You are on page 1of 22

Mathematical Biosciences 334 (2021) 108567

Contents lists available at ScienceDirect

Mathematical Biosciences
journal homepage: www.elsevier.com/locate/mbs

Review
Evolution of mathematical models of cardiomyocyte electrophysiology
Bogdan Amuzescu a,b ,∗, Razvan Airini a,b , Florin Bogdan Epureanu a,b , Stefan A. Mann c ,
Thomas Knott d , Beatrice Mihaela Radu a,b
a
Department of Anatomy, Animal Physiology and Biophysics, Faculty of Biology, University of Bucharest, 91–95 Splaiul
Independentei, Bucharest 050095, Romania
b
Life, Environmental and Earth Sciences Division, Research Institute of the University of Bucharest (ICUB), 91–95 Splaiul
Independentei, Bucharest 050095, Romania
c Cytocentrics Bioscience GmbH, Nattermannallee 1, 50829 Cologne, Germany
d CytoBioScience Inc., 3463 Magic Drive, San Antonio, TX 78229, USA

ARTICLE INFO ABSTRACT


MSC: Advanced computational techniques and mathematical modeling have become more and more important to the
92-08 study of cardiac electrophysiology. In this review, we provide a brief history of the evolution of cardiomyocyte
92C30 electrophysiology models and highlight some of the most important ones that had a major impact on our
92C37
understanding of the electrical activity of the myocardium and associated transmembrane ion fluxes in normal
92C40
and pathological states. We also present the use of these models in the study of various arrhythmogenesis
92C42
92C45
mechanisms, particularly the integration of experimental pharmacology data into advanced humanized models
for in silico proarrhythmogenic risk prediction as an essential component of the Comprehensive in vitro
Keywords:
Proarrhythmia Assay (CiPA) drug safety paradigm.
Cardiac action potential
Cardiomyocyte
Cardiomyocyte electrophysiology
Mathematical models
Hodgkin–Huxley models
Markov models

Contents

1. Introduction ...................................................................................................................................................................................................... 1
2. Purkinje fiber cardiomyocyte models ................................................................................................................................................................... 6
3. Atrial cardiomyocyte models............................................................................................................................................................................... 8
4. SAN cardiomyocyte models ................................................................................................................................................................................ 10
5. AVN cardiomyocyte models ................................................................................................................................................................................ 11
6. Ventricular cardiomyocyte models....................................................................................................................................................................... 11
7. Mathematical models of stem cell-derived cardiomyocytes .................................................................................................................................... 12
8. Stability and bifurcation analysis of cardiomyocyte models ................................................................................................................................... 13
9. Use of cardiomyocyte models in arrhythmogenesis and cardiac safety pharmacology studies.................................................................................... 14
10. A note on computational methods ....................................................................................................................................................................... 15
11. Conclusions and perspectives .............................................................................................................................................................................. 16
Declaration of competing interest ........................................................................................................................................................................ 16
Acknowledgments .............................................................................................................................................................................................. 16
References......................................................................................................................................................................................................... 16

1. Introduction (Table 1), and the complexity of these models increased progres-
Starting with the 1960s an impressive number of mathematical sively since then as more and more experimental data became avail-
models of cardiomyocyte electrophysiology have been developed able. Mathematical modeling of excitatory systems featuring sustained

∗ Corresponding author at: Department of Anatomy, Animal Physiology and Biophysics, Faculty of Biology, University of Bucharest, 91–95 Splaiul Independentei,
Bucharest 050095, Romania.
E-mail address: bogdan@biologie.kappa.ro (B. Amuzescu).

https://doi.org/10.1016/j.mbs.2021.108567
Received 7 October 2020; Received in revised form 10 January 2021; Accepted 4 February 2021
Available online 16 February 2021
0025-5564/© 2021 Elsevier Inc. All rights reserved.
B. Amuzescu, R. Airini, F.B. Epureanu et al. Mathematical Biosciences 334 (2021) 108567

Abbreviations hiPSC-CM Human induced pluripotent stem


cell-derived cardiomyocyte
ACh Acetylcholine
IBM-PC International Business Machines personal
AF Atrial fibrillation
computer
AGOS API Generator for ODE Solution
IP3R Inositol triphosphate receptor
ALGOL Algorithmic language
JSR Junctional sarcoplasmic reticulum
AP Action potential
LC Limit cycle
APD Action potential duration
LRd Luo–Rudy dynamic model
API Application programming interface
LQT long QT syndrome
ATP Adenosine triphosphate
M Midmyocardial ventricular cardiomyocyte
AVN Atrioventricular node
MathML Mathematics Markup language
BR Beeler–Reuter model
MDP Maximum diastolic potential
CaM Calmodulin
MNT McAllister–Noble–Tsien model
CaMKII Calcium/calmodulin-dependent protein ki-
MS-DOS Microsoft Disk Operating System
nase II
NCX Sodium-calcium exchanger
CaN Calcineurin
NSR Network sarcoplasmic reticulum
CellML Cell Markup Language
ODE Ordinary differential equation
CESE Cell Electrophysiology Simulation Environ-
PB Priebe–Beuckelmann model
ment
RDF Resource Description Framework
CICR Calcium-induced calcium release
RP Resting potential
CiPA Comprehensive in vitro Proarrhythmia As-
RV Right ventricle
say
RyR Ryanodine receptor
CMISS Continuum Mechanics, Image analysis, Sig-
nal processing and System identification SA Sinoatrial
CPVT Catecholaminergic polymorphic ventricular SAN Sinoatrial node
tachycardia SERCA Sarcoendoplasmic reticulum calcium pump
CV Conduction velocity SR Sarcoplasmic reticulum
CUDA Compute Unified Device Architecture TdP Torsades-de-pointes
CVODE Variable coefficient ODE solver written in C TEA Tetraethylammonium
DAD Delayed afterdepolarization TP Turbo Pascal (programming language)
DAE Differential algebraic equation TRPM4 Transient receptor potential melastatin-
DLSODE/DLSODA Double-precision Livermore Solver for related 4 ion channel
ODEs/with Automated method switching TTX Tetrodotoxin
for stiff/nonstiff problems XML eXtensible Markup Language
EAD Early afterdepolarization
EASE Electronic Analog and Simulation Equip-
ment
as analytically tractable simplified versions of the highly successful
ECC Excitation–contraction coupling
Hodgkin–Huxley (HH) model [7]. Unfortunately, in this group of mod-
ECG Electrocardiography
els there is no direct link between electrical activity and physiological
EGTA Ethylene glycol-bis(𝛽-aminoethyl
processes important for the generation and propagation of excitation
ether)-N,N,N′ ,N′ -tetraacetic acid
such as the opening and closing of ion channels.
EP Equilibrium point
In 1949, Hodgkin and Katz showed that the amplitude and speed of
FDA Food and Drug Administration action potential upstroke varies with extracellular Na+ concentration,
ginSODA GPU integrator for Systems of ODEs with concluding that the upward phase of the nerve impulse is produced by
Automated stiff/nonstiff switching a specific increase in membrane permeability for Na+ ions. A few years
GPU Graphics processing unit later Hodgkin and Huxley separated the total transmembrane ion flow
hERG Human ether-à-go-go related gene into the Na+ and K+ components and formulated the equations describ-
HH Hodgkin–Huxley model ing how these currents vary depending on the membrane potential and
hESC-CM Human embryonic stem cell-derived car- time. In a series of five articles published in the Journal of Physiology
diomyocyte (London) in 1952, they described a mathematical model based on
previously obtained experimental data using the equivalent electrical
circuit of the nerve cell membrane (Fig. 1A). The main relationship
underlying this mathematical model (the current equation):
oscillations similar to those involved in pacemaking begins with the 𝑑𝑉 ∑
𝐼𝑚 (𝑡) = 𝐶𝑚 𝑚 + 𝐼𝑖
studies of van der Pol and van der Mark. They published in 1928 𝑑𝑡

the first mathematical description of the heartbeat as a relaxation 𝐼𝑖 = 𝐼𝑁𝑎 + 𝐼𝐾 + 𝐼𝐿 = 𝑔 𝑁𝑎 𝑚3 ℎ(𝑉𝑚 − 𝑉𝑁𝑎 ) + 𝑔 𝐾 𝑛4 (𝑉𝑚 − 𝑉𝐾 )
oscillation [1]. This study underpins the development of a group of
‘‘cell’’ models represented by a minimum number of equations and vari- +𝑔 𝐿 (𝑉𝑚 − 𝑉𝐿 )
ables and described by terms such as excitability, stimulus threshold is a first order non-linear ordinary differential equation linking the
and refractory period. The most widely used models in propagation transmembrane potential 𝑉m to the total transmembrane current sur-
studies were the FitzHugh–Nagumo [2,3], Aliev–Panfilov [4], Roger– face density 𝐼𝑚 (t ), present in all cardiomyocyte electrophysiology mod-
McCulloch [5], and Mitchell–Schaeffer [6] models, all of them derived els. I 𝑖 , I 𝑁𝑎 , I 𝐾 , I 𝐿 represent generic and specific (for Na+ , K+ and

2
B. Amuzescu, R. Airini, F.B. Epureanu et al. Mathematical Biosciences 334 (2021) 108567

Table 1
Brief description of cardiomyocyte electrophysiology mathematical models of first, second and third generation, including their type, species and heart territory, main features,
explained phenomena, parent models, and AP shapes.
Name Type Main features AP shape Explained phenomena Parent
(year) model
First generation models
Monocompartmental
HH-like formulations
Based on microelectrode impalement and voltage-clamp data in multicellular preparations
Built to study the role of one or a few ion current components
Purkinje fibers
Noble 1962 HH-like Introduces delayed and inward rectifier K+ I-V relationship in Purkinje fibers Hodgkin–Huxley
generic current (𝐼K2 and 𝐼K1 ); HH regenerative repolarization 1952
model with modified gating kinetics pacemaking
McAllister–Noble–Tsien HH-like Two 𝐼K delayed rectifier components (𝐼X1 and Notched (spike-and-dome) AP Noble 1962
(MNT) 1975 generic 𝐼X2 ); slow inward (Ca2+ ), transient and effects of current pulses and [Ca2+ ]o
background currents on diastolic depolarization
Ventricular
Beeler–Reuter (BR) 1977 HH-like Slow inward Ca2+ current (𝐼CaL ); fast and slow freq.-APD; repolarization by fast MNT 1975
generic 𝐼Na recovery from inactivation (𝑚3 hj gating voltage-clamp during AP plateau;
kinetics) sustained/damped oscillations
Drouhard–Roberge 1987 HH-like Improved Na+ current model with realistic AP phase 0 maximal slope BR 1977
generic activation and inactivation kinetics comparable to experiment; fixed
firing threshold for RP-90 to-70 mV
Sino-atrial
Yanagihara 1980 HH-like Includes 𝐼f , effects of ACh, epinephrine, steady Slow inward (Ca2+ ) current plays a BR 1977
rabbit external current on pacemaking major role and K+ current almost no
role in sino-atrial pacemaking
Irisawa–Noma 1982 HH-like Includes novel experimental data for slow Simulates pacemaking well, but exp. Yanagihara 1980
rabbit inward Ca2+ current (𝐼CaL ) activation at I-V and AP fitted only with artificial
potential above-60 mV ‘‘window’’ 𝐼si (in fact 𝐼CaT )
Bristow–Clark 1982 HH-like Modification of MNT1975 model currents Mimicks well experimental data MNT 1975
rabbit (voltage dependence of gating variables) to fit reproduces free pacemaking and
exp. data effects of injected current
Reiner–Antzelevitch 1985 HH-like 𝐼K2 replacement by 𝐼f pacemaker current-less Reproduces pacemaker Bristow–Clark
rabbit important for phase 4 depolarization than for phase-resetting and annihilation to 1982
phase reset subthreshold/electrotonic stimuli
Second generation models
Monocompartmental
HH-like formulations
Based on patch-clamp/single-channel data from isolated cardiomyocytes
Include Ca2+ dynamics, some ion pumps and exchangers
Ventricular

Luo–Rudy I 1991 HH-like 6 dynamic/background currents; 𝐼Na mh kinetics Wenckebach BR 1977


generic from Ebihara–Johnson; [K+ ]o dependence of 𝐺K patterns/alternans/supernormal
and 𝐺K1 excitability (chaotic activity) at
normal/low [K+ ]o
Atrial
Rasmusson 1990 HH-like 3 dynamic+6 background currents; Role of Ca2+ buffers in the positive Hilgemann–
bullfrog internal & cleft ion concentrations; Ca2+ force–frequency relationship; NCX & Noble
dynamics without SR Na/K pump - small role in repol. 1987

(continued on next page)

leak) ion current surface densities, C 𝑚 transmembrane capacitance use in systems biology approaches and modeling of complex excitable
surface density, 𝑔 𝑁𝑎 , 𝑔 𝐾 and 𝑔 𝐿 the maximal conductance surface systems at almost seven decades since its appearance [11,12].
densities of the respective ion current components, m, h and n the The first models of cardiomyocytes tried to reproduce the experi-
open probabilities of Na+ activation and inactivation and K+ activa- mentally recorded shape of the action potential. Since the potentials
tion gates, respectively, and V 𝑁𝑎 , V 𝐾 , V 𝐿 the reversal potentials of of a nerve fiber act to encode information as quickly as possible,
these ion current components. HH introduced the concept of voltage- they last only a few ms. The action potential of a human ventricular
dependent activation and inactivation gates and developed kinetic cardiomyocyte may last 400 ms. This extended duration is required
differential equations describing the time and potential dependence of for excitation-contraction coupling (ECC), a series of molecular events
these gates [8]. Although the authors initially regarded their work as a initiated by depolarization wave progression along the t tubule system,
partial failure to uncover the specific molecular mechanisms underlying triggering diadic Ca2+ -induced Ca2+ release necessary for actin–myosin
nerve excitability [9,10], the HH model proved highly accurate in interaction [16]. FitzHugh (1960) showed that within the HH for-
quantitatively reproducing a wealth of experimental data. It was also malism repolarization can be delayed by reducing the amplitude and
insightful in terms of revealing structural details of transmembrane ion speed of K+ current activation and the rate of Na+ current inacti-
transport systems, correctly predicting the four-subunit stoichiometry vation [17], consistent with the experimental results of Tasaki and
of voltage-dependent Na+ and K+ ion channels, as well as easy to use, Hagiwara on squid giant axons internally perfused with tetraethylam-
robust and computationally effective, a fact that explains its widespread monium (TEA) [18]; these changes slowed repolarization and created a

3
B. Amuzescu, R. Airini, F.B. Epureanu et al. Mathematical Biosciences 334 (2021) 108567

Table 1 (continued).
Name Type Main features AP shape Explained phenomena Parent
(year) model
sino-atrial
Rasmusson HH-like Similar to Rasmusson 1990 atrial cell model; Pacemaking cannot be attributed to a Hilgemann–
1990 bullfrog includes a diffusion-limited extracellular single current; it is due to lack of 𝐼K Noble
space (cleft) background+effect of 𝐼Ca , 𝐼K , 𝐼b 1987
Third generation models
Multicompartmental, including ion diffusion and buffering, release and reuptake from stores, exchangers, pumps, modulatory and regulatory effects (e.g.
adrenergic/cholinergic)
HH-like and/or Markov formulations
Built to study complex effects of the interplay between structural/functional features
Often include complex experimental datasets, ion microfluorimetry, pharmacological effects
Trend towards human/humanized models
Purkinje fibers
DiFrancesco– HH-like Replacement of hyperpolarization-inhibited [Na+ ]i changes by [Na+ ]o variation MNT 1975
Noble generic 𝐼K2 with 𝐼f ; NCX and Na/K pump; SR or Na/K pump block; effect of
1985 release/ uptake stores current pulses during phase 2/4
Aslanidi 2009 HH-like 𝐼NaL , 𝐼CaL , 𝐼CaT , 𝐼to , 𝐼Kr , 𝐼Ks , 𝐼K1 densities AP shape and rate dependence; too Hund–Rudy
dog and kinetics from experimental data fast or too slow conduction through 2004
validated by RP, 𝑉peak ,dV /d 𝑡max of AP P-V junction is not safe Benson 2008
Stewart 2009 HH-like Inclusion of 𝐼f and 𝐼sus and modification of Reproduces AP shape, overdrive Ten Tusscher–
human 𝐼K1 and 𝐼to in ten Tusscher–Panfilov 2004 suppression, effects of ion current Panfilov
model density changes on AP 2004
Atrial
Hilgemann– HH-like Includes exchangers and pumps, Ca2+ buffers, Reproduces [Ca2+ ]I and [Ca2+ ]o DiFrancesco–
Noble rabbit SR Ca2+ release & uptake, an electrogenic transients, Ca2+ dynamics during Noble
1987 NCX steady/non-steady stimulation 1985
Earm–Noble HH-like Explores Ca2+ -activated SR Ca2+ release and Reproduces nonlinear relation Hilgemann–
1990 rabbit NCX inflow during late plateau; uses 𝐼to between Ca2+ current and Ca2+ Noble
inactivation release, effect of Fura-2 buffering 1987
Lindblad 1996 HH-like Ion currents based on whole cell Provides accurate simulations of Demir 1994
rabbit voltage-clamp data; Na+ , K+ , Ca2+ dynamics, whole-cell voltage-clamp data; fits
SR release/uptake APs at 35 ◦ C 0.2–3.0 Hz stim. rate
Courtemanche HH-like Specific formulations of 𝐼K , 𝐼Na , 𝐼Ca based on Reproduces AP shape, APD Luo–Rudy 1994
1998 human human atrial myocytes data, pumps and rate-dependent adaptation, 𝐼CaL and
exchangers NCX inhibition, 𝐼to variations
Nygren 1998 HH-like Based on averaged voltage-clamp data Proves that the role of 𝐼sus in AP can Lindblad 1996
human recorded from isolated single be modulated by 𝐼CaL , 𝐼sus , 𝐼Kr ;
myocytes;bulk&cleft external space 𝐼sus changes in diseases like AF
Ramirez 2000 HH-like Includes 𝐼K , 𝐼Na , 𝐼Ca , 𝐼Cl based on Realistic APD rate dependency;
dog measurements in canine atrial myocytes; central role of 𝐼CaL inactivation in
junctional & network SR rate adaptation; AF remodeling
Grandi 2011 HH&Markov Redefined current densities (lower 𝐼K1 , 𝐼NCX , Chronic AF model with shorter AP, Grandi–Bers
human 𝐼NaK , removed 𝐼toslow , higher 𝐼tofast , inserted reduced APD shortening at high rate; 2009
𝐼Kur ); altered SERCA 𝐼Kur block -> adrenergic EAD
Davies 2014 HH&Markov 𝐼CaL , 𝐼tofast , 𝐼Kur , 𝐼Kss , 𝐼k1 from atrial myo- Successfully reproduces murine atrial Bondarenko–
mouse cytes experimental data; AP, RP, 𝑉peak ,dV /d 𝑡max , APD and Rasmusson
fibroblast–cardiomyocyte coupling 2D model rate dependence 2004

(continued on next page)

depolarization plateau. FitzHugh’s approach main flaw was the energy- conductance surface densities, driving forces, and open probabilities of
intensive way of generating a plateau; the ion conductances were so various independent activation and inactivation gates, evolving under
large that, during each action potential, the Na+ and K+ ion gradients kinetic equations driven by voltage-dependent rates. Despite their sim-
would have to degrade at a rate of at least one order of magnitude plicity, these early models fit surprisingly well action potentials (APs)
larger than the real one. In contrast to this predicted behavior, Weid- and voltage-clamp currents experimentally recorded via microelectrode
mann’s experiments on cardiomyocytes in the 1950s [19] showed that impalement in Purkinje fibers, ventricular or atrial cardiomyocytes.
transmembrane conductance is very low during the action potential Mathematical models developed after 1989 are more complex (Figs. 2
plateau. Later it was discovered that this decrease in conductance was
and 3). They include additional sets of differential equations for de-
due to the special features of the inward rectifying K+ current (𝐼K1 ).
scribing the kinetics of ion pumps and exchangers, as well as calcium
Ronald Wilders divided the period from the first cardiomyocyte
uptake and release processes in subcellular compartments [20]. Some
model developed by Denis Noble in 1962 until 2007 into two stages:
models that emerged after the 1990s, particularly those developed by
early electrophysiology models of cardiomyocytes, starting in the early
the group of Prof. Yoram Rudy, include more detailed ion current
‘60s and lasting until 1988, and advanced or detailed models that
emerged after 1989 [20]. The early models include: Noble, 1962 [21], kinetics based on single-channel patch-clamp recordings.
McAllister–Noble–Tsien [22], Beeler–Reuter 1977 [23], DiFrancesco– The tremendous success of HH-like models relies on a powerful
Noble [24] and Hilgemann–Noble [25] models. These are based on combination of fundamental simplicity in formulation and computa-
mathematical formulations of specific ion current components similar tional effectiveness, explaining why these models are still preferred
to those of the HH model, including a few variables such as maximal and dominate cardiomyocyte and neuronal electrophysiology modeling

