Stimulation of recombinant Ca_v3.2, T-type, Ca²⁺ channel currents by CaMKIIγ_C

Abstract

Molecular cloning of low-voltage activated (LVA) T-type calcium channels has enabled the study of their regulation in heterologous expression systems. Here we investigate the regulation of Ca_v3.2 α₁-subunits (α1H) by calcium- and/or calmodulin-dependent protein kinase II (CaMKII). 293 cells stably expressing α1H were transiently transfected with CaMKIIγ_C. Using the whole-cell recording configuration, we observed that elevation of pipette free Ca²⁺ (1 μm) in the presence of CaM (2 μm) increases T-type channel activity selectively at negative potentials, evoking an 11 mV hyperpolarizing shift in the half-maximal potential (V_1/2) for activation. The V_1/2 of channel inactivation is not altered by Ca²⁺/CaM. These effects reproduced modulation observed in adrenal zona glomerulosa cells. The potentiation by Ca²⁺/CaM was dependent on the co-expression of CaMKIIγ_C and required Ca²⁺/CaM-dependent kinase activity. Peptide (AIP) and lipophilic (KN-62) protein kinase inhibitors prevented the Ca²⁺/CaM-induced changes in channel gating without altering basal Ca_v3.2 channel activity (27 nm free Ca²⁺) as did replacing pipette ATP with adenylyl imidodiphosphate (AMP-PNP), a non-hydrolysable analogue. CaMKII-dependent potentiation of channel opening resulted in significant increases in apparent steady-state open probability (P_o) and sustained channel current at negative voltages. Under identical conditions, CaMKII activation did not regulate the activity of Ca_v3.1 channels, the first cloned member (α1G) of the T-type Ca²⁺ channel family. Our results provide the first evidence for the differential regulation of two members of the Ca_v3 family by protein kinase activation and the first report reconstituting CaMKII-dependent regulation of any cloned Ca²⁺ channel.

Low voltage activated (LVA) Ca²⁺ channel currents (T-type) were extensively characterized in vertebrate sensory neurones (Carbone et al. 1984; Fedulova et al. 1985; Nowycky et al. 1985; Swandulla & Armstrong, 1988). Since the initial description, whole-cell and single-channel LVA currents have been described in a large variety of excitable and non-excitable cells (Bean, 1989; Huguenard, 1996). Although native LVA currents are kinetically diverse, all LVA channels activate at lower membrane voltages, inactivate more rapidly, deactivate more slowly and are metabolically more stable than high voltage activated (HVA) Ca²⁺ channels. Molecular cloning studies have identified two HVA Ca²⁺ channel families, Ca_v1 (L-type) and Ca_v2 (P-/Q-type, N-type, R-type) and one LVA Ca²⁺ channel family, Ca_v3 (T-type), that is the most dissimilar, sharing only 25 % sequence identity with HVA channel family members (Catterall, 2000). Unlike HVA Ca²⁺ channels that have been purified and shown to be complexes of α1-, α2-δ-, β- and in some cases γ-subunits; LVA Ca²⁺ channels have not been purified. Expression of the pore-forming α1-subunit alone reconstitutes functional LVA channels that exhibit the voltage dependencies and kinetics of native LVA channels (Cribbs et al. 1998; Perez-Reyes et al. 1998).

Like HVA channels, LVA channels are targets of hormonal modulation, although the molecular mechanisms for the regulation of these channels are incompletely understood. LVA current is inhibited by atrial natriuretic peptide in adrenal glomerulosa cells (Barrett et al. 1991; McCarthy et al. 1993) by angiotensin II (type 2 receptor) in NG108-15 cells (Buisson et al. 1995), and by dopamine in retinal horizontal cells (Pfeiffer-Linn & Lasater, 1993), adrenal glomerulosa cells (Osipenko et al. 1994) and in a variety of pituitary cell preparations (lactotrophs; Lledo et al. 1990), pars intermedia cells (Nussinovitch & Kleinhaus, 1992) and melanotrophic cells (Keja et al. 1992)). In addition, LVA current is enhanced by endothelin-1 in ventricular myocytes (Furukawa et al. 1992) and portal vein cells (Inoue et al. 1990). Less consistent effects among preparations have been observed with other hormones. Noradrenaline increases T-type Ca²⁺ channel currents in portal vein (Pacaud et al. 1987) yet inhibits these currents in sensory neurones (Bean, 1989), while the modulation of LVA current by angiotensin II via the type 1 receptor changes with development. Angiotensin II potentiates LVA current at negative potentials in neonatal bovine adrenal glomerulosa cells (Cohen et al. 1988; McCarthy et al. 1993; Lu et al. 1996) yet reduces it in adult glomerulosa cells (Rossier et al. 1995). In a few cases, the cellular mediators of these hormone-induced changes in channel activity have been identified, yet across preparations, a hormone-elicited change in the activity of any specific kinase does not mediate a consistent change in channel gating. In part, this heterogeneity of response might be attributable to the existence of multiple members of the T-type Ca²⁺ channel family, Ca_v3.1 (α1G), Ca_v3.2 (α1H), Ca_v3.3 (α1I) (Lee et al. 1999), multiple subtypes of each receptor that couple to different signalling molecules, or the concurrent activation by a single receptor subtype of multiple signalling cascades that induce opposing changes in channel gating (Pemberton et al. 2000).

Our laboratory has demonstrated that Ca_v3.2 is the predominant T-type Ca²⁺ channel family member expressed in the adrenal zona glomerulosa of two genera, rat and bovine (Schrier et al. 2001). In the adrenal glomerulosa cell, activation of CaMKII induces a 10 mV-hyperpolarizing shift in the voltage dependence of activation of Ca_v3.2 channels (Lu et al. 1994; Chen et al. 1999). Underlying this change in activation gating is an increase in channel open probability, the result of an increase in the number of active sweeps and channel re-openings (Barrett et al. 2000). The regulation of Ca_v3.2 channels by CaMKII in adrenal glomerulosa cells in the excised patch by membrane-associated kinase or by exogenous-recombinant kinase that is constitutively active (Barrett et al. 2000), indicated that regulation was confined to a local area and prompted us to assay whether this regulation of Ca_v3.2 channels could be reconstituted in a heterologous expression system.

