The presence of ca²⁺ channel β subunit is required for mitogen-activated protein kinase (mapk)-dependent modulation of α1b ca²⁺ channels in cos-7 cells

Abstract

In rat sensory neurones, voltage-dependent calcium channels (VDCCs), including the N-type, are tonically up-regulated via Ras/mitogen-activated protein kinase (MAPK) signalling. To determine whether VDCC β subunit is involved in this process, the role of the four neuronal βs (β1b, β2a, β3, β4) in MAPK-dependent modulation of α1B (Ca_v2.2, N-type) Ca²⁺ channels has been examined in COS-7 cells. MAPK is exclusively activated by MAPK kinase (MEK), and here, acute application of a MEK-specific inhibitor UO126, significantly inhibited peak α1B Ca²⁺ channel current (I_max) within a period of 5–10 min, regardless of which β subunit was co-expressed (25-50%, P < 0.01). With β2a however, the percentage inhibition of I_max was less than that observed with any other β (ANOVA: F_3,34 = 6.48, P < 0.01). UO126 also caused a hyperpolarising shift (6 ± 1 mV, P < 0.001) in the voltage dependence of β2a current activation, such that inhibition occurred only at depolarised potentials (> +5 mV) whereas at more negative potentials the current amplitude was enhanced. A marked change in β2a current kinetics, perceived either as decreased activation or increased inactivation, was also associated with UO126 application. A similar effect of UO126 on β4 current kinetics was also observed. The β2a-specific effects of UO126 on current inhibition and voltage dependence of activation were abolished when α1B was co-expressed with de-palmitoylated β2a(C3,4S), in which amino terminal cysteines 3 and 4 had been mutated to serines. In the absence of β subunit, UO126 had no effect on α1B Ca²⁺ channel current. Together, these data suggest an absolute requirement for β in MAPK-dependent modulation of these channels. Since β subunits vary both in their temporal expression and localisation within neurones, β subunit-dependent modulation of N-type Ca²⁺ channels via MAPK could provide an important new mechanism by which to fine-tune neurotransmitter release.

Voltage-dependent Ca²⁺ channels (VDCCs) are multi-subunit proteins composed of a pore-forming α1 subunit, a regulatory cytoplasmic β subunit, and a largely extracellular α2δ subunit (reviewed in Dolphin, 1998). To date, at least ten α1 subunit genes have been identified, four β and three α2δ-subunits, many of which also have multiple splice variants (Birnbaumer et al. 1994; Ertel et al. 2000). Consequently, there is enormous potential for functional and structural diversity of VDCCs. This is reflected in the wide variety of cellular processes which are dependent on Ca²⁺ influx through this group of channels, e.g. muscle contraction, neurotransmitter release and gene expression, as well as Ca²⁺-mediated processes of cell growth (reviewed in Berridge, 1998). An understanding of the mechanisms involved in regulation of VDCCs is therefore of crucial importance.

Several signalling molecules involved in the control of cell growth, notably the small G-protein Ras, receptor tyrosine kinases and phosphatidylinositol 3-kinase, are known to play a role in regulating ion channels, including VDCCs (Pollack & Rane, 1996; Fitzgerald & Dolphin, 1997; Blair et al. 1999). However, the mechanisms involved are not well understood. It was recently reported that Ras-dependent up-regulation of VDCCs in rat dorsal root ganglion (DRG) neurones requires activation of the mitogen-activated protein kinase (MAPK/extracellular signal regulated kinase, ERK) pathway (Fitzgerald, 2000). In these cells, pharmacological suppression of endogenous MAPK activity by application of a MAPK kinase (MEK)-specific inhibitor, UO126, caused marked inhibition of Ca²⁺ channel current within 10 min (Fitzgerald, 2000). This time course for inhibition was suggested to be consistent with modulation of existing channels rather than, or at least in addition to, any gene expression-mediated effects which are perhaps more usually associated with the MAPK pathway. Furthermore, since all components of calcium channel current in DRGs were apparently up-regulated via this pathway, it was also suggested that there may be a common mechanism for interaction at the channel level.

The biophysical and pharmacological properties of VDCCs are primarily characterised by the pore forming α1 subunit. The β subunit, and to a lesser extent α2δ, have a regulatory function with profound influence on expression levels and biophysical properties, notably the voltage dependence and current kinetics of activation and inactivation (Birnbaumer et al. 1998; Walker & De Waard, 1998). Potentially therefore, any or all of these subunits could be targeted by Ras/MAPK signalling to effect tonic up-regulation of Ca²⁺ channels. The aim of the present study was to determine whether β subunit is involved in the MAPK-dependent modulation of α1B (Ca_v2.2) Ca²⁺ channels, equivalent to the N-type channels that constitute a major component of current in rat DRGs. Here it is demonstrated that in common with native N-type VDCCs, α1B Ca²⁺ channels expressed in COS-7 cells also exhibit MAPK-dependent tonic up-regulation of current density that can be inhibited by acute application of the MEK-specific inhibitor, UO126. This effect is observed in the presence of any of the four neuronal β subunits (β1b, β2a, β3, β4) but not in the absence of β. These data suggest an absolute requirement for β subunit in MAPK-dependent modulation of α1B Ca²⁺ channels.

Methods

Materials

The following cDNAs were used: rabbit α1B (GenBank accession number D14157); rat α2δ1 (neuronal splice variant, M86621); rat β1b (X11394); rat β2a (M80545); rat β3 (M88751); rat β4 (L02315); de-palmitoylated β2a (C3,4S), in which cysteines 3 and 4 were mutated to serine (Stephens et al. 2000); dominant negative MEK1 (Cowley et al. 1994); and mut-3 green fluorescent protein (mut-3 GFP, U73901). All cDNAs were subcloned into pMT2 (Swick et al. 1992).

