Sunrise, Mérida cable car station, Venzuela

SunriseMérida,

Venezuela

Biological timing for days and for years

Andrew Millar, Professor of Systems BiologyUniversity of Edinburgh

Biological clocks in theory and experiments www.amillar.org

Current:Kieron EdwardsKirsten KnoxJohn O’NeillLaura DixonQian XingAdrian Thomson

Ozgur AkmanKevin StratfordTreenut SaithongOxana SorokinaAlexandra PokhilkoCarl Troein

Ruth Bastow (GARNet)

Past:Simon ThainKamal SwarupRuth BastowHarriet McWattersShigeru HananoSeth DavisMandy Dowson-DayGiovanni MurtasNeeraj SalathiaMaria ErikssonAnthony HallAlex MortonBoris ShulginNickiesha BromleyVictoria HibberdMegan SouthernDomingo SalazarPaul BrownJames LockeLaszlo Kozma-Bognar

CollaboratorsIgor Goryanin (Informatics)EPCCIPCR / WSBC -Matthew Turner (UW Physics)David Rand (UW Maths)Bärbel Finkenstädt (Statistics)David Broomhead (UMIST)

Anthony Hall (Liverpool)Peter Lumsden (C. Lancs)Ferenc Nagy (Szeged)F.-Y. Bouget (Banyuls)

Funding: BBSRC, Gatsby, EPSRC, EU

One of six UK Centres for Sys Bio

Focus on Dynamic Modelling• 2009: C.H. Waddington building

Research Focus: Systems Modelling

• Universal requirement, currently limiting• Theoretical developments

Flexible, modular, automated• Informed by biological projects

three pilot CSBE projects + linked awards

Existing knowledge

High-resolution data

High-throughput data

Static models

Kinetic models

New knowledge

Graphical Notation

Network Inference

Process Algebras Model

analysisTheoretical projects

Widely Applicable

Experimental projects

Existing knowledge

High-resolution data

High-throughput data

New knowledge

Static models

Kinetic models

Systems Biology Software Infrastructure

Kinetic Parameter Facility

Clock

RNAInterferon

Synthetic biology

Systems Biology Research

ROBuST(SABR)

ONDex(SABR)

MiniClock(ANR)

BioGrid(BBR)

PlPortal(BBR)

Biological clocks are sophisticated

White bar = sleepEach day is plotted

twice

9-month record

The Human Body Clock in action: sleep/wake rhythms

Body clock in humans

• Test under constant environmentHalley Base, British Antarctic Survey

The Human Body Clock Is Slow

Minimal circadian system

Ubiquitous properties of circadian clocks:• ~24 period• Entrained by light/dark cycles• Temperature compensated• Negative feedback in oscillator• Intracellular

A B CD

E G

FWW

Input OutputOscillator OvertRhythms

24h

Circadian clockwork (Drosophila)

Panda et al., Nature 2002.

Circadian clock in mammals

Brain

3h 15h

Clock

WT

Gen

e ex

pres

sion

Pacemaker centre (SCN)

Schibler

Clock cells

Herzog

Firin

g ra

te

Gene network

BMAL1

Clock

Rev-erbα Tim

Per 1,2 (3)

Cry 1, 2

CLKBMAL CK1ε

Pittendrigh & Daan

Circadian clocks – summary 1

• Circadian clocks in all organisms share common regulatory properties (high level)

• Clocks drive 24-h rhythms in many important biological processes

• Circadian oscillator (rhythm generator) includes negative feedback loops of gene regulation

• Most clock protein sequences differ widely between cyanobacteria, plants, fungi and animal clocks

the casein kinases are shared… but so is the proteasome…

The clock gene circuit across species

• Similar, interlocked loops (even in cyano’s)

• Varying numbers of genes

• Varying light input (not shown)

• Why so complex?

Why such complexity?

• Multiple genes• Multiple loops• Multiple clocks

>1 per cell?

Systems biology seeks simple principles

• Flexibility- Cycle to cycle:

phase and waveform- In evolution

Photo by Martin Lange

• Multiple genesFrq isoforms

• Multiple loops• Multiple clocks

>1 per cell?

Systems biology seeks simple principles

• Robustness- Stochastic noise- Temperature compensation

Photo by Martin Lange

Why such complexity?

seed

3 mm

Circadian clock in Arabidopsis thaliana

• Model plant for genetics• Clock controls growth and flowering time

Aerial dry biomass

20 24 28LD cycle (h)Dodd et al.

Science 2005

Genome-wide circadian rhythms

~ 15% RNA transcripts rhythmic in white light

> 3,000 genes of 22,000 on array

• Functional clustering• 68% of rhythmic

transcripts also stress-regulated, (Kreps et al. 2002)

Edwards et al. Plant Cell 2006

Time in LL (h) 26 30 34 38 42 46 50 54 58 62 66 70 74 0h

Mutant plants identify genes in the clockwork

?

