Burst pacemaker activity of the sinoatrial node in sodium–calcium exchanger knockout mice

Significance

The sinoatrial node (SAN) generates cardiac pacemaker activity through the interplay of membrane ionic currents and intracellular calcium cycling. SAN dysfunction is a common disorder that usually requires implantation of costly electronic pacemakers. To study the role of intracellular calcium regulation by the sodium/calcium exchanger (NCX) in SAN pacing, we generated an atrial-specific NCX knockout mouse. The SAN beating pattern in these mice is abnormal, with bursts of activity interrupted by frequent pauses reminiscent of clinical SAN disease. We found that cellular calcium accumulation was responsible for this abnormal beating pattern, underscoring the importance of NCX-mediated calcium efflux to normal pacing. We propose that burst firing is a common feature of SAN dysfunction caused by elevated cytoplasmic calcium.

Abstract

In sinoatrial node (SAN) cells, electrogenic sodium–calcium exchange (NCX) is the dominant calcium (Ca) efflux mechanism. However, the role of NCX in the generation of SAN automaticity is controversial. To investigate the contribution of NCX to pacemaking in the SAN, we performed optical voltage mapping and high-speed 2D laser scanning confocal microscopy (LSCM) of Ca dynamics in an ex vivo intact SAN/atrial tissue preparation from atrial-specific NCX knockout (KO) mice. These mice lack P waves on electrocardiograms, and isolated NCX KO SAN cells are quiescent. Voltage mapping revealed disorganized and arrhythmic depolarizations within the NCX KO SAN that failed to propagate into the atria. LSCM revealed intermittent bursts of Ca transients. Bursts were accompanied by rising diastolic Ca, culminating in long pauses dominated by Ca waves. The L-type Ca channel agonist BayK8644 reduced the rate of Ca transients and inhibited burst generation in the NCX KO SAN whereas the Ca buffer 1,2-Bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid (acetoxymethyl ester) (BAPTA AM) did the opposite. These results suggest that cellular Ca accumulation hinders spontaneous depolarization in the NCX KO SAN, possibly by inhibiting L-type Ca currents. The funny current (I_f) blocker ivabradine also suppressed NCX KO SAN automaticity. We conclude that pacemaker activity is present in the NCX KO SAN, generated by a mechanism that depends upon I_f. However, the absence of NCX-mediated depolarization in combination with impaired Ca efflux results in intermittent bursts of pacemaker activity, reminiscent of human sinus node dysfunction and “tachy-brady” syndrome.

Physiological heart rhythm originates in the sinoatrial node (SAN), a cluster of specialized pacemaker cells located on the endocardial surface of the right atrium (RA). SAN dysfunction (SND) leads to serious arrhythmias characterized by pathological pauses, often alternating with rapid heart rates or atrial fibrillation (1). Each year in the United States, close to 200,000 patients affected with SAN disease require surgical implantation of an electronic pacemaker (2). Therefore, advances in our understanding of SAN pacemaker activity are essential for developing new therapies to avoid this costly procedure and its related morbidity.

In SAN pacemaker cells, action potentials (APs) are thought to be triggered by spontaneous diastolic depolarization (SDD) produced by a coupled system of cellular “clocks” (3). The first clock, known as the “membrane clock,” initiates SDD in response to inward funny current (I_f) carried mostly by HCN4 channels (4) although other ion channels, like voltage-dependent Ca channels, have also been implicated (5). The second (and more controversial) clock is referred to as the “Ca clock.” This clock produces a depolarizing current in late diastole when local Ca released by ryanodine receptors (RyRs) on the sarcoplasmic reticulum (SR) is extruded by the electrogenic sodium–calcium exchanger (NCX) (3). Both clocks are then synchronized by the opening of L-type Ca channels (LTCCs) during the upstroke of the AP. Ca entry through LTCCs also refills the SR, thereby priming the system for its next release cycle (3).

Several studies have demonstrated the importance of NCX in the mechanism of spontaneous SAN cell depolarization (2, 3, 6–8). However, NCX is also the dominant Ca efflux mechanism in cardiomyocytes (9), and thus an essential mechanism for maintaining cellular Ca balance. How the impairment of Ca efflux could affect SAN automaticity has not been investigated.

To study how NCX contributes to SAN pacemaker activity, we previously created an atrial-specific NCX knockout (KO) mouse, where NCX is totally eliminated from the atria, including the SAN (2). These mice live into adulthood and are healthy despite a lack of P waves on electrocardiogram (ECG). Atrial electrograms (EGMs) are silent, consistent with atrial quiescence, and isolated NCX KO SAN cells lack spontaneous APs despite normal I_f (2). Nevertheless, it is possible to pace the atria and single SAN cells from these mice. Although we found no evidence of spontaneous atrial depolarization in the NCX KO mouse, we did not examine SAN activity directly.

We developed a method for dynamic 2D imaging of intracellular Ca and voltage mapping in the intact explanted mouse SAN. This method takes advantage of the unique properties of the mouse SAN, which is only a few cell layers thick. Using this approach, we found that the intact NCX KO SAN tissue depolarizes spontaneously, contrary to our findings in isolated single cells. However, this pacemaker activity is characterized by bursts of APs alternating with pauses, a pattern reminiscent of the clinical “tachy-brady” syndrome of SND (1). We found that, when intracellular Ca is normal, the complex interplay between I_f and other ion channels is sufficient to generate bursts of APs despite the complete absence of NCX. However, rising intracellular Ca during bursts, combined with the lack of NCX-mediated depolarization, terminates spontaneous pacing. This oscillation of intracellular Ca explains the burst pacing pattern of the NCX KO SAN.

Results

Electrical Activity in the Intact SAN/Atrial Tissue.

