Abstract
Calcium influx into any cell requires fine tuning to guarantee the correct balance between activation of calcium-dependent processes, such as muscle contraction and neurotransmitter release, and calcium-induced cell damage. G protein-coupled receptors play a critical role in negative feedback to modulate the activity of the CaV2 subfamily of the voltage-dependent calcium channels, which are largely situated on neuronal and neuro-endocrine cells. The basis for the specificity of the relationships among membrane receptors, G proteins, and effector calcium channels will be discussed, as well as the mechanism by which G protein-mediated inhibition is thought to occur. The inhibition requires free Gβγ dimers, and the cytoplasmic linker between domains I and II of the CaV2 α1 subunits binds Gβγ dimers, whereas the intracellular N terminus of CaV2 α1 subunits provides essential determinants for G protein modulation. Evidence suggests a key role for the β subunits of calcium channels in the process of G protein modulation, and the role of a class of proteins termed “regulators of G protein signaling” will also be described.
I. Introduction
Voltage-gated calcium channels (VGCCs1) play a major role both in the normal functioning and in the patho-physiology of neurons and other excitable cells. Although they are also found at low levels in nonexcitable cells, their presence has been said to define an excitable cell (Hille, 2001). They were first identified in crustacean muscle by Fatt and Katz (1953), and subsequently extensively studied by a number of groups, including Hagiwara and Takahashi (1967). VGCCs were first classified according to their biophysical properties into low- and high-voltage-activated (LVA and HVA) channels (Carbone and Lux, 1984). Further study, with the additional aid of pharmacological tools, led to the classification of certain HVA channels as “long-lasting” or L-type channels, which were sensitive to the 1,4-dihydropyridine (DHP) class of drugs, and present in skeletal muscle, heart, smooth muscle, and neurons (Hess et al., 1984; Nowycky et al., 1985a).
In neurons it was clear that a component of the HVA calcium current was not L-type, for example, in Purkinje cells in the cerebellum (Hillman et al., 1991) and at presynaptic terminals (see Stanley and Atrakchi, 1990 for example). These additional current components were subclassified with the aid of several invaluable toxins. Two additional subtypes of calcium channel were thus identified: N-type channels, sensitive to ω-conotoxin GVIA (Nowycky et al., 1985b; McCleskey et al., 1987) and P-type channels, sensitive to ω-agatoxin IVA (Mintz et al., 1992). Another ω-agatoxin IVA-sensitive current component was subsequently identified in cerebellar granule cells and termed Q-type current (Randall and Tsien, 1995), but these two components are now combined as P/Q. There is also a residual or R-type calcium current component that is resistant to DHPs and the N and P/Q channel toxins (Randall and Tsien, 1995).
A. Molecular Subtypes of Calcium Channel
The molecular basis for the physiological subtypes of VGCCs was clarified after the identification of the subunits of voltage-gated calcium channels. This era started with the purification of skeletal muscle calcium channel complex also called the DHP receptor, which consisted of α1, α2, β, δ, and γ subunits (Flockerzi et al., 1986; Hosey et al., 1987; Takahashi et al., 1987; Chang and Hosey, 1988; Hymel et al., 1988).
After identification of individual subunits, the sequencing of peptides derived from these subunits formed the basis for the subsequent identification and cloning of the cDNA for the DHP receptor, initially from skeletal muscle (Tanabe et al., 1987) and subsequently from heart by homology with the skeletal muscle sequence (Mikami et al., 1989) (Fig. 1A). The α1 subunits have 24 putative transmembrane segments, arranged into four homologous domains, with intracellular linkers and N and C termini (Fig. 1A). Ten different α1 subunits have been cloned that have specialized functions and distributions (for review, see Ertel et al., 2000) (Fig. 1B). Those that are of most concern to this review are the CaV2 subfamily of HVA calcium channels that shows classical modulation by G proteins, comprising CaV2.1 or α1A, the molecular counterpart of P/Q-type calcium channels (Mori et al., 1991), CaV2.2 or α1B (Dubel et al., 1992), the molecular counterpart of N-type calcium channels, and CaV2.3 or α1E (Soong et al., 1993), thought to contribute to the molecular counterpart of the R-type calcium current component (Piedras-Rentería and Tsien, 1998) (Fig. 1B, boxed).
In the case of the N- and P/Q-type as well as the L-type HVA calcium channels, the CaVα1 subunit has been shown to copurify with an intracellular β subunit (CaVβ) (Liu et al., 1996; Scott et al., 1996). Four β subunits have been cloned (β1–4), with β1a being the skeletal muscle isoform of β1 (Ruth et al., 1989), β2 being cloned initially from cardiac muscle (Perez-Reyes et al., 1992), β3 present in cardiac and smooth muscle and neuronal tissue (Castellano et al., 1993b), and β4 cloned from brain (Castellano et al., 1993a). A number of splice variants have been identified, with one particular splice variant of β2, β2a, being N-terminally palmitoylated in certain species, giving it distinctive properties (Chien et al., 1996).
HVA calcium channels also copurify with an extracellular CaVα2 subunit, which is attached by S-S bonds to a transmembrane δ subunit (Tanabe et al., 1987; Chang and Hosey, 1988; Witcher et al., 1993; Liu et al., 1996). Four α2δ subunits have been cloned (Ellis et al., 1988; Klugbauer et al., 1999; Barclay et al., 2001; Qin et al., 2002).
Skeletal muscle calcium channels also copurify with a γ1 subunit (Takahashi et al., 1987). Whether any of the recently cloned novel γ-like subunits (γ2–8) (Fig. 1B) are tightly associated with other types of VGCCs remains controversial (Letts et al., 1998; Black and Lennon, 1999; Klugbauer et al., 2000; Kang et al., 2001; Moss et al., 2002; Tomita et al., 2003).
II. Role of CaVβ Subunits in Calcium Channel Function
The intracellular CaVβ subunits have marked effects on the properties of HVA α1 subunits (CaV1 and CaV2 families), including trafficking of calcium channel complexes to the plasma membrane and modification of kinetic and voltage-dependent properties (Singer et al., 1991; De Waard et al., 1994; Chien et al., 1995; Brice et al., 1997; Bichet et al., 2000). My group has shown that the converse also applies, in that antisense-induced knockdown of CaVβ subunits from native neurons results in a reduction of the amplitude of endogenous calcium currents and slowed kinetics of activation (Berrow et al., 1995; Campbell et al., 1995b).
Most research indicates that all CaVβ subunits (except truncated splice variants described recently (Hibino et al., 2003; Hullin et al., 2003) increase the functional expression of HVA α1 subunits (for a recent review, see Dolphin, 2003). In theory, this could be attributable to effects on a number of channel properties, an increase in the open probability of the channel, an increase in single-channel conductance, an increase in the number of functional channels inserted into the plasma membrane, or a combination of several of these processes. Initial studies did not agree whether there was an increase in number of channels at the plasma membrane, measured as charge moved in isolated gating currents, with either no increase (Neely et al., 1993) or an increase being reported (Josephson and Varadi, 1996). Much early work on the roles of CaVβ subunits in calcium channel expression was performed in Xenopus oocytes, but these cells are now known to contain an endogenous Xenopus β subunit that complicates the interpretation of these results (Tareilus et al., 1997; Canti et al., 2001). This endogenous CaVβ subunit was found to be both necessary and able to traffic at least some heterologously expressed CaV channels to the plasma membrane, since if endogenous β subunit expression was reduced or eliminated by injection of β3 antisense oligonucleotides, CaV expression was largely lost (Tareilus et al., 1997; Canti et al., 2001).
In COS-7 cells, small currents were observed when CaV2.1, CaV2.2, and CaV2.3 were expressed alone, but exogenous β subunits all increased CaV2.1, CaV2.2, and CaV2.3 maximum conductance about 10-fold (Berrow et al., 1997; Stephens et al., 1997; Meir and Dolphin, 1998; Stephens et al., 2000). COS-7 cells do contain mRNA for endogenous β subunits (Meir et al., 2000), but the protein for corresponding β subunits was not detectable by immunocytochemistry (Meir et al., 2000), although a low level might be found if higher sensitivity detection methods were used. Thus, at the moment there are no expression systems that definitively contain no CaVβ subunits that can be used conclusively to answer the question as to whether HVA calcium channels can be trafficked to the plasma membrane without a β subunit.
