Arachidonic acid inhibition of L-type calcium (Ca_V1.3b) channels varies with accessory Ca_Vβ subunits

In the nervous system, voltage-gated L-type (L-) Ca²⁺ channels are composed of several proteins: the pore-forming Ca_Vα₁ subunit, through which Ca²⁺ ions pass, and accessory Ca_Vβ and α₂δ subunits (Catterall, 2000). Neurons in the brain express two isoforms of the L-channel Ca_Vα₁ subunit: Ca_V1.2 and Ca_V1.3 (Hell et al., 1993). The Ca_V1.3 isoform plays a role in gene expression (Gao et al., 2006; Zhang et al., 2006), exocytosis (Brandt et al., 2005), and membrane excitability (Brandt et al., 2003; Olson et al., 2005), depending on the cell type and localization.

L-channel activity is inhibited by signal transduction pathways downstream of neurotransmitters, including certain types of dopamine (Wikstrom et al., 1999; Banihashemi and Albert, 2002; Olson et al., 2005), glutamate (Chavis et al., 1994), serotonin (Cardenas et al., 1997; Day et al., 2002), and acetylcholine receptors (Pemberton and Jones, 1997; Bannister et al., 2002; Liu et al., 2006). Activation of these G protein–coupled receptors (GPCRs) also releases arachidonic acid (AA; C20:4) (Axelrod et al., 1988; Lazarewicz et al., 1992; Yehuda et al., 1998; Tang et al., 2006). Our laboratory has documented that endogenous AA release is necessary for muscarinic M1 receptor (M1) inhibition of L-current in superior cervical ganglion (SCG) neurons (Liu et al., 2006). Moreover, exogenously applied AA inhibits L-current in SCG neurons similarly to M₁R agonists (Liu et al., 2006). The Ca_V1.3b L-channel isoform has been detected and cloned from SCG neurons (Lin et al., 1996), suggesting that endogenous AA modulates Ca_V1.3b.

The mechanism by which AA acts downstream of GPCR activation to inhibit L-current remains incompletely characterized. Single-channel recordings from SCG indicate that AA decreases the open probability of L-channels by increasing the dwell time in a closed state with no effect on unitary channel conductance (Liu and Rittenhouse, 2000). Similar findings of AA affecting closed states have been reported for the T-type (T-) Ca²⁺ channel, Ca_V3.1 (Talavera et al., 2004). A second family member, Ca_V3.2, is also inhibited by AA but via a leftward shift in holding potential–dependent inactivation (Zhang et al., 2000). Additionally, both T-channel studies reported increases in the rate of fast inactivation after AA, whereas our study on whole cell SCG L-current revealed no such changes (Liu et al., 2001). One obvious difference between T- and L-channels is that T-channels lack the recognition sequence in the I-II linker for binding Ca_Vβ subunits (Arias et al., 2005), whereas Ca_Vβ binding to L-channels fine-tunes their kinetics and voltage dependence of activation and inactivation (Singer et al., 1991; Hering et al., 2000; Kobrinsky et al., 2004). Whether certain Ca_Vβ subunits block kinetic changes elicited by AA or whether Ca_V1.3 lacks a homologous site that confers the kinetic changes is unknown.

Therefore, to examine the extent of AA's actions on L-channel activity, we tested whether coexpression of Ca_V1.3b with different Ca_Vβ subunits accounts for the lack of kinetic changes observed by AA inhibition of whole cell L-current in SCG neurons. We show that AA inhibits Ca_V1.3b currents expressed in human embryonic kidney (HEK) 293 cells by stabilizing channels in a closed state. Inhibition occurs regardless of which Ca_Vβ subunit is coexpressed; however, the magnitude of inhibition produced and whether kinetic changes occur after AA depend on the Ca_Vβ subunit. Physiologically, the tissue-specific expression of different Ca_Vβ subunits is predicted to affect the response of L-channels after the stimulation of certain GPCRs or exposure to high levels of free AA, such as during ischemia.

The HEK cell line, stably transfected with the M1 muscarinic receptor (HEK M1; provided by E. Liman, University of Southern California, Los Angeles, CA, and originally transfected by Peralta et al., 1988) was propagated at 37°C with 5% CO₂ in Dulbecco's MEM (DMEM)/F12 supplemented with 10% FBS, 1% G418, 0.1% gentamicin, and 1% HT supplement (Invitrogen). Cells were passaged once they became 80% confluent.

HEK M1 cells were placed in 12-well plates at ∼60–80% confluence and transfected with a 1:2:1 molar ratio of Ca_V1.3b, Ca_Vβ, and α₂δ-1 subunits (Xu and Lipscombe, 2001) using Lipofectamine PLUS (Invitrogen) according to the manufacturer's instructions. Constructs for Ca_V1.3b (+exon11, Δexon32, and +exon42a; GenBank accession no. AF370009), Ca_Vβ₃ (no. M88751), and α₂δ-1 (no. AF286488) were provided by D. Lipscombe (Brown University, Providence, RI). Constructs for Ca_Vβ_1b (GenBank accession no. X61394), Ca_Vβ_2a (no. M80545), and Ca_Vβ₄ (no. L02315) were provided by E. Perez-Reyes (University of Virginia, Charlottesville, VA). Ca_Vβ_2a C3,4S was constructed by M. Hosey's lab (Chien et al., 1996) and supplied by A. Fox (University of Chicago, Chicago, IL). The Ca_Vβ_2e (AF423192) construct was provided by H.M. Colecraft (Columbia University, New York, NY), and the Ca_Vβ_2aC3,4S-CD8 construct was provided by P. Charnet (Centre National de la Recherche Scientifique, Paris, France). For all transfections, 0.5 µg DNA was used per well of a 12-well plate. Green fluorescent protein cDNA was <10% of the total cDNA used. Cells were washed with DMEM, and the DNA mixture was added dropwise to each well, gently rocked, and then incubated for 1 h at 37°C in a 5% CO₂ incubator. Supplemented media, without antibiotics, was returned to the cells to bring the volume up to 1 ml (normal growth medium volume). After 2 h, cells were washed with full media. Cells were washed a final time 2 h later, and 10 mM MgSO₄ was added. Cells were transferred 24–72 h after transfection using 2 mM EDTA in 1× PBS to poly-l-lysine–coated coverslips; recording began 1 h after transfer.

