Mechanistic Insights Into Very Slow Conduction in Branching Cardiac Tissue
A Model Study
It is known that branching strands of cardiac tissue can form a substrate for very slow conduction. The branches slow conduction by acting as current loads drawing depolarizing current from the main strand (“pull” effect). It has been suggested that, upon depolarization of the branches, they become current sources reinjecting current back into the strand, thus enhancing propagation safety (“push” effect). It was the aim of this study to verify this hypothesis and to assess the contribution of the push effect to propagation velocity and safety. Conduction was investigated in strands of Luo-Rudy dynamic model cells that branch from either a single branch point or from multiple successive branch points. In single-branching strands, blocking the push effect by not allowing current to flow retrogradely from the branches into the strand did not significantly increase the branching-induced local propagation delay. However, in multiple branching strands, blocking the push effect resulted in a significant slowing of overall conduction velocity or even in conduction failure. Furthermore, for certain slow velocities, the safety factor for propagation was higher when slow conduction was caused by branching tissue geometry than by reduced excitability without branching. Therefore, these results confirm the proposed “pull and push” mechanism of slow, but nevertheless robust, conduction in branching structures. Slow conduction based on this mechanism could occur in the atrioventricular node, where multiple branching is structurally present. It could also support reentrant excitation in diseased myocardium where the substrate is structurally complex.
- slow conduction
- discontinuous conduction
- source-to-load mismatch
- atrioventricular node
- mathematical model
It is well-known that slow conduction of the cardiac action potential (AP) is essential for the function of the atrioventricular (AV) node.1 However, when slow conduction occurs in the working myocardium under pathological conditions, it can precipitate life-threatening reentrant arrhythmias.2
Conduction velocity (θ) is determined primarily by membrane excitability (ie, the magnitude and kinetics of depolarizing currents) and by the degree of intercellular gap-junctional coupling.1,3–6 In the AV node, both reduced excitability (conduction is based predominantly on the L-type Ca2+ current, ICa,L) and reduced coupling are the major mechanisms of very slow conduction.1 During acute ischemia, slow conduction in the myocardium is mostly due to hyperkalemia-induced membrane depolarization, which acts to inactivate sodium channels,7,8 and reduced intercellular coupling.9 In aging and fibrotic myocardium, and probably in the AV node as well, nonuniform anisotropy contributes to discontinuous slow conduction in the direction transverse to myocardial fibers.10,11
Conduction is also influenced by the presence of structural inhomogeneities within cardiac tissue. Such inhomogeneities can create an electrical source-to-load mismatch, eg, a misbalance between the current provided by a smaller mass of tissue (source) and the current necessary to bring to threshold a larger mass of tissue downstream (load). As shown previously, source-to-load mismatch can induce a local slowing of conduction or conduction block.12–15 It is hypothesized that, in addition to reduced excitability and reduced intercellular coupling, structurally-based source-to-load mismatch can also contribute to slow conduction in the AV node and in structurally complex myocardium. In the AV node, so-called “dead-end” pathways could play this role by shunting a fraction of electrotonic current.16,17 Furthermore, in parts of the AV node consisting of myocardial fascicles interspersed with collagen,18,19 branchings of these fascicles could form sites of source-to-load mismatch. In the myocardium, structural complexities can develop during chronic infarction: the infarct border zone consists of interconnected islands and strands of surviving myocytes intermingled with scar tissue and is known to support very slow conduction.20
The effects of branching on conduction were investigated recently in patterned cultures of neonatal rat ventricular myocytes.21 In preparations consisting of multiple successive strands branching from a main strand, slow conduction resulted from the repetition of source-to-load mismatch at the successive branching points. The main hypothesis suggested by these experiments was that branches would have a dual effect: (1) by acting as a current sink, they slow conduction (“pull” effect); (2) once depolarized, they would act as a current source, reinjecting current back into the main strand (“push” effect), stabilizing conduction. The pull effect is in accordance with the classical source-to-load mismatch concept. The push effect challenges the notion that a current load cannot switch to become a current source for the upstream tissue that caused its excitation. However, the existence of the push effect and its contribution to conduction could not be demonstrated and evaluated specifically in the experiments.
