This article focuses on the regional heterogeneity of the mammalian sinoatrial (SA) node in terms of cell morphology, pacemaker activity, action potential configuration and conduction, densities of ionic currents (iNa, iCa,L, ito, iK,r, iK,s and if), expression of gap junction proteins (Cx40, Cx43 and Cx45), autonomic regulation, and ageing. Experimental studies on the single SA node cell to the whole animal are reviewed. The heterogeneity is considered in terms of the gradient model of the SA node, in which there is gradual change in the intrinsic properties of SA node cells from periphery to centre, and the alternative mosaic model, in which there is a variable mix of atrial and SA node cells from periphery to centre. The heterogeneity is important for the dependable functioning of the SA node as the pacemaker for the heart, because (i) via multiple mechanisms, it allows the SA node to drive the surrounding atrial muscle without being suppressed electrotonically; (ii) via an action potential duration gradient and a conduction block zone, it promotes antegrade propagation of excitation from the SA node to the right atrium and prevents reentry of excitation; and (iii) via pacemaker shift, it allows pacemaking to continue under diverse pathophysiological circumstances.
Time for primary review 42 days.
The sinoatrial (SA) node is the pacemaker of the heart. More than any other tissue in the heart (apart from the atrioventricular node) the SA node is a complex tissue and its function depends on this complexity. This is the central thesis that this review will attempt to establish. From the study of the electrophysiology of single SA node cells, in particular, the pacemaking of the SA node is reasonably well understood. However, the relatively simple scenario that has emerged from these studies cannot account for the undoubted complexity of the SA node — for such well known phenomena as the marked heterogeneity of electrical activity throughout the SA node, the non-radial spread of the action potential from the leading pacemaker site in the SA node, the block of conduction from the leading pacemaker site towards the atrial septum, and pacemaker shift. We will attempt to show that these phenomena are related to the problems of (i) the relatively small SA node having to drive the large mass of atrial muscle without it being suppressed by the more hyperpolarized atrial muscle, (ii) protecting the SA node from reentry and invasion from action potentials from outside the SA node, and (iii) how pacemaking in the SA node must continue under diverse circumstances.
There have been many books and reviews on the SA node [1–9]. The first comprehensive monograph on SA node function and morphology was published by Brooks and Lu in 1972 . One review includes the interesting history of research into the SA node , one focuses more on the SA node of the dog and human  and another focuses on the ionic currents of the SA node .
2 Anatomy of the SA node
2.1 Gross architecture
Fig. 1 shows a schematic diagram of a SA node–atrial muscle preparation from the rabbit . In all species, the SA node is located in the right atrium at the junction of the crista terminalis (a thick band of atrial muscle at the border of the atrial appendage) with venous tissue — the superior and inferior vena cava, and the intercaval region between the two great veins [4,10–16]. The size of the SA node varies in different species and is reviewed by Opthof [4,17]. In the human, published photographs show the SA node beneath the epicardial surface of the crista terminalis [18,19]. In other species, at least part of the SA node lies in the intercaval region. In the cat, the SA node from the intercaval region can rise up the epicardial face of the crista terminalis before terminating . In the dog, the SA node in the intercaval region appears to abut against the crista terminalis . In the rabbit, SA node tissue from the intercaval region rises up the endocardial face of the crista terminalis and terminates at the right branch of the sinoatrial ring bundle (Fig. 1). Similarly, in the monkey SA node tissue rises up the endocardial face of the crista terminalis . In most species (rabbit, rat, guinea-pig, cat, dog, pig, monkey [12–16,20–22]), perhaps even human , but apparently not cow , the SA node may extend from the superior to the inferior vena cava.
Within the intercaval region, the SA node may occupy the entire thickness between the endocardium and epicardium as in the rabbit, guinea-pig and monkey [12,13,16]. However, in the human, dog and pig there is a layer of atrial muscle between the SA node and endocardium [15,18–20]. The atrial muscle in the intercaval region together with extensive connective tissue (see below) is thought to protect the region against the high wall stresses.
One question to be considered in this review is how the relatively small SA node can drive the large mass of atrial muscle surrounding it without being suppressed by it. Theoretically this is difficult . It is possible that this is in part achieved by restricting the connections between the SA node and atrial muscle, because contact between the two tissues over a large area would result in greater suppression of the SA node pacemaker activity. For this reason it is of interest to know how the SA node is connected to the atrial muscle; for example, is the SA node electrically connected to the atrial muscle throughout their common boundary? It is possible that for much of the overlap between the SA node and atrial muscle there is connective tissue separating the two. Such a connective tissue barrier has been described in the human and cow  (but not in human according to Anderson and Ho ), rabbit  and monkey . The connective tissue barrier in the rabbit is shown schematically in Fig. 2 as the thick black band.
2.2 Fine architecture
A characteristic feature of the SA node is much connective tissue, mainly collagen and fibroblasts , although the extent of connective tissue is species-dependent and varies from 50% in the rabbit, guinea-pig and rat to 75–90% in the cat . Another characteristic feature of the SA node under the light microscope is a high density of nuclei; in the monkey, the density in the SA node is approximately double that in the atrial muscle . The high density of nuclei reflects the fact that SA node cells are small as compared to the surrounding atrial cells. The cells in the centre of the SA node are reported to be 5–10 μm in diameter in the human and dog , roughly spindle-shaped, 25–30 μm long and <8 μm in diameter in the rabbit , 20–30 μm long with an irregular profile in cross section and a diameter of <8 μm in the guinea-pig , <10 μm in diameter (often only 4–6 μm) in the cat , 40 μm long and 4–8 μm in diameter in the pig , and irregular and spindle-shaped with a diameter of ∼7 μm in the monkey . In contrast, atrial cells in these species are ∼100 μm in length and 15–20 μm in diameter. In the centre of the SA node, the cells are poorly organised and described as ‘interweaving’ in the human, dog and rabbit [12,26].
In the human and dog, in the centre of SA node, there are characteristic ‘P’ cells (or ‘typical nodal’ cells), which are believed to be the leading pacemaker cells . The cells are not only small as reviewed above but also ‘empty’. This is because they contain only a few poorly organised myofilaments (running in all directions and not organised into myofibrils). In the human and dog, typical nodal cells also contain fewer and randomly distributed mitochondria and little sarcoplasmic reticulum . Empty cells are characteristic of the SA node in a number of species: rabbit , guinea-pig , cat , pig  and monkey . However, the cells are less empty in the cat, pig and monkey, and they contain more and better organised myofilaments than in the other species . In rabbit , guinea-pig , cat , pig  and monkey , the presence of numerous caveolae on the cell membrane has been noted. The absence of myofilaments and mitochondria can be understood in a specialised tissue like the SA node, but the presence of numerous caveolae has not been explained. Could they increase the area of membrane available to receive neurotransmitters? An abundance of glycogen granules has been noted in the rabbit .
In the rabbit, there is a gradual transition in cell type over several millimetres from the centre of the SA node (the leading pacemaker site) in all directions to the periphery of the SA node, where the SA node meets the atrial muscle; there is no distinct border between the SA node and atrial muscle [12,28]. From the centre to the periphery of the SA node, the shape and arrangement of the cells become more regular and the myofilaments more numerous and better organised. In panels A and B of Fig. 1 the stippled area shows the region of small interweaving SA node cells, the yellow area (without stipples) shows the region of larger transitional SA node cells and the dashed yellow line shows the extent of peripheral SA node tissue lying over the atrial muscle of the crista terminalis. The myofilament content is least in the region of small interweaving cells and gradually increases in all directions from this region . In the human and dog, there is evidence of a similar gradual transition in cell morphology [21,26]. James et al.  described the nodal cell as the main cell in the SA node in the human and dog, but in addition to these there are ‘transitional cells’ with an increasing number of myofilaments and mitochondria. All stages of transition can be found, with some cells being similar to typical nodal cells and others similar to atrial cells. Some transitional cells are half nodal cell like and half atrial cell like. The transitional cells are located between the nodal cells and the atrial cells and James et al.  proposed that conduction from the nodal cells to the atrial cells occurs via the transitional cells. However, in the cat SA node, Opthof et al.  described a sharp transition between nodal cells in the SA node and atrial cells outside. Consistent with the description above of a gradual transition in cell type from the centre to the periphery, cells of various sizes can be isolated from the rabbit SA node . Isolated cells can also have various morphologies: Denyer and Brown  and Verheijck et al.  classified rabbit SA node cells into elongated spindle, spindle and spider cells — spider cells, unlike elongated spindle and spindle cells, have multiple cytoplasmic projections.
3 Electrical coupling in the SA node
3.1 Conduction velocity and space constants
Based on mathematical modelling, Joyner and van Capelle  concluded that some degree of electrical uncoupling of the cells within the SA node may be an essential design feature to protect the leading pacemaker site from the atrial muscle. The conduction velocity within the SA node is very low (0.03–0.05 m/s — Section 4.1) compared to that in the surrounding atrial muscle (∼1 m/s). Conduction velocity is determined by a variety of factors, but one of the most important is the coupling conductance between cells. The low conduction velocity in the SA node is one line of evidence of poor electrical coupling in the SA node. The space constant (the distance over which an electrotonic potential, caused by intracellular current injection, decays to 1/e of its original value) is a measure of electrical coupling between cells. In the rabbit SA node, it has been measured to be between 465 and 828 μm parallel to the crista terminalis and between 205 and 310 μm perpendicular to it [31–35]. The difference in space constants in the two directions can explain the anisotropy in the SA node, i.e. the higher conduction velocity parallel to the crista terminalis than perpendicular to it (see Section 4.1). The difference in the space constants in the two directions is probably the result of the orientation of cells in the rabbit SA node — the cells tend to be arranged in parallel with the crista terminalis [12,28]. The space constant of SA node is lower than that of atrial muscle  and this is again consistent with less electrical coupling in the SA node than in the atrial muscle.
