See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/273516089

Mathematical Models of Cardiac Pacemaking Function

Article in Frontiers in Physics · October 2013

DOI: 10.3389/fphy.2013.00020

CITATIONS READS

22 1,767

4 authors, including:

Mary M Maleckar Aslak Tveito

Allen Institute for Cell Science Simula Research Laboratory
79 PUBLICATIONS 2,117 CITATIONS 249 PUBLICATIONS 3,809 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Aslak Tveito on 26 April 2016.

The user has requested enhancement of the downloaded file.

 REVIEW ARTICLE
published: 30 October 2013
PHYSICS doi: 10.3389/fphy.2013.00020

Mathematical models of cardiac pacemaking function

Pan Li , Glenn T. Lines , Mary M. Maleckar and Aslak Tveito *
Center for Biomedical Computing, Simula Research Laboratory and Center for Cardiological Innovation, Oslo University Hospital, Oslo, Norway

Edited by: Over the past half century, there has been intense and fruitful interaction between
Leonid L. Rubchinsky, Indiana experimental and computational investigations of cardiac function. This has led, for
University - Purdue University, USA
example, to more profound understanding of cardiac excitation-contraction coupling;
Reviewed by:
how it works, as well as how it fails. However, many lines of inquiry remain
Elena Tolkacheva, University of
Minnesota, USA unresolved, among them the initiation of each heartbeat. The sinoatrial node, a cluster
Xiaopeng Zhao, University of of specialized pacemaking cells in the right atrium of the heart, spontaneously generates
Tennessee, USA an electro-chemical wave that spreads through the atria and the cardiac conduction system
*Correspondence: to the ventricles, initiating the contraction of cardiac muscle essential for pumping blood
Aslak Tveito, Simula Research
to the body. Despite the fundamental importance of this primary pacemaker, this process
Laboratory, Martin Linges vei 17,
Fornebu, P.O. Box 134, 1325, Oslo, is still not fully understood, and ionic mechanisms underlying cardiac pacemaking function
Norway are currently under heated debate. Several mathematical models of sinoatrial node cell
e-mail: [email protected] membrane electrophysiology have been constructed as based on different experimental
data sets and hypotheses. As could be expected, these differing models offer diverse
predictions about cardiac pacemaking activities. This paper aims to present the current
state of debate over the origins of the pacemaking function of the sinoatrial node. Here, we
will specifically review the state-of-the-art of cardiac pacemaker modeling, with a special
emphasis on current discrepancies, limitations, and future challenges.

Keywords: mathematical modeling, cardiac automaticity, pacemaker, ion channels, electrophysiology

1. INTRODUCTION been under heated debate, with seemingly exclusive hypotheses

Complex biological systems rely on numerous yet well- proposed by different groups (5). Following this debate, a num-
coordinated interactions at various scales to achieve proper ber of mathematical models of cardiac pacemaker cells have been
physiological function (1). In the heart, for example, complex sig- developed (3, 6). These models are based on different experimen-
naling pathways regulate chemical processes within the cytosol, tal data, different hypotheses, and, not surprisingly, they result in
cardiac ion channels mediate membrane fluxes, and nexus junc- diverse predictions about cardiac pacemaker activity (7). A sin-
tions between cardiac myocytes allow the tissue to function as a gle model or hypothesis has, thus far, been unable to provide
syncytium. Although governing principles describing e.g., kinet- convincing explanations which address all experimental findings.
ics for each component may be considered relatively simple, the Here, we will review the state-of-the-art of mathematical mod-
dynamics of the ensemble is anything but. The processes involved els of cardiac pacemaking function, with a special emphasis on
are highly non-linear and may be almost impossible to dissect current discrepancies, limitations and future challenges. It is
without the aid of mathematical modeling and computer simu- hoped that this review will allow readers a quick grasp of the
lations (2). Current cardiac models range from subcellular struc- core of the debate, and potentially attract a greater number of
tures to the entire organ and from aforementioned ion channels researchers with interdisciplinary backgrounds to tackle present
to signaling pathways (a variety of spatial scales), from Purkinje challenges.
cells to ventricular myocytes (a variety of subtypes of cardiac
cells), and from rat to human (a variety of species) (3). The inter- 2. MODELS OF CARDIAC PACEMAKING CELLS
actions between these diverse models and associated experimental In the healthy human heart, the sinoatrial node (SAN) is the
measurements have greatly advanced our basic understanding of primary pacemaker which spontaneously generates APs to ini-
cardiac physiology and pathophysiology. tiate each heartbeat. The location and size of SAN vary among
Yet after decades of modeling and experimental studies, one different species; in human, the SAN is typically located both lat-
of the most important questions in cardiac physiology remains erally and inferior to the terminal crest (crista terminalis), and
unanswered: what keeps the heart ticking (4)? The sinoatrial has been found to take the form of a cigar, with an estimated
node, a cluster of specialised pacemaking cells in the right atrium length of 20 mm and width of 5 mm (8, 9). Morphologies of
of the heart, spontaneously generates an electro-chemical wave isolated SAN cells can be either elongated and spindle-shaped
known as the action potential (AP). The AP spreads through- (about 150 μm long) or spider-shaped (about 50 μm long) (10).
out the atria and the cardiac conduction system to the ventricles, These cells are electrically coupled via gap junctions; the spatial
initiating the contraction of cardiac muscle essential for pump- distribution of gap junctions in the SAN is sparse as compared
ing blood to the body. The underlying mechanism of this cardiac to other cardiac tissues. This leads to relatively weak intercel-
spontaneous pacemaking activity (known as automaticity) has lular coupling and slow, non-uniform conduction through the

www.frontiersin.org October 2013 | Volume 1 | Article 20 | 1

Li et al. Modeling cardiac pacemaker function

SAN (11). Cardiac pacemaking is robust, however: in the case can be modeled via an electrical circuit, where the membrane
of SAN failure, secondary [atrioventricular node (AVN)] and is represented as a capacitance, and the voltage-gated channels
tertiary (Purkinje fiber network) pacemakers will ensure repeti- permitting the passage of ions through this membrane are
tive ventricular contractions (see Figure 1) (12). All pacemaker represented as electrical conductances.
tissues share certain cellular characteristics which distinguish
them from non-pacemaking cardiac myocytes; e.g., pacemak-
ing cells lack a T-tubular system (13) and share similar tissue-
specific expression levels of various ion channels (14, 15). It is
likely that they share also share similar underlying pacemaking
mechanisms.
Many models of cardiac pacemaking cells have been devel-
oped since the 1960s. A recent publication noted twenty-one
such mathematical models of pacemaking cells, including nine
Purkinje cell models, eleven SAN cell models and an AVN cell
model (3). All mathematical models of cardiac cellular electro-
physiology are based, at least in part, on the seminal electro-
physiological work of Hodgkin and Huxley in the giant squid
axon (16), which quantified the ionic mechanisms underlying
the neuronal AP. Based on their work, the cellular AP can be
conceptualized as a momentary, active change in the transmem-
brane electrical potential (the difference between intracellular and
extracellular electrical potentials) of an excitable membrane that
occurs when it is stimulated. In short, an AP relies on a “depo-
larization” phase, classically mediated by inward sodium current,
wherein the membrane is positively polarized with respect to
its resting state, followed by “repolarization,” wherein outward
potassium current(s) return the membrane to rest. A cardiac
pacemaking cell additionally experiences a “diastolic depolariza-
tion” (DD) phase, wherein the transmembrane potential becomes
progressively depolarized through time in the absence of stimu-
lus, eventually triggering an AP; see Figure 2B. For an automatic,
pacemaking cell (and indeed any cardiac cell), these processes

FIGURE 2 | (A) Model schematic of the Noble 1962 model, where the
FIGURE 1 | The cardiac conduction system and the cascade of cardiac Purkinje cell membrane is represented as an electrical circuit [Figure modified
pacemaking tissues; figure modified from (17). Green arrows indicate from (18)]. (B) Simulated AP and ion currents using the Noble 1962 Purkinje
the cardiac activation sequence during sinus rhythm. SAN, sinoatrial node; cell model to illustrate the IK 2 “decay” hypothesis. Black arrows indicate the
AVN, atrioventricular node; PF, Purkinje fibre; RA, right atrium; LA, left depolarization and repolarization phases during an AP. Blue arrows indicate
atrium; RV, right ventricle; LV, left ventricle; TV, tricuspid valve; MV, mitral the shift between inward and outward current balance during the DD that
valve. finally triggers a spontaneous AP.

