New Determinant for the CaVβ2 Subunit Modulation of the CaV1.2 Calcium Channel*

Cavβ subunits support voltage gating of Cav1.2 calcium channels and play important role in excitation-contraction coupling. The common central membrane-associated guanylate kinase (MAGUK) region of Cavβ binds to the α-interaction domain (AID) and the IQ motif of the pore-forming α1C subunit, but these two interactions do not explain why the cardiac Cavβ2 subunit splice variants differentially modulate inactivation of Ca2+ currents (ICa). Previously we described β2Δg, a functionally active splice variant of human Cavβ2 lacking MAGUK. By deletion analysis of β2Δg, we have now identified a 41-amino acid C-terminal essential determinant (β2CED) that stimulates ICa in the absence of Cavβ subunits and conveys a +20-mV shift in the peak of the ICa-voltage relationship. The β2CED is targeted by α1C to the plasma membrane, forms a complex with α1C but does not bind to AID. Electrophysiology and binding studies point to the calmodulin-interacting LA/IQ region in the α1C subunit C terminus as a functionally relevant β2CED binding site. The β2CED interacts with LA/IQ in a Ca2+- and calmodulin-independent manner and need LA, but not IQ, to activate the channel. Deletion/mutation analyses indicated that each of the three Cavβ2/α1C interactions is sufficient to support ICa. However, β2CED does not support Ca2+-dependent inactivation, suggesting that interactions of MAGUK with AID and IQ are crucial for Ca2+-induced inactivation. The β2CED is conserved only in Cavβ2 subunits. Thus, β2CED constitutes a previously unknown integrative part of the multifactorial mechanism of Cavβ2-subunit differential modulation of the Cav1.2 calcium channel that in β2Δg occurs without MAGUK.

Voltage-gated Ca v 1.2 calcium channels couple membrane depolarization to excitation in a wide variety of cells. The voltage gating, or membrane potential-dependent opening and closing of a channel, is associated with conformational changes in the pore-forming (␣ 1 ) subunit (1). Ca v 1.2 channels require auxiliary ␣ 2 ␦ and ␤ (Ca v ␤) subunits to integrate the functional channel into the plasma membrane (PM) 3 and facilitate voltage gating of the current (2). How Ca v ␤ subunits mediate these functions and what are the Ca v ␤-specific determinants are important questions to be answered. Members of the Ca v ␤ family are structurally divergent. Four different Ca v ␤ subunit genes code for ␤ 1 -␤ 4 subunit variants, some of which are alternatively spliced. Cytosolic Ca v ␤ subunits bind to the 18-amino acid ␣ 1 -interaction domain (AID) of the cytoplasmic linker between internal repeats I and II of the pore-forming ␣ 1 subunit (Fig. 1), stimulate the Ca 2ϩ channel current (I Ca ), and shift the currentvoltage (I-V) curve to more negative voltages (3,4). The AID is conserved between the Ca v 1 and Ca v 2 subfamilies of Ca 2ϩ channels. It is located in close proximity to the transmembrane segment IS6 that is a part of the pore domain (5) implicated in voltage-dependent inactivation of the channel (6,7). A common central region of Ca v ␤ subunits has structural similarity with the membrane-associated guanylate kinase (MAGUK) motif (8). When co-expressed with the ␣ 1C subunit, the Ca v ␤ MAGUK domain increased Ba 2ϩ current (I Ba ) amplitude and shifted the steady-state activation (9). Confirming tight binding of the central Ca v ␤ domain to the ␣ 1C subunit, diffraction studies revealed structural patterns that were implicated in interaction with the AID (10 -12). However, variant-specific regulatory properties of Ca v ␤ appear to be AID-independent. Although different Ca v ␤ subunits have MAGUK, they modulate Ca 2ϩ channels with individual characteristic variations. For example, the primary cardiac ␤ 2a subunit did not fully substitute the ␤ 1a subunit in skeletal muscle EC coupling although it restored activation of I Ca and gating of Ca 2ϩ transients (13). Unlike other Ca v ␤ subunits, ␤ 2a endows the distinct cardiac phenotype by not