Distinct localization and modulation of Cav1.2 and Cav1.3 L-type Ca2+ channels in mouse sinoatrial node

Authors


A. Lee: Department of Molecular Physiology and Biophysics, University of Iowa, 5–610 Bowen Science Building 51 Newton Road, Iowa City, IA 52242, USA. Email: amy-lee@uiowa.edu

Key points

  • • In the sinoatrial node (SAN), Cav1 voltage-gated Ca2+ channels mediate L-type currents that are essential for normal cardiac pacemaking.
  • • Both Cav1.2 and Cav1.3 Ca2+ channels are expressed in the SAN but how their distinct properties affect cardiac pacemaking is unknown.
  • • Here, we show that unlike Cav1.2, Cav1.3 undergoes voltage-dependent facilitation and colocalizes with ryanodine receptors in sarcomeric structures.
  • • By mathematical modelling, these properties of Cav1.3 can improve recovery of pacemaking after pauses and stabilize SAN pacemaking during excessively slow heart rates.
  • • We conclude that voltage-dependent facilitation and colocalization with ryanodine receptors distinguish Cav1.3 from Cav1.2 channels in the SAN and contribute to the major impact of Cav1.3 on pacemaking.

Abstract  Dysregulation of L-type Ca2+ currents in sinoatrial nodal (SAN) cells causes cardiac arrhythmia. Both Cav1.2 and Cav1.3 channels mediate sinoatrial L-type currents. Whether these channels exhibit differences in modulation and localization, which could affect their contribution to pacemaking, is unknown. In this study, we characterized voltage-dependent facilitation (VDF) and subcellular localization of Cav1.2 and Cav1.3 channels in mouse SAN cells and determined how these properties of Cav1.3 affect sinoatrial pacemaking in a mathematical model. Whole cell Ba2+ currents were recorded from SAN cells from mice carrying a point mutation that renders Cav1.2 channels relatively insensitive to dihydropyridine antagonists. The Cav1.2-mediated current was isolated in the presence of nimodipine (1 μm), which was subtracted from the total current to yield the Cav1.3 component. With strong depolarizations (+80 mV), Cav1.2 underwent significantly stronger inactivation than Cav1.3. VDF of Cav1.3 was evident during recovery from inactivation at a time when Cav1.2 remained inactivated. By immunofluorescence, Cav1.3 colocalized with ryanodine receptors in sarcomeric structures while Cav1.2 was largely restricted to the delimiting plasma membrane. Cav1.3 VDF enhanced recovery of pacemaker activity after pauses and positively regulated pacemaking during slow heart rate in a numerical model of mouse SAN automaticity, including preferential coupling of Cav1.3 to ryanodine receptor-mediated Ca2+ release. We conclude that strong VDF and colocalization with ryanodine receptors in mouse SAN cells are unique properties that may underlie a specific role for Cav1.3 in opposing abnormal slowing of heart rate.

Abbreviations 
AP

action potential

BB

blocking buffer

Cav

voltage-gated Ca2+ channel

CDF

Ca2+-dependent facilitation

CDI

Ca2+-dependent inactivation

DHP

dihydropyridine

PDZ

post-synaptic density/zonula occludens

HCN4KiT-Cre

double transgenic mice with targeted deletion of Cav1.2 in the SAN

KB

K+ buffer

LCS

low calcium solution

NIM

nimodipine

PM

plasma membrane

RYR

ryanodine receptor

SAN

sinoatrial node

SANDD

SAN dysfunction and deafness syndrome

SR

sarcoplasmic reticulum

VDF

voltage-dependent facilitation

VDI

voltage-dependent inactivation

WT

wild-type

Introduction

Normal cardiac rhythmicity depends on the intrinsic properties, modulation and subcellular localization of ion channels in the heart. Voltage-gated Cav1 channels (Cav1.2 and Cav1.3) mediate L-type Ca2+ currents (ICa,L) that play distinct roles in different cardiac cell types. Expressed throughout the heart, Cav1.2 regulates action potential (AP) duration and excitation–contraction coupling in the ventricle (Klugbauer et al. 2002). Cav1.3 is largely absent from the ventricles but is highly expressed in atria, atrioventricular node and sinoatrial node (SAN) (Marger et al. 2011; Zhang et al. 2011). Loss-of-function of Cav1.3 both in mice and humans causes ‘sick sinus’ syndrome characterized by severe bradycardia (Platzer et al. 2000; Le Scouarnec et al. 2008; Baig et al. 2011). Thus, factors that regulate Cav1.3 may have a major impact on pacemaking.

Compared to Cav1.2, Cav1.3 activates more rapidly and at more negative membrane potentials (Koschak et al. 2001; Xu & Lipscombe, 2001). These properties allow Cav1.3 to contribute more significantly than Cav1.2 to the diastolic depolarization in SAN cells (Mangoni et al. 2006a). However, Cav1.3 differs in other ways from Cav1.2 that may further determine the larger impact of Cav1.3 on sinoatrial pacemaking. In transfected HEK293T cells, Cav1.3 undergoes significantly greater Ca2+-dependent inactivation (CDI) and less voltage-dependent inactivation (VDI) than Cav1.2 (Tadross et al. 2010). In addition, Cav1.3 interacts selectively with multiple post-synaptic density/zonula occludens (PDZ) domain containing proteins that regulate the function and localization of these channels in various cell types. For example, harmonin controls Cav1.3 current density by enhancing ubiquitination and proteosomal degradation of these channels in auditory hair cells (Gregory et al. 2011). Interactions with Shank in neurons allows for Cav1.3 modulation by G-protein coupled receptors and coupling to cAMP response element-binding protein-dependent transcription (Olson et al. 2005; Zhang et al. 2005, 2006). Erbin and densin promote voltage- and Ca2+-dependent facilitation (CDF) (Calin-Jageman et al. 2007; Jenkins et al. 2010), respectively, of Cav1.3 but not Cav1.2. Whether such Cav1.3-specific modulation occurs in SAN cells is unknown but necessary for understanding the privileged role of Cav1.3 in regulating heart rhythm.

