We began our study by expressing Ca_V1.2 channels in tsA-201 cells and recording elementary Ca_V1.2 currents from cell-attached patches. Currents were elicited with a step depolarization to −30 mV with Ba²⁺ or Ca²⁺ as the charge carrier (Figure 1A,B). The mean amplitudes of i_Ca and i_Ba were 0.50 ± 0.02 (n = 6) and 1.45 ± 0.01 pA (n = 6), respectively. All-points histograms revealed that multi-channel openings were more likely with Ca²⁺ than with Ba²⁺. Accordingly, the activity (nP_o), defined as the number of channels (n) times the open probability (P_o), of Ca_V1.2 channels within a patch, was significantly higher with Ca²⁺ (0.24 ± 0.10) than Ba²⁺ (0.02 ± 0.01; Figure 1C). Closer inspection of the multi-channel openings revealed that, with Ca²⁺ as the charge carrier, multiple channels frequently opened together instantaneously and subsequently closed together. For example, in the enlarged trace in Figure 1B, four channels opened simultaneously, followed by the opening of four additional channels. The eight channels remained open for a time (∼11.5 ms), then all closed simultaneously. This apparent coordinate gating of multiple Ca_V1.2 channels was not observed when Ba²⁺ was used as the charge carrier. These results challenge the long-held and generally accepted view that individual Ca_V1.2 channels gate independently, and instead strongly suggest that these channels frequently exhibit ‘cooperative gating’. An additional implication is that Ca²⁺ itself increases the probability of cooperative Ca_V1.2 channel gating.

Figure 1

Download asset Open asset

To test whether this hypothesis holds true in primary cells, we recorded Ca_V1.2 Ca²⁺ sparklet activity in freshly isolated adult ventricular myocytes (Figure 1D and Video 1). A multi-Gaussian fit to the all-points histogram of the calibrated Ca²⁺ signal obtained with 20 mM [Ca²⁺]_o revealed quantal amplitudes of 36.8 ± 4.2 nM (Figure 1E), in excellent agreement with the ∼38 nM reported previously for Ca_V1.2 Ca²⁺ sparklets in arterial smooth muscle cells and tsA-201 cells (Navedo et al., 2005, 2006). Consistent with our single-channel data, we frequently observed multi-quantal Ca²⁺ sparklets corresponding to the simultaneous opening of several Ca_V1.2 channels. We next applied a coupled Markov chain model to determine whether these Ca²⁺-influx events were solely attributable to stochastic, independently gating Ca_V1.2 channels or instead reflected cooperative channel gating. The model assigns a coupling coefficient (κ) value to each Ca²⁺ sparklet site, ranging from 0 for independently gating channels to 1 for channels that gate exclusively in a cooperative manner (Chung and Kennedy, 1996; Navedo et al., 2010). Using κ > 0.1 as a threshold for cooperative gating (Navedo et al., 2010), we found that the majority of Ca²⁺ sparklet sites displayed cooperative or ‘coupled’ gating behavior, consistent with our hypothesis. Together, these data suggest that functional coupling of Ca_V1.2 channels is a Ca²⁺-dependent phenomenon.

Video 1

Download asset

If the trigger for Ca_V1.2 channel coupling were a local elevation in [Ca²⁺]_i resulting from the opening of a single channel, then the efficacy of this signal in recruiting a nearby channel would be critically dependent on the distance separating the channels. Using super-resolution nanoscopy (Folling et al., 2008), we examined the spatial organization of endogenous Ca_V1.2 channels in ventricular myocytes (Figure 2A–C) and heterologously expressed Ca_V1.2 channels in tsA-201 cells (Figure 3A–C). Our data clearly show that Ca_V1.2 channels formed clusters along the sarcolemmal Z-lines of ventricular myocytes (Figure 2A,B) and throughout the plasma membrane (PM) of tsA-201 cells (Figure 3A,B). The average area occupied by a Ca_V1.2 channel cluster was 2555 ± 82 nm² in ventricular myocytes (n = 5; Figure 2C) and 2190 ± 20 nm² in tsA-201 cells (n = 9; Figure 3C).

Figure 2

Download asset Open asset

Figure 3 with 1 supplement see all

Download asset Open asset

Since Ca_V1.2 channel clusters were localized to the t-tubule regions of ventricular myocytes where the junctional SR (jSR) comes into close apposition to the myocyte PM, one might predict that the channel clusters would similarly localize to PM-adjacent ER structures in tsA-201 cells. To test this idea, we co-expressed mCherry-sec61β (a general ER marker (Zurek et al., 2011)) in tsA-201 cells together with Ca_V1.2 channels. Super-resolution imaging revealed that Ca_V1.2 channel cluster distribution was not restricted to regions of the surface membrane where the ER was located (Figure 3—figure supplement 1A). Instead, these channels were broadly distributed along the PM surface. It is possible that the lack of organization of Ca_V1.2 channels along ER junctions in tsA-201 cells reflects a lack of contact points or excessive distance between the ER and the PM. In ventricular myocytes, the membrane binding protein junctophilin-2 (JPH2) has been suggested to tether the jSR to the t-tubule membrane (Takeshima et al., 2000). Thus, we attempted to anchor the ER to the PM by co-transfecting tsA-201 cells with JPH2. However, even in the presence of JPH2, the Ca_V1.2 channel cluster distribution was not limited to PM-ER junctions in the manner that they are in cardiomyocytes (Figure 3—figure supplement 1B). These data suggest that Ca_V1.2 channel clustering occurs independently of SR/ER microdomains.

