Ca²⁺-dependent modulation of voltage-gated Ca²⁺ channels: analysis in native and heterologous expression systems

Abstract

Background

Voltage-gated (Ca_v) Ca²⁺ channels are multi-subunit complexes that play diverse roles in a wide variety of tissues. A fundamental mechanism controlling Ca_v channel function involves the Ca²⁺ ions that permeate the channel pore. Ca²⁺ influx through Ca_v channels mediates feedback regulation to the channel that is both negative (Ca²⁺-dependent inactivation, CDI) and positive (Ca²⁺-dependent facilitation, CDF).

Scope of Review

This review highlights general mechanisms of CDI and CDF with an emphasis on how these processes have been studied electrophysiologically in native and heterologous expression systems.

Major Conclusions

Electrophysiological analyses have led to detailed insights into the mechanisms and prevalence of CDI and CDF as Ca_v channel regulatory mechanisms. All Ca_v channel family members undergo some form of Ca²⁺-dependent feedback that relies on CaM or a related Ca²⁺ binding protein. Tremendous progress has been made in characterizing the role of CaM in CDI and CDF. Yet, what contributes to the heterogeneity of CDI/CDF in various cell-types and how Ca²⁺-dependent regulation of Ca_v channels controls Ca²⁺ signaling remain largely unexplored.

General Significance

Ca²⁺ influx through Ca_v channels regulates diverse physiological events including excitation-contraction coupling in muscle, neurotransmitter and hormone release, and Ca²⁺-dependent gene transcription. Therefore, the mechanisms that regulate channels, such as CDI and CDF, can have a large impact on the signaling potential of excitable cells in various physiological contexts.

Voltage-gated (Ca_v) Ca²⁺ channels are multi-subunit complexes that play diverse roles in a wide variety of tissues. Multiple Ca_v channels have been characterized (Ca_v1.x-Ca_v3.x, Table 1), which are comprised mainly of a pore-forming α₁ subunit and for Ca_v1 and Ca_v2 channels, auxiliary Ca_vβ and α₂δ subunits [1]. Ca_v channels are generally named according to the identity of the α₁ subunit such that Ca_v1.2 channels are those containing the α₁1.2 subunit. Mutations in the genes encoding Ca_v α₁ cause severe human disorders including migraine, deafness, epilepsy, autism, cardiac arrhythmia, malignant hyperthermia and periodic paralysis [2–4]. Ca_v channels mediate an inward Ca²⁺ current that not only depolarizes the cell membrane potential but also provides an intracellular Ca²⁺ signal that can activate gene transcription, protein phosphorylation, and neurotransmitter release. Therefore, factors that modulate Ca_v channel properties can significantly affect cellular excitability and signal transduction.

Table 1.

Classification and expression of voltage-gated Ca²⁺ channels¹

CavX.X	α1 subunit2	Current3	Primary tissue distribution
1.1	α1S	L-type	Skeletal muscle
1.2	α1C	L-type	Heart, nervous system, endocrine glands
1.3	α1D	L-type	Heart, nervous system, endocrine glands, inner ear
1.4	α1F	L-type	Retina, T-cells
2.1	α1A	P/Q-type	Nervous system
2.2	α1B	N-type	Nervous system
2.3	α1E	R-type3	Nervous system
3.1	α1G	T-type	Nervous system, heart
3.2	α1H	T-type	Nervous system, heart, adrenal gland
3.3	α1I	T-type	Nervous system

¹See [134] for references;

²according to former nomenclature, currently “α₁x.x”;

³classical nomenclature based on native currents;

³Ca_v2.3 does not account for all R-type currents in neurons [135, 136]

A fundamental mechanism controlling Ca_v channel function involves the Ca²⁺ ions that permeate the channel pore. Ca²⁺ influx through Ca_v channels mediates feedback regulation to the channel that is both negative (Ca²⁺-dependent inactivation, CDI) and positive (Ca²⁺-dependent facilitation, CDF). CDI and CDF have been described mainly for “high voltage-activated” Ca_v1 and Ca_v2 Ca²⁺ channels both in native cell-types and heterologous expression systems. CDI and some forms of CDF depend on calmodulin (CaM) binding to the pore-forming Ca_v α₁ subunit. This review will highlight current understanding of the general mechanisms of CDI and CDF with an emphasis on how these processes have been studied electrophysiologically in native and heterologous expression systems.

Ca²⁺-dependent inactivation (CDI)

Inactivation represents a non-conducting state of many ion channels that is favored by multiple mechanisms. For Ca_v channels, sustained or very positive depolarization promotes voltage-dependent inactivation (VDI). However, Ca_v channel inactivation can also occur via a Ca²⁺-dependent mechanism, which is evident as a faster decay of the Ca_v current when Ca²⁺ rather Ba²⁺ is used as the charge carrier during sustained depolarizations (Fig. 1a). This was first demonstrated in voltage clamp recordings of Paramecium [5, 6], in which Ca²⁺ influx through Ca_v channels regulates reorientation of cilia and swimming in reverse [7]. In addition to faster inactivation of Ca²⁺ currents (I_Ca) compared to Ba²⁺ currents (I_Ba), inactivation correlated with the amplitude of the peak I_Ca. This hallmark of CDI is readily studied in double-pulse voltage-protocols, where the effects of a conditioning (inactivation-inducing) prepulse are reflected in the amplitude of a subsequent test current evoked at a single voltage. A plot of the normalized test I_Ca amplitude vs. prepulse voltage is bell-shaped, with greatest inactivation at the prepulse voltage eliciting the maximal inward Ca²⁺ current (Fig. 1b). By contrast, inactivation of I_Ba using this protocol should be relatively minor, increasing monotonically with the membrane potential. For such measurements of CDI, the duration of the conditioning prepulse should be short enough to promote greater CDI than VDI, which generally occurs on a longer timescale for most Ca_v1 and Ca_v2 channels.

Figure 1.

