Voltage-gated Mobility of the Ca2+ Channel Cytoplasmic Tails and Its Regulatory Role*

Transient increase in intracellular free Ca2+ concentration generated by the voltage-gated Cav1.2 channels acts as an important intracellular signal. By using fluorescence resonance energy transfer combined with patch clamp in living cells, we present evidence for voltage-gated mobility of the cytoplasmic tails of the Cav1.2 channel and for its regulatory role in intracellular signaling. Anchoring of the C-terminal tail to the plasma membrane caused an inhibition of its state-dependent mobility, channel inactivation, and CREB-dependent transcription. Release of the tail restored these functions suggesting a direct role for voltage-gated mobility of the C-terminal tail in Ca2+ signaling.

determined after background subtraction. Corrected intensity of FRET (I FRET c) was calculated (20) as (I FRET Ϫ I ECFP ϫ 58.5% Ϫ I EYFP ϫ 11.5%). In the patch clamp experiments, the acquisitions of fluorescence ranging from 50 to 300 ms were obtained with simultaneous recording of the current at the indicated conditioning voltages under steady state. In some cases, the images were adjusted pixel-by-pixel using the reference channel of regions of interest. With the acceptor photobleaching, we used an excitation filter 436/10 nm and an emission filter 470/30 nm for ECFP and an excitation filter 500/10 nm and emission filter 535/20 nm for EYFP. The apparent efficiency of FRET was calculated as (I ECFP* Ϫ I ECFP )/I ECFP* , where I ECFP and I ECFP* Ϫ intensities of ECFP fluorescence before and after acceptor (EYFP) photobleaching, respectively. Normalization of the corrected FRET ratios against EYFP or ECFP fluorescence showed very similar values (Table I). Corrected FRET and bleed-through values were obtained by standard approach (20,21) according to Equation 1, where a and b are, respectively, the norm of the percentage of ECFP and EYFP bleed-through under the FRET filter set, and I FRET , I ECFP and I EYFP are intensities in each region of interest under FRET, ECFP, and EYFP filter sets, respectively. For the double-labeled channels, we have experimentally determined that a ϭ 58.5% and b ϭ 11.5%. Background subtraction was carried out using the standard approach (e.g. Ref. 22). In each image, we determined the average fluorescence from the cell-free regions. Under our experimental conditions, the average fluorescence measured from these specified regions was not different from the fluorescence measured with unused substrate covered with the bath solution only. Variability of the background fluorescence ranged from 9 to 14, whereas those of the experimental cell fluorescence varied from 50 to 250 units of the same arbitrary scale. To obtain ratio of images, 10 was added to each pixel of the image corrected by the background subtraction to avoid division by zero. Plotting of corrected FRET images was carried out according to standard and well established procedures (for example, see Refs. 20 and 23). Because our images were obtained from the same cells with the same level of ECFP expression, there was no need to additionally normalize FRET images against ECFP. Under these conditions, we focused only on the relative changes of fluorescence at different holding potentials. Comparison of the ratios of corrected FRET (ϩ40 mV/Ϫ90 mV) with those additionally normalized against EYFP, ECFP, and square root of their product according to Xia and Liu (24) (see Table I) clearly demonstrates that additional normalization did not significantly change the results of corrected FRET ratio. This result was obtained probably because the ratio of ECFP to EYFP in experiments with the double-labeled channels was always equal to 1.
CREB-dependent Transcription Activation-The plasmids YKIDN and KIXCN coding for KID and KIX domains (25), respectively, were co-expressed (1:1) in COS1 cells with other cDNAs at 2.5:1 ratio to cDNA coding for an ␣ 1C subunit. To monitor transcriptional activation under voltage clamp conditions, we used the perforated patch clamp technique (26). ␤-Escin (20 M, Sigma) was added to the pipette solution containing potassium gluconate 120, NaCl 10, MgATP 2.5, HEPES 5, and KCl, 20; pH 7.2. External solution contained (in mM) NaCl, 140; KCl, 5.4, MgCl 2 1, HEPES 5, CaCl 2 2.0, and glucose 5.5, pH 7.4. Depolarization to ϩ20 mV from the holding potential of Ϫ90 mV was applied to elicit maximum Ca 2ϩ current. To monitor changes of free Ca 2ϩ concentration, cells were loaded with Fluo-4 by incubation for 20 min at 37°C in 5 M Fluo-4 AM (Molecular Probes) added to the external solution. For de-esterification of the probe, cells were incubated for another 20 min before experiment. Ca 2ϩ measurements were performed at 20 -22°C using confocal microscope PCM2000 (Nikon, Inc.) with excitation by an argon laser at 488 nm and recording of the fluorescence emission at Ͼ515 nm.

