performed (89). The results showedthat the inhibition of glycogen synthase ki-nase 3 beta (GSK3β) shortens the period ofmammalian clock cells, superseding a previousproposal about the function of GSK3β thatwas based on the period-lengthening effects

of lithium (94, 95). This evidence supportedthe use of a chemical-biological approach inprobing the fundamental processes of the mam-malian circadian clock.

To systematically identify the fundamen-tal processes involved in determining the pe-riod length of mammalian clocks, Isojima et al.(99) tested 1260 pharmacologically active com-pounds for their effect on period length inmouse and human clock cell lines, NIH3T3and U2OS, and found 10 compounds thatmost markedly lengthened the period of bothclock cell lines and affected both the central(SCN) and peripheral (mouse embryonic fi-broblast) circadian clocks. Most compounds in-hibited CKIε or CKIδ activity, and the siRNAknockdown of Csnk1e or Csnk1d exhibited greatperiod-lengthening effects—more than 28 h incircadian time (CT). Moreover, the combinato-rial knockdown of Csnk1e and Csnk1d additivelylengthened the period of circadian oscillationsto over 30 h in CT, and high-dose pharmaco-logical inhibition of the CKIε/δ activity length-ened the circadian (∼24-h) period to an al-most circabidian (∼48-h) period (Figure 2b).Inhibition of CKIε/δ activity led to decelera-tion of the PER2 degradation rate, which isregulated by CKIε-dependent phosphorylation(100). Moreover, the period length of the circa-dian oscillation is strongly correlated with thePER2 stability under the administration of theCKI inhibitor. These data suggest that CKIε/δ-dependent phosphorylation on the PER2 pro-tein is an important period-determining pro-cess in the mammalian circadian clock.

As these studies have shown, both the per-turbation of components’ quantity via cDNA,siRNA, and chemical libraries and the perturba-tion of components’ quality (parameter change)via functional genomics are important for theidentification of complex and dynamic biologi-cal systems, such as that of the mammalian cir-cadian clock.

ANALYSIS OF CLOCKSTo derive the design principle of a system ofinterest, one must validate the behavior of a

586 Ukai · Ueda

Ann

u. R

ev. P

hysio

l. 20

10.7

2:57

9-60

3. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by U

nive

rsity

of T

okyo

on

02/1

4/10

. For

per

sona

l use

onl

y.


predicted system through an accurate mea-surement with several types of perturbations(system analysis). In this section, we describethe quantitative analysis of identified period-determining biochemical reactions in the abovesection, which is probably important for deter-mining the temperature compensation of themammalian circadian clock.

A 24-h period that protects against environ-mental changes, such as in temperature and nu-trition, is one of the most intriguing aspectsof the circadian clock. The robustness of thisprotection against temperature change isknown as temperature compensation. Typicalbiochemical reactions, such as enzymatic reac-tions, show temperature dependency, which isrepresented by a Q10 value of approximatelytwo. In contrast, the period of circadian rhythmis independent of or compensated against tem-perature change (i.e., the Q10 value of the pe-riod is approximately one). The importanceof temperature compensation in poikilothermscan be easily understood. This aspect of themammalian circadian clock had been contro-versial, but in 2003 it was confirmed in culturedmammalian cells (101, 102). Temperature com-pensation has proved to be one of the generalcharacteristics of circadian clocks, appearing inspecies ranging from cyanobacteria to humans.Despite our increasing knowledge of the molec-ular mechanism of circadian clocks, however, itis difficult to explain how circadian clocks sus-tain such a constant period against temperaturechange.

To explain this phenomenon, theoreticalstudies (103, 104) have proposed a balanced re-action model (Figure 3a) in which (a) increas-ing kinetic parameters of some reactions in anegative feedback loop shorten the period and(b) other parameters prolong the period. If theperiod-accelerating and period-deceleratingreactions were equally sensitive to tempera-ture, the frequency of the circadian oscillationwould increase with an increase in temperature.To reconcile this issue, the balance reactionmodel proposes that the period-acceleratingreaction(s) are less temperature sensitive thanthe period-decelerating reaction(s) and, hence,

Q10: the change in therate of a process as aresult of increasing thetemperature by 10◦C

that these two sets of enzymatic reactions arebalanced to maintain a constant circadian os-cillation. These theoretical explanations seemplausible, but the balance of effects on the pe-riod length between basic reactions can be eas-ily broken by perturbations, such as inhibitorsand point mutations on clock proteins. Further-more, these theoretical models are inconsistentwith the fact that many circadian clock mutantsshow diverse periods but sustain temperaturecompensation. How do circadian clocks acquirerobustness against temperature change?

The circadian rhythm of KaiC (a circadianclock protein) phosphorylation can be reconsti-tuted by only three Kai proteins, KaiA, KaiB,and KaiC, in a test tube, demonstrating that thisrhythm is the central oscillator of the cyanobac-terial circadian clock (72). In this test tube, theoscillation period of KaiC phosphorylation isunaffected by temperature change despite theoccurrence of biochemical reactions. This ob-servation indicates that the robustness of circa-dian oscillation against temperature differencein cyanobacterial cells depends on the biochem-ical properties of three clock proteins. Althoughthe mechanism of the KaiC phosphorylationcycle remains unclear, these findings suggest arobust reaction model (Figure 3a) for temper-ature compensation in circadian clocks, at leastin cyanobacteria.

Isojima et al. (99) demonstrated that CKIε/δactivity on the PER2 protein is one of themost potent period-determining processes inthe mammalian circadian oscillator. If potentperiod-determination processes were highlysensitive to temperature, it would be very diffi-cult, if not impossible, to maintain temperaturecompensation over the entire circadian period.Thus, Isojima et al. (99) investigated the tem-perature dependency of this process in livingclock cells and found that (a) the degradationrate of mPER2, which is regulated by CKIε-dependent phosphorylation (100), and (b) theperiod length were completely temperature in-sensitive in cellulo (Figure 3b,c). These find-ings imply that the period-determination pro-cess is remarkably robust against temperaturedifferences.


Ann

u. R

ev. P

hysio

l. 20

10.7

2:57

9-60

3. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by U

nive

rsity

of T

okyo

on

02/1

4/10

. For

per

sona

l use

onl

y.