supporting facilitation of the Ca 2ϩ channel current by a depolarizing prepulse (14). This general picture was further detailed by FRET microscopy combined with patch clamp that demonstrated differential voltage-dependent rearrangement of Ca v ␤ subunits vis à vis the ␣ 1C subunit N terminus (15). Unlike the Ca v ␤ 1a subunit, Ca v ␤ 2 exhibited no such mobility. These and other findings show that a number of Ca v ␤ functions do not rely on AID as a main site of regulation and may involve other determinants (16,17). Thus, identification of functional motifs that are unique for different Ca v ␤ subunits may give an important insight into the functional specificity of the Ca v ␤-dependent modulation. One feasible approach is to explore the naturally occurring Ca v ␤ splice variants (18). In line with this was the discovery of two new functionally active small splice variants of the human cardiac ␤ 2 subunit lacking the central domain (19). These ␤ 2f and ␤ 2⌬g subunits share a 153amino acid distal C-terminal region common to all known "large" Ca v ␤ 2 subunits (␤ 2a -␤ 2e ) (20) suggesting that this region may have a role of an essential Ca v ␤ 2 determinant. Our attention to this region of the ␤ 2 subunit was stimulated by the finding that ␤ 2⌬g supports I Ca on co-expression with ␣ 1C and ␣ 2 ␦ in Ca v ␤-free COS1 cells. Because large and small Ca v ␤ 2 splice variants convey sharply different inactivation kinetics, it seems apparent that, in addition to MAGUK, there is a C-terminal determinant (defined here as ␤ 2 CED) that is common only to Ca v ␤ 2 subunits and thus may contribute to the Ca v ␤ 2specific tuning of the channel modulation by large Ca v ␤ 2 subunits. In the case of the small Ca v ␤ 2 subunits, ␤ 2 CED may play the key regulatory role. This intriguing possibility prompted us to locate ␤ 2 CED and characterize the properties of ␤ 2 CEDmodulated Ca 2ϩ channels that rely on ␤ 2 CED-dependent, MAGUK-independent modulation.
Immunoprecipitation Analysis-Human embryonic kidney (HEK) 293 cells were used for the IP analysis because of high expression efficiency. To exclude endogenous Ca v 1.2 subunits from the analysis, we expressed only tagged subunits and used antibodies against the tags. For IP-Western blot analysis, Ϸ80% confluent early passage HEK293 cells in 100-mm culture dishes were transfected with selected plasmids (for details, see figure legends) using Effectene (Qiagen) according to the manufacturer's instructions. 72 h after transfection, cells were harvested and washed 3 times with phosphate-buffered saline. To improve the yield and stability of co-IP, cells were subjected to a standard cross-linking reaction (22,23) by incubation with the cell-permeant thiol-cleavable reagent dithiobis(succinimidyl propionate) (1 mM) (Pierce) at room temperature for 30 min. Cross-linking was stopped by incubation of cells with 20 mM Tris-HCl (pH 7.5) for 15 min. Cells were lysed with a Cel-Lytic M lysis reagent (450 l/plate, Sigma) containing a protease inhibitor mixture (Sigma, 1/100 dilution) supplemented with 1 mM phenylmethylsulfonyl fluoride. To ensure direct interaction, the microsomal fraction was used for co-IP experiments involving ␣ 1C . The 80-l aliquots of total lysates were kept to verify the expression of each protein (see "input" on immunoblots). Co-IP was performed with the selected antibodies according to the manufacturer's instructions. Briefly, co-IP with anti-FLAG antibody (Ab) was performed with 40 l/reaction of a monoclonal EZ View TM ANTI-FLAG M2 affinity gel (Sigma) at 4°C overnight, and the immunoprecipitates were eluted by incubation with 10 g of 3ϫ FLAG peptide (Sigma) in 100 l of TBS solution (pH 7.4) at 4°C for 1 h. The co-IP with anti-LC Ab was performed with 5 l of a Living Colors Full-Length A.V. polyclonal Ab (Clontech) using Protein A-agarose (Sigma) as carrier (overnight at 4°C), and the immunoprecipitates were eluted by boiling for 5 min at 95°C. Dithiobis(succinimidyl propionate) was cleaved by incubation of the co-immunoprecipitate