Motivated by the experimental observations, the aim of the present study was to gain deeper insight into the mechanism of slow conduction in branching structures. Specifically, the experimental observations were reproduced in networks of Luo-Rudy dynamic (LRd) ventricular model cells in order to (1) demonstrate the existence of the push effect, (2) quantify the current flowing retrogradely from the branches into the main strand (push current), (3) assess the contribution of the push effect/push current to the velocity and stability of conduction, and (4) characterize the conduction of premature APs in branching structures.
Materials and Methods
The LRd cell model,22 with its updates published in detail previously,23 was used to construct branching networks of cardiac cells reproducing the geometry of the cultures used in the experiments.21 Whereas the original LRd cell has a long cylindrical shape (100 μm length, 22 μm diameter), cultured rat heart cells are shorter (≈60 μm) and flattened against the growth surface.24 In order to use the LRd cell with dimensions closer to those of the cultured cells, the LRd cell was reshaped into a prism (58.8×58.8×11 μm) with identical surface-to-volume ratio and intracellular compartmentalization. Basic characteristics of conduction were assessed in an unbranched strand of 121 cells. The effects of a single branching on conduction were studied in this strand by connecting a pair of sealed-end branches of a predefined length (L) to its middle cell. The effects of multiple successive branchings were studied in strands of at least 121 cells emanating pairs of sealed-end branches of a predefined length and at predefined distances between branchings along the entire main strand. Previous work5 showed that spatial discretization into elements equal to the cell length (cell considered isopotential) leads to <0.5% error in computed θ compared with high-resolution subcellular discretization. Therefore, a discretization length equal to the cell length (58.8 μm) was used. The myoplasmic resistance (150 Ω · cm) was therefore lumped with the gap-junctional resistance (1.5 Ω · cm2) and the extracellular resistance was considered negligible.5,15 Membrane potential (Vm) was computed using the Crank-Nicholson algorithm25 at time-steps of 0.005 ms. Ionic currents and concentrations were recomputed at every time-step and integrated using the Euler method.
Conduction was examined in normal and elevated extracellular potassium ([K+]o=4.5 and 14.8 mmol/L, respectively). Elevated [K+]o, which was the same as that used experimentally,6,21 mimicked hyperkalemia during acute ischemia7,8 and created excitability conditions similar to those of the AV node (ICa,L-based conduction).1 Initial conditions were obtained from a single cell, which was paced for 10 seconds at 2 Hz. In order to reproduce θ, dVm/dtmax, and the space constant measured experimentally in the cell cultures for similar [K+]o,6,21,24,26 the sodium current (INa) was reduced by 50% and ICa,L was increased by 100%. In the unbranched model strand for normal [K+]o, θ was 38.7 cm/s, dV/dtmax was 121.4 V/s, and the space constant was 446 μm. In elevated [K+]o, these values were 14.5 cm/s, 16.0 V/s, and 314 μm, respectively.
If not specified otherwise, the results correspond to the first elicited AP (control, S1). Propagation of premature APs (S2), elicited at predefined S1S2 intervals, was studied in specific simulations. In further simulations, the specific contribution of the push current to conduction was investigated by not allowing retrograde current flow from the branches into the main strand. At every time step and for every branch, if Vm at the branch point was less than in the first cell of the branch, the corresponding intercellular conductance was set to 0; otherwise, it was kept at its nominal value.
The safety (stability) of propagation was quantified with the safety factor (SF).5 The SF is defined for every cell as the ratio of the charge generated by the cell to the charge required for its excitation. The fraction of SF above 1 represents the excess of charge generated over the charge required and quantifies the margin of safety. A recently generalized definition of the SF was used.27 The minimal SF in a given network was considered the determinant of the overall safety of propagation.