3.2 Connexin phenotypes in the SA node
Ultrastructural studies using the electron microscope have demonstrated the presence of gap junctions in the SA node of rabbit, mouse, rat, bat, mole, dog, sheep, cow, monkey and man, although they are sparse and smaller in size compared with those in the surrounding atrial muscle . Masson-Pévet et al. , using electron microscopy, estimated that gap junctions along the cell border at the leading pacemaker site in the rabbit SA node occupy 0.2% of the cell border. From this, assuming a typical SA node cell is 6 μm in diameter and 20 μm in length with a surface area of 1000 μm2, they calculated the total surface area of gap junctions to be 2 μm2. The area of individual gap junctions in the same study was 0.87×10−2 μm2. Therefore, the number of gap junctions in the cell membrane of one primary pacemaker cell can be estimated to be ∼230 from the morphological observations.
Each gap junction comprises clusters of serially linked hemichannels (connexons) contributed by the two apposing cell membranes, giving a way for small molecules (<1 kDa) to pass between the two cell interiors . Each connexon is composed of six transmembrane proteins called connexins, a multigene family of conserved proteins, of which at least 13 members are known in mammals [40,41]. In the heart, mRNA for several connexins has been detected: Cx37, Cx40, Cx43, Cx45 and Cx46 [42–44]. Gap junction channels made from different connexin types in vitro are reported to show distinct unitary conductance, ionic selectivity and molecular permeability properties [45,46]. Cx43 is ubiquitous and abundant in the working myocardium (atrial and ventricular muscle) [39,47,48]. This is the not the case in the specialised conducting tissues of the heart [49–51].
Many immunohistochemical studies have focused on connexin phenotypes in the SA node, but the results have been inconsistent and conflicting. Anumonwo et al.  described Cx43-containing gap junctions in the rabbit SA node. Trabka-Janik et al.  also showed clear labelling of hamster SA node, the location of which had been confirmed by action potential mapping, with Cx43 antibodies. Other studies on rabbit, rat, guinea-pig, dog, cow and human, however, have failed to detect Cx43 in the SA node [21,24,47,50,54–57]. In the rabbit, Coppen et al.  found that Cx43 expression in the centre of the SA node is negligible compared to that in the surrounding atrial muscle, and that Cx40 and Cx45 are expressed in the Cx-43-negative area (Fig. 2). The dimensions and quantities of the Cx40 and Cx45 spots observed were much smaller than those of Cx43 spots in the atrial muscle, and this is consistent with the size and frequency of SA node cell gap junctions as revealed by electron microscopy (see above). Verheule et al.  also showed the absence of Cx43 in the pacemaker cells in the central part of the rabbit SA node, but they demonstrated the presence of Cx40 and Cx46 gap junctions in those pacemaker cells. Davis et al. [50,56] demonstrated the presence of Cx40 and Cx45 in the dog and human SA node region where Cx43 was undetectable.
Although discrepancies between different studies may be due to species differences in connexin expression in the SA node, the discrepancies may also be due to technical problems. Recognition of a specific immunohistochemical signal may be more difficult in SA node cells than in atrial and ventricular cells, because the gap junctions are smaller. The mean number of channels in each SA node gap junction in rabbits is estimated to be ∼90 channels . This is close to the limit for detection by immunofluorescence . Therefore, the absence of labelling does not necessarily mean the absence of the connexin. Complex anatomical architecture at the junction of the SA node and atrial muscle (interdigitations) as described below will make the situation unclear; the positive Cx43 labelling reported by some investigators might reflect gap junctions in atrial cells between layers of SA node cells. Unreliability of antibodies (cross reaction with other types of connexin) may also lead to false conclusions .
Anumonwo et al.  and Verheule et al.  measured macroscopic gap junction conductance (Gc) of rabbit SA node cell pairs and obtained variable Gc with average values of 2600 and 7500 pS, respectively. Single gap junction channel conductance was estimated to be 40–60 pS by Anumonwo et al.  and 133–241 pS by Verheule et al. . From these values, SA node pacemaker cells were considered to be coupled by only 30–60 functional gap junction channels. This is less than the ∼230 gap junctions per cell estimated from electron microscopy (see above), but there are several possible reasons for this (the number of gap junction channels per cell will be greater than the number of gap junction channels between any two cells; some gap junction channels may not be functional). Because of the high membrane resistance of SA node cells (∼1 GΩ), Anumonwo et al.  suggested that the minimal Gc required for the synchronisation of excitation would be low, ∼140 pS (∼three gap junction channels). In support of this suggestion, Cai et al.  in a modelling study determined that approximately four gap junction channels of 50 pS are needed for frequency entrainment. In experiments using an external circuit that couples two cells that are not physically connected, Verheijck et al.  demonstrated Gc-dependent electrical behaviour of coupled rabbit SA node cells: as Gc was progressively increased the cells exhibited (i) independent pacemaking, (ii) complex activity with mutual interactions, (iii) entrainment of action potential frequency at a 1:1 ratio with different action potential waveforms, and (iv) 1:1 frequency entrainment with virtually identical action potential waveforms. The critical value of Gc for 1:1 frequency entrainment was only 130–500 pS. These observations suggest that SA node pacemaker cells can be synchronised with a low Gc provided by a limited number (<10) of gap junction channels.
3.3 Spatial distribution of connexins in and around the SA node
Joyner and van Capelle  suggested that a gradual increase of intercellular coupling from the centre of the SA node towards the periphery is important for proper functioning of the SA node: in a computer model, such a change in intercellular coupling allowed the SA node to show pacemaking and drive the surrounding atrial muscle, whereas with an abrupt change in intercellular coupling the SA node was unable to drive the atrium or the SA node was made quiescent by the atrial hyperpolarizing load. In order to shed light on this issue, ten Velde et al.  examined the spatial distribution of Cx43 in atrial muscle surrounding and abutting the guinea-pig SA node. They used an immunohistochemical marker (anti-α smooth muscle actin, α-SMA) that specifically cross-reacts with guinea-pig SA node cells together with Cx43 antibody to label previously electrophysiologically mapped SA node. There was no gradual increase in the Cx43 labelling density at the border between the SA node and atrial muscle. Instead, there was an intermingling (or interdigitation) of strands of atrial cells (Cx43-positive but α-SMA-negative) and SA node cells (Cx43-negative but α-SMA-positive) . A similar interdigitating arrangement of bundles of Cx43-negative SA node and Cx43-positive atrial cells at the periphery of the SA node was reported in a immunohistochemical study on rat, cow and human hearts . Such interdigitation of well-coupled atrial cells and poorly coupled SA node cells might be important to ensure that SA node pacemaker cells are shielded from the hyperpolarizing influence of atrial muscle and yet able to drive the atrial muscle.
In the study of Kwong et al. , the dog SA node was composed mainly of Cx43-negative cells (many of them Cx40-positive), but amongst these were bundles of Cx43-positive SA node cells (also Cx40- and Cx45-positive). The Cx43-positive bundles appeared to abut atrial cells. In the same study, complementary data were obtained from dissociated cells from the dog SA node — two main populations of pacemaker cells were identified: 30–35% of cells expressed Cx43, Cx40 and Cx45 and ∼55% of cells expressed only Cx40 (the remaining 10–15% of cells had no detectable connexin expression). In the rabbit SA node, Coppen et al.  found that most boundaries between Cx43-positive cells and Cx43-negative (but Cx40- and Cx45-positive) SA node cells were sharply delineated, and no extensive interdigitation between the two cell types was apparent. Instead, in the periphery of the SA node both Cx43 and Cx45 were expressed (yellow region on crista terminalis in Fig. 2).
Cx43 in the dog SA node and in the periphery of the rabbit SA node could serve an important role as the preferential conduction pathway for the propagation of the action potential from the centre of the SA node to the atrial muscle. Such differential spatial expression of gap junction protein might provide the structural substrate for the putative transitional zone proposed by Joyner and van Capelle  to enable the SA node to drive the atrial muscle surrounding it and yet not be suppressed by it. More extensive experimental studies using a combination of electrophysiology and immunohistochemistry will be required to substantiate this possibility.
4 The functioning of the intact SA node
4.1 Activation sequence and regional differences in the action potential in the rabbit
The leading pacemaker site and the activation sequence in the rabbit SA node are shown in Fig. 1A (activation sequence can be seen as a movie at http://www.leeds.ac.uk/bms/staff/boyett/). In Fig. 1A the isochrones show the time in milliseconds for the action potential to conduct from the leading pacemaker site. The leading pacemaker site (asterisk) is just a small fraction of the total area of the SA node (Fig. 1A). Bleeker et al.  estimated that the action potential is first initiated in an area of 0.1 mm2 comprising ∼5000 cells. This is just ∼1% of the total area of the SA node [4,12]. The leading pacemaker site is located in the centre of the SA node and is typically 0.5–2 mm from the crista terminalis in the intercaval region in the area of small interweaving cells (Fig. 1A) . From here the action potential propagates preferentially in an oblique cranial direction towards the crista terminalis. Bleeker et al.  calculated the conduction velocity to be 2–8 cm/s or less around the leading pacemaker site. In experiments in which extracellular potentials were recorded from the endocardial surface, Yamamoto et al.  calculated the conduction velocity near the leading pacemaker to be 4.5 cm/s parallel to the crista terminalis and 3.0 cm/s perpendicular to it. In the periphery of the SA node, conduction was more rapid: the conduction velocity was 49.7 cm/s parallel to the crista terminalis and 36.3 cm/s perpendicular to it . The outcome of this is that the action potential arrives at the crista terminalis as a broad wavefront (Fig. 1A). In relation to the problem of the SA node driving the atrial muscle, the arrival of the action potential at the atrial muscle as a broad wavefront may have advantages, because if the action potential arrived at the atrial muscle at a point, it is possible that the action potential could be more easily suppressed by the atrial muscle. Fig. 1A also shows that conduction towards the interatrial septum is blocked. Activation of the interatrial septum must wait for the action potential to conduct around the top or bottom of the block zone (dark grey area in Fig. 1A). This block zone is thought to be physiologically important — it may have a protective function and help protect the SA node from reentry and invasion by action potentials from outside of the SA node .