Expression of constructs

COS-7 cells were grown to confluency in Dulbecco's modified Eagle's medium (DMEM) containing 10 % newborn calf serum plus 50 U ml⁻¹ penicillin-50 mg ml⁻¹ streptomycin (Gibco/Life Technologies, Paisley, UK). Cells were transiently transfected with α1B-, β-, α2δ1- and mut-3 GFP-pMT2 constructs, using the GenePORTER transfection reagent (QBiogene, Harefield, Middlesex, UK). A master mix of cDNA was made in the ratio 15, 5, 5, 1 μg of α1B-, β-, α2δ1- and mut-3 GFP, respectively. In experiments in the absence of β subunit, blank pMT2 vector was substituted for β, to maintain the total complement of cDNA in the correct ratio. Briefly, COS-7 cells were plated onto 35 mm dishes to give approximately 40–50 % confluency. For each dish of cells, 1 ml of transfection mix was made up in serum-free DMEM without antibiotics. In one-half of the transfection volume, 4 μl of DNA master mix was added, and in the other half 20 μl of GenePORTER was included. After adding the cDNA to the GenePORTER, the transfection mix was vortexed, then left at room temperature (20-22 °C) for 2 h. For transfection of plated cells, the culture medium was replaced with 1 ml of transfection mix. Cells were then incubated at 37 °C for 4 h. Following this, transfection was terminated by adding 1 ml of culture medium containing 20 % serum plus antibiotics, and incubating the cells at 37 °C. Cells were maintained at 37 °C for 48 h after transfection, and prior to recording were replated using a non-enzymatic cell dissociation medium (Sigma, UK), then maintained at 25 °C for a further 1 to 4 h before recording.

Electrophysiology and solutions

Electrophysiological recordings were made from fluorescent COS-7 cells expressing the mut-3 GFP reporter gene. The whole-cell configuration of the patch-clamp technique (Hamill et al. 1981) was used to record barium currents in the following solutions. The internal solution contained (mm): caesium aspartate 140, MgCl₂ 2, CaCl₂ 0.1, Hepes 20, EGTA 5, K₂ATP 1; adjusted to pH 7.2 with CsOH, adjusted to 310 mosmol l⁻¹ with sucrose. The external solution contained (mm): TEABr 160, MgCl₂ 1, KCl 5, NaHCO₃ 1, Hepes 10, glucose 4, BaCl₂ 10; adjusted to pH 7.4 with Tris-base, adjusted to 320 mosmol l⁻¹ with sucrose. In the absence of any β subunit, it was necessary to use 20 mm Ba²⁺ in order to obtain robust, measurable currents. All experiments were performed at room temperature (20-22 °C). In some cells, depending on which β subunit was co-expressed, a time-dependent run-up or run-down of peak I_Ba occurred during the initial period after establishing whole-cell mode. All the data presented here were therefore taken from currents that had been allowed to stabilise for at least 10 min before any measurements were made.

Patch pipettes of resistance 2–4 MΩ were pulled from thin-walled borosilicate glass tubing, fire-polished and coated with Sigmacote (Sigma). An Axopatch-1D amplifier (Axon Instruments, Union City, CA, USA) was used for recordings which were filtered at 2 kHz (4-pole Bessel filter) and digitised at 5–10 kHz using a Digidata 1200 A/D converter (Axon Instruments, USA). The cells were held at a potential of −80 mV where the holding current was less than −0.05 nA and series resistance normally less than 10 MΩ. Currents were recorded with the cell capacitance compensated and leak currents subtracted using a P/4 subtraction protocol. Series resistance was compensated up to 80 % and only cells that were adequately space clamped, as judged by a graded current activation, were used. The liquid junction potential was 6 mV; values in the figures and text have not been corrected for this. Data acquisition and analysis was performed using pCLAMP software (version 6.0.4, Axon Instruments) and Origin version 5.0 (OriginLab Corp., Northampton, MA, USA).

Drug or control solutions were applied under continuous perfusion. The MEK-specific inhibitor, UO126 (Promega, Southampton, UK) and its inactive analogue, UO124 (Calbiochem, Nottingham, UK), were first dissolved in DMSO and then added to external solution to give a final concentration of 20 μM. There were no adverse effects of DMSO (0.01 %) on the α1B Ca²⁺ currents expressed (data not shown). These reagents were stored as aliquots of stock solution at −20 °C, diluted and used freshly each day.

All data are presented as the mean ± s.e.m. Unless stated otherwise, statistical analysis was carried out by Student's paired t test, using 95 % confidence limits (Statistica version 4.0, StatSoft Inc., Tulsa, OK, USA). Where more than two factors were tested, one-way ANOVA was used as appropriate.

Results

Effect of UO126 on α1B Ca²⁺ current density

In initial experiments, α1B was co-expressed with auxiliary β1b subunit, in COS-7 cells. The inward Ba²⁺ currents recorded from these cells were typically high voltage activated, peaking at between +10 and +20 mV, with a midpoint of voltage-dependent activation, V_50,act, of 2 ± 2 mV (n = 5, Table 1). To determine whether this heterologously expressed α1B could be modulated by MAPK in the same way as native N-type channels in DRGs, MEK inhibitor UO126, was acutely applied to cells. UO126 is an established and highly specific inhibitor of MEK, that binds to activated MEK1 and 2 to prevent the subsequent activation of MAPK1 and 2 (De Silva et al. 1998; Favata et al. 1998). Pre-incubation of intact cells, including COS-7 cells, with 10–50 μM UO126 for 15 min has previously been shown to block MAPK activity (Favata et al. 1998), and in DRGs, whole-cell Ca²⁺ channel current amplitude is reduced by up to 45 % in the presence of this drug (Fitzgerald, 2000). Here, acute application of UO126 to α1B,β1b-expressing cells caused a marked inhibition of maximum peak current density, I_max, that reached a maximum of −50 ± 5 % within 5–10 min of application (n = 11, P < 0.01; Fig. 1A, Table 1). A distinctive sigmoidal time course for UO126-induced current inhibition was observed, as shown in Fig. 1A. This effect was reversible upon washout with control solution for approximately 10 min (data not shown). Over a similar time period, the inactive analogue UO124, which does not inhibit MEK activity even at concentrations of 100 μM, had no significant effect on current density (Favata et al. 1998; Fig. 1B, Table 2). On average, I_max decreased by −12 ± 3 % (n = 11) in the presence of UO124. The slight current decrease observed in some UO124-treated cells was similar to normal levels of current run-down seen in cells maintained in control solutions for a similar period of time (data not shown). In COS-7 cells that also overexpressed a dominant negative MEK1 mutant, in which serines 217 and 221 had been mutated to alanine (MEK1(S217,221A), Cowley et al. 1994), the effect of UO126 on α1B,β1b was significantly reduced, as expected assuming the level of basal MAPK activity had been reduced. In control cells, I_max was inhibited by −54 ± 8 % (n = 5) compared with inhibition of −19 ± 5 % (n = 8, P < 0.01) in mutant MEK1-expressing cells. These results suggest that in common with whole cell Ba²⁺ currents recorded in DRG neurones, endogenous MAPK activity in COS-7 cells also tonically up-regulates α1B Ca²⁺ channel current.