Electrodes were pulled from borosilicate glass capillary tubes (Drummond Scientific Company). Each electrode tip was fire polished to a diameter of ∼1 µm to give the pipette a resistance of 2–4 MΩ. Whole cell currents were recorded at room temperature (20–24°C) with an Axon 200B patch clamp amplifier (MDS Analytical Technologies). Currents were filtered at 5 or 20 kHz and digitized at five times the filter cut-off frequency of the four-pole Bessel filter. Data were acquired using Signal software (Cambridge Electronic Design) and stored for later analysis on a personal computer. For time course experiments, a holding potential of −60 or −90 mV was used, and currents were measured at a test potential of −10 mV for 40 ms at 5-s intervals. For holding potential–dependent inactivation experiments, a holding potential of −60 mV was used and the “conditioning” potential varied from −100 to +50 mV for 2.2 s, immediately followed by a test potential of −10 mV for 40 ms.

The pipette solution consisted of (in mM): 125 Cs-aspartate, 10 HEPES, 0.1 BAPTA, 5 MgCl₂, 4 ATP, and 0.4 GTP brought to pH 7.50 with CsOH. High resistance seals were established in Mg²⁺ Tyrode's consisting of (in mM): 5 MgCl₂, 145 NaCl, 5.4 KCl, and 10 HEPES brought to pH 7.50 with NaOH. Once a seal was established and the membrane ruptured, the Tyrode's solution was exchanged for an external bath solution consisting of (in mM): 125 NMG-aspartate, 20 Ba²⁺, and 10 HEPES brought to pH 7.50 with CsOH. AA (all-cis 5,8,11,14-eicosatetraynoic acid; NuCheck Prep), oleic acid (NuCheck Prep), and ETYA (BioMol) were dissolved in 100% EtOH and stored under nitrogen as stock solutions at −70°C. Working dilutions were made fresh daily by diluting stock solutions at least 1:1,000 in external bath solution. A final concentration of 10 µM AA was used unless otherwise noted. Palmitic acid (PA) was prepared daily as a 50-mM stock solution in EtOH and diluted to a final concentration of 10 µM in external bath solution. BSA (fraction V, heat shock, fatty acid ultra-free; Roche) was diluted 1 mg/ml directly in bath solution. All chemicals were purchased from Sigma-Aldrich unless otherwise noted.

Linear leak and capacitive currents were subtracted and maximal inward current amplitudes were measured after the onset of the test pulse (peak current). Percent inhibition was calculated as

where I_CTL is the average of five peak inward current values before the application of test material, and I_INHIB is the average of five peak inward current values 1 min after the application of test material (unless otherwise noted). Difference current was measured as

Time to peak (TTP) was measured using a trough-seeking function in Signal within the test pulse duration. Current remaining was measured from an average of five individual sweeps per recording using the equation:

where I_R is the current remaining at the end of a 40-ms (r40) or 2.2-s (r2200) test pulse, I_END is the value at the end of the test pulse, and I_PEAK is the maximum inward current measured during the test pulse. Conductance was calculated from a modified Ohm's Law equation:

where I_PEAK is the peak current at each test potential, V_m is the test potential, and V_rev is the reversal potential. Relative conductance (G/G_max)-voltage (V_m) curves were plotted and fit using the Boltzmann equation:

where G_max is the maximal conductance, G_min is the minimal conductance, V_m is the test potential, V_m1/2 is the voltage at half-maximal conductance, and k is the slope factor. Holding potential–dependent inactivation was plotted as normalized current during the test pulse against the conditioning pulse voltage. The inactivation curve was fit using the Boltzmann equation:

where V_m is the conditioning pulse and V_inact1/2 is the voltage where half the maximal number of channels has inactivated. The rate of inactivation was measured by fitting the current traces to a biexponential function yielding time constants for a fast (τ_fast) and slow (τ_slow) component:

where A_fast and A_slow are the current amplitudes with the respective time constants, and A₀ is any remaining current. The time courses for recoveries from fast and slow inactivation were fit to the same biexponential function yielding time constants for τ_fast and τ_slow components.

Data are presented as mean values ± SEM. Data were analyzed statistically using either a one-way ANOVA followed by a Tukey multiple comparison post-hoc test or a two-tailed paired or unpaired Student's t test. Statistical significance was set at P ≤ 0.05. Analysis programs included Signal (Cambridge Electronic Design), Excel (Microsoft), and Origin (OriginLab).

Fig. S1 shows that 10 μM oleic acid had no significant effect on β2a- and β3-containing channels. Fig. S2 shows that no significant correlations among control current amplitude, control r40, and percent inhibition by 10 μM AA were found for Ca_V1.3 currents coexpressed with any of the β-subunits. Correlations were calculated for recordings using a holding potential of −90 or −60 mV. Fig. S3 shows that rate of activation is faster in the presence of AA when fitting current traces to the equation: ⁠. Ca_V1.3 currents, regardless of the Ca_Vβ subunit coexpressed, were best fitted to one exponential. Figs. S1–S3 are available.

To determine if AA inhibits Ca_V1.3b activity, peak current amplitudes were measured from −90 mV to a test pulse of −10 mV every 5 s before and after the application of 10 µM AA to cells transiently transfected with Ca_V1.3b, β₃, α₂δ-1, and green fluorescent protein. AA produced a 51 ± 8% inhibition of current after 1 min (n = 7; P < 0.001). We then tested whether inhibition occurred in a concentration-dependent manner and found that the magnitude of inhibition increased with increasing AA concentration (Fig. 1, A and B). To stay well below the critical micelle concentration, concentrations >10 µM AA were not tested. In measuring current inhibition, we noticed an apparent acceleration in activation kinetics that was most pronounced with 10 µM AA (Fig. 1 B) and was apparent when current traces were normalized to the end of the 40-ms test pulse (Fig. 1 C). The increased rate of activation (Fig. 1 D, top), measured as TTP, maximally decreased after 1 min, whereas current inhibition (Fig. 1 D, bottom) developed over the span of 3 min. The application of 1 mg/ml BSA, which binds free fatty acid, reversed both the kinetic changes and current inhibition, indicating that channels were specifically modulated as opposed to nonspecific rundown of channel activity over time.

The change in TTP was unanticipated because native L-current in SCG neurons exhibits no apparent kinetic change when inhibited by AA (Liu et al., 2001). Additionally, whole cell L-current in SCG neurons is non-inactivating as opposed to the relatively rapid inactivating current observed with unmodulated transfected Ca_V1.3b, β₃ currents (Fig. 1 B, control [CTL]). Non-inactivating current kinetics are characteristic of channels associated with β_2a, which is uniquely palmitoylated, tethering it to the membrane and conferring different inactivation properties (Chien et al., 1996). Therefore, we tested whether the AA-induced TTP change also occurs with β_2a. Fig. 1 E (bottom) shows that 10 µM AA significantly inhibited current from cells transfected with Ca_V1.3b, α₂δ, and β_2a by 31 ± 7% after 1 min (n = 8; P < 0.001). However, no change in TTP was observed (Fig. 1 E, top). After 3 min, current was inhibited by 56 ± 15% (n = 5; P < 0.001), suggesting that currents from β_2a-containing channels can be inhibited to a similar extent as β₃-containing channels after 1 min. The application of BSA recovered inhibited current. These differences suggest that different Ca_Vβ subunits uniquely tune Ca_V1.3b modulation by AA.