and input samples with 5% ␤-mercaptoethanol at 100°C for 5 min or with 50 mM dithiothreitol at 37°C for 30 min (only for ␣ 1C ) before SDS-PAGE. SDS-PAGE and immunoblotting with the indicated antibodies were performed according to standard protocols. The following primary antibodies were used: anti-FLAG M2 monoclonal Ab (2 g/ml, Sigma) for the FLAG-tagged proteins, Living Colors Full-Length A.V. monoclonal Ab (0.5 g/ml, Clontech) for the fluorescent tagged proteins, and streptavidin-horseradish peroxi-  B, representative trace of I Ca generated in response to V t ϭ ϩ30-mV applied from V h ϭ Ϫ90 mV to COS1 cells transfected by EYFP N -␣ 1C,77 and ␣ 2 ␦ subunits in the absence of Ca v ␤ subunits. No current was observed between 0 and ϩ50 mV. C, superimposed traces of the maximal I Ca through the EYFP N -␣ 1C,77 /␣ 2 ␦ channel with ␤ 2 CED (trace 1) or ECFP N -␤ 2 CED (trace 2) recorded at V t ϭ ϩ40 mV and normalized to the same amplitude. The ␣ 1C , ␣ 2 ␦, and ␤ 2 CED subunits were co-expressed in a 1:1:1 molar ratio. No significant difference in the kinetics of the currents was observed.
dase (1/1000 dilution, Invitrogen) for the biotin-tagged proteins. Nitrocellulose membrane (Invitrogen) was used for immunoblot analysis of ␣ 1C co-IP experiments and polyvinylidene difluoride membrane (Invitrogen) was used for all other studies. Electrophysiology-The Effectene kit (Qiagen) was used for transfection of COS1 cells as described previously (15) under conditions optimized for a total amount of 0.2 g of DNA per 35-mm Petri dish. Constructs were expressed in a 1:1 molar ratio. COS1 cells were grown on poly-D-lysine-coated coverslips (MatTek) in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. Whole-cell patch clamp recordings were performed as described (24) (15). Voltage protocols were generated and data were digitized, recorded, and analyzed using pClamp 8.1 software (Axon Instruments). Test pulses were applied at 15-s intervals from the holding potential V h ϭ Ϫ90 mV. Currents were filtered at 1 kHz, sampled at 2.5-5 kHz, and corrected for leakage using an on-line P/4 subtraction protocol. At the end of experiments, channels were routinely tested for sensitivity of I Ca to the inhibition by the dihydropyridine blocker PN200-110 (see examples). I-V curves were obtained by step depolarization to test potentials in the range of Ϫ60 to ϩ90 mV (with 10-mV increments) applied from V h . Steady-state inactivation curves were measured with conditioning pulses (1 s) applied from V h ϭ Ϫ90 mV up to Ϫ60 to ϩ40 mV with 10-mV increments followed by a 100-ms test pulse. Peak current amplitudes were normalized to the maximal value. Averaged I-V curves were fit with the equation: where G max is maximum conductance, E rev is reversal potential, V 0.5,act the voltage at 50% of the current (I) activation, and k I-V the slope factor. Steady-state inactivation curves were fitted by a Boltzmann function: where V is the conditioning pulse voltage; V 0.5,in is the voltage at half-maximum of inactivation, k i is a slope factor, a and b are fractions of noninactivating and inactivating components of the current, respectively. To estimate the time constant of inactivation, currents were fitted with the Chebyshev method according to the standard exponential function, where I i is the amplitude of the inactivating component of the current, is the time constant of inactivation, and I 0 is the non-inactivating component of the current. Statistical values are given as mean Ϯ S.E. Error bars in the figures are S.E., n, number of tested cells. Differences were considered significant if Student's t test showed p Ͻ 0.05. Imaging-Cell images were recorded with a 14-bit Hamamatsu digital camera C9100-12 mounted on the Nikon epifluorescent microscope TE200 (60 ϫ 1.2 N.A. objective) equipped with multiple filter sets (Chroma Technology, Rockingham, VT). Excitation light was delivered by a 175-watt Xenon lamp. Images were obtained and analyzed using C-Imaging software program (Compix, Sewickley, PA).