To determine the number of channels within Ca_V1.2 clusters, we injected mice with an adeno-associated virus serotype 9 (AAV9) designed to express photo-activatable-GFP–tagged Ca_vβ_2a, and examined ventricular myocyte Ca_V1.2 clusters 5 wk later using single-particle photobleaching (Ulbrich and Isacoff, 2007). Ca_Vβ_2a is a palmitoylated peripheral membrane protein that binds to the α₁ pore-forming subunit of Ca_V1.2 with a 1:1 stoichiometry (Dalton et al., 2005); therefore, photo-activation of this protein with 405-nm light provides a fluorescent marker of Ca_V1.2 channels. Single Ca_V1.2 were identified and excited using total internal reflection fluorescence (TIRF) microscopy. The number of channels in each cluster was determined by continuous photobleaching and counting of stepwise decreases in fluorescence intensity (Figure 2D–F). A preponderance (47%) of Ca_V1.2 clusters displayed 1 to 6 stepwise decreases in fluorescence (Figure 2F). A single photobleaching step was observed in only 1% of the spots analyzed. Indeed, the mean number of bleaching steps per cluster was 7.91 ± 0.23 (n = 25). This suggests that Ca_V1.2 channels preferentially cluster in groups of about 8 channels in adult ventricular myocytes. We did not discriminate cluster size based on location; thus, our estimate of 8 channels/cluster combines dyadic and extra-dyadic populations.

Similar stepwise photobleaching experiments were performed on enhanced green fluorescent protein (EGFP)-tagged Ca_V1.2 channels heterologously expressed in tsA-201 cells (Figure 3D–F). TIRF imaging showed that Ca_V1.2 clusters displayed a mean of 5.07 ± 0.15 (n = 10) discrete bleaching steps. Taken together with the data from ventricular myocytes, these findings suggest that the formation of multi-channel clusters is a fundamental property of Ca_V1.2 channels with important implications for Ca²⁺ signaling.

To investigate the mechanisms regulating Ca_V1.2-Ca_V1.2 interactions in living cells (Figure 4), we applied a bimolecular fluorescence complementation approach using Ca_V1.2 channels fused with either the N- or C-terminus of the split-Venus fluorescent protein system to yield Ca_V1.2-VN155(I152L) and Ca_V1.2-VC155, respectively. In isolation, VN155(I152L) and VC155 are non-fluorescent; however, when brought into close proximity by interacting proteins, they can reconstitute a full, fluorescent Venus protein. Thus, Venus fluorescence can be used to report spontaneous interactions between adjacent Ca_V1.2 channels, as depicted in Figure 4A. In cells expressing Ca_V1.2-VN155(I152L) and Ca_V1.2-VC155 channels, Venus fluorescence at −80 mV was very low, suggesting that Ca_V1.2-Ca_V1.2 channel interactions are rare at this hyperpolarized membrane potential. To determine whether an increase in [Ca²⁺]_i is sufficient to induce Ca_V1.2-Ca_V1.2 channel interactions, we loaded tsA-201 cells with DMNP-EDTA (caged Ca²⁺) via the patch pipette and induced flash photolysis of DMNP-EDTA with pulses of 405-nm light while holding cells at −80 mV. Photolysis of DMNP-EDTA induced a transient increase in [Ca²⁺]_i and a concomitant increase in Venus fluorescence (Figure 4—figure supplement 1), demonstrating that an elevation in [Ca²⁺]_i is indeed sufficient to induce Ca_V1.2-Ca_V1.2 interactions.

Figure 4 with 5 supplements see all

Download asset Open asset

To determine if Ca²⁺ influx via Ca_V1.2 channels is required for channel interactions, we depolarized cells and recorded Venus fluorescence and membrane currents in the presence of Ba²⁺ or Ca²⁺. Cells were dialyzed with an intracellular solution containing 10 mM EGTA to restrict the local [Ca²⁺]_i signal to about 1 μm from the channel and maintain very low global [Ca²⁺]_i. With Ba²⁺ in the external solution, depolarization evoked currents (I_Ba) over a wide range of potentials, but Venus fluorescence was very low at all membrane potentials (Figure 4B,F,G and Figure 4—figure supplement 2A–C). After switching to a Ca²⁺-containing external solution, application of the same voltage protocol activated currents (I_Ca) and induced graded increases in Venus fluorescence (Figure 4C,F,G and Figure 4—figure supplement 2D–F). The fluorescence-voltage and I_Ca conductance-voltage relationships were sigmoidal. The normalized conductance and Venus fluorescence exhibited similar voltage dependencies (Figure 4—figure supplement 2E,F). Taken together with super-resolution, photobleaching and DMNP-EDTA data, these findings suggest that local and global [Ca²⁺]_i signals produced by Ca²⁺ influx via Ca_V1.2 channels are required for physical interactions between adjacent channels within a cluster.

We next investigated the mechanisms underlying Ca²⁺-dependent coupling of Ca_V1.2 channels, focusing on CaM since this protein binds Ca²⁺, associates with Ca_V1.2 C-terminal pre-IQ and IQ domains, and is involved in CDI and CDF of Ca_V1.2 channels (Eckert and Tillotson, 1981). Dialysis with an intracellular solution containing the CaM inhibitory peptide MLCKp (0.1 μM) prevented Venus reconstitution during membrane depolarization (Figure 4D,F,G), without altering the rate of inactivation of I_Ca (Figure 4—figure supplement 3). Expression of a mutant CaM that does not bind Ca²⁺ (CaM₁₂₃₄) also prevented Ca_V1.2-VN-Ca_V1.2-VC fusion upon membrane depolarization (Figure 4E,F,G). However, unlike MLCKp, CaM₁₂₃₄ did slow the rate of I_Ca inactivation (Figure 4—figure supplement 3), suggesting that two functionally distinct CaM molecules are involved in CDI and Ca_V1.2-to-Ca_V1.2 channel interactions.