Like cell morphology (Section 2.2) the form of the action potential is not uniform within the SA node. The diversity of action potentials has been extensively studied in the rabbit SA node [12,28,62–64]. In the centre of the SA node, the upstroke of the action potential is slow (<10 V/s), the action potential overshoot is low (<10 mV), the action potential is long (∼150 ms), the maximum diastolic potential is low (−60 to −70 mV) and the pacemaker potential is steep. From the centre of the SA node to the periphery of the SA node and then atrial muscle of the crista terminalis, there is a gradual increase in the action potential upstroke velocity and overshoot and maximum diastolic potential and a gradual decrease in action potential duration and steepness of the pacemaker potential. It has already been commented on that there is a progressive increase in myofilament content from the centre to the periphery of the SA node (Section 2.2) and Masson-Pévet et al.  showed that in the rabbit there is an inverse correlation between the slope of the pacemaker potential and volume density of myofilaments. In the periphery (but not centre) of the SA node, the action potential can also have a notch [62,63].
The changes in the action potential are complex and occur in two dimensions. The action potential duration is greatest at or near the leading pacemaker site and it decreases in all directions from this as shown in Fig. 3A (position of leading pacemaker site shown by asterisk in Fig. 3A). The distribution of action potential duration is roughly similar to the activation sequence (Fig. 1A). Thus, there is a downward gradient in action potential duration along the conduction pathway in and around the SA node. Because of this marked gradient in action potential duration, repolarization in the SA node occurs in the opposite direction to depolarization in the SA node [61,65] (illustrated by a movie at http://www.leeds.ac.uk/bms/staff/boyett/). A downward gradient in action potential duration along the conduction pathway appears to be a general rule in the heart (other examples being crista terminalis vs. atrial appendage, Purkinje fibres vs. ventricular muscle, ventricular sub-endocardium vs. sub-epicardium and ventricular base vs. apex) and is thought to be a protective mechanism to help prevent reentry . It is interesting that the action potential duration gradient in the SA node is substantially greater than that elsewhere in the heart .
Down the centre of the intercaval region, as well as the maximum for action potential duration (Fig. 3A) there is also a maximum for pacemaker slope (Fig. 3C) and minima for action potential peak, diastolic potential and upstroke velocity (Fig. 3B) . However, the parameters are not distributed in the same way. The leading pacemaker site (asterisk in Fig. 3A–C) occurs in the region of maximum action potential duration as already mentioned (Fig. 3A) and maximum pacemaker slope (Fig. 3C), but not minimum action potential peak, diastolic potential and upstroke velocity (Fig. 3B). In Fig. 3 the block zone is shown by the thick black line — the block zone occurs in the region of minimum action potential peak and upstroke velocity (Fig. 3B), but not maximum action potential duration (Fig. 3A) and pacemaker slope (Fig. 3C).
In the block zone, a spectrum of action potentials can be seen: they are usually small and slow and, although some have a normal appearance, the action potentials often show two components [12,34,34,65,66]. In the block zone in the inferior part of the SA node, depolarizations with amplitudes of only a few tens of millivolts, rather than action potentials, and even stable resting potentials have been observed (Fig. 4) . Interestingly, the resting potentials can be ∼−75 mV , whereas the resting potential of the SA node at the leading pacemaker site (when spontaneous activity is stopped) is ∼−40 mV . The block of conduction in this zone could be the result of poor electrical coupling between cells or poor excitability. The results above are suggestive of poor excitability and this was also suggested by the study of Bleeker et al.  (see also Opthof et al. ). Bleeker et al.  showed that the block zone in the rabbit is a region of complete block rather than just slow conduction, because if the action potential was prevented from propagating around the region of block (by cutting the tissue superior and inferior to the block zone) the action potential was seen to fail to propagate across the block zone — it died out. They showed that the two component action potentials often seen in the block zone are the result of the collision of two wavefronts — the one attempting to propagate across the block zone and the one that has travelled around the region of block and is invading the block zone from the other direction. Finally, Bleeker et al.  showed that the space constant of the block zone is similar to that elsewhere in the SA node — this suggests that electrical coupling is at least as good as elsewhere in the SA node. Support for this comes from the observation that there is no break in connexin expression in the block zone .
4.2 Activation sequence and regional differences in the action potential in other species
Preferential conduction from the leading pacemaker site in the oblique cranial direction and conduction block towards the septum are also seen in the cat  and pig . In the guinea-pig, a similar pattern is seen, although the block zone is differently placed . In the monkey, the activation sequence is different in that the action potential from the leading pacemaker site propagates preferentially in an oblique caudal direction to the crista terminalis . However, the effect is the same — the action potential arrives at the crista terminalis as a broad wavefront. In the monkey, as in the other species, there is conduction block from the leading pacemaker site to the septum . In the guinea-pig , cat , pig  and monkey  similar regional differences in the action potential are seen to those in the rabbit, although two component action potentials in the block zone are unusual in the cat  and absent in the guinea-pig . In the guinea-pig  and pig , but not the cat , there is an inverse correlation between the slope of the pacemaker potential and volume density of myofilaments in SA node cells .
Compared to other species, much less is known of the activation sequence and regional differences in the action potential of the SA node in the dog and human. In the dog, extracellular electrodes on the epicardial surface have been used to map the spread of electrical activity . Similar recordings have been made intraoperatively in patients . In patients, the position of the apparent origin of the action potential measured in this way is highly variable (from close to the superior vena cava to close to the inferior vena cava) and there can even be multiple widely separated simultaneous points of origin. A similar picture is seen in the dog . However, it is unlikely that pacemaking in the dog and human is radically different from that in the small mammals. In both the dog and human, the SA node tissue is embedded in atrial muscle (but presumably electrically isolated from it) (see Section 2.1) and extracellular electrode recording at the tissue surface cannot discriminate between potentials resulting from SA node tissue and potentials resulting from atrial muscle (Section 5). Bromberg et al.  recorded the activation sequence from the epicardial surface of dog hearts in the way described and then made intracellular recordings from the SA node (in the adult dog, this is difficult and the epicardium had to be first removed from above the SA node tissue). It was shown that the activation sequence of the SA node tissue (recorded with intracellular microelectrodes) was not correlated with the activation sequence determined by extracellular electrodes . It is likely that the origin (even multiple origins) of the action potential when mapped using extracellular electrodes represents the exit point from the SA node to the atrial muscle and not the leading pacemaker site . The nature of these ‘exit points’ is unknown. The true activation sequence of the SA node of the dog (or human) has not been determined (because of technical difficulties) and presents a future challenge. However, Bromberg et al.  and earlier Woods et al.  made some intracellular recordings from the dog SA node and from these it can be tentatively concluded that the activation sequence and regional differences in the action potential (e.g. higher upstroke velocity in the periphery of the SA node) are not fundamentally different from those in small mammals. There is only one report of intracellular recordings of action potentials from the adult human SA node .
5 Extracellular potentials from the SA node
Any difference in transmembrane potential between electrically coupled cells should cause a current to flow through the cell interior, across the cell membrane and through the extracellular fluid (‘volume conductor’), giving rise to a potential gradient in the extracellular fluid.
5.1 Animal experiments
The extracellular potential changes associated with the pacemaker activity of the SA node were first recorded by Cramer et al.  from rabbit atria using a unipolar electrode (0.5 mm in diameter) connected to a high-gain direct-coupled amplifier. They found a slow negative wave (slope, −30 to −90 μV/s) and a more rapid negative wave (slope, −400 to −800 μV/s) closely correlated to the pacemaker potential and upstroke of leading pacemaker cells. This was confirmed by Haberl et al.  who also found variations of the extracellular potentials dependent on the length of SA node conduction time. Similar extracellular potentials were recorded by Cramer et al.  from the dominant pacemaker site of the dog SA node in vitro as well as in vivo with the use of low frequency (0.1 to 50–100 Hz) band pass filters. In all of these early reports, the SA node potentials were interrupted by large high-frequency deflections as cells in the surrounding atrium depolarized. In the SA node region, the amount of volume conductor is small and the electrical coupling between cells is weak compared to those in atrial muscle [52,74]. Extracellular potential changes localised to the SA node are, therefore, easily masked by relatively large far-field potentials from the atrial muscle near the SA node. This inhibits study of the morphology of the extracellular potentials in relation to the spread of excitation in and around the SA node.
Recently, Yamamoto et al.  reported a better technique to record extracellular potentials in the SA node. They used modified bipolar electrodes with a tip diameter of 0.1 mm. The indifferent electrode was placed 1 mm above the recording site to minimise far-field potentials by a high level of common-mode rejection. Fig. 5 shows recordings from the endocardial surface of the rabbit SA node. Extracellular potentials showed a variety of morphologies. In a small area near the leading pacemaker site, a slow negative wave was preceded by a gradual increase of the negativity, and was normally followed by a second slow negative wave (Fig. 5B: m). At the periphery of the SA node superior and inferior to the leading pacemaker site, slow positive/negative waves were recorded (Fig. 5B: e, t). On the septal side of the SA node with slow conduction, long, slow, positive waves were recorded (Fig. 5B: f, n). In the atrial muscle surrounding the SA node, the extracellular potentials showed a sharp positive wave followed by a short negative wave (Fig. 5B: j, o, u). When transmembrane action potentials were recorded simultaneously with the leading pacemaker-type extracellular potentials, the initial slow negative wave coincided with the upstroke phase of the action potential, whereas the second slow negative wave coincided with the repolarization phase of the action potential, and the gradual increase of negativity at the end of the electrical diastole corresponded with the terminal phase of the diastolic depolarization. These characteristics are consistent in part with previous reports [64,72,73], but the interruption of the waves by sharp deflections reflecting atrial activity (far-field potentials) was minimal or negligible. This made it possible to interpret a wide variety of morphologies of extracellular potentials in terms of the local current generated at the recording site. For instance, the cells at the leading pacemaker site in the rabbit SA node depolarize first but repolarize last because of a prominent gradient of action potential duration from the centre of the SA node, through the periphery of the SA node to the atrial muscle (Section 4.1). Consequently, the leading pacemaker site may play a role as a current source not only during depolarization but also during repolarization. This may result in the characteristic dual negative waves in the extracellular potentials. Pacemaker shifts induced by vagal nerve stimulation or pharmacological block of the Na+ current, iNa, or the L-type Ca2+ current, iCa,L, were also shown to result in changes in the morphology of the extracellular potentials . These facts suggest that the endocardial extracellular potentials recorded in and around the SA node under appropriate conditions may provide useful information, helping recognition of the leading pacemaker site and alterations of the conduction pattern and excitability of mammalian SA node .
