A Mechanism for the Direct Regulation of T-Type Calcium Channels by Ca²⁺/Calmodulin-Dependent Kinase II

Introduction

Low-voltage-activated (LVA), T-type, Ca²⁺ channels are important determinants of neuronal excitability (McCormick, 1992; Huguenard, 1996; Eilers and Konnerth, 1997; Hausser et al., 2000). They support the forward propagation of distal dendritic inputs in neocortical (Markram and Sakmann, 1994) and CA1 hippocampal pyramidal neurons (Magee and Johnston, 1995a,b) and mediate low-threshold spikes that support somatic burst firing in thalamocortical (Huguenard and McCormick, 1992; Destexhe et al., 1996) and CA3 hippocampal pyramidal neurons (Fisher et al., 1990; Migliore and Shepherd, 2002). Both dendritic boosting and neuronal bursting supported by LVA channels amplify subthreshold stimuli (Hausser et al., 2000). LVA Ca²⁺ channels can also facilitate long-term changes in neuronal plasticity as demonstrated in the dorsal horn (Ikeda et al., 2003) and CA1 pyramidal neurons (Su et al., 2002). Thus, LVA Ca²⁺ channel activity can contribute to the complex and diverse firing behavior of neurons.

LVA Ca²⁺ channels belong to the Ca_v3 family, of which there are three family members (Ca_v3.1, Ca_v3.2, and Ca_v3.3: α_1G, α_1H, and α_1I) (Cribbs et al., 1998; Perez-Reyes et al., 1998; Lee et al., 1999). Little is known about the regulation of this Ca²⁺ channel class, in contrast to high voltage-activated Ca²⁺ channels that are well-established targets of G-proteins and kinases (Catterall, 2000). One defined mode of LVA channel regulation depends on the activity of Ca²⁺/CaM-dependent protein kinase II (CaMKII), an important regulator of ion channel activity (Soderling and Derkach, 2000; Hudmon and Schulman, 2002). Our previous work established that CaMKII modulates α_1H but not α_1G channel gating, preferentially increasing current at subthreshold potentials (Wolfe et al., 2002). Because the activity of multimeric CaMKII depends on Ca²⁺/CaM binding and intersubunit autophosphorylation, CaMKII activity can extend beyond the lifetime of the Ca²⁺ signal and enable the kinase to sum and thereby respond to small repeated Ca²⁺ transients (Soderling and Derkach, 2000; Hudmon and Schulman, 2002). Thus, CaMKII-dependent regulation of LVA channel activity could persist during weak synaptic stimulation.

CaMKII is ubiquitously distributed throughout the CNS and PNS and is a major constituent of postsynaptic densities of excitatory synapses (Kennedy et al., 1983; Kelly et al., 1984; Braun and Schulman, 1995; Liu and Jones, 1996; Hudmon and Schulman, 2002). The widespread localization of this kinase (Braun and Schulman, 1995) and LVA channels (Perez-Reyes, 2003) in both dendritic and somatic compartments (Liu and Jones, 1996; Shen et al., 1998; Thiagarajan et al., 2002) and their independent roles in modulating synaptic transmission suggested a direct mechanism for the regulation of α_1H channels by CaMKII that would be of general importance to neuronal function.

Here, we show that the regulation of α_1H channels by CaMKII depends on structural determinants contained within the intracellular linker connecting channel domains II and III. A serine residue (Ser¹¹⁹⁸) in the H_II-III linker is a preferred CaMKII phosphorylation site and is critical to this modulation. These data identify the α_1H Ca²⁺ channel as a direct effector for CaMKII and provide a new CaMKII-dependent mechanism for enhancing neuronal excitability.