Four genes in the Ca_Vβ family (β₁, β₂, β₃, and β₄) and several splice variants have been identified to date (Birnbaumer et al., 1998; Takahashi et al., 2004). To test the effects of other Ca_Vβ family members on the modulation of Ca_V1.3b by AA, we compared representative current traces for Ca_V1.3b currents coexpressed with β_1b, β_2a, β₃, or β₄ in the absence (CTL) or presence of 10 µM AA (Fig. 2 A). After a 1-min exposure to AA, the magnitude of inhibition was 58 ± 6% for β_1b (n = 7; P < 0.001) and 44 ± 4% for β₄ (n = 6; P < 0.001) (Fig. 3 A). Thus, AA inhibits Ca_V1.3b regardless of the Ca_Vβ subunit coexpressed, but the magnitude of inhibition is less for β_2a and is not accompanied by kinetic changes. The finding that β_2a-containing channels exhibit a profile of modulation identical to our studies of native L-current (Liu et al., 2001) suggests that Ca_V1.3 in SCG neurons primarily couples to β_2a.

To visualize differences in kinetics, individual traces were normalized to the end of the 40-ms test pulse (Fig. 2 B). When the inhibited current was subtracted from CTL current, the difference current revealed a relatively non-inactivating shape for currents with each of the Ca_Vβ subunits (Fig. 2 C). The loss of non-inactivating current after AA is similar to whole cell difference currents recorded in SCG neurons and is consistent with AA-stabilizing channels in closed or inactivated states, thus removing inhibited channels from contributing to the whole cell current. This result suggests that the current actually inhibited by AA inactivates very little, whereas the residual current displays more inactivation for β_1b-, β₃-, or β₄-containing channels.

Single-channel analysis of L-current inhibition by AA revealed an increase in null sweeps and a decrease in open probability (Liu and Rittenhouse, 2000), suggesting that AA increases the number of channels dwelling in a non-conducting conformation: either closed or inactivated. Additionally, first latency, the time for a single channel to transition from closed to open in response to a test pulse, increases from ∼30 to 90 ms (Liu and Rittenhouse, 2000), indicating that when an L-channel is affected by AA, the time to transition from closed to open increases. These results support a model where AA stabilizes a closed conformation rather than an inactivated conformation because inactivated channels will not respond to voltage until sufficient time has transpired for channels to recover.

Many kinetic models display closed Ca²⁺ channels in at least four closed conformations representing the movement of each voltage sensor from domains I–IV of the Ca_Vα₁ subunit in response to changes in membrane potential to the open conformation (Patil et al., 1998; Serrano et al., 1999). For example, at a holding potential of −90 mV, none of the voltage sensors have moved and channels reside primarily in a deep closed state (Marks and Jones, 1992); at a holding potential of −60 mV, the voltage sensors have moved in response to depolarization and channels reside in closed conformations closer to the open state. Therefore, we tested whether holding cells at −60 mV would alter the magnitude of Ca_V1.3b inhibition by AA compared with holding cells at −90 mV. AA suppressed Ca_V1.3b currents with β_1b, β₃, or β₄ by 34 ± 6%, 26 ± 7%, or 26 ± 4%, respectively (Fig. 3 B), but with β_2a, only by 12 ± 6% (n = 7–13; P < 0.001 for all Ca_Vβ subunits compared with CTL; P < 0.05 for all Ca_Vβ subunits comparing current inhibition from −90 to −60 mV). From −60 mV, TTP continued to be unaffected by AA application for channels with β_2a and decreased for channels with β₃ (Fig. 3 C). Overall, current inhibition by AA was greater for all Ca_Vβ subunits recorded with a holding potential of −90 mV than with −60 mV, consistent with AA stabilizing a deep closed state (Liu and Rittenhouse, 2000).

Next, we examined whether channels need to enter the open conformation for AA inhibition of Ca_V1.3b currents. Here, we used a protocol with short 10-ms test pulses and no depolarization for 1 min in the absence or presence of AA (Fig. 3 D). The protocol was designed to maintain channels in a closed conformation; the shorter test pulse should minimize the amount of inactivation. Additionally, no depolarization for 1 min should bring channels into a resting, closed conformation. The 10-ms test pulse from −60 mV had no effect on the ability of AA to inhibit currents from β₃-containing channels. In fact, the shorter 10-ms test pulse showed more inhibition after 1 min (35%; Fig. 3 E) than recordings with 40-ms test pulses (26%; Fig. 3, B and E). In the absence of test pulses, AA inhibited 34% of the current (n = 7; P < 0.001 compared with CTL current before the application of AA). Moreover, BSA reversed inhibition to CTL peak amplitudes (not depicted). Overall, these experiments suggest that Ca_V1.3b does not need to open to become inhibited by AA.

To address the specificity of AA inhibition and the possible role of AA metabolites, a nonhydrolyzable form of AA, ETYA, was tested for inhibition of Ca_V1.3b currents. From a holding potential of −60 mV, current was measured 1 min after the application of 30 µM ETYA. Fig. 4 A shows representative individual sweeps for each of the Ca_Vβ subunits before or after ETYA. To visualize differences in kinetics, individual traces were normalized to the end of the 40-ms test pulse (Fig. 4 B). ETYA inhibited currents for β_1b-, β_2a-, β₃-, or β₄-containing channels 32 ± 8%, 14 ± 4%, 24 ± 5%, or 22 ± 6%, respectively (n = 5–8; P < 0.001 compared with CTL) (Fig. 4 C). ETYA inhibition did not significantly differ from AA inhibition from a holding potential of −60 mV shown in Fig. 3 B (β_1b, P ≥ 0.89; β_2a, P ≥ 0.81; β₃, P ≥ 0.78; β₄, P ≥ 0.59). To test whether the effects of ETYA and AA were additive, 30 µM ETYA was added for 1 min, 30 µM ETYA + 10 µM AA was added for another minute, and current inhibition was measured. Current was further inhibited by ETYA + AA for a total inhibition of 49 ± 7% (n = 4; P < 0.001 compared with CTL; P < 0.05 compared with ETYA alone) for Ca_V1.3b currents with β₃. This magnitude of inhibition is similar to that measured by 10 µM AA alone after application for 2 min (not depicted). These results suggest that inhibition is the result of AA itself and not a metabolite. In contrast to polyunsaturated AA or ETYA, 10 µM oleic acid (C18:1), a monounsaturated fatty acid, had no significant effect on currents from channels containing β_2a (n = 4; P ≥ 0.66) or β₃ (n = 3; P ≥ 0.58) (Fig. S1), indicating that inhibition requires certain aspects of fatty acid structure.

