Functional roles of cytoplasmic loops and pore lining transmembrane helices in the voltage-dependent inactivation of HVA calcium channels

Abstract

Voltage-dependent inactivation of calcium channels is a key mechanism for regulating intracellular calcium levels and neuronal excitability. In sodium and potassium channels, the molecular determinants that govern fast inactivation involve pore block by a cytoplasmic gating particle. As we discuss here, there is an increasing body of evidence that is consistent with a qualitatively similar inactivation mechanism in high-voltage-activated calcium channels. Work from a number of laboratories has implicated both cytoplasmic regions and the pore-lining S6 transmembrane helices in the inactivation process. Together with our recent findings, this leads us to propose a model in which the intracellular domain I–II linker region acts as a ‘hinged lid’ that physically occludes the pore by docking to the cytoplasmic ends of the S6 segments. We further propose that the ancillary calcium channel β subunits differentially modulate inactivation kinetics by binding to and thereby regulating the mobility of the putative inactivation gate. Indeed, additional evidence suggests that the carboxy terminus, amino terminus and domain III–IV linker regions of the channel modulate inactivation rates through interactions with the I–II linker per se, or indirectly via the ancillary β subunits. Taken together, the fast voltage-dependent inactivation of calcium channels appears reminiscent of that of sodium channels, but appears to show a more complex regulation through intramolecular interactions between the putative inactivation gate and other cytoplasmic regions.

Calcium influx through voltage-gated calcium channels triggers a number of intracellular responses, including contraction of skeletal (Adams et al. 1990), cardiac (Tanabe et al. 1990) and smooth muscle (Klockner & Isenberg, 1985), calcium-dependent gene transcription (Dolmetsch et al. 2001), the activation of calcium-dependent enzymes (Borodinsky et al. 2003; Lilienbaum & Israel, 2003), the release of neurotransmitters from presynaptic nerve termini (Wheeler et al. 1994; Atwood & Karunanithi, 2002) and the secretion of hormones (Kurjak et al. 2002). A number of different types of calcium channel have been identified in native cells; these have been classified into low-voltage-activated (LVA) and high-voltage-activated (HVA) channels (Nowycky et al. 1985). The LVA channels (or T-type calcium channels) activate at hyperpolarized potentials and typically display a transient current waveform. HVA channels require larger membrane depolarizations to open, and have been classified further, based on their pharmacological and biophysical characteristics, into L-, P-, Q-, R- and N-type channels (Fox et al. 1987; Llinas et al. 1989; Zhang et al. 1993). l-type calcium channels support long-lasting currents that are blocked by dihydropyridines, N-type channels are selectively inhibited by ω-conotoxin GVIA, and P- and Q-type channels differ in their current kinetics, but are both inhibited by the American funnel web spider toxin ω-agatoxin IVA, albeit with different affinities. R-type channels were originally defined based on their insensitivity to these blockers.

Molecular cloning has resulted in the identification of the molecular identities of native calcium current subtypes (for review, see Stea et al. 1995). Based on these studies, we now know that HVA calcium channels are heteromultimers comprised of a pore-forming α₁ subunit, plus ancillary β, α₂-δ and γ subunits, whereas LVA calcium channels appear to contain only the α₁ subunit (for review, see Catterall, 2000). To date, 10 different types of calcium channel α₁ subunits have been identified and shown to represent the major types of native calcium channels. Ca_v1.1, Ca_v1.2, Ca_v1.3 and Ca_v1.4 calcium channel α₁ subunits encode different types of l-type channel (Tanabe et al. 1987; Mikami et al. 1989; Williams et al. 1992; Koschak et al. 2003), different splice isoforms of Ca_v2.1 channels encode P- and Q-type channels (Bourinet et al. 1999), Ca_v2.2 and Ca_v2.3 encode N-type and R-type channels (Dubel et al. 1992; Niidome et al. 1992), respectively, and Ca_v3.1, Ca_v3.2 and Ca_v3.3 channels are members of the family of LVA (or T-type) calcium channels (Cribbs et al. 1998; Perez-Reyes et al. 1998; Lee et al. 1999a; McRory et al. 2001). The α₁ subunit is comprised of four homologous repeats, each containing six transmembrane-spanning helices plus a re-entrant p-loop motif (see Fig. 1; Catterall, 2000). The four domains are linked through large cytoplasmic loops that are capable of interacting with a number of regulatory proteins, including the calcium channel β subunit. Indeed, although the α₁ subunits are capable of forming functional channels, their functional properties (including activation and inactivation kinetics, surface expression and second-messenger regulation) are modulated by the ancillary subunits. To date, four genes encoding β subunits, four genes encoding α₂-δ subunits and eight genes encoding different types of γ subunits have been identified, and their effects on α₁ subunit function have been characterized in expression systems.