Table 1.

Effect of UO126 on the biophysical properties of β1B Ca²⁺ channels expressed in COS-7 cells

	Imax (pA pF−1)			V50,act (mV)			k (mV)
β subunit	Control	UO126	n	Control	UO126	n	Control	UO126	n
β1b	−99 ± 20	−49 ± 11**	11	2 ± 2	2 ± 3	5	3 ± 1	6 ± 1**	5
β2a	−91 ± 18	−71 ± 17**	12	5 ± 1	−1 ± 1***	8	4 ± 1	4 ± 1	8
β2a(C3,4S)	−74 ± 14	−48 ± 9**	11	5 ± 4	3 ± 3	9	4 ± 0.4	5 ± 0.4	9
β3	−80 ± 21	−42 ± 10**	7	6 ± 2	4 ± 3	7	4 ± 1	6 ± 1	7
β4	−70 ± 14	−36 ± 10**	6	4 ± 3	2 ± 2	5	5 ± 0.2	6 ± 0.3*	5
(-)β	−6 ± 1	−6 ± 1	6	29 ± 5	27 ± 2	4	11 ± 2	10 ± 1	4

I_max is the maximum peak current density. Individual current density–voltage plots were fitted with a Boltzmann function:

where V_rev is the reversal potential, V_50,act is the voltage for half-maximal activation of current, g is the conductance and k is the slope factor. Statistical analysis used Student's paired t test. Asterisks denote statistically significant differences as follows:

^*P < 0.05

^**P < 0.01

^***P < 0.001. n is the number of cells tested for each treatment.

Figure 1. MAPK kinase (MEK) inhibitor UO126 rapidly reduces α1B Ca²⁺ current regardless of which neuronal β subunit is co-expressed.

A, representative time course showing the effect of UO126 on α1B co-expressed with β1b. Inset shows traces obtained at time points 1 and 2. B, representative time course and associated current traces showing the lack of effect of the inactive analogue, UO124, on α1B,β1b current. C, percentage inhibition of maximum peak α1B current density (I_max) following application of UO126 (20 μM, ▪), and the inactive analogue UO124 (20 μM, □). Asterisks denote significant differences in percentage inhibition of I_max in the presence of UO126 compared with UO124, where ** P < 0.01, *** P < 0.001 (unpaired t test). Despite there being significant inhibition of current in the presence of all four β subunits, the inhibitory effect of UO126 was significantly less for β2a compared with the other βs (ANOVA: F_3,34 = 6.48, P < 0.01). Currents were evoked using 100 ms test pulses to +10 mV from a holding potential, V_h, of −80 mV.

Table 2.

Effect of UO124 on the biophysical properties of β1B Ca²⁺ channels expressed in COS-7 cells

	Imax (pA pF−1)			V50,act (mV)			k (mV)
β subunit	Control	UO124	n	Control	UO124	n	Control	UO124	n
β1b	−52 ± 10	−44 ± 7	11	5 ± 2	7 ± 3	8	5 ± 1	6 ± 1	8
βb2a	−122 ± 43	−120 ± 40	5	9 ± 5	6 ± 6	4	4 ± 1	5 ± 1	4
β3	−65 ± 23	−53 ± 19	4	10 ± 2	9 ± 1	3	6 ± 1	6 ± 0.3	3
βb4	−23 ± 9	−20 ± 8	7	13 ± 2	10 ± 1	5	6 ± 0.3	6 ± 0.3	5

All parameters are as previously defined in the legend to Table 1.

Heterogeneity in the β subunits of native N-type calcium channels has been demonstrated in rat brain (Scott et al. 1996), and in DRGs it is not known which β subunits are functionally expressed, nor which β subunit(s) is associated with N-type VDCCs in these neurones. Therefore, to determine whether UO126-induced inhibition of α1B current was dependent on a specific β subunit, α1B was co-expressed with each of the other neuronal βs, β2a, β3 and β4 and the effects of UO126 and its inactive analogue UO124, compared. These data are summarised in Fig. 1C and in Tables 1 and 2. In common with the effect on α1B co-expressed with β1b, so in the presence of any of the other three neuronal β subunit isoforms, peak current density, I_max, was also significantly inhibited by between −25 and −50 %, following application of UO126 (Fig. 1C, Table 1). As with β1b, sigmoidal time courses for inhibition of current, usually between 5 and 10 min, were observed in the presence of β2a, β3 or β4 and the effect reversible upon washout with control solution (data not shown). In contrast, the inactive analogue UO124, had no significant effect on current density in the presence of any of the β subunits (Fig 1C, Table 2). Although UO126 caused marked current inhibition regardless of which β subunit was co-expressed, in the presence of β2a, the percentage inhibition of I_max (-25 ± 4 %, n = 12) was significantly lower than that observed with any other β (ANOVA: F_3,34 = 6.48, P < 0.01).

Effect of UO126 on the biophysical properties of α1B Ca²⁺ current

Different β subunits exert profoundly different effects on the biophysical properties of VDCCs. In particular, the current kinetics and voltage dependence of current activation and inactivation are known to vary according to which β subunit is expressed (e.g. De Waard & Campbell, 1995; Birnbaumer et al. 1998; Walker & De Waard, 1998). The effects of UO126 on the biophysical properties of α1B current were therefore also compared in the presence of each of the four neuronal β subunit isoforms.