Here we show that the activation of CaMKIIγ_C in 293 cells stably expressing Ca_v3.2 channels induces a hyperpolarizing shift in the voltage dependence of channel activation that is not accompanied by a concomitant change in inactivation gating, mimicking changes in gating previously reported in native glomerulosa cells. Regulation in 293 cells is dependent upon the expression of active kinase and surprisingly, is not observed with Ca_v3.1 channels. Our results provide the first evidence for regulation of a Ca_v3 family member by protein kinase activity and for the differential regulation by CaMKII of two Ca²⁺ channel family members.

METHODS

Cell culture and transfection

293 cells (human embryonic kidney cells, American Type Culture Collection, Manassas, VA, USA) stably expressing either human Ca_v3.2 channels (α1H), lines HFH or AH13 (Cribbs et al. 1998) or Ca_v3.1 (α1G) channels, line Q39 (Cribbs et al. 2000) maintained under G418 selection (400 μg ml⁻¹; Gibco/BRL, Rockville, MD, USA), were cultured in DMEM/F-12 medium (Gibco/BRL) containing 100 U ml⁻¹ penicillin and 100 μg ml⁻¹ streptomycin. Stable cell lines were transiently transfected with plasmids that encode green fluorescent protein (GFP) and CaMKIIγc under the control of a cytomegalovirus promoter. We obtained cDNAs for: pGreenLantern (GFP) in pcDNA3 (Gibco/BRL) and porcine CaMKIIγc in pRc/CMV (H. Singer, Albany, NY, USA). The expression plasmids were co-transfected in 35 mm diameter culture dishes by CaPO₄ precipitation at a ratio of 6:1 (μg CaMKIIγ_C plasmid: μg GFP plasmid; 7 μg total) or with the Effectene transfection reagent (Qiagen, Valencia, CA, USA) at a ratio of 4:1 (1 μg total). Cells were seeded onto glass coverslips 48–72 h after transfection for recording.

Solutions

Solutions for whole-cell patch-clamp experiments were designed to eliminate potassium current contamination and minimize T-type channel run-down. The pipette (intracellular) solution contained (mm): 115 CsCl, 1 tetrabutylammonium chloride, 1 MgCl₂, 5 Mg-ATP, 1 Li-GTP, 20 Hepes, pH 7.2 and 11 BAPTA, as well as 2 μm calmodulin and CaCl₂ (see below). The external solution contained (mm): 127 TEACl, 10 CaCl₂, 0.5 MgCl₂, 10 Hepes, 5 dextrose, and 32 sucrose, pH 7.4. Free [Ca²⁺]_i was calculated using the ligand-binding program EQ-CAL (Biosoft, Ferguson, MO, USA) with the following dissociation constant (K_d) values: 190 nm BAPTA (for 150 mm ionic strength) and 33, 25, 0.7 and 0.4 μm calmodulin. With 11 mm BAPTA and 2 μm calmodulin in the pipette solution, added CaCl₂ fixed the free Ca²⁺ at 27 nm (0.9 mm) or 1 μm (8.8 mm) (Lu et al. 1994).

Stock solutions of autocamtide-2-related inhibitory protein (AIP) (500 μm), adenylyl imidodiphosphate AMP-PNP (100 mm) were diluted daily with the internal solution to final concentrations of 2 μm and 10 mm, respectively. Stock solutions of KN-62 (10 mm in DMSO) and KN-04 (10 mm in DMSO) were diluted to a final concentration of 3 μm each with external solution immediately prior to use. RU-16 was the generous gift of Andy Czernik (Rockefeller University, NY, USA) and baculovirus expressed recombinant porcine CaMKIIγ_C the generous gift of Roger Colbran (Vanderbilt University, Nashville, TN, USA).

Current recording and analysis

Whole-cell channel activity was recorded with either an EPC-7 or an Axopatch 200A amplifier. Patch electrodes were fabricated from borosilicate glass (N-51A (KIMAX-51), Kimble Scientific, Vineland, NJ, USA) with resistances that ranged from 1.8 to 3.0 MΩ. Adherent cells were transferred to a recording chamber that was perfused continuously by gravity at a rate of 0.5 ml min⁻¹. All experiments were performed at room temperature (22–24 °C).

Current traces were acquired at 12.5 kHz and filtered at 2.5 kHz with an eight-pole low-pass Bessel filter (Frequency Devices, Haverhill, MA, USA). Data acquisition was performed using pCLAMP 6.0 (Axon Instruments, Foster City, CA, USA). Fast capacitative transients were minimized on line by the patch-clamp analogue compensation. Residual capacitative and leak currents were removed digitally using scaled hyperpolarizing steps of one-fourth amplitude (P/N 4).

Data analysis was performed using Clampfit (Axon Instruments) and Kaleidagraph (Synergy Software, Reading, PA, USA) software. The dependence of channel activation on voltage was determined using Ca²⁺ channel tail currents in response to 15 test depolarizations in 5 mV increments (-60 to +10 mV; 10.4 ms) from a holding potential of −90 mV upon repolarization to −60 mV (45 ms). Interpulsing time was 6 s to allow for recovery from inactivation. The dependence of channel inactivation on voltage was determined by holding at various potentials (−90 mV to −20 mV, in 5 mV intervals) for 6 s followed by a brief depolarizing test pulse to +20 mV (8 ms) to open available channels, followed by repolarization to −60 mV. Doubling the holding time to 12 s did not augment inactivation indicating the attainment of apparent steady-state inactivation. Tail currents were fitted to a single exponential plus a constant using the Chebyshev algorithm. We blanked the first 250 μs of the 45 ms fitting region to eliminate any possible contamination with residual capacitative current. The current amplitudes used to construct activation and inactivation curves corresponded to minimum values obtained from the exponential function at the end of the blanking period. We determined the voltage dependence of activation or inactivation of T-type channels by plotting the relative amplitude of the tail current vs. the test or holding (prepulse) potential. For each cell, the data set (15 test or holding potentials in 5 mV increments from −60 to +10 mV (activation) or −90 to −20 mV (inactivation)) was fitted to a Boltzmann distribution, given by the equations:

and

where k is the slope factor (mV/e-fold change), V_t is the test potential, V_p is the prepulse, and I_max is the maximum current. Activation data fitted significantly better to a Boltzmann distribution raised to the second power. Half-activation potential was calculated as: V_0.5 = 0.882(k) + V_0.25. We evaluated statistical differences using either Student' unpaired two-tailed t test, Mann-Whitney unpaired two-tailed non-parametric test, or Newman-Keuls non-parametric one-way analysis of variance, where appropriate. We assumed significance when P ≤ 0.05.

