Calmodulin kinase is functionally targeted to the action potential plateau for regulation of L-type Ca²⁺ current in rabbit cardiomyocytes

Abstract

L-type Ca²⁺ current (I_Ca−L) triggers Ca²⁺ release from the sarcoplasmic reticulum (SR) and both SR and I_Ca−L are potential sources of intracellular Ca²⁺ (${\text{Ca}}_{\mathrm{i}}^{2+}$) for feedback regulation of I_Ca−L. ${\text{Ca}}_{\mathrm{i}}^{2+}$ bound to calmodulin (Ca²⁺–CaM) can inhibit I_Ca−L, while Ca²⁺–CaM can also activate Ca²⁺–CaM-dependent protein kinase II (CaMK) to increase I_Ca. However, it is not known whether I_Ca−L or the SR is the primary source of Ca²⁺ for I_Ca−L regulation. The L-type Ca²⁺ channel C terminus is implicated as a critical transduction element for I_Ca−L responses to Ca²⁺–CaM and CaMK, and the C terminus undergoes voltage-dependent steric changes, suggesting that ${\text{Ca}}_{\mathrm{i}}^{2+}$ control of I_Ca−L may also be regulated by cell membrane potential. We developed conditions to separately test the relationship of Ca²⁺–CaM and CaMK to I_Ca−L and SR ${\text{Ca}}_{\mathrm{i}}^{2+}$ release during voltage clamp conditions modelled upon time and voltage domains relevant to the cardiac action potential. Here we show that CaMK increases I_Ca−L after brief positive conditioning pulses, whereas Ca²⁺–CaM reduces I_Ca−L over a broad range of positive and negative conditioning potentials. SR Ca²⁺ release was required for both Ca²⁺–CaM and CaMK I_Ca−L responses after strongly positive conditioning pulses (+10 and +40 mV), while ${\text{Ca}}_{\mathrm{i}}^{2+}$ from I_Ca−L was sufficient for Ca²⁺–CaM during weaker depolarizations. These findings show that I_Ca−L responses to CaMK are voltage dependent and suggest a new model of L-type Ca²⁺ channel regulation where voltage-dependent changes control I_Ca−L responses to Ca²⁺–CaM and CaMK signalling.

L-type Ca²⁺ current (I_Ca−L) is a prominent feature of the cardiac action potential plateau where it triggers release of Ca²⁺ from the sarcoplasmic reticulum (SR) to modulate contraction (Tanabe et al. 1990), Ca²⁺-dependent gene transcription (Song et al. 2002) and is a source of inward current for arrhythmia-initiating afterdepolarizations (January et al. 1988; Wu et al. 2002). I_Ca−L is regulated by cytoplasmic Ca²⁺ (${\text{Ca}}_{\mathrm{i}}^{2+}$) through binding to the ${\text{Ca}}_{\mathrm{i}}^{2+}$-sensing protein calmodulin (Ca²⁺–CaM) (Peterson et al. 1999; Zuhlke et al. 1999), and Ca²⁺–CaM can bind to the l-type Ca²⁺ channel C terminus and inactivate I_Ca−L (Peterson et al. 1999; Zuhlke et al. 1999). However, Ca²⁺–CaM can also activate the Ca²⁺–CaM-dependent protein kinase II (CaMK) (Hudmon & Schulman, 2002), and CaMK increases I_Ca−L by shifting l-type Ca²⁺ channels into a high opening probability gating mode, through a phosphorylation-dependent mechanism (McCarron et al. 1992; Dzhura et al. 2000). The site of CaMK action for increasing I_Ca−L is uncertain, but CaMK can bind to the l-type Ca²⁺ channel C terminus in a region that overlaps with known Ca²⁺–CaM binding domains (Hudmon et al. 2002). Thus, the l-type Ca²⁺ channel C terminus appears to play a critical role for transducing I_Ca−L responses to ${\text{Ca}}_{\mathrm{i}}^{2+}$. In this regard, it is of interest that the ${\text{Ca}}_{\mathrm{i}}^{2+}$-sensing L-type Ca²⁺ channel C terminus is mobilized by cell membrane depolarization (Kobrinsky et al. 2003), suggesting the possibility that I_Ca−L regulation by ${\text{Ca}}_{\mathrm{i}}^{2+}$ may also be voltage dependent.

${\text{Ca}}_{\mathrm{i}}^{2+}$ can increase as a direct consequence of I_Ca−L, or through secondary release of Ca²⁺ from the SR in heart. Some studies suggest that SR Ca²⁺ release is more important than I_Ca−L for activating CaMK, because CaMK-dependent phosphorylation of the SR regulatory protein phospholamban is reduced 50–90% when SR Ca²⁺ release is prevented by ryanodine (Kuschel et al. 1999; Bartel et al. 2000). On the other hand, ${\text{Ca}}_{\mathrm{i}}^{2+}$ from I_Ca−L is sufficient for engaging Ca²⁺–CaM-dependent I_Ca−L inactivation mechanisms in non-cardiac cells (Zuhlke et al. 1999; Peterson et al. 1999) where SR Ca²⁺ release is not anticipated to contribute to ${\text{Ca}}_{\mathrm{i}}^{2+}$. Furthermore, it remains unknown whether ${\text{Ca}}_{\mathrm{i}}^{2+}$ for Ca²⁺–CaM-dependent I_Ca−L inactivation and for activating CaMK is primarily from I_Ca−L or the SR in cardiomyocytes during dynamic changes in cell membrane potential, as occur in the working heart.