The increase in current density brought about by CaVβ subunits can be attributed to a number of effects on biophysical properties as well as the important influence on trafficking. All CaVβ subunits hyperpolarize the voltage dependence of activation of all HVA VGCCs (Fig. 2), whereas all, except the β2a splice variant that is N-terminally palmitoylated, hyperpolarize the voltage dependence of steady-state inactivation (Birnbaumer et al., 1998). Where it has been studied, the β subunits all produce an increase in mean open time, which is at least in part due to a hyperpolarizing shift in the voltage dependence of the mean open time (Wakamori et al., 1999; Meir et al., 2000). Although CaVα1 subunits contain inherent determinants of voltage-dependent inactivation (Zhang et al., 1994; Herlitze et al., 1997; Cens et al., 1999; Spaetgens and Zamponi, 1999), association with different CaVβ subunit isoforms dictates their overall kinetics of inactivation (Olcese et al., 1994; Meir and Dolphin, 2002). At the whole-cell level, the inactivation rate was affected in the following order (highest first) β3 > β1b > β4 > β2 subunits. Retardation of inactivation has been shown to be particularly dramatic for the palmitoylated CaVβ2a subunit expressed with CaV1.2 (Chien and Hosey, 1998), CaV2.2 (Bogdanov et al., 2000; Stephens et al., 2000), or CaV2.3 (Qin et al., 1998).
A. Binding of CaVβ to the α1 I-II Linker
CaVβ subunits have been found to bind with very high affinity to the cytoplasmic intracellular linker between domains I and II of all HVA calcium channels, via an 18-amino acid motif called the α interaction domain (AID) on the I-II linker (Pragnell et al., 1994). The AID sequence of rabbit CaV2.2 is QQIERELNGYLEWIFKAE, and the consensus sequence present in both CaV1.x and CaV2.x subfamilies is QQxExxLxGYxxWIxxxE.
A 41-amino acid sequence (BID) on the CaVβ subunit was identified as the minimal motif required to influence α1 subunit expression and to bind to the α1 subunit (De Waard et al., 1994, 1996). The consensus sequence of BID is K—E—PYDVVPSMRP—LVGPSLKGYEVTDMMKQALFDF; the underlined serines are consensus protein kinase C (PKC) phosphorylation sites. The residues in bold have been identified as particularly important for binding to CaVα1 subunits (Walker and De Waard, 1998). This small BID sequence alone can produce an increase in calcium current density, albeit not to the same extent as the full-length protein (De Waard et al., 1994). The affinity between CaVβ subunits and a I-II linker fusion protein has been measured to be between 5 and 60 nM (De Waard et al., 1994), but has been proposed to be state-dependent (De Waard et al., 1995; Canti et al., 2001). In one study (De Waard et al., 1995), no dissociation was seen for β1b from the CaV2.1 I-II linker fusion protein after 10 h, but this may reflect a technical difficulty of overlay assays, because the bound protein may become directly anchored to the membrane. In our own binding studies using surface plasmon resonance, the affinity of β3 for the GST fusion protein of the I-II linker of CaV2.2 was about 20 nM, and the koff was 5.2 × 10-3 s-1 (Canti et al., 2001). We found similar data (Fig. 3A) for β1b binding to the I-II linker of both CaV2.2 and CaV1.3 (Bell et al., 2001).
We have studied the in vivo concentration dependence of the effects of CaVβ subunits. Our evidence supports the hypothesis that there are two distinct binding processes for β subunits on CaV2.2 (Canti et al., 2001). One is a high-affinity process related to the effect of CaVβ on the maximum conductance of CaV2.2, presumably involving its trafficking to the plasma membrane, whose affinity corresponds closely to the in vitro affinity for the I-II linker (∼17 nM), which is coincidentally the concentration of endogenous β3 estimated to exist in Xenopus oocytes (Canti et al., 2001). The second process is of lower affinity (KD ∼120 nM), associated with the voltage-dependent effects of the β subunit, for example steady-state inactivation. One explanation for the discrepancy in these two calculated affinities is that a single binding site undergoes a marked reduction in affinity for CaVβ subunits once the CaVα1 subunits have been trafficked from the endoplasmic reticulum and are inserted in the polarized plasma membrane. Alternatively, one might postulate the coexistence of two separate CaVβ subunit binding sites on each CaV2.2 molecule, but the binding of two CaVβ subunits has not been demonstrated directly (Canti et al., 2001). Whichever hypothesis is correct, it is highly likely that the CaVβ subunit interacts with other domains on the CaVα1 subunit as well as the I-II linker.
B. Binding of CaVβ Subunits to the N and C Termini of CaVα1 Subunits
Two other β subunit interaction sites have been identified on various α1 subunits on the C terminus (Qin et al., 1997; Walker et al., 1998) and the N terminus (Walker et al., 1999; Stephens et al., 2000). These appear to be of lower affinity and may be selective for certain CaVβ subunits. Whether they represent part of a single complex β subunit binding pocket made up of the I-II linker and the N and C termini remains to be established. However, the binding site found for β2a on the C terminus of CaV2.3 appeared to involve binding to the BID domain of β2a, the same as that which binds to the CaVα1 I-II linker, making it an alternative, rather than an additional site, for an individual CaVβ subunit (Qin et al., 1997). Walker et al. (1999) showed that the N terminus of CaV2.1 interacted with CaVβ4 and β2a but not β1b or β3. The region of β4 involved was within its C terminus (amino acids 446–482). The C terminus of CaVβ4 also bound to the C terminus of CaV2.1. The N and C termini of CaV2.1 were found to occupy overlapping binding sites that were mutually exclusive, but either could bind in combination with binding to the I-II linker (Walker et al., 1999). This group also showed that CaVβ4 produced a smaller hyperpolarizing shift of CaV2.1 currents than did CaVβ3, and that this differential was due to the CaV2.1 N terminus. Stephens et al. (2000) showed that CaV2.2 N-terminal residues in the same region as the essential site for G protein modulation (see Section IV.D.) were involved in retardation of inactivation kinetics by β2a. Palmitoylated β2a has been suggested to retard inactivation by tethering the I-II linker so that it cannot mediate inactivation (Restituito et al., 2000; Stotz et al., 2000), but our data show an additional role for the N terminus of CaV2.2.
III. Modulation of Calcium Channels
There are several means by which VGCCs may be both up- and down-regulated by second messenger pathways, for example by phosphorylation (Nunoki et al., 1989; Dolphin, 1999; Catterall, 2000). These include regulation by kinases, for example, up-regulation of cardiac L-type channels by cyclic AMP-dependent protein kinase (Reuter, 1987) and regulation by protein kinase C (Stea et al., 1995). In this review, however, I shall concentrate on the classical G protein pathway and describe first how this was examined in native neurons.
For the neuronal channels, particularly N- and P/Q-types, a major mechanism of inhibitory modulation occurs via the activation of heterotrimeric G proteins by seven transmembrane G protein-coupled receptors (GPCRs). GPCR activation was first found to reduce action potential duration in dorsal root ganglion neurons in the 1970s (Dunlap and Fischbach, 1978). Subsequently, this effect was found to result from inhibition of voltage-gated calcium channels (Dunlap and Fischbach, 1981). Such modulation has since been observed in many types of neuron, including superior cervical ganglion neurons (Ikeda and Schofield, 1989) and submucosal neurons (Surprenant et al., 1990).