Western blotting

A screen of available CaMKII antibodies revealed that RU-16 was capable of detecting recombinant CaMKIIγc. This antibody was developed originally towards the peptide CaMKIIβ-(506–531) but targets a region with high homology among all CaMKII isoforms. Cells were transfected as described above for patch clamping and 48 h after transfection were homogenized in (mm): 20 Hepes, 5 MgCl₂, 1 EDTA, 10 % sucrose, and 100 μm phenyl methyl sulfonyl fluoride. Lysates (8 μg per 1ane) were resolved by electrophoresis on 10 % SDS-polyacrylamide gels and proteins were transferred onto PVDF membranes (Millipore, Bedford, MA, USA) at 40 mA overnight at 22 °C in glycine (200 mm), Tris (25 mm), SDS (0.1 %) and MeOH (20 %). Membranes were blocked (5 % milk/TBS-T, 1 h; room temperature), incubated with primary antibody (1:1000 in 5 % milk/TBS-T, 1 h), washed (6× 5 min TBS-T), incubated with secondary antibody (donkey, anti-rabbit, HRP-Amersham) 1:1250 in 5 % milk/TBS-T, 30 min) and washed again (6×5 min TBS-T) prior to visualization with ECL (Renaissance, DuPont-New England Nuclear, Boston, MA, USA) on autoradiographic film (XOMAT-AR, Kodak, Rochester, NY, USA).

PCR detection of the CaMKIIγ_C transcript

Neonatal bovine adrenal glomerulosa cells were prepared by collagenase digestions as described previously (Chen et al. 1999) from adrenal glands collected by Florida Biologicals (V. K. Daniels, Tampa, FL, USA) obtained from a local slaughterhouse. Total glomerulosa cell RNA was prepared using RNeasy Mini Kit (Qiagen, Valencia, CA, USA) according to the manufacturer' protocol, and 1 μg of total RNA was reverse transcribed using SuperScript II RT (Gibco/BRL). A set of primers: forward 5′-TGA CCA TCA ACC CCG-3′; reverse 5′-GTG AGG CGG ATG TAA GC-3′, were designed to recognize highly conserved regions in the regulatory and association domains in all CaMKII gene families that spanned the intervening variable regions and these were used in PCR reactions. Reactions were performed at 94 °C for 3 min, followed by 30 cycles at 94 °C for 1 min, at 55 °C for 45 s, at 72 °C for 1 min, and at an additional 72 °C for 5 min. PCR products were subcloned into the TA cloning vector pCR2.1 (Invitrogen, Carlsbad, CA, USA) before transformation into competent cells (INVαF’ Invitrogen, Rockville, MD, USA). Positive clones were detected by colour and identified through restriction enzyme digestion before sequencing. Sequences were compared to CaMKII sequences in GenBank.

RESULTS

Regulation of T-channel activity in 293 cells expressing Ca_v3.2 channels

In adrenal glomerulosa cells, activation of CaMKII induces a 10 mV hyperpolarizing shift in the voltage dependence of activation of T-type Ca²⁺ channels. Because in this cell type, Ca_v3.2 is the predominant channel isotype, we stably expressed T-type Ca²⁺ channels composed of full-length human Ca_v3.2 α1-subunits in 293 cells to study the regulation of this channel by CaMKII. Ca²⁺ currents were measured in the whole-cell voltage-clamp configuration using internal solutions that either promoted (1 μm free calcium; 2 μm calmodulin) or impaired (27 nm free calcium) the activity of CaMKII. Representative currents elicited near the half-maximal (−30 mV) and maximal potentials (+10 mV) for Ca_v3.2 channel activation are shown in Fig. 1A. At these and other test potentials (−60 to +10 mV), inward currents and tail currents (V_{repolarization} = −60 mV), recorded with internal solutions that activated CaMKII, were indistinguishable from those recorded with non-activating solutions (Fig. 1B). These data indicate that in this stable cell line an elevation in pipette Ca²⁺ in the presence of CaM did not potentiate current at negative potentials nor shift the half-maximal potential for activation (V_1/2) of Ca_v3.2 channels. Boltzmann fits to normalized tail current amplitudes yielded similar V_1/2 values of: 27 nm Ca²⁺, −23.7 ± 0.4 mV(n= 14 cells), and 1 μm Ca²⁺, −26.8 ± 0.2 mV(n = 9 cells). Therefore, the response of Ca_v3.2 channels in 293 cells differs markedly from that in adrenal glomerulosa cells where CaMKII activity modulates channel gating.

Figure 1. Elevation of pipette Ca²⁺/CaM fails to increase Ca²⁺ currents in 293 cells expressing Ca_v3.2 channels.

Slowly deactivating tail currents were elicited at −60 mV after various depolarizing test pulses (V_t = −60 to +10 mV; 10 ms) from a holding potential of −90 mV. A, sample currents at V_t = −30 and +10 mV from two cells with intracellular free Ca²⁺ fixed at 27 nm Ca²⁺ (○) or 1 μm Ca²⁺ + 2 μm CaM (•). Note that the two cells have similar maximal currents (I_max). B, voltage dependence of activation. Normalized amplitude of slowly deactivating tail current (I/I_max) plotted (means ± s.e.m.) vs. V_t for two data sets. Data were fitted to a squared Boltzmann distribution yielding half-maximal activation potentials of: 27 nm Ca²⁺, V_1/2 = −23.7 ± 0.4 mV (k = 10.2, r = 0.99; n = 14 cells) and 1 μm Ca²⁺, V_1/2 = −26.8 ± 0.2 mV * (k = 10.2, r = 0.99, n = 9 cells, where k is the slope factor and r is the regression coefficient. *Not statistically different from 27 nm Ca²⁺ (by Student' unpaired t test).

