STIM1–Ca²⁺ signaling modulates automaticity of the mouse sinoatrial node

Significance

Each heartbeat originates in the sinoatrial node (SAN), a collection of specialized cardiomyocytes (SANCs), which exhibit rhythmic action potentials and spontaneous Ca²⁺ transients. We have found that the Ca²⁺ sensor protein stromal interaction molecule 1 (STIM1) is enriched in the SANCs. Here we show that STIM1 Ca²⁺ signaling is important in SANCs to maintain the Ca²⁺ content of intracellular Ca²⁺ stores and that this contributes to maintaining the regular sinus rhythm of the heart.

Abstract

Cardiac pacemaking is governed by specialized cardiomyocytes located in the sinoatrial node (SAN). SAN cells (SANCs) integrate voltage-gated currents from channels on the membrane surface (membrane clock) with rhythmic Ca²⁺ release from internal Ca²⁺ stores (Ca²⁺ clock) to adjust heart rate to meet hemodynamic demand. Here, we report that stromal interaction molecule 1 (STIM1) and Orai1 channels, key components of store-operated Ca²⁺ entry, are selectively expressed in SANCs. Cardiac-specific deletion of STIM1 in mice resulted in depletion of sarcoplasmic reticulum (SR) Ca²⁺ stores of SANCs and led to SAN dysfunction, as was evident by a reduction in heart rate, sinus arrest, and an exaggerated autonomic response to cholinergic signaling. Moreover, STIM1 influenced SAN function by regulating ionic fluxes in SANCs, including activation of a store-operated Ca²⁺ current, a reduction in L-type Ca²⁺ current, and enhancing the activities of Na⁺/Ca²⁺ exchanger. In conclusion, these studies reveal that STIM1 is a multifunctional regulator of Ca²⁺ dynamics in SANCs that links SR Ca²⁺ store content with electrical events occurring in the plasma membrane, thereby contributing to automaticity of the SAN.

Sinus rhythm of the heart is set by specialized cardiomyocytes located in the sinoatrial node (SAN). These cardiomyocytes (SANCs) lack a resting membrane potential but generate a sinus impulse after spontaneous diastolic depolarization triggers an action potential (AP). Automaticity is achieved in the SANCs by the simultaneous activation of diastolic currents during membrane depolarization and the spontaneous release of Ca²⁺ from internal stores (1–3). Several recent studies show that maintenance of sarcoplasmic reticulum (SR) Ca²⁺ stores is critically important for SAN automaticity, as is evident from genetic studies involving patients and mice that have leaky SR Ca²⁺ stores (4). Mutations in the ryanodine receptor (RYR2) or calsequestrin genes that result in spontaneous Ca²⁺ release cause catecholamiergic polymorphic ventricular tachycardia (CPVT) (5, 6). In addition to ventricular arrhythmias, these patients also develop sinus node dysfunction and bradycardia, which frequently requires permanent pacemaker insertion. Computational studies further reinforce the idea that SAN dysfunction results from leaky RYR2-containing Ca²⁺ stores (5). These studies emphasize the importance of Ca²⁺ signaling in the automaticity of SANCs. Given the emerging role of store-operated Ca²⁺ entry (SOCE) in excitable cells, we asked here whether stromal interaction molecule 1 (STIM1) plays a major role in regulating the Ca²⁺ signaling and automaticity of SANCs.

SANs are structurally and functionally heterogeneous, exhibiting differences in shape and size that correspond to differences in electrophysiological features (7). It is believed that this heterogeneity is required to establish regional zones within the SAN for impulse generation by pacemakers. Under resting conditions, clusters of SANCs serve as the dominant pacemaker by firing APs at rates faster than subsidiary pacemakers located in adjacent aspects of the SAN and atria. However, under different chronotropic conditions, as occurs with autonomic stimulation, the dominant pacemaking sites can shift to different regions within the SAN, where subsidiary pacemakers can slow or speed up the heart rate (HR) (8). Dominant pacemakers are established by differences in ion channel distribution that create regional differences in electrophysiological properties such as the maximal diastolic potential (MDP) and the rate of diastolic depolarization. In addition, local intracellular factors also contribute to pacemaking by integrating currents generated at the plasma membrane with the rhythmic release of Ca²⁺ from the SR. Given the importance of Ca²⁺ dynamics to diastolic depolarization of the SAN, we hypothesize that multiple mechanisms must be available to replenish internal Ca²⁺ stores. Refilling internal Ca²⁺ stores in SANCs is known to involve Ca²⁺ entry via voltage-gated Ca²⁺ channels (9). Here we propose that Ca²⁺ store refilling also requires SOCE.

In nonexcitable cells such as lymphocytes, the molecular mechanisms underlying SOCE have been well characterized and shown to require STIM1, a single-pass endoplasmic reticulum (ER) membrane protein that serves as the sensor of SR/ER Ca²⁺ store content. When Ca²⁺ stores are depleted, STIM1 molecules in the SR/ER membrane interact with plasma membrane Ca²⁺ channels, such as Orai1, to initiate Ca²⁺ entry into the cytoplasm and consequent refilling of internal Ca²⁺ stores via SR/ER Ca²⁺ pumps. SOCE is therefore an attractive candidate to link the Ca²⁺ and membrane clocks that underlie SANC automaticity. Roles for STIM1-dependent SOCE have been suggested in the heart (10–14). Here, we show that STIM1 is selectively expressed in the SAN where it activates Ca²⁺ entry via Orai1 channels and thereby modulates the Ca²⁺ signals required for pacemaking activity of the SAN. These findings build on an emerging theme, that STIM1 is a multifunctional signaling molecule, to show that STIM1 regulates several aspects of Ca²⁺ signaling in SANCs.

Results

STIM1 Expression in the Mouse SAN.

A first clue that STIM1 might regulate SAN function came from analysis of the expression pattern of the STIM1–LacZ fusion protein in the hearts of STIM1–β-galactosidase reporter mice (STIM1^+/Lacz) (15). Surprisingly, very little STIM1 expression was detected in atrial or ventricular chambers. However, we did detect STIM1 in the SAN and coronary arteries of adult mouse heart (Fig. 1 A–C). STIM1–LacZ staining in whole mount hearts revealed STIM1 localization at the junctional region of the superior vena cava and the right atrial appendage, where the SAN is located (Fig. 1 B and C). Serial sections of the heart from STIM1^+/LacZ mice demonstrated that the bulk of STIM1–LacZ staining extended from the edge of the crista terminalis to the interatrial septum, in a pattern consistent with the SAN area of the mouse heart (Fig. 1D and Fig. S1 A and B). This was further confirmed by detecting HCN4, a marker of the SAN, in the same region in which STIM1 was expressed (Fig. S1 C–F). Given that STIM1 activates Orai1 channels, we used Orai1 specific antibodies to show that Orai1 is also expressed in the SAN, in a pattern similar to HCN4 (Fig. 1 E–G). Because STIM1 was enriched in the SAN, we asked whether mice lacking STIM1 exhibited a phenotype attributable to changes in cardiac pacemaking. Indeed, we found that 6-wk-old unanesthetized gene-trapped STIM1^−/− mice exhibited slower HRs than their WT littermates, as measured by electrocardiography [WT HR 688 ± 14 beats per minute (bpm) versus KO HR 621 ± 21 bpm; P < 0.05, n = 4 mice each genotype].

Fig. 1.

STIM1 is expressed in the mouse SAN. (A) Whole mount photographs prepared from hearts of adult STIM1^+/LacZ mice (4 mo old) with STIM-LacZ staining. (B) An enlarged view of STIM1-LacZ staining in adult heart to show the highest level of LacZ staining in the SAN. (C) The SAN was dissected out from surrounding atrial tissues and was stained with LacZ. (D) Paraffin section of whole mount of the SAN from C. LacZ staining (blue) extended from the edge of cristae terminalis to interatrial septum. Section was counterstained with Eosin. (E) Orai1 specific antibodies detected Orai1 expression (green) in sections prepared from STIM1^fl/fl SAN preparations. (F) The SAN area is defined by colabeling for HCN4 (red). (G) Colocalization of Orai1(green) with HCN4 (red) in the SAN. (Scale bar, 200 µm.)