RESULTS
Acceptor Photobleaching Assay-Confirming earlier findings (13,27), fusion of the green fluorescent protein variants to the cytoplasmic N and/or C termini did not compromise channel function. The electrophysiological properties of the (EYFP) N -␣ 1C,77 -(ECFP) C channel were similar to those of the wild-type channel (13) except of the acceleration by ϳ15% of the Ba 2ϩ current inactivation (Table II). A direct FRET acceptor photobleaching assay provided strong evidence of FRET in the (EYFP) N -␣ 1C,77 -(ECFP) C channel (Fig. 1). Illumination of the transfected COS1 cells for 15 min with a mercury lamp caused on average Ͼ90% photobleaching of EYFP. The resulting increase in donor fluorescence (shown as yellow-green and yellowred signals, Fig. 1A, c and e, respectively) is a direct demonstration of FRET (11). As a positive control, under the same conditions, FRET was observed (Fig. 1A, d and f) with the co-expressed mixture of the EYFP-PH and ECFP-PH domains of phospholipase C␦1 (28). In both cases FRET was confined to the plasma membrane region where the functional channel molecules reside. We found that the mixture of co-expressed (1:1) single N-and C-tail-labeled channel isoforms did not show substantial intermolecular FRET (Fig. 1B) indicating that with the double-labeled ␣ 1C channels we recorded predominantly intramolecular FRET.
Voltage-dependent FRET in the Inactivated State-FRET is the result of long range dipole-dipole interactions between fluorophores. FRET depends on the distance (r 6 ) between the donor and acceptor fluorophore and, to a lesser extent, the relative orientation ( 2 ) of the dipoles (12). To determine the differences in relative proximity and/or angular orientation of the tagged ␣ 1C tails in the functionally distinct resting, conducting, and inactivated states of the channel, we combined FRET imaging with voltage clamp. We used quantitative analysis of reversibly changing FRET in place of acceptor photobleaching as the fluorophores became irreversibly damaged by photobleaching. Simultaneous monitoring of the Ba 2ϩ current provided a direct assessment of the Ca v 1.2 channel's transition into the resting, conducting, or inactivated state prior to FRET imaging. The images were acquired with a three-filter set system, successfully used elsewhere (20,29), for sensitized acceptor emission. The recombinant channel, composed of the wildtype ␣ 1C,77 and accessory ␤ 1 and ␣ 2 ␦ subunits, was stabilized in either the resting or inactivated state by whole-cell voltage clamp at Ϫ90 or ϩ40 mV, respectively. Depolarization to ϩ40 mV was selected to evoke maximum activation of the channels under conditions when the state of inactivation could be monitored. Thus, the amplitudes of the Ba 2ϩ currents (Fig. 2d) were less than 50% of the maximum ones. The analysis of FRET images of a cell expressing the (EYFP) N -␣ 1C,77 -(ECFP) C chan-  (19) nel ( Fig. 2A) showed that the sequential transitions between the resting (Ϫ90 mV, Fig. 2A, c) and the inactivated states of the channel (ϩ40 mV, Fig. 2A, b) produced a fully reversible increase in FRET: (I FRET c) ϩ40 mV /(I FRET c) Ϫ90 mV ϭ 2.9 Ϯ 0.7 (n ϭ 24). The steady state inactivation analysis indicated that ϳ7% of the Ba 2ϩ -conducting (EYFP) N -␣ 1C,77 -(ECFP) C channels remain available (Table II), suggesting that Ն93% of the channels are inactivated and Յ7% remain in the resting closed state at the end of the depolarization pulse shown in Fig. 2A. Since the resting state is characterized by low FRET, this fraction has only slightly contributed to the measurement of FRET. Taken together, these data indicate that the cytoplasmic tails of the channel in the inactivated state are rearranged into a conformation that gives greater FRET as compared with the closed resting state. The electrophysiological properties of the ␣ 1C channel are not significantly changed by the deletion of the N-terminal tail (30). In contrast, the proximal half of the C-terminal tail is essential for channel function (13,30,31). A fusion of ECFP to the full size, 662-amino acid C-terminal tail of ␣ 1C,77 did not alter the electrophysiological properties of the (EYFP) N -␣ 1C,77F -(ECFP) C channel (data not shown), but reduced efficiency of FRET determined in the acceptor photobleaching assay to 0.19 Ϯ 0.02 (n ϭ 29), thus indicating that apparent separation of the fluorophores has increased. However, the intensity of FRET in the inactivated state (ϩ40 mV) of this full-length (EYFP) N -␣ 1C,77F -(ECFP) C channel was not significantly different as compared with the truncated channel.
