Authors: Konstantin Hennis, Chiara Piantoni, Martin Biel, Stefanie Fenske, Christian Wahl-Schott
Categories: Reviews, action potentials, autonomic nervous system, heart rate, myocytes, cardiac, sinoatrial node, 10013, 10014
Source: Circulation Research
Loss or dysregulation of the normally precise control of heart rate via the autonomic nervous system plays a critical role during the development and progression of cardiovascular disease—including ischemic heart disease, heart failure, and arrhythmias. While the clinical significance of regulating changes in heart rate, known as the chronotropic effect, is undeniable, the mechanisms controlling these changes remain not fully understood. Heart rate acceleration and deceleration are mediated by increasing or decreasing the spontaneous firing rate of pacemaker cells in the sinoatrial node. During the transition from rest to activity, sympathetic neurons stimulate these cells by activating β-adrenergic receptors and increasing intracellular cyclic adenosine monophosphate. The same signal transduction pathway is targeted by positive chronotropic drugs such as norepinephrine and dobutamine, which are used in the treatment of cardiogenic shock and severe heart failure. The cyclic adenosine monophosphate–sensitive hyperpolarization–activated current (If) in pacemaker cells is passed by hyperpolarization-activated cyclic nucleotide-gated cation channels and is critical for generating the autonomous heartbeat. In addition, this current has been suggested to play a central role in the chronotropic effect. Recent studies demonstrate that cyclic adenosine monophosphate–dependent regulation of HCN4 (hyperpolarization-activated cyclic nucleotide-gated cation channel isoform 4) acts to stabilize the heart rate, particularly during rapid rate transitions induced by the autonomic nervous system. The mechanism is based on creating a balance between firing and recently discovered nonfiring pacemaker cells in the sinoatrial node. In this way, hyperpolarization-activated cyclic nucleotide-gated cation channels may protect the heart from sinoatrial node dysfunction, secondary arrhythmia of the atria, and potentially fatal tachyarrhythmia of the ventricles. Here, we review the latest findings on sinoatrial node automaticity and discuss the physiological and pathophysiological role of HCN pacemaker channels in the chronotropic response and beyond.
Keywords: action potentials; autonomic nervous system; heart rate; myocytes, cardiac; sinoatrial node
A regular and rhythmic heartbeat is one of the most fundamental processes of life. Around embryonic day 36, the developing heart starts beating and the heartbeat does not stop until the person dies. Remarkably, the rhythmicity of cardiac contractions is controlled with high precision, making the normal heartbeat almost perfectly regular on palpation, with physiological baseline fluctuations known as heart rate variability.^1^ This regularity persists even if the heart rate (HR) changes due to respiration or a change in activity level. The body’s need for blood supply varies depending on physical exertion or emotional activity. To meet such varying demands, cardiac output is mainly regulated by adjusting the HR, which, in turn, is controlled by the autonomic nervous system (ANS): increased activation of the sympathetic nervous system accelerates, while increased activation of the parasympathetic nervous system slows down the heart. This mechanism is termed chronotropic effect and enables smooth changes in HR, without rapid fluctuations or rhythm disturbances. Three components are associated with the chronotropic response, which are (1) the actual HR, (2) the variability of HR at a given time, and (3) the initiation site of pacemaker activity (Figure 1). When the activity state changes from rest to physical activity and back, all 3 parameters are changed by the ANS and need to be considered.
Another remarkable characteristic of cardiac physiology is the automaticity of the heart. The autonomous heartbeat is generated by specialized cardiac tissue in the heart itself, without the need for external stimulation. In this process, the sinoatrial node (SAN) acts as the primary pacemaker (Figure 1A and 1B). It is located in the right atrium of the heart,^2^ with the central part (or head) of the node being localized caudally to the orifice of the superior caval vein (right superior caval vein in rodents). Laterally, it is limited by the crista terminalis, and, medially, it extends toward the interatrial septum. From the head region, the peripheral SAN (or tail) extends toward the inferior caval vein. The longitudinal axis of the SAN can be identified by the SAN artery, which runs in an axial direction through the head and tail and supplies blood to the SAN. The blood supply to the SAN is independent from that of the neighboring atrium and constitutes a selective blood-SAN barrier between the artery and the node. Functionally, the SAN contains specialized cardiomyocytes called pacemaker cells that spontaneously generate rhythmic action potentials (APs). These APs are also called pacemaker potentials (Figure 1D) and serve as the source of strictly timed electrical discharges that propagate along the cardiac conduction system and ultimately trigger the contraction of all atrial and ventricular myocardial cells.^3^ The special ability of SAN pacemaker cells to generate spontaneous APs is enabled by slow diastolic depolarization (SDD; Figure 1D). During this phase of the pacemaker cycle, ion channels and transporters carry net inward currents that slowly depolarize the membrane until the voltage threshold for the next AP is reached. The slope of SDD, which is altered during the chronotropic response, and the repolarization velocity and AP duration determine the chronotropic state of the SAN and regulate the frequency of the heartbeat.^4,5^
Intuitively, one would assume that all pacemaker cells in the SAN fire simultaneously and at the same frequency to generate a common network rhythm. However, recent research has shown that there are subdivisions of pacemaker cell regions within the SAN characterized by different firing rates.^6^ The region with the fastest firing frequency has been termed the leading pacemaker region of the SAN.^7,8^ The presence of a leading pacemaker region has been confirmed in the SAN of the mouse,^9^ rabbit,^10^ dog,^11^ and human.^12^ During HR acceleration on increased activation of the sympathetic nervous system, the leading pacemaker region shifts upwards, while during deceleration of HR on increased activation of the vagus nerve, it shifts downwards toward the atrioventricular junction^7,8^ (Figure 1B). Thus, the fastest HR is generated in a leading pacemaker region in the upmost part of the SAN, while the slowest HR arises in lower parts of the SAN. This observation has even led to the conclusion that instead of being described as the head and tail region of the SAN; at least in rats and humans, there are 2 competing right atrial pacemakers located near the superior vena cava (superior SAN) and near the inferior vena cava (inferior SAN)^13^ coinhabiting the SAN. This conclusion is further supported by a recent publication demonstrating that in mice, the microvasculature of the superior central SAN is more densely organized with more and closer contacts of capillaries to SAN pacemaker cells than the inferior peripheral SAN.^6^ Moreover, there is a sharp transition within the blood supply, which marks the border between the 2 parts of the SAN. This difference in the architecture of the microvasculature is also paralleled by the differential anatomy of autonomic inputs.^14^ There is evidence that in humans^15–17^ and goats,^18^ the SAN is surrounded by an extensive paranodal area with properties of both nodal and atrial tissues. This paranodal area has been proposed to adopt a role in pacemaking, which could explain the widespread distribution of the leading pacemaker site (wandering pacemaker) in the human right atrium as observed in clinics. Alternatively, it could facilitate the exit of APs from the SAN to the atrial muscle.^17^ To date, functional studies are lacking, and, therefore, the function of this area remains unclear.^17^ In addition to the SAN and possibly as part of the paranodal area, in rats,^19,20^ dogs,^21,22^ and humans^23^ an anatomically and functionally confined subsidiary atrial pacemaker has been described; it is located superior to the inferior vena cava and inferior to the crista terminalis. The subsidiary atrial pacemaker region is characterized by its own leading pacemaker region, which is independent of the SAN and gives rise to slower HR than the SAN.
Interestingly, even in the leading pacemaker region of the SAN, there is heterogeneity with respect to firing behavior.^24,25^ This includes nonfiring or dormant pacemaker cells, which have only lately become better characterized experimentally^6,25–31^ (Table 1). Most recent studies revealed that nonfiring pacemaker cells fulfill important physiological functions in the SAN and are necessary to ensure smooth HR transitions during the chronotropic response.^4,25,32,33^ Understanding the mechanisms that regulate firing behavior and the transition between firing and nonfiring is of great physiological and pathophysiological importance and represents an increasingly upcoming research topic in cardiac cellular electrophysiology.
This review is dedicated to the role of pacemaker channels in the chronotropic response in health and disease. It is based on an analysis of >230 in vivo and in vitro studies in different mammalian species (178 references) and humans (50 references) and also modeling studies (17 references). In each section, special emphasis is placed on demonstrating the relationships between the physiology and pathophysiology of humans and other mammals and formulating conclusions that are generally valid for humans. To this end, we will outline how the ANS innervates the SAN to initiate the chronotropic response (Figure 1). We then will introduce the hyperpolarization-activated cyclic nucleotide-gated cation channel (HCN channel) family (Figure 2) as the molecular correlate of the pacemaker current (If) and present the cellular mechanisms by which these channels, together with other transmembrane ion channels and Ca^2+^ handling proteins, interact to generate the autonomous firing of pacemaker cells at the single cell level (Figure 3). Further on, we will introduce the concept of nonfiring cells in the SAN and review the mechanisms underlying the nonfiring mode (Figures 4 and 5). This is followed by a discussion of how pacemaker channels regulate the chronotropic response in the heart. The focus will be on the 3 components associated with this response, the actual HR, the variability of HR at a given time, and the anatomic site of the leading pacemaker region. Then, 2 models will be presented, which explain the interaction between firing and nonfiring cells in the SAN network (Figure 6), and how HCN channels contribute to this interaction and, thereby, regulate the chronotropic response. Finally, we will discuss how diseases affect the pacemaking of the heart.
