Spontaneous, Pro-Arrhythmic Calcium Signals Disrupt Electrical Pacing in Mouse Pulmonary Vein Sleeve Cells

The pulmonary vein, which returns oxygenated blood to the left atrium, is ensheathed by a population of unique, myocyte-like cells called pulmonary vein sleeve cells (PVCs). These cells autonomously generate action potentials that propagate into the left atrial chamber and cause arrhythmias resulting in atrial fibrillation; the most common, often sustained, form of cardiac arrhythmia. In mice, PVCs extend along the pulmonary vein into the lungs, and are accessible in a lung slice preparation. We exploited this model to study how aberrant Ca2+ signaling alters the ability of PVC networks to follow electrical pacing. Cellular responses were investigated using real-time 2-photon imaging of lung slices loaded with a Ca2+-sensitive fluorescent indicator (Ca2+ measurements) and phase contrast microscopy (contraction measurements). PVCs displayed global Ca2+ signals and coordinated contraction in response to electrical field stimulation (EFS). The effects of EFS relied on both Ca2+ influx and Ca2+ release, and could be inhibited by nifedipine, ryanodine or caffeine. Moreover, PVCs had a high propensity to show spontaneous Ca2+ signals that arose via stochastic activation of ryanodine receptors (RyRs). The ability of electrical pacing to entrain Ca2+ signals and contractile responses was dramatically influenced by inherent spontaneous Ca2+ activity. In PVCs with relatively low spontaneous Ca2+ activity (<1 Hz), entrainment with electrical pacing was good. However, in PVCs with higher frequencies of spontaneous Ca2+ activity (>1.5 Hz), electrical pacing was less effective; PVCs became unpaced, only partially-paced or displayed alternans. Because spontaneous Ca2+ activity varied between cells, neighboring PVCs often had different responses to electrical pacing. Our data indicate that the ability of PVCs to respond to electrical stimulation depends on their intrinsic Ca2+ cycling properties. Heterogeneous spontaneous Ca2+ activity arising from stochastic RyR opening can disengage them from sinus rhythm and lead to autonomous, pro-arrhythmic activity.


Introduction
Throughout a typical human lifetime, a coordinated 'cardiac cycle' of atrial and ventricular contraction is repeated over a billion times [1]. The cardiac cycle is initiated by the sino-atrial node (SA node) located within the right atrial wall. The SA node generates action potentials (APs) that sweep over the atrial and ventricular chambers to cause them to contract and pump blood [2]. This contraction is mediated by Ca 2+ increases within cardiac myocytes via the well-known process of 'excitation-contraction coupling' (EC-coupling). Essentially, membrane depolarisation opens voltage-operated Ca 2+ channels (VOCCs) to allow Ca 2+ influx into the 'dyadic' cleft between the sarcolemma and the sarcoplasmic reticulum (SR; the myocyte Ca 2+ store). This Ca 2+ signal is amplified by Ca 2+ -induced Ca 2+ release via ryanodine receptors (RyRs) on the SR, thus causing a global Ca 2+ signal that triggers actin-myosin filament interaction and myocyte contraction [3]. After each AP, myocytes become refractory to further electrical stimulation for tens to hundreds of milliseconds, and during this time Ca 2+ is returned to diastolic levels by Ca 2+ -ATPases on the sarcolemma and SR, and Na + /Ca 2+ exchangers on the sarcolemma [4]. In humans, the normal sinus rhythm is ,60 beats per minute, but during atrial fibrillation (AF) the atrial chambers can display activity in excess of 300 beats per minute due to aberrant electrical signals [5]. Because the ventricular chambers are largely responsible for pumping blood, AF is not immediately life threatening, but a loss of atrial function can lead to fainting and chest pain. Moreover, blood clotting resulting from stagnant or turbulent blood in the atrial chambers greatly increases the risk of stroke [6].
It is believed that increased 'automaticity' (spontaneous depolarisation of myocytes), acute 'triggered activity' (spontaneous electrical events following recovery from an action potential) or 're-entry circuits' (return of an electrical impulse to cardiac cells following a refractory period) contribute to AF. The underlying causes of all these pro-arrhythmic conditions are not fully understood, but substantial evidence has implicated spurious Ca 2+ signals as a likely cause [6][7][8]. Spontaneous Ca 2+ signals occurring during the recovery from a previous AP, or during the quiescent diastolic phase, can depolarise the sarcolemma and potentially trigger an ectopic AP or alter the refractoriness of myocytes relative to their neighbours [9].
