Anisotropic conduction block and reentry in neonatal rat ventricular myocyte monolayers

Abstract

Anisotropy can lead to unidirectional conduction block that initiates reentry. We analyzed the mechanisms in patterned anisotropic neonatal rat ventricular myocyte monolayers. Voltage and intracellular Ca (Ca_i) were optically mapped under the following conditions: extrastimulus (S1S2) testing and/or tetrodotoxin (TTX) to suppress Na current availability; heptanol to reduce gap junction conductance; and incremental rapid pacing. In anisotropic monolayers paced at 2 Hz, conduction velocity (CV) was faster longitudinally than transversely, with an anisotropy ratio [AR = CV_L/CV_T, where CV_L and CV_T are CV in the longitudinal and transverse directions, respectively], averaging 2.1 ± 0.8. Interventions decreasing Na current availability, such as S1S2 pacing and TTX, slowed CV_L and CV_T proportionately, without changing the AR. Conduction block preferentially occurred longitudinal to fiber direction, commonly initiating reentry. Interventions that decreased gap junction conductance, such as heptanol, decreased CV_T more than CV_L, increasing the AR and causing preferential transverse conduction block and reentry. Rapid pacing resembled the latter, increasing the AR and promoting transverse conduction block and reentry, which was prevented by the Ca_i chelator 1,2-bis oaminophenoxy ethane-N,N,N′,N′-tetraacetic acid (BAPTA). In contrast to isotropic and uniformly anisotropic monolayers, in which reentrant rotors drifted and self-terminated, bidirectional anisotropy (i.e., an abrupt change in fiber direction exceeding 45°) caused reentry to anchor near the zone of fiber direction change in 77% of monolayers. In anisotropic monolayers, unidirectional conduction block initiating reentry can occur longitudinal or transverse to fiber direction, depending on whether the experimental intervention reduces Na current availability or decreases gap junction conductance, agreeing with theoretical predictions.

cardiac tissue is inherently anisotropic, with faster conduction velocity (CV) along fiber direction than across it (24, 29, 31), a property implicated in arrhythmogenesis. Spach et al. (31) showed that anisotropy can lead to preferential longitudinal conduction block initiating reentry, due to a lower safety factor for longitudinal than transverse conduction. Other investigators, however, found preferential transverse conduction block in anisotropic tissue (7, 8, 22, 24, 26), particularly when gap junction conductance was pharmacologically decreased by heptanol (8). Although theoretical mechanisms for both preferential longitudinal block (due to reduced safety factor) (27, 31) and preferential transverse block [due to reduced gap junction conductance (13) and certain low excitability conditions (8)] have been proposed, the experimental evidence for these mechanisms in intact three-dimensional (3D) tissue is somewhat conflicting. For example, in Spach's classic studies, preferential longitudinal block leading to reentry occurred in aged myocardium with nonuniform anisotropy and discontinuous conduction due to fibrosis, but not in young myocardium with uniform anisotropy. On the other hand, Koura et al. (15) reported that preferential longitudinal block in young tissue converted to preferential transverse block with aging. Finally, whereas interventions that decreased Na current availability in anisotropic tissue caused preferential longitudinal conduction block in some studies (7, 31), other studies showed preferential transverse block under low excitability conditions (7, 16).

A unified mechanistic interpretation of these findings is made difficult by differences in pacing protocols, animal models, and methods. Some studies used premature extrastimulus testing (S1S2 pacing) to induce conduction block, whereas others used incremental rapid pacing (8, 30, 31). In addition, many of these studies have been performed in 3D tissue in which fiber direction rotates, so that the apparent direction of conduction block on the surface may be influenced by subsurface events. Although anisotropic conduction block and reentry has also been studied in cryoablated or infarcted ventricles with only a thin layer of surviving epicardium (10, 24, 26, 32), these preparations are not truly two-dimensional (2D) and effects of injured subsurface tissue cannot be completely excluded.