We dissected and visualized the entire SAN/atrial tissue, including the SAN, RA, and left atrium (LA), using a stereomicroscope with low magnification (7×). In the wild-type (WT), we observed coordinated SAN and atrial contractions (Movie S1). In contrast, NCX KO SANs displayed intermittent contractile activity, and atria were quiescent (Movie S2). To determine whether this mechanical activity was caused by depolarizations, we performed optical voltage mapping. WT tissues exhibited rhythmic APs originating in the SAN (Fig. S1) and spreading synchronously to both atria (Fig. 1A). In the NCX KO, we instead recorded bursts of APs originating in the SAN (Fig. S1). Bursts were interrupted by pauses, which significantly reduced the average AP frequency (Fig. 1B).

Fig. S1.

Leading region of pacemaker activity recorded with voltage mapping (colored dots) marked on an example of sinoatrial node (SAN) tissue in (A) WT (n = 8) and (B) NCX KO (n = 6). Although the anatomical features of the KO SAN are clear and suggest that their spontaneous activity originates in the SAN, we cannot absolutely exclude the presence of intermittent latent pacemaker foci outside the classical confines of the SAN in NCX KO. CT, crista terminalis; IAS, interatrial septum; IVC, inferior vena cava; LA, left atrium; RA, right atrium; SVC, superior vena cava.

Fig. 1.

Optical voltage mapping of explanted SAN/atrial tissue. (A) Isochronal voltage maps from WT and NCX KO tissue (Left) and corresponding optical recordings of APs from discreet locations (Right). (B) AP frequency in NCX KO (n = 11) and WT tissue (n = 8). ***P < 0.001, unpaired t test.

In 6 of 11 NCX KO tissues, electrical activity originating in the SAN did not excite the atria, suggestive of SAN exit block (Fig. 1A). In five SANs, we observed partial depolarization of the atria or intermittent conduction block. In those cases where the NCX KO SAN depolarized the atria, conduction velocity was slower than in WT (time from SAN to RA depolarization: KO, 55 ± 2 ms, n = 4; WT, 14 ± 1 ms, n = 6; P < 0.001, unpaired t test). These results explain the lack of P waves on ECG in NCX KO mice.

Because SAN enlargement (WT, 10 ± 1 mm², n = 20; KO, 14 ± 1 mm², n = 13; P < 0.05, unpaired t test) and atrial remodeling are evident in our NCX KO mouse (Fig. S1) (2), we reasoned that abnormal conduction could be attributed to fibrosis (10, 11) or changes in connexins (Cxs) (12). Indeed, Masson’s trichrome stain showed increased fibrosis in the SAN, RA, and LA of NCX KO compared with WT (Fig. S2 A and B). Quantitative PCR (qPCR) revealed a significant decrease in Cx40 whereas there was no change in Cx45 or Cx43 (Fig. S2C).

Fig. S2.

Fibrosis and connexin expression in the NCX KO SAN and atria. (A) Representative light micrographs of Masson’s trichrome staining of the sinoatrial node (SAN), right atrium (RA), and left atrium (LA) in WT (Upper) and KO (Lower). Blue areas correspond to fibrosis. (B) Summary plots of percent area with fibrosis in six WT and four NCX KO tissues. (C) Expression level of Cx40, Cx43, and Cx45 measured by qPCR, relative to the housekeeping gene GAPDH. *P < 0.05, **P < 0.01, ***P < 0.001 by unpaired t test.

Ca Transients in the Entire SAN Tissue.

To characterize Ca transients in the intact SAN from WT and NCX KO mice, we loaded the SAN/atrial tissue with the Ca dye Cal-520 AM (13) and used high-speed 2D confocal microscopy at physiological temperature to image a large portion of the SAN (14) (Fig. 2 and Movies S3 and S4). In WT, Ca transients occurred simultaneously throughout the SAN (Fig. 2A) and were synchronized with contractions (Movie S3). The Ca transient rate (313 ± 12 transients per min, n = 28) was comparable with the spontaneous rate of Langendorff perfused hearts (15). Transients had rapid upstrokes (Table S1), consistent with Ca release elicited by depolarization (14). We observed similar Ca transients, synchronized with contractions, in NCX KO SANs (Fig. 2 and Movie S4). However, their firing pattern was characterized by bursts interrupted by pauses, so that the average Ca transient rate (106 ± 8 transients per min, n = 40) was reduced compared with WT (P < 0.001, unpaired t test). Ca transients in KOs also exhibited a longer time to peak than WT (Right Insets of Fig. 2B and Table S1). Still, there was no difference in amplitude or decay rate (Table S1), consistent with our observations in paced NCX KO ventricular myocytes (16).

Fig. 2.

Ca dynamics in the intact SAN tissue. (A) Time series of 2D confocal images of Ca [Cal-520 fluorescence in arbitrary units (a.u.); 50-Hz acquisition speed] in WT and NCX KO SANs. (B) Spatial average of fluorescence normalized to baseline (F/F₀) over the entire field of observation in WT and NCX KO SANs. (Right Insets) Ca transients on a 400-ms time scale. (C) Distribution of Ca transient rates in WT compared with the burst firing (BF) rates in NCX KO.

Table S1.

Ca transient parameters recorded in WT and NCX KO

Parameter	WT	NCX KO	Unpaired t test
Time to peak, ms	13.3 ± 0.4 (n = 23)	22.1 ± 1.0 (n = 20)	p < 0.001
ΔF/F0	1.2 ± 0.1 (n = 23)	1.0 ± 0.1 (n = 20)	NS
Tau, ms	33 ± 2 (n = 13)	29 ± 2 (n = 14)	NS

To confirm that the Ca transients described here corresponded to electrical activity, we simultaneously recorded tissue depolarization using three sets of electrodes surrounding the SAN/atrial preparation (Fig. S3C). This approach produced bipolar EGMs (17) showing 100% concordance between depolarization and Ca transients in the SAN of both genotypes. EGMs confirmed the absence of electrical activity during pauses in the KO (Fig. S3 A and B).

Fig. S3.