Materials and Methods

Preparation of human α_1Gand α_1Hchimeric and mutant constructs. All constructs were prepared using a PCR-subcloning strategy in which PCR products were verified by DNA sequencing. A truncated α_1H sequence (AF051946 version 1.0) was extended by PCR to add a sequence encoding for 50 amino acids using 5′-GACAGCCCTAGGGACAC-3′ [5′-CGAGAATGCACTCGAGCTACACGGGGTCATC-3′ (AF051946 modified)]. The II-III linker regions were amplified from α_1H (AF051946 modified) and α_1G (AF190860) (Cribbs et al., 2000) were generated by PCR (α_1H, 5′-TCCTGTACAACGGCATGG-3′ and 5′-CTCTCTAGATATCAGCGTCTCCACC-3′; α_1G, 5′-CCTGTACAATGGTATGGCCTCC-3′ and 5′-CTCTCTAGACACGTGGTCGAACATCTTG-3′) and subcloned into pUC18 at the SmaI-XbaI site, resulting in pUC18-H_II-III and pUC18-G_II-III. The PCR reaction introduced an upstream BsrGI site in the H_II-III. The II-III linker of α_1G was isolated from pUC18-G_II-III by BsrGI-PmlI digestion and inserted into pUC18-H_II-III at the corresponding site to replace H_II-III, resulting in pUC18-G_II-III-1. Finally, the BsrGI-EcoRV fragment containing the II-III linker of α_1G was isolated from the last construct and subcloned back to pcDNA3-α_1H, resulting in construct H-G_II-III. Similarly, preparation of the construct G-H_II-III involved an intermediate construct pUC18-H_II-III-1 that was generated by subcloning a PCR (5′-CCAATGCATCTTTTCGGCTGCAAGTTC-3′ and 5′-CCATCTAGACACGTGATCAAACATCTTG-3′) product of H_II-III into pUC18 at the SmaI-XbaI site, in which an NsiI site was introduced upstream of H_II-III. The H_II-III linker was isolated from an NsiI-PmlI digestion of pUC18-H_II-III-1 and subcloned into pUC18-α_1G replacing the original II-III linker, resulting in pUC18-G-H_II-III. The chimearic channel was subcloned into pcDNA3.1(-) at the EcoRI-KpnI site, resulting in the construct G-H_II-III. All mutations were introduced into the intermediate constructs pUC18-H_II-III or pUC18-H_II-III-1 using QuikChange XL Site-Directed Mutagenesis Kit (catalog #200517; Stratagene, La Jolla, CA). Last, the construct G-H_Cterminus was prepared by fusing the 5′ section of α_1H and the 3′ section of α_1G at the PmlI site in pcDNA3.

Preparation of glutathione S-transferase fusion proteins. The wild-type DNA construct of the glutathione S-transferase (GST)-H_II-III fusion protein was generated by subcloning a PCR (5′-CATGGATCCCAGGCGGAGGGCGATG-3′ and 5′-CAAGATATCCGTGGGTCCAGGGACTC-3′) product of α_1H (α_1H3117-3696) into pGEX-2T at the BamHI-SmaI site. The mutations S1198A and/or S1153A were introduced by PCR. All PCR products were verified by DNA sequencing.

In vitro CaMKIIγc phosphorylation. GST-H_II-III fusion proteins were purified from transformed Escherichia coli BL21 cell lysates by binding to 200 μl of 50% (w/v) slurry of glutathione-Sepharose 4B (Amersham Biosciences, Arlington Heights, IL). After repeated washes with PBS, the GST-H_II-III fusion proteins were eluted from the beads with 20 mm reduced glutathione in a Tris-HCl buffer, pH 7.4. Purity of the isolated proteins was assessed by SDS-PAGE, and concentration was determined using a Coomassie Protein Assay (catalog #1856209; Pierce, Rockford, IL). It was estimated from Coomassie staining of SDS-PAGE samples that 50% of the final protein was full length. Phosphorylation of the GST-H_II-III fusion proteins was performed in a final volume of 50 μl containing 1 μm fusion protein, 1 mm DTT, 10 mm HEPES, 0.5 mm CaCl₂, 2 μm calmodulin, 10 mm MgCl₂, 0.1% Triton X-100, 1 mg/ml BSA, and 100 μm [³²P]ATP (10 mCi/ml). Reaction was initiated by the addition of 30 nm purified recombinant porcine CaMKIIγc, and samples were incubated at 37°C for up to 15 min. Reactions were terminated by addition of 20 μl of 4× SDS sample buffer containing 500 mm EGTA, followed by incubation at 95°C for 5 min before separation by SDS-PAGE. Gels were Coomassie stained, and, after autoradiography, bands corresponding to phosphorylated GST-H_II-III fusion proteins were cut from the gel, and radioactivity was assessed using liquid scintillation counting.