Figure 1. Schematic representation of the calcium channel α₁ (pore-forming) and ancillary (α₂-δ, β and γ) subunits.

There are 10 known α₁ subunits, which are divided into three families based on their sequence identity: Ca_v1 (1.1–1.4), Ca_v2 (2.1–2.3) and Ca_v3 (3.1–3.3). The α₁ subunit consists of four domains (I–IV), each containing six transmembrane segments (S1–S6), with S4 acting as the putative voltage-sensor, and the p-loops between S5 and S6 segments forming the ion-selective pore. The Ca²⁺ channel β subunits (1–4) are cytoplasmic proteins that associate with the domain I–II linker region of the α₁ subunit. The α₂-δ subunits (1–4) are derived from a single gene, but are cleaved post-translationally into α₂ (extracellular) and δ (membrane-spanning) subunits that are linked via disulphide bonds. The γ subunits (1–8) consist of four membrane-spanning regions. Inset: representation of a functional channel in vivo, consisting of a pore-forming α₁ subunit, and a single member from each of the three classes of ancillary subunits.

Significance of calcium channel inactivation

In response to a prolonged membrane depolarization, calcium channels, like many other types of ion channel, enter a non-conducting inactivated gating state. This prevents the breakdown of calcium gradients, ensures the temporal and spatial precision of calcium signals in response to membrane depolarization, and is an important mechanism by which the accumulation of excessive, cytotoxic levels of intracellular calcium is prevented (Choi, 1988; Orrenius et al. 1989; Orrenius & Nicotera, 1994). Specifically, the inactivation of calcium currents in nerve terminals appears to contribute to the short-term depression of neurosecretion (Branchaw et al. 1997; Forsythe et al. 1998). In neurones, the voltage-dependent inactivation of T-type calcium channels is a key determinant of pattern behaviour and pacemaker activity (Chemin et al. 2001, 2002). Moreover, many clinically active calcium channel blockers display an increased affinity for inactivated calcium channels (Sun & Triggle, 1995; Roullet et al. 1999; Jimenez et al. 2000). Finally, in P/Q-type calcium channels, inactivation behaviour is altered by naturally occurring point mutations linked to disorders such as familial hemiplegic migraine (Kraus et al. 1998, 2000; Wappl et al. 2002), thus underlining the importance of the inactivation process for CNS function. As a consequence, it is not surprising to note that voltage-gated calcium channels are capable of undergoing multiple types of inactivation processes. l-type channels (with the exception of Ca_v1.4 channels; McRory et al. 2004) undergo a pronounced speeding of inactivation kinetics in response to an increase in intracellular calcium concentrations (Peterson et al. 1999; Zuhlke et al. 1999). Although it was once believed that this calcium-dependent inactivation process is a unique feature of l-type channels, more recent patch-clamp experiments carried out in the presence of low concentrations of intracellular calcium buffers have revealed calcium-dependent inactivation in P/Q-type and other HVA calcium channels (Lee et al. 1999a; DeMaria et al. 2001; Soong et al. 2002; Liang et al. 2003). In contrast, T-type calcium channels do not appear to support calcium-dependent inactivation. The second means of inactivating calcium entry is through prolonged membrane depolarization. This voltage-dependent inactivation process has both fast and slow components, with fast inactivation occurring over several tens to hundreds of milliseconds, and slow inactivation requiring much more prolonged membrane depolarization (∼1 min), and remains a poorly understood process (Sokolov et al. 2000; Shi & Soldatov, 2002). The ensuing sections will be concerned predominantly with fast, voltage-dependent inactivation of HVA calcium channels, with a particular focus on the underlying channel structural determinants.