The absence of pharmacological agents that unequivocally distinguish between Cav1.2 and Cav1.3 poses a major challenge to the characterization of these Cav1 channels in their native contexts. However, genetically modified mice (Cav1.2 DHP) in which a single point mutation eliminates the high sensitivity of Cav1.2 to dihydropyridines (DHPs), have proven useful in this regard (Sinnegger-Brauns et al. 2004). Thus, in cells from Cav1.2 DHP mice that express both Cav1.2 and Cav1.3, DHP antagonists spare the Cav1.2-mediated L-type current. In whole cell patch-clamp recordings from SAN cells isolated from these mice, we identified voltage-dependent properties that distinguish Cav1.2 and Cav1.3. In addition, we uncovered differences in the subcellular localization of Cav1.2 and Cav1.3 in SAN cells by immunofluorescence. In a mathematical model of mouse SAN cells, we demonstrate how these unique properties of Cav1.3 may be physiologically relevant in controlling pacemaking.

Methods

Ethical approval

All procedures involving animals were approved by the Institutional Animal Care and Use Committee at the University of Iowa, University of Erlangen, and the University of Montpellier. These procedures were in accordance with National Institutes of Health guidelines.

Preparation of isolated mouse sinoatrial node cells

SAN cells were isolated from 2- to 6-month-old male or female Cav1.2 DHP mice (Sinnegger-Brauns et al. 2004). Age-matched wild-type (WT) mice were used as controls. Mice were anaesthetized by inhalation of isoflurane and depth of anaesthesia assessed by loss of hindlimb reflex. Beating hearts were removed from anaesthetized animals and immediately transferred to pre-warmed (37°C) Tyrode solution (in mm: 140 NaCl, 5 Hepes, 5.4 KCl, 1.8 CaCl2, 1 MgCl2, 5.5 glucose, pH 7.35 with NaOH). The SAN was removed and washed with low Ca2+ solution (LCS; in mm: 140 NaCl, 5 Hepes, 5.4 KCl, 0.2 CaCl2, 0.5 MgCl2, 5.5 glucose, 1.2 KH2PO4, 50 taurine, 1 mg ml−1 bovine serum albumin, pH 6.9 with NaOH), cut into 10–20 fragments, and then incubated in LCS containing proteases (in U ml−1: 229 collagenase, 1.9 elastase (both Worthington Biochemical, Lakewood, NJ, USA), 0.9 protease type XIV (Sigma-Aldrich, St. Louis, MO, USA), pH 6.9 with NaOH) at 37°C for 20–25 min, shaken gently every 5 min. The tissue was washed three times with high K+ buffer (KB, in mm: 100 potassium glutamate, 5 Hepes, 20 glucose, 25 KCl, 10 potassium aspartate, 2 MgSO4, 10 KH2PO4, 20 taurine, 5 creatine, 0.5 EGTA, pH 7.35 with KOH). The tissue was triturated to release SAN cells, which were kept at 4°C until experiments performed the same day.

Electrophysiological recordings

Isolated SAN cells (in KB) were transferred to a glass coverslip in a recording chamber and left to settle for 15–20 min before recording. Upon wash-in of Tyrode solution, rhythmically beating SAN cells were identified for recording. Glass recording electrodes (2–4 MΩ) were filled with intracellular solution containing (in mm: 135 CsCl2, 10 Hepes, 10 EGTA, 1 MgCl2, 4 Mg-ATP, pH 7.2 with CsOH). After establishing whole cell configuration Tyrode solution was replaced with extracellular recording solution (in mm: 130 TEA-Cl, 1 MgCl2, 25 Hepes, 4 BaCl2, 10 4-aminopyridine, pH 7.3 with TEA-OH). Holding voltage was set to −60 mV. Currents were recorded with an EPC-9 patch-clamp amplifier using PULSE software by HEKA Electronics (Lambrecht/Pfalz, Germany). Leak and capacitive transients were calculated and subtracted using a P/4 protocol. Data were analysed using IGOR Pro software (WaveMetrics, Lake Oswego, OR, USA). Averaged data represent mean ± standard error of the mean.

Generation of sinoatrial node-specific CaV1.2 knockout mice

Double transgenic mice with targeted deletion of Cav1.2 in the SAN (CaV1.2flox/flox, HCN4KiTCre/+) were generated by crossing floxed CaV1.2 (Seisenberger et al. 2000) and HCN4KiT-Cre (Hoesl et al. 2008). Cre-mediated recombination was induced by administration of tamoxifen (1 mg 25 mg−1, i.p.) for five consecutive days.

Immunofluorescence

Isolated SAN cells in KB solution were left to settle for 30 min on glass-bottom microwell dishes (35 mm Petri dish, 14 mm microwell) coated with fibronectin (0.05 μg μl−1 in Ham's F-10 media). Cells were fixed with 2% paraformaldehyde for 15 min, washed three times with phosphate-buffered saline (PBS), and blocked for 1 h in blocking buffer (BB: 0.075% Triton-X 100, 5% goat serum, in PBS). To block non-specific sites, cells were incubated with rabbit IgG (1:50 in BB) overnight at 4°C and after washing three times with PBS (15 min each), cells were incubated with goat anti-rabbit Fab fragments (1:25 in BB) for 2 h at room temperature. Cells were washed and incubated overnight at 4°C with the appropriate primary antibodies diluted in BB: mouse-anti-ryanodine receptor (RYR) type 2 (1:1000, Thermo Scientific, Waltham, MA, USA), rabbit polyclonal antibodies against Cav1.2 (1:2000; Tippens et al. 2008) or Cav1.3 (1:500; Gregory et al. 2011). The specificity of these antibodies for labelling Cav1.2 and Cav1.3 in SAN was confirmed using tissue and cells from mice lacking either channel (Supplementary Fig. 1). In some experiments, commercially available rabbit polyclonal Cav1.3 antibodies (Alomone Labs, Jerusalem, Israel) were used, which produced qualitatively similar results as the Cav1.3 antibodies we characterized previously (Gregory et al. 2011). Following washes, cells were incubated with secondary antibodies (Alexa 488 goat-antirabbit and Alexa 568 goat-anti-mouse, 1:1000 in BB) for 2 h at room temperature, washed three times with PBS and mounted on glass microscope slides with Dako fluorescence mounting medium (Carpinteria, CA, USA). Images were acquired with a Fluoview (FV1000, Olympus) confocal microscope on an upright microscope (BX61WI) using a 100× oil-immersion objective. Line scan analysis of immunofluorescence intensity (Fig. 5) was performed with Fluoview software (Olympus).