FastSlow

25°C 35°C

Robust

Balanced reaction theory

Robust reaction theory

Pi in

corp

orat

ion

(m

ol m

ol−1

CKI

)

Time (min)

25°C35°C

a

b c d

Step A Step B

Step C

Step D

Step E Oscillator

Time (min)0 120 240 27 37

Temperature (˚C)32 0 50 100 150 200 250

mPE

R2::L

UC

(%)

100

50

0

36

24

12Per

iod

Leng

th (h

)0

50

100

150

200

Figure 3System analysis of clocks. (a) Two theoretical models of temperature compensation of the circadian clock.Steps A through E indicate the basic molecular reactions of circadian clock components. In the balancedreaction theory, increasing kinetic parameters of some basic reactions (magenta) shorten period, and those ofother basic reactions (blue) lead to prolonging the period. These effects are offset so that the period issustained. In the robust reaction model, such as the cyanobacterial circadian clock, temperaturecompensation of the circadian clock is caused by reactions of which kinetic parameters are independent( green). (b) The temperature dependency of decay for the mPER2::LUC bioluminescence in mPer2Lucmouse embryonic fibroblasts (MEFs). The degradation of the mPER2::LUC protein was monitored afterthe addition of cycloheximide to MEFs. The time-course data of each sample were normalized to anapproximate function in which time point zero was 100%. Each value represents the mean ± SEM (standarderror of the mean) of the normalized data. The lines represent an approximated curve in which y = 100 attime zero and y = 50 at the averaged half-life time. The blue dots and line indicate the data at 27◦C, greenindicates 32◦C, and magenta indicates 37◦C (N = 23). (c) Temperature compensation in the period length ofmPer2Luc MEFs. The graph indicates the mean ± SEM. The gray broken line indicates the approximatedline described by the equation y = 19.02 + 0.097x, and the Q10 value between 27◦C and 37◦C calculatedfrom the equation is 0.957. (d ) Temperature dependency of the $CKIε(wt) phosphorylation activity for theβTrCP-peptide substrate. Assays were performed at 25◦C (blue) and 35◦C (magenta). Abbreviations:$CKIε(wt), catalytic domain of wild-type (wt) casein kinase I epsilon (CKIε); Q10, the change in the rate ofa process caused by increasing the temperature by 10◦C; βTrCP, β-transducin repeats–containing protein.

To examine the biochemical foundation un-derlying the observed temperature insensitiv-ity, Isojima et al. (99) analyzed the phospho-rylation activity of CKIε and CKIδ in vitro.They used synthetic peptide substrate de-rived from the putative β-transducin repeats–containing protein (βTrCP)-binding region ofmouse PER2 and the catalytic domain of wild-type CKIε, lacking the C-terminal regulatory

domain [$CKIε(wt)], to prevent the confusionthat could result from the autophosphorylationof this regulatory domain and the subsequentrepression of CKIε kinase activity. This use ofthe catalytic domain was also justified by evi-dence that CKIε is kept in a dephosphorylated,active state in vivo (105). Under this experimen-tal condition, $CKIε(wt) phosphorylated thepeptide substrate at similar rates both at 25◦C

588 Ukai · Ueda

Ann

u. R

ev. P

hysio

l. 20

10.7

2:57

9-60

3. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by U

nive

rsity

of T

okyo

on

02/1

4/10

. For

per

sona

l use

onl

y.


and at 35◦C, indicating a strong temperatureinsensitivity (Q10 = 1.0) (Figure 3d ). Simi-lar temperature insensitivity was also observedwith $CKIδ(wt) at a Q10 of 1.2. However, theQ10 of $CKIε/δ(wt) activity for two clock-unrelated substrates—casein, a conventionalprotein substrate, and CK1tide, a commerciallyavailable peptide substrate for CKIε/δ, whichis a peptide substrate already phosphorylated attwo residues on the N terminus of the CKIε/δphosphorylation site—was 1.6 for casein and1.4 for CK1tide. This finding indicates that thetemperature insensitivity depends substantiallyon the substrate. Earlier and Isojima et al.’sreports (99, 106) indicated that preincubatingfull-length CKIε with ATP repressed CKIε en-zymatic activity ∼15-fold and 8-fold at 25◦Cand 35◦C, respectively (99), probably as a resultof the autophosphorylation of the C-terminalregulatory domain. Isojima et al. (99) investi-gated the effect of autophosphorylation of full-length CKIε on temperature insensitivity; theirresults indicated that autophosphorylation in-creases temperature sensitivity, suggesting thatCKI-dependent phosphorylation can be tem-perature insensitive, depending both on thesubstrate and on the state of the enzyme. More-over, the expression of exogenous full-lengthmPER2 protein and the catalytic domain ofCKIε in NIH3T3 cells recapitulated the tem-perature insensitivity of degradation rate. Onthe basis of these experimental data, Isojimaet al. (99) proposed that CKIε/δ-dependentphosphorylation is probably a temperature-insensitive period-determining process in themammalian circadian clock.

The evidence for the existence of atemperature-insensitive reaction in vitro and incellulo, with possible implications for tempera-ture compensation in circadian clocks, suggeststhe surprising capability of CKIε/δ-dependentphosphorylation. Therefore, the remainingchallenge is to obtain the atomic resolutionmodel of this temperature-insensitive reaction.

In addition to the robustness of the cir-cadian clock against temperature differences,mammalian circadian clocks also seem to have

Singularity:the suppression ofcircadian rhythms by acritical perturbation

additional interesting dynamical characteris-tics. Recently, Dibner et al. (107) manipulatedthe general transcriptional rates by using a spe-cific inhibitor of RNA polymerases II and III;they suggested that mammalian circadian oscil-lators are resilient to large fluctuations in gen-eral transcription rates. The authors also sug-gested that PER1 has an important functionin transcription compensation (107). As thesereports have shown, static perturbation of thesystem of interest and subsequent quantitativemeasurements of its dynamics yield importantresults that will aid our understanding of thesystem.

CONTROL OF CLOCKSSystem control aims to regulate the target sys-tem toward the desired state through the pre-cise perturbation of its components. To achievethis regulation, one must develop an assay sys-tem that can be controlled in a dynamic andquantitative manner.

The circadian clock is known to be entrain-able by external cues such as light. Informationobtained from light is transmitted to the cir-cadian clock through sensing mechanisms con-taining photoreceptors, and as a result of lightpulses the dynamics of the clock system changedrastically. In this section, we describe the in-tentional control of the oscillating clock systemin individual cultured cells via artificial light-sensing mechanisms (108, 109). We also explainhow Ukai et al. (108) and Pulivarthy et al. (109)independently applied this photoperturbationsystem to a longstanding and unsolved biolog-ical phenomenon known as the singularity be-havior of circadian clocks.

Circadian clocks exhibit various dynamicproperties, making them difficult to elucidatewithout quantitative perturbation and precisemeasurement of their dynamics. One of themost fundamental but still unsolved dynami-cal properties of circadian clocks is singularitybehavior, in which robust circadian rhythmicitycan be abolished after a certain critical stimulus,such as a light or temperature pulse applied at


Ann

u. R

ev. P

hysio

l. 20

10.7

2:57

9-60

3. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by U

nive

rsity

of T

okyo

on

02/1

4/10

. For

per

sona

l use

onl

y.


the appropriate timing and strength. Since thefirst report of singularity behavior in Drosophilapseudoobscura by Winfree (110), circadian clocksingularities have been experimentally observedin various organisms, including unicells suchas Gonyaulax (111), Euglena (112), and Chlamy-domonas (113); fungi (114); insects (115); plants(116, 117); and mammals (118, 119). This ev-idence suggests that this behavior is a sharedproperty of an extremely broad range of circa-dian clocks (4).