Simultaneous recordings of Ca and voltage in WT and NCX KO SAN/atrial tissue. (A) Ca transients recorded using rapid 2D confocal microscopy (Upper) and simultaneous recordings of EGMs (Lower) using external electrodes in a WT SAN/atrial preparation. Recordings are presented with standard (Left) and expanded (Right) time scales. (B) Ca (Upper) and voltage (Lower) recordings from a NCX KO, where the EGM signals are generated only by SAN activity. (C) Electrode placement surrounding the ex vivo SAN/atrial tissue. The position of these electrodes reflects the Einthoven triangle used for human ECG. Only lead I is reported in A and B. SVC and IVC stand for superior and inferior vena cava, respectively.

In ∼80% of NCX KO SAN tissues (n = 21 out of 27), we observed low amplitude Ca transients during the decay phase of the preceding transients. These “early aftertransients” (EATs) were suggestive of early afterdepolarizations (EADs) (Fig. S4A, red arrows). In contrast, only 14% of WT SANs showed rare EATs. Simultaneous recording of EGMs showed that EATs corresponded to voltage changes. Using optical voltage mapping, we observed events suggestive of EADs (red arrows, Fig. S4B) in 75% of NCX KO tissues (n = 8) whereas there were none in WT (n = 8).

Fig. S4.

Early aftertransients (EATs) and early afterdepolarizations (EADs) in the NCX KO SAN. (A) Simultaneous recording of Ca fluorescence (Upper) and EGMs (Lower) using the same methods as Fig. S3, showing irregular spontaneous activity of an NCX KO SAN that exhibited frequent EATs (red arrows). Insets (Right) show expanded views of the simultaneous recordings. Inset 1 shows a normal Ca transient and depolarization whereas Inset 2 shows a Ca transient with an EAT. (B) Optical voltage recording of NCX KO SAN activity showing EADs (red arrows).

L-Type Ca Current in NCX KO SAN Cells.

We previously reported that NCX KO SAN cells have reduced L-type Ca current (I_Ca,L) amplitude (2). To determine whether Ca-dependent inactivation of I_Ca,L is responsible for the reduction in amplitude, we recorded I_Ca,L in isolated SAN cells dialyzed with a pipette solution containing either a low or high concentration of the Ca buffer 1,2-Bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid (BAPTA, 1 or 10 mM). Under low buffering conditions, the amplitude of I_Ca,L at a test potential of −5 mV was decreased in KO cells by ∼50% compared with WT (Fig. 3A), confirming our previous results (2). Under high buffering conditions, I_Ca,L was nearly identical in WT and KO (Fig. 3B). These findings suggest that I_Ca,L reduction in the KO is attributable to Ca-dependent inactivation, likely caused by high subsarcolemmal Ca in the absence of NCX, and similar to our results in ventricular NCX KO cells (18). This reduction in I_Ca,L would help to balance cellular Ca, given the reduced Ca efflux capacity in NCX KO SAN, and could also explain the slowed upstroke of Ca transients recorded in KO tissues.

Fig. 3.

L-type Ca current (I_Ca,L) in isolated SAN cells. (A) Representative I_Ca,L recordings (Upper) and average current-voltage plots (Lower) from 9 WT (black) and 12 NCX KO (gray) single SAN cells, buffered by a low concentration of BAPTA (1 mM) in the patch pipette. (B) I_Ca,L recorded under high concentration of BAPTA (10 mM) in WT (n = 6) and KO (n = 7) SAN cells. ***P < 0.001, *P < 0.05 by two-way ANOVA with Bonferroni’s posttest.

Burst Pattern of Automaticity in NCX KO SAN Tissue.

The alternation of bursts and pauses observed in the NCX KO (Fig. 2B and Fig. S3B) is an unexpected pattern of SAN pacemaking, never observed in WT, and distinct from the arrhythmic bradycardia characterizing most SAN models of impaired depolarization (7, 19, 20). To further investigate this pattern, we defined a burst as any train of three or more Ca transients (21) followed by a pause (14). The average number of Ca transients per burst was 11 ± 1 (n = 25 SANs). Bursts (on average, lasting 2.4 ± 0.2 s, n = 25 SANs) occupied approximately half of the recording period in NCX KOs whereas the balance consisted of either pauses (lasting 2.1 ± 0.1 s, n = 25 SANs) or slow irregular activity (8.1 ± 1.3% of the total recording) (Fig. S4A).

As indicated above, the average frequency of Ca transients (including the pauses) in the NCX KO SAN was slower than in the WT. However, the rate of transients within each burst, here indicated as burst-firing (BF) rate (5.7 ± 0.5 Hz, n = 25 SANs) (22), was as fast as the rate measured in WT (5.0 ± 0.2 Hz, n = 22 SANs; P = NS, by unpaired t test). Nevertheless, the BF rate distribution of NCX KO was wider than the rate distribution in WT and skewed toward higher frequencies, as in tachy-brady syndrome (Fig. 2C). In ∼70% of bursts (from 27 NCX KO SANs), there was also a gradual slowing (by 56 ± 3%) of the Ca transient frequency within the burst (Fig. 4 A and B), a phenomenon reminiscent of spike adaptation in neurons (23, 24).

Fig. 4.

Diastolic Ca increase and spike adaptation in the NCX KO SAN tissue. Gradual (A) and sudden (B) increase in diastolic Ca during bursts, followed by gradual (A) and rapid (B) return to baseline Ca during pauses. Lower Insets highlight Ca increase during the first burst in A and B. Note Ca transients slowing during the bursts (spike adaptation). (C) Prolonged bursts with a high burst-firing (BF) rate are associated with higher postburst diastolic Ca and longer pauses compared with the shorter bursts in the tracing. (D) Percent increase in diastolic Ca corresponding to BF rate for those spontaneous bursts (∼80% of total) where diastolic Ca increased (R² = 0.82, P < 0.01; n = 419 bursts in 25 NCX KO SANs).

Intracellular Ca During Spontaneous Bursts and Electrical Stimulation.

Previous studies have speculated that dynamic changes in intracellular Ca could cause burst behavior in SAN and neurons (10, 23, 25); however, this concept has not been rigorously tested. In NCX KO SANs, we found that diastolic Ca rose in ∼80% of bursts, either gradually (Fig. 4A) or suddenly (Fig. 4B). The extent of the Ca increase in these cases was proportional to the BF rate (Fig. 4D), and the longest and fastest bursts were associated with the longest pauses (Fig. 4C). Copious intracellular Ca waves were always present in confocal recordings during pauses (Fig. 2A, Fig. S5, and Movie S4).