Electrophysiology. Whole-cell currents were recorded from adherent HEK293 cells using a standard patch electrode voltage-clamp method (Lu et al., 1994). To eliminate K⁺ currents and fix free [calcium], we used internal solutions containing the following (in mm): 115 CsCl, 1 tetrabutylammonium chloride, 1 MgCl₂, 5 Mg-ATP, 1 Li-GTP, 20 HEPES, pH.7.2 (adjusted with CsOH), and 11 BAPTA; added CaCl₂ fixed the free Ca²⁺ at 27 nm (0.9 mm) or 1 μm (8.8 mm) with 2 μm CaM (Lu et al., 1994). HEK293 cells were superfused with a solution containing the following (in mm): 127 TEA-Cl, 10 CaCl₂, 0.5 MgCl₂, 10 HEPES, 5 dextrose, and 32 sucrose, pH 7.4 (adjusted with CsOH). Currents were filtered at 2.5 kHz and sampled at 12.5 kHz; leak subtraction was performed on line using P/4 protocol. Activation gating was determined using tail currents in response to 15 test depolarizations in 5 mV increments (-60 to +10 mV; 10.4 msec) from a holding potential of -90 mV during repolarization to -60 mV (45 msec). Interpulsing time was 6 sec to allow for recovery from inactivation. Tail currents were fitted to a single exponential plus a constant using the Chebyshev algorithim in pClamp 6.0 software (Axon Instruments, Foster City, CA). As described previously (Wolfe et al., 2002), activation gating fitted significantly better to a Boltzmann distribution raised to the second power: [I/I_max = 1/(1 + exp[(V_0.25 - V_t)/k])², where k is the slope factor (mV/e-fold change), V_t is the test potential, and I_max is the maxium current. Half-activation potential was calculated as follows: V_0.5 = 0.882(k) + V_0.25. Data obtained with the CaMKII-activating internal solution were fitted to the sum of two Boltzmanns. Averaged data sets obtained with activating or non-activating solutions were analyzed concurrently to obtain a single set of parameters for the non-modulated component that optimally describes both sets of data and another set of parameters for the second modulated component. Parameter estimations were performed by a nonlinear least-squares procedure (Johnson and Frasier, 1985) using boot strapping to determine confidence limits (Efron and Tibshirani, 1993). To evaluate the goodness-of-fit to a single versus a double Boltzmann distribution, the sum of the squared residuals was calculated and mapped into a probability by calculating a Z score for the runs test (Straume and Johnson, 1992).

