Effect of intracellular Ca2+ and action potential duration on L-type Ca2+ channel inactivation and recovery from inactivation in rabbit cardiac myocytes

Ca(2+) current (I(Ca)) recovery from inactivation is necessary for normal cardiac excitation-contraction coupling. In normal hearts, increased stimulation frequency increases force, but in heart failure (HF) this force-frequency relationship (FFR) is often flattened or reversed. Although reduced sarcoplasmic reticulum Ca(2+)-ATPase function may be involved, decreased I(Ca) availability may also contribute. Longer action potential duration (APD), slower intracellular Ca(2+) concentration ([Ca(2+)](i)) decline, and higher diastolic [Ca(2+)](i) in HF could all slow I(Ca) recovery from inactivation, thereby decreasing I(Ca) availability. We measured the effect of different diastolic [Ca(2+)](i) on I(Ca) inactivation and recovery from inactivation in rabbit cardiac myocytes. Both I(Ca) and Ba(2+) current (I(Ba)) were measured. I(Ca) decay was accelerated only at high diastolic [Ca(2+)](i) (600 nM). I(Ba) inactivation was slower but insensitive to [Ca(2+)](i). Membrane potential dependence of I(Ca) or I(Ba) availability was not affected by [Ca(2+)](i) <600 nM. Recovery from inactivation was slowed by both depolarization and high [Ca(2+)](i). We also used perforated patch with action potential (AP)-clamp and normal Ca(2+) transients, using various APDs as conditioning pulses for different frequencies (and to simulate HF APD). Recovery of I(Ca) following longer APD was increasingly incomplete, decreasing I(Ca) availability. Trains of long APs caused a larger I(Ca) decrease than short APD at the same frequency. This effect on I(Ca) availability was exacerbated by slowing twitch [Ca(2+)](i) decline by approximately 50%. We conclude that long APD and slower [Ca(2+)](i) decline lead to cumulative inactivation limiting I(Ca) at high heart rates and might contribute to the negative FFR in HF, independent of altered Ca(2+) channel properties.     

Calmodulin is the Ca2+ sensor for Ca2+ -dependent inactivation of L-type calcium channels

Elevated intracellular Ca2+ triggers inactivation of L-type calcium channels, providing negative Ca2+ feedback in many cells. Ca2+ binding to the main alpha1c channel subunit has been widely proposed to initiate such Ca2+ -dependent inactivation. Here, we find that overexpression of mutant, Ca2+ -insensitive calmodulin (CaM) ablates Ca2+ -dependent inactivation in a "dominant-negative" manner. This result demonstrates that CaM is the actual Ca2+ sensor for inactivation and suggests that CaM is constitutively tethered to the channel complex. Inactivation is likely to occur via Ca2+ -dependent interaction of tethered CaM with an IQ-like motif on the carboxyl tail of alpha1c. CaM also binds to analogous IQ regions of N-, P/Q-, and R-type calcium channels, suggesting that CaM-mediated effects may be widespread in the calcium channel family.     

Intracellular Cs+ activates the PKA pathway,revealing a fast,reversible, Ca2+-dependent inactivation of L-type Ca2+ current

Inactivation of the L-type Ca²⁺ current (I_CaL) was studied in isolated guinea pig ventricular myocytes with different ionic solutions. Under basal conditions, I_CaL of 82% of cells infused with Cs⁺-based intracellular solutions showed enhanced amplitude with multiphasic decay and diastolic depolarization-induced facilitation. The characteristics of I_CaL in this population of cells were not due to contamination by other currents or an artifact. These phenomena were reduced by ryanodine, caffeine, cyclopiazonic acid, the protein kinase A inhibitor H-89, and the cAMP-dependent protein kinase inhibitor. Forskolin and isoproterenol increased I_CaL by only 60% in these cells. Cells infused with either N-methyl-D-glucamine or K⁺-based intracellular solutions did not show multiphasic decay or facilitation under basal conditions. Isoproterenol increased I_CaL by 200% in these cells. In conclusion, we show that multiphasic inactivation of I_CaL is due to Ca²⁺-dependent inactivation that is reversible on a time scale of tens of milliseconds. Cs⁺ seems to activate the cAMP-dependent protein kinase pathway when used as a substitute for K⁺ in the pipette solution. L-type calcium current; calcium-dependent inactivation; facilitation; phosphorylation; cesium     