The GPCRs typically involved in this type of modulation include α2-adrenoceptors, μ and δ opioid receptors, GABA-B receptors (Fig. 4, A and B), and adenosine A1 receptors (Dunlap and Fischbach, 1978; Dolphin et al., 1986; Scott and Dolphin, 1986). The key features that typify this inhibition are a slowing of the current activation kinetics, which is thought to be due to a time- and depolarization-dependent recovery from voltage-dependent inhibition (Bean, 1989). The voltage dependence is manifested by a shift to more depolarized potentials of the current activation-voltage relationship and the loss of inhibition at large depolarizations, because of a shift from “reluctant” to “willing” channels (Bean, 1989). Removal of inhibition can also be induced by a depolarizing prepulse applied immediately before the test pulse (Ikeda, 1991). Additional mechanisms that are not voltage-dependent have also been described in various cell types manifested by a scaled reduction in the current and an inability of a depolarizing prepulse to reverse this component of the inhibition (for example, see Diversé-Pierluissi and Dunlap, 1993).
N- and P/Q-type calcium channels support synaptic transmission and are concentrated at nerve terminals. P/Q-type channels are most important for transmitter release at central terminals (Takahashi and Momiyama, 1993), although N-type channels are also present and particularly contribute earlier in development. In contrast, N-type channels are more prevalent in peripheral nerve terminals and are largely responsible for synaptic transmission in autonomic and sensory terminals (Mochida et al., 1996; Koh and Hille, 1997). Modulation of these channels by activation of GPCRs has been shown to occur both in cell bodies (Holz et al., 1986; Scott and Dolphin, 1986; Dolphin and Scott, 1987; Ikeda, 1991) and at presynaptic terminals (Takahashi et al., 1998). This mechanism may be responsible for at least some of the presynaptic inhibition of synaptic transmission mediated by a wide variety of GPCRs in many areas of the nervous system (Dolphin and Prestwich, 1985; Man-Son-Hing et al., 1989; Toth et al., 1993). Activation of GPCRs such as the GABA-B receptor will reduce calcium entry into presynaptic terminals via VGCCs by the same mechanism that is observed in cell bodies (Fig. 4A), and the effect should also be frequency- and potential-dependent. Inhibition will be reduced during a high-frequency train as a result of the voltage dependence of the inhibitory modulation. Relief of inhibition of calcium currents, evoked by action potential-like voltage waveforms, has been reported during high-frequency trains (Williams et al., 1997; Brody and Yue, 2000) and may contribute to the modulation of presynaptic inhibition according to input frequency.
IV. Inhibitory Coupling between G Proteins and Voltage-Gated Calcium Channels in Native Tissue
The role of G proteins in the inhibition of calcium currents by GPCR activation was first demonstrated some years later (Holz et al., 1986; Scott and Dolphin, 1986). It was also identified that there was a direct membrane-delimited link between G protein activation and N- or P/Q-type calcium channel inhibition. A key experiment enabling this idea to be accepted was the finding that inhibition of calcium channels recorded in the cell-attached patch mode only occurs when the receptor agonist is present in the patch pipette, and not when it bathes the remainder of the cell membrane (Forscher et al., 1986). This indicates that the inhibitory process is very localized and that a soluble second messenger is not involved. In most studies, it is found that this direct linkage only applies to the CaV2 family of calcium channels, but additional non-voltage-dependent pathways, which may be direct or via down-stream soluble intracellular messengers also occur in certain cell types and may additionally apply to L-type channels (Hille, 1992; Dunlap and Ikeda, 1998) and also to T-type channels (Wolfe et al., 2003).
A. The G Protein Subunits Involved in the Direct Inhibitory Modulation of Native and Heterologously Expressed Calcium Channels
The modulation of neuronal VGCCs in most native neurons is usually mediated by receptors coupled to pertussis toxin-sensitive G proteins (Gi and Go subtypes) (Holz et al., 1986; Scott and Dolphin, 1986). The response to an agonist can be mimicked by nonhydrolyzable analogs of GTP such as guanosine 5′-O-(3-thiotriphosphate) (GTPγS) (Fig. 4C) (Dolphin and Scott, 1987) and by photoactivation of a caged GTP analog (Dolphin et al., 1988). The effect of GTP analogs is, as expected, more extensive than that of receptor agonists and is irreversible because G proteins are permanently activated. To provide evidence that the effect of agonists is mediated by G proteins, initial experiments showed that the effect of agonists is enhanced and made irreversible by a low concentration of GTPγS (Scott and Dolphin, 1986) and prevented by a GDP analog such as guanosine 5′-O-(2-thiodiphosphate) (Holz et al., 1986; Dolphin and Scott, 1987).
To identify which G proteins are involved in receptormediated inhibition of calcium channels in native systems (such as dorsal root ganglion neurons and sympathetic neurons), a number of studies were performed with blocking antibodies and antisense oligonucleotides complementary to G protein α subunits, which showed that Gαo was primarily responsible for the response (McFadzean et al., 1989; Baertschi et al., 1992; Campbell et al., 1993; Menon-Johansson et al., 1993). However, others found that both Gαi and Gαo were involved (Ewald et al., 1989), and in several studies, Gs- or Gq-coupled receptors produced similar modulation (Shapiro and Hille, 1993; Golard et al., 1994; Zhu and Ikeda, 1994). This led to the hypothesis that the species involved was the moiety common to all these G proteins, Gβγ rather than any particular Gα, and this was subsequently found to be the case (Herlitze et al., 1996; Ikeda, 1996), although previously other groups had directly investigated the involvement of Gβγ in calcium channel modulation and not found any effect of its infusion (Hescheler et al., 1987). However, there was a clear precedent for an effector role for Gβγ in the G protein-activated potassium channels (GIRKs). Although for many years a controversy reigned concerning which G protein subunit was responsible for modulation of the native GIRKs (e.g., Yatani et al., 1987), they were eventually shown conclusively to be activated by Gβγs (Logothetis et al., 1987; Kurachi et al., 1989; Clapham and Neer, 1993). Furthermore, most Gβγ combinations tested except transducin (Gβ1γ1) are similarly effective (Wickman et al., 1994; Yamada et al., 1997). From the work of two groups (Herlitze et al., 1996; Ikeda, 1996), it became clear that transfection of either primary neurons or cell lines with Gβγ subunits mimicked agonist effects and led to tonic inhibition of the calcium current, which could be transiently reversed by a depolarizing prepulse, applied just before the test pulse, a hallmark of voltage-dependent inhibition of these channels.
Taking examples for illustration from our own work (Meir et al., 2000), the effect of coexpression of Gβγ with N-type calcium channels can be seen both at the whole-cell and at the single-channel level (Figs. 5A, and 6A). When CaV2.2 was coexpressed in COS-7 cells with β2a and Gβ1γ2, the whole-cell currents were very small and slowly activating but were markedly enhanced by a depolarizing prepulse (Fig. 5A, left panel). The time constant of activation of the peak current at 0 mV in the presence of Gβγ was about 27 ms, compared with less than 5 ms in the presence of the Gβγ binding domain of β-adrenergic receptor kinase 1 (β-ARK1) to bind any free endogenous Gβγ. At the single-channel level, the slow activation of CaV2.2/β2a channels was seen as a marked prolongation of the latency to first opening (Fig. 6A, compare traces in right panel with Gβγ with those in the left panel with β-ARK1). Indeed, there were many instances where no openings were observed to a test pulse. The latency to first opening was significantly reduced by a large depolarization (Fig. 6A, right panel). Gβγ overexpression also occluded modulation by agonist (Ikeda, 1996). The Gβγ-mediated inhibition presents a picture that is very similar to that of agonist-mediated inhibition in terms of slowed activation and reversal by a large depolarizing prepulse. This is illustrated in an example from work by my own group, showing the effect of quinpirole-mediated activation of a coexpressed D2-dopamine receptor on the same channel combination (Fig. 7A) (Meir et al., 2000).
The involvement of Gβγ dimers as mediators of the G protein-signaling pathway does not call into question the finding by many groups that Gαo is involved in receptor-mediated inhibition in many native systems (Ewald et al., 1989; McFadzean et al., 1989; Campbell et al., 1993; Menon-Johansson et al., 1993; Degtiar et al., 1996), because Go is present in very high concentrations, particularly in neurons. Thus, in the absence of the Gαo subunit, GPCR-mediated signaling will be markedly attenuated since it depends on the G protein heterotrimer.