Cells expressing α1B with either β1b or β3, were found to produce currents with the fastest inactivation kinetics, followed by β4, then β2a (Fig. 2A). This agrees with earlier work by Stephens et al. (2000), also in COS-7 cells. Under control conditions, the percentage of inactivation over a 200 ms test pulse to +10 mV, Δ% 200, was −62 ± 7 % for β1b and −57 ± 8 % for β3 (Fig. 2A and B). Values of Δ% 200 were calculated as (I_peak - I_end)/I_peak, where (-) indicates a decrease in current relative to I_peak (Fig. 2A and C), Although the percentage of inactivation of β1b and β3 currents appeared to increase slightly in the presence of UO126, indicated by more negative values of Δ% 200 (Fig. 2A), these effects were not significant, and a similar tendency towards increased inactivation was also observed following application of UO124 (Fig. 2C). Comparison of the current density-voltage relationships before and after UO126 application, revealed that the voltage dependence of current activation was unaffected in cells expressing either β1b or β3. Thus, the voltage for half-maximal activation of current, V_50,act, was not significantly shifted in either case (Table 1). In the presence of β1b however, the slope factor k did increase by 3 mV (P < 0.05) following application of UO126, indicating a decrease in the voltage sensitivity of these channels (Table 1). A similar, though not statistically significant trend was also observed with β3.

Figure 2. The effect of UO126 on the kinetics of α1B Ca²⁺ current is dependent on the specific β subunit expressed.

A, percentage change in peak current over 200 ms depolarising test pulses to +10 mV from V_h −80 mV (Δ% 200), before (▪) and after application of 20 μM UO126 (□). Δ% 200 was derived as follows; for inactivating ((I_peak - I_end)/I_peak), and activating currents ((I₂₀ - I_end)/I₂₀), where I₂₀ is the current at 20 ms. I₂₀ corresponds approximately to the point at which fast activation switches to slow activation. Values of Δ% 200 are shown either as (-) or (+) according to whether I_end is decreased or increased relative to I_peak/I₂₀. Asterisks denote significant differences between Δ% 200 before and after application of UO126, where ** P < 0.01, *** P < 0.001. B, example traces recorded from β1b- and β2a-expressing cells before and after application of UO126. C, percentage change in Δ% 200 measured before (▪) and after application of the inactive analogue, UO124 (). D, linear regression analyses for β4- and β2a-expressing cells. Da, regression of Δ% 200 (Control) vs. Δ% 200 (UO126) in β4-expressing cells, where r = 0.94 and P < 0.05. Db, regression of Δ% 200(control) vs. Δ% 200 (UO126) for β2a gives values of r = 0.61, P < 0.05.

In the majority of β4-expressing cells, Ba²⁺ currents typically exhibited partial inactivation during 200 ms depolarising pulses to +10 mV. However in a few cells, currents showed either no inactivation or even slight activation at the end of the test pulse (Fig. 2D). Consequently, the effects of UO126 on β4 current kinetics were most easily compared in terms of the percentage change in current amplitude over 200 ms. This value was measured as (I₂₀ - I_end)/I₂₀, where I₂₀ is the current at 20 ms into the pulse, and corresponds approximately to the point at which fast activation switches either to slow activation or to inactivation. For comparison of β4 with β1b/β3, this percentage change in current is also referred to as Δ% 200, where (-) indicates current decrease and (+) denotes increased current at I_end relative to I₂₀ (Fig. 2). Importantly, regardless of the current kinetics observed under control conditions, UO126 always shifted Δ% 200 to more negative values, and hence decreased Ba²⁺ entry, as indicated by regression analysis of Δ% 200 (control) vs. Δ% 200 (UO126) (r = 0.94, P < 0.05; Fig. 2Da). On average, UO126 caused a marked increase in inactivation from −2 ± 8 to −28 ± 9 % (n = 5, P < 0.001) when measured at +10 mV. This trend was observed consistently throughout the voltage range tested (+5 to +50 mV, data not shown). In spite of a tendency in some cells for inactive UO124 also to induce a slight negative shift in Δ% 200 measured at +10 mV, the effect was not statistically significant (Fig. 2C). Whilst no shift in the midpoint of voltage-dependent activation, V_50,act, was observed following UO126 application, the slope factor k did increase slightly suggesting some decrease in the voltage sensitivity of these channels (Table 1). The effects of UO126 on current inactivation were further investigated by comparing steady-state inactivation in the absence or presence of inhibitor. Due to the difficulty of recording from individual cells for prolonged periods, β4-expressing cells were pre-incubated for 30 min either in normal growth medium (control) or in growth medium plus UO126, prior to recording Ba²⁺ currents. Both the current-voltage relationships and steady-state inactivation curves obtained from individual cells within each treatment group were then compared. As expected, Δ% 200 measured at I_max exhibited a significant decrease in the presence of UO126, from 31 ± 16 % (n = 6) to −12 ± 2 % (n = 6). However, comparison of the steady-state inactivation curves indicated that in spite of a tendency for UO126 to induce a slight hyperpolarising shift in voltage-dependent inactivation (data not shown), neither the midpoint of voltage-dependent inactivation, V_50,inact, nor the slope factor, k, differed significantly between the two groups of cells. On average, V_50,inact in control cells was −42 ± 4 mV compared with −50 ± 4 mV in UO126-treated cells, whilst k averaged 13 ± 1 mV vs. 14 ± 1 mV in the absence or presence of UO126, respectively.