Results

Activation of CaMKII induces an 11 mV hyperpolarizing shift in the half-activation potential (V_1/2) of native and recombinant α_1H channels (Lu et al., 1994; Schrier et al., 2001; Wolfe et al., 2002). This regulation depends on active kinase (Lu et al., 1994; Barrett et al., 2000) and is not observed with human recombinant α_1G channels (Ca_v3.1) (Monteil et al., 2000) stably expressed in HEK293 cells (Wolfe et al., 2002). To test for the differential regulation of these two Ca_v3.0 family members in a transient transfection paradigm, we cotransfected CaMKII with either the α_1H or the α_1G channel construct. Channel activity was measured using internal solutions that either promote (1 μm free Ca²⁺; 2 μm CaM) or impair (27 nm free Ca²⁺) CaMKII activation (Wolfe et al., 2002). Ca²⁺/CaM potentiates α_1H currents at potentials at which channel activation is incomplete (-60 to 0 mV) (Fig. 1b). Neither currents recorded at maximally activating voltages (V_test =+10 mV; V_r = -60 mV) nor currents recorded from α_1G channels (Fig. 1b,e) are increased with CaMKII-activating internal solutions. The gating behavior of α_1H channels recorded with CaMKII-activating solutions is poorly fitted by a single squared Boltzmann function but is fitted significantly better (p < 0.05) by a double Boltzmann function that defines two channel populations. One population displays properties indistinguishable from channels recorded with impaired CaMKII activity [termed non-modulated (N-M)], whereas the other population [termed modulated (M)] displays a more hyperpolarized half-activation potential (V_1/2) and an increased voltage sensitivity (k) (Fig. 1c, inset). Concurrent analysis of the averaged data sets reveals that CaMKII activation modulates α_1H but not α_1G channels, inducing a -14.3 ± 1.6 mV (p < 0.05) shift in the half-activation potential and an Δ -4.5 ± 1.4 mV/e-fold (p < 0.05) increase in the voltage sensitivity of gating, as indicated by the reduction in slope factor (k). A concomitant change in the midpoint and the apparent steepness of gating suggests that CaMKII may change the work required to open the channel, possibly by reducing the electrostatic surface potential sensed by the voltage-sensing module (Hille, 2001). On average, CaMKII modulates 45 ± 9% of the α_1H channel population; however, in any given cell, modulation can be near complete or remain submaximal. An additional analysis of the regulation of α_1H channels shows that the CaMKII-modulated channel population in individual cells may be as large as 99% or as small as 21% of the total channel population (46 ± 5%). We interpret this variability as a byproduct of transient transfection because the expression of GFP, one cotransfectant, is also variable. Nonetheless, the regulation of α_1H but not α_1G channels in transient transfection suggests that structural determinants in the channel protein may underlie CaMKII-induced regulation.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Effect of CaMKII activity on the voltage dependence of activation of T-type Ca²⁺ channels. a, d, Schematic representation of α_1H (Ca_v3.2) and α_1G (Ca_v3.1) channels. b, e, Whole-cell current traces recorded from α_1H (b) or α_1G (e) channels transiently expressed with CaMKII in HEK293 cells. Sample currents at V_test = -55, -45, -35, -25, -10, and +10 mV recorded with intracellular free Ca²⁺ fixed at 27 nm Ca²⁺ (open symbols) to prevent CaMKII activation or 1 μm Ca²⁺ plus 2 μm CaM (filled symbols). Currents were normalized (I_max = 2.5 nA) to illustrate differences at hyperpolarized V_m between channel subtypes recorded with CaMKII activating solutions. c, f, Relative conductance (normalized tail current amplitudes, I/I_max) plotted (means ± SEM) versus V_t. Data sets were fitted to squared Boltzmann functions; single functions provided good fits to low Ca²⁺ data (open symbols) for α_1H (c) [V_1/2 = -23.2 ± 0.3mV; k = 10.9 ± 0.4 mV/e-fold change; n = 17 (27 nm)] and to both low- and high-Ca²⁺ data for α_1G channels (f) [V_1/2 = -29.2 ± 0.6 mV; k = 9.4 ± 0.4 mV/e-fold change; n = 16 (27 nm) and 11 (1 μm Ca²⁺)]. Fitting to the sum of two functions was significantly better (p < 0.05) for the high-Ca²⁺ data for α_1H. Inset shows the contributions of the two components fitted to the high-Ca²⁺ α_1H data set (T); one component is identical to the low-Ca²⁺ fit and is designated non-modulated (N-M), whereas the second modulated component (M) activates at more hyperpolarized voltages [M: V_1/2 = -37.5 ± 0.3mV; k = 6.4 ± 1.5 mV/e-fold change; n = 15 (1 μm Ca²⁺)].