We controlled CaMK activity with a Ca²⁺–CaM-independent form of CaMK and a CaMK inhibitory buffer (IB), previously shown to prevent I_Ca−L increases by endogenous CaMK (Wu et al. 2001a), and inhibited Ca²⁺–CaM with an inhibitory peptide (Wu et al. 2001b) or Ba²⁺ substitution to separately control CaMK and Ca²⁺–CaM activity in ventricular myocytes, in order to differentially test the effects of ${\text{Ca}}_{\mathrm{i}}^{2+}$ from I_Ca−L and SR for regulating I_Ca−L availability. I_Ca−L was measured in cardiomyocytes using a non-steady-state inactivation protocol that mimicked time and voltage conditions present during the cardiac action potential. Our findings support the novel concept that CaMK regulation of I_Ca−L in cardiomyocytes depends upon cell membrane potential. Both ${\text{Ca}}_{\mathrm{i}}^{2+}$ from I_Ca−L and the SR can recruit Ca²⁺–CaM for I_Ca−L inactivation, but SR Ca²⁺ release is required for CaMK effects, while SR Ca²⁺ release also predominates for Ca²⁺–CaM-dependent inactivation at strong depolarizations. A new model of voltage- and ${\text{Ca}}_{\mathrm{i}}^{2+}$-dependent I_Ca−L regulation is proposed.

Methods

Electrophysiology

Whole-cell mode configuration was used for voltage clamping isolated rabbit ventricular myocytes according to previously published methods (Wu et al. 1999a). Ventricular myocytes were isolated from New Zealand White rabbits killed by pentobarbital (50 mg kg⁻¹, i.v.) overdose prior to excising the heart. The Vanderbilt University Animal Care Committee approved all experiments. Cells were held at –80 mV for >5 min for adequate dialysis with pipette solution before initiating experiments. I_Ca−L was conditioned by stepping the cell membrane from –80 mV to +40 mV in 10 mV increments from 30 to 500 ms at 0.1 Hz, and peak I_Ca−L was measured at a test potential of +10 mV (Fig. 1) and expressed as relative current. In some experiments I_Ca−L inactivation was quantified as the fraction of residual inward current present at the end of 30 ms (R₃₀) and 80 ms (R₈₀) conditioning pulses. All experiments were performed at 24°C. Adding Cs⁺ and TEA and reducing Na⁺ and K⁺ in the pipette and bath solutions eliminated Na⁺ and K⁺ currents. A Ca²⁺-activated Cl⁻ conductance (I_Cl,Ca), known to be activated by SR Ca²⁺ release in rabbit ventricular myocytes (Wu & Anderson, 2000) at cell membrane potentials more positive than +20 mV (Wu et al. 1999b), was most clearly seen as a transient outward current in response to voltage command steps to +30 and +40 mV. I_Cl,Ca was eliminated by niflumic acid (10–20 μm, data not shown) and thapsigargin (Fig. 6A), and was not present at the test command potential of +10 mV. Thus, I_Cl,Ca was not likely to have significantly contributed to I_Ca−L measurements because the relationship of relative current (see below) to conditioning potential was not affected by niflumic acid (data not shown), and the relationship between relative current and R₃₀ and R₈₀ was not altered during positive voltage commands in IB (see below, Fig. 5A and B) or IB and thapsigargin (Fig. 7A and B). However, we cannot rule out the possibility that I_Cl,Ca did contribute to CaMK and Ca²⁺–CaM effects, especially at voltage command steps to +30 and +40 mV (Wu et al. 1999b). Elimination of the residual current by nifedipine (10 μm) or Cd²⁺ (100 μm) confirmed that the identity of active inward current was I_Ca−L (data not shown). The control pipette (intracellular) solution was (mm): CsCl 120.0, EGTA 10.0, Hepes 10.0, tetraethylammonium chloride (TEA) 10.0, phosphocreatine 5.0, CaCl₂ 3.0, MgATP 1.0, NaGTP 1.0, and pH was adjusted to 7.2 with 1.0 n CsOH. The ability of endogenous CaMK to facilitate I_Ca−L was eliminated by addition of an inhibitory buffer (IB) solution (see below) (Wu et al. 1999a,2001a). The bath (extracellular) solution was NMDG 137.0, CsCl 25.0, Hepes 10.0, glucose 10.0, CaCl₂ (or BaCl₂) 1.8, MgCl₂ 0.5, and pH was adjusted to 7.4 with 12 n HCl. Ryanodine (10 μm) or thapsigargin (1 μm) were added to the bath solution for some experiments. Myocyte contraction was eliminated under these conditions (data not shown).

Figure 1. Non-steady-state inactivation voltage clamp protocol reveals increases in relative L-type Ca²⁺ current (I_Ca−L) after brief, positive conditioning steps.