Fig. S5.

Ca waves in NCX KO SAN tissue. (Left) A still frame from a 2D confocal image of Ca fluorescence in a wide field of the explanted NCX KO SAN tissue, showing two selected cells (marked by red dashed lines) intersected by the scanning line (white dashed line). Bright areas represent waves. (Right) Confocal line scan showing a burst of Ca transients followed by a pause. During the pause, numerous Ca waves were generated. Ca waves are not apparent during the rapid burst of Ca transients. The green fluorescence intensity indicates Ca concentration in arbitrary units (a.u.).

Because the rise of Ca in the SAN of NCX KOs correlated with burst termination, we hypothesized that abnormal Ca handling in the absence of NCX might be responsible. To examine this possibility in greater detail, we electrically stimulated both KO and WT SANs at increasing rates (Fig. 5) while recording diastolic Ca. In both WT and KO, faster rates led to proportional increases of diastolic Ca. However, the effect was much more pronounced in KO (Fig. 5 A and B). We next measured corrected SAN recovery times (cSNRTs) (26) in response to different pacing rates in WT and KO SANs. cSNRT is a common clinical test to investigate SAN dysfunction (27) and is defined as the period of quiescence after rapid pacing, normalized to the pacing rate (11, 26). In both WT and KO, we found that cSNRT was directly proportional to the rate of stimulation (Fig. 5B). However, cSNRT was prolonged in the KO compared with WT at all pacing rates tested. The longer cSNRT in KO coincided with the exaggerated increases in diastolic Ca associated with pacing (Fig. 5). Furthermore, return of diastolic Ca to baseline took much longer in NCX KO compared with WT, a phenomenon we also observed during long spontaneous bursts (Fig. 4C). These results indicate that the NCX KO SAN has a reduced capacity for recovery after rapid pacing compared with WT, consistent with the absence of NCX-mediated Ca efflux.

Fig. 5.

Diastolic Ca and sinus node recovery time (SNRT) of SAN/atrial tissue. (A) Spontaneous Ca transients in WT and NCX KO before and after 20 s of external pacing at 6, 8, and 14 Hz. (B) Plot showing the steeper linear rate dependence of diastolic Ca in KO (n = 4) compared with WT (n = 4). (C) Rate dependence and exponential fits of the corrected SNRT (cSNRT) in WT and KO. P < 0.001 by two-way ANOVA for WT vs. KO.

Intracellular Ca and Pacemaker Activity of NCX KO SAN.

To further investigate the relationship between cell Ca and burst termination, we lowered intracellular Ca using the acetoxymethyl ester (AM) form of the Ca buffer BAPTA. In WT tissues, incubation with a low dose of BAPTA AM (2 µM) for at least 30 min caused a decrease in the average AP rate, consistent with a slowing of SDD as shown previously (28) (Fig. 6 A and B). However, in NCX KO SANs, we obtained the opposite effect, with a decrease in the period of electrical silence [indicated as quiescent (Q) time] (Fig. 6). Although BAPTA improved NCX KO pacemaker activity by decreasing Q time, it also decreased BF rate in seven of eight NCX KO SANs (Fig. 6C).

Fig. 6.

Intracellular Ca modulation of SAN pacemaker activity. (A) Optical recordings of spontaneous APs in WT (Upper) and NCX KO (Lower) SAN tissue, before (Left) and after (Right) incubation with BAPTA AM (2 µM). (B) Changes in the overall rate of APs in WT (n = 5) and NCX KO (n = 7) with BAPTA. (C) Percent change in burst-firing (BF) rate, quiescent (Q) time and bursts per minute (B/min) in KO SAN (n = 7) after incubation with BAPTA. (D) Confocal recordings of Ca transients in WT (Upper) and NCX KO (Lower) SAN tissue, before (Left) and after (Right) perfusion with BayK (1 µM). (E) Changes in the overall rate of transients in WT (n = 7) and NCX KO (n = 6) with BayK. (F) Percent change, same parameters as in C, but in response to BayK. *P < 0.05, **P < 0.01, ***P < 0.001, unpaired t test.

Next, we increased Ca entry using the LTCC agonist BayK8644 (BayK, 1 µM) (29). BayK caused an increase in Ca transient frequency in WT SAN, consistent with prior studies (6, 19). However, in KO, we obtained the opposite response (Fig. 6 D and E), with an overall reduction in rate caused by a decrease in the number of bursts and an increase in Q time (Fig. 6F). Thus, our results with BayK are the exact opposite of what we obtained with BAPTA, consistent with an inability of NCX KO SANs to cope with additional Ca influx.

NCX KO Pacemaker Activity Is Responsive to I_f Inhibition and β-Adrenergic Stimulation.

To investigate the mechanism initiating spontaneous APs in the intact NCX KO SAN, we applied the specific I_f inhibitor ivabradine (IVA) to WT and KO SAN tissues. IVA (3 µM) caused a significant decrease in the overall rate of Ca transients in both WT and NCX KO (Fig. 7 A and B), similar to what has been reported previously in normal rabbit SAN tissue (30). In seven of eight NCX KO SANs, the inhibition of I_f not only reduced the BF rate, but also decreased the number of bursts during each recording (Fig. 7 B and C). Unlike BAPTA AM, which also reduced BF rate, IVA did not decrease Q time, suggesting that IVA suppressed impulse generation without altering cellular Ca.

Fig. 7.

