Regulation of voltage-gated calcium channels by the ER calcium sensor STIM1

Abstract

The endoplasmic reticulum (ER) Ca²⁺ sensor STIM1, best-known for its essential role in triggering influx of extracellular Ca²⁺ via Ca²⁺-release-activated channels when ER stores become depleted, unexpectedly also regulates Ca²⁺ entry through voltage-gated Ca²⁺ channels. In response to a drop in ER luminal Ca²⁺ level, this ER membrane-spanning sensor can contact voltage-gated Ca²⁺ channels in the plasma membrane and thereby inhibit Ca²⁺ influx through them. This previously unappreciated, interaction between ER Ca²⁺ level and magnitude of Ca²⁺ influx via voltage-gated Ca²⁺ channels may turn out to powerfully impact Ca²⁺ signaling in excitable cells, including neurotransmitter release, structural and functional postsynaptic plasticity, and transcription factor translocation.

Introduction

Ca²⁺ is the lingua franca of intracellular communication. In neurons, transduction of electrical activity into important cytoplasmic Ca²⁺ signals is accomplished in large part through the activity of voltage-gated calcium channels (Ca_Vs). In keeping with such broad capability, the activity of Ca_Vs is made highly adjustable by assembling these channels within macromolecular signaling modules replete with regulatory machinery. Among the many forms of such regulation are several involving Ca²⁺ and Ca²⁺ signaling proteins, such as Ca²⁺-dependent inactivation of Ca_Vs, Ca_V-initiated Ca²⁺-induced Ca²⁺ release (CICR) from intracellular stores, and retrograde control of Ca_Vs by Ca²⁺-release channels (1).

In 2010, two groups reported the surprising and transformative finding that Ca_Vs are subject to strong regulation by the ER Ca²⁺ sensor, stromal interaction molecule-1 (STIM1) (2, 3). Previous work had shown that STIM1 is expressed in a wide range of tissues, including prominently and most famously in T lymphocytes and in other non-excitable tissues of lung, liver, kidney, as well as in excitable tissues of heart, skeletal muscle, smooth muscle, glandular tissue, and throughout the brain. This work had also shown that STIM1 is a single-pass ER membrane protein bearing a Ca²⁺ binding EF-hand in its N-terminal intraluminal region which allows STIM1 to report ER Ca²⁺ concentration ([Ca²⁺]_ER) (4, 5). Following release of Ca²⁺ from ER stores, for example in response to Gq-coupled receptor/inositol 1,4,5-trisphosphate receptor activity, the resulting depletion of Ca²⁺ from the ER results in unloading of Ca²⁺ from STIM1’s EF hand, activation and oligomerization of STIM1, and extension of the cytosolic domains of this complex across the gap between the ER and plasma membrane (PM). In breakthrough work that revealed the fundamental mechanism of stores-operated Ca²⁺ entry (SOCE) (Box1), the C-terminal domain of the extended STIM1 complex contacts and activates Ca²⁺ release-activated channels (CRAC) embedded in the plasma membrane (PM) (6–8), which allows influx of Ca²⁺ that both acts as a cytosolic signal and replenishes ER stores. It now appears that this same mechanism, involving STIM1 or its relative STIM2, is largely responsible for SOCE and regulation of [Ca²⁺]_ER in neurons (9–11) and other excitable cells (12). The 2010 reports from the Dolmetsch and Gill research groups that STIM1 regulates Ca_Vs as well as CRAC channels were wholly unexpected not only because Ca_Vs, in addition to CRAC channels, were discovered to be regulated by STIM1, but also because Ca²⁺ influx via Ca_Vs was found to be attenuated, rather than activated, by interaction with STIM1.

The ability of STIM1 to physically engage its PM signaling partners is supported by regions of close apposition of cortical ER with the PM. The intermembrane gap of such ER-PM junctions has been found to range from ~10 to 17 nm (13–15). This gap is narrow enough that the extended conformation of the cytosolic C-terminal domain of STIM1 can reach across it to the PM. In considering the dynamics, localization and restriction of subcellular Ca²⁺ signals it should be noted that ER-PM junctions are tiny, thin spaces—only ~0.1% the diameter of neuronal somata, ~1% the diameter of a dendrite, ~2% the diameter of a dendritic spine, and a little narrower than the neuronal synaptic cleft (~20 nm).