A, a schematic representation of the voltage clamp protocol used for this study. Peak I_Ca−L was measured at a +10 mV test pulse, after conditioning steps from –80 to +40 mV in 10 mV steps, lasting from 30 to 500 ms. B, conditioning steps (30 ms) from +20 to +40 mV progressively increase available I_Ca−L at the test pulse of +10 mV (indicated by arrows). The conditioning prepulse potential is labelled above each panel. A transient outward Ca²⁺-activated Cl⁻ current is superimposed on I_Ca−L during some conditioning steps positive to +10 mV (see Methods for details) (Wu et al. 1999b).

Figure 6. CaM and CaMK require SR ${\text{Ca}}_{\mathrm{i}}^{2+}$ after positive conditioning pulses.

A, superimposed current tracings displayed as in Fig. 3A. Cells were dialysed with IB in all panels (IB), but ryanodine (IB, Ryan) or thapsigargin (IB, Thaps) were added to the bath solution to inactivate SR Ca²⁺ release in some experiments. Both ryanodine (B, Ryan) and thapsigargin (C, Thaps) significantly increased available I_Ca−L after 80 ms conditioning prepulses in cells dialysed with IB to inhibit endogenous CaMK. Pre-pulse potentials (abscissa) are plotted against peak I_Ca−L normalized to the maximum value (ordinate). D, normalized peak I_Ca−L values from B and C are plotted against conditioning prepulses to +20 mV for a range of prepulse durations (abscissa). No significant differences in peak I_Ca−L were present between ryanodine- and thapsigargin-treated cells. *P < 0.05, ^†P < 0.01 and ^‡P < 0.001 for panels B–D. E, addition of Ca²⁺-independent CaMK failed to increase relative I_Ca−L in cells dialysed with IB and treated with thapsigargin after positive conditioning prepulses (IB + CaMK + Thaps). F, the Ca²⁺–CaM inhibitory peptide 290–309 failed to increase relative I_Ca−L in cells dialysed with IB and treated with thapsigargin (IB + Thaps + 290–309) after positive conditioning prepulses. ^‡P < 0.001, ^†P < 0.01, *P < 0.05 for comparisons in panels E and F.

Figure 5. The effect of CaMK on relative and residual I_Ca−L.

The effect of CaMK on relative I_Ca−L (recorded during test pulses to +10 mV as in Fig. 1, shown as open circles) after conditioning prepulses to +20 mV (A and B) and –20 mV (C and D) and the residual I_Ca−L at the end of 30 ms (R₃₀, A and C) and 80 ms (R₈₀, B and D) conditioning prepulses (filled circles). The experimental pipette solutions are labelled below C and D, but are also valid for A and B. Cells were dialysed with control pipette solution (Control), IB solution (IB), IB containing a Ca²⁺-independent form of CaMK (IB + CaMK) or IB containing the CaM binding peptide 290–309 (IB + 290–309) and all data are from Figs 2 and 3. Significant differences in residual current (R₃₀ and R₈₀, *P < 0.05) or relative current (**P < 0.05) between the IB condition and Control, IB + CaMK or IB + 290–309 are indicated. Significant differences between residual currents and relative current measurements for each experimental condition are shown (^†P < 0.05).

Figure 7. The effect of CaMK and Ca²⁺–CaM on relative I_Ca−L and R₃₀ and R₈₀ in cells treated with thapsigargin.

The data are displayed as in Fig. 5 except thapsigargin was included in the bath solution and cells were dialysed with IB (IB + thapsigargin), IB and Ca²⁺-independent CaMK (IB + thapsigargin + CaMK) or the Ca²⁺–CaM inhibitory peptide 290–309 (IB + thapsigargin + 290–309). The experimental conditions are labelled below C and D, but are also valid for A and B. The data sets were previously displayed for relative current in Fig. 6E and F. Significant differences between the IB + thapsigargin and other groups for residual current (R₃₀ and R₈₀, *P < 0.05) and relative current (**P < 0.05) are indicated. Significant differences between residual currents and relative currents for each experimental condition are shown (^†P < 0.01).

Approaches to controlling CaM and CaMK activity

A recombinant monomeric truncation mutant of the mouse CaMK α isoform (amino acid residues 1–380) was expressed using baculovirus and then purified using CaM agarose affinity chromatography (Brickey et al. 1990). This CaMK was stored in IB (50 mm Hepes, pH 7.5, 1 mm EDTA, 1 mm DTT, 50% (v/v) glycerol, 10% (v/v) ethylene glycol) and activated by autophosphorylation in a 100 μl reaction containing 50 mm Hepes, pH 7.5, 2 mm magnesium acetate, 1.5 mm CaCl₂, 18 μm CaM, 2 mm DTT and 100 μm ATPγ S. The reaction was initiated by addition of the (1–380) CaMK (9 μmol l⁻¹ final subunit concentration), incubated at 30°C for 10 min, and stopped by the addition of EDTA (10 mm). IB was used without added CaMK, or after inactivation of enzymatically active CaMK by heating, to observe CaMK-independent effects of manipulating Ca²⁺–CaM and SR Ca²⁺ release. IB prevents I_Ca−L facilitation by endogenous CaMK (Wu et al. 2001a) and was used to separate CaMK from Ca²⁺–CaM activity and SR Ca²⁺ release. Ca²⁺–CaM-dependent autophosphorylation of the CaMK produces a constitutively active species that can phosphorylate substrates in the absence of Ca²⁺–CaM. Ca²⁺-independent activity was typically 35–50% of total activity in the presence of Ca²⁺–CaM using the peptide substrates syntide-2 or autocamtide. Constitutively active CaMK and IB were diluted 10-fold in the pipette solution (0.9 μm final) for use in voltage clamp studies and its activity confirmed in vitro, as described (Wu et al. 1999a). This dilution was chosen to approximate the physiological CaM kinase activity in heart (∼1–2 μm) derived from percentage yield calculations during purification (Iwasa et al. 1986; Gupta & Kranias, 1989).