Characterization of CDI of Ca_v2.1 channels in voltage-clamp recordings. a) Comparison of inactivation of I_Ca (black trace) and I_Ba (grey trace) in HEK293T cells transfected with Ca_v2.1 (α₁2.1, β_2A and α2δ). Currents are evoked by 1-s pulses from −80 mV to 0 mV. Extracellular solution contains 10 mM Ca²⁺ or Ba²⁺ and intracellular solution contains 0.5 mM EGTA. Right, inactivation is measured by dividing residual current amplitude (I_res) by the peak amplitude (I_peak). b,c) Double pulse protocol for measuring CDI. Left, I_Ca is evoked by a 10-ms test pulse after a conditioning prepulse to various voltages as indicated. Right, normalized I_Ca for the test and prepulse are plotted against prepulse voltage. With physiological Ca²⁺ buffering (0.5 mM EGTA, b) but not with higher levels of Ca²⁺ buffering (10 mM BAPTA, c), maximal CDI of the test current occurs at prepulse voltages evoking peak inward current.

In their recordings of Paramecium, Brehm and Eckert also showed that strong buffering of intracellular Ca²⁺ greatly inhibited inactivation of I_Ca, suggesting that Ca²⁺ accumulation in the cell was necessary for CDI ([5, 6]; Fig. 1b,c). However, as will be discussed, the extent to which Ca²⁺ chelators suppress CDI varies between Ca_v channel classes. Nevertheless, these classical studies of Paramecium defined approaches for characterizing Ca_v channel CDI, which has proven a fundamental form of negative feedback in various cell types and organisms [8–11]. The mechanisms and properties of CDI vary between Ca_v channel classes and between cell-types, which may have important physiological consequences.

Ca_v1.2

Ca_v1 channels mediate L-type Ca²⁺ currents in muscle, nerve, and endocrine cells (Table 1). Of the four classes of Ca_v1 channels, Ca_v1.2 is the most ubiquitously expressed [12]. Ca_v1.2 regulates excitation-contraction coupling in the heart, which may explain why genetic inactivation of this channel in mice causes embryonic lethality [13]. In ventricular myocytes, Ca_v1.2 mediates L-type I_Ca that contributes to the plateau depolarization (phase 2) of the action potential. Electrophysiological recordings of these cells show that Ca_v1.2 undergoes strong CDI [14–17]. Single-channel analyses of rat myocytes support a model in which Ca²⁺ entry through Ca_v1.2 channels shifts their gating to a mode with lower open probability [18, 19]. The fact that Ca_v1.2 CDI can be measured in single-channel recordings suggests that CDI requires local Ca²⁺ near the channel pore rather than global Ca²⁺ elevations due to the opening of multiple open channels.

It is now well-established that CDI involes calmodulin (CaM) binding to the cytoplasmic C-terminal domain (CT) the Ca_v1.2 α₁ subunit (α₁1.2) [20]. CaM binds to a consensus site, the IQ-domain, in the proximal CT of α₁1.2. Ca²⁺ ions permeating the channel pore bind to the associated CaM, which produces a conformational change supporting CDI. The molecular details underlying CaM interactions with Ca_v channels have been covered in a several excellent reviews [20–22] and will not be a focus here. In whole-cell recordings, CaM inhibitors such as calmidazolium do not prevent CDI of Ca_v1.2 in cardiac myocytes [19], but such inhibitors may not antagonize CaM that is tightly associated with some effectors. Overexpression of CaM mutants that cannot bind Ca²⁺ not only inhibit CDI of Ca_v1.2 in ventricular myocytes, but also significantly prolongs cardiac action potentials [23]. Since excessively long action potentials can lead to cardiac arrhythmia, the ability of CDI to control action potential duration may protect against aberrant cardiac excitability.

CDI has also been characterized for L-type currents in neuronal and neuroendocrine cells that likely express both Ca_v1.2 and Ca_v1.3 [24]. However, the functional significance of Ca_v1 CDI in these cell-types is not entirely clear. Ca_v1 channels regulate Ca²⁺-dependent gene transcription which depends on CaM binding to the IQ-domain, but not on Ca_v1 CDI [25]. Under pathological conditions, such as in epilepsy, Ca_v1 CDI may be neuroprotective by preventing excitotoxic Ca²⁺ overloads in select neurons [26].

Ca_v1.3

While Ca_v1.2 and Ca_v1.3 may be expressed in many of the same cell-types in the brain and heart, Ca_v1.3 activates more rapidly and at negative voltages compared to Ca_v1.2 [27–30]. These properties allow its contribution to threshold depolarizations that promote spontaneous firing in the sinoatrial node, although Ca_v1.2 is also expressed in this tissue [31, 32]. In addition, Ca_v1.3 mediates ~90% of the whole-cell I_Ca in cochlear inner hair cells [33, 34]. Presynaptic Ca_v1.3 channels at the specialized “ribbon synapses” in inner hair cells conduct Ca²⁺ ions that trigger glutamate release at the first synapse in the auditory pathway [33]. The importance of Ca_v1.3 for hearing and cardiac pacemaking is underscored by the sinus bradycardia and deafness phenotypes in mice lacking Ca_v1.3 and humans with a loss-of-function mutation in the CACNA1D gene encoding α₁1.3 [35].

While Ca_v1.3 exhibits intense CDI when expressed heterologously, CDI is more limited in inner hair cells [28, 29, 34]. Whole-cell and perforated patch recordings from auditory hair cells in various species reveal I_Ca that inactivates only ~30–40% in 1 second while I_Ca in HEK293T cells transfected with Ca_v1.3 inactivates 80–90% within the same timeframe (Fig. 2a,b), although the extent of CDI in hair cells can vary with temperature and between species [36–38]. Differences in CDI of Ca_v1.3 between hair cells and other cell-types could be accounted for by alternative splicing of α₁1.3 which yields one variant lacking the IQ domain that exhibits little CDI. Since this variant is mainly expressed in outer hair cells, it is unlikely to account for the slow CDI in inner hair cells [39]. An alternate hypothesis is that Ca²⁺-binding proteins related to CaM oppose CDI of Ca_v1.3 in inner hair cells (See below).