Frontiers in Physics | Computational Physics October 2013 | Volume 1 | Article 20 | 2

Li et al. Modeling cardiac pacemaker function

Instead of providing complete review of the many extant pace- Functionally, INa is a fast inward current mediated by Na+ ions
maker cell models, we will instead categorize these based on their responsible for membrane depolarization, IK1 is the inward rec-
implied pacemaking hypotheses. For each mechanistic hypoth- tifier K + current that repolarizes the transmembrane potential to
esis, we will present and comment on the most representative its resting state, and IK2 is the K + -mediated pacemaker current.
mathematical models. The so-called “decay” hypothesis of pacemaking, as mediated by
IK2 , surmises that, during DD, the outward IK2 gradually decays
2.1. THE IK2 -BASED MODELS and “unmasks” inward sodium current, INa , to trigger a sponta-
The first cardiac cell model to describe a Purkinje cell AP and neous AP (as shown in Figure 2B). The membrane will repolarize
its automaticity was developed by Noble in 1962 (18). Following following the elicited AP via the outward IK2 current, and so on.
Hodgkin-Huxley formalism, the cell is represented as an electrical This cyclical activity is an oscillator known as the “membrane
circuit, where the membrane is represented as a capacitance, and oscillator,” as it describes cycles in the transmembrane potential,
voltage-gated ion channels are represented as electrical conduc- or simply as the membrane “clock” (M-clock).
tances (Figure 2A): In 1975, McAllister, Noble and Tsien constructed an updated
model of a cardiac Purkinje cell (the MNT model) based on newly
dVm available experimental data (19), and reformulated the pacemaker
Cm = − (INa + IK + IL ) . (1) current IK2 as:
dt
Here Cm is the membrane capacitance, Vm is the transmembrane IK2 = IK2 (Vm ) · s(Vm , t) (15)
potential, and three types of ionic currents are considered: a Na+
current (INa ), a K + current (IK ) and a non-specific leak (IL ) Here, IK2 is the maximum IK2 current, and s is the activation gate
current. More specifically, of IK2 . This reformulation was based on a number of experimen-
  tal studies that further characterized IK2 . For instance, the newer
INa = 400 m3 h + 0.14 (Vm − 40) , (2) data revealed that the role of IK2 during the AP plateau had been
overestimated, so an additional current Ix was introduced to rep-
where m and h are the activation and inactivation gates respec- resent the additional small contribution to the AP plateau from
tively, and ions other than K + . The kinetics of IK2 were also found to have
certain resemblance to those of IK1 , i.e., inward rectification fol-
dm
= αm (1 − m) − βm m, (3) lowing instantaneous changes in potential. In addition, for any
dt particular potential, it was suggested that IK2 was directly pro-
dh portional to a first-order variable (see Equation 5), instead of a
= αh (1 − h) − βh h, (4)
dt fourth-order variable as in the Noble 1962 model (see Equation
−0.1 (Vm + 48) (4)) (20).
αm = Vm +48
, (5) For over a decade, until the late 1970s, the IK2 “decay” hypoth-
e− 15 − 1 esis was the basis for modeling cardiac pacemaking activities. It
0.12 (Vm + 8) was supported by an array of experimental and theoretical papers;
βm = Vm +8
, (6)
e 5 −1 e.g., evidence suggesting that the catecholaminergic regulation of
Vm +90
IK2 could lead to the acceleration of the pacing rate induced by
αh = 0.17e− 20 , (7) sympathetic stimulation (21).
1
βh = . (8) 2.2. If -BASED HYPOTHESIS AND MODELS
− Vm10+42
e +1 In 1979, Brown et al. reported a novel current and its involve-
Furthermore, ment in the generation and catecholamine-induced acceleration
of SAN pacemaking activities (22). Know as the “funny” cur-
IK = IK1 + IK2 , (9) rent, If , it+ is +
considered so in a number of ways: (1) it has a
 Vm +90 Vm +90
 mixed Na -K current; (2) it activates on hyperpolarization, or
IK1 = IK1 (Vm ) = 1.2e− 50 + 0.015e 60 (Vm + 100), (10) negative deflection from the resting membrane potential (most
voltage-gated channels activate upon depolarization); (3) it has
IK2 = IK2 (Vm , t) = 1.2n (Vm , t)4 · (Vm + 100), (11) a very slow kinetics. All these properties are well-suited for If to
function as a pacemaker current; however, they share no similar-
where n is the activation gate, and ities with the properties of IK2 . While IK2 is an outward current
activated on depolarization, If is an inward current activated on
dn
= αn (1 − n) − βn n, (12) hyperpolarization. The community considered the simultaneous
dt evolution of two distinct yet mechanistically opposed pacemak-
−0.0001(Vm + 50) ing functions in the heart quite unlikely. Thus, the discovery of If
αn = , (13)
− Vm10+50 in SAN cells led to a series of experimental and modeling stud-
e −1
Vm +90
ies aimed at radical reinterpretation of IK2 (23). It was confirmed
βn = 0.002e− 80 . (14) that, although IK2 has a reversal potential near the expected K +

www.frontiersin.org October 2013 | Volume 1 | Article 20 | 3

Li et al. Modeling cardiac pacemaker function

equilibrium potential, it is, in fact, not a pure K + current. With generally-called hyperpolarization-activated cyclic nucleotide-
the application of Ba2+ , a K + current blocker, IK2 was converted gated (HCN) channels (26). They found that some features of
into an inward, hyperpolarization-activated current with a mixed HCN channel gating do not comply with classic Hodgkin-Huxley
Na+ and K + permeability, similar to that of If (24). kinetics, such as sigmoidal channel activation and deactivation.
In 1985, DiFrancesco and Noble published an updated Thus, they proposed an allosteric Markovian scheme of voltage
Purkinje cell model (the DN model) based on these new findings gating of HCN channels. In this scheme, HCN channels are con-
about If and the new interpretation of IK2 as the primary mecha- sidered to be composed of four subunits each. Each subunit is
nism underlying the M-clock (25). An activation gate, y, for If was gated independently by voltage, and possesses one voltage sensor
introduced, corresponding to the fully deactivated state of IK2 in that can be either in the “willing” state or the “reluctant” state.
the MNT model, so that These assumptions result in a 10-state model, with 5 open and
5 closed states. Using different sets of rate constants, the model
y (Vm , t) = 1 − s (Vm , t) , (16) is capable of quantitative representation of the voltage gating of
different HCN isoforms, and thus permits an interpretation of
and If is formulated as various kinetic properties of these isoforms. However, this sophis-
ticated HCN gating scheme has not yet been incorporated into
  Kc whole SAN cell models to investigate the cellular mechanisms
If = y(Vm , t) · Gf ,K (Vm − EK ) + Gf ,Na (Vm − ENa ) , (17)
Kc + Km underlying pacemaking activities.
In 2012, Severi et al. developed an updated model (the SD
where Kc is the cleft K + concentration, Km is half activation con- model) of a rabbit SAN cell based on the most recent exper-
centration, and EK and ENa are the reversal potentials for K + imental data, with a updated representation of intracellular
and Na+ , respectively. Instead of the decay of an outward cur- Ca2+ dynamics (27). If remains the major pacemaker cur-
rent (see Figure 2B), the gradual activation of the inward If gives rent in the Severi model, and it is formulated using a similar
rise to DD until reaching the voltage threshold for INa activation, Hodgkin-Huxley scheme as in the DN Purkinje cell model. As
as predicted by the DN model (see Figure 3). predicted by the SD model (see Figure 4), If and the Na+ −
In 2001, Altomare et al. combined both experimental and the- Ca2+ exchange current INCX are of similar size during DD.
oretical approaches to illustrate the limitation of Hodgkin-Huxley Yet If gradually increases to promote depolarization, while the
formalism in describing If , the current originating from the now inward INCX decays slightly over time. In addition, the model