Although singularity behavior has beenwidely observed, little is known about its un-derlying mechanisms. Because such behaviorswere experimentally observed at the multicelllevel (i.e., the collective behavior of unicells orthe physiological or locomotor activity of mul-ticellular organisms), two alternative single-cell-level mechanisms have been proposed toexplain their collapse to singularity: (a) arrhyth-micity of individual clocks (Figure 4a) and(b) desynchronization of individual rhythmi-cally oscillating clocks (Figure 4b) (115, 120).In the former mechanism, individual clocksbecome arrhythmic; that is, the amplitude ofthe individual cells is substantially attenuatedby the application of the critical light pulse. Incontrast, in the latter mechanism of desynchro-nization, the phases of individual clocks arediversified by the critical light pulse. Althoughboth mechanisms can explain substantialsuppression of the multicell-level amplitude ofcircadian rhythm, the dynamical properties ofthe two fundamentally differ: The oscillationsof individual cells are impaired in the arrhyth-micity mechanism, whereas individual cellsmaintain their oscillations in the desynchro-nization mechanism. Importantly, althoughmany researchers have observed multicell-levelsingularity behaviors in various organisms,it remains unknown whether arrhythmicityor desynchronization of individual clocksunderlies the singularity behavior of circadianclocks.

Determination of the underlying mecha-nism for singularity behaviors of circadianclocks may require adjustable perturbation,

given that the ability of a critical stimulus todrive circadian clocks into singularity dependson its timing and strength. Various stimuli,such as reagents and temperature, have beenreported to directly reset mammalian cellularclocks (7–9, 101, 121–126). Unfortunately, it isdifficult, although not impossible, for these fac-tors to achieve the requisite flexibility in tim-ing and strength. In contrast to perturbationsachieved by the use of reagents or temperaturechange, photoperturbation provides an idealrange of adjustability in timing and strength.Whereas most mammalian cells cannot senselight, recent studies showed that mammaliancells (specifically Neuro-2a and HEK293 cells)become photoresponsive following the intro-duction of an exogenous Gαq protein–coupledphotoreceptor, melanopsin (also known asOpn4 ) (127, 128). It was reported that photo-stimulation of melanopsin triggers a release ofintracellular calcium mediated through the Gαqprotein signaling pathway; importantly, thereare several reports that mammalian cellularclocks can be reset by this pathway via a releaseof intracellular calcium (123, 124). These re-sults suggest that melanopsin-dependent pho-toperturbation may enable the adjustable andquantitative perturbation of mammalian cel-lular clocks by changing intracellular calciumlevels.

To experimentally reveal the underlyingmechanism of the singularity behavior ofmammalian cells, Ukai et al. (108) syn-thetically implemented photoresponsive mam-malian cells by exogenously introducingmelanopsin (Figure 4c). To continuously andquantitatively monitor the effect of photoper-turbation on the state of cellular clocks, theauthors devised a high-throughput monitoringsystem with a light-exposure unit. Using thissystem, Ukai et al. revealed that a critical lightpulse drives cellular clocks into a singularity be-havior in which robust circadian rhythmicitycan be abolished after a certain stimulus. The-oretical analysis and subsequent single-cell-level observation consistently predicted anddirectly observed that desynchronization of

590 Ukai · Ueda

Ann

u. R

ev. P

hysio

l. 20

10.7

2:57

9-60

3. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by U

nive

rsity

of T

okyo

on

02/1

4/10

. For

per

sona

l use

onl

y.


a

b

c

Gaq

PLCβ4

PKC

DAG

IP3Ca2+ stores

b g

Melanopsin

Photoresponsive NIH3T3

Synthetic implementation of photoresponsiveness

Melanopsin

Circadian clock

Photoinsensitive NIH3T3

Circadian clock

Ca2+

Biol

umin

esce

nce

Time (h)

Time (h)

Biol

umin

esce

nce

PIP2

Figure 4System control ofclocks. (a,b) Diagramsof two alternativesingle-cell-levelmechanisms formulticell-levelsingularity behavior.(a) Arrhythmicity and(b) desynchronizationof individual cellularclocks. (c) Syntheticimplementation ofphotoresponsivenesswithin mammalianclock cells. Schematicrepresentation ofmelanopsin-dependentphotoresponsiveNIH3T3 cells and theknown Gαq signalingpathway.Abbreviations: DAG,diacylglycerol; IP3,inositol triphosphate;PIP2,phosphatidylinositol4,5-bisphosphate;PLCβ4, phospholipaseC beta 4; PKC,protein kinase C.

individual cellular clocks underlies this sin-gularity behavior. Ukai et al. (108) also con-structed a theoretical framework to explain whysingularity behaviors have been experimentally

observed in various organisms, and they pro-posed desynchronization as a plausible mecha-nism for the observable singularity of circadianclocks.


Ann

u. R

ev. P

hysio

l. 20

10.7

2:57

9-60

3. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by U

nive

rsity

of T

okyo

on

02/1

4/10

. For

per

sona

l use

onl

y.


In an independent study, Pulivarthy et al.(109) implemented the same photoreceptor(Opn4 ) into the immortalized fibroblast de-rived from Per2LUC knockin mice and createdPer2LUC;Opn4 cell lines. Using this cell line,Pulivarthy et al. observed a 40% amplitude re-duction of PER2::LUC bioluminescence levelsof individual cells under the experimental con-dition inducing singularity-like behavior. Thisreduction had not been observed in Ukai et al.’sstudy using Per2-Luc and Bmal1-Luc. How-ever, Pulivarthy et al. and Ukai et al. simultane-ously observed the remarkable desynchroniza-tion of individual cellular oscillations. Althoughthere are some differences between the twostudies, which could be due to different meansof readouts, each observed desynchronizationof individual cellular oscillations, which thusseems to underlie the singularity behavior ofmammalian circadian clocks. Importantly, thesein cellulo and in silico findings are further sup-ported by in vivo observations by Ukai et al.(108) that desynchronization actually underliesthe multicell-level amplitude decrease in ratSCN that is induced by the critical light pulses,which in turn can predispose organisms’ loco-motor activity to transient amplitude decrease.