Selection of Appropriate Expression System-Electrophysi-
ological studies of Ca 2ϩ channels are traditionally based on the use of HEK293 cells. However, several independent careful evaluations have shown that these cells contain endogenous Ca 2ϩ channel subunits and exhibit I Ca at a level of 1-3 pA/pF (25,26). Thus, HEK293 cells can be used safely for the functional analysis of recombinant Ca 2ϩ channels only when the amplitude of the current is large enough to ignore the contribution of the endogenous channels. To avoid this problem, in this study we used COS1 cells because they lack endogenous Ca v 1.2 subunits (27): (a) Western blot analysis with anti-␣ 1C Ab revealed no detectable endogenous ␣ 1C in non-transfected COS1 cells ( Fig. 2A, lane 1), and (b) no Ca 2ϩ channel activity was observed in COS1 cells expressing recombinant ␣ 1C and ␣ 2 ␦ subunits (Fig. 2B) in contrast to cells that were co-transfected with ␤ 2d , ␤ 2⌬g , or ␤ 2 CED (see below). This experiment unambiguously shows that Ca v 1.2 calcium channels are silent in the absence of Ca v ␤ and that activity of the Ca 2ϩ channel demonstrated in HEK293 cells in the absence of exogenous Ca v ␤ subunits may be due to endogenous channels. The absence of the functional Ca v ␤ subunits in COS1 cells that follows from the data in Fig. 2B was further confirmed by co-IP analysis with recombinant FLAG N -␣ 1C /␣ 2 ␦ that revealed a lack of detectable endogenous monkey ␤ 1 , ␤ 2 , ␤ 3 , and ␤ 4 subunits in COS1 cells (28). Kinetics parameters and voltage dependence of activation and inactivation of I Ca and I Ba through the ␣ 1C,77 / ␣ 2 ␦/␤ 1a channel measured in COS1 cells were consistent with data obtained in other expression systems (15). An important advantage of COS1 cells is their relatively slow division rate that allows for better control over efficiency of expression and assembly of the Ca v 1.2 channel subunits of different size. However, HEK293 cells were more appropriate for co-IP-Western blot analysis of the recombinant tagged channel proteins in our study because they provide higher efficiency of expression, whereas endogenous subunits were undetectable with streptavidin and Abs to FLAG and Venus (ECFP) tags by Western blot analysis and fluorescence microscopy.
Confirming previous observations (19), when co-expressed with ␣ 1C,77 and ␣ 2 ␦ subunits, ␤ 2⌬g was appreciably accumulated in PM (Fig. 3A, panels b and c), stimulated inward I Ca with an average maximal amplitude of 80 Ϯ 15 pA (n ϭ 45; Fig. 3A, panel d), and co-immunoprecipitated with ␣ 1C subunit (Fig. 3F, lane 1). Sequential deletion of the ␤ 2⌬g subunit (Fig. 3, B-E) revealed that calcium channel activity is associated with the distal quarter of the ␤ 2⌬g sequence. Only the ␤ 2⌬g fragments containing the distal C-terminal regions 83-164 (Fig. 3C) and 124 -164 (Fig. 3E) induced the current when co-expressed with ␣ 1C,77 /␣ 2 ␦ (panels d) and directly interacted with ␣ 1C , as evident from the marked accumulation in PM (panels b and c) and Western blot analysis of co-IP with microsomal ␣ 1C (Fig. 3F,  lanes 3 and 5). Taken together, results of this analysis show that the C-terminal sequence of 41 amino acids of the Ca v ␤ 2 subunit (␤ 2 CED) represents a previously unknown determinant that may have a role in calcium channel modulation. Amino acid alignment revealed (Fig. 4) that ␤ 2 CED is conserved in Ca v ␤ 2 and shares only a subtle homology with the other Ca v ␤ subunits.