While spontaneous Ca 2+ signaling and electrical events can arise within the atrial chambers themselves [10], a clinically-recognised source of pro-arrhythmic signals are pulmonary vein sleeve cells (PVCs) that form sheaths surrounding the large pulmonary veins [11]. PVCs are present in all mammalian cardiovascular systems and utilise the same EC-coupling machinery as atrial and ventricular myocytes (described above). Even though PVCs are developmentally and anatomically distinct from atrial myocytes, both cell types are in electrical continuity [12]. Thus, PVCs are entrained by sinus rhythm because APs arising in the SA node would sweep across the atrial chambers, propagate out of the heart into the PVCs and cause them to contract. However, electrical mapping has demonstrated that ectopic activity can arise within PVCs [13], and propagate into the left atrial chamber. Moreover, ablation procedures that electrically isolate pulmonary veins from the posterior wall of the left atrium (the border of the two tissues) are highly successful in treating acute and sustained AF [14,15], supporting the notion of PVC-initiated arrhythmias. The conditions leading to the generation of arrhythmic pacemaking sites within PVC sheaths is not understood, but is likely to involve the development of spontaneous Ca 2+ signals [16].
The extent of the PVC sheath varies between animal species. Here, we took advantage of the situation in mice, where the PVCs extend from the left atrium to the pulmonary veins within the lung for several branching generations [17]. In previous work, we demonstrated the utility of lung slices for studying airway smooth muscle physiology [18]; lung slices provide an intact, multi-cellular preparation that retains in situ organizational and physiological characteristics of the lung, and are viable for many days. Because the PVCs are disconnected from the left atrial chamber, they no longer receive APs arising from the SA node. This electrical isolation allows the spontaneous Ca 2+ signaling capacity of the PVCs to be evident without a background of SA node-evoked events. As required, the PVCs within a lung slice can be activated by application of electric field stimulation (EFS). In the present study, we characterised spontaneous and EFS-induced Ca 2+ transients in PVCs to determine whether these two processes use similar signaling mechanisms. Moreover, we explored the hypothesis that the inherent spontaneous Ca 2+ signals within PVCs could corrupt their electrical entrainment, and thereby lead to pro-arrhythmic outcomes.

Preparation of Lung Slices
For the preparation of lung slices [18] 8-12 week-old female BALB/C mice were killed by intraperitoneal Nembutal (pentobarbital sodium) injection, as approved by the Institutional Animal Care and Use Committee of the University of Massachusetts Medical School. After opening the chest cavity, the trachea was cannulated and lungs were inflated by injecting ,1.0 ml of low melting point agarose (37uC, 1.8% in sHBSS; Life Technologies, Carlsbad, CA). Agarose was flushed out of the airways and into the alveoli by injection of 0.3 ml of air and the agarose was gelled by applying cold sHBSS to the lungs. The stiffened lungs were cut into ,180 mm thick lung slices using a VF-300 microtome (Precisionary Instruments Inc., Greenville, NC). The presence of PVCs was confirmed by visual inspection at low magnification using the following criteria: the vein is not located next to an airway, spontaneous fibrillations and/or striations are visible in the cells surrounding the vein. Slices were kept in Dulbecco's Modified Eagle Medium supplemented with antibiotics and antimycotics (Life Technologies) and 10% fetal bovine serum and kept in a humidified tissue culture incubator at 37uC with 10% CO 2 . Slices were used within 48 h of preparation, apart from experiments investigating the viability of PVCs. For imaging experiments, lung slices were mounted between 2 cover-glasses separated with silicon grease on a custom-built Perspex support, which allowed for a continuous superfusion of the lung slices with experimental solutions.

Imaging Experiments
Imaging experiments were performed on custom-built microscopes [19]. Video Savant software (IO Industries, CA) was used for recording images and controlling the timing of solution changes. Contractions were measured using a Nikon Diaphot 300 microscope with a 10x objective (N.A. 0.3) at 15, 30 or 120 images per second. Ca 2+ imaging was performed on a custom-built 2photon system (excitation wavelength of 800 nm), based on an Olympus IX71 microscope, using an oil immersion UApo 340 40x (N.A. 1.3) or PlanApo60x (N.A. 1.42) objective. Lung slices were loaded in the dark with 20 mM Oregon Green BAPTA-1 AM (Acetoxymethyl ester; Life Technologies) in presence of 0.1% Pluronic F127 (Life Technologies) and 200 mM sulfobromophthalein in sHBSS at 30uC for 1 hour, followed by a de-esterification period of at least 30 min at 30uC in the presence of sulfobromophthalein. Image acquisition was performed using Video Savant software at 30 images per second. Throughout the experiment, lung slices were perfused with sHBSS using a gravity-fed system. The solution exchange was controlled by Video Savant software. The 2-photon excitation wavelength was set by tuning the laser. The excitation light was reflected using a FF665-Di02-25636 dichroic (Semrock, Rochester, NY), and the emitted fluorescence light was filtered using a short-pass barrier filter FF01-680/SP-25 (Semrock).