Reciprocal control of voltage-gated Ca²⁺ channels by STIM1

Neuronal Ca_Vs are composed of three subunits, one α₁, one α₂δ and one β subunit. These channels are functionally dominated by their single α₁ subunit, which forms the permeation pathway for Ca²⁺, detects membrane depolarization arising from action potentials or excitatory postsynaptic potentials, and undergoes conformational changes from closed to open to inactivated states. Neuronal Ca_Vs are variously specialized for neurotransmitter release (Ca_V2 subfamily, corresponding to P/Q-, N- and R-type channels), regulation of gene expression (Ca_V1, corresponding to L-type channels) and postsynaptic/somatic Ca²⁺ signaling (Ca_V1 and Ca_V2), or repetitive firing (Ca_V1.3 and the T-type Ca_V3 channels) (Table 1) (16). While depolarization is the only natural stimulus that directly causes Ca_Vs to open, their probability of opening is modulated by many factors including interaction with their own accessory α₂δ and β channel subunits (α₂δ1–4, β1–4), channel phosphorylation by multiple types of kinases, or binding of regulatory factors such as calmodulin or the βγ subunits of G_i/o proteins.

Table 1.

Neuronal voltage-gated Ca²⁺ channel subunits.

Subfamily	Functional type	Systematic name	α1 subunit
cav1	L-type	cav1.2	α11.2
	L-type	cav1.3	α11.3
cav2	P/Q-type	cav2.1	α12.1
	N-type	cav2.2	α12.2
	R-type	cav2.3	α12.3
Cav3	T-type	cav3.1	α13.1
	T-type	cav3.2	α13.2
	T-type	cav3.3	α13.3

Dolmetsch and colleagues (2) and Gill and colleagues (3) added to the already extensive list of Ca_V regulatory processes when they discovered that depletion of [Ca²⁺]_ER and consequent activation of STIM1 can attenuate the activity of voltage-gated Ca²⁺ channels. Using rat cultured cortical neurons (endogenous Ca_V1-,3 channels), the A7r5 vascular smooth muscle cell line (endogenous Ca_V1 channels), and HEK293 cells which lack endogenous Ca_Vs but were transfected with Ca_V1.2 L-type Ca²⁺ channels, these pioneering groups found that diverse pharmacological interventions resulting in depletion of ER Ca²⁺ stores attenuated depolarization-evoked intracellular Ca²⁺ ([Ca²⁺]_i) elevation and also reduced the amplitude of voltage-clamped Ca²⁺ current. Linking stores depletion-mediated inhibition of Ca_V activity to STIM1 were two pieces of evidence: first, overexpression of STIM1 in cortical neurons led to reduction in depolarization-driven rises in [Ca²⁺]_i, and reduced or eliminated Ca_V current in Ca_V1.2-transfected HEK293 cells. Second, STIM1 defective in Ca²⁺-binding (STIM1-D76A mutation) and hence chronically active, when overexpressed in A7r5 cells or Ca_V1.2-transfected HEK293 cells, caused constitutive reduction in Ca_V1.2 activity (3).

In cells with a well-characterized mechanism of STIM1 action, albeit with CRAC channels, would STIM1 also suppress Ca_V1.2 activity? Park et al. 2010 (2) tested this by transfecting Jurkat T lymphocytes with Ca_V1.2 subunits (α₁1.2, α₂δ1 and β1b), and found no depolarization-evoked rise in [Ca²⁺]_i, as though endogenous STIM1 had chronically shut off the exogenous Ca_Vs. In contrast, Jurkat T lymphocyte mutants that lack functional STIM1 displayed, when transfected with Ca_V1.2 subunits, depolarization-driven elevation in [Ca²⁺]_i. Overexpression of STIM1 in these mutants restored wild-type functionality, eliminating Ca_V1.2-dependent [Ca²⁺]_i rises. Finally, knockdown of STIM1 in wild-type Jurkat T lymphocytes allowed Ca_V1.2-dependent [Ca²⁺]_i rises, confirming that STIM1 suppresses Ca_V1.2 activity in these cells, as it appeared to do in cortical neurons and other cells