Ca(2+)-dependent inactivation of cardiac L-type Ca2+ channels does not affect their voltage sensor

Inactivation of currents carried by Ba2+ and Ca2+, as well as intramembrane charge movement from L-type Ca2+ channels were studied in guinea pig ventricular myocytes using the whole-cell patch clamp technique. Prolonged (2 s) conditioning depolarization caused substantial reduction of charge movement between -70 and 10 mV (charge 1, or charge from noninactivated channels). In parallel, the charge mobile between -70 and -150 mV (charge 2, or charge from inactivated channels) was increased. The availability of charge 2 depended on the conditioning pulse voltage as the sum of two Boltzmann components. One component had a central voltage of -75 mV and a magnitude of 1.7 nC/microF. It presumably is the charge movement (charge 2) from Na+ channels. The other component, with a central voltage of approximately - 30 mV and a magnitude of 3.5 nC/microF, is the charge 2 of L-type Ca2+ channels. The sum of charge 1 and charge 2 was conserved after different conditioning pulses. The difference between the voltage dependence of the activation of L-type Ca2+ channels (half-activation voltage, V, of approximately -20 mV) and that of charge 2 (V of -100 mV) made it possible to record the ionic currents through Ca2+ channels and charge 2 in the same solution. In an external solution with Ba2+ as sole metal the maximum available charge 2 of L-type Ca2+ channels was 10-15% greater than that in a Ca(2+)-containing solution. External Cd2+ caused 20-30% reduction of charge 2 both from Na+ and L-type Ca2+ channels. Voltage- and Ca(2+)-dependent inactivation phenomena were compared with a double pulse protocol in cells perfused with an internal solution of low calcium buffering capacity. As the conditioning pulse voltage increased, inactivation monitored with the second pulse went through a minimum at about 0 mV, the voltage at which conditioning current had its maximum. Charge 2, recorded in parallel, did not show any increase associated with calcium entry. Two alternative interpretations of these observations are: (a) that Ca(2+)- dependent inactivation does not alter the voltage sensor, and (b) that inactivation affects the voltage sensor, but only in the small fraction of channels that open, and the effect goes undetected. A model of channel gating that assumes the first possibility is shown to account fully for the experimental results. Thus, extracellular divalent cations modulate voltage-dependent inactivation of the Ca2+ channel. Intracellular Ca2+ instead, appears to cause inactivation of the channel without affecting its voltage sensor.     

Modulating L-type calcium current affects discontinuous cardiac action potential conduction.

We have used pairs of cardiac cells (i.e., one real guinea pig ventricular cell and a real-time simulation of a numerical model of a guinea pig ventricular cell) to evaluate the effects on action potential conduction of a variable coupling conductance in combination with agents that either increase or decrease the magnitude of the L-type calcium current. For the cell pairs studied, we applied a direct repetitive stimulation to the real cell, making it the "leader" cell of the cell pair. We have demonstrated that significant delays in action potential conduction for a cell pair can occur either with a decreased value of coupling conductance or with an asymmetry in size such that the follower cell is larger than the leader cell. In both conditions we have shown that isoproterenol, applied to the real cell at very low concentrations, can reversibly decrease the critical coupling conductance (below which action potential conduction fails) for a cell pair with fixed cell sizes, or, for a fixed value of coupling conductance, increase the maximum allowable asymmetry in cell size for successful conduction. For either of these effects, we were able to show that treatment of the real cell with BayK 8644, which more specifically increases the magnitude of the L-type calcium current, was able to mimic the actions of isoproterenol. Treatment of the leader cell of the cell pair (the real cell) with nifedipine, which selectively lowers the magnitude of the L-type calcium current, had effects opposite those of isoproterenol or BayK 8644. The actions of nifedipine, isoproterenol, and BayK 8644 are all limited to conditions in which the conduction delay is on the order of 5 ms or more, whether this delay is caused by limited coupling conductance or by asymmetry in size of the cells. This limitation is consistent with the time course of the L-type calcium current and suggests that the effects of calcium channel blockers or beta-adrenergic blocking drugs, in addition to being selective for regions of the heart that depend on the L-type calcium current for the upstroke of the action potential, would also be somewhat selective for regions of the heart that have discontinuous conduction, either normally or because of some pathological condition.     