To determine the molecular basis for this regulation, we generated chimeric channel constructs from wild-type α_1H (Ca_v3.2) and α_1G (Ca_v3.1) channels. Replacement of the H_II-III linker with the G_II-III linker abolishes CaMKII-dependent regulation (Fig. 2a-c) [H/G_II-IIIlinker: V_1/2 = -22.4 ± 0.3 mV, k = 11.4 ± 0.5 mV; n = 15 (27 nm) and 11 (1 μm Ca²⁺)], suggesting that the determinants for the regulation of activation gating by CaMKII are located in the H_II-III linker. Despite the absence of regulation, other channel properties remain unchanged. Neither the voltage dependency of inactivation (-65 ± 0.2 mV) nor the τ of deactivation (3.9 ± 0.3 msec at -60 mV) of the H/G_II-III linker chimera differs from that of α_1H wild-type channels (-64.5 ± 0.2 mV, 4.2 ± 0.5 msec; NS).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

The II-III cytoplasmic linker of α_1H channels confers modulation in cells coexpressing CaMKII. a, d, Schematic representation of channel chimeras in which the II-III linker plus a single transmembrane helix is swapped. b, e, Sample currents at V_t = -55, -45, -35, -25, -10, and +10 mV recorded with solutions that impair (open symbols) or promote (filled symbols) CaMKII activation as presented in Figure 1, b and e. c, f, Normalized tail current amplitudes (I/I_max) plotted (means ± SEM) versus V_t. Single squared Boltzman functions provided good fits for both low- and high-Ca²⁺ data for H/G_{II-III linker} (c) [V_1/2 = -22.4 ± 0.3 mV; k = 11.4 ± 0.5 mV; n = 15 (27 nm) and 11 (1 μm Ca²⁺)] and low-Ca²⁺ data (open symbols) for G/H_{II-III linker}. High Ca²⁺ data for G/H _{II-III linker} fitted significantly better (p < 0.05) to the sum of two functions (filled symbols) (f) [G/H_{II-III linker}, M: V_1/2 = -37.8 ± 0.6 mV; k = 5.3 ± 0.6 mV; n = 17 (1 μm Ca²⁺)]. Inset shows the fitted contribution of modulated (M) and non-modulated (N-M) channels to the voltage dependence of activation during CaMKII simulation (T). T, Sum of both fitted components for high Ca²⁺ data.

On the basis of the previous studies, we predicted that the H_II-III linker, when transferred to the α_1G backbone, would result in a gain of regulation and tested for CaMKII-dependent regulation of this G/H_II-IIIlinker chimera. CaMKII activity shifts the half-activation potential of this chimera by -9.4 ± 0.4 mV (p < 0.05) and also increases the voltage sensitivity by Δ -3.3 ± 0.6 mV/e-fold change (p < 0.05). On average, 75 ± 11% of the G chimeric channel population is modulated, displaying a half-activation potential that is nearly identical to that recorded for CaMKII-modulated α_1H wild-type channels [G/H_II-III linker: M, V_1/2 = -37.8 ± 0.6 mV, k = 5.3 ± 0.6 mV, n = 17 (1 μm Ca²⁺)]. An additional analysis of the regulation of G-chimeric channels shows that the CaMKII-modulated channel population in individual cells may vary from 89 to 46% of the total channel population (74 ± 4%). This broad distribution differs but overlaps that observed for wild-type α_1H channels. The singular importance of the H_II-III linker to the CaMKII-dependent change in channel gating was corroborated by the construction of a second G/H chimera that replaces a large segment of the α_1G channel from IIIS1 to the C terminus (residues 1294-2353), with the corresponding regions from α_1H channels. CaMKII-dependent regulation is not conferred to this G/H_Cterminus chimera (Fig. 3c). On the basis of these findings, we conclude that residue(s) critical for CaMKII-dependent regulation are located in the H_II-III linker.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Effect of single point mutations on CaMKII-induced changes in channel gating. a, Schematic representation of α_1H chimeras in which CaMKII-consensus motifs on the H_II-III domain linker are mutated (RAPA¹¹⁵³ and RAEA¹¹⁹⁸), alone and in combination, and α_1G chimeras with transferred α_1H domains. b, Sample currents at V_t = -55, -45, -35, -25, -10, and +10 mV recorded with solutions that impair (open symbols) or promote (filled symbols) CaMKII activation as presented in Figure 1, b and e. c, Fitted values for half-activation potential; open bars indicate low-Ca²⁺ fitted parameters (open bars), and filled bars indicate high-Ca²⁺ fitted parameters. Values are mean ± SE; ^*p < 0.05; n indicates number of cells. Normalized current waveforms recorded at V_t = -35 mV (10 msec, V_{repolarization} of -60 mV) with free Ca²⁺ fixed at 27 nm Ca²⁺ (open symbols) or 1 μm Ca²⁺ plus 2 μm CaM (filled symbols).