Propagation in Networks With Single Branching
Figure 1 illustrates the effects of single branching on propagation in the situation of short branches (L=2 cells, Figure 1A) and long branches (L=10 cells, Figure 1B). In both situations, the source-to-load mismatch resulted in local slowing of propagation proximal to the branch point (Figure 1, panel I), and thus in a conduction delay across the branching site (Figure 1, panel IV). Because the magnitude of the source-to-load mismatch depended on L, this delay was, as expected, longer for L=10 cells.
The activation pattern of short branches was very different from that of long branches. Short branches were activated virtually simultaneously (Figure 1A, panel II) and earlier than the distal portion of the strand. Furthermore, because of the reflection of electrotonic current from the sealed ends, Vm in the branches was greater than Vm at the branch point (main strand), and the gradient of Vm between the main strand and the branches was reversed. On the contrary, activation of long branches was gradual (Figure 1B, panel II), in a manner similar to the distal segment of the main strand, and accelerated only in close vicinity to the sealed branch ends. In networks with short branches, the reversal of the Vm gradient resulted in a push current (Figure 1, panel III), ie, in a retrograde electrotonic current flowing from the branches into the strand. In networks with long branches, the reflection of current at the sealed ends occurred too far from the branch point and no push current was observed.
Figure 2 shows the behavior of the peak push current as a function of L. In both normal and elevated [K+]o, this current first increased with increasing L, reached a maximum, and decreased toward 0. In normal [K+]o, the push current had a significant magnitude for L ranging from 1 to 7 cells (≈60 to 400 μm). The maximum push current was observed for L=2 cells and its peak amplitude (185 μA/μF) was comparable to that of peak INa (234 μA/μF). In elevated [K+]o, INa was 99.8% inactivated and propagation was supported by ICa,L. The push current was important over a broad range of L and its maximum (57 μA/μF), obtained at L=5 cells, was even larger than ICa,L itself (21.6 μA/μF).
The large magnitude of the push current suggests that this current can provide a major contribution to impulse propagation in the main strand. To assess this contribution, the push current was specifically blocked. This is illustrated in Figure 3 for L=2 and normal [K+]o. Surprisingly, this intervention resulted in only a very small prolongation of the overall conduction delay. This almost negligible effect is explained by the fact that, at the time of the onset of the push current, Vm at the branch point and immediately distal to it in the main strand had already undergone almost complete depolarization (Vm>0). Therefore, although the push current was large, it occurred too late to contribute to the activation of downstream cells within the main strand.
The dependence of the conduction delay on L is shown by the solid curves in Figure 4A. In close agreement with experimental results,21 the delay increased with increasing branch length and reached a plateau at ≈1 ms in normal [K+]o and at ≈6 ms in elevated [K+]o, respectively. Block of the push current induced only a negligible prolongation of the delay (dotted curves indicated by arrows, almost superimposed on the solid curves). This minimal effect was only present for shorter L, consistent with the observation that the push current is smaller for longer branches (see Figure 2).
The finding that the contribution of the push effect is small seems in disagreement with the experimental study.21 In the experiments, the push effect was suppressed by a pharmacological block of INa and ICa,L confined to the branches. This intervention resulted in a significant prolongation of the delay in normal [K+]o and conduction block in elevated [K+]o for L=960 μm. To address this discrepancy, the same protocol was reproduced computationally by setting INa and ICa,L to 0 inside the branches (Figure 4A, dashed curves). Consistent with the experiments, this led to a significant prolongation of the delay in normal [K+]o and to conduction block for L>700 μm in elevated [K+]o. Detailed analysis (not shown) revealed that blocking the depolarizing membrane currents in the branches indeed suppressed the push current. However, slowing or block of conduction was caused mainly by an increase of the pull effect. Because the branches could not generate inward transmembrane currents for local depolarization, they constituted an increased electrotonic load drawing a larger pull current during a prolonged period of time.
Figure 4B shows the dependence of the conduction delay on L for premature APs elicited during the relative refractory period of the control AP. Reduced excitability during that period (incomplete recovery from inactivation of INa and ICa,L) led to prolongation of the delay for all L. For moderate prematurity, no conduction blocks occurred. For more premature impulses, conduction failed at the branch point when L exceeded a critical length. With increasing prematurity, this critical length became progressively shorter.