To thoroughly analyze any changes in the overall closed to open transitions, we examined the effect of AA on peak current over a range of test potentials (−100 mV to +50 mV in 10-mV steps). In I-V plots, all currents, regardless of the Ca_Vβ subunit expressed with Ca_V1.3b, activated at −50 mV, peaked at −10 mV, and reversed at +50 mV, similar to previous reports of Ca_V1.3 activating at low voltages (Koschak et al., 2001; Xu and Lipscombe, 2001). AA inhibited currents at all voltages regardless of the Ca_Vβ subunit (Fig. 5 A). G-V plots revealed no shift in the activation curve by AA compared with CTL plots (Fig. 5 B), consistent with inhibited channels becoming unavailable to open (stabilized closed) and thus not contributing to the G-V relationship. Therefore, AA inhibition is not the result of altering the voltage sensitivity of Ca_V1.3b.

Because these initial measurements are consistent with a model where AA stabilizes a deep closed state, we next sought to understand the observed changes in activation and inactivation described in Figs. 1 and 2. We first quantified current remaining at the end of the 40-ms test pulse (r40), respectively, for Ca_V1.3b with each Ca_Vβ subunit. For currents elicited from a holding potential of −90 to −10 mV, r40 (Fig. 6 A) significantly decreased after a 1-min exposure to AA for currents containing β_1b, β₃, or β₄, but not β_2a (n = 6–8; P < 0.05 for β_1b, β₃, and β₄; P ≥ 0.48 for β_2a). Although less inhibition by AA was observed at −10 mV from a holding potential of −60 mV, the decrease in r40 was still apparent for channels with β_1b, β₃, and β₄; no kinetic changes were observed for channels with β_2a (n = 7–13; P < 0.05 for β_1b, β₃, and β₄; P ≥ 0.48 for β_2a) (Fig. 6 B). BSA reversed both the kinetic change and current inhibition regardless of holding potential (not depicted). Additionally, we found no correlations between CTL current amplitude, CTL r40, or percentage of AA-induced inhibition of Ca_V1.3b with any Ca_Vβ subunit (Fig. S2).

TTP was also measured across a range of test potentials from a holding potential of −60 mV. After AA application, TTP was generally faster for β_1b-, β₃-, or β₄-containing channels across voltages (Fig. 6 C). In contrast, TTP for Ca_V1.3b with β_2a, which is slower than with the other Ca_Vβ subunits, was only faster in the presence of AA at −20 mV. To confirm these data, activation time constants were measured at multiple test potentials before and after 1 min of AA (Fig. S3). The activation of Ca_V1.3b was best fitted with one exponential (τ_ACT) for all Ca_Vβ subunits, similar to previous monoexponential fits of Ca_V1.3 with β₃ (Koschak et al., 2001). The τ_ACT for Ca_V1.3b, β_2a currents was approximately threefold slower than that for the other Ca_Vβ subunits. Overall, AA decreased the mean τ_ACT across voltages for Ca_V1.3b currents with all Ca_Vβ subunits, consistent with the decrease in TTP after AA. TTP kinetic changes were also apparent after AA with the shorter 10-ms test pulses (not depicted). These findings suggest that although the kinetic changes observed with β_1b-, β₃-, or β₄-containing channels occur with inhibition over the range of voltages tested, the TTP change produced by AA is not the result of a shift in voltage dependence of activation.

Kinetic changes produced by AA were also mimicked by ETYA. When comparing CTL currents to currents exposed to ETYA, TTP significantly decreased for all Ca_Vβ subunits (n = 5–8; P < 0.05) (Fig. 6 D). Significant decreases in r40 were also observed for Ca_V1.3b with β_1b, β₃, and β₄, but not with β_2a (Fig. 6 E). For Ca_V1.3 with β₃, TTP and r40 significantly decreased for ETYA + AA compared with CTL, but it did not significantly differ from ETYA alone. Thus, in addition to inhibition, the kinetic changes observed after AA application are not produced by an AA metabolite.

At least two explanations may account for the kinetic changes and the decrease in current amplitude after AA. In the first scenario, after exposure to AA, all of the channels shift to more reluctant gating where a greater voltage step is required to open channels. If this is the case, the activation curve will exhibit a rightward shift to more positive voltages after AA. Moreover, if AA shifts all channels toward reluctant gating, we would expect to observe a slower TTP. This scenario seems unlikely because no shift in the G-V curve occurs (Fig. 5 B), and TTP is faster at −10 mV in the presence of AA compared with CTL currents, despite our previous findings that AA increases dwell times in closed conformations and increases first latency to 90 ms (Liu and Rittenhouse, 2000). Because the test pulse duration for the whole cell current recordings was 40 msec, the inhibited channels are unlikely to open.

In a second scenario, at least two populations of L-channels are present under CTL conditions: one population of channels is reluctant to open (reluctant closed channel conformation [RC]), whereas another population is willing to open (willing closed channel conformation [WC]), with the RC transition time to the open channel conformation (O) slower than the WC→O transition time (Jones and Elmslie, 1997). This description of reluctant gating has been described previously for Ca_V2 channels modulated by G proteins (Bean, 1989) or by phosphatidylinositol-4,5-bisphosphate (PIP₂) binding (Wu et al., 2002). CTL TTP can be expressed as the sum of RC→O plus WC→O transition times. After AA, a greater percentage of channels reside in the RC conformation longer and, therefore, will not open during the 40-ms test pulse due to the observed first latency increase to 90 ms (Liu and Rittenhouse, 2000). Consequently, the residual current displays only the faster WC→O transition. This model is consistent with the decreased TTP after AA. Similarly, RC channels show slower inactivation. Therefore, if RC channels are inhibited and WC channels remain active, an apparent increase in inactivation would occur. This model fits the observed kinetic changes in TTP and r40 after AA.