The α_1H channel is a large 259 kDa membrane protein containing 22 sites that conform to the minimal consensus sequence for CaMKII phosphorylation, RXXS/T (Soderling, 1996). We considered each site as a putative substrate for CaMKII using criteria established on the basis of the rank order of phosphorylation efficiency (V_max/K_m) of a combinatorial peptide library (White et al., 1998). The ¹¹⁹²LRRAESL¹²⁰⁰ recognition motif, located in the 275 residue H_II-III linker, contains hydrophobic residues at -5 and +1, a non-basic amino acid residue at -3, and an alanine (Ala) at the -2 position, determinants that are known to enhance substrate recognition (White et al., 1998). Notably, no other motif in the channel protein has as many advantageous residues for substrate recognition, and this motif is not found in the α_1G sequence. To determine whether the LRRAES¹¹⁹⁸L recognition motif on the H_II-III linker is a site of CaMKII phosphorylation, we expressed GST fusion proteins containing ∼70% of the H_II-III linker that includes Ser¹¹⁹⁸ as well as Ser¹¹⁵³, a strong but less optimal substrate site (¹¹⁴⁷LGRAPSLK¹¹⁵⁶). Recombinant CaMKII (at 30 nm) phosphorylates the full-length wild-type linker protein (47 kDa) as well as a proteolytic fragment (44 kDa) lacking the RAES¹¹⁹⁸ motif that resolves as a lower band of a doublet on SDS-PAGE (Fig. 4a,b). The full-length wild-type linker protein is phosphorylated to a stoichiometry of 1.7 mol of PO₄ incorporated per moles of protein at 1 min, indicating that more than a single residue is phosphorylated, and reaches saturation at 15 min, to attain a final stoichiometry of 2.3. Despite the presence of five CaMKII phosphorylation motifs on the H_II-III linker protein, the majority of incorporated phosphate at 1 and 15 min is distributed between Ser¹¹⁹⁸ and Ser¹¹⁵³. The combined mutation of these residues to Ala reduces phosphorylation by ∼80%. At 1 min, Ser¹¹⁹⁸ is preferentially phosphorylated to full stoichiometry as determined by mutational analysis. In contrast, mutation of Ser¹¹⁵³, which is embedded in a less optimal substrate recognition motif, does not result in significant loss of PO₄ incorporation at 1 min. Longer incubation allows for the slower phosphorylation of the RAPS¹¹⁵³ motif and the recognition of other poorer substrates. Because rapid stoichiometric phosphorylation of a motif is a signature feature of a preferred kinase substrate, our findings suggested that the phosphorylation of Ser¹¹⁹⁸ on the H_II-III linker could be a critical event that mediates CaMKII-dependent changes in channel gating.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Phosphorylation of GST-domain II-III linker by CaMKII. A, GST fusion proteins containing residues 1039-1232 of domain linker II-III (wild-type or mutated at RAPA¹¹⁵³ and/or RAEA¹¹⁹⁸) were phosphorylated with purified recombinant CaMKIIγ_C in the presence of [γ-³²P]ATP. Reactions were subjected to SDS-PAGE analysis. Representative autoradiograph showing phosphorylation state at 15 min (n = 2). Autophosphorylated CaMKIIγ_C resolves above the linker proteins. b, Analysis of phosphorylation stoichiometry of full-length linker protein (47 kDa) at 1 and 15 min of incubation (mean ± SE). The apparent phosphorylation stoichiometry of the RAES¹¹⁹⁸ or RAPS¹¹⁵³ motifs is indicated by the difference in phosphorylation state between the single-mutated (hatched or stripped bars) and double-mutated (open bars) fusion proteins. Phosphate incorporation into mutant GST-linker fusion proteins were compared with wild-type using ANOVA with post hoc Dunnett's testing, in which significance was p < 0.05.