Electrophysiological Properties of the ␤ 2 CED-supported Channel-In COS1 cells expressing EYFP N -␣ 1C,77 and ␣ 2 ␦ subunits, the average maximal amplitude of I Ca decreased from 647 Ϯ 34 pA (n ϭ 48) with ␤ 2d to 120 Ϯ 25 pA (n ϭ 48) with ␤ 2 CED. Fig. 5A shows a family of representative traces of I Ca evoked by a stepwise depolarization in the range of Ϫ20 to ϩ60 mV applied from V h ϭ Ϫ90 mV. The currents were almost completely inhibited by the specific L-type calcium channel blocker (ϩ)PN200-110 (Fig. 5B, traces a). An interesting feature of these currents is the presence of a large slow component of inactivation that is unusual for the Ca 2ϩ -conducting Ca v 1.2 channels. Analysis of the steady-state inactivation curve (Fig.  5C) showed that at the end of a 1-s conditioning pulse 14.5 Ϯ 1.7% (n ϭ 12) of the peak I Ca remained non-inactivated. Analysis of ϳ50 expressing cells revealed that I Ca evoked by V t between Ϫ20 and 0 mV are better fitted with two exponentials and showed a prominent fast component of inactivation (see Table 1). The latter property can be better appreciated from the exemplar I Ca traces (recorded at Ϫ20 and Ϫ10 mV) in Fig. 5B that have larger amplitude than the representative currents in Fig. 5A. However, at V t Ն 10 mV, the decay of I Ca was better fitted by a single exponential. The large sustained current (I 0 , Table 1) is characteristic for all shown voltages and may be indicative of the inhibited slow inactivation (6,24).
Co-plotting of I-V and -V curves (Fig. 5D) showed that when ␤ 2d in the channel (Table 1) was replaced with ␤ 2 CED, inactivation of I Ca became slower on stronger depolarization and did not depend on the size of the current. The corresponding lack of U-shaped -V dependence is evidence that ␤ 2 CED does not support CDI (30). To further characterize modulation of inactivation of the Ca v 1.2 channel by ␤ 2 CED, we tested the effect of replacement of Ca 2ϩ for Ba 2ϩ in the bath medium on kinetics of the current decay. When Ba 2ϩ is the charge carrier, Ca v 1.2 channels inactivate by a voltagedependent mechanism (2). The I-V relationship for I Ba (V 0.5,act ϭ 41.3 Ϯ 4.3, n ϭ 31, Fig. 5E) was shifted to more positive potentials as compared with I Ca (V 0.5,act ϭ 19.8 Ϯ 1. 8, n ϭ 48, Fig. 5D). A ϳ10-mV positive shift of the steadystate inactivation curve was also observed on replacement of Ca 2ϩ (V 0.5,in ϭ 13.9 Ϯ 1.1) for Ba 2ϩ (V 0.5,in ϭ 24.2 Ϯ 3.3) in the bath medium (Fig. 5C), whereas the voltage dependence of availability of the ␤ 2 CED-modulated channel was increased to 50.6 Ϯ 2.9% (n ϭ 24) with Ba 2ϩ as the charge carrier. These data suggest that the inactivating fraction of channels is reduced in Ba 2ϩ because of increased voltage dependence of availability of the ␤ 2 CED channel.
. Homology alignment of ␤ 2 CED with human Ca V ␤ subunits.
Conserved residues are shown in black boxes. Blast analysis revealed that ␤ 2 CED is 100% conserved in the C termini of all known human Ca V ␤ 2 subunits (not shown), whereas only ␤ 1b2 (M92303) shows 43% of homology with ␤ 2 CED. No substantial homology was seen with ␤ 1b1 (M92302), ␤ 3 (X76555), ␤ 4 (U95020), ␤ 1a (not shown), or other human proteins as revealed by a general blast analysis. Thus, ␤ 2 CED is unique for Ca v ␤ 2 subunits. Having revealed these unusual inactivation properties of the ␤ 2 CED channel, we tested for CDI by calculating the f factor (31), which is the difference between the r 50 values, or fractions of I Ca and I Ba remaining at the end of a 50-ms depolarization. The 50-ms window was selected to accurately account for the relatively fast decay of I Ca in the range of Ϫ20 to ϩ10 mV (Fig.  5A). Confirming the result of the -V analysis (Fig. 5D), no U-shaped dependence of r 50 on V t was observed for I Ca , whereas those for I Ba was almost flat (Fig. 5F). A sharp difference between r 50 values for I Ba and I Ca at lower voltages reflects a switch from an apparent biexponential voltage-dependent inactivation at Ϫ20 to 0 mV to a predominantly single-component mechanism at V t Ն 10 mV (see above). One could argue that this result may be due to a specific level of intracellular Ca 2ϩ buffering in our experiment. However, CDI in Ca v 1.2 calcium channels exhibits low sensitivity to intracellular Ca 2ϩ buffers (32,33). The possibility that ␤ 2 CED evokes CDI only in the narrow voltage range of Ϫ20 to 0 mV is doubtful because Ca 2ϩ currents at these potentials are relatively small, and contribution of the voltage-dependent inactivation is probably greater. Making no further assumptions about the nature of this property, we compared inactivation of I Ca and I Ba near the maximum of I-V curves (Fig. 5, D and E), where CDI should be the most prominent (30). The difference f between r 50 values at ϩ30 to ϩ50 mV was close to zero (f ϭ 0.04 Ϯ 0.02 at ϩ40 mV, Fig. 5F). This result is consistent with the lack of CDI and explains the slow kinetics of I Ca decay in the ␤ 2 CED channel by the lack of the negative feedback regulation of inactivation by the permeating Ca 2ϩ ions.