In common with other α1 subunits, notably α1A (De Waard & Campbell, 1995), α1C and α1E (Qin et al. 1998), co-expression of β2a with α1B is known to produce currents with significantly retarded voltage-dependent inactivation when compared with other VDCC β subunits (e.g. Canti et al. 2000; Stephens et al. 2000). Thus, over the 200 ms test pulses used here, little or no current decay was observed in the presence of β2a, and in most cells the currents were still activating at the end of the test pulse (Fig. 2A and B). Activation of α1B in the presence of β2a is typically composed of a fast component followed by a slow component of activation (Fig. 2B). In those β2a-expressing cells that exhibited a significant proportion of this slowly activating current, the relative size of this component of current tended to vary both within and between batches of cells. Consequently, the effect of UO126 on β2a current kinetics was also compared in terms of the percentage change in current amplitude over 200 ms depolarising test pulses, as previously defined in relation to β4. Comparison of Δ% 200 at +10 mV, measured before and after application of UO126 indicated a significant change in Δ% 200 from 7 ± 6 to −17 ± 6 % (n = 8, P < 0.01; Fig. 2A and B). In the majority of cells, the slow component of activation was thus replaced by partial inactivation. However, in the few cells that exhibited a particularly large component of slow activation under control conditions, the effect of UO126 was perceived as a marked decrease in activation. Nonetheless, the shift to more negative values of Δ% 200 following UO126 application was seen in every cell (r = 0.61, P < 0.05; Fig. 2Db) and furthermore, the effect was consistently observed throughout the voltage range tested (+5 to +50 mV, data not shown). As with β4, the overall effect of UO126 on β2a current kinetics was therefore to reduce Ba²⁺ entry. In contrast, the application of UO124 had no significant effect on the current kinetics (Fig. 2C). It was possible that the effect of UO126 on the current kinetics may partially be due to the hyperpolarising shift of 6 ± 1 mV (n = 8, P < 0.001), in V_50,act that was also specifically associated with UO126 application to β2a-expressing cells, (Table 1, Fig. 3A). However, comparison of Δ% 200 measured at I_max before (5 ± 5 %, n = 8) and after UO126 application (-20 ± 4 %, P < 0.01), indicated that the effect of UO126 on current kinetics was independent of the shift in the current-voltage relationship. Importantly, the hyperpolarising shift in the current-voltage relationship also indicated that on average, current inhibition occurs only at depolarised potentials greater than +5 mV, whereas current is enhanced at more negative voltages. Indeed, when UO126 was applied to cells stepped to 0 mV, an average increase in current density of 61 ± 25 % (n = 3) was observed. This increase in current was consistent with that predicted from the current-voltage plots shown in Fig. 3A. To examine further the voltage dependence of activation, conductance-voltage data, derived from current density-voltage plots, was normalised to give the activation curves shown in Fig. 3B. These activation curves also indicated that half-maximal activation of current was significantly shifted from 6 ± 1 mV in control conditions to 0.1 ± 1 mV in the presence of UO126 (P < 0.01). The slope (k) however, was unaffected (control = 4 ± 1 mV, UO126 = 5 ± 1 mV), implying no change in the voltage sensitivity of these channels. The application of UO124 to α1B co-expressed with β2a had no effect on the voltage dependence of current activation (Table 2). Steady-state inactivation in the absence or presence of UO126 was also compared using the same approach as previously outlined for β4. In agreement with data obtained from individual cells, comparison of the current-voltage plots recorded from cells pre-incubated in either control or UO126-containing growth media also indicated a significant reduction in Δ% 200 following treatment with UO126. The average value of Δ% 200 measured at I_max in control cells was 85 ± 20 % (n = 8) compared with 28 ± 11 % (n = 8, P < 0.05) in the presence of UO126. Although comparison of the steady-state inactivation curves suggested a slight tendency for increased inactivation in UO126 pre-treated cells, particularly at potentials between −20 and +20 mV, neither the midpoint for inactivation, V_50,inact, nor the slope factor, k, were significantly different from controls (Fig. 3C). V_50,inact for controls was −8 ± 3 mV compared with −13 ± 2 mV in the UO126-treated cells, whilst k, averaged 7 ± 2 mV (control) vs. 5 ± 2 mV (UO126).

Figure 3. Effects of UO126 on voltage dependence of current activation and inactivation and current kinetics compared in cells expressing wild type β2a vs. de-palmitoylated β2a(C3,4S).

A, average normalised current density-voltage relationships obtained before (▪) and after application of 20 μM UO126 (▪) to a wild type β2a-expressing cell (n = 8). Continuous lines indicate the Boltzmann fits to the averaged curves. Boltzmann fits were derived using the equation provided in the legend to Table 1. B, conductance-voltage relationships were derived from individual current density voltage plots and fitted with a Boltzmann function of the form: G/G_max = 1/(1 + exp((V_50,act - V)/k), where G_max is the maximum conductance. The peak conductance G at each test potential was calculated from the corresponding peak current, I, as follows: (G = I/(V - V_rev)). V₅₀, k and V_rev are as previously defined in the legend to Table 1. C, averaged steady-state inactivation curves recorded from wild type β2a-expressing cells pre-incubated for 30 min in either control growth medium (▪, n = 8) or growth medium + 20 μM UO126 (▪, n = 8). Individual curves were obtained from 100 ms duration test pulses to 10 mV, preceded by a 10 s pre-pulse to the potential given (-100 to +30 mV), from V_h −80 mV. Peak current values were normalised and the fits shown are to a Boltzmann function of the form: I/I_max = 1/(1 + exp(V_t - V_50,inact)/k)), where I_max is the maximum current, V_50,inact is the midpoint of voltage-dependent inactivation, V_t is the conditioning potential and k is the slope factor. D, histogram showing comparison of the midpoint for voltage-dependent activation, V_50,act, derived from current-voltage plots obtained from cells expressing wild type β2a and de-palmitoylated β2a(C3,4S), in which N-terminal cysteines 3 and 4 were mutated to serine. V_50,act values under control conditions (▪) were compared with values of V_50,act in the presence of UO126 (□). Significant differences in V_50,act are indicated by asterisks, where *** P < 0.001. Currents were evoked using 200 ms depolarising steps in 5 mV intervals, from V_h of −80 mV. E, histogram comparing Δ% 200 measured at +10 mV under control conditions (▪) and in the presence of UO126 (□), for wild type β2a and β2a(C3,4S). Asterisks denote significant differences in Δ% 200, where ** P < 0.01.