I_f and β-adrenergic modulation of SAN pacemaker activity. (A) Confocal recordings of spontaneous Ca transients in a WT (Upper) and NCX KO (Lower) SAN, before (Left) and after (Right) perfusion with ivabradine (IVA, 3µM). (B) Changes in the overall rate of Ca transients in WT (n = 6) and NCX KO (n = 8) with IVA. (C) Percent change in burst-firing (BF) rate, quiescent time (Q) time, and bursts per minute (B/min) in NCX KO SAN tissue (n = 7) after IVA. (D) Spontaneous Ca transients in WT (Upper) and NCX KO (Lower) SAN tissue, before (Left) and after (Right) stimulation with ISO (1 µM). (E) Changes in the overall rate of Ca transients in WT (n = 5) and NCX KO (n = 11) with ISO. (F) Percent change in the same parameters as in C for the 6 NCX KO tissues (out of 11 tested) clearly responding to ISO (1 µM). *P < 0.05, **P < 0.01, unpaired t test.

The β-adrenergic pathway is an essential regulator of SAN pacemaker activity, which also results in Ca loading of the cell by increasing I_Ca,L and SR Ca ATPase (SERCA) uptake. β-adrenergic stimulation with isoproterenol (ISO, 1 µM) caused the expected rate increase in WT, but had no effect on the overall rate in NCX KO (Fig. 7 D and E). However, ISO caused an increase in both BF rate and Q time in 6 out of 11 NCX KO SANs (Fig. 7 D and F).

Discussion

As the dominant Ca efflux mechanism in cardiomyocytes, NCX ensures Ca balance by removing precisely the amount of Ca that enters the cell with each beat (9). In SAN pacemaker cells, NCX-mediated extrusion of Ca during late diastole also generates a depolarizing current in response to local Ca release by RyRs, a mechanism we and others have previously shown to be important for normal SAN cell automaticity (2, 6, 7). Although the contribution of NCX to late SDD has been widely investigated, less attention has been focused on the importance of NCX-mediated Ca efflux during SAN activity.

We found that the intact NCX KO SAN generated spontaneous APs/Ca transients that were mostly confined to the SAN region. This SAN “exit block” explains why P waves are absent in ECGs recorded from NCX KO mice, which instead display both accelerated and slow junctional rhythm (2). We also found an abnormal pacemaker pattern within the NCX KO SAN, characterized by bursts of transients at a physiological rate interrupted by pauses. This pattern is very different from the bradycardia that typifies most SAN models of impaired depolarization (7, 19, 20). Bursts in the NCX KO SAN led to increased diastolic Ca and Ca waves, consistent with abnormal Ca handling. External pacing led to similar increases in diastolic Ca associated with extremely long cSNRT. Consistent with these findings, Ca buffering with BAPTA AM improved pacemaker activity whereas increased Ca loading with BayK further inhibited it. These results indicate that automaticity in the NCX KO SAN is directly suppressed by impaired Ca efflux. We also found that SAN automaticity was sensitive to the I_f blocker IVA, as well as the β-agonist ISO, confirming the importance of funny channels and cAMP-dependent phosphorylation in SAN pacing, even in the absence of NCX.

Because isolated SAN cells from the NCX KO have no spontaneous APs (2), we were surprised to find any pacemaker activity in the intact SAN. Why single cell activity should differ so much from tissue behavior is uncertain, but such differences have been described in other organs such as pancreas (31). A similar mismatch between single cell and SAN tissue behavior has been described in a tamoxifen-induced HCN4 KO mouse (32). In our previous study (2), isolated NCX KO SAN cells had normal diastolic Ca, likely because these cells were electrically silent. Further study will be needed to explain the different behavior in cells and tissue, but we speculate that the more physiological cellular network of the intact SAN reinforces I_f-mediated depolarization.

Our results differ somewhat from the two other mouse models of NCX KO SAN (6, 7). Both of these models exhibit a much less severe phenotype in SAN cell function than our mice (2), which may be due to incomplete ablation of NCX. Nevertheless, there are certain similarities: Gao et al. (6) showed an impaired response to ISO and Bay K, as we did; Herrmann et al. (7) reported bursts alternating with pauses in ECG traces, comparable with our NCX KO SAN pattern.

Impaired Conduction to the Atria.

Voltage mapping in the NCX KO SAN revealed disorganized depolarizations, along with delayed or blocked conduction to the atria. There are several potential explanations for abnormal atrial conduction. Atrial remodeling occurs in our NCX KO mice (2), and we found that atrial fibrosis is increased compared with WT (Fig. S2). Fibrosis could interfere with conduction between myocytes (10), particularly in the setting of Cx40 reduction (Fig. S2). The cause of increased fibrosis in the atria of these mice is uncertain, but Ca overload has been implicated in the activation of apoptotic pathways in myocytes leading to fibrosis (11). A similar etiology has been suggested for the fibrotic changes and disordered conduction in the SAN and atria of calsequestrin KO (Casq2^−/−) mice (10). Whether Ca loading is also responsible for the decreased Cx40 in NCX KO atria is uncertain. However, Cx40 ablation is known to impair SAN/atria conduction (12). Loss of SAN cells through apoptosis in the NCX KO tissue, combined with the atrial enlargement, could also increase source/sink mismatch (11), thereby restricting spread of conduction to the atria.

Ca Accumulation in the NCX KO SAN.

In the absence of NCX, Ca removal is mediated by the only alternative Ca efflux mechanism, the plasma membrane Ca pump (PMCA) (33). Although this protein is highly expressed in NCX KO atria (2), it is a much slower transport mechanism compared with NCX. Thus, it is not surprising that bursts of spontaneous pacemaker activity in the KO led to significant increases in diastolic Ca followed by Ca waves and cessation of automaticity. Spontaneous automaticity returned once Ca recovered to baseline, presumably through the activity of PMCA. External pacing in the NCX KO SAN effectively resulted in the same response: a significant increase in diastolic Ca associated with an extremely long cSNRT compared with WT. The duration of the cSNRT was directly proportional to the pacing rate and consequent level of Ca accumulation. Presumably the increased diastolic Ca further inactivated an already compromised I_Ca,L in the KO SAN or, less likely, in latent extranodal pacemakers, thus contributing to the prolonged cSNRT by inhibiting depolarization. I_Ca,L inactivation by elevated diastolic Ca is also thought to cause pauses in RyR2^R4496C mice (14), a model of catecholaminergic polymorphic ventricular tachycardia (CPVT). Bursts of pacemaker activity followed by pauses, associated with a rise and fall of Ca, have also been observed in the SANs of another CPVT model, Casq2^−/− mice (10). These mice have abnormal SR Ca release and periods of elevated diastolic Ca corresponding to pauses. Finally, SAN cells isolated from heterozygous ankyrin-B KO mice, which have reduced NCX and Ca_v1.3 expression, also show a burst pattern attributed to impaired Ca homeostasis (34). Thus, burst firing seems to be a common feature caused by elevated cytoplasmic Ca and compromised I_Ca,L. Possibly Ca_v1.3 is also decreased in NCX KO although macroscopic I_Ca is fully restored by Ca buffering with BAPTA.