4
B. Amuzescu, R. Airini, F.B. Epureanu et al. Mathematical Biosciences 334 (2021) 108567

Table 1 (continued).
Name Type Main features AP shape Explained phenomena Parent
(year) model
sino-atrial
Noble 1989 HH-like Based on patch-clamp exps. on isolated cells Features pacemaking, but with 𝐼K DiFrancesco–
rabbit (Denyer&Brown 1990) scaled up 2.7-fold; 𝐼ε−NaK and 𝐼bNa Noble
new equations for 𝐼f and 𝐼K probably overestimated 1985
Wilders 1991 HH-like Current equations based on published Reproduces spontaneous pacemaking, DiFrancesco–
generic single-cell voltage-clamp data; includes 𝐼CaT AP parameters, voltage-clamp and Noble
I-V data 1985
Demir 1994 HH-like Two 𝐼Na recovery components, new 𝐼f Fits well voltage-clamp&AP data; DiFrancesco–
rabbit equations; a diffusion-limited external space assesses role of 𝐼CaT , 𝐼b , pumps, Noble
(cleft) exchangers on pacemaker rate 1985
Dokos 1996 HH-like Dynamic modulation of 𝐼f and 𝐼CaL by ACh; Strong ACh-induced Wilders 1991
rabbit improved 𝐼K(ACh) , second-order kinetics of hyperpolarization with
ACh action phase-dependent slowed-down
pacemaking
Zhang 2000 HH-like New formulations of several ion current Reproduces center/periphery AP
rabbit components; SA node center/periphery shape & pacemaking, effects of 𝐼Na ,
current differences 𝐼CaL , 𝐼CaT , 𝐼to , 𝐼Kr , 𝐼Ks , 𝐼f blockers
Sarai 2003 HH-like Includes contraction, a sustained inward Reproduces effects of [Ca2+ ]o and
(Kyoto model: rabbit & Na/K current-𝐼st ; current densities from [K+ ]o , 𝐼CaL and 𝐼st changes on AP
Matsuoka guinea pig experimental data and pacemaking rate
2003)
Lovell 2004 HH&Markov Spatial gradient model of SA node AP Reproduces a wide range of
rabbit heterogeneity;7 time-dependent currents: experimental AP waveforms from
3/4-states Markov models central & peripheral SA node
Kurata 2002 HH-like Subsarcolemmal diffusion space, new 𝐼st , 𝐼Kr Simulates 𝐼CaL , 𝐼Kr , 𝐼st voltage-clamp
rabbit activation kinetics, 𝐼CaL Ca2+ and data, AP and currents waveforms
voltage-dep. inactivation during AP-clamp
Kurata 2008 HH-like Model for central & peripheral SA node cells; Reproduces bifurcation behavior for Zhang 2000
rabbit includes internal Ca2+ dynamics, 𝐼CaL , 𝐼Kr , and steady external
vagal/adrenergic effects current, 𝐼Na peripheral pacemaking
Atrio-ventricular nodal (AVN)
Inada–Boyett HH-like Models for atrio-nodal, nodal, nodal-His cells Atrio-ventricular nodal reentry after Kurata 2002
2009 rabbit with realistic AP shapes and refractoriness premature beat; AVN block during Zhang 2000
atrial fibrillation (AF)
Ventricular
Luo–Rudy HH-like 𝐼CaL , 𝐼NaCa , 𝐼NaK ,SR Ca2+ release/ uptake, 𝐼CaL V -dependent inactivation,AP& Luo–Rudy 1991
1994 guinea pig network&junctional NSR & JSR, currents, phase 2/3EAD, DAD,
(Luo–Rudy SR&myoplasm Ca2+ buffering post-extrasystolic potentiation
dynamic)
Shaw 1995 HH-like Simulation of unidirectional propagation in a Exploration of time and voltage Luo–Rudy 1994
guinea pig 1D cardiac fiber using the Luo–Rudy 1994 vulnerable window for unidirectional
model block/reentry arrhythmias
Zeng 1995 HH-like Uses the Luo–Rudy 1994 model with Mechanisms of EAD and DAD: 𝐼CaL Luo–Rudy 1994
guinea pig modified CICR for EAD&DAD; recovery&reactivation, no matter
+ BayK8644/Cs+ /isoproterenol what current induces EAD
Jafri 1998 HH&Markov Detailed model of Ca2+ regulation: includes Simulates well Ca2+ transients during Luo–Rudy 1994
guinea pig subspace, Keizer-Levine RyR & AP, with higher peak [Ca2+ ] in
mode-switching 𝐼CaL model subspace vs. bulk myoplasm

(continued on next page)

at more that 7 decades since their introduction. Their main com- closed (C), inactivated (I), or blocked (B) (Fig. 1B). In addition, model-
petitors are Markov gating models, developed since the emergence dependent fitting algorithms of idealized single-channel recordings in
of the patch-clamp technique and the first single-channel recordings stationary or non-stationary conditions using hidden Markov modeling
of nicotinic acetylcholine receptors of skeletal muscle end-plates per- and maximum log-likelihood methods allow comprehensive analysis
formed by Bert Sakmann and Erwin Neher in the late 1970s [26,27]. of entire sequences of events and search of optimal models [14,30].
Analysis of single-channel recordings, developed with the essential Markov models and state diagrams are more accurate for simulating
complex ion channel gating behaviors such as fast and slow voltage-
contribution of Sir David Colquhoun, is comprised of a series of steps
dependent inactivation of Na+ and Ca2+ channels, the late bursting
including preprocessing, idealization, plotting dwell-time and all-points
mode and late I Na [31], Ca2+ -dependent inactivation of I CaL [32], mode
histograms, model-dependent fitting, etc. [28,29]. Under the assump-
switching and voltage hysteresis of I f and its antiarrhythmic effects on
tions of Markov chain of random independent events, it can be shown pacemaking [33], allosteric effects in activation of ion channel sub-
that the lifetimes of a certain state, corresponding to a particular con- units [34], state-dependent drug binding [35], etc. Despite their high
formation of the ion channel molecule, are exponentially distributed, fidelity in reproducing ion channel state transitions, Markov models
therefore mono- or multiexponential fits of dwell-time histograms of add supplementary differential equations and parameters to the car-
a conductance level obtained from single-channel traces provide es- diomyocyte model, therefore requiring extra computational resources
timates of transition rates between different states such as open (O), and increased computation times, and sometimes not all rates of a

5
B. Amuzescu, R. Airini, F.B. Epureanu et al. Mathematical Biosciences 334 (2021) 108567

Table 1 (continued).
Name Type Main features AP shape Explained phenomena Parent
(year) model
Noble 1998 HH-like Iadic space with increased Ca2+ Reproduces diadic space Ca2+ Noble 1991
guinea pig accumulation, 𝐼Kr , 𝐼Ks , length- and accumulation during AP plateau;
tension-dependent processes study of 𝐼Na and 𝐼NaK inhibitors
Priebe– HH-like 𝐼CaL , 𝐼to , 𝐼Kr , 𝐼Ks , 𝐼K1 based on human data, Cellular electric properties in heart Luo–Rudy 1994
Beuckelmann human modified Ca2+ SR release/uptake and failure:longer ventricular AP&EAD by
(PB) 1998 buffering 2+
lower 𝐼Kr & spont. Ca release
Winslow 1999 HH-like Jafri 1998 model with modified 𝐼to , 𝐼Kr , 𝐼Ks , Reproduces normal & failing AP & Jafri 1998
dog 𝐼K1 , 𝐼CaL , 𝐽up ; heart failure modeled by lower Ca2+ transient; used to estimate
𝐼to , 𝐼K1 , 𝐼NaCa , 𝐽up 𝐼NaCa &𝐽up in failing myocytes
Dumaine 1999 HH-like Endocardial, LV & RV epicardial modified Effects on AP of accelerated 𝐼Na Luo–Rudy 1994
guinea pig Luo–Rudy 1994 model with mutant T1620M inact.-rapid RV repol. explaining ST
Na+ channel elevation in V1-V3 in Brugada
Clancy–Rudy HH&Markov Wild-type and 𝛥KPQ mutant Na+ channel Arrhythmogenic consequences of Luo–Rudy 1994
1999 guinea pig Markov models included in Luo–Rudy 𝛥KPQ mutation: APD prolongation
dynamic model and EADs
Viswanathan HH-like Subepicardial/M/subendocardial models with Pause-induced EADs by 𝐼CaL recovery Luo–Rudy 1994
-Rudy 1999 guinea pig fixed 𝐼Kr & variable 𝐼Ks density; models of & 𝐼NaCa ; M cells: propen- sity for
LQT1,2,3 EADs, higher in LQT1/2/3
Faber–Rudy HH-like Luo–Rudy dynamic model with 2-phase APD shortening with rapid Luo–Rudy 1999
2000 guinea pig re-formulation of Ca2+ dynamics: 𝐼NaK , pacing, by high 𝐼Ks & [Na+ ]i with
𝐼NaCa ,CICR, 𝐼K(Na) ; high [Na+ ]i & [Ca2+ ]i high 𝐼NaCa &𝐼NaK
Viswanathan– HH-like LQT1/2 (reduced 𝐼Ks / 𝐼Kr ) & LQT3 M cells: higher EAD susceptibility in Vishwanathan-
Rudy guinea pig (increased 𝐼Na late) poorly coupled tissue; 𝐼Kr block/late Rudy 1999
2000 subepicardial/M/subendocardial models 𝐼Na increase APD differences
Greenstein HH&Markov 𝐼to fast (Kv4.3) and slow (Kv1.4) human and Effects of 𝐼to density on AP shape Winslow 1999
2000 dog canine Markov models included in Winslow and duration: bimodal relationship
1999 model between 𝐼to fast and APD
Clancy–Rudy HH&Markov Wild-type/mutant (T474I, R56Q, N629D) Arrhythmogenic consequences of Vishwanathan-
2001 guinea pig hERG1 Markov models included in Rudy hERG1 mutations: APD prolongation Rudy 1999
1999 model and EADs
Puglisi–Bers HH-like Modified conductances for 𝐼Na , 𝐼K1 , 𝐼Kr , 𝐼Ks , Reproduces AP shape and Luo–Rudy 1994
2001 rabbit 𝐼Kp , 𝐼CaT , 𝐼NaCa ; 𝐼to kinetics from Winslow voltage-clamp data for normal/failing
1999; new 𝐼Cl(Ca) rabbit ventricular myocyte
Pandit 2001 HH-like Subendocardial/subepicardial rat ventricular Reproduces AP shape, Ca2+ fluxes, Demir 1994
rat cardiomyocyte model; currents form Demir Ca2+ transient in
models subendocardial/subepicardial
myocytes
Bernus 2002 HH-like Reduced version of PB 1998 model for 6-variable model with fixed [Ca2+ ]i PB 1998
human subendocardial/subepicardial/M human reproduces APD and CV restitution
ventricular myocytes properties; 2D spiral waves

(continued on next page)

complex Markov model can be accurately estimated from experimental the ion pumps to restore the transmembrane gradients at the end of
data. the AP. Thus, metaphorically speaking, the paradoxical functioning of
𝐼K1 represents ‘‘Nature’s pact with the Devil’’, as Denis Noble himself
2. Purkinje fiber cardiomyocyte models inspiredly described this phenomenon in a review [40], a process which
allows generating long APs with at least one thousand times less energy
At the beginning of 1960s Denis Noble and Otto Hutter proved waste than by using the classical Na+ and K+ currents of the HH model
the existence of two types of K+ channels in Purkinje cardiac fibers: with modified kinetics but without 𝐼K1 , as in the initial theoretical
the inward rectifier current 𝐼K1 , and the delayed rectifier current 𝐼K approaches of FitzHugh and Noble in 1960. However, the introduction
[36,37]. In 1962 Noble published the first cardiomyocyte electrophysi- of 𝐼K1 rendered repolarization a much more sensitive process, because
ology model. The Noble 1962 model was an adapted version of the HH small current variations can change the delicate balance between out-
model created in order to reproduce the cardiac action potential and ward and inward currents during AP plateau, disturbing or completely
pacemaker behavior. impeding repolarization [41]. Even though the equations of the Noble
For the inward rectifier K+ current 𝐼K1 Noble fitted his own equa- 1962 model were incomplete, by omitting (or rather incorporating in
tions including the dependence of this current on external K+ con- the Na+ current equations) the L-type Ca2+ current (𝐼CaL ), important
centration [38], while for the delayed rectifier K+ current 𝐼K he used during the plateau phase, but still unknown at that time and discovered
slower versions of K+ current equations from the HH model, due to at the end of the ‘60s [42], and by the lack of characterization of the
the fact that a process that lasts a few milliseconds in the nerve late components of the Na+ current, K+ current kinetics was success-
needs hundreds of milliseconds in the heart. For the Na+ current fully reproduced. In addition, the slow decay of 𝐼K allowed the model to
Noble used the HH equations because Na+ current inactivation occurs reproduce the pacemaker behavior, generating transmembrane voltage
similarly in the myocardium [39]. The Noble 1962 model emphasizes traces very similar to the experimentally recorded ones.
the balance imposed by evolution in the development of long APs in The McAllister–Noble–Tsien 1975 model is based on the Noble
cardiomyocytes; by reducing the K+ ions outflow during depolarization, 1960 and Noble 1962 models. This model used for the first time
𝐼K1 allows to some small Na+ and Ca2+ inward currents to maintain experimental data obtained with advanced voltage-clamp protocols to
the long depolarized AP plateau with a smaller ATP consumption by derive rate equations. It introduced the ‘‘slow inward’’ calcium current

6
B. Amuzescu, R. Airini, F.B. Epureanu et al. Mathematical Biosciences 334 (2021) 108567

Table 1 (continued).
Name Type Main features AP shape Explained phenomena Parent
(year) model
Hund–Rudy HH-like Includes subspace compartment, dynamic Reproduces APD adaptation, 𝐼CaT Luo–Rudy 1994
2004 dog effects of CaMKII on several ion currents and increase at shorter cycle length by
fluxes CaMKII phosphorylation
Shannon–Bers HH&Markov Subsarcolemmal & junctional subspace, Reproduces the nonlinear Luo–Rudy 1994
2004 rabbit realistic Ca2+ buffering, SR release and dependence of ECC gain (𝐽rel ∕𝐼CaL )
reuptake and fractional SR release on SR load
Ten Tusscher– HH-like Transmural 𝐼to , 𝐼Kr , 𝐼Ks gradients APD and CV restitution PB 1998
Panfilov human realistic Ca2+ transients and positive staircase spiral waves in 2D model Courtemanche
2004 1998
Iyer–Mazhari– HH&Markov Markov models for 𝐼Na , 𝐼CaL , 𝐼to fast and 𝐼to APD restitution, [Ca2+ ]SR -frequency,
Winslow human slow , 𝐼Kr , 𝐼Ks extrasystolic restitution and
2004 post-extrasystolic potentiation
Bondarenko– HH&Markov Markov models for 𝐼Na , 𝐼CaL , 𝐼Kr ; other Realistic simulation of AP shape from
Rasmusson mouse currents, Ca2+ transients, buffering & apex&septum, voltage-clamp data,
2004 dynamics like in experiments Ca2+ transients
Faber–Rudy HH&Markov Markov models for L-type Ca2+ and Ryr2 Realistic 𝐼CaL (whole-cell & single- Luo–Rudy 1994
2007 guinea pig channels, diadic subspace, realistic Ca2+ channel), APD & 𝐼CaT restitution,
buffering & dynamics variable gain of SR Ca2+ release
Mahajan 2008 HH&Markov Shannon–Bers 2004 model with Markov 𝐼CaL Reproduces 𝐼CaL Ca2+ and Shannon–Bers
rabbit model and [Ca2+ ]i cycling from Shiferaw et voltage-dependent inactivation, 2004
al. 2003 APD&Ca transient alternans &
restitution
Saucerman– HH&Markov Shannon–Bers 2004 model integrating Only for internal Ca2+ Predicts that CaM is highly activated Shannon–Bers
Bers rabbit kinetic models of Ca2+ binding on CaM, dynamics during AP plateau in diadic cleft 2004
2008 CaMKII, CaN &not in bulk myoplasm
Grandi–Bers HH&Markov Subsarcolemmal&diadic subspace, humanized epi&endo models reproduce AP&Ca2+ Shannon–Bers
2009 human ion channel&transporter models, transmural transients, APD adaptation and 2004
gradients restitution
Livshitz–Rudy HH-like Luo–Rudy dynamic&Hund–Rudy models with With biphasic charge-conserving Luo–Rudy1994
2009 guinea pig DAE, removed singularities, charge stimuli models reach steady-state and Hund–Rudy
dog conservation eqn. 1D propagation alternans 2004
O’Hara–Rudy HH-like Includes human ventricular myocyte data for Reproduces AP shape (subendo-, Luo–Rudy 1994
2011 human ion channels& transporters, detailed effects of subepi- and M-type), rate
CaMKII depen-dence of APD,[Na+ ]i ,[Ca2+ ]i ,
EADs
Li–Dutta– HH&Markov O’Hara–Rudy 2011 model including 𝐼Kr Introduces cqInward (integral of O’Hara–Rudy
Colatsky human Markov model and multiple ion channel 𝐼Nalate + 𝐼CaL over AP duration) as 2011
2017 inhibition data for 12 drugs torsadogenic risk predictor

All graphs of simulations for AP shape, generated with OpenCOR, CESE, MatLab® , Myokit, or C++ scripts, have a duration of 3 s (unless otherwise specified) and amplitude range
from −100 to +50 mV; the isoelectric line is marked with a dotted line.
Abbreviations: HH - Hodgkin-Huxley model; MNT - McAllister–Noble–Tsien model; BR - Beeler–Reuter model; PB- Priebe–Beuckelmann model; DAE - differential algebraic equation;
AVN - atrio-ventricular node; SA - sino-atrial (node); AP - action potential; APD - action potential duration; RP- resting potential; EAD - early afterdepolarization; DAD - delayed
afterdepolarization; AF - atrial fibrillation; CV - conduction velocity; LQT - long QT syndrome; ECC - excitation–contraction coupling; SR - sarcoplasmic reticulum; JSR - junctional
sarcoplasmic reticulum; NSR - network sarcoplasmic reticulum; CICR - calcium-induced calcium release; ACh - acetylcholine; CaMKII - calcium-calmodulin-dependent kinase II;
CaM - calmodulin; CaN - calcineurin; hERG - human ether-à-go-go related gene; RyR - ryanodine receptor; NCX - Na+ /Ca2+ exchanger; SERCA - sarcoendoplasmic reticulum Ca2+
pump; RV - right ventricle; M - midmyocardial ventricular cardiomyocyte.

(𝐼si ), based on experimental results of Harald Reuter, who provided intracellular calcium signaling system. At the same time a similar 𝐼f
the first description of the cardiac calcium current [42], and the current contributing to slow diastolic depolarization and pacemaker ac-
voltage-clamp experiments of Noble and Richard Winyu Tsien, who tivity was identified in collagenase-dissociated isolated Purkinje fibers
discovered multiple K+ current components with slow kinetics [43,44] of different species approached via two-microelectrode voltage-clamp
called 𝐼X1 and 𝐼X2 (known today as 𝐼Kr and 𝐼Ks ) [45], along with [49].
the much faster inactivating transient outward current 𝐼qr and various Newer Purkinje cell models included improved biophysical descrip-
background currents [22]. Although slow diastolic depolarization and tion of several ion current components, such as I f and a sustained
pacemaking were explained by activation during depolarization of an (plateau) 4-aminopyridine-sensitive K+ current named I sus , with a volt-
outward current instead of deactivation of an inward current [46], the age dependence similar to I to but non-inactivatable. Thus, Stewart
mathematical description of this current and the subsequent behavior of et al. [50] developed a Purkinje cell model based on the ten Tusscher–
the model were adequate, succeeding to accurately reproduce particu- Panfilov 2006 ventricular model [51] and including experimental data
larly the chronotropic effects of brief current pulses applied externally from human heart failure myocardium [52], succeeding to demonstrate
during slow diastolic depolarization (phase 4) of pacemaker Purkinje the role of I f in pacemaking, to explore all-or-nothing repolarization
cells. The model explained the slow changes of conductance around and overdrive suppression, to evidence I to conductance density differ-
the resting potential as the result of a K+ current with slow gating, 𝐼K2 . ences between human and canine Purkinje fibers and their effect on
The DiFrancesco and Noble 1985 model [24] incorporated the AP, and to assess consequences of heart failure-induced electrical re-
𝐼f current (funny, queer, anomalous rectifier) discovered by Dario modeling [50]. Other models, like Li and Rudy 2011 [53] and Trovato
DiFrancesco [47] to replace the role of I K2 used in the McAllister– et al. 2020 [54], take into account the lack of t tubules in Purkinje cells,
Noble–Tsien model [48]. This model incorporated for the first time the which renders Ca2+ diffusion the dominant process of Ca2+ cycling, and
changes in intracellular and extracellular ion concentrations and the the triple-layer spatial distribution of ryanodine receptors RyR1, RyR2,

7
B. Amuzescu, R. Airini, F.B. Epureanu et al. Mathematical Biosciences 334 (2021) 108567

Fig. 1. Hodgkin–Huxley and Markov models. A. The equivalent electrical circuit of squid giant axon and the simplified set of equations of the Hodgkin–Huxley model [8]; B.
Markov models based on transitions between states with predefined rates; as an example, the state diagram of the Lu–Vandenberg 2001 model for hERG channel gating [13],
with added state-dependent drug block (C - closed states, O - open state, I - inactivated state, B - blocked states); single-channel traces simulated with the Lu–Vandenberg 2001
model with default parameters at + 40 mV (top trace - 1 s total duration, bottom trace - 0.1 s total duration); dwell time histograms of the conducting and non-conducting level
durations for a 10-s single-channel simulation with the above-mentioned model; traces and histograms were obtained with the QuB software package [14,15].

and inositol triphosphate receptors IP3R1 that induces biphasic Ca2+ atrial cardiomyocytes appeared in 1998 [59,60]. Ion currents were
transients and affects Purkinje cell electrophysiology (rate dependence) formulated using HH equations and incorporated for the first time the
and arrhythmic properties. atrial-specific ultrarapid inactivating K+ current component 𝐼Kur . The
model of Nygren et al. [60] also included the persistent plateau current
3. Atrial cardiomyocyte models 𝐼sus , deemed to regulate AP duration [61], and the Courtemanche et al.
1998 model presented formulations for Ca2+ fluxes at SR level: 𝐽SERCA ,
The first (rabbit) atrial cardiomyocyte model was developed in 𝐽leak , 𝐽tr . Despite similarities, these models show differences in AP
1987 by Hilgemann and Noble [25] based on the DiFrancesco–Noble morphology and rate dependence [62].
1985 model. This model pays special attention to Ca2+ dynamics. Later, Ramirez et al. developed the first canine atrial model with
Built starting from experimental data, the model simulates several pro- HH formulations of major ion currents, and including realistic L-type
cesses: Ca2+ ions influx through Ca2+ channels, sarcoplasmic reticulum Ca2+ channel Ca2+ -dependent inactivation kinetics that exerts a tight
(SR) Ca2+ release, Ca2+ - dependent activation of the sodium-calcium control on AP plateau level and duration in this species [63]. The model
exchanger 1 (NCX1) and Ca2+ - dependent inactivation of 𝐼CaL . allows simulation of the effects of atrial tachycardia on AP, such as AP
Three years later Earm and Noble developed a new rabbit atrial cell shortening and loss of adaptation rate.
model [55]. This was the first model based on recordings in isolated Grandi et al. developed a model of human atrial cardiomyocytes in
cardiac cells rather than multicellular cardiac tissue. In the same year 2011 with detailed validation of rate-dependent Ca2+ dynamics based
Rasmusson and co. have developed the only bullfrog atrial model [56]. on new experimental data [64]. This model succeeded to simulate some
This model included the dynamic intracellular Ca2+ changes as ap- electrophysiological consequences of atrial fibrillation and 𝛽-adrenergic
proached in the Hilgemann–Noble 1987 model, although it did not receptors stimulation. The Grandi et al. 2011 model showed that block-
feature a functional SR. A merit of this model is represented by the ac- ing 𝐼Kur increases intracellular Ca2+ and can cause arrhythmogenic
curate description of extracellular Ca2+ depletion and K+ accumulation early afterdepolarizations (EAD) during adrenergic stress, reproducing
effects due to limited diffusion. experimental observations. In 2014 Davies et al. developed a model of
In 1996, Lindblad et al. published a model of rabbit atrial cardiomy- atrial cardiomyocytes in mice to reproduce atrial remodeling and ar-
ocytes [57]. Based on the sinoatrial node (SAN) cardiomyocyte model rhythmogenesis associated with atrial fibrillation, using Markov models
developed by Demir et al. [58], this model describes ion currents using to formulate the main ion channel current equations: 𝐼Na , 𝐼CaL , 𝐼Kr , and
HH-type formulations and has a very complex subcellular structure studied the effects of cardiomyocyte–fibroblast electrical coupling on
with a double SR compartment, nucleus, mitochondria and a cytoso- AP shape, conduction velocity and reentry in a 2D tissue model [65].
lic space that contains Ca2+ buffers. The first two models of human Other recent multicompartment mouse atrial cardiomyocyte models

8
B. Amuzescu, R. Airini, F.B. Epureanu et al. Mathematical Biosciences 334 (2021) 108567

Fig. 2. Phylogenetic tree of evolution of cardiomyocyte electrophysiology models (MNT: McAllister–Noble–Tsien 1975 model; SAN: sino-atrial node).