Propagation in Networks With Multiple Branching
In consistency with the experimental observations,21 the succession of source-to-load mismatches in the model with multiple branching resulted in an overall slow conduction. We defined θ as the average velocity in the main strand. Figure 5A depicts the behavior of θ as a function of L and of the distance between successive branch points (D). In both normal and elevated [K+]o, for a given D, θ decreased with increasing L and reached a steady value. For a given L, θ decreased with decreasing D (ie, with increasing spatial frequency of branching). In elevated [K+]o, very slow conduction in the range of a few cm/s was observed for the ICa,L-based AP. The slowest θ was ≈2 cm/s (≈14% of θ in unbranched strands).
Figure 5B compares conduction patterns of control and premature APs in a network with L=15 cells and D=10 cells in normal [K+]o. Conduction of the control AP was characterized by equal delays at successive branch points. The premature AP elicited at a S1S2 interval of 210 ms was blocked at the first branch point. When elicited at 220 ms, it underwent a very long delay at the first branch point; therefore, conduction resumed in less refractory tissue beyond this point. Because of the dependence of the conduction delays on refractoriness (see Figure 4B), the delays at the next branch points were progressively shorter as the AP propagated in progressively less refractory tissue. Conduction was thus characterized by major initial slowing, which partially compensated for the prematurity. Distally in the network, conduction was similar to that of a premature AP elicited later, eg, at 240 ms. This phenomenon was observed for networks with D and L over the entire range studied.
The contribution of the push effect to conduction was further investigated by blocking the push current. Figure 6 illustrates the behavior of θ as a function of L and D when the push current was blocked at all branch points. θ in the absence of the push effect was normalized relative to θ in its presence. In normal [K+]o (Figure 6A), for D=2 cells, the block of the push effect resulted in a reduction of θ by up to ≈50% (for L=7 or 8 cells). For L>8 cells, propagation could not be supported when the push effect was suppressed. For D=4, 6, 8, and 10 cells, the reduction of θ was smaller (up to ≈10%) and only for short L. In elevated [K+]o (Figure 6B), the effects of suppressing the push current were very prominent; the reduction of θ ranged up to ≈50%, and for D=2, 4, and 6 cells propagation could not be supported in networks with relatively long branches.
The finding that the push effect provides such a large contribution to conduction is in sharp contrast to the observation that its contribution is insignificant in the strand with single branching. The data presented in Figure 7 explain why this contribution becomes important in the setting of multiple branching. Conduction along a network with L=4 cells and D=4 cells was simulated under conditions of elevated [K+]o (solid traces). In contrast to the single branching situation, the onset and the peak of the push current at a given branching site occurred at a time when downstream cells, located within a distance of electrotonic influence, were still undergoing depolarization. This delayed depolarization was due to the slowed conduction in the presence of multiple branches that acted to increase the electrical load. When the push current was blocked at the branching site under consideration (dashed traces), Vm at the corrersponding branch point was significantly depressed by ≈10 mV during the late phase of its upstroke. This depression resulted in a decreased driving force for current to the distal cells and, consequently, to further delay of their activation. Therefore, because the push current was occurring at the time of distal depolarization, it contributed to propagation by maintaining the late phase of the upstroke at the branch point and consequently the driving force for current to the depolarizing distal cells.
Safety Factor Considerations
Whereas the action of branches as current loads leads to slow conduction, it was hypothesized that the action of branches as current sources would contribute to the safety of propagation. Consequently, conduction in multiple branching structures would be slow but nevertheless stable.21 This issue was addressed by computing the safety factor for propagation (SF). Figure 8 compares the SF during slow conduction due to branching geometry to the SF during slow conduction due to reduced membrane excitability in the absence of branching. In nonbranching strands, SF and θ both decreased while maintaining a linear relationship when excitability was reduced (reduction of INa and/or ICa,L; straight thin lines in Figure 8).5 In branching networks, SF also decreased with θ. However, beyond a certain degree of conduction slowing, SF became higher when slow conduction was due to branching than when slow conduction was due to reduced excitability. This indicates that at these slow velocities, branching is a more robust mechanism for supporting slow conduction than reduced excitability.