This lack of modulation of Ca_v3.2 channels could indicate that the signalling cascade utilized by CaMKII to regulate Ca_v3.2 channel activity in adrenal glomerulosa cells is absent in 293 cells. However, the regulation of Ca_v3.2 channels by CaMKII previously described occurs close to sites of Ca²⁺ entry since recombinant CaMKII that is constitutively active increases channel activity in patches excised from glomerulosa cell membranes, and endogenous CaMKII that co-segregates with patches of excised membranes increases unitary currents when activated by Ca²⁺/CaM. Alternatively, the lack of modulation of Ca_v3.2 channels in 293 cells could indicate that the isoform of CaMKII that transduces activity in adrenal glomerulosa cells is not expressed or is not properly co-localized.

CaMKIIγ_C transcripts are detected in bovine adrenal glomerulosa cells

To date, four separate genes encoding for CaMKII have been identified (α β, γ, δ). Each gene shares a high degree of sequence identity, yet because of alternative exon usage, gives rise to multiple splice variants that differ primarily in regions (variable region: V1, V2 and V3) upstream of the kinase' association domain (Braun & Schulman, 1995). Since these flanking sequences and the C-terminal association domain facilitate multimerization of kinase monomers and tethered non-kinase target proteins, αCaMKII-association protein (αKAPs), the expressed isoform of the kinase, may dictate its subcellular location, and thus its capacity to regulate ion channel activity (Bayer et al. 1998). To identify the CaMKII mRNA transcript(s) expressed in bovine adrenal glomerulosa cells, we used RT-PCR with primers targeted to highly conserved regions in the regulatory and association domains to amplify products that spanned the variable regions. These primers were designed to hybridize with sequences across CaMKII gene families and across genera. Two PCR products of 599 and 554 bp were amplified. The first clone (599 bp) lacked two sequence insertions encoding 21 (V1) and 23 (V2) amino acids, each of which has been shown to be absent in CaMKIIγ_C isoforms (Singer et al. 1997). The nucleotide sequence of this bovine PCR product is 93 % identical to published CaMKIIγ_C-subunit sequences for pig, rabbit and rat, and has a predicted amino acid sequence that is 100 % identical, identifying it as bovine CaMKIIγ_C (bγ_C). A second clone (554 bp) lacked the V1 and V2 insertions as well as the intervening sequence (AAKSLLNKKSDGGVK) between these two regions that is conserved among all reported CaMKII subtypes. Its sequence was otherwise identical to that of bγ_C and on this basis it was identified as a putatively new γ-subunit variant (bγ_X) (Fig. 2). The identification of the γ subtype of CaMKII in bovine adrenal glomerulosa cells is consistent with its demonstrated expression in numerous peripheral tissues (Tobimatsu & Fujisawa, 1989; Mayer et al. 1994; Singer et al. 1997).

Figure 2. Identification of two partial CaMKIIγ_C transcripts in bovine glomerulosa cells.

Deduced amino acid sequences from two bovine glomerulosa cell RT-PCR products (γ_C and γ_X) amplified using bovine glomerulosa cell RNA and primers to highly conserved regions in the regulatory and association domains in all CaMKII gene families that spanned the intervening variable regions. The bovine sequences were aligned showing deletion regions (V1 and V2) absent in CaMKIIγ subunits previously designated as γ_C. The conserved domain separating the V1 and V2 regions was absent in γ_X. The sequences have been deposited in GenBank with the following accession numbers: bγ_C, AF 389986; and bγ_X, AF389987.

CaMKIIγ_C activity modulates T-channel gating

To test whether the lack of modulation of Ca_v3.2 channel currents by Ca²⁺/CaM was the result of insufficient expression of CaMKIIγ_C, 293 cells stably expressing human Ca_v3.2 α1-subunits were co-transfected with the cDNA for CaMKIIγ_C and GFP. In GFP-expressing cells, the voltage dependence of activation (Fig. 3A and B) and inactivation (Fig. 3C and D) of Ca_v3.2 channels was assessed using voltage command protocols that elicited tail currents. Elevation of pipette Ca²⁺ and CaM increased inward Ca_v3.2 channel currents and the associated tail currents at depolarizing test potentials, but not at potentials positive to 0 mV (Fig. 3A). The I_max at 27 nm was 2.5 ± 0.4 nA (n = 17 cells) vs. that at 1 μm which was 2.4 ± 0.3 nA (n = 19 cells; n.s.) The selective increase in current at negative potentials resulted in an 11 mV hyperpolarizing shift in the V_1/2 of activation. The half-activation potential shifted from −23.1 ± 0.3 mV (n = 21) to −34.0 ± 0.5 mV (n = 21), P < 0.05 with Ca²⁺/CaM. The magnitude of the change in activation gating was comparable to that induced by CaMKII activation in adrenal glomerulosa cells. In contrast, elevation of pipette Ca²⁺/CaM did not alter currents evoked by apparent steady-state inactivation voltage protocols (Fig. 3C). The half-maximal potential for inactivation remained unchanged at −67.1 ± 0.2 mV (n = 11) in the presence of Ca²⁺/CaM (Fig. 3D). A lack of modulation by Ca²⁺/CaM of the voltage dependence of inactivation of Ca_v3.2 channels was also observed in adrenal glomerulosa cells. Thus, the heterologous expression of CaMKIIγ_C in 293 cells stably expressing Ca_v3.2 channels enabled regulation by Ca²⁺/CaM, suggesting that CaMKIIγ_C is poorly expressed in 293 cells. Western blots using a CaMKII antibody (RU-16) generated against a peptide sequence that is common to all CaMKII isoforms supported this conclusion (Fig. 3B inset). No detectable CaMKIIγ_C protein expression was observed in untransfected 293 cells compared with the robust protein expression in cells transfected with the CaMKIIγ_C plasmid.

Figure 3. Elevation of pipette Ca²⁺/CaM induces a hyperpolarizing shift in the V_1/2 of activation of Ca_v3.2 channels in 293 cells expressing CaMKIIγ_C.