Fig. 3. Evolution of cardiomyocyte electrophysiology mathematical models complexity, expressed by the number of model variables, parameters and constants.

including realistic t tubule system-SR coupling, Ca2+ dynamics, 𝛽- All three pioneering human atrial cardiomyocyte models (Courte-
adrenergic signaling and sometimes Markov models for certain ion manche 1998 [59], Nygren 1998 [60], and Grandi 2011 [64]) were
currents were developed by Asfaw et al. 2020 [66] and Zhang et al. subsequently developed in series of models incorporating increasing
2020 [67]. levels of structural and functional complexity [68]. The Courtemanche

9
B. Amuzescu, R. Airini, F.B. Epureanu et al. Mathematical Biosciences 334 (2021) 108567

model was modified by the group of Michael Colman by incorporating by ingenious modification of the McAllister–Noble–Tsien 1975 Purkinje
different experimental datasets to reproduce chronic atrial fibrillation fiber model, removing or adjusting the voltage-dependent properties
(AF)-induced remodeling with changes in ion current densities, internal of some current components; their small-scale model, including only
Ca2+ handling and AP shape, resulting in the Colman 2013 model [69], 5 ion currents, accurately reproduced AP shape [85]. A change to
and further by inclusion of realistic t tubule and dyadic architecture the Bristow–Clark model was brought by Reiner and Antzelevitch, by
in the Colman 2017 model [69]. Another contribution of the Colman replacing the hyperpolarization-induced deactivation of the outward
group is incorporation in the Colman 2013 and Courtemanche models K+ current 𝐼K2 with activation of 𝐼f ; the model was subsequently used
of gain-or-loss-of-function mutations in the KCNA5 gene encoding the to study phase resetting and annihilation by subthreshold external
main subunit of K+ channels contributing I Kur to simulate different stimuli occurring via interaction of 𝐼si and 𝐼K [86]. The Irisawa–
types of familial AF [70]. The Nygren 1998 model was developed by Noma 1982 model also represents a refinement of the Yanagihara
Mary Maleckar et al. by improvement in ion current formulations and 1980 model, including new equations for 𝐼si , 𝐼K and 𝐼f [87], but
inclusion of the acetylcholine-activated K+ current I K(ACh) and used to it reproduces experimental I-V plots only by artificially introducing
investigate effects of cardiomyocyte–fibroblast electrical coupling [71], ‘‘voltage window’’ current properties for 𝐼si by shifting in opposite
and subsequently by the group of Jussi Koivumäki by simulation of directions the voltage dependencies of activation and inactivation to
heterogeneity of SR Ca2+ release and evidencing the role of internal overlap over a voltage interval, and at the price of an abnormally large
Na+ accumulation at faster beat rates in adaptation of cardiac electrical background current [88]. All these early SA node electrophysiology
activity [72]. The Grandi 2011 model was expanded by Voigt, Heijman models rely on two-microelectrode voltage-clamp recordings in rabbit
et al. by including a spatial representation of Ca2+ dynamics via 2 μm- SA node small multicellular preparations, comprised of approximately
wide longitudinal and 1 μm-wide transversal divisions of the atrial 100 pacemaker cells, performed in the 1970s by Irisawa and Noma
cardiomyocyte and a stochastic model of ryanodine receptors RyR2 (reviewed by [89]).
Ca2+ release; the model was used to demonstrate the role of RyR2 The Noble & Noble 1984 SA node model, a modification of the
disregulation and sarcoendoplasmic reticulum Ca2+ pump (SERCA2a) DiFrancesco–Noble 1985 Purkinje cell model, achieved the transition
activation in promoting delayed afterdepolarizations (DAD) and trig- to single-cell SA node models. The model was developed separately
gered activity in paroxysmal AF [73]. Extensions of both the Colman for central (maximum diastolic potential - MDP - between −65 and
2013 model and Grandi 2011 model were rendered onto realistic 3-D −60 mV) and peripheral (MDP −75 mV) SA node cells; like its parent
atrial anatomy models and could be successfully used to study cell- model, it incorporates Na+ /Ca2+ exchanger and Na+ /K+ pump cur-
level and reentry mechanisms in AF initiation and maintenance, and rents, and succeeds to reproduce the smaller influence of [K+ ]o on
the effects of electrical remodeling on atrial arrhythmogenesis [68,69, pacemaking rate compared to Purkinje fibers [90]. A model of similar
73–78]. complexity, incorporating some ion exchanger and pump currents but
All these new computational approaches to atrial electrophysiology, not dynamics of internal calcium was developed in 1990 in parallel
particularly larger-scale propagation models, provided new insights with the atrial cell model based on recordings in isolated bullfrog sinus
into the reentry mechanisms of atrial arrhythmias; some of them con- venosus [91].
firmed pioneering studies such as atrial electrical mapping demonstrat- Due to the difficulties in maintaining a normal shape and physiolog-
ing reentry circuits in atrial tachycardia [79], or proposed spiral waves ical properties of isolated SA node cardiomyocytes, the first successful
in sleeves of ectopic atrial tissue at the origin of pulmonary veins as experiments on dissociated rabbit SA pacemaker cells were performed
a mechanism of paroxysmal atrial fibrillation [80]. A comprehensive only in 1990 [92]. These results opened the door for new SA node
approach to these topics can be found in [81]. To date in excess of models based on single-cell data, the first of them being the Noble–
300 studies have applied mathematical atrial cardiomyocyte models at DiFrancesco–Denyer 1989 model [93]. A refined version of this model
different scales to reveal intricate details of atrial electrophysiology and was issued by Wilders et al. in 1991, incorporating single-cell size
arrhythmogenesis, many of them demonstrating the important roles and membrane capacitance data, as well as realistic current surface
of internal Ca2+ dynamics in these phenomena [62]. Future efforts densities and kinetics, including 𝐼CaT , 𝐼NaCa and 𝐼NaK [88].
in this field will focus on multi-scale approaches with tissue struc- Subsequent SA node electrophysiology mathematical models be-
tural and electrical heterogeneity due to remodeling, including effects came progressively more realistic and detailed, incorporating exten-
of autonomic innervation and mechano-electrical feedback, complex sively new experimental data obtained on isolated cellular prepara-
macromolecular interactions and description of signaling pathways and tions. The 1994 model of Demir, Clark, Murphey & Giles features
modulatory effects, as well as exploring the effects of cell-to-cell vari- several subcellular compartments (bulk and cleft extracellular space,
ability, and will heavily rely on enhanced and extended experimental release and uptake sarcoplasmic reticulum - SR - components), intra-
datasets [62,82]. cellular and SR Ca2+ buffers and adequate calcium dynamics, new
formulations for 𝐼f , 𝐼CaT , and a dual kinetics of 𝐼Na recovery from inac-
4. SAN cardiomyocyte models tivation, reproducing satisfactorily instantaneous I-V plots of multiple
ion currents [58]. The model was updated in 1999 to include effects of
The first model of pacemaker activity of the sino-atrial (SA) node sympathetic and parasympathetic activation [94]. The 1993 and 1996
was developed in 1980 by Yanagihara and co. This model included models of Dokos, Celler and Lovell also incorporate realistic details for
only four dynamic currents described by HH-like equations: the slow the acetylcholine-activated K+ current (𝐼K(ACh) ) and reproduce well the
inward Ca2+ current (𝐼si ), a time-dependent delayed rectifier K+ cur- vagal stimulation-triggered SA node phase resetting [95,96]. The same
rent (𝐼K ), a voltage-dependent Na+ current (𝐼Na ), and the ‘‘funny’’ group developed in 2004 a spatial gradient SA tissue model accounting
hyperpolarization-activated current (𝐼f ), together with a time- for AP heterogeneity, with smooth transition of parameters from center
independent ‘‘leak’’ current (𝐼l ) [83]. Despite its simplicity, the model to periphery [97]. The group of Mark Boyett also built an advanced
reproduced fairly well spontaneously generated APs, the effects of rabbit SA node model with variants for central and peripheral cells,
acetylcholine and epinephrine on slow diastolic depolarization (phase including new experiment-based formulations for several ion currents,
4) and SA node beating rate, restitution of pacemaking by a steady notably a 4-aminopyridine-sensitive current including a transient out-
externally applied hyperpolarizing current after its cessation by a ward and a sustained component (𝐼to and 𝐼sus ), as well as intracellular
depolarizing current, as well as 𝐼K inhibition by 5 mM external Ba2+ Ca2+ , Na+ and K+ dynamics [98]. The Kyoto ventricular and SA node
without significant changes in pacemaking, in good agreement with ex- cardiomyocyte model [99] uses simplified 3/4-states Markov formu-
perimental results of the same group [84]. In 1982, David Bristow and lations for different ion currents, an excitation–contraction coupling
John Clark generated another successful SA node mathematical model model, and remarkably only for SA node a ‘‘sustained inward’’ mixed

10
B. Amuzescu, R. Airini, F.B. Epureanu et al. Mathematical Biosciences 334 (2021) 108567

Na+ and K+ current (𝐼st ) with a reversal potential between + 10 and currents were mathematically expressed by HH-like equations. The
+ 30 mV, identified experimentally since 1995 [100], with properties model also includes a basic equation describing intracellular calcium
such as selective permeability to monovalent cations and activation concentration dynamics.
by calcium overlapping those of TRPM4, shown to be present and In 1991, the first two models of guinea-pig ventricular cardiomy-
influence pacemaking in the SA node [101,102]. The features of the ocytes were published: one developed by Ching–Hsing Luo and Yoram
Kyoto SA model, particularly the effects of 𝐼CaL and 𝐼st on pacemaking Rudy [124] and the other by Noble’s group [125], included in the
rate, were further explored by the same group [103]. Kharche et al. HEART software package. The Luo–Rudy I model is an update of
published in 2011 a realistic mouse SAN model based on own experi- the Beeler–Reuter 1977 model with equations adapted from other
mental data in isolated SAN pacemaker cells, including specific kinetics models to describe various ion current components. This model uses
for different I Na (Nav1.5 and Nav1.1) and I CaL (Cav1.1, Cav1.2, Cav1.3) HH-like equations to describe six ion currents: three time-dependent
isoforms [104]. (Na+ , slow inward, time-dependent K+ ) and three background cur-
Another remarkable set of SA node models are those developed by rents (time-independent K+ , plateau, and background). The rapid Na+
Kurata et al.., a mathematics-oriented Japanese group [105,106]. Their current incorporates a slow inactivation process and an adequate max-
models incorporate accurate experiment-based estimates of different imum conductance surface density. The activation (m) parameters and
ion current components present in rabbit SA node cells, using HH- inactivation (h and j) steady-state values are adapted from the Ebihara–
like formulations, including the 𝐼st component of the Kyoto model, Johnson 𝐼Na model based on experimental voltage-clamp data [126],
deemed to play an essential role in pacemaking, as well as detailed while the slow inactivation time constant 𝜏j , as well as the slow
calcium dynamics and a subsarcolemmal subspace occupying 1% of the inward (Ca2+ ) current are from the BR model. The time-dependent K+
cell volume and featuring higher systolic Ca2+ concentrations due to current (𝐼K ) is controlled by a time-dependent activation gate (X ) and a
limited diffusion. A major merit of their approach is the use of stability time-independent inactivation gate (𝑋i ) with inward-rectifying proper-
and bifurcation analysis methods: detection of equilibrium points (EP) ties. There are three properties that differentiate the time-independent
and limit cycles (LC), construction of codimension-1 and 2 bifurcation potassium current (𝐼K1 ) in the Luo–Rudy I model from the same current
diagrams to detect saddle–node and Hopf bifurcation points, and study- in the BR model: the square-root conductance dependence on [K+ ]o and
ing EP stability by computing the eigenvalues of the Jacobian matrix the high selectivity for K+ (both discovered by [127]) and the inactiva-
of the non-linear dynamic system of ordinary differential equations tion gate K1 identified by Kurachi in single-channel experiments [128].
of the model linearized around these EP. Thus, by varying different Since this current is completely inactivated during depolarization, the
parameters such as an externally applied steady current (𝐼bias ) or the model was supplemented with two more currents: a [K+ ]o -insensitive
conductance densities of 𝐼CaL and 𝐼Kr they could identify intervals for plateau current simulating the single-channel properties of the plateau
these parameters with positive real part of eigenvalues that correspond current measured by Yue and Marbán [129] and a background current
to unstable EP and spontaneous pacemaking behavior [106,107]. (𝐼b ). Although derived from single-channel experiments, the conduc-
The group of Stefano Severi also succeeded to produce a series tances, gating kinetics and reversal potentials of these three currents
of comprehensive SAN pacemaker cell models for rabbit [108] and were adjusted using a parameter estimation technique to fit the time-
human [109], and used them to explore effects on pacemaking of independent whole-cell K+ current measured by Sakmann and Trube
certain factors such as I f properties [109,110], changes in external Ca2+ for different [K+ ]o values [130]. Although the Luo–Rudy I model does
concentration [111,112], or mutations in caveolin gene [113]. not include detailed Ca2+ dynamics, it proved to be very realistic.
A subsequent model issued in 1994 by the same authors [131],
5. AVN cardiomyocyte models known as the Luo–Rudy II or Luo–Rudy dynamic (LRd) model, features
major improvements in internal Ca2+ handling, including a detailed
A series of models of the rabbit atrio-nodal, nodal and nodo-Hissian and realistic description of SR Ca2+ release and reuptake as well as
cells of the atrio-ventricular node (AVN) were developed by Inada, myoplasmic and SR Ca2+ buffering. The model was used to evaluate
Hancox, Zhang and Boyett [114]. These models were incorporated into the relative contributions of Na+ /Ca2+ exchanger and non-specific
pseudo-linear multicellular models including SAN, atrial and AVN com- Ca2+ -activated currents to phase-3 early afterdepolarizations (EADs), to
ponents, with two distinct atrio-nodal pathways (fast and slow), used to distinguish them from phase-2 (plateau) EADs occurring by secondary
study AVN reentry arrhythmogenesis and ventricular protection during activation of L-type Ca2+ channels (𝐼CaL ), and to study SR Ca2+ dy-
atrial fibrillation. A large number of subsequent AVN model studies namics during postextrasystolic potentiation [132]. It also represented
explored AP propagation in physiological and pathological conditions the starting point for many other ventricular cardiomyocyte single-cell
and the effect of different ion currents on AV conduction properties: and multicellular models and studies. Thus, Shaw and Rudy used the
Climent et al. 2011 [115], Inada et al. 2013 [116], Li et al. 2014 [117], LRd model to build homogeneous or nonhomogeneous 1D propagation
Ryzhii and Ryzhii 2015 [118], Cheng et al. 2016 [119], Jackowska- models from gap junction-coupled cell models and studied the time
Zduniak and Foryś 2016 [120], Temple et al. 2016 [121], Hulsmans and voltage vulnerable window of unidirectional block resulting from
et al. 2017 [122], Wallman & Sandberg 2018 [123]. premature stimulation [133]. Zeng and Rudy used LRd models with
modified Ca2+ -induced Ca2+ release (CICR) to demonstrate that the L-
6. Ventricular cardiomyocyte models type Ca2+ current (𝐼CaL ) exerts the main depolarizing effect during an
EAD irrespective of its triggering factor, and that Ca2+ release from
In 1977, George W. Beeler Jr. and Harald Reuter published in stores does not necessarily occur during EADs [134]. Dumaine et al.
the Journal of Physiology the first mathematical model of ventricular studied with modified LRd models including 𝐼to to simulate subendo-
cardiomyocyte electrophysiology. The McAllister–Noble–Tsien model cardial, midmyocardial and subepicardial cardiomyocytes the effects of
was the basis for the extension to ventricular muscle cells developed SCN5 A T1620M point mutation, identified in patients with Brugada
in the Beeler–Reuter (BR) model. Since not all ion current components syndrome, resulting in Na+ currents with shifted voltage dependence of
of a Purkinje fiber are found in ventricular cardiomyocytes, the BR activation and faster inactivation kinetics, demonstrating significantly
model is less complex than the McAllister–Noble–Tsien model. The faster subepicardial AP repolarization that may explain very well the ST
Beeler–Reuter model started from experimental voltage-clamp data that segment elevation in right ventricular ECG leads [135]. Viswanathan
separated four ion currents: the fast excitatory inward Na+ current and Rudy obtained modified versions of the LRd model by varying
𝐼Na , a slow inward current primarily carried by calcium ions - 𝐼S , a 𝐼Kr and 𝐼Ks surface densities to reproduce AP shapes and rate adap-
time-independent inward K+ current - 𝐼K1 , and a voltage- and time- tations specific for subendocardial, midmyocardial and subepicardial
dependent outward current 𝐼X1 , primarily carried by K+ ions. All these ventricular cardiomyocytes [136], and also used these models to study

11
B. Amuzescu, R. Airini, F.B. Epureanu et al. Mathematical Biosciences 334 (2021) 108567

long QT (LQT) 1, 2 and 3 syndromes [137,138]. Clancy and Rudy studying various prorrhythmogenic conditions and arrhythmias, as hu-
also used these modified LRd model versions with Markov models of man experimental data are becoming more easily available. A first
Na+ channels to assess the consequences of mutations 𝛥KPQ associated attempt to generate a human ventricular cardiomyocyte model was
with LQT3 [139] and 1795insD resulting in either LQT3 or Brugada taken by Priebe and Beuckelmann in 1998, by adjusting the LRd model
syndrome [140], or with Markov models of hERG K+ channels to formulations for several ion currents in agreement with parameters
explore congenital LQT2 [141] (reviewed in [142]). Faber and Rudy derived from human ion channels and transporters experimental data,
used the LRd model with modified equations for SR Ca2+ release (𝐼rel ), and an internal Ca2+ dynamics modified to reproduce Ca2+ transients
Na+ /Ca2+ exchanger (𝐼NaCa ) and Na+ -activated K+ current (𝐼K(Na) ) to recorded in isolated human cardiomyocytes from patients with terminal
assess the effects of internal Na+ overload on APD [143]. In 2007 heart failure [158]. The Bernus 2002 model represents a reduced ver-
they built an improved guinea pig ventricular cardiomyocyte model sion of the Priebe–Beuckelmann 1998 model featuring a few dynamic
including realistic Markov models of L-type Ca2+ channels and RyR2 variables and fixed internal calcium concentration, but succeeding to
SR release channels, succeeding to reproduce important experimental reproduce AP duration and conduction velocity (CV) restitution prop-
features such as the fractional SR Ca2+ release and voltage-dependent erties [159]; due to its computational effectiveness it was extensively
variable gain of SR Ca2+ release [32]. Livshitz and Rudy used the up- used in large-scale multicellular modeling. The Iyer–Mazhari–Winslow
dated LRd models [137] and Hund–Rudy model [144] in a differential 2004 subepicardial ventricular model includes experimental data from
algebraic system of equations including electrical charge conservation human cardiomyocytes or human ion channels in heterologous expres-
with a charge-conserving current stimulus and removing singularities sion systems, with Markov formulations for the fast voltage-dependent
due to discontinuities in function definitions to assess evolution to Na+ current, L-type Ca2+ current, RyR2, 𝐼to (fast and slow), 𝐼Kr , 𝐼Ks ,
steady-state in isolated and 1D propagation models [145]. and reproduces a variety of phenomena such as spiral wave genera-
Another landmark model based on LRd was the Jafri–Rice–Winslow tion in 2D cardiomyocyte sheets [160]. Ten Tusscher, Panfilov et al.
1998 model [146]; it included a diadic subspace, a Keizer–Levine also developed humanized ventricular cardiomyocyte models for dif-
Markov model for RyR2, a two-tier mode-switching Markov model of ferent layers (subepicardial, subendocardial, and M) using transmural
L-type Ca2+ channel with explicit Ca2+ -dependent inactivation, and conductance density gradients for 𝐼to , 𝐼Kr , 𝐼Ks and ion current for-
constituted the basis for a series of subsequent developments. The mulations adapted from other models, succeeding to reproduce the
Winslow 1999 canine ventricular cardiomyocyte model represents an positive staircase phenomenon for internal Ca2+ transient amplitudes
updated version including modified formulations of 𝐼to , 𝐼Kr , 𝐼Ks , 𝐼K1 , with increasing pacing frequency and reentry arrhythmias such as 2D
𝐼CaL , 𝐽up , used to assess effects of changed current densities in heart
spiral waves [161]. Grandi, Pasqualini and Bers developed a human
failure [147]. A subsequent update was developed by Greenstein et al.
ventricular cardiomyocyte model [31] adapted from the Shannon-Bers
by introduction of Markov models for the human and canine variants
2004 rabbit ventricular model, including transmural gradients of Ca2+
of I to fast (Kv4.3) and I to slow (Kv1.4) [148].
handling proteins, new formulations for 𝐼to fast and slow, and Markov
A series of successful cardiomyocyte models was developed by the
models for some ion channels like RyR2 and a human 𝐼Ks model
group of Donald M. Bers. Thus, the Puglisi-Bers 2001 model represents
from [162]. However, the most successful human cardiomyocyte ven-
an update of the LRd model with modified kinetics and surface con-
tricular models for subendocardial, subepicardial, and midmyocardial
ductance densities for several ion current components, most notably
(M) cells are those developed by O’Hara, Virág, Varró and Rudy in
a HH-like 𝐼to formulation from Winslow et al. [147]; the model was
2011 starting from extensive experimental data from dissociated human
embedded in a user-friendly graphical interface, the LabHEART soft-
ventricular cardiomyocytes, with new formulations and inclusion of
ware [149]. The Shannon–Bers 2004 model includes many components
modulatory effects of CaMKII for many ion currents [163]. Due to
from the LRd model and Puglisi-Bers 2001 update, combined with an
its computational effectiveness resulting from exclusive use of HH-like
original SR Ca2+ release/reuptake [150]. Mahajan et al. completed in
formulations and a variable-step numeric integration algorithm, as well
2008 the Shannon-Bers 2004 model with an original Markov model of
as its ability to accurately reproduce proarrhythmogenic conditions
the L-type Ca2+ channel and improved internal Ca2+ dynamics [151].
such as EADs, the O’Hara–Rudy 2011 model became a new standard
Saucerman and Bers issued in 2008 another update of the Shannon-
Bers 2004 model by including kinetic models of Ca2+ binding on and reference model for in silico cardiac safety drug testing.
calmodulin (CaM), calcium/calmodulin-dependent protein kinase II
(CaMKII), and the calcium/calmodulin-dependent protein phosphatase 7. Mathematical models of stem cell-derived cardiomyocytes
calcineurin (CaN) [152].
A first mouse ventricular cardiomyocyte model, with distinct ver- Human embryonic stem cell-derived cardiomyocytes (hESC-CMs)
sions for apical and septal cardiomyocytes, based on own experi- and human induced pluripotent stem cell-derived cardiomyocytes
mental data and including detailed subcellular compartments, Ca2+ (hiPSC-CMs) have many applications in disease modeling, cell ther-
buffers and internal dynamics, was built in 2004 by Bondarenko, apy, drug screening and personalized medicine [164]. Differentiation
Szigeti, Bett, Kim and Rasmusson [153]. Another mouse ventricular of hiPSCs into cardiomyocytes is an important method for in vitro
cardiomyocyte model, developed by Morotti et al. in 2014 based on modeling of cardiac diseases such as inherited cardiomyopathies and
the Soltis and Saucerman 2010 rabbit ventricular model [154], suc- arrhythmias [165]; some examples include hypertrophic cardiomy-
ceeded to reproduce the internal Na+ -induced Ca2+ overload with opathies [166], dilated cardiomyopathy [167], Barth syndrome [168],
subsequent CaMKII activation, creating an arrhythmogenic positive or long QT (LQT) syndromes [169,170].
feedback loop [155]. Negroni et al. 2015 produced a rabbit ventricu- Computational models can be used to interpret experimental find-
lar cardiomyocyte electrophysiology and contractility model including ings in iPSC-CMs, provide mechanistic insights, and translate these
electrophysiology components from the Soltis and Saucerman 2010 findings to adult cardiomyocyte electrophysiology. In 2013 Paci M.
model as well as 𝛽-adrenergic effects [156]. Moreno et al. [157] de- et al. published the first models of ventricular and atrial-like hiPSC-
veloped a Markov model for the differential action of ranolazine on CMs [171]. These models were developed by combining different hu-
peak and late I Na , separately for wild-type and 𝛥KPQ inactivation- man or guinea pig HH-like formulations for different ion current com-
removed mutant channels, and further explored its pharmacological ponents with realistic conductance surface densities and Boltzmann
effects by inclusion of the I Na Markov models in several ventricular distributions of voltage-dependent activation and inactivation with
cardiomyocyte models [157]. parameters adjusted according to experimental data from hiPSC-CM
There is a trend in cardiomyocyte electrophysiology modeling to obtained by the group of Craig January [172]. For the ventricular
switch from animal models of different species (guinea pig, dog, rab- model, they also studied the effects of replacement of iPSC-CM-adapted
bit, rat or mouse) to human models, deemed to be more useful in ion current equations with human adult ventricular HH-like equations