The SAN is densely innervated by the ANS^34,35^ (Figure 1A). Cell bodies of preganglionic cardiac parasympathetic neurons are primarily residing in the nucleus ambiguus within the brain stem. Reportedly, only a minority of cardiac parasympathetic neurons are located in the dorsal motor nucleus of the vagus and these neurons control ventricular inotropy and excitability, but not HR. Axons from neurons in the nucleus ambiguus form the cardiac branches of the vagus nerve, which bilaterally innervate cholinergic neurons in the cardiac ganglia located on the epicardial surface. Postganglionic fibers arising from these ganglia in turn densely innervate the SAN,^36^ atrioventricular node, and myocardial tissue. The parasympathetic nervous system regulates HR on a fast time scale of milliseconds,^25,37–42^ allowing for beat-to-beat control of HR.
The cell bodies of preganglionic cardiac sympathetic neurons are located in the intermediolateral cell columns of the upper 4 to 7 thoracic segments of the spinal cord.^34^ These neurons receive input from forebrain centers.^34,36^ Axons arising from these neurons project to the superior cervical, cervicothoracic, and thoracic ganglia. The axons from these postganglionic neurons form the sympathetic cardiac nerves on both sides, which project to cardiac plexuses and also directly to the SAN. In contrast to the parasympathetic nervous system, the sympathetic nervous system controls HR on a slower time scale, within seconds.^37–42^
Furthermore, intrinsic cardiac ganglia located around the pulmonary veins modulate the activity of ANS (Figure 1). In the mouse, it has been reported that cholinergic neurons in the inferior pulmonary vein-ganglionated plexus and noradrenergic neurons in the craniomedial part of the right stellate ganglion project to the SAN.^43^ Furthermore, vagal afferents have been shown to enhance parasympathetic and reduce sympathetic efferent outflow to the heart via central mechanisms. In pig and human hearts, the right atrial ganglionated plexus mediates primarily bilateral vagal inputs to exert profound functional control of the SAN and the leading pacemaker region.^44^
The balance between the 2 branches of the ANS can be estimated by comparing the resting HR in vivo (basal HR)^45^ and the HR in the absence of autonomic input (intrinsic HR, measured in the denervated heart or with pharmacological inhibition of the ANS).^45–48^ In small animals, the basal HR is high, for example, around 500 bpm in mice.^49–51^ By contrast, the intrinsic HR in mice is substantially lower than the basal HR. This indicates a significant sympathetic tone in the mouse and other small animals.^51^ In humans and also dogs, the basal HR is much lower, for example, around 60 to 100 bpm in humans. In those species, basal HR arises under pronounced vagal tone,^51^ and pharmacological blockade of autonomic input significantly accelerates the HR.
When the ANS stimulates SAN pacemaker cells to increase or decrease their firing rate, this is achieved by changing the duration of the pacemaker cycle, in particular by changing the SDD rate, as well as the repolarization velocity and AP duration. These effects are mainly caused by altering the intracellular concentration of the second messenger cyclic adenosine monophosphate (cAMP), which is the central element to determine the chronotropic state of the SAN. Increased activation of the sympathetic nervous system leads to the release of the neurotransmitter noradrenaline from nerve terminals in the SAN. Noradrenaline binds to Gs protein-coupled beta-1 adrenergic receptors (β1 adrenoceptors/ADRB1) at the surface of pacemaker cells, causing the Gαs subunit to activate adenylyl cyclases (ACs). Subsequently, ACs catalyze the production of cytosolic cAMP, which, in turn, activates a plethora of downstream targets (Figure 3), many of which are involved in the SAN pacemaker process. Conversely, increased activation of the parasympathetic nervous system causes the release of acetylcholine that binds to Gi protein-coupled M2 muscarinic receptors (CHRM2), thereby inhibiting cAMP synthesis by ACs. The possible mechanisms by which these changes in cytosolic cAMP ultimately accelerate or decelerate the heart^52^ include direct cAMP effects and PKA (protein kinase A)- and CaMKII (Ca^2+^/calmodulin-dependent protein kinase II)-dependent phosphorylation of downstream targets and will be discussed in the ANS Initiates the HR Increasing or Decreasing Effect of the Chronotropic Response section.
The chronotropic response is the result of changes in the SAN pacemaker process, which itself is based on a multitude of subcellular mechanisms (Figure 3). In this section, we highlight the mechanisms that control the 2 modes of sinoatrial pacemaker cells, that is, the firing mode characterized by spontaneous pacemaker potentials and the nonfiring mode, which is required for stabilizing the HR, especially during chronotropic responses. To make the mechanisms underlying spontaneous firing more comprehensible, a concept of 2 interconnected clocks has been established.^53^ In this concept, the term membrane clock refers to cyclic activation and deactivation/inactivation of ionic currents generated by ion channels and transporters in the cell membrane, whereas the term calcium clock describes subcellular periodic uptake and release of Ca^2+^ from the sarcoplasmic reticulum (SR).^53–55^ However, because there are proteins that can be assigned to both clocks, for example, the NCX1 (sodium-calcium exchanger 1), and because the 2 clocks influence each other as coupled oscillators, it would be more precise to speak of a coupled clock system.^56,57^ Nevertheless, for reasons of conformity with the relevant literature, we will use the traditional terms membrane clock and calcium clock in this review.
To achieve spontaneous firing, a complex sequence of ion movements across the cell membrane is required, which together determine the pacemaker potential composed of the SDD, the AP upstroke, and the repolarization phase.^4,5,33^ HCN channels (Figure 2) represent a central element of the membrane clock and are considered key drivers of SDD. The channels pass a noninactivating mixed Na^+^/K^+^ current (If)^58^ and remain open throughout the pacemaker cycle.^59^ Beginning at the maximum diastolic potential (MDP), If initiates slow depolarization of the membrane.^5,58,60,61^ In mammals, there are 4 HCN channel isoforms, termed HCN1-4.^5,58,62,63^ The distinctive feature of these channels is their activation by membrane hyperpolarization. Moreover, HCN channel activity is regulated by cAMP (see ANS Initiates the HR Increasing or Decreasing Effect of the Chronotropic Response section),^64^ and the 4 HCN isoforms differ in their activation and deactivation kinetics and in their sensitivity to cAMP.^32,62^ HCN4 (hyperpolarization-activated cyclic nucleotide-gated cation channel isoform 4) is highly expressed in SAN pacemaker cells^65,66^ and is characterized by the slowest kinetics and highest cAMP sensitivity among the 4 subtypes.^62,67^ Throughout the human SAN, HCN4 is the main isoform and is evenly distributed alongside HCN1 and HCN2,^68^ while no clear evidence of HCN3 expression has been reported.^16^ Similarly, in mice, HCN4 is the predominant isoform expressed in the entire SAN.^25,69^ HCN1, however, is almost exclusively present in the head region of the murine SAN,^69^ and the expression profiles of HCN2 and HCN3 have not yet been conclusively clarified. In addition to enabling spontaneous firing, HCN channels are involved in functional intercellular crosstalk between pacemaker cells of the SAN^32^ (see HCN Channels Stabilize SAN Function During the Chronotropic Response section). Moreover, HCN channels are expressed in secondary pacemaker structures further downstream in the cardiac conduction system.^65,66^ In the subsidiary atrial pacemaker region of the rat, transcripts of HCN1, HCN2 and HCN4 have been reported^19,70^ with higher levels of HCN2 and HCN1 and similar levels of HCN4 compared with the SAN.^19,70^ In the human paranodal area, transcript levels of HCN1 and HCN4 have been reported to be lower and transcripts of HCN2 higher compared with the SAN.^16^ In the murine atrioventricular node, similar to expression in the SAN, HCN4, HCN1, and (with restriction to few individual cells) HCN2 have been detected.^65^ Interestingly, 2 mouse studies with an inducible heart-specific knockout of HCN4^71^ or conditional silencing of complete cardiac If^72^ report severe impairment of atrioventricular conduction up to complete atrioventricular block. While this indicates an important role of HCN4 in atrioventricular node function, no changes in atrioventricular conduction are reported in other HCN4 mouse models.^25,73–76^ Furthermore, HCN channels seem to be involved in the generation of escape rhythms below the atrioventricular node.^77^ In the bundle of His, HCN4 appears to be the only isoform expressed,^65^ while in the bundle branches and Purkinje fibers, HCN4, HCN1, and small amounts of HCN2 are expressed.^65,78,79^ In this review, we specifically focus on HCN channels in the SAN and mechanisms relevant to the chronotropic effect. For further reading on HCN channels in secondary pacemaker structures, please refer to references ^65,80–83^.