To provide a better understanding of how various factors predispose anisotropic tissue to preferential longitudinal or transverse conduction block, the goal of this study was to compare systematically the effects of pacing modality and various pharmacological interventions on conduction block and initiation of reentry in the same anisotropic preparation. To avoid contamination of mapping data by unrecorded subsurface events, we chose a genuinely 2D tissue model, cultured neonatal rat ventricular myocyte (NRVM) monolayers, in which the degree and pattern of anisotropy could be readily controlled by micropatterning techniques (1).

METHODS

Monolayer preparation.

NRVM monolayers were prepared as previously described (4, 5) and plated on 22 × 22 mm² plastic cover slips. Briefly, hearts from 2- to 3-day-old Sprague-Dawley rats were extracted, trimmed of atrial and fat tissue, and digested with collagenase (0.02%; Worthington) and pancreatin (0.06%; Sigma). Myocytes were isolated from fibroblasts and other cells using a Percoll (Pharmacia) gradient (4, 5). Cells were plated with a density of one million cells per cover slip. Cover slips were either not abraded (isotropic, n = 37), or microabraded with 15-μm lapping paper (anisotropic, n = 80) (1). Unidirectional anisotropy was created by abrading the entire cover slip in one direction (n = 67). To create bidirectional anisotropy, abrasions were performed in different directions on the two halves of the cover slip, with the transition in the center (n = 13).

Monolayers were exposed to heptanol (0.5–1 mM, n = 7) or tetrodotoxin (TTX, 5–10 μM, n = 9) by directly adding these drugs to the superfusate. In some experiments, monolayers were incubated with 100 μmol/l 1,2-bis oaminophenoxy ethane-N,N,N′,N′-tetraacetic acid, n = 9 (BAPTA-AM) and 0.02% pluronic (Molecular Probes, Eugene, OR) for 30–40 min.

Optical mapping system.

As previously described (4, 5), monolayers were stained by immersion in oxygenated Tyrode's solution at 37°C, containing the fluorescent voltage dye RH-237 (5 μM for 5 min) and/or the Ca dye rhod 2-AM (5 μM for 40 min) plus 0.016% (wt/wt) pluronic (Molecular Probes) and then placed in a perfusion bath at 37°C. Fluorescence was excited by two light sources (each with 4 light-emitting diodes; Luxeon, Ontario, Canada) filtered at 540 ± 20 nm. The emitted fluorescence was separated using a dichroic mirror (at 630 nm) directed to two separate cameras with their corresponding emission filters (715 nm for RH-237 and 585 nm for rhod 2/rhod FF). Single-dye staining with rhod 2 (n = 68) or RH-237 (n = 29) confirmed the absence of cross talk between voltage and Ca signals, and simultaneous voltage and Ca mapping was performed in 20 specimens. Data were recorded using electron-multiplying, back-illuminated, cooled charge-coupled device cameras (Photometrics Cascade 128+), with 171 × 171 μm spatial resolution at 0.6–5 ms/frame. Signals were digitized with 16 bits of precision.

Pacing protocols.

Monolayers were paced at 2 Hz with a Grass stimulator delivering unipolar point stimuli (10 V), using one of two computer-generated pacing protocols: 1) incremental pacing, initially at 340 ms cycle length, decremented by 20 ms every eight beats until reaching 140 ms cycle length or 2) S1S2 pacing, in which an eight-beat train with a basic cycle length of 500 ms (S1) was followed by a premature extrastimulus ranging from 220-ms to 140-ms by 20-ms decrements (S2).

Immunohistochemistry.

See supplemental data for this article, which may be found on the American Journal of Physiology: Heart and Circulatory Physiology website.

Data analysis.

Raw data were processed with custom-written software as previously described (3, 4). Action potential duration (APD) was measured at 80% repolarization. CV in the longitudinal and transverse directions (CV_L and CV_T, respectively) and anisotropy velocity ratio (AR = CV_L/CV_T) were estimated from the activation times using voltage or intracellular Ca (Ca_i) fluorescence with custom-written software. See supplemental methods.