Nevertheless, a number of studies have further concluded that there is a specific role for Gα subunits. In some cell types, a marked specificity of different Gαβγ combinations for signaling pathways between different receptors and calcium channels has been demonstrated (Kleuss et al., 1991; Degtiar et al., 1996). These findings might also be reconciled with the evidence that most Gβγ subunits are able to transduce the signal to calcium channels (Ikeda, 1996; Garcia et al., 1998; Ruiz-Velasco and Ikeda, 2000) by interpreting that a specific G protein heterotrimer combination may selectively couple to a particular receptor in intact cells, and the selectivity is therefore largely at the receptor-G protein interaction step or due to segregation into different compartments, rather than due to Gβγ specificity. In one study, however, it was concluded that there was an effector role for Gαi3 subunits (Furukawa et al., 1998b). In this study, either Gα or Gβγ moieties were coexpressed in Xenopus oocytes with CaV2.2 or CaV2.1 channels and the μ-opioid receptor. Receptor-mediated inhibition was found to be enhanced by coexpression of Gα, which was therefore said to mediate this inhibition; but a more likely explanation of this finding is that expression of exogenous Gα increases the amount of Gαβγ available for coupling to the receptor (Jeong and Ikeda, 1999; Canti and Dolphin, 2003) rather than that a specific Gα mediates the response. In another study in chick sensory neurons, it was originally suggested that following activation of α2-adrenoceptors, Gβγ dimers were responsible for the voltage-independent inhibition via activation of PKC and activated Gα for voltage-dependent inhibition via an unknown second messenger (Diversé-Pierluissi and Dunlap, 1993; Diversé-Pierluissi et al., 1995). It is unknown whether different pathways might be activated in avian neurons. Interpretation of the role of phosphorylation in mediating a pathway must always bear the proviso that phosphorylation might also occlude receptor-mediated effects by receptor down-regulation.
I have further examined whether Gα subunits play any role in mediating calcium channel inhibition, by the use of receptor-Gα fusion proteins. I found that there was no difference between α2 adrenoreceptor-Gαo and -Gαi tandems and the wild-type α2 adrenoreceptor in their ability to support G protein-mediated inhibition of N-type calcium channels in an expression system, and also no difference in the voltage dependence of the inhibition (Fig. 8), despite the fact that no Gα amplification would occur in the case of the tandems (Bertaso et al., 2003). This is in contrast to the selective inhibition by Go rather than Gi in sympathetic neurons (Delmas et al., 1999), which may depend on localization within discrete membrane compartments in native cells (Delmas et al., 2000). Thus, coexpression studies in heterologous systems provide information on what is possible, but studies in native cells, using different means of interfering with the signal transduction pathway or its intermediates, provide information on what actually happens in any given cell. Both types of study are essential to place the correct interpretation on data from native cells.
One Gβ subunit, Gβ5, when overexpressed in sympathetic neurons was less effective than other Gβ subunit combinations at producing G protein modulation of VGCCs (Ruiz-Velasco and Ikeda, 2000). This may be because Gβ5 preferentially interacts with certain regulators of G protein signaling (RGS) proteins with G protein γ-like domains, rather than with Gγ subunits themselves (Snow et al., 1998), and also couples selectively to the Gq family of Gα subunits (Fletcher et al., 1998).
B. Voltage Dependence of G Protein Modulation of Calcium Channels
It is commonly accepted that the relief of G protein-mediated inhibition by depolarization is a result of rapid dissociation of Gβγ dimers from the channel at depolarized potentials. This process is thought to be strongly voltage- and time-dependent (see Figs. 5A and 6A for examples) and, as suggested above, is also believed to cause the slow relaxation observed in response to a test pulse (Jones and Elmslie, 1997). Furthermore, re-establishment of inhibition after a prepulse, during a period at the holding potential, is likely to result from rebinding of Gβγ dimers. Whether these processes actually result in physical dissociation and reassociation between the G protein subunits and channel remains formally to be established. However, the finding that the rate of reblock is dependent on the concentration of activated G protein is consistent with this view (Elmslie and Jones, 1994; Stephens et al., 1998a; Zamponi and Snutch, 1998). The interpretation of these results is that the process involves binding from the pool of free Gβγ dimers.
C. The Role of the CaVα1 I-II Linker in G Protein Modulation of CaV2 Calcium Channels
The combination of two findings, 1) that Gβγ dimers are the mediators of inhibitory modulation, and 2) that there is a functional interaction between CaVβ subunits and Gβγ (Campbell et al., 1995), led a number of groups to examine the intracellular I-II linker in detail. Gβγ dimers have been found previously to bind to sites on type 2 adenylyl cyclase and phospholipase C β2 (Chen et al., 1995), which have a characteristic central motif consisting of QXXER. Whereas this motif is not necessarily indicative of a functional Gβγ binding site, it is also found to occur in the I-II loop of CaV2.1, CaV2.2, and CaV2.3, intriguingly within the binding site described for the VGCC β subunit (QQIERELNGY–WI-KAE) (Pragnell et al., 1994). Furthermore, it is modified in the cognate region in L-type channels (QQLEEDL-GY–WITQ-E).
It is clear that Gβγ binds to the I-II linker of G protein-modulated calcium channels. This has been shown by several groups using overlay assays (De Waard et al., 1997; Zamponi et al., 1997) and also by the use of a surface plasmon resonance-based system to measure reversible binding, where the on- and off-rates can be measured, showing an affinity of Gβγ for the I-II linker of CaV2.2 of 62 nM (Fig. 3B) (Bell et al., 2001). Furthermore, the I-II linker of the L-type channel CaV1.3 does not bind Gβγ, although it will bind CaVβ subunits (Bell et al., 2001). The residues in the I-II linker critical for Gβγ binding have been mapped (De Waard et al., 1997). As predicted, the AID part of the linker is one important domain, and some residues of the QQIER sequence were shown to be essential for Gβγ binding (De Waard et al., 1997).
The role of the I-II linker Gβγ binding site in the process of G protein modulation remains controversial. Initial electrophysiological studies that supported a major role for the I-II linker used peptides derived from the I-II linker region in the patch pipette and found that they blocked G protein modulation (Herlitze et al., 1997; Zamponi et al., 1997). However, peptides alone do not prove that the CaVα1 I-II loop is the site of modulation but rather indicate whether the peptides bind to Gβγ and can therefore effectively compete for this mediator. Chimeric and mutant channels have also been made between those channels that showed the greatest G protein modulation, such as CaV2.2, and those that exhibited no or less modulation, in an attempt to define the regions involved in this process. Three groups found that the I-II linker was key to Gβγ modulation (De Waard et al., 1997; Herlitze et al., 1997; Zamponi et al., 1997). Zamponi et al. (1997) made chimeras between CaV2.1 and CaV2.2 by putting the I-II linker of CaV2.2 into CaV2.1, which resulted in an increased modulation, with CaVβ1b as the coexpressed β subunit. However, both the channels used in this study were G protein modulatable. De Waard et al. (1997) found that a mutation that prevented Gβγ binding (R → E in QQIER) also prevented inhibitory modulation of CaV2.1/β4 by GTPγS injection into oocytes, although in this study only a small amount of inhibition was observed with GTPγS even in the wild-type CaV2.1. However, Herlitze et al. (1997) mutated the QQIER sequence in CaV2.1 to that in CaV1.2, which is QQLEE, and found a reduction of modulation of the channel coexpressed with β1b by GTPγS, but not an abolition. Interestingly, in this study, they observed that CaV2.1 with the sequence QQIEE showed increased, rather than decreased, modulation by GTPγS, in contrast to the results of De Waard et al. (1997).