During SDD in SAN cells, the mainly HCN4-mediated If (Figure 3A through 3C) depolarizes the membrane to a range sufficient for activation of voltage-gated T-type and L-type Ca^2+^ channels^5,84,85^ (Figure 3). CaV3.1 (CACNA1G) is the predominant T-type Ca^2+^ channel isoform in the adult mammalian SAN and, together with CaV3.2 (CACNA1H), constitutes ICa,T that activates at negative voltages around −60 mV^86^ (Figure 3D). At more positive voltages, L-type Ca^2+^ channels are concomitantly activated and contribute to the late SDD and AP upstroke phase (Figure 3E). ICa,L is comprised of CaV1.3 (CACNA1D) and CaV1.2 (CACNA1C)-mediated currents that are activated one after the other due to their distinct activation thresholds (test potential at which 5% of the maximum current is activated^87^) of −45 and −25 mV, respectively.^84,88,89^ In addition, SAN cells express voltage-gated Na^+^ channels that have been shown to play a role in pacemaking and impulse conduction (Figure 3F). The presence of tetrodotoxin-sensitive (neuronal) NaV1.1 (SCN1A) and NaV1.3 (SCN3A) channels^90^ and the tetrodotoxin-insensitive (cardiac) NaV1.5 (SCN5A) isoform^91^ has been reported in the mouse. In the human SAN, NaV1.6 (SCN8A) is additionally expressed and contributes to intranodal conduction.^92^ Furthermore, a sustained inward Na^+^ current (Ist) has been found in SAN cells.^93–96^ This current is activated at negative membrane potentials and persists throughout the range of SDD, bearing the possibility of a functional contribution to pacemaking. Only recently, the L-type CaV1.3 channel was identified as an obligatory molecular component of Ist in SAN cells.^94,97,98^ In the next phase of the pacemaker cycle, the positive membrane potential causes both inactivation of voltage-gated Na^+^ and Ca^2+^ currents and activation of voltage-gated K^+^ currents, which are responsible for membrane repolarization and return to the MDP (Figure 3G through 3J). There are 3 distinct voltage-gated K^+^ currents, that is, the rapidly activating (IKr) and slowly activating (IKs) delayed rectifier K^+^ currents and the transient outward current (Ito).^5^ In the SAN, IKr is generated by channels composed of the ERG1 alpha subunit (KCNH2),^99^ possibly in interaction with its accessory subunit MiRP1 (KCNE2). IKs channels consist of the KV7.1/KvLQT1 alpha subunit (KCNQ1) in combination with minK (KCNE1).^100^ However, the expression of both IKr and IKs seems to underlie profound species-dependent differences. Ito expression is variable across the SAN and appears to be more important for AP repolarization in peripheral than in central regions.^101^ While KV4.2 (KCND2) and KV4.3 (KCND3), as well as KV1.4 (KCNA4), have been suggested to be possible candidates encoding Ito in the SAN,^101–103^ the actual subunit composition is not yet known. After reaching the MDP, the time- and voltage-dependent decay of K^+^ currents allows the inward currents to initiate membrane depolarization and can, thus, be described as another key mechanism of early SDD.^104^ In addition, SAN cells express several inwardly rectifying K^+^ channels responsible for IKACh/IKAdo, IKATP (Figure 3A), and IK1. IKACh/IKAdo is based on the expression of the G protein-coupled K^+^ channels GIRK1 and GIRK4, which are composed of Kir3.1 (KCNJ3) and Kir3.4 (KCNJ5), respectively.^5,84,105^ Upon activation of Gi protein-coupled M2 muscarinic receptors or A1R and A3R adenosine receptors, these channels are directly opened by binding of the Gβγ subunit to the channel complex.^105,106^ The corresponding outward K^+^ current significantly hyperpolarizes the MDP and slows down the firing rate of SAN cells during parasympathetic activity.^107,108^ IKATP in the SAN is generated by channels consisting of the pore-forming Kir6.2 (KCNJ11)^109^ or Kir6.1 (KCNJ8) subunit^110^ and the regulatory SUR2A (ABCC9) subunit that is important for the channel’s reactivity to the metabolic state of the cell.^111^ KATP channels are blocked by ATP and open when intracellular ATP concentrations drop. The current is, thus, activated when the metabolic rate is low, leading to a slowdown of pacemaker activity that potentially protects the heart from ischemic damage.^109,110^ The background K^+^ current IK1 is mainly generated by Kir2.1 (KCNJ2) and Kir2.3 (KCNJ4) channels.^16^ It is small or absent in human SAN^16^ but moderately expressed in murine SAN cells,^5,30^ although at significantly lower densities than in the working myocardium. In ventricular and atrial cardiomyocytes, IK1 is considered to stabilize the resting membrane potential and counteract spontaneous firing, thereby generating the driving force for inward Na^+^ and Ca^2+^ currents during the AP. The greatest influence on the resting membrane potential of excitable cells is exerted by the sodium-potassium pump (Na^+^/K^+^-ATPase), which is also significantly involved in setting the MDP of SAN cells to the range of −60 mV^5,112^ (Figure 3K). This is achieved by carrying a steady-state net outward current (Ip). However, also a voltage-dependent behavior of Ip has been described, which, together with its intrinsic Na^+^ dependency, could cause a periodic change in Ip, possibly contributing to SAN cell automaticity.^5,113^ More recently, the family of Ca^2+^-activated K^+^ channels (KCa) has been suggested to influence pacemaking in the SAN^84^ (Figure 3A). These channels are categorized into voltage-independent small-conductance (SK) and intermediate-conductance channels, as well as voltage-dependent big-conductance channels. KCa channels have distinct sensitivities to Ca^2+^ conferred by the interaction of CaM (calmodulin) with SK and intermediate-conductance channels^114,115^ or by direct binding of Ca^2+^ to big-conductance channels.^116,117^ The 3 SK channel isoforms, SK1 (KCa2.1/KCNN1), SK2 (KCa2.2/KCNN2), and SK3 (KCa2.3/KCNN3),^118,119^ as well as the intermediate-conductance channel (SK4, KCa3.1/KCNN4)^120^ and big-conductance channel (KCa1.1/KCNMA1),^121^ are expressed in the SAN and translate changes in intracellular Ca^2+^ into repolarizing currents, thereby affecting the AP duration, MDP, and SDD of pacemaker cells.^122^ Another protein regulated by intracellular Ca^2+^ is the sodium-calcium exchanger 1 (NCX1/SLC8A1), a transporter located in the cell membrane whose activity is indispensable for SAN automaticity.^31,123,124^ NCX1 transports 1 Ca^2+^ ion and 3 Na^+^ ions in opposite directions. The reversal potential of NCX in cardiac myocytes is around −20 mV,^125,126^ thereby facilitating the outward transport of Ca^2+^ coupled with inward transport of Na^+^ (forward mode) in the diastolic range of membrane potentials, which is more negative than −20 mV. Consequently, NCX1 activity causes a net inward current (INCX) during late SDD that increases when subsarcolemmal Ca^2+^ is elevated (Figure 3L). Moreover, members of the superfamily of transient receptor potential channels are expressed in the SAN^84,127^ (Figure 3A). TRPC channels are involved in store-operated Ca^2+^ entry,^128,129^ whereas TRPM4 activity directly contributes to the baseline firing rate^130–132^ and TRPM7 indirectly influences pacemaking by regulating the expression of HCN4.^127,133^
Different groups have recently demonstrated that several components of the membrane clock, and, consequently, the HR are regulated by stretch (for review see^134^). In large mammals with inherently slow HR, stretch increases HR, while in smaller mammals such as the mouse and rat with resting HR of ≥500 bpm, SAN stretch can increase or decrease the beating rate.^134–137^ Stretch-induced changes in beating frequency occur in isolated hearts,^138,139^ isolated right atria,^140–142^ SAN tissue,^137^ and individual SAN cells,^143^ all of which lack autonomic innervation. These and other experiments (for details, please see reference ^134^) suggest that the chronotropic response of the heart to stretch is, at least in part, intrinsic to the SAN.^134,144^ The response of the SAN to stretch occurs faster than and complementary to the positive chronotropic response induced by the ANS and occurs on a beat-to-beat time scale. The cellular mechanisms that elicit the chronotropic response to SAN stretch involve stretch-activated ion channels and have been summarized recently.^134,145^ Some of these channels are closed under baseline conditions and require a stretch to be activated (ClC2, ClC3, SWELL1, PIEZO1/2, TREK-1).^144^ Others are integral components of the membrane clock, and their mechanosensitivity additionally modulates the contribution of these channels to SDD. For example, the activation, deactivation, and amplitude of the currents generated by HCN2 pacemaker channels in heterologous expression systems have been shown to be altered by mechanical stimulation.^146,147^ In addition, many other ion channels and transporters in the heart have been shown to be mechanosensitive, including INCX, ICa,L, ICa,T, IKr, and IKs.^148^ Thus, mechanosensitive gating of ion channels supports the pacemaker function to adapt to changes in hemodynamic load on a beat-to-beat basis, which intrinsically occur on a fast time scale.^134,144^ Finally, there is also evidence of an interaction between the control of HR by the ANS and the mechanosensitivity of the SAN. Specifically, when the vagus nerve is stimulated, the SAN stretch-induced increase in beating rate is greater than without nerve stimulation. Conversely, the decrease in beating rate in response to vagus nerve stimulation is attenuated following SAN stretch in both SAN tissue preparations and whole animals.^149,150^