12
B. Amuzescu, R. Airini, F.B. Epureanu et al. Mathematical Biosciences 334 (2021) 108567

of the O’Hara–Rudy 2011 model on AP shape, excitability and spon- chaotic systems, stochastic processes and Markov chains of independent
taneous pacemaking, and reproduced by simulation the effects on AP random events, a.s.o. A landmark was set in 1960 by Richard FitzHugh
of some well-known ion current blockers such as tetrodotoxin (TTX), via successful use of an analog electronic computer to generate phase
E4031 and nifedipine, in good agreement with results of patch-clamp space portraits of the complete Hodgkin–Huxley nerve excitability
experiments on hiPSC-CM [172,173]. The in silico tests of pharmaco- model and of reduced HH models, to plot trajectories, to find equi-
logical effects of ion channel blockers were continued in a subsequent librium points at intersection of nullclines of model variables such as
study [174]. These encouraging results led to the idea of adjusting or transmembrane voltage (V ) and open probability of Na+ current acti-
‘‘tailoring’’ default parameters of a hiPSC-CM model according to data vation gate (m) and assess their stability in different conditions [17].
obtained in an individual patch-clamp experiment on hiPSC-CM to get This achievement led to the development of the FitzHugh–Nagumo
a better prediction of pharmacological effects on AP duration [175]. model [2,3], a reduced analytically tractable version of the HH model
Another interesting development is automated parameter optimization featuring essentially the same dynamic behaviors with equilibrium
of a hiPSC-CM model using a weighted cost function algorithm to repro- states of monostability, bistability and sustained oscillations (stable
duce the experimentally recorded AP and Ca2+ transient wave [176]. limit cycles) depending on model parameters, and to the understanding
This updated Paci 2018 ventricular hiPSC-CM model was used to gener- of the fact that any dynamic model of an excitable system, no matter
ate by parameter variability in silico models of long QT syndromes LQT1 how complex, can be transformed into an equivalent FitzHugh–Nagumo
and LQT2, produced by loss-of-function mutations in ion channels model with only two variables: the transmembrane voltage v and a slow
passing 𝐼Ks and 𝐼Kr , respectively [177]. Koivumäki et al. developed gating variable w [188].
an improved hiPSC-CM ventricular model by using the membrane ion As more complex and realistic models of cardiomyocyte electro-
current equations of the Paci 2013 ventricular model combined with physiology were generated, they were subjected to detailed stability
the cell geometry and Ca2+ dynamics equations of a mouse embryonic and bifurcation analysis studies. Thus, Teresa R. Chay and Young
ventricular cardiomyocyte model [178]; the enhanced model was used Seek Lee performed automated bifurcation analysis via continuation
to simulate changes associated with Brugada syndrome, LQT syndrome methods with the software package AUTO developed by Prof. E. Doedel
and catecholaminergic polymorphic ventricular tachycardia (CPVT), at Concordia University, Montréal [189,190] of the Beeler–Reuter 1977
concluding that immature Ca2+ handling in hiPSC-CM prevents accu- ventricular cardiomyocyte model, obtaining by varying a single pa-
rate in silico modeling of the arrhythmogenic consequences of these rameter (a steady externally applied current) a complex codimension-1
diseases [179]. Another recent study comparing simulation results bifurcation diagram featuring multiple fold bifurcation points, as well
using the Paci 2018 ventricular hiPSC-CM model with experimental as Hopf bifurcation points generating branches of stable and unstable
extracellular field potential recordings showed that the model failed periodic solutions (limit cycles); near the Hopf points, adequately timed
to accurately reproduce interbeat intervals and durations of repolar- brief stimuli triggered phase resetting from quiescence to stable oscil-
ization, presumably due to the approximate equations for the ‘‘funny’’ lations [191]. Vinet and Roberge performed a similar stability analysis
hyperpolarization-activated current [180]. using the AUTO software of a modified Beeler–Reuter model [192]
A remarkable attempt to produce a ‘‘population’’ hiPSC-CM model in conditions simulating myocardial ischemia (high extracellular K+
was recently performed at University of California Davis by Divya concentration and increased Ca2+ [slow inward] current density under
Kernik et al.: ion current and Ca2+ dynamics equations were adapted stimulatory effect of catecholamines) [193]. Gibb et al. simulated using
from previous models and several model parameters such as conduc- the Luo–Rudy 1991 model the propagation of sustained oscillations
tance densities, gating rates, RyR rates, SR buffering affinities and rates appeared on the plateau phase of the action potential, representing re-
were optimized relative to different experimental datasets, followed by peated early afterdepolarizations (EAD), from one patch of ventricular
generation of populations of models by introducing parameter vari- myocardium to a neighboring one, where they were able to induce
ability according to experimental distributions, thus reproducing fairly repeated action potentials in conditions of increased tissue volume
well the experimental phenotypic variability of AP morphology [181]. resistivity specific for myocardial ischemia or infarction; they were
The Kernik 2019 hiPSC-CM population of models was further used to able to prove using bifurcation diagrams built with the AUTO software
explore and classify arrhythmogenic consequences of 40 point mutants that phase 2 EADs represent periodic oscillatory solutions occurring
of KCNQ1, the gene encoding the main subunit of slow delayed-rectifier near Hopf bifurcation points, and suggested that K+ channels blocking
K+ channels contributing I Ks [182]. drugs may enhance EAD synchronization in ventricular myocardium,
Although a number of studies have pointed out the problems posed exerting proarrhythmic effects [194]. Using the same Luo–Rudy 1991
by the immature morphological and electrical phenotype of iPSC-CM, ventricular cardiomyocyte model, Tran et al. identified a region in
most notably the low level of I K1 current density due to less developed the parameter space of the model where oscillations of a so-called
t tubule system [183–185] and high propensity to spontaneous pace- ‘‘fast subsystem’’ [a simplified model with only three variables: trans-
making by improper SR Ca2+ reuptake and NCX activity [184,186], membrane voltage V and open probabilities of ‘‘slow inward’’ Ca2+
our direct experience with commercial ventricular-enriched stem cell- current activation (d) and inactivation (f ) gates, but with fixed d and
derived cardiomyocyte preparations like Cor.4U® or Pluricyte® CM f time constants] appeared when the real part of two complex conju-
from Ncardia proved that when cultured in optimal conditions, at high gated eigenvalues of the 3 × 3 Jacobian matrix became positive; these
density to achieve spontaneously beating confluent monolayers and oscillations representing EADs increased progressively in amplitude
allowing sufficient time (≥ 5 days since replating) for maturation results until a homoclinic bifurcation occurred when the trajectory reached
in about 40% cardiomyocytes with adequate RP and AP shape suitable the middle branch of saddle points, sending the system on the lower
for pharmacology studies, particularly upon externally applying faint branch of stable node equilibrium points at resting potential [195].
hyperpolarizing steady currents up to 30 pA [173,187]. In a subsequent study, the same authors demonstrated using recur-
rence plots that irregular EAD occurrence follows deterministic chaos
8. Stability and bifurcation analysis of cardiomyocyte models patterns, and EAD synchronization in myocardial tissue can generate
complex arrhythmias such as polymorphic ventricular tachycardia and
Cardiomyocyte electrophysiology models at single-cell level, to- torsades-de-pointes (TdP) [196]. In a similar analysis of the Luo–Rudy
gether with neuronal models, represented during the past half-century a 1991 model with the MATCONT software [197,198] we proved that
fertile field for test and development of advanced applied mathematics EADs occur near the Hopf bifurcation point on the upper branch (on V
concepts and tools such as phase space analysis of non-linear (ordinary) axis) of equilibrium points when oscillations reach a homoclinic orbit
differential equations (ODE) systems, linear stability analysis and bi- intersecting a saddle point on the middle branch; we also assessed
furcation theory of non-linear dynamic systems, theory of deterministic vector field discontinuities at singular points resulting from definition

13
B. Amuzescu, R. Airini, F.B. Epureanu et al. Mathematical Biosciences 334 (2021) 108567

on intervals of some functions included in the model equations, we generation results from a delicate balance between multiple inhibitory
computed the eigenvalues of the Jacobian matrix of the complete 8- effects on outward and inward cardiac ion currents [210]. This finding
variable linearized system and found the region in parameter space prompted a paradigm shift in cardiac safety drug testing from the
leading to sustained oscillations due to positive real part (for vari- current ‘‘hERG-centric’’ to the new ‘‘CiPA (Comprehensive in vitro
able steady externally applied current), and identified the region of Proarrhythmia Assay)’’ paradigm [211]. CiPA proposes a three-step in-
sustained oscillations for a three-dimensional parameter space with depth mechanistic approach based on: 1. study of inhibitory effects of a
dimensions of externally applied current, calcium (slow inward) and drug on multiple human cardiac ion channels expressed in heterologous
time-dependent K+ current conductances [199]. cell lines; 2. use of these specific pharmacological inhibition data
Stability of cardiomyocyte models should not be regarded only as together with an advanced human cardiomyocyte electrophysiology
a purely mathematical property resulting from time evolution of the model to predict proarrhythmogenic effects such as EAD occurrence;
dynamic differential equations system, but it needs to be interpreted in 3. validation of these predictions via experiments on human induced
biological terms as an emergent feature resulting from the complex in- pluripotent stem cell-derived cardiomyocytes. Recently, our group in-
terplay of various membrane current components and ion concentration troduced enhanced methods for patch-clamp recordings on hiPSC-CM
changes in cellular subcompartments. Thus, phenomena like bistability to combine the three CiPA steps in one assay for in vitro detection of
or phase resetting of ventricular cardiomyocytes are essentially a con- the proarrhythmogenic liability of drug candidates [187,212].
sequence of the highly non-linear voltage dependence of multiple ion A number of in silico studies have addressed the second step of
currents including the inward rectifier I K1 and fast delayed rectifier I Kr the CiPA paradigm. Thus, a research group at US FDA (Food and
K+ currents and their role in controlling refractoriness, as proved by ex- Drug Administration) produced a modified version of the O’Hara–
ploring the origin of Wenckebach periodicity and beat-to-beat alternans Rudy 2011 human ventricular cardiomyocyte model by replacing the
of the Luo–Rudy I model [124], or the analysis of APD90 variability standard HH-like formulation for 𝐼Kr (hERG) with a Markov hERG
of electrically coupled or uncoupled Luo–Rudy I models and stabilizing model including a physiological and a pharmacological component, and
effects of I Na inhibitor TTX and Ca2+ buffering with EGTA vs. destabiliz- used experimental inhibition data (half-inhibitory concentrations - IC50
ing effects of I Kr inhibition with L-691,121 demonstrated by Zaniboni and Hill coefficients) for several ion currents (𝐼Na peak and late, 𝐼CaL ,
et al. 2000 [200]. The same study highlights the relationship between 𝐼to , 𝐼K1 , 𝐼Ks ) as well as rates for the pharmacological component of
transmembrane resistance (Rm ) evolution during AP plateau (phase 2) the hERG Markov model obtained by direct model-dependent fitting
and repolarization variability and the importance of Rm monitoring of macroscopic hERG current recordings to assess computationally
along this sensitive period, a technique pioneered by Silvio Weidmann the proarrhythmogenic risk for a panel of 12 compounds [35]. They
since the 1950s [19] and used recently to improve the results of model proposed as proarrhythmogenic risk predictor cqInward, the integral
parameter optimization programs using genetic algorithms [201] by over the duration of a complete pacing cycle (2 s) of two main in-
taking into account the all-or-nothing repolarization window during ward (depolarizing) ion currents, 𝐼CaL and 𝐼Na late. In a subsequent
which dynamic Rm becomes negative [202]. This negative dynamic study [213], the same group proposed an even better risk predictor,
Rm during early repolarization resulting from an imbalance between 𝑄net (integral over the duration of a pacing cycle of six cardiac ion
repolarizing currents and increased inward depolarizing currents is currents: 𝐼CaL + 𝐼Na late + 𝐼Kr + 𝐼Ks + 𝐼K1 + 𝐼to ), demonstrating
one major trigger of system oscillations known as EADs, as proved it is more reliable than many other predictors based on analysis of
recently by Trenor et al. [203,204]. Beyond AP phase 2 EADs generated action potential and intracellular calcium concentration waveforms.
essentially by I CaL reactivation or its disequilibrium with repolarizing 𝑄net seems to accurately reflect the robustness of a cardiomyocyte
currents, another important EAD triggering mechanism is represented model against EAD generation, but an uncertainty quantification anal-
by spontaneous SR Ca2+ releases due to higher Ca2+ uptake rates, ysis showed that it provided ‘‘safe’’ predictions for the 12-compound
as proved recently by Kurata et al. [205]. The same phenomenon, panel only in the range of 1x to 10x 𝐶max (maximal therapeutic free
produced by the chaotic dynamics of SR Ca2+ sparks [206], plays plasma concentration) [214].
an important role in phase 3 EADs, DADs and their arrhythmogenic An alternative human ventricular cardiomyocyte model extensively
effects at both ventricular and atrial level, including AF initiation used for in silico proarrhythmogenic risk testing, for computation of 𝑄net
and maintenance, particularly following heart failure-induced electrical and other predictors, was developed recently by Tomek et al. starting
remodeling [64,68,73,74,157,207]. from the O’Hara–Rudy 2011 model by including an original Markov
model of L-type Ca2+ channels, the Lu–Vandenberg 2001 Markov model
9. Use of cardiomyocyte models in arrhythmogenesis and cardiac of hERG channels [13], and updated models of 𝐼Na , 𝐼Cl(Ca) and 𝐼K1 , fol-
safety pharmacology studies lowed by application of a multiobjective genetic algorithm to generate
populations of models [215]. The model was calibrated and extensively
The problem of EAD origin is not one of pure theoretical inter- validated by APD restitution and rate dependence, torsadogenic risk
est. Since a large number of pharmacological compounds are able prediction for 62 drugs, reproduction of pathological states such as hy-
to trigger EADs and thus to exert proarrhythmogenic effects, cardiac perkaliemia and hypertrophic cardiomyopathy, ECG simulations using
safety drug testing may benefit from applying mathematical models whole-heart models. Both 𝑄net predicting models are incorporated in
and computational tools to explore such effects. It was shown that TdP the Virtual Assay® software [216].
represents the most frequent drug-induced arrhythmia and its occur- A new idea gaining more and more momentum in computational
rence is correlated to prolongation of QT interval on electrocardiogram cardiac safety pharmacology is that individual cardiomyocyte vari-
and block of hERG K+ ion channels that pass 𝐼Kr , the main ion current ability, for example that induced by stochastic I Ks current fluctua-
component producing repolarization (phase 3) of the ventricular action tions [217], may exert significant effects on repolarization, AP shape
potential [208]. It is estimated that 40 to 50% of new pharmaceutical and arrhythmogenesis, and that multivariate regression analysis ap-
compounds in research phases exert hERG inhibitory effects that may plied to mathematical cardiomyocyte models and construction of pop-
result in fatal arrhythmias in humans [209]. Consequently, cardiac ulations of models can boost proarrhythmogenic risk prediction [218].
safety drug testing standards adopted by regulatory agencies worldwide Subsequent studies, particularly those of the Mirams group at Ox-
rely on direct in vitro assessment of hERG inhibitory effects or QT ford, explored the extrinsic and intrinsic sources of variability and
prolongation effects explored in clinical trials, or by in vivo or in vitro introduced uncertainty analysis and quantification methods for im-
methods. However, a key study performed in 2013 showed that hERG proving model-based predictions [219–221]. Further approaches used
blocking effects alone are not sufficient to predict the proarrhythmo- experimental pharmacological effects on AP parameters recorded in
genic (or ‘‘torsadogenic’’) risk of a compound, and propensity to EAD hiPSC-CM combined with multivariate regression and simulations using

14
B. Amuzescu, R. Airini, F.B. Epureanu et al. Mathematical Biosciences 334 (2021) 108567

populations of human adult ventricular cardiomyocytes models gen- numeric computing eventually won the race of cardiac electrophysiol-
erated by random variability in ion conductance surface densities to ogy modeling, although patch-clamp amplifiers used for cardiomyocyte
perform clinically relevant proarrhythmogenic risk predictions [222]. recordings can be considered advanced analog computers.
The population-of-models method, with parameters generated either Numeric integration methods evolved in parallel with the com-
randomly or, better, estimated from experimental distributions, was puting gear used to run simulations of the models. Thus, Hodgkin
successfully used in recent studies for hiPSC-CM models [181], ventric- and Huxley performed numeric integration for their nerve excitability
ular [215,216], atrial [68,78], or Purkinje cell [54] models for drugs model [8] with advanced linear multistep integration methods [10]
proarrhythmogenic risk prediction, as well as for exploring arrhythmo- such as Adams [233,234] or Hartree [235]. Later, Denis Noble used
genic effects of cardiac channelopathies [177,182,223]. a classical Runge–Kutta method with a time step of 100 μs [21]. How-
An interesting approach is to include in cardiomyocyte models ever, as complexity of the models and the number of ODEs increased,
used for proarrhythmogenic risk predictions quantitative data ob- these numeric integration methods became less effective. Therefore,
tained from molecular dynamics simulations. Thus, Ramasubramanian Rush and Larsen proposed in 1978 an alternate numeric integration
& Rudy succeeded to correlate structural conformational changes of algorithm [236] applicable to first order ODEs for gating variables, of
a KCNQ1 + KCNE1 (subunits of slow delayed rectifier K+ channels the general type:
passing 𝐼Ks ) molecular model obtained via Monte-Carlo methods with 𝑑𝑦
experimentally recorded subconductance states [224]. Xu & Rudy = 𝛼𝑦 (1 − 𝑦) − 𝛽𝑦 𝑦
𝑑𝑡
introduced voltage-and-time-dependent conducting state probabilities where 𝛼𝑦 and 𝛽𝑦 are the opening and closing rates of the gate y, gener-
of such models into an O’Hara–Rudy 2011 model to simulate AP shape ally defined by exponential functions of membrane voltage representing
at different pacing frequencies and under adrenergic stimulation [225]. Boltzmann distributions for assemblies of particles with two distinct
Yang et al. used different protonation states of dofetilide and mox- energy levels. If these two rates are assumed to be constant during
ifloxacin in molecular dynamics simulations with a conducting state the integration step 𝛥t, the gating variable y would follow a first-order
model of hERG1 channels (passing 𝐼Kr ) to estimate free energy barriers monoexponential relaxation kinetics towards a steady-state value:
and blocking/unblocking rates for a direct pore block model, and 𝛼𝑦
further included these data into O’Hara–Rudy models of isolated and 𝑦∞ =
𝛼𝑦 + 𝛽𝑦
unidimensional-string cardiomyocytes to assess EAD occurrence and
effects on QT interval [226]. with a time constant:
Although some general principles for validation of proarrhyth- 1
𝜏𝑦 =
mia risk prediction models based on in silico studies have been out- 𝛼𝑦 + 𝛽𝑦
lined [219,227], the problem is far from being completely solved and and the value of y at the time t + 𝛥t could be computed starting from
new developments will certainly occur, possibly including quantitative the previous value at time t with the formula:
assessments based on dynamic systems linear stability analysis and
bifurcation theory methods. 𝑦(𝑡 + 𝛥𝑡) = 𝑦∞ − (𝑦∞ − 𝑦(𝑡)) exp(−𝛥𝑡∕𝜏𝑦 )