Slowly deactivating tail currents were elicited at −60 mV, after various depolarizing test pulses (V_t = −60 to +10 mV; 10 ms) from a holding potential of −90 mV (activation) or after depolarizations (+20 mV, 8 ms) from various prepulse potentials (V_P = −90 to −20 mV) lasting 6 s (inactivation). A, activation. Sample currents at V_t = −55, −45, −35, −25, −5 or +5 mV from two cells with intracellular free Ca²⁺ fixed at 27 nm (○) or 1 μm + 2 μm CaM (•). Note the larger amplitude of currents at hyperpolarized potentials recorded with free Ca²⁺ fixed at 1 μm (•). B, voltage dependence of channel activation. Relative amplitude of tail currents (means ± s.e.m.) plotted vs. V_t for two data sets. Data were fitted to a squared Boltzmann distribution, yielding half-maximal activation potentials of: 27 nm Ca²⁺, V_1/2 = −23.1 ± 0.3 mV (k = 11.2, r = 0.99; n = 21 cells); and 1 μm Ca²⁺, V_1/2 = −34.0 ± 0.5 mV * (k = 10.2, r = 0.99, n = 21 cells). * Statistically different from 27 nm Ca²⁺ (P < 0.05 by Mann-Whitney unpaired non-parametric test). Inset, immunoblot of cell lysates from Ca_v3.2-expressing cells transfected with: CaMKIIγ_C and GFP (γ_C transfected), or GFP alone (untransfected). Recombinant CaMKIIγ_C protein served as standard. C, inactivation. Sample currents at V_P = −90, −70, −60, −55, −50 and −40 mV from two cells with intracellular free Ca²⁺ fixed at 27 nm (▵) or 1 μm + 2 μm CaM (▾). D, voltage dependence of channel inactivation. Relative amplitude of tail current (means ± s.e.m.) plotted vs. V_p for two data sets. Data were fitted to a Boltzmann distribution, yielding half-maximal inactivation potentials of: 27 nm Ca²⁺, V_1/2 = −67.1 ± 0.2 mV (k = 6.6, r = 0.99, n = 10 cells); and 1 μm Ca²⁺, V_1/2 = −67.2 ± 0.2 mV * (k = 6.8, r = 0.99, n = 9 cells). * Not statistically significant from 27 nm Ca²⁺ (by Student' unpaired t test).

Based on the recorded change in the voltage dependence of activation, it was not surprising that Ca²⁺/CaM also altered the current-voltage relationship for Ca_v3.2 channels. Figure 4A shows the mean Ca_v3.2 current elicited upon depolarization to −50 mV from cells recorded with CaMKII-activating (n = 11) or non-activating (n = 10) internal solutions. Although peak current was detected at equivalent times during the 37 ms test depolarization to −50 mV (33 vs. 29 ms; values not significantly different), Ca_v3.2 channel current was enhanced by 150 % with Ca²⁺/CaM. An enhancement of peak inward current by Ca²⁺/CaM was observed within the voltage range of −60 to −35 mV (Fig. 4B) without significant increases recorded at potentials positive to −35 mV.

Figure 4. Ca²⁺/CaM shifts the current-voltage relationship of Ca_v3.2 channels to negative potentials.

A, mean traces (n = 10 and 11) at V_t = −50 mV show inward currents elicited from cells with pipette Ca²⁺ fixed at 27 nm (○) or 1 μm (•). B, I-V relationship. Peak inward current normalized to cell capacitance plotted vs. test potential for cells recorded with CaMKII-activating (n = 11 cells) or non-activating solutions (n = 10 cells). Note that with CaMKII-activating solutions Ca_v3.2 currents are enhanced within the range of −60 to −35 mV.

The participation of CaMKIIγ_C in the regulation of Ca_v3.2 channels by Ca²⁺/CaM was evaluated further using selective inhibitors of the kinase. Autocamtide-2-related inhibitory peptide (AIP), a synthetic peptide containing residues of the autoinhibitory domain of CaMKII and a selective inhibitor of CaMKII, was used to block regulation. The inclusion of this peptide in the patch pipette (2 μm) abolished the increase in Ca_v3.2 channel activity evoked by Ca²⁺/CaM (Fig. 5A). In the presence of Ca²⁺/CaM, AIP maintained the half-maximal potential for channel activation at −22.7 ± 0.3 mV (n = 11), a value that was indistinguishable from non-modulated channels (-23.1 ± 0.3 mV, Fig. 4B). In addition, Ca_v3.2 channel currents were not reduced by AIP when recording with a non-activating internal solution (27 nm Ca²⁺); the V_1/2 remained unchanged at −21.5 ± 0.3 mV(n = 12). These data indicate that AIP selectively blocks the Ca²⁺-induced increase in Ca_v3.2 channel activity and further implicates CaMKII as the mediator of this change in channel gating.

Figure 5. Autocamtide-2-related inhibitory protein (AIP), a specific peptide inhibitor of CaMKII, blocks Ca²⁺/CaM-induced potentiation of Ca_v3.2 currents in 293 cells expressing CaMKIIγ_C.

A, representative traces at V_t = −30 mV and +10 mV show tail currents (V_r= −60 mV) elicited from two cells with 2 μm AIP in the pipette solution. Intracellular free Ca²⁺ was fixed at 27 nm (○) or 1 μm (•). Note that in the presence of AIP, Ca²⁺/CaM did not potentiate Ca_v3.2 currents. B, voltage dependence. Relative amplitude of tail current (means ± s.e.m.) plotted vs. V_t for two data sets. Half-maximal activation potentials were calculated as in Fig. 3B; 27 nm Ca²⁺ + AIP, V_1/2 = −21.5 ± 0.3 mV (k = 12.4, r = 0.99, n = 12 cells); and 1 μm Ca²⁺ + AIP, V_1/2 = −22.7 ± 0.5 mV * (k = 12.4, r = 0.99, n = 11 cells). * Not statistically significant from 27 nm Ca²⁺ + AIP (by Student' unpaired t test). Dashed lines show comparison fits to data points obtained at 27 nm and 1 μm Ca²⁺ without AIP in the pipette, data from Fig. 3B.