To elucidate the underlying mechanism ofsingularity behaviors, Winfree (115, 120) con-ducted a two-pulse experiment that revealed theunclocklike behavior of the circadian clock. Inthis experiment, the critical light pulse inducingthe singularity behavior seemed to decrease theamplitude of the circadian clock without affect-ing its frequency. This finding seemed to con-tradict both the tenets of the simple limit-cyclemodel and its prediction that arrhythmicityunderlies the singularity behavior. To explainthis unclocklike behavior, Winfree proposedthe so-called clockshop hypothesis, in whichan organism-level circadian clock consists ofmultiple circadian oscillators with substantialfluctuations, and predicted that the desyn-chronization of individual circadian oscillatorsunderlie the singularity behavior (multicell-level amplitude decrease). He was unable to testthis prediction, however, as there was no way toobserve single-cell-level circadian rhythmicity

at that time (1975). The studies by Ukaiet al. (108) and Pulivarthy et al. (109) provedWinfree’s prediction on the desynchronization,at least in the mammalian circadian system,more than 30 years after it was originallyproposed (120).

DESIGN OF CLOCKSThe final step in the systems-biological pro-cess is system design, the reconstruction anddesign of a dynamical system based on the de-sign principles revealed through the efforts ofsystem identification, system analysis, and sys-tem control. In Escherichia coli, various dynam-ical systems, including the toggle switch (129),the repressilator (130), the metabolic oscilla-tor (131), and the tunable oscillator (132), havebeen designed and implemented. Tigges et al.(133) recently designed and implemented tun-able oscillators in mammalian cells. Such a syn-thetic approach can be applied to a complex anddynamic system such as a mammalian circadianclock and will enhance its understanding. A syn-thetic approach is especially useful in validat-ing the hypothesis derived from the identifiedstructure and observed dynamics.

A possible synthetic approach is to extendthe in cellulo cycling assay system describedabove (27) and utilize it as a physical simu-lator, which allows implementation of artifi-cial transcriptional circuits to mimic transcrip-tional circuits of mammalian circadian clocksand, hence, to test the design principles of nat-ural systems. To this end, Ukai-Tadenuma et al.(134) have developed an in cellulo system withwhich to validate the sufficiency of the compo-nents of a natural circadian phase-controllingmechanism. Alternatively, a more radical syn-thetic approach, such as an in vitro reconstruc-tion of a mammalian circadian clock, could beundertaken. We describe these two syntheticapproaches in the following sections.

Design of Transcriptional CircuitsThe network structure comprising clock genesand CCEs has been comprehensively described.

592 Ukai · Ueda

Ann

u. R

ev. P

hysio

l. 20

10.7

2:57

9-60

3. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by U

nive

rsity

of T

okyo

on

02/1

4/10

. For

per

sona

l use

onl

y.


However, the dynamic principles governingthis transcriptional circuit remain elusive. Akey issue concerning the logic of the mam-malian circadian clock is how the expressionpeaks (phases) of circadian oscillating genesare determined. To understand the molecularlogic of the phase-controlling system(s), Ukai-Tadenuma et al. (134) focused on the threemain CCEs: the E/E′-box, the D-box, and theRRE. The transcriptional activator DBP acti-vates gene expression via the D-box, whereasthe transcriptional repressor E4BP4 repressesgene expression (Figure 2a). Dbp is regulatedby the E-box, the morning control element;E4bp4, however, is regulated by the RRE, thenighttime control element. On the basis ofthis transcriptional circuit information, the in-vestigators hypothesized that a morning ac-tivator and a nighttime repressor determinethe daytime transcriptional output mediatedthrough the D-box. Similarly, RRE activators(e.g., Rorα) are expressed during the daytimeunder the control of the D-box, and the RRErepressors (e.g., RevErbα) seem to be stronglyinfluenced by a morning element (the E′-box;Figure 2a). These observations led to the hy-pothesis that a daytime activator and a morningrepressor can specify the nighttime transcrip-tional output mediated through the RRE. Totest this hypothesis, Ukai-Tadenuma et al. (134)adopted a synthetic approach to physically sim-ulate an identified structure, and they observedthe resulting dynamics through use of artificialtranscriptional circuits.

To design and implement artificial transcrip-tional circuits in mouse NIH3T3 cells, whichhave a self-oscillating circadian clock, Ukai-Tadenuma et al. (134) developed an in vitro cy-cling assay system composed of three compo-nents: (a) an artificial activator (a destabilizedGAL4-VP16 fusion protein, dGAL4-VP16),(b) an artificial repressor (a destabilized GAL4protein, dGAL4), and (c) an output reportergene (dLuc) driven by a minimal TATA boxfused with four tandem repeats of the upstreamactivator sequence (UAS; the GAL4-bindingsequence) (Figure 5a). Because the artificial ac-tivator and repressor are driven by the SV40

UAS: upstreamactivator sequence

basic promoter fused with three tandem repeatsof CCEs, the expression timing of these artifi-cial regulators is controlled via the morning (E′-box), daytime (D-box), and nighttime (RRE) el-ements. The authors reasoned that if and only ifthe phases of the transcriptional activator(s) andrepressor(s) are acceptable determinants of thephase of the downstream transcriptional out-put, then it should be possible to generate thenatural phases using synthetic transcriptionalregulators and promoters. If this proved to betrue, it would be very significant, as transcrip-tion factors are regulated by various posttran-scriptional mechanisms—including translation,phosphorylation, ubiquitination, sumoylation,and nuclear transportation—that are thoughtto contribute at least in part to the controlof phases of the downstream transcriptionaloutputs.

Using the in vitro cycling assay system as aphysical simulator, Ukai-Tadenuma et al. (134)tested the hypothesis for daytime output byexamining the dynamic behavior of the tran-scriptional output generated from the compe-tition between an artificial morning activatorcontrolled via E′-box and a nighttime repressorcontrolled via RRE. The phases of the activa-tor and repressor in this experiment were de-tected at CT4.0 ± 0.29 (n = 2) and CT17.1 ±0.37 (n = 2). The transcriptional output drivenby these regulators exhibited circadian oscil-lation with a phase at CT7.7 ± 0.85 (n =2), which is very close (≤1.0 h) to the cor-responding natural daytime (D-box) phase,CT8.7 ± 1.13 (Figure 5b). These results sug-gest that morning activation and nighttime re-pression are sufficient to determine the day-time transcriptional output. Importantly, high-amplitude circadian oscillation was not ob-served when only the morning activator or onlythe nighttime repressor was expressed. The au-thors concluded that both morning activationand nighttime repression are necessary for day-time output.

Moreover, the hypothesis for nighttime out-put was tested through examination of an-other artificial circuit consisting of an artificialdaytime activator under D-box control and a


Ann

u. R

ev. P

hysio

l. 20

10.7

2:57

9-60

3. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by U

nive

rsity

of T

okyo

on

02/1

4/10

. For

per

sona

l use

onl

y.