The β2a subunit is unique amongst the β subtypes in that its properties are dependent on the palmitoylation of two amino- (N-) terminal cysteine residues not found on any other β subunits (Chien et al. 1996). Functionally, mutation of these cysteines to serines prevents palmitoylation at these sites and partially reverses the distinguishing features of β2a regulation (Qin et al. 1998; Restituito et al. 2000; Stephens et al. 2000). Given the β2a-specific effects associated with UO126 application, the drug was also tested on α1B channels co-expressed with mutant β2a, (β2a(C3,4S)), in which serine had been substituted for cysteine at positions 3 and 4 to prevent palmitoylation. In contrast to the effect of UO126 on wild type β2a, no shift in the current density-voltage relationship was observed with β2a(C3,4S) in the presence of UO126, as indicated by the lack of effect on the V_50,act derived from current density-voltage plots (Table 1, Fig. 3D). Also, comparison of wild type β2a and β2a(C3,4S) current recordings made at the same time, showed that inhibition in β2a(C3,4S) cells (-28 ± 5 %, n = 11) was significantly greater than in β2a-expressing cells (-15 ± 6 %, n = 7; P < 0.05). Furthermore, the percentage inhibition of β2a(C3,4S) was not significantly different from that seen with β1b, β3 or β4 recorded from the same batches of cells (ANOVA: F_3,25 = 0.74, P = 0.5). Together these data indicate that the UO126-induced shift in voltage-dependence of activation and the reduced level of inhibition seen with β2a relative to other β isoforms, are somehow dependent on palmitoylation of the N-terminal cysteine residues of β2a. As previously noted, β2a(C3,4S) partially reverses the slow activation associated with wild type β2a (Qin et al. 1998; Stephens et al. 2000). In agreement with this, comparison of Δ% 200 measured at I_max for mutant β2a(C3,4S) Δ% 200 = −6 ± 6 %, n = 16) vs. wild type β2a cells (Δ% 200 = 27 ± 8 %, n = 16; P < 0.01) showed that cells expressing the de-palmitoylated mutant did not exhibit the strong activation normally associated with co-expression of wild type β2a in these cells. However, as with wild type β2a, so in β2a(C3,4S)-expressing cells, UO126 did cause negative shifts in Δ% 200 at each test potential, as indicated by regression analysis of Δ% 200 (control) vs. Δ% 200 (UO126), where r = 0.80 and P < 0.01 (data not shown). When measured at +10 mV, Δ% 200 in control conditions averaged 5 ± 10 % (n = 9), whereas in the presence of UO126, Δ% 200 was −27 ± 4 % (P < 0.01; Fig. 2D and E). Thus, the effect of UO126 on the current kinetics is not dependent on β2a palmitoylation.

Effect of UO126 on α1B expressed in the absence of β subunit

This study has shown that UO126 exerts a number of β subunit-specific effects on the biophysical properties of α1B currents. To confirm whether MAPK-dependent modulation of α1B channels is dependent on the presence of β subunit, UO126 was also tested on α1B channels expressed in the absence of β. In agreement with the reports of other workers (e.g. Canti et al. 2000; Stephens et al. 2000), cells expressing α1B minus β, produced very small rapidly activating and inactivating Ba²⁺ currents with I_max of −6 ± 1 pA pF⁻¹ (see Table 1). No endogenous Ba²⁺ currents were seen in either non-transfected or mock-transfected COS-7 cells (Fig. 4A). Application of UO126 to α1B minus β-expressing cells had no effect on the current density (Fig. 4). Furthermore, no shift in V_50,act, derived from current-voltage plots, was observed (Table 1), and nor was there any significant change in the current kinetics, as indicated by the lack of effect on Δ% 200 values measured before and after UO126 application. Under control conditions Δ% 200 was −70 ± 8 % (n = 4), whereas in the presence of UO126 the value was −75 ± 11 %. Thus, UO126 was completely without effect on α1B currents in the absence of β (Fig. 4). Importantly however, α1B,β1b currents recorded from the same batch of cells, were inhibited by −54 ± 8 % (n = 5) following application of UO126. Together these data suggest that the presence of β subunit is necessary for MAPK-dependent modulation of α1B Ca²⁺ channels.

Figure 4. In the absence of βsubunit, application of UO126 has no effect α1B Ca²⁺ current.

A, typical inward currents recorded in a COS-7 cell transiently transfected with α1B in the absence of β subunit, (-)β. These currents were evoked from a holding potential, V_h of −80 mV by means of depolarising steps to −15, 0, +20, +40 and +55 mV. Using the same voltage protocol no inward currents were observed in either mock-transfected (expression vector only) or non-tranfected cells (data not shown). B, a representative time course showing the lack of effect of UO126 on α1B Ca²⁺ current in the absence of β subunit. Inset shows current traces recorded at time points 1 and 2, as indicated. C, histogram comparing percentage inhibition of I_max in the presence of UO126, for α1B current in the presence of β1b and in the absence of β subunit, (-)β. D, average current density-voltage relationships before (▪) and after (▪) application of 20 μM UO126 to α1B Ca²⁺ channels, in the absence of β subunit. Dashed lines indicate Boltzmann fits to the current-voltage plots. Currents were evoked from V_h −80 mV, using 200 ms depolarising pulses in 5 mV steps.

Discussion

It was previously reported in DRG neurones that Ras-dependent tonic up-regulation of VDCCs involves activation of the MAPK (ERK) pathway (Fitzgerald, 2000). Pharmacological inhibition of endogenous MAPK activity in these cells causes rapid reduction of current density, a process that has been suggested to result from direct modulation of VDCCs (Fitzgerald, 2000). As yet, the mechanism(s) involved in MAPK-dependent modulation of these channels is not understood. Given the particular importance of Ca²⁺ channel β subunit in regulating expression levels and modifying the biophysical properties of α1 subunits, a role for β in MAPK-dependent regulation of VDCCs would seem likely. The aim of the present study was therefore to determine whether the presence of neuronal β subunit is required for MAPK-dependent modulation of α1B Ca²⁺ channels, equivalent to the neuronal N-type channels which constitute a major component of whole-cell current in rat DRGs (Fitzgerald, 2000). Here it is demonstrated that in common with native neuronal N-type channels, the heterologously expressed α1B Ca²⁺ channels also exhibit tonic up-regulation via MAPK-dependent signalling. MAPK-dependent up-regulation of current density is observed regardless of which β subunit is co-expressed but is not seen in the absence of β, implying an absolute requirement for β subunit in this process. Additional β subunit-specific effects on the voltage dependence of activation (β2a) and the current kinetics (β2a and β4) not only provide further support for the suggestion that MAPK signalling exerts non-transcriptional effects on VDCCs, but also suggests a means to provide some specificity of cellular function.