Trifluoperazine enhancement of Ca2+-dependent inactivation of L-type Ca2+ currents inHelix aspersa neurons

The effects of trifluoperazine hydrochloride (TFP), a calmodulin antagonist, on L-type Ca²⁺ currents (L-type ICa²⁺) and their Ca²⁺-dependent inactivation, were studied in identifiedHelix aspersa neurons, using two microelectrode voltage clamp. Changes in [Ca²⁺]_i were measured in unclamped fura-2 loaded neurons. Bath applied TFP produced a reversible and dose-dependent reduction in amplitude of L-type ICa²⁺ (IC₅₀=28 μM). Using a double-pulse protocol, we found that TFP enhances the efficacy of Ca²⁺-dependent inactivation of L-type ICa²⁺. Trifluoperazine sulfoxide (50 μM), a TFP derivative with low calmodulin-antagonist activity, did not have any effects on either amplitude or inactivation of L-type ICa²⁺. TFP (20 μM) increased basal [Ca²⁺]_i from 147±37 nM to 650±40nM (N=7). The increase in [Ca²⁺]_i was prevented by removal of external Ca²⁺ and curtailed by depletion of caffeine-sensitive intracellular Ca²⁺ stores. Since TFP may also block protein kinase C (PKC), we tested the effect of a PKC activator (12-O-tetradecanoyl-phorbol-13-acetate) on L-type Ca²⁺ currents. This compound produced an increase in L-type ICa²⁺ without enhancing Ca²⁺-dependent inactivation. The results show that 1) TFP reduces L-type ICa²⁺ while enhancing the efficacy of Ca²⁺-dependent inactivation. 2) TFP produces an increase in basal [Ca²⁺]_i which may contribute to the enhancement of Ca²⁺-dependent inactivation. 3) PKC up-regulates L-type ICa²⁺ without altering the efficacy of Ca²⁺ dependent inactivation. 4) The TFP effects cannot be attributed to its action as PKC blocker.     

Analysis of the modal hypothesis of Ca2+-dependent inactivation of L-type Ca2+ channels

A kinetic model of Ca2+-dependent inactivation (CDI) of L-type Ca2+ channels was developed. The model is based on the hypothesis that postulates the existence of four short-lived modes with lifetimes of a few hundreds of milliseconds. Our findings suggest that the transitions between the modes is primarily determined by the binding of Ca2+ to two intracellular allosteric sites located in different motifs of the CI region, which have greatly differing binding rates for Ca2+ (different k(on)). The slow-binding site is controlled by local Ca2+ near a single open channel that is consistent with the "domain" CDI model, and Ca2+ binding to the fast-binding site(s) depends on Ca2+ arising from distant sources that is consistent with the "shell" CDI model. The model helps to explain numerous experimental findings that are poorly understood so far.     

Extremely slow inactivation of the ion channels formed by transfected α2 of L-type Ca2+ channelsof L-type Ca2+ channels

Barium currents through ion channels formed by α1-subunit of L-type Ca²⁺ channel (I _α1) were recorded from cultured chinese hamster ovary (CHO) cells. The cells were stably transfected with either a cardiac or a smooth muscle (SM) variant of α1-subunit. TheI _α1 in both cases exhibited similar fast voltage-dependent activation kinetics and slow apparent inactivation kinetics. With 10 mM Ba²⁺ in the bath solution,I _α1 was activated at potentials more positive than −40 mV, peaked between 0 and +10 mV, and reversed at about +50 mV. In addition to slow apparent inactivation of inward current, both subunits provided an extremely slow voltage-dependent inactivation at potentials more positive than −100 mV, with half-maximum inactivation at −43.4 mV for cardiac and −41.4 mV for SM α1-subunits. The onset of inactivation as well as recovery from this process were within a time range of minutes. The voltage dependence of steady-state inactivation could be fitted by the sum of two Boltzmann's equations with slope factors of about 12 mV and 5 mV. A less sloped component has its midpoints at −75.6 and −63.7 mV, and a steeper component has its midpoints at −42.8 and −37.7 mV for cardiac and SM α1-subunits, respectively. Relative contribution of the steeper component was higher in both subunits (0.86 and 0.66 for cardiac and SM subunits, respectively). For comparison, the inactivation curves for 5-sec-long conditioning prepulses could be fitted by single Boltzmann's distribution with a 20 mV more positive midpoint and a slope factor of about 13 mV. In contrast to the steady-state inactivation curves, they showed considerable overlap with the steady-state activation curve. Our results reflect functional consequences of known sequence differences between α1-subunits of the cardiac and SM L-type Ca²⁺ channels and could be used in structural modeling of Ca²⁺ channel gating. In addition, they show that depolarization-induced window current has a transient nature and decays with the development of extremely slow inactivation. This is the first demonstration that slow inactivation of the L-type Ca²⁺ channel is an intrinsic property of its α1-subunits.     