Quantitation of colocalization within a selected region of interest was performed with the JACoP plug-in for ImageJ (Bolte & Cordelieres, 2006). To restrict analysis to colocalization of Cav1 and RYR2 labelling within the plasma membrane and sarcomeric structures, the signal threshold intensity for both was manually set to levels that just eliminated fluorescence signals outside these regions. Use of the Costes automatic threshold calculation tended to overestimate colocalization as it included signal intensities outside plasma and sarcomeric membranes. Using the manually set thresholds, Mander's colocalization coefficient was determined by the JACoP software. This coefficient varied from 0 to 1, the former corresponding to non-overlapping signals and the latter reflecting 100% colocalization between the two fluorophores. As Mander's colocalization coefficient depends on the accuracy with which the threshold is determined, Pearson's correlation coefficient was also measured, which is generally insensitive to inaccuracies of the threshold setting. Similar results were obtained with both procedures.

To compare the subcellular distribution of colocalized RYR2 with Cav1.2 or Cav1.3, we performed three line scans per cell through RYR-labelled sarcomeric structures. The signal from each line scan was subdivided into plasma membrane and intracellular signals and the ratio of the two was calculated as the relative intensity. Using this method, a plasma membrane/intracellular ratio of 1 would indicate similar signal intensities in both areas. A ratio less than 1 would signify greater intracellular compared to plasma membrane labelling and vice versa for a ratio greater than 1 (Supplementary Fig. 2).

Computational analysis

We performed numerical simulations of pacemaker activity by modifying a model of mouse SAN automaticity that we developed previously (Mangoni et al. 2006b). This previous model fairly reproduced the behaviour of membrane-bound ion channels and automaticity of mouse SAN pacemaker cells; however, it did not include calculation of intracellular ion homeostasis and compartments underlying Ca2+ handling (Mangoni et al. 2006b). To generate a new model of mouse SAN cell automaticity calculating Ca2+ handling, we used the set of model equations in Mangoni et al. (2006b) to simulate the behaviour of ion channels of the plasma membrane and added calculations of intracellular compartments and ion concentrations according to Kurata et al. (2002). The general structure of the model, including the organization of intracellular Ca2+ handling compartments and membrane-bound ion channels and RYRs, is shown in Supplementary Fig. 3.

The volumes of the subsarcolemmal space, as well as that of sarcoplasmic reticulum (SR) and non-SR compartments were defined as in Kurata et al. (2002). Intracellular Na+ and K+ concentrations, as well as Ca2+ handling were also calculated according to Kurata et al. (2002). We used the set of equations previously published by Maltsev & Lakatta (2009) to calculate RYR-dependent Ca2+ release from the SR. To simulate co-distribution between Cav1.3 channels and RYRs, our model assumed that 71% of the total Cav1.3-mediated ICa,L current density contributed to the Ca2+ concentration of the subsarcolemmal compartment (SSCa). The residual 29% of Cav1.3-mediated ICa,L, the Cav1.2-mediated ICa,L, Cav3.1-mediated ICa,T and background Ca2+ current (Ib(Ca)) contributed to Ca2+ of the bulk cytosol ([Ca2+]i) (Kurata et al. 2002; Maltsev & Lakatta, 2009), as well as to Ca2+ load of the SR network (Maltsev & Lakatta, 2009). Consequently, Cav1.2- and Cav3.1-dependent Ca2+ entries did not control RYR-dependent Ca2+ release directly. In this respect, our model differs from that of Kurata et al. (2002) and Kharche et al. (2011), which assume that Cav1.3- and Cav1.2-mediated ICa,L as well as Cav3.1-mediated ICa,T contribute to SSCa and can collectively trigger RYR-dependent Ca2+ release. All the other parameters controlling membrane-bound ion channel densities and ionic pumps were set as in the model by Mangoni et al. (2006b).

To calculate VDF of Cav1.3 channels we modelled VDF by directly modifying the equations generating Cav1.3-mediated ICa,L in the mouse SAN automaticity model. Indeed, we reasoned that during simulations of pacemaker activity, the preceding AP amplitude would condition Cav1.3- and Cav1.2-mediated ICa,L similarly to the prepulse in a voltage-clamp experiment, while the diastolic interval would determine the time after the prepulse. As we calculated the VDF model parameters by fitting the dependency of VDF from the prepulse amplitude and time after prepulse in voltage-clamp experiments conducted on native SAN cells we expected that the model would reproduce the behaviour of Cav1.3- and Cav1.2-mediated ICa,L at different SAN cell firing rates. Specifically, to compute Cav1.3-mediated ICa,L, we assumed that the probability of Cav1.3 to be in the open state is the result of combined effects of VDF, VDI and CDI. VDF of Cav1.3-dependent Ca2+ current (ICa,L) was modelled by adding a variable for VDF, f(f), that affected ICa,L independently from VDI and CDI according to the equations:

display math
display math
display math
display math

where gmax is the maximal current conductance, V is the membrane voltage, Vrev is the current reversal potential, t is the time, f(d) is the activation variable, f(I) and f(Ca) are the inactivation variables corresponding to VDI and CDI respectively, F(v) is the dependence of VDF on the prepulse voltage (Fig. 2A, Supplementary Fig. 3A) and F(t) is the dependence of VDF on the recovery interval (Figs 2B and S3B). A, B, A1, A2 and A3 are constant factors; s is a slope factor. The f(f) variable was omitted in calculations of Cav1.2-mediated ICa,L activation (Fig. 1, this study) and inactivation (Mangoni et al. 2003) time constants experimentally measured in SAN cells were used to model Cav1.3-mediated ICa,L. The maximal conductance of Cav1.3- and Cav1.2-mediated ICa,L was set according to measurements in extracellular 2 mm Ca2+ native SAN cells (Mangoni et al. 2003). The model satisfactorily reproduced our experimental characterization of Cav1.3 VDF (Figs 2 and 6). We assumed that SSCa directly controlled the open probability of RYRs and the CDI inactivation variable f(Ca) of Cav1.3-mediated ICa,L. This assumption is experimentally verified (Neco et al. 2012; see below). In the main code of the SAN model we added an expression to apply simulated voltage commands when needed. The user in simulations of cellular automaticity can deactivate this function.