However, care has to be taken in extrapolating the results obtained in rabbits to other species. The rabbit SA node is a superficial structure just beneath the endocardium, whereas the SA node of other species including human can be embedded within atrial tissue (Section 2.1). In these species, the extracellular potentials recorded from the endocardial surface will be the result of conduction from the exit point from the SA node and as a result the earliest extracellular activation may not correspond to the position of the leading pacemaker site as already discussed (Section 4.2).
5.2 Clinical implications
Recording of extracellular potentials from the SA node through catheter electrodes was first introduced into clinical electrophysiology in the 1980s [76–79]. This technique was expected to have an advantage over indirect atrial pacing methods in assessment of conduction disturbances from the SA node to the surrounding atrial muscle [80–82]. Identification of the SA node potential could also be important for preservation of normal pacemaker activity in catheter ablation procedures to treat reentrant atrial tachyarrhythmias. The potential usefulness of the technique has been, however, limited by large far-field potentials from atrial muscle close to the SA node [83,84]. This is probably due to the use of a relatively large catheter electrode tip (diameter >1 mm), and a large distance between the recording site and the indifferent electrode (10–15 mm). A closer setting of the indifferent electrode just above a recording tip of smaller size would result in a better isolation of potentials localised to the SA node as described in Section 5.1. The SA node potential recorded from patients through catheter electrodes is usually monitored in the reversed-polarity to see upward going potentials at the time of depolarization. This convention originally employed by Cramer et al. [72,73] and Haberl et al.  may be misleading. Extracellular potential waves are caused by local current through the extracellular fluid in association with a propagation of excitation; an action potential approaching the recording site from upstream produces a positive wave, whereas an action potential going away from the recording site downstream produces a negative wave. Accordingly, an alteration of the activation sequence can produce a fundamental change of the extracellular potentials even though a change of the transmembrane action potential may be minimal . We propose that the extracellular potentials recorded in and around the SA node should be presented using the correct polarity to facilitate the translation of their morphology into two-dimensional propagation patterns of excitation.
Introduction of new technology for more precise topological recognition of recording sites, such as catheter-based electroanatomical mapping using a magnetic field [85,86], may also help to draw more valuable information from the SA node extracellular potentials for understanding the complex pathophysiology of human SA node dysfunction.
6 Regional differences as studied using small balls
In 1982 we developed the technique of isolating a strand of small ball-like tissue specimens from the centre to the periphery of the rabbit SA node . This is an extension of a technique for the preparation of small balls of SA node tissue for double-microelectrode voltage clamp originally employed by Noma and Irisawa in 1976 . Fig. 6A shows a diagram of a preparation from the rabbit that included the whole SA node from which six balls were made from the sites indicated by the filled circles. Fig. 6B compares the characteristics of the action potentials from the six balls (closed circles) with those of action potentials recorded from the same tissue before the dissection (open circles); all of the data are plotted as a function of the distance from the sinoatrial ring bundle (the border between the SA node and atrial muscle). The two sets of data are similar and this suggests that the small balls are viable and show similar properties to tissue in the intact SA node. An important corollary of this observation is that the continuous gradient of action potential characteristics from the centre to the periphery of the SA node (Fig. 3) is the result of regional differences in the tissue and not the result of electrotonic interaction between the SA node and the surrounding atrial muscle.
6.1 Spontaneous action potentials
In balls from the periphery, the action potential upstroke velocity is faster, the action potential overshoot is greater, the maximum diastolic potential is more negative, the action potential is shorter, and the spontaneous rate is faster than in balls from the centre (Fig. 7) [63,67,87,89,90]. Most of these differences (except of spontaneous rate) are concordant with the regional differences in action potential configuration in the intact SA node (Fig. 3). In a recent study, in which the action potential characteristics of small balls of tissue from more superior and more inferior regions as well as the periphery and centre of the rabbit SA node were compared, the periphery–centre differences described above were shown to be just one component of a complex two-dimensional variation in these parameters . In the small balls of tissues (especially from the periphery), there was a decrease in maximum upstroke velocity, an increase of action potential duration and a decrease in spontaneous rate from the more superior tissue to the more inferior tissue. These changes (except of spontaneous rate) are again concordant with regional differences in the intact SA node.
The one discrepancy between small balls of SA node tissue and the intact SA node concerns pacemaker activity. The intrinsic pacemaker activity of small balls of tissue taken from different regions of the SA node is greater in balls from the periphery than in balls from the centre (Fig. 7). This paradox, which was also shown by Opthof et al.  in small pieces of rabbit SA node tissue, can be explained by the electrotonic influence of the atrial muscle. The periphery of the SA node is connected to a large mass of atrial muscle in the crista terminalis through gap junctions. The pacemaker depolarization in peripheral SA node cells will, therefore, be reduced by the hyperpolarizing current flowing from the non-pacemaking atrial cells, which have more negative diastolic potentials. The cells near the centre of the SA node may be subjected to less electrotonic interference from the non-pacemaking atrial muscle because of their greater distance from the crista terminalis and the poorer electrical coupling between cells. Evidence to support this idea was presented by Kirchhof et al. . They showed that, when the atrial muscle was cut away from the SA node, the leading pacemaker site shifted from the centre to the periphery of the SA node and there was an increase in the spontaneous rate of the preparation. In a modelling study using a massively parallel computer, Winslow et al.  examined the spontaneous activation pattern in the SA node and surrounding atrial muscle in a multicellular model, in which the regional differences in electrical activity of small balls of SA node tissue  were incorporated. When the SA node was not connected to the non-pacemaking atrial cell network, the action potential was initiated in the periphery of the node, but when it was connected to the atrial cell network, the leading pacemaker site was shifted from the periphery to the centre of the SA node and there was a decrease in the spontaneous rate. Similar results were obtained by Boyett et al.  using a multicellular model of the SA node and surrounding atrial muscle. In experiments using an external circuit that mimics the gap junction conductance (Gc), Watanabe et al.  demonstrated that the spontaneous activity of rabbit SA node cells was easily inhibited when they were connected to a membrane model (resistance–capacitance circuit) of an atrial cell even at relatively low Gc, >580 pS/cell (less than measured Gc — Section 3.2).
In the intact SA node of rabbit, Kerr et al.  found that the amplitude of a premature action potential deceased progressively from the periphery to the centre of the SA node. They, therefore, suggested that there is a progressive gradation of refractoriness from the periphery to the centre. Kodama and Boyett  applied premature stimuli to small balls of rabbit SA node tissue. The strength–interval curve was shifted upwards and to the right if the ball was more distant from the crista terminalis and closer to the centre of the SA node. The restitution curves of upstroke velocity and amplitude of premature action potentials showed a similar rightward shift, indicating slower recovery of excitability in balls from closer to the centre of the SA node. Such gradation of refractoriness is most likely the consequence of the different ionic currents responsible for excitation in the periphery and the centre of the SA node: iNa and iCa,L, respectively, as described in later sections. iCa,L requires a much longer time for repriming (reactivation) as compared to iNa[97,98].
6.2 Block of iNa and iCa,L
In the intact SA node, there is a gradual decline in the upstroke velocity of the action potential from the periphery to the centre (Section 4). Lipsius and Vassalle  and Kreitner  observed that in the intact SA node of the guinea-pig and rabbit tetrodotoxin (TTX), a selective iNa blocker, reduced the upstroke velocity in the periphery but not in the centre. In contrast, Verheijck et al. , Kodama et al.  and Yamamoto et al.  showed that, in the intact SA node of the rabbit, nifedipine, a potent iCa,L blocker, inhibited or abolished the action potential in the centre, whereas the action potential in the periphery was well preserved in the presence of the compound.
Kodama et al.  investigated the effect of block of iNa by TTX on small balls of rabbit SA node tissue. In the example shown in Fig. 8A, in the peripheral ball (left), the maximum upstroke velocity of the action potential was 100 V/s, and TTX reduced it to 5 V/s. TTX also reduced the take-off potential from −61 to −40 mV and this resulted in a substantial increase in cycle length. In the central ball, the maximum upstroke velocity was 4 V/s and TTX had no discernible effect on electrical activity (Fig. 8A, right). Under control conditions there were decreases in the take-off potential and maximum upstroke velocity in balls from the periphery to the centre (Fig. 7). In the presence of TTX, these gradients were abolished due to a significant reduction of the take-off potential and the maximum upstroke velocity in balls from the periphery . Kodama et al.  also investigated the effect of block of iCa,L by nifedipine on small balls of rabbit SA node tissue. Verheijck et al.  have shown that nifedipine (5 μM) selectively blocks iCa,L of rabbit SA node cells without affecting iNa, T-type Ca2+ current (iCa,T), delayed rectifier K+ current (iK) and hyperpolarization-activated current (if). In the example shown in Fig. 8B, on application of nifedipine, the action potential of the central ball was abolished (Fig. 8B, right). In marked contrast, nifedipine failed to abolish the action potential in the peripheral ball; instead the cycle length was shortened (Fig. 8B, left). This was the result of a shortening of the action potential and an increase in the pacemaker slope. In the presence of 2 μM nifedipine, on average, the rate was increased in ball A by 21%, increased or decreased in ball B, decreased in balls C and D by 86 and 78%, respectively, and abolished in ball E. Zaza et al.  showed in rabbit SA node cells that nifedipine-sensitive current is outward during the pacemaker potential, possibly as a result of a decrease in a Ca2+-activated outward K+ current. This finding is compatible with the observation in Fig. 8B, and may help explain the chronotropic effect of nifedipine.