This simple model predicts that AA would only need to stabilize one or more deep closed states to confer the observed changes in channel gating; no change in fast or slow inactivation need occur. The above data rule out changes in fast inactivation. Therefore, we tested whether AA changes slower forms of inactivation. To test AA's effect on holding potential–dependent inactivation, a protocol was used in which channels were exposed to a series of “conditioning pulses” ranging from −100 to +50 mV (in 10-mV steps) for 2.2 s, and then repolarized to the test potential of −10 mV for 40 ms. Representative current traces for Ca_V1.3 with β_2a (Fig. 7, A and C) or β₃ (Fig. 7, B and D) are shown before (CTL) or after exposure to AA for 1 min. The long, 2.2-s conditioning pulse highlights the striking differences in inactivation kinetics between β_2a- and β₃-containing channels. Inactivation currents for Ca_V1.3 with β_1b, β₃, or β₄ were best fit with two exponentials (τ_INACT-fast and τ_INACT-slow), whereas Ca_V1.3 β_2a currents were best fit with one exponential (τ_INACT-slow). Fits of inactivation are shown overlying the representative traces (Fig. 7, C and D). Fig. 8 (A and B) summarizes the inactivation time constants for voltages from −20 to +40 mV. AA significantly decreased τ_INACT-fast at positive test potentials (0 to +40 mV) for β_1b-, β₃-, or β₄-containing currents. As a general trend, AA did not effect τ_INACT-slow for β_1b-, β₃-, or β₄-containing currents (Fig. 8 B), although AA decreased τ_INACT-slow for currents with β_1b at +20 mV, β_2a at 0 and +20 mV, β₃ at +20 to +40 mV, and β₄ at 0 mV (n = 5–7; P < 0.05). Overall, Ca_V1.3 currents with β_1b, β₃, or β₄ appear to fast inactivate more rapidly in the presence of AA, consistent with the r40 data.

To measure the voltage sensitivity of holding potential–dependent inactivation, currents were normalized to peak inward current during the test pulse and plotted against the conditioning test potential (Fig. 8 C). The CTL inactivation curve for Ca_V1.3 currents with β_2a was shifted ∼10 mV more positive than the other Ca_Vβ subunits. From fits to the Boltzmann equation (Fig. 8 C), we found that AA had no effect on the V_inact1/2 of the Ca_V1.3 currents coexpressed with β_1b (CTL, −41.1 ± 1.2 mV; AA, −44.1 ± 0.8 mV), β_2a (CTL, −29.8 ± 0.7 mV; AA, −30.4 ± 0.8 mV), β₃ (CTL, −38.5 ± 0.8 mV; AA, −41.9 ± 1.1 mV), or β₄ (CTL, −38.5 ± 0.8 mV; AA, −40.8 ± 0.4 mV). Similar values for CTL Ca_V1.3 β₃ V_inact1/2 have been reported (Scholze et al., 2001). These results show that AA has no effect on the voltage dependence of holding potential–dependent inactivation for Ca_V1.3 currents with any Ca_Vβ subunit.

Closer inspection of the Ca_V1.3 inactivation curves for β_1b-, β₃-, or β₄-containing channels revealed that inactivation occurred at negative test potentials before channels began to open, suggesting that channels were inactivating from a closed state similar to T- (Frazier et al., 2001) and N-type Ca²⁺ channels (Patil et al., 1998). For example, at the conditioning pulse of −60 mV, ∼10–15% of the channels were inactivated even though the channels do not open at this voltage. At the conditioning pulse of −90 mV, only 5% of the channels were inactivated. The slope between −100 and −60 mV, before channels begin to activate, does not appear to be an artifact or an accumulation of inactivation because testing longer intervals between the final test pulse and the following conditioning pulse had no effect on the voltage dependence of the inactivation curves with and without AA (not depicted). Regardless, this observed inactivation profile does not change in the presence of AA.

Although the voltage sensitivity did not shift after AA, significantly fewer β_2a-containing channels were available to open per voltage (−10 to +50 mV) in the presence of AA (Fig. 8 C). This finding correlates with the amount of current remaining at the end the 2.2-s conditioning pulse (r2200). Here, AA had no effect on the r2200 for β_1b-, β₃-, or β₄-containing channels because >90% of CTL and inhibited current had already undergone fast inactivation. However, AA decreased the r2200 for β_2a-containing channels (Fig. 8 A, inset in box). Therefore, over a longer time scale, currents with β_2a subunits also displayed more inactivation in the presence of AA. By decreasing the number of channels available to open during a given test pulse, fast inactivation of β_1b-, β₃-, or β₄-containing channels was accentuated, whereas holding potential–dependent inactivation of β_2a-containing channels was accentuated by AA. Thus, AA may act via a common mechanism of stabilizing deep closed states independently of which Ca_Vβ subunit is coexpressed with Ca_V1.3b.

Because an apparent increase in channels residing in closed conformations could be due to an increase in recovery time from the inactivated state, we next examined recovery from inactivation using two separate protocols. First, peak current was measured every 5 s after the inactivation protocol in the absence and presence of AA (Fig. 9 A). Current recoveries were fit to a single exponential time constant and did not significantly differ between CTL and AA for Ca_V1.3 with any Ca_Vβ subunit (Fig. 9 B). Second, in another test for changes in recovery from inactivation, the test pulse was placed at various time points (Δt) after a 1-s conditioning pulse (Fig. 9 C). In this protocol, we chose a 1-s conditioning pulse to +20 mV based on where maximal holding potential inactivation occurred (Fig. 8 C, β₃). Recovery from inactivation was best fitted with two time constants: τ₁ = 291 ms and τ₂ = 13.6 s. Recovery from inactivation in the presence of AA also was best fitted with two time constants: τ₁ = 371 ms and τ₂ = 17.2 s (Fig. 9 D). This experiment allowed for the measurement of recovery from fast inactivation, which was overlooked in fitting the recovery from the time courses during which currents were evoked every 5 s (Fig. 9, A and B). Consistent with the first protocol, the time constants were not significantly different between CTL and AA (Fig. 9 D). However, the fractional amplitudes A₁ and A₂ associated with the recovery time constants revealed an important difference. The CTL values for A₁ and A₂ were 0.52 and 0.34, respectively, whereas the values for AA were 0.67 and 0.18, respectively. The finding suggests that in the presence of AA, there are more channels recovering from fast inactivation and fewer channels recovering from slow inactivation. Therefore, AA does not change the rate constant between inactivated and closed conformations, but rather alters the number of channels recovering from fast and holding potential–dependent inactivation.