Given the apparently essential role of Ca²⁺/CaM, we next examined the importance of IQ and pre-IQ motif CaM-binding sites for Ca_V1.2-Ca_V1.2 channel interactions. To do so, we mutated critical residues in each segment and used bimolecular fluorescence complementation to evaluate changes in channel coupling ability. Ca_V1.2-VN and Ca_V1.2-VC channels with an I1654E mutation in their IQ domains were able to fuse upon membrane depolarization (Figure 4—figure supplement 4). This isoleucine residue is crucial for CaM binding to the IQ motif. Previous in vitro studies have reported that the I1654E (or the human homolog I1624E) mutation decreases the affinity of the IQ motif for CaM by ∼100-fold (Zühlke et al., 1999, 2000). Thus, our results suggest that CaM binding to the IQ-motif is not required for Ca_V1.2 channel coupling. To investigate the role of the pre-IQ motif in channel interactions, we exchanged a 33-amino-acid segment of the pre-IQ domain for 33 non-identical amino acids (Figure 4—figure supplement 4A). A similar segment exchange performed on human Ca_V1.2 channels expressed in tsA-201 cells was previously shown to impair channel clustering on the PM and reduce P_o compared to wild-type (WT) channels (Kepplinger et al., 2000). Interestingly, the pre-IQ ‘swap’ mutation rendered Ca_V1.2-VN and Ca_V1.2-VC channels incapable of functional coupling (Figure 4—figure supplement 4C). Collectively, these results suggest that pre-IQ domains are required for Ca²⁺/CaM-dependent Ca_V1.2 channel oligomerization. Additional credence is given to this finding by the previously published crystal structure of dimeric cardiac L-type Ca²⁺ channels showing two pre-IQ helices bridged by two Ca²⁺/CaMs (Fallon et al., 2009).

We next examined the spatial distribution of Ca_V1.2(pre-IQ swap) channels in tsA-201 cells co-transfected with mCherry-sec61β (to permit visualization of the ER). Surprisingly, Ca_V1.2(pre-IQ swap) channels still formed clusters in tsA-201 cells despite their inability to functionally interact (Figure 4—figure supplement 5A). Indeed, under identical imaging conditions (i.e., using the same fixative and TIRF penetration depth), Ca_V1.2(pre-IQ swap) channel cluster areas were not significantly different from those of WT channels. As noted above for WT channels, the Ca_V1.2(pre-IQ swap) channel cluster distribution was not limited to PM-ER junctions, even in the presence of JPH2 (Figure 4—figure supplement 5A,B). These results suggest that, while the physical proximity of Ca_v1.2 channels is necessary for channel interactions, it is not sufficient for functional coupling of the channels.

An important prediction of our findings is that the functional coupling produced by Ca²⁺-induced Ca_V1.2-Ca_V1.2 channel interactions manifests as enhanced channel activity. To test this, we exploited the fact that Venus reconstitution is irreversible, recording Ca_V1.2 sparklets (Wang et al., 2001; Navedo et al., 2005) before and after Ca²⁺-induced Ca_V1.2 coupling during membrane depolarization. Prior to membrane depolarization, Ca_V1.2 sparklet activity (nP_s) was low (0.002 ± 0.001). However, after depolarization, nP_s increased ∼40-fold (0.085 ± 0.022) and Ca_V1.2 sparklet density increased almost 10-fold (Figure 5A,C,D and Video 2; n = 5 cells). Membrane depolarization failed to significantly alter nP_s or Ca_V1.2 sparklet density in cells co-expressing the Ca²⁺-insensitive CaM₁₂₃₄ mutant (Figure 5B–D; n = 6 cells). These data indicate that the post-depolarization augmentation of Ca_V1.2 sparklet activity resulted from a Ca²⁺/CaM-dependent increase in the activity of previously active sites as well as the emergence of new, high-activity Ca²⁺ sparklet sites that appear to lack CDI. These findings suggest that Ca_V1.2 channel interactions increase Ca²⁺ influx and, consequently, total conductance by increasing the extent to which the channels are functionally coupled.

Figure 5

Download asset Open asset

Video 2

Download asset

Having demonstrated that physical Ca_V1.2-to-Ca_V1.2 interactions functionally couple adjoining channels to enhance channel activity, we investigated the dynamics of these interactions. Bimolecular fluorescence complementation experiments are unable to provide information about channel-interaction dynamics, since Venus reconstitution is irreversible (Kerppola, 2006). Instead, we sought to detect fluorescence resonance energy transfer (FRET) between EGFP and red fluorescent protein (RFP)-tagged Ca_V1.2 channels as a function of [Ca²⁺]_i. Flash photolysis of DMNP-EDTA (at −80 mV) induced a transient increase in [Ca²⁺]_i and enhanced the Ca_V1.2-RFP/Ca_V1.2-EGFP FRET ratio (FRETr; Figure 6A). The time courses of [Ca²⁺]_i and FRETr were similar (Figure 6B), with a time-to-peak of 605.8 ± 98.8 ms for [Ca²⁺]_i and 531.5 ± 57.5 ms for FRETr. The [Ca²⁺]_i-FRETr relationship was sigmoidal, yielding a FRETr_1/2 at a [Ca²⁺]_i of approximately 250 nM (Figure 6C). Both [Ca²⁺]_i and FRETr traces followed triple exponential decay kinetics; for [Ca²⁺]_i, the decay time constants (τ) for fast, intermediate and slow components were 0.62 ± 0.15, 2.19 ± 0.40 and 3.15 ± 0.26 s, respectively, and the corresponding values for FRETr were 0.64 ± 0.48, 1.09 ± 0.97 and 3.71 ± 0.73 s (n = 5). The time to decay to 50% of the peak (T_50%) was 2.65 ± 0.22 s for [Ca²⁺]_i and 0.63 ± 0.23 s for FRETr. The complex multi-component decay kinetics of each trace reflects the multiple Ca²⁺ binding sites and affinities of the two lobes of CaM for Ca²⁺ (Faas et al., 2011; Faas and Mody, 2012).

Figure 6

Download asset Open asset

Concurrent recordings of currents and FRETr during membrane depolarization in the presence of Ca²⁺ (with 10 mM EGTA in the pipette solution) showed that depolarization to +10 mV evoked a transient increase in FRETr (Figure 6D). The increase in FRETr appeared biphasic, with a large amplitude spike followed by a lower level plateau that was sustained for the duration of the increase in [Ca²⁺]_i. These data suggest that Ca_V1.2-Ca_V1.2 interactions are dynamic and regulated by local and global changes in [Ca²⁺]_i. Notably, the increase in FRETr persisted after the Ca²⁺ current had decayed to baseline, indicating that channels remained coupled for a time in the absence of a stimulus.