FIGURE 3 | Simulated AP and ionic currents using the DiFrancesco FIGURE 4 | Simulated AP and ionic currents using the SD SAN model
Noble (DN) 1985 Purkinje cell model to illustrate the If hypothesis. to illustrate the If hypothesis. Blue arrows indicate the changes of If
Blue arrows indicate that the hyperpolarization-activated If during the during the DD finally triggers a spontaneous AP, while the contribution of
diastolic depolarization (DD) finally triggers a spontaneous AP. INCX is less important.

Frontiers in Physics | Computational Physics October 2013 | Volume 1 | Article 20 | 4

Li et al. Modeling cardiac pacemaker function

predicts that complete blockade of If will lead to cessation of cell concentrations, instead of a function of channel gating per
automaticity. Hodgkin-Huxley formalism (31). Along with this novel approach,
they presented a simple model of an SAN cell, where, as in the YNI
2.3. ICa -BASED HYPOTHESIS AND MODELS model, ICa is the pacemaker current driving cellular automaticity.
Shortly after the discovery of If in the SAN by (22), Yanagihara,
Noma and Irisawa published the very first SAN cell model (the 2.4. Ca2+ -CLOCK-BASED HYPOTHESIS AND MODELS
YNI model) (22, 28). One major difference between SAN and Motivated by the formidable success of the Hodgkin-Huxley
Purkinje cells is the expression level of Na+ channels. While a model, mathematical modeling of cardiac cells was first
strong INa is critical for Purkinje cells to maintain a rapid AP strongly focused on the cell membrane. Later, experi-
upstroke velocity and secure the robustness of electric conduction ments revealed that internal Ca2+ machinery is of great
(15), INa is very small in SAN cells, especially those localized at the importance for the action potential of cardiac cells and is
SAN center (29). In contrast to the If hypothesis (see Figure 3), essential for understanding excitation-contraction coupling.
the YNI model suggests that changes in K + current during DD are In addition to the aforementioned membrane oscillator (M-
quite moderate, and the major pacemaking current is ICa instead clock), an internal oscillator associated with Ca2+ cycling
of If . They propose that ICa plays a major role in generating both across the sarcoplasmic reticulum (SR) membrane (or
the AP and the pacemaker potential. Ca2+ “clock”) exists in all cardiac cells. In fact, this internal
The formulation of ICa in the YNI model is modified from clock can work independently of the sarcolemma (32).
Beeler and Reuter’s model of ventricular myocytes (30): Although various experimental studies had reported a possible
involvement of the Ca2+ -clock in cardiac pacemaking activi-
ICa = d (Vm , t) · f (Vm , t) · ICa (Vm ) (18) ties (33–37), it was only recently that mathematical models of
the SAN with integrated SR membranes and thus Ca2+ -clocks
where d and f represent the activation and inactivation of ICa , become available. In contrast to earlier membrane-delimited
respectively. As shown in Figure 5, ICa is the largest inward models of the SAN cell, in 2009, Maltsev and Lakatta (the ML
current and dominates both DD and AP upstroke, while the model) first developed a model of a coupled membrane- and
predicted contributions from If and INa are relatively small. Ca2+ -clock to address the ionic mechanism of cardiac pacemak-
Two decades later, Endresen et al. presented an alternative ing (see Figure 6) (32). Importantly, they quantitatively described
approach wherein membrane potential (Vm ) of cells could be the contribution of spontaneous Ca2+ release during late DD to
derived as a function of intracellular and extracellular ionic SAN pacemaking, by triggering an inward Na+ − Ca2+ exchange
current (INCX ). They concluded that only a coupled system of
membrane- and Ca2+ -clocks offers both the robustness and flexi-
bility that are required to maintain normal pacemaking function.
The ML model predicts that the most important pacemaker cur-
rent during late DD is the inward INCX , instead of If . Updated
versions of the ML model have recently become available to
accomodate additional experimental results (38, 39).
As shown in Figure 7, INCX is the leading current during the
DD, and is secondary to dynamic changes in [Ca]i and the Ca2+ -
clock. Without the Ca2+ -clock, the M-clock alone is not capable
of maintaining normal pacemaking function, as predicted by the
ML model (27).

3. THE “Ca2+ -CLOCK vs. If ” DEBATE

Although many ionic currents in pacemaker cells have been pro-
posed as potentials pacemaker currents, the present debate is
primarily split between two popular candidates: the Ca2+ -clock
and If (4). These two candidate hypotheses can be well repre-
sented by the ML and the SD models, respectively, and there exists
experimental evidence in support of or in opposition to each
hypothesis.
The expression level of HCN channels is much higher in pace-
making tissue as compared to elsewhere in heart (40); it is thus
reasonable to associate the tissue specificity of If with its role in
pacemaking function (23). During beta-adrenergic stimulation,
If can easily adjusted its working range to promote a faster heart
FIGURE 5 | Simulated AP and ionic currents using the YNI SAN model
rate by local cAMP-dependent regulation (41). Furthermore, If
to illustrate the ICa hypothesis. Blue arrows indicate that the changes in
ICa during diastolic depolarization finally trigger a spontaneous AP.
does not consume energy (at least directly), whereas the Ca2+ -
clock hypothesis relies on the SR Ca2+ pump to re-cycle Ca2+

www.frontiersin.org October 2013 | Volume 1 | Article 20 | 5

Li et al. Modeling cardiac pacemaker function

FIGURE 6 | Model schematic of the ML SAN cell model, figure rate global variable; RyR: cardiac ryanodine receptor; AP: action
modified from (32). Cai : Ca2+ in bulk cytosol; Casub : Ca2+ in the potential; IKr : rapid delayed rectifier current; Ist : sustained current; INCX :
subspace; SERCA: SR Ca2+ pump; CanSR : Ca2+ in the network SR NCX current; ICaT : T-type Ca2+ current; If : hyperpolarization-activated or
(nSR); CajSR : Ca2+ in the junctional SR (jSR); jSRCarel : SR Ca2+ release funny current.

