Down-regulation of Voltage-gated Ca2+ Channels by Neuronal Calcium Sensor-1 Is β Subunit-specific*

Neuronal Ca2+sensor protein-1 (NCS-1) is a member of the Ca2+ binding protein family, with three functional Ca2+ binding EF-hands and an N-terminal myristoylation site. NCS-1 is expressed in brain and heart during embryonic and postnatal development. In neurons, NCS-1 facilitates neurotransmitter release, but both inhibition and facilitation of the Ca2+ current amplitude have been reported. In heart, NCS-1 co-immunoprecipitates with K+channels and modulates their activity, but the potential effects of NCS-1 on cardiac Ca2+ channels have not been investigated. To directly assess the effect of NCS-1 on the various types of Ca2+ channels we have co-expressed NCS-1 inXenopus oocytes, with CaV1.2, CaV2.1, and CaV2.2 Ca2+ channels, using various subunit combinations. The major effect of NCS-1 was to decrease Ca2+ current amplitude, recorded with the three different types of α1 subunit. When expressed with CaV2.1, the depression of Ca2+ current amplitude induced by NCS-1 was dependent upon the identity of the β subunit expressed, with no block recorded without β subunit or with the β3 subunit. Current-voltage and inactivation curves were also slightly modified and displayed a different specificity toward the β subunits. Taken together, these data suggest that NCS-1 is able to modulate cardiac and neuronal voltage-gated Ca2+ channels in a β subunit specific manner.

Ca 2ϩ entry through voltage-gated Ca 2ϩ channels is essential for various cellular processes that include muscle contraction, pacemaker activity, synaptic transmission, or gene expression. Several types of Ca 2ϩ channels have been characterized (T, L, N, P/Q, and R) that appear to play a specific role in each of these functions. These channels share a common architecture composed of a major ␣ 1 subunit (for which ten genes are known) tightly associated with regulatory subunits ␣ 2 -␦ (four different genes), ␤ (four genes), and possibly ␥ (eight genes) in a functional multimeric complex (1)(2)(3)(4). This molecular diversity, further expanded by the existence of several splice variants for each of these genes, produces a large number of possible Ca 2ϩ channel subunit combinations with different pharmacological and biophysical properties and specific cellular and subcellular localization (5). The precise regulation of the Ca 2ϩ influx in response to various physiological situations is further controlled by several regulatory mechanisms, working at different levels, including channel expression, localization, or activity, via additional interactions with modulatory proteins. Several Ca 2ϩ -dependent feedback mechanisms sense incoming Ca 2ϩ ions to finely tune channel activity to the cellular Ca 2ϩ demands and prevent cytotoxic Ca 2ϩ overload (6). These mechanisms use Ca 2ϩ -sensing proteins and are specific of a given type of Ca 2ϩ channel. It has been shown, for example, that Ca 2ϩ -dependent inactivation of the L-type Ca 2ϩ channel (encoded by the Ca V 1.2 ␣ 1 subunit) is governed by a Ca 2ϩ -driven interaction between calmodulin and the C-terminal tail of the channel ␣ 1 subunit (7)(8)(9)(10). A similar functional interaction also appears to exist on the P/Q-type Ca 2ϩ channel (encoded by the CaV2.1 subunit), the major channel type involved in synaptic transmission in the mammalian central nervous system (11)(12)(13).
However, a possible involvement of NCS-1 in expression and regulation of voltage-gated Ca 2ϩ channels has also been proposed. Overexpression of a dominant-negative mutant of NCS-1, which displays impaired Ca 2ϩ -dependent conformational changes (19), or direct loading of presynaptic nerve terminals with NCS-1 suggested that voltage-independent inhibition, as well as activity-dependent facilitation of P/Q-type Ca 2ϩ channels (Ca V 2.1), could be controlled by NCS-1, possibly via direct protein-protein interactions (18). Effects on N-type Ca 2ϩ channel (Ca V 2.2) properties have also been reported (20), and the expression of NCS-1 in mammalian cardiac myocytes and subsequent effect on K ϩ channel expression (26) gave rise to the possibility that NCS-1 may regulate multiple types of Ca 2ϩ channels and other voltage-dependent ion channels, not only in neurons (27).