The method was successfully tested by Rush and Larsen for the
10. A note on computational methods Hodgkin–Huxley and McAllister–Noble–Tsien models with fixed inte-
gration step of 20 μs and even with adaptive integration step in the
Cardiac electrophysiology models have evolved not only in terms range of 10 μs to 1 ms, according to the change in transmembrane volt-
of the number of variables (describing ion channels and transporters) age dV /dt. Subsequently the Rush–Larsen numeric integration method
they incorporate (Fig. 1), but also in terms of the complexity of the was widely used in simulations of cardiomyocyte electrophysiology
equations through which these variables are expressed, as more and models based on Hodgkin–Huxley-like formulations of ion channel
more experimental data became available. gating kinetics; it is still used in the O’Hara–Rudy 2011 model with
Computational tools have evolved tremendously during the last an adaptive or constant time step [163]. However, for Markov type ion
70 years. Hodgkin and Huxley (1952) had to compute APs generated channel gating models the direct Euler tangent line method [237] is
by their models ‘by hand’, using a Brunsviga mechanical computing preferred with smaller time steps, generally between 1 and 10 μs.
machine, a process that took Andrew Huxley 8 h to generate only 8 ms The first cardiac modeling software, named OXSOFT HEART, ap-
of electrical activity [10]. In 1960–1961 Denis Noble also used the peared in 1984 [55,238]. Starting with the 1990s a lot of software pack-
famous Brunsviga Model 20 mechanical calculator to compute the rapid ages and computational environments like LabHEART (http://www.
upstroke phase of his model, i.e. ∼2 ms of cardiac electrical activity, labheart.org/) [149], CESE (http://cese.sourceforge.net/) [239,240],
and only after long persistence he managed to get access to a Ferranti iCell (http://ssd1.bme.memphis.edu/icell/) [241], OpenCOR (http://
Mercury computer at University College London, with a top speed of www.opencor.ws/) [242], Jsim (http://www.physiome.org/jsim/)
10 kFlops, for his cardiac modeling work [21,228,229]. For these first [243], AGOS (API generator for ODE Solution) (http://www.fisiocomp.
simulations performed on an electronic valve computer, Noble used di- ufjf.br/) [244], Cardiac Electrophysiology Web Lab (https://travis.cs.
rectly machine code scripts on a perforated paper tape; later, he coded ox.ac.uk/FunctionalCuration/index.html) [245], a.s.o., and libraries
his models in the ALGOL programming language before switching to and applications for finite element modeling and 3D visualization like
Turbo Pascal (TP) on an IBM PC running MS-DOS [48]. In the 1950s Virtual Cell (http://vcell.org/run-vcell-software) [246] or CMISS (http:
American researchers were also using digital electronic computers for //www.cmiss.org/) were developed and used to perform simulations
simulations with the Hodgkin–Huxley model [230–232]. However, use of action potentials and ion currents under a variety of relevant
of the centralized computing facilities of that time was very incon- conditions.
venient, and often delays of one week or more occurred between Certainly, these user-friendly tools with attractive and easy to
sending the code and receiving computation results. To speed up this use graphic interfaces boosted access to cardiac electrophysiology
process, Richard FitzHugh successfully used a Berkeley EASE analog computations, offering even to inexperienced users the opportunity
electronic computer for running the Hodgkin–Huxley model, com- to select models and run simulations with different combinations
posed of 40 operational amplifiers, 6 diode functions generators, and of parameters, stimulus protocols and integration steps. However,
5 servo-multipliers, which printed quasiinstantaneously the dynamics many professional researchers prefer compiling and running model
of transmembrane potential and gating variables on an X-Y chart plot- scripts written in C/C++, Python, or developed for scientific and
ter [17]. In spite of the elegance and simplicity of analog computing, mathematical software platforms such as R, MatLab® , Maple™ or

15
B. Amuzescu, R. Airini, F.B. Epureanu et al. Mathematical Biosciences 334 (2021) 108567

Mathematica. For example, MatLab® provides effective numeric in- Acknowledgments


tegration routines like ode45 for nonstiff ODE systems, ode15s and
ode23s for stiff systems, while the ginSODA library based on the This study was funded from Competitiveness Operational
DLSODA and DLSODE algorithms developed by Alan Hindmarsh and Programme 2014–2020 project P_37_675 (contract no. 146/2016),
Linda Petzold at Lawrence Livermore National Laboratory detects Priority Axis 1, Action 1.1.4, co-financed by the European Funds for
automatically the errors and switches between solvers for stiff and Regional Development and Romanian Government funds. The authors
non-stiff ODE systems. Some advanced mathematics softwares allow gratefully acknowledge Prof. Yoram Rudy for insightful comments, as
automated numeric or symbolic computation of the Jacobian ma- well as Cornelia Dragomir and Geanina Haralambie for support.
trix of an ODE system, accelerating speed and effectiveness of ODE
solvers. A similar method is used by the CVODE (Variable coefficient References
ODE written in C) solver [247], implemented in platforms like Open-
COR and Chaste [248]. Many of these platforms use CellML (https: [1] B. van der Pol, J. van der Mark, The heartbeat considered as a relaxation
oscillation and an electrical model of the heart, Lond. Edinburgh Dublin Physiol.
//www.cellml.org/) [249], an open multiplatform standardized format
Mag. J. Sci. 6 (1928) 763–775.
for biological ODE/DAE systems models based on XML (eXtensible [2] R. Fitzhugh, Impulses and physiological states in theoretical models of Nerve
Markup Language) [250] and MathML [251], with RDF/XML (https:// Membrane, Biophys. J. 1 (1961) 445–466, http://dx.doi.org/10.1016/s0006-
www.w3.org/TR/REC-rdf-syntax/) annotations. Several platforms like 3495(61)86902-6.
OpenCOR, Chaste, its component PyCml (https://chaste.cs.ox.ac.uk/ [3] J.S. Nagumo, S. Arimoto, S. Yoshizawa, An active pulse transmission line
simulating nerve axon, Proc. IRE 50 (1962) 2061–2071, http://dx.doi.org/10.
cellml/), Jsim and Myokit (http://myokit.org/) [252] perform auto- 1109/JRPROC.1962.288235.
mated conversion of CellML models into C/C++ or Python™ scripts. [4] R.R. Aliev, A.V. Panfilov, A simple two-variable model of cardiac excitation,
All these flexible, open, communicating and interchangeable formats, Chaos Solitons Fractals 7 (1996) 293–301, http://dx.doi.org/10.1016/0960-
platforms and libraries create a favorable environment for the devel- 0779(95)00089-5.
[5] J.M. Rogers, A.D. McCulloch, A collocation–Galerkin finite element model of
opment of a generous idea in complex systems computational biology,
cardiac action potential propagation, IEEE Trans. Biomed. Eng. 41 (1994)
the Human Physiome Project [48]. 743–757, http://dx.doi.org/10.1109/10.310090.
In addition, Myokit performs automated CellML model conversion [6] C.C. Mitchell, D.G. Schaeffer, A two-current model for the dynamics of cardiac
into MatLab, CUDA and OpenCL scripts, offering parallel computing membrane, Bull. Math. Biol. 65 (2003) 767–793, http://dx.doi.org/10.1016/
facilities for GPU (graphics processing unit) architectures that can be S0092-8240(03)00041-7.
[7] M. Boulakia, M.A. Fernández, J.-F. Gerbeau, N. Zemzemi, A coupled system
exploited to run large-scale cardiac electrophysiology simulations at
of PDEs and ODEs arising in electrocardiograms modelling, Appl. Math. Res.
multicompartmented cellular, tissue or organ level [253–256]. eXpress 2008 (2008) 1–28, http://dx.doi.org/10.1093/amrx/abn002.
[8] A.L. Hodgkin, A.F. Huxley, A quantitative description of membrane current and
11. Conclusions and perspectives its application to conduction and excitation in nerve, J. Physiol. 117 (1952)
500–544, http://dx.doi.org/10.1113/jphysiol.1952.sp004764.
[9] A.L. Hodgkin, Chance and design in electrophysiology: an informal account of
Cardiomyocyte electrophysiology mathematical modeling evolved certain experiments on nerve carried out between 1934 and 1952, J. Physiol.
steadily over 60 years, in parallel with improvement of experimental 263 (1976) 1–21, http://dx.doi.org/10.1113/jphysiol.1976.sp011620.
methods and computational techniques. Looking retrospectively, it is [10] A. Huxley, From overshoot to voltage clamp, Trends Neurosci. 25 (2002)
amazing how the earliest models [21,124] were able to predict a variety 553–558, http://dx.doi.org/10.1016/s0166-2236(02)02280-4.
[11] H. Markram, E. Muller, S. Ramaswamy, M.W. Reimann, M. Abdellah, C.A.
of experimental phenomena with minimal mathematical complexity.
Sanchez, A. Ailamaki, L. Alonso-Nanclares, N. Antille, S. Arsever, G.A. Kahou,
Extrapolating an idea formulated by Wulfram Gerstner for neuron mod- T.K. Berger, A. Bilgili, N. Buncic, A. Chalimourda, G. Chindemi, J.D. Courcol,
eling [257], ‘‘toy models’’ may still be mathematically attractive and F. Delalondre, V. Delattre, S. Druckmann, R. Dumusc, J. Dynes, S. Eilemann, E.
easier to approach with analytical methods, offering profound theoret- Gal, M.E. Gevaert, J.P. Ghobril, A. Gidon, J.W. Graham, A. Gupta, V. Haenel,
ical explanations for essential phenomena like spontaneous pacemaking E. Hay, T. Heinis, J.B. Hernando, M. Hines, L. Kanari, D. Keller, J. Kenyon,
G. Khazen, Y. Kim, J.G. King, Z. Kisvarday, P. Kumbhar, S. Lasserre, J.V.
and arrhythmogenesis mechanisms.
Le Be, B.R. Magalhaes, A. Merchan-Perez, J. Meystre, B.R. Morrice, J. Muller,
Examining this long evolution of cardiomyocyte electrophysiol- A. Munoz-Cespedes, S. Muralidhar, K. Muthurasa, D. Nachbaur, T.H. Newton,
ogy models and their surprising diversity, we can identify several M. Nolte, A. Ovcharenko, J. Palacios, L. Pastor, R. Perin, R. Ranjan, I. Riachi,
trends: progressive shift from models relying on animal cardiomyocyte J.R. Rodriguez, J.L. Riquelme, C. Rossert, K. Sfyrakis, Y. Shi, J.C. Shillcock, G.
experiments to humanized or completely human models, inclusion Silberberg, R. Silva, F. Tauheed, M. Telefont, M. Toledo-Rodriguez, T. Trankler,
W. Van Geit, J.V. Diaz, R. Walker, Y. Wang, S.M. Zaninetta, J. DeFelipe,
of realistic data on internal Ca2+ dynamics and buffering, deriving
S.L. Hill, I. Segev, F. Schurmann, Reconstruction and simulation of neocortical
population-average models or population-of-models using large ex- microcircuitry, Cell 163 (2015) 456–492, http://dx.doi.org/10.1016/j.cell.2015.
perimental datasets comprised of current-clamp, voltage-clamp and 09.029.
Ca2+ microfluorimetry recordings [163,176,181], exploring the pa- [12] C. Meunier, I. Segev, Playing the devil’s advocate: is the Hodgkin–Huxley
rameter space of models by applying experimental distributions of model useful?, Trends Neurosci. 25 (2002) 558–563, http://dx.doi.org/10.
1016/s0166-2236(02)02278-6.
key features like conductance surface densities [175], using advanced [13] Y. Lu, M.P. Mahaut-Smith, A. Varghese, C.L. Huang, P.R. Kemp, J.I. Vanden-
human cardiomyocyte models according to the principles of the CiPA berg, Effects of premature stimulation on HERG K(+) channels, J. Physiol. 537
paradigm [211] to explore in silico proarrhythmogenic effects of drugs (2001) 843–851, http://dx.doi.org/10.1111/j.1469-7793.2001.00843.x.
[35,213]. Moreover, cardiomyocyte models can be used in massively [14] F. Qin, A. Auerbach, F. Sachs, Maximum likelihood estimation of aggregated
Markov processes, Proc. Biol. Sci. 264 (1997) 375–383.
parallel computational approaches to explore mechanisms of proar-
[15] F. Qin, A. Auerbach, F. Sachs, Estimating single-channel kinetic parameters from
rhythmogenic events like EADs, DADs and ventricular arrhythmias idealized patch-clamp data containing missed events, Biophys. J. 70 (1996)
that they trigger [256,258–260]. They also offer a basis for exploring 264–280.
functional consequences of mutations in genes encoding ion channels [16] D.M. Bers, Cardiac excitation-contraction coupling, Nature 415 (2002) 198–205,
and transporters subunits or associated proteins in the study of cardiac http://dx.doi.org/10.1038/415198a.
[17] R. Fitzhugh, Thresholds and plateaus in the Hodgkin–Huxley nerve equations,
channelopathies in combination with experiments on patient-specific J. Gen. Physiol. 43 (1960) 867–896, http://dx.doi.org/10.1085/jgp.43.5.867.
induced pluripotent stem cell-derived cardiomyocytes [177,261–263]. [18] I. Tasaki, S. Hagiwara, Demonstration of two stable potential states in the squid
giant axon under tetraethylammonium chloride, J. Gen. Physiol. 40 (1957) 859,
Declaration of competing interest http://dx.doi.org/10.1085/jgp.40.6.859.
[19] S. Weidmann, Effect of current flow on the membrane potential of cardiac
muscle, J. Physiol. 115 (1951) 227–236, http://dx.doi.org/10.1113/jphysiol.
The authors declare that they have no known competing finan- 1951.sp004667.
cial interests or personal relationships that could have appeared to [20] R. Wilders, Computer modelling of the sinoatrial node, Med. Biol. Eng. Comput.
influence the work reported in this paper. 45 (2007) 189–207, http://dx.doi.org/10.1007/s11517-006-0127-0.

16
B. Amuzescu, R. Airini, F.B. Epureanu et al. Mathematical Biosciences 334 (2021) 108567

[21] D. Noble, A modification of the Hodgkin–Huxley equations applicable to purk- [48] D. Noble, A. Garny, P.J. Noble, How the Hodgkin–Huxley equations inspired
inje fibre action and pace-maker potentials, J. Physiol. 160 (1962) 317–352, the Cardiac Physiome Project, J. Physiol. 590 (2012) 2613–2628, http://dx.doi.
http://dx.doi.org/10.1113/jphysiol.1962.sp006849. org/10.1113/jphysiol.2011.224238.
[22] R.E. McAllister, D. Noble, R.W. Tsien, Reconstruction of the electrical activity [49] G. Callewaert, E. Carmeliet, J. Vereecke, Single cardiac Purkinje cells:
of cardiac Purkinje fibres, J. Physiol. 251 (1975) 1–59, http://dx.doi.org/10. general electrophysiology and voltage-clamp analysis of the pace-maker cur-
1113/jphysiol.1975.sp011080. rent, J. Physiol. 349 (1984) 643–661, http://dx.doi.org/10.1113/jphysiol.1984.
[23] G.W. Beeler, H. Reuter, Reconstruction of the action potential of ventricular sp015179.
myocardial fibres, J. Physiol. 268 (1977) 177–210, http://dx.doi.org/10.1113/ [50] P. Stewart, O.V. Aslanidi, D. Noble, P.J. Noble, M.R. Boyett, H. Zhang,
jphysiol.1977.sp011853. Mathematical models of the electrical action potential of Purkinje fibre cells,
[24] D. DiFrancesco, D. Noble, A model of cardiac electrical activity incorporating Philos. Trans. A Math. Phys. Eng. Sci. 367 (2009) 2225–2255, http://dx.doi.
ionic pumps and concentration changes, Phil. Trans. R. Soc. Lond. B 307 (1985) org/10.1098/rsta.2008.0283.
353–998, http://dx.doi.org/10.1098/rstb.1985.0001. [51] K.H. Ten Tusscher, O. Bernus, R. Hren, A.V. Panfilov, Comparison of electro-
[25] D.W. Hilgemann, D. Noble, Excitation-contraction coupling and extracellular physiological models for human ventricular cells and tissues, Prog. Biophys.
calcium transients in rabbit atrium: reconstruction of basic cellular mechanisms, Mol. Biol. 90 (2006) 326–345, http://dx.doi.org/10.1016/j.pbiomolbio.2005.05.
Proc. R. Soc. B 230 (1987) 163–205, http://dx.doi.org/10.1098/rspb.1987. 015.
0015. [52] W. Han, L. Zhang, G. Schram, S. Nattel, Properties of potassium currents in
[26] O.P. Hamill, A. Marty, E. Neher, B. Sakmann, F.J. Sigworth, Improved patch- Purkinje cells of failing human hearts, Am. J. Physiol. Heart Circ. Physiol. 283
clamp techniques for high-resolution current recording from cells and cell-free (2002) H2495–H2503, http://dx.doi.org/10.1152/ajpheart.00389.2002.
membrane patches, Pflugers Arch. 391 (1981) 85–100. [53] P. Li, Y. Rudy, A model of canine purkinje cell electrophysiology and Ca(2+)
[27] E. Neher, B. Sakmann, Single-channel currents recorded from membrane of cycling: rate dependence, triggered activity, and comparison to ventricular my-
denervated frog muscle fibres, Nature 260 (1976) 799–802. ocytes, Circ. Res. 109 (2011) 71–79, http://dx.doi.org/10.1161/CIRCRESAHA.
[28] D. Colquhoun, A.G. Hawkes, The principles of stochastic interpretation of 111.246512.
Ion-Channel Mechanisms, in: B. Sakmann, E. Neher (Eds.), Single-Channel [54] C. Trovato, E. Passini, N. Nagy, A. Varró, N. Abi-Gerges, S. Severi, B.
Recording, second ed., Plenum Press, New York and London, 1995, pp. Rodriguez, Human purkinje in silico model enables mechanistic investigations
397–482. into automaticity and pro-arrhythmic abnormalities, J. Mol. Cell Cardiol. 142
[29] D. Colquhoun, F.J. Sigworth, Fitting and statistical analysis of single-channel (2020) 24–38, http://dx.doi.org/10.1016/j.yjmcc.2020.04.001.
records, in: B. Sakmann, E. Neher (Eds.), Single-Channel Recording, second ed., [55] Y.E. Earm, D. Noble, A model of the single atrial cell: relation between
Plenum Press, New York and London, 1995, pp. 483–588. calcium current and calcium release, Proc. R. Soc. B 240 (1990) 83–96,
[30] D. Colquhoun, C.J. Hatton, A.G. Hawkes, The quality of maximum likelihood http://dx.doi.org/10.1098/rspb.1990.0028.
estimates of ion channel rate constants, J. Physiol. 547 (2003) 699–728. [56] R.L. Rasmusson, J.W. Clark, W.R. Giles, K. Robinson, R.B. Clark, E.F. Shibata,
[31] E. Grandi, F.S. Pasqualini, D.M. Bers, A novel computational model of the D.L. Campbell, A mathematical model of electrophysiological activity in a
human ventricular action potential and Ca transient, J. Mol. Cell. Cardiol. 48 bullfrog atrial cell, Am. J. Physiol. 259 (1990) H370–H389, http://dx.doi.org/
(2009) 112–121, http://dx.doi.org/10.1016/j.yjmcc.2009.09.019. 10.1152/ajpheart.1990.259.2.H370.
[32] G.M. Faber, J. Silva, L. Livshitz, Y. Rudy, Kinetic properties of the cardiac [57] D.S. Lindblad, C.R. Murphey, J.W. Clark, W.R. Giles, A model of the action
L-type Ca2+ channel and its role in myocyte electrophysiology: a theoretical potential and underlying membrane currents in a rabbit atrial cell, Am. J.
investigation, Biophys. J. 92 (2007) 1522–1543, http://dx.doi.org/10.1529/ Physiol. 271 (1996) H1666–H1696, http://dx.doi.org/10.1152/ajpheart.1996.
biophysj.106.088807. 271.4.H1666.
[33] R. Männikkö, S. Pandey, H.P. Larsson, F. Elinder, Hysteresis in the voltage [58] S.S. Demir, J.W. Clark, C.R. Murphey, W.R. Giles, A mathematical model of
dependence of HCN channels: conversion between two modes affects pacemaker a rabbit sinoatrial node cell, Am. J. Physiol. 266 (1994) C832–C852, http:
properties, J. Gen. Physiol. 125 (2005) 305–326, http://dx.doi.org/10.1085/jgp. //dx.doi.org/10.1152/ajpcell.1994.266.3.C832.
200409130.Epub2005Feb14. [59] M. Courtemanche, R.J. Ramirez, S. Nattel, Ionic mechanisms underlying human
[34] C. Altomare, A. Bucchi, E. Camatini, M. Baruscotti, C. Viscomi, A. Moroni, D. atrial action potential properties: insights from a mathematical model, Am. J.
DiFrancesco, Integrated allosteric model of voltage gating of HCN channels, J. Physiol. 275 (1998) H301–H321, http://dx.doi.org/10.1152/ajpheart.1998.275.
Gen. Physiol. 117 (2001) 519–532, http://dx.doi.org/10.1085/jgp.117.6.519. 1.H301.
[35] Z. Li, S. Dutta, J. Sheng, P.N. Tran, W. Wu, K. Chang, T. Mdluli, D.G. Strauss, T. [60] A. Nygren, C. Fiset, L. Firek, J.W. Clark, D.S. Lindblad, R.B. Clark, W.R. Giles,
Colatsky, Improving the in silico assessment of Proarrhythmia risk by combining Mathematical model of an adult human atrial cell: the role of K+ currents in
hERG (Human Ether-a-go-go-Related Gene) channel-drug binding kinetics and repolarization, Circ. Res. 82 (1998) 63–81, http://dx.doi.org/10.1161/01.res.
multichannel pharmacology, Circ. Arrhythm. Electrophysiol. 10 (2017) 004628, 82.1.63.
http://dx.doi.org/10.1161/circep.116.004628. [61] Z. Wang, B. Fermini, S. Nattel, Sustained depolarization-induced outward
[36] A.E. Hall, O.F. Hutter, D. Noble, Current–voltage relations of Purkinje fibres in current in human atrial myocytes. Evidence for a novel delayed rectifier
sodium-deficient solutions, J. Physiol. 166 (1963) 225–240, http://dx.doi.org/ K+ current similar to Kv1.5 cloned channel currents, Circ. Res. 73 (1993)
10.1113/jphysiol.1963.sp007102. 1061–1076, http://dx.doi.org/10.1161/01.res.73.6.1061.
[37] O.F. Hutter, D. Noble, Rectifying properties of heart muscle, Nature 188 (1960) [62] J. Heijman, P. Erfanian Abdoust, N. Voigt, S. Nattel, D. Dobrev, Computational
495, http://dx.doi.org/10.1038/188495a0. models of atrial cellular electrophysiology and calcium handling, and their
[38] D. Noble, Electrical properties of cardiac muscle attributable to inward going role in atrial fibrillation, J. Physiol. 594 (2016) 537–553, http://dx.doi.org/
(anomalous) rectification, J. Cell. Comp. Physiol. 66 (1965) 127–136, http: 10.1113/jp271404.
//dx.doi.org/10.1002/jcp.1030660520. [63] R.J. Ramirez, S. Nattel, M. Courtemanche, Mathematical analysis of canine
[39] B. Amuzescu, A. Corlan, G. Nistor, Modelarea Matematică a Electrofiziologiei atrial action potentials: rate, regional factors, and electrical remodeling, Am.
Cardiace, Editura Universitatii din Bucuresti, Bucharest, 2012. J. Physiol. Heart Circ. Physiol. 279 (2000) H1767–H1785, http://dx.doi.org/
[40] D. Noble, Modeling the heart, Physiology (Bethesda) 19 (2004) 191–197, 10.1152/ajpheart.2000.279.4.H1767.
http://dx.doi.org/10.1152/physiol.00004.2004. [64] E. Grandi, S.V. Pandit, N. Voigt, A.J. Workman, D. Dobrev, J. Jalife, D.M.
[41] A.L. Muzikant, R.C. Penland, Models for profiling the potential QT prolongation Bers, Human atrial action potential and Ca2+ model: sinus rhythm and chronic
risk of drugs, Curr. Opin. Drug Discov. Dev. 5 (2002) 127–135. atrial fibrillation, Circ. Res. 109 (2011) 1055–1066, http://dx.doi.org/10.1161/
[42] H. Reuter, The dependence of slow inward current in Purkinje fibres on CIRCRESAHA.111.253955.
the extracellular calcium-concentration, J. Physiol. 192 (1967) 479–492, http: [65] L. Davies, J. Jin, W. Shen, H. Tsui, Y. Shi, Y. Wang, Y. Zhang, G. Hao, J.
//dx.doi.org/10.1113/jphysiol.1967.sp008310. Wu, S. Chen, J.A. Fraser, N. Dong, V. Christoffels, U. Ravens, C.L. Huang, H.
[43] D. Noble, R.W. Tsien, The kinetics and rectifier properties of the slow potassium Zhang, E.J. Cartwright, X. Wang, M. Lei, Mkk4 is a negative regulator of the
current in cardiac Purkinje fibres, J. Physiol. 195 (1968) 185–214, http://dx. transforming growth factor beta 1 signaling associated with atrial remodeling
doi.org/10.1113/jphysiol.1968.sp008454. and arrhythmogenesis with age, J. Am. Heart Assoc. 3 (2014) 000340.
[44] D. Noble, R.W. Tsien, Outward membrane currents activated in the plateau [66] T.N. Asfaw, L. Tyan, A.V. Glukhov, V.E. Bondarenko, A compartmentalized
range of potentials in cardiac Purkinje fibres, J. Physiol. 200 (1969) 205–231, mathematical model of mouse atrial myocytes, Am. J. Physiol. Heart Circ.
http://dx.doi.org/10.1113/jphysiol.1969.sp008689. Physiol. 318 (2020) H485–H507, http://dx.doi.org/10.1152/ajpheart.00460.
[45] M.C. Sanguinetti, N.K. Jurkiewicz, Two components of cardiac delayed rectifier 2019.
K+ current. Differential sensitivity to block by class III antiarrhythmic agents, [67] H. Zhang, S. Zhang, W. Wang, K. Wang, W. Shen, A mathematical model of
J. Gen. Physiol. 96 (1990) 195–215, http://dx.doi.org/10.1085/jgp.96.1.195. the mouse atrial myocyte with Inter-Atrial electrophysiological heterogeneity,
[46] D. Noble, Modelling the heart: insights, failures and progress, Bioessays 24 Front. Physiol. 11 (2020) 972, http://dx.doi.org/10.3389/fphys.2020.00972.
(2002) 1155–1163, http://dx.doi.org/10.1002/bies.10186. [68] M. Vagos, I.G.M. van Herck, J. Sundnes, H.J. Arevalo, A.G. Edwards, J.T.
[47] D. DiFrancesco, A new interpretation of the pace-maker current in calf Purkinje Koivumäki, Computational modeling of electrophysiology and pharmacotherapy
fibres, J. Physiol. 314 (1981) 359–376, http://dx.doi.org/10.1113/jphysiol. of atrial fibrillation: Recent advances and future challenges, Front. Physiol. 9
1981.sp013713. (2018) 1221, http://dx.doi.org/10.3389/fphys.2018.01221.