against the concentration gradient across the SR membrane (42). have been implemented, e.g., combined voltage and Ca clamp-
These data would suggest If to be the primary pacemaker current ing, and these novel numerical approaches could be helpful to
in the SAN. However, there are experimental data to suggest that address the “Ca2+ -clock vs. If ” discussion in the realm of cardiac
pharmacological or genetic blockade of If only moderately alters pacemaking function.
the pacing rate, and neither blockade terminates the pacemaking
activity completely (43). 3.1. COMPARISON OF MODEL FORMULATIONS
Spontaneous Ca2+ release has been experimentally observed The ML and SD models offer diverse predictions about car-
during DD of SAN cells; the involvement of a Ca2+ -clock in diac pacemaking activities (e.g., the effect of If blockade), yet
pacemaking activities seems hard to dispute. One ultrastructural their model formulations have much in common. For exam-
property that all cardiac pacemaking tissues lack is a t-tubular ple, ion channels in both models are described using Hodgkin-
system, which is known to have major impact on subcellular Huxley formalism, with similar channel gating schemes. The
Ca2+ cycling. Could such lack of a defined t-tubular system be Ca2+ cycling subsystem is essentially identical between these two
considered characteristic of cardiac pacemaker tissues, as is the models. A major difference between ML and SD, however, is
expression of HCN channels? During beta-adrenergic stimula- the experimental data used to fit ion channel properties (27). In
tion, both the SR Ca2+ pump and RyR are phosphorylated to Figure 8, significant differences in If are shown between the two
accommodate the faster pacing rate, as is the case for If (44). models. The If channel availability (yss2 ) is much higher at a given
The major evidence against the Ca2+ -clock hypothesis is that pacemaking voltage range, e.g., at −60 mV, SD yss2 is about 2.5
either RyR blockade/knockout or the application of BAPTA to times the size of ML yss2 , while the difference in rate constants
suppress intracellular Ca2+ cycling does not have major effects on is much smaller. The difference in channel kinetics is consistent
pacemaking activities. Indeed, in some cases, the heart rate accel- with the difference in the functional role of If during APs. In
erates (45, 46). ATP consumption is likely to be substantial if basal Figure 9, phase plots of INCX and If against membrane voltage
pacemaking activities were mostly Ca2+ -clock driven (42). (V) are shown. It is straightforward to observe that, at pacemak-
In addition, it should be noted that similar questions concern- ing voltage ranges; i.e., −60 to −40 mV, that SD If can be up to
ing the respective roles of membrane and Ca2+ dynamics have two times larger than in ML. On the contrary, SD INCX is about
been raised for models of cardiac myocytes as well (47, 48). At half the size of that in ML in the pacemaking voltage range, yet
fast pacing rates, cardiac myocytes can develop alternans, chara- larger as compared to the ML INCX during AP upstroke. In other
terized by beat-to-beat alternations in both AP duration (APD) words, Ca2+ -mediated INCX is important to promote DD in the
and Ca transient. However, it remains unclear if APD alternans is ML model, and to enhance AP upstroke in the SD model.
driven by Ca2+ alternans or vice versa, since membrane and Ca2+ When there is little difference in model formulation, the exper-
subsystems are bidrectionally coupled, and it is experimentally imental data that was used for validation, especially the prop-
challenging to identify which subsystem is the follower. To resolve erties of key ion channels or receptors, will largely determine
this “chicken or the egg” debate, some computational protocols the ultimate behavior of a model. If such discrepancies between

Frontiers in Physics | Computational Physics October 2013 | Volume 1 | Article 20 | 6

Li et al. Modeling cardiac pacemaker function

debate between If and the Ca2+ -clock. On the other hand, incon-
sistencies among experimental measurements are inevitable,
assuming natural variations in experimental conditions, tissue
preparations, and existing limitations of experimental protocols.
So how can mathematical models be more helpful when there
are existing uncertainties in the experiments, e.g., the steady-state
activation of If (Figure 8)(left)? One option is to include model
constraints that can be quantitatively characterized by additional
experiments at tissue level, for example, the responses of SAN
and atrial tissue excitation patterns to If or Ca2+ blockade. These
tissue-level constraints are likely to help in evaluation of existing
single SAN cell models. Another option could be to keep uncer-
tain parameter values (e.g., the half-activation voltage of If ) open
within experimentally reported ranges, then identify the optimal
value(s) that can produce the best pacemaking performance in
terms of robustness, stability and flexibility.
As mentioned earlier, both the If and Ca2+ -clock hypothe-
ses are facing supporting and opposing experimental evidence.
It thus seems reasonable to hypothesize that these two mecha-
nisms are not necessarily exclusive. Their “redundant” coexistence
may be important to maintain the robustness of cardiac pace-
making function against possible disturbances through one’s
lifetime. This idea is supported by the fact that neither If
or Ca2+ blockade eliminates pacemaking activities entirely.
But it is also possible that these independent pacemaking
oscillators in the heart are assigned with different, complemen-
tary roles. It has recently been suggested that, while If is essential
for maintenance of the basal heart rate, the Ca2+ -clock is more
important for regulation of the heart rate during beta-adrenergic
stimulation (49). It may be worthwhile to take a step back from
the debate, and consider the question as a design problem: if one
could implement any theoretical pacemaker cell for the heart,
which would be ideal, a If -based M-clock, or a Ca2+ -clock, or
both? In other words, is it possible to define numerical protocols
to evaluate and compare the pacemaking performance of theoret-
ical pacemaker cells? And, if so, what are the ultimate properties
we are looking for in a pacemaker cell?
In 2013, Maltsev and Lakatta employed a different approach
to probe some of these questions (39). Based on the original ML
2009 model, they identified the minimal set of model components
(i.e., membrane ion currents and SR Ca2+ -dynamics - termed
simply the Ca2+ -clock) required to generate robust, flexible and
energy-efficient cardiac pacemaking function. Extensive parame-
ter sensitivity analyses were performed to evaluate the pacemak-
ing performance of 13 different model types, each with either
4 or 5 model components. They found that a minimal model
FIGURE 7 | Simulated AP and ionic currents using the ML SAN model
to illustrate the Ca clock hypothesis. Blue arrows indicate that the
with 4 components (ICaL + IKr + INCX + Ca2+ -clock) is capable
changes in INCX during the diastolic depolarization finally trigger a of producing the full range of autonomic modulation. Inclusion
spontaneous AP, while the contribution of If is much less. Inward INCX of If or ICaT increased model robustness, yet reduced model flex-
during the DD is secondary to spontaneous Ca release from the SR ibility. In addition, they quantified and compared the energy
(ISRCarel ), which gives rise to subspace and myoplasmic Ca concentrations
efficiency of the derived minimal model against earlier models,
([Ca]ss and [Ca]i ) elevation during the DD.
and suggested that by keeping a small Na+ influx, cell energy
required to maintain Na+ homeostasis can be largely reduced
mathematical models are primarily the result of different experi- to accommodate the energy associated with SR Ca2+ pumping.
mental data used to constrain these models in the first place, it is These numerical findings provided a number of valuable insights
logical to conclude that mechanistic insights gleaned using these into basal cardiac pacemaking function. More importantly, how-
models may be limited, especially in consideration of the current ever, these numerical approaches (parameter sensitivity analysis

www.frontiersin.org October 2013 | Volume 1 | Article 20 | 7

Li et al. Modeling cardiac pacemaker function

2 ) (left) and activation rate constant (y

FIGURE 8 | Differences in If steady state activation (yss tau ) (middle) between the ML and SD model. (right) the
ratios between SD and ML yss 2 , and between SD and ML y
tau .