Figure 2.

CDI of Ca_v1 channels. I_Ca (black) or I_Ba (grey) were evoked by 1-s depolarizing pulses in HEK293T cells transfected with Ca_v1.3 (a) or Ca_v1.4 (c) or in mouse inner hair cells (IHC), which express predominantly Ca_v1.3 (b).

Ca_v1.4

While the consensus site(s) for binding to CaM are relatively conserved between Ca_v1 channel subclasses, some intriguing differences in CDI have been noted. For example, Ca_v1.4, which is the primary Ca_v channel in retinal photoreceptors, undergoes little CDI [40, 41] (Fig. 2c). Similar to the role of Ca_v1.3 in inner hair cells, the ability of Ca_v1.4 to support sustained presynaptic L-type I_Ca helps maintain tonic glutamate release at ribbon synapses. CaM can bind to the Ca_v1.4 α₁ subunit IQ domain, but this interaction may be disrupted by an inhibitory domain (ICDI: inhibitor of CDI) in the distal C-terminus [42, 43]. This domain is weakly conserved in other Ca_v1 channel subclasses, such that its transfer to the α₁1.2 or α₁1.3 nearly nullifies the normally strong CDI of Ca_v1.2 and Ca_v1.3 [42, 43]. Multiple mutations in the CACNA1F gene encoding α₁1.4 cause congenital stationary night blindness type 2 [2], one of which (K1591X) causes premature truncation of α₁1.4 and deletion of the ICDI. As a consequence, Ca_v1.4 (K1591X) shows enhanced CDI [42], which could cause vision deficits due to a loss-of-function in photoreceptor transmission.

Ca_v1.1

In skeletal muscle, Ca_v1.1 channels are the “dihydropyridine receptors (DHPRs)” that mediate excitation/contraction coupling. The voltage-sensing function of Ca_v1.1 is directly coupled to activation of Ca²⁺ release by ryanodine receptors in the sarco-endoplasmic reticulum, which initiates muscle contraction. Although Ca²⁺ conductance by Ca_v1.1 is not required for excitation/contraction coupling, Ca_v1.1 channels do mediate measurable L-type I_Ca in skeletal muscle cells [44]. Compared to Ca_v1.2, Ca_v1.1 undergoes modest CDI [45], which may be due to sequence differences in the IQ-domains that impair CaM binding and/or the affinity of bound CaM for Ca²⁺ [46–48]. In particular, Y1657 and K1662 in the rabbit cardiac α₁1.2 are replaced by H1532 and M1537 in the rabbit skeletal muscle α₁1.1. Substituting Y1657 and K1662 in α₁1.2 with the corresponding residues of α₁1.1 weaken CaM binding and eliminate CDI of Ca_v1.2 channels in transfected HEK293T cells [46]. However, additional residues that distinguish the IQ domain of α₁1.1 and α₁1.2, while not directly contacting CaM, alter its Ca²⁺ binding affinity. When bound to the IQ-domain of α₁1.2, the C- and N-terminal lobes of CaM bind Ca²⁺ with ~5-fold higher affinity than when bound to the α₁1.1 IQ-domain [47]. This latter result could explain why Ca_v1.1 CDI is weaker and more strongly inhibited by BAPTA than Ca_v1.2 CDI [45]. Sequence differences in the α₁1.1 IQ domain are probably not the only determinants of slow CDI in skeletal muscle. The expression of recombinant Ca_v1.2 channels in dysgenic myotubes, which lack functional Ca_v1.1 channels, results in L-type currents that lack CDI [12]. Thus, as for Ca_v1.3 in cochlear hair cells, factors that inhibit CDI of Ca_v1 channels may be endogenously expressed in skeletal muscle. The significance of slow CDI for Ca_v1.1 is unknown but may be important for preventing refractoriness of the voltage-sensing capabilities of Ca_v1.1 for excitation/contraction coupling [46].

Ca_v2 channels

Previous efforts to characterize CDI of Ca_v2 channels by heterologous expression did not reveal CDI [49] in part because of high concentrations of intracellular EGTA that are typically used for electrophysiological recordings of Ca_v1 channels. However, with a relatively low (0.5 mM) concentration of EGTA in intracellular recording solutions, CDI of Ca_v2.1 channels in transfected HEK293T cells is evident as significantly faster inactivation of I_Ca compared to I_Ba [50] (Fig. 1a). High concentrations (10 mM) of EGTA or BAPTA, which spare CDI of Ca_v1.2 and Ca_v1.3, completely block Ca_v2.1 CDI in transfected cells (Fig. 1c, [51, 52]). These results suggest that Ca_v2.1 CDI depends on global elevations in Ca²⁺ while Ca_v1.2 CDI depends on local Ca²⁺ elevations near the channel pore. Consistent with this hypothesis, unlike for Ca_v1.2 [18], CDI for Ca_v2.1 is not evident in single channel recordings [53]. In heterologous expression systems, the local Ca²⁺ sensitivity of Ca_v1.2 and Ca_v1.3 CDI contrasts with the global Ca²⁺ sensitivity of CDI for all Ca_v2 channels [54].

Similar to Ca_v1.2 and Ca_v1.3, CDI for Ca_v2 channels involves CaM binding to an IQ-like domain in the Ca_v2 α₁ C-terminal domain. For Ca_v2.1, an additional CaM-binding domain (CBD) C-terminal to the IQ-like domain is also involved. Mutations or deletions of both of these sites impact CDI [50–52, 55]. Besides this difference in CaM binding sites between Ca_v1.2 and Ca_v2.1, CDI for these channels relies on different Ca²⁺-binding lobes of CaM. This conclusion is based on electrophysiological experiments where CaM mutants unable to bind Ca²⁺ in the N- or C-lobes were co-transfected with Ca_v channels in HEK293T cells. These studies demonstrate a requirement for the CaM C-lobe for Ca_v1.2 CDI whereas the CaM N-lobe is needed for Ca_v2.1 CDI [52, 55, 56]. Structural analyses of CaM bound to α₁1.2, α₁2.1, α₁2.2, or α₁2.3 peptides containing the IQ domain indicate that distinctions in how CaM associates with key residues within and/or outside of the IQ domain may underlie the differences in Ca_v1.2 and Ca_v2 CDI [57–60].