In addition to the membrane clock, periodically occurring subcellular Ca^2+^ dynamics of the calcium clock play an important role in cardiac pacemaking^26,28,56^ (Figure 3A and 3L through 3O). During diastole, the opening of ryanodine receptor 2 (RyR2/RYR2) generates rhythmic local calcium release (LCR) from the SR,^151–153^ resulting in a periodic increase in the subsarcolemmal Ca^2+^ concentration. The question of whether LCRs are triggered by ICa,L or occur independently has been investigated experimentally.^153–157^ In permeabilized rabbit SAN cells and in intact rabbit SAN cells under acute voltage-clamp conditions, it has been shown that rhythmic LCRs occur spontaneously and do not require membrane currents per se.^153,157^ In contrast, an important role of ICa,L in generating RyR2-dependent Ca^2+^ release was demonstrated in murine pacemaker cells.^154–156^ Regardless of the basic requirements for the occurrence of LCRs, there is evidence that the kinetics of membrane currents and intracellular Ca^2+^ cycling are coupled to each other and are synchronized during changes in HR across different species.^158^ Functionally, RyR2-mediated LCRs have an impact on the pacemaker cycle via NCX1. When subsarcolemmal local Ca^2+^ is elevated (Figure 3N), more Ca^2+^ is extruded from the cell and the net inward current INCX increases. In this way, INCX translates LCRs into changes in membrane potential and significantly contributes to depolarization during the late (exponential) phase of SDD^56^ (Figure 3L). In SAN cells, LCRs are strictly regulated, and their timing and magnitude are influenced by the amount of Ca^2+^ stored in the SR. During the AP, intracellular Ca^2+^ increases (Figure 3M) due to calcium-induced calcium release. The SR/endoplasmic reticulum Ca^2+^-ATPase (SERCA2a/ATP2A2) contributes to removing Ca^2+^ from the cytosol after AP termination, thereby refilling the SR Ca^2+^ stores and enabling LCRs during the next pacemaker cycle.^159^ SERCA activity is, in turn, regulated by phospholamban (PLN/PLN), a small pentameric protein, which, when unphosphorylated, directly binds to and inhibits SERCA^160^ (Figure 3A). In addition, reverse NCX activity (Ca^2+^ entry coupled with Na^+^ extrusion) at positive membrane potentials during the AP has been suggested to influence the SR Ca^2+^ content and, hence, the calcium clock function.^161^ Furthermore, intracellular calcium dynamics are influenced by the Ca^2+^-mobilizing second messenger inositol-1,4,5-triphosphate.^162–164^ Type II inositol-1,4,5-triphosphate receptor (IP3R2/ITPR2) is the predominant isoform located on the SR of murine SAN cells^163,164^ and releases Ca^2+^ when inositol-1,4,5-triphosphate is present (Figure 3A). This has been shown to have a direct effect on pacemaker function, possibly through activation of NCX1, other Ca^2+^-dependent currents, or stimulation of Ca^2+^-activated ACs.^162^ In addition, pacemaking depends on the function of calsequestrin 2 (CASQ2/CASQ2), the main Ca^2+^ binding protein in the SR of SAN cells. CASQ2 has been shown to be important for ensuring robust diastolic Ca^2+^ release, which is necessary for maintaining a normal resting HR and regular rhythmicity.^165^ Whether the proteins of the calcium clock—just like their counterparts of the membrane clock—are regulated by stretch is currently under investigation and debate. While some studies do not report that calcium Clock proteins are activated by stretch,^144^ others do.^134,136,166^
Given that the mechanisms of the membrane clock and calcium clock all influence SDD and, thereby, the pacemaker activity, the question arises as to how important the individual contributions are. At least for the membrane clock and calcium clock, this issue was investigated on the transcript and protein level. Using high-resolution mass spectrometry, a recent study identified 7000 proteins in the SAN and the neighboring atrial myocardium. Strikingly, the expression of 575 proteins differs between the 2 tissues. Using single-nucleus RNA sequencing of sinus node biopsies, measured protein abundances could be attributed to specific cell types. The data reveal significant differences between the SAN and the atrial myocardium in ion channels responsible for the membrane clock, but not in calcium clock proteins. Furthermore, the expression of PKA and CaMKII was more abundant in the atrial muscle, and the PDEs (phosphodiesterases) detected were either uniformly expressed (PDE1A) or more highly expressed (PDE3A and PDE5E) in the atrial muscle compared with the SAN. Together, these findings suggest that specific ion channels of the membrane clock are primarily responsible for pacemaking.^167^
Given the broad knowledge about how autonomic firing at the single cell level is generated, it came as a surprise that single pacemaker cells can enter a nonfiring state and switch back and forth between firing and nonfiring (Figures 4 and 5). In the following, we provide the results of a comprehensive analysis of the literature and a detailed comparison of studies in which nonfiring pacemaker cells have been reported (Tables 1 through 3). The baseline for the discovery of nonfiring cells is the observation that acute preparations of single SAN cells are typically heterogeneous, morphologically and functionally, with only a fraction of around 10% to 30% of cells spontaneously beating with visible contractions or firing.^168,169^ Only recently, researchers have started to closely investigate cells without apparent automaticity (Table 1). The respective studies report the characterization of quiescent pacemaker cells isolated from guinea pig SAN,^26,29^ arrested human SAN cells,^28^ and dormant^27^ and nonfiring^25^ murine cells. Importantly, nonfiring pacemaker cells were not only found in single-cell preparations but also in intact tissue explants containing the whole SAN region.^24,25^ Together, these observations strengthen the suggestion that rather than being experimental artifacts, these cells have an important physiological function in the SAN.^4,25,32,33^
Continuously firing SAN cells (Figure 4A) usually have a stable and hyperpolarized MDP of −50 to −70 mV. The membrane potential of dormant cells, however, varies substantially between the different studies. In addition, some results show initially firing cells with intermittent episodes of nonfiring, whereas, in others, the cells are initially dormant and need to be stimulated to show rhythmic firing. By comparing the characteristics of the respective cells, 3 different groups could be identified.
The first group comprises initially firing cells that spontaneously and reversibly switch to nonfiring (Figures 4B and 5A). Before entering the nonfiring mode, the MDP of firing cells slowly hyperpolarizes by ≈8 mV until firing ceases.^25^ During nonfiring episodes, the membrane potential remains at hyperpolarized values in the range of −60 mV.^25^ This is accompanied by a slow membrane depolarization until firing is reinitiated. Under baseline (unstimulated) conditions, the transition to nonfiring is rare but can be augmented by the application of the muscarinic receptor agonist carbachol (Figure 4C). Likewise, nonfiring is completely abolished when cells are stimulated with the beta-adrenoceptor agonist isoproterenol.^25^ This confirms the crucial role of ANS input in controlling the firing behavior of SAN cells. Similar to this observation, burst-firing cells with long silent periods at hyperpolarized membrane potentials, interrupted by short periods of high-frequency AP firing, have been reported.^6^ The possible biophysical mechanisms underlying hyperpolarized nonfiring cells will be described in HCN Channels Stabilize SAN Function During the Chronotropic Response section.