Both WT and NCX KO transiently accelerated their spontaneous rate after external pacing (Fig. 5A). We speculate that this increased rate may be a consequence of activation of the adenylate/cAMP pathway by Ca (35).

Similar to RyR2^R4496C mice, I_Ca,L was reduced at baseline in NCX KO SAN cells. This decrease in I_Ca,L could be mitigated by buffering intracellular Ca with BAPTA, a response we have also observed in isolated NCX KO ventricular myocytes (18). Because buffering intracellular Ca in single patch-clamped SAN cells could restore I_Ca,L, we tried the same approach in intact SAN tissue using the cell permeable Ca buffer, BAPTA AM. This treatment improved SAN pacemaker activity in the NCX KO by decreasing quiescent time. Conversely, increasing Ca entry using the LTCC agonist BayK suppressed NCX KO pacemaker activity by increasing quiescent time. β-adrenergic stimulation of NCX KO SAN with ISO increased the rate during the burst (BF rate) but, paradoxically, reduced overall rate because of increased Ca loading and consequent prolonged quiescent time (Fig. 7D).

Parallel Mechanisms Between NCX KO Pacemaker Activity and Burst Firing in Neurons and Chromaffin Cells.

Burst firing is a common property of different cell types, including pancreatic islet cells (31), chromaffin cells (36), and several types of neurons (29, 37, 38), but, to our knowledge, it has not been studied in the SAN. Burst-firing activity can occur in neurons during normal physiological function (38), after pharmacological stimulation (37), or as a consequence of Parkinson’s disease (39). In these systems, Ca modulation of specific ion channels [e.g., I_Ca,L, Small K (SK), Big K (BK)] causes burst termination (29, 36–38) whereas Ca-activated transient receptor potential melastatin (TRPM) channels (29, 40) are important for maintaining burst firing. Similar to our findings in the NCX KO SAN, buffering intracellular Ca in neuronal systems can alter burst firing (37).

In the NCX KO SAN, we found that the frequency of Ca transients during bursts gradually slowed until burst termination (Fig. 4). This slowing is reminiscent of spike adaptation (23, 24), a phenomenon well-described in neuronal and chromaffin cells and thought to be caused by activation of Ca-dependent K currents (22, 23, 36). Big K (41) and TRPM4 channels (40) are present in the SAN, and SAN dysfunction occurs after ablation of Small K channels (42). Thus, NCX KO SAN burst firing may be modulated by the same Ca-activated channels found in neurons and chromaffin cells. Ca activation of TRPM4 channels may also contribute to the EADs we recorded in the NCX KO SAN preparation (43).

Conclusion

Despite the lack of NCX-mediated spontaneous diastolic depolarization (i.e., a broken Ca clock), we found that I_f and I_Ca,L are capable of initiating bursts of APs in the SAN of atrial-specific NCX KO mice. Our results suggest that, in the absence of NCX-mediated Ca efflux, Ca entry during APs accumulates until the level is sufficiently high to terminate the burst. This termination is likely mediated in part by Ca-dependent inactivation of I_Ca,L. Other Ca-activated repolarizing currents (e.g., SK or BK) (41, 42) may also contribute to burst termination although this possibility was not tested.

In conclusion, our results suggest a new mechanism of pacemaker dysfunction caused by NCX ablation and consequent impaired Ca efflux. This mechanism is complementary to the impairment of SDD caused by the absence of NCX-mediated depolarization. Our study also highlights a remarkably common Ca-dependent burst-firing behavior that is present in both neurons and dysfunctional SAN, with many similarities to the clinical tachy-brady syndrome (1).

Methods

Detailed methods can be found in SI Methods. All animal procedures were reviewed and approved by the Institutional Animal Care and Use Committee of Cedars-Sinai Medical Center.

SI Methods

Generation of Atrial-Specific NCX KO Mice.

We have previously described the generation of atrial specific NCX1 knockout (KO) mice (2). Briefly, we produced the mice by crossing NCX1 exon 11 floxed mice (16) with sarcolipin Cre mice (2). NCX1 KO mice (referred to throughout this paper as NCX KO mice) are viable into adulthood. The animals used in this study were between 10 and 16 wk of age and were healthy. We used males and females in the ratio 70% to 30% but did not observe any sex-specific differences in rhythm. All animal procedures were reviewed and approved by the Institutional Animal Care and Use Committee of the Cedars-Sinai Medical Center. Animals were housed under standard conditions and allowed access to food and water ad libitum.

Intact SAN/Atrial Preparation.

We removed hearts via thoracotomy from heparinized (300 U i.p.) NCX KO mice or NCX floxed littermates (referred to throughout the paper as WT) anesthetized with isoflurane. We then separated the atria from the ventricles leaving the sinoatrial node (SAN) intact. The entire SAN/atrial explanted tissue was placed in heparinized (10 U/mL) modified Tyrodes solution heated to 36 °C. The Tyrodes solution contained 140 mM NaCl, 5.4 mM KCl, 5 mM Hepes, 5.5 mM glucose, 1 mM MgCl₂, and 1.8 mM CaCl₂ (pH adjusted to 7.4 with NaOH) and served as the control solution for all experiments. We used a stereomicroscope (SZX16; Olympus) with low magnification (7×) to transilluminate and visualize directly the isolated SAN/atrial preparation. We identified the SAN region using the borders of the superior and inferior vena cava, the crista terminalis, and the interatrial septum as landmarks (14). The SAN/atrial preparation including right and left atria (RA and LA) was pinned to the bottom of an optical chamber (Fluorodish, FD35PDL-100; WPI) coated with ∼2 mm of clear Sylgard (Sylgard 184 Silicone elastomer kit; Dow Corning). To maintain the SAN in a flat plane, we pinned the atrial preparation as shown in Movies S1 and S2. To avoid any interference from secondary pacemaker tissue, we removed the AV node from the preparation.