Figure 1.

Pharmacological isolation of IBa mediated by Cav1.2 and Cav1.3 in Cav1.2 DHP mouse sinoatrial node (SAN) cells 
A, phase contrast image of a representative mouse SAN cell used for whole cell patch-clamp recordings. B, effects of nimodipine (NIM, 1 μm) on IBa in wild-type (WT, n = 3) mouse SAN cells. C, same as in B except with Cav1.2 DHP SAN cells (n = 23). The residual current in the presence of NIM (Cav1.2) was subtracted from the total current to yield the Cav1.3 component. In B and C, representative IBa traces and voltage protocol (left) and current–density (pA pF−1) vs. voltage (mV) relation (right) are shown.

Figure 2.

Cav1.3 currents activate faster and inactivate more slowly than Cav1.2 currents in mouse sinoatrial node cells 
A, Cav1.2 and Cav1.3-mediated IBa was evoked and isolated as in Fig. 1B. Current traces (left) were fit with a single exponential function (smooth black line). The resulting time constants (τ) were plotted against test voltage (right, n = 22). B, inactivation of IBa evoked by 500 ms test pulses to a test voltage of −10 mV. Ires/Ipeak = current amplitude at the end of the pulse normalized to peak amplitude, plotted for test voltages of −30 and −10 mV. *= 0.01, one-way ANOVA, n = 8.

Slowing of pacemaking was obtained by simulating activation of the ATP-dependent K+ current (IKATP). IKATP is expressed in mouse SAN pacemaker cells and is activated by lowering the intracellular ATP concentration during ischaemia (Fukuzaki et al. 2008). Calculations were performed in the Jsim environment for integration of differential equations (http://nsr.bioeng.washington.edu/jsim/). The integration step was set to 200 μs. Simulations were analysed using the Graph Prism software (ver. 5.03).

Results

Cells dissociated from the SAN of Cav1.2 DHP mice exhibited properties similar to those described previously for mouse SAN cells (Mangoni & Nargeot, 2001). The cells selected for recording were generally spindle-shaped (Fig. 1A) and were spontaneously beating in physiological saline (Tyrode's solution). Similar to values reported previously for mouse primary pacemaker SAN cells (Mangoni & Nargeot, 2001), the mean capacitance of Cav1.2 DHP SAN cells was 32.4 ± 1.6 pA pF−1 (n = 23). Compared to WT SAN cells, Cav1.2 DHP SAN cells exhibited a similar range of IBa densities (5–33 pA pF−1 for Cav1.2 DHP vs. 12–33 pA pF−1 for WT), which was consistent with previous results indicating that Cav1.2 current density was not significantly altered by the DHP knock-in mutation (Sinnegger-Brauns et al. 2004; Zhang et al. 2007).

In the present study, we focused on Ba2+ currents (IBa) rather than Ca2+ currents (ICa) to restrict our analysis to voltage- rather than Ca2+-dependent modulation of Cav1 channels (Christel & Lee, 2012 for review). To isolate IBa mediated by Cav1.2 and Cav1.3 in Cav1.2 DHP SAN cells, we used nimodipine (NIM, 1 μm) a DHP antagonist of Cav1 channels. In SAN cells from WT mice, NIM nearly abolished IBa (up to 95%, Fig. 1B), consistent with the dominant contribution of Cav1.2 and Cav1.3 channels to the whole cell L-type current in these cells (Mangoni et al. 2003; Marionneau et al. 2005). In contrast, the same concentration of NIM reduced IBa only ∼58% in Cav1.2 DHP SAN cells, which was consistent with a reduced sensitivity of Cav1.2 DHP channels to NIM (Fig. 1C). We interpreted the residual current that was not blocked by NIM as largely mediated by Cav1.2. Subtraction of this NIM insensitive IBa from total IBa yielded the Cav1.3 component (Fig. 1C).

The presence of a modest DHP-insensitive component (∼5%) of the whole cell IBa in WT SAN cells (Fig. 1B) could be problematic if it contributed more significantly to IBa in Cav1.2 DHP cells as it would be interpreted as ‘Cav1.2’ using our NIM subtraction protocol. However, if this protocol effectively dissected Cav1 components of IBa, the properties of the NIM-insensitive current and the difference current should match those expected of Cav1.2 and Cav1.3, respectively. For example, in heterologous expression systems, Cav1.3 activates more rapidly and at more negative voltages than Cav1.2 (Koschak et al. 2001; Scholze et al. 2001; Xu & Lipscombe, 2001). In our experiments, Boltzmann fits of current–voltage relations revealed that compared to Cav1.2 currents, Cav1.3 currents activated at more negative voltages (V1/2 = −25.4 ± 1.4 mV for Cav1.3 vs. −15.7 ± 1.4 mV for Cav1.2; n = 22, < 0.001) in Cav1.2 DHP SAN cells (Fig. 1C). To compare the activation kinetics of Cav1.2 and Cav1.3 currents, we fit the rising phase of IBa evoked by various voltages with a single exponential function. With this analysis, Cav1.3 currents activated significantly faster than Cav1.2 currents (up to 30%, < 0.01; Fig. 2A). Based on findings that recombinant Cav1.3 channels undergo less VDI than Cav1.2 (Koschak et al. 2001), we also compared inactivation of Cav1.2 and Cav1.3 IBa. For these experiments, a long (500 ms) test pulse was used to evoke IBa and time constants for inactivation determined from a single exponential fit of the IBa decay. As expected, Cav1.3 currents inactivated significantly more slowly (∼36% at −10 mV, P = 0.01; Fig. 2B) than Cav1.2. Taken together, these results validated our strategy for isolating Cav1.2 and Cav1.3 from the total IBa in Cav1.2 DHP SAN cells, which was applied for the remainder of the study.