The second group of cells is initially dormant, has a relatively depolarized membrane potential of −30 to −40 mV, and develops rhythmic firing only on hyperpolarization during sustained beta-adrenergic stimulation^27,29^ (Figure 4D). Such cells were found in wild type (WT) preparations from mice and guinea pigs^27,29^ (Table 1). Similar observations were made in human cells although the responder rate to isoproterenol stimulation was considerably lower.^28^ Importantly, viability is clearly confirmed because tonic firing can be induced in responding cells by application of isoproterenol (100 nmol/L–1 µmol/L). The cellular mechanism underlying the depolarized membrane potential during nonfiring has not been investigated. It is possible that ion conductances with a more positive reversal potential around −35 mV become dominant during nonfiring. In line with this, a background current has been postulated (reviewed in reference^54^) with an equilibrium potential between −30 and −40 mV. The selective activation of this current could explain depolarized nonfiring pacemaker cells. Another explanation would be that the integrity of these pacemaker cells is compromised.
The third group comprises depolarized dormant cells that do not show changes in membrane potential and fail to elicit APs during beta-adrenergic stimulation (nonresponder; Figure 4E).^28,29^ While these cells still generate LCRs under baseline conditions and show increased membrane current densities (ICa,L, IK, and If) during isoproterenol application,^29^ it is not clear why automaticity cannot be restored. Therefore, they may have been damaged during cell isolation so that (in the words of the authors) arrested cells may represent SANCs that have lost functional clock coupling, possibly because of the cell isolation procedure.^28^
To ensure reliable characterization of nonfiring cells, certain experimental quality criteria need to be fulfilled. In the first group of cells, the switch between the 2 modes happens at a slow time scale. Therefore, spontaneous activity must be recorded over a long period to reliably detect, investigate, and characterize nonfiring. Moreover, it is essential to ensure that cytosolic cAMP is kept close to physiological levels during measurements by performing current-clamp experiments with the perforated patch-clamp method. To this end, a pore-forming compound (Figure 4F) is used, which provides electrical access to the intracellular compartment without breaking the membrane patch. Consequently, the integrity of the cytoplasm is largely preserved by preventing dilution of the intracellular milieu, washout of cellular factors including cAMP, and also marked rundown of ion channel currents. In this respect, amphotericin B is superior to other compounds such as beta-escin due to the considerably smaller pore size. Beta-escin pores are significantly larger, resulting in the permeation of even high-molecular-weight compounds.^170,171^ In future experiments, firing and nonfiring need to be exactly monitored and documented over long time periods with several cycles covered in one recording. Such sufficiently long recordings should confirm that the cells can switch back and forth between the 2 modes spontaneously or pharmacologically triggered, thereby confirming cell viability.
Altogether, these recent data suggest that there may be different manifestations of nonfiring in SAN cells, characterized by depolarized versus hyperpolarized membrane potentials and by stimulated versus spontaneous switch to rhythmic firing (Figure 4). To minimize selection bias and optimize validity, future studies should consider the different behaviors of SAN cells, clearly specify the population studied, and ideally use consistent nomenclature for the different modes.
The second messenger cAMP is the key determinant of the chronotropic state of the SAN, and it is used to be generally believed that cAMP-dependent regulation (CDR) of HCN4 is the primary mechanism for mediating the frequency aspect of the chronotropic response.^60,61,64,175^ However, besides HCN channels, several cAMP receptors have been identified in the SAN, including other proteins of the membrane clock and calcium clock (Figure 3). HCN4 is activated by membrane hyperpolarization and carries a depolarizing inward current during SDD. Furthermore, the channel contains a cyclic nucleotide-binding domain in its intracellular C-terminus^176^ that makes it directly responsive to cytosolic cAMP levels (Figure 2). The traditional concept assumes that when the activity of the sympathetic nervous system increases, cAMP-binding to the cyclic nucleotide-binding domain of HCN4 and subsequent increase in If would be responsible for accelerating SDD, thereby increasing the firing frequency of pacemaker cells and, consequently, increasing HR (positive chronotropic effect). On the other hand, a reduction in HCN4 activity due to reduced cytosolic cAMP caused by parasympathetic signaling would be responsible for slowing down SDD and, thereby, reducing HR (negative chronotropic effect). This theory became widely accepted in cardiac physiology but did not remain undisputed. Although recent data from isolated single cells support the view that the increase in If following beta-adrenergic stimulation is responsible for at least part of the concomitant increase in firing rate^177^ in vitro, this concept has never been conclusively confirmed in vivo. In fact, during the past 20+ years, different research groups have been investigating the cardiac function of several mouse models targeting HCN4^25,71–76,172,178^ (Table 2). The results demonstrate that despite the lack of responsiveness to cAMP or the complete absence of HCN4, the adult animals still showed a normal increase in HR, clearly refuting the longstanding textbook knowledge of HCN4 function as the main driver of the chronotropic response in the SAN. This implies that although changes in intracellular cAMP regulate HCN4 activity and If, this particular regulation is not the cause of altering SDD and mediating HR transitions but must serve a different purpose. Only recently, it was discovered that CDR of HCN4 plays an important role in controlling the firing and nonfiring mode of pacemaker cells, which is crucial for stabilizing the HR and leading pacemaker region, and dampening ANS input during the chronotropic response.^4,25,32,33^ In HCN Channels Stabilize SAN Function During the Chronotropic Response section, we will propose a model explaining the underlying mechanism.
If CDR of HCN4 is not the main mechanism underlying ANS-induced changes in HR, frequency regulation must be controlled by other ion channels and transporters in pacemaker cells. Beta-adrenergic signaling increases cytosolic cAMP and, thereby, activates PKA,^179^ which, in turn, phosphorylates and activates a variety of proteins of the membrane clock and calcium clock (Figure 3). These downstream targets of PKA include voltage-gated T-type CaV3.1 channels,^180,181^ L-type CaV1.3^182^ and CaV1.2 channels,^183,184^ NaV1.5 channels,^185,186^ RyRs,^187^ PLN,^159,188^ Na^+^/K^+^-ATPase,^189^ and delayed rectifier K^+^ channels.^190,191^ While IKs is clearly augmented by PKA-dependent phosphorylation, there is no agreement on the beta-adrenergic modulation of IKr.^191^ Moreover, it was recently discovered that the mechanism by which PKA positively regulates ICa,L in ventricular myocytes during the inotropic response is based on phosphorylation of the small G protein Rad and subsequent disinhibition of CaV1.2.^192^ It is, therefore, possible that members of the RGK GTPase family such as Rad or Rem are also essential to the chronotropic effect by regulating CaV1.2 and CaV1.3 channels in SAN cells.^193,194^ In parallel, sympathetic activity stimulates CaMKII,^195–197^ subsequently activating L-type CaV1.3^182^ and CaV1.2 channels,^174,183^ NaV1.5 channels,^185,186^ PLN,^198^ SERCA,^199^ and RyR2.^198^ Moreover, the resulting elevated diastolic Ca^2+^ levels positively regulate TRPM4,^131,132^ KCa channels,^118,120,121^ NCX1,^56,200,201^ and Ca^2+^-activated ACs (AC isoform 1 [AC1] and AC8).^173,202,203^ When studied individually, these proteins have all been shown to influence the firing rate of pacemaker cells and play a role in the chronotropic response. However, it is evident that there is a high degree of redundancy or mutual cooperation involved in this process, and the relative contribution of each mechanism to efficient autonomic regulation of HR remains to be determined.
Besides the increasing or decreasing effect on HR by the ANS, 2 more components of the chronotropic response need to be considered (Figure 1): the stability of the HR at a given time point, which is characterized by a specific variability or fluctuation over time (heart rate variability), and the stability of the anatomic position of the leading pacemaker region within the SAN. Together, the 3 components of the chronotropic response are independently regulated by the ANS. The main function of HCN channels is to dampen and stabilize the electrical activity of the SAN network in time and at a given anatomic position within the SAN at basal conditions and during modulation by the ANS. Thereby, HCN channels ensure smooth changes in HR without rapid fluctuations or even rhythm disturbances in time and with regard to anatomic localization. The detailed mechanisms underlying these HCN channel functions and, in particular, the mechanisms by which these channels regulate the nonfiring and firing of SAN pacemaker cells are presented in a model (Figure 5).
The finding that during sufficiently long current-clamp recordings, a substantial fraction of murine WT cells shows transient episodes of nonfiring before returning to rhythmic firing^25^ (Figure 5A) can be explained by dynamic mode shifts that can specifically be induced in HCN channels.^4,25,32,33^ Mode shifts are the cause of voltage-dependent changes in the position of the activation depolarized membrane potentials induce a leftward shift of the activation curve toward more negative voltages (mode x) and hyperpolarized membrane potentials induce a rightward shift of the activation curve toward more positive voltages^4,25,32,33,204–208^ (mode y; Figure 5B). Because in HCN4 channels, these mode shifts occur slowly, the mechanism could drive the slow alternations between firing and nonfiring in the following way.