To avoid deterioration of the slices, or temperature gradients associated with superfusing cells, the experimental data shown in this study were obtained at room temperature (20uC). However, electrical pacing, contraction and spontaneous Ca 2+ transients were evident at 20uC and 37uC (c.f. fig. S1 and Results section). The average frequencies of spontaneous Ca 2+ transient activity were not significantly different at either temperature (2.360.2 and 2.560.2 Hz, 20uC and 37uC, respectively; see fig. S1C). Moreover, the effects of pharmacological reagents and experimental maneuvers (caffeine, ryanodine, KCl) were similar at both temperatures. A consequence of studying the slices at room temperature was that the electrical pacing rate was reduced from ,8 Hz in vivo to #5 Hz. We speculate that the lack of effect of temperature on the frequency of spontaneous Ca 2+ signals relates to the counterbalancing processes of Ca 2+ release and reuptake being equally affected.

Electric Field Stimulation (EFS)
EFS (40 V, 10 ms duration) was applied using a Grass Instruments SD9 stimulator and platinum electrodes (0.0059 wire thickness, A-M systems), placed at the opposite ends of the imaging chamber. The EFS conditions were empirically determined by finding the minimal voltage condition that would reliably pace PVCs with no appreciable run-down of the responses.

Statistics
Statistical analysis was performed in GraphPad Prism 6 (GraphPad software Inc., La Jolla, CA, USA), using a Mann-Whitney U-Test (2 datasets) or a Kruskall-Wallis Test followed by a Dunn's post-hoc analysis (.2 datasets) to test for statistically significant differences by comparing the initial values the first 30 s in sHBSS at the start of the experiment to the values at the indicated time periods. P values ,0.05 were considered significant, *P,0.05, **P,0.01, ***P,0.001. Data are shown as mean 6 s.e.m.

PVCs Respond to Electrical Pacing and Show Spontaneous Ca 2+ Signals in the Absence of Stimulation
To explore Ca 2+ signaling mechanisms within PVCs, lung slices were loaded with the Ca 2+ -sensitive indicator Oregon Green BAPTA-1 and examined with real-time 2-photon microscopy. PVCs displayed rapid, transient Ca 2+ increases in response to electric field stimulation (EFS; Fig. 1). Most PVCs followed such pacing for long periods with consistent responses (n .50 slices; maximum observation time 17 minutes). Disruption of electrical stimulation by stopping EFS (Fig. 1A), application of the VOCC inhibitor nifedipine (100 mM; Fig. 1B) or removal of extracellular Ca 2+ (Fig. 1C) inhibited the pacing-induced Ca 2+ responses and provoked the occurrence of spontaneous Ca 2+ transients. These data illustrate two key aspects of PVC Ca 2+ signaling: the regular Ca 2+ signals underpinning physiological EC-coupling, and proarrhythmic, spontaneous Ca 2+ transients.
We were mindful that damage to PVCs during preparation of lung slices could induce the spontaneous Ca 2+ signalling that we observed. However, the slices remained viable, with no obvious visual indication of slice deterioration, in culture for 3 days. The average frequencies of spontaneous Ca 2+ transients were not statistically different up to 2 days after preparing the lung slices (Fig. 1D). In addition, PVCs could be electrically paced by EFS for 2 days after slice preparation (later days were not examined). PVCs consistently loaded with the Ca 2+ -sensitive indicator to the same level for 3 days post preparation (Fig. 1E). Moreover, the Ca 2+ indicator was stably retained within PVCs up to 3 days post preparation. Active PVCs did not label with propidium iodide (PI; fig. S2). These data indicate that the PVCs were vital and not damaged during slice preparation, and that they did not rapidly de-differentiate in culture. To ensure consistency, all data in the present study were obtained using slices within 48 hours after preparation. Atrial myocytes within slices of the right atrial chamber, prepared using the same buffer solutions and conditions as for PVCs, could also be paced using EFS, but did not show prominent spontaneous Ca 2+ transients ( fig. S3). These observations indicate that PVCs and atrial myocytes have a different propensity for spontaneous Ca 2+ signaling.