An explicit implication of our results is that physical Ca_V1.2-Ca_V1.2 interactions are critical for CDF in ventricular myocytes. To test this hypothesis, we investigated whether Ba²⁺ permeation and inhibition of CaM with MLCKp—both of which prevent Ca_V1.2-Ca_V1.2 interaction (see above)—decreases or eliminates CDF in ventricular myocytes. Because Ca²⁺/CaM-dependent kinase II (CaMKII) has been implicated in Ca²⁺ current facilitation (Anderson et al., 1994; Xiao et al., 1994; Yuan and Bers, 1994), we included 10 mM EGTA in the patch pipette to maintain global [Ca²⁺]_i < 50 nM, which is below the threshold (>200 nM) for activation of this kinase (Miller and Kennedy, 1986). Experiments were performed in cells dialyzed with 100 nM or 1 μM MLCKp. The rationale for using these two concentrations of MLCKp is that, whereas 100 nM MLCKp inhibits CaM (Török and Trentham, 1994; Török et al., 1998), we found that it does not change CaMKII activity (p < 0.05). However, we found that 1 μM MLCKp inhibits CaM and eliminates CaMKII activity. Thus, using these two MLCKp concentrations, we can selectively dissect the contribution of CaM and locally activated (i.e., near the channel pore) CaMKII.

We used a three-step protocol (Figure 7A) to record facilitated Ca_V1.2 currents as previously described (Poomvanicha et al., 2011). Briefly, cells were held at −80 mV and I_Ca was elicited with a 200-ms control pulse (V₁) to 0 mV. Cells were held again at −80 mV for 10 s followed by a 200-ms prepulse to +80 mV (V_pre). In light of our FRET results suggesting that Ca_V1.2 channels remain coupled for a time in the absence of a stimulus, we held the cells at −80 mV for a variable interpulse interval of 0.1–1.6 s before beginning the second 200-ms test pulse (V₂) to 0 mV. Fast Na⁺ currents were inactivated with a 50-ms step to −40 mV applied prior to each control (V₁) or test pulse (V₂). The ratio of I_Ca resulting from test and control pulses (I₂/I₁) was used as a measure of facilitation (I₂/I₁ > 1) and recovery from inactivation (I₂/I₁ ≈ 1).

Figure 7

Download asset Open asset

In control cells (no peptide, 2 mM external Ca²⁺), this protocol induced I_Ca facilitation (i.e., I₂/I₁ > 1) (Figure 7B). Note, however, that the magnitude of I_Ca facilitation varied with interpulse duration, reaching a peak at a V_pre-V₂ interval of 0.3 s. Longer V_pre-V₂ intervals induced progressively less I_Ca facilitation. Indeed, I₂/I₁ was statistically indistinguishable from 1 at interpulse durations longer than 1.2 s, suggesting no I_Ca facilitation at stimulation rates >0.8 Hz.

In the presence of extracellular Ca²⁺, intracellular dialysis with 0.1 μM MLCKp suppressed I_Ca facilitation at interpulse intervals of 0.1–0.3 s (n = 10 for 0.1 s intervals; n = 5 for 0.2 and 0.3 s intervals), but control levels of facilitation returned with intervals of 0.4–1 s (n = 5–9, p < 0.05; Figure 7B). Dialysis with a 10-fold higher concentration of MLCKp (1 μM) eliminated I_Ca facilitation at all interpulse intervals (n = 4; Figure 7B). Finally, superfusion of ventricular myocytes with Ba²⁺ was as effective as dialysis with 1 μM MLCKp in preventing I_Ca facilitation (Figure 7B–D).

We performed a detailed analysis of our data to gain more insights into the relationship between Ca_V1.2 channel coupling and I_Ca facilitation during high frequency stimulation. The amplitude of I₂ in the paired pulse facilitation would depend, in part, on the degree of recovery from inactivation. A previous study has determined that the time course of I_Ca recovery from inactivation follows a single exponential function with a τ_recovery of about 375 ms (Blaich et al., 2010). Our FRET data in Figure 6D suggest that the number of coupled Ca_V1.2 channels will fade exponentially with an estimated τ_decoupling of 333 ms upon repolarization. In Figure 7E (left), we show a simulation of the time-course of I_Ca recovery and Ca_V1.2 channel decoupling (middle) during repolarization using exponential functions with these τ_decoupling and τ_recovery values. Note that the time-course of I_Ca facilitation data obtained from myocytes superfused with Ca²⁺ is well described by the product of the recovery and decoupling functions (Figure 7E; right). This analysis is consistent with a model in which I_Ca facilitation during high frequency stimulation is directly proportional to the number of coupled channels and the number of channels available for activation.

These data support the view that I_Ca facilitation in ventricular myocytes depends on Ca²⁺ influx and CaM, but has CaMKII-dependent and independent components. In combination with our FRET and split-Venus data, these findings suggest that Ca²⁺ augments Ca_V1.2 channel activity at least in part by increasing the number of functionally coupled channels.

WT C57BL/6 mice were euthanized with a single lethal dose of sodium pentobarbital delivered via intraperitoneal injection, as approved by the University of Washington Institutional Animal Care and Use Committee (IACUC). Ventricular myocytes were isolated as previously described (Dixon et al., 2012). Briefly, the heart was excised and rinsed with cold 150 μM EGTA digestion buffer containing 130 mM NaCl, 5 mM KCl, 3 mM Na-pyruvate, 25 mM HEPES, 0.5 mM MgCl₂, 0.33 mM NaH₂PO₄, and 22 mM glucose. The aorta was cannulated for Langendorff perfusion, and the coronary arteries were subsequently perfused with warmed (37°C) 150 μM EGTA digestion buffer until they were cleared of blood. The perfusate was then switched to digestion buffer (no EGTA) supplemented with 50 μM CaCl₂, 0.04 mg/ml protease (XIV), and 1.4 mg/ml collagenase (type 2; Worthington Biochemical, Lakewood, NJ) for 8–10 min. The ventricles were then cut away from the atria, sliced, and placed in 37°C digestion buffer supplemented with 0.96 mg/ml collagenase, 0.04 mg/ml protease, 100 μM CaCl₂, and 10 mg/ml bovine serum albumen (BSA). Gentle agitation was applied using a transfer pipette until the ventricles dissociated. The cells were then allowed to pellet by gravity for 15–20 min, after which they were washed in enzyme-free digestion buffer supplemented with 10 mg/ml BSA and 250 μM CaCl₂, pelleted once more, and finally resuspended at room temperature in Tyrode's solution containing 140 mM NaCl, 5 mM KCl, 10 mM HEPES, 10 mM glucose, 2 mM CaCl₂, and 1 mM MgCl₂; pH was adjusted to 7.4 with NaOH.