effects of cAMP on channel properties. Furthermore, Vs is

described as:
 [cAMP] 
1+ Ko
Vs[cAMP] = v · ln [cAMP]
(20)
1+ Kc

where Ko and Kc are the cAMP dissociation constants when

the channel is open and closed, respectively. As local [cAMP]
increases, the channel open probability curve will shift to more
positive voltages, e.g., about 14 mV shift with 10 μM [cAMP]
FIGURE 9 | Differences in INCX (left) and If (right) between the ML and (41), thus increasing the current availability during DD, and
SD model. Computer simulations are performed over a number of accelerating the pacing rate.
pacemaking cycles using default parameter settings. Later, a more complete βAR regulation system was modeled
in ventricular myocytes, then incorporated into existing SAN cell
models (see Figure 10) (50–52). These models describe both the
and quantitative pacemaking performance evaluation) should βAR signaling cascade and changes to electrophysiological sub-
prove useful toward reaching a generalized theory of cardiac strates after phosphorylation, and have been very useful tools to
pacemaking. quantify the ionic mechanisms underlying the rate adaptation of
SAN, e.g., using the lead potential (VL ) analysis proposed by Sarai
4. MODELING β-ADRENERGIC REGULATION OF THE SAN et al. (53). More recently, βAR system models that describe the
Under the conditions of exercise or stress, heart rate increases complexity of localized signaling domains have become available
to accommodate the needs of the body, achieved via β- (54). This level of detail has not been included in most recent SAN
adrenergic regulation (βAR) of the SAN. During βAR, there cell models, e.g., the ML and SD models, yet very well might be a
are a variety of functional modifications of ion channels critical piece leading to resolution of the Ca2+ -clock vs If debate.
and transporters, and the properties of both membrane and
Ca2+ -clocks are altered. It is thus quite difficult to evaluate 5. TISSUE LEVEL COMPLICATIONS OF SAN PACEMAKING
the contribution of each ion current to heart rate regulation FUNCTION
experimentally during βAR. The SAN achieves its physiological function via electrical cou-
It has long been known that If can be activated by direct pling with atrial tissue (29, 55). To illustrate the effects of SAN-
binding of cyclic AMP (cAMP), a second messenger impor- atrial coupling on pacemaking patterns, a simplified 2D SAN
tant in βAR. This voltage-independent regulation of If is model is implemented on a 1 by 1 cm grid, with the elliptically
achieved by shifting the steady state activation curve accord- shaped SAN in the center (half-axes 3 and 1 mm, see Figure 11A).
ing to the local concentration of cAMP. In 1999, DiFrancesco Here, we use the SD model (27), and the model developed by
proposed an allosteric reaction model to describe the dual Malecker et al. (56), to describe the electrophysiological prop-
activation of If by voltage hyperpolarization and cAMP (41). erties of SAN and atrial cells, respectively. Electric diffusion in
Specifically, the steady state open probability of If (Po ) is the whole 2D tissue is anisotropic, with atrial tissue fibre orien-
described as: tation pointing 30◦ off from the SAN long axis. The conductivity
is 1.2 m/cm along the fibre direction, and 0.25 ms/cm trasversely
1 (57, 58). The cell-area-to-volume ratio is 3000 cm−1 . When the
Po = Vm −V1/2 −Vs
(19)
1+e v coupling is weak (about 40% of the control value), atrial tissue
fails to respond to every SAN beat; when the coupling is strong
where V1/2 is the half-activation voltage, v is the inverse slope ( the control value), SAN and atrial excitations are in synchrony
factor, and Vs is a cAMP-dependent term that represents the with a slower pacing rate (Figure 11B); when the coupling is too

Frontiers in Physics | Computational Physics October 2013 | Volume 1 | Article 20 | 8

Li et al. Modeling cardiac pacemaker function

FIGURE 10 | Model schematic of a SAN cell model with integrated βAR system, figure modified from (52).

FIGURE 11 | (A) Model schematic of a simplified 2D SAN model. proximal (at 0 cm) and distal (at 0.5 cm) sites correspond to the
(B) Space-time plots of SAN and atrial excitation patterns when central SAN cell and peripheral atrial cell respectively, indicated as
the SAN is either weakly or well coupled to atrial tissue. The black dots in (A).

strong (more than 250% of the control value), the whole 2D tissue enhanced electrotonic (electrical sink) effects. With either If or
became quiescent. Figure 12 further illustrates the relationship Ca2+ blockade (by adding strong Ca2+ buffers), beats-per-minute
between tissue pacing frequency and SAN-atrial coupling, and the (BPM) of both the SAN and atrial tissue are largely reduced.
effects of If or Ca2+ blockade. Specifically, when the SAN-atrial Proper coupling to non-pacemaking tissue is clearly a delicate bal-
coupling is below a certain threshold, the SAN beats much faster ance and is essential to achieve robust and rhythmic whole-heart
than surrounding atrial tissue and the stimulus-to-response ratio excitation.
is increased; when the coupling is increased, the SAN and atrial To further illustrate the importance and complexities of SAN-
tissue activate in concert, yet at slower rates. When the coupling atrial interaction, we implemented a 3D atrial model to quantify
is above a certain threshold, the SAN remains quiescent due to the effects of weak SAN-atrial coupling on cardiac excitation

www.frontiersin.org October 2013 | Volume 1 | Article 20 | 9

Li et al. Modeling cardiac pacemaker function

FIGURE 12 | The relationship between pacing rates (in

beats-per-minute (BPM)) and the SAN-atrial electric coupling, and the
effects of If or Ca blockade. Solid lines: the central SAN cell. Dashed lines:
the peripheral atrial cell.
FIGURE 13 | Simulated irregular SAN and atrial excitation patterns
induced by the weak SAN-atrial coupling in a 3D atrial model. Only 3
patterns. The 3D mesh of right atria and SAN is a courtesy of (#1, #4, and #7) out of 7 SAN excitations successfully activate the atrial
tissue.
Dr Edward J. Vigmond, Université de Bordeaux. The mesh com-
prised 315986 nodes, and 1065333 tetrahedral elements with an
average edge length of 100 μm [please see (59) for more details]. system (13). Without the t-tubular system, the coupling between
Same cell models are used as in the 2D model (27, 56). The membrane and Ca2+ subsystmes is relatively weak, since most
conductivity is zero in the block zone adjacent to the SAN, and local Ca2+ -release machineries are not located in conjunction
1 ms/cm elsewhere. Within 500 ms of simulation time, there are with the cell membrane, and do not directly respond to changes
seven spontaneous SAN excitations, while only three of these in membrane polarization. Thus consideration of the ultrastruc-
successfully trigger atrial excitation (#1, #4, and #7). In addi- tural details of SAN cells is critical to a quantitative understanding
tion, chaotic dynamics at the SAN pacemaking site and dyssyn- of SAN pacemaking function. The importance of the spatial hier-
chronous SAN excitations are observed (#2, #3, #5, #6); see archy of subcellular Ca2+ patterns in determining the behaviors
Figure 13. The complex tissue dynamics seen in this example of Ca2+ -clock, i.e., from localized Ca2+ sparks to global Ca2+
highlight the difficulty of employing only single-cell studies to waves, is now clear. Despite this fact, the subcellular heterogene-
discern SAN pacemaking function. 3D tissue effects are important ity of SAN cells has not yet been carefully considered in model
to consider in any study of the cardiac pacemaking mechanism, construction and spatially distributed models are not yet available
and may serve as additional constraints to evaluate SAN single (66, 67).
cell models. For example, when coupled to atrial tissue, SAN
cell models should be capable of producing robust and rhythmic 6. SUMMARY
atrial excitations at the complete range of physiological pac- The underlying mechanism of cardiac pacemaking activity has
ing rates. However, none of the current SAN cell models have historically been under debate, often with seemingly exclusive
been systematically evaluated with respect to these tissue-level hypotheses proposed by different groups. In this review, we have
constraints. presented these hypotheses and discussed the developments in
Current multi-scale models of the SAN integrate anatomi- mathematical models of cardiac pacemaking function in detail.
cal details of SAN heterogeneity and fiber direction. This per- A first group of models supports a driving M-clock (mem-
mits quantitative description of the functional heterogeneity and brane oscillator). For over a decade until the late 1970s, the IK2
synchronization within the SAN, investigation of the role of “decay” hypothesis was the basis for cardiac pacemaker mod-
SAN dysfunction in atrial arrhythmias, and characterization of eling (Noble 1962 SAN model and its iterations). In 1979, the
the complex interactions among SAN, atrial tissue and, more, discovery of If (HCN channels) in SAN cells led to a series of
recently AVN tissue (9, 59–65). Yet most of these models are con- experimental and modeling studies aimed at radical reinterpre-
structed by scaling-up existing compartmental single cell models tation of the IK2 -based hypothesis, leading to the debut of the
to tissue and organ levels based on either simplified or realistic If -based hypothesis. The DN Purkinje cell model (1985) relies
geometries of the SAN. However, the subcellular heterogeneity of on If as the current underlying the M-clock. Further experi-
SAN cells is much more pronounced as compared to ventricular mental data regarding sarcolemmal currents and Ca2+ cycling
or even atrial myocytes, due to the lack of a well-defined t-tubular are both incorporated into the recent (2012) SD model; If and