Two other groups found that the presence of an I-II linker from a G protein-modulatable channel was either not essential for G protein modulation or not the most critical region (Zhang et al., 1996; Page et al., 1997, 1998; Canti et al., 1999). The results of Zhang et al. (1996) showed that neither the I-II linker from CaV2.1 nor that from CaV1.2 reduced modulation in a CaV2.2 backbone when coexpressed with β1. In agreement, the results of Page et al. (1997) showed that insertion of the I-II linker of CaV2.2 into a CaV2.3 construct (RbEII) (Soong et al., 1993), which had a truncated N terminus, did not restore the G protein modulation shown by CaV2.2/β1b. Furthermore, Canti et al. (1999) showed that the I-II linker of CaV2.2 inserted into a CaV1.2 backbone and coexpressed with β2a did not allow any G protein modulation (Fig. 9). Both groups found that domain I of G protein-modulated channels was a key region in the process of G protein modulation (Zhang et al., 1996; Stephens et al., 1998b). The discrepancies do not appear to be due to the different Gβγ dimers or CaVβ subunits used, because the different groups have coexpressed channels with a variety of different CaVβ subunits, and the Gβγ combination, where used, was Gβ1γ2 or Gβ2γ3.
Thus, the I-II linker of G protein-modulated channels has a clear ability to bind Gβγ (Fig. 3B), although there appear to be no motifs in the I-II linker whose presence, despite being essential for high-affinity binding to the I-II linker, is absolutely essential for the process of modulation of CaV2.x channels by these Gβγ dimers.
D. The Essential Role of the CaVα1 N Terminus in G Protein Modulation
In reconstituted systems, consisting minimally of a VGCC α1/β combination, with or without an α2δ and either an endogenous or an expressed G protein-coupled receptor, classical G protein modulation could be demonstrated for CaV2.2 and, to a lesser extent, for CaV2.1, but modulation of CaV2.3 was only observed by some groups (Yassin et al., 1996; Meza and Adams, 1998) but not by my group using a particular rat CaV2.3 clone (Page et al., 1997). My group subsequently identified the reason for this discrepancy; one of the initial clones from rat brain, rbEII (Soong et al., 1993), had a truncated 5′ coding region, commencing at the second methionine, and this showed no G protein modulation (Page et al., 1998). PCR extension of the N terminus formed a full-length rat CaV2.3 clone homologous to the rabbit and human clones, and this full-length CaV2.3 is strongly G protein-modulated (Page et al., 1998). To examine whether the requirement for an intact N terminus was a general conclusion, not limited to CaV2.3, my group further showed that partial truncation of the highly homologous N terminus of CaV2.2 abolished its ability to be G protein modulated (Page et al., 1998). Furthermore, a chimeric channel consisting of only the cytoplasmic N terminus of CaV2.2 in a rat CaV1.2 backbone showed all the elements of classical G protein modulation, whereas CaV1.2 did not (Canti et al., 1999) (Fig. 9). We then extended this study to show that an 11-amino acid motif YKQSIAQRART in CaV2.2 N terminus that is also highly conserved in CaV2.1 and CaV2.3 was essential for G protein modulation. Within this sequence, mutation of either YKQ or RAR to AAA abolished G protein modulation (Canti et al., 1999). Elements of this 11-amino acid motif were also involved in interaction with CaVβ subunits, because deletion of this motif in CaV2.2 or mutation of certain residues countered the CaVβ2a-mediated retardation of inactivation (Stephens et al., 2000).
E. Basis for the Selectivity of Calcium Current Inhibition by Transmembrane G Protein-Coupled Receptors
Activation of the Gq family of G proteins does not produce typical voltage-dependent inhibition of N-type calcium channels despite the production of Gβγ dimers, but instead produces a non-voltage-dependent inhibition (Kammermeier et al., 2000; Bertaso et al., 2003). The reason for this is unclear, but a number of hypotheses have been put forward. One possibility is that, as stated above, Gαq frequently interacts with Gβ5 in native tissues, and Gβ5 is unique among the G proteins in that it does not interact with most Gγ subunits (Zhou et al., 2000). A second possibility is that the inhibition is via Gαq itself or a downstream effector. Indeed, the voltage-dependent inhibitory modulation of calcium currents resulting from activation of the pertussis toxin-sensitive G protein pathway or expression of Gβγ dimers can be reversed by coactivation of Gαq (Zamponi et al., 1997; Simen et al., 2001; Bertaso et al., 2003). Gαq activates phospholipase C and downstream signal transduction events including PKC. There is a threonine, Thr422, in a PKC consensus phosphorylation site just C-terminal to the AID motif in the sequence KRAATKKSR within the I-II linker of rat CaV2.2. It has been proposed that phosphorylation by PKC of this threonine residue counteracts Gβγ binding to the I-II linker and thus counters inhibitory modulation (Zamponi et al., 1997; Hamid et al., 1999). However, this was subsequently found only to hold true for Gβ1 and not other Gβ subunits (Cooper et al., 2000). Furthermore, the sequence is not completely conserved in rabbit CaV2.2, which has alanine in place of threonine at the equivalent residue 422 (KRAAAKKSR in rabbit CaV2.2), although the same phenomenon of cross talk between G protein modulation and PKC activation occurs (Bertaso et al., 2003). Thus, the site(s) of phosphorylation by PKC responsible for the reversal of G protein modulation is not yet certain (Bertaso et al., 2003). It has been shown recently that Gαq binds to the C terminus of N-type calcium channels, to which PKC was also found to bind, and this colocalization may facilitate the phosphorylation of the N-type calcium channel (Simen et al., 2001). It is also possible that, if there is a reduction in membrane levels of phosphatidylinositol (4,5)-bisphosphate (PIP2) resulting from phospholipase C activation, this may play a role in the reversal or attenuation of G protein modulation, since PIP2 has been shown to regulate calcium channels (Wu et al., 2002). However, it remains controversial whether PIP2 levels are reduced substantially after activation of G protein-coupled receptors linked to phospholipase C, because no global change was observed in the heart (Nasuhoglu et al., 2002), but the reduction may be localized and also linked to increased synthesis of PIP2 (Zhao et al., 2001).
F. Is There a Role for the C Terminus in Calcium Current Inhibition by G Protein-Coupled Receptors?
A number of studies have suggested that the C terminus of certain CaV2 calcium channels is essential for their G protein modulation and indeed binds either Gβγ or Gα subunits (Qin et al., 1997; Furukawa et al., 1998a,b; Kinoshita et al., 2001). However, my group has found that the chimeric constructs containing domain I of CaV2.2 and the entire final three domains and C terminus of CaV1.2 is strongly G protein-modulated (Stephens et al., 1998b). Furthermore, truncation of the C terminus of CaV2.2 including the region homologous with that found to bind Gβγ in CaV2.3 (Qin et al., 1997) did not affect G protein modulation by GTPγS (Meza and Adams, 1998), and a similar truncation of CaV2.2 was found by another group to reduce, but not to abolish modulation by somatostatin receptor activation (Hamid et al., 1999). Thus, any direct role for the C terminus is probably a minor one.
V. Essential Role of Cavβ Subunits in G Protein Modulation of Calcium Channels
The identification of a QQIER motif, known to bind Gβγ dimers in other proteins, in the region of the α1 subunit I-II linker where the CaVβ subunit binds, suggested that the CaVβ subunit might be involved in G protein modulation.
A. Initial Evidence for the Role of CaVβ Subunits in G Protein Modulation in Native Neurons
To investigate the involvement of CaVβ subunits in G protein modulation, I first developed an antisense strategy to deplete dorsal root ganglion neurons of their CaVβ subunits by microinjection of an antisense oligonucleotide complementary to a region common to all β subunits (Berrow et al., 1995). Antisense knockdown of the CaVβ subunit by about 90% resulted in an enhancement of the ability of the GABA-B receptor agonist (-)-baclofen to inhibit the residual currents (Campbell et al., 1995b). We hypothesized from these results that there might be a functional interaction between activated G protein and VGCC β subunit for interaction with the relevant channels (Campbell et al., 1995b).