Voltage Mapping.

To analyze voltage changes in the SAN/atrial tissue, the entire preparation, including SAN, LA, and RA, was loaded by immersing the tissue in a Tyrodes solution containing the voltage-sensitive indicator RH237 (10 µM; Biotium) for at least 30 min at room temperature (20–22 °C). To maintain proper oxygenation of the tissue, a micromagnet agitated the solution during loading. After the loading step, the tissue was washed in dye-free Tyrodes for 15 min. During this step, the temperature was slowly increased to 34–36 °C. The SAN/atrial tissue was then constantly perfused at 34–36 °C and imaged by high speed optical voltage mapping (1 or 2 ms per frame) on a MiCAM Ultima-L complementary metal oxide semiconductor (CMOS) camera (100 × 100-pixel CMOS sensor, 10 × 10 mm) (SciMedia). This camera was mounted on a THT microscope, with two objectives (2× and 1.6×) that generated a field of view of 12.5 × 12.5 mm. A 150-W halogen light system with built-in shutter (SciMedia) was used as an excitation light source for the voltage dye. The filter set included a 531/50-nm excitation filter, 580-nm dichroic mirror, and 580 long-pass emission filter. To avoid motion artifacts, we blocked mechanical activity using blebbistatin (1.5–5 µM; Tocris Bioscience). We usually limited our recording times to 32.768 s (16,384 frames at 2 ms per frame) to avoid phototoxic effects of the dyes. Optical raw data were analyzed using dedicated software from the camera developer, BVAna Analysis Software (Brainvision). We used the background picture of the SAN/atrial preparation obtained with the CMOS camera to measure the SAN area (bounded by the crista terminalis and interatrial septum) in mm². Measurements were obtained by three independent observers, blinded to genotype, using ImageJ software (NIH image).

Fibrosis Quantification.

To quantify the amount of fibrosis in the SAN, RA, and LA, the atrial region of the heart was fixed overnight in phosphate-buffered 10% (wt/vol) formaldehyde (pH 7.4), and embedded in paraffin. Then 5-µm sections were cut parallel to the epicardial surface and stained for Masson’s trichrome as previously described (10). Briefly, we scanned each tissue preparation with a Leica SCN400 Slide Scanner. Digital images of whole cross-sections of the sample were saved for analysis. Between 10 and 15 random fields from each histological area (i.e., SAN, RA, LA) were analyzed at 400× magnification. The number of fields studied was enough to include the vast majority of the tissue. We used ImageJ with a custom thresholding macro to determine the ratio of fibrosis (blue) to cardiac tissue (red). Average results were expressed as percentage of fibrosis area per total tissue area.

Name	Amplicon size	Primer	Sequence (5′ → 3′)	Length	Location	Annealing temperature
Cx40	136	Forward	GGTCCACAAGCACTCCACAG	20	39–58	58.3
		Reverse	CTGAATGGTATCGCACCGGAA	21	174–154	57.3
Cx43	176	Forward	TCTGTGCCCACACTCCTGTA	20	256–275	58
		Reverse	TTGCCGTGTTCTTCAATCCCA	21	431–411	57.2
Cx45	245	Forward	AGATCCACAACCATTCGACAT	21	35–57	54.8
		Reverse	TCCCAGGTACATCACAGAGGG	21	279–259	58.1
GAPDH	169	Forward	TCACCACCATGGAGAAGGC	19	539–557	57.3
		Reverse	GCTAAGCAGTTGGTGGTGCA	20	688–707	58.3

RNA Preparation and qPCR.

Atrial cardiomyocytes were isolated by enzymatic digestion with Langendorff perfusion, using our ventricular myocyte protocol (18). Total RNA from two pools of five mice each was isolated and DNase-treated using an RNeasy Fibrous Tissue Mini Kit (Qiagen). The first strand cDNA was synthesized from 1 µg of the total RNA using an RT² First Strand Kit (Qiagen). PCR reactions were then mixed with RT² SYBR Green ROX qPCR Mastermix (Qiagen) and carried out in a 96-well plate using the 7900 HT Fast-Time PCR system (Applied Biosystems, Life Technologies brand). The gene primers used (IDT) are listed above. Data were collected by SDS 2.3 software (Applied Biosystems, Life Technologies brand) and analyzed using the threshold cycle (C_t) relative quantification method. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) reference gene was used for normalizing the data. The expression 2^-ΔCt corresponds to the ratio of each gene expression versus GAPDH.

Ca Imaging of the SAN.

To record cellular Ca, we immersed the SAN/atrial tissue in Tyrodes containing the Ca-sensitive indicator Cal-520/AM (10 µM; AAT Bioquest) (13) and Pluronic F-127 (0.13%; Invitrogen) for 30–45 min at 20–22 °C. To maintain proper oxygenation of the tissue, a micromagnet agitated the solution during loading. We chose Cal-520 because its signal-to-noise ratio is much higher than the more commonly used fluo-4 (13), allowing lower doses of the indicator to avoid Ca buffering. After the loading step, we washed the tissue with dye-free Tyrodes for 15 min before imaging while also increasing the temperature to 34–36 °C. We used the xyt mode (2D) of a Leica TCS-SP5-II (Leica Microsystems Inc.) to image intra- and intercellular Ca dynamics in the SAN region (14). We used 488-nm excitation and >505-nm emission with a 10× objective (N PLAN 10×/0.25; Leica) and scan speeds ranging from 36 to 5 ms per frame depending on the field size. The fluorescence intensity (F) proportional to Ca concentration was analyzed after background subtraction and normalized to baseline fluorescence, F₀ (F/F₀). To maintain a bath temperature of 34–36 °C, we used a temperature-controlled perfusion system (SHM-8, Warner Instruments), together with a warming chamber (QE-1; Warner Instruments).