Mechanisms of fast inactivation in sodium and potassium channels

More than two decades worth of research into the inactivation mechanisms of sodium and potassium channels (i.e. channels that are structurally homologous to the calcium channel α₁ subunit) provide an excellent framework towards gaining insights into the mechanism of calcium channel inactivation. For example, certain types of potassium channels, such as Shaker B, inactivate through a ‘ball and chain’ mechanism (Armstrong et al. 1973) in which the amino terminus of the α subunit contains a ∼20-amino-acid cluster (i.e. ‘ball’) of hydrophobic and polar domains that is tethered to the remainder of the α subunit via a stretch of about 50 residues (i.e. ‘chain’, see Fig. 2A). Upon channel opening, the ball region enters the inner vestibule of the pore from the cytoplasmic end of the channel to occlude current flow, hence inactivating the channel. Deletion of this region dramatically slows voltage-dependent inactivation, and intracellular addition of synthetic N-terminal peptides to inactivation-deficient Shaker B mutants restores inactivation, consistent with open channel block (Hoshi et al. 1990; Zagotta et al. 1990; Demo & Yellen, 1991). Certain potassium channel β subunit isoforms (e.g. K_v β_1.1 and K_v β₃) contain regions that resemble gating balls, and when coexpressed with certain types of non-inactivating K_v α subunits such as K_v1.1 or K_v1.2, resulting in a rapidly inactivating phenotype channels (Heinemann et al. 1996; Robertson, 1997).

Figure 2. Possible methods of inactivation of the potassium channel (A) and the sodium channel (B).

A, ‘ball and chain’ inactivation of the potassium channel. Each subunit of the tetrameric voltage-gated potassium (K_v) channel has an N-terminal cluster (ball) of ∼20 amino acids tethered to the N-terminal tail of an α subunit that enters the inner vestibule of the channel postopening and physically occludes the pore, thus inactivating the channel. Any of the four available N-terminal ‘balls’ (or a ‘ball’ from an associated β subunit) is capable and sufficient to occlude the pore. B, ‘hinged-lid’ inactivation of the sodium channel. Upon opening, the III–IV linker region folds into the inner vestibule and physically occludes the pore, effectively inactivating the channel.

In sodium channels, inactivation occurs via a qualitatively similar mechanism; however, in this case the intracellular loop connecting domains III and IV of the sodium channel α subunit acts as a ‘hinged lid’ to block the channel pore (West et al. 1992; Eaholtz et al. 1994; Kellenberger et al. 1997; see Fig. 2B). Cleavage of this region via proteolytic enzymes (Armstrong et al. 1973) or intracellular application of functional antibodies against this region removes inactivation (Vassilev et al. 1988, 1989). Indeed, site-directed mutagenesis of a cluster of three hydrophobic amino acid residues in the III–IV linker region (I1488, F1489, M1490; IFM) to glutamine abolishes fast inactivation (West et al. 1992). In analogy with experiments on potassium channels, intracellular application of a pentapeptide containing the IFM motif region partially restores inactivation in inactivation-deficient sodium channel mutants (Eaholtz et al. 1994). Also similar to potassium channels, coexpression of ancillary β1 subunits modulates the inactivation kinetics of the α subunit (McCormick et al. 1999).

Taken together, these data indicate that there are many qualitative similarities in the mechanisms by which sodium and potassium channels undergo fast, voltage-dependent inactivation, including pore block by a cytoplasmic gating particle and regulation by ancillary subunits. As outlined in the ensuing sections, our current understanding of fast inactivation of calcium channels may in many ways resemble that of the families of sodium and potassium channels; however, the molecular details of calcium channel inactivation appear to be much more complex.

What are the determinants of α₁ subunit structural inactivation in HVA channels?