B. The Involvement of CaVβ Subunits in G Protein Inhibition of Heterologously Expressed Calcium Channels
The role of CaVβ subunits in G protein inhibition of expressed calcium channels has now been extensively examined (Bourinet et al., 1996; Qin et al., 1997; Roche and Treistman, 1998; Canti et al., 2000, 2001; Meir et al., 2000). In initial studies in Xenopus oocytes, there was reported to be less or even a complete loss of G protein inhibition following coexpression of a β subunit (Bourinet et al., 1996; Qin et al., 1997), although these studies only examined inhibition at a single potential and need to be interpreted with caution because of the presence of an endogenous oocyte CaVβ subunit. The result was interpreted in terms of a competition or displacement of CaVβ by Gβγ at an overlapping binding site (Bourinet et al., 1996). However, since β subunits shift the calcium current activation to more hyperpolarized potentials, it is inappropriate to measure G protein-mediated inhibition at a single potential. By studying the voltage dependence of receptor-mediated inhibition in Xenopus oocytes, my group has shown that this is a bell-shaped curve, peaking at about the voltage for 50% current activation (Canti et al., 2000). In the absence of coexpressed β subunits, the maximum amount of inhibition induced by activation of the coexpressed dopamine D2 receptor was about 70% at -10 mV. This curve is hyperpolarized in the presence of coexpressed β subunits, and the peak inhibition observed with β1b, β3, and β4 was little changed at 70, 62, and 59%, although it occurred at -20 mV, whereas with β2a coexpression, maximal inhibition was modestly reduced, being 51% at -10 mV (Canti et al., 2000). Thus, it is likely that this cannot represent a simple competition between CaVβ subunits and Gβγ dimers, but the interaction is dynamic and depends on the membrane voltage.
C. Does Gβγ Displace CaVβ Subunits?
In this section, the evidence will be assessed for two different views concerning the mechanism of inhibition of calcium channels by G proteins: 1) that CaVβ subunits do not dissociate during this process, which involves an allosteric rearrangement of the α1-β interaction associated with the voltage-dependent binding and unbinding of Gβγ:CaVα1-CaVβ + Gβγ ⇌ CaVα1-β-Gβγ as proposed in Meir et al. (2000); or 2) that G protein modulation is favored in the absence of Caβ subunits and opposed by the presence of β subunits, indicating that CaVβ and Gβγ compete for a single binding site on the α1 subunit, as in the reaction: CaVα1-CaVβ + Gβγ ⇌ CaVα1-Gβγ + CaVβ as proposed by Bourinet et al. (1996) and Qin et al. (1997). Such a reaction would either require transient formation of an intermediate ternary CaVα1-β-Gβγ complex or require that CaVβ dissociates before Gβγ binds, if they bind to the same site.
If Gβγ binding either displaced CaVβ or allosterically resulted in the physical dissociation of CaVβ, then the effects of the two species should oppose one another. Indeed, in many respects Gβγ dimers do appear to have the opposite effect from CaVβ subunits on calcium channel properties. All CaVβ subunits shift calcium channel activation to more hyperpolarized potentials (for review, see Birnbaumer et al., 1998) and Gβγ has the opposite effect (Bean, 1989). However, all CaVβ subunits except palmitoylated β2a hyperpolarize the steady-state inactivation by about 30 mV (Canti et al., 2000). In contrast, where it has been studied, little or no effect of G protein activation or Gβγ dimers has been observed on steady-state inactivation (Bean, 1989; Meir and Dolphin, 2002). This suggests that Gβγ does not simply displace CaVβ subunits and prevent their interaction with the channel.
For the CaV2 subfamily of channels, prepulse facilitation of G protein-modulated channels is thought to involve Gβγ unbinding from the channel, induced by depolarization, as described above. This finding can be used to test the hypothesis that there is a competition between Gβγ and CaVβ subunits, since if Gβγ unbinds during a prepulse, then CaVβ might be expected to bind in its place. If so, the rate of facilitation would be directly dependent on CaVβ concentration in the cytosol. Indeed, the rate of facilitation during a prepulse was markedly increased by the heterologous expression of all CaVβ subunits (Roche and Treistman, 1998; Canti et al., 2000), which might be construed as supporting this view.
My group therefore developed a means of testing this hypothesis by expressing increasing amounts of CaVβ3 cDNA with a constant amount of CaV2.2 cDNA in Xenopus oocytes. We first showed that there was a linear relationship between β3 cDNA injected and β3 protein expressed (Canti et al., 2001). The β3 subunit was used in these experiments because it is almost identical to the endogenous Xenopus β3 present in oocytes (Tareilus et al., 1997). We then performed an intracellular doseresponse curve for CaVβ subunits to examine the concentration dependence of the effect of β subunits to increase the facilitation rate (Canti et al., 2001). This experiment is illustrated in Fig. 10. At high CaVβ concentrations in oocytes (between about 20 and 100 ng of β3/oocyte), the facilitation rate during the depolarizing prepulse can be fit by a single fast exponential (Fig. 10, A and B), which we have interpreted as corresponding to the Gβγ off-rate from the channel that has CaVβ bound. The reason for this interpretation is that with these concentrations of β3 coexpressed, the steady-state inactivation of the CaV2.2 channels is fit by a single Boltzmann function, which is hyperpolarized compared with that for CaV2.2 expressed without a β subunit (Fig. 10C).
At intermediate CaVβ concentrations, the facilitation rate is not well fit by a single exponential (Canti et al., 2001), but can be fit by the sum of the same invariant fast exponential plus a slow exponential (Fig. 10, A and B). The value of the fast time constant, interpreted above as the Gβγ off-rate, is invariant over 100-fold change of β3 concentration and is therefore highly unlikely to involve any process requiring actual binding of β3 from the bulk solution. However, one aspect of the process does show a dependence on CaVβ concentration. The proportion of current showing the fast facilitation rate (Fig. 10D) shows exactly the same dependence on CaVβ protein concentration as the proportion of current with a hyperpolarized steady-state inactivation (see the biphasic steady-state inactivation curves at intermediate CaVβ concentrations in Fig. 10C). This agrees with our interpretation that the fast time constant of facilitation represents the behavior of a population of channels that has CaVβ bound. Reciprocally, the proportion of current showing a slow time constant of facilitation decreases as CaVβ concentration is increased, as does the proportion of current with a depolarized steady-state inactivation (Fig. 10C). We interpret this component as representing CaV channels without a bound CaVβ. Unlike the component with the fast time constant of facilitation, this slow component of facilitation has a time constant that does vary with CaVβ concentration (Fig. 10D). We have interpreted this finding as representing CaVβ subunit binding to the population of free channels during the depolarizing prepulse, after which Gβγ then unbinds rapidly with the invariant fast time constant (Canti et al., 2001). From this study we conclude that there is not a simple competition between Gβγ and CaVβ subunits, but rather that under normal circumstances, Gβγ dissociates at depolarized potentials from and rebinds at hyperpolarized potentials to channels that have CaVβ bound. Only under circumstances when CaVβ is limiting, which might rarely occur in native tissues, does CaVβ bind with higher affinity during depolarization and thence induces Gβγ unbinding.
Thus, CaVβ must have a higher affinity for the channels in their depolarized state. This phenomenon of depolarization-dependent displacement of the α1-β equilibrium toward increased CaVβ binding would only be observed under conditions where CaVβ is limiting, and it remains unknown whether this is ever the case in native tissues. In agreement with this interpretation, we could still observe significant tonic facilitation under conditions where Gβγ was minimized for CaV2.2 channels expressed in oocytes without a CaVβ subunit, in sharp contrast to the lack of tonic facilitation in the additional presence of exogenous β1b (Fig. 11) (Canti et al., 2000).
It was found that the reinhibition rate following a prepulse (for the CaV2.2/β1b/α2δ-1 combination) was increased as the concentration of Gβγ protein in the patch pipette was increased (Zamponi and Snutch, 1998), but in my group, we have observed that this rate is also increased slightly by overexpression of any CaVβ subunit together with CaV2.2, compared with the rate in the presence only of endogenous oocyte CaVβ (Canti et al., 2000). If there were direct competition between Gβγ and CaVβ for an overlapping binding site, then CaVβ should unbind during this process, before Gβγ rebinds. During the process of reinhibition following a prepulse, the depolarizing shift in activation and consequently the inhibition observed would result from CaVβ unbinding, rather than Gβγ binding, and therefore should not be dependent on Gβγ concentration. Elevation of CaVβ would be expected to slow the overall reinhibition rate, which is the opposite of what is observed. The conclusion must be that Gβγ preferentially rebinds to the CaVα1-β combination, and this species predominates when CaVβ is overexpressed.