• Day-time, Negative regulation - CCA1/LHY, myb TFs• Night-time, Positive regulation – TOC1 +++, ‘PRR’• Build model to test potential for regulation

Alabadi et al., 2001

WT = wild type (normal)

dLHYm = vT TOCn4 - vLCLHYm

dt kT + TOCn4 k1 + LHYm

Building kinetic models: Matthew Turner, James Locke

• O.D.E. models• Global parameter search

KPF now measuring

• Combine genes with similar function

CCA1 and LHY genesLHY/CCA1 TOC1

1-Loop

Overview of the plant clock model

• Central LHY/CCA1 – TOC1 loop Alabadi et al. 2001 (expt)

• Evening TOC1 – Y (GI) Locke et al. 2005 (model)

• Morning PRR9/PRR7 - LHY/CCA1 loop Farré et al. 2005, Salome et al. 2005 (expts). 3-loop and 4-loop models, Locke et al. and Zeilinger et al. 2006

• Similar genes combined for parsimony• Several other processes/components to locate in model

X

LHY/ CCA1 TOC1Y (GI)PRR9/

PRR7

Morning Evening

ZTL

Complex clocks allow flexible timing

• Idea - Pittendrigh, Roenneberg, et al.“The necessity to accommodate changing conditions or seasons might explain the complexity of the circadian clock.”

• Physiological observation: changing relative phase

Pharbitis nilonly photoperiod rhythm tracks dusk

• Potentially flexible mechanism(s)Test clock components first

LHY/CCA1

PRR7/PRR9

X

Y (GI) TOC1

3-Loop 72

78

84

90

96

102

108

114

120

72 78 84 90 96

Photoperiod (h)

Peak

Pha

se (h

)

NBmax

CAB RNA

Stomata

Luciferase reporter at work

• Luciferase protein (LUC) produces light in fireflies• Also works in plants that contain the LUC gene• Amount of light determined by Control part of plant gene

• Method widely used to study fly, mouse, fungus clocks

+camera

Control Luciferase gene

LUC reporter plants in LD, more photoperiods

• Distinguish acute light and circadian effects• Models reveal dynamics of regulation

0

2

4

6

8

0 12 24 36 48 60 72

Time (h)

Nor

mal

ised

Exp

ress

ion 18:06

15:09

12:12

09:15

06:18

03:21

CCA1:LUC

0

2

4

6

8

0 12 24 36 48 60 7

Time (h)

2

GI:LUCLD cycle

Phase changes in the single and interlocked models

0

0.5

1

1.5

2

2.5

576 600

Time (hours)

TOC

1 m

RN

A

3

6

9

12

15

18

00.20.40.60.8

11.21.41.6

576 600

CC

A1 m

RN

A

0

0.1

0.2

0.3

0.4

0.5

120 144

TOC

1 m

RN

A

0

0.5

1

1.5

2

2.5

3

120 144

CC

A1 m

RN

A

0

2

4

6

8

10

12

14

16

18

3 6 9 12 15 18

Photoperiod (h)

Tim

e of

pea

k ex

pres

sion

TOC1CCA1Dusk

Locked to dawn

0

2

4

6

8

10

12

14

16

18

3 6 9 12 15 18

Photoperiod

Tim

e of

pea

k ex

pres

sion

Changes with dusk

0 12 24

0 12 24

LHY/CCA1 TOC1

1-Loop

LHY/CCA1

X

Y (GI) TOC1

2-Loop

00.20.40.60.8

11.21.41.6

456 480

CC

A1

mR

NA

Phase changes in the 3-loop model

0

0.5

1

1.5

2

2.5

456 480

Time (hours)

TOC

1 m

RN

A3

6

9

12

15

18

• 3-loop (or 4-loop) circuit allows flexibility in phase setting• Dusk affects evening genes but morning genes track dawn

Effect of light via Y activating TOC1, then evening feedback

TOC1

X

LHY/ CCA1

PRR7/ PRR9 Y (GI)

0 12 24

ZTL

CCA1

0 12 24

TOC10

3

6

9

12

15

18

3 6 9 12 15 18

Photoperiod

Tim

e of

pea

k ex

pres

sion

Photoperiod

Peak phase

TOC1

CCA1

PRR7/9

Summary

• Clock mRNAs show flexibility of phase3-loop circuit + 2 light inputs may offer some advantage

• But only within limits… no dusk-tracking genes ?Evening-expressed clock genes rarely track dusk (eg TOC1)If they do, this appears to be a direct light response (eg GI)Exception: CCA1 rise in short photoperiods…a morning gene!

• Major role for output pathways and/or other regulators in tuning phase and waveform.

• Wild-type data support the 3-loop model structure

Modelling the photoperiod switch

• Proposed circuit cf.molecular data

• External coincidence model

• Rhythm of CO RNA coincides with light in LD > SD.FT

Vegetative

Flowering

Model 4

CO

Clo

ck

Mod

el 2

TOC1 ≈ CO

Simple Clock 2-loop Clock

TOC1

LHY

Y

TOC1

LHY

X

Model 1

Model 1: CO plus light activate FT

Data set 1 Data set 2

SD

LD

CO

FTFT

CO

FT

CO

PCO

• Two data sets from Kay lab papers• Model fits data better than data sets fit each other

Model 2: Circadian regulation of CO

• CO waveform changes between SD and LD

• Clock must delay phase in LD• Two-loop clock model fits best

• CO increases in afternoon, especially in LD

Requires FKF1

• Model 2 fits fkf1 mutant data

2D

WT

fkf1

CO mRNA

SD

LD

FKF1-dependent increase in CO RNA

• Add second aCO gene

• Fit transcriptionamount and duration to match the wild-type COwaveform

• FKF1 gives higher transcription for ~6h in LD• Smaller effect in SD

Summary

• External coincidence model holds• Details have been elaborated

Circadian phase delayed in LD (2-loop clock model required)Morning gate for FT activationEvening activation of CO by FKF1

• Modelling quantifies these effects• All depend upon photoperiod

FT

Vegetative

Flowering

CO

FKF1

Y

TOC1

LHY

X

Data needed!