The CaMK inhibitory peptide AC3-I (KKALHRQEAVDCL, IC₅₀∼3 μm) (Braun & Schulman, 1995) (Macromolecular Resources, Fort Collins, CO, USA) is a modified CaMK substrate, which inhibits endogenous and thiophosphorylated constitutively active CaMK. AC3-I was included in the pipette solution at a final concentration of 20 μm. The CaM binding peptide 290–309 (Calbiochem) is modelled on the CaM binding domain of CaMK, and inhibits Ca²⁺–CaM signalling generally, and was added to the pipette solution at a final concentration of 50 μm. CaMK, 290–309 and AC3-I were dialysed into cells for 5–10 min prior to initiating experiments.

Statistics

The null hypothesis was evaluated with Student's t test or ANOVA, as appropriate. Bonferroni's correction was applied for multiple comparisons.

Results

Enhanced I_Ca−L availability following brief, positive conditioning pulses

Brief, positive conditioning prepulses progressively increased relative I_Ca−L availability (Fig. 1), while IB significantly reduced relative I_Ca−L availability (Fig. 2A–C), compared to control pipette solution, suggesting that endogenous CaMK can increase I_Ca−L under action potential plateau conditions. The increase in I_Ca−L in response to brief (30–130 ms), positive conditioning prepulses was lost at longer (500 ms) prepulse durations when Ca²⁺ was the charge carrier (Fig. 2D), but persisted when Ba²⁺ was substituted for Ca²⁺ (Fig. 2E), in cells dialysed with IB. The persistence of Ca²⁺-dependent reduction in I_Ca−L availability in IB indicates that Ca²⁺–CaM-dependent inactivation is operative under these experimental conditions. These findings show that Ca²⁺ critically determines I_Ca−L availability during time and voltage conditions present during the ventricular action potential plateau, and serve as a starting point for dissecting the role of Ca²⁺–CaM, CaMK and SR Ca²⁺ release in determining I_Ca−L availability in cardiac myocytes.

Figure 2. A CaMK inhibitory buffer (IB) reduces I_Ca−L at action potential plateau potentials.

A, current tracings recorded in control pipette solution with Ca²⁺ as charge carrier (Control), IB pipette solution with Ca²⁺ as charge carrier (IB) and IB pipette solution with Ba²⁺ as charge carrier (Ba²⁺, IB) are shown. Currents in response to +20 mV (open arrows) and +30 mV (filled arrow heads) conditioning prepulses are superimposed and normalized to the peak I_Ca−L at +20 mV for comparison. B, topology plots showing the combined effects of conditioning prepulse potentials (–80 to +40 mV) and conditioning prepulse durations (30–500 ms) on relative peak Ca²⁺ current (I_Ca) recorded during the test potential step to +10 mV in cells (n= 3 for each group) dialysed with control solution (Control) or IB solution (IB). The grid lines on the topology plots indicate conditioning potentials in +10 mV increments (evenly spaced from –80 to +40 mV) and conditioning prepulse durations (evenly spaced at 30, 80, 130, 190, 300 and 500 ms). C, peak I_Ca−L after 80 ms conditioning prepulses from –80 to +40 mV (abscissa) in cells dialysed with control buffer (n= 5) that permits activation of endogenous CaMK and with IB (n= 7, see Methods for details) that prevents activation of endogenous CaMK. This IB data set is also used in Figs 3–6 and this Control data set is used in Figs 4 and 5 for other comparisons. D and E show a family of non-steady-state inactivation relationships for a single cardiomyocyte dialysed with IB using (D) Ca²⁺ or (E) Ba²⁺ as charge carrier. The duration of the conditioning prepulse is indicated for each of the isochrones in C and D. Peak I_Ca−L is normalized to the maximum value in all panels. *P < 0.05, ^‡P < 0.001.