The different membrane potentials during firing and nonfiring (Figure 5A) have a long-lasting effect, meaning that the cell changes its firing behavior in a history-dependent fashion. An important prerequisite for this scenario is that the activation and deactivation kinetics of HCN4 are substantially slower than the duration of the pacemaker cycle. When the cell is firing, If can, therefore, be considered almost constant during the pacemaker potential with only minor oscillations around a constant mean. Consequently, both voltage-dependent activation and dynamic mode shifts are determined by the mean membrane potential, which suddenly changes at the beginning and end of nonfiring (Figure 5A). During nonfiring, the relatively hyperpolarized membrane potential increases the activation of HCN channels, slowly driving the cell back toward firing. This effect is potentiated by an additional slow shift of the activation curve toward the right (mode y in Figure 5B) so that more HCN channels are open at a given membrane potential and the cell membrane can be depolarized faster. At a critical point, the threshold is reached and the cell returns to rhythmic firing. The effect persists for a while during depolarization (firing) until it slowly declines, the activation curve shifts back toward the left (mode shift to mode x in Figure 5B), fewer HCN channels are open at a given membrane potential, and the MDP slowly hyperpolarizes, eventually leading to the next nonfiring episode. In addition, the binding of cAMP influences the position of the activation curve and facilitates channel opening (Figure 5C). This means that the intrinsic biophysical effect is further regulated and superimposed by the ANS, which, thereby, controls the cell’s firing behavior through sympathetic and parasympathetic signaling.
Quiescent/dormant/nonfiring SAN cells have only recently been identified in WT preparations (Table 1), most likely because nonfiring occurs rarely in the WT, or because experimenters did not further investigate a pacemaker cell, which was nonfiring. In contrast, nonfiring pacemaker cells have been reported before in the context of genetic mouse models, including knockout and loss-of-function mutations in HCN4 (Table 2). Interestingly, intermittent or permanent nonfiring due to impaired or absent HCN4 function was typically accompanied by hyperpolarized membrane potentials,^25,73,74,76^ which could reflect a more frequent occurrence and exaggeration of the spontaneous nonfiring described in WT cells. If, thus, provides a depolarization reserve in SAN cells that prevent overshooting hyperpolarization and keep the membrane potential in a range in which diastolic depolarization and tonic firing are, in principle, possible.^73^ Within this range of membrane potentials, mode shifts in HCN4 cause the voltage-dependent hysteresis that could induce the alternations between firing and nonfiring (Figure 5A and 5B). In addition, the effect is further fine-tuned by the CDR of HCN4 and, thereby, controlled by the ANS. Activation of HCN4 by cAMP contributes to terminating nonfiring episodes (Figure 5C) and maintaining cellular automaticity, while a reduction in HCN4 activity promotes nonfiring. In line with this, transient episodes of nonfiring are heavily increased when the CDR of HCN4 is missing.^25^ Similar effects were observed in AC1 knockout mice.^173^ AC1 is an important downstream target of beta-adrenergic signaling and forms functional microdomains with HCN4. It is, therefore, well possible that AC1 provides cAMP required for the regulation of HCN4. In addition, dysfunction of several other proteins has been associated with quiescent SAN cells. Some prominent examples include CaMKII,^174^ PKA,^157^ CaV1.3,^27^ and NCX1^31^ (Table 3).
The main function of the SAN is to ensure a rhythmic heartbeat and to enable smooth HR transitions. This raises the important question of how firing and nonfiring cells interact with each other and how the fraction/number of nonfiring cells is controlled to obtain a stable rhythm in the network of the SAN. The intact SAN is a heterogeneous tissue composed of various cell types connected to each other and embedded in an extracellular matrix of connective tissue, mainly consisting of elastin and collagen fibers.^17,209^ The different cell types include pacemaker cells and atrial myocytes,^210^ sympathetic and parasympathetic neurons,^211^ and nonexcitable cells such as fibroblasts^212^ and macrophages.^213^ Recently, additional components (telocytes and peripheral glia cells), which also belong to the group of nonexcitable cells and further increase the heterogeneity of the SAN, have been discovered.^14,24^ This heterogeneity has even led to the comparison of the SAN with brain-like structures^33,57,214^ or networks, composed of neuron-like cellular components.^215,216^ Within this network, pacemaker cells are weakly electrically coupled to each other via gap junctions. In the central SAN, gap junctions are formed by the connexins Cx30.2 and Cx45 (low conductance), while there is no expression of Cx43 (intermediate conductance) and only little expression of Cx40 (high conductance). In the periphery of the SAN at the border to the surrounding atrial myocardium, Cx43 and Cx45 are expressed.^217^ Through the weak electrical coupling, a mutual and long-lasting interaction process between neighboring firing and nonfiring cells will take place, which has been termed tonic entrainment.^4,25,32,33^ Nonfiring cells are more hyperpolarized and will electrotonically draw flows of cations, including Ca^2+^, from more depolarized firing cells to which they are coupled via gap junctions. As a result, nonfiring cells will slightly depolarize and neighboring firing cells will slightly hyperpolarize. When a new equilibrium is reached, the common firing rate will be decreased and a bradycardic network rhythm will emerge.^25,33^ In this way, nonfiring cells may function as brakes in the SAN network and, continuously, inhibit the activity of surrounding firing cells. In addition, tonic inhibition appears to increase during vagal activity and decrease during beta-adrenergic stimulation, probably because more cells switch to the nonfiring or firing mode, respectively.^4,25^ While this process is important for setting the baseline HR and stabilizing SAN network activity, it is not involved in changing HR, that is, mediating the frequency aspect of the chronotropic effect.^4,25,32,33^ Like in neuronal networks, inhibitory control of excitability may be essential in the SAN to ensure a robust function of the pacemaker process but requires a stable balance between nonfiring cells (inhibition) and firing cells (excitation). Under physiological conditions, this balance is controlled by the CDR of HCN4, which is particularly important to dampen and fine-tune the HR-lowering effect of the parasympathetic nervous system.^25,75^ In contrast, when activation of HCN4 by cAMP is missing or disturbed, the fraction of nonfiring pacemaker cells strongly increases. As a consequence, impulse formation in the central SAN and propagation to the atrial myocardium are delayed. Such overshooting inhibition is responsible for SAN dysfunction and secondary arrhythmia, which predominantly occurs during chronotropic responses.^25^ Taken together, the newly discovered role of HCN4 in the SAN is to regulate the number of nonfiring cells, which is critical for stabilizing the network rhythm and protecting the SAN during ANS-induced HR transitions.^4,25,32,33^ At the same time, the stabilizing action of HCN channels will also stabilize the leading pacemaker region in a given anatomic location.
As pointed out above, nonfiring cells with depolarized membrane potential seem to be present in the SAN. Initially, the existence of such cells in the sinus node has been postulated, and their role has been investigated in computer simulation studies of the sinus node.^218^ Subsequently, depolarized nonfiring pacemaker cells have been experimentally confirmed (Table 1). In addition, nonexcitable cells characterized by depolarized membrane potential such as fibroblasts or macrophages have been described, which could also interact with SAN pacemaker cells via gap junctions in the SAN network. Regardless of whether excitable or nonexcitable, in an intact SAN network a depolarized nonfiring cell, if connected via gap junctions, would induce analog current flow from nonfiring to firing cells as described above, but in the reverse direction, thereby depolarizing firing cells and hyperpolarizing the particular nonfiring cell. This would be expected to accelerate and also to stabilize the network rhythm. The described scenario for tonic entrainment between depolarized nonfiring and neighboring firing cells needs to be experimentally confirmed in future experiments.