PVC Organization, Morphology and Contraction
Pulmonary veins with a PVC myocardial sheath are found in mouse lung slices cut from the hilus region, where the pulmonary veins exit the lungs, to a distance approximately half-way through the lung lobe towards the periphery of the lungs ( Fig. 2A). The myocardial sheath is absent in more peripheral lung slices. The pulmonary veins (also known as intrapulmonary veins within the lung) are easily identified by their lone appearance within the lung alveoli tissue, as compared to the consistently paired association of a pulmonary artery with an airway ( Fig. 2A). The PVCs surrounding each vein could be identified using bright-field or phase-contrast optics as relatively large cells, which often displayed regular cellular striations that are typical of myocytes (Figs. 2B i -D i ). Immunostaining for the cardiac isoforms of RyR (types 2 and 3) revealed a striated distribution of these Ca 2+ channels (Figs. 2B ii and C ii , respectively) within the PVCs. These striations were 1.860.1 mm apart (n = 5 regions per slice, 6 slices), similar to the RyR (z-line) distribution in cardiac myocytes [22]. The membrane dye di-8-ANNEPS predominantly stained only the periphery of PVCs in .98% of cells, indicating that PVCs do not express an extensive transverse-tubule system (Figs. 2D i and D ii ). Collectively, these data suggest that PVCs structurally resemble rodent atrial myocytes [22] [23].

Contraction of Pulmonary Veins
In addition to their location and morphology, PVCs surrounding pulmonary veins could be readily identified by their prominent spontaneous contractions (Fig. 2E). The nature of this contractile activity was heterogeneous. In some veins, repetitive ripples of contraction propagated asynchronously along individual PVCs (frequency range 0.6-1.3 Hz). In other veins, multiple PVCs displayed more rapid, uncoordinated contractile activity. The numbers of PVCs involved ranged from a limited sub-population of cells, to all the PVCs surrounding the vein.
In many lung slices, asynchronous spontaneous PVC contractile activity could be coordinated into synchronous contractions by EFS, with a consequent reduction in the lumen size of the pulmonary vein during each contraction (Figs. 2E i-iii ; Video S1). During spontaneous contractions, the relative reduction in pulmonary vein lumen size was 0.260.1%, which increased significantly to 6.360.7% (P,0.01) during the EFS-induced contraction (n = 6 slices). Synchronised contraction only occurred during EFS; spontaneous contractile activity resumed with the termination of EFS.

Spontaneous Ca 2+ Signaling within PVCs and Entrainment by EFS
The overwhelming majority of PVCs displayed spontaneous Ca 2+ activity in the absence of EFS (n .250 slices). The spontaneous Ca 2+ transients were heterogeneous, and ranged from localized Ca 2+ increases (Ca 2+ sparks) to Ca 2+ waves that  across the cells is represented vertically while time progresses horizontally. Propagating Ca 2+ waves are evident as diagonal lines. Spontaneous Ca 2+ waves can be seen to originate in the centre of cell 1 (asterisk), and propagate bi-directionally to the poles of the cell (arrowheads). The Ca 2+ waves stopped at the boundary between cells 1 and 2 (dashed line). Cell 2 had its own intrinsic rhythm, and displayed Ca 2+ waves that sometimes temporally coincided with those of cell 1.
EFS-evoked Ca 2+ transients are evident in fig. 3A as vertical lines that coincide with EFS pulses. EFS restored the effect of sinus rhythm by evoking Ca 2+ signals (and triggering contraction) within the PVCs. However, the efficacy of each EFS pulse varied on a pulse-to-pulse basis, and ranged from inducing a whole-cell Ca 2+ signal to having no effect. For example, in fig. 3A, only on 3 occasions did EFS pulses cause a whole-cell Ca 2+ signal simultaneously in both cells (at pulses 9, 11, 14; both Cell 1 and 2 are marked '+'). The other EFS pulses correlated with one of three types of responses; i) only one cell showed a whole-cell response (pulse 1, Cell 1 is marked '+'), ii) one or both cells showed an incomplete response (pulses 3-7, 12, 13 and 15, marked '6') or iii) neither cell responded (pulses 2, 8 and 10). A plausible explanation for variable effects of EFS is that the PVCs enter a refractory period after a Ca 2+ rise during which they cannot respond to another Ca 2+ -releasing stimulus. Therefore, if a cell responds to EFS with a whole-cell response, the subsequent spontaneous Ca 2+ signal is inhibited or delayed. Conversely, if a spontaneous Ca 2+ wave occurs just prior to an EFS pulse, the cell cannot respond with a Ca 2+ transient. In essence, Ca 2+ signals evoked by EFS delayed the timing of spontaneous Ca 2+ activity and vice versa.
We found that the average latency between an EFS-evoked Ca 2+ signal and a subsequent spontaneous Ca 2+ event was correlated with the frequency of the intrinsic spontaneous Ca 2+ activity. The latency was longer in cells with a sparse spontaneous Ca 2+ activity (,1 Hz) as compared to cells with more frequent spontaneous Ca 2+ activity (.1 Hz, fig. 3C, left). An explanation for these observations is that cells with higher frequencies of spontaneous Ca 2+ signals have a greater capacity for restoring SR Ca 2+ content. This would enable PVCs to recover more quickly from an EFS-induced Ca 2+ pulse, as well as predispose the PVCs to show more frequent spontaneous Ca 2+ signals. Interestingly, there was no similar relationship governing the interval between a spontaneous Ca 2+ signal and a subsequent Ca 2+ pulse induced by EFS (Fig. 3C, right). This is likely due to the fact that an EFS pulse triggers a Ca 2+ influx signal to promote internal Ca 2+ release. The EFS-induced Ca 2+ signals may therefore be less dependent on SR Ca 2+ refilling in comparison to the spontaneous Ca 2+ events.