Trypsin increases availability and open probability of cardiac L-type Ca2+ channels without affecting inactivation induced by Ca2+.

The patch-clamp technique was employed to investigate the response of single L-type Ca2+ channels to the protease trypsin applied to the intracellular face of excised membrane patches from guinea pig ventricular myocytes. Calpastatin and ATP were used to prevent run-down of Ca2+ channel activity monitored with 96 mM Ba2+ as charge carrier in the presence of 2.5 microM (-)-BAYK 8644. Upon application of trypsin (100 micrograms/ml) channel activity was enhanced fourfold and remained elevated upon removal of trypsin, as expected of a proteolytic, irreversible modification. The trypsin effect was not mediated by a proteolytic activation of protein kinases, as evidenced by the insensitivity of this effect to protein kinase inhibitors. Trypsin-modified Ca2+ channels exhibited the usual run-down phanomenon upon removal of calpastatin and ATP. In ensemble average currents trypsin-induced changes of channel function are apparent as a threefold increase in peak current and a reduction in current inactivation. At the single channel level these effects were based on about a twofold increase in both Ca2+ channels' availability and open probability. Neither the actual number of channels in the patch nor their unitary conductance as well as reversal potential was changed by trypsin. The Ca(2+)-induced inactivation was not impaired, as judged by a comparable sensitivity of trypsin-modified Ca2+ channels to intracellular Ca2+. Similarly, trypsin treatment did not affect the sensitivity of Ca2+ channels to phenylalkylmine inhibition. The observed alterations in channel function are discussed in terms of possible structural correlates.     

Molecular basis of calmodulin tethering and Ca2+-dependent inactivation of L-type Ca2+ channels.

Ca(2+)-dependent inactivation (CDI) of L-type Ca(2+) channels plays a critical role in controlling Ca(2+) entry and downstream signal transduction in excitable cells. Ca(2+)-insensitive forms of calmodulin (CaM) act as dominant negatives to prevent CDI, suggesting that CaM acts as a resident Ca(2+) sensor. However, it is not known how the Ca(2+) sensor is constitutively tethered. We have found that the tethering of Ca(2+)-insensitive CaM was localized to the C-terminal tail of alpha(1C), close to the CDI effector motif, and that it depended on nanomolar Ca(2+) concentrations, likely attained in quiescent cells. Two stretches of amino acids were found to support the tethering and to contain putative CaM-binding sequences close to or overlapping residues previously shown to affect CDI and Ca(2+)-independent inactivation. Synthetic peptides containing these sequences displayed differences in CaM-binding properties, both in affinity and Ca(2+) dependence, leading us to propose a novel mechanism for CDI. In contrast to a traditional disinhibitory scenario, we suggest that apoCaM is tethered at two sites and signals actively to slow inactivation. When the C-terminal lobe of CaM binds to the nearby CaM effector sequence (IQ motif), the braking effect is relieved, and CDI is accelerated.     

Interplay of voltage and Ca-dependent inactivation of L-type Ca current

Inactivation of L-type Ca channels (LTCC) is regulated by both Ca and voltage-dependent processes (CDI and VDI). To differentiate VDI and CDI, several experimental and theoretical studies have considered the inactivation of Ba current through LTCC (I_Ba) as a measure of VDI. However, there is evidence that Ba can weakly mimic Ca, such that I_Ba inactivation is still a mixture of CDI and VDI. To avoid this complication, some have used the monovalent cation current through LTCC (I_NS), which can be measured when divalent cation concentrations are very low. Notably, I_NS inactivation rate does not depend on current amplitude, and hence may reflect purely VDI. However, based on analysis of existent and new data, and modeling, we find that I_NS can inactivate more rapidly and completely than I_Ba, especially at physiological temperature. Thus VDI that occurs during I_Ba (or I_Ca) must differ intrinsically from VDI during I_NS. To account for this, we have extended a previously published LTCC mathematical model of VDI and CDI into an excitation-contraction coupling model, and assessed whether and how experimental I_Ba inactivation results (traditionally used in VDI experiments and models) could be recapitulated by modifying CDI to account for Ba-dependent inactivation. Thus, the view of a slow and incomplete I_NS inactivation should be revised, and I_NS inactivation is a poor measure of VDI during I_Ca or I_Ba. This complicates VDI analysis experimentally, but raises intriguing new questions about how the molecular mechanisms of VDI differ for divalent and monovalent currents through LTCCs.     