In addition to tonic entrainment, a phasic entrainment mechanism has been postulated,^32,219–222^ and experimental evidence has been provided.^223^ Phasic entrainment is faster than tonic entrainment; it contributes to synchronizing spontaneous firing on a beat-to-beat time scale.^32,219–221^ The process is illustrated in Figure 6A in which 2 hypothetical pacemaker cells interact with each other. It is assumed that 1 fast-firing cell (marked with F in Figure 6A) is weakly coupled to a slow-firing cell (marked with S in Figure 6A) and both cells fire all-or-none suprathreshold pacemaker potentials. With weak coupling between cells (solid traces), pacemaker potential firing of 1 cell is conducted to the other but attenuated by high-resistance gap junctions so it produces only subthreshold membrane depolarizations in the other cell. In turn, these subthreshold depolarizations either accelerate or delay the next firing of that cell, depending on timing.^221^ For example, if the second pacemaker potential of the faster firing cell F occurs at a phase close to the middle of the intrinsic period of S, it induces a shortening of the next pacemaker potential of S, accelerating the firing of this cell. Similarly, this second pacemaker potential of cell S occurs at an early phase of cell F and induces a prolongation of the next pacemaker cycle of F, slowing down the firing of cell F. The model predicts that by similar interactions as outlined for the 2 cells in Figure 6A, the entire SAN including all pacemaker cells can be entrained to operate at one and the same frequency. The common network frequency would lie in between the frequency that the fastest and slowest cells would generate if they were isolated cells. The model would furthermore predict that phase shifts would arise between individual cells, which would mimic the classical propagation of excitation. However, rather than sequentially activating one cell after the other, there would be apparent conduction resulting from phase shifts between cells. The role of HCN channels in phasic entrainment would be to contribute to synchronization by dampening depolarizing or hyperpolarizing influences and stabilizing the membrane potential in the subthreshold diastolic voltage range.^32,58,69,224^
Recently, it was shown that some assumptions of the concept of phasic entrainment are not met in the SAN network, and, therefore, the model had to be modified. For example, the cellular heterogeneity of the SAN is considerably higher than initially postulated.^14,24^ There even might be at least 2 separate weakly coupled independent meshworks of HCN4^+^/Cx43^–^/F-actin^–^ cells on the one hand and Cx43^+^/F-actin^+^/HCN4^–^ cells on the other hand.^24^ In addition, not all pacemaker cells of the SAN fire with the same frequency, rather there are several clusters with differing frequencies giving rise to multicentric SAN activation.^225,226^ In line with this, anatomically distinct subdivisions such as the superior and inferior SAN can be distinguished,^6,13^ which could provide an explanation for pacemaker shifts on activation of the SAN by the parasympathetic or sympathetic nervous system.^13,226^ Further insight into the entrainment process was provided in a recent study, using elaborated imaging techniques.^14,24^ This study postulated that not only suprathreshold all-or-none pacemaker potentials can induce the entrainment process but also subthreshold fluctuations in membrane potential differing in amplitudes, frequency, and kinetics. The subthreshold fluctuations in membrane potential could either arise from LCRs from the SR, which lead to subthreshold Ca^2+^ signals, which, in turn, are transformed into corresponding depolarizing inward currents by the NCX, thereby inducing subthreshold oscillations in membrane potential. In addition, there is evidence for subthreshold fluctuations in membrane potential that are not triggered by local intracellular Ca^2+^ signals.^6^ Regardless of the underlying mechanism, subthreshold fluctuations in membrane potential could become integrated and, eventually, reach the threshold for firing in this particular cell or induce firing in neighboring cells by entrainment.^24^ In conclusion, the classical entrainment model and also its modification have in common that regular pacemaking results from the reduction of electrical heterogeneity present at the level of single cells.
Interestingly, an alternative conceptual model, termed the stochastic resonance model,^227,228^ postulates exactly the opposite of the entrainment model, that is, that electrical heterogeneity of single pacemaker cells is the motor of regular pacemaking. According to this model, pacemaker cells, in addition to firing suprathreshold APs (firing mode), produce stochastic low-amplitude voltage fluctuations in a range of 4 to 17 mV^6^ (noise; Figure 6B), which, on their own, cannot trigger firing. In the network of the SAN, where firing and nonfiring cells are weakly connected to each other via gap junctions, voltage noise could add to and thereby amplify weak signals (periodic oscillations) at their peaks and increase the probability of crossing the threshold for firing (Figure 6C). It would be expected that successively increasing noise would increase the firing frequency up to a resonance peak at the natural frequency of the firing pacemaker cells and then, when the noise dominates the signal, would decrease firing performance, thus giving rise to a resonance curve (Figure 6D). In the context of the stochastic resonance model, noise would have a positive role in signal processing. This effect could indeed be so efficient that a hypothetical SAN network consisting of only passive pacemaker cells, which only produce subthreshold oscillations, would be integrated to eventually produce suprathreshold network rhythm. This has recently been shown in a modeling study for the SAN.^229^ However, in the SAN, nonfiring cells with subthreshold oscillations (eg, inferior SAN) would be expected to be coupled to more periodically active SAN pacemaker cells that fire pacemaker potentials (eg, superior SAN). The noisy nonfiring SAN pacemaker cells would increase the strength and periodicity of tonically firing SAN pacemaker cells and, thereby, the robustness of pacemaker activity. In other words, random subthreshold voltage noise would have stabilizing antiarrhythmic effects on SAN pacemaker activity. Furthermore, all SAN cells are important in driving the system. However, the contribution of the 2 types of cells, that is, the noisy cells and the periodically oscillating cells, would be different. A loss of noisy nonfiring cells with subthreshold fluctuations in membrane potential would reduce the robustness of the system to synchronize and adapt. In contrast, the loss of periodically firing cells from the intrinsically oscillating cells would completely abolish SAN activity.
The resonance model could also explain how the superior SAN, which fires at a high frequency, and the inferior SAN, which fires at a lower frequency, interact.^6,13^ According to the model, most SAN pacemaker cells in the inferior region would generate stochastic subthreshold voltage fluctuations or rare single APs. These signals would only lead to periodic pacemaking when coupled to more periodically firing SAN pacemaker cells, that is, voltage oscillators of the superior SAN. Under these conditions, random subthreshold voltage fluctuations and APs are integrated by stochastic resonance to increase the probability that superior SAN pacemaker cells reach the AP threshold. Like in dendritic integration in the brain, noisy SAN pacemaker cells would increase the strength and periodicity of tonically firing cells and the overall pacemaker activity of the SAN network. For this model to work, both are the presence of heterogeneity and the presence of the inferior SAN. A recent study shows that loss of heterogeneity or the electrical separation of the inferior from the superior SAN will decrease HR and HR regularity.^13^
Robust pacemaking is central to cardiac physiology, and disturbances in this process give rise to cardiac disease. The most prominent disease of the SAN is sinus node dysfunction (SND), also known as sick sinus syndrome. SND is defined as a group of arrhythmias caused by malfunction of the SAN pacemaker process.^84,230–233^ These include bradycardia, sinus arrest, sinoatrial block, alternations of bradycardia and tachycardia known as tachy-brady syndrome, and chronotropic incompetence.^231^ Importantly, chronotropic incompetence is defined as the inability to adequately increase HR to meet elevated metabolic demands.^46,48^ This is reflected by not reaching the same maximum HR as healthy individuals during exercise, while the range of frequency regulation, albeit at lower absolute HRs, may be preserved. SND is generally considered to be an age-related pathology that develops secondary to degenerative fibrosis of the pacemaker tissue. However, there are other forms of SND, including the genetically caused idiopathic primary SND, and forms that arise secondary to cardiovascular or systemic disease. Typical symptoms of SND are palpitations, dizziness, presyncope, and syncope,^230,232^ but the disease can also occur asymptomatically. Only patients with untreated symptomatic SND have a high risk of deterioration to cardiovascular events such as atrial fibrillation, heart failure, and systemic thromboembolism. This is also reflected by the finding that age-related SND and chronotropic incompetence are associated with an increased risk of cardiovascular death and overall mortality.^84^ In a broader context, disturbances in HR regulation by the ANS generally play a significant role in the development and progression of cardiovascular disease. In support of this, there is a well-established link between ANS dysfunction (and subsequent HR dysregulation) and various cardiovascular diseases such as myocardial infarction, congestive heart failure, and arterial hypertension.^234^ In addition, pathological input from the ANS can potentially cause secondary SND,^84^ and primary SND can be worsened by ANS imbalance.^84^ Furthermore, bradycardia and SND can develop after cardiac transplantation.^233^ In hypertension, an increased HR will further increase blood pressure, leading to the worsening of the disease and, as a consequence, a higher risk of myocardial infarction and stroke.^235^ Moreover, there is an important link between chronic heart failure (CHF) and SND^84^: both diseases can arise as a consequence of the other, meaning that CHF can develop secondary to SND and vice versa. In this context, atrial fibrillation due to SND can lead to or worsen CHF. Furthermore, tachy-brady syndrome due to SND can have worsening effects in compensated CHF, possibly leading to decompensation and progression to terminal stages of the disease. Similarly, myocardial ischemia can be worsened by tachy-brady syndrome but can also be an important driver of SND. For example, acute stenosis or thrombosis of the SAN artery and myocardial infarction favor SND.^84^ To date, the only definitive treatment option for SND is the implantation of an electronic pacemaker device.^236^