β_2a uniquely stands apart from other Ca_Vβ subunits because it contains two palmitoyl groups at cysteine residues in the N terminus (Chien et al., 1996), conferring unique non-inactivating properties to Ca_V current (Qin et al., 1998). To test whether the non-inactivating properties of β_2a-containing channels are responsible for less AA inhibition, we tested another β₂ splice variant, β_2e, which is not palmitoylated, yet still has non-inactivating properties. After exposure to AA for 1 min, β_2e channels were inhibited 24 ± 8% (n = 6; P ≥ 0.26 compared with β_2a) (Fig. 10 A). Even though percent inhibition by AA approximately doubled for β_2e compared with β_2a, the difference was not significant. Therefore, we tested a β_2a subunit in which amino acid residues 3 and 4 have been mutated from cysteines to serines (β_2a C3,4S) resulting in a β_2a subunit that cannot be palmitoylated (Restituito et al., 2000). After exposure to AA for 1 min, β_2a C3,4S channels were inhibited 32 ± 5% (n = 6; P < 0.05 compared with β_2a) (Fig. 10 A). To examine whether the loss of the palmitoyl groups or the loss of non-inactivating properties resulted in more inhibition by AA, we tested a β_2a C3,4S subunit with a CD8 sequence attached. CD8 is a transmembrane protein, and thus attaching CD8 to the N terminus of β_2a C3,4S restores the non-inactivating properties (Restituito et al., 2000). After exposure to AA for 1 min, β_2a C3,4S-CD8 channels were inhibited 44 ± 9% (n = 9; P < 0.05 compared with β_2a) (Fig.10 A). Fig. 10 B shows the r40 for currents with β_2e, β_2a C3,4S, and β_2a C3,4S-CD8 before and after AA application. Representative sweeps for each condition are displayed in Fig. 10 C. β_2e and β_2a C3,4S-CD8 currents both display non-inactivating properties yet are inhibited more by AA than β_2a channels, suggesting that the palmitoyl groups of β_2a, and not the non-inactivating properties, are responsible for decreased AA inhibition of Ca_V1.3b.

A compelling idea is that palmitoyl groups of β_2a may compete for a “fatty acid” or “AA” binding site on Ca_V1.3. Therefore, we tested if excess PA would block AA inhibition. A 1-min exposure to 10 µM PA inhibited β_2a C3,4S currents by 8 ± 4% (n = 3; P < 0.05 compared with 1 min exposure of β_2a C3,4S currents to 10 µM AA). Cells were then preincubated in 10 µM PA in Tyrode's solution before establishing a seal. After breakthrough, bath solution containing 10 µM PA was washed in and a baseline current was measured before adding 10 µM PA + 10 µM AA. After a 1-min exposure to 10 µM AA, currents were inhibited 11 ± 7% (n = 3; P < 0.05 compared with 1 min exposure of β_2a C3,4S currents to 10 µM AA). Fig. 11 A is a summary bar graph of mean inhibition by AA. Fig. 11 B shows representative sweeps for each condition. A summary bar graph for current remaining for each condition is shown in Fig. 11 C. Overall, these results suggest that the palmitoyl groups of β_2a or excess PA may compete with AA for an inhibitory site on Ca_V1.3b.

Here, we show that AA inhibits Ca_V1.3b, a short form of Ca_V1.3 with a truncated C terminus, ruling out its importance in modulation by AA. Because we had previously characterized AA's effects on native L-current from SCG neurons (Liu and Rittenhouse, 2000; Liu et al., 2001, 2006), we used Ca_V1.3b, which was cloned from SCG neurons (Xu and Lipscombe, 2001), in these studies. Using ETYA, a nonmetabolizable form of a fatty acid similar to AA, we show that inhibition and kinetic changes of Ca_V1.3 are due to AA itself, and not a metabolite (Figs. 4 and 6, D and E). Current inhibition occurs regardless of the Ca_Vβ subunit coexpressed. However, the magnitude of inhibition after 1 min by either AA or ETYA was less with coexpression of β_2a than β_1b, β₃, or β₄. In addition, AA decreased r40, a measure of fast inactivation, and TTP of channels containing β_1b, β₃, or β₄ (Figs. 1, 2, and 6). In contrast, β_2a-containing channels showed no differences in TTP after exposure to AA. These complex changes are manifested in the presence of AA due to the influence of different Ca_Vβ subunits coexpressed with Ca_V1.3b on gating kinetics. Because the kinetic changes do not correlate with percent inhibition by AA over time (Figs. 1 D and 3 C), AA may act at multiple sites on L-channels.

AA inhibits native L-current in a variety of systems: cardiac muscle (Petit-Jacques and Hartzell, 1996; Xiao et al., 1997; Liu, 2007), skeletal muscle T-tubule membranes (Oz et al., 2005), smooth muscle (Shimada and Somlyo, 1992), SCG (Liu and Rittenhouse, 2000; Liu et al., 2001), photoreceptors (Vellani et al., 2000), and retinal glial cells (Bringmann et al., 2001). No consensus of mechanism exists as to how AA inhibits L-channel activity, whether directly, through AA metabolites, or via AA activating a phosphatase. Varying effects of AA on L-current may be due to expression of different L-Ca_Vα₁ subunits. For example, ventricular myocytes and smooth muscle cells primarily express Ca_V1.2, whereas SCG neurons, atrial myocytes, and cochlear hair cells express Ca_V1.3 (Lin et al., 1996; Platzer et al., 2000; Mangoni et al., 2003; Catterall et al., 2005). Additionally, Ca_V1.2 and Ca_V1.3 undergo alternative splicing (Hell et al., 1993; Safa et al., 2001), which may contribute to the differences in L-current kinetics observed with varying cell types. Notably, the lack of AA-induced kinetic changes in cardiac myocytes and SCG neurons may be due to high levels of β_2a subunit mRNA found in these cells (Lin et al., 1996; Birnbaumer et al., 1998).

In this study of whole cell recombinant Ca_V1.3b L-current, we found changes induced by AA complementary to our previous single-channel data. Analysis of unitary L-channel gating in SCG neurons revealed that AA-induced inhibition had no effect on mean open time or on unitary current amplitude (Liu and Rittenhouse, 2000). Consistent with this finding, the results from three experiments indicate that the open conformation of Ca_V1.3b is not necessary for inhibition by AA. First, test pulse length did not affect the magnitude of current inhibition by AA (Fig. 3 E). In fact, reducing the test pulse length increased the amount of AA-induced current inhibition. Second, β_2a-containing currents, which do not exhibit fast inactivation and therefore gate open and closed longer, exhibited less inhibition (Figs. 1–5). Third, Ca_V1.3b channels exhibited robust inhibition when held closed for 1 min during AA application and then allowed to open (Fig. 3 E). Thus, AA does not confer inhibition by binding to the open state.

