Abstract
Ca2+ channel inactivation is a key element in controlling the level of Ca2+ entry through voltage-gated Ca2+ channels. Interaction between the pore-forming α1 subunit and the auxiliary β subunit is known to be a strong modulator of voltage-dependent inactivation. Here, we demonstrate that an N-terminal membrane anchoring site (MAS) of the β2a subunit strongly reduces α1A(CaV2.1) Ca2+ channel inactivation. This effect can be mimicked by the addition of a transmembrane segment to the N terminus of the β2a subunit. Inhibition of inactivation by β2a also requires a link between MAS and another important molecular determinant, the β interaction domain (BID). Our data suggest that mobility of the Ca2+channel I–II loop is necessary for channel inactivation. Interaction of this loop with other identified intracellular channel domains may constitute the basis of voltage-dependent inactivation. We thus propose a conceptually novel mechanism for slowing of inactivation by the β2a subunit, in which the immobilization of the channel inactivation gate occurs by means of MAS and BID.
- P/Q type Ca2+ channels
- CaV2.1
- β subunit
- inactivation mechanism
- palmitoylation
- membrane anchoring
- I–II loop
Ca2+channels β subunits are intracellular proteins associated in vivo with high voltage-activated Ca2+channel α1 subunits, which finely tune many of their electrophysiological and kinetic properties (Berrow et al., 1997). Ten different genes encode voltage-gated Ca2+ channel α1subunits (Randall and Benam, 1999). All of them, except the T-type Ca2+ channels (Randall and Benam, 1999), can interact with one of four different β subunits (β1–β4) (Birnbaumer et al., 1998; Walker and De Waard, 1998). Within the same type of pharmacologically defined Ca2+ channels, these β subunits represent a major determinant of variability in channel properties. β1, β3, and β4 subunits induce hyperpolarizing shifts in the activation and inactivation properties of these channels and accelerate their activation kinetics and voltage- and Ca2+-dependent inactivation (Varadi et al., 1991; Neely et al., 1993; Sather et al., 1993; Stea et al., 1993; Jones et al., 1998). The β2a subunit plays a different role because it slows down voltage- and Ca2+-dependent inactivation (Sather et al., 1993; Stea et al., 1994; Jones et al., 1998) and is unable to confer prepulse facilitation to the L-type Ca2+ channel (Cens et al., 1996). Sequence homology analysis of these different β subunits reveals the existence of two well conserved domains in their primary sequence (C1, C2), surrounded by more variable regions in which alternative splicing occurs (V1–V3) (Fig. 1) (Perez Reyes and Schneider, 1994). The conserved site of interaction between the α1 and the β subunits has been mapped to the beginning of the C2 domain of β (De Waard et al., 1994; Walker and De Waard, 1998). This site interacts with a consensus β subunit-binding sequence [α interaction domain (AID)] localized on the loop connecting domain I and II of the α1 subunit (Pragnell et al., 1994; De Waard et al., 1995). The specificity of the effects of the β2a subunit is mediated, as least in part, by two cysteines located at the N terminal end of some isoforms of the β2a subunits (Fig. 1) (Chien et al., 1996; Qin et al., 1998). Similar to αs GTP-binding proteins, these cysteines are post-translationally modified through the addition of thioester-bound palmitic acids and thus allow a membrane association of the β2a subunit (Chien et al., 1998) and regulation of the inactivation of the α1E subunit (Qin et al., 1998). However, the partial data available have not yet allowed the elucidation of the molecular mechanisms by which these palmitoylated cysteines slow inactivation.
A, Schematic representation of the Ca2+ channel β subunit. V1,V2, and V3 represent regions of variable sequences among the different β subunits. C1 andC2 are regions of high homology. Drawing has been scaled according to β2a sequence, and the boxrepresents the V1 sequence. Amino acid alignment of the N-terminal tail of the β1 and β2 subunits. Note the presence of the two Cys residues in the rat β2subunit. B, Left, Rapidly and slowly inactivating Ba2+ currents recorded from oocytes expressing the α1A plus α2-δ calcium channel subunits with, respectively, the β1 or the β2 subunits. Inactivation was quantified by the percent of inactivation measured at the end of a 2.5 sec test pulse to +10 mV.Right, Confocal immunofluorescent images of β subunit-transfected tsA 201 cells were obtained after fixation and immunohistochemical staining using a β-com primary antibody. Note the membrane localization of the β2a subunit while the β1 subunit is localized to the cytoplasm.