Several mutations in ion channels have been identified as the cause of familial forms of primary SND. For a comprehensive analysis of all the human mutations found to be causative of SND, please refer to references ^84,233,237^. Here, we specifically focus on SAN diseases caused by mutations in HCN4 (Table 4). Overall, HCN4 mutations have been linked to sinus bradycardia, chronotropic incompetence, inappropriate sinus tachycardia, and noncompaction cardiomyopathy. The link between changes in If and the HR of patients with associated HCN4 mutations has recently been confirmed in silico.^238^ Noncompaction cardiomyopathy is a congenital disease with not yet fully clarified pathogenesis. It is generally thought to arise from disturbed compaction of the trabeculated myocardium during embryogenesis, which takes place at around 5 to 8 weeks of gestation in humans.^239–241^ In line with this, there is evidence from mouse studies that HCN4 is expressed during early embryonic development in cardiac progenitors of the first heart field and is only later downregulated in the working myocardium.^83,241–244^ Accordingly, HCN4 could have an important signaling function in the normal compaction process, and dysfunction of HCN4 might interfere with molecular mechanisms required for cardiac development.^239,241^ This may cause hypertrabeculation that ultimately leads to noncompaction cardiomyopathy.^241^
Given the important role of HCN channels in these and other physiological and pathophysiological conditions, great efforts have been made to explore the channels as targets for therapeutic intervention.^245–247^ Besides cardiac expression, HCN channels play important roles in neuronal excitability and are linked to diseases of the nervous system such as epilepsy, neuropathic pain, and affective disorders.^58,246,248–250^ Accordingly, several established anticonvulsant, analgesic, anesthetic, and antidepressant drugs have been shown to interact with HCN channels,^246,248,249,251^ which may account for at least part of their therapeutic effect. Because the expression profiles of the 4 HCN channel isoforms vary between different tissues and cell types, there is ongoing work to identify subtype-specific compounds as potential drug candidates.^250–254^ For example, a series of brain-penetrant HCN1-specific inhibitors have recently been identified that enhance working memory while lacking cardiovascular side effects, qualifying them as possible candidates for the treatment of cognitive dysfunction in brain disorders.^255^ Moreover, there is ongoing research on peptides based on TRIP8b, an auxiliary subunit that regulates surface expression and modulates the cAMP sensitivity of HCN channels in the brain.^256–259^ This includes the development of small molecules targeting the TRIP8b-HCN interaction,^260^ as well as the design of shortened membrane-permeable versions of TRIP8b as tools for potential therapeutic use.^261^ Given the crucial role of HCN channels in the development of neuronal and cardiac disease, the search for HCN-targeting drugs continues to be the subject of intensive research.^262–266^
To date, the only clinically approved drug that specifically targets HCN channels is ivabradine.^267–270^ It does not cross the blood-brain barrier and is, therefore, used as a selective bradycardia-inducing agent.^246,271^ It is currently approved by the US Food and Drug Administration and the European Medicines Agency for the treatment of CHF with systolic dysfunction^267,269,272,273^ and the European Medicines Agency for the symptomatic treatment of chronic stable angina pectoris in patients with coronary artery disease.^269,274–276^ Ivabradine enters and blocks the pore of HCN channels from the intracellular side, thereby reducing If and decreasing SDD and the firing rate of pacemaker cells, allowing for a dose-dependent reduction in HR.^277–280^ It can access HCN4 only when the pore is opened by hyperpolarization.^279,280^ The resulting use dependency enables increased efficiency of ivabradine at higher HRs,^280,281^ which has been proven beneficial in patients with normal sinus rhythm at initial HRs of ≥70–75 bpm.^269,282,283^ Interestingly, at the cellular level, the effects on SDD caused by blocking the pore of the channel (ie, slowdown of SDD^280^) are different from those caused by loss or blockade of CDR (ie, induction of nonfiring but normal SDD during firing mode^25^). The reason for this apparent inconsistency could be that at therapeutic doses, only part of the channels are blocked, and only with further blockade, the cells would switch to the nonfiring mode. In support of this conclusion, even a partial block of If is sufficient to explain the clinically relevant HR-slowing effect of ivabradine.^284^ In fact, a clinically relevant ivabradine concentration of 3 μM slowed the spontaneous AP firing rate by 15% in isolated rabbit SAN pacemaker cells.^284^ Importantly, at a membrane potential of −60 mV, which roughly corresponds to the MDP, ivabradine (3 μM) reduced the If amplitude by 40%.^284^ However, in such experiments, repetitive activation/deactivation cycles were necessary to reach a substantial current block, and there was only a little block (around 6%) observed during steady-state activation with a single test pulse to −100 mV.^279,280^ As there is evidence that HCN channels remain open throughout the pacemaker cycle,^59^ the actual reduction in If under physiological conditions might be even smaller. Furthermore, part of the HR-reducing effect could also occur on the network level of the whole SAN due to an increased fraction of pacemaker cells entering the nonfiring mode. Successively increasing block larger than 40% would induce bradycardia and, in addition, HR fluctuations and sinus pauses. Eventually, a 100% block of HCN4 would be expected to completely stop the spontaneous beating of SAN cells. This conclusion is supported by results reported in studies using inducible knockout of HCN4 in adult mice, which resulted in a large fraction of quiescent isolated cells.^73,74^ Such a level of reduction in HCN4 activity is not possible with ivabradine because loss of selectivity will already occur at doses, which do not block 100% of HCN channels.^285–287^ Moreover, in addition to quantitative dose-dependent effects, there could be qualitative differences in the effects caused by reducing the current amplitude (knockout of the channel or blocking the pore with ivabradine) and by shifting the activation curve, that is, blocking or switching off CDR of the channel. This could be explained by the fact that different structural domains within the channel are targeted (pore region versus cyclic nucleotide-binding domain) and need to be further investigated in future studies. Importantly, such a qualitative difference should be considered in the development of future drugs that interfere with cAMP modulation of If (eg, TRIP8b-related molecules).
Given the limited treatment options available, different strategies have been used to develop biological pacemakers that could potentially replace the electronic devices currently used for the treatment of SND and other arrhythmias. For reviews and the most recent progress, please refer to references ^288–292^. In line with the focus of this review, it is worth mentioning that many of the approaches used to generate a biological pacemaker involve HCN channels. Some strategies are based on in situ injection of viral constructs of wild-type or mutated HCN2, HCN1, or chimeric (HCN212) genes in canine left atrium, left bundle branch, or rat subsidiary atrial pacemaker.^70,77,293–296^ Overexpression of solely the HCN channel induces spontaneous pacemaker activity at the site of injection. This fits well with the new mechanistic insights from native pacemaker cells that a certain level of If is important to maintain the firing mode and keep nonfiring at an appropriate balance. Other approaches involve genetically engineered cardiac or noncardiac cells to express HCN channels. Indeed, it has been shown that overexpression of HCN1 or HCN4 channels could modify ventricular myocyte monolayers and mesenchymal stem cells to become cardiac pacemaker cells.^297,298^ Transplantation of HCN2- or HCN4-mesenchymal stem cells induced stable firing activity in vivo for up to 6 weeks, displaying morphology adaptation to the site of injection.^299–303^ To date, the most promising approach to enhance pacemaker activity and autonomic response involves combining HCN channels with other genes, such as ADRB2 (beta[2]-adrenergic receptor), Ca^2+^-stimulated AC1, dominant-negative inward rectifier K^+^ channel (Kir2.1), or SkM1 (skeletal muscle Na^+^ channel 1).^304–308^ In particular, the HCN2/SkM1 gene hybrid approach achieved results that closely align with clinical requirements.^290,304^ Another promising approach to reprogram ventricular myocytes into pacemaker cells is the reexpression of transcription factors, such as Shox2, Tbx3, Tbx5, and Tbx18, which are physiologically involved in the differentiation of embryonic SAN cells.^237,290,309^ The expression of ≥1 of these transcription factors downregulates genes specific for the working myocardium and influences genes relevant for pacemaking.^310–314^ In view of the complexity of sinoatrial pacemaking and the differences in the various promising approaches to develop biological pacemakers, it will be interesting to see in which direction the field will progress in the future.
The chronotropic response is a complex physiological process by which HR is adapted to changing physical needs. The ANS densely innervates the SAN and regulates the membrane clock and calcium clock proteins to accelerate or decelerate the heart. Only recently, it has become more recognized that nonfiring or dormant pacemaker cells exist in the SAN and play an important role during the chronotropic response. Regulation of HCN4 by cAMP contributes to controlling the number of nonfiring cells rather than changing the frequency of firing cells. In this way, pacemaker channels do not contribute substantially to the frequency-increasing effect of the chronotropic response, while they do control other important aspects of the chronotropic effect, such as stabilization of basal HR and HR transitions, and also dampening of ANS input and limitation of maximal HR reduction by the vagal nerve. Dysfunction of this normally precise mechanism has important implications for a range of physiological and pathophysiological conditions, including the development and progression of cardiovascular disease.
This work was supported by the German Research Foundation (DFG) grants WA 2597/3-2, FE 1929/2-2, FE 1929/1-1, BI 484/5-2, SFB870 B05, and TRR152 P28.
None.