By reproducing the characteristics of conduction in patterned cardiac cell cultures, this study provided mechanistic insight into the proposed “pull and push” mechanism of slow conduction.21 In essence, it demonstrated that a structurally-based current load can convert to a source, and that, if properly timed, this conversion can enhance conduction and make it more robust.
The Push Current Plays a Similar Role to ICa,L During Propagation That Involves Long Local Delays
In situations where INa-based conduction involves long propagation delays (several milliseconds), the role of ICa,L becomes critical by providing a crucial contribution to sustain AP propagation.5,15,28 During these delays, INa inactivates within ≈1 ms, but ICa,L progressively increases to its peak amplitude. By maintaining the plateau in proximal tissue, ICa,L maintains the driving force for depolarizing current to distal tissue. Similarly, in multiple branching tissue, conduction is sufficiently slowed so that the push current contributes to propagation by maintaining the plateau in proximal tissue at a time when distal tissue is not yet fully depolarized. Therefore, in the presence of sufficiently long local conduction delays, any delayed current that supports the plateau of the AP will provide a significant contribution to propagation. This principle is independent of whether this current is a transmembrane inward current or an electrotonic current flowing from neighboring tissue.
The Safety of Slow Conduction
A previous modeling study compared the effects of reduced INa with those of reduced gap-junctional coupling on SF in one-dimensional cell strands.5 Progressive reduction of INa resulted in progressive conduction slowing from 54 down to ≈17 cm/s (69% slowing) with a concomitant reduction of the SF until propagation failed. In contrast, during reduction of gap-junctional coupling, conduction slowing was first accompanied by a seemingly paradoxical increase of the SF. With further reduction of coupling, SF dropped abruptly when θ was <2 cm/s (≈97% slowing), until block occurred at <1 cm/s (>99% slowing). The SF increase was explained by the fact that with reduced coupling, a larger fraction of the depolarizing charge provided by transmembrane currents during the activation of a given cell accumulated in the cell because less charge was leaking downstream. Similarly, it could be hypothesized that in branching structures, charge would accumulate in the cells of a given branch, and that SF would increase as branching becomes more extensive. However, SF actually decreased with increased branching. This is explained by the fact that the large load imposed by distal branches increased in parallel with charge accumulation. In contrast, in the poorly coupled strand, the large junctional resistance decreased the downstream load. Nevertheless, the push current returned a fraction of this accumulated charge to the main strand. When L was long enough to provide a substantial push current, SF was higher in branching structures than in unbranched structures where an equal degree of conduction slowing was obtained by reducing excitability.
Frequency-Dependence of Propagation in Branching Structures
Two essential functions of the AV node are to protect the ventricles from atrial tachyarrhythmias and to regulate the delay between atrial and ventricular excitations.29 In the present study, propagation of premature APs across single branchings was characterized by long delays or even conduction failure. In multiple branching structures, propagation of premature APs was either blocked or markedly delayed. These impairments were caused by an increased source-to-load mismatch (less source current during the relative refractory period). In multiple branching structures, the push effect contributed to the conduction of premature impulses as well (block of the push current resulted in further conduction slowing or in the precipitation of conduction block, data not shown). Although the long-term behavior during rapid pacing was not assessed, these observations suggest that source-to-load mismatch in single or multiple branching structures might be involved in blocking or delaying the conduction of premature atrial impulses through the AV node and, possibly, in the establishment of characteristic beat-to-beat patterns (eg, intermittent block, Wenckebach periodicities). Furthermore, these phenomena might determine cycle length variations and spontaneous termination of reentrant arrhythmias involving strong head-tail interactions in discontinuous myocardium.