Voltage-dependent FRET in the Conducting State of the Ca v 1.2 Channel-Rapid spontaneous channel inactivation complicates direct FRET imaging of the channel in the transient conducting state. Replacement of the cytoplasmic accessory ␤ 1 subunit with ␤ 2a (32) significantly slowed inactivation (compare Ba 2ϩ current traces on d in Fig. 2, A and B) and thus

(A) and fluorescent labeled (EYFP) N -␣ 1C,77 -(ECFP) C (B), and (EYFP) N -␣ 1C,IS-IV -(ECFP) C (C) Ca 2ϩ channels
The ␣ 1C subunits were co-expressed with ␤ 1 and ␣ 2 ␦ accessory subunits. Ba 2ϩ currents were elicited by 600-ms test pulses to ϩ10 mV from a holding potential of Ϫ90 mV. The bath medium contained 20 mM Ba 2ϩ . The time constants of the Ba 2ϩ current inactivation fast and slow were determined by two-exponential fitting. (The approximated slow values are presented solely to reflect the fact that slow inactivation is completely inhibited in the labeled ␣ 1C,IS-IV channel.) V max , voltage for the peak current, and V 0.5 , voltage at 50% of the Ba 2ϩ current activation were determined from the current-voltage relationship. V 0.5in , the voltage at half-maximum of inactivation, and I sust , fraction of non-inactivating current, were determined from the fitting of steady-state inactivation curves by Boltzmann function. Steady-state inactivation curves were measured using a two-step voltage clamp protocol. A 1-s conditioning pulse was applied at 30-s intervals with 10-mV increments from the holding potential V h ϭ Ϫ90 mV followed by a 250-ms test pulse to ϩ10 mV. Peak current amplitudes were normalized to maximum value. All fittings were performed according to Ref. 33. Number of tested cells is shown in parentheses. allowed the channel to be maintained in a predominantly conducting state. Voltage-dependent FRET was measured with the (EYFP) N -␣ 1C,77 -(ECFP) C /␤ 2a channel at the end of a 3-s depolarization at ϩ40 mV. Under these conditions, the Ba 2ϩ current activation was almost maximal, whereas inactivation was minimal with respect to the time of FRET acquisition (marked by the red bar above the current traces). The voltage-dependent increase of FRET (3.8 Ϯ 0.8, n ϭ 15, see Fig. 2B) resulting from the transition of the channels from the resting (Ϫ90 mV) to a predominantly conducting state (ϩ40 mV) was found not to be significantly different from that determined for the inactivated state (2.9 Ϯ 0.7, Fig. 2A). Thus, the folding of the channel tails characterized by FRET in the ensemble of predominantly conducting channels was on average closer to Image areas indicated by arrows were digitally magnified to demonstrate confinement of FRET to the plasma membrane. Note that in every case the Ba 2ϩ current recordings (d) provide evidence of the channel state achieved prior to FRET image acquisitions (marked by red bars). FRET at holding potential V h ϭ Ϫ90 mV was recorded for the same duration of time before and after the depolarization pulse. those determined for the inactivated than those determined for the resting state of the same channel.
This finding was independently confirmed by the study of the ␣ 1C,IS-IV channel that is deprived of slow inactivation by mutations introduced in the pore region (14). The (EYFP) N -␣ 1C,IS-IV -(ECFP) C channel expressed in COS1 cells showed the characteristic sustained Ba 2ϩ current (Fig. 2C, d). The amplitude of the Ba 2ϩ current through the ␣ 1C,IS-IV channel was smaller than those through the ␣ 1C,77 channel. Similar changes were observed earlier for other isoforms of the Ca v 1.2 channel with impaired Ca 2ϩ -induced inactivation (33). This was found to be due to a lower open probability and a 10 -15% reduction in single channel conductance (13). In the stable conducting (ϩ40 mV) state of the ␣ 1C,IS-IV channel (Fig. 2C, b), the corrected  Fig. 2A, b) suggesting further reduction in the apparent distance between the labeled tails and/or dipole reorientation of the fluorophores (12) leading to an increase of FRET as compared with the resting state (Ϫ90 mV, see Fig. 2C, c). Taken together, these data point to distinct rearrangements of the ␣ 1C tails associated with voltage gating.