Fig. S1.

STIM1 expression in the SAN. (A and B) Micrographs of sections taken from STIM1 gene trap mice. LacZ histochemistry was performed to define STIM1 expression. LacZ was concentrated in the superior vena cava (SVC) near the interatrial septum (IAS) and right atrium (RA). (C–F) Cardiac-specific deletion of STIM1 was achieved by crossing exon 2 floxed mice with MHC–Cre transgenic mice. STIM1 antibodies detected STIM1 in the STIM1^fl/fl hearts (D) but not in the cSTIM1^−/− heart (C). STIM1 staining was concentrated in the SAN, as indicated by the staining of HCN4 in parallel sections of cSTIM1^−/− (E) and STIM1^fl/fl hearts (F). (G and H) STIM1 deletion from the SAN was confirmed by Western blotting. SAN isolated from WT and cSTIM1^−/− hearts were immunoblotted with antibodies. HCN4 identified SAN purity. Cre confirms the presence of the recombinase in SAN. We did not detect changes in the expression of HCN4 or Orai1 channels.

STIM1 Regulates a Store-Operated Ca²⁺ Current in SANCs.

Because constitutive deletion of STIM1 in mice leads to a severe skeletal muscle phenotype and early lethality (15), we used site specific recombination Cre/lox technology to delete STIM1 specifically in the heart. Cardiac-specific STIM1^−/− mice (cSTIM1^−/−) were generated by crossing mice carrying the α-MHC–Cre transgene with a mouse line in which exon 2 of the Stim1 gene was flanked by lox P sites (16). The cSTIM1^−/− mice appeared to develop normally and exhibited no obvious differences from control STIM1^fl/fl mice. The cSTIM1^−/− SAN lacked STIM1, as shown by immunostaining (Fig. S1 C–F). Moreover, STIM1 and Cre immunoblotting from heart lysates revealed a threefold reduction in STIM1 expression, indicating that STIM1 expression in non-Cre-expressing cells remained intact (Fig. S1 G and H). We did not detect a difference in the expression of HCN4 or Orai1 in the cSTIM1^−/− SAN (Fig. S1G).

To determine whether STIM1 in SANCs activates a store-operated Ca²⁺ current (I_soc) and SOCE, we isolated SANCs from the hearts of STIM1^fl/fl and cSTIM1^−/− mice and used the whole-cell patch clamp recording technique and protocols previously established for measuring I_soc in nonexcitable cells (Fig. 2 A–C). SANCs were held at a membrane potential of 0 mV and subjected to a standard ramp protocol from +100 mV to −100 mV. Inclusion of BAPTA (10 mM) in the patch pipette to indirectly deplete SR Ca²⁺ stores activated an inwardly rectifying Ca²⁺ current with a positive reversal potential (V_rev = 48 ± 6mV, n = 8) in STIM1^fl/fl SANCs (Fig. 2C). As in nonexcitable cells, this inwardly rectifying current was inhibited by Gd³⁺ (20 μM) or by preincubation with BTP-2 (500 nM). This current was completely absent in the cSTIM1^−/− SANCs. Together, these results indicate that the inward Ca²⁺ current displays the defining characteristics of Ca²⁺ release-activated current (I_CRAC) found in nonexcitable cells such as lymphocytes, and that it required STIM1 for activation (Fig. 2B). Consistent with these electrophysiological results, Fura-2 Ca²⁺ imaging showed that SOCE was present in STIM1^fl/fl SANCs but nearly absent in cSTIM1^−/− SANCs (Fig. 2 D–F). We also found that removal of external Ca²⁺ from the bath led to a rapid decrease in cytosolic Ca²⁺ levels, which is consistent with a rapid Ca²⁺ extrusion mechanism. The physiological importance of STIM1 in SANCs was further evidenced by the dramatic decrease in Ca²⁺ release from internal SR Ca²⁺ stores that was triggered by 10 mM caffeine and CPA in cSTIM1^−/− SANCs compared with control STIM1^fl/fl SANCs, indicating a significant decrease in SR Ca²⁺ store content (Fig. 2E).

Fig. 2.

STIM1 regulates I_soc and SOCE in mouse SANCs. (A) A representative time course for activation of I_soc by BAPTA (10 mM) from SANCs. I_soc was defined as the Gd³⁺-sensitive current at −80 mV and was plotted from STIM1^fl/fl SANCs in the absence (black) and presence (blue) of BTP-2 (500 nM) and from cSTIM1^−/− SANCs (red). (B) Average I_soc amplitudes from STIM1^fl/fl (black, n = 8), cSTIM1^−/− (red, n = 9), and BTP-2 treated STIM1^fl/fl SANCs (blue, n = 5). (C) Current–voltage (I-V) relationship of I_soc indicated inward rectification and a positive reversal potential. (D) Ca²⁺ transients and SOCE from SANCs using Fura-2 Ca²⁺ imaging. The perfusate was changed from a 2-mM Ca²⁺ to a 0-mM Ca²⁺ solution (beginning at arrows), and SR Ca²⁺ stores were depleted by the addition of caffeine (10 mM) plus CPA (30 μM). SOCE was then determined by perfusion with 10 mM Ca²⁺ in the continued presence of CPA. (E and F) Quantification of the SR Ca²⁺ release elicited by caffeine plus CPA (E) and SOCE (F) for SANCs from STIM1^fl/fl (n = 5 cells from three mice) and cSTIM1^−/− mice (n = 11 cells from three mice). Data are represented as mean ± SEM; ***P < 0.001, **P < 0.01; Student’s t test for unpaired variances.

STIM1 Resides in the SR of SANCs.

To understand if STIM1 might activate SOCE in SANCs through a mechanism similar to that in lymphocytes and other nonexcitable cells, we used immunofluorescence to analyze the molecular components of the SOCE machinery under store-replete and store-depleted conditions. HCN4 was used to identify SANCs and outline the surface membrane. Profile analysis of fluorescence intensity (Fig. 3A, Right) indicated that Orai1 colocalized with HCN4 channels, suggesting that Orai1 was expressed in the plasma membrane of SANCs. STIM1 was localized in a transverse pattern typical of the Z line and also in a linear pattern underneath the sarcolemma (Fig. 3B). This staining pattern strongly resembles that described for RYR2 in SANCs, which defines the SR compartment (17, 18). Profile analysis validated that, in resting SANCs, STIM1 was present both in the subsarcolemmal compartment and in the Z-line compartment (Fig. 3B, Right). Interestingly, SR Ca²⁺ store depletion using the SR/ER Ca²⁺ ATPase (SERCA) inhibitor cyclopiazonic acid (CPA, 30 μM) did not alter this distribution of STIM1 in SANCs (n = 9, Fig. 3 C and D). Based on these observations, we conclude that STIM1 is positioned in the SR membrane at the Z-line and also in the subsarcolemmal SR, where it can interact with Orai1 channels to rapidly activate SOCE. These findings further suggest that STIM1 movement is limited in SANCs. Expression of STIM1 underneath the plasma membrane is likely to provide ready access to Orai1 channels to activate Ca²⁺ entry.

Fig. 3.

Immunostaining of STIM1 and Orai1 in SANCs. (A) (Left) Immunostaining for Orai (green) and HCN4 (red) in SANCs. Profile analysis (Right) demonstrates colocalization on the plasma membrane. (B) (Left) Immunostaining for STIM1 (green) and HCN4 (red) in SANCs. STIM1 localization is consistent with localization in the SR, as indicated by profile analysis (Right, average from nine SANCs). (C) (Left) STIM1 (green) and HCN4 (red) expression in SANCs in response to 30-µm CPA treatment. Profile analysis (Right) is average from nine CPA-treated SANCs. (D) Relative staining of STIM1 in peripheral verses central regions of SANCs with and without CPA treatment. Data were averaged from nine SANCs for each condition. Red dashed line (C, Right) represents areas used to calculate peripheral and central expression of STIM1. White boxes across SANCs show examples of regions from which pixel counts were taken. (Scale bar, 10 µm.)