In a first step to explore this possibility, we have co-expressed NCS-1 with three different types of Ca 2ϩ channel, Ca V 1.2, Ca V 2.1, and Ca V 2.2, associated with different combinations of auxiliary ␤ subunit, and measured the resulting Ba 2ϩ and Ca 2ϩ currents. These combinations are likely to be expressed in different cell types where they represent potential targets for NCS-1 effects. Our goal was to explore the effect of NCS-1 on both Ca 2ϩ channel expression and properties and to provide a first description of the molecular requirements necessary for NCS-1 effects on Ca 2ϩ channel that may help in the understanding of the precise mode of action of this Ca 2ϩ -binding protein. We have, however, focused this study on the P/Qtype (Ca V 2.1) Ca 2ϩ channels, which seem to be a primary target in various cell types (28). Our results show that NCS-1 down-regulates expression of L-, N-, and P/Q-type Ca 2ϩ channels in a ␤ subunit-specific manner and induces minor modifications of the electrophysiological properties of the channel. We provide evidence of direct functional effects of NCS-1, in addition to modifications in the expression level and/or trafficking of the channels to the membrane.

EXPERIMENTAL PROCEDURES
Materials and Oocyte Preparation-The following cDNA were used, and the GenBank TM accession number is provided: Mutations NCS-1 E120Q and NCS-1 G2A have been described previously (29). Ca 2ϩ channel subunits were subcloned into the pmt2 vector, whereas NCS-1 and its mutants were subcloned into pcDNA3 (Invitrogen).
Xenopus laevis oocyte preparation and injection were performed as described previously (30). Each oocyte was injected with 5-10 nl of a cDNA mixture containing the ␣1ϩ␣2␦ϩ␤ϩNCS-1 cDNAs at ϳ0.3 ng/nl with a ratio of 1:2:3:1. When one or more of these cDNAs was omitted, cDNA concentrations were kept constant by addition of the appropriate volume of deionized water. Oocytes were kept for 2 to 4 days before recordings at 18°C and under gentle agitation.
Electrophysiology-Whole-cell Ba 2ϩ currents were recorded under two-electrode voltage clamp using the GeneClamp 500 amplifier (Axon Instruments, Union City, CA). Current and voltage electrodes (less than 1 megohm) were filled with 3 M KCl, pH 7.2, with KOH. Ba 2ϩ and Ca 2ϩ current recordings were performed after injection of BAPTA (ϳ50 nl of the following (in mM): BAPTA-free acid (Sigma), 100; CsOH, 10; HEPES, 10; pH 7.2 with CsOH) using the following bathing solution (in mM): BaOH/CaOH, 10; TEAOH, 20; NMDG, 50; CsOH, 2; HEPES, 10; pH 7.2, with methanesulfonic acid. Currents were filtered and digitized using a DMA-Tecmar Labmaster and subsequently stored on a Pentium-based personal computer using the pClamp software (version 6.02; Axon Instruments). Ba 2ϩ or Ca 2ϩ currents were recorded during a 400-ms test pulse from Ϫ80 to ϩ10 mV. Current amplitudes were measured at the peak of the current. Comparisons of averaged amplitudes between batches were always made with amplitudes measured the same day after injection. Comparisons between similar experiments were made by normalizing all averaged amplitudes with respect to the control current amplitude set as 100%. Isochronal steady-state inactivation curves (2.5 s of conditioning voltage followed by a 400-ms test pulse to ϩ10 mV) were fitted using the equation, where I is the current amplitude measured during the test pulse at ϩ10 mV for conditioning voltages varying from Ϫ80 to ϩ50 mV, I max is the current amplitude measured during the test pulse for a conditioning voltage to Ϫ80 mV, V in is the potential for half-inactivation, V is the voltage, k is the slope factor, and R in is the proportion of non-inactivating current. Current to voltage curves were fitted using the equation, where I is the current amplitude measured during voltage steps varying from Ϫ80 to ϩ50 mV, I max is the peak current amplitude measured at the minimum of the current-voltage curve, G is the normalized macroscopic conductance, E rev is the apparent reversal potential, V act is the potential for half-activation, V is the value of the voltage step, and k is a slope factor. Inactivation kinetics were estimated by fitting Ba 2ϩ current decay with two exponential components using the equation, where I is the current amplitude, t is the time, 1 , 2 , A 1 , and A 2 represent the time constants and amplitudes of the two compo-nents, and C is a constant. The proportion of the slow time constant (% 2 ) is the ratio A 2 /(A 1 ϩA 2 ).