When expressed in the absence of ancillary subunits, barium currents carried by various types of HVA calcium channel display pronounced voltage-dependent inactivation, suggesting that the primary structural determinants underlying the inactivation process reside in the α₁ subunit itself. The first attempt at defining key structural determinants of calcium channel inactivation came from chimeric studies by Zhang et al. (1994). The authors created hybrid calcium channels between rat Ca_v2.1 and marine ray Ca_v2.3 and concluded, based on their electrophysiological characterization, that the domain IS6 segment was responsible for the difference in inactivation kinetics of these two calcium channel subtypes. Subsequently, a number of additional studies have implicated S6 segments in each of the remaining three transmembrane domains. Substituting the domain II or domain IIIS6 segments from rapidly inactivating Ca_v2.3 channels into slowly inactivating Ca_v1.2 channels is sufficient to induce a rapidly inactivating phenotype (Stotz et al. 2000). Familial hemiplegic migraine mutations in the IIS6 and IV S6 regions alter the inactivation kinetics of Ca_v2.1 calcium channels (Kraus et al. 1998, 2000). Artificial point mutations in the IIS6, IIIS6 and IVS6 regions (Hering et al. 1996, 1998; Berjukow et al. 2001; Stotz & Zamponi, 2001) also mediate dramatic increases in the inactivation of l-type calcium channels. For example, Berjukow et al. 2001) reported that substitution of a single methionine residue in the IVS6 region with glutamine increases inactivation rates by as much as 75-fold. Similarly, Stotz & Zamponi (2001) showed that replacement of a single phenylalanine residue in the IIS6 region dramatically speeds up the inactivation of rat Ca_v1.2 calcium channels. Taken together, these findings indicate that all four S6 segments in the calcium channel α₁ subunit contribute to the inactivation process. Furthermore, Spaetgens & Zamponi (1999) observed that all four transmembrane domains contribute to determining the overall voltage-dependence of inactivation, supporting the notion that calcium channel inactivation may arise from a global structural change in the entire channel. We note that a global phenomenon of pore collapse has been proposed as a mechanism of C-type/slow inactivation of potassium and sodium channels (Hoshi et al. 1991; Lopez-Barneo et al. 1993; Lopez et al. 1994; Richmond et al. 1998; Vilin et al. 1999).

In contrast, a number of additional observations provide support for a classical hinged-lid-type mechanism of inactivation. One key feature of a hinged-lid/ball-and-chain mechanism is the involvement of cytoplasmic regions. The first striking evidence supporting the involvement of a cytoplasmic linker region came from observations with Ca_v2.1 calcium channel splice variants (Bourinet et al. 1999). Insertion of a single valine residue in the Ca_v2.1 I–II linker dramatically slows the rate of inactivation of this channel, effectively converting it from a Q-type to a P-type phenotype. Consistent with a key role of the I–II linker region in the inactivation process, substitution of the I–II linker of Ca_v2.3 into the Ca_v1.2 channel results in a significant increase in inactivation rates, whereas a partial substitution of this region has the opposite effect, virtually abolishing inactivation (Stotz & Zamponi, 2001). Furthermore, several other mutagenesis studies also support the involvement of the I–II linker in determining inactivation rates (Herlitze et al. 1997; Berrou et al. 2001). Finally, Cens et al. (1999) reported that overexpression of I–II linker peptides accelerates the inactivation kinetics of expressed Ca_v2.1 channels. Taken together, these findings raise the possibility that the domain I–II linker might perhaps act as a cytoplasmic gating particle. If so, how can this be reconciled with data implicating the S6 segments? One simple possibility is a mechanism in which the S6 segments may be involved in forming the docking site for the gating particle, similar to what has been proposed for sodium channels (McPhee et al. 1994). This will be discussed in greater detail below.

In addition to the domain I–II linker, several other cytoplasmic regions have been linked to the inactivation process. Data from Soldatov et al. (1998) and Sandoz et al. (2001) suggest that the C-terminal region may, in addition to being a key determinant of calcium-dependent inactivation, also regulate the voltage-dependent inactivation of calcium channels. In addition, truncation of the C-terminal region dramatically speeds the inactivation of Ca_v1.2 channels (Stotz & Zamponi, 2002). Stephens et al. (2000) suggested an additional involvement of the N-terminal region in determining inactivation rates. Geib et al. (2002) reported that the inactivation kinetics of Ca_v2.1 channels are slowed when an intramolecular interaction between the domain I–II and the domain III–IV linker regions are disrupted via mutagenesis. This effect, however, was only seen in the absence of the calcium channel β subunit. Finally, deletion of part of the domain II–III linker region of human Ca_v2.2 channels through alternate splicing, while having little effect on inactivation kinetics, results in a depolarizing shift in the half-inactivation potential (Kaneko et al. 2002). Hence, voltage-dependent inactivation of HVA calcium channels involves both pore-lining transmembrane helices and all of the major intracellular domains of the α₁ subunit.