Statistics.

Data are shown as means ± SD. Statistical significance was assessed using Chi square and Student's t-tests, with a P value <0.05 considered significant. The investigation conformed with the Guide for the Care and Use of Laboratory Animals published by the United States National Institutes of Health. The use and care of the animals in these experiments were approved by the Chancellor's Animal Research Committee at the University of California Los Angeles.

RESULTS

Conduction properties in isotropic vs. anisotropic monolayers.

In unpatterned monolayers paced at 2 Hz (Fig. 1A), isochronal maps obtained from simultaneously acquired voltage and Ca_i recordings demonstrated isotropic propagation patterns (Fig. 1A, n = 11). CV in orthogonal directions averaged 24 ± 10 cm/s (4); APD and Ca_i transient duration averaged 126 ± 12 and 174 ± 14 ms, respectively. Isotropic monolayers maintained 1:1 conduction up to a pacing frequency of 7 Hz, with CV progressively slowing as pacing frequency increased, consistent with CV restitution (5). CV measurements using voltage or Ca_i signals showed excellent agreement (Fig. 1B) and were therefore considered equivalent in subsequent analysis.

Fig. 1.

Isotropic neonatal rat ventricular myocyte (NRVM) monolayers. A: representative simultaneous voltage (V) and intracellular Ca (Ca_i) traces and corresponding isochronal maps during pacing at 2 Hz in an isotropic monolayer. D1 and D2 indicate orthogonal directions along which conduction velocity (CV) was measured. Arrows indicate longitudinal (D1) and transverse (D2) directions, with the length of the arrow proportionate to CV in that direction. CV_L, longitudinal CV; CV_T, transverse CV; F_Ca, Ca_i fluorescence; F_V, voltage fluorescence. B: rate dependence of CV obtained from Ca_i (red) and voltage signals (black) in isotropic monolayers.

In anisotropic monolayers paced at 2 Hz (n = 19) (Fig. 2), CV_L was significantly faster than CV_T, averaging 37 ± 10 and 18 ± 9 cm/s, respectively (P < 0.05). The AR averaged 2.1 ± 0.8. APD and Ca_i transient duration averaged 156 ± 9 and 280 ± 37 ms, respectively, and were prolonged compared with isotropic monolayers (P < 0.05), as reported previously (3).

Fig. 2.

Effects of reduced Na current availability and gap junction coupling on the anisotropy ratio (AR). A: isochronal maps of the S1 and S2 beats at the coupling intervals indicated in an anisotropic monolayer (with the fiber orientation and stimulus site indicated by the schematic diagram on the right). Longitudinal CV (CV_L, cm/s), transverse CV (CV_T, cm/s), and AR are indicated below. Note that, despite slower CV_L and CV_T during the S2 beat, the AR is unchanged. Arrows indicate longitudinal and transverse fiber directions, with the length of the arrow proportionate to CV in that direction. S1S2, pacing in which an 8-beat train with a basic cycle length of 500 ms (S1) was followed by a premature extrastimulus ranging from 220 to 140 ms by 20-ms decrements (S2). B: isochronal map after tetrodotoxin (TTX, 5 μM) during pacing at 2 Hz. CV_T and CV_L slowed proportionately, and AR did not change. C: isochronal map after heptanol (0.5 mM). CV_T decreased more than CV_L, causing the AR to increase. D: CV_T (open circles), CV_L (solid circles), and AR (blue) during S1S2 pacing (n = 5), with S1 at 500 ms (2 Hz) and S1S2 varied from 220 to 140 ms (4.5–7 Hz). Note that the AR remained stable. E: average values of CV_T (white bars), CV_L (black bars), and AR (blue bars) under control conditions and after TTX or heptanol during pacing at 2 Hz. Bars indicate means + 1 SD. Inset: fiber orientation and stimulus site.