A second CaM-binding determinant (NSCaTE) in the N-terminal domain of α₁1.2 and α₁1.3, that is not present in Ca_v2 α₁ subunits, contributes to the difference in Ca²⁺ buffer sensitivity of CDI for Ca_v1 and Ca_v2 channels. Transfer of this region to the Ca_v2.2 α₁ subunit (α₁2.2) allows for significant CDI in the presence of 10 mM BAPTA. The underlying mechanism may involve distinct interactions of the N- and C- terminal lobes of CaM with the Ca_v1 α₁ NSCaTE and IQ domain, respectively [61].

Although intensely studied in heterologous expression systems, Ca_v2 CDI has been infrequently characterized for Ca_v2 channels in neurons. Ca²⁺ influx through Ca_v2 channels, particularly Ca_v2.1 and Ca_v2.2, initiates neurotransmitter release at many central synapses. Fluorometric analysis of Ca²⁺ influx in rat brain synaptosomes suggested CDI of P-type currents, which was suppressed by decreasing the extracellular Ca²⁺ concentration [62]. CDI of presynaptic Ca²⁺ channels was also demonstrated in patch clamp recordings of presynaptic nerve terminals in the rat neurohypophysis [63] and for P-type currents at the Calyx of Held synapse in the rat auditory brainstem [64]. Because of the steep dependence of neurotransmitter release on presynaptic Ca²⁺ concentrations, Ca_v2 channel CDI could significantly inhibit neurotransmission. In support of this hypothesis, Ca_v2.1 CDI can cause short-term depression at the Calyx of Held and in transfected sympathetic neurons [64, 65].

Ca²⁺-dependent facilitation (CDF)

Facilitation is an enhanced form of channel opening that can be achieved through multiple mechanisms. CDF is a positive feedback regulation of further Ca²⁺ entry through Ca_v1 and Ca_v2 channels (CDF), which can significantly impact cell physiology in a variety of contexts. Although other forms of facilitation have been described for various Ca_v family members, CDF has mainly been described for native and heterologously expressed Ca_v1.2, Ca_v1.3, Ca_v2.1, and Ca_v3 channels.

Ca_v1.2

In cardiac ventricular myocytes, Ca_v1.2 CDF helps amplify intracellular Ca²⁺ signals coupled to contraction of the heart [66–68]. During a train of depolarizations, I_Ca progressively increases in amplitude, reaching a steady state by the fifth pulse [69]. Facilitation of I_Ca is Ca²⁺-dependent in that it can be induced by photolysis of caged-Ca²⁺ compounds [70] and is abolished by substitution of extracellular Ca²⁺ with Ba²⁺ or Sr²⁺ [69]. CDF can still be observed with 20 mM EGTA or 5 mM BAPTA in intracellular recording solutions, suggesting that Ca_v1.2 CDF depends on very local increases in Ca²⁺ near the channel [69].

Like CDI, CDF for Ca_v1.2 channels also depends on CaM, which activates CaM-dependent protein kinase II (CaMKII). Two mechanisms have been proposed for the role of CaMKII in Ca_v1.2 CDF. In the first, CaMKII binds to the C-terminal domain of α₁1.2 near the IQ-domain, which may position the kinase for phosphorylating two nearby serine residues (S1512/S1570 in the mouse cardiac α₁1.2 [71, 72]. In transgenic mice in which these residues were mutated to alanine, L-type I_Ca shows slower recovery from inactivation and enhanced steady state inactivation, which is consistent with a loss of CDF. CDF was not completely abolished in ventricular myocytes from these mice, suggesting that other processes contribute to CDF (see below). In vivo telemetry in these mice revealed electrocardiograms with shorter QT intervals indicative of faster repolarization of cardiac action potentials. Since excessive CaMKII activity such as in heart failure causes QT prolongation and arrhythmia [73], the results suggest that Ca_v1.2 CDF, while normally beneficial for improving the force-frequency relationship of excitation/contraction coupling, may contribute to heart disease under pathological conditions.

A second route by which CaMKII enhances CDF is through interactions with the auxiliary Ca_v β_2A subunit. CaMKII binds to and phosphorylates Ca_v β_2A at threonine 498 (rat β_2A), which enhances a mode of gating (mode 2) characterized by increased Ca_v1.2 channel open probability [74–76]. Overexpression of wild-type Ca_v β_2A in ventricular myocytes causes Ca²⁺ overload and early after-depolarizations [77], which are premature depolarizations between cardiac action potentials that can cause arrhythmia. Moreover, rapid pacing, which would induce CDF, caused premature death of Ca_v β_2A-overexpressing ventricular myocytes. By contrast, overexpression of Ca_v β_2A with mutation of threonine 498 to alanine (or leucine 493 mutation that prevents CaMKII binding to Ca_v β_2A) reduced Ca_v1.2 Ca²⁺ entry and early afterdepolarizations and also prevent Ca_v1.2 mode 2 gating [77]. These results confirm that Ca_v1.2 CDF can contribute to abnormal excitability that leads to cardiac arrhythmia and a key role for CaMKII/Ca_v β_2A interactions in this process.