The Gill and Dolmetsch groups also showed, using HEK293 cells transfected with Ca_V1.2 and STIM1, that immunoprecipitation of STIM1 coimmunoprecipitated Ca_V1.2, and vice versa, suggesting that these two proteins can interact directly. Using neuroblastoma cells that express both of these partners natively, STIM1 was shown to be associated with immunoprecipitated Ca_V1.2 as well. Molecular dissection of this interaction revealed that STIM1’s cytosolic CRAC activation domain (CAD; residues 342–448), which binds and activates Orai-based CRAC channels, also interacts with Ca_V1.2 channels, though in this case inhibiting channel activity. Thus deletion of CAD from STIM1 prevented STIM1 suppression of [Ca²⁺]_i rises, whereas expression of the CAD peptide alone strongly suppressed Ca_V1.2 activity (2, 3). Moreover, immunoprecipitation of CAD coimmunoprecipitated Ca_V1.2, indicating direct interaction (2).

Within STIM1’s CAD domain, substitution of alanine at two adjacent positions, L₃₄₇ and Q₃₄₈, prevented STIM1 activation of CRAC channels (17), and this double substitution mutant also failed to inhibit Ca_V1.2 channels (3). In contrast, STIM1Δ441–448 is incapable of activating CRAC channels, but this mutation retains the ability to inhibit Ca_V1.2. Thus the site on STIM1 that inhibits Ca_V1.2 overlaps but is not identical to that which activates Orai-based CRAC channels (3).

Which parts of Ca_V1.2 are contacted by STIM1? In HEK293 cells expressing the Ca_V1.2 α₁ subunit alone, stores depletion was still able to suppress Ca²⁺ influx, showing that the channel accessory α₂δ and β subunits are not needed for depletion-driven inhibition of Ca_V1.2 activity (3). Functional requirement for the Ca_V1.2 C-terminal domain was demonstrated by the finding that Ca_V1.2 lacking its C-terminal domain supported depolarization-dependent rises in [Ca²⁺]_i that were unaffected by the presence of STIM1 (2). Coimmunoprecipitation experiments with HEK293 cells overexpressing STIM1, or its CAD domain, along with individual intracellular domains of Ca_V1.2 revealed that STIM1 binds to a region in the cytoplasmic distal C-terminal of Ca_V1.2, encompassed by residues 1809–1908 (2). This specific region of the Ca_V1.2 C terminal region is distinct from other C-terminal sites, such as those that bind calmodulin, calmodulin kinase II, and A-kinase anchoring proteins; sites involved in protein kinase A-dependent enhancement of channel activity, including sites of Ser/Thr phosphorylation; or domains involved in Ca²⁺-dependent inactivation.

Finally, in cultured hippocampal neurons, overexpression of STIM1 greatly reduced the fraction of Ca_V1.2 channels on the cell surface (2) via a dynamic-dependent mechanism. Thus STIM1 can acutely inhibit Ca_V1.2 activity via stores depletion and it can also chronically reduce Ca_V1.2 activity by promoting cellular internalization of these channels.

STIM1 inhibition of L-type Ca_Vs in neurons

In addition to its key role in non-excitable cells, STIM1 is clearly involved in neuronal ER Ca²⁺ signaling and Ca²⁺ homeostasis. For example, in cerebellar Purkinje neurons, STIM1 has been shown to mediate mGluR1 activation of TRPC3 channels and thereby support slow excitatory synaptic transmission that is involved in cerebellar motor control (10). STIM1 is present throughout neurons, as ER pervades dendritic shafts, projects into large dendritic spines in a labile fashion, and also helps form a synaptopondin-bearing structure in the base of spines known as the spine apparatus (18–21). And it turns out, STIM1 exerts strong modulatory control over Ca_V activity in neurons in order to regulate diverse cellular functions.