We have shown previously that Cav1.3 channels in transfected HEK293T cells undergo voltage-dependent facilitation (VDF) (Calin-Jageman et al. 2007). Because L-type currents in rabbit SAN cells also undergo VDF, we tested if VDF was a property of Cav1.2 and/or Cav1.3 in mouse SAN cells. Using a modified voltage protocol for measuring VDF, we compared IBa evoked before (P1) and after (P2) a conditioning prepulse (Fig. 3). VDF due to the conditioning prepulse should manifest as an increase in the P2 compared to P1 current. Rather than facilitation, both Cav1 currents exhibited VDI in that the P2 current amplitude was always less than that for P1 (Fig. 3). Consistent with our kinetic analyses (Fig. 2B), VDI was significantly weaker for Cav1.3 compared to Cav1.2 (∼26% at +80 mV prepulse voltage, < 0.001). Notably, the ratio of P2/P1 current amplitude was relatively invariant with prepulse voltage except for between +60 and +80 mV, where P2/P1 for Cav1.3 significantly increased (Fig. 3). This suggested that VDF might have been occluded by VDI such that overt VDF was not observed with this protocol.

Figure 3.

Cav1.3 and Cav1.2 undergo VDI in mouse sinoatrial node (SAN) cells 
Top, voltage protocol. Test currents were evoked by 50 ms test pulses before (P1) and after (P2) a 200 ms conditioning prepulse. Middle, representative IBa evoked by P1 (black trace) and P2 (grey trace). Bottom, fractional current (P2/P1) was plotted against prepulse voltage. *< 0.001, two-way ANOVA for group and voltage, Tukey post hoc test, n = 9.

To follow-up on this possibility, we sought to resolve VDF from VDI by measuring IBa during recovery from inactivation. The rationale for this approach was based on our previous findings that Cav channels may recover faster from inactivation than facilitation, such that facilitation can be measured once channels have fully recovered from inactivation (Lee et al. 1999, 2000). For these experiments, we compared P2 and P1 current amplitudes at varying intervals after a conditioning prepulse to +80 mV. Consistent with a role for Cav1.3 VDF in offsetting the effects of VDI, the full amplitude of the Cav1.3 current was restored within ∼250 ms of the prepulse, at which time the Cav1.2 current was still ∼20% inactivated. At longer recovery intervals (∼450 ms), Cav1.3 showed small but measurable VDF (∼5%) while Cav1.2 currents remained ∼15% inactivated (Fig. 4). These results demonstrate that Cav1.3 channels in mouse SAN cells undergo VDF, a property that is not shared by Cav1.2.

Figure 4.

Cav1.3 but not Cav1.2 undergoes voltage-dependent facilitation during recovery from inactivation 
Top, voltage protocol. Test currents were evoked by 50 ms test pulses before (P1) and at variable intervals (Δt) after (P2) a 200 ms conditioning prepulse to +80 mV. Middle, representative IBa evoked by P1 (black trace) and P2 (grey trace). Bottom, fractional current was plotted against prepulse to P2 interval. *< 0.01, #< 0.001 for Cav1.3 vs. Cav1.2 by one-way anova and Tukey post hoc test, n = 8.

In addition to differences in their intrinsic properties and regulation, distinct localization of Cav1.2 and Cav1.3 may also affect their respective contributions to SAN function. Therefore, we compared the subcellular distribution of Cav1.2 and Cav1.3 in isolated mouse SAN cells. We used Cav1.2 and Cav1.3 antibodies, which we have shown to specifically recognize the corresponding channels in rodent brain (Tippens et al. 2008; Gregory et al. 2011). To verify that these antibodies are equally specific for immunofluorescence of mouse SAN, we compared immunostaining in tissue obtained from WT mice and those lacking Cav1.3 (Platzer et al. 2000) or Cav1.2. As full body knockout of Cav1.2 causes embryonic lethality (Seisenberger et al. 2000), we used mice with inducible, targeted deletion of Cav1.2 in the SAN (see Methods section). In both fixed SAN tissue and isolated SAN cells, there was strong immunofluorescence corresponding to Cav1.2 and Cav1.3 antibodies in samples obtained from WT but not knockout mice (Supplementary Fig. 1). These results validated the use of these antibodies for reporting the localization of Cav1.2 or Cav1.3 in SAN cells.

In isolated SAN cells, Cav1.3 but not Cav1.2 antibodies labelled distinct bands of regular periodicity (Supplementary Fig. 1, Fig. 5), a pattern resembling the localization of RYR2 reported in SAN cells (Rigg et al. 2000). As in ventricular myocytes, RYRs mediate Ca2+-induced Ca2+ release in SAN cells (Rigg et al. 2000; Bogdanov et al. 2001), which subsequently could depend on Ca2+ influx through Cav1.3 channels. To determine if Cav1.3 colocalized with RYR2, we performed double labelling with mouse monoclonal anti-RYR2 antibodies and either Cav1.2 (Fig. 5AE) or Cav1.3 (Fig. 5FJ) antibodies. Both Cav1.2 and Cav1.3 were strongly colocalized with RYR2 based on coefficients for Mander's overlap (0.71 ± 0.08 for Cav1.3 and 0.85 ± 0.07 for Cav1.2, n = 3 each) and Pearson's correlation (0.74 ± 0.09 for Cav1.3, 0.71 ± 0.01 for Cav1.2). Unexpectedly, Cav1.3 strongly colocalized with RYR2 both at the sarcolemma and in the sarcomeric structures, while Cav1.2 mainly colocalized with RYR2 in the sarcolemmal membrane. Line scan analyses confirmed the coincidence of Cav1.3 and RYR2 immunofluorescence throughout the SAN cell (Fig. 5J), while Cav1.2 labelling most intensely overlapped with RYR2 at the delimiting sarcolemmal membrane (Fig. 5E).