In this work, we have expressed several mutated forms of the β1 and β2 subunits inXenopus oocytes and tsA 201 cells and analyzed in parallel (1) their effects on the inactivation of P/Q type Ca2+ channels, and (2) their subcellular localization when expressed alone. Our results strongly suggest that the β2a subunit acts as an anchor for the Ca2+ channel I–II loop proposed to be an inactivation particle, thus reducing inactivation. We show that several intracellular domains of the α1A(CaV2.1) channel can interact with the I–II loop and are thus potential receptor sites for the inactivation particle.
MATERIALS AND METHODS
Preparation of mutated β subunits.The following calcium channel subunits were used: α1A (Starr et al., 1991), β1b (Pragnell et al., 1991), β2a (Perez Reyes et al., 1992), and α2-δ. All of these cDNAs were inserted into the pMT2 expression vector (Stea et al., 1994). Chimeras were produced by a classical two-step PCR approach (Cens et al., 1998). βTF1 and β chimera were finally digested using EcoRI andXbaI and subcloned into pBluescript (Stratagene, La Jolla, CA) before sequencing (DiDeoxy Terminator technology; Applied Biosystems, Foster City, CA). Constructs were subsequently subcloned into pMT2 for injection and expression.
Point mutants were obtained by PCR following commercial mutagenesis kit instructions (Quick Change site-directed mutagenesis kit; Stratagene) and using the following sense and antisense primers: β2C3,4S AS, ATG TAC CAG CCC GGA GGA CTG CAT GAA GAG GTG G; β2 C3S AS, ATG TAC CAG CCC GCA GGA CTG CAT GAA GAG GTG G; β2 C4S AS, ATG TAC CAG CCC GGA GCA CTG CAT GAA GAG GTG G; β2R9–11A AS, GGA CAC CCG TAC TGC CGC GGC ATG TAC CAG CCC G; β2R10A AS, CCG TAC TGC CGC GCG ATG TAC; and β2R13A AS, CCA TAG GAC ACC GCT ACT CGC CGG.
For the chimera CD8-β2aC3,4S construction, the β2a was amplified from the pMT2 vector by PCR, using a forward primer containing the double cysteine mutation. The following primers were used: forward, 5′-CGCGGATCCCAGTCCTCCGGGCTGGTACATCGCCGGCGAGTACGG-3′, and reverse, 5′-ACGTGAATTCTTGGCGGATGTATACATCCCTGTTCCACTCGCCGAC-3′, containing BamHI or EcoRI restriction sites, respectively.
The PCR product was purified and subcloned in frame into theBamHI and EcoRI sites of the pcDNA3-CD8-βARK-Myc vector after removing the βARK insert. This vector was generously provided by Dr. J. Lang (Geneva University, Geneva, Switzerland).
In vitro translation and binding of glutathioneS-transferase fusion proteins.35S-labeled α1AI–II loop was synthesized by coupled in vitro transcription and translation (TNT; Promega, Madison, WI). Purified glutathioneS-transferase (GST) fusion proteins (250 nm each) were immobilized to glutathione agarose beads (Sigma-Aldrich, Saint Quentin Fallavier, France) by 30 min incubation in TBS (25 mm Tris and 150 mm NaCl, pH 7.4) and 0.1% Triton X-100. Binding was initiated by addition of the35S-labeled α1AI–II loop (2 μl), and this final mixture was incubated overnight at 4°C. Beads were washed four times with binding buffer, and associated35S-labeled α1A I–II loop was analyzed by SDS-PAGE and autoradiography.
Cell transfection and immunofluorescence. tsA 201 cells were maintained in DMEM (Life Technologies, Rockville, MD) containing 10% fetal bovine serum and 1% penicillin–streptomycin at 37°C in 5% CO2. Transfections were performed using Superfect according to the protocols of Qiagen (Hilden, Germany), 1 d after plating the cells on poly-l-ornithine-treated 35 mm Petri dishes. Plasmid cDNA(s) (5 μg) was used for each transfection with an incubation time of 2 hr. Forty-eight hours later, cells were fixed and permeabilized using PBS supplemented with 4% paraformaldehyde and 0.05% Triton X-100 (20 and 10 min, respectively). After an incubation of 1 hr in 3% PBS plus BSA, cells were incubated an additional 1 hr with the primary polyclonal antibody β-com (Pichler et al., 1997), washed three times in PBS, and incubated 1 hr with the secondary anti-rabbit goat antibody conjugated to CY-3 (Sigma-Aldrich). After three other washes, cells were mounted and viewed on a conventional or a confocal immunofluorescent microscope. Confocal microscopy was performed at the CRIC (Center Régional d'Imagerie Cellulaire) facilities.