We hypothesised that increased EFS frequency might overcome the confounding effects of spontaneous Ca 2+ signals and provide a more reliable entrainment of PVCs. However, elevated EFS frequencies only partially improved entrainment, and moreover caused markedly heterogeneous cell responses. For example, fig. 4 shows the responses of two adjacent PVCs to EFS applied at 2 and 3 Hz. Prior to EFS, both cells displayed independent spontaneous Ca 2+ signals (Fig. 4A). Cell 1 was fully paced (without any remaining spontaneous activity, FP) with 2 Hz (Fig. 4B ii ) and 3 Hz EFS. However, in response to 3 Hz, Cell 1 displayed 'alternans' (alternating large and small Ca 2+ transients, fig. 4C ii ). Conversely, Cell 2 was only partially paced (PP) with 2 Hz EFS (Fig. 4B iii ), but was fully paced (FP) with 3 Hz EFS without alternans (Fig. 4C iii ). These observations indicate that while increasing the frequency of EFS could sometimes entrain spontaneous Ca 2+ signals, other forms of pro-arrhythmic activity, e.g. alternans, could ensue.  Figure 4D quantitatively summarizes the relationship between the frequency of EFS and the pattern of PVC Ca 2+ signaling. It is evident that PVCs were more successfully paced as the EFS frequency increased. However, not all PVCs were paced (Fig. 4D i ). PVCs that were partially paced were more evident at lower EFS frequencies (Fig. 4 D ii ), whilst higher EFS frequencies were most effective for complete entrainment (Fig. 4D iii ). However, the higher EFS frequencies were also capable of generating alternans (Fig. 4D iv ). We suggest that the heterogeneity of EFS-induced Ca 2+ signaling arises from the relative timing of an EFS pulse with respect to the intrinsic spontaneous Ca 2+ signaling activity; it is easier to fully pace cells which show little spontaneous activity, and more difficult to pace cells with a high frequency of spontaneous activity. In support of this conclusion, we observed that the minimal EFS frequency required for fully paced cells correlated with the intrinsic frequency of spontaneous Ca 2+ signals (Fig. 4E). Those cells with more frequent spontaneous Ca 2+ signals required higher EFS frequencies. Moreover, at all EFS frequencies tested, the cells that could be fully paced had a lower average frequency of spontaneous Ca 2+ signals compared to those cells that could not be fully paced. With 1 Hz EFS, for example, fully paced PVCs had an average spontaneous Ca 2+ signal frequency of 0.760.1 Hz, whereas those cells that could not be fully paced by 1 Hz EFS had an average spontaneous Ca 2+ signal frequency of 1.960.2 Hz (n = 20, 5 slices). We suggest that all PVCs have unique periodicities of spontaneous Ca 2+ signaling, and this determines the cells' refractory periods. In some cells, the frequency of spontaneous Ca 2+ signaling is so high that EFS is ineffective. It should be noted that whilst there is a clear impact of the frequency of spontaneous Ca 2+ signals on the pacing of PVCs, this is not the only factor. We also observed cells with similar frequencies of spontaneous Ca 2+ signaling responding differently to the same EFS pulse. This implies heterogeneity in the responsiveness of the cells to EFS in addition to the effects of spontaneous Ca 2+ signaling.

Ca 2+ Re-addition Induces Rapid Ca 2+ Waves within PVCs and Prevents Electrical Synchronisation
Superfusion of non-paced PVCs with Ca 2+ -free medium caused a progressive reduction in the amplitude and frequency of spontaneous Ca 2+ signals (Fig. 5A), and of the basal cytosolic Ca 2+ concentration, until the PVCs became quiescent (quantified in figs. 5F i-iv ). In 58% of the PVCs, the spontaneous Ca 2+ activity was fully inhibited within 5 min of Ca 2+ removal (mean time to inhibition: 209.5615.3 s). This inhibition of the spontaneous Ca 2+ signals also correlated with the loss of spontaneous PVC contractions. These data indicate i) that spontaneous Ca 2+ signals arise via Ca 2+ release from intracellular stores, ii) that PVCs have a substantial capacity for recycling Ca 2+ , and iii) that PVCs require Ca 2+ influx to sustain spontaneous Ca 2+ activity and contraction.