To broaden our understanding of the limited movement of STIM1 within the SR, we examined STIM1 in other excitable cells. We chose neonatal primary cardiomyocytes (NCMs) because they beat spontaneously and exhibit many of the electrophysiological features typical of SANCs (19). NCMs were isolated from newborn STIM1^fl/fl and cSTIM1^−/− mouse hearts. Like cSTIM1^−/− SANCs, NCMs from cSTIM1^−/− hearts also showed no SOCE (Fig. S2A). We reintroduced STIM1 into the cSTIM1^−/− NCMs by infecting them with an adenovirus that carries rat STIM1–cherry red and found that this restored both SOCE and SR Ca²⁺ refilling in cSTIM1^−/− NCMs (Fig. S2B). Again, the pattern of STIM1–cherry red fluorescence in the NCMs exhibited minimal change on a beat-to-beat basis (Movie S1) or following CPA-mediated Ca²⁺ store depletion (Fig. S2C). Based on these findings, we conclude that Ca²⁺ store depletion in NCMs does not significantly alter STIM1 positioning in the SR. These findings support the notion that STIM1 movement is limited in striated muscle where dense sarcomeric material may restrict STIM1 diffusion in the SR.

Fig. S2.

STIM1 directly influences SOCE and SR Ca²⁺ store content in NCMs. (A) NCMs were isolated from neonatal (P2) STIM1^fl/fl and cSTIM1^−/− mouse hearts and cultured at 37 °C. SOCE was measured with Fura-2 Ca²⁺ imaging using a standard SOCE protocol. Adenoviral delivery of STIM1–cherry red, but not of LacZ, to cSTIM1^−/− NCMs rescued both SOCE and SR Ca²⁺ content. (B) (Left) To assess the Ca²⁺ stores of NCMs, whole-cell NCX1 current recordings were made from STIM1^fl/fl, cSTIM1^−/−, and STIM1-cherry infected cSTIM1^−/− NCMs after treatment with caffeine and CPA. Peak NCX1 currents (Right) indicate a significant reduction in NCX1 currents from cSTIM1^−/− NCMs (red). Reintroduction of STIM1-cherry to the cSTIM1^−/− NCMs (blue) partially rescued the SR Ca²⁺ stores. (C) Confocal images obtained from a STIM1-cherry infected cSTIM1^−/− NCMs before and after caffeine/CPA induced store depletion. Data are mean ± SEM.

Loss of STIM1 in SANCs Markedly Reduced Spontaneous Ca²⁺ Transients and SR Ca²⁺ Store Content.

Having established that STIM1-dependent SOCE is operational in SANCs, we questioned whether deletion of STIM1 from SANCs would alter the critical Ca²⁺ transients that occur simultaneously with the spontaneous APs. High-speed confocal Ca²⁺ imaging of SAN preparations revealed a significant reduction in the amplitude of spontaneous Ca²⁺ transients in the cSTIM1^−/− SAN compared with the STIM1^fl/fl SAN (Fig. 4 A and B). This could be caused by decreased SR Ca²⁺ content and thus decreased Ca²⁺ release. To determine whether the decrease in Ca²⁺ transient amplitude might be due to changes in the expression of Ca²⁺-handling proteins, we carried out Western blot analysis of lysates prepared from SAN tissue from STIM1^fl/fl and cSTIM1^−/− hearts. The analysis revealed no significant changes in the expression of NCX1, Cav1.2, SERCA2a, and HCN4 (Fig. 4 C and D).

Fig. 4.

Loss of STIM1 reduced spontaneous Ca²⁺ transients, but did not alter Ca²⁺-handling protein expression in cSTIM1^−/− SAN. (A) Representative traces of spontaneous Ca²⁺ transients recorded from STIM1^fl/fl (black) and cSTIM1^−/− (red) SAN preparations. (B) Quantification of Ca²⁺ transient amplitudes from STIM1^fl/fl (WT, n = 7) and cSTIM1^−/− (KO, n = 8) SAN. (C) Amplitude of spontaneous Ca²⁺ transients before and after isoproterenol stimulation. (D) Expression of HCN4 and several key Ca²⁺-handling proteins in STIM1^fl/fl and cSTIM1^−/− SAN. (E) No significant differences were detected between genotypes. Data are represented as mean ± SEM; ***P < 0.001 **P < 0.01, *P < 0.05; Student’s t test for unpaired variances.

What are the consequences of reduced Ca²⁺ store refilling in cSTIM1^−/− SANCs? It has been proposed that activation of inward current (I_NCX1) through the sarcolemmal Na⁺/Ca²⁺ exchanger during removal of cytosolic Ca²⁺ released from internal stores is a critical link coupling the Ca²⁺ clock (release of Ca²⁺ from the SR) and the membrane clock (inward depolarizing current through NCX1) in SANCs. Here we hypothesized that STIM1-dependent Ca²⁺ signaling is also a component of the link between these two clock processes. To test this idea, we used 20 mM caffeine to release Ca²⁺ from the SR in the presence of 30 µM CPA to block the SERCA2a pump, which activates I_soc (Fig. 2 D–F and Fig. S2B). We then measured I_NCX1, whose identity was confirmed by block with 5 mM Ni²⁺. The integrated I_NCX1 current, which represents the SR store Ca²⁺ content and SOCE, was markedly reduced in cSTIM1^−/− SANCs compared with STIM1^fl/fl SANCs (Fig. 5 A–C), consistent with our Ca²⁺ imaging data showing reduced Ca²⁺ release from SR Ca²⁺ stores in cSTIM1^−/− SANCs. Further, in STIM1^fl/fl SANCs, we observed a persistent I_NCX1 following SR store depletion (Fig. 5D). We hypothesized that this persistent NCX1 current was triggered by STIM1-dependent Ca²⁺ entry through Orai1 channels. Supporting our hypothesis, the persistent NCX1 current was absent in cSTIM1^−/− SANCs (Fig. 5D). We confirmed that loss of STIM1 in SANCs is responsible for depleting SR Ca²⁺ stores, because reintroduction of STIM1 into the cSTIM1^−/− NCMs restored the ability to replenish the SR Ca²⁺ stores (Fig. S2B). These studies indicate that there is an intimate relationship between STIM1, I_NCX1, and I_soc. We next assessed the steady-state I_NCX1 density and found that I_NCX1 was increased in cSTIM1^−/− SANCs (Fig. 5 E and F). We also found that there was a change in the reversal potential for the I_NCX1 in cSTIM1^−/− SANCs.

Fig. 5.

SR Ca²⁺ stores and NCX1 current in cSTIM1^−/− SANCs. (A) NCX1 current was measured as a function of SR Ca²⁺ load in STIM1^fl/fl (Top, black trace) andcSTIM1^−/− (Middle, red trace) SANCs. SR refilling was inhibited by CPA (30 µM) and SR Ca²⁺ store was depleted with caffeine (20 mM). Current block by Ni²⁺ (5 mM) was used to confirm the identity of the NCX1 current (Bottom, blue trace). Dashed lines indicate the basal level of NCX1 current before store depletion. (B–D) Quantitative analysis of NCX1 currents following SR Ca²⁺ store depletion with CPA plus caffeine as in A. (B) Peak NCX1 current was significantly reduced in cSTIM1^−/− SANCs. (C) Quantification of SR Ca²⁺ store content and SOCE-mediated Ca²⁺ entry measured by integrating NCX1 currents (30 s) activated by Ca²⁺ store depletion with CPA plus caffeine as in A. NCX1 currents were significantly reduced in cSTIM1^−/− SANCs. (D) The persistent NCX1 current activated by I_soc with Ca²⁺ store depletion was also markedly reduced in the cSTIM1^−/− SANCs, consistent with a loss of I_soc. (E) Average I-V relationships of steady state NCX1 currents from the STIM1^fl/fl (n = 8) and cSTIM1^−/− (n = 9) SANCs. (F) The total NCX1 current density at −60 mV was greater in cSTIM1^−/− SANCs (n = 9) than in STIM1^fl/fl SANCs (n = 8). Data are represented as mean ± SEM; ***P < 0.001 and **P < 0.01; Student’s t test for unequal variances.