The absolute requirement for β in MAPK-dependent modulation of α1B, plus the fact that current inhibition was commonly observed in the presence of any β, suggests the likelihood of a common mechanism to account for the decrease in current density. As already mentioned, β subunits exert a repertoire of effects on VDCC α1 subunits. As regards its influence on current amplitude, β acts as a chaperone to traffic VDCCs to the cell membrane (Chien et al. 1995; Brice et al. 1997), and can also act directly on α1 subunits to alter the gating properties of channels (Neely et al. 1993; Kamp et al. 1996). In response to hormonal or protein kinase-dependent stimuli, rapid translocation of ion channels from intracellular compartments to the cell membrane, with a concurrent increase in current amplitude, has been reported to occur within minutes in some systems (Strong et al. 1987; Kanzaki et al. 1999). There is some possibility therefore, that MAPK signalling could increase current density via β subunit-dependent trafficking of channels to the membrane. However, the β2a- and β4-specific effects of UO126 on the voltage dependence of activation and the current kinetics, are more consistent with a direct modulation via MAPK, of α1B channels already at the membrane.

Based on the assumption that MAPK-dependent up-regulation of α1B channels, as observed here, does not involve gene expression-mediated effects or protein trafficking, the most likely influence of MAPK would be to cause phosphorylation of these channels. Whether MAPK modulates α1B channels directly, or via some other downstream signalling components however, cannot be determined on the basis of the data presented here. Several protein kinases are known to modulate VDCCs. Consensus sites for both protein kinase A (PKA) and protein kinase C (PKC) are found on VDCC α1 (including α1B) and β subunits (e.g. Nastainczyk et al. 1987; Puri et al. 1997; Zamponi et al. 1997). Also, full length α1B is a substrate for calcium/calmodulin-dependent kinase II (CaMKII; Hell et al. 1994). Although each of these kinases can regulate MAPK activity, there is little evidence to date that they can also act as downstream effectors of MAPK (Cobb, 1999; Sweatt, 2001). Furthermore, the kinases which are known to be activated via MAPK, e.g. S6 kinase and the MAPK interacting kinases 1 and 2 (Cohen, 1997), have no known role in ion channel regulation. Thus, although some involvement of downstream effectors cannot be ruled out, direct phosphorylation of α1B by MAPK itself is perhaps more likely. Indeed, several α1 and β subunits, including α1B and all four neuronal βs do possess putative MAPK consensus sites, as indicated either by optimal consensus sequences of Pro-x-Ser/Thr-Pro, where x is any amino acid, or, the minimum sequence requirement, Ser/Thr-Pro (Cobb, 1999). Although there is no evidence to date that direct MAPK phosphorylation of VDCCs occurs, in vitro phosphorylation of MAPK consensus sites on the A-type K⁺ channel, K_v4.2, has been shown (Adams et al. 2000). Thus, direct MAPK phosphorylation of VDCCs may be possible. The principal VDCC β-binding site is a conserved motif within the intracellular linker between transmembrane domains I and II (I-II loop) of α1 subunits (α1-interaction domain, AID) (Pragnell et al. 1994). A complimentary β-interaction domain, BID, is common to all β subunits (De Waard et al. 1994). Since α1B has MAPK consensus sites within the I-II loop, albeit not overlapping with the AID, and additional consensus sites close to the BID are common to all four βs, it is tempting to speculate that the AID-BID interaction is of critical importance in MAPK-dependent regulation of these channels. MAPK phosphorylation at these putative consensus sites on α1B and/or β subunits would be consistent with the fact that MAPK-dependent modulation of current is observed regardless of which β subunit is present but is lost in the absence of β.

In addition to the common effect of UO126 on current density, several β subunit-specific effects were also observed. Although UO126-induced current inhibition was seen with all βs, the effect was significantly reduced with β2a compared with any other β. Also specifically associated with β2a, a hyperpolarising shift in the current-voltage relationship was observed following inhibition of MAPK. Furthermore, marked changes in the current kinetics, perceived either as increased inactivation or decreased activation, were also seen in the presence of both β2a and β4. De-palmitoylation of β2a partially reverses the characteristic properties normally associated with wild type β2a. In the present study, co-expression of α1B with de-palmitoylated β2a(C3,4S), abolished the UO126-induced shift in V_50,act and increased the percentage of current inhibition to levels comparable with that seen in the presence of the other βs. The change in current kinetics however, was unaffected. The β 2a-specific effects of UO126 were thus shown to be dependent on β2a palmitoylation. Exactly how palmitoylation of β2a modifies MAPK-dependent modulation of α1B is unclear but palmitoylation is known to target certain proteins to the membrane, and to alter various protein-protein interactions (Resh, 1999). Indeed, β2a's slow inactivation is suggested to involve membrane tethering via its palmitoylated cysteines, of the α1 I-II loop, which normally acts as an inactivation particle (Cens et al. 1999; Hering et al. 2000; Restituito et al. 2000). In this model, mobility of the I-II loop which is essential for inactivation, is prevented, so allowing the channel to remain open (Restituito et al. 2000). Given the suggestion that MAPK might phosphorylate either the α1B I-II loop and/or the β subunits, then membrane anchoring of the I-II loop via β2a's palmitoylated N-terminus, may partially occlude phosphorylation sites which would otherwise be accessible in the presence of any other β. In such a scenario, the basal level of channel phosphorylation might then be reduced compared with that of other βs, resulting in less inhibition of current by UO126. This would be consistent with β2a(C3,4S)‘s alleviation of the reduced UO126-induced current inhibition seen with wild type β2a.

Palmitoylation of β2a is not usually associated with any effect on the voltage dependence of current activation. However, following application of UO126, a hyperpolarising shift in the current-voltage relationship was observed in β2a-expressing COS-7 cells. This effect was found to be dependent on N-terminal palmitoylation of β2a but without further investigation it is difficult to speculate on the precise mechanism involved. In comparison with other βs that lack palmitoylated residues, β2a binding to α1B may alter the channel conformation sufficiently to enable MAPK to influence voltage-dependent activation. Such a process may involve recruitment of MAPK and/or other kinases to the cell membrane via β2a's palmitoylated N-terminus. Physiologically, hyperpolarisation of the current-voltage relationship will lower the threshold of α1B channel activation to the extent that at negative membrane potentials, Ca²⁺ channel current will be enhanced when MAPK activity is reduced. In the presence of de-palmitoylated β2a(C3,4S), inhibition of current was observed throughout the voltage range tested. Thus, voltage-dependent enhancement of current appears to be independent of β2a current inhibition