An important feature of CDF to consider in whole-cell recordings is the activation of CaMKII by repetitive Ca²⁺ spikes, which stimulates its autophosphorylation at threonine 286 and autonomous activity independent of Ca²⁺/CaM [78]. If L-type I_Ca is the Ca²⁺ source for CaMKII activation, autonomous activity should be more efficiently generated by short repetitive pulses rather than sustained depolarizations. In rat ventricular myocytes, CDF can be evoked with 150-ms pulses from −80 to 0 mV delivered at a frequency of 0.5 Hz. With this protocol, I_Ca increases in amplitude ~20% by the 4^th pulse [76]. In theory, double-pulse voltage protocols such as those used to measure CDF of Ca_v2.1 (see below) could also be applied for analyses of Ca_v1.2 CDF. However, the voltage of the conditioning prepulse should not be so depolarized as to evoke voltage-dependent facilitation (VDF). Prepulses up to +160 mV have been used to demonstrate a role for CaMKII in Ca_v1.2 facilition [72]. However, little Ca²⁺ influx would be generated during such positive prepulses, due to the small driving force for Ca²⁺ influx near the reversal potential for Ca²⁺ in most electrophysiological recordings. Thus, the measured facilitation with such protocols is likely VDF rather than CDF [72].

While both CDF and CDI characterize Ca_v1.2 channels in cardiac myocytes, recombinant Ca_v1.2 channels expressed in Xenopus oocytes and HEK 293 cells show primarily CDI and very little CDF, even during repetitive stimuli [79, 80]. CDF for heterologously expressed Ca_v1.2 channels is only evident when CDI is inhibited, such as with α₁1.2 subunits containing I-A mutation in the IQ-domain [79]. Coexpression of the CaM-like protein, CaBP1, which also inhibits Ca_v1.2 CDI, also allows for Ca_v1.2 CDF that is independent of CaMKII [80, 81]. Thus, the presence of overt CDF that is not masked by CDI in cardiac myocytes may depend on cell-type specific differences in the processing of Ca_v1.2 channels or on cellular factors which are more abundant in cardiac cells than in heterologous expression systems.

Ca_v1.3

CaMKII is also implicated in CDF of Ca_v1.3 but through a mechanism distinct from that for Ca_v1.2. Ca_v1.3 channels do not exhibit CDF when expressed alone or with CaMKII in HEK293T cells. However, when cotransfected with CaMKII and densin-180, a CaMKII-interacting protein highly enriched at excitatory synapses in the brain [82, 83], Ca_v1.3 channels undergo CDF during repetitive depolarizations [84]. Densin-180 binds via its PDZ domain to the distal C-terminus of α₁1.3 and therefore may help scaffold CaMKII to the Ca_v1.3 channel complex. All three proteins coimmunoprecipitate from brain lysates, which suggests that densin 180 may permit CaMKII-dependent Ca_v1.3 CDF in neurons [84]. CaMKII activation in dendritic spines is mediated by Ca_v1 channels during stimulation protocols that produce long-term potentiation [85]. Thus, densin and CaMKII association with Ca_v1.3 may augment Ca²⁺ signals that trigger CaMKII activation and participation in long-term synaptic plasticity.

Ca_v2.1

At the Calyx of Held synapse, Ca_v2.1 channels initially undergo CDF and then CDI during high-frequency depolarizations, which can influence short-term facilitation and depression of the excitatory postsynaptic responses [64, 86–89]. This facilitation is Ca²⁺-dependent in that it is only seen for I_Ca and not I_Ba and reduced by high intracellular EGTA [86]. Similar biphasic CDF and CDI was also reported for Ca_v2.1 channels transfected into superior cervical ganglion neurons [65]. Here, CDF and CDI were also associated with short-term facilitation and depression of synaptic transmission [65]. The neurophysiological significance of short-term facilitation due to CDF is not entirely clear, but may ensure reliable transmission by offsetting depression [88].

Ca_v2.1 channels in cerebellar Purkinje neurons also undergo CDF as well as CDI during stimulation with action potential waveforms (Fig. 3a,b; [90–94]), but CDF tends to balance the effects of CDI in a way that maintains relatively constant I_Ca amplitudes [92]. In Purkinje neurons, P/Q-type I_Ca is strongly coupled to Ca²⁺-activated K⁺ channels that regulate Purkinje cell firing rates [95]. Fluctuations in I_Ca due to overt CDF or CDI may cause instabilities in firing that could prevent normal correlations of firing rate with synaptic activation [92]. The significance of CDF was also shown for Ca_v2.1 channels in presynaptic parallel fiber inputs to Purkinje neurons. Here, Ca_v2.1 channels bearing a mutation associated with familial hemiplegic migraine in humans (S218L) do not undergo CDF because they may be in a basally facilitated state. As a consequence, there is no short-term synaptic facilitation at this synapse [94]. Thus, the current data suggest that CDF plays a functionally more significant role for Ca_v2.1 in presynaptic terminals than in postsynaptic compartments of neurons.

Figure 3.

Characterization of CDF using repetitive and double pulse voltage-clamp protocols. (a) I_Ca is evoked in mouse cerebellar Purkinje neurons by a train of action potential (AP) waveforms at 100 Hz. CDF is evident as an initial increase in the amplitude of I_Ca above the baseline level (dashed line). (b) CDF is measured by plotting the peak amplitude of test currents normalized to the first in the train for I_Ca(filled circles) or I_Ba (open circles). Shown are results from AP trains in mouse cerebellar Purkinje neurons and Ca_v2.1-transfected HEK293T cells, and square test pulses (−80 mV to +10 mV, 100 Hz) for Ca_v2.1-transfected HEK293T cells. (c) Double-pulse protocol measures CDF of Ca_v2.1 in transfected HEK293T cells. Left, I_Ca was evoked by test currents evoked before (P1) or after (P2) a conditioning prepulse (Pre). The prepulse-induced current is not sampled in the representative current trace. The ratio of P2:P1 current amplitude (Fractional current) is plotted against prepulse voltage for I_Ca (filled circles) and I_Ba (open circles).