Recent studies demonstrate that STIM1 regulates voltage-gated Ca²⁺ currents via a direct physical interaction with Cav1.2 (20, 21). In contrast to those works where silencing STIM1 led to greater L-type Ca²⁺ currents (I_Ca-L), we found that the current density of I_Ca-L was reduced in cSTIM1^−/− SANCs compared with STIM1^fl/fl SANCs (Fig. S3A). Interestingly, viral replacement of rat STIM1 in cSTIM1^−/− SANCs was able to rescue I_Ca-L (Fig. S4). Therefore, the reduction in I_Ca-L may be due to the depletion of SR Ca²⁺ stores in SANCs from cSTIM1^−/− mice, as has been described for SANCs harboring mutations in mouse RyR2 that cause CPVT (5, 22, 23). We also extended our investigation to include T-type Ca²⁺ currents (I_Ca-T). Our results indicated that there was no significant difference in the T-type Ca²⁺ current densities measured in cSTIM1^−/− and STIM1^fl/fl SANCs (Fig. S3B). Taken together, these studies demonstrate that STIM1 signaling regulates three membrane currents, I_soc, I_NCX1, and I_Ca-L, either directly or indirectly, and thereby plays a critical role in controlling cytosolic Ca²⁺ homeostasis and the refilling of SR Ca²⁺ stores in SANCs. These findings are likely to underlie the regulation of SAN pacemaking by STIM1-dependent Ca signaling.

Fig. S3.

Loss of STIM1 in SANCs did not change T-type Ca²⁺ currents but significantly reduced L-type Ca²⁺ currents. (A) (Top) Representative L-type Ca²⁺ currents (I_Ca-L) recorded from STIM1^fl/fl and cSTIM1^−/− SANCs. The currents were activated by 500-ms depolarizing steps from a holding potential of −50 mV to +60 mV, with an interval of 10 mV. The I_Ca-L was defined as 5-µM nifedipine-sensitive currents. I-V curves (Bottom) of I_Ca-L were shown from STIM1^fl/fl (black, n = 11) and cSTIM1^−/− (red, n = 9) SANCs. I_Ca-L was significantly smaller in In cSTIM1^−/− SANCs than in STIM1^fl/fl SANCs at −20 mV, −10 mV, 0 mV, +10 mV, and +20 mV. (B) (Top) Representative T-type Ca²⁺ currents recorded from STIM1^fl/fl and cSTIM1^−/− SANCs. SANCs were held at −40 mV, and T-type Ca²⁺ currents were activated by a 100-s depolarizing step (Inset) from −80 mV to −40 mV. (Bottom) The summarized data of I_Ca-T averaged from STIM1^fl/fl (n = 9) and cSTIM1^−/− (n = 9) SANCs. No significant difference was found between the two groups. Data are mean ± SEM. Significance was indicated by asterisk (*P < 0.05 and **P < 0.01).

Fig. S4.

(A) Rescue of I_Ca-L currents in cSTIM1^−/− SANCs by reintroduction of rat STIM1-cherry. Cultured SANCs were held at −60 mV, and I_Ca-L currents were activated by a 250-ms depolarizing step from −60 mV to 0 mV. Representative current recordings are from cultured cSTIM1^−/− (purple) and STIM1^fl/fl (blue) SANCs infected with β-gal, and from cSTIM1^−/− SANCs infected with STIM1-cherry adenovirus (rSTIM1; red). The currents were completely blocked by nifedipine (5 µM). (B) I-V relationship of I_Ca,L in SANCs of STIM1^fl/fl (filled blue circles, n = 7), cSTIM1^−/− (open blue circles, n = 8), and cSTIM1^−/− SANCs with rSTIM1 (filled red circles, n = 4) (*P < 0.05 vs. WT; ⁺ P < 0.05 vs. +rSTIM1). (C) Averaged values of peak I_Ca-L in SANCs of STIM1^fl/fl (blue, n = 7), cSTIM1^−/− (open blue bar, n = 8), and cSTIM1^−/− SANCs with STIM1 (filled red, n = 4).

SAN Function in cSTIM1^−/− Mice.

Having established that STIM1 regulates Ca²⁺ signaling and the electrophysiological properties of SANCs, we next examined the physiological consequences of STIM1 deletion for SAN function in cSTIM1^−/− mice. We recorded surface electrocardiograms (ECGs) from 16-wk-old unanesthetized mice and found a significant reduction in the HR of cSTIM1^−/− mice compared with STIM1^fl/fl mice (673 ± 11 bpm for STIM1^fl/fl (n = 16), versus 610 ± 23 bpm for cSTIM1^−/− mice (n = 22); P < 0.01) (Fig. 6A). To investigate potential causes for the reduced HRs, we performed detailed invasive electrophysiological testing with ECG analysis in anesthetized mice (Table 1). Here, we also observed a relatively slower HR in cSTIM1^−/− mice but no differences in conduction intervals between groups, thereby excluding defects in atrioventricular (AV) conduction. These data imply that the relative bradycardia in the cSTIM1^−/− mice derives from a SAN defect rather than a conduction system defect (Fig. 6B).

Fig. 6.

The cSTIM1^−/− mice exhibit SAN dysfunction during cardiac electrophysiological testing. (A) HRs were determined from conscious STIM1^fl/fl (n = 16) and cSTIM1^−/− (n = 22) mice. (B) Surface ECG was measured from anesthetized cSTIM1^−/− and STIM1^fl/fl mice. (C) Optical mapping of Ca²⁺ signaling from STIM1^fl/fl SAN (Top) and cSTIM1^−/− SAN (Bottom). In these images, the blue color represents the earliest point of activation of the membrane potential, as indicated by the asterisks. Red traces to the right of images indicate the time course of the Ca²⁺ transients at the earliest activation point. This analysis reveals the variability in both the firing rate and the origin of impulse in cSTIM1^−/− SAN. Data are represented as mean ± SEM; *P < 0.05; Student’s t test.

Table 1.

ECG variables change with carbachol in the STIM1^fl/fl and cSTIM1^−/−mice

	STIM1fl/fl (n = 8)		Knockout (n = 8)
Parameter	Baseline	Carbachol	Baseline	Carbachol
PR, ms	38.8 ± 0.6	37.4 ± 1.0	35.2 ± 2.4	30.3 ± 4.6
P duration, ms	15.1 ± 1.4	9.4 ± 1.2	12.6 ± 2.3	11.4 ± 2.0
QRS, ms	10.4 ± 0.8	11.6 ± 0.8	11.9 ± 0.6	15.6 ± 1.9*
QTc, ms	66.6 ± 4.8	79.6 ± 11.8	67.4 ± 5.6	65.6 ± 6.3
HR, bpm	548 ± 18	478 ± 38*	458 ± 34†	282 ± 30*,‡

^*P < 0.05 compared carbachol with baseline in each group.

^†P < 0.05 compared KO baseline with wild type baseline.

^‡P < 0.05 compared with wild type with carbachol.