Several independent experiments (N, number of batches of oocytes injected) were performed, always including a control group without NCS-1 expressed. These experiments always gave the same statistical result, and the total number of recordings (from n oocytes of these N experiments) are thus presented. All values are presented as mean Ϯ S.E., and comparison between groups of oocytes were evaluated using a Student's t test, with a statistical significance set at the p value Ͻ0.05.

Expression of NCS-1 in Xenopus
Oocytes-Previous works on chromaffin, human embryonic kidney 293, and COS-7 cells have shown that NCS-1 was endogenously expressed at nonnegligible levels (24,28). 2 In non-injected X. laevis oocytes, the level of expression of the endogenous NCS-1 was barely detectable in Western-blots and much lower than in human embryonic kidney 293 and COS-7 cells (see Fig. 1A). Thus X. laevis oocytes are a system of choice to study the functional effect of NCS-1 on voltage-gated Ca 2ϩ channels. In these oocytes, injection of the cDNA coding for rat NCS-1 led, as expected, to a massive expression of a protein of a molecular mass of ϳ23 kDa, in accordance with the theoretical molecular mass of 2 A. Jeromin, unpublished observations. NCS-1 (21.9 kDa; see Fig. 1B). In the following experiments, the oocytes used for current recordings were collected after recordings and submitted to a similar Western blot analysis to ensure that the NCS-1 protein was properly expressed.
Down-regulation of Ca V 1.2, Ca V 2.1, and Ca V 2.2 Expression by NCS-1-In a first set of experiments, the effects of NCS-1 were tested on Ba 2ϩ currents flowing through L-, N-, and P/Q-type Ca 2ϩ channels. For each Ca 2ϩ channel type, this was done by co-injecting a mixture of cDNA containing ␤ 2 , ␣ 2 -␦, and the appropriate ␣ 1 Ca 2ϩ channel subunits (Ca V 1.2, Ca V 2.2, or Ca V 2.1, respectively) with either NCS-1 cDNA or water into two different batches of oocytes. After 2 to 4 days of incubation, Ba 2ϩ current amplitudes were recorded from the two batches of oocytes injected the same day, during a single 400-ms-long depolarizing step to ϩ10 mV from a holding potential of Ϫ80 mV and compared. Under these conditions, a clear decrease in the averaged Ba 2ϩ current amplitude was seen upon co-expression of NCS-1 (see Fig. 2A). This effect was most pronounced in oocytes expressing the Ca V 2.2 Ca 2ϩ channel, where the averaged Ba 2ϩ current amplitude recorded in the batch of oocytes co-injected with rat NCS-1 cDNA (N ϭ 1 experiment, n ϭ 38 oocytes) was only 10% of the control current amplitude recorded from oocytes co-injected with H 2 O instead of the NCS-1 cDNA (N ϭ 1, n ϭ 35). However, a similar effect was also found when NCS-1 was co-injected with Ca V 1.2 (N ϭ 3, n ϭ 71; 52% of the control amplitude n ϭ 70) or Ca V 2.1 Ca 2ϩ channel subunits (N ϭ 7, n ϭ 206; 46% of the control amplitude n ϭ 162).