Frontiers in Physics | Computational Physics October 2013 | Volume 1 | Article 20 | 10

Li et al. Modeling cardiac pacemaker function

thus the M-clock remains the major pacemaking mechanism. The A number of natural further directions extends the trajec-
YNI model published in 1980, in contrast, suggests that changes tory of pacemaker cell model development. Critically, a pro-
in K + current during DD are quite moderate, and that the major posed Markovian scheme for HCN channels was put forward by
pacemaking current is ICa instead of If . Although focused on Altomare in 2001, but has yet to be incorporated into whole cell
the sarcolemmal current, this first suggestion of the importance models. Furthermore, subcellular heterogeneity of SAN cells is
of Ca2+ to cardiac cell automaticity may be seen as prescient; much more pronounced as compared to other cardiac cell types;
in addition to the M-clock, the SR-membrane-dependent Ca2+ - consideration of ultrastructural details of SAN cells may be essen-
clock is also involved for cardiac pacemaking. In 2009, Maltsev tial to a quantitative understanding of SAN pacemaking function.
and Lakatta debuted the first coupled M- and Ca2+ -clock mathe- In addition, under conditions of exercise or stress, our heart rate
matical model (ML) and suggested that a coupled system offered increases to accommodate the needs of the body, achieved via βAR
both the flexibility and robustness required to maintain normal regulation of the SAN. This level of detail has not been included
pacemaking function. ML predicts that the most important pace- in the most recent SAN cell models, yet very well may also be a
maker current during late DD is the inward INCX , rather than If , critical piece leading to resolution of the M- vs Ca2+ -clock anal-
as for SD. It is reasonable to conjecture that these two mecha- ysis. Finally, it is important to consider that the SAN is not an
nisms may not be mutually exclusive. Indeed, a recent study shows isolated cell; rather it achieves its physiological function via elec-
that these could be complementary, and together generate robust, trical coupling with atrial tissue. Our simulations showed that,
flexible, and energy-efficient cardiac pacemaking (39). when coupling between SAN and atrial cells is weak, atrial tissue
ML and SD may offer diverse predictions about cardiac pace- fails to respond to every SAN beat. However, when coupling is
making activities, yet their model formulations have much in strong, SAN and atrial excitations are in synchrony with a slower
common. Their distinct emergent behaviors result from the pacing rate. When coupling is above a certain threshold, the SAN
diverse data sets upon which they were built. It is actually elemen- remains quiescent due to enhanced electrotonic effects. Clearly,
tary, from a methodological perspective, to incorporate new data an appreciation of both the SAN’s electrophysiology as well as its
into these models. When new experimental knowledge becomes delicate relationship to surrounding tissue is essential to under-
available, it is thus tempting to augment extant models by default. standing the mechanisms of robust and rhythmic whole-heart
However, it is essential to first consider how model expansion excitation.
leading to a more comprehensive model will help us to fur-
ther understand SAN function. Alternatively, is it possible to ACKNOWLEDGMENTS
use employ novel experimental data as additional model con- This work is supported by the Research Council of Norway,
straints to simplify existing models? Indeed, as compared to through which Simula is supported via its partnership in the
model expansion, model reduction can be more challenging (68). Center for Cardiological Innovation.

REFERENCES background. Circ Res. (2010) (1998) 82:604–12. doi: 10.1161/ (2012) 110:1525–37. doi: 10.1161/
1. Noble D. Systems biology and the 106:1791–7. doi: 10.1161/ 01.RES.82.5.604 CIRCRESAHA.111.260026
heart. Biosystems (2006) 83:75–80. CIRCRESAHA.110.218875 12. Irisawa H. Comparative phys- 18. Noble D. A modification of the
doi: 10.1016/j.biosystems.2005. 8. Dobrzynski H, Boyett MR, iology of the cardiac pacemaker Hodgkin–Huxley equations appli-
05.013 Anderson RH. New insights mechanism. Physiol Rev. (1978) cable to Purkinje fibre action and
2. Kitano H. Computational sys- into pacemaker activity pro- 58:461–98. pacemaker potentials. J Physiol.
tems biology. Nature (2002) moting understanding of sick 13. Brette F, Orchard C. T-tubule (1962) 160:317–52.
420:206–10. doi: 10.1038/ sinus syndrome. Circulation function in mammalian cardiac 19. McAllister RE, Noble D, Tsien
nature01254 (2007) 115:1921–32. doi: 10.1161/ myocytes. Circ Res. (2003) R. Reconstruction of the
3. Noble D, Garny A, Noble PJ. CIRCULATIONAHA.106.616011 92:1182–92. doi: 10.1161/01.RES. electrical activity of cardiac
How the Hodgkin–Huxley 9. Seemann G, Höper C, Sachse FB, 0000074908.17214.FD Purkinje fibres. J Physiol. (1975)
equations inspired the cardiac Dössel O, Holden AV, Zhang H. 14. Musa H, Lei M, Honjo H, Jones 251:1–59.
Physiome Project. J Physiol. Heterogeneous three-dimensional SA, Dobrzynski H, Lancaster MK, 20. Noble D, Tsien R. The kinet-
(2012) 590:2613–28. doi: 10.1113/ anatomical and electrophysiologi- et al. Heterogeneous expression ics and rectifier properties of the
jphysiol.2011.224238 cal model of human atria. Philos of Ca2+ handling proteins in rab- slow potassium current in cardiac
4. Billman GE. Does the ‘cou- Trans R Soc A Math Phys Eng Sci. bit sinoatrial node. J Histochem Purkinje fibres. J Physiol. (1968)
pled clock’ make the heart tick? (2006) 364:1465–81. doi: 10.1098/ Cytochem. (2002) 50:311–24. doi: 195:185–214.
Cardiovasc. Res. (2012) 96:343–4. rsta.2006.1781 10.1177/002215540205000303 21. Hauswirth O, Noble D, Tsien
doi: 10.1093/cvr/cvs300 10. Dobrzynski H, Li J, Tellez 15. Boyden PA, Hirose M, R. Adrenaline: mechanism of
5. Lakatta EG, DiFrancesco D. What J, Greener I, Nikolski V, Dun W. Cardiac Purkinje action on the pacemaker poten-
keeps us ticking: a funny current, a Wright S, et al. Computer cells. Heart Rhythm (2010) tial in cardiac Purkinje fibers.
calcium clock, or both? J Mol Cell three-dimensional reconstruc- 7:127–35. doi: 10.1016/j.hrthm. Science (1968) 162:916–7. doi:
Cardiol. (2009) 47:157–70. doi: tion of the sinoatrial node. 2009.09.017 10.1126/science.162.3856.916
10.1016/j.yjmcc.2009.03.022 Circulation (2005) 111:846–54. 16. Hodgkin AL, Huxley AF. A quan- 22. Brown H, DiFrancesco D,
6. Wilders R. Computer modelling doi: 10.1161/01.CIR.0000152100. titative description of membrane Noble S. How does adrenaline
of the sinoatrial node. Med Biol 04087.DB current and its application to con- accelerate the heart? Nature
Eng Comput. (2007) 45:189–207. 11. Kwong KF, Schuessler RB, Green duction and excitation in nerve. J (1979) 280:235–36. doi: 10.1038/
doi: 10.1007/s11517-006-0127-0 KG, Laing JG, Beyer EC, Boineau Physiology. (1952) 117:500. 280235a0
7. Noble D, Noble PJ, Fink M. JP, et al. Differential expres- 17. Munshi NV. Gene regulatory 23. DiFrancesco D. The role of the
Competing oscillators in sion of gap junction proteins in networks in cardiac conduction funny current in pacemaker activ-
cardiac pacemaking historical the canine sinus node. Circ Res. system development. Circ Res. ity. Circ Res. (2010) 106:434–46.