The re-addition of extracellular Ca 2+ (1.3 mM) did not restore the initial pattern of spontaneous Ca 2+ activity, but rather induced an entirely different form of Ca 2+ signaling. Within 40 to 60 s of Ca 2+ re-addition, PVCs began to display Ca 2+ waves that progressively increased in amplitude, frequency and spatial propagation (Figs. 5A, B ii ; quantified in figs. 5C i-iv ). Moreover, these Ca 2+ waves had a longer duration than the original spontaneous activity (compare Figs. 5B iii and B i ; fig. 5C iv ). Consequently, the cytosolic Ca 2+ concentration did not have time to recover between wave fronts, so the diastolic Ca 2+ level became elevated. Typically, all the PVCs surrounding a pulmonary vein displayed Ca 2+ waves in response to re-addition of extracellular Ca 2+ (100% of slices; n = 12 slices). The Ca 2+ waves were associated with strong PVC contractions, but the contractions were not coordinated between cells with the result that the layer of PVCs fibrillated asynchronously. Significantly, the rapid Ca 2+ waves and strong contractions persisted for some time after Ca 2+ re-addition, and did not return to the original pattern of spontaneous activity within 15 minutes. The spontaneous Ca 2+ waves in PVCs under control conditions were predominantly intracellular. The Ca 2+ removal/re-addition experiments presented a situation in which intracellular Ca 2+ waves frequently turned into intercellular Ca 2+ waves (an example is shown in Fig. S4).
Application of EFS either prior to, or during, Ca 2+ re-addition did not prevent the PVCs from developing long-lasting Ca 2+ waves ( Fig. 5D and E; compare spontaneous Ca 2+ transients in Fig. 5E i with the increased duration Ca 2+ waves in Fig. 5E iii ; Video S3). Even slices showing an initial high degree of entrainment to EFS (Fig. 5 E i ) were driven to the fibrillated state that was unresponsive to EFS (Figs. 5 E ii and E iii ) by removal and re-addition of extracellular Ca 2+ . Similarly, EFS could not re-synchronize the contractile activity of the PVCs following re-addition of extracellular Ca 2+ (Figs. 5 F i-iv ). Re-addition of a lower extracellular Ca 2+ concentration (500 mM instead of 1.3 mM), slowed the onset of the rapid Ca 2+ waves and contractions, but the PVCs eventually reached a similar state of fibrillation that was unresponsive to EFS.
Our data indicate that PVCs can progress to an unrecoverable, EFS-insensitive form of Ca 2+ signaling that is characterized by Ca 2+ waves with a longer duration and a persistent elevation of cytosolic Ca 2+ . Moreover, other maneuvers (see below) that promoted Ca 2+ influx or altered Ca 2+ homeostasis could also cause PVCs to display a rapid, EFS-resistant, form of Ca 2+ signaling.
The fact that spontaneous Ca 2+ signals continue for some time in the absence of extracellular Ca 2+ suggests that these Ca 2+ signals depend on Ca 2+ release from internal stores. Consistent with this idea, we observed that caffeine (a RyR agonist; 1 mM) increased the frequency of spontaneous Ca 2+ transients and typically caused a sustained elevation of cytosolic Ca 2+ (Fig. 6A and B, quantified in E i and E ii ). PVCs with EFS showed a similar caffeine-evoked increase in spontaneous Ca 2+ activity, and consequent loss of entrainment in 47% of the cells (Fig. 6C and D, quantified in E iii ). The effects of caffeine were reversible within 30-60 s (Fig. 6A ii and   B ii ), and the PVCs could be re-entrained by EFS (compare responses between 30 to 60 seconds in Fig. 6 C i /D i with C ii /D ii ). Ryanodine (a RyR antagonist; 10 mM) inhibited the spontaneous Ca 2+ signals within ,150 s (Figs. S5A and B) and completely blocked EFS-evoked Ca 2+ signals within ,100 s in an irreversible manner (Figs. S5C and D). We attempted to use a maximal caffeine concentration to provoke emptying of the SR Ca 2+ store. This is a commonly-employed method for assessing the SR Ca 2+ load in cardiomyocytes. However, we found that 20 mM caffeine did not produce the same large, monophasic Ca 2+ responses in PVCs that it evokes in isolated cells. Rather, application of 20 mM caffeine caused an acceleration of Ca2+ transients, leading to a persistent increase of the cytosolic Ca 2+ concentration (Fig. S6).