Role of the Voltage-gated Mobility of the C-terminal Tail in the Ca v 1.2 Channel Inactivation-To investigate a regulatory role of the voltage-gated rearrangements of the ␣ 1C C-terminal tail, the tail was anchored to the plasma membrane via the PH domain of phospholipase C␦1 (Fig. 3). This domain binds specifically to phosphatidylinositol bisphosphate in the inner leaflet of the plasma membrane. Hydrolysis of phosphatidylinositol bisphosphate by activation of phospholipase C induces redistribution of the PH domain to the cytoplasm (15,34).
The (EYFP) N -␣ 1C,77 -(PH-ECFP) C channel with anchored Ctail was co-expressed in COS1 cells with epidermal growth factor (EGF) receptor to permit a release of the PH-tagged C-terminal tail by activating phospholipase ␥ and phosphatidylinositol bisphosphate hydrolysis in response to exposure of the cell to EGF (Fig. 3A). The Ba 2ϩ current through the channel with the anchored C-terminal tail was activated in the characteristic range of membrane potentials but exhibited a very slowly inactivating component of the Ba 2ϩ current (Fig.  3B, b). Membrane trapping of the N-terminal tail did not alter the inactivation properties of the channel (data not shown).
Very little, if any, FRET was observed with the anchored C-terminal tail in both the conducting (Fig. 3B, c) and the resting (d) states of the channel. The release of the tail by EGF treatment irreversibly restored the ability of the channel to inactivate fully as can be seen from the complete decay of the current (Fig. 3C, b) and increased FRET in the inactivated state (Fig. 3C, c). The ratio (I FRET c) ϩ40 mV /(I FRET c) Ϫ90 mV ϭ 1.9 Ϯ 0.9 (n ϭ 5, p Ͻ 0.05) was not different to that observed with the (EYFP) N -␣ 1C,77 -(ECFP) C ( Fig. 2A) and (EYFP) N -␣ 1C,77F -(ECFP) C (not shown) channels. Thus, limitations imposed on free movement of the C-terminal tail of the ␣ 1C channel affect inactivation properties as well as the associated FRET. When released, the C-tail appears to assume a functional conformation as determined by both the return of the normal inactivation properties of the channel and voltage-dependent FRET. This precludes a re-insertion of the PH domain into the membrane. Even a 30-min washout period, the longest that we were able to achieve without losing the voltage clamp, did not restore the membrane association of the PH domain distinguished by the properties of the channel with the trapped tail.
Role of the Voltage-gated Mobility of the C-terminal Tail in Regulation of CREB-dependent Transcription-The amplitude of the voltage-gated mobility of the Ca v 1.2 channel C-terminal tail may be sufficient to have a role in Ca 2ϩ signal transduction. It has been shown previously (8) that Ca v 1.2 channels are important for Ca 2ϩ -induced activation of cAMP-responsive element-binding protein (CREB)-dependent transcription. In this work, to study the role of the voltage-gated mobility of the Ca 2ϩ channel C-terminal tail for transcriptional activation, we investigated the interaction between KID and KIX domains of CREB and co-activator CREB-binding protein under voltage clamp conditions by monitoring FRET between (EYFP)-KID and (ECFP)-KIX, both containing nuclear localization sequences (25). Use of perforated patch clamp technique (26) allowed us to preserve the integrity of the cytoplasmic content of the cell, crucial in retaining components of signaling cascade involved in CREB-dependent transcription. We found that when the C-terminal tail of the ␣ 1C,77 -(PH) C channel was anchored via the PH domain to the plasma membrane (Fig. 4A, a), activation of the channel by depolarization to ϩ20 mV increased intracellular free Ca 2ϩ concentration ([Ca 2ϩ ] i ) detected by the free Ca 2ϩ indicator Fluo-4 ( Fig. 4A, a, see image on the lower panel). However, stimulation of the sustained Ca 2ϩ conductance of the anchored channel by depolarization (Fig. 4A, d) did not cause strong activation of CREB-dependent transcription despite the presence of the CaM-binding IQ motif of the C-tail 1624 -1635 (35), previously identified as important for CREB activation (8). Release of the C-terminal tail of the channel, induced at Ϫ90 mV by ACh stimulation of the coexpressed type 1 muscarinic acetylcholine receptor (M1AChR), was accompanied by activation of the inositol 1,4,5-trisphosphate-dependent Ca 2ϩ release but did not induce substantial activation of CREB-dependent transcription (Fig. 4A, b). It was the release of the C-terminal tail of the channel combined with the activation of Ca 2ϩ conductance by