Targeting the late component of the cardiac L-type Ca2+ current to suppress early afterdepolarizations

Early afterdepolarizations (EADs) associated with prolongation of the cardiac action potential (AP) can create heterogeneity of repolarization and premature extrasystoles, triggering focal and reentrant arrhythmias. Because the L-type Ca²⁺ current (I_Ca,L) plays a key role in both AP prolongation and EAD formation, L-type Ca²⁺ channels (LTCCs) represent a promising therapeutic target to normalize AP duration (APD) and suppress EADs and their arrhythmogenic consequences. We used the dynamic-clamp technique to systematically explore how the biophysical properties of LTCCs could be modified to normalize APD and suppress EADs without impairing excitation–contraction coupling. Isolated rabbit ventricular myocytes were first exposed to H₂O₂ or moderate hypokalemia to induce EADs, after which their endogenous I_Ca,L was replaced by a virtual I_Ca,L with tunable parameters, in dynamic-clamp mode. We probed the sensitivity of EADs to changes in the (a) amplitude of the noninactivating pedestal current; (b) slope of voltage-dependent activation; (c) slope of voltage-dependent inactivation; (d) time constant of voltage-dependent activation; and (e) time constant of voltage-dependent inactivation. We found that reducing the amplitude of the noninactivating pedestal component of I_Ca,L effectively suppressed both H₂O₂- and hypokalemia-induced EADs and restored APD. These results, together with our previous work, demonstrate the potential of this hybrid experimental–computational approach to guide drug discovery or gene therapy strategies by identifying and targeting selective properties of LTCC.     

Influence of channel subunit composition on L-type Ca2+ current kinetics and cardiac wave stability

Previous studies have demonstrated that the slope of the function relating the action potential duration (APD) and the diastolic interval, known as the APD restitution curve, plays an important role in the initiation and maintenance of ventricular fibrillation. Since the APD restitution slope critically depends on the kinetics of the L-type Ca(2+) current, we hypothesized that manipulation of the subunit composition of these channels may represent a powerful strategy to control cardiac arrhythmias. We studied the kinetic properties of the human L-type Ca(2+) channel (Ca(v)1.2) coexpressed with the alpha(2)delta-subunit alone (alpha(1C) + alpha(2)delta) or in combination with beta(2a), beta(2b), or beta(3) subunits (alpha(1C) + alpha(2)delta + beta), using Ca(2+) as the charge carrier. We then incorporated the kinetic properties observed experimentally into the L-type Ca(2+) current mathematical model of the cardiac action potential to demonstrate that the APD restitution slope can be selectively controlled by altering the subunit composition of the Ca(2+) channel. Assuming that beta(2b) most closely resembles the native cardiac L-type Ca(2+) current, the absence of beta, as well as the coexpression of beta(2a), was found to flatten restitution slope and stabilize spiral waves. These results imply that subunit modification of L-type Ca(2+) channels can potentially be used as an antifibrillatory strategy.     

Angiotensin II-induced increase of T-type Ca2+ current and decrease of L-type Ca2+ current in heart cells

The effect of angiotensin II (Ang II) on the T- and L-type calcium currents (I(Ca)) in single ventricular heart cells of 18-week-old fetal human and 10-day-old chick embryos was studied using the whole-cell voltage clamp technique. Our results showed that in both, human and chick cardiomyocytes, Ang II (10(-7)M) increased the T-type calcium current and decreased the L-type I(Ca). The effect of Ang II on both types of currents was blocked by the AT1 peptidic antagonist, [Sar1, Ala8] Ang II (2 x 10(-7)M). Protein kinase C activator, phorbol 12,13-dibutyrate, mimicked the effect of Ang II on the T- and L-type calcium currents. These results demonstrate that in fetal human and chick embryo cardiomyocytes Ang II affects the T- and L-type Ca2+ currents differently, and this effect seems to be mediated by the PKC pathway.     