These results show regional differences in the role of iNa and iCa,L in pacemaker activity of the SA node and Kodama et al.  suggested that iNa is responsible for the action potential upstroke in the periphery, whereas iCa,L is responsible in the centre. This can explain why the take-off potential and upstroke velocity are higher in the periphery and lower in the centre (Fig. 7B) (the threshold of iNa is higher than that of iCa,L), why block of iNa affects pacemaker activity in the periphery but not the centre, and why block of iCa,L abolishes pacemaker activity in the centre and not the periphery. However, the two currents do not play identical roles: in the centre, block of iCa,L leads to the abolition of the action potential, whereas in the periphery, block of iNa does not lead to the loss of the action potential . This is presumably because in the periphery iCa,L takes over the role played by iNa after block of iNa (iCa,L is present in both the centre and periphery).
In the rabbit SA node, Verheijck et al.  reported that there exists a substantial number of atrial cells (as well as SA node cells) even in the centre of the SA node, and the ratio of atrial cells:SA node cells gradually increases from the centre of the SA node (ratio, 41:59) to the periphery (ratio, 63:27). Atrial cells in the SA node has also been reported by others [21,57,60]. Verheijck et al.  also reported that morphological differences in the SA node cells are not associated with differences in action potential configuration and pacemaker activity (contrary to our own findings — Section 8). Based on these data, Verheijck et al.  hypothesized a ‘mosaic model’ (Fig. 9) in which there is a random mix of the two cell types and the ratio of atrial cells:SA node cells increases from the centre to the periphery of the SA node. If iNa is present in atrial cells but not in SA node cells, this will explain the regional variation in the roles of iNa and iCa,L. Our observations on single cells isolated from the rabbit SA node, however, suggest a ‘gradient model’ (Fig. 9), because we do not observe a significant number of atrial cells amongst cells isolated from the SA node and we do find morphological differences in SA node cells are associated with differences in action potential configuration and pacemaker activity (Section 8). According to the gradient model, there is a progressive regional variation in the properties of SA node cells from the centre to the periphery (Fig. 9). According to this model, the regional variation in the roles of iNa and iCa,L can be explained by a decline in the expression of Na+ channels from the periphery to the centre. Evidence for this is presented in Section 8. A greater density of iNa in the periphery may be functionally important: it may protect the SA node from the hyperpolarizing influence of the surrounding atrial muscle, because hyperpolarization will lead to a reduction of inactivation of iNa and this may help overcome the effects of hyperpolarization. iNa in the periphery may also help the SA node to drive the atrial muscle: using a multicellular model of the SA node, Zhang et al.  showed that the elimination of iNa from the periphery of the SA node resulted in SA node exit block (i.e. the SA node was no longer able to drive the atrial muscle).
6.3 Block of 4-AP-sensitive current
It is well known that transient outward current (ito) is present in atrial and ventricular cells and plays an important role in the early repolarization phase of the action potential . ito is known to be blocked by 4-aminopyridine (4-AP). ito is also known to be present in the SA node [30,105–108]. Boyett et al.  examined the effects of 5 mM 4-AP on small balls of rabbit SA node tissue. 4-AP increased the duration of the action potential and the increase was greater in peripheral tissue than in central tissue: 42–63% in balls A and B from the periphery and 21–22% in balls D and E from the centre. 4-AP also altered the spontaneous cycle length, increasing it in peripheral tissue (by 13–28% in balls A and B) but decreasing it in central tissue (by 5–26% in balls D and E). There is an increase in action potential duration from the periphery to the centre under normal conditions (Fig. 7) and it has been argued that this is a protective mechanism (Section 4.1) — this increase in action potential duration was no longer significant in the presence of 4-AP , which suggests that 4-AP-sensitive current is, in part at least, responsible for it. In peripheral tissue, action potentials with notches are often observed (a brief period of rapid repolarization after the action potential upstroke perhaps followed by a second period of depolarization); the notches were abolished by 4-AP in the study of Boyett et al. .
Boyett et al.  also examined regional differences in the effects of 4-AP in the superior–inferior direction in four strands of balls of tissue. In transitional/central balls (C and D), the average percentage increase in action potential duration was significantly greater in tissue from a more inferior region: 23% in strand 1 (superior) and 86% in strand 4 (inferior). 4-AP caused no significant change in cycle length in balls from strands 1–3, but significantly increased cycle length (by 30%) in balls from strand 4 from the more inferior region. These results suggest that the 4-AP sensitive currents (itrans and isus: ito and possibly iK,ur — Section 8.2) play a more important role in the periphery of the SA node than the centre, and in the more inferior region of the SA node than the more superior region. There are two possible reasons for the regional differences in the effects of 4-AP. First, the density of 4-AP-sensitive current may be greater in SA node cells in the periphery than in the centre (gradient model). Secondly, because the maximum diastolic potential is more negative in the periphery of the SA node than in the centre, the voltage-dependent inactivation of ito during diastole is expected to be less in the periphery. Voltage-clamp experiments on rabbit SA node cells by Honjo et al.  and Lei et al.  provide evidence for the former interpretation (Section 8.2).
6.4 Block of iK,r
There are three types of delayed rectifier K+ current: ultra-rapid (iK,ur), rapid (iK,r) and slow (iK,s). iK,r and iK,s at least are known to be present within the SA node [105,109–111]. In the intact SA node of the rabbit, Verheijck  showed that 0.2–1 μM E-4031, a selective iK,r blocker, reduced the upstroke velocity of the action potential, increased the action potential duration, reduced the maximum diastolic potential, reduced the slope of the pacemaker potential and increased cycle length; similar results have been obtained by others [111,112]. Verheijck  also reported that small tissue specimens containing the primary pacemaker cells (i.e. central tissue) were more susceptible to the inhibitory effects of E-4031 than large preparations including both the SA node and surrounding atrial muscle.
Kodama et al.  examined the effects of low (0.1 μM) and high (1 μM) concentrations of E-4031 on small balls of rabbit SA node tissue (examples of recordings are shown in Fig. 10). Near complete block of iK,r by 1 μM E-4031 [105,111] caused prolongation of the action potential followed by the cessation of spontaneous activity in all balls studied, regardless of whether they were from the periphery or centre of the SA node. A difference between peripheral and central tissue emerged when iK,r was partially blocked [105,111] by the lower concentration (0.1 μM) of E-4031: pacemaking continued in peripheral balls, but again ceased in central balls (Fig. 10). In the presence of 0.1 μM E-4031, the action potential peak and maximum diastolic potential were generally reduced in all balls, but the reduction was greatest in central balls: on average, the percentage decrease of the action potential amplitude was 28% in ball A from the periphery and 93% in ball E from the centre. The inhibitory effect of 0.1 μM E-4031 on the rate of spontaneous activity was also greatest in the central balls: on average the percentage decrease of rate was 39% in ball A from the periphery and 88% in ball E from the centre. Kodama et al.  also examined differences in the response to E-4031 in the superior–inferior direction in four strands of balls of tissue. In transitional/central balls (B, C and D), percentage decreases of action potential amplitude and spontaneous rate in response to 0.1 μM E-4031 were greater in more inferior strands. For instance, on average, percentage decreases in action potential amplitude and spontaneous rate in strand 1 from the more superior part of the SA node were 40 and 34%, respectively, whereas the corresponding values for strand 4 from the more inferior region were 77 and 79%, respectively.
In an immunocytochemical study on the distribution of ERG protein (responsible for iK,r) in the ferret heart, Brahmajothi et al.  demonstrated that labelling of ERG in the intercaval region distant from the crista terminalis is much less than that in the intercaval region abutting the crista terminalis. If the distribution of the ERG protein is the same in the rabbit SA node, it is possible that a lower density of iK,r in the centre of the SA node than in the periphery may underlie the higher sensitivity to 0.1 μM E-4031 of the centre than the periphery. Further evidence of a decrease in the density of iK,r from the periphery to the centre is given in Section 8.2. The higher sensitivity to 0.1 μM E-4031 of the inferior region of the rabbit SA node could also be the result of a lower iK,r density in the inferior region. For the moment, however, no evidence is available to support this interpretation.
It may appear paradoxical that because the density of iNa, ito and if is lower in the centre of the SA node (Section 8.2), block of the currents produces smaller effects on electrical activity in the centre of the SA node (Sections 6.2, 6.3, 6.5) [89,90] and yet a lower density of iK,r in the centre means that block of the current (by E-4031) produces a greater effect. This is a consequence of the different roles of the currents in electrical activity. iNa, ito and if are not required for spontaneous activity to persist; spontaneous activity can continue after complete block of the currents. Block of the currents results in changes in electrical activity and the magnitude of the changes is proportional to the density of the currents. However, iK,r is required for spontaneous activity to persist (as indicated by the fact that complete block of iK,r abolishes spontaneous activity). It follows from this that a minimal density of iK,r is required to sustain spontaneous activity. If the density of iK,r is less in central tissue, a smaller fraction of iK,r will need to be blocked to abolish spontaneous activity.
6.5 Block of if
Although it is widely accepted that if plays a role in pacemaker activity in the heart, the importance of if in pacemaking is controversial. In the SA node, DiFrancesco and co-investigators [114,115] have argued that if may be exclusively responsible for the pacemaker potential, whereas other investigators [116–118] have argued that its role may be minor.
Kreitner  was the first to suggest that the role of if may vary in different regions of the SA node, because block of if by Cs+ had little effect on the pacemaker potential in central tissue from the rabbit SA node, whereas it significantly decreased the slope of the pacemaker potential in peripheral tissue. This was confirmed by Nikmaram et al.  who examined the effects of three blockers of if: Cs+, zatebradine and ZD-7288. On average, the three blockers decreased the spontaneous rate of the intact SA node of the rabbit by 12, 16 and 13%, respectively. The decrease of pacemaker slope was maximal (69–120%) in the periphery and least (22–25%) in the centre. Nikmaram et al.  also examined the effects of Cs+ block of if on small balls of rabbit SA node tissue: Cs+ reduced the pacemaker slope and the spontaneous rate and the effects were greater in peripheral balls than central balls. On average, Cs+ decreased the pacemaker slope by 47% in ball A, 38% in ball B, 21% in ball C and 17% in ball D. On average, Cs+ decreased the spontaneous rate by 19% in ball A, 18% in ball B, 9% in ball C and 7% in ball D.