Our model for the functional interplay between Gβγ dimers and CaVβ subunits does not support the idea that there is a simple competition between CaVβ and Gβγ for binding to the channel, or that CaVβ dissociates from the channel during G protein modulation, but rather that under normal conditions where the channels all have a CaVβ bound, Gβγ allosterically disrupts the effect of CaVβ on CaVα1 channels. Conversely, depolarization, such as occurs during a prepulse, results in a state-dependent conformational change between CaVα1 and CaVβ, which decreases the stability of Gβγ binding.
The observation that at some potentials G protein modulation is enhanced in oocytes in the absence of overexpressed CaVβ in Xenopus oocytes (Bourinet et al., 1996), or following antisense depletion of CaVβ subunits in sensory neurons (Campbell et al., 1995b), may be explained as follows. The slowed current activation in the presence of Gβγ is one of the main components of G protein-mediated inhibition, and is a reflection of the fact that Gβγ-bound channels either do not open upon depolarization until Gβγ dissociates, or show a very brief reluctant opening (Patil et al., 1996; Lee and Elmslie, 2000). Since a reduction of CaVβ produces slowing of overall facilitation rate of G protein-modulated channels during the prepulse, then the same will be true for the smaller degree of Gβγ unbinding that occurs during the test pulse. This is likely to be the reason that a reduction in CaVβ levels results in the observation of enhanced inhibition during a test pulse, or at least a shift in the voltage dependence of inhibition to more depolarized potentials (Canti et al., 2000). Conversely, an elevation of CaVβ reduces the inhibition observed. It should be noted that the direct effects of CaVβ subunits on inactivation during the test pulse, while not directly influencing the process of G protein modulation, will also influence the net amount of modulation exhibited (Meir et al., 2000; Meir and Dolphin, 2002).
In an expression system (COS-7 cells) in which (unlike Xenopus oocytes) no endogenous CaVβ subunit protein was detected by immunocytochemistry (Meir et al., 2000), coexpression of Gβγ with CaV2.2 in the absence of CaVβ resulted in calcium channel currents that were rapidly activating and not facilitated by a prepulse. This was observed both at the whole-cell level (Fig. 5B), and at the single-channel level (Fig. 6B). However, Gβγ did produce a small reduction in the current amplitude, compared with currents recorded in the absence of Gβγ. For all these sets of currents, their depolarized activation clearly showed that no CaVβ was associated. The additional presence of heterologously expressed CaVβ subunits was required for the relief of Gβγ-mediated inhibition by a depolarizing prepulse (Figs. 5A, 6A, and 7A) (Meir et al., 2000).
At the single-channel level, in the cell-attached patch mode, when only one channel is present, the effect of Gβγ and CaVβ can be compared in the absence of the confounding effect of CaVβ on the number of channels expressed. The main effect of coexpression CaVβ2a on CaV2.2 channel properties is an increase in the mean open time and a hyperpolarizing shift in the latency to first opening (Meir et al., 2000). For the model in which CaVβ is displaced by Gβγ to be correct, the currents in the presence of Gβγ should display the same properties as those in the absence of CaVβ, that is, the combination CaV2.2α1/Gβγ. This is not the case as the main effect of Gβγ in the presence of CaVβ is an increased latency to first opening. For the competition hypothesis to be true, if it is assumed that the N-type channels do not open before Gβγ unbinds, then before the first opening when the CaV2.2α1/β/Gβγ combination is coexpressed, only Gβγ should be bound; however, the channels do not show the same properties as the CaV2.2α1/Gβγ combination (Meir et al., 2000).
We concluded from that study that CaVβ subunits were essential for the process of facilitation or Gβγ dissociation. In the same system, receptor-mediated inhibition via activation of the D2 dopamine receptor was also examined. It was much reduced in the absence of coexpressed CaVβ subunits (Fig. 7B), and reversal of this inhibition by a 100-ms prepulse was lost, implying that in the absence of CaVβ subunits, Gβγ dimers are able to bind and produce a small non-voltage-dependent inhibition of the CaV2.2 current, but their unbinding is not influenced by voltage (Meir et al., 2000).
D. Potential Overlap of Determinants for CaVβ Subunit and Gβγ Subunit Function
There is overlap in the determinants for G protein modulation and CaVβ binding or function for all three of the sites discussed above: the I-II linker, the C terminus, and the N terminus of CaV2 calcium channels. The Gβγ binding site on the CaVα1 subunit intracellular I-II loop (De Waard et al., 1997; Zamponi et al., 1997) partially coincides with binding sites for auxiliary CaVβ subunits (Pragnell et al., 1994). However, the main amino acids that are critical for CaVβ subunit interaction are not within but adjacent to the QxxER consensus sequence implicated in Gβγ binding (Herlitze et al., 1996; De Waard et al., 1997).
Within the N terminus of CaV2.2 between amino acids 45 and 55, four individual point mutations (S48A, I49A, R52A, and R54A) were isolated, which significantly compromised modulation of CaV2.2 by G proteins (Canti et al., 1999). My group has subsequently shown that both the CaV2.2-R52,54A and CaV2.2-R52A constructs also exhibited compromised β2a-mediated retardation of inactivation as did CaV2.2-Q47A, which was shown previously to undergo normal Gβγ modulation (Stephens et al., 2000). Taken together with our initial study that identified this site (Canti et al., 1999), the results indicate that the CaV2.2 amino terminus contributes determinants for both CaVβ subunit and Gβγ dimer function. However, the differentiating effect of CaV2.2-Q47A suggests that although the overall region involved may partially coincide, the determinants are not identical.
A partial overlap in CaVβ subunit and Gβγ binding sites has also been proposed for the CaV2.3 carboxyl-terminal site (Qin et al., 1997). However, whereas deletion of the majority of this CaV2.3 site affected Gβγ modulation, it allowed retention of full sensitivity to β2a (Qin et al., 1997), suggesting that this is not the prime mediator of the β subunit response (see also Jones et al., 1998).
VI. Molecular Mechanism of G Protein-Mediated Inhibition
It is still not understood how Gβγ binding to the CaVα1 subunit results in inhibition of the calcium current. It is first necessary to establish the nature of the inhibitory effect at the level of a single calcium channel. It was initially suggested either that the G protein-bound channels opened with different gating properties (Kasai and Aosaki, 1989) or that there was a modal shift in gating (Delcour et al., 1993; Delcour and Tsien, 1993). However, the view that now prevails is that G protein-bound channels are reluctant to open, and that dissociation of bound G protein from the channel is required to convert them into willing channels (Bean, 1989; Elmslie et al., 1990). The simplest model would involve opening only the free and not the G protein-bound channel (for review, see Dolphin, 1991). At the single-channel level there is a prolonged latency to first opening in the presence of an agonist (Patil et al., 1996). However, once a channel had opened, no difference was observed on subsequent open probability or gating pattern compared with nonmodulated channels (Patil et al., 1996). This result is in agreement with the hypothesis that the delay to first opening is due to dissociation of the Gβγ dimers from the channel, allowing it to open, and that the G protein-bound channel does not open even with large depolarizations. This indicates either that the Gβγ binding is itself strongly voltage- or state-dependent or that Gβγ binds to a site on the channel that produces a voltage-dependent inhibition. More recently, it has been suggested that reluctant or Gβγ-bound channels can open, albeit with a low probability (Lee and Elmslie, 2000).
It is still unclear how many Gβγ dimers are required to bind to each CaV2 channel to produce inhibition. Models have suggested that more than one activated G protein may be bound per channel in a cooperative manner (Boland and Bean, 1993), but more recently it was suggested that a single Gβγ was bound per channel (Zamponi and Snutch, 1998).