Xenopus oocyte preparation and injection. Xenopusoocyte preparation and injection (5–10 nl of α1, α1 plus β, or α1 plus α2δ plus β cDNAs at ∼ 0.3 ng/nl) were performed as described previously (Cens et al., 1996). Oocytes were then incubated for 2–7 d at 19°C under gentle agitation before recording.
Electrophysiological recordings. Whole-cell Ba2+ currents were recorded under two-electrode voltage clamp using a GeneClamp 500 amplifier (Axon Instruments, Burlingame, CA). Current and voltage electrodes (<1 MΩ) were filled with CsCl 2.8 m and BAPTA 10 mm, pH 7.2 with CsOH. Ba2+ current recordings were performed after injection of BAPTA [∼50 nl of (in mm): 100 BAPTA-free acid (Sigma-Aldrich), 10 CsOH, and 10 HEPES, pH 7.2 with CsOH using one or two 40–70 msec injection at 1 bar] in the following bath solution (in mm): 10 BaOH, 20 TEAOH, 50 NMDG, 2 CsOH, and 10 HEPES, pH 7.2 with methanesulfonic acid. Ba2+ current amplitudes were usually in the range of 1–5 μA. Currents were filtered and digitized using a DMA-Tecmar Labmaster (Tecmar Inc., Longmont, CO) and subsequently stored on a IPC 486 personal computer using version 6.02 of the pClamp software (Axon Instruments). Ba2+ currents were recorded during a typical 2.5 sec duration test pulse from −80 to +10 mV. Current amplitudes were measured at the peak of the current (I1) and at the end of the pulse (I2). The percentage of inactivation was calculated as the ratio (I1 −I2)/I1. Pseudo steady-state inactivation (2.5 sec of conditioning depolarization followed by a 400 msec test pulse to +10 mV) was fitted using the following equation:
where I is the current amplitude measured during the test pulse at +10 mV for conditioning depolarizations varying from −80 to +50 mV, Imax is the current amplitude measured during the test pulse for a conditioning depolarization of −80 mV, R is the proportion of non-inactivating current, V is the conditioning depolarization, V0.5 is the half-inactivation potential, and k is a slope factor. Similar inactivation curves were also performed using 7.5 sec conditioning depolarizations without significant differences in the calculated V0.5. All values are presented as mean ± SD of n determinations. A Student's t test was used at p = 0.05 to determine the significance of the difference between the two means.
RESULTS
Coexpression of the α1A and α2-δ subunits with the neuronal β1b in Xenopus oocytes resulted in a Ba2+ current that inactivated by >90% after 2.5 sec (93 ± 2%) (Fig.1B). A similar experiment performed with the β2a subunit produced Ba2+ currents with very slow inactivation kinetics leading to only 40 ± 9% inactivation at the end of the pulse (Fig. 1). Because expression of the α1A subunit alone gave currents with fast inactivation (Mangoni et al., 1997), we interpreted this slowing in the presence of the β2a subunit as a blocking of the normal inactivation mechanism. Moreover, the β1b subunit that provided rapidly inactivating currents (when expressed with the α1A subunit) was localized throughout the cytoplasm (when expressed alone), whereas the β2a subunit that produced slow currents was localized close to the membrane (Fig. 1B). We and others have shown that the V1 domain of the β2asubunit (in which palmitoylation occurs) (Fig. 1) is the main determinant of the β2a subunit-induced regulation of voltage- and Ca2+-dependent inactivation, as well as of the inhibition of facilitation of the α1C Ca2+ channel (Olcese et al., 1994; Cens et al., 1998, 1999a,b; Qin et al., 1998). However, although palmitoylation of the β2asubunit has been shown recently to be involved in both membrane association of the subunit and slowing of the inactivation kinetics, the causal relation between these phenomena has not been studied.