The Ca 2ϩ channel ␤ subunit is an important determinant of the final Ca 2ϩ current amplitude observed at the oocyte surface membrane and regulates many of its electrophysiological properties (4,31,32). Hence, we tested the role of the ␤ subunit on the effects of NCS-1. Using the same experimental approach, we co-expressed NCS-1 with the Ca V 2.1 Ca 2ϩ channel either with the ␣ 2 -␦ subunit alone or with the ␣ 2 -␦ and one of the four ␤ subunits (␤ 1 -␤ 4 ). Again, in each case, the resulting current amplitude was compared with the current amplitude recorded in oocytes injected with the same combination of Ca 2ϩ channel subunits but without NCS-1. Interestingly, when NCS-1 was co-expressed with Ca V 2.1 without ␤ subunit, almost no effect on current amplitude was observed (N ϭ 2, n ϭ 32 and 26 for control). A similar result was also found upon co-expression of the ␤ 3 subunit (N ϭ 12, n ϭ 198, with control n ϭ 198), whereas co-expression of NCS-1 with the Ca V 2.1 and ␤ 1 , ␤ 2 , or ␤ 4 subunit decreased the expressed Ba 2ϩ current amplitude to between 25 and 45% of their respective control values (see Fig.  2B, N ϭ 3, 7, 2 and n ϭ 100, 206, 71 and n ϭ 65, 162, 84 for controls, respectively). The lack of effect of NCS-1 on the Ca V 2.1 Ca 2ϩ channel subunit, expressed alone or with the ␤ 3 subunit, could not be attributed to a deficit of NCS-1 protein in these oocytes, because a robust expression of NCS-1 was also detected in these oocytes by Western blot (see bottom of Fig.  2B). On the other hand, to ensure that the decrease of the Ba 2ϩ current amplitude obtained upon co-expression of NCS-1 with the ␤ 1 , ␤ 2 , or ␤ 4 subunit was not because of a deleterious effect on the expression of any cloned protein, we analyzed the level of expression of the Ca V 2.1 protein in these conditions by Western blot. Histogram showing the effects of NCS-1 on Ba 2ϩ current amplitude recorded in oocytes injected with the ␣ 2 -␦, the ␤ 2 , and Ca V 1.2, Ca V 2.1, or Ca V 2.2 Ca 2ϩ channel subunits, with or without NCS-1 cDNA, are shown. Peak Ba 2ϩ currents were recorded 2-4 days after injection during a test pulse to ϩ10 mV from a holding potential of Ϫ80 mV and expressed as % of control (averaged current amplitude recorded with the same subunit combination but without NCS-1 expressed). *, significantly different from control (p Ͻ 0.05); n.s., not significantly different. B, histogram showing the effects of NCS-1 on Ba 2ϩ current amplitude recorded on oocytes injected with the Ca V 2.1, the ␣ 2 -␦, and the ␤ 1 , ␤ 2 , ␤ 3 , ␤ 4 , or no ␤ Ca 2ϩ channel subunits, with or without NCS-1 cDNA. Peak Ba 2ϩ currents were recorded 2-5 days after injection during a test pulse to ϩ10 mV from a holding potential of Ϫ80 mV. *, significantly different from control (p Ͻ 0.05); n.s., not significantly different. In A and B, Western blot analyses of the expression of NCS-1 in the oocytes used for the recordings are shown at the bottom. Each lane was loaded with ϳ3 oocytes. C, Western blot of total proteins obtained from oocytes injected with, from left to right, the Ca V 2.1ϩ␣ 2 -␦ϩ␤ 2 cDNAs, the Ca V 2.1ϩ␣ 2 -␦ϩ␤ 2 ϩNCS-1 cDNAs, or water, probed with an anti-Ca V 2.1 antibody. Each lane was loaded with ϳ 3oocytes. Note the presence of a specific band at 250 kDa of similar density in oocytes injected with the Ca V 2.1 subunit with or without NCS-1. The cross-reactive band at ϳ160 kDa, also found in non-injected oocytes, was used as a control of the loading charge.
ties. Current-voltage curves and isochronal inactivation curves were constructed from currents recorded in oocytes expressing the Ca V 2.1 subunit with the ␣ 2 -␦ alone or with the ␤ 1 , ␤ 2 , or ␤ 3 subunits, in the presence or absence of NCS-1. These recordings were performed using either Ba 2ϩ or Ca 2ϩ ions as extracellular permeant cations to track any Ca 2ϩ -specific modulation.