To investigate the mechanistic basis for the SAN defect in the cSTIM1^−/− mice, we performed optical mapping of Ca²⁺ transients in isolated SAN preparations from STIM1^fl/fl and cSTIM1^−/− mice during spontaneous sinus rhythm. For STIM1^fl/fl SAN preparations (5/5 mice), spontaneous firing emerged from the edge of the crista terminalis near the superior vena cava (SVC) and propagated in a pattern typically described for the SAN (Fig. 6C, Top and Movie S2). In contrast, two distinct activation patterns with varied cycle lengths were noted for the cSTIM1^−/− SAN, indicating a shift in the primary pacemaking site. Specifically, at short cycle lengths, origination of the pacemaker occurred near the junction of the SVC and the crista terminalis. The impulses conducted rapidly, with no difference in conduction time between cSTIM1^−/− and STIM1^fl/fl SAN (Fig. 6C, Middle and Movie S3). However, at longer cycle lengths, the impulse initiation site shifted toward the atrial septum, resulting in the propagation of the impulse in the opposite direction (Fig. 6C, Bottom and Movie S4). Spontaneous pacemaker shifting was apparent in four of five cSTIM1^−/− SAN preparations, indicating that impulse generation is unstable in cSTIM1^−/− mice. Therefore, SANCs lacking STIM1 might be unable to efficiently refill stores, resulting in Ca²⁺ clock dysfunction and unstable impulse generation.

Cholinergic Sensitivity and STIM1–SOCE in Mouse SAN.

Parasympathetic tone is regulated by cholinergic signaling, which affects pacemaker function, enhances pacemaker shifting, slows the HR by altering the hyperpolarization-activated cation current (I_f) and the muscarinic receptor activated K⁺ current (I_K,Ach), and reduces Ca²⁺ transients (24–26). We next examined whether deletion of STIM1 altered the sensitivity of the SAN in response to cholinergic stimulation. A single dose of the cholinergic agonist, carbachol (CCh, 100 µg/kg) significantly slowed the HR and induced sinus arrest in hearts of cSTIM1^−/− mice, whereas this same CCh dose only minimally slowed the HR in STIM1^fl/fl mice (Fig. 7A and Table 1). Increased cholinergic sensitivity was also evident during invasive electrophysiological testing of cSTIM1^−/− mice (Table 2). When STIM1^fl/fl and cSTIM1^−/− mice were subjected to rapid atrial pacing for 30 s at a 100-ms cycle length, the time interval required for the SAN to recover after termination of rapid atrial pacing (corrected sinus node recovery time, cSNRT) was not different between cSTIM1^−/− mice and STIM1^fl/fl mice at baseline. After treatment with CCh, however, the cSNRT was significantly prolonged in cSTIM1^−/− mice compared with STIM1^fl/fl mice (Fig. 7B and Table 2). In contrast to the exaggerated cholinergic response, there was no difference in HR between cSTIM1^fl/fl and cSTIM1^−/− mice following treatment with isoproterenol (ISO), a β-adrenergic agonist (Fig. 7A, Top).

Fig. 7.

The cSTIM1^−/− mice exhibit altered cholinergic responsiveness. (A) Effects of i.p. injection of ISO (0.5 mg/kg, Top) or CCh (100 µg/kg, Bottom) on HR. ECGs were recorded from anesthetized mice. The i.p. injection of ISO increased the HR in cSTIM1^−/− mice by 68% (n = 5) and in STIM1^fl/fl mice by 48% (n = 10). CCh induced profound bradycardia in cSTIM1^−/− mice (n = 8) compared with STIM1^fl/fl (n = 8). (B) The cSTIM1^−/− mice exhibited prolonged SAN recovery time (SNRT) following rapid atrial pacing for 30 s. Surface ECGs showed the HR changes evoked by intracardiac pacing (indicated by *) and the P wave following termination of rapid pacing from STIM1^fl/fl and cSTIM1^−/− mice (Top). Quantification of the SNRT for STIM1^fl/fl mice (n = 5) and cSTIM1^−/− mice (n = 5) with and without CCh (100 µg/kg) (Bottom). (C) Dose–response curve of the effect of CCh on SAN firing rate in vitro revealed a significant leftward shift in the cSTIM1^−/− SAN (n = 5) and in the STIM1^fl/fl SAN treated with BTP-2 (10 μM). EC₅₀ was 0.32, 0.13, and 0.12 µM, respectively. (D) Effect of BTP-2 on CCh sensitivity of the STIM1^fl/fl and cSTIM1^−/− SAN firing rate. Data are represented as mean ± SEM ***P < 0.001, **P < 0.01 and *P < 0.05. Student’s t test.

Table 2.

Electrophysiological effects of Carbachol in the STIM^fl/fl and cSTIM1^−/−mice

	STIM1fl/fl (n = 5)		cSTIM1−/−mice (n = 5)
Parameter	Baseline	Carbachol	Baseline	Carbachol
AH, ms	27.0 ± 3.6	28.2 ± 3.9	31.4 ± 2.7	32.8 ± 7.9
Hd, ms	6.2 ± 1.6	7.6 ± 1.8	7.4 ± 0.5	7 ± 1.8
HV, ms	10.8 ± 2.0	11.6 ± 1.8	12.6 ± 1.3	12.2 ± 3.0
AVWCL, ms	79.6 ± 4.0	75.6 ± 7.9	81.6 ± 6.5	81.2 ± 9.3
AVNERP, ms	51.4 ± 8.1	54.8 ± 5.7	52 ± 7.2	53.5 ± 3.4
AERP, ms	34.8 ± 6.1	24.4 ± 4.3*	36 ± 5.8	20.4 ± 5.5*,†,‡
cSNRT, ms	34.4 ± 14.8	41 ± 7.0	37.4 ± 29.5	142.75 ± 24.3*,†,‡

AERP, atrial effective refractory period; AH, atrial-His conduction time; AVNERP, effective refractory period of the atrioventricular node; AVWCL, AV Wenchebach cycle length; cSNRT, corrected sinus node recovery time; Hd, His duration; HV, His-ventricular conduction time.

^*P < 0.05 compared with baseline in each group.

^†P < 0.05 compared with WT baseline.

^‡P < 0.05 compared with WT carbacol.

To gain additional insight into the enhanced cholinergic sensitivity, we recorded ECGs during the administration of the cholinergic receptor antagonist atropine (1 mg/kg) to anesthetized STIM1^fl/fl and cSTIM1^−/− mice. We observed a significant increase in the HR from 375 ± 16 bpm to 537 ± 26 bpm (n = 12, P < 0.005) in STIM1^fl/fl mice and from 310 ± 15 bpm to 510 ± 28 bpm (n = 9, P < 0.001) in cSTIM1^−/− mice. Application of β-adrenergic receptor antagonist propranolol (1 mg/kg) in the presence of atropine produced a comparable reduction in the HRs of STIM1^fl/fl and cSTIM1^−/− mice (Fig. S5). Together, these data demonstrate that vagal tone was enhanced whereas sympathetic tone was unaltered in mice lacking cardiac STIM1 expression.

Fig. S5.

HR changes in anesthetized STIM1^fl/fl (n = 12) and cSTIM1^−/− (n = 9) mice under basal conditions and following autonomic blockade. The i.p. injection of atropine (1 mg/kg) increased HR, which was partially reversed by the addition of propranolol (1 mg/kg) to slow down HR both in STIM1^fl/fl and cSTIM1^−/− mice. Asterisk indicates the significance (KO vs. WT, P < 0.05).

To determine if these changes in cholinergic sensitivity in cSTIM1^−/− mice were intrinsic to the SAN, we measured the effects of CCh on spontaneous pacemaking of SAN tissue in vitro, which lacks neural input, from STIM1^fl/fl and cSTIM1^−/− mice. The cSTIM1^−/− SAN tissue was more sensitive to CCh than the STIM1^fl/fl SAN, resulting in a leftward shift in dose–response curves for AP firing rate (Fig. 7C). To verify that this increase in cholinergic sensitivity resulted from loss of STIM1-dependent SOCE and not from abnormal development in the absence of STIM1, we analyzed the effects of CCh on STIM1^fl/fl SAN firing following acute inhibition of SOCE with BTP-2 (10 µM). Pharmacologic block of SOCE shifted the dose–response curve leftward, similar to the cSTIM1^−/− SAN (Fig. 7C). Moreover, in the presence of BTP-2, the STIM1^fl/fl SAN firing rate was reduced by 10%, and became more sensitive to CCh compared with untreated controls (Fig. 7D). When these studies are interpreted alongside recent work from others showing that Ca²⁺ signaling in the SANCs is subject to cholinergic control (25), we suggest that STIM1-dependent SOCE is required to refill internal Ca²⁺ stores and acts as a backup pacemaker to oppose the negative chronotropic effects of acetylcholine.