depolarization that was essential to re-establish the voltage-gated signaling sufficient for CREB-dependent transcription activation (Fig. 4A, c). The fact that this response was inhibited by co-expression of CaM 1234 (Fig. 4B), a Ca 2ϩ -insensitive analogue of CaM retaining the affinity to CaM-binding sites (36), points to a critical role of CaM in signal transduction by the voltage-gated mobile C-terminal tail of the ␣ 1C channel. Selective disruption of the apo-CaM-binding site 1572-1598 (37-39) on the C-terminal tail of the ␣ 1C channel (33) (Fig. 4C) abolished activation of CREB-dependent transcription despite unrestricted mobility of the C-terminal tail and the presence of CaM and IQ motif (8). These results, for the first time, demonstrated the role of the Ca 2ϩ channel C-tail as a voltage-gated mobile carrier of the signal for CREB-dependent transcriptional activation (Fig.  4D). This signal appears to be associated with apo-CaM binding to the C-tail and may involve its Ca 2ϩ loading (9) and translocation to the IQ motif (8,36), a Ca 2ϩ -filled CaM-binding site on the C-tail. DISCUSSION In this report, we provide strong evidence that the C-terminal cytoplasmic tail is a functionally important moving part of the voltage-gated Ca 2ϩ channel. Previously implicated as cytoplasmic channel elements that respond to voltage gating were the cytoplasmic domain between repeat II-III of the skeletal muscle ␣ 1S subunit (40), inactivation gates of Na ϩ channel (41), and the ball-and-chain (N-type) inactivation determinant of K ϩ channel (7,42). The state dependence of FRET in the ␣ 1C channel is associated with conformational refolding of the cytoplasmic parts in the resting, conducting, and inactivated states. FRET increased upon transition from the resting to the inactivated or conducting states of the Ca 2ϩ channel suggesting gating-dependent conformational mobility of the channel tails. The anchoring of the C-terminal tail to the plasma mem-(C) by disruption of the apoCaM-binding site in the free C-tail of the ␣ 1C,77L channel (33), which lost Ca 2ϩ -dependent inactivation property as the result of this mutation. Upper panels show voltage and ACh application protocol. Twelve steps of depolarization from Ϫ90 mV to ϩ20 mV, 10-mV increment for 1-s pulse with 10-s intervals were applied, and a 100-ms FRET images a-c were recorded at the end of the last depolarization before (a) or after (c) 5-min application of 5 M ACh. b shows effect of 5 M ACh applied at Ϫ90 mV. Left panels, phase-contrast images of expressing cells with the shadow of patch pipette; scale bars, 4 m. Free Ca 2ϩ measurements with Fluo-4 (lower panels) were carried out under the same conditions and show that the activation of CREB-induced transcription did not depend strongly on free intracellular Ca 2ϩ unless the C-tail was released. Schematic diagrams depict positions of the ␣ 1C C-tail vis-à -vis the plasma membrane. D, histogram of FRET intensity ratio of images c and b (n ϭ 3, p Ͻ 0.01). brane impaired both the channel inactivation and FRET until fully recovered with its release. Thus, the state-dependent mobility of the cytoplasmic C-terminal tail is essential for regulation of the Ca 2ϩ channel.
In addition to relative proximity of the tagged ␣ 1C tails, their angular orientation may contribute to FRET because the environment of the fluorophores in the (EYFP) N -␣ 1C -(ECFP) C channels may be structured and their segmental motions may not independently randomize the orientations ( 2 Յ 2/3) (12). The 2 factor adds to the uncertainties complicating interpretation of the FRET measurements in terms of translational distances and is of particular concern when dipoles become oriented perpendicular to one another ( 2 ϭ 0). Because FRET was consistently observed, such perpendicular orientation seems unlikely. In fact the corresponding voltage-dependent movement of the ␣ 1C cytoplasmic C-tail is sufficient to reach other targets and be involved in signal transduction. An important regulatory signal associated with the C-terminal tail is CaM, which supports Ca 2ϩ -induced inactivation of the Ca 2ϩ channel (35,37,38). It is possible that the C-terminal tail mobility may, in a state-dependent manner, provide a coordinated transfer of CaM between the channel pore inner mouth, where it is loaded by Ca 2ϩ when the channel opens, and protein targets like CaM-activated protein kinase (8), where Ca 2ϩ /CaM exerts signaling. This hypothesis may be generalized to the role of CaMcarrying mobile tails in other Ca 2ϩ channels (29,43).