The mechanisms underlying CDF have been studied extensively for Ca_v2.1 channels in transfected HEK293T cells [52, 55]. Ca_v2.1 CDF depends on CaM binding to the IQ-domain of α₁2.1, with a reliance primarily on the C-lobe of CaM in contrast to the CaM N-lobe dependence of CDI [52, 55, 57, 59]. In transfected HEK293T cells, Ca_v2.1 CDF can be measured using 2–5-ms square pulses or action potential waveforms at a frequency of 100 Hz (Fig. 3b). CDF is plotted as the amplitude of each test current normalized to that for the first in the train. CDF can also be measured with double-pulse protocols with short (~20–50 ms) conditioning prepulses that evoke significant inward I_Ca (i.e., −80 mV to +10 mV). With this protocol, facilitation can be measured as the test current amplitude or charge integral normalized to a current in the absence of a prepulse. When plotted against prepulse voltage, facilitation of I_Ca, but not I_Ba, should be bell-shaped, reflecting its dependence on the amount of inward I_Ca during the prepulse (Fig. 3c).

In single-channel recordings, Ca_v2.1 CDF is associated with an enhanced open probability of I_Ca above that seen for I_Ba, rather than an acceleration of activation kinetics [53]. The detection of CDF but not CDI in single-channel recordings [53] suggests a different reliance on local vs. global Ca²⁺ signals for CDF and CDI, respectively. Consistent with this hypothesis, CDF, unlike CDI, is spared by 10 mM EGTA in whole-cell patch clamp recordings of transfected HEK293T cells [23, 55]. Analyses of CDF of transfected Ca_v2.1 channels would therefore benefit by high Ca²⁺ buffering conditions which would limit the competing effects of CDI. However, since the decay of CDF is significantly accelerated by high Ca²⁺ buffering in transfected cells [51], paired-pulse protocols should employ a relatively short interval between the conditioning prepulse and the test pulse for maximal CDF to be achieved.

Ca_v3

Ca_v3 channels are the major Ca_v channels in adrenal glomerulosa cells which secrete aldosterone in response to angiotensin II and physiological increases in K⁺ [96]. Angiotensin II stimulates Ca²⁺ release from intracellular stores, activation of CaMKII, and potentiation of Ca_v channel opening by K⁺-induced depolarization [97]. The facilitation involves CaMKII binding to and phosphorylation of the cytoplasmic domain linking domains 2 and 3 of α₁3.2 [98]. Phosphorylation of S1198 (human α₁3.2) negatively shifts voltage-dependent activation of Ca_v3.2 channels in HEK293 cells. This residue is not present in α₁3.1 and accounts for the absence of CaMKII regulation of Ca_v3.1 [98, 99]. Cell-attached single-channel recordings of T-type currents in adrenal glomerulosa cells show that raising the bath Ca²⁺ concentration enhances Ca_v3.2 channel open probability, which is accompanied by increased frequency, rather than longer duration, of channel openings [100]. In addition, Ca_v3.2 CDF is blocked by CaMKII inhibitors and mimicked by constitutively active CaMKII. Since CDF would cause Ca_v3.2 channels to open at more negative membrane potentials, CDF may help set the level of steady state Ca²⁺ influx and angiotensin II-stimulated aldosterone secretion. In heart failure, which is characterized by heightened activation of the renin-angiotensin system, Ca_v3.2 CDF may exacerbate the pathological consequences of excessive aldosterone secretion in heart failure, which include structural and functional changes in the heart and vasculature. Since Ca_v3.2 channels are highly expressed in the brain [101], CDF may also contribute to the roles of Ca_v3.2 in regulating firing patterns in multiple classes of neurons.

Factors that influence CDI and CDF in native cell-types

CDI and CDF, even for a given Ca_v subclass, can be highly variable between cell-types (Fig. 2a,b). Multiple factors can influence the extent to which Ca_v channels undergo CDI and CDF, which can complicate mechanistic inquiries as to their relative significance.

Alternative splicing of Ca_v α₁

Post-translational splicing increases the functional diversity afforded by the 10 genes encoding Ca_vα₁ subunits in the human, rat, and mouse genomes. Because of the large size of these genes (~50 exons), the number of splice variants possible for a single Ca_v α₁ subunit may be greater than 1000 [102]. Like the α₁1.3 splice variant lacking the IQ domain in outer hair cells [39], all of the splice variations that have been found to impact CDI or CDF involve alterations in the C-terminal domain. For example, alternative splicing at exon 37 (human α₁2.1) produces two variants (EFa and EFb) with amino acid substitutions in an EF-hand like domain upstream from the CaM binding IQ-like domain and CBD [103]. The EFa but not the EFb variant exhibits CDF [104]. When combined with alternative splicing out of the final exon 47, EFb causes CDF that is blocked by high concentrations of BAPTA only in cells with large current density. Apparently, exclusion of exon 47 permits CDF in EFb containing Ca_v2.1 channels that is driven by global, rather than local, increases in Ca²⁺ [104]. How the absence of exon 47 when combined with EFb transforms the Ca²⁺-sensitivity of Ca_v2.1 CDF remains to be elucidated.

The distal C-terminal domain of α₁1.3 has also been shown to regulate CDI of Ca_v1.3. Alternative splicing of exons 42 and 42A produces α₁1.3 variants that extend >100 amino acids beyond the IQ-domain (exon 42) or are truncated just after the IQ-domain (exon 42A) [28, 105]. When compared in transfected HEK293T cells, the short exon 42A variant exhibits stronger CDI than the long exon 42 variant. FRET experiments suggested that the exon 42 sequence interfered with CaM binding to the proximal α₁1.3 C-terminal region including the IQ-domain [106]. Thus, this alternatively spliced exon plays a similar role in α₁1.3 as the distal CT in α₁1.4 [42, 43], except that the inhibitory effect on CDI is significantly less for Ca_v1.3 than for Ca_v1.4.