Morning Daytime

Luc-PEST

dGAL4

Luciferin+ATP

Bioluminescence

dGAL4-VP16 dLuc

CCE × 3 − SV40 promoter

dGAL4-VP16

Gal4-VP16-PEST

dGAL4

Gal4-PEST

Output from the artificial network

Artificial transcriptional repressor

Artificial transcriptional activator

Artificial target promoter

Morning Daytime

Nighttime

4.0 7.7

17.1

3.9 8.9

a

b c

dGAL4

dGAL4-VP16

dGAL4 dGAL4-VP16

CCE × 3 − SV40 promoter

UAS × 4 − CMV promoter

Nighttime

16.6

Repression Activation

Repression Repression Activation

Activation

Figure 5System design of clocks. (a) The artificial transcriptional system. dGAL4-VP16 and dGAL4 were used as theactivator and the repressor, respectively. These transcriptional factors were expressed under the control ofthree tandem repeats of a clock-controlled element (CCE): the E′-box of Per2, the D-box of Per3, and theRRE (RevErbA/ROR-binding element) of Bmal1. The artificial activator and repressor competitively bindthe four tandem repeats of the upstream activator sequences (UAS) in the artificial promoter to regulate theoutput reporter gene dLuc. NIH3T3 cells were transiently transfected with plasmids harboring the activator,repressor, or reporter, and the reporter activity was monitored under a real-time bioluminescence-measuringsystem, which allowed observation of this artificial circuit’s dynamic behavior. (b and c) Proof by synthesis ofdaytime and nighttime transcriptional regulations. Synthesis of (b) daytime and (c) nighttime expressionsfrom two different artificial transcriptional circuits: (b) the morning activator under E′-box control and thenighttime repressor under RRE control and (c) the daytime activator under D-box control and the morningrepressor under E′-box control. The schemes summarize the timing of the peaks (i.e., phases) of promoteractivity. The activator ( green oval and arrow), repressor (magenta oval and arrow), and output reporter areindicated with their phases in circadian time (numbers).

594 Ukai · Ueda

Ann

u. R

ev. P

hysio

l. 20

10.7

2:57

9-60

3. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by U

nive

rsity

of T

okyo

on

02/1

4/10

. For

per

sona

l use

onl

y.


morning repressor under E′-box control. Thephases of the transcriptional activator and re-pressor in this experiment were at CT8.9 ±0.28 (n = 2) and CT3.9 ± 0.13 (n = 2). Thetranscriptional output driven by the regulatorsexhibited circadian oscillation with a phase atCT16.6 ± 1.04 (n = 2), which is very close(≤1.0 h) to the corresponding natural nighttimephase (CT17.0 ± 0.81) (Figure 5c). These re-sults also suggest that daytime activation andmorning repression are necessary and suffi-cient to generate the circadian nighttime out-put, which implies that the input phases of tran-scriptional regulators can determine the phasesof transcriptional output. These findings led tothe idea that various combinations of transcrip-tional regulators with CCEs for the three basiccircadian phases (morning, daytime, and night-time) may generate not only the basic phases butalso other phases. Indeed, Ukai-Tadenuma et al.(134) succeeded in generating various phasesthough simple combinations of the transcrip-tional regulators of the three basic circadianphases.

These experiments showed that the tran-scriptional regulation of upstream transcrip-tion factors can determine the phase of thedownstream output. This study presents a syn-thetic approach to the proof by synthesis oftranscriptional logic, which provides us witha new strategy with which to (a) investigatethe requirements for identified componentsand/or their interactions and (b) reveal as-yet-unidentified components or interactions.For example, although the authors success-fully (re)generated two basic phases, daytimeand nighttime—as well as additional phasesnear the subjective noon, dawn, dusk, and latenight—from the three basic circadian phases,we have not yet been able to regenerate the ba-sic morning phase, which is expected to be reg-ulated directly or indirectly by the three basicphases to maintain circadian oscillations. Thus,morning transcriptional regulation remains amissing link in the mammalian clock system.Given that a strong repressor at evening phaseseems indispensable to the reconstruction of the

morning phase, a candidate transcription fac-tor could be CRY1, a transcriptional repres-sor expressed during the evening. However,the expression-regulation mechanism of theCry1 gene remains unknown at present becausemore than one DNA element seems to be in-volved in this expression. Hence, the challengeof synthesizing evening and morning expres-sion, as we work toward the complete recon-struction of the transcriptional circuits underly-ing mammalian circadian clocks, remains to besolved.

Reconstitution ofPosttranscriptional CircuitsIn the preceding section, we described the de-sign of the transcriptional circuits of the mam-malian circadian clock. The proof by synthesisof transcriptional circuits is useful because tran-scriptional feedback repression is important forcircadian clock function, as we discussed in thesection entitled The Role of Negative Feed-back. However, the period length of circadianclock is largely determined by posttranscrip-tional circuits, as we discussed in the section en-titled The Identification of the Rate-LimitingProcess. Therefore, synthetic approaches basedon biochemistry toward posttranscriptional cir-cuits will be necessary for attaining a com-plete understanding of the mammalian circa-dian clock. Among such synthetic approaches,one of the most radical and fundamental isto perform an in vitro reconstruction, fromscratch, of the mammalian circadian clock.

Indeed, biochemical studies on the circadianclock of cyanobacteria, the simplest organismsto possess such a clock, revealed the sufficiencyof posttranscriptional circuits in the circadianclock (72). Interestingly, although ubiquitousmolecular behaviors concerning the circadianclock, such as negative feedback regulationof clock genes, circadian oscillation of ac-cumulation of RNA and clock proteins, andphosphorylation of clock proteins, are observedin cyanobacteria, the robust circadian oscilla-tion of the phosphorylation state of KaiC was


Ann

u. R

ev. P

hysio

l. 20

10.7

2:57

9-60

3. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by U

nive

rsity

of T

okyo

on

02/1

4/10

. For

per

sona

l use

onl

y.


reconstituted via a mixture of only threecyanobacterial clock proteins and ATP in a testtube (72). The circadian oscillation of KaiCphosphorylation was therefore demonstratedto be the central oscillator of the cyanobac-terial circadian clock. This elegant study onthe cyanobacterial circadian clock evoked theimportance of using a biochemical approachto the circadian clock to answer fundamen-tal questions, such as those of autonomousoscillation and temperature compensation.

As described in the section entitled Iden-tification of Clocks, the clock-related geneshave been comprehensively identified in mam-mals, as well as in other model organisms suchas Drosophila and Neurosphora (3, 4, 70, 71).Furthermore, Isojima et al. (99) have discov-ered that CKIε/δ-dependent phosphorylationexhibits temperature insensitivity in vitro, as docyanobacteria (72). Although it is not knownwhether a temperature-insensitive reverse reac-tion (i.e., dephosphorylation) exists in the mam-malian circadian clock, its discovery would be agreat step toward the complete reconstructionof the autonomous oscillator in the mammaliancircadian clock.