To strengthen this conclusion, we tested the ability of other pharmacological agents to abrogate activation. An obligatory role for kinase activity in the observed regulation was evaluated by replacing ATP in the patch pipette with AMP-PNP, a non-hydrolysable analogue of ATP. As illustrated in Fig. 6, AMP-PNP failed to support the potentiation of Ca_v3.2 channel current by Ca²⁺/CaM and the V_1/2 remained at −26.2 ± 0.3 mV(n = 14) in the presence of Ca²⁺/CaM. Finally, the effect of a CaMKII inhibitor with a mechanism of action which differed from that of AIP was assessed. Unlike AIP that inhibits the active kinase by competing at each of two substrate-binding sites (autophosphorylation substrate site, exogenous substrate site), KN-62 prevents the activation of the kinase by competing with CaM for binding to the kinase. Preincubation with KN-62 (3 μm, 30 min) completely prevented Ca²⁺/CaM from shifting the V_1/2 of activation of Ca_v3.2 channels. This effect was specific to inhibition of kinase activity because the inactive analogue, KN-04, did not prevent Ca²⁺/CaM from activating the channel (Fig. 6). Taken together, the above experiments indicate that activation of CaMKII induces a change in the activation gating properties of Ca_v3.2 channels that mimic those induced by CaMKII in adrenal glomerulosa cells.

Figure 6. CaMKII activity is required for Ca²⁺/CaM to modify the gating of Ca_v3.2 channels.

V_1/2 of activation (means ± s.e.m., n = number of cells) was calculated for each cell recorded with 1 μm or 27 nm Ca²⁺ in the absence and presence of kinase inhibitors in the bath or pipette solutions. Dashed line indicates baseline V_1/2 determined in Fig. 3B at 27 nm Ca²⁺ in the absence of inhibitors. * Statistically significant from 27 nm Ca²⁺ (P < 0.05 by non-parametric one-way ANOVA). AMP-PNP (3 μm), poorly hydrolysable ATP analogue; AIP (2 μm), specific CaMKII-peptide inhibitor; and KN-62 (3 μm), membrane-permeable CaMKII-inhibitor, significantly reduced Ca²⁺-induced gating change. KN-04 (3 μm), inactive CaMKII-inhibitor, preserved gating change. Neither AIP nor KN-04 (not shown) changed control V_1/2 of activation at 27 nm Ca²⁺.

CaMKIIγ_C increases steady-state window current by modifying Ca_v3.2 channel gating

The impact of the CaMKII-evoked change in channel gating on steady-state Ca_v3.2 current was evaluated by determining the predicted window current associated with each channel state. The area of overlap in the activation and inactivation curves identifies a range of voltages over which channels are predicted to be open in the steady state. Since inactivation was unaltered by CaMKII activity, the more hyperpolarized V_1/2 of activation for CaMKII-modulated channels increased the fraction of Ca_v3.2 channels that are predicted to be open at negative membrane potentials (Fig. 7A). As shown in Fig. 7B, the effect of CaMKII on steady-state open probability (P_o) was evident over a wide range of negative membrane potentials (−90 to −20 mV) but was maximal at −55 mV. At −55 mV, steady-state P_o increased by 295 % (Fig. 7B). The relationship between steady-state open probability (m_∞²h_∞) and channel current is defined by the equation I = ((m_∞²h_∞)gN (V_t - V_r)), where, m_∞ and h_∞ are the voltage dependencies of activation and inactivation respectively, g is the unitary conductance, N is the number of channels, V_t is the test potential and V_r is the reversal potential. To estimate steady-state whole-cell current for each channel state we used a maximal open probability of 0.25 (Shuba et al. 1991), a unitary conductance of 4.7 pS in 10 mm Ca²⁺ (Balke et al. 1992), a Ca_v3.2 channel population of 13 400 in stably transfected 293 cells, and a theoretical reversal potential (V_r) of +52 mV. The latter value was used to estimate physiological steady-state current in lieu of the thermodynamic equilibrium potentials of +162 and +116 mV for non-activating and CaMKII-activating Ca²⁺ solutions, respectively, in order to account for the potassium ion permeability of Ca²⁺ channels in physiological solutions and Ca²⁺ channel rectification at positive membrane voltages (Hille, 1992). As illustrated in Fig. 7C, CaMKII-modulated channels are predicted to carry 270 % more current at −55 mV than non-modulated channels (52 vs. 14 pA). Thus, an 11 mV shift in activation gating evoked by CaMKII is estimated to greatly increase steady-state Ca²⁺ channel current.

Figure 7. CaMKII activation increases Ca_v3.2 window current.

A, predicted voltage range for observable Ca_v3.2 window currents. Overlap of Boltzmann distributions describing the voltage dependence of Ca_v3.2 activation (m_∞²) and inactivation (h_∞) determined from data in Fig. 3 defines voltage ranges for discernible window currents. Data were obtained with non-activating (continuous lines) or CaMKII-activating (dashed lines) pipette solutions. Note, the more hyperpolarized voltage range defined for CaMKII-activated cells. B, estimated steady-state channel open probability (P_o). Theoretical steady-state P_o (m_∞²h_∞) for Ca_v3.2 channels plotted vs. membrane potential for control (continuous line) and CaMKII-activated (dashed line) cells. Note, at −55 mV CaMKII activity increases P_o 295 %. C, predicted steady-state window current. Theoretical steady-state current calculated as: I = ((m_∞²h_∞)g N (V_t - V_r)), assuming g = 4.7 pS in 10 mm Ca²⁺, N = 13 400 Ca_v3.2 channels per cell, and that V_reversal = +52 mV. Note that CaMKII activity is predicted to increase Ca_v3.2 steady-state current by 295 % at −55 mV and 220 % at −80 mV. D, Ca²⁺ channel currents. Representative traces averaged from two cells at V_t = −60, −55, −50, −45, −35 and −30 mV. Intracellular free Ca²⁺ was fixed at 27 nm (upper panel) or 1 μm (lower panel). Note sustained Ca²⁺ channel current at 600 ms. Scale bar applies to both plots. E, measured channel open probability (P_o). Slowly deactivating tail currents (V_r = −90 mV) elicited following a 600 ms depolarization (V_t as in D) were averaged among cells with intracellular free Ca²⁺ fixed at 27 nm (○, n = 7 cells) or 1 μm (•, n = 8 cells). Relative amplitude of averaged tail current plotted vs. V_t for two data sets. The tail I_max (determined at V_r = −90 mV following V_t to +20 mV for 10 ms) at 27 nm = 3423 pA (n = 7 cells) and at 1 μm = 2897 pA (n = 8 cells). Note that at −55 mV CaMKII activity increases P_o by 195 %. F, maintained Ca²⁺ channel currents. Representative traces averaged from two cells at V_t = −50 and −45 mV where channel open probability is greatest. Intracellular free Ca²⁺ was fixed at 27 nm (upper panel) or 1 μm (lower panel). Scale bar applies to both plots. Both inward current and tail current amplitudes are increased by CaMKII.