As presented in Table I, in the presence of Ca 2ϩ ions and in the absence of ␤ subunit (␣ 1A ϩ␣ 2 -␦ subunits), no modifications in the activation and inactivation parameters were observed upon co-expression of NCS-1. This lack of effect was also noted upon co-expression of the ␤ 1 or ␤ 3 subunit (see Fig. 3 and Table  I). Interestingly, when co-expressed with Ca V 2.1 and the ␤ 2a subunit, NCS-1 significantly depolarized the current-voltage curve (V act ϭ Ϫ5.0 and Ϫ0.8 mV without and with NCS-1 respectively, p Ͻ 0.05) and reduced inactivation (R in ϭ 69 and 56% respectively, p Ͻ 0.05). Similar effects were also found in the presence of extracellular Ba 2ϩ (Table II) and thus were not Ca 2ϩ -dependent.
We then analyzed the effects of NCS-1 on Ba 2ϩ current kinetics. Ca V 2.1 current inactivation could be approximated by an exponential decaying phase, best described using a fast ( 1 ) and a slow ( 2 ) component. None of these components appeared to be significantly affected by expression of the Ca 2ϩ -binding protein NCS-1 (Fig. 4). This lack of effect was found for channels co-expressed with the ␤ 1 or the ␤ 2 subunit and in the presence of either extracellular Ba 2ϩ or Ca 2ϩ . Neither the time constants ( 1 and 2 ) nor their respective amplitude (% 2 ) were changed at all voltages examined (see Fig. 4). Moreover, no effect on channel activation and reactivation were observed upon co-expression of NCS-1, whether the Ca V 2.1 subunit was expressed with the ␣ 2 -␦ alone or with any of the four ␤ subunits (not shown). Therefore, although both channel expression and channel properties seemed to be regulated by NCS-1 in a ␤ subunit-specific manner, they appear to require different subunit arrangements, i.e. expression was modified when ␤ 1 , ␤ 2 , or ␤ 4 subunits were expressed, whereas modifications in channel properties were only recorded in the presence of the ␤ 2 subunit.
To get some insight into the possible molecular determinants involved in NCS-1 effects, the same experiments, with the ␤ 2a subunit, were conducted using two mutants of NCS-1. NCS-1 E120Q , with its third EF-hand disrupted, showed impaired Ca 2ϩ -dependent conformational changes (19) but was still able to bind cellular proteins. NCS-1 G2A , a myristoylation-deficient mutant of NCS-1, relocalized NCS-1 from the perinuclear region to the cytosol (29).
Co-expression of either NCS-1 E120Q or NCS-1 G2A had the same effect as wild-type NCS-1 on the current-voltage curve of the Ca V 2.1ϩ␣ 2 -␦ϩ␤ 2 Ca 2ϩ channel (i.e. a small but significant positive shift of ϳ5 mV; see Fig. 5B and Table III). They differed, however, in their effects on the inactivation curve. Although the NCS-1 E120Q completely suppressed the effect of NCS-1 on the residual current (R in ; see Fig. 5 and Table III), co-expression of NCS-1 G2A left this parameter unchanged but suppressed the shift in V in induced by wild-type NCS-1. These two mutants also had different actions on current amplitudes. Indeed, whereas NCS-1 E120Q reduced the Ba 2ϩ current amplitude when compared with control currents, recorded the same day in oocytes not injected with NCS-1, NCS-1 G2A had no consistent effect, although a small, albeit not significant, reduction could be noted (see Fig. 5A). In the case of NCS-1 E120Q , however, this effect was less marked than when recorded in oocytes co-injected with the wild-type NCS-1, suggesting that both Ca 2ϩ -dependent conformational changes, and possibly myristoylation, participated to the observed effects. It should be noted that none of these mutants had any effect on the functional properties of the Ca V 2.1ϩ␣ 2 -␦ Ca 2ϩ channels, expressed alone or with the ␤ 1 or ␤ 3 subunits, whereas the effects   Table I) and decreased the residual current (marked by an asterisk on the inactivation curve). This effect was only seen with the ␤ 2 subunit. on current amplitude were reproduced with the ␤ 1 subunit (data not shown). DISCUSSION Recent studies have extended the area of expression of NCS-1 from the nervous system to neuroendocrine cells and even cardiac myocytes (21,26,33). In these cell types, NCS-1 modulates synaptic transmission (14,22,34), secretion (21,25), or cellular excitability (18,20) via complex processes