Finally, to test the functionality of the RAES¹¹⁹⁸ motif, we mutated Ser¹¹⁹⁸ in the full-length channel protein and assayed for CaMKII-dependent regulation of channel activity. Mutation of Ser¹¹⁹⁸ to Ala abrogates CaMKII-dependent modulation of α_1H channels (Fig. 3c). Neither the half-activation potential nor the voltage sensitivity differs from non-modulated wild-type α_1H channels [V_1/2 = -21.8 ± 0.4 mV, k = 11.5 ± 0.5 mV; n = 17 (1 μm Ca²⁺); NS]. In contrast, mutant α_1H channels harboring a Ser¹¹⁵³ to Ala mutation retain CaMKII-dependent modulation (M: V_1/2 = -35.4 ± 1.9 mV; k = 5.3 ± 2.0 mV/e-fold change; 34 ± 11%; p < 0.05).

Modification of protein function by phosphorylation can be the result of the accommodation of the bulk of the phosphate moiety (Busch et al., 2002) or the introduction of negative charge (Buchbinder et al., 1997). We mutated Ser¹¹⁹⁸ to glutamine (Gln) to test first for the possible importance of the loss of serine hydroxyl interactions during phosphate incorporation. Gln¹¹⁹⁸ failed to change either the half-activation potential or the voltage sensitivity of this mutant α_1H channel [V_1/2 = -23.5 ± 1.2 mV; k = 9.4 ± 0.3 mV; n = 10 (27 nm)]. Moreover, as observed previously with mutation of the RAES¹¹⁹⁸ motif, the loss of Ser¹¹⁹⁸ prevents regulation by CaMKII [V_1/2 = -25.2 ± 0.7 mV; k = 9.4 ± 0.2 mV; n = 11 (1 μm)]. In a similar manner, mutation of Ser¹¹⁹⁸ to glutamic acid (Glu) creates a channel with gating properties that mimic those of the Gln¹¹⁹⁸ mutant. Both the V_1/2 of activation and the voltage sensitivity of this mutant channel also remain unaltered in the absence and presence of CaMKII activation [V_1/2 = -24.8 ± 0.8 mV, k = 9.8 ± 0.4 mV, n = 12 (27 nm); V_1/2 = -23.3 ± 0.5 mV, k = 9.7 ± 0.4 mV, n = 11., (1 μm)], suggesting that the monoanionic carboxylate group of Glu cannot replace the dianionic phosphate of phosphoserine in reproducing either the exact charge balance or the phosphopeptide positioning that is required to change channel gating (Buchbinder et al., 1997). Collectively, our mutagenesis studies provide support for our in vitro phosphorylation data and identify the key role played by Ser¹¹⁹⁸ in the observed changes in α_1H channel gating induced by CaMKII.