17
B. Amuzescu, R. Airini, F.B. Epureanu et al. Mathematical Biosciences 334 (2021) 108567

[69] M.A. Colman, O.V. Aslanidi, S. Kharche, M.R. Boyett, C. Garratt, J.C. Hancox, [93] D. Noble, D. DiFrancesco, J.C. Denyer, Ionic mechanisms in normal and
H. Zhang, Pro-arrhythmogenic effects of atrial fibrillation-induced electrical abnormal cardiac pacemaker activity, in: J.W. Jacklet (Ed.), Neuronal and
remodelling: insights from the three-dimensional virtual human atria, J. Physiol. Cellular Oscillators, Marcel Dekker, New York, 1989, pp. 59–85.
591 (2013) 4249–4272, http://dx.doi.org/10.1113/jphysiol.2013.254987. [94] S.S. Demir, J.W. Clark, W.R. Giles, Parasympathetic modulation of sinoatrial
[70] M.A. Colman, H. Ni, B. Liang, N. Schmitt, H. Zhang, In silico assessment node pacemaker activity in rabbit heart: a unifying model, Am. J. Physiol. 276
of genetic variation in KCNA5 reveals multiple mechanisms of human atrial (1999) H2221–H2244, http://dx.doi.org/10.1152/ajpheart.1999.276.6.H2221.
arrhythmogenesis, PLoS Comput. Biol. 13 (2017) e1005587, http://dx.doi.org/ [95] S. Dokos, B. Celler, N. Lovell, Ion currents underlying sinoatrial node pacemaker
10.1371/journal.pcbi.1005587. activity: a new single cell mathematical model, J. Theoret. Biol. 181 (1996)
[71] M.M. Maleckar, J.L. Greenstein, N.A. Trayanova, W.R. Giles, Mathematical 245–272, http://dx.doi.org/10.1006/jtbi.1996.0129.
simulations of ligand-gated and cell-type specific effects on the action potential [96] S. Dokos, B.G. Celler, N.H. Lovell, Modification of DiFrancesco-Noble equations
of human atrium, Prog. Biophys. Mol. Biol. 98 (2008) 161–170, http://dx.doi. to simulate the effects of vagal stimulation on in vivo mammalian sinoatrial
org/10.1016/j.pbiomolbio.2009.01.010. node electrical activity, Ann. Biomed. Eng. 21 (1993) 321–335, http://dx.doi.
[72] J.T. Koivumäki, T. Korhonen, P. Tavi, Impact of sarcoplasmic reticulum calcium org/10.1007/bf02368625.
release on calcium dynamics and action potential morphology in human atrial [97] N.H. Lovell, S.L. Cloherty, B.G. Celler, S. Dokos, A gradient model of cardiac
myocytes: a computational study, PLoS Comput. Biol. 7 (2011) e1001067, pacemaker myocytes, Prog. Biophys. Mol. Biol. 85 (2004) 301–323, http://dx.
http://dx.doi.org/10.1371/journal.pcbi.1001067. doi.org/10.1016/j.pbiomolbio.2003.12.001.
[73] N. Voigt, J. Heijman, Q. Wang, D.Y. Chiang, N. Li, M. Karck, X.H.T. Wehrens, [98] H. Zhang, A.V. Holden, I. Kodama, H. Honjo, M. Lei, T. Varghese, M.R.
S. Nattel, D. Dobrev, Cellular and molecular mechanisms of atrial arrhythmo- Boyett, Mathematical models of action potentials in the periphery and center
genesis in patients with paroxysmal atrial fibrillation, Circulation 129 (2014) of the rabbit sinoatrial node, Am. J. Physiol. Heart Circ. Physiol. 279 (2000)
145–156, http://dx.doi.org/10.1161/CIRCULATIONAHA.113.006641. H397–H421, http://dx.doi.org/10.1152/ajpheart.2000.279.1.H397.
[74] J. Heijman, N. Voigt, S. Nattel, D. Dobrev, Cellular and molecular electrophys- [99] S. Matsuoka, N. Sarai, S. Kuratomi, K. Ono, A. Noma, Role of individual ionic
iology of atrial fibrillation initiation, maintenance, and progression, Circ. Res. current systems in ventricular cells hypothesized by a model study, Jpn. J.
114 (2014) 1483–1499, http://dx.doi.org/10.1161/CIRCRESAHA.114.302226. Physiol. 53 (2003) 105–123, http://dx.doi.org/10.2170/jjphysiol.53.105.
[75] M.W. Krueger, A. Dorn, D.U. Keller, F. Holmqvist, J. Carlson, P.G. Platonov, [100] T. Mitsuiye, Y. Shinagawa, A. Noma, Sustained inward current during pace-
K.S. Rhode, R. Razavi, G. Seemann, O. Dössel, In-silico modeling of atrial maker depolarization in mammalian sinoatrial node cells, Circ. Res. 87 (2000)
repolarization in normal and atrial fibrillation remodeled state, Med. Biol. Eng. 88–91, http://dx.doi.org/10.1161/01.res.87.2.88.
Comput. 51 (2013) 1105–1119, http://dx.doi.org/10.1007/s11517-013-1090-1. [101] M. Demion, P. Bois, P. Launay, R. Guinamard, TRPM4, a Ca2+-activated
[76] G. Seemann, C. Höper, F.B. Sachse, O. Dössel, A.V. Holden, H. Zhang, nonselective cation channel in mouse sino-atrial node cells, Cardiovasc. Res.
Heterogeneous three-dimensional anatomical and electrophysiological model of 73 (2007) 531–538, http://dx.doi.org/10.1016/j.cardiores.2006.11.023.
human atria, Philos. Trans. A Math. Phys. Eng. Sci. 364 (2006) 1465–1481, [102] T. Hof, C. Simard, R. Rouet, L. Salle, R. Guinamard, Implication of the TRPM4
http://dx.doi.org/10.1098/rsta.2006.1781. nonselective cation channel in mammalian sinus rhythm, Heart Rhythm. 10
(2013) 1683–1689, http://dx.doi.org/10.1016/j.hrthm.2013.08.014.
[77] E. Grandi, D. Dobrev, J. Heijman, Computational modeling: What does it
[103] N. Sarai, S. Matsuoka, S. Kuratomi, K. Ono, A. Noma, Role of individual ionic
tell us about atrial fibrillation therapy?, Int. J. Cardiol. 287 (2019) 155–161,
current systems in the SA node hypothesized by a model study, Jpn. J. Physiol.
http://dx.doi.org/10.1016/j.ijcard.2019.01.077.
53 (2003) 125–134, http://dx.doi.org/10.2170/jjphysiol.53.125.
[78] H. Ni, A. Fogli Iseppe, W.R. Giles, S.M. Narayan, H. Zhang, A.G. Edwards,
[104] S. Kharche, J. Yu, M. Lei, H. Zhang, A mathematical model of action potentials
S. Morotti, E. Grandi, Populations of in silico myocytes and tissues reveal
of mouse sinoatrial node cells with molecular bases, Am. J. Physiol. Heart Circ.
synergy of multiatrial-predominant K(+) -current block in atrial fibrillation, Br.
Physiol. 301 (2011) H945–H963, http://dx.doi.org/10.1152/ajpheart.00143.
J. Pharmacol. 177 (2020) 4497–4515, http://dx.doi.org/10.1111/bph.15198.
2010.
[79] M.A. Allessie, F.I. Bonke, F.J. Schopman, Circus movement in rabbit atrial
[105] Y. Kurata, I. Hisatome, S. Imanishi, T. Shibamoto, Dynamical description
muscle as a mechanism of tachycardia, Circ. Res. 33 (1973) 54–62.
of sinoatrial node pacemaking: improved mathematical model for primary
[80] C.W. Zemlin, B.G. Mitrea, A.M. Pertsov, Spontaneous onset of atrial fibrillation,
pacemaker cell, Am. J. Physiol. Heart Circ. Physiol. 283 (2002) H2074–H2101,
Physica D 238 (2009) 969–975, http://dx.doi.org/10.1016/j.physd.2008.12.
http://dx.doi.org/10.1152/ajpheart.00900.2001.
004.
[106] Y. Kurata, H. Matsuda, I. Hisatome, T. Shibamoto, Regional difference in
[81] A.G. Kléber, Y. Rudy, Basic mechanisms of cardiac impulse propagation and
dynamical property of sinoatrial node pacemaking: role of na+ channel cur-
associated arrhythmias, Physiol. Rev. 84 (2004) 431–488, http://dx.doi.org/
rent, Biophys. J. 95 (2008) 951–977, http://dx.doi.org/10.1529/biophysj.107.
10.1152/physrev.00025.2003.
112854.
[82] D.M. Bers, E. Grandi, Human atrial fibrillation: insights from computational
[107] Y. Kurata, I. Hisatome, S. Imanishi, T. Shibamoto, Roles of L-type Ca2+
electrophysiological models, Trends Cardiovasc. Med. 21 (2011) 145–150, http:
and delayed-rectifier k+ currents in sinoatrial node pacemaking: insights from
//dx.doi.org/10.1016/j.tcm.2012.04.004.
stability and bifurcation analyses of a mathematical model, Am. J. Physiol.
[83] K. Yanagihara, A. Noma, H. Irisawa, Reconstruction of sino-atrial node pace- Heart Circ. Physiol. 285 (2003) H2804–H2819, http://dx.doi.org/10.1152/
maker potential based on the voltage clamp experiments, Jpn. J. Physiol. 30 ajpheart.01050.2002.
(1980) 841–857, http://dx.doi.org/10.2170/jjphysiol.30.841. [108] S. Severi, M. Fantini, L.A. Charawi, D. DiFrancesco, An updated computational
[84] K. Yanagihara, H. Irisawa, Potassium current during the pacemaker depolar- model of rabbit sinoatrial action potential to investigate the mechanisms of
ization in rabbit sinoatrial node cell, Pflugers Arch. 388 (1980) 255–260, heart rate modulation, J. Physiol. 590 (2012) 4483–4499, http://dx.doi.org/
http://dx.doi.org/10.1007/bf00658491. 10.1113/jphysiol.2012.229435.
[85] D.G. Bristow, J.W. Clark, A mathematical model of primary pacemaking cell in [109] A. Fabbri, M. Fantini, R. Wilders, S. Severi, Computational analysis of the hu-
SA node of the heart, Am. J. Physiol. 243 (1982) H207–H218, http://dx.doi. man sinus node action potential: model development and effects of mutations,
org/10.1152/ajpheart.1982.243.2.H207. J. Physiol. 595 (2017) 2365–2396, http://dx.doi.org/10.1113/JP273259.
[86] V.S. Reiner, C. Antzelevitch, Phase resetting and annihilation in a mathematical [110] E. Ravagli, A. Bucchi, C. Bartolucci, M. Paina, M. Baruscotti, D. DiFrancesco, S.
model of sinus node, Am. J. Physiol. 249 (1985) H1143–H1153, http://dx.doi. Severi, Cell-specific Dynamic Clamp analysis of the role of funny If current
org/10.1152/ajpheart.1985.249.6.H1143. in cardiac pacemaking, Prog. Biophys. Mol. Biol. 120 (2016) 50–66, http:
[87] H. Irisawa, A. Noma, Pacemaker mechanisms of rabbit sinoatrial node cells, in: //dx.doi.org/10.1016/j.pbiomolbio.2015.12.004.
L.N. Bouman, H.J. Jongsma (Eds.), Cardiac Rate and Rhythm: Physiological, [111] A. Loewe, Y. Lutz, N. Nagy, A. Fabbri, C. Schweda, A. Varro, S. Severi, Inter-
Morphological, and Developmental Aspects, Martinus Nijhoff, London, 1982, species differences in the response of sinus node cellular pacemaking to changes
pp. 35–51. of extracellular calcium, in: Annu Int Conf IEEE Eng Med Biol Soc, Vol. 2019,
[88] R. Wilders, H.J. Jongsma, A.C. van Ginneken, Pacemaker activity of the rabbit 2019, pp. 1875–1878, http://dx.doi.org/10.1109/EMBC.2019.8857573.
sinoatrial node. A comparison of mathematical models, Biophys. J. 60 (1991) [112] A. Loewe, Y. Lutz, D. Nairn, A. Fabbri, N. Nagy, N. Toth, X. Ye, D.H. Fuertinger,
1202–1216, http://dx.doi.org/10.1016/S0006-3495(91)82155-5. S. Genovesi, P. Kotanko, J.G. Raimann, S. Severi, Hypocalcemia-induced slowing
[89] H. Irisawa, H.F. Brown, W. Giles, Cardiac pacemaking in the sinoatrial node, of human sinus node pacemaking, Biophys. J. 117 (2019) 2244–2254, http:
Physiol. Rev. 73 (1993) 197–227, http://dx.doi.org/10.1152/physrev.1993.73. //dx.doi.org/10.1016/j.bpj.2019.07.037.
1.197. [113] G. Campostrini, M. Bonzanni, A. Lissoni, C. Bazzini, R. Milanesi, E. Vezzoli,
[90] D. Noble, S.J. Noble, A model of sino-atrial node electrical activity based on M. Francolini, M. Baruscotti, A. Bucchi, I. Rivolta, M. Fantini, S. Severi,
a modification of the DiFrancesco-Noble (1984) equations, Proc. R. Soc. B 222 R. Cappato, L. Crotti, J.S. P, D. DiFrancesco, A. Barbuti, The expression of
(1984) 295–304, http://dx.doi.org/10.1098/rspb.1984.0065. the rare caveolin-3 variant T78m alters cardiac ion channels function and
[91] R.L. Rasmusson, J.W. Clark, W.R. Giles, E.F. Shibata, D.L. Campbell, A mathe- membrane excitability, Cardiovasc. Res. 113 (2017) 1256–1265, http://dx.doi.
matical model of a bullfrog cardiac pacemaker cell, Am. J. Physiol. 259 (1990) org/10.1093/cvr/cvx122.
H352–H369, http://dx.doi.org/10.1152/ajpheart.1990.259.2.H352. [114] S. Inada, J.C. Hancox, H. Zhang, M.R. Boyett, One-dimensional mathematical
[92] J.C. Denyer, H.F. Brown, Rabbit sino-atrial node cells: isolation and electro- model of the atrioventricular node including atrio-nodal, nodal, and nodal-his
physiological properties, J. Physiol. 428 (1990) 405–424, http://dx.doi.org/10. cells, Biophys. J. 97 (2009) 2117–2127, http://dx.doi.org/10.1016/j.bpj.2009.
1113/jphysiol.1990.sp018219. 06.056.