Cys3 and Cys4 of β2 are major determinants for membrane localization and slow inactivation.A, Localization and nature of the different mutations made in the V1β2 domain.β2 refers to the wild-type sequence of the rat β2a subunit. B, Mutation of Cys(3,4) of the β2a subunit to Ser (labeledC3,4S in B) induced a marked increase in inactivation (n = 21) and a localization of the β2a subunit to the cytoplasm (see E). Mutation of Arg(9,10,11) to Ala (labeled R9–11A in Fig.3B; n = 13) has a similar, albeit reduced in amplitude, effect: acceleration of the inactivation kinetics and cytoplasmic localization of the mutated β2a subunit (see E). Traces labeledβ2 show current kinetics recorded with the wild-type β2a subunit (similar totrace labeled β2in Fig. 1B) for comparison. * indicates statistically different from β2. C, Mutation of only one Cys to Ser (either Cys3, n = 14; or Cys4, n = 7; labeled C3S andC4S, respectively) is sufficient to produce an acceleration of current kinetics and a delocalization of the mutated β2a subunit to the cytoplasm (see E). * indicates statistically different from β2.D, However, mutation of only one Arg (either Arg(10)Ala,n = 19; or Arg(13)Ala, n = 9; labeled R10A and R13A, respectively) has no effect on current kinetics (left) and membrane localization (right) of the β subunit (E). E, Confocal immunofluorescent images of tsA 201 cells transfected with the different mutated β2a subunits.
Simultaneous mutations of Cys3 and Cys4 to Ser residues (designated β2C3,4S) (Fig. 2A) has been shown to prevent β2a subunit palmitoylation inXenopus oocytes and to accelerate α1E Ca2+ channel inactivation (Chien et al., 1998; Qin et al., 1998). When expressed in oocytes with the α1A subunit, we also observed this increase in inactivation (87 ± 13% compared with 40% for β2a) (Fig. 2B). Figure2E shows that expression of this mutated subunit was diffuse in the cytoplasm. Interestingly, expression of single Cys mutations of the β2a subunit, either Cys3 (β2C3S) or Cys4 (β2C4S), produced a similar effect on both the Ca2+ channel inactivation (85 ± 7 and 87 ± 17%, respectively) and the intracellular localization of the β2a subunit (Fig. 2E), suggesting that both Cys residues need to be present for a correct palmitoylation of the β2a subunit. Mutation of positively charged residues Arg10 and Arg13 of the β2a subunit (β2R10A and β2R13A, respectively) (Fig.2D) had no effect on channel inactivation (30 ± 11 and 47 ± 20%) or on β subunit localization, which remained, for both mutations, membrane localized (Fig. 2E). However, when the three Arg residues at positions 9, 10, and 11 of the β2a subunit were replaced by Ala residues (mutant β2R9–11A) (Fig. 2), the membrane association of the β2a subunit was lost (Fig.2B,E). Recordings of Ca2+ currents in the presence of this mutant showed that the inactivation kinetics of the α1A Ca2+ channel were significantly faster than with the β2asubunit (51 ± 22%). Therefore, additional sites other than the two Cys are important for the proper membrane association and slowing of inactivation induced by the β2a subunit. If we postulate that membrane association of the β2a subunit, when expressed alone, was attributable to palmitoylation of Cys3 and Cys4, mutation R9–11A could affect either the formation and the stability of the thioester bond or the association with the plasma membrane through modification of electrostatic interactions as already shown in the case of Src (Murray et al., 1998). Indeed, the double mutant (β2C3,4S-R9–11A), which lacks both the two Cys residues and the three Arg residues (Fig. 2) and therefore cannot be palmitoylated, displayed current inactivation kinetics and subcellular localization identical to the βC3,4S mutant (94 ± 2% and cytoplasm localization; data not shown).