Single-channel analysis of L-current in SCG neurons also revealed that AA decreased channel open probability by increasing the number of null sweeps. Moreover, when channels were active, AA decreased the number of openings and increased first latency and mean closed time (Liu and Rittenhouse, 2000). Together, the changes in the unitary activity indicate that AA stabilizes L-channels in one or more nonconducting states: either closed or inactivated conformations.

We hypothesize that AA decreases channel availability by increasing the number of L-channels stabilized in a deep closed state, as opposed to an inactivated state, based on several findings. First, we observe more AA inhibition from channels held at a membrane potential of −90 mV than at −60 mV (Fig. 3, A and B); at a membrane potential of −90 mV, channels undergo less closed state inactivation than at −60 mV (Fig. 8 C, β_1b, β₃, and β₄). Second, the voltage sensitivity of holding potential–dependent inactivation does not change with AA (Fig. 8 C). Third, when measuring both fast and slow recovery for β₃-containing channels, we observe no change in recovery from inactivation in the presence of AA (Fig. 9), suggesting that the transition rate from inactivated to closed conformations remains unchanged. Moreover, inactivated channels regardless of voltage will not respond to positive test pulses until given enough time at a negative holding potential to recover. On the other hand, stabilizing a deep closed state would require more depolarized voltage and, hence, more time to transition the channel to an open state, consistent with increases in first latency (Liu and Rittenhouse, 2000). The first latency of 30 ms fits well with the notion that L-channels in SCG neurons couple to β_2a. Liu and Rittenhouse (2000) showed that in the presence of AA, this time increases to 90 ms. In these single-channel experiments, the effect of AA on a single channel's activity is directly monitored. Here, we hypothesize that when measuring whole cell current in the presence of AA, the observed current arises from channels not inhibited by AA.

Our hypothesis that AA modulates channels in deep, closed conformations, but not channels that are in a conformation near to the open state, is reminiscent of the two populations of channels described as willing and reluctant. Historically, willing-reluctant gating terminology has been used to describe gating that occurs when G proteins bind to high voltage–activated Ca²⁺ (Ca_V2) channels (Bean, 1989). G protein G_βγ subunits can bind Ca_V2 α₁ subunits and inhibit Ca_V2 current by stabilizing “reluctant gating” conformations. This inhibition can be relieved by a strong depolarizing prepulse (typically to +120 mV), releasing the G_βγ subunit and allowing the channel to occupy “willing gating” conformations. Although CTL Ca_V1.3 currents display facilitation after a strong depolarizing prepulse, the mechanism is not mediated by G_βγ binding Ca_V1.3 (Bell et al., 2001; Safa et al., 2001) and remains unknown. Determining whether AA promotes facilitation under certain conditions as well as inhibition will require further detailed experimentation.

More recently, the willing-reluctant terminology has been used to describe how PIP₂ increases channel availability of the Ca_V2 family (Wu et al., 2002; Gamper et al., 2004). PIP₂ has been hypothesized to act at two unidentified sites termed the “S” and “R” sites on Ca_V2 channels (Wu et al., 2002). When PIP₂ binds to the high affinity S site, channel activity is stabilized and whole cell currents are protected from rundown. When PIP₂ binds to the lower affinity R site, channels open more slowly during weak depolarizations, similar to the reluctant gating observed with G proteins (Wu et al., 2002; Michailidis et al., 2007). Exogenous application of AA appears to exert the opposite effects of PIP₂'s dual actions on Ca_V2.2 channels where AA decreases current magnitude but increases the activation and inactivation kinetics (unpublished data). These two actions of AA are distinguishable both biophysically and pharmacologically (Barrett et al., 2001; Liu et al., 2001). Because AA is a product of PIP₂ breakdown by several phospholipases, these findings raise an interesting possibility that AA also acts at two analogous sites on Ca_V1.3 to antagonize PIP₂'s putative actions on L-current (Michailidis et al., 2007). A similar bidirectional relationship between PIP₂ and AA has been documented for Kir and K_V channels in which PIP₂ and AA are hypothesized to act at distinct sites (Oliver et al., 2004; Tucker and Baukrowitz, 2008). Whether exogenously applied AA competitively displaces PIP₂ or acts directly or indirectly to alter a distinct site on Ca_V1.3 channels to elicit inhibition has not yet been determined; however, free AA interacting at the same location(s) of PIP₂'s fatty acid tail(s) to confer inhibition is an appealing idea in its simplicity.

In addition to current inhibition, AA decreases TTP and the τ_ACT becomes faster. We present the following scenario to explain our interpretation of this result. At rest, channels reside in a series of closed conformations whereby τ_ACT describes the rate of channels transitioning from different closed conformations to the open conformation: RC + WC→O. Under CTL conditions, RC channels may have PIP₂ bound to the R site as well as the S site (Wu et al., 2002), resulting in channels available to open but at a slower rate of transition to the open conformation than WC channels. However, the overall rate will be the average of RC + WC channels and is dependent on the proportion of channels occupying each closed conformation. If AA stabilizes a deep, closed conformation, RC channels, which transition through the closed states more slowly than WC channels, would be more susceptible to inhibition (Fig. 12 A). Thus, AA does not cause reluctant or willing gating, but rather AA is more likely to act on reluctant gating channels.

If AA selectively stabilizes RC channels in a closed state, making them unlikely to open during a test pulse, only the WC channels, with the faster activation rate, will open and determine the whole cell current kinetics. The decreases in TTP (i.e., peak inward current occurs faster) after exposure to AA is consistent with this model. Notably, in experiments with β_2a-containing channels, TTP changes are not observed with 40-ms test pulses as this time frame is not long enough to account for the changes in activation; however, TTP changes are observed for β_2a in the 2.2-s pulses that are long enough to account for first latency changes from 30 to 90 ms observed in SCG neurons. In several experiments, the membrane potential was held at −60 mV, near the threshold for opening (−50 mV) for Ca_V1.3 (Xu and Lipscombe, 2001). At this voltage, more channels would be predicted to reside in conformations closer to the open state. We observed less inhibition at −60 mV than with a holding potential of −90 mV, at which a greater percentage of channels would reside in deeper closed conformations (Fig. 12 B). Thus, in the presence of AA, the observed smaller current with a faster τ_ACT arises from willing channels residing closer to the open state and “above” the closed state(s) stabilized by AA. It will be interesting to confirm these kinetic changes produced by AA for each of the subunit combinations in future studies of single-channel gating.