The Push Effect In Vivo: Limitations of the Present Study
In extrapolating the above findings and considerations to conduction in vivo, the present study suffers from the following limitations: (1) this study assumes that the branches have a sealed end; (2) both the AV node and structurally complex myocardium present a much more elaborate and irregular topology than the prototypical networks studied here; and (3) no histological information is available concerning the detailed branching patterns and their spatial scales and distribution in such structures. In vivo, it is likely that emanating branches merge with adjacent tissue (eg, in nonuniform anisotropic bundles where branches form side-to-side connections between myocardial fascicles).30 The pull and push effects will obviously be modulated by the activation of parallel fascicles or adjacent tissue. This modulation will further depend on the global activation pattern, ie, on the manner in which the wavefront enters into and spreads through the complex tissue and on the geometry of the excitation wave. Although the present study considers propagation in relatively simple structures, it shows that the push effect contributes to propagation over a wide range of L and D in structures with multiple branching. The documented presence of such structures in cardiac tissue supports the hypothesis that the pull and push effects are fundamental mechanisms in the establishment of slow conduction in vivo. However, elucidation of the detailed effects of more complex structures on conduction will necessitate further investigation.
Another limitation is that the cell model used is of a ventricular cell, and a similar study should be conducted with an AV-nodal cell model to support conclusions regarding the AV node. Nevertheless, at sufficiently elevated [K+]o, the ventricular cell AP upstroke is supported by ICa,L and mimics this aspect of AV-nodal cells. It is therefore likely that using an AV-nodal cell model would lead to observations similar to those reported in this study.
Physiological and Clinical Implications
The pull and push effects can act to increase the robustness of slow conduction in cardiac tissue that contains complex structures and structural heterogeneities. Such substrates exist in the AV node and in the healed infarct. It is likely that this mechanism plays an important role in the slow conduction through the AV node (a desired property essential to normal function) and in slow conduction in the infarct border zone that can lead to reentrant arrhythmias.
This work was supported by fellowships from the Swiss National Science Foundation and the Swiss Foundation for Stipends in Biology and Medicine (to J.P.K.), and by grants R01-HL49054 and R37-HL33343 from the National Institutes of Health, National Heart, Lung and Blood Institute (to Y.R.).
Original received June 25, 2001; revision received August 17, 2001; accepted August 29, 2001.
Meijler FL, Janse MJ. Morphology and electrophysiology of the mammalian atrioventricular node. Physiol Rev. 1988; 68: 608–647.
Wit AL, Hoffman BF, Cranefield PF. Slow conduction, reentry, and the mechanism of ventricular arrhythmias in myocardial infarction. Bull N Y Acad Med. 1971; 47: 1233–1234.
Cranefield PF. The Conduction of the Cardiac Impulse. New York, NY: Futura; 1975.
Joyner RW. Effects of the discrete pattern of electrical coupling on propagation through an electrical syncytium. Circ Res. 1982; 50: 192–200.
Shaw RM, Rudy Y. Ionic mechanisms of propagation in cardiac tissue: roles of the sodium and L-type calcium currents during reduced excitability and decreased gap junction coupling. Circ Res. 1997; 81: 727–741.
Rohr S, Kucera JP, Kléber AG. Slow conduction in cardiac tissue, I: effects of a reduction of excitability versus a reduction of electrical coupling on microconduction. Circ Res. 1998; 83: 781–794.
Kléber AG, Janse MJ, Wilms-Schopmann FJG, Wilde AAM, Coronel R. Changes in conduction velocity during acute ischemia in ventricular myocardium of the isolated porcine heart. Circ Res. 1986; 73: 189–198.
Shaw RM, Rudy Y. Electrophysiologic effects of acute myocardial ischemia: a mechanistic investigation of action potential conduction and conduction failure. Circ Res. 1997; 80: 124–138.
Vermeulen J, Tan H, Rademaker H, Schumacher C, Loh P, Opthof T, Coronel R, Janse M. Electrophysiologic and extracellular ionic changes during acute ischemia in failing and normal rabbit myocardium. J Mol Cell Cardiol. 1996; 28: 123–131.