Using cultured hippocampal neurons, Dittmer et al. discovered that the neurotransmitter glutamate, acting through N-methyl-D-aspartate (NMDA) receptor-dependent depletion of ER Ca²⁺ stores, could initiate STIM1 inhibition of L-type Ca²⁺ current (22) (Figure 1). In their experiments, responses of voltage-clamped L-type current to bath applied glutamate (15 s), or of simultaneously-imaged cytoplasmic and ER Ca²⁺ to uncaging of glutamate (60 laser flashes in 1 min) near a single dendritic spine, were measured. In either case, NMDA receptors initiated CICR and neuronal depolarization, the latter in turn activating L-type Ca²⁺ channels and hence additional CICR. Completing the negative feedback loop, stores depletion-activated STIM1 inhibited subsequent Ca²⁺ entry through L-type Ca²⁺ channels. RNAi knockdown of STIM1 prevented inhibition, and inhibition was restored by co-expression of an RNAi-insensitive STIM1.This group also used Förster resonance energy transfer (FRET) to gauge interaction between overexpressed, fluorescent protein-labeled Ca_V1.2 and STIM1, which showed that the time course of interaction between these partners reflected the time course of inhibition of the channel. Previous work had shown that Ca²⁺ influx through NMDA receptors can trigger CICR from the ER, including in dendritic spines (23–26). It seems possible, or even likely however, that other pathways that release Ca²⁺ from the ER may, in other experimental conditions or neuronal cell types, also initiate STIM1 inhibition of L-type channels.

Figure 1.

Activation by glutamate of STIM1 feedback inhibition of L-type Ca_Vs in neurons. (a) Activation of NMDA receptors (NMDAR) by glutamate (plus glycine). NMDAR activity (step 1) depolarizes the plasma membrane, activating Ca_V1 channels (L channels) and thereby (step 2) generating Ca²⁺ influx. NMDAR activity also directly provides (step 2) influx of Ca²⁺. Listed on the right of the panel are approximate concentrations of Ca²⁺ in the extracellular space, and prior to influx and release of Ca²⁺, in the ER-PM junctional space and in the ER luminal space. (b) Combined influx of Ca²⁺ via NMDARs and Ca_V1 channels (step 2) triggers a large release of Ca²⁺ from stores (step 3, CICR process), resulting in loss of Ca²⁺ from its binding site within STIM1’s ER lumen-localized EF hand and thus activation of STIM1. Activated STIM1 translocates along the ER membrane, clusters, binds to Ca_V1 channels, and (step 4) reduces the probability of Ca_V1 channel opening in response to subsequent depolarizing stimuli.

Singh and colleagues studied STIM1 regulation of Ca_V1.3-dependent rhythmic firing in dopaminergic neurons of the substantia nigra (27). In these neurons, stores depletion reduced Ca²⁺ current and neuronal firing rate, and also activated TRPC1 channels, which are known to be able to function as stores-operated channels (12). In TRPC1 knock-out neurons, firing rate was elevated above wild-type level, and it could not be reduced by stores depletion. In addition, SOCE was largely eliminated and stores depletion could no longer inhibit Ca²⁺ current. Anti-STIM1 was found to coimmunoprecipitate Ca_V1.3, and stores depletion increased anti-STIM1 pull-down of Ca_V1.3.

Using SH-SY5Y neuroblastoma cells, these workers showed that knockdown of TRPC1 strongly attenuated Ca²⁺ influx following stores depletion, as in dopamine neurons, consistent with the idea that TRPC1 channels function in both of these kinds of cells as a SOC (27). TRPC1 knockdown also increased Ca²⁺ current amplitude, but in cells in which Ca_V1.3 was also knocked down, TRPC1 knockdown had no effect on Ca²⁺ current amplitude, indicating that TRPC1 inhibited Ca_V1.3 activity. Like TRPC1 knockdown, STIM1 knockdown strongly attenuated TRPC1-mediated Ca²⁺ influx following stores depletion, and increased Ca²⁺ current density. Store depletion promoted increased interaction among all three of STIM1, TRPC1 and Ca_V1.3, as measured in reciprocal coimmunoprecipitation experiments. The results suggest that store depletion drives STIM1 binding and activation ofTRPC1, and TRPC1 additionally facilitates STIM1 interaction with, and inhibition of, Ca_V1.3. This is reminiscent of the observation by Gill and colleagues that Orai-based CRAC channels facilitated STIM1 interaction with Ca_V1.2 (3). In the case of TRPC1, however, it is curious that block of Ca²⁺ flux through TRPC1 channels also blocked depletion-mediated, STIM1-dependent inhibition of Ca_V1.3, as though Ca²⁺ itself is involved. In any event, STIM1- and TRPC1-dependent inhibition of Ca_V1.3 channels is proposed to be neuroprotective, by reducing the firing frequency and thus Ca²⁺ load in dopaminergic neurons (27).