that include regulation of key enzymes for membrane transport (15,17,23,24,35) and specific regulation of various ion channels (16, 18 -20, 28). K ϩ channels were the first ion channel target to be characterized in expression systems (26,36). Co-expres-sion studies described an up-regulation of channel expression and activity, specifically recorded with the K V 4 K ϩ channel family (26,36). Modifications in the expression and properties of other ion channels have also been reported. In endocrine cells and neurons P/Q-type and N-type Ca 2ϩ channels are clearly affected (18 -20, 28), but no effect on L-type channels has been reported so far (20). These results suggest the existence of specific effects among the different voltage-activated Ca 2ϩ channel types or among different tissues. However, the molecular basis of this specificity, and in particular the role of  FIG. 5. Effects of NCS-1 E120Q and NCS-1 G2A on Ca V 2.1 expression and properties. A, effects of wild-type NCS-1, NCS-1 E120Q , and NCS-1 G2A mutants on Ca V 2.1 channel expression. Oocytes were injected with Ca V 2.1ϩ␣ 2 -␦ϩ␤ 2a subunit cDNAs with or without the appropriate NCS-1 mutants, and currents were measured in 10 mM Ba 2ϩ during a test depolarization to ϩ10 mV, as described for Fig. 2. Note that NCS-1 and NCS-1 E120Q both decreased current amplitude significantly (p Ͻ 0.05). B, oocytes were injected with Ca V 2.1ϩ␣ 2 -␦ϩ␤ 2a subunit cDNAs with (open circle) or without (open square; Control) the appropriate NCS-1 mutants, NCS-1 E120Q in the left panel and NCS-1 G2A in the right panel. Currents were measured in 10 mM extracellular Ba 2ϩ as described for Fig. 3. Both mutations slightly shifted the current-voltage curve in the positive direction, but only NCS-1 E120Q seemed to be able to reverse the decrease in the residual current (R in ; see Table III for V act , V in , and R in values). the subunit composition of the channel, in the observed effects remains unknown.
In the present work, using heterologous expression of defined Ca 2ϩ channel subunits, we show that NCS-1 has two major effects on high voltage-activated Ca 2ϩ channels: (1) a decrease in the current amplitude that is ␤ subunit-specific but observed with Ca V 1.2, Ca V 2.1, and Ca V 2.2 Ca 2ϩ channels; and (2) a small modification in the activation and inactivation parameters of the Ca V 2.1, only seen when the ␣ 1 subunit is expressed with the ␤ 2a subunit.
Effect of NCS-1 on Ca 2ϩ Current Amplitude-Effects of NCS-1 on P/Q-and N-type Ca 2ϩ channel current amplitudes have been documented over the past years but always based on experiments using either dominant negative mutants or overexpression of wild-type NCS-1 in native cells constitutively expressing the protein (18 -20). Our study is therefore the first to analyze the effects of NCS-1 on multiple types of Ca 2ϩ channel in the same environment, using heterologous expression of defined Ca 2ϩ channel subunits. In neurons, NCS-1 has been shown to clearly increase N-type Ca 2ϩ channel expression (20), with significant modifications of the electrophysiological parameters. The opposite effects have been found on P/Q-type Ca 2ϩ channels in bovine chromaffin cells, where a dominant negative mutant of NCS-1 increased Ca 2ϩ current amplitude (19) without effect on L-type Ca 2ϩ channels. Using cDNA injection, we show that all three channel types are down-regulated by co-expression of NCS-1. Our experimental conditions and our analysis of the expression level of the Ca V 2.1 subunit (Fig. 2C) suggest that this effect must take place after protein synthesis and could involve modifications in the correct folding and trafficking of the channel complex to the plasma membrane and/or regulation of the channel activity per se, taking place after insertion of the channel into the plasma membrane. Membrane expression of high voltage-activated Ca 2ϩ channels is known to rely on a short sequence located within the intracellular loop connecting the homologous domains I and II of the ␣ 1 subunit (37). This sequence, located close to the ␤ subunit binding site (AID), acts as a retention signal in the endoplasmic reticulum and thus reduces trafficking of the ␣ 1 subunit to the membrane, just like a