In the presence of either β2a or β4, the inhibition of endogenous MAPK caused a marked change in the current kinetics, with the overall effect being to reduce Ba²⁺ influx through these channels at any given potential. Voltage-dependent inactivation of VDCCs is known to contribute to the control of action potential duration. Under physiological conditions, with Ca²⁺ as the charge carrier, such modulation of the current kinetics in the presence of either β2a or β4, may therefore influence the control of this process. Interestingly, in DRGs, Fitzgerald (2000) also reported a slight trend for increased inactivation of whole-cell current when Ras/MAPK was down-regulated, and vice versa. On the basis of the data presented here for β2a and β4, it was not possible to distinguish the effects of UO126 on activation from those on inactivation of these channels. Comparison of β2a and β4 currents recorded from cells pre-incubated either in the presence or absence of UO126, suggested that steady-state inactivation was largely unaffected by inhibition of MAPK. However, the slight hyperpolarisation of both the β2a and β4 inactivation curves obtained from UO126-treated cells vs. controls, suggests that at least some influence of MAPK on steady-state inactivation cannot be completely ruled out. A more detailed examination of α1B activation/ inactivation in the presence of UO126 is therefore required to determine the precise mechanism(s) responsible for these effects. In terms of possible molecular determinants that might influence the effects of MAPK on activation/inactivation, it is important to note that the current kinetics in both wild type β2a and de-palmitoylated β2a(C3,4S) were affected in the same way. Thus, N-terminal palmitoylation of β2a does not appear to be a requirement. Since the kinetics of both β2a and β4 currents were similarly affected by UO126 it is possible that both βs may share some common structural element which is of importance in regulating channel kinetics. Several sites on α1 and β subunits are known to influence VDCC inactivation kinetics (reviewed in Hering et al. 2000). For instance, regions within the I-II and II-III loops as well as the N- and C-termini of α1 subunits are particularly important. Furthermore, the N- and C-termini of β2a have also been reported to play a role in controlling the rate of inactivation (Tareilus et al. 1997; Qin et al. 1998). Intriguingly, as well as the conserved MAPK consensus sites common to all βs, β2a and β4 also have additional sites that map to their C- and N-termini, respectively, raising the possibility that interaction of these sites with α1B might influence the inactivation kinetics.

Possible biological implications

The results of this study have important implications for our understanding of VDCC regulation. The majority of reports implicating MAPK in up-regulation of VDCCs suggest that current enhancement results from gene expression-mediated synthesis of new channels (e.g. Lei et al. 1998; Baldelli et al. 2000). Consequently, MAPK-dependent modulation of α1B Ca²⁺ channels already at the plasma membrane, as implied by the findings of the present study, constitutes a novel mechanism to regulate neuronal Ca²⁺ influx.

Specific VDCC subtypes tend to be associated with specific cellular processes. Ca²⁺ influx through presynaptic N-type (α1B) Ca²⁺ channels is specifically linked with neurotransmitter release (Yu et al. 1992; Gruner & Silva, 1994). The results reported here suggest that depending on which β subunit is co-expressed with α1B, MAPK signalling could potentially regulate Ca²⁺ influx by altering channel conductance (all βs), by altering activation/inactivation kinetics (β2a, β4), or by shifting the voltage-dependent activation of channels (β2a). Furthermore, VDCC β subunits vary in their temporal expression and can be differentially localised across specific brain regions and within the same neurone (Vance et al. 1998; Wittemann et al. 2000). Consequently, β subunit-specific regulation of N-type channels via MAPK could potentially allow channels expressed during different stages of development and in different regions of the brain/cells, to be differentially modulated by the same neurotransmitter. In addition, several upstream activators of MAPK, including receptor tyrosine kinases, heterotrimeric G-proteins and Ca²⁺ influx via NMDA receptors and L-type VDCCs, have now been identified in neurones (Cobb et al. 1999; Sweatt, 2000). MAPK-dependent modulation of N-type channels within particular cell types is therefore likely to involve extremely complex signalling networks. Such intricacy in the potential mechanisms by which MAPK can regulate Ca²⁺ entry via these channels would therefore provide a novel system for fine-tuning of neurotransmitter release and hence synaptic transmission.

At the molecular level, the data presented here demonstrate that β binding to α1B is a critical requirement for MAPK-dependent modulation of N-type channels. It is suggested that the AID-BID interaction between the α1B I-II loop and β subunits is of particular importance in this process. Since putative MAPK phosphorylation sites are located within the α1B I-II loop and at sites close to the BID that are common to all neuronal βs, it is possible that α1B and/or β subunits may be directly phosphorylated by MAPK. Experiments are currently in progress to confirm whether these putative MAPK consensus sites are directly involved in α1B modulation. Importantly however, the possibility that additional downstream signalling components may also be involved, or that cross-talk between different signalling pathways might occur, cannot be ruled out. As previously mentioned, N-type VDCCs are substrates for PKA, PKC and CaMKII (Nastainczyk et al. 1987; Hell et al. 1994; Zamponi et al. 1997). Furthermore, these channels are also inhibited by direct interaction of G-protein βγ subunits with sites in the I-II loop and C-terminus of α1B that also bind Ca²⁺ channel β subunits (Qin et al. 1997; Tareilus et al. 1997; Zamponi et al. 1997). The presence of β subunit antagonises G-protein inhibition of α1B and there is some evidence that specific β subunits can differentially regulate certain aspects of G-protein modulation of these channels (e.g. Canti et al. 2000; Feng et al. 2001). The α1B I-II loop has also been identified as an integration centre for cross-talk between PKC and G-protein modulation (Hamid et al. 1999). Since β binding is apparently essential for modulation of α1B via MAPK, and interaction of β with the α1B I-II loop is likely to be of primary importance in this process, some overlapping influence of MAPK on the modulation of these channels by G-proteins and/or protein kinases may be possible. The answers to these, and other such questions relating to the potential cross-talk between MAPK and other signalling pathways that influence N-type Ca²⁺ channels require clarification. Nonetheless, the results presented here suggest that MAPK-dependent modulation of α1B Ca²⁺ channels could provide an important and novel mechanism for fine-tuning neurotransmitter release. In the wider context, given the diversity of Ca²⁺ channel α1 subunits so far identified and the possibility that many of these might also be regulated via MAPK signalling (Fitzgerald, 2000), numerous Ca²⁺-dependent processes within cells could be regulated via this common pathway.