To strengthen this prediction, we measured Ca²⁺ channel currents during long test depolarizations, and used tail currents to assess channel open probability. CaMKII-activating solutions (Fig. 7D, lower panel) selectively increased peak currents from 45 to 200 % at negative test potentials (-60 to −30 mV) without increasing the maximal currents recorded at +20 mV (not shown), and at potentials negative to −40 mV CaMKII continued to augment the small Ca²⁺ currents remaining at 600 ms. We used tail currents to amplify these remaining currents (Fig. 7F) and quantify channel open probability (I_tail/I_tailmax). As is evident in Fig. 7E, CaMKII increased steady-state P_o over a wide range of membrane voltages and induced a maximal effect at −55 mV. However, the relative magnitude of the increase effected by CaMKII-activating solutions was approximately 2-fold, reduced from the 3-fold predicted increase in steady-state channel open probability. Comparison of the measured and estimated steady-state P_o distributions indicates that the reduction in efficacy was solely attributable to an underestimation of the extent of channel opening in cells recorded with non-activating solutions, a possible consequence of the determination of the voltage dependence of activation by isochronal tail currrent measurements. Nonetheless, out data further support the prediction that CaMKII induces a substantial increase in Ca²⁺ channel window current (Fig. 7F).

CaMKIIγ_C does not potentiate the activity of Ca_v3.1 channels

At present, the T-type Ca²⁺ channel family consists of three family members: Ca_v3.1, 3.2 and 3.3, also known as α1G, α1H and α1I, respectively. These channels share a 90 % identity across their transmembrane domains but are only 40 % identical when the entire channel sequences are compared. As sequence divergence across intracellular loops and termini could confer altered susceptibility to regulation, we tested whether CaMKII modified the gating of Ca_v3.1 channels. 293 cells stably expressing full-length human Ca_v3.1 α1-subunits were co-transfected with the cDNA for CaMKIIγ_C and GFP. In GFP-expressing cells, CaMKII activation failed to increase Ca_v3.1 currents (Fig. 8A) and did not shift the half-activation potential (Fig. 8B). The V_1/2 of activation of Ca_v3.1 channels determined with non-activating solutions was −30.1 ± 0.3 mV remaining indistinguishable from that determined with kinase-activating solutions (-32.1 ± 0.3 mV; n.s.). As illustrated by the Western blot inserted in Fig. 8B, the lack of modulation of Ca_v3.1 by CaMKIIγ_C could not be attributed to poor expression of the kinase construct since robust protein expression of CaMKIIγ_C was detected in the transfected Ca_v3.1 stable cell line using a pan-CaMKII antibody (RU-16). Thus, our data indicate that CaMKIIγ_C differentially regulates T-type Ca²⁺ channel α1-subunits.

Figure 8. Elevation of pipette Ca²⁺/CaM fails to increase Ca_v3.1 currents.

A, representative traces at V_t −30 mV, +10 mV in two cells with intracellular Ca²⁺ fixed at 27 nm (○) or 1 μm (•). Note Ca²⁺/CaM did not potentiate Ca_v3.1 currents. B, voltage dependence. Relative amplitude of tail current (means ± s.e.m.) plotted vs. V_t for two data sets. Half-maximal potentials were calculated as above; 27 nm Ca²⁺, V_1/2 = −30.1 ± 0.3 mV (k = 8.2, r = 0.99, n = 12 cells); and 1 μm Ca²⁺, V_1/2 = −32.1 ± 0.3 mV * (k = 7.9, r = 0.99, n = 10 cells). * Not statistically different (by Student' unpaired t test). Inset, immunoblot of cell lysates from Ca_v3.1-expressing cells transfected with: CaMKIIγ_C and GFP (γ_C transfected), or GFP alone (untransfected). Recombinant CaMKIIγ_C protein served as standard. Note inadequate expression of CaMKIIγ_C cannot account for the lack of Ca_v3.1 regulation by CaMKIIγ_C.

DISCUSSION

We have provided evidence for the regulation of Ca_v3.2 channels by CaMKII. Activation of CaMKIIγ_C transfected into 293 cells stably expressing full-length Ca_v3.2 α1-subunits increases channel current at negative test potentials, shifting the half-maximal potential for channel activation by −11 mV. This regulation is effected by Ca²⁺/CaM in the patch pipette and depends upon the expression of CaMKIIγ_C. Inhibition of CaMKII activity with peptide or lipophilic CaMKII antagonists prevents the modification of channel gating, as does replacement of pipette ATP with AMP-PNP. Channel activity measured with non-activating solutions remains unaltered by similar manipulations. Despite the robust expression of CaMKIIγ_C in 293 cells stably expressing Ca_v3.1 α1-subunits, the activity of Ca_v3.1 channels did not depend upon the state of activation of CaMKII. Thus, unexpectedly, Ca_v3.1 channels are not the molecular substrates for functional regulation by CaMKII.

CaMKII regulation of Ca²⁺ channels

Our data provide the first evidence for the regulation of a Ca_v3 family member by protein kinase activation and the first report reconstituting CaMKII-dependent regulation of any cloned Ca²⁺ channel. In heart and smooth muscle, native HVA Ca²⁺ channels are regulated by CaMKII, although the molecular basis for this regulation remains unclear (Gurney et al. 1989; McCarron et al. 1992). In a variety of cardiac preparations, repetitive test depolarizations induces a frequency-dependent potentiation of L-type HVA current (facilitation) that depends upon Ca²⁺ entry (Anderson et al. 1994; Yuan & Bers, 1994; Vinogradova et al. 2000; Zuhlke et al. 2000). This autoregulation of whole cell L-type channel current can be disrupted by inhibiting CaMKII activity (Anderson et al. 1994; Yuan & Bers, 1994; Vinogradova et al. 2000) or by preventing Ca²⁺/CaM binding to the IQ motif within the C-terminal cytoplasmic tail of the channel (Zuhlke et al. 2000), suggesting that two molecular events may be required for Ca²⁺-induced potentiation of Ca_v1.2 current. In contrast, studies using an engineered Ca²⁺/CaM-independent CaMKII show that CaMKII activity itself is sufficient to induce a high open probability mode of single channel gating that is characterized by long openings (mode 2) (Dzhura et al. 2000) whereas Ca²⁺/CaM alone failed to support mode 2 gating. Thus, direct Ca²⁺/CaM binding to Ca_v1.2 channels might facilitate channel regulation by CaMKII operating in its Ca²⁺/CaM-dependent state. In our studies, a role for Ca²⁺/CaM-binding to Ca_v3.2 channels was not evaluated directly. However, elevation of Ca²⁺/CaM in the absence of CaMKIIγ_C transfection did not increase Ca_v3.2 channel activity. Furthermore, a requirement for Ca²⁺/CaM binding, in addition to CaMKII activation, is not likely since the identified Ca²⁺/CaM-binding IQ motif in Ca_v1.2 channels is lacking in the Ca_v3 channel family (Lee et al. 1999).