Synthetic approaches based on cell biologyand biochemistry can validate the sufficiencyof transcriptional and posttranscriptional cir-cuits for the dynamical property of interest. Fol-lowing these approaches will lead to a deeperunderstanding of the design principles of themammalian circadian clock.

PERSPECTIVESIn this review, we focused primarily on the in-tracellular mechanisms of the mammalian cir-cadian clock. However, the mammalian cir-cadian system is organized in a hierarchy ofcellular oscillators and exhibits various in-tercellular dynamical properties derived frompopulations of clock cells. One example ofsuch a property is singularity behavior (desyn-chronization of multiple circadian clock cells),discussed in the section entitled Analysis ofClocks. Other examples, which are not dis-cussed herein, include the synchronization of

multiple clock cells within the SCN (135–137) as well as the intertissue hierarchical cou-pling between master clock (SCN) and slave(peripheral) clocks (3, 138).

Synchronization of multiple clock cellswithin the SCN is one of the most importantcharacteristics that distinguish the SCN fromperipheral clocks. Coherent circadian rhythms,probably due to intercellular couplings of in-dividual clock cells in the SCN, are generatedin the SCN (135–137). Interestingly, not onlyare intercellular interactions among individualclock cells in the SCN important for the cou-pling of multiple circadian oscillators, they arealso necessary for maintaining self-sustainedcircadian oscillations of individual clock cells(135). Indeed, self-sustained oscillations of in-dividual cells in the SCN are highly sensitive tointercellular neural communications (135) andcAMP signaling (92). Furthermore, these inter-cellular coupling–dependent self-sustained os-cillations within the SCN seem able to rescuethe molecular defects of individual clock cells.For example, Liu et al. (139) recently discov-ered that Per1 and Cry1 are required for self-sustainable oscillations in peripheral clock cellsand in neurons dissociated from the SCN. Theauthors also found that intercellular couplingsin the SCN can compensate for those moleculardeficiency.

Intercellular communications in the mam-malian clock system are not limited to thesynchronization of clock cells in the SCN.Intertissue hierarchical coupling betweenmaster clock (SCN) and slave clocks playsan important role in generation orchestratedcircadian oscillations throughout the body. Infact, Kornmann et al. (140) discovered—underthe condition that transcription of the essentialcore clock gene Bmal1 is repressed in liver—that 31 genes, including core clock gene Per2,exhibited robust circadian oscillations irrespec-tive of whether the liver clock was running.In contrast, in liver explants cultured in vitro,circadian cycles of Per2 were only observedwhen hepatocyte oscillators were operational.These findings indicate that circadian oscilla-tions of gene expression observed in the liver

596 Ukai · Ueda

Ann

u. R

ev. P

hysio

l. 20

10.7

2:57

9-60

3. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by U

nive

rsity

of T

okyo

on

02/1

4/10

. For

per

sona

l use

onl

y.


of intact animals without functional hepatocyteoscillations are driven by systemic cues.

These intercellular dynamical propertieswould be difficult to understand withoutsystems-biological approaches and techniquessuch as tissue-specific, conditional, and quanti-tative perturbation and subsequent quantitativemeasurement. Kornmann et al. adopted suchconditional perturbation technique, in whichREVERBa is conditionally overexpressed andrepresses the transcription of Bmal1 in adoxycycline-dependent manner. Integrative re-search that makes use of quantitative pertur-bations with high spatial and temporal reso-lutions, quantitative measurements with highaccuracy, and appropriate theoretical analysiswill enable us to elucidate unsolved systems-level questions about complex and dynamicphenomena.

CONCLUSIONFollowing identification of key clock genes, de-mand for a higher-order understanding of thedesign principles of the mammalian circadianclock has increased. In this review we have de-scribed several steps, beginning with compre-hensive identification of clock components andtheir networked interactions (system identifi-cation) and quantitative analysis (system anal-ysis) of temperature-insensitive reactions andleading to the ability to control existing sys-tems toward the desired state (system control)and to design new systems according to an un-derstanding of structure and underlying dy-namical principles (system design). We stronglybelieve that it is time to fully integrate thesesystems-biological approaches to solve system-level questions.

SUMMARY POINTS

1. Systems biology can be defined as biology after identification of key gene(s) and containsfour steps: system identification, system analysis, system control, and system design.

2. System identification refers to identification of the whole network structure throughfunctional genomics and comparative genomics.

3. System analysis involves prediction and validation to derive the design principle throughthe accurate measurement and (static) perturbation of network dynamics.

4. System control involves repair and control of the network state so as to achieve thedesired state through the precise perturbation of its components.

5. System design refers to the reconstruction and design of systems according to the designprinciples derived from the identified structure and observed dynamics.

DISCLOSURE STATEMENTThe authors are not aware of any affiliations, memberships, funding, or financial holdings thatmight be perceived as affecting the objectivity of this review.

ACKNOWLEDGMENTSThis work is supported by an intramural grant-in-aid from the RIKEN Center for Develop-mental Biology (CDB); an intramural grant-in-aid from the Genomic Science Center, RIKEN;the Director’s Fund from CDB; RIKEN Strategic Programs; the RIKEN Research Collabora-tions with Industry Program; the National Project on Protein Structural and Functional Analyses;a Genome Network Project grant from the Japanese Ministry of Education, Culture, Sports,


Ann

u. R

ev. P

hysio

l. 20

10.7

2:57

9-60

3. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by U

nive

rsity

of T

okyo

on

02/1

4/10

. For

per

sona

l use

onl

y.


Science and Technology; KAKENHI (grant-in-aid for Scientific Research) on Priority Areas Sys-tems Genomics from the Ministry of Education, Culture, Sports, Science and Technology ofJapan; a research and development project of the Industrial Science and Technology Programfrom the New Energy and Industrial Technology Development Organization (NEDO); a grant-in-aid for scientific research on “Development of Basic Technology to Control Biological SystemsUsing Chemical Compounds” from the NEDO; a grant-in-aid for strategic programs in researchand development from the Institute of Physical and Chemical Research; a grant-in-aid from theUehara Memorial Foundation; and research fellowship number 17-4169 of the Japan Society forthe Promotion of Science for Young Scientists.