Overlaying of I Ca and I Ba traces, scaled to the same amplitude, is a common approach to estimate the contribution of CDI and voltagedependent mechanisms in inactivation of the channel. In Fig. 6 we compared inactivation properties of ␤ 2 CED-and ␤ 2d -modulated channels by superimposing I Ca and I Ba traces near the maximum of I-V curves. At V t between ϩ20 and ϩ40 mV, where the currents are larger, the ␤ 2d -modulated I Ca inactivated notably faster than I Ba (Fig.  6A) due to CDI. In contrast, we observed a matching decay of I Ca and I Ba for the ␤ 2 CED-modulated channel (Fig. 6B) confirming the lack of CDI. We next compared decays of I Ca and I Ba sampled near the maximum of I-V curves. Because of CDI, the ␤ 2d -modulated I Ca inactivated appreciably faster than with ␤ 2 CED (Fig. 6C). However, I Ba through both ␤ 2 CED-and ␤ 2d -modulated channels recorded at the same test voltages showed a very similar decay (Fig. 6D) indicating striking similarity of the voltage-dependent inactivation of the channels. Thus, the Ca 2ϩ /Ba 2ϩ test confirmed lack of CDI in the ␤ 2 CED-modulated channel. Lack of CDI is an unusual property that was not previously observed in Ca v 1.2 channels with native pore-forming ␣ 1C subunits. One can see that I Ca inactivated much faster than I Ba in the case of ␤ 2d , but not ␤ 2 CED, thus confirming that ␤ 2 CED does not support CDI. C and D, comparison of I Ba (C) and I Ca (D) through the ␣ 1C,77 /␣ 2 ␦ channels modulated by ␤ 2 CED or ␤ 2d . The traces were normalized to the maximum amplitude of I Ca through the ␣ 1C,77 /␣ 2 ␦/␤ 2 CED (C) and ␣ 1C,77 /␣ 2 ␦/␤ 2d (D) channels. Unlike CDI, voltage-dependent inactivation of the channel supported by ␤ 2 CED was not markedly changed. Currents were recorded at V t ϭ ϩ20, ϩ30, and ϩ40 mV applied from V h ϭ Ϫ90 mV. Ca v ␤ 2 Subunit C-terminal Determinant JUNE 6, 2008 • VOLUME 283 • NUMBER 23 Effect of ␤ 2 CED Deletion from ␤ 2d on the Ca v 1.2 Calcium Channel-To better understand the functional impact of ␤ 2 CED on Ca v ␤ 2 subunit modulation of Ca v 1.2 channels, the distal 41-amino acid sequence (identical to those of ␤ 2⌬g (124 -164)) was genetically deleted from ␤ 2d . Modulation of the Ca v 1.2 channel by the obtained deletion mutant ␤ 2d⌬CED was compared with those of ␤ 2d (Fig. 7). Similar to other Ca v ␤ subunits, ␤ 2d facilitated large I Ca through the Ca v 1.2 channel. Distinct features of I Ca through the ␤ 2d channel include: 1) a relatively large sustained component of the current that comprised ϳ35% of the peak current at the end of a 600-ms depolarization pulse (Fig. 7A), and 2) a prominent shift of the maximum I-V curve from a typical value of ϩ20 to ϩ30 mV (15) to ϩ40 mV (Fig. 7C, closed circles). As expected, ␤ 2d supported CDI and showed a U-shaped -V dependence of I Ca peaked near the maximum of I-V curve (Fig. 7C, open circles).