www.frontiersin.org October 2013 | Volume 1 | Article 20 | 11

Li et al. Modeling cardiac pacemaker function

doi: 10.1161/CIRCRESAHA.109. of sinoatrial pacemaker cells do rabbit SA node. J Physiol. (1999) S. Simulation analysis of intra-
208041 not require membrane depolar- 515:367–76. doi: 10.1111/j.1469- cellular Na+ and Cl- homeostasis
24. DiFrancesco D. A new interpre- ization. Circ Res. (2004) 94:802–9. 7793.1999.367ac.x during β1-adrenergic stimulation
tation of the pace-maker current doi: 10.1161/01.RES.0000122045. 42. Matsuoka S, Sarai N, Jo H, Noma of cardiac myocyte. Prog Biophys
in calf Purkinje fibres. J Physiol. 55331.0F A. Simulation of ATP metabolism Mol Biol. (2008) 96:171–86.
(1981) 314:359–76. 34. Vinogradova TM, Maltsev VA, in cardiac excitation–contraction doi: 10.1016/j.pbiomolbio.2007.
25. DiFrancesco D, Noble D. A Bobdanov KY, Lyashkov AE, coupling. Prog Biophys 07.005
model of cardiac electrical activ- Lakatta EG. Rhythmic Ca2+ Mol Biol. (2004) 85:279–99. 52. Himeno Y, Sarai N, Matsuoka S,
ity incorporating ionic pumps oscillations drive sinoatrial nodal doi: 10.1016/j.pbiomolbio.2004. Noma A. Ionic Mechanisms
and concentration changes. cell pacemaker function to make 01.006 underlying the positive
Philos Trans R Soc Lond B Biol the heart tick. Ann NY Acad 43. Bucchi A, Barbuti A, DiFrancesco chronotropy induced by β1-
Sci. (1985) 307:353–98. doi: Sci. (2005) 1047:138–56. doi: D, Baruscotti M. Funny current adrenergic stimulation in
10.1098/rstb.1985.0001 10.1196/annals.1341.013 and cardiac rhythm: insights guinea-pig sinoatrial node
26. Altomare C, Bucchi A, Camatini 35. Méry A, Aimond F, Ménard C, from HCN knockout and trans- cells: a simulation study. J Physiol
E, Baruscotti M, Viscomi C, Mikoshiba K, Michalak M, Pucéat genic mouse models. Front Sci. (2008) (0):0801180027. doi:
Moroni A, et al. Integrated M. Initiation of embryonic Physiol. (2012) 3:240. doi: 10.2170/physiolsci.RP015207
allosteric model of voltage gat- cardiac pacemaker activity by 10.3389/fphys.2012.00240 53. Sarai N, Matsuoka S, Kuratomi
ing of HCN channels. J Gen inositol 1, 4, 5-trisphosphate– 44. Vinogradova TM, Bogdanov S, Ono K, Noma A. Role
Physiol. (2001) 117:519–32. doi: dependent calcium signaling. Mol KY, Lakatta EG. β-Adrenergic of individual ionic current sys-
10.1085/jgp.117.6.519 Biol Cell. (2005) 16:2414–23. doi: stimulation modulates ryan- tems in the SA node hypothe-
27. Severi S, Fantini M, Charawi LA, 10.1091/mbc.E04-10-0883 odine receptor Ca2+ release sized by a model study. Jpn
DiFrancesco D. An updated com- 36. Lakatta EG, Vinogradova T, during diastolic depolarization J Physiol. (2003) 53:125–34. doi:
putational model of rabbit sinoa- Lyashkov A, Sirenko S, Zhu W, to accelerate pacemaker activity 10.2170/jjphysiol.53.125
trial action potential to investi- Ruknudin A, et al. The integration in rabbit sinoatrial nodal cells. 54. Heijman J, Volders PG, Westra
gate the mechanisms of heart rate of spontaneous intracellular Ca2+ Circ Res. (2002) 90:73–9. doi: RL, Rudy Y. Local control
modulation. J Physiol. (2012) cycling and surface membrane 10.1161/hh0102.102271 of β-adrenergic stimulation:
590:4483–99. doi: 10.1113/jphys- ion channel activation entrains 45. Lappi MD, Billman GE. Effect effects on ventricular myocyte
iol.2012.229435 normal automaticity in cells of of ryanodine on ventricular electrophysiology and Ca tran-
28. Yanagihara K, Noma A, Irisawa the heart’s pacemaker. Ann NY fibrillation induced by myocar- sient. J Mol Cell Cardiol. (2011)
H. Reconstruction of sino-atrial Acad Sci. (2006) 1080:178–206. dial ischaemia. Cardiovasc 50:863–71. doi: 10.1016/j.yjmcc.
node pacemaker potential based doi: 10.1196/annals.1380.016 Res. (1993) 27:2152–9. doi: 2011.02.007
on the voltage clamp experiments. 37. Vinogradova TM, Lyashkov AE, 10.1093/cvr/27.12.2152 55. Garny A, Kohl P, Hunter PJ, Boyett
Jpn J Physiol. (1980) 30:841. doi: Zhu W, Ruknudin AM, Sirenko 46. Himeno Y, Toyoda F, Satoh H, MR, Noble D. One-Dimensional
10.2170/jjphysiol.30.841 S, Yang D, et al. High basal Amano A, Cha CY, Matsuura rabbit sinoatrial node mod-
29. Zhang H, Holden A, Kodama protein kinase A–dependent H, et al. Minor contribution els. J Cardiovasc Electrophysiol.
I, Honjo H, Lei M, Varghese phosphorylation drives rhythmic of cytosolic Ca2+ transients to (2003) 14:S121–S132. doi:
T, et al. Mathematical mod- internal Ca2+ store oscillations the pacemaker rhythm in guinea 10.1046/j.1540.8167.90301.x
els of action potentials in the and spontaneous beating of pig sinoatrial node cells. Am 56. Maleckar MM, Greenstein JL,
periphery and center of the cardiac pacemaker cells. Circ Res. J Physiol Heart Circ Physiol. Giles WR, Trayanova NA. K+
rabbit sinoatrial node. Am J (2006) 98:505–14. doi: 10.1161/ (2011) 300:H251–H261. doi: current changes account for
Physiol Heart Circ Physiol. (2000) 01.RES.0000204575.94040.d1 10.1152/ajpheart.00764.2010 the rate dependence of the
279:H397–H421. Available online 38. Yaniv Y, Stern MD, Lakatta EG, 47. Tolkacheva EG, Zhao X. action potential in the human
at: http://ajpheart.physiology. Maltsev VA. Mechanisms of Nonlinear dynamics of peri- atrial myocyte. Am J Physiol
org/content/279/1/H397 beat-to-beat regulation of car- odically paced cardiac tissue. Heart Circul Physiol. (2009)
30. Beeler GW, Reuter H. diac pacemaker cell function by Nonlin Dyn. (2012) 68:347–63. 297:H1398–H1410. doi: 10.1152/
Reconstruction of the action Ca cycling dynamics. Biophys doi: 10.1007/s11071-011-0241-x ajpheart.00411.2009
potential of ventricular myocar- J. (2013) 105:1551–61. doi: 48. Krogh-Madsen T, Christini 57. Klepfer RN, Johnson CR, Macleod
dial fibres. J Physiol. (1977) 10.1016/j.bpj.2013.08.024 DJ. Nonlinear dynamics in RS. The effects of inhomogeneities
268:177–210. 39. Maltsev VA, Lakatta EG. cardiology. Ann Rev Biomed and anisotropies on electrocar-
31. Endresen L, Hall K, Høye J, Numerical models based on Eng. (2012) 14:179–203. doi: diographic fields: a 3-D finite-
Myrheim J. A theory for the mem- a minimal set of sarcolemmal 10.1146/annurev-bioeng-071811- element study. Biomed Eng
brane potential of living cells. Eur electrogenic proteins and an 150106 IEEE Trans. (1997) 44:706–19.
Biophys J . (2000) 29:90–103. doi: intracellular Ca clock generate 49. Gao Z, Chen B, Joiner MlA, doi: 10.1109/10.605427
10.1007/s002490050254 robust, flexible, and energy- Wu Y, Guan X, Koval OM, 58. Henriquez CS, Muzikant AL,
32. Maltsev VA, Lakatta EG. efficient cardiac pacemaking. et al. If and SR Ca release both Smoak CK. Anisotropy, fiber
Synergism of coupled subsar- J Mol Cell Cardiol. (2013). contribute to pacemaker activ- curvature, and bath loading
colemmal Ca2+ clocks and 59:181–95. doi: 10.1016/j.yjmcc. ity in canine sinoatrial node effects on activation in thin and
sarcolemmal voltage clocks 2013.03.004 cells. J Mol Cell Cardiol. (2010) thick cardiac tissue preparations.
confers robust and flexible 40. Shi W, Wymore R, Yu H, Wu 49:33–40. doi: 10.1016/j.yjmcc. J Cardiovasc Electrophysiol. (1996)
pacemaker function in a novel J, Wymore RT, Pan Z, et al. 2010.03.019 7:424–44. doi: 10.1111/j.1540-
pacemaker cell model. Am J Distribution and prevalence 50. Saucerman JJ, Brunton LL, 8167.1996.tb00548.x
Physiol Heart Circ Physiol. (2009) of hyperpolarization-activated Michailova AP, McCulloch AD. 59. Muñoz MA, Kaur J, Vigmond
296:H594–H615. doi: cation channel (HCN) mRNA Modeling β-adrenergic control EJ. Onset of atrial arrhythmias
10.1152/ajpheart.01118.2008 expression in cardiac tissues. of cardiac myocyte contrac- elicited by autonomic modula-
33. Vinogradova TM, Zhou YY, Circ Res. (1999) 85:e1–e6. doi: tility in silico. J Biol Chem. tion of rabbit sinoatrial node
Maltsev V, Lyashkov A, Stern 10.1161/01.RES.85.1.e1 (2003) 278:47997–8003. doi: activity: a modeling study. Am
M, Lakatta EG. Rhythmic ryan- 41. DiFrancesco D. Dual allosteric 10.1074/jbc.M308362200 J Physiol Heart Circ Physiol.
odine receptor Ca2+ releases modulation of pacemaker (f) 51. Kuzumoto M, Takeuchi A, Nakai (2011) 301:H1974–H1983. doi:
during diastolic depolarization channels by cAMP and voltage in H, Oka C, Noma A, Matsuoka 10.1152/ajpheart.00059.2011