CaMK increases I_Ca−L at action potential plateau conditions

Previous work has shown that IB prevented CaMK mediated I_Ca increases in cardiac myocytes (Wu et al. 1999a, 2001a), strongly suggesting that CaMK inhibition was a critical feature of IB actions at I_Ca−L. On the other hand, the persistence of Ca²⁺-dependent I_Ca−L inactivation (Fig. 2A, D and E) suggested IB did not disrupt Ca²⁺–CaM signalling, generally. We created a topographical surface plot of relative I_Ca−L availability to better illustrate the effects of IB over a wide range of conditioning times and voltages (Fig. 2B). These plots reveal the functional targeting of significant IB actions to time and voltage durations relevant to the action potential plateau (P < 0.05 for all intergroup comparisons from conditioning potentials between 0 and +40 mV and conditioning pulse durations from 30 to 130 ms). In order to more thoroughly test the concept that IB was selective for CaMK and that reduction in I_Ca−L at action potential plateau potentials in IB was due to CaMK, we dialysed a Ca²⁺ independent form of CaMK into myocytes in the presence of IB (Fig. 3B). Ca²⁺–CaM-independent CaMK significantly restored reduced I_Ca−L availability in IB (Fig. 3B), indicating that inhibition of endogenous CaMK was the critical IB effect. The effects of Ca²⁺–CaM-independent CaMK on I_Ca−L were specifically due to the enzymatic activity of the exogenous kinase, as they were significantly reduced by coadministration of the CaMK inhibitory peptide AC3-I (data not shown), and eliminated by heat inactivation of the added CaMK (Fig. 3B). CaMK effects were confined to brief conditioning pulses (30–130 ms), and were lost after longer conditioning prepulses (190–500 ms, Fig. 4), confirming that CaMK actions on I_Ca−L were targeted to time and voltage domains relevant to the cardiac action potential plateau.

Figure 3. The effects of Ca²⁺–CaM and CaMK on I_Ca−L.

A, the left panels show superimposed current tracings (as in Fig. 2A) in response to conditioning prepulses to +20 mV (open arrows) and +40 mV (filled arrows) in cells dialysed with IB (IB) or a combination of IB and the Ca²⁺–CaM inhibitory peptide 290–309 (IB + 290–309). The right panel shows summary findings for peak I_Ca−L after 80 ms conditioning prepulses from –80 to +40 mV (abscissa) in cells treated as described in the left panels. B, the left panels show superimposed current tracings labelled as in A (above), but these cells were dialysed with IB and a Ca²⁺-independent form of CaMK (IB + CaMK) or IB and heat-inactivated CaMK (IB + heated CaMK). The right panel shows summary findings for peak I_Ca−L (as in A), but from cells treated as in the left panels. Dialysis with a Ca²⁺–CaM-independent form of CaMK that is resistant to IB increases relative I_Ca−L (80 ms prepulses) compared to IB alone after positive conditioning pulses. CaMK activity is ablated by heat inactivation and peak I_Ca−L is normalized to the maximum value. *P < 0.05, ^†P < 0.01, ^‡P < 0.001 for A and B.

Figure 4. The effect of exogenous Ca²⁺–CaM-independent CaMK on I_Ca−L availability after positive conditioning prepulses.

A–D, each shows peak I_Ca−L responses to conditioning prepulses of varying durations (abscissa) at four positive membrane potentials (indicated above each plot). Cells were dialysed with IB to inhibit endogenous CaMK activity in the presence (filled diamonds, n= 5) or absence (open squares, n= 7) of Ca²⁺-independent CaMK that is resistant to IB. Filled circles represent the difference in relative I_Ca−L in IB + CaMK and IB alone and the contiuous lines are exponential fits of these differences. *P < 0.05, ^†P < 0.01.

We next considered the relationship between I_Ca−L inactivation during brief conditioning voltage pulses (R₃₀ and R₈₀) and relative I_Ca−L availability (relative current) in the subsequent test pulse (Fig. 1), in order to better understand the apparent voltage dependence of CaMK signalling to l-type Ca²⁺ channels (Dzhura et al. 2000). At a positive conditioning potential (+20 mV) R₃₀ and R₈₀ accurately predicted relative current under control conditions, in IB and in IB after CaMK replacement (Fig. 5A and B). CaMK replacement significantly restored R₃₀, R₈₀ and relative current under these conditions. In contrast, IB significantly reduced R₃₀ compared to relative current after a weakly depolarizing conditioning step to –20 mV, but CaMK replacement failed to equalize R₃₀ and relative current (Fig. 5C). IB significantly reduced both R₈₀ and relative current after a –20 mV conditioning potential, but R₈₀ was significantly less than relative current even under control conditions (Fig. 5D). These comparisons between I_Ca−L inactivation and relative (I_Ca−L) current suggest that a greater fraction of L-type Ca²⁺ channels were available to open at the +10 mV test voltage following inactivation of I_Ca−L during the –20 mV than the +20 mV conditioning step. Taken together, results from experiments using both CaMK inhibition and replacement strategies indicate that CaMK can significantly increase I_Ca−L by a time- and voltage-dependent mechanism in cardiac myocytes.

Ca²⁺–CaM reduces I_Ca−L availability at action potential plateau potentials

The finding that cells dialysed with IB retained Ca²⁺-dependent inactivation (Figs 2A, D and E) suggested that Ca²⁺–CaM remained operative under these experimental conditions. Cellular dialysis with the Ca²⁺–CaM binding peptide 290–309 did significantly increase relative I_Ca−L availability. In contrast to CaMK that was only effective after positive conditioning potentials (Figs 3B and 5), 290–309 was effective after weakly and strongly depolarizing conditioning potentials (Figs 3A and 5). These data show that endogenous Ca²⁺–CaM was a significant signal transduction element for grading I_Ca−L in the presence of IB. However, 290–309 did not fully increase I_Ca−L availability to levels present with Ba²⁺ as charge carrier (Fig. 2E), perhaps indicating that a constitutively bound pool of CaM (Erickson et al. 2001; Pitt et al. 2001) was inaccessible to the peptide.