tsA-201 cells (Sigma–Aldrich, St. Louis, MO) were cultured in Dulbecco's Modified Eagle Medium (DMEM; Gibco-Life Technologies, Grand Island, NY) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin at 37°C in a humidified 5% CO₂ atmosphere and passaged every 3–4 day. Cells were transiently transfected at ∼70% confluence using jetPEI (Polyplus Transfection, New York, NY) transfection reagent and plated onto the appropriate coverglass ∼12 hr before experiments. For super-resolution imaging and photobleaching experiments, cells were plated onto 22 mm no. 1.5 coverslips (Thermo Fisher Scientific, Waltham, MA). For all other experiments, cells were plated onto 25 mm no. 1 coverslips (Thermo Fisher Scientific). Plasmids used in this study include pcDNA clones of the pore-forming subunit of rabbit Ca_V1.2 (α_1c) and rat auxillary subunits Ca_Vα₂δ and Ca_Vβ₃ (kindly provided by Dr Diane Lipscombe; Brown University, Providence, RI). Standard PCR techniques were used to fuse the carboxyl tail of Ca_V1.2 to different proteins depending on the experimental approach: for FRET experiments, to EGFP or tagRFP; for bimolecular fluorescence complementation, to either the N-fragment (VN) or the C-fragment (VC) of the Venus protein (27097, 22011; Addgene, Cambridge, MA) (Kodama and Hu, 2010); for photobleaching experiments, to the monomeric GFP variant GFP(A206K) (Zacharias et al., 2002), kindly provided by Dr Eric Goaux (Vollum Institute, Portland, OR). Ca²⁺-insensitive CaM₁₂₃₄ was a gift from Dr Johannes Hell (UC Davis, CA). Mutant rabbit Ca_V1.2(I1654E), analogous to human I1624E (Zühlke et al., 1999), was used in experiments designed to elucidate the role of the IQ motif in channel interactions. This single point mutation (I1672E) was introduced using the QuikChange II XL Site-Directed Mutagenesis kit (Agilent Technologies, Santa Clara, CA). The role of the pre-IQ motif in channel interactions was examined by exchanging a 33-amino-acid segment, as illustrated in Figure 4—figure supplement 4A. The resultant Ca_V1.2(pre-IQ swap) was used in whole-cell patch-clamp and bimolecular fluorescence complementation experiments. The general ER marker mCherry-Sec61β was a gift from Gia Voeltz (Addgene plasmid # 49155). Finally, JPH2-GFP was used to tether the ER to the PM in tsA-201 cells.

Ca²⁺ currents were recorded in the whole-cell voltage-clamp or cell-attached patch configurations using borosilicate patch pipettes with resistances of 3–6 μΩ for tsA cells and 2–3 μΩ for cardiomyocytes. Currents were sampled at a frequency of 20 kHz, low-pass–filtered at 2 kHz using an Axopatch 200B amplifier, and acquired using pClamp 10.2 software (Molecular Devices, Sunnyvale, CA). All membrane potentials referred to herein have been corrected for liquid junction potential. All experiments were performed at room temperature (22–25°C).

For whole-cell current recordings from tsA-201 cells, pipettes were filled with a Cs-based internal solution containing 87 mM Cs-aspartate, 20 mM CsCl, 1 mM MgCl₂, 10 mM HEPES, 10 mM EGTA and 5 mM MgATP, adjusted to pH 7.2 with CsOH. In experiments requiring dialysis of the calmodulin inhibitory peptide MLCKp (EMD Millipore, Darmstadt, Germany; 208735), this chemical was added to the pipette solution. Cells were continuously superfused throughout experiments with our regular external solution containing 5 mM CsCl, 10 mM HEPES, 10 mM glucose, 113 mM NMDG, 1 mM MgCl₂ and 20 mM CaCl₂, adjusted to pH 7.4 with HCl. For experiments in which Ba²⁺ was used as the charge carrier in place of Ca²⁺, the perfusate contained 5 mM CsCl, 10 mM HEPES, 10 mM glucose, 140 mM NMDG, 1 mM MgCl₂ and 2 mM BaCl₂, adjusted to pH 7.4 with HCl. Current-voltage relationships were obtained by subjecting cells to a series of 300-ms depolarizing pulses from a holding potential of −80 mV to test potentials ranging from −60 to +80 mV. The voltage dependence of conductance was obtained by converting the resultant currents to conductances using the equation, G = I_Ca/[test pulse potential − reversal potential of I_Ca], normalizing (G/Gmax), and plotting conductance vs the test potential.

Cell-attached patch, single-channel currents (i_Ca) were recorded from tsA-201 cells superfused with high K⁺ solution to fix the membrane potential at ∼0 mV. The solution had the following composition: 145 mM KCl, 2 mM MgCl₂, 0.1 mM CaCl₂, 10 mM HEPES and 10 mM glucose; pH was adjusted to 7.3 with KOH. Pipettes were filled with a solution containing 10 mM HEPES and either 110 mM CaCl₂ or 110 mM BaCl₂; pH was adjusted to 7.2 with CsOH. The dihydropyridine agonist BayK-8644 (500 nM) was included in the pipette solution to promote longer channel open times. A voltage-step protocol from a holding potential of −80 mV to a depolarized potential of −30 mV was used to elicit currents. The single-channel event-detection algorithm of pClamp 10.2 was used to measure single-channel opening amplitudes and nP_o, and to construct all-points histograms.