The ␤ 2d⌬CED -modulated channel generated large inward I Ca (average maximal amplitude 466 Ϯ 160 pA, n ϭ 12) in response to depolarization in a characteristic range of membrane potentials (Fig. 7B). Similar to the ␤ 2d -modulated channel, decay of I Ca at all tested potentials was better fitted by a single exponential. The -V relation had a distinct U-shape indicating that deletion of the ␤ 2 CED from the ␤ 2d subunit did not compromise CDI (Fig. 7D). The maximal inactivation rate of I Ca through the ␤ 2d⌬CED -modulated channel was faster ( ϭ 43 Ϯ 5 ms at ϩ10 mV) than that modulated by ␤ 2d ( ϭ 59 Ϯ 6 ms at ϩ20 mV). The voltage dependence of activation (V 0.5,act ) and inactivation (V 0.5,in ) of the ␤ 2d⌬CED -modulated channel were shifted by ϳ12 (Fig. 7, C and D) and 5 mV (Fig. 7, E and F) to more negative voltages, respectively, suggesting that deletion of ␤ 2 CED endows the channel a higher voltage sensitivity (for statistics, see figure legend). Taken together, these results and data in Fig. 5 point to a synergy between MAGUK and ␤ 2 CED, but they may act independently as modulators of the Ca v 1.2 channel.
At present, the prevailing view suggests that AID is a constitutive binding site for all known Ca v ␤ subunits. To further test whether deletion of ␤ 2 CED from the ␤ 2d subunit interferes with binding to AID, we used the I-II AID peptide (amino acids 418 -455 of the ␣ 1C,77 subunit) that harbors AID (amino acids 428 -445) in its central part and retains binding affinity to Ca v ␤ subunits (34). To ease IP and detection, I-II AID was tagged at the N terminus with the monomeric mVenus protein (29). The mVenus N -labeled I-II AID was co-expressed with FLAG N -tagged ␤ 2d (Fig. 7G, lane 1) or ␤ 2d⌬CED (lane 2) in HEK293 cells. Consistent with the results of electrophysiological experiments (Fig. 7, A-F), Western blot analysis of co-IP showed that deletion of CED from ␤ 2d did not compromise binding of ␤ 2d⌬CED to I-II AID as compared with ␤ 2d (Fig. 7G).
␤ 2 CED Supports Ca v 1.2 Channels in the Absence of AID-When mVenus N -␤ 2 CED and Biotin N -I-II AID were co-expressed in HEK293 cells (Fig. 8A), ␤ 2 CED was identified (lane 1) on Western blot by monoclonal anti-LC Ab in both the immunoprecipitated (top left panel) and input (top right panel) fractions. However, I-II AID was detected on the blot by streptavidin only in the input fraction (bottom panel). Thus, co-IP analysis indicates that ␤ 2 CED does not bind to AID. These data suggest that ␤ 2 CED exerts its modulation of the Ca v 1.2 channel through interaction with a site(s) other than AID.