Discussion

The present data provide a molecular explanation for how CaMKII differentially regulates LVA Ca²⁺ channels. We described a remarkably simple way by which CaMKII changes α_1H activation gating. Here, we identify a single serine residue (Ser¹¹⁹⁸) within the II-III cytoplasmic linker that is unique to α_1H channels, phosphorylated by CaMKII, and critical for channel regulation. This mechanism for Ca²⁺ channel regulation differs from that reported for α_1C (Ca_v1.2) channels. Feedforward regulation (facilitation) of α_1C channels can be supported by a CaMKII-independent mechanism that requires CaM binding to an IQ motif (Zuhlke et al., 2000), a domain absent from the T-type Ca²⁺ channel family, and/or by a CaMKII-dependent mechanism that may require cytoskeletal intermediates (Yuan and Bers, 1994; Dzhura et al., 2000, 2002; Wu et al., 2001). However, the critical substrate(s) targeted by CaMKII remains undetermined. Our data support previous observations showing that the opening frequency of single α_1H channels in excised patches can be increased by membrane-associated CaMKII (Barrett et al., 2000). These studies showed that the participation of a cytosolic kinase cascade is not required for CaMKII-dependent regulation and implied the possibility of a direct modulation of α_1H Ca²⁺ channels by CaMKII. Our data provide the first evidence for direct regulation of the pore-forming subunit of a voltage-dependent Ca²⁺ channel by CaMKII.

The II-III cytoplasmic linker of the pore-forming channel subunit of other voltage-gated Ca²⁺ channels forms important structural associations with signaling effectors. In the Ca_v1 family, the II-III linker of α_1S couples the voltage sensor to the ryanodine-sensitive Ca²⁺ release channel (Tanabe et al., 1988, 1990), whereas in the Ca_v2 family, the II-III linker facilitates efficient delivery of Ca²⁺ to sites of neurotransmitter release by binding SNARE (soluble N-ethylmaleimide-sensitive factor attached protein receptor) proteins that concomitantly modify inactivation gating (Sheng et al., 1994; Catterall, 2000; Jarvis and Zamponi, 2001). Our data highlight a role for the II-III linker of T-type Ca²⁺ channels and show that, in the Ca_v3 family, the II-III linker is an important site for the control of activation gating. Recent studies from our laboratory indicate that the H_II-III linker also transduces current inhibition by directly binding G-protein β₂-containing βγ-dimers (Wolfe et al., 2003). Thus, in the Ca_v3 family, the H_II-III linker may serve as a center for integrating effects of multiple stimuli.

Although mRNA transcripts encoding each of the T-type Ca²⁺ channel subtypes are widely expressed throughout the CNS and PNS, the pattern of expression of each channel gene is unique, albeit overlapping (Talley et al., 1999). α_1H mRNA is expressed at high levels in restricted regions of the hippocampus (pyramidal cell layers and granule layer of the dentate gyrus), cerebral cortex (a subset of layer V neocortical cells), and thalamus (reticular cells), as well as in sensory ganglia, dorsal horn neurons (external lamina), and the olfactory bulb. In contrast to other members of the Ca_v3 Ca²⁺ channel family whose currents are blocked by Ni²⁺ with an IC₅₀ that overlaps that for block of HVA Ca²⁺ channels, α_1H channels are inhibited by a Ni²⁺ concentration that is 20- to 50-fold lower (IC₅₀ of 10 μm) (Perez-Reyes, 2003).

The Ni²⁺ sensitivity of low-threshold, slowly deactivating Ca²⁺ channel currents has revealed the specific importance of the α_1H channel isoform in the following: hippocampal-neocortical dendritic low-threshold Ca²⁺ spike initiation (Markram and Sakmann, 1994; Magee and Johnston, 1995b), regenerative potential conduction, and somatic burst firing (Gillessen and Alzheimer, 1997; Larkum et al., 2001), as well as in both synaptic (lamina I projection neurons of the dorsal horn) and nonsynaptic (CA1 pyramidal neurons) forms of activity-dependent potentiation (Su et al., 2002; Ikeda et al., 2003). Given the broad distribution of α_1H channels and CaMKII in nervous tissue, the operation of the direct phosphorylation-dependent regulatory mechanism described here may be widespread, supporting the enhanced participation of α_1H channels in these diverse neuronal settings.