Altogether, these data suggest that multiple sites can affect the palmitoylation level and the cellular localization of the β2a subunit. In all cases, however, the membrane localization of the β subunit was always associated with a marked slowing of the inactivation kinetics, suggesting that this feature represents a key element for slow inactivation. We tested this idea by forcing the membrane localization of the subunit by addition of a transmembrane segment at the N terminal tail of the double Cys-mutated β subunit. The resulting chimera (CD8-β2C3,4S) (Fig.3A) possessed the transmembrane sequence of the CD8 protein in the N-terminal position. We expressed this chimeric protein in Xenopus oocytes along with the α1A and α2δ subunits for current recordings or alone in tsA 201 cells to study its cellular localization. As shown in Figure 3B, addition of the CD8 sequence produced a very marked slowing of the inactivation kinetics (33 ± 14% compared with 87 ± 12% to the β2C3,4S mutant). In other terms, the forced membrane localization of the protein restored a key regulation that is normally seen with the wild-type β2a subunit but not with the double Cys mutated form of the protein. This characteristic slowing in inactivation induced by the CD8-β2C3,4S chimera was quite expectedly correlated with a membrane association of the protein as detected by immunofluorescence staining using either an anti-β (Fig.3C) or an anti-CD8 antibody (data not shown). Similar results were found with the CH4 subunit, a chimeric β1 subunit with the first V1 domain replaced by the homologous domain of the β2 subunit. The slow current inactivation (Olcese et al., 1994) and membrane localization induced by CH4 (Fig. 3B) is only attributable to the V1 domain of β2, because a β1 subunit from which this domain has been deleted (β1TF1) kept its cytoplasmic localization and still induced fast inactivation (88 ± 4% of inactivation; data not shown). Thus, mutation of Cys3 and Cys4 in CH4 accelerates current inactivation and localizes the subunit to the cytoplasm, whereas addition of the CD8 segment restores both slow inactivation and membrane localization (Fig.3B,C, CH4-C3,4S,CD8- CH4-C3,4S). The latter results suggest that the β2a subunit needs a membrane anchor to reduce channel inactivation.
Addition of a membrane-spanning sequence at the N-terminal end of β2C3,4S slows inactivation. A, Schematic representation of the CD8 constructions. The ectomembrane and transmembrane domains of the CD8 receptor were fused to the N-terminal end of the β2C3,4S subunit, giving the chimeric CD8-β2C3,4S subunit. Similarly CD8-CH4-C3,4S was constructed using the mutated Cys3,4S CH4 chimera (in which the V1 domain of β1 was replaced by V1β2).B, Ba2+ currents recorded from oocytes expressing the α1A and α2-δ subunits with the β2C3,4S, the CD8-β2C3,4S, the CH4, the CH4-C3,4S, or the CD8-CH4-C3,4S subunits. Mutation Cys(3,4)Ser in β2 induced a rapidly inactivating currents (labeled C3,4S; n = 21), correlated with a cytoplasmic localization, as seen in Figure 3. Addition of the CD8 receptor transmembrane segment (CD8-β2C3,4S;n = 8) restored the slow inactivation typically seen with the wild-type β2a subunit. * indicates statistically different from β2C3,4S. Similar effects were found with the slowly-inactivating CH4 chimera (see right panel). C, Confocal images of a middle plane of tsA 201 cells expressing the various chimera and immunostained with an anti-β antibody. After expression in cells, the CD8-β2C3,4S chimera was also targeted to the membrane as the slowly inactivating wild-type β2a, CH4, and CD8-CH4-C3,4S subunits. The CH4-C3,4S subunit was localized to the cytoplasm.
Another important feature of the α1A channel properties recorded in the presence of the β2asubunit is the shift in inactivation toward more depolarized values. Analysis of the inactivation properties of channels expressed with a β2a subunit evidenced a depolarizing shift of ∼10–20 mV in the half-inactivation potential (E0.5) compared with channels containing a β1 subunit. This difference was also found in our experimental conditions (BAPTA-injected oocytes) between oocytes expressing an α1A plus α2-δ plus β1b or α1A plus α2-δ plus β2a subunit combination (E0.5 of −32 ± 5 and −13 ± 6 mV for β1b and β2a, respectively) (Fig.4) . β subunit constructs that induced slowly inactivating currents also shifted theE0.5 toward positive values (more than −20 mV) (Fig. 4, β2, βCH4,R10A, R13A,CD8-β2C3,4S,CD8-CH4-C3,4S). Conversely, β subunits producing fast inactivation (β1b, β2C3,4S, β 2C3S, and CH4-C3,4S) generated more hyperpolarized steady-state inactivation curves (E0.5 of less than −20 mV). Interestingly, expression of the β2R9–11A subunit, which induced currents characterized by a moderate percentage of inactivation, had an E0.5 value intermediate between those observed for the β1band the β2a subunits. The β2C4S mutant was the only exception to this set of observations. The localization of this subunit was cytoplasmic and it induced rapidly inactivating currents (Fig. 3), but, contrary to expectations, the voltage dependence of inactivation was depolarized, similar to the β2a subunit (−16 ± 1 mV). However, we conclude that, overall, a strong correlation appears to exist between the ability of these subunits to induce currents with slow kinetics and depolarized inactivation and their membrane localization.