In contrast to the effects of 290–309 on I_Ca−L availability, peak I_Ca−L during the conditioning pulse in IB + 290–309 (4.1 ± 0.4 pA pF⁻¹, n= 5) was significantly less than peak I_Ca−L recorded in IB alone (7.1 ± 0.9 pA pF⁻¹, n= 7) or in IB + CaMK (8.1 ± 0.5 pA pF⁻¹, n= 5), whereas IB + 290–309 did not decrease peak I_Ca−L compared to control solution (6.2 ± 0.5 pA pF⁻¹, n= 5).

SR Ca²⁺ release selectively reduces I_Ca−L after brief, positive conditioning pulses

The results of experiments so far show that the CaMK-dependent component of ${\text{Ca}}_{\mathrm{i}}^{2+}$ signalling to L-type Ca²⁺ channels is voltage dependent (Figs 3B and 4); however, they do not distinguish between I_Ca−L and SR Ca²⁺ release as dominant sources of signalling ${\text{Ca}}_{\mathrm{i}}^{2+}$ for grading I_Ca−L. Both ryanodine and thapsigargin significantly increased I_Ca−L after positive conditioning potentials in the presence of IB, indicating that Ca²⁺-induced Ca²⁺ release is present under these experimental conditions (Fig. 6A–C). Reduction in SR Ca²⁺ release by either ryanodine or thapsigargin targeted I_Ca−L increases over similar cell membrane potential (Fig. 6B and C) and temporal domains (Fig. 6D) as CaMK (Figs 3B and 4), suggesting the possibility that activation of endogenous CaMK is predominately due to SR Ca²⁺ release, and that Ca²⁺–CaM-dependent inactivation relies on SR ${\text{Ca}}_{\mathrm{i}}^{2+}$ under action potential plateau conditions.

CaMK actions at I_Ca−L are determined by SR Ca²⁺ release

Previous studies suggest SR Ca²⁺ is important for activating endogenous CaMK in cardiac myocytes (Kuschel et al. 1999; Wu et al. 1999a, 2001b; Bartel et al. 2000), so we reasoned that Ca²⁺-independent CaMK would circumvent the requirement for activation of endogenous CaMK by SR Ca²⁺ release. Surprisingly, thapsigargin-treated cells did not show increases in I_Ca−L availability at action potential plateau potentials in response to Ca²⁺-independent CaMK, but did show a significant depolarizing shift in I_Ca−L availability in response to weaker depolarizations (Fig. 6E). This result was in striking contrast to the significant increases in I_Ca−L after positive conditioning pulses when myocytes with intact SR Ca²⁺ release were supplemented with Ca²⁺-independent CaMK (Figs 3B and 4). In contrast to CaMK-dependent increases in R₃₀, R₈₀ and relative I_Ca−L at +20 mV (Fig. 5A and B), CaMK replacement was ineffective for increasing these parameters after thapsigargin at +20 mV (Fig. 7A and B). CaMK replacement also failed to evoke a consistent response in R₈₀ and relative current after a –20 mV conditioning step (Fig. 7D). These data underscore the close relationship between SR Ca²⁺ release and L-type Ca²⁺ channel function, and suggest the possibility that the Ca²⁺-independent form of CaMK acts to increase I_Ca−L availability through a SR-dependent mechanism.

SR Ca²⁺ release significantly determines Ca²⁺–CaM responses after strong depolarizations

In contrast to CaMK (Fig. 3B), Ca²⁺–CaM effects on I_Ca−L appear to be voltage independent because they operate over a broad range of physiological conditioning potentials (Fig. 3A). SR Ca²⁺ release does contribute to Ca²⁺-dependent I_Ca−L inactivation (Balke & Wier, 1991; Wu et al. 2001b), and reduction of SR Ca²⁺ release by ryanodine (Fig. 6B) or thapsigargin (Fig. 6C) significantly increased relative I_Ca−L in cells dialysed with IB only after positive conditioning steps, suggesting that the source of activator ${\text{Ca}}_{\mathrm{i}}^{2+}$ for Ca²⁺–CaM-dependent inactivation may vary in a voltage-dependent manner in cardiac myocytes. In order to better understand the contribution of I_Ca−L and SR to Ca²⁺–CaM for regulating I_Ca−L responses, we dialysed 290–309 into myocytes after thapsigargin. Ca²⁺–CaM inhibition with 290–309 significantly increased relative I_Ca−L (Fig. 6F) and R₃₀ (Fig. 7C) and R₈₀ (Fig. 7D) at –20 mV, but not at +20 mV (Fig. 7A and B), supporting previous findings that I_Ca−L is sufficient for Ca²⁺–CaM-dependent inactivation, but suggesting that normal SR Ca²⁺ release significantly determines Ca²⁺–CaM signalling to L-type Ca²⁺ channels at strongly depolarized conditioning potentials, present during the action potential plateau.