Figure 5.

Confocal micrographs showing double labelling of RYR2 and Cav1.2 (A–E) or Cav1.3 (FJ) 
Regions of colocalization are yellow in the merged images (C, D, H and I). E and J, line scan analysis through region marked with dashed lines in D and I showing signals for RYR2 (red) and Cav1.2 or Cav1.3 (green).

To quantitatively estimate the difference in the subcellular distribution of Cav1.2 and Cav1.3, we calculated the ratio of plasma membrane to intracellular Cav1 immunofluorescence from line scans through RYR2-labelled sarcomeric structures (see Methods and Supplementary Fig. 2). By this metric, values less than 1 indicate greater signal intensity in intracellular compared to plasma membrane regions. This analysis indicated stronger intensity of Cav1.3 labelling in sarcomeric structures than on the plasma membrane (ratio = 0.86 ± 0.03, n = 3). In contrast, Cav1.2 labelling was far more intense on the plasma membrane than intracellularly (ratio = 2.44 ± 0.10, n = 3). When compared to values obtained for RYR2 (ratio = 0.95 ± 0.04 and 0.99 ± 0.04 for Cav1.3 and Cav1.2 double-labelled groups, respectively), only the ratio obtained for Cav1.2 was significantly different from that of Cav1.3 and RYR2 (< 0.001 by ANOVA and post-hoc Bonferroni test). Taken together, these results verify a distinct subcellular distribution of Cav1 channels and suggest that Cav1.3 is better positioned than Cav1.2 to promote RYR Ca2+ release in mouse SAN cells.

The unique localization and VDF exhibited by Cav1.3 and not Cav1.2 could significantly influence the important role of Cav1.3 in SAN pacemaking (Mangoni et al. 2003). Furthermore, no pharmacological or molecular tools are currently available to selectively inhibit VDF of Cav1.3 channels, a condition that prevents direct experimental investigation of the physiological role of VDF in cardiac pacemaker activity. Computational models have proven useful in predicting the impact of alterations in ionic currents on mouse SAN pacemaking (Mangoni et al. 2006b). To predict the impact of VDF on pacemaker activity, we thus used a numerical model of mouse SAN automaticity. We assumed that Cav1.3 VDF operates on channel gating in a way that is independent from VDI, CDI, and CDF. Experimental evidence showed that enhanced RYR-dependent Ca2+ release promotes CDI of SAN ICa,L in a mouse model of congenital polymorphic ventricular tachycardia (Neco et al. 2012). Consequently, we assumed that RYRs also contributed with Cav1.3-mediated ICa,L to the subsarcolemmal Ca2+ concentration, which in turn directly controlled CDI. As Cav1.3 channels preferentially colocalized with RYR2 in comparison to Cav1.2 channels (Fig. 5), the model included Cav1.3-mediated ICa,L as the main contributor of Ca2+ entry in the subsarcolemmal space. By quantitative analyses of our colocalization data (Fig. 5), we estimated ∼71% of total Cav1.3 immunofluorescence colocalized with that for RYRs. Therefore, we assumed that 71% of Cav1.3 channels contributed to subsarcolemmal Ca2+ concentration coupled to RYR-dependent Ca2+ release and that the remaining 29% contributed mainly to cytoplasmic Ca2+ concentration along with Cav1.2 and Cav3.1 channels (see Methods).

The model satisfactorily fitted experimental data on VDF of Cav1.3-mediated ICa,L as a function of prepulse amplitude or recovery time (Fig. 6A and B). In voltage-clamp simulations, we simulated Cav1.3-mediated ICa,L with a prepulse amplitude to +30 mV, a voltage close to the maximum positive voltage reached by the simulated SAN AP. Cav1.3-mediated ICa,L was then measured at 0 mV (Fig. 6C). Simulations of voltage-clamp experiments showed that VDF increased Cav1.3-mediated ICa,L during recovery from inactivation (Fig. 6C). Consistently with our experimental findings (Fig. 4), Cav1.3-mediated ICa,L amplitude was higher in the model containing VDF for each recovery time tested (from 14.2% at 66 ms, to 23.8% at 800 ms recovery time).

Figure 6.

Numerical modelling of VDF in mouse sinoatrial node cells 
A and B, fitting by the model of P2/P1 ratio as a function of the conditioning prepulse voltage (A) or recovery interval (B). Voltage protocols were the same as in Figs. 3 and 4. C, simulation of voltage-clamp experiments showing recovery from inactivation of Cav1.3-mediated ICa,L with (grey line) or without (black line) voltage-dependent facilitation. Dashed lines indicate zero current level and maximal ICa,L during recovery from inactivation. For the voltage-clamp protocol (left), the 200 ms conditioning prepulse was set to +30 mV.

The numerical model of mouse SAN pacemaker activity, which included VDF and preferential coupling of Cav1.3 channels to RYRs, faithfully reproduced SAN APs and the main ionic currents involved in the diastolic depolarization (Fig. 7A). The hyperpolarization-activated current If was activated in the first phase of the diastolic depolarization and its density increased in parallel with the decay of IKr (Fig. 7A). The fast Na+ current (INa) was modelled as the sum of the tetrodotoxin-sensitive and tetrodotoxin-insensitive component as previously described (Lei et al. 2004; Mangoni et al. 2006b). The density of INa during diastolic depolarization was similar to that of Cav1.3-mediated ICa,L (data not shown). Cav1.3-mediated ICa,L began to activate about midway between the maximum diastolic potential and the threshold of the following AP upstroke, while the Cav1.2-mediated ICa,L was activated during the AP upstroke (Fig. 7A). In comparison to a model excluding Cav1.3 channel VDF, Cav1.3-mediated ICa,L density was increased (25%) at a basal diastolic interval of 278 ms (Fig. 7C and D). As a consequence of augmented Cav1.3-mediated ICa,L during pacemaking (Fig. 7C), the model predicted a small but measurable positive chronotropic effect on basal pacing rate compared to a model lacking VDF (2%; Fig. 7D).