Computer Modeling of Action Potentials.

To provide additional support for our experimental observations on the role of STIM1-dependent SOCE in SAN function, we used mathematical modeling to analyze the effects of I_soc on APs of the SAN (Fig. 8A). It should be noted that there is no available mathematical model for SAN APs that includes I_soc, an inwardly rectifying current activated during rhythmic Ca²⁺ release in the SAN. We therefore introduced our experimentally acquired data of I_soc (conductance 0.05 nS/pF), as well as a background K⁺ conductance (I_bk) and inwardly rectifying K⁺ current (I_K1) into a model based on the rabbit SAN (Table 3). The introduction of I_soc to this mathematical model resulted in an ∼10% increase in the SAN firing rate (Fig. 8A). These findings are consistent with the decrease in SAN firing rate found with inhibition of I_soc with BTP-2 or STIM1 deletion from the SAN (Fig. 7 C and D). We also examined the output of this model by varying the Ca²⁺ pump (SERCA2a) rate to mimic SR Ca²⁺ store depletion. Changing I_soc as a function of Ca²⁺ pump activity influenced both the MDP of APs and the SAN firing rate (Fig. 8 B and C), suggesting that I_soc is involved in setting the MDP and the initiation of diastolic depolarization.

Fig. 8.

Computational modeling of SAN APs in the presence or absence of I_soc. (A) Computation of I_soc during AP (Top) and SAN AP dynamics in the presence and absence of I_soc (Bottom). SERCA2a pump was set at maximal rate of 0.04 mM/ms. (B) The direct interplay between maximal conductance of I_soc (g_soc, nS/pF) and cycle length (CL, ms) is shown at different rates of SERCA2a Ca²⁺ uptake (Pup). SERCA2a pumping was set at 0.04, 0.012 and 0.006 mM/ms. (C) AP changes at different SERCA pumping rates. MDP became more negative when g_soc was removed at the reduced SERCA2a pump rates.

Table 3.

Model parameters/variables

Model parameters/variables	Definition	Values/initial values
$g_{SOC}$	conductance of Isoc	0.05 nS/pF
$V_m$*	transmembrane potential	−65 mV*
$E_{SOC}$	reversal potential for Isoc	45 mV
$x$*	activation gating variables for Isoc	0*
$C_m$	cell membrane capacitance	32 pF
$x_{\infty }$*	steady level of the gating variable	—
${\tau }_{SOC}$	time constant of Isoc	1000 ms
$K_m$	half maximal activation of Isoc	0.02 mM
$C{a}_{jSR}$*	junctional SR Ca2+ contents at given times	0.029 mM*
$g_{K1}$	conductance of IK1	0.0324 nS/pF
$g_{bK}$	conductance of background K+ current	0.001 nS/pF
${\left[K^{+}\right]}_o$	extracellular K+ concentration	5 mM
$E_K$	reversal potential for IK1 and IbK	−89 mV

^*These parameters were altered to mimic experimental conditions.

Discussion

For the first time, to our knowledge, we show a role for STIM1 and SOCE in the regulation of SAN pacemaking. Cardiac specific deletion of STIM1 or pharmacologic inhibition of Orai1 channels in mice caused a reduction in the SR Ca²⁺ stores, which manifests as a relative bradycardia, enhanced cholinergic sensitivity, and pacemaker shifting. These studies therefore establish the importance of STIM1 as a regulator of Ca²⁺ dynamics and electrophysiological function in SANCs and, thus, of cardiac pacemaking of the SAN. To provide mechanistic insight into the reduction in SR Ca²⁺ store content in cSTIM1^−/− SANCs, we show that STIM1 is involved in the regulation of Ca²⁺ entry pathways. Notably, we provide the first evidence, to our knowledge, that I_CRAC is present in mouse SANCs. We also show that STIM1-dependent SOCE functionally interacts with NCX1 and that the augmented NCX1 activity in cSTIM1^−/− SANCs accelerates the HR to partially compensate for the relative bradycardia in these mice. Based on our modeling studies, we further propose that this SOCE–NCX1 interaction is likely to influence AP parameters, including diastolic depolarization rate and MDP. Taken together, our data support the notion that STIM1 is of fundamental importance to the Ca²⁺ dynamics of SANCs. Here, STIM1 and SOCE are activated by the rhythmic release of Ca²⁺ from the SR, which activates Orai1 to refill the SR stores. In this way, STIM1 ensures the fidelity of SAN Ca²⁺ dynamics and the integrity of the cardiac pacemaker.

In the present work, we tested the hypothesis that STIM1 contributed to SAN automaticity by regulating Ca²⁺ homeostasis in a manner similar to that seen in skeletal muscle (15) and lymphocytes (27). Ca²⁺ imaging studies have previously implicated SOCE in the SAN (28). Although SOCE was attributed to transient receptor potential channels (TRPC) channels and even Orai1 channels, no membrane current measurements were made to validate this hypothesis. Here, we used whole-cell patch clamp recording of single SANCs to characterize I_soc. Depletion of SR Ca²⁺ stores in SANCs activated I_soc that shared many features with the well-characterized Ca²⁺ release-activated Ca²⁺ current (CRAC) currents described in lymphocytes. I_soc in SANCs exhibited inward rectification of the I-V, a positive reversal potential, and pharmacologic inhibition by BTP-2 and Gd³⁺ (29, 30). Given that I_soc in SANCs is similar to I_CRAC and that Orai1 is expressed in these cells, we conclude that I_soc activated by STIM1 is mediated by Orai1 channels in SANCs.

Unlike in lymphocytes and other nonexcitable cells, SANCs have an extensive repertoire of ionic currents that orchestrate membrane excitability and the properties of the AP. As a result, we hypothesize that STIM1 and Orai1 interact with other membrane currents, as we propose for the interaction between I_soc and I_NCX1. An interesting observation in the present studies is that loss of STIM1 in cSTIM1^−/− SANCs chronically decreased L-type Ca²⁺ currents. Previous studies have shown that STIM1 physically interacted with Cav1.2 channels to suppress L-type Ca²⁺ currents in cell lines (20, 21). However, in SANCs, we have no evidence showing that the decrease in L-type Ca²⁺ currents results from a direct interaction between STIM1 and Cav1.2. Our studies also reveal enhanced vagal tone in cSTIM1^−/− mice, which would reduce the basal activity of L-type Ca²⁺ channels. Taken together, our findings support the idea that depletion of SR Ca²⁺ stores of cSTIM1^−/− SANCs induced a Ca²⁺-dependent decrease in L-type Ca²⁺ currents, much as in CPVT mice (31, 32).

In lymphocytes and other nonexcitable cells, STIM1 is subject to oligomerization and migration in the ER in response to Ca²⁺ store depletion, where STIM1 migrates from the perinuclear ER to the cortical ER beneath the plasma membrane (33). STIM1 then interacts with and activates Orai1 channels resident in the plasma membrane. In the present study, we show that STIM1 is positioned in the SR membrane at the Z-line structures and near the sarcolemma, but that it does not undergo movement within the SR in response to store depletion. This localization suggests that, in SANCs, STIM1 is located in the subsarcolemmal SR membranes where it can rapidly activate Ca²⁺ entry by Orai1 on a beat-to-beat basis. Meanwhile, the population of STIM1 located in the SR that attaches to the Z line may regulate SERCA activity on a beat-to-beat basis, as was described during skeletal muscle development (34) and recently was shown for cardiomyocytes by our group. In this way, STIM1 could target key elements of Ca²⁺ signaling in SANCs including Orai channels, SERCA, and NCX1. These findings are likely to underlie the regulation of SAN pacemaking by STIM1-dependent Ca²⁺ signaling. Although the mechanistic basis of this restricted mobility is not yet known, potential mechanisms may involve posttranslational modification of STIM1, temperature sensitivity, or binding of STIM1 to static structures such as intermediate filaments that are enriched excitable cells.