Effects of decreasing Na current availability vs. gap junction conductance in anisotropic monolayers.

When Na current availability was decreased by introducing premature stimuli during pacing at a cycle length of 500 ms (S1S2 pacing), CV_L and CV_T decreased proportionately as the S2 beat became more premature in five anisotropic monolayers (Fig. 2A). The AR did not change significantly, remaining between 2.0 and 2.1 up to the point at which the S2 stimulus failed to capture near 140 ms (Fig. 2D). Before complete loss of S2 capture, localized longitudinal conduction block close to the pacing site developed in 80% of the specimens during S1S2 pacing, as shown in Fig. 3A, frequently initiating reentry (defined as one or more full rotations). Transverse conduction block appearing before longitudinal conduction block was never observed.

Fig. 3.

Anisotropic conduction block and reentry. A: isochronal maps before and after TTX administration (10 μM) during S1S2 pacing (with the fiber orientation and stimulus site indicated by the schematic diagram on the right). Before TTX, the premature S2 beat (middle) developed longitudinal conduction block near the pacing site (black line) but did not initiate reentry. After TTX (right) longitudinal conduction block after a premature extrastimulus at 220 ms initiated reentry. Nos. at bottom indicate CV_L (in mm/s), CV_T (in mm/s), and AR. B: isochronal maps before and after heptanol (1 mM) during pacing at 2Hz. Heptanol slowed CV and increased the AR, resulting in transverse conduction block (middle), which initiated reentry (right). Inset, fiber orientation and stimulus site.

Na current availability was also reduced in nine anisotropic monolayers by exposure to the Na channel blocker TTX (5–10 μM). During pacing at 2 Hz, CV_L and CV_T decreased proportionately, and the AR did not change significantly (2.05 ± 0.1, P > 0.05) (Fig. 2, B and E). After TTX administration (n = 9), the incidence of longitudinal conduction block inducing either figure-eight or single-loop reentry during S1S2 pacing increased to 100% of specimens (P < 0.05) (Fig. 3A and Supplemental Online Movie 2).

To reduce gap junction conductance, seven anisotropic monolayers were exposed to 0.5–1 mM heptanol (Fig. 2, C and E). During pacing at 2 Hz, heptanol significantly slowed CV_L by 23 ± 1% (P < 0.05) and CV_T by 69 ± 1% (P < 0.05), leading to a significant increase in the AR from 2.1 ± 0.1 to 2.9 ± 0.3 (P < 0.001). In six of the seven monolayers, preferential slowing of CV_T led to transverse conduction block before longitudinal block, initiating reentry (Fig. 3B). Longitudinal conduction block preceding transverse block as a mechanism initiating reentry after heptanol was not observed. Similar results were obtained in 10 preparations when the pacing site was moved from the edge of the cover slip to the center (see Supplemental Fig. 1), indicating that tissue border effects were not critical.

Effects of incremental pacing in anisotropic monolayers.

During rapid incremental pacing, Na current availability becomes compromised as diastole becomes too short for Na channels to recover fully from inactivation. In addition, gap junction conductance may decrease because of rate-related accumulation of Ca_i (17, 20), protons, and/or other factors. In anisotropic monolayers in which the pacing cycle length was decreased by 20 ms every eight beats, both CV_L and CV_T decreased proportionately with minor changes in the AR up to a pacing frequency of 4 Hz, consistent with reduced Na current availability. At pacing frequencies above 4 Hz, however, CV_L decreased less than CV_T, such that the AR progressively increased, reaching 2.5–2.9 at 7–10 Hz pacing (P < 0.05) (Fig. 4). Reentry was initiated by rapid pacing in 16 of 19 specimens (84%). In 13 of these 16 specimens (81%), reentry was initiated by preferential transverse conduction block, which developed at an average pacing frequency of 5.6 ± 0.9 Hz (Fig. 4A and Supplemental Online Movie 3). In the other three specimens, preferential longitudinal direction conduction block initiated reentry. Figure 4, B and C, summarizes the rate-dependent changes in CV_L, CV_T, and AR, analyzed using either voltage (n = 8) or Ca_i fluorescence (n = 9). Unlike isotropic cardiac monolayers (4), spatially discordant alternans was only infrequently observed before conduction block in anisotropic monolayers such that the site of initial conduction block occurred near a nodal line in only 16% of cases. These findings suggest that, in anisotropic monolayers, the pacing rate threshold for anisotropic conduction block is typically reached before the onset of spatially discordant alternans.