Ca_v β subunits

Because inactivation of Ca_v channels can occur by voltage- or Ca²⁺-dependent mechanisms (VDI or CDI), factors that minimize the former can enhance detection of the latter. For example, CDI is quite minimal in Ca_v2.1 channels that contain the Ca_vβ_1b subunit due to the strong VDI that typifies this subunit composition [107]. Inclusion of the Ca_vβ_2A, which produces Ca_v2.1 channels that show significantly weaker VDI than with Ca_vβ_1b, results in robust CDI [51]. Considering the heterogeneous expression patterns of different Ca_vβ subunits [108], the magnitude of Ca_v2.1 CDI may vary between tissues and cell-types.

For Ca_v1.2 and Ca_v1.3, CDI is less dependent on the identity of the Ca_vβ subunit than for Ca_v2.1 channels. However, Ca_v1.2 CDF due to CaMKII may require Ca_vβ₂ or Ca_vβ₁ subunit. Biochemical studies indicate that autophosphorylated CaMKII stably associates with Ca_vβ₂ and β₁ but not β₃ or β₄ [74]. Although CaMKII can phosphorylate all four Ca_v β subunits to a similar extent, a conserved LXRXXS/T motif present in Ca_vβ₂ and β₁ but not β₃ or β₄ may be required for functionally relevant phosphorylation of Ca_vβ which ensures stability of the Ca_vβ/CaMKII complex [74]. To what extent Ca_vβ subunits influence the extent of Ca_v1.2 CDF awaits electrophysiological comparisons of CDF of Ca_v1.2 channels bearing distinct Ca_vβ subunits.

CaBPs

Due to their similarities with CaM, some neuron-specific CaBPs can substitute for CaM in binding to Ca_v channels, which can significantly alter CDF and CDI [109]. Like CaM, these CaBPs have four EF-hand Ca²⁺ binding domains, at least one of which is nonfunctional [110]. Other structural distinctions including N-terminal myristoylation and a longer central helical domain may confer some CaBPs with the ability to differentially modulate Ca_v channels compared to CaM [111]. For example, CaBP1 is colocalized to some extent with presynaptic Ca_v2.1 channels. Like CaM, CaBP1 binds to the CBD of α₁2.1, but does not support CDF and causes intense inactivation independent of Ca²⁺ [112]. Ca_v2.1 also can interact with other CaBPs, such as neuronal Ca²⁺ sensor-1 (NCS-1), which has inhibitory effects on P/Q-type current amplitude in Xenopus oocytes and adrenal chromaffin cells [113]. However, injection of purified NCS-1 protein into presynaptic nerve terminals at the Calyx of Held synapse promotes CDF of P/Q-type currents, and activity-dependent facilitation P/Q-type currents at this synapse can be prevented by injection of NCS-1-inhibitor peptides [114]. Therefore, the effects of NCS-1 on Ca_v2.1 may depend on the cellular and subcellular context in which they are coexpressed. A third CaBP that modulates Ca_v2.1 is visinin-like protein-2 (VILIP-2), which is highly expressed in the neocortex and hippocampus and undergoes Ca²⁺-dependent association with the plasma membrane in neurons and other cell-types [115]. When cotransfected with Ca_v2.1 in mammalian cells, VILIP-2 does not affect CDF, but inhibits CDI [116]. These effects of VILIP-2 may involve displacement of CaM from the CBD, although both CBD and IQ-like domain of α₁2.1 are required for the association of VILIP-2 with the channel [116]. How VILIP-2 and CaBP1 can have such opposing effects on Ca_v2.1 function is not entirely clear, but may involve structural distinctions between the two CaBPs and/or how they interact with Ca_v2.1 [117]. The remarkably divergent actions of CaBPs on Ca_v2.1 may increase the range of presynaptic P/Q-type Ca²⁺ signals, thus contributing to heterogeneous forms of short-term plasticity at different synapses.

Ca_v1 channels are also differentially regulated by CaBPs. Unlike its inhibitory effects on Ca_v2.1, CaBP1 prolongs Ca_v1.2 Ca²⁺ currents and completely abolishes CDI [80, 118]. Structural analyses indicate that the opposing actions of CaM and CaBP1 on Ca_v1.2 may depend on differences in how the two interact with the IQ-domain [81]. In addition, CaBP1 binds to a site in the N-terminal domain of the α₁1.2, which can affect both CDI and VDI [118, 119]. Interestingly, a CaBP1 variant, caldendrin, causes more modest suppression of Ca_v1.2 CDI than CaBP1 through interactions solely with the IQ-domain and not the N-terminal site in α₁1.2 [120]. CaBP1 and caldendrin associate and colocalize with Ca_v1.2 in somatodendritic domains of neurons [80, 120], and so may help fine-tune and boost postsynaptic Ca_v2.1 Ca²⁺ signals that regulate gene transcription and neuronal excitability.

CaBP1 is highly expressed in inner hair cells where Ca_v1.3 channels exhibit anomalously slow CDI at room temperature compared to in transfected cells at room temperature [29, 34, 121, 122]. Consistent with a role for CaBP1 in the mechanism, coexpression of CaBP1 with Ca_v1.3 in HEK293T cells strongly inhibits CDI [121, 122]. In addition to CaBP1, CaBP2, 4, and 5 are also expressed in inner hair cells. However, when Ca_v1.3 is cotransfected with each of these CaBPs in HEK293T cells, the effects of CaBP1 on slowing CDI and also VDI most closely reproduce the slowly inactivating properties of Ca_v1.3 in inner hair cells. In addition, genetic inactivation of CaBP4 in mice has only a minor effect on CDI and no impact on hearing [121]. Confirmation of the role of CaBP1 in CDI suppression awaits similar analyses in inner hair cells from CaBP1 null mice.