Membrane association of the β subunit shifts the steady-state inactivation relationship. Left, Typical steady-state inactivation curves recorded from oocytes expressing α1A plus α2-δ plus β1(labeled β1) or α1A plus α2-δ plus β2(labeled β2) subunits. Half-inactivation potentials were −32 and −12 mV for β1b and β2a, respectively.Right, Half-inactivation potentials obtained from oocytes expressing the α1A plus α2-δ subunits with different mutated or chimeric β subunits (n = 5–20). β subunits that induced slowly, β2a-like, inactivating currents induced depolarized half-inactivation potentials. Shadowed boxes underline the values and SDs of the half-inactivation potentials for β1b and β2a subunits.
We then analyzed the effects of the N terminal tail of β2a directly on Ca2+ channel inactivation. This was done by injecting, 4 hr before recording, the V1β2peptide (corresponding to the first 16 amino acids of the β2a subunit) (Fig.5) into oocytes expressing the α1A, α2-δ, and β1TF1 subunit combination. The final concentration of the peptide was estimated to be ∼0.1 mm. Ba2+ currents recorded from both control noninjected and V1β2-injected oocytes inactivated rapidly, as seen on the superimposed traces and histograms in Figure 5 (94 ± 2 and 88 ± 4% for injected and control oocytes, respectively). Injection of the peptide therefore did not modify the inactivation kinetics of the currents, although the combined presence of β1TF1 and the V1β2 peptide mimicked the βCH4 chimera, which produced slowly inactivating currents (Fig. 3). These results suggest that the presence of the N-terminal tail of β2in the cell is not sufficient, by itself, to induce slowing of inactivation. A physical link between the membrane-anchored N-terminal tail of β2 (MAS) and the rest of the β subunit appears to be required to observe a slowing in inactivation. This link could be expected to restrict the mobility of the α1 Ca channel I–II loop (AI–AII) (Fig.6), which interacts directly with the β subunit. In these conditions, formation of the inactivated state of the channel may involve the binding of the I–II loop to a receptor site located on the intracellular side of the channel. We therefore searched for such potential interactions between the I–II loop and other intracellular domains of the channel. Different N- and C-terminal segments, as well as the II–III and the III–IV intracellular loops of the α1A subunit, were constructed as GST fusion, produced, and purified (Fig. 6B, Coomassie blue-stained SDS-PAGE analysis of the GST fusion protein used). The different GST fusion proteins were then immobilized on glutathione agarose beads, incubated with in vitro-translated35S-labeled α1AI–II loop, washed, and analyzed by SDS-PAGE and autoradiography. As seen in Figure 6C, the α1A I–II loop interacts with several intracellular domains, including the loop connecting domain III to IV (III–IV loop) and the N-terminal (NT6) and C-terminal (CT1, CT6) tails of the α1 subunit. These sequences may thus be involved in the formation of the receptor site of the inactivation particle.
Perfusion of V1β2a peptide does not accelerate inactivation. We injected the V1β2a peptide (50 mm in H20; final intra oocyte concentration of ∼0.1 mm; see Fig. 2A for sequence) into oocytes expressing the α1A and α2-δ subunit with the β1b subunit truncated in this V1 domain (β1TF1). The combination (V1β2a plus β1TF1) corresponds to the two parts of chimera β-CH4, which induced slowly inactivating current when expressed with α1A subunit. Currents were recorded during a typical test pulse to +10 mV in V1β2a-injected (n = 6) and noninjected oocytes (n = 12). No statistical differences were seen in the inactivation kinetics between these two batches of oocytes, indicating that the V1β2a peptide had no direct effect on channel inactivation.
The α1A Ca channel I–II loop interacts with multiple intracellular domains of the channel. Schematic localization (A) and Coomassie blue-stained SDS-PAGE (B) of the purified GST proteins fused to N-terminal (NT1, NT4,NT5, NT6), C-terminal (CT1, CT3, CT6) or intracellular loop (AII-III, AIII-IV,AID) sequences of the α1A subunit.C, Specific association of 35S-labeled I–II loop with N- and C-terminal sequences and III–IV loop of the channel.I, Input (2 μl of in vitro translated35S-labeled I–II loop); GST, control GST.