Depolarising PVCs by elevating the extracellular KCl concentration (from 5.3 mM to 50 mM) progressively increased the frequency of spontaneous Ca 2+ signals (Figs. 7A i and B i , quantified in fig. 7F i ) and stimulated the occurrence of spontaneous events in PVCs that were previously silent (Figs. 7A ii and B ii ). KCl treatment had a similar effect to that of caffeine and induced rapid Ca 2+ waves superimposed on an elevated cytosolic Ca 2+ level, which were reversible upon washout (Fig. 7F ii ). The effects of KCl on PVCs during EFS were similar; KCl induced the progressive development of rapid Ca 2+ waves on an elevated Ca 2+ baseline to the point that synchronised Ca 2+ signals and contractions induced by EFS were prevented (Figs. 7C and D, F iii ; contractions in figs. 7E i-iv ). Pacing was lost after 1961.2 s of KCl application. As with caffeine, the rapid Ca 2+ waves evoked by KCl could be reversed upon washout and entrainment re-established 67.163.6 s after washout (Fig. 7D ii ).
Nifedipine (100 mM) stopped EFS-induced pacing within 63626.1 s (Figs. S7A and B) and contraction (Fig. S7C). While inhibiting the EFS-induced Ca 2+ signals, nifedipine promoted the re-occurrence of spontaneous Ca 2+ events which had been previously suppressed by EFS (Figs. 1B and S7B i -B iv ). In the absence of EFS, nifedipine reduced the frequency of spontaneous Ca 2+ signals in a reversible manner (Figs. S7D and E). This latter effect indicates that VOCCs provide a source of Ca 2+ influx to support or trigger spontaneous Ca 2+ events. Verapamil (another L-type VOCC blocker; 100 mM) had similar effects to nifedipine (data not shown).

Discussion
In the present study, we characterised spontaneous and EFSinduced Ca 2+ signals within PVCs, and examined the effect of spontaneous Ca 2+ signaling on the ability of the PVCs to respond to sustained electrical pacing. The use of the lung slice preparation for examining PVCs has the advantages of easy preparation, a cellular architecture resembling the in situ organization, compatibility with microscopy and a viability lasting several days. Spontaneous Ca 2+ signals have previously been recorded in PVCs from various species [24][25][26], but their influence on pacing has not been resolved. A critical aspect of PVCs that was apparent in this study was the heterogeneity between cells in terms of Ca 2+ handling. PVCs display repetitive spontaneous Ca 2+ release events, but the frequencies and spatial properties of those events are unique for each cell. Correspondingly, PVCs show heterogeneous responses to electrical pacing such that neighboring cells can become, and remain, desynchronised. Such heterogeneity is known to serve as a basis for cardiac arrhythmias, and this may underlie the generation of arrhythmic events within pulmonary veins.
PVCs play an important role in the genesis and maintenance of AF [15,[27][28][29]. Electrical mapping studies have pinpointed PVCs as a source of phase 4 depolarisations (i.e. APs during the normally quiescent diastolic period) [30] that can pervade the atria via the junction of the atrium and PV at the left atrial antrum. Ablation procedures that prevent the electrical continuity between PVCs and atrial myocytes are highly effective in treating AF, but can have serious side-effects [31]. Despite the confirmation of PVCs as a source of ectopic activity, the events that occur within these cells to generate arrhythmic signals are less clear [11]. With respect to the mechanisms generating arrhythmias, numerous studies, using a range of cardiac cell types, have demonstrated that spurious Ca 2+ signals are a potent source of arrhythmic activity [32]. Moreover, 3-dimensional electrical mapping of pulmonary veins identified spontaneous activity occurring as discrete focal events, consistent with the notion that cellular Ca 2+ signals may be involved [33].
Under normal conditions, contractile myocytes and conductive cells within the heart experience a Ca 2+ signal during the passage of each AP. At the end of an AP, cytosolic Ca 2+ must recover to diastolic levels in order for the cells to reset and be fully responsive to the next stimulation. Typically, there is a diastolic period of hundred(s) of milliseconds between each AP where Ca 2+ remains at the resting level of ,100 nM. Consequently, spontaneous Ca 2+ signals may corrupt this normal activity and recovery cycle of cardiac cells in a number of ways [32]. For example, Ca 2+ causes the activation of electrogenic ion transporters, such as Na + /Ca 2+ exchange or Ca 2+ -dependent chloride flux that leads to cellular depolarisation. Ca 2+ signals occurring during the recovery of an AP, or in the normally quiescent diastolic period, give rise to 'triggered activity' such as early-or delayed-after-depolarisations (often denoted 'EADs' and 'DADs' respectively) [34]. If EADs and DADs occur with sufficient magnitude, they can cause a cell to depolarise and trigger an AP that propagates to neighboring cells. It is not known how many PVCs would need to act in concert to be a focus for ectopic AP generation. Most likely, coordination of multiple cells is required to provide a sufficient signal for a propagating wave of depolarisation. However, even those EADs and DADs that do not reach the threshold for triggering an AP can disrupt the cyclical activation of myocytes by rendering them refractory, and thereby causing them to lose synchrony with their neighbours; a putative mechanism for generating electrical reentry circuits. In the longer term, spontaneous Ca 2+ signals can alter gene expression, and lead to deleterious phenotypic remodelling of myocytes [35].