In addition to these CaM-like CaBPs, Ca²⁺ binding proteins such as parvalbumin (PV) and calbindin D-28k (CB), which serve more as Ca²⁺ buffers [123], also regulate Ca_v CDI. PV and CB are highly expressed in Purkinje neurons where I_Ca is largely mediated by Ca_v2.1 [124]. Since Ca_v2.1 CDI is sensitive to Ca²⁺ buffers that dampen global Ca²⁺ elevations [51], one would expect that PV and CB might similarly naturally suppress CDI. This hypothesis was partially supported in transfected HEK293T cells [125]. If PV and CB inhibited CDI in Purkinje neurons, then CDI should be increased in neurons from mice with genetic inactivation of the genes encoding PV and CB. However, Purkinje neurons isolated from mice lacking PV and CB showed a decrease in Ca_vβ_2A expression and consistently, increased VDI but not CDI compared to in wild-type neurons [90]. This alteration may have been a compensatory response to prevent excessive Ca_v2.1 Ca²⁺ influx in neurons lacking the Ca²⁺ buffering capacity of PV and CB. Thus, whether Ca²⁺ buffering proteins limit CDI of neuronal Ca_v2.1 channels remains to be proven.

Despite evidence that Ca_v1 CDI in transfected HEK293T cells is relatively insensitive to Ca²⁺ buffers [54, 61], CDI of Ca_v1 channels in thalamocortical neurons was shown to be inhibited by inclusion of PV and CB in intracellular whole-cell recording solutions in isolated thalamocortical neurons [126]. Further evidence for a role for Ca²⁺ buffering proteins in regulating Ca_v CDI comes from recordings of surviving hippocampal granule cells isolated from the dentate gyrus of patients with mesial temporal lobe epilepsy. These cells have reduced levels of CB, which correlates with increased CDI of Ca_v1 L-type currents [26]. Unlike for Ca_v1 channels in transfected cells, Ca_v1 CDI in these surviving granule cells is inhibited by high intracellular concentrations of BAPTA or EGTA [127]. Why Ca_v1 CDI is sensitive to Ca²⁺ buffers in these neurons but not in HEK293T cells is unclear, but could be related to the presence of Ca_v1 α₁ splice variants that may lack the NSCaTE [61]. Alternatively, CaBPs with distinct Ca²⁺ binding affinities compared to CaM might compete with CaM for binding to the Ca_v1 IQ domain or NSCaTE in neurons. Regardless of the mechanism, the ability of PV and/or CB to suppress Ca_v1 CDI in neurons may be necessary for supporting Ca²⁺ signals of sufficient duration to activate signaling pathways underlying activity-dependent synaptic plasticity [128]. Moreover, pathological decreases in neuronal Ca²⁺ buffering, such as in temporal lobe epilepsy [26], may be neuroprotective in increasing Ca_v1 CDI and reducing excitotoxic Ca²⁺ overloads.

Other cellular factors

Cells such as myocytes and neurons have complex subcellular features that could also significantly impact the extent to which Ca_v channels undergo CDI and CDF. This seems to be particularly true in skeletal and cardiac muscle cells in which Ca_v1 channels are tightly clustered at junctional membranes in association with intracellular ryanodine receptors (RYRs) in the sarco-endoplasmic reticulum. In cardiac myocytes, Ca²⁺ influx through Ca_v1.2 channels enhances Ca²⁺-induced Ca²⁺ release through activation of RYRs. Due to the tight spatial coupling of RYRs and Ca_v1.2 in junctional membranes, Ca_v1.2 CDI is enhanced by RYR-mediated Ca²⁺ signals [129]. Such functional coupling between RYRs and Ca_v1.2 may also occur in neurons [130], since blockade of RYRs in thalamocortical neurons decreased Ca_v1 CDI [131].

The cytoskeleton has also been implicated as a determinant of CDI in neurons [132, 133]. It is hypothesized that Ca²⁺ influx through Ca_v channels destabilizes cytoskeletal elements, which promotes CDI. In support of this hypothesis, pharmacological stabilization of microfilaments and microtubules reduce Ca_v CDI in neurons [127, 133]. How cytoskeletal dynamics participate in CDI regulation is unclear, but may involve Ca_v α₁ interactions with Ca_v β subunits and/or cytoskeleton-associated proteins [24].

Summary and remaining questions

Electrophysiological analyses have led to detailed insights into the mechanisms and prevalence of CDI and CDF as Ca_v channel regulatory mechanisms. All Ca_v channel family members undergo some form of Ca²⁺-dependent feedback that relies on CaM or a related Ca²⁺ binding protein. Tremendous progress has been made in characterizing the role of CaM in CDI and CDF of Ca_v1 and Ca_v2 channels at the molecular and atomic scales. Yet, numerous ambiguities remain. First, what determines the variability in CaM-dependent CDI/CDF in native cell types? At one extreme, Ca_v1 channels in photoreceptors and inner hair cells have mechanisms that prevent CDI, to allow for sustained Ca²⁺-dependent exocytosis required for sensory transmission. By contrast, intense CaM-dependent CDI of Ca_v1 channels in cardiac myocytes is necessary for curtailing cardiac action potentials and maintaining cardiac rhythmicity. Molecular and biochemical analyses of Ca_v splice variation and Ca_v/protein interactions may reveal mechanisms underlying the heterogeneous presentations of Ca_v CDI/CDF in a variety of cell-types.

Second, how does CDI/CDF control cellular excitability and other Ca²⁺ signaling events? Due to the depolarizing influence of Ca_v currents, CDI could directly inhibit neuronal excitability. However, due to the tight coupling of Ca_v channels with Ca²⁺-dependent K⁺ (K_Ca) channels, CDF could also inhibit excitability through stronger activation of K_Ca channels. Detailed consideration of CDI/CDF within computational models of cellular and network excitability is a necessary first step towards understanding the broader impact of these forms of Ca_v channel modulation in defined physiological contexts.

Highlights.

• Electrophysiological analysis have revealed detailed mechanisms of Ca_v channel modulation.

• Ca_v channels undergo CDI/CDF that relies on Ca²⁺ binding proteins like calmodulin.

• The heterogeneity of CDI/CDF in various cell-types remains largely unexplored.