Based on our findings, we provide a potential mechanism by which STIM1 influences the rate of diastolic depolarization of SANCs. NCX1 is activated late in diastole by local Ca²⁺ release and produces an inward current that contributes to diastolic depolarization (35, 36). NCX1 is also activated during systole when Ca²⁺ transients peak in response to emptying of SR Ca²⁺ stores. Normally, with each AP, Ca²⁺ transients decay as Ca²⁺ recycles back to the SR stores via SERCA2a and is rapidly extruded across the plasma membrane via NCX1. In the current studies, we show that the steady-state NCX1 current was augmented in the cSTIM1^−/− SANCs despite no change in NCX1 protein levels. These data suggest that cytosolic Ca²⁺ was preferentially extruded by NCX1 in cSTIM1^−/− SANCs, owing in part to chronic depletion of the SR Ca²⁺ stores. In other words, a much larger portion of the Ca²⁺ transient is extruded by NCX1 in cSTIM1^−/− SANCs than in STIM1^fl/fl SANCs. Because it is well established that depletion of SR Ca²⁺ stores in SANCs slows the HR, the increase in I_NCX1 in cSTIM1^−/− SANCs is likely to accelerate the SAN firing rate and thereby partially compensate for the loss of I_soc. Thus, the compensatory increase in NCX1 activity may account for the relatively small change in the HR of cSTIM1^−/− mice.

Summary

In conclusion, our results demonstrate that STIM1 serves a fundamental role in SANCs, where it activates I_soc, refills SR Ca²⁺ stores, and regulates specific membrane currents (I_Ca-L and I_NCX1). STIM1 therefore influences cardiac pacemaking by linking the Ca²⁺ and the membrane clocks. Deleting STIM1 from the SAN causes depletion of SR Ca²⁺ store, slows the HR, increases responsiveness to cholinergic stimulation, and destabilizes primary pacemaker sites in the SAN. These findings are consistent with the SAN dysfunction that has been reported to occur in humans and mice with leaky Ca²⁺ stores. These studies thus provide the initial evidence for the physiologic function of STIM1 in the SAN. Because of the specific expression of STIM1 in the SAN and sparing of atrial and ventricular cardiomyocytes, targeting SOCE might be a therapeutic strategy to treat SAN dysfunction and atrial arrhythmias.

Materials and Methods

Animals.

STIM1–lacZ mice were genotyped as previously described (15). STIM1^fl/fl mice (C57BL/6) were generated as previously described (37). The α-MHC (Cardiac specific) Cre transgenic mice (α-MHC–Cre) (C57BL/6) were obtained from Jackson Laboratories. To generate cardiac-specific knockout of STIM1, αMHC–Cre transgenic mice were bred with founder STIM1^fl/fl mouse and the progeny STIM1^fl/-; α-MHC–Cre^+/− were then back-crossed with STIM1^fl/fl mice. All mice were maintained in pathogen-free barrier facilities at Duke University and were used in accordance with protocols approved by the Division of Laboratory Animal Resources and Institutional Animal Care & Use Committee at Duke University.

STIM1-LacZ Histochemistry, Immunoblotting, and Immunohistochemistry.

To analyze STIM1 expression using STIM1–LacZ gene trap mice (STIM1^+/LacZ), hearts were isolated from appropriately aged mice and then fixed in 2% paraformaldehyde with 0.2% glutaraldehyde for 20 min at room temperature. Hearts were then washed in rinse solution (5 mM EGTA, 0.01% deoxycholate, 0.02% Nonidet P-40, 2 mM MgCl₂) and stained in LacZ staining solution (5 mM K₃Fe(CN)₆, 5 mM K₄Fe(CN)₆, 5 mM EGTA, 0.01% deoxycholate, 0.02% Nonidet P-40, 2 mM MgCl₂, 1 mg⋅mL^–1 X-gal solution) overnight at room temperature. Paraffin sections were cut using standard methods and counterstained with Eosin. For antibody staining and immunoblotting, hearts were isolated and stored at −80 °C until sectioning or blotting. Polyclonal STIM-1 (Protein Tech group or sigma) and Orai1 (Alomone) antibodies were used at 1:250, and polyclonal HCN4 (Sigma) antibody was used at 1:500. After washing, staining with the secondary antibody, and counterstaining with DAPI, slides were mounted in Vectashield (Vector Laboratories).

SAN Preparation and Single SANC Isolation.

The heart was excised quickly from the anesthetized mouse and immersed into oxygenated Tyrode (TYR) solution containing (in millimolars) NaCl 140, KCl 5.4, MgCl₂ 1.05, NaH₂PO₄ 0.33, CaCl₂ 1.8, Hepes 5, Glucose 10, and pH 7.4. To isolate single SANCs, the SAN was dissected out from surrounding atrial tissues and cut into several small pieces. The SAN was then enzymatically digested in Ca²⁺-free TYR solution containing BSA 0.2%, Collagenase (Type II; Worthington) 0.25 mg/mL, and Elastase 0.2 mg/mL The digestion step was carried out in a shaking incubator at 37 °C. The single SAN cells were released in modified KB solution by gentle titration with a glass transfer pipette. KB solution includes KCl 85, K Glutamate 20, KH₂PO₄ 20, Taurine 20, EGTA 0.5, Glucose 20, Creatine 5, Succinic Acid 5, Pyruvic Acid 5, MgSO₄ 5, Hepes 5, and pH 7.2. The isolated SANCs were stored in KB solution at 4 °C for at least 2 h before experiments. Those SANCs having a spindle or spider-like shape, beating in TYR, and showing I_f currents were chosen for patch clamp experiments.

Electrophysiology.

For ionic current recordings from single SANCs, the cells were plated in a Warner perfusion chamber. Patch pipette was pulled from borosilicate glass capillaries (Sutter, Inc.) with a Sutter P-87 micropipette puller. The pipette usually had a resistance of 4–6 MΩ when filled with pipette solution. The liquid junction potential was nulled before gigaseal formation. Whole-cell currents were recorded using an Axon AXOPatch 200B amplifier connected to a Digidata 1322A digital converter. Whole-cell currents were filtered at 2 KHz and sampled at a rate of 10 kHz. The data were collected through Clampex software 9.0 (Mol Device) and stored on a computer for further analysis. All experiments were carried out at room temperature, except those specified.

Ca²⁺ Imaging.

To measure spontaneous Ca²⁺ transients, SAN tissue, dissected from the lateral right atrium, was loaded with 5 µM fluo-3 AM for 20–45 min at room temperature. The SAN preparation was pinned down to a recording chamber and washed with TYR solution for 20 min. Ca²⁺ transients were then imaged with a high-speed confocal microscope (Noran Oz) equipped with a 40× water immersion fluorescence objective (Olympus). The Ca²⁺ images were acquired at 30 frames per second, and data were analyzed with Metamorph imaging processing software (Molecular Devices). Ca²⁺ imaging experiments were done at 30 °C. To measure SOCE, SANCs were loaded with Fura-2 (1 μM) as previously described (15).

Optical Mapping of Ca²⁺ Transients in SA Node.

Ca²⁺ transients in the SAN regions were optically recorded during sinus rhythm. Briefly, isolated SAN tissue was placed in a temperature-controlled (37 °C) bath, stained by superfusion of a Ca²⁺-sensitive indicator, rhod-2 AM (Invitrogen; 5 µM in DMSO) for 20 min, and illuminated using a 520 ± 30 nm LED excitation light source (LEX2, Scimedia). The resulting Rhod-2 fluorescence signal (wavelength, ʎem = 580 ± 40 nm) was focused on a CMOS camera (Ultima, Scimedia; 100 × 100 pixels), and CaT movies during SAN activity were recorded with a high spatial (30 × 30 µm pixel size) and temporal (2-KHz frame rate) resolution. The CaT amplitude in each pixel was normalized between 0 and 1, and the activation time of CaT was defined as the moment of its maximum upstroke velocity [(dF/dt)max].