These results demonstrate that if plays a greater role in pacemaker activity in the periphery than in the centre. Incidentally, even in the periphery, the contribution of if to pacemaker activity is relatively small, and if is clearly not the only current involved. The different contribution of if in the periphery and centre can again be explained by both the mosaic and gradient models (Fig. 9). However, in Section 8.2 indirect evidence is presented in favour of the gradient model and of a decrease in the density of if from the periphery to the centre. A greater density of if in the periphery can be considered to be a protective mechanism, because it will protect the SA node from the hyperpolarizing influence of the surrounding atrial muscle, since hyperpolarization leads to further activation of if, and this will oppose and possibly overcome the hyperpolarizing influence.
7 Pacemaker shift
As highlighted by Schuessler et al. , the origin of the action potential in the SA node is not static — it is dynamic and changing according to the prevailing conditions. In humans, changes in P wave morphology occur spontaneously , during sinus arrhythmia , with vagal stimulation , with exercise , and during myocardial infarction . The changes in P wave morphology probably result from changes in the sequence of atrial muscle activation arising from a change in the exit site from the SA node [3,70,122,123]. The variability of exit sites from the SA node into the atrial muscle in humans has already been noted (Section 4.2). Furthermore, in humans, spontaneous changes in exit sites and changes in response to overdrive have been seen . In the dog, the exit site has been correlated with the rate : the rate was changed by sympathetic and parasympathetic nerve stimulation or agonists and antagonists and when the rate was increased the site was shifted in the superior direction and when the rate was decreased it was shifted in the inferior direction. It is likely that all of these changes result from changes in the position of the leading pacemaker site within the SA node in response to the changing conditions of the SA node.
Pacemaker shift has been principally studied in the rabbit heart. It occurs in response to a large number of different interventions: premature electrical stimulation , overdrive, sympathetic and parasympathetic stimulation [126,127], ACh , adrenaline , cardiac glycosides , nifedipine , 4-AP , E-4031 , a change in temperature [130,131], and a change in extracellular Na+, K+, Ca2+[130,131] and Cl−. Fig. 11, adapted from Opthof et al. , summarises the directions of the pacemaker shifts in response to the different interventions.
The leading pacemaker site is the site showing the fastest pacemaker activity. In response to an intervention that decreases pacemaker activity, the leading pacemaker site is expected to shift to the site of which the intrinsic pacemaker activity is least depressed. Peripheral balls are more resistant than central balls to nifedipine and E-4031 and thus the leading pacemaker site is expected to shift to the periphery; central balls are more resistant than peripheral balls to 4-AP and thus the leading pacemaker site is perhaps expected to shift further away from the periphery; superior balls are more resistant than inferior balls to 4-AP and E-4031 and thus the leading pacemaker site is expected to shift in the superior direction. The pacemaker shifts observed in response to nifedipine, 4-AP and E-4031 are consistent with these predictions (Fig. 11). This has been confirmed by Zhang et al.  using a mathematical multicellular model of the rabbit SA node and surrounding atrial muscle incorporating the likely regional differences in ionic current density (Section 8), in particular iNa in the periphery but not the centre and a higher density of iK,r in the periphery than in the centre: block of iCa,L or iK,r caused a shift of the leading pacemaker site from the centre to the periphery of the SA node as is observed experimentally (Fig. 11). Pacemaker shift within the SA node can be considered as a back-up phenomenon: when the pacemaker activity at the normal leading pacemaker site is inhibited, another site can immediately take over as the leading pacemaker. We suggest that the heterogeneity of ionic currents within the SA node (Section 8) provides multiple pacemaker mechanisms within the SA node, which ensures that the system is robust by providing an immediate backup in case of failure of one mechanism.
8 Electrical heterogeneity of single SA node cells
The finding that regional differences in the action potential are observed in small balls of SA node tissue as well as the intact SA node rules out the possibility that the regional differences are the result of electrotonic interaction between SA node and atrial muscle. However, the results from the small balls of tissue cannot be used to discriminate between the mosaic and gradient models (Fig. 9). Data from single SA node cells can help to discriminate: according to the gradient model, SA node cells should show a range of electrical activities comparable to that of the intact SA node. As already discussed, in the case of the rabbit SA node the size of pacemaker cells gradually increases from the centre towards the periphery (Section 2.2). This was recently confirmed by Lei et al.  who measured dimensions of single cells isolated from different regions of the rabbit SA node using confocal laser scanning microscopy: cells isolated from centre were small (length, 51.2±2.4 μm; width, 10.0±0.3 μm, n=58) and non-striated, while cells isolated from periphery were significantly larger (length, 87.8±2.7 μm; width, 11.5±0.3 μm, n=60, P<0.01) and 38% of them were striated. According to the gradient model, action potential characteristics and ionic current densities should be dependent on cell size and should follow regional differences in electrical activity.
8.1 Spontaneous action potentials
Verheijck et al. [29,100] reported that the rabbit SA node comprises three morphologically distinct nodal cell types (elongated spindle, spindle and spider). They reported that the electrical activity of SA node is heterogeneous, but there is no correlation with cell type (morphologically identical cells can show markedly different electrical activity; see also Oei et al. ). Partly because of this, they suggested that the gradual transition of electrical activity from the centre of the SA node to the periphery is the result of a gradual increase in the ratio of atrial cells:SA node cells from the centre to the periphery (mosaic model — Fig. 9). In contrast, Nathan  reported two types of pacemaker cells (type I and II) isolated from rabbit SA node and maintained for 1–3 days in culture. Type I cells were spindle-shaped and weakly beating and the action potential had a low maximum upstroke velocity (∼3 V/s). These features are characteristic of the centre. Type II cells were perhaps originally rod-shaped (although they rounded up in culture) and strongly beating and the action potential had a high maximum upstroke velocity (∼32 V/s). These features are characteristics of the periphery. These results are consistent with the gradient model (Fig. 9).
Honjo et al.  investigated the relationship between electrical activity and the size of rabbit SA node cells as measured by cell capacitance, Cm. In their study, SA node cells of different size showed different action potential configurations (Fig. 12A). Take-off potential, maximum upstroke velocity, action potential amplitude and maximum diastolic potential were significantly correlated with Cm; these parameters were greater in larger SA node cells (Fig. 12B). For example, relatively small SA node cells with Cm<30 pF had a maximum upstroke velocity of 10 V/s or less, whereas some larger cells had a maximum upstroke velocity of >60 V/s. The slope of the pacemaker potential and cycle length were also correlated significantly with Cm; the pacemaker potential was steeper and the intrinsic spontaneous activity was faster in larger cells (Fig. 12B). The electrical activity of small cells (presumably from the centre) is comparable to that of the centre of the intact SA node or balls of central tissue and the electrical activity of large cells (presumably from the periphery) is comparable to that of the periphery of the intact SA node or balls of peripheral tissue (see Figs. 3, 7 and 12). On the basis of these results, we tentatively suggest that the gradient model (Fig. 9) represents the most likely organisation of the SA node.
8.2 iNa, iCa,L, 4-AP-sensitive current, iK,r, iK,s and if
In order to understand the ionic mechanisms underlying the cell size-dependent variations in electrical activity of rabbit SA node cells, Honjo, Lei and their co-investigators [106,108,138,139] investigated the relationship between various ionic currents and cell size. Honjo et al.  observed in rabbit SA node cells that the densities of iNa and if were significantly correlated with Cm (Fig. 13A, G). In the experiments of Nathan  type II (putative peripheral), but not type I (putative central), cultured rabbit SA node cells (Section 8.1) possessed iNa. This is consistent with the work of Honjo et al. , although the work of Honjo et al.  suggests that there is a gradual change in cell type rather than just two cell types. The most extensive data concerning iCa,L is that of Lei et al.  and a significant correlation was observed between the density of iCa,L and Cm (Fig. 13B). Honjo et al.  and Lei et al.  examined 4-AP-sensitive current in rabbit SA node cells. The 4-AP-sensitive current was composed of transient and sustained components, itrans and isus[106,108]. The electrophysiological and pharmacological properties of itrans are similar to those of ito in other tissues. The sustained component (isus) could be a non-inactivating component of ito or a 4-AP sensitive ultra-rapid delayed rectifier K+ current, iK,ur. In experiments at room temperature, the density of isus was greater in larger cells, giving rise to a significant correlation with Cm, whereas the density of itrans was not significantly correlated with Cm. In experiments at 35°C, however, the densities of both itrans and isus were significantly correlated with Cm and their densities were greater in larger cells (Fig. 13C, D) . Lei et al.  separated the delayed rectifier K+ current in rabbit SA node cells into iK,r and iK,s with the aid of the specific blockers, E-4031 and chromanol 293B. The densities of iK (iK,r plus iK,s), E-4031-sensitive current and 293B-insensitive current (iK,r), and E-4031-insensitive current and 293B-sensitive current (iK,s) were correlated with Cm (Fig. 13E, F). Densities of the currents were all greater in larger cells. These observations suggest that cell size-dependent differences in the density of iNa, iCa,L, 4-AP-sensitive current (including ito), iK,r, iK,s and if are involved in regional differences in electrical activity within the rabbit SA node and the densities of all the currents are less in the centre than in the periphery. The putative regional differences in current density are consistent with the regional differences in the response of small balls of rabbit SA node tissue to TTX, nifedipine, 4-AP, E-4031, Cs+, zatebradine and ZD-7288 (Section 6).
It should be noted that, in contrast to the studies above, the data of Wilders and co-workers [140,141] suggest that in rabbit SA node cells of variable size the input conductance and density of major ionic currents (iCa, iK, if) are not proportional to Cm. However, Wilders et al.  include data for single SA node cells with Cm >115 pF. This value might represent clusters of SA node cells, because it is much larger than the maximum capacitance of rabbit SA node cells (∼65 pF). In any case, more experimental studies will be required to solve the discrepancy.