surface expression brake. This brake is usually removed by the ␣ 1 /␤ subunit association, which occurs in the reticulum. The fact that the decrease in current amplitude induced by NCS-1 was only observed when ␤ 1 , ␤ 2 , or ␤ 4 subunits were expressed suggests that NCS-1 may interfere with this mechanism. A similar effect on channel expression has been recently reported (38) for the small GTPase Kir/gem, which binds directly to the ␤ subunit, thus preventing the ␣ 1 /␤ association and restoring the expression brake imposed by the retention signal. Direct specific binding of NCS-1 to this site and/or to the ␤ subunit is thus an attractive hypothesis to explain the reduction of the current amplitude but needs to be further explored by additional experiments designed to directly test the biochemical interactions between these subunits. Such a mechanism, however, may not be exclusive, and other pathways, acting directly or indirectly on channel activity, such as Src-dependent inhibition (28), or direct modulation of the Gprotein-coupled receptor pathways (19), may also exist. Although in our recording conditions, the Src-dependent and G-protein pathways could be discarded (the Src kinase inhibitor, PP1, had no effect; data not shown), it is worth noting that G-protein ␤␥ subunits and the Ca 2ϩ channel ␤ subunit possess very close binding sites on the main ␣ 1 Ca 2ϩ channel subunit (39 -41), leaving open the possibility that interactions between the ␣ 1 subunit on one hand, and G-protein, Ca 2ϩ channel ␤ subunit, and NCS-1 on the other hand, could be mutually exclusive and/or under the control of tissue-specific conditions. The I-II loop would therefore acts as a cross-road for cellspecific regulations.
Interestingly, NCS-1 had no effect on Ca V 2.1 Ca 2ϩ channel co-expressed with the ␤ 3 subunit, the ␤ subunit that displays the lowest affinity for the AID (42), and which also shows the weakest potency to increase current amplitude in expression systems (43). In neurons, where the N-type Ca 2ϩ channel is predominantly associated with the ␤ 3 subunit (44) the increase in vesicular transport and membrane trafficking induced by NCS-1, acting through the activation of the phosphatidylinositol 4-kinase ␤ (45), may overcome the effect on the ␣ 1 retention signal, poorly masked/removed by the ␤ 3 subunit, and lead to the observed overexpression of N-type Ca 2ϩ channels (20). In this scenario, the tissue specificity of the effects of NCS-1 on Ca 2ϩ channels should thus be critically dependent on the subunit composition of the channel. Whether NCS-1 acts directly or indirectly on the Ca 2ϩ channel ␣ 1 or ␤ subunit requires further investigation.
The fact that mutants that prevented Ca 2ϩ binding (NCS-1 E120Q ) and protein myristoylation (NCS-1 G2A ) both decreased the effects of NCS-1 on channel expression underlines the requirement for a fully functional NCS-1 protein to record these effects. Mutation of glutamate 120 to glutamine (NCS-1 E120Q mutant) disrupts a high affinity Ca 2ϩ binding site and impairs Ca 2ϩ -dependent conformational changes (19). This mutant acts as a dominant negative mutant for the regulation of the P/Q channel in chromaffin cells (19). In Xenopus oocytes the NCS-1 E120Q mutant appeared also less potent than wildtype NCS-1. However, a clear and significant decrease in current amplitude was nevertheless observed, suggesting that the mutation did not completely suppress NCS-1 activity. These differential effects constitute another argument in favor of the presence of different mechanisms working in chromaffin cells or Xenopus oocytes to regulate Ca 2ϩ channels, i.e. removal of a voltage-independent inhibition versus down-regulation of channel trafficking, and suggest that Ca 2ϩ -dependent changes are absolutely necessary only for the removal of the voltage-independent inhibition of the P/Q channels (19,28). Preserved interactions of NCS-1 E120Q with cellular proteins can be an argument for the dominant negative effect of the mutant, as suggested (19), but can also constitute an interesting area of investigation to identify preserved interactions that may be still functional and involved in a Ca 2ϩ -independent down-regulation of the channel activity.