Single SAN Cell Isolation.

We excised the SAN tissue visually by cutting along the crista terminalis and the interatrial septum in prewarmed (36 °C) Tyrodes solution. The tissue obtained was immersed into a “low-Ca-low-Mg²⁺” solution containing 140 mM NaCl, 5.4 mM KCl, 0.5 mM MgCl₂, 0.2 mM CaCl₂, 1.2 mM KH₂PO₄, 50 mM taurine, 5.5 mM d-glucose, 1 mg/mL BSA, and 5 mM Hepes-NaOH (adjusted to pH 6.9 with NaOH) for 2 min. Then we transferred the tissue into the low-Ca-low-Mg²⁺ solution containing a highly purified form of collagenase (Liberase TM; 229 U/mL; Roche), as a single digestive enzyme. Digestion was carried out for 15–20 min at 36 °C. To stop the digestion process, the SAN was washed using a modified “Kraftbrühe” (KB) medium containing 70 mM l-glutamic acid, 20 mM KCl, 80 mM KOH, 10 mM KH₂PO₄, 10 mM taurine, 1 mg/mL BSA, and 10 mM Hepes-KOH (adjusted to pH 7.4 with KOH). Single cells were isolated from the SAN tissue by manual agitation in KB solution at 36 °C for 2 min (14). To recover the automaticity of the SAN cells, Ca was gradually reintroduced to a final concentration of 1.8 mM (2).

Electrogram Recording of the SAN/Atrial Tissue.

Electrograms (EGMs) (17) of the SAN/atrial tissue were recorded using three electrodes (Fig. S3C), immersed in the same bath as the SAN/atrial tissue and pinned in close proximity to reduce electrical impedance. The position of the electrodes reflects the Einthoven triangle used for human ECG recordings. Therefore, this combination of electrodes generated three bipolar and three unipolar leads (I, II, and III, and aVR, aVL, and aVF, respectively). For data acquisition, we used the ML870 PowerLab 8/30, connected to two Animal Bio FE 136 Amplifiers (ADInstruments Inc.). EGM raw data were analyzed using dedicated software (Lab Chart 7 PRO; ADInstruments).

I_Ca,L Recording on Whole-Cell Patch Configuration.

We recorded I_Ca,L using the whole-cell patch clamp technique. We prepared electrodes from borosilicate pipettes (TW150F-3; WPI) using a Sutter P-97 electrode puller. Experiments were carried out using an Axopatch 200B amplifier (Molecular Devices), controlled by pClamp software (version 10.3; Molecular Devices, Inc.). The electrode resistance was ∼2 MΩ when the pipette was filled with 110 mM CsCl, 30 mM TEACl, 10 mM NaCl, 0.5 mM MgCl₂, 5 mM MgATP, 5 mM creatine phosphate, and 10 mM Hepes (adjusted to pH 7.2 with KOH). Then 1 mM or 10 mM BAPTA (A4926; Sigma-Aldrich) was included in the internal solution when required by the protocol (Fig. 3). The bath solution contained 136 mM NaCl, 5.4 mM CsCl, 0.33 mM NaH₂PO₄, 10 mM Hepes, 1 mM MgCl₂, 1.8 mM CaCl₂, 10 mM glucose, and 10 µM TTX (pH adjusted to 7.4 with NaOH). The bath temperature was 36 °C. Series resistance and capacitance were compensated. Single SAN cells were depolarized from −55 to −40 mV for 50 ms to inactivate Na⁺ current. Then we evoked Ca current by applying a family of depolarizing voltage steps from −55 to +75 mV in 10-mV increments (150 ms). I_Ca,L was obtained as the difference current before and after application of 20 µM nifedipine (N7634; Sigma-Aldrich).

Electrical Stimulation of the SAN/Atrial Tissue.

To stimulate the SAN/atrial tissue, we used two electrodes located close to, but not in contact with, the superior and inferior vena cava regions (Fig. S3C). These electrodes were connected to a MyoPacer Field Stimulator (IonOptix). A 20-s train of pulses at different frequencies was applied. We used bipolar stimulation with 40-V output and 0.2-ms pulse width.

Data Analysis and Statistics.

To analyze the burst-firing pattern of the NCX KO SAN, we defined pauses as any absence of activity between two transients (or APs) that was longer than twice the average cycle length (CL) (14) and/or three times the first quartile CL. CLs longer than 1 s were considered pauses without further consideration (37). To calculate the extent of spike adaptation, we compared the first and last cycle length (CL) of each burst (23). We reported the change as percent increase of CL.

To quantify the diastolic Ca increase during spontaneous bursts, we took the diastolic fluorescence at the end (F_end) and at the beginning (F₀) of the spontaneous burst to obtain the amplitude increase of the diastolic Ca [e.g., (F_end − F₀)/F₀]. Then we normalized it to the average amplitude of spontaneous Ca transients during the preceding recording [(F_peak − F₀)/F₀]. We used the same strategy to quantify the diastolic Ca increase during rapid electrical stimulation.

All data were expressed as mean ± SEM, unless otherwise indicated. Student’s t test (paired or unpaired) and two-way ANOVA with Bonferroni’s multicomparison test were used for statistical analysis where appropriate, as indicated in each experiment. Prism 6 (GraphPad Software, Inc.) was used for all statistical analysis. *P < 0.05 was considered statistically significant; NS stands for nonsignificant.

Reagents.

Unless otherwise specified, all chemicals and reagents indicated in the text are from Sigma-Aldrich or Fisher. Ivabradine (IVA, SML0281-10MG), isoproterenol (ISO, I5627-5G), and BayK8644 (BayK; B133-1MG) were purchased from Sigma-Aldrich whereas BAPTA AM (B-6769) was purchased from Invitrogen.