To record I_Ca from isolated ventricular myocytes, cells were initially perfused with Tyrode's solution. Once the whole-cell configuration was successfully established, the external solution was replaced with one containing 5 mM CsCl, 10 mM HEPES, 10 mM glucose, 140 mM NMDG, 1 mM MgCl₂ and either 2 mM CaCl₂ or 2 mM BaCl₂, adjusted to pH 7.4 with HCl. The pipette was filled with the Cs-based internal solution described above. Facilitation of I_Ca was measured using a triple-pulse protocol consisting of two identical test pulses (V₁ and V₂), separated by a conditioning pulse to +80 mV (V_pre), as previously described (Poomvanicha et al., 2011) and illustrated in Figure 7A. The currents elicited by V₁ and V₂ were referred to as I₁ and I₂, and the ratio between their peaks (I₂/I₁) was used as a measure of facilitation.

To measure Ca²⁺ sparklets in tsA-201 cells or ventricular myocytes, cells were patch-clamped in whole-cell mode and held at a hyperpolarized potential of −80 mV to increase the driving force for Ca²⁺ entry. Sub-sarcolemmal Ca²⁺ signals were monitored by dialyzing cells with 200 μM Rhod-2 via the patch pipette and continuously perfusing them with the external solution described above containing 20 mM CaCl₂ and 10 mM EGTA, a relatively slow Ca²⁺ buffer. With this buffer/indicator combination, Ca²⁺ entering the cell via membrane Ca_v1.2 channels binds to the relatively fast Ca²⁺ indicator Rhod-2 to generate a fluorescent signal, and the excess EGTA rapidly buffers Ca²⁺, restricting the signal to the point of entry. Sub-sarcolemmal Ca²⁺ signals (Ca²⁺ sparklets) were captured using a through-the-lens TIRF microscope built around an Olympus IX-70 inverted microscope equipped with an oil-immersion ApoN 60×/1.49 NA TIRF objective and an Andor iXON CCD camera. Images were acquired at 100 Hz using TILLvisION imaging software (TILL Photonics, FEI, Hillsboro, OR). Sparklets were detected and analyzed using custom software written in MATLAB (Source code 1). Rhod-2 fluorescence signals were converted to Ca²⁺ concentration units using the F_max equation (Maravall et al., 2000). The activity of Ca²⁺ sparklets was determined by calculating the nP_s of each Ca²⁺ sparklet site, where n is the number of quantal levels and P_s is the probability that a quantal Ca²⁺ sparklet event is active. A detailed description of this analysis can be found in Navedo et al. (Navedo et al., 2005, 2006).

The degree of coupling between single Ca_V1.2 channels or Ca²⁺ sparklet sites was assessed by further analyzing single-channel and sparklet recordings using a binary coupled Markov chain model (Source code 2), as first described by Chung and Kennedy (1996) and previously employed by our group (Navedo et al., 2010; Cheng et al., 2011; Dixon et al., 2012). The custom program, written in the MATLAB language, assigns a coupling-coefficient (κ) to each record, where κ can range from 0 (purely independently gating channels) to 1 (fully coupled channels). Elementary event amplitudes were set at 0.5 pA for i_Ca, 1.5 pA for i_Ba, and 38 nM for Ca²⁺ sparklets.

Transfected tsA-201 cells expressing the relevant Ca_V1.2 channel constructs were plated onto poly-L-lysine–coated #1.5 coverslips (Thermo Fisher Scientific) the day before fixation. For immunostaining, cells were fixed by incubating for 10 min in ice-cold methanol, then washed with PBS and blocked for 1 hr at room temperature in 50% SEA BLOCK (Thermo Fisher Scientific) and 0.5% vol/vol Triton X-100 in PBS (blocking buffer). The pore-forming subunit of Ca_V1.2 was probed with a rabbit polyclonal primary antibody (anti-CNC1; kindly provided by Drs William Catterall and Ruth Westenbroek [Hulme et al., 2003]), diluted to 5 μg/ml in diluted blocking buffer (20% SEA BLOCK, 0.5% Triton X-100), by incubating overnight at 4°C. The following morning, cells were washed extensively, receiving three washes with PBS and five washes with diluted blocking buffer. Cells were then incubated for 1 hr at room temperature with Alexa Fluor 647-conjugated donkey anti-rabbit secondary antibody (2 μg/ml; Molecular Probes–Life Technologies) in diluted blocking buffer. Cells were finally washed thoroughly with PBS and mounted for imaging. Native Ca_v1.2 channels in freshly isolated adult ventricular myocytes were immunostained in an identical manner except that plating procedures differed. Specifically, myocytes were plated onto laminin and poly-L-lysine–coated #1.5 coverslips and allowed to adhere for 1 hr before fixation.

For double-labeling experiments used to examine the co-localization of Ca_V1.2 channels and the ER, tsA-201 cells transfected with Ca_V1.2 channels (WT or pre-IQ swap mutants) and mCherry-Sec61β were fixed in 3% paraformaldehyde and 0.1% glutaraldehyde in PBS for 10 min followed by extensive washing in PBS and reduction for 5 min in ∼0.1% sodium borohydride in water to reduce background fluorescence. Cells were washed and blocked as described above and were then incubated in primary antibody solution containing 5 μg/ml rabbit anti-CNC1 and 2 μg/ml rat monoclonal anti-mCherry (Life Technologies) for 1 hr at room temperature. Excess primary antibody was removed by three washes with PBS and five washes with diluted blocking buffer. Cells were then incubated for 1 hr at room temperature with Alexa Fluor 647-conjugated donkey anti-rabbit and Alexa Fluor 568-conjugated chicken anti-rat secondary antibodies (2 μg/ml each; Molecular Probes–Life Technologies) in diluted blocking buffer. Cells were finally washed thoroughly with PBS and mounted for imaging. In some experiments, tsA-201 cells co-expressing JPH2-GFP were fixed and stained as per the double-staining protocol described above. JPH2-GFP was not immunostained, but simply imaged in TIRF mode by exciting the GFP tag.