DISCUSSION
Ca2+ channel inactivation influences not only cellular excitability but also different Ca-dependent pathways leading to contraction, secretion, synaptic activation, or gene transcription. A large number of mutational studies have pointed out the important role played by multiple elements of the α1 pore-forming subunit on channel inactivation (Zhang et al., 1994; de Leon et al., 1995; Klockner et al., 1995;Parent et al., 1995; Adams and Tanabe, 1997; Herlitze et al., 1997;Sokolov et al., 1999; Spaetgens and Zamponi, 1999). It has been shown, for example, that the IS6, I–II loop, and C-terminal part of the α1 subunit are critical determinants for inactivation kinetics (Zhang et al., 1994; Herlitze et al., 1997;Bourinet et al., 1999; Cens et al., 1999a; Spaetgens and Zamponi, 1999). Overall, these studies strongly suggest that several interactions occur between multiple structural motifs of the channel during the transition between the open and the inactivated states of the channel. Additionally, the β subunits have also been shown to be directly involved in the modulation of channel inactivation and are therefore likely to represent key components of the molecular mechanism that leads to channel inactivation.
Our first finding that palmitoylation of Cys3 and Cys4 of the β2a subunit is responsible for the β2a-induced slowing of inactivation of the P/Q type Ca2+ channel extends previous reports of the effects of β2a palmitoylation on α1E and α1CCa2+ channels (Chien et al., 1996; Qin et al., 1998). The slowing of α1ACa2+ channel inactivation was correlated to a membrane localization of the β2a subunit when expressed alone in tsA 201 cells. Both effects were suppressed by mutation of a single Cys residue (3 or 4), showing for the first time that the two Cys residues are necessary for slowing inactivation and producing membrane localization. Similarly, whereas mutation of single positively charged residues (Arg to Ala) at positions 9 and 13 was without effect on channel inactivation (Fig. 2), similar mutations of the three contiguous Arg at positions 9–11 strongly reduced channel inactivation and membrane localization. This effect may arise from structural modifications in the N-terminal tail of β2a preventing subunit palmitoylation. Alternatively, electrostatic interactions between the acidic phospholipids and the cluster of basic residues could provide the energy necessary for membrane association of the subunit, as already shown for Src (Murray et al., 1998). In this case, hydrophobic and electrostatic interactions may act in synergy to target the β2a subunit to the membrane. Further experimental analysis of the level of palmitoylation of the β2R9–11A mutant will help to clarify the mechanism of targeting of the β2 subunit. However, it should be noted that the inactivation kinetics recorded with the β2R9–11A subunit, although being faster than those recorded with the wild-type β2a subunit, were still slower than with the nonpalmitoylated, cytoplasmically localized β2C3,4S subunit, suggesting that functional differences still exist between these two subunits. Our results define a minimal sequence (or MAS) for β2a subunit targeting, in which Cys3, Cys4, and adjacent Arg (10 and 12) are key residues: M-CC—–R-R–.
Another important clue to understand the role of the β2a subunit in the modulation of inactivation is provided by the βCH4 chimera. In the βCH4 chimera, addition of the first 16 amino acids of β2a, in place of the V1 domain of β1b (Fig.1A), induced palmitoylation (Chien et al., 1998), membrane localization of the chimera, and slowing of inactivation in coexpression experiments. This β2a-V1 sequence can therefore act as a MAS independently of any other β2a-specific sequences. However, effects of deletion of this MAS domain in the β2a subunit, which prevents palmitoylation, induces relocalization of the subunit to the cytoplasm, and acceleration of inactivation, are not compensated by injection of the V1 domain of β2a (Fig. 5). The latter results suggest that V1/MAS cannot act on its own to modulate channel inactivation but rather needs to be linked to a structural element common to β1 (see results with βCH4) and β2a. One obvious candidate is the β interaction domain (BID) sequence (De Waard et al., 1994) located on C2, conserved in all β subunits, and directly responsible for the interaction between α1 and β subunits. In our view, palmitoylation of Cys3,4 and binding to BID would act in concert to attenuate channel inactivation. The slowing in inactivation kinetics observed with the CD8-βC3,4S and CD8-CH4C3,4S chimera suggests in fact that the role of palmitic acid is solely to provide an MAS to the β subunit because addition of a transmembrane segment appears equally effective. The direct involvement of the I–II loop of the α1A subunit in channel inactivation (Herlitze et al., 1997; Bourinet et al., 1999; Spaetgens and Zamponi, 1999) leads us to propose that this sequence may behave as an inactivation particle, directly responsible for channel occlusion after opening. We suggest that, owing to its two functional domains, the β2a subunit acts as a molecular groom for the channel inactivation gate and immobilizes this inactivation particle by linking it to the membrane (Fig. 7). The model is compatible with the effect of the CD8-β2C3,4S chimera, the effects of mutation on the I–II loop, and the acceleration of α1Achannel inactivation recorded during overexpression of this loop (Cens et al., 1999a). Such an immobilization of the inactivation particle in the open configuration should increase the free energy necessary to reach the inactivated state. It is thus expected to shift the steady-state inactivation curve toward depolarized potential values. This effect is indeed recorded with the β2asubunit [half-inactivation potential (E0.5), 15 mV more depolarized than with β1] but also with its functional analog the CD8-β2C3,4S or the βCH4 chimera. Conversely, release of the inactivation particle produces a shift of voltage-dependent inactivation toward more hyperpolarized potentials (see mutations β2C3,4S and β2C3S). The depolarized inactivation potential recorded with the rapidly inactivating β2C4S subunit cannot be explained in this context. This particular mutation may prevent inactivation occurring from the closed state of the channel as opposed to other rapidly inactivating channels, without modifying the stability of the inactivated state. Similar modifications have already been reported for mutated Na+channels (Hartmann et al., 1994). The complete understanding of the effect of this mutation awaits recordings at the single-channel level and the knowledge of the palmitoylation state of all these mutant subunits.