8.3 Other currents
It is possible that there are regional differences in other ionic currents. Noma and co-investigators [142,143] have shown the presence of sustained inward current, ist, in rabbit and guinea-pig SA node and argued that ist plays an important role in pacemaker activity. ist is activated on depolarization in the pacemaker potential range, carried mainly by Na+, enhanced by sympathetic agonists and blocked by Ca2+ antagonists . In the rabbit, Guo et al.  reported that ist was present in tapering spindle-shaped SA node cells (perhaps central cells) but not in small rod shaped ‘transitional’ cells (i.e. cells transitional between tapering spindle-shaped SA node cells and atrial cells; perhaps peripheral cells). In addition to the currents considered in detail, other currents have also been shown to be present in the sinoatrial node: iCa,T[144,145]; Na+–Ca2+ exchange current, iNaCa; background inward current carried mainly by Na+, ib,Na; Na+–K+ pump current, ip; ACh-activated K+ current, iK,ACh; ATP-sensitive K+ current, iK,ATP; ATP-activated cation current ; stretch-activated anion current . It is not known whether any of these currents vary regionally within the SA node. It has recently become apparent that intracellular Ca2+ can modulate pacemaker activity, perhaps for example by regulating Na+–Ca2+ exchange current, iNaCa[146,153–155]. A recent report showing that the expression of Ca2+-handling proteins (L-type Ca2+ channel, Na+–Ca2+ exchanger, ryanodine receptor, SERCA2) declines from the periphery to the centre of the rabbit SA node  raises the possibility that intracellular Ca2+ regulation of pacemaking may vary between the periphery and centre.
9 Computer modelling
Various biophysically-detailed mathematical models have been developed of rabbit (species for which most data are available) SA node action potentials based on data from voltage clamp experiments. The first models were developed by Yanagihara et al.  and Noble and Noble  and subsequent models [141,159–161] were based on the earlier models. The action potentials generated by all models are reasonable approximations of action potentials recorded experimentally.
Most models are of a ‘typical’ SA node action potential. Noble and Noble  and Winslow et al.  developed separate models for central and peripheral SA node cells, but the models were based on speculation rather than experimental data concerning regional differences. More recently, Zhang et al.  developed models of action potentials in central and peripheral rabbit SA node cells. Typical central and peripheral cells were assumed to be small and large with Cm of 20 and 65 pF, respectively. The models were based on data on the experimentally determined relationships between current densities and Cm (Fig. 13) — the chosen densities of iNa, iCa,L, ito (itrans), isus, iK,r, iK,s and if were close to experimentally-determined densities for cells with Cm of 20 and 65 pF and were assumed to be less in the central cell. The action potentials generated by the models are similar to those recorded experimentally both from small and large rabbit SA node cells (Fig. 14A) and from central and peripheral balls of tissue from the rabbit SA node (see Figs. 14A and 7A). The models were tested by comparing the effects of block of ionic currents on the models and small balls of tissue from the centre and periphery of the SA node (Section 6). In all cases (block of iNa, iCa,L, ito, iK,r, iK,s, if), the effects were qualitatively comparable. This work shows that the current densities measured in small cells assumed to be from the centre and large cells assumed to be from the periphery are appropriate to explain the form of the action potential and the response of the action potential to a range of ion channel blockers in the centre and periphery of the SA node.
Multicellular models are being used to explore the behaviour of the intact SA node. Most uses of multicellular models of the SA node are considered in other sections. Computer modelling has been used to consider the optimal size of the SA node — if it is too small, it will fail to drive the surrounding atrial muscle . It is known that conduction from the leading pacemaker site in the centre of the SA node proceeds to the crista terminalis, whereas conduction in the direction of the atrial septum is blocked (Fig. 1A). The conduction block is not the result of the absence of electrical coupling — it has been shown that it is the result of poor excitability (Section 4.1) and Zhang et al.  have proposed that it is the result of the absence of iCa,L. Using a multicellular model of the SA node, Zhang et al.  showed that such an absence of iCa,L will result in conduction block (Fig. 14B) — this is because the block zone being close to the centre of the SA node is likely to lack iNa as well and, therefore, there is no inward current to support an action potential. Because of the lack of inward current in the block zone there is decremental conduction and, furthermore, the collision of two wavefronts (one propagating from the centre towards the atrial septum and the other propagating around the block zone and then retrogradely propagating into the block zone) leads to two component depolarizations as is observed experimentally (see Figs. 14B and 4A).
10 Autonomic regulation
10.1 Cholinergic modulation
The most important action of ACh, released from parasympathetic vagal nerves, on the heart is a decrease in the heart rate, a negative chronotropic effect. How ACh slows the heart rate is not resolved. ACh affects the heart by binding to the M2 muscarinic receptor. In the rabbit SA node, the binding of ACh to the receptor leads to the activation of the ACh-activated K+ current, iK,ACh, and a hyperpolarizing shift in the if activation curve . It has been argued that if rather than the activation of iK,ACh is responsible for the slowing of rate especially at low ACh concentrations, because the inhibition of if occurs at lower ACh concentrations than the activation of iK,ACh and ACh can slow rate without producing any of the changes expected from significant activation of iK,ACh (increase of the maximum diastolic potential and shortening of the action potential) . Boyett et al.  have taken a contrary view to this, because block of if by Cs+ or zatebradine does not abolish the effect of ACh or vagal stimulation on the rabbit SA node — it can increase it. On the other hand, Ba2+ (a blocker of iK,ACh, although not a specific one) does greatly reduce it. On the basis of computer modelling, Boyett et al.  have suggested that the slowing of rate by ACh is primarily the result of the activation of iK,ACh. The decrease in rate per se is expected to increase if and this will antagonise the slowing, but the antagonism is minimized by the hyperpolarizing shift of if activation curve . In the rabbit SA node, ACh has also been reported to inhibit iCa,L[165,166], but such a decrease in iCa,L under basal conditions (i.e. in the absence of β-stimulation) has not been seen by other investigators, e.g. Honjo et al. . It is unlikely that an inhibition of iCa,L is responsible for the chronotropic effect of ACh, because inhibition of iCa,L by nifedipine does not necessarily lead to a decrease in rate — it can lead to an increase (Fig. 8B) .
A different scenario has been presented by Hirst and co-investigators [168–172]. They reported that, in the toad, vagal stimulation and exogenously applied ACh both caused hyperpolarization in quiescent preparations (activity stopped by nifedipine) and, whereas Ba2+ blocked the hyperpolarization caused by exogenously applied ACh, it did not block that caused by vagal stimulation [170,172]. Furthermore, vagal stimulation caused a decrease in membrane conductance, but exogenously applied ACh an increase . They concluded that whereas exogenously applied ACh may act by activating iK,ACh, ACh released from vagal nerves acts by inhibiting background inward current, presumably background inward Na+ current . However, this conclusion is difficult to reconcile with the evidence above for the rabbit SA node. In addition, low intensity vagal stimulation in rabbit SA node has been shown to increase membrane conductance . Finally, Noble et al.  has shown that modulation of rate by changes in background inward current is not effective, because although a decrease in background inward current will tend to decrease rate, the resultant increase in if will minimise such a decrease.
The ACh-activated K+ channel is a heteromultimer of GIRK1 and GIRK4. In the GIRK4 knockout mouse produced by Wickman et al. , heart rate regulation was abnormal — heart rate variability was abolished and they concluded that iK,ACh is responsible in total for fast beat-to-beat regulation of the heart rate. Furthermore, the bradycardia in response to baroreceptor stimulation (via the α-agonist, methoxamine) was halved. This study suggests that vagal control does involve iK,ACh, although care should be taken in extrapolating the data to larger animal species.
The effects of vagal stimulation vary regionally. In 1934, Brown and Eccles  first showed in the cat that brief vagal stimulation results in a complex change in heart rate: an initial decrease in rate, followed by a relative or absolute acceleration and, finally, a secondary decrease in rate. This observation has been confirmed many times since 1934, including in the human [177,178]. Kodama et al.  have suggested that this complex response is in part the result of regional differences in the response to vagal stimulation. They observed that in the centre of the rabbit SA node (leading pacemaker site) the principal effects of vagal stimulation (hyperpolarization and action potential shortening) were short lasting, whereas elsewhere in the SA node (peripheral, superior and inferior regions) and especially in the surrounding atrial muscle the principal effects (hyperpolarization and action potential shortening) were >10 times longer as a result of a lower concentration of acetylcholinesterase (enzyme that breaks down ACh). Kodama et al.  noted that the primary decrease in rate following brief vagal stimulation had a similar time course to the effects on the centre of the SA node and the secondary decrease had a similar time course to the effects on the atrial muscle. They suggested that whereas the primary decrease is a direct effect of vagal stimulation on the centre of the SA node, the secondary decrease is the result of the longer lasting effects of vagal stimulation on the surrounding atrial muscle and the electrotonic suppression of the SA node by the atrial muscle. They assumed that the principal effects of ACh (on both the SA node and atrial muscle) were the result of the activation of iK,ACh. The activation of iK,ACh in the atrial muscle is expected to increase the membrane conductance of the atrial muscle and, therefore, the suppressive effect of the atrial muscle on the SA node. Consistent with this hypothesis, when they cut the atrial muscle away from the SA node, the secondary decrease in rate following brief vagal stimulation was much reduced. This hypothesis that the atrial muscle can control the pacemaking of the SA node has been supported by other studies: first, in a study in which a rabbit SA node cell was coupled to a passive model of an atrial cell, increasing the conductance of the model atrial cell to mimic the activation of iK,ACh by ACh decreased the spontaneous rate of the SA node cell  (see also ). Secondly, in a multicellular model of the SA node and atrial muscle, the activation of iK,ACh in the atrial muscle by ACh slowed the spontaneous rate of the SA node as predicted .