Coverslips were mounted with MEA-GLOX (for double staining) or β-ME-GLOX imaging buffer on glass depression slides (neoLab, Heidelberg, Germany) and sealed with Twinsil (Picodent, Wipperfürth, Germany). The imaging buffers contained TN buffer (50 mM Tris pH 8.0, 10 mM NaCl), GLOX oxygen scavenging system (0.56 mg/ml glucose oxidase, 34 μg/ml catalase, 10% wt/vol glucose), and either 100 mM MEA (cysteamine) or 142 mM 2-mercaptoethanol (β-ME). Excess imaging buffer was blotted away before application of the Twinsil sealant. This is particularly important with the β-ME-GLOX imaging buffer as the Twinsil will not set if it contacts this buffer.

GSD super-resolution images of Ca_V1.2 channels in fixed ventricular myocytes or Ca_V1.2 channels and mCherry-Sec61β in fixed tsA-201 cells were generated using a Leica SR GSD 3D system. The system is built around a Leica DMI6000 B TIRF microscope and is equipped with a Leica oil-immersion HC PL APO 160×/1.43 NA super-resolution objective, four laser lines (405 nm/30 mW, 488 nm/300 mW, 532 nm/500 mW, and 642 nm/500 mW), and an Andor iXon3 EM-CCD. Images were collected in TIRF mode at a frame rate of 100 Hz for 20,000–100,000 frames using Leica Application Suite (LAS AF) software. Ca_V1.2 cluster area sizes were determined using binary masks of the images in ImageJ/Fiji.

tsA-201 cells expressing Ca_V1.2 channels tagged at their C-terminus with monomeric GFP were fixed in 4% paraformaldehyde (10 min) and imaged in TIRF mode on the Leica 3D-GSD system described above using a 160×/1.43 NA objective. The core Leica DMI6000 B TIRF microscope in this system is capable of functioning outside of GSD SR imaging mode as a conventional TIRF microscope. Cells were illuminated with 488-nm laser light, and image stacks of 2000 frames were acquired at 30 Hz. The first five images after the shutter was opened were averaged, and a rolling-ball background subtraction was applied using ImageJ/Fiji. This image was then low-pass filtered with a 2-pixel cut-off and high-pass filtered with a 5-pixel cut-off (see Figures 2D and 3D). Thresholding was then applied to identify connected regions of pixels that were above threshold. The ImageJ/Fiji plugin ‘Time Series Analyzer v2.0’ was then used to select 4 × 4 pixel regions of interest (ROIs) centered on the peak pixel in each spot. Next, z-axis intensity profiles (where z is time) from these ROIs were examined over the entire image stack. To facilitate the identification of bleaching steps, the signal-to-noise ratio was improved by applying a 5-pixel rolling-ball background subtraction, a median filter (1 pixel radius), and a 10 frame moving average. Bleaching steps were then manually counted.

Identical procedures were performed on cardiomyocytes expressing photo-activatable, GFP-tagged Ca_Vβ₂ auxiliary subunits. This subunit binds to the pore-forming α₁ subunit of the channel with a 1:1 stoichiometry; thus, expression of this protein in cardiomyocytes represents a photo-activatable fluorescent marker of Ca_V1.2 channels. Since adult cardiomyocytes are impervious to chemical transfection, we used adeno-associated virus serotype 9 (AAV9) to transfer this gene into mice via retro-orbital injection, a strategy that has been successfully used by others to transfer cardiac genes in mice (Fang et al., 2012). The AAV9-packaged Ca_Vβ₂-PA-GFP was engineered from Ca_Vβ₂-PA-GFP pcDNA by Vector Biolabs. Mice were sacrificed 5 week after retro-orbital injection. Successful gene transfer was confirmed by photo-activating the Ca_Vβ₂-PA-GFP with 405 nm laser light. Prior to photo-activation, no GFP fluorescence emission was detected upon excitation with 488 nm laser light, but after photo-activation, robust GFP fluorescence emission was observed in the z-lines of isolated ventricular myocytes (data not shown). For stepwise photobleaching experiments, GFP was photo-activated prior to starting movie recordings.

Spontaneous interactions of Ca_V1.2 channels were assayed using bimolecular fluorescence complementation. In these experiments, Ca_V1.2 channels were tagged at their C-terminus with non-fluorescent N- (VN_{(1-154, I152L)}) or C-terminal (VC_{(155-238, A206K)}) halves of a ‘split’ Venus fluorescent protein. When Ca_V1.2-VN and Ca_V1.2-VC are brought close enough together to interact, the full Venus protein can fold into its functional, fluorescent confirmation. The magnitude of Venus fluorescence emission therefore provides an indicator of Ca_V1.2 interactions. Venus fluorescence was monitored in tsA-201 cells expressing Ca_V1.2-VN and Ca_V1.2-VC using TIRF microscopy, as described above (‘recording of Ca²⁺ sparklets’). Transfected cells were identified by co-expression of tagRFP or, in experiments requiring Ca²⁺ imaging with Rhod-2, by weak initial Venus expression.

The relationship between membrane voltage and Ca_V1.2 interactions was obtained by subjecting patch-clamped cells in whole-cell mode to a series of 9-s depolarizations from a holding potential of −80 mV to test potentials ranging from −60 to +80 mV. Maturation of newly reconstituted Venus protein takes some time, hence the long depolarizing pulse (Nagai et al., 2002). Cells were superfused throughout with the 20 mM Ca²⁺ external solution described above. A TTL pulse generated by the TILL imaging system was used to trigger the onset of each voltage sweep. The cell TIRF footprint was illuminated using 491-nm light throughout each voltage sweep, and PM-localized Venus fluorescence emission was monitored in a TIRF movie acquired at a rate of 100 Hz. The final six frames of each movie were averaged to generate a single image for each voltage sweep. These images were median filtered (1 pixel radius) then divided by the −60 mV image to obtain Venus F/F₀ images. Finally, the images were smoothed and pseudo-colored using the ‘red hot’ lookup table in ImageJ/Fiji. The Ca²⁺ dependence of Ca_V1.2 interactions was tested by performing the aforementioned procedure first with 2 mM Ba²⁺ as the conducting ion, then switching the perfusate to our regular external solution described above (20 mM Ca²⁺) and running the procedure again on the same cell (see Figure 4—figure supplement 2).