Fig. 4.

Effects of rapid pacing. A: isochronal maps during incremental rapid pacing in an anisotropic monolayer (with the fiber orientation and stimulus site indicated by the schematic diagram on the right). Shown are beats at 2 and 7 Hz without conduction block, 7 Hz with transverse conduction block, and subsequent reentry. CV_L (cm/s), CV_T (cm/s), and AR are indicated below, showing the increase in AR at 7 Hz. B and C: rate dependence of CV_T (open circles), CV_L (solid circles), and AR (blue line) obtained from F_V (n = 8) or F_Ca (n = 9) recordings, respectively. Note that CV_L and CV_T diverge more rapidly after 4Hz, leading to an increase of the AR. Values are means + 1 SD.

To investigate whether the increase in AR during incremental rapid pacing at >5 Hz might be explained by reduced gap junction conductance from rate-related Ca_i accumulation, we preloaded anisotropic monolayers with BAPTA-AM to buffer Ca_i, as confirmed by a marked suppression of the amplitude of raw Ca_i transients (Fig. 5, C and D). ΔF/F averaged 6 ± 2% in 14 control preparations vs. 0.5 ± 0.2% in 9 BAPTA-AM-loaded preparations (P < 0.05). After BAPTA-AM, APD prolonged from 148 ± 6 to 230 ± 54 ms (P < 0.05) during pacing at 2 Hz. Most of specimens treated with BAPTA could not be paced faster than 5 Hz before 2:1 conduction block occurred, at which point there was no significant change in AR. In nine specimens, however, APD adaptation allowed pacing up to 7 Hz. In these specimens, CV_L and CV_T decreased proportionately as rate increased, and the AR did not change significantly (Fig. 5B). Preferential transverse conduction block was not observed, and reentry was induced in only two of nine (22%) specimens, always because of preferential longitudinal conduction block.

Fig. 5.

Effects of 1,2-bis oaminophenoxy ethane-N,N,N′,N′-tetraacetic acid (BAPTA-AM) during incremental rapid pacing. A: isochronal maps during incremental rapid pacing at 2 Hz (left) and 7 Hz (right) in an anisotropic monolayer (with the fiber orientation and stimulus site indicated by the schematic diagram below). CV_L (cm/s), CV_T (cm/s), and AR are indicated below. B: rate dependence of CV_T (open circles), CV_L (solid circles), and AR (blue line) showing that BAPTA-AM prevented the increase in AR as rate increased. Values are means + 1 SD. C: confirmation of effective Ca_i buffering by BAPTA-AM. Red traces show raw Ca_i fluorescence (shown as ΔF/F_Ca × 100) at the onset of 2:1 pacing block (2:1 bl) during the rapid pacing protocol, under control (left) and BAPTA-AM-loaded conditions (right). BAPTA markedly attenuated the Ca_i transient amplitude as well as the decrease in diastolic Ca_i during the long diastolic interval after 2:1 block. Black trace at top right shows the simultaneous voltage fluorescence trace. D: bar graph summarizing the amplitude of the systolic Ca_i transient and the decrease in diastolic Ca_i during the onset of 2:1 block under the two conditions. Values are means + 1 SD.