Ϫ6 Mutation of the myristoylation site, NCS-1 G2A, is known to affect subcellular localization of NCS-1 (29), with minor modifications in the protein structure (46). The tendency of the NCS-1 G2A mutation to decrease Ca 2ϩ current amplitude was not found statistically significant. This may simply reflect a small decrease in the availability of the NCS-1 G2A mutant at early stages of protein assembly, because of the loss of the perinuclear localization (29) or because of more indirect effects, related to mutationinduced modifications in the degree of cooperativity in Ca 2ϩ binding (46). The construction and testing of the double mutant NCS-1 G2A/E120Q should help to solve this issue.
Effect of NCS-1 on Ca 2ϩ Current Properties-Beyond the decrease in current amplitude, we also noted that NCS-1 could specifically affect the electrophysiological properties of the Ca V 2.1 channel, but only when co-expressed with the ␤ 2a subunit. Increased inactivation and a positive shift of the currentvoltage curve were the most significant changes. These effects were seen in the presence of extracellular Ba 2ϩ or Ca 2ϩ and therefore, in our conditions, did not seem to be Ca 2ϩ -dependent. The fact that the decrease in current amplitude and the modifications in the channel properties did not have the same ␤ subunit specificity suggested different underlying mechanisms. Direct interactions between Ca 2ϩ channels and NCS-1 at the plasma membrane have not been reported so far, but the presence of NCS-1 in synaptic-like microvesicles and its colocalization in presynaptic terminals with the proteins of the soluble N-ethylmaleimide-sensitive factor attachment protein receptor complex, VAMP (45) and syntaxin (18), also known to interact with P/Q Ca 2ϩ channels (1), suggest that NCS-1 is appropriately located for these potential functional interactions. Indeed, a previously published work (18) reported an enhancement of the facilitation of the P/Q-type Ca 2ϩ channel in the Calyx of Held, occurring via an acceleration of current activation. A similar frequency-dependent facilitation was also apparent in oocytes (data not shown), but the large membrane capacitance of the oocytes did not allow a precise analysis of the modifications in the activation kinetics. Interestingly, in the Calyx of Held, these effects were recorded during acute perfusion of NCS-1 and could be occluded by perfusion of a Cterminal peptide of NCS-1, suggesting direct interactions with the channel (18). A large hydrophobic crevice in this region has been proposed as a potential site for target recognition (15).
Alternatively, it has also been proposed that NCS-1 could activate the same targets as calmodulin (35), a regulatory element of the Ca V 1.2 and Ca V 2.1 Ca 2ϩ channels, responsible for the Ca 2ϩ sensitivity of inactivation and facilitation (10,12,13,47,48). The binding site for calmodulin on these channels is positioned on their intracellular C-terminal tails and is constituted of multiple, Ca 2ϩ -dependent, or constitutive microsites. The same sites can accommodate CABP1, another Ca 2ϩ -binding protein, which is able to displace calmodulin from its sites, to induce a Ca 2ϩ -independent positive shift of the currentvoltage curve accompanied by an acceleration of the inactivation kinetics (11). These effects are reminiscent of the action of NCS-1 on Ca 2ϩ channels in Xenopus oocytes. However, several questions need to be answered before one can speculate that NCS-1 and CABP1 share the same regulatory pathway. For example, CABP1 has only been tested on Ca V 2.1 channels containing the ␤ 2a subunit, and we have no information, for the moment, on its ␤ subunit specificity.
In conclusion, we demonstrate here that NCS-1-dependent regulation of high threshold Ca 2ϩ channels includes modifications in the current amplitude and the electrophysiological properties. These modifications are specifically modulated by the ␤ subunit of the Ca 2ϩ channel. Our results, together with previously published work on NCS-1, CABP1, calmodulin, and Ca 2ϩ channels, suggest that these effects could take place at two sites on the ␣ 1 subunit known for their regulatory role, the I-II loop and the C-terminal tail of the channel. These sites also interact with the Ca 2ϩ channel ␤ subunit and the G␤␥ subunits of the G-protein suggesting that regulation via G-protein-coupled receptors may also be affected. These results provide a possible molecular explanation for the tissue and Ca 2ϩ channel type specificity of NCS-1, which can be tested biochemically and pharmacologically. They also suggest that down-regulation of Ca 2ϩ channel activity may not be restricted to neuronal or neuroendocrine cells but may also take place in cardiac myocytes, where NCS-1 and the L-type Ca 2ϩ channels are expressed (26).