Discussion

CaMK, I_Ca−L and SR Ca²⁺ release

The present studies use a combination of approaches to inhibit endogenous CaMK-dependent I_Ca−L increases (with IB) or control CaMK activity (with exogenous Ca²⁺–CaM independent CaMK) while SR Ca²⁺ release is preserved or eliminated (with ryanodine or thapsigargin) in cardiomyocytes. The central finding of these experiments is that CaMK increases in I_Ca−L are functionally targeted over time and voltage domains that are directly relevant to the cardiac action potential plateau. This finding suggests CaMK actions are analogous to protein kinase A, which can also regulate L-type Ca²⁺ channels by a voltage-dependent mechanism (Sculptoreanu et al. 1993). Previous investigations established that CaMK can directly increase L-type Ca²⁺ channel openings in excised membrane patches from ventricular myocytes (Dzhura et al. 2000), but these experiments did not test for temporal- or voltage-dependent features of CaMK signalling and could not measure effects of SR Ca²⁺ release. The dramatic loss of CaMK signalling effects after elimination of SR Ca²⁺ release (Figs 6E and F and 7) shows those CaMK actions are reliant upon SR Ca²⁺ release. One possible unifying explanation for these observations is that an important CaMK action on I_Ca−L in intact myocytes may be indirect, via modulation of SR Ca²⁺ release (Li et al. 1997; Wu et al. 2001a). The concept that CaMK exerts important actions on I_Ca−L by influencing SR Ca²⁺ release may reconcile earlier reports showing that I_Ca−L increases were linked to dynamic reduction in SR Ca²⁺ release (Delgado et al. 1999), but were eliminated by inactivation of SR Ca release (Wu et al. 1999a, 2001b). Interestingly CaMK slightly, but significantly, increased relative I_Ca−L availability at weakly depolarized potentials in the absence of SR Ca²⁺ release (Figs 6E and 7D), raising the possibility that CaMK may be capable of regulating I_Ca−L by a SR-independent pathway under these voltage clamp conditions. The finding that I_Ca−L availability responses to CaMK (Fig. 6E) and 290–309 (Fig. 6F) are very similar in thapsigargin, after weak and strong depolarizations, suggests that in the absence of SR Ca²⁺ release, CaMK and Ca²⁺–CaM may compete for shared molecular machinery, such as the L-type Ca²⁺ channel C terminus. Thus, the present studies add important new information to our understanding of how CaMK may contribute to I_Ca−L regulation in the working heart.

Ca²⁺–CaM, I_Ca−L and SR Ca²⁺ release

CaM is a ubiquitous ${\text{Ca}}_{\mathrm{i}}^{2+}$-sensing protein that is required for ${\text{Ca}}_{\mathrm{i}}^{2+}$-dependent I_Ca−L inactivation in cardiac L-type Ca²⁺ channels (Peterson et al. 1999; Zuhlke et al. 1999). CaMK and CaM colocalize with L-type Ca²⁺ channels and ryanodine receptors (Wu et al. 1999a; Pate et al. 2000; Balshaw et al. 2001; Pitt et al. 2001; Dzhura et al. 2002; Erickson et al. 2003), and are capable of regulating both of these proteins. Ca²⁺–CaM also activates other proteins, including CaMK, so that myriad effects potentially complicate interpretation of Ca²⁺–CaM inhibition experiments. The present experiments used IB dialysis and a Ca²⁺–CaM inhibitory peptide (290–309) to separately control Ca²⁺–CaM and CaMK signalling. These findings support the concept that Ca²⁺–CaM reduces available I_Ca−L in a voltage-independent manner (Fig. 3A). However, the critical source of ${\text{Ca}}_{\mathrm{i}}^{2+}$ for Ca²⁺–CaM is determined by cell membrane voltage because ryanodine (Fig. 6B) and thapsigargin (Fig. 6C) only increased I_Ca−L at plateau potentials and because 290–309 was ineffective at increasing I_Ca−L at plateau potentials in the absence of SR Ca²⁺ release (Figs 3A and 6F). In contrast, CaM sequestration with the 290–309 peptide enhanced I_Ca availability after weakly depolarizing prepulses, independent of SR Ca²⁺ release (Figs 6F and 7D and E), suggesting that I_Ca−L alone is a sufficient source of ${\text{Ca}}_{\mathrm{i}}^{2+}$ for Ca²⁺–CaM-dependent inactivation of I_Ca−L at weakly depolarized cell membrane potentials. The finding that Ca²⁺–CaM competition by 290–309 significantly increased relative currents and slowed I_Ca−L inactivation in IB (Fig. 5) is consistent with a recent report showing marked slowing of I_Ca−L inactivation and action potential prolongation in cardiomyocytes transfected with Ca²⁺-binding deficient, dominant-negative CaM mutants (Alseikhan et al. 2002). A potential limitation to studies with 290–309 is highlighted by the finding that peak I_Ca−L during the conditioning pulse was reduced in IB + 290–309 compared to IB alone, raising the possibility that an outward current, such as I_Cl,Ca, may be activated by 290–309 and complicate I_Ca−L measurements during this experimental condition. Taken together, these results reveal the interdependence of CaM, SR Ca²⁺ release and cell membrane potential and add to other recent work highlighting the importance of CaM as a ${\text{Ca}}_{\mathrm{i}}^{2+}$-driven signalling element for regulating I_Ca in heart.