Figure 7.

Effects of Cav1.3 voltage-dependent facilitation (VDF) on ICa,L and sinoatrial node (SAN) cell pacemaking in a mathematical model 
A, numerical simulations, including Cav1.3 VDF of SAN action potentials (APs), ICa,L (Cav1.3, black line; Cav1.2-mediated, grey line), IKr, and If. B, comparison between pacemaker activity (APs), Na+/Ca2+ exchanger current (INCX), subsarcolemmal Ca2+ concentration (SSCa), and SR Ca2+ release (SRCa) in our model. C, Cav1.3-mediated ICa,L calculated with VDF (grey line) or without VDF (black line). D, corresponding pacemaker activity with VDF (grey line) or without VDF (black line).

To gain further insights into the potential impact of VDF in regulating aberrant pacemaking, we modelled the behaviour of Cav1.3-mediated ICa,L upon slowing of pacemaker activity by activation of currents through KATP channels (IKATP, see Methods; Fig. 8A and B). IKATP inhibits SAN automaticity, which may protect against myocardial damage due to ischemic insults (Fukuzaki et al. 2008). In rabbit SAN cells, pharmacological activation of IKATP robustly slows pacemaking. Arrest of automaticity can also be observed at high IKATP densities (Fukuzaki et al. 2008). We thus modelled the impact of IKATP activation by moderately increasing KATP conductance (0.1–0.15 pA pF−1 in the diastolic depolarization range). The effect of IKATP activation was a transient pause in automaticity that recovered to regular steady-state pacemaking. However, pacemaker activity recovered faster in the model with VDF (within 828 ms; Fig. 8A) than without VDF (within 2005 ms; Fig. 8B). During the first beat upon recovery of pacemaking, Cav1.3-mediated ICa,L was 36.2% larger in the model with VDF than without VDF (Fig. 8CE). After recovery of automaticity, steady-state pacemaker activity was slower than in control simulations with no IKATP activation in the model including VDF and excluding VDF. Compared to control conditions, steady-state pacemaker activity was slowed by 18% in the model including VDF and by 20% in the model excluding VDF (from 278 to 338 ms including VDF and from 283 to 350 ms excluding VDF; Fig. 8G), which would constitute a significant reduction in heart rate in vivo. At steady state, Cav1.3-mediated ICa,L was 17% larger in the model including VDF than in the model excluding VDF (Fig. 8F), which is consistent with a role for VDF in promoting full recovery from inactivation (Figs 4 and 6). Slowing of pacemaker activity itself induces recovery of ICa,L from inactivation, which increases Cav1.3-mediated ICa,L even without VDF (44%; Fig. 8H). However, with VDF, slowing of pacemaking was reduced and pacemaker activity was measurably faster (3.4%) than that of the model excluding VDF (Fig. 8G). Taken together, our results demonstrate that Cav1.3 VDF may operate as a positive chronotropic factor that may regulate Ca2+ signals and excitability of SAN cells.

Figure 8.

Activation of IKATP transiently stopped automaticity in a model including VDF (A) and excluding VDF (B) 
Arrows indicate the onset of IKATP activation. Pacemaking resumed after pause. At steady state, pacemaker activity was slowed by 18% in the model including VDF and by 20% in the model excluding VDF. C and D show corresponding Cav1.3-mediated ICa,L, while dashed arrows indicate Cav1.3-mediated ICa,L at the first beat upon resuming of automaticity. E and F, enlargement of Cav1.3-mediated ICa,L at the first beat and steady state upon resuming of automaticity in the model including (grey line) and excluding VDF (black line). Traces represent superimposed Cav1.3-mediated ICa,L during steady-state pacemaker activity after IKATP activation in the model including (grey line) and excluding VDF (black line). Steady-state pacemaker activity (G) and Cav1.3-mediated ICa,L (H) after activation of IKATP in the model including VDF (grey line) or after abolition of VDF (black line).

Discussion

Our study provides multiple lines of evidence for the functional specialization of Cav1 channels in SAN cells. First, Cav1.3 undergoes VDF particularly during slow heart rate, as channels recover from inactivation. Second, Cav1.3 channels are strongly colocalized with RYRs. Third, Cav1.2 channels undergo comparatively little VDF and are not colocalized with sarcomeric RYRs. Thus, despite contributing ∼42% of the whole cell ICa,L, Cav1.2 channels are less capable than Cav1.3 to restore normal pacemaking. Fourth, a numerical model that included preferential coupling of Cav1.3 channels to RYR-dependent Ca2+ release suggests that VDF can promote normal SAN pacemaking. We conclude that differences in VDF and subcellular localization further contribute to the non-overlapping roles of Cav1.2 and Cav1.3 in controlling sinus rhythm.

Voltage-dependent facilitation of Cav1.3 in sinoatrial node cells

Cav1 channels undergo frequency-dependent facilitation in multiple cardiac tissues (Lee, 1987; Richard et al. 1990; Mangoni et al. 2000). The resulting increase in ICa,L is due in part to CDF. For Cav1.2, Ca2+ influx augments channel open probability through calmodulin (CaM)-dependent protein kinase II (CaMKII) (Dzhura et al. 2000; Grueter et al. 2006). VDF of Cav1.2 can also involve CaMKII, but unlike CDF, is seen when Ba2+ is used as the permeant ion (Lee et al. 2006). While both CDF and VDF have been characterized for Cav1.2 in ventricular myocytes, little is known regarding these processes for Cav1.3 in the heart. We focused on IBa rather than ICa, as our goal was to compare VDF of SAN Cav1 currents. With Ca2+ as the charge carrier, L-type currents undergo CDI, due to CaM, as well as CDF (see Christel & Lee, 2012 for review), both of which would hinder analyses of VDF. Because Ba2+ substitutes poorly for Ca2+ in binding to CaM (Wang, 1985), voltage-dependent modulation of IBa can be studied in isolation from Ca2+-dependent modulation.

