A systems biology approach to modelling the heart

Denis NobleUniversity of Oxford

WUN Lecture

14th December 2005

NOBLE, D (2002) Science 295, 1678-1682.

The problem

There are around 25,000 genes in the Human Genome Coding for over 100,000 proteins We know the function of only a very small proportion – genes or proteins

Yet most biological functions depend on many genes & proteins interacting

Illustrative calculation

Assume each function depends on 2 genes(absurd, but still instructive)

Total number of possible ‘functions’ would be0.5 x 25,000 x 24,999

= 312,487,500 With more realistic assumptions about # of genes in each

function, the figures are huge : at 100/function (~ 1.5 e302); for all combinations (~ 2 e166713)

10289

1072403 !

We will only read the “Book of Life” by understanding how genes interact in co-operative ways in complex systems

This means we must unravel & understand biological complexity – i.e. complement reduction with integration

Experimentally and in simulation

The solution

The reductionist causal chain

organism organtissue

cellular sub-

cellularpathwaysprotein

gene

I know one approach that will fail, which is to start with genes, make proteins from them and to try to build things bottom-up

Sydney Brenner, 2001

NOBLE, D (2002) Nature Reviews Molecular Cell Biology 3, 460-463.

Unravelling complexityNeed to work in an integrative way at all levels:

organism organtissue

cellular sub-cellularpathwaysprotein

gene

There are feed-downs as well as upward between all these levels

Systems leveltriggers ofcell signalling

Systems levelcontrols ofgene expression

Protein machineryreads genes

Physiological systems & function

Top-down

Middle-out

Bottom-up

Molecular data & mechanisms

systems organs

tissues cells

pathways organelles

Sydney BrennerNobel Prize2001Middle-out!!

Modeling the heart

Working to connect many levels:

Genetic mutationsProtein function

Cellular machineryTissue function

Whole organ : the Virtual Heart

Modeling the heart

Cell models

1960 : analysis of K+ channels in the heart

-25-50-75-100mV

1

2

-1

-2

current

IK1

IK

Replotted from Hall, Hutter & Noble, 1963Journal of Physiology, 166, 225-240

Model Construction 1960

INa

IK1IK

ICl

Purkinje fibre Channels

Noble, 1960Nature 188, 495-497

Nature’s pact with the devil:

1. The good news

The inward rectifier potassium currentiK1 (Kir2.x)

guarantees energy conservation

Nature’s pact with the devil:

2. The bad news

The delayed potassium currentiKr (HERG + MinK)

is highly promiscuous

Model Construction 2000

INa

IClIK1 IK Ito

ICaChannels

I Na/K

I NaCaNa/H Na/HCO3 Cl/OH

Cl/HCO3

Carriers

Ca

pH

ATP

Glucose

Fatty Acids

Amino Acids

H/Lactate

SubstratesAng II1

2

NO

ßM

Receptors

Example of protein interaction in a cell model Reconstructing the heart’s pacemaker

Sinus rhythm generated by ion channel interaction

ICaL

IKr

Em

If is example of fail-safe ‘redundancy’

Rhythm abolished when interaction prevented

Acceleration of sinus rhythm by adrenaline

If

All 3 protein levels up-regulated

SA node model – ibNa & if

Example of ‘gene knock-out’

Em

If

IbNa

20% 40% 60% 80% 100%

Noble, D., J. C. Denyer, H.F. Brown. & D DiFrancesco (1992). Proc Royal Society B 250: 199-207.

Disease insightModelling arrhythmias

Mutations in various ionic channels can predispose to repolarization failure

This simulation is of a sodium channelmis-sense mutation responsible foridiopathic ventricular fibrillation

Sodium channel molecular structureFour transmembrane domains each with six subunits

Heart sodium channel mutationsgreen : IVF mutations red : long QT mutations

(Chen et al, Nature, 19 March 1998)

Expressed sodium channel kinetics(Chen et al, Nature, 19 March 1998)

Computer model prediction

• Sodium channel missense mutation

• 12 and 18 mV voltage shifts

• Using digital cell ventricular model

12 mVshift18 mVshift

This approach has now been used for a substantial number of gene manipulations in heart cells and can account for genetic susceptibility to fatal cardiac arrhythmia

Including interactions with drugs causing long QT and arrhythmia in clinical trials

Genetic typing to screen out those susceptible to drugs causingQT problems is therefore a foreseeable possibility

Noble D (2002) Unravelling the genetics and mechanisms of cardiac arrhythmia. Proc Natl Acad Sci USA 99, 5755-6

Unravelling genetics of arrhythmia

This example shows

effect of 90% block of IK alone (pure class III)

effect of additional 20% block of Ica,L

Multiple site drugs:QT prolongation without arrhythmia?

• Normal action potential

• Block of IK alone

• Partial block of ICaL

Connecting levels

Incorporation of cellular models into organ models

Multi-level simulations use engineering principles: select lower-level models

using higher-level insights

Noble D (2002) Modelling the heart: from genes to cells to the whole organ. Science 295, 1678-1682

Physiome Sciences

Peter Hunter – the Aucklandmodel ventricle

Launch of ThePhysiome ProjectOf IUPShttp://www.physiome.org.nz

movies

Spread of excitation wave in whole ventricle model

Model – Smith et al Experiment – Nash et al

Spread of excitation wave in whole ventricle model

Model – Smith et al Experiment – Nash et al

Simulation of heart failure EADs leading toTorsades de Pointes arrhythmia

Rai Winslow

Impact-induced arrhythmia Li, Kohl & Trayanova, 2004

Bi-domain modelling with full ionic cell modelsincluding stretch-activated channels

Breakdown of re-entrant arrhythmia into fibrillationSimulation – Panfilov,

Experiment – Witkowsky

Ischaemia ProjectNic Smith – coronary circulation model

INaCa

0 100 200 300

Time (seconds)

20

10

0

[Na+]i

(mM)

Voltage

[Na+ ]i

[Ca2+]i( M)

0

-20

-40

0

-100

1.0

0

INaCa

(nA)

Voltage(mV)

[Ca2+ ]i

Ca oscillator activated as Ca oscillator activated as [Na][Na]ii rises rises

Human cell model

TEN TUSSCHER, NOBLE, D., NOBLE, P. J. & PANFILOV (2004).

A model of the human ventricular myocyte. American Journal of Physiology

286, H1573-1589.

Detailed channel, transporter and SR equations,but computationally very efficient

Human ventricular cell modelClass III induced EAD

[K]o reduced from 5.4 mM to 2.7 mMThen iKr blocked by 90%

mV

seconds

[K]o reduced

iKr blocked by 90%

Re-entrant arrhythmia inhuman model

TEN TUSSCHER & PANFILOV (2004).