A key finding in the present study was the propensity of PVCs to perpetually display spontaneous Ca 2+ signals even in the presence of electrical pacing. It would be reasonable to expect that EFS could extinguish spontaneous Ca 2+ signals by providing regular command pulses, and normalising SR Ca 2+ content. Consistent with this idea, we observed that EFS could dampen spontaneous Ca 2+ release, in particular in those cells with relatively low spontaneous Ca 2+ activity. However, the effectiveness of electrical pacing was determined by the frequency of spontaneous Ca 2+ signals. In essence, if spontaneous Ca 2+ release occurred close to the EFS frequency then pacing was either partially successful or unsuccessful. Similar observations have been made using model stimulations of spontaneously active atrial myocytes [36]. The tendency of PVCs to show spontaneous Ca 2+ signals would certainly contribute to heterogeneous electrical responses within a pulmonary vein [16]. Such heterogeneity could lead to the generation of focal activity that may eventually propagate to the heart and disturb the cardiac cycle. Moreover, events that change the occurrence of spontaneous Ca 2+ release, such as alteration of Ca 2+ influx or RyR sensitivity can further decrease the ability of APs to control PVCs and enhance the likelihood of proarrhythmic events. Suppressing spontaneous Ca 2+ signaling in PVCs may therefore provide a plausible target to prevent the inception of AF. Other studies have also suggested clamping of RyR activity as a putative mechanism for controlling pulmonary vein-induced arrhythmogenesis [37]. Our data suggest that VOCC-mediated Ca 2+ influx is necessary for EFS, but also provides a source of Ca 2+ for spontaneous Ca 2+ signaling. In the present study, we did not characterise all putative Ca 2+ influx mechanisms, in particular those activated by removal/re-addition of extracellular Ca 2+ , but multiple Ca 2+ influx pathways may be involved, as in smooth muscle [38].
The reason why PVCs have a high propensity to show spontaneous Ca 2+ signals is unclear. In previous studies of PVCs [39] and atrial myocytes [40] InsP 3 Rs were identified as triggers for spontaneous Ca 2+ signaling, but these channels did not play a detectable role within the PVCs used in this study. Since SR Ca 2+ content is known to regulate RyR activation [41], a plausible explanation is that PVCs have a relatively high SR Ca 2+ content so that their RyRs open spontaneously. In support of this notion, experiments designed to increase SR Ca 2+ content using KCl, or RyR sensitivity using caffeine, increased spontaneous Ca 2+ signaling. In addition, it is evident that PVCs have sufficient SR Ca 2+ to allow spontaneous Ca 2+ signals to occur for several minutes in a Ca 2+ -free medium. All manoeuvres that provoked PVCs to show more rapid spontaneous events also made the PVCs resistant to electrical pacing. We therefore suggest that if the SR Ca 2+ content, Ca 2+ influx or RyR sensitivity are sufficiently enhanced, spontaneous Ca 2+ activity will be accelerated and the PVCs become insensitive to EFS. Video S1 Spontaneous and EFS induced contraction of a pulmonary vein. Phase contrast image of a cross section through a pulmonary vein, showing a small amount of spontaneous contraction around the whole circumference of the vein. After application of EFS a stronger contraction of the pulmonary vein is observed in response to every electric pulse. Image acquisition rate 15 frames per second. (MP4) Video S2 Spontaneous and EFS induced Ca 2+ signals in PVCs. The video shows spontaneous Ca 2+ signaling in Oregon-Green BAPTA-1 loaded PVCs. The spontaneous Ca 2+ waves either stay intracellular or travel through several PVCs. In response to EFS a simultaneous whole cell Ca 2+ increase is seen in all PVCs. The spontaneous activity continues in-between the EFS pulses. Image acquisition rate 30 frames per second. (MP4) Video S3 Ca 2+ signals during a Ca 2+ removal and readdition experiment. The first part of the video shows the spontaneous Ca 2+ signals in PVCs before Ca 2+ removal from the superfusion medium. The observed Ca 2+ waves are predominantly intracellular. The second part of the video shows the absence of spontaneous Ca 2+ signals after 5 minutes in Ca 2+ free sHBSS (100 mM EGTA). A gradual increase in Ca 2+ transients after readdition of 1.3 mM Ca 2+ is shown in the third part of the video. In contrast to the spontaneous activity seen in the first part, most of the Ca 2+ waves after Ca 2+ re-addition are intercellular Ca 2+ waves, travelling though several cells. Strong, uncoordinated contractions develop during the Ca 2+ re-addition period. Image acquisition rate 30 frames per second. (MP4)