Regulation of inactivation kinetics by extracellular metal ions and second messengers

A recent study by Beedle et al. (2002) has revealed that the inactivation kinetics of transiently expressed Ca_v2.1 and Ca_v1.2, but not other calcium channel subtypes, is regulated by extracellular application of trivalent metal ions. In the presence of nanomolar concentrations of yttrium, these channels underwent a pronounced speeding of the inactivation kinetics, an effect that was independent of the open channel blocking action of these ions. The effects were attenuated following manipulations that antagonize inactivation (i.e. coexpression of the rat β_2a subunit or structural modifications in the I–II linker region). The existence of an external regulation site that can bind multivalent ions is consistent with previous suggestions by Zamponi & Snutch (1996) and Zamponi et al. (1996), and indicates that extracellular regions of the channel may be coupled allosterically to the inactivation machinery.

An observation from the Findlay laboratory showing that β-adrenergic modulation regulates voltage-dependent inactivation of native cardiac l-type calcium channels (Findlay, 2002a,b,c) is perhaps more easily explained in the context of prior structure versus function data. The author showed that β-adrenergic receptor activation slowed down voltage-dependent inactivation, but that this effect was abolished by carbachol. There are a number of phosphorylation sites in the C-terminal region of the channel, perhaps providing a structural explanation for these observations, considering that C-terminal truncations alter voltage-dependent inactivation of the channels. Site-directed mutagenesis of putative protein kinase A (PKA) sites would be required in future experiments to elucidate the underlying molecular determinants. Nonetheless, these data show that intracellular messenger pathways can dynamically regulate the inactivation characteristics of l-type calcium channels.

Determinants of LVA calcium channel inactivation

Compared with HVA calcium channels, the molecular determinants that underlie the inactivation of LVA calcium channels have been well characterized. In perhaps the most detailed study to emerge to date, Staes et al. (2001) reported that the C-terminus region of the Ca_v3.1 channel is important for inactivation, whereas the III–IV linker and N-terminal regions (i.e. the structures involved in sodium and potassium channel inactivation, respectively) do not appear to play a major role. The authors did not, however, examine a putative role for the I–II linker. In contrast, Marksteiner et al. (2001) implicated the domain IIIS6 region as an important inactivation determinant of Ca_v3.1 channels, thus suggesting some overlap in the mechanism of LVA and HVA calcium channel inactivation. More detailed structure–functional analysis will be required to substantiate such as possibility.

Ancillary subunits regulate calcium channel inactivation

It is now well established that ancillary β subunits are important regulators of voltage-dependent inactivation of HVA calcium channels (Lacerda et al. 1991; Varadi et al. 1991), which is consistent with the observation that the primary β subunit interaction site is located in the domain I–II linker region of the α₁ subunit (Pragnell et al. 1994). As a general rule of thumb, coexpression of HVA channels with β_1b or β₃ subunits accelerates inactivation, whereas β₄ subunits result in inactivation kinetics that are as slow as those seen with the α₁ subunit alone. The rat and human β_2a subunits dramatically slow inactivation kinetics (Isom et al. 1994) due to palmitoylation of two cysteine residues in the N-terminus region of this subunit and subsequent tethering of this region to the plasma membrane (Olcese et al. 1994; Qin et al. 1996, 1998; Hurley et al. 2000; Restituito et al. 2000; Feng et al. 2001). Additional interactions between the N-terminus of the calcium channel α₁ subunit and second variable region of β_2a subunit also affect inactivation characteristics (Qin et al. 1996; Stephens et al. 2000). Inactivation of the α₁ subunit is also modulated, although to a lesser degree, by α₂-δ and γ subunits. Coexpression of different α₂-δ subunits with Ca_v1.2 and Ca_v2.3 calcium channels alters the kinetics and the voltage-dependence of inactivation (Klugbauer et al. 1999). The effects of the γ subunit are more subtle, resulting in an enhancement of a slowly inactivating current component seen with Ca_v2.1 calcium channels (Rousset et al. 2001). Neither their interaction sites with the calcium channel α₁ subunit nor the molecular mechanisms by which these two ancillary subunits regulate inactivation kinetics are understood.