Frontiers in Physics | Computational Physics October 2013 | Volume 1 | Article 20 | 12

Li et al. Modeling cardiac pacemaker function

60. Michaels DC, Matyas EP, Jalife model study. PLoS Comput microscopic dyadic function. Received: 29 August 2013; accepted:
J. Mechanisms of sinoatrial Biol. (2010) 6:e1001041. doi: Biophys J. (2011) 100:2904–12. 08 October 2013; published online: 30
pacemaker synchronization: a new 10.1371/journal.pcbi.1001041 doi: 10.1016/j.bpj.2011.05.031 October 2013.
hypothesis. Circ Res. (1987) 64. Chandler N, Aslanidi O, 67. Nivala M, de Lange E, Rovetti Citation: Li P, Lines GT, Maleckar MM
61:704–14. doi: 10.1161/01.RES. Buckley D, Inada S, Birchall R, Qu Z. Computational and Tveito A (2013) Mathematical
61.5.704 S, Atkinson A, et al. Computer modeling and numerical models of cardiac pacemaking func-
61. Garny A, Noble D, Kohl three-dimensional anatom- methods for spatiotemporal tion. Front. Physics 1:20. doi: 10.3389/
P. Dimensionality in car- ical reconstruction of the calcium cycling in ventricular fphy.2013.00020
diac modelling. Progress human sinus node and a novel myocytes. Front Physiol. (2012) This article was submitted to
Biophys Mol Biol. (2005) paranodal area. Anat Record 3:114. doi: 10.3389/fphys.2012. Computational Physics, a section
87:47–66. doi: 10.1016/j. (2011) 294:970–9. doi: 10.1002/ 00114 of the journal Frontiers in Physics.
pbiomolbio.2004.06.006 ar.21379 68. Kohl P, Hunter P, Winslow R. Copyright © 2013 Li, Lines, Maleckar
62. Chandler NJ, Greener ID, Tellez 65. Podziemski P, Zebrowski JJ. Model interactions: it is the and Tveito. This is an open-access
JO, Inada S, Musa H, Molenaar A simple model of the right simple, which is so difficult. Prog article distributed under the terms
P, et al. Molecular architec- atrium of the human heart Biophys Mol Biol. (2011) 107:1–3. of the Creative Commons Attribution
ture of the human sinus node with the sinoatrial and atri- doi: 10.1016/j.pbiomolbio.2011. License (CC BY). The use, distribution
insights into the function of the oventricular nodes included. J 07.003 or reproduction in other forums is per-
cardiac pacemaker. Circulation Clin Monitor Comput. (2013) mitted, provided the original author(s)
(2009) 119:1562–75. doi: 10.1161/ 27:481–98. doi: 10.1007/s10877- Conflict of Interest Statement: The or licensor are credited and that the
CIRCULATIONAHA.108.804369 013-9429-6 authors declare that the research was original publication in this journal is
63. Oren RV, Clancy CE. 66. Gaur N, Rudy Y. Multiscale conducted in the absence of any com- cited, in accordance with accepted aca-
Determinants of heterogene- modeling of calcium cycling mercial or financial relationships that demic practice. No use, distribution or
ity, excitation and conduction in cardiac ventricular myocyte: could be construed as a potential con- reproduction is permitted which does
in the sinoatrial node: a macroscopic consequences of flict of interest. not comply with these terms.

www.frontiersin.org October 2013 | Volume 1 | Article 20 | 13

View publication stats