To address whether AA stabilizes inactivated as well as closed conformations, we measured whole cell Ca_V1.3 currents under various inactivation protocols in the absence and presence of AA. Although r40 decreased for β_1b-, β₃-, and β₄-containing channels, the difference current was relatively non-inactivating. Because Ca_V1.3, β_2a currents only undergo slow inactivation, the decrease in r2200 revealed that there is a common mechanism for inhibition by AA that occurs regardless of the β subunit that is coexpressed. AA may preferentially inhibit the RC population of channels, and thus for β_2a-containing channels, the contribution of reluctant gating channels undergoing slow inactivation is decreased. Therefore, the percentage of WC channels, which remain available to open and available to undergo inactivation, is greater in the normalized holding potential–dependent inactivation curves. These findings suggest that when Ca_V1.3b exhibits willing gating, L-channels will be less susceptible to inhibition by AA. Taken together, AA decreases the contribution of reluctant gating channels to the whole cell current, resulting in an increased proportion of the remaining current displaying inactivation kinetics of willing channels. Moreover, a shift in percentage of willing versus reluctant channels is manifested in the current kinetics in a Ca_Vβ subunit–dependent manner after AA; fast inactivation increases for β_1b-, β₃-, and β₄-containing channels, whereas holding potential–dependent inactivation increases for β_2a-containing channels.

Why is less inhibition of Ca_V1.3b current observed after 1 min with AA for channels coexpressed with β_2a? One possibility is that the site of inhibition by AA occurs on the Ca_Vα₁ subunit. In this case, the palmitoylated β_2a may alter the conformation of the channel to noncompetitively regulate the availability of an AA binding pocket. Second, residues in the S6 transmembrane domains affect inactivation, suggesting an overall change in the pore structure (Stotz et al., 2004), a concept that has been compared with pore collapse or C-type inactivation describing slow inactivation in Na⁺ and K⁺ channels (Lopez et al., 1994; Ogielska et al., 1995; Vilin et al., 1999). Channels coassembled with β_2a have a higher open probability than β_1b or β₃ (Colecraft et al., 2002) and would provide greater access for AA to contact these pore residues, promoting greater holding potential–dependent inactivation at more depolarized voltages.

Indeed, AA, or molecules with AA moieties in their structure including anandamide, can occupy the binding site for L-channel agonists and antagonists, such as dihydropyridines (DHPs) (Shimasue et al., 1996; Jarrahian and Hillard, 1997). DHPs interact with several critical amino acid residues in the pore of Ca_V1.2 and Ca_V1.3 (Sinnegger et al., 1997; Wappl et al., 2001; Liu et al., 2003; Yamaguchi et al., 2003). Thus, pore residues overlapping with the DHP site may constitute a site for AA interaction. However, if AA interacted with pore residues, coexpression of Ca_V1.3b with β_2a should increase rather than decrease the magnitude of Ca_V1.3 current inhibition by AA. Alternatively, slow open-channel pore block by AA is a third possible mechanism of inhibition that would explain an increase in the rate of inactivation as well as a decreased TTP. This possibility seems unlikely because AA inhibits channels held closed in the absence of test potentials. Regardless, a role for Ca_Vβ subunits modulating the AA site, whether by occupying a DHP site in the pore or by pore block, is plausible because these accessory subunits can fine-tune the affinity of L-channel blockers (Hering, 2002).

A fourth possibility is that two sites of AA action could exist, and either the palmitoyl group(s) or the inability of β_2a to fast inactivate prevents AA interaction with Ca_V1.3. In fact, the time course of inhibition of Ca_V1.3b, β_2a currents appears slower than with the other Ca_Vβ subunits (Fig. 1 E, top), suggesting that palmitoylated β_2a may antagonize AA binding. Our findings suggest that the palmitoyl groups of β_2a either directly or indirectly impede AA's ability to inhibit channel activity (Figs. 10 and 11). We propose a model whereby two AA modulatory or fatty acid sites exist on Ca_V1.3b. We have included two sites in our model based on a previous determination that AA inhibited T-channels with a Hill coefficient of 1.6 (Talavera et al., 2004). T-channel expression requires no other accessory subunit, suggesting that at least two sites of interaction exist on Ca_Vα subunits. When Ca_V1.3b colocalizes with β_1b, β₃, or β₄, both sites are available. When Ca_V1.3b colocalizes with β_2a, the palmitoyl groups interfere or occupy at least one of the sites, resulting in less inhibition by AA (Fig. 12 C).

Ca_V1.3 channels colocalize with large-conductance Ca²⁺-activated K⁺ channels (BK channels) in the brain (Grunnet and Kaufmann, 2004). Increases in AA levels may directly affect Ca²⁺-activated K⁺ channel activity, resulting in an enhancement of K⁺ current (Denson et al., 2000; Clarke et al., 2002; Sun et al., 2007). The influence of AA on Ca²⁺ influx and K⁺ efflux, as well as several other ion channels (Meves, 2008), could profoundly alter membrane excitability. Additionally, proteins with two adjacent palmitoyl groups, such as β_2a, may localize to cholesterol-rich microdomains (Pike, 2003), which may in turn alter Ca_V1.3 susceptibility to modulation by AA.

In certain pathophysiological conditions, such as ischemia and stroke, concentrations of AA, comparable to those used in the present study, are released in the brain (Lipton, 1999). Brain regions with high expression of β_2a subunits (Birnbaumer et al., 1998; McEnery et al., 1998) may be less affected by the cytotoxic effects of AA release downstream of neurotransmitter activation because β_2a-containing Ca_V1.3b currents show less inhibition. Conversely, AA could increase cell survival by inhibiting Ca²⁺ influx because large sustained influxes of Ca²⁺, such as that of the non-inactivating β_2a current, can be cytotoxic (Lipton, 1999). Similarly, currents from β_1b-, β₃-, or β₄-containing Ca_V1.3b channels show more AA inhibition; this overall depression of Ca²⁺ influx may decrease excitotoxicity. Whether AA acts indirectly or by binding to Ca_V1.3b or the Ca_Vβ subunit remains unanswered. Because AA is hydrophobic, if inhibition is due to a direct binding effect, Ca_V1.3 with its 24 membrane–spanning segments seems the more likely candidate as opposed to the cytosolic Ca_Vβ subunit. Future studies involving mutational analyses of Ca_Vα or the Ca_Vβ subunits should provide these answers.

The authors would like thank Liwang Liu, John Heneghan, and Tora Mitra-Ganguli for their assistance and discussions; TzuFeng Chung for maintaining the HEK M1 cell line; and José Lemos, William Kobertz, Haley Melikian, Elizabeth Luna, and Edward Perez-Reyes for thoughtful critique of the manuscript.

This work was supported by National Institutes of Health grants (NS34195 and NS07366) and funding from UMASS Medical School.

Lawrence G. Palmer served as editor.