Roles for STIM1 regulation of Ca_Vs in neuronal physiology

Beyond limiting Ca²⁺ load and thereby minimizing the risk of cell death in dopaminergic neurons (27), in what other physiological tasks is STIM1 control of Ca_Vs involved? Dittmer and colleagues (22) found that Ca_V1.2-dependent nuclear translocation of the NFATc3 transcription factor was attenuated by STIM1 inhibition of L-type channels, suggesting that this negative feedback pathway may be involved in control of neuronal gene expression. They also found, in spines stimulated via glutamate uncaging at a frequency that initiates spine enlargement and long-term potentiation of post-synaptic responses (28), that spine ER volume was increased in a STIM1- and L channel-dependent manner. This response was initiated by activation of NMDA receptors, in agreement with previous findings (25).This aligns with findings that STIM1 knockout impairs long-term potentiation (29), and that STIM1 overexpression impairs long-term depression and improves contextual learning (30).

The evidence to date indicates that STIM1 directly regulates (inhibits) only L-type Ca_Vs in neurons, but will that prove true? Ryan and colleagues have reported that STIM1, activated by stores depletion, reduces action potential-driven presynaptic, Ca_V2-dependent Ca²⁺ influx and synaptic vesicle exocytosis (31). How activated STIM1 is coupled to Ca_V2-dependent Ca²⁺ influx, and hence exocytosis, is unexplored, but direct interaction with Ca_V2 channels is an attractive hypothesis advanced by these authors. In this proposed mechanism, as Ca_V2 channels circulate between release zones and other regions of the presynaptic terminal, activated STIM1 could bind and trap those traveling outside of release zones, thereby reducing the number of channels available to occupy release zones, reducing action potential-driven Ca²⁺ influx, and reducing synaptic vesicle exocytosis.

Future directions

Specifics of the mechanism of interaction of STIM proteins with Ca_Vs remain to be elucidated: How does STIM1 binding to a floppy C-terminal tail inhibit opening of Ca_Vs, and might other binding domains be involved? Are there ER-PM junctions specialized to support STIM1 inhibition of Ca_Vs? Other questions in need of answers include: what kinds of Ca_Vs can be inhibited (directly) by STIM1 other than the known Ca_V1.2 and Ca_V1.3 L-type channels? In neurons, how might gene expression controlled by nuclear translocation of the NFAT transcription factors be coordinately, and reciprocally, regulated by STIM1-Orai versus STIM1-Ca_V1.2? How do Ca_V1.2 and STIM1 work together to control ER content in spines undergoing synaptic plasticity? By what mechanism(s) might STIM1 regulate presynaptic Ca_V2 function? The answers to these and other stimulating questions await.

Box 1.

Molecular components that support stores-operated Ca²⁺ entry (SOCE)

IP3R1, IP3R2, IP3R3	inositol trisphosphate receptors, release ER Ca2+
RYR1, RyR2, RyR3	ryanodine receptors, also release ER Ca2+
STIM1, STIM2	high, and low, affinity sensors of [Ca2+]ER; activate stores-operated channels (SOCs)
Orai1, Orai2, Orai3	Ca2+-release activated Ca2+ (CRAC) channels, a class of SOCs
TRPC1, other TRPCs can function as STIMI -activated SOCs

Highlights.

• Key in stores-operated Ca²⁺ entry, STIM1 also controls voltage-gated Ca²⁺ channels

• STIM1, sensing a drop in ER Ca²⁺, binds and inhibits voltage-gated Ca²⁺ channels

• STIM1 and voltage-gated Ca²⁺ channels together control dendritic spine ER content

• STIM1 tunes voltage-gated Ca²⁺ channel-dependent transcription factor translocation

• STIM1 inhibits transmitter release, perhaps by voltage-gated Ca²⁺ channel inhibition