LITERATURE CITED

1. Kitano H. 2002. Systems biology: a brief overview. Science 295:1662–642. Kitano H. 2002. Computational systems biology. Nature 420:206–103. Reppert SM, Weaver DR. 2002. Coordination of circadian timing in mammals. Nature 418:935–414. Dunlap JC, Loros JJ, DeCoursey PJ, eds. 2004. Chronobiology: Biological Timekeeping. Sunderland, Mass.:

Sinauer5. Panda S, Hogenesch JB, Kay SA. 2002. Circadian rhythms from flies to human. Nature 417:329–356. Yamazaki S, Numano R, Abe M, Hida A, Takahashi R, et al. 2000. Resetting central and peripheral

circadian oscillators in transgenic rats. Science 288:682–857. Balsalobre A, Damiola F, Schibler U. 1998. A serum shock induces circadian gene expression in mam-

malian tissue culture cells. Cell 93:929–378. Yagita K, Tamanini F, Van Der Horst GT, Okamura H. 2001. Molecular mechanisms of the biological

clock in cultured fibroblasts. Science 292:278–819. Akashi M, Nishida E. 2000. Involvement of the MAP kinase cascade in resetting of the mammalian

circadian clock. Genes Dev. 14:645–4910. Vollmers C, Panda S, DiTacchio L. 2008. A high-throughput assay for siRNA-based circadian screens

in human U2OS cells. PLoS One 3:e345711. King DP, Zhao Y, Sangoram AM, Wilsbacher LD, Tanaka M, et al. 1997. Positional cloning of the

mouse circadian clock gene. Cell 89:641–5312. Bunger MK, Wilsbacher LD, Moran SM, Clendenin C, Radcliffe LA, et al. 2000. Mop3 is an essential

component of the master circadian pacemaker in mammals. Cell 103:1009–1713. Reick M, Garcia JA, Dudley C, McKnight SL. 2001. NPAS2: an analog of clock operative in the mam-

malian forebrain. Science 293:506–914. Bertolucci C, Cavallari N, Colognesi I, Aguzzi J, Chen Z, et al. 2008. Evidence for an overlapping role

of CLOCK and NPAS2 transcription factors in liver circadian oscillators. Mol. Cell. Biol. 28:3070–7515. Bae K, Jin X, Maywood ES, Hastings MH, Reppert SM, Weaver DR. 2001. Differential functions of

mPer1, mPer2, and mPer3 in the SCN circadian clock. Neuron 30:525–3616. Zheng B, Albrecht U, Kaasik K, Sage M, Lu W, et al. 2001. Nonredundant roles of the mPer1 and mPer2

genes in the mammalian circadian clock. Cell 105:683–9417. Van Der Horst GT, Muijtjens M, Kobayashi K, Takano R, Kanno S, et al. 1999. Mammalian Cry1 and

Cry2 are essential for maintenance of circadian rhythms. Nature 398:627–3018. Vitaterna MH, Selby CP, Todo T, Niwa H, Thompson C, et al. 1999. Differential regulation of mam-

malian period genes and circadian rhythmicity by cryptochromes 1 and 2. Proc. Natl. Acad. Sci. USA96:12114–19

19. Lowrey PL, Shimomura K, Antoch MP, Yamazaki S, Zemenides PD, et al. 2000. Positional synteniccloning and functional characterization of the mammalian circadian mutation τ. Science 288:483–92

20. Xu Y, Padiath QS, Shapiro RE, Jones CR, Wu SC, et al. 2005. Functional consequences of a CKIδmutation causing familial advanced sleep phase syndrome. Nature 434:640–44

21. Sato TK, Panda S, Miraglia LJ, Reyes TM, Rudic RD, et al. 2004. A functional genomics strategy revealsRora as a component of the mammalian circadian clock. Neuron 43:527–37

598 Ukai · Ueda

Ann

u. R

ev. P

hysio

l. 20

10.7

2:57

9-60

3. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by U

nive

rsity

of T

okyo

on

02/1

4/10

. For

per

sona

l use

onl

y.


22. Preitner N, Damiola F, Lopez-Molina L, Zakany J, Duboule D, et al. 2002. The orphan nuclear re-ceptor REV-ERBα controls circadian transcription within the positive limb of the mammalian circadianoscillator. Cell 110:251–60

23. Wuarin J, Schibler U. 1990. Expression of the liver-enriched transcriptional activator protein DBPfollows a stringent circadian rhythm. Cell 63:1257–66

24. Mitsui S, Yamaguchi S, Matsuo T, Ishida Y, Okamura H. 2001. Antagonistic role of E4BP4 and PARproteins in the circadian oscillatory mechanism. Genes Dev. 15:995–1006

25. Honma S, Kawamoto T, Takagi Y, Fujimoto K, Sato F, et al. 2002. Dec1 and Dec2 are regulators of themammalian molecular clock. Nature 419:841–44

26. Zylka MJ, Shearman LP, Weaver DR, Reppert SM. 1998. Three period homologs in mammals: differ-ential light responses in the suprachiasmatic circadian clock and oscillating transcripts outside of brain.Neuron 20:1103–10

27. Ueda HR, Chen W, Adachi A, Wakamatsu H, Hayashi S, et al. 2002. A transcription factor responseelement for gene expression during circadian night. Nature 418:534–39

28. Gekakis N, Staknis D, Nguyen HB, Davis FC, Wilsbacher LD, et al. 1998. Role of the CLOCK proteinin the mammalian circadian mechanism. Science 280:1564–69

29. Kume K, Zylka MJ, Sriram S, Shearman LP, Weaver DR, et al. 1999. mCRY1 and mCRY2 are essentialcomponents of the negative limb of the circadian clock feedback loop. Cell 98:193–205

30. Grundschober C, Delaunay F, Puhlhofer A, Triqueneaux G, Laudet V, et al. 2001. Circadian regulationof diverse gene products revealed by mRNA expression profiling of synchronized fibroblasts. J. Biol.Chem. 276:46751–58

31. Storch KF, Lipan O, Leykin I, Viswanathan N, Davis FC, et al. 2002. Extensive and divergent circadiangene expression in liver and heart. Nature 417:78–83

32. Akhtar RA, Reddy AB, Maywood ES, Clayton JD, King VM, et al. 2002. Circadian cycling of the mouseliver transcriptome, as revealed by cDNA microarray, is driven by the suprachiasmatic nucleus. Curr.Biol. 12:540–50

33. Duffield GE, Best JD, Meurers BH, Bittner A, Loros JJ, Dunlap JC. 2002. Circadian programs oftranscriptional activation, signaling, and protein turnover revealed by microarray analysis of mammaliancells. Curr. Biol. 12:551–57

34. Panda S, Antoch MP, Miller BH, Su AI, Schook AB, et al. 2002. Coordinated transcription of keypathways in the mouse by the circadian clock. Cell 109:307–20

35. Oishi K, Miyazaki K, Kadota K, Kikuno R, Nagase T, et al. 2003. Genome-wide expression analysis ofmouse liver reveals CLOCK-regulated circadian output genes. J. Biol. Chem. 278:41519–27