Between the primary and secondary decreases in rate following brief vagal stimulation there is an acceleration of rate. In isolated rabbit SA node preparations, this acceleration cannot be attributed to the non-adrenergic, non-cholinergic vagal cardiac accelerator system [181,182], because it is abolished completely by atropine. In the centre of the SA node, brief vagal stimulation leads to a hyperpolarization as mentioned above, but this is followed by a depolarization (up to ∼10 mV; smaller in peripheral, superior and inferior regions of the SA node) [179,183]. The depolarization is abolished by block of if by Cs+ or zatebradine, which suggests that the depolarization is the result of the activation of if by the preceding hyperpolarization. The depolarization could be responsible for the acceleration as proposed by Spear et al. , and in support of this Boyett et al.  showed that Cs+ abolishes or reduces the acceleration.
Vagal stimulation and ACh cause pacemaker shift (Section 7). Such a shift has been seen in the rabbit [128,184] and dog . In the rabbit, vagal stimulation has been reported to cause a shift in the superior direction [183,184], whereas exogenous ACh has been shown to cause a shift in the inferior direction and towards the periphery, except at high concentrations when there is a shift in the superior direction (but also towards the periphery) . In the dog, vagal stimulation causes a shift in the inferior direction . The cause of the pacemaker shift is not known, but Duivenvoorden et al.  suggested that a decrease in the space constant in response to vagal stimulation may contribute to the shift by reducing the electrical coupling between cells. Different densities of muscarinic receptors  and vagal innervation  may also underlie the pacemaker shift. To investigate this issue, Mackaay et al. [126,128] separated the rabbit SA node into superior and inferior parts and showed that the inferior part is less sensitive to ACh; such a decrease in sensitivity can explain the pacemaker shift to the inferior part of the SA node that is usually seen in the rabbit and dog (see above).
10.2 β-Adrenergic modulation
Noradrenaline released from the sympathetic nerves increases the heart rate, a positive chronotropic effect. The mechanism underlying the chronotropic effect of noradrenaline is unresolved. Noradrenaline affects the heart by binding to the β-adrenergic receptor. In SA node, this leads to a depolarising shift in the if activation curve, a potentiation of iCa,L, a potentiation of iK, a hyperpolarizing shift in the iK activation curve, an acceleration of the deactivation of iK and a potentiation of ist[142,143,166,187]. The relative importance of the different actions is not known. Block of if by Cs+ has been reported to have no effect on the chronotropic effect of β-adrenergic receptor stimulation in the rabbit SA node , but block of if by zatebradine has been reported to inhibit the chronotropic effect of β-adrenergic receptor stimulation in the pig . Choate et al.  concluded that the effects of sympathetic nerve stimulation and exogenously applied noradrenaline on the guinea-pig SA node are different and whereas the effects of exogenously applied noradrenaline may be the results of the actions described above, the effects of sympathetic nerve stimulation may be the result of an enhancement of background inward current (however see comments in Section 10.1 on modulation of rate by background inward current). β-adrenergic receptor stimulation causes pacemaker shift (Section 7). A shift has been seen in the rabbit in the inferior direction  and the dog in the superior direction . Mackaay et al.  separated the rabbit SA node into superior and inferior parts and showed that the inferior part is more sensitive to β-adrenergic receptor stimulation; such a difference in sensitivity can explain the pacemaker shift to the inferior part of the SA node that is seen in the rabbit (see above). Beau et al.  showed differences in the density of β-receptors in different regions of the SA node of the dog; such differences may be involved in pacemaker shift.
11 Deterioration in the function of the SA node with ageing
The heterogeneity of the SA node may be important in the effects of ageing on the SA node. With age in humans the function of the SA node deteriorates: the intrinsic heart rate (i.e. the heart rate in the absence of autonomic nerve activity) declines and SA node conduction time increases (SA node exit block can occur) . Sinus node dysfunction may be related to the age-related deterioration. Alings et al.  have shown that similar age-related changes occur in the rabbit and cat. In the centre of the SA node is a region in which the upstroke velocity of the action potential is low (Fig. 3B), presumably as a result of low or no expression of the Na+ channel (Section 8.2). In the rabbit and cat SA node, Alings et al.  showed that the total area of SA node does not change with age, but the region in which the upstroke velocity is low increases in area, i.e. the upstroke velocity decreases in the periphery of the SA node to a value similar to that in the centre. The age-related decrease in the upstroke velocity of the action potential could account for the age-related decrease in the conduction velocity. A possible reason for the decrease in the upstroke velocity in the periphery is down regulation of the Na+ channel. This is not unreasonable: the Na+ channel in the centre of the SA node of the neonatal rabbit heart disappears by the time the animal is a young adult . Zhang et al.  showed that, in a multicellular model of the SA node incorporating regional differences in iNa etc., elimination of iNa from the periphery of the SA node slowed pacemaker activity and increased SA node conduction time; it could also result in SA node exit block. These changes are similar to the changes with ageing. Alings et al.  also observed an age-related increase in action potential duration in the rabbit and cat SA node and this raises the possibility that there are age-related changes in the density of other ionic currents.
The major conclusion from this review is that the SA node is a complex and non-uniform tissue. It is also tentatively concluded that the gradient model represents the most likely organisation of the rabbit SA node. Our working hypothesis is shown in Fig. 15. From the crista terminalis (CT) to the leading pacemaker site (LPS): (i) the volume density of myofilaments declines to a minimum (Fig. 15i); (ii) the upstroke velocity declines (Fig. 15ii) as a result of a decrease in the density of Na+ channels (Fig. 15iv); (iii) the action potential peak becomes less positive for the same reason and perhaps also as a result of a decrease in the density of L-type Ca2+ channels (Fig. 15v); (iv) the action potential duration increases markedly to a maximum as a result of decreases in the density of iK,r (Fig. 15vi) and ito (not shown in Fig. 15) channels and despite the decrease in the density of L-type Ca2+ channels (Fig. 15v); (v) the maximum diastolic potential becomes less negative, initially as a result of the loss of iK,1 channels (Fig. 15viii) and then as a result of a decline in the density of iK,r channels (Fig. 15vi); and (vi) the slope of the pacemaker potential increases to a maximum (Fig. 15iii) solely as a result of a gradual decline in the electrotonic influence of atrial muscle and despite a decline in the density of if channels (Fig. 15vii). In summary, at the leading pacemaker site pacemaking is vigorous, because the suppressive influence of the atrial muscle has declined and the density of if channels, although it has declined, is still sufficient to sustain pacemaking. Furthermore, even though the density of Na+ channels has dropped to zero, the action potential at the leading pacemaker site is sufficient to support non-decremental conduction, because the density of L-type Ca2+ channels is still sufficiently high (Fig. 15v).
From the leading pacemaker site (LPS) to the block zone (Blk): (i) the volume density of myofilaments begins to increase from its minimum (Fig. 15i); (ii) the upstroke velocity declines further (Fig. 15ii) perhaps as a result of a further decline in the density of Ca2+ channels (Fig. 15v); (iii) the action potential peak declines further for the same reason; (iv) the action potential duration declines from its maximum perhaps as a result of the continued decline in the density of Ca2+ channels (Fig. 15v) and an increase again in the density of iK,r (Fig. 15vi) and ito channels; (v) the diastolic potential becomes more negative again perhaps as a result of an increase in the density of iK,1 channels again (Fig. 15viii); and (vi) the slope of the pacemaker potential declines from its maximum (Fig. 15iii) perhaps as a result of a further decline in the density of if channels (Fig. 15vii) and an increase in the density of iK,1 channels (Fig. 15viii). In summary, in the block zone, possibly because of a continued decline in the density of Ca2+ channels and an increase in the density of iK,1 channels and the continuing absence of Na+ channels, the ‘action potential’ is no longer sufficient to support non-decremental conduction and conduction block is the result.
The non-uniformity of the SA node is vital for its normal functioning in many ways: There are still many gaps in our knowledge of the SA node. Although much is known about central–peripheral differences in the SA node, it is clear that there are important superior–inferior differences and yet little is known about these [63,65,67]. Although much is known about the SA node in small animals, relatively little is known about the human SA node, for example about its activation sequence, expression of ion channels and regional differences etc. . The heterogeneity of the SA node may be important in dysfunction of the SA node and this will be important to test.
Protection of the SA node from the hyperpolarizing influence of the surrounding atrial muscle. Possible mechanisms identified for this are: restricting the area of contact between the SA node and atrial muscle (Section 2.1); poor electrical coupling within the SA node (Section 3); a leading pacemaker site in the centre of SA node separated from the atrial muscle by the periphery of the SA node (Section 4.1); interdigitations between SA node and atrial tissue at the border between the two tissues (Section 3.3); a high density of if and iNa in the periphery of the SA node (Sections 6.2, 6.5).
Assistance to the SA node to drive the surrounding atrial muscle. Possible mechanisms identified for this are: an optimal size of SA node (Section 9); interdigitations between SA node and atrial tissue at the border between the two tissues (Section 3.3); the action potential arriving at atrial muscle as a broad wavefront (Section 4.1); iNa in the periphery of the SA node (Section 6.2); better electrical coupling in the periphery of the SA node (Section 3.3).
Protection of the SA node from invasion by action potentials from outside the SA node. Possible mechanisms identified for this are: longest action potential being at the leading pacemaker site in the centre of the SA node (Section 4.1) as a result of a particular pattern of expression of ion channels at the centre (Sections 6 and 8); block zone around a substantial fraction of the SA node (Section 4.1).
Ability of the SA node to continue to function as the pacemaker under diverse conditions through pacemaker shift. Pacemaker shift is the result of differences in pacemaker activity and its control in different regions. Possible mechanisms identified for this are: regional differences in the expression of ion channels (Section 7); regional differences in innervation (Section 10).
Pacemaker activity and intercellular communication. Boca Raton, FL: CRC Press, 1998.
Verheule S. Distribution and physiology of mammalian cardiac gap junctions. Dissertation, University of Utrecht, 1999.
Verheijck EE. Pacemaker currents in the sino-atrial node. PhD thesis, University of Amsterdam, 1994.
Opthof T. The mammalian sinoatrial node. PhD thesis, University of Amsterdam, 1986.