Using SAN cells from Cav1.2 DHP mice, we found that VDF for Cav1.3 IBa,L is largely consistent with previously described VDF of ICa,L in rabbit SAN cells (Mangoni et al. 2000). Because it was not possible to completely isolate Cav1.3 VDF from VDI, VDF was only evident following very strong depolarizations (+80 mV; Fig. 4). Our interpretation that VDF is partially obscured by VDI at more negative prepulse voltages is supported by the overt facilitation of Cav1.3 currents seen at longer intervals after the conditioning prepulse (Fig. 3), assuming that recovery is faster for VDI than VDF. In addition, VDF of ICa,L was apparent in simulations of voltage-clamp recordings at less positive voltages (+30 mV), compatible with native SAN APs (Fig. 6) and simulated SAN pacemaker activity (Figs 7 and 8).

In comparing the total IBa,L to the individual Cav1.3 and Cav1.2 components, it is clear that one function of Cav1.3 VDF is to balance the strong VDI of Cav1.2 in SAN cells (Figs 3 and 4). Cav1.2 currents showed significant VDI following depolarizing prepulses (∼30%), which was offset (∼10%) by the limited VDI and stronger VDF of Cav1.3 currents (Fig. 3). Cav1.3 VDF also allowed full recovery of IBa,L within 0.5 s after a strong depolarization (Fig. 4), thus stabilizing IBa,L from reductions due to Cav1.2 VDI.

Our finding that Cav1.3 but not Cav1.2 undergoes overt VDF during recovery from inactivation (Fig. 2B) was unexpected considering that VDF is also a property of Cav1.2 in transfected cells (Lee et al. 2006). VDF of SAN IBa,L may involve molecular determinants and protein interactions specific for Cav1.3. For Cav1.3, the distal C-terminal domain of the long exon 42-containing α1 subunit contains a type I PDZ binding sequence and autoinhibitory module that suppresses VDF. Binding to PDZ domain-containing proteins such as erbin relieves autoinhibition, resulting in robust VDF in transfected HEK293T cells. Cav1.2 channels are not subject to such regulation by PDZ domain-containing proteins (Calin-Jageman et al. 2007). Whether interactions with erbin or related proteins underlie Cav1.3-selective VDF in SAN cells remains to be elucidated.

Cav1.3/ryanodine receptor apposition in sinoatrial node cells

The strong colocalization of Cav1.3 with RYR2 (Fig. 5) in SAN cells may be relevant for the functional role of RYR-mediated Ca2+ release in pacemaking (Vinogradova et al. 2002). During the late phase of diastolic depolarization, RYR-mediated Ca2+ release promotes activation of a depolarizing current due to Na+/Ca2+ exchange, which accelerates reaching the threshold of the SAN AP upstroke. Close apposition of Cav1.3 with RYRs may facilitate SR Ca2+ release as ICa,L stimulates RYR open probability. In this respect, numerical simulations predicted that the slope of rise of diastolic RYR-dependent Ca2+ release increased as a function of Cav1.3-mediated ICa,L half-activation voltage (data not shown). The coupling of this SR Ca2+ release to the depolarizing influence of Na+/Ca2+ exchange should accelerate attainment of the threshold for AP firing of SAN cells (Vinogradova et al. 2002).

Cav1.3 channels have been shown to physically associate with RYR2 in the nervous system (Ouardouz et al. 2003; Kim et al. 2007). However, in neurons, Cav1.3 activation promotes RYR2-mediated Ca2+ release independent of Ca2+ influx, suggesting that Cav1.3 channels may trigger voltage-dependent activation of RYR2 analogous to Cav1.1 channels and skeletal RYR1 (Paolini et al. 2004). Similar characterization of Cav1.3/RYR interactions will reveal additional insights as to how Ca2+-induced Ca2+ release affects automaticity in SAN cells.

Cav1.3 and arrhythmia

The importance of Cav1.3 for normal cardiac rhythmicity is now well established. Loss-of-function mutations in the human CACNA1D gene encoding Cav1.3 cause SAN dysfunction and deafness syndrome (SANDD) (Baig et al. 2011). Patients with SANDD present with profound deafness and sinus bradycardia, similar to mice with genetic inactivation of CACNA1D (Platzer et al. 2000). Patients with SANDD also exhibit atrioventricular block and dissociated rhythms, which can be explained by findings that Cav1.3 is necessary for automaticity of atrioventricular nodal cells (Marger et al. 2011). Human mutations in the gene encoding the multifunctional scaffolding protein, ankyrin B, cause reductions in Cav1.3 current density in SAN cells and atrial myocytes, which is associated with sinus bradycardia and atrial fibrillation, respectively (Le Scouarnec et al. 2008; Cunha et al. 2011). With respect to defects in Cav1.3 causing SAN dysfunction, loss of rapid activation kinetics and negative activation thresholds of Cav1.3 weakens the diastolic depolarization, thus slowing pacemaking (Mangoni et al. 2003). By experimental and computational approaches, we establish that VDF and potential coupling with RYRs are additional features of Cav1.3 that support its role in the generation of heart rhythm.

Appendix

Author contributions

The experiments in this study were performed at the University of Iowa, University of Montpellier, and University of Erlangen. Specific contributions are as follows: conception and design of experiments: C.J.C., P.M., S.H., M.E.M. and Amy Lee; collection, analysis, and interpretation of data: C.J.C., N.C., P.M., S.H., M.E.M. and A.L.; drafting the article or revising it critically for important intellectual content: C.J.C., M.E.M., S.H., A.L., J.S. and A.L. All authors approved the final version of the manuscript.

Acknowledgements

Supported by the NIH (DC009433, HL087120 to A.L.; T32007121 to C.J.C.), the Carver Research Program of Excellence (to A.L.), Deutsche Forschungsgemeinschaft (GRK 333 to A. Ludwig), Austrian Science Fund (P20670 to J.S.), ANR-2010-BLAN-1128-01 and ANR-09-GENO-034 (to M.E.M.). The IGF group is a member of the Laboratory of Excellence ‘Ion Channel Science and Therapeutics’ supported by a grant from ANR. The authors thank Dr Yuejin Wu for advice with SAN cell recording and Dr Mark Anderson for comments on the manuscript.

Ancillary