36. Grechez-Cassiau A, Panda S, Lacoche S, Teboul M, Azmi S, et al. 2004. The transcriptional repressorSTRA13 regulates a subset of peripheral circadian outputs. J. Biol. Chem. 279:1141–50

37. Rudic RD, McNamara P, Reilly D, Grosser T, Curtis AM, et al. 2005. Bioinformatic analysis of circadiangene oscillation in mouse aorta. Circulation 112:2716–24

38. Oishi K, Amagai N, Shirai H, Kadota K, Ohkura N, Ishida N. 2005. Genome-wide expression analysisreveals 100 adrenal gland–dependent circadian genes in the mouse liver. DNA Res. 12:191–202

39. Ben-Shlomo R, Akhtar RA, Collins BH, Judah DJ, Davies R, Kyriacou CP. 2005. Light pulse–inducedheme and iron-associated transcripts in mouse brain: a microarray analysis. Chronobiol. Int. 22:455–71

40. Gery S, Gombart AF, Yi WS, Koeffler C, Hofmann WK, Koeffler HP. 2005. Transcription profiling ofC/EBP targets identifies Per2 as a gene implicated in myeloid leukemia. Blood 106:2827–36

41. Menger GJ, Lu K, Thomas T, Cassone VM, Earnest DJ. 2005. Circadian profiling of the transcriptomein immortalized rat SCN cells. Physiol. Genomics 21:370–81

42. Lemos DR, Downs JL, Urbanski HF. 2006. Twenty-four-hour rhythmic gene expression in the rhesusmacaque adrenal gland. Mol. Endocrinol. 20:1164–76

43. Oster H, Damerow S, Hut RA, Eichele G. 2006. Transcriptional profiling in the adrenal gland revealscircadian regulation of hormone biosynthesis genes and nucleosome assembly genes. J. Biol. Rhythms21:350–61

44. Yang S, Wang K, Valladares O, Hannenhalli S, Bucan M. 2007. Genome-wide expression profiling andbioinformatics analysis of diurnally regulated genes in the mouse prefrontal cortex. Genome Biol. 8:R247


Ann

u. R

ev. P

hysio

l. 20

10.7

2:57

9-60

3. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by U

nive

rsity

of T

okyo

on

02/1

4/10

. For

per

sona

l use

onl

y.


45. Porterfield VM, Piontkivska H, Mintz EM. 2007. Identification of novel light-induced genes in thesuprachiasmatic nucleus. BMC Neurosci. 8:98

46. Reddy AB, Maywood ES, Karp NA, King VM, Inoue Y, et al. 2007. Glucocorticoid signaling synchronizesthe liver circadian transcriptome. Hepatology 45:1478–88

47. Menger GJ, Allen GC, Neuendorff N, Nahm SS, Thomas TL, et al. 2007. Circadian profiling of thetranscriptome in NIH/3T3 fibroblasts: comparison with rhythmic gene expression in SCN2.2 cells andthe rat SCN. Physiol. Genomics 29:280–89

48. Zvonic S, Ptitsyn AA, Kilroy G, Wu X, Conrad SA, et al. 2007. Circadian oscillation of gene expressionin murine calvarial bone. J. Bone Miner. Res. 22:357–65

49. McCarthy JJ, Andrews JL, McDearmon EL, Campbell KS, Barber BK, et al. 2007. Identification of thecircadian transcriptome in adult mouse skeletal muscle. Physiol. Genomics 31:86–95

50. Miller BH, McDearmon EL, Panda S, Hayes KR, Zhang J, et al. 2007. Circadian and CLOCK-controlledregulation of the mouse transcriptome and cell proliferation. Proc. Natl. Acad. Sci. USA 104:3342–47

51. Dupre SM, Burt DW, Talbot R, Downing A, Mouzaki D, et al. 2008. Identification of melatonin-regulated genes in the ovine pituitary pars tuberalis, a target site for seasonal hormone control.Endocrinology 149:5527–39

52. Fukuhara C, Tosini G. 2008. Analysis of daily and circadian gene expression in the rat pineal gland.Neurosci. Res. 60:192–98

53. Bailey MJ, Coon SL, Carter DA, Humphries A, Kim JS, et al. 2009. Night/day changes in pinealexpression of >600 genes: central role of adrenergic/cAMP signaling. J. Biol. Chem. 284:7606–22

54. Lemos DR, Downs JL, Raitiere MN, Urbanski HF. 2009. Photoperiodic modulation of adrenal glandfunction in the rhesus macaque: effect on 24-h plasma cortisol and dehydroepiandrosterone sulfaterhythms and adrenal gland gene expression. J. Endocrinol. 201:275–85

55. Lipshutz RJ, Fodor SP, Gingeras TR, Lockhart DJ. 1999. High density synthetic oligonucleotide arrays.Nat. Genet. 21:20–24

56. Suzuki Y, Taira H, Tsunoda T, Mizushima-Sugano J, Sese J, et al. 2001. Diverse transcriptional initiationrevealed by fine, large-scale mapping of mRNA start sites. EMBO Rep. 2:388–93

57. Hogenesch JB, Gu YZ, Jain S, Bradfield CA. 1998. The basic helix-loop-helix PAS orphan MOP3forms transcriptionally active complexes with circadian and hypoxia factors. Proc. Natl. Acad. Sci. USA95:5474–79

58. Ueda HR, Hayashi S, Chen W, Sano M, Machida M, et al. 2005. System-level identification of tran-scriptional circuits underlying mammalian circadian clocks. Nat. Genet. 37:187–92

59. Yoo SH, Ko CH, Lowrey PL, Buhr ED, Song EJ, et al. 2005. A noncanonical E-box enhancer drivesmouse Period2 circadian oscillations in vivo. Proc. Natl. Acad. Sci. USA 102:2608–13

60. Falvey E, Marcacci L, Schibler U. 1996. DNA-binding specificity of PAR and C/EBP leucine zipper pro-teins: A single amino acid substitution in the C/EBP DNA-binding domain confers PAR-like specificityto C/EBP. Biol. Chem. 377:797–809

61. Harding HP, Lazar MA. 1993. The orphan receptor Rev-ErbAα activates transcription via a novelresponse element. Mol. Cell. Biol. 13:3113–21

62. Paquet ER, Rey G, Naef F. 2008. Modeling an evolutionary conserved circadian cis-element. PLoSComput. Biol. 4:e38

63. Nakahata Y, Yoshida M, Takano A, Soma H, Yamamoto T, et al. 2008. A direct repeat of E-box-likeelements is required for cell-autonomous circadian rhythm of clock genes. BMC Mol. Biol. 9:1

64. Kumaki Y, Ukai-Tadenuma M, Uno

systems biology of mammalian circadian clocks · systems biology of mammalian circadian clocks...

Documents