18
B. Amuzescu, R. Airini, F.B. Epureanu et al. Mathematical Biosciences 334 (2021) 108567

[115] A.M. Climent, M.S. Guillem, Y. Zhang, J. Millet, T.N. Mazgalev, Functional [136] P.C. Viswanathan, R.M. Shaw, Y. Rudy, Effects of IKr and IKs hetero-
mathematical model of dual pathway AV nodal conduction, Am. J. Physiol. geneity on action potential duration and its rate dependence: a simulation
Heart Circ. Physiol. 300 (2011) H1393–H1401, http://dx.doi.org/10.1152/ study, Circulation 99 (1999) 2466–2474, http://dx.doi.org/10.1161/01.cir.99.
ajpheart.01175.2010. 18.2466.
[116] S. Inada, T. Ono, N. Shibata, M. Iwata, R. Haraguchi, T. Ashihara, K. Mitsui, [137] P.C. Viswanathan, Y. Rudy, Pause induced early afterdepolarizations in the
M.R. Boyett, H. Dobrzynski, K. Nakazawa, Simulation study of complex action long QT syndrome: a simulation study, Cardiovasc. Res. 42 (1999) 530–542,
potential conduction in atrioventricular node, in: Annu Int Conf IEEE Eng Med http://dx.doi.org/10.1016/s0008-6363(99)00035-8.
Biol Soc, Vol. 2013, 2013, pp. 6850–6853, http://dx.doi.org/10.1109/EMBC. [138] P.C. Viswanathan, Y. Rudy, Cellular arrhythmogenic effects of congenital and
2013.6611131. acquired long-QT syndrome in the heterogeneous myocardium, Circulation 101
[117] J. Li, S. Inada, J.E. Schneider, H. Zhang, H. Dobrzynski, M.R. Boyett, Three- (2000) 1192–1198, http://dx.doi.org/10.1161/01.cir.101.10.1192.
dimensional computer model of the right atrium including the sinoatrial and [139] C.E. Clancy, Y. Rudy, Linking a genetic defect to its cellular phenotype in a
atrioventricular nodes predicts classical nodal behaviours, PLoS One 9 (2014) cardiac arrhythmia, Nature 400 (1999) 566–569, http://dx.doi.org/10.1038/
e112547, http://dx.doi.org/10.1371/journal.pone.0112547. 23034.
[118] E. Ryzhii, M. Ryzhii, Formation of second-degree atrioventricular blocks in the [140] C.E. Clancy, Y. Rudy, Na(+) channel mutation that causes both brugada and
cardiac heterogeneous oscillator model, in: Annu Int Conf IEEE Eng Med Biol long-QT syndrome phenotypes: a simulation study of mechanism, Circulation
Soc, Vol. 2015, 2015, pp. 4491–4494, http://dx.doi.org/10.1109/EMBC.2015. 105 (2002) 1208–1213, http://dx.doi.org/10.1161/hc1002.105183.
7319392.
[141] C.E. Clancy, Y. Rudy, Cellular consequences of HERG mutations in the long
[119] H. Cheng, J. Li, A.F. James, S. Inada, S.C. Choisy, C.H. Orchard, H. Zhang,
QT syndrome: precursors to sudden cardiac death, Cardiovasc. Res. 50 (2001)
M.R. Boyett, J.C. Hancox, Characterization and influence of cardiac background
301–313, http://dx.doi.org/10.1016/s0008-6363(00)00293-5.
sodium current in the atrioventricular node, J. Mol. Cell Cardiol. 97 (2016)
[142] B. Amuzescu, B. Istrate, S. Musat, Channelopathies and heart disease,
114–124, http://dx.doi.org/10.1016/j.yjmcc.2016.04.014.
in: K.A.S.e. al., (Ed.), Cardiac Arrhythmias: From Basic Mechanisms To
[120] B. Jackowska-Zduniak, U. Foryś, Mathematical model of the atrioventricular
State-of-the-Art Management, Springer, London, 2013, pp. 95–129.
nodal double response tachycardia and double-fire pathology, Math. Biosci. Eng.
[143] G.M. Faber, Y. Rudy, Action potential and contractility changes in [Na(+)](i)
13 (2016) 1143–1158, http://dx.doi.org/10.3934/mbe.2016035.
overloaded cardiac myocytes: a simulation study, Biophys. J. 78 (2000)
[121] I.P. Temple, S.J. Logantha, M. Absi, Y. Zhang, E. Pervolaraki, J. Yanni, A.
2392–2404, http://dx.doi.org/10.1016/S0006-3495(00)76783-X.
Atkinson, M. Petkova, G.M. Quigley, S. Castro, M. Drinkhill, H. Schneider, O.
Monfredi, E. Cartwright, M. Zi, T.T. Yamanushi, V.S. Mahadevan, A.M. Gurney, [144] T.J. Hund, Y. Rudy, Rate dependence and regulation of action potential and
E. White, H. Zhang, G. Hart, M.R. Boyett, H. Dobrzynski, Atrioventricular calcium transient in a canine cardiac ventricular cell model, Circulation 110
node dysfunction and ion channel transcriptome in pulmonary hypertension, (2004) 3168–3174, http://dx.doi.org/10.1161/01.CIR.0000147231.69595.D3.
Circ. Arrhythm. Electrophysiol. 9 (2016) e003432, http://dx.doi.org/10.1161/ [145] L. Livshitz, Y. Rudy, Uniqueness and stability of action potential models during
CIRCEP.115.003432. rest, pacing, and conduction using problem-solving environment, Biophys. J. 97
[122] M. Hulsmans, S. Clauss, L. Xiao, A.D. Aguirre, K.R. King, A. Hanley, W.J. (2009) 1265–1276, http://dx.doi.org/10.1016/j.bpj.2009.05.062.
Hucker, E.M. Wülfers, G. Seemann, G. Courties, Y. Iwamoto, Y. Sun, A.J. Savol, [146] M.S. Jafri, J.J. Rice, R.L. Winslow, Cardiac Ca2+ dynamics: the roles of
H.B. Sager, K.J. Lavine, G.A. Fishbein, D.E. Capen, N. Da Silva, L. Miquerol, ryanodine receptor adaptation and sarcoplasmic reticulum load, Biophys. J. 74
H. Wakimoto, C.E. Seidman, J.G. Seidman, R.I. Sadreyev, K. Naxerova, R.N. (1998) 1149–1168, http://dx.doi.org/10.1016/S0006-3495(98)77832-4.
Mitchell, D. Brown, P. Libby, R. Weissleder, F.K. Swirski, P. Kohl, C. Vinegoni, [147] R.L. Winslow, J. Rice, S. Jafri, E. Marbán, B. O’Rourke, Mechanisms of altered
D.J. Milan, P.T. Ellinor, M. Nahrendorf, Macrophages facilitate electrical Con- excitation-contraction coupling in canine tachycardia-induced heart failure, II:
duction in the Heart, Cell 169 (2017) 510–522, http://dx.doi.org/10.1016/j. model studies, Circ. Res. 84 (1999) 571–586, http://dx.doi.org/10.1161/01.res.
cell.2017.03.050. 84.5.571.
[123] M. Wallman, F. Sandberg, Characterisation of human AV-nodal properties using [148] J.L. Greenstein, R. Wu, S. Po, G.F. Tomaselli, R.L. Winslow, Role of the calcium-
a network model, Med. Biol. Eng. Comput. 56 (2018) 247–259, http://dx.doi. independent transient outward current I(to1) in shaping action potential
org/10.1007/s11517-017-1684-0. morphology and duration, Circ. Res. 87 (2000) 1026–1033, http://dx.doi.org/
[124] C.H. Luo, Y. Rudy, A model of the ventricular cardiac action potential. Depolar- 10.1161/01.res.87.11.1026.
ization, repolarization, and their interaction, Circ. Res. 68 (1991) 1501–1526, [149] J.L. Puglisi, D.M. Bers, LabHEART: an interactive computer model of rabbit
http://dx.doi.org/10.1161/01.res.68.6.1501. ventricular myocyte ion channels and Ca transport, Am. J. Physiol. Cell
[125] D. Noble, S.J. Noble, G.C. Bett, Y.E. Earm, W.K. Ho, I.K. So, The role of sodium- Physiol. 281 (2001) C2049–C2060, http://dx.doi.org/10.1152/ajpcell.2001.281.
calcium exchange during the cardiac action potential, Ann. New York Acad. Sci. 6.C2049.
639 (1991) 334–353, http://dx.doi.org/10.1111/j.1749-6632.1991.tb17323.x. [150] T.R. Shannon, F. Wang, J. Puglisi, C. Weber, D.M. Bers, A mathematical
[126] L. Ebihara, E.A. Johnson, Fast sodium current in cardiac muscle. A quantitative treatment of integrated Ca dynamics within the ventricular myocyte, Biophys.
description, Biophys. J. 32 (1980) 779–790, http://dx.doi.org/10.1016/S0006- J. 87 (2004) 3351–3371, http://dx.doi.org/10.1529/biophysj.104.047449.
3495(80)85016-8. [151] A. Mahajan, Y. Shiferaw, D. Sato, A. Baher, R. Olcese, L.H. Xie, M.J. Yang,
[127] B. Sakmann, G. Trube, Conductance properties of single inwardly rectifying P.S. Chen, J.G. Restrepo, A. Karma, A. Garfinkel, Z. Qu, J.N. Weiss, A rabbit
potassium channels in ventricular cells from guinea-pig heart, J. Physiol. 347 ventricular action potential model replicating cardiac dynamics at rapid heart
(1984) 641–657, http://dx.doi.org/10.1113/jphysiol.1984.sp015088. rates, Biophys. J. 94 (2008) 392–410, http://dx.doi.org/10.1529/biophysj.106.
[128] Y. Kurachi, Voltage-dependent activation of the inward-rectifier potassium 98160.
channel in the ventricular cell membrane of guinea-pig heart, J. Physiol. 366
[152] J.J. Saucerman, D.M. Bers, Calmodulin mediates differential sensitivity of
(1985) 365–385, http://dx.doi.org/10.1113/jphysiol.1985.sp015803.
CaMKII and calcineurin to local Ca2+ in cardiac myocytes, Biophys. J. 95
[129] D.T. Yue, E. Marbán, A novel cardiac potassium channel that is active and
(2008) 4597–4612, http://dx.doi.org/10.1529/biophysj.108.128728.
conductive at depolarized potentials, Pflügers Arch. 413 (1988) 127–133, http:
[153] V.E. Bondarenko, G.P. Szigeti, G.C. Bett, S.J. Kim, R.L. Rasmusson, Computer
//dx.doi.org/10.1007/bf00582522.
model of action potential of mouse ventricular myocytes, Am. J. Physiol. Heart
[130] B. Sakmann, G. Trube, Voltage-dependent inactivation of inward-rectifying
Circ. Physiol. 287 (2004) H1378–H1403, http://dx.doi.org/10.1152/ajpheart.
single-channel currents in the guinea-pig heart cell membrane, J. Physiol. 347
00185.2003.
(1984) 659–683, http://dx.doi.org/10.1113/jphysiol.1984.sp015089.
[154] A.R. Soltis, J.J. Saucerman, Synergy between CaMKII substrates and Î2 -
[131] C.H. Luo, Y. Rudy, A dynamic model of the cardiac ventricular action potential.
adrenergic signaling in regulation of cardiac myocyte Ca(2+) handling, Biophys.
I. Simulations of ionic currents and concentration changes, Circ. Res. 74 (1994)
J. 99 (2010) 2038–2047, http://dx.doi.org/10.1016/j.bpj.2010.08.016.
1071–1096, http://dx.doi.org/10.1161/01.res.74.6.1071.
[132] C.H. Luo, Y. Rudy, A dynamic model of the cardiac ventricular action potential. [155] S. Morotti, A.G. Edwards, A.D. McCulloch, D.M. Bers, E. Grandi, A novel
II. Afterdepolarizations, triggered activity, and potentiation, Circ. Res. 74 (1994) computational model of mouse myocyte electrophysiology to assess the synergy
1097–1113, http://dx.doi.org/10.1161/01.res.74.6.1097. between na+ loading and CaMKII, J. Physiol. 592 (2014) 1181–1197, http:
[133] R.M. Shaw, Y. Rudy, The vulnerable window for unidirectional block in //dx.doi.org/10.1113/jphysiol.2013.266676.
cardiac tissue: characterization and dependence on membrane excitability and [156] J.A. Negroni, S. Morotti, E.C. Lascano, A.V. Gomes, E. Grandi, J.L. Puglisi,
intercellular coupling, J. Cardiovasc. Electrophysiol. 6 (1995) 115–131, http: D.M. Bers, 𝛽-Adrenergic effects on cardiac myofilaments and contraction in an
//dx.doi.org/10.1111/j.1540-8167.1995.tb00763.x. integrated rabbit ventricular myocyte model, J. Mol. Cell Cardiol. 81 (2015)
[134] J. Zeng, Y. Rudy, Early afterdepolarizations in cardiac myocytes: mechanism 162–175, http://dx.doi.org/10.1016/j.yjmcc.2015.02.014.
and rate dependence, Biophys. J. 68 (1995) 949–964, http://dx.doi.org/10. [157] J.D. Moreno, P.C. Yang, J.R. Bankston, E. Grandi, D.M. Bers, R.S. Kass, C.E.
1016/S0006-3495(95)80271-7. Clancy, Ranolazine for congenital and acquired late INa-linked arrhythmias: in
[135] R. Dumaine, J.A. Towbin, P. Brugada, M. Vatta, D.V. Nesterenko, V.V. silico pharmacological screening, Circ. Res. 113 (2013) e50, http://dx.doi.org/
Nesterenko, J. Brugada, R. Brugada, C. Antzelevitch, Ionic mechanisms re- 10.1161/CIRCRESAHA.113.301971-e61.
sponsible for the electrocardiographic phenotype of the brugada syndrome are [158] L. Priebe, D.J. Beuckelmann, Simulation study of cellular electric properties in
temperature dependent, Circ. Res. 85 (1999) 803–809, http://dx.doi.org/10. heart failure, Circ. Res. 82 (1998) 1206–1223, http://dx.doi.org/10.1161/01.
1161/01.res.85.9.803. res.82.11.1206.

19
B. Amuzescu, R. Airini, F.B. Epureanu et al. Mathematical Biosciences 334 (2021) 108567

[159] O. Bernus, R. Wilders, C.W. Zemlin, H. Verschelde, A.V. Panfilov, A computa- [178] T. Korhonen, R. Rapila, V.P. Ronkainen, J.T. Koivumäki, P. Tavi, Local Ca2+
tionally efficient electrophysiological model of human ventricular cells, Am. J. releases enable rapid heart rates in developing cardiomyocytes, J. Physiol. 588
Physiol. Heart Circ. Physiol. 282 (2002) H2296–H2308, http://dx.doi.org/10. (2010) 1407–1417, http://dx.doi.org/10.1113/jphysiol.2009.185173.
1152/ajpheart.00731.2001. [179] J.T. Koivumäki, N. Naumenko, T. Tuomainen, J. Takalo, M. Oksanen, K.A.
[160] V. Iyer, R. Mazhari, R.L. Winslow, A computational model of the human Puttonen, Š. Lehtonen, J. Kuusisto, M. Laakso, J. Koistinaho, P. Tavi, Structural
left-ventricular epicardial myocyte, Biophys. J. 87 (2004) 1507–1525, http: immaturity of human iPSC-derived cardiomyocytes: In silico investigation of
//dx.doi.org/10.1529/biophysj.104.043299. effects on function and disease modeling, Front. 9 (2018) http://dx.doi.org/10.
[161] K.H. ten Tusscher, D. Noble, P.J. Noble, A.V. Panfilov, A model for human ven- 3389/fphys.2018.00080.
tricular tissue, Am. J. Physiol. Heart Circ. Physiol. 286 (2004) H1573–H1589, [180] P. Kügler, G. Rast, B.D. Guth, Comparison of in vitro and computational exper-
http://dx.doi.org/10.1152/ajpheart.00794.2003. iments on the relation of inter-beat interval and duration of repolarization in a
[162] J. Silva, Y. Rudy, Subunit interaction determines IKs participation in cardiac specific type of human induced pluripotent stem cell-derived cardiomyocytes,
repolarization and repolarization reserve, Circulation 112 (2005) 1384–1391, PLoS One 14 (2019) 2019, http://dx.doi.org/10.1371/journal.pone.0221763.
http://dx.doi.org/10.1161/CIRCULATIONAHA.105.543306. [181] D.C. Kernik, S. Morotti, H. Wu, P. Garg, H.J. Duff, J. Kurokawa, J. Jalife,
[163] T. O’Hara, L. Virág, A. Varró, Y. Rudy, Simulation of the undiseased human J.C. Wu, E. Grandi, C.E. Clancy, A computational model of induced pluripotent
cardiac ventricular action potential: model formulation and experimental val- stem-cell derived cardiomyocytes incorporating experimental variability from
idation, PLoS Comput. Biol. 7 (2011) 26, http://dx.doi.org/10.1371/journal. multiple data sources, J. Physiol. 597 (2019) 4533–4564, http://dx.doi.org/10.
pcbi.1002061. 1113/jp277724.
[164] L. Thorrez, M. Sampaolesi, The future of induced pluripotent stem cells for [182] D.C. Kernik, P.C. Yang, J. Kurokawa, J.C. Wu, C.E. Clancy, A computa-
cardiac therapy and drug development, Curr. 17 (2011) 3258–3270, http: tional model of induced pluripotent stem-cell derived cardiomyocytes for high
//dx.doi.org/10.2174/138161211797904154. throughput risk stratification of KCNQ1 genetic variants, PLoS Comput. Biol.
[165] P. Dell’Era, P. Benzoni, E. Crescini, M. Valle, E. Xia, A. Consiglio, M. Memo, 16 (2020) e1008109, http://dx.doi.org/10.1371/journal.pcbi.1008109.
Cardiac disease modeling using induced pluripotent stem cell-derived human [183] D.T. Du, N. Hellen, C. Kane, C.M. Terracciano, Action potential morphology of
cardiomyocytes, World J. Stem. Cells 7 (2015) 329–342, http://dx.doi.org/10. human induced pluripotent stem cell-derived cardiomyocytes does not predict
4252/wjsc.v7.i2.329. cardiac chamber specificity and is dependent on cell density, Biophys. J. 108
[166] F. Lan, A.S. Lee, P. Liang, V. Sanchez-Freire, P.K. Nguyen, L. Wang, L. Han, (2015) 1–4, http://dx.doi.org/10.1016/j.bpj.2014.11.008.
M. Yen, Y. Wang, N. Sun, O.J. Abilez, S. Hu, A.D. Ebert, E.G. Navarrete, [184] W.R. Giles, D. Noble, Rigorous phenotyping of cardiac iPSC preparations
C.S. Simmons, M. Wheeler, B. Pruitt, R. Lewis, Y. Yamaguchi, E.A. Ashley, requires knowledge of their resting potential(s), Biophys. J. 110 (2016)
D.M. Bers, R.C. Robbins, M.T. Longaker, J.C. Wu, Abnormal calcium handling 278–280.
properties underlie familial hypertrophic cardiomyopathy pathology in patient- [185] S.S. Parikh, D.J. Blackwell, N. Gomez-Hurtado, M. Frisk, L. Wang, K. Kim,
specific induced pluripotent stem cells, Cell Stem. Cell 12 (2013) 101–113, C.P. Dahl, A. Fiane, T. Tonnessen, D.O. Kryshtal, W.E. Louch, B.C. Knollmann,
http://dx.doi.org/10.1016/j.stem.2012.10.010. Thyroid and glucocorticoid hormones promote functional T-tubule development
[167] C.W. Siu, Y.K. Lee, J.C. Ho, W.H. Lai, Y.C. Chan, K.M. Ng, L.Y. Wong, K.W.
in human-induced pluripotent stem cell-derived cardiomyocytes, Circ. Res. 121
Au, Y.M. Lau, J. Zhang, K.W. Lay, A. Colman, H.F. Tse, Modeling of lamin A/C
(2017) 1323–1330.
mutation premature cardiac aging using patient-specific induced pluripotent
[186] J.J. Kim, L. Yang, B. Lin, X. Zhu, B. Sun, A.D. Kaplan, G.C. Bett, R.L. Rasmusson,
stem cells, Aging 4 (2012) 803–822, http://dx.doi.org/10.18632/aging.100503.
B. London, G. Salama, Mechanism of automaticity in cardiomyocytes derived
[168] G. Wang, M.L. McCain, L. Yang, A. He, F.S. Pasqualini, A. Agarwal, H. Yuan, D.
from human induced pluripotent stem cells, J. Mol. Cell Cardiol. 81 (2015)
Jiang, D. Zhang, L. Zangi, J. Geva, A.E. Roberts, Q. Ma, J. Ding, J. Chen, D.Z.
81–93, http://dx.doi.org/10.1016/j.yjmcc.2015.01.013.
Wang, K. Li, J. Wang, R.J. Wanders, W. Kulik, F.M. Vaz, M.A. Laflamme, C.E.
[187] S.A. Mann, J. Heide, T. Knott, R. Airini, F.B. Epureanu, A.-F. Deftu, A.-T. Deftu,
Murry, K.R. Chien, R.I. Kelley, G.M. Church, K.K. Parker, W.T. Pu, Modeling
B.M. Radu, B. Amuzescu, Recording of multiple ion current components and
the mitochondrial cardiomyopathy of Barth syndrome with induced pluripotent
action potentials in human induced pluripotent stem cell-derived cardiomy-
stem cell and heart-on-chip technologies, Nat. Med. 20 (2014) 616–623, http:
ocytes via automated patch-clamp, J. Pharmacol. Toxicol. Methods 100 (2019)
//dx.doi.org/10.1038/nm.3545.
106599, http://dx.doi.org/10.1016/j.vascn.2019.106599.
[169] I. Itzhaki, L. Maizels, I. Huber, L. Zwi-Dantsis, O. Caspi, A. Winterstern, O.
[188] J.D. Murray, Hodgkin–Huxley Theory of Nerve Membranes: FitzHugh–Nagumo
Feldman, A. Gepstein, G. Arbel, H. Hammerman, M. Boulos, L. Gepstein,
Model, Mathematical Biology. I. an Introduction, third ed., Springer Verlag,
Modelling the long QT syndrome with induced pluripotent stem cells, Nature
New York Berlin Heidelberg, 2002, pp. 239–244.
471 (2011) 225–229, http://dx.doi.org/10.1038/nature09747.
[189] E. Doedel, Auto: a program for the automatic bifurcation analysis of au-
[170] A. Moretti, M. Bellin, A. Welling, C.B. Jung, J.T. Lam, L. Bott-Flugel, T. Dorn,
tonomous systems. in: Proc. Manitoba Conf. Num. Math. Comput. 10th Congr.
A. Goedel, C. Höhnke, F. Hofmann, M. Seyfarth, D. Sinnecker, A. Schömig,
Num. Winnipeg, Canada, 1981, pp. 265–284.
K.L. Laugwitz, Patient-specific induced pluripotent stem-cell models for long-
[190] E. Doedel, J. Kernévez, AUTO: Software for Continuation Problems in Ordinary
QT syndrome, New Engl. J. Med. 363 (2010) 1397–1409, http://dx.doi.org/10.
Differential Equations with Applications, California Institute of Technology,
1056/NEJMoa0908679.
Applied Mathematics, 1986.
[171] M. Paci, J. Hyttinen, K. Aalto-Setälä, S. Severi, Computational models of
ventricular- and atrial-like human induced pluripotent stem cell derived car- [191] T.R. Chay, Y.S. Lee, Phase resetting and bifurcation in the ventricular my-
diomyocytes, Ann. Biomed. Eng. 41 (2013) 2334–2348, http://dx.doi.org/10. ocardium, Biophys. J. 47 (1985) 641–651, http://dx.doi.org/10.1016/S0006-
1007/s10439-013-0833-3. 3495(85)83960-6.
[172] J. Ma, L. Guo, S.J. Fiene, B.D. Anson, J.A. Thomson, T.J. Kamp, K.L. Kolaja, [192] J.P. Drouhard, F.A. Roberge, Revised formulation of the Hodgkin–Huxley
B.J. Swanson, C.T. January, High purity human-induced pluripotent stem cell- representation of the sodium current in cardiac cells, Comput. Biomed. Res.
derived cardiomyocytes: electrophysiological properties of action potentials 20 (1987) 333–350, http://dx.doi.org/10.1016/0010-4809(87)90048-6.
and ionic currents, Am. J. Physiol. Heart Circ. Physiol. 301 (2011) 2, http: [193] A. Vinet, F.A. Roberge, A model study of stability and oscillations in the
//dx.doi.org/10.1152/ajpheart.00694.2011. myocardial cell membrane, J. Theoret. Biol. 147 (1990) 377–412, http://dx.
[173] O. Scheel, S. Frech, B. Amuzescu, J. Eisfeld, K.H. Lin, T. Knott, Action potential doi.org/10.1016/S0022-5193(05)80495-3.
characterization of human induced pluripotent stem cell-derived cardiomyocytes [194] W.J. Gibb, M.B. Wagner, M.D. Lesh, Effects of simulated potassium blockade
using automated patch-clamp technology, Assay. Drug Dev. Technol. 12 (2014) on the dynamics of triggered cardiac activity, J. Theoret. Biol. 168 (1994)
457–469, http://dx.doi.org/10.1089/adt.2014.601. 245–257, http://dx.doi.org/10.1006/jtbi.1994.1106.
[174] M. Paci, J. Hyttinen, B. Rodriguez, S. Severi, Human induced pluripotent stem [195] D.X. Tran, D. Sato, A. Yochelis, J.N. Weiss, A. Garfinkel, Z. Qu, Bifurcation
cell-derived versus adult cardiomyocytes: an in silico electrophysiological study and chaos in a model of cardiac early afterdepolarizations, Phys. Rev. Lett.
on effects of ionic current block, Br. J. Pharmacol. 172 (2015) 5147–5160, 102 (2009) 258103, http://dx.doi.org/10.1103/physrevlett.102.258103.
http://dx.doi.org/10.1111/bph.13282. [196] D. Sato, L.H. Xie, A.A. Sovari, D.X. Tran, N. Morita, F. Xie, H. Karagueuzian,
[175] C.L. Lei, K. Wang, M. Clerx, R.H. Johnstone, M.P. Hortigon-Vinagre, V. Zamora, A. Garfinkel, J.N. Weiss, Z. Qu, Synchronization of chaotic early afterdepolar-
A. Allan, G.L. Smith, D.J. Gavaghan, G.R. Mirams, L. Polonchuk, Tailoring izations in the genesis of cardiac arrhythmias, Proc. Natl. Acad. Sci. USA 106
mathematical models to stem-cell derived cardiomyocyte lines can improve (2009) 2983–2988.
predictions of drug-induced changes to their electrophysiology, Front. Physiol. [197] A. Dhooge, W. Govaerts, Y.A. Kuznetsov, MATCONT: A MATLAB package for
8 (2017) http://dx.doi.org/10.3389/fphys.2017.00986. numerical bifurcation analysis of ODEs, ACM Trans. Math. Software 29 (2003)
[176] M. Paci, R.P. Pölönen, D. Cori, K. Penttinen, K. Aalto-Setälä, S. Severi, J. 141–164, http://dx.doi.org/10.1145/779359.779362.
Hyttinen, Automatic optimization of an in silico model of human iPSC derived [198] A. Dhooge, W. Govaerts, Y.A. Kuznetsov, W. Mestrom, A.M. Riet, B. Sautois,
cardiomyocytes recapitulating calcium handling abnormalities, Front. Physiol. MATCONT and CL_MATCONT: Continuation Toolboxes in MATLAB, Universiteit
9 (2018) 709, http://dx.doi.org/10.3389/fphys.2018.00709. Gent, Belgium, and Utrecht University, The Netherlands, 2006.
[177] M. Paci, S. Casini, M. Bellin, J. Hyttinen, S. Severi, Large-scale simulation of [199] B. Amuzescu, A. Georgescu, G. Nistor, M. Popescu, I. Svab, M.L. Flonta,
the phenotypical variability induced by loss-of-function long QT mutations in A.D. Corlan, Stability and sustained oscillations in a ventricular cardiomyocyte
human induced pluripotent stem cell cardiomyocytes, Int. J. Mol. Sci. 19 (2018) model, Interdiscip. Sci. 4 (2012) 1–18, http://dx.doi.org/10.1007/s12539-012-
http://dx.doi.org/10.3390/ijms19113583. 0116-y.