ECG and Intracardiac Electrophysiology Studies.

HRs from conscious mice were obtained from ECG data. Briefly, mice were first anesthetized with isoflurane, and ECG electrodes were then placed on the backs of the mice. After recovery from anesthesia, ECGs were recorded from conscious, ambulatory mice, and HRs were analyzed. Invasive mouse electrophysiology studies were performed as previously described (16).

Computational Modeling.

For modeling studies, previously described Voltage–Ca²⁺ clock model of a pacemaker cell was modified by simply adding an inward SOCE current (I_SOC) (38). Because we observed pacemaking shifting in the cSTIM1^−/− SAN, an inward rectifier K⁺ current (I_K1) and a background K⁺ current (I_bK), which were also described in mouse SA node models, were also added in the model. Model equations, parameters, and variables for I_SOC, I_K1, and I_bK are shown below. Based on our experimental measurements, the maximal conductance ($g_{SOC})$ and the time constant (${\tau }_{SOC})$ of I_SOC were set at 0.05 pA/pF and 1,000 ms, respectively.

$$I_{SOC}=C_m*g_{SOC}*\left(V_m-E_{SOC}\right)*x$$

$$\frac{dx}{dt}=(x_{\infty }-x)/{\tau }_{SOC}$$

$$x_{\infty }=1-\left(\frac{C{a}_{jSR}^4}{K_m^4+C{a}_{jSR}^4}\right)$$

$$I_{K\mathit{1}}=C_m*g_{K1}*{\left[K^{+}\right]}_o*(V_m-E_K)/([1+e^{0.071*}(V_m-E_K)]*({\left[K^{+}\right]}_o+0.229))$$

$$I_{bK}=C_m*g_{bK}*(V_m-E_K).$$

Data Analysis.

Averaged results are presented as mean ± SEM. For in vivo studies, based on a power calculation using a 95% confidence interval and assuming a 20% difference in the magnitude of the effect between groups, a sample size of n (mice, SAN tissue, or cells) for each group was between 8 and 20 to achieve a statistical power of 80% (beta = 20%). Mice (WT vs. KO) were allocated randomly for ECG recording, in vivo electrophysiology, and tissue optical mapping. Otherwise, no randomization was performed. The principal operator was blinded to the genotypes for in vivo procedures. Comparisons were made among sex- and age-matched controls for several independent litters. No data were excluded from analysis except those obtained from mice that died periprocedurally. A paired Student t test was used to compare measurements made in the same mice or in the same cells under two different conditions. An unpaired Student t test was used to compare two independent groups. All data points represent biologic replicates. For analysis of I-V curves, the currents at each membrane potential were averaged and plotted as mean ± SEM. At each membrane potential, means were tested for significance using an unpaired Student t test. A value of P < 0.05 was considered significant.

ECG and Intracardiac Electrophysiology Studies

In brief, mice were anesthetized using 250 mg/kg avertin following institutional guidelines. Anesthetized mice were placed on a heating pad (37 °C) for temperature control. Surface ECG recordings were obtained with s.c. placed 29-gauge needle electrodes connected to an ML138 Octal Bioamp (ADInstruments) and a Powerlab 16/30 acquisition system (ADInstruments) in both forelimbs and hindlimbs to create a Lead I and Lead II configuration. A cutdown of the internal jugular vein was performed, and an EPR-800 1.1F octapolar catheter (Millar Instruments) was advanced into the right atrium, atrioventricular junction, and right ventricle under intracardiac electrogram guidance. Programmed stimulation was performed at twice diastolic threshold using the Powerlab 16/30 Voltage stimulator (ADInstruments). Surface and intracardiac ECGs were sampled at 2 kHz, filtered between 0.3 Hz and 1 kHz, and analyzed with LabChart Pro-7.2 software (ADInstruments). Cardiac cycle intervals (milliseconds) were averaged from five consecutive PQRS complexes. The SNRT was measured after applying a 30-s atrial pacing train at a cycle length of 100 ms. The SNRT was defined as the interval between the last captured atrial pacing stimulus and the onset of first return sinus beat. The cSNRT was defined as the SNRT minus the baseline cycle length before rapid atrial pacing. The atrial effective refractory period (AERP) was defined as the longest S1–S2 coupling interval for atria that failed to demonstrate atrial capture with S2, with S1 representing the drive cycle and S2 representing the progressively premature stimulus. The effective refractory period of the AV node was defined as the longest S1–S2 coupling interval at which the premature stimulation delivered to the atrium is followed by a His potential but not by a QRS complex. Between stimulations, the mice were allowed to recover for at least 30 s.

Electrophysiology

For I_SOC measurements, the ruptured patch clamp technique was used. Pipette solution contained (in millimolars) NaCl 5, Aspartate acid 140, CsOH, 140, Mg·ATP 4, BAPTA 10, Hepes 10, and pH 7.2. The SANC was held at 0 mV, and I_SOC was activated by a standard ramp voltage protocol from −100 mV to 100 mV at a slope of 1 mV/ms. Voltage ramps were repeated every 3 s, and the initial two traces immediately after seal rupture were used as baseline. Because of the high input resistance of SANCa, a −5-mV, 20-ms seal test step was added to the end of voltage ramp protocol to monitor changes in seal resistance. Leak subtraction was performed during off-line data analysis. SR Ca²⁺ store depletion was achieved by including 10 mM BAPTA in the pipette solution and/or by adding 2 μM thapsigargin; 10 mM TEA and 5 mM Cs⁺ were added to external perfusion solution to inhibit potassium currents and I_f current.

Steady-state NCX1 currents were recorded using whole-cell patch clamp technique; 10 mM TEA, 5 mM CsCl, and 10 µM verapamil were added to external TYR solution to inhibit voltage gated calcium currents, potassium currents, and the hyperpolarization-activated current I_f. The pipette solution included (in millimolars) CsCl 65, NaCl 20, creatine phosphate 5, Mg·ATP 2, tetraethylammonium chloride 20, EGTA 20, Hepes 10, and pH 7.2. The membrane currents were activated by a ramp voltage protocol from −100 mV to 100 mV with a rate of 1 mV/ms. The protocol was repeated every 10 s. NCX1 current was dissected out by blockage with 5 mM Ni²⁺. I_NCX was therefore defined as the 5-mM Ni²⁺-sensitive currents.

To measure Ca²⁺ entry and SR Ca²⁺ store content, SANCs were held at −60 mV and perfused with TYR solution containing 10 µM verapamil, 2 mM CsCl, and 0.2 mM BaCl₂. NCX1 currents were recorded in response to a 30-s perfusion of 20 mM caffeine and 30 µM CPA by fast position perfusion solution near the cell (Warner Instrument SF-77B). SR Ca²⁺ store content as well as Ca²⁺ entry following the SR Ca²⁺ store depletion were measured by integrating NCX1 currents. The pipette solution contained (in millimolars) Aspartate acid 140, CsOH 140, Mg·ATP 2, EGTA 0.2, creatine phosphate 5, TEA·Cl 10, Hepes 10, and pH 7.2.

To measure I_Ca-L currents, the perforated patch clamp technique was used to minimize the current rundown. The pipette solution contained (in millimolars) Aspartate acid 140, CsOH 140, Mg·ATP 2, EGTA 11, ATP·Na₂ 2, CaCl₂ 1, Hepes 10, and pH 7.2. Amphotericin B was freshly added to pipette solution before each experiment, and its final concentration in pipette solution was 300 µg/mL SANCs were perfused with normal TYR solution. After gigaseal formation, the external solution was switched to a Na⁺-free solution containing (in millimolars) TEA chloride 140, MgCl₂ 1.05, CaCl₂ 1.8, CsCl 5, Glucose 10, Hepes 5, and pH 7.4. I_Ca-L currents were evoked by a 500-ms depolarization step from a holding potential of −50 mV to 60 mV with an interval of 10 mV. I_Ca-L currents were defined as 5-µM nifedipine-sensitive currents.