To document that diastolic Ca_i increased significantly during rapid pacing, and was blunted by BAPTA-AM, we compared the diastolic Ca_i fluorescence during maximum pacing and immediately after the onset of 2:1 block (Fig. 5, C and D). Under control conditions (n = 14), diastolic Ca_i fluorescence decreased significantly during the long diastolic interval after 2:1 block (ΔF/F −1.8 ± 0.6%), whereas, in BAPTA-AM-loaded preparations (n = 9), the decrease was minimal (−0.1 ± 0.1%, P < 0.05). In summary, the BAPTA-AM results suggest that Ca_i accumulation plays a role in preferentially slowing CV_T to increase the AR and promote transverse conduction block at high pacing frequencies, presumably by decreasing gap junction conductance (30), although other factors, such as acidosis, may also contribute.

Stability of reentry.

In isotropic monolayers, rapid pacing induced reentry in 16% of preparations (5). Once initiated, reentry was always spatiotemporally unstable, with the core of the rotor meandering through the tissue until the rotor self-terminated at a tissue boundary. In anisotropic monolayers, the incidence of rapid pacing-induced reentry increased to 84% of specimens (P < 0.05). Reentry was also spatiotemporally unstable (Fig. 6, C and D), leading to self-termination. Although reentrant rotors drifted slowly, they often remained in the nearly same position for several beats, allowing the core size to be estimated. During reentry in anisotropic monolayers, the core of the rotor was elliptical, elongated along the direction of underlying fibers (long axis 0.78 ± 0.1 mm, short axis 0.25 ± 0.05 mm, P < 0.001). The core area was significantly smaller compared with isotropic monolayers (0.63 ± 0.16 vs. 0.87 ± 0.12 mm², P < 0.05). As shown in the lower tracing in Fig. 6, Ca_i at the core during reentry was intermediate between normal diastolic and systolic levels.

Fig. 6.

Initiation of meandering reentry in anisotropic monolayer. Sequential isochrone maps of a paced beat without conduction block (A), a paced beat with transverse conduction block initiating reentry (C), and a subsequent beat during spiral wave reentry in which the spiral core has meandered to the upper region (D). No significant Ca transient alternans was present (B), and the AR (CV_L/CV_T) for the last paced beat before conduction block was 2.5. Traces at the bottom show Ca fluorescence at sites labeled 1 and 2 in the isochronal maps. When the spiral wave was initiated at site 1, the Ca_i fluorescence level in the spiral wave core was intermediate between the systolic and diastolic levels observed later after the core had meandered to a different region. At site 2, similar behavior occurred late in the trace when core meandered to that site. Schematic diagram on the right shows the fiber orientation and stimulus site.

Because rotors have been reported to anchor to the base of papillary muscles at which fiber direction changes abruptly (14), we fabricated bidirectionally anisotropic monolayers (Supplemental Fig. 3S) to investigate how abrupt changes in fiber direction (>45°) affected the stability of reentry. In 13 bidirectionally anisotropic monolayers, rapid pacing led to progressive slowing of conduction and preferentially transverse conduction block at a distance from the pacing site (Supplemental Fig. 2S), initiating reentry which became anchored in the transition zone of fiber direction change (Fig. 7). Reentry initiation by preferential transverse conduction block was thus similar to that in unidirectionally anisotropic tissue. In bidirectionally anisotropic tissue, the main difference was that, instead of self-terminating, reentry became stable in 77% of specimens, with the core of the rotor anchored within 2 mm of the transition zone between different fiber orientations (Fig. 7 and Supplemental Online Movie 5). Histologically, cells in the transition zone region were less aligned than in the adjoining uniformly anisotropic regions (Supplemental Fig. 3S), but no macroscopic defects that might serve as anchors for rotors were obvious.

Fig. 7.

Anchored reentry in a bidirectionally anisotropic monolayer. A: schematic showing fiber direction (lines) and sites of representative tracings shown in C. B: Ca_i fluorescence traces recorded at sites 1–4 during reentry. The core of the spiral wave was located at site 4 near the fiber direction transition zone. C: isochronal map during reentry. D: trajectory of the spiral wave tip during 35 cycles of reentry, showing that the core remains stationary at site 4.