Channel regulation by phosphorylation

Although L-type Ca²⁺ channels are regulated by CaMKII activity and are suitable substrates of the kinase in vitro (Jahn et al. 1988), the modulatory phosphorylation sites have yet to be identified. In contrast, β-adrenergic regulation of native cardiac Ca²⁺ channels is well documented in vitro (Jahn et al. 1988) and in vivo (Haase et al. 1993, 1996) as is the in vitro modulation of cloned Ca_v1.2 channels (Sculptoreanu et al. 1993a, b) by protein kinase A (PKA). Three functionally important phosphorylation sites have been identified. Although the relative importance of these sites remains in dispute, PKA-induced phosphorylation of one residue on the α1-subunit C-terminal tail (Gao et al. 1997) and/or two residues on the β2_a-subunit (Bunemann et al. 1999) controls the ion conductance activity of Ca_v1.2 channels. Similarly, phosphorylation of two residues on the N-terminus of α1-subunits of cardiac Ca_v1.2 channels underlies the selective PKC-induced inhibition of the cardiac isoform (McHugh et al. 2000). Thus, phosphorylation of the channel complex underlies the regulation of Ca_v1.2 channel activity by kinases. Lacking the conserved, consensus β-subunit binding motif (α1-β interaction domain) that is present in the I-II loop of all HVA channels (De Waard et al. 1996), the biophysical properties of Ca_v3 family members are not substantively altered by β-subunits (Dolphin et al. 1999). Thus, if modulation follows paradigms similar to those described above, the α1-subunit of Ca_v3.2 channels is most probably a substrate of CaMKII. Within Ca_v3.2 channels there are at least 19 potential phosphorylation sites that conform to the minimal CaMKII substrate recognition motif (R-X-X-S/T) (Soderling, 1996), which are distributed among the N- and C-termini, and the intracellular loops linking each of the four domains. Among these sites, many are unique to Ca_v3.2 channels, as they are absent in Ca_v3.1 channels. Since the gating properties of Ca_v3.1 channels were not altered by CaMKII activity, these unique sites are candidates for mediating functionally important changes in channel activity. However, this hypothesis remains to be proven, particularly in view of the fact that the PKA-induced changes in the phosphorylation state of β2_a-subunits that mediate ion conductance activity occur on non-classical PKA consensus residues (Gerhardstein et al. 1999).

CaMKII-channel interactions

Our data indicate that CaMKIIγ_C displays selectivity in regulating Ca_v3.2 channel activity. CaMKIIα also shows selectivity in co-localizing with and binding to NMDA-glutamate receptors. NR2B, but not NR2A or NR1, channel subunits act as targeting subunits for autophosphorylated-CaMKIIα, bringing it to the membrane (Strack & Colbran, 1998). Non-conserved amino acids within the core-binding domain of NR2B are critical for high affinity binding (Strack et al. 2000). Yet, monomeric CaMKIIα mutants that lack the ability to multimerize, but retain fully functional catalytic activity, are not able to bind to NR2B subunits, providing evidence that the COOH-terminal association domain of CaMKII is also an important determinant in channel targeting (Strack et al. 2000). In preliminary studies, we have observed that the Ca²⁺/CaM-dependent potentiation of Ca_v3.2 current is not supported by all isoforms of CaMKII tested (results not shown). Since only the catalytic and regulatory domains of CaMKII are largely conserved across all isoforms, differential regulation of Ca_v3.2 channels by CaMKII subtypes could indicate differential targeting to the channel.

Physiological relevance of gating shift

The physiological consequence of the CaMKIIγ_C-induced gating shift reported here is to extend the voltage range over which steady-state window current occurs. Our laboratory has previously reported that in physiological K⁺ solutions (2–5 mm) and temperatures (37 °C) the adrenal glomerulosa cell maintains a membrane potential that is close to the K⁺ equilibrium potential E_K (Chen et al. 1999). As these potentials overlap only minimally with the foot of the activation curve determined at 37 °C with physiological Ca²⁺ as the charge carrier (1.25 mm), shifts in the half-maximal potential for channel gating induced by agonists will greatly alter at negative membrane potentials the magnitude of Ca²⁺ channel flux that is necessary for sustaining stimulated aldosterone production. We predict here that activation of CaMKIIγ_C will increase steady-state Ca_v3.2-channel current by ∼2-fold at membrane voltages near E_K. We note that this prediction could be an underestimate because of uncorrected charge screening by 10 mm Ca²⁺ or an overestimate because the squared Boltzmann function used to fit the activation curve may over-predict channel activity at the foot of the activation curve. Nonetheless, our measurements of sustained Ca²⁺ channel current following a 600 ms depolarization indicate that CaMKII increases Ca²⁺ channel window current at membrane voltages positive to −60 mV and further support the prediction of an increase in steady-state current at membrane voltages near E_K. Interestingly, CaMKII also increases L-type Ca²⁺ channel steady-state window current in sinoatrial node cells; however, in this paradigm the underlying gating change is a depolarizing shift in the half-maximal potential for inactivation (Vinogradova et al. 2000). Thus, by amplifying Ca²⁺-channel window current, modulation of Ca²⁺ channel activity by CaMKII can play an important regulatory role in physiological processes as diverse as aldosterone production in non-excitable cells and pacemaker activity in excitable cells.