Proposed mechanism for β2a-induced slowing of inactivation. In our scheme, the β2a subunit works as a rigid link between the membrane (via a palmitic acid anchor) and the inactivating particle (via its BID domain). When this link is broken (e.g., in mutation β2C3,4S, in a physiological situation producing a depalmitoylation, or in the case of a nonpalmitoylated β subunit), the inactivating particle (I–II loop) can move freely and produce the typical fast inactivation.
As expected, none of these mutations affected the interaction between the α1A and the β subunit, because all are membrane localized when expressed with the α1Asubunit (data not shown). The GST pull-down experiment, shown in Figure6, suggests multiple possible receptor sites for this inactivation particle. N-terminal and C-terminal sequences have already been shown to be involved in voltage- or calcium-dependent channel inactivation or regulation by β subunits. Interestingly, interaction of the I–II loop appears to be stronger with the III–IV loop, the most conserved connecting loop among high voltage-activated Ca channels (α1A, α1B, α1C, and α1E; >70% of similarity). These channels all display a slowing of inactivation by the β2a subunit. This loop is therefore a prime candidate for future studies on the inactivation mechanism, although other intracellular or intrapore domains may also directly participate in the process of inactivation.
The sensitivity of the thioester bond linking the palmitic acid to the β2a subunits makes this site a potentially important pathway for the regulation of Ca2+ entry into cells. Reducing agents or membrane receptor activation are known to modify the palmitoylated state of some signaling proteins (Gα-protein and β-adrenergic receptor) (Casey, 1995; Peterson and Catterall, 1995). Recently, nitrosylation by nitric oxide of thiol groups on cysteine residues of the β-adrenergic receptor has been reported (Adam et al., 1999). Whether a similar pathway can affect palmitoylation of the β2a subunit in vivo is not known but could potentially be important in physiological or pathological situations during ischemia or oxidative stress, for example. This phenomena would be restricted to β2a-containing channels and would decrease calcium entry by promoting channel inactivation. In this respect, our results open new perspectives for further studies of the control of calcium channel inactivation by β2a subunits, in particular in their native cellular environment.
Footnotes
This work was supported by Centre National de la Recherche Scientifique, Institut National de la Santé et de la Recherche Médicale, the Association Française contre les Myopathies, the Association pour la Recherche contre le Cancer, the Fondation pour la Recherche Médicale, Ligue Nationale contre le Cancer, and the Groupe de Reflexion sur la Recherche Cardiovasulaire. We thank Drs. T. Snutch and E. Perez-Reyes for kindly providing calcium channel cDNAs, Dr. J. Streissnig for β-com antibody, Drs. J. Mery and A. Chavanieux for peptide synthesis, and Drs. N. Morin, I. Lefèvre, A. Gouin-Charnet, J.-C. Labbé, and N. Lautredau (Centre for Research on Innovation and Competition) for invaluable technical help during the course of these experiments.
Correspondence should be addressed to Dr. Pierre Charnet, Centre de Recherches de Biochimie Macromoléculaire, Centre National de la Recherche Scientifique, Unité Propre de Recherche 1086, Institut Fédératif de Recherche 24, 1919 Route de Mende, 34293 Montpellier Cedex 05, France. E-mail: charnet{at}crbm.cnrs-mop.fr.