DISCUSSION

Anisotropy is known to predispose cardiac tissue to unidirectional conduction block initiating reentry (24, 28, 31). Many experimental studies in anisotropic tissue preparations have documented preferential conduction block occurring either longitudinal or transverse to fiber direction (7, 8, 22, 24, 26, 28, 30, 31). However, factors related to 3D tissue geometry, fiber rotation, nonuniform anisotropy due to tissue fibrosis, discontinuous conduction, variable pacing protocols, and possible influences of unmapped subsurface events have complicated the interpretation of the underlying mechanism in some of these experiments. For this reason, we used the same anisotropic 2D tissue preparation to systematically evaluate how different pacing and pharmacological interventions promote unidirectional conduction block and reentry. In contrast to the often variable findings in the aforementioned experimental studies, we find good general agreement with theoretical predictions (8, 13, 23, 25, 31). When reduced Na channel availability is the primary factor (as in extrastimulus testing or TTX), block is preferentially longitudinal. When reduced gap junction conduction is the primary factor (as in rapid pacing or heptanol), block is preferentially transverse. We find that both forms of anisotropic conduction block are equally likely to initiate reentry. In addition, we make the novel observation that, unlike isotropic monolayers (5), conduction block and initiation of reentry do not depend on amplification of electrophysiological dispersion by spatially discordant repolarization alternans. That is, the onset of anisotropic conduction block almost always occurred before the onset of spatially discordant APD alternans. We make the novel observation that, when anisotropy is unidirectional, reentry is typically nonsustained, with the rotor core meandering to the edge of the tissue and self-terminating. However, when anisotropy is bidirectional, rotors typically anchor to within 2 mm of the fiber direction transition zone, creating sustained reentry. Although the precise mechanism is unclear, this finding may explain why rotors are prone to anchoring at the base of papillary muscles in intact heart muscle, where abrupt fiber direction changes are common (14). On the other hand, others have described stable anchored reentrant rotors in monolayers without bidirectional anisotropy (2). The reasons for this discrepancy are unclear.

Implications in 3D ventricular tissue.

Fiber direction rotates in 3D tissue, so that the apparent direction of conduction block mapped exclusively on the surface may not reflect the influence of unmapped subsurface events. This may also explain some of the discrepancies in anisotropic conduction block (longitudinal vs. transverse) in 3D tissue. Schalij et al. (26) found that transverse conduction near the site of stimulation was interrupted by epicardial breakthrough of faster intramural wavefronts in rabbit ventricles. After endocardial cryoablation, however, the AR increased, and rapid pacing induced preferential transverse conduction block initiating reentry. Thus, in normal ventricle, the ∼180° fiber rotation across the ventricular wall may effectively cause a “short circuit,” protecting the heart from anisotropic conduction block and reentry. In this setting, conduction block may need to await the onset of spatially discordant repolarization alternans at even faster heart rates to initiate reentry (19, 21).

In contrast, anisotropic conduction block may come into play when there is no significant “short-circuiting” effect of rotational anisotropy, as in postischemic surviving epicardial ventricular layers or in regions of the atria. These observations may thus explain why anisotropic conduction block and reentry are much more common in ischemic heart disease and the atria.

Differences between S1S2 and rapid incremental pacing.

Our findings also indicate that the pacing protocol plays a critical factor in the mechanism of anisotropic conduction block and reentry. Premature extrastimuli (S1S2 pacing) induced preferential longitudinal block by reducing Na current availability without affecting the AR, whereas incremental rapid pacing induced preferential transverse block by increasing the AR. Spach et al. (30) previously reported similar effects of premature extrastimuli vs. rapid pacing on the AR in intact canine ventricular tissue. The increase in the AR induced by rapid pacing was also exacerbated by ouabain, leading them to speculate that gap junction conductance was reduced by Ca_i accumulation during rapid pacing (9, 11, 19). Rapid pacing has been shown to rapidly increase diastolic Ca_i levels (17), which some studies show can reduce intercellular coupling within seconds (20), although others have reported a delay of one to several minutes (6, 18). In computer simulations, Shaw and Rudy (27) demonstrated that, when intercellular coupling is reduced, elevated Ca_i may also promote conduction block by facilitating Ca-induced inactivation of the L-type Ca current. Our findings are consistent with both mechanisms, since preloading anisotropic monolayers with BAPTA-AM to suppress rate-related Ca_i accumulation prevented the increase in AR and markedly reduced the incidence of conduction block and reentry.