A molecular model of HVA calcium channel inactivation

How can all of the observations described in the previous sections be reconciled into one all-encompassing model of inactivation? Structure versus function studies have consistently implicated the S6 segments as well as intracellular regions (see Fig. 3). Of the latter, structural alterations in the I–II linker appear to mediate the greatest effects. Together with the observation that overexpression of I–II linker peptides results in increased inactivation of Ca_v2.1 channels, a role for this region as a putative hinged-lid gating particle is an attractive possibility. Within the confines of such a model, the S6 segments (which probably line the inner vestibule of the pore) could form part of the docking site for the inactivation gate. Consistent with this idea, there is a ring of highly conserved residues at the innermost part of the four S6 regions, which when mutated in individual S6 segments, results in the slowing of inactivation (Shi & Soldatov, 2002). Moreover, mutations in multiple S6 segments produce additive effects. It is also interesting to note that mutations in the domain IIS6 region, which speed the rate of inactivation, do not affect the rate of recovery from inactivation (Stotz & Zamponi, 2001), which indicates that these mutations selectively affect the rate of entry into the inactivated conformation, but not its stability. Based on these findings, Stotz & Zamponi (2001) proposed that in response to membrane depolarization, the S6 segments might undergo a slight conformational change that allows the domain I–II linker gating particle to dock at the cytoplasmic ends of the S6 segments (Fig. 4). Mutations in the S6 regions that speed the rate of inactivation would then simply promote the conformational switch without affecting the stability of the interaction with the I–II linker. Application of trivalent metal ions such as yttrium (Beedle et al. 2002) may perhaps also promote such a putative conformational change in the S6 segments.

Figure 3. Regions of the calcium channel α₁ subunits that have been implicated in inactivation (shown in blue).

Figure 4. A model for calcium channel inactivation (A) reconciled with data from other laboratories (B).

A, possible model for calcium channel inactivation. Following membrane depolarization and channel opening, the S6 segments (represented as helices) undergo a conformational change that unmasks docking sites (noted in green at the base of the S6 helices) for the inactivation gate (red asterisks) formed by the domain I–II linker. B, reconciliation of the model presented in A with data by a number of other laboratories (see main text). Left panel: membrane insertion of the palmitoylated N-terminus of the rat β_2a subunit restricts the mobility of the inactivation gate. Middle panel: interactions between the calcium channel β subunit with the N-terminus and/or the C-terminus region of the α₁ subunit indirectly impairs I–II linker function as the inactivation gate. Right panel: interactions between the I–II linker and the III–IV linker, occurring in the absence of a β subunit, directly regulate inactivation.

Assuming that the domain I–II linker acts as a physical gating particle, how can one account for the effects of other intracellular regions and/or of the ancillary subunits? Since the calcium channel β subunit physically interacts with the I–II linker region, anchoring of this subunit to the plasma membrane, as is the case for rat β_2a, would be expected to restrict the mobility of the I–II linker, thus slowing inactivation (Fig. 4). A similar consequence would be expected from the reported interactions between the I–II and III–IV linker regions (Geib et al. 2002). In contrast, the effects of the N-terminus and C-terminal regions (such as, for example, those due to PKA-dependent phosphorylation) may arise indirectly from their interactions with the β subunit (Stephens et al. 2000; Sandoz et al. 2001; Fig. 4).

In summary, the model presented in Fig. 4 can accommodate virtually all of the structure–function data reported in the literature to date. Although this model will need to be rigorously tested, we suggest that the basic mechanism of calcium channel inactivation is qualitatively similar to that seen with other types of voltage-dependent ion channels, thus underlining the fundamental importance of the inactivation process in the physiology of excitable cells.