20
B. Amuzescu, R. Airini, F.B. Epureanu et al. Mathematical Biosciences 334 (2021) 108567

[200] M. Zaniboni, A.E. Pollard, L. Yang, K.W. Spitzer, Beat-to-beat repolarization [220] R.H. Johnstone, E.T.Y. Chang, R. Bardenet, T.P. de Boer, D.J. Gavaghan,
variability in ventricular myocytes and its suppression by electrical coupling, P. Pathmanathan, R.H. Clayton, G.R. Mirams, Uncertainty and variability in
Am. J. Physiol. Heart Circ. Physiol. 278 (2000) H677–H687, http://dx.doi.org/ models of the cardiac action potential: Can we build trustworthy models?, J.
10.1152/ajpheart.2000.278.3.H677. Mol. Cell Cardiol. 96 (2016) 49–62, http://dx.doi.org/10.1016/j.yjmcc.2015.
[201] Z. Syed, E. Vigmond, S. Nattel, L.J. Leon, Atrial cell action potential parameter 11.018.
fitting using genetic algorithms, Med. Biol. Eng. Comput. 43 (2005) 561–571, [221] G.R. Mirams, P. Pathmanathan, R.A. Gray, P. Challenor, R.H. Clayton, Uncer-
http://dx.doi.org/10.1007/BF02351029. tainty and variability in computational and mathematical models of cardiac
[202] E. Pouranbarani, R. Weber Dos Santos, A. Nygren, A robust multi-objective physiology, J. Physiol. 594 (2016) 6833–6847, http://dx.doi.org/10.1113/
optimization framework to capture both cellular and intercellular properties JP271671.
in cardiac cellular model tuning: Analyzing different regions of membrane [222] J.Q.X. Gong, E.A. Sobie, Population-based mechanistic modeling allows for
resistance profile in parameter fitting, PLoS One 14 (2019) e0225245, http: quantitative predictions of drug responses across cell types, NPJ Syst. Biol.
//dx.doi.org/10.1371/journal.pone.0225245. Appl. 4 (2018) http://dx.doi.org/10.1038/s41540-018-0047-2.
[203] B. Trenor, K. Cardona, L. Romero, J.F. Gomez, J. Saiz, S. Rajamani, L. [223] M. Paci, E. Passini, S. Severi, J. Hyttinen, B. Rodriguez, Phenotypic variability
Belardinelli, W. Giles, Pro-arrhythmic effects of low plasma [K(+)] in human in LQT3 human induced pluripotent stem cell-derived cardiomyocytes and their
ventricle: An illustrated review, Trends Cardiovasc. Med. 28 (2018) 233–242, response to antiarrhythmic pharmacologic therapy: An in silico approach, Heart
http://dx.doi.org/10.1016/j.tcm.2017.11.002. Rhythm 14 (2017) 1704–1712, http://dx.doi.org/10.1016/j.hrthm.2017.07.026.
[224] S. Ramasubramanian, Y. Rudy, The structural basis of IKs ion-channel activa-
[204] B. Trenor, K. Cardona, J. Saiz, D. Noble, W. Giles, Cardiac action potential
tion: Mechanistic insights from molecular simulations, Biophys. J. 114 (2018)
repolarization revisited: early repolarization shows all-or-none behaviour, J.
2584–2594, http://dx.doi.org/10.1016/j.bpj.2018.04.023.
Physiol. 595 (2017) 6599–6612, http://dx.doi.org/10.1113/JP273651.
[225] J. Xu, Y. Rudy, Effects of b-subunit on gating of a potassium ion channel:
[205] Y. Kurata, K. Tsumoto, K. Hayashi, I. Hisatome, Y. Kuda, M. Tanida, Multiple
Molecular simulations of cardiac IKs activation, J. Mol. Cell Cardiol. 124 (2018)
dynamical mechanisms of phase-2 early afterdepolarizations in a human ven-
35–44, http://dx.doi.org/10.1016/j.yjmcc.2018.10.003.
tricular myocyte model: Involvement of spontaneous SR Ca (2+) release, Front.
[226] P.C. Yang, K.R. DeMarco, P. Aghasafari, M.T. Jeng, J.R.D. Dawson, S. Bekker,
Physiol. 10 (2020) 1–22, http://dx.doi.org/10.3389/fphys.2019.01545.
S.Y. Noskov, V. Yarov-Yarovoy, I. Vorobyov, C.E. Clancy, A computational
[206] Y. Shiferaw, M.A. Watanabe, A. Garfinkel, J.N. Weiss, A. Karma, Model of
pipeline to predict cardiotoxicity: From the atom to the rhythm, Circ. Res.
intracellular calcium cycling in ventricular myocytes, Biophys. J. 85 (2003)
126 (2020) 947–964, http://dx.doi.org/10.1161/circresaha.119.316404.
3666–3686, http://dx.doi.org/10.1016/S0006-3495(03)74784-5.
[227] Z. Li, G.R. Mirams, T. Yoshinaga, B.J. Ridder, X. Han, J.E. Chen, N.L.
[207] Y. Shiferaw, G.L. Aistrup, W.E. Louch, J.A. Wasserstrom, Remodeling promotes Stockbridge, T.A. Wisialowski, B. Damiano, S. Severi, P. Morissette, P.R. Kowey,
proarrhythmic disruption of calcium homeostasis in failing atrial myocytes, M. Holbrook, G. Smith, R.L. Rasmusson, M. Liu, Z. Song, Z. Qu, D.J. Leishman,
Biophys. J. 118 (2020) 476–491, http://dx.doi.org/10.1016/j.bpj.2019.12.012. J. Steidl-Nichols, B. Rodriguez, A. Bueno-Orovio, X. Zhou, E. Passini, A.G.
[208] W.S. Redfern, L. Carlsson, A.S. Davis, W.G. Lynch, I. MacKenzie, S. Palethorpe, Edwards, S. Morotti, H. Ni, E. Grandi, C.E. Clancy, J. Vandenberg, A. Hill,
P.K. Siegl, I. Strang, A.T. Sullivan, R. Wallis, A.J. Camm, T.G. Hammond, M. Nakamura, T. Singer, L. Polonchuk, A. Greiter-Wilke, K. Wang, S. Nave, A.
Relationships between preclinical cardiac electrophysiology, clinical QT interval Fullerton, E.A. Sobie, M. Paci, F. Musuamba Tshinanu, D.G. Strauss, General
prolongation and torsade de pointes for a broad range of drugs: evidence for principles for the validation of proarrhythmia risk prediction models: An
a provisional safety margin in drug development, Cardiovasc. Res. 58 (2003) extension of the CiPA in silico strategy, Clin. Pharmacol. Ther. 107 (2019)
32–45, http://dx.doi.org/10.1016/s0008-6363(02)00846-5. 102–111, http://dx.doi.org/10.1002/cpt.1647.
[209] D. Noble, Computational models of the heart and their use in assessing the [228] D. Noble, Cardiac action and pacemaker potentials based on the Hodgkin–
actions of drugs, J. Pharmacol. Sci. 107 (2008) 107–117, http://dx.doi.org/10. Huxley equations, Nature 188 (1960) 495–497, http://dx.doi.org/10.1038/
1254/jphs.cr0070042. 188495b0.
[210] J. Kramer, C.A. Obejero-Paz, G. Myatt, Y.A. Kuryshev, A. Bruening-Wright, J.S. [229] D. Noble, A description of cardiac pacemaker potentials based on the Hodgkin–
Verducci, A.M. Brown, MICE models: superior to the HERG model in predicting Huxley equations, J. Physiol. 154 (1960) 64–65P, http://dx.doi.org/10.1038/
Torsade de Pointes, Sci. Rep. 3 (2013) http://dx.doi.org/10.1038/srep02100. 188495b0.
[211] P.T. Sager, G. Gintant, J.R. Turner, S. Pettit, N. Stockbridge, Rechanneling [230] K.S. Cole, Membrane excitation of the Hodgkin–Huxley axon; preliminary
the cardiac proarrhythmia safety paradigm: a meeting report from the Cardiac corrections, J. Appl. Physiol. 12 (1958) 129–130, http://dx.doi.org/10.1152/
Safety Research Consortium, Am. Heart J. 167 (2014) 292–300, http://dx.doi. jappl.1958.12.1.129.
org/10.1016/j.ahj.2013.11.004. [231] K.S. Cole, H.A. Antosiewicz, P. Rabinowitz, Automatic computation of nerve
[212] B. Amuzescu, B.M. Radu, D.F. Mihailescu, S.A. Mann, EP3546935 / 2019 - excitation, J. Soc. Ind. Appl. Math. 3 (1955) 153–172.
method for in vitro detection of the proarrhythmogenic risk of a drug candidate [232] R. Fitzhugh, H.A. Antosiewicz, Automatic computation of nerve excitation-
on human induced pluripotent stem cell-derived cardiomyocytes (HIPSC-CM), detailed corrections and additions, J. Soc. Ind. Appl. Math. 7 (1959)
EPO application number 18465611.4, filed 24.10.2018, 2018. 447.
[213] S. Dutta, K.C. Chang, K.A. Beattie, J. Sheng, P.N. Tran, W.W. Wu, M. Wu, D.G. [233] E.T. Whittaker, G. Robinson, The Calculus of Observations, Blackie, London,
Strauss, T. Colatsky, Z. Li, Optimization of an in silico cardiac cell model for 1924, chapter 14.
proarrhythmia risk assessment, Front 8 (2017) 616, http://dx.doi.org/10.3389/ [234] X. Yang, D.A. Ralescu, Adams method for solving uncertain differential equa-
fphys.2017.00616. tions, Appl. Math. Comput. 270 (2015) 993–1003, http://dx.doi.org/10.1016/
j.amc.2015.08.109.
[214] K.C. Chang, S. Dutta, G.R. Mirams, K.A. Beattie, J. Sheng, P.N. Tran, M. Wu,
[235] D.R. Hartree, A practical method for the numerical solution of differential
W.W. Wu, T. Colatsky, D.G. Strauss, Z. Li, Uncertainty quantification reveals
equations, Mem. Proc. Manchester Lit. Phil. Soc. 77 (1932-33) 91–107.
the importance of data variability and experimental design considerations for
[236] S. Rush, H. Larsen, A practical algorithm for solving dynamic membrane
in silico proarrhythmia risk assessment, Front. Physiol. 8 (2017) http://dx.doi.
equations, IEEE Trans. Biomed. Eng. 25 (1978) 389–392, http://dx.doi.org/10.
org/10.3389/fphys.2017.00917, eCollection 2017.
1109/TBME.1978.326270.
[215] J. Tomek, A. Bueno-Orovio, E. Passini, X. Zhou, A. Minchole, O. Britton, C.
[237] W.E. Boyce, R.C. DiPrima, Elementary Differential Equations and Boundary
Bartolucci, S. Severi, A. Shrier, L. Virág, A. Varró, B. Rodriguez, Development,
Value Problems, 7th edn., John Wiley & Sons, Inc., New York, 2001, chapter
calibration, and validation of a novel human ventricular myocyte model in
8.
health, disease, and drug block, Elife 8 (2019) e48890, http://dx.doi.org/10.
[238] D. Noble, A. Varghese, P. Kohl, P. Noble, Improved guinea-pig ventricu-
7554/elife.48890.
lar cell model incorporating a diadic space, IKr and IKs, and length- and
[216] E. Passini, O.J. Britton, H.R. Lu, J. Rohrbacher, A.N. Hermans, D.J. Gallacher, tension-dependent processes, Can. J. Cardiol. 14 (1998) 123–134.
R.J.H. Greig, A. Bueno-Orovio, B. Rodriguez, Human in silico drug trials [239] S. Missan, T.F. McDonald, CESE: Cell electrophysiology simulation environment,
demonstrate higher accuracy than animal models in predicting clinical pro- Appl. Bioinform. 4 (2005) 155–156, http://dx.doi.org/10.2165/00822942-
arrhythmic cardiotoxicity, Front. Physiol. 8 (2017) 668, http://dx.doi.org/10. 200504020-00011.
3389/fphys.2017.00668. [240] S. Vascotto, S. Missan, C. La, CESE plus: a computational approach for
[217] E. Pueyo, A. Corrias, L. Virág, N. Jost, T. Szél, A. Varró, N. Szentandrássy, P.P. accelerating electrophysiology research, Lab. Focus 13 (2009) 8–11.
Nánási, K. Burrage, B. Rodríguez, A multiscale investigation of repolarization [241] S.S. Demir, Interactive cell modeling web-resource, icell, as a simulation-based
variability and its role in cardiac arrhythmogenesis, Biophys. J. 101 (2011) teaching and learning tool to supplement electrophysiology education, Ann.
2892–2902, http://dx.doi.org/10.1016/j.bpj.2011.09.060. Biomed. Eng. 34 (2006) 1077–1087, http://dx.doi.org/10.1007/s10439-006-
[218] A.X. Sarkar, D.J. Christini, E.A. Sobie, Exploiting mathematical models to 9138-0.
illuminate electrophysiological variability between individuals, J. Physiol. 590 [242] A. Garny, P.J. Hunter, OpenCOR: a modular and interoperable approach to
(2012) 2555–2567, http://dx.doi.org/10.1113/jphysiol.2011.223313. computational biology, Front. Physiol. 6 (2015) http://dx.doi.org/10.3389/
[219] M.R. Davies, K. Wang, G.R. Mirams, A. Caruso, D. Noble, A. Walz, T. Lave, fphys.2015.00026.
F. Schuler, T. Singer, L. Polonchuk, Recent developments in using mechanistic [243] E. Butterworth, B.E. Jardine, G.M. Raymond, M.L. Neal, J.B. Bassingthwaighte,
cardiac modelling for drug safety evaluation, Drug Discov. Today 21 (2016) JSim, an open-source modeling system for data analysis, F1000 Res. 2 (2013)
924–938, 10./j.drudis.2016.02.003. http://dx.doi.org/10.12688/f1000research.2-288.v1, eCollection 2013.

21
B. Amuzescu, R. Airini, F.B. Epureanu et al. Mathematical Biosciences 334 (2021) 108567

[244] C.B. Barbosa, R.W. dos Santos, R.M. Amorim, L.N. Ciuffo, F. Man- [254] E. Bartocci, E.M. Cherry, J. Glimm, R. Grosu, S.A. Smolka, F.H. Fenton, Toward
froi, R.S. Oliveira, F.O. Campos, A transformation tool for ODE based Real-Time Simulation of Cardiac Dynamics, CMSB 2011: Proceedings of the 9th
models, in: V.N. Alexandrov, G.D. van Albada, P.M.A. Sloot, J. Don- ACM International Conference on Computational Methods in Systems Biology,
garra (Eds.), Computational Science – ICCS 2006. ICCS 2006. Lecture Paris, France, September 21-23, New York, NY, USA, 2011, pp. 103–110.
Notes in Computer Science, vol. 3991, Springer, Berlin, Heidelberg, [255] M. Nivala, E. de Lange, R. Rovetti, Z. Qu, Computational modeling and
2006, pp. 68–75, http://dx.doi.org/10.1007/11758501_14. numerical methods for spatiotemporal calcium cycling in ventricular myocytes,
[245] A.C. Daly, M. Clerx, K.A. Beattie, J. Cooper, D.J. Gavaghan, G.R. Mirams, Re- Front. Physiol. 3 (2012) 114, http://dx.doi.org/10.3389/fphys.2012.00114.
producible model development in the cardiac electrophysiology web lab, Prog. [256] E. Van Nieuwenhuyse, G. Seemann, A.V. Panfilov, N. Vandersickel, Effects of
Biophys. Mol. Biol. 139 (2018) 3–14, http://dx.doi.org/10.1016/j.pbiomolbio. early afterdepolarizations on excitation patterns in an accurate model of the
2018.05.011. human ventricles, PLoS One 12 (2017) e0188867, http://dx.doi.org/10.1371/
[246] J. Schaff, C.C. Fink, B. Slepchenko, J.H. Carson, L.M. Loew, A general compu- journal.pone.0188867.
tational framework for modeling cellular structure and function, Biophys. J. 73 [257] W. Gerstner, H. Sprekeler, G. Deco, Theory and simulation in neuroscience,
(1997) 1135–1146, http://dx.doi.org/10.1016/S0006-3495(97)78146-3. Science 338 (2012) 60–65, http://dx.doi.org/10.1126/science.1227356.
[247] S.D. Cohen, A.C. Hindmarsh, CVODE, a stiff/nonstiff ODE solver in C, Comput. [258] E. de Lange, Y. Xie, Z. Qu, Synchronization of early afterdepolarizations and
Phys. 10 (1996) 138–143, http://dx.doi.org/10.1063/1.4822377. arrhythmogenesis in heterogeneous cardiac tissue models, Biophys. J. 103
[248] J. Cooper, R.J. Spiteri, G.R. Mirams, Cellular cardiac electrophysiology mod- (2012) 365–373, http://dx.doi.org/10.1016/j.bpj.2012.06.007.
eling with Chaste and CellML, Front. Physiol. 5 (2015) http://dx.doi.org/10. [259] C.Y. Ko, M.B. Liu, Z. Song, Z. Qu, J.N. Weiss, Multiscale determinants of
3389/fphys.2014.00511. delayed afterdepolarization amplitude in cardiac tissue, Biophys. J. 112 (2017)
[249] A. Garny, D.P. Nickerson, J. Cooper, R. Weber dos Santos, A.K. Miller, S. 1949–1961, http://dx.doi.org/10.1016/j.bpj.2017.03.006.
McKeever, P.M. Nielsen, P.J. Hunter, Cellml and associated tools and tech- [260] Z. Song, C.Y. Ko, M. Nivala, J.N. Weiss, Z. Qu, Calcium-voltage coupling in the
niques, Phil. Trans. A Math. Phys. Eng. Sci. 366 (2008) 3017–3043, http: genesis of early and delayed afterdepolarizations in cardiac myocytes, Biophys.
//dx.doi.org/10.1098/rsta.2008.0094. J. 108 (2015) 1908–1921, http://dx.doi.org/10.1016/j.bpj.2015.03.011.
[250] T. Bray, J. Paoli, C.M. Sperberg-McQueen, E. Maler, Extensible Markup Lan- [261] H. Abriel, E. de Lange, J.P. Kucera, G. Loussouarn, M. Tarek, Computational
guage (XML)1.0. (5th Edn), 2000, available online at: www.w3.org/TR/REC- tools to investigate genetic cardiac channelopathies, Front. Physiol. 4 (2013)
xml. 390, http://dx.doi.org/10.3389/fphys.2013.00390.
[251] R. Ausbrooks, S. Buswell, D. Carlisle, S. Dalmas, S. Devitt, A. Diaz, M. [262] J. Bai, K. Wang, Y. Liu, Y. Li, C. Liang, G. Luo, S. Dong, Y. Yuan, H.
Froumentin, R. Hunter, P. Ion, M. Kohlhase, R. Miner, N. Poppelier, B. Smith, Zhang, Computational cardiac modeling reveals mechanisms of ventricular
N. Soiffer, R. Sutor, S. Watt, Mathematical markup language (MathML) version arrhythmogenesis in long QT syndrome type 8: CACNA1C R858H mutation
2.0 (second edition), 2001, available online at: www.w3.org/TR/MathML2. linked to ventricular fibrillation, Front. Physiol. 8 (2017) http://dx.doi.org/10.
[252] M. Clerx, P. Collins, E. de Lange, P.G. Volders, Myokit: A simple interface 3389/fphys.2017.00771, eCollection 2017.
to cardiac cellular electrophysiology, Prog. Biophys. Mol. Biol. 120 (2016) [263] A. Sadrieh, L. Domanski, J. Pitt-Francis, S.A. Mann, E.C. Hodkinson, C.A. Ng,
100–114, http://dx.doi.org/10.1016/j.pbiomolbio.2015.12.008. M.D. Perry, J.A. Taylor, D. Gavaghan, R.N. Subbiah, J.I. Vandenberg, A.P. Hill,
[253] N. Altanaite, J. Langguth, GPU-Based Acceleration of Detailed Tissue-Scale Multiscale cardiac modelling reveals the origins of notched t waves in long
Cardiac Simulations, GPGPU-11: General Purpose GPUs, February 24–28, 2018, QT syndrome type 2, Nature Commun. 5 (2014) http://dx.doi.org/10.1038/
Association for Computing Machinery, Vienna, Austria, New York, NY, USA, ncomms6069.
2018, p. 9.

22

You might also like