Functional coupling of voltage-dependent L-type Ca2+ current to Ca2+-activated K+ current in pituitary GH3 cells

Ca2+-activated K+ currents (I(K(Ca)) can contribute to action potential repolarization and after-hyperpolarization in GH3 cells. In this study, we examined how the activation of I(K(Ca) at the cellular level could be functionally coupled to Ca2+ influx through L-type Ca2+ channels. A 30-msec Ca2+ influx step to 0 mV was found to exhibit substantial contribution of Ca2+ influx through the activation of I(Ca,L) to the activation of I(K(Ca)). A bell-shaped relationship between the conditioning potentials and the integrated I(K(Ca)) was observed, suggesting that the magnitude of integrated I(Ca,L) correlates well with that of integrated I(K(Ca)) in the same cell. A linear relationship of integrated I(Ca,L) and integrated I(K(Ca)) was found with a coupling ratio of 69+/-7. The value of the coupling ratio was unaffected by the presence of Bay K 8644 or nimodipine, although these compounds could effectively affect the amplitudes of both I(K(Ca)) and I(Ca,L). However, tetrandrine could decrease the coupling ratio. Paxilline or intracellular Ca2+ buffer with EGTA decreased the coupling ratio, while apamin had no effect on it. Interestingly, phorbol 12-myristate 13-acetate also reduced the coupling ratio significantly, whereas thapsigargin increased this value. Thus, the present study indicates that the activation of I(K(Ca)) during brief Ca2+ influx, which is inhibited by paxilline, is coupled to Ca2+ influx primarily through the L-type channels. The selective modulation of I(K(Ca)) by second messengers or Ca2+ release from internal stores may affect the coupling efficiency and hence cellular excitability.     

Modulation of L-type Ca2+ channels in neonatal rat heart by a novel Ca2+ channel agonist

L-type Ca2+ channels are essential in triggering the intracellular Ca2+ release and contraction in heart cells. In this study, we used patch clamp technique to compare the effect of two pure enantiomers of L-type Ca2+ channel agonists: (+)-CGP 48506 and the dihydropyridine (+)-SDZ-202 791 in cardiomyocytes from rats 2-5 days old. The predominant Ca2+ current activated by standard step pulses in these myocytes was L-type Ca2+ current. The dihydropyridine antagonist (+)-PN200-110 (5 microM) blocked over 90% of Ca2+ currents in most cells tested. CGP 48506 lead to a maximum of 200% increase in currents. The threshold concentration for the CGP effect was at 1 microM and the maximum was reached at 20 microM. SDZ-202 791 had effects in nanomolar concentrations and a maximum effect at about 2 microM. The maximal effect of (+)-SDZ-202 791 was a 400% increase in the amplitude of Ca2+ currents and was accompanied by a 10-15 mV leftward shift in the voltage dependence of activation. CGP 48506 increased the currents equally at all voltages tested. Both compounds slowed the deactivation of tail currents and lead to the appearance of slowly activating and slowly deactivating current components. However, SDZ-202 791 had larger effects on deactivation and CGP 48506 had larger effect on the rate of Ca2+ current activation. The effect of SDZ-202 791 was fully additive to that of CGP 48506 even after maximum concentrations of CGP. This observation suggests that the two Ca2+ channel agonists may act at two different sites on the L-type Ca2+ channel. We suggest that CGP 48506 would be a potential cardiotonic agent without the deleterious proarrhythmic effects attributable to the dihydropyridine agonists.     

Ca2+ modulation of Ca2+ release-activated Ca2+ channels is responsible for the inactivation of its monovalent cation current

The Ca(2+) release-activated Ca(2+) (CRAC) channel is the most well documented of the store-operated ion channels that are widely expressed and are involved in many important biological processes. However, the regulation of the CRAC channel by intracellular or extracellular messengers as well as its molecular identity is largely unknown. Specifically, in the absence of extracellular divalent cations it becomes permeable to monovalent cations with a larger conductance, however this monovalent cation current inactivates rapidly by an unknown mechanism. Here we found that Ca(2+) dissociation from a site on the extracellular side of the CRAC channel is responsible for the inactivation of its Na(+) current, and Ca(2+) occupancy of this site otherwise potentiates its Ca(2+) as well as Na(+) currents. This Ca(2+)-dependent potentiation is required for the normal functioning of CRAC channels.