Reconstitution of the Venus protein is irreversible; thus, once channels spontaneously interact, they remain fused together. We exploited this feature of the bimolecular fluorescence complementation assay in experiments designed to investigate the physiological effects of Ca_V1.2 interactions. In these experiments, transfected tsA-201 cells expressing the split-Venus–tagged channels were patch-clamped in whole-cell mode and dialyzed with 200 μM Rhod-2 via the pipette. Ca²⁺ sparklets were recorded while holding the cell at −80 mV. The depolarizing step protocol described above was performed, and Venus fluorescence was monitored as before. Increases in Venus fluorescence emission during the protocol provide an indication of Ca_v1.2 interactions. Holding cells once more at −80 mV, Ca²⁺ sparklet activity was recorded from the irreversibly fused Ca_V1.2 channels.

For this set of experiments, tsA-201 cells were transfected with expression constructs for Ca_V1.2-EGFP and Ca_V1.2-tagRFP. FRET from Ca_V1.2-EGFP (donor) to Ca_V1.2-tagRFP (acceptor) was measured on an Olympus Fluoview 1000 (FV1000) confocal laser-scanning microscope equipped with an Olympus APON 60×/1.49 NA oil-immersion objective. A 473-nm diode laser was used to excite the sample. Emitted light was separated with an SDM560 beam splitter, collected with BA490-540 and BA575-675 emission filters, and detected by a photomultiplier tube. The resultant raw donor and acceptor images were corrected for background and bleed-through of GFP emission into the RFP channel. In separate experiments using cells expressing only Ca_V1.2-tagRFP, bleed-through was determined to be ∼16%. Intensity measurements from the corrected images, referred to as GFPg and RFPg respectively (where g refers to excitation of GFP with 473 nm light), were extracted using Metamorph or ImageJ software. FRET was expressed as the ratio, FRETr = RFPg/GFPg. In some experiments, the time course of FRETr was monitored during photolysis of caged Ca²⁺. In others, the time course of FRETr was recorded simultaneously with whole-cell Ca_V1.2 currents. A TTL pulse generated by the FV1000 system at the onset of imaging was used to trigger a voltage protocol consisting of a 500-ms depolarizing pulse from a holding potential of −80 to +10 mV. Finally, the Ca²⁺ dependence of FRET between Ca_V1.2 channels was monitored while perfusing cells with external solutions containing 5 mM CsCl, 10 mM HEPES, 10 mM glucose, 140 mM NMDG, 1 mM MgCl₂ and increasing concentrations (0, 25, 50, 100, 200, 300, 400, 800 and 5000 nM) of CaCl₂. The solutions were adjusted to pH 7.4, and 1 μM ionomycin was added to equilibrate intracellular and extracellular Ca²⁺.

tsA cells were perfused with a ‘zero calcium’ external solution containing 5 mM CsCl, 10 mM HEPES, 10 mM glucose, 140 mM NMDG, 1 mM MgCl₂, and 2 mM BaCl₂ (pH. 7.4). Cells were patch-clamped in whole-cell mode as previously described (Tadross et al., 2013) using an internal solution containing 5 mM CsCl, 40 mM HEPES, 135 mM CsMeSO₃, 1 mM citrate and 1.6 mM CaCl₂, adjusted to pH 7.4 and frozen into aliquots. On the day of experiments, the Ca²⁺ cage DMNP-EDTA (2 mM; Invitrogen, D6814) was added. Mg²⁺ was omitted from the pipette solution since DMNP-EDTA can also cage Mg²⁺. Internal solutions were supplemented with 0.25 μM PI(4,5)P₂ (Avanti Polar Lipids, Alabaster, AL; 840046P) and 0.5 μM okadaic acid (LC Laboratories, Woburn, MA; O-2220) to stabilize I_Ca (Tadross et al., 2013). This enabled us to record stable currents and achieve successful uncaging of Ca²⁺, as assayed with the Ca²⁺-sensitive dyes, Fluo-4 AM or Rhod-2 AM (see Figure 6A and Figure 4—figure supplement 1). Uncaging experiments were performed on an Olympus FV1000 confocal microscope equipped with a SIM scanner unit that permits simultaneous laser light uncaging (100% 405 nm for two frames) and recording of Venus, Fluo-4, Rhod-2 or EGFP/tagRFP FRET fluorescence emission.

CaMKII activity was measured using the ‘SignaTECT Calcium/Calmodulin-Dependent Protein Kinase Assay System’ (Promega Corporation, Madison, WI). Mouse hearts were homogenized in a lysis buffer solution containing 20 mM Tris–HCl (pH 8.0), 2 mM EDTA, 2 mM EGTA, 2 mM DTT, PhosSTOP phosphatase inhibitor cocktail (Roche Diagnostics GmbH, Mannheim, Germany) and cOmplete, Mini protease inhibitor cocktail (Roche Diagnostics). Hearts were subjected to three 5 s pulses at 12,000–17,000 rpm using a PRO200 Homogenizer Unit (Pro Scientific, Oxford, CT). The homogenate was centrifuged at 2000 rpm for 10 min and the resultant supernatant was collected and assayed for CaMKII activity as per the manufacturers instructions. To determine the effect of MLCKp on CaMKII activity, 0.1 or 1 μM MLCKp was added to the reaction.

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

We thank Dr Joshua Vaughan for help with super-resolution imaging procedures and analyses. Drs William Catterall and Ruth E Westenbroek provided antibodies against Ca_V1.2. The work was supported by grants from the NIH (HL085870, HL085686 and NS077863) and AHA (14GRNT18730054).

Animal experimentation: This study was performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All of the animals were handled according to approved institutional animal care and use committee (IACUC) protocols (#3374-01) of the University of Washington.

© 2015, Dixon et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.