Limitations.

A limitation of our study is that we could not measure absolute levels of Ca_i during the rapid pacing protocol because of photobleaching and nonratiometric properties of rhod 2 (12). However, from the raw fluorescence traces, we were able to confirm that BAPTA-AM markedly suppressed both the amplitude of the Ca_i transient and the decrease in diastolic Ca_i with sudden rate slowing (Fig. 5, C and D). On the other hand, we cannot exclude contributions of alternative mechanisms causing the increase in AR leading to transverse conduction block during rapid pacing. Rapid pacing may lower intracellular pH, which can decrease gap junction conductance, and BAPTA-AM may blunt intracellular acidosis by decreasing metabolic requirements during rapid pacing. In addition, BAPTA-AM also exhibits effects on the APD. Other potentially important factors include alterations in other ionic concentrations, APD differences for longitudinal vs. transverse conduction, wavefront curvature effects, or positioning of unipolar extracellular stimulating electrodes.

The pacing and pharmacological interventions used in this study also have some limitations. Although S1S2 pacing and TTX reduce Na current availability without affecting gap junction conductance, they also affect APD and possibly other electrophysiological properties that could influence anisotropic conduction block. Heptanol and rapid pacing, on the other hand, suppress Na current availability as well as gap junction conductance. However, since neither S1S2 pacing nor TTX altered the AR, decreased Na current availability by heptanol or rapid pacing cannot explain the observed increase in the AR or preferential transverse conduction block.

The role of fibrosis in promoting anisotropic reentry also deserves comment. Spach et al. (29) described preferential longitudinal conduction block after a premature stimulus in older fibrotic heart tissue with nonuniform anisotropy, whereas conduction failure occurred simultaneously in both longitudinal and transverse directions in young anisotropic tissue. Our model resembled the former case, possibly because fibroblasts contaminate NRVM monolayers to some extent . In this study, we required isochrone maps to exhibit grossly uniform propagation to exclude macroscopic conduction barriers. However, isochrones did not form mathematically pure ellipses, likely because of the effects of microscopic heterogeneities caused by fibroblast proliferation. Likewise, unidirectional conduction block initiating reentry was often not symmetrical, inducing single rotors rather than figure eight rotors, as expected if the anisotropic monolayers were microscopically homogeneous. In bidirectionally anisotropic monolayers, we speculate that tissue microheterogeneities may be involved in anchoring reentrant rotors at transition zones of sudden fiber direction change, for which we otherwise have no clear explanation. Our finding that rotors anchored to the interfaces may be relevant to observations in intact cardiac muscle that reentrant rotors anchored to the base of papillary muscles, where fiber direction changes abruptly (14).

Finally, neonatal myocytes are not electrophysiologically identical to adult myocytes, such that extrapolation of our findings to adult intact cardiac tissue must be made with caution. Nevertheless, it is encouraging that our findings closely replicate theoretical predictions based on generic cardiac electrophysiological properties.

GRANTS

This work was supported by National Heart, Lung, and Blood Institute Grants P50 HL-52319, P01 HL-078931, and R01 HL-103662 to J. N. Weiss, American Heart Association Grants 0625048Y to C. de Diego, 0365133Y and 0565149Y to M. Valderrábano, and 0755043Y to F. Chen, and by a Spanish Society of Cardiology Grant to C. de Diego.

DISCLOSURES

We have no conflicts of interests to disclose.