The relationship between I_Ca−L inactivation and availability

The present experiments show that it is possible to distinguish between SR Ca²⁺, CaM and CaMK signalling effects on relative I_Ca−L availability in cardiac myocytes. Both CaM (Fig. 3A) and CaMK (Fig. 3B) can separately regulate I_Ca−L availability under non-steady-state conditions. Relative I_Ca−L availability is closely related to I_Ca−L inactivation after positive conditioning pulses (Figs 5A and B, and 7A and B). This relationship is consistent with the concept that I_Ca−L inactivation (R₃₀ and R₈₀) during the conditioning pulse directly determines I_Ca−L availability during the subsequent test pulse. Experiments to inhibit or replace CaMK (Fig. 5), eliminate SR Ca²⁺ release, or reduce Ca²⁺–CaM (Fig. 7) did not alter this relationship at +20 mV. In contrast, R₃₀ was significantly less than relative I_Ca−L after CaMK inhibition with IB (Fig. 5C), suggesting that CaMK could reduce this measure of I_Ca−L inactivation at –20 mV without altering the pool of L-type Ca²⁺ channels available for opening in response to the +10 mV test pulse. SR Ca²⁺ also reduced R₃₀ and R₈₀ during CaMK inhibition at –20 mV (compare IB in Fig. 5C and D with IB + thapsigargin in Fig. 7C and D). However, ${\text{Ca}}_{\mathrm{i}}^{2+}$ from I_Ca−L was sufficient for significant Ca²⁺–CaM actions on R₃₀ and R₈₀ at –20 mV, because these measures were both increased by 290–309 in the combined presence of IB and thapsigargin (Fig. 7D).

A model for voltage- and ${\text{Ca}}_{\mathrm{i}}^{2+}$-dependent regulation of I_Ca−L in heart

The recent finding that the L-type Ca²⁺ channel C terminus undergoes significant voltage-dependent movement (Kobrinsky et al. 2003) provides a potentially important context for understanding our findings. The C terminus is now accepted to be richly endowed with ${\text{Ca}}_{\mathrm{i}}^{2+}$ sensing machinery, and three distinct Ca²⁺–CaM binding domains have been identified (Zuhlke et al. 1999; Pate et al. 2000; Pitt et al. 2001). On the other hand, the C terminus is also capable of binding activated CaMK (Hudmon et al. 2002). Given that both Ca²⁺–CaM and CaMK can converge upon the C terminus and given that the C terminus is significantly mobile over the cell membrane potential ranges used in our study, it is possible that the C terminus could be variably positioned to differentially respond to ${\text{Ca}}_{\mathrm{i}}^{2+}$ from I_Ca−L (when positioned near the pore region) or the SR (when positioned outside of the pore region). Based upon these considerations and upon our finding that SR Ca²⁺ release was required for complete Ca²⁺–CaM and CaMK actions at +20 mV (Figs 5 and 7), we hypothesize that the C terminus moves away from the pore region during strong depolarizations. Because ${\text{Ca}}_{\mathrm{i}}^{2+}$ from I_Ca−L was sufficient for Ca²⁺–CaM at –20 mV, we further hypothesize that the C terminus is close enough to the pore region during weak depolarizations to sense ${\text{Ca}}_{\mathrm{i}}^{2+}$ directly from I_Ca−L (Fig. 8).

Figure 8. Schematic depiction of the hypothesized relationship between cell membrane potential, L-type Ca²⁺ channel C terminus motion and ${\text{Ca}}_{\mathrm{i}}^{2+}$‘sensing’ by CaM and CaMK.

The L-type Ca²⁺ channel pore forming subunit (α_1C) is shown as a pair of open rectangles with a central pore. The C terminus protrudes into the cytoplasmic space from the right rectangle and the pore region and SR Ca²⁺ release channel (ryanodine receptor, dark trapezoid) are en face. Ca²⁺–CaM is depicted as a pair of stippled circles linked by a curved segment and activated CaMK is shown as a thick bar with curved ends bound to Ca²⁺–CaM. According to the hypothesized model, strong depolarizations motivate the C terminus to move away from the α_1C pore so that ${\text{Ca}}_{\mathrm{i}}^{2+}$ sensing is primarily from the SR. Weak depolarizations leave the C terminus in the vicinity of the pore where sensed ${\text{Ca}}_{\mathrm{i}}^{2+}$ is directly from I_Ca−L. Inactivation of SR Ca²⁺ release (by thapsigargin) results in significant impairment of ${\text{Ca}}_{\mathrm{i}}^{2+}$ sensing through CaM and CaMK during strong depolarizations, while ${\text{Ca}}_{\mathrm{i}}^{2+}$ from I_Ca−L is sufficient for Ca²⁺–CaM at weak depolarizations in the absence of SR Ca²⁺ release.

Our findings show that CaMK responses are very different during non-steady-state time and voltage domains that approximate action potential plateau conditions than during steady-state conditions. While most studies have understandably focused on the effects of prolonged voltage clamp command pulses, in order to measure steady-state responses, steady-state behaviours may not always be relevant to the physiology of the action potential. These experiments highlight the importance of considering non-steady-state measurements for understanding the effects of cellular signals on ionic current responses.