Spach MS, Dolber PC. Relating extracellular potentials and their derivatives to anisotropic propagation at a microscopic level in human cardiac muscle: evidence for electrical uncoupling of side-to-side fiber connections with increasing age. Circ Res. 1986; 58: 356–371.
Spach M, Josephson M. Initiating reentry: the role of nonuniform anisotropy in small circuits. J Cardiovasc Electrophysiol. 1994; 5: 182–209.
Mendez C, Mueller WJ, Urguiaga X. Propagation of impulses across the Purkinje fiber-muscle junction in the dog heart. Circ Res. 1970; 26: 135–150.
Rohr S, Salzberg BM. Characterization of impulse propagation at the microscopic level across geometrically defined expansions of excitable tissue: multiple site optical recording of transmembrane voltage (MSORTV) in patterned growth heart cell cultures. J Gen Physiol. 1994; 104: 287–309.
Goldstein S, Rall W. Changes of action potential shape and velocity for changing core conductor geometry. Biophys J. 1974; 14: 731–757.
Wang Y, Rudy Y. Action potential propagation in inhomogenous cardiac tissue: safety factor considerations and ionic mechanism. Am J Physiol. 2000; 278: H1019–H1029.
Anderson RH, Janse MJ, Van Capelle FJL, Billette J, Becker AE, Durrer D. A combined morphological and electrophysiological study of the atrioventricular node of the rabbit heart. Circ Res. 1974; 35: 909–922.
Weidmann S. Heart electrophysiology. Physiol Rev. 1974; 36: 155–169.
Woods WT, Sherf L, James TN. Structure and function of specific regions in the canine atrioventricular node. Am J Physiol. 1982; 243: H41–H50.
Racker DK, Kadish AH. Proximal atrioventricular bundle, atrioventricular node, and distant atrioventricular bundle are distinct anatomic structures with unique histological characteristics and innervation. Circulation. 2000; 101: 1049–1059.
De Bakker JMT, Van Capelle FJL, Janse MJ, Tasseron S, Vermeulen JT, Dejonge N, Lahpor JR. Slow conduction in the infarcted human heart: zigzag course of activation. Circulation. 1993; 88: 915–926.
Kucera JP, Kléber AG, Rohr S. Slow conduction in cardiac tissue, II: effects of branching tissue geometry. Circ Res. 1998; 83: 795–805.
Luo CH, Rudy Y. A dynamic model of the cardiac ventricular action potential, I: Simulations of ionic currents and concentration changes. Circ Res. 1994; 74: 1071–1096.
Faber GM, Rudy Y. Action potential and contractility changes in [Na+]i overloaded cardiac myocytes: a simulation study. Biophys J. 2000; 78: 2392–2404.
Rohr S, Schölly DM, Kléber AG. Patterned growth of neonatal rat heart cells in culture. Morphological and electrophysiological characterization. Circ Res. 1991; 68: 114–130.
Crank J, Nicholson P. A practical method for numerical evaluation of solutions of partial differential equations of the heat conduction type. Proc Cambridge Phil Soc. 1947; 43: 50–67.
Jongsma HJ, van Rijn HE. Electrotonic spread of current in monolayer cultures of neonatal rat heart cells. J Membr Biol. 1972; 9: 341–360.
Azene EM, Trayanova NA, Warman E. Wave front-obstacle interactions in cardiac tissue: a computational study. Ann Biomed Eng. 2001; 29: 35–46.
Rohr S, Kucera JP. Involvement of the calcium inward current in cardiac impulse propagation: induction of unidirectional conduction block by nifedipine and reversal by Bay K 8644. Biophys J. 1997; 72: 754–766.
Meijler FL, Jalife J. AV node function during atrial fibrillation.In: Mazgalev TN, Tchou PJ, eds. Atrial-AV Nodal Electrophysiology: A View From the Millenium. Armonk, NY: Futura Publishing Company; 2000: 251–268.
Spach M, Dolber P, Heidlage J. Properties of discontinuous anisotropic propagation at a microscopic level. Ann N Y Acad Sci. 1990; 591: 62–74.