The mechanism by which bound Gβγ prevents the channel from opening is also unknown. In one study, the effect of G protein activation on voltage sensor movement in CaV2.2 was examined. This revealed that GTPγS produced a depolarizing shift in the voltage dependence of charge movement that could be reversed by a large depolarizing prepulse and also induced the appearance of a slow component of “on” gating charge (Jones et al., 1997). The greatest effect was the large separation on the voltage axis between gating charge movement and channel opening. Thus, Gβγ is acting both to slow voltage sensor movement and to inhibit the subsequent transduction of this movement into channel opening (Jones et al., 1997). In partial agreement with this, from a study exploiting a spontaneous mutation in CaV2.2 channels, G177E in domain IS3, which converts channels into a form that behaves as if it is tonically G protein-modulated in the absence of Gβγ dimers, it was suggested that the normal role of Gβγ dimers is voltage sensor trapping (Zhong et al., 2001).
There are a number of potential mechanisms whereby the N terminus might exert its essential role in G protein modulation. Three possibilities will be considered here. The first possibility is that, bearing in mind the effect of the N terminus of CaV2.x channels on the actions of CaVβ subunits, it might form part of a complex CaVβ subunit binding pocket, into which Gβγ dimers could intercalate. However, the interaction of the N terminus with β subunits is unlikely to be of high affinity; as in a yeast two-hybrid assay, the N terminus did not interact with CaVβ subunits or with the I-II linker of CaV2.2 (Canti et al., 2001), although as described above (Section II.B.), the N terminus of CaV2.1 has been shown to bind to β4 subunits in overlay assays (Walker et al., 1999).
The second possible mechanism is that the N-terminal motif identified in CaV2.x channels might also form the effector of G protein modulation, as suggested in Canti et al. (1999). It might, for example, create a blocking particle, in a manner somewhat analogous to the ball and chain model of potassium channel block by the N terminus (MacKinnon et al., 1993). However, the N-terminal motif is not at the extreme N terminus, because it represents amino acids 44 to 55 in CaV2.2, and amino acids N-terminal to this motif (1–44 in CaV2.2) are not required for G protein modulation (Canti et al., 1999). The sequence of the N-terminal ball peptide in KV1.1 (Shaker B) is MAAVAGLYGLGEDRQHRKKQ, and there are no similarities with the N-terminal motif of CaV2.2 (YKQSIAQRART), apart from the presence of a number of essential positively charged residues and a KQ motif. In the case of KV channels, the ball peptide inserts into the inner vestibule of the pore of the open channel and produces inactivation by open-channel block. If this type of action is involved in G protein modulation of CaV2 channels, the inhibition is not an open-channel block mechanism; rather it is both retarding voltage sensor movement and preventing channel opening in response to the voltage sensor movement, and the effect is relieved by a large depolarization. One might envisage that the N-terminal motif is held in place, for example, to anchor the voltage sensor(s) by the binding of a Gβγ dimer, and its association is weakened by an altered interaction between the CaVα1 and CaVβ subunits induced by depolarization.
A third possibility is that the relevant part of the N terminus may form a PIP2 binding site, since RAR is similar to motifs containing positively charged residues in GIRK channels involved in binding the negatively charged head groups of PIP2, resulting in membrane association. In GIRK channels, PIP2 binding is thought to lead to association of residues on the N and C termini with the inner surface of the plasma membrane, producing a channel conformation that favors activation. The interaction of these regions of GIRK channels with PIP2 is found to be stabilized by Gβγ, and the presence of PIP2 in the membrane is a prerequisite for Gβγ modulation (Huang et al., 1998; Logothetis and Zhang, 1999; Zhang et al., 1999). If such a mechanism were to occur for calcium channel modulation, PIP2 might be expected to coregulate the channel together with Gβγ dimers. Indeed, this has been studied (Wu et al., 2002), and PIP2 was found to have a dual effect, both preventing calcium channel rundown in patches and producing an inhibitory modulation. Whether the inhibitory effect of PIP2 directly interacts with Gβγ modulation remains unclear.
VII. Recovery from G Protein-Mediated Inhibition
The speed of termination of GPCR-mediated inhibition of calcium currents depends on a number of factors. The rates of onset and offset of an agonist-mediated response are slower than the dissociation and reassociation rates of the activated G protein-channel complex, obtained by the prepulse protocol (Zhou et al., 1997). The onset of agonist-mediated inhibition using 10 μM noradrenaline was found to have a time constant of 0.7 s, whereas for reinhibition following a prepulse, the time constant was about 0.2 s. This difference is accounted for by the time taken for agonist binding to the receptor and for G protein activation. The rate of recovery from the agonist-mediated response (time constant of approximately 6 s for 10 μM noradrenaline) is very much slower than the facilitation rate, representing the dissociation of Gβγ measured during the depolarizing prepulse (Zhou et al., 1997). This discrepancy may be explained both by the off-rate of the agonist, which for some drugs may be very slow, the slower dissociation of Gβγ from the calcium channel at polarized rather than depolarized potentials (the basis for the voltage dependence of inhibition), and by the slow decay of the free Gβγ concentration, which determines the rebinding rate. This will depend on lateral diffusion in the membrane and reassociation of Gβγ and Gα-GDP, which will in turn be dependent on the rate of hydrolysis of activated Gα-GTP to Gα-GDP. The intrinsic hydrolysis rates of most heterotrimeric G proteins are much slower than the measured off-rate of this response. A family of RGS proteins has been identified that stimulates the GTPase activity of the Gα moiety of specific heterotrimeric G proteins (Watson et al., 1996; Dohlman and Thorner, 1997; García-Palomero et al., 2001). Overexpression of certain RGS proteins accelerated the off-rate of the response (Jeong and Ikeda, 2000). They also slowed the rate of recovery of inhibition after prepulse facilitation, indicating that they had reduced the level of free Gβγ by increasing the Gα-GDP available to rebind Gβγ dimers. Endogenous RGS proteins are likely to be involved in recovery from inhibition as expression of an RGS-insensitive Gαo in sympathetic neurons resulted in a dramatic slowing of the rate of recovery of calcium currents after inhibition by noradrenaline (Jeong and Ikeda, 2000). Of interest in this regard, my group has shown that the GTPase activity of Gαo in neuronal membranes is blocked by an antibody against CaVβ subunits (Campbell et al., 1995a). It remains to be determined whether there is an association between RGS proteins or G protein subunits and CaVβ subunits. It will also be fascinating to examine whether GPCRs are included in such macromolecular signaling complexes.
VIII. Conclusion
A number of experiments indicate that the CaV2 calcium channel α1 I-II linker is involved in the modulation of the CaV2 family of calcium channels by Gβγ dimers. However, several pieces of evidence suggest that this is not the main site involved in mediating the effects of Gβγ, since the N terminus is essential in this regard (Page et al., 1998; Canti et al., 1999). Goals for the future include elucidation of the molecular mechanism of modulation by Gβγ dimers since there is still little understanding of the way in which G protein binding is converted into an effect on latency of channel opening (Patil et al., 1996). It will also be of interest to evaluate whether the G protein α subunit plays a role in terminating the signal transduction process, which may be the case for GIRKs (Schreibmayer et al., 1996), and to examine the exact mechanism of the functional interaction between CaVβ subunits and Gβγ dimers in the inhibition of the CaV2 family of calcium channels.
Acknowledgments
The work from my laboratory described in this review has largely been funded by grants from the Wellcome Trust. I thank and acknowledge the contribution of members of my group past and present, from Rod Scott, with whom I started this work, through Liz Fitzgerald, Nick Berrow, Alon Meir, Karen Page, Carles Canti, Damien Bell, and many others who have made substantial contributions.
Footnotes
-
↵1 Abbreviations: VGCC, voltage-gated calcium channel; HVA, high-voltage-activated (channel); DHP, 1,4-dihydropyridine; AID, α interaction domain; PKC, protein kinase C; GST, glutathione S-transferase; GPCR, G protein-coupled receptor; GTPγS, guanosine 5′-O-(3-thiotriphosphate; GIRK, G protein-activated potassium channel; β-ARK1, β-adrenergic receptor kinase 1; RGS, regulators of G protein signaling; PIP2, phosphatidylinositol (4,5)-bisphosphate.
-
DOI: 10.1124/pr.55.4.3.
- The American Society for Pharmacology and Experimental Therapeutics
References
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