Electrophysiological experiments showed that, despite the inhibited binding between MAGUK and AID (Fig. 8B), ␤ 2d facilitated I Ca through the mutated ␣ 1C,77AIDM channel that showed little, if any, inactivation (Fig. 8C). As it is shown in Fig.  2B, no current could be detected in COS1 cells expressing ␣ 1C / ␣ 2 ␦ in the absence of Ca v ␤. The AID mutation did not inhibit conductance completely, but reduced the amplitude of the maximal I Ca induced by ␤ 2d to 131 Ϯ 25 pA (n ϭ 5) suggesting that the channel activation outside of AID by the full-size Ca v ␤ 2 does occur, but is less effective than that with the participation of intact AID. The same conclusion was obtained with ␤ 2d⌬CED , ␤ 2⌬g , and ␤ 2 CED. When ␤ 2d was substituted for ␤ 2d⌬CED (Fig.  8D), we observed a functionally active channel that exhibited a slowly inactivating I Ca with an average maximum amplitude of 150 Ϯ 24 pA (n ϭ 6). Similar to ␤ 2⌬g (Fig. 8E), inactivation of a large fraction of I Ca was inhibited by substitution of ␤ 2d for ␤ 2 CED (Fig. 8F), whereas the average amplitude of the peak I Ca through the ␣ 1C,77AIDM channel was smaller (40 Ϯ 15 pA, n ϭ 3). No appreciable modulation of the ␣ 1C,77AIDM channel (i.e. zero I Ca ) was observed in the absence of Ca v ␤ (Fig. 8G). Thus, inhibition of the MAGUK domain binding to AID did not abolish the sensitivity of the channel to ␤ 2 CED. These data confirm that essential regulatory properties of Ca v ␤ are AID-independent (17) and show that ␤ 2 CED can serve as a weak I-II linkerindependent activator of the Ca v 1.2 channel even when AID is mutated causing large conformational changes in the I-II loop.
Analysis of ␤ 2 CED Interaction with the LA/IQ Region of ␣ 1C -A meaningful characterization of the ␤ 2 CED modulation of the channel requires identification of its functional target. A recent report (36) demonstrated that the N-terminal domain of MAGUK in Ca v ␤ binds to the ␣ 1C subunit C-terminal region (amino acids 1571-1636 in ␣ 1C,77 ) that is involved in CaMmediated CDI regulation and includes LA and IQ loci of interaction with apo-CaM and Ca 2ϩ /CaM, respectively (37) (Fig. 1). Because ␤ 2 CED does not support CDI, we tested ␤ 2d (containing ␤ 2 CED) and ␤ 2d⌬CED (lacking ␤ 2 CED) for binding to LA/IQ. Co-IP analysis showed (Fig. 9) that LA/IQ binds ␤ 2d and ␤ 2d⌬CED independently on the presence of I-II AID ; moreover, ␤ 2d and ␤ 2d⌬CED bound to LA/IQ do not form triple complexes with I-II AID .
We then tested ␤ 2 CED for binding to LA/IQ. The FLAG Ntagged LA/IQ domain was co-expressed with mVenus N -␤ 2 CED (Fig. 10) in the presence (lanes 1 and 2) or absence of ECFP N -CaM (lanes 3 and 4). To assess for Ca 2ϩ dependence of binding, cells were permeabilized for external Ca 2ϩ before co-IP by incubating with ionophore ionomycin (5 M) in the bath medium containing 2 mM EGTA (lanes 1 and 3) or 2 mM Ca 2ϩ (lanes 2 and 4). Co-IP analysis confirmed that ␤ 2 CED binds to LA/IQ independently on Ca 2ϩ or co-expressed CaM.
Is LA/IQ the only region necessary for ␤ 2 CED action? To answer this question, we deleted LA/IQ from the ␣ 1C,77 subunit and co-expressed the resulting ␣ 1C,77⌬LK mutant with ␣ 2 ␦ and ␤ 2d or ␤ 2 CED (Fig. 11A). ␤ 2d modulated the channel via MAGUK/AID interaction and induced I Ca with an average maximal amplitude ϳ90 Ϯ 25 pA (panel a, n ϭ 3). Under the same conditions, with ␤ 2 CED no appreciable I Ca was observed on cell depolarization (panel b). This result suggests that the LA/IQ region is the only functional target of the ␣ 1C subunit where ␤ 2 CED may exert its action. To test whether LA or IQ determinants of CDI are essential for modulation of the channel by ␤ 2 CED, we examined the effect of ␤ 2 CED on the ␣ 1C,77 mutants lacking IQ (Fig. 11B, ␣ 1C,77L ) or LA (Fig. 11C, ␣ 1C,77K ) determinants defined in Fig. 1. In the absence of Ca v ␤, none of the tested channels showed appreciable I Ca in response to V t ϭ ϩ20 mV applied from V h ϭ Ϫ90 mV for 600 ms (Fig. 11, B and C, panels a). Co-expression of ␤ 2 CED induced I Ca only with ␣ 1C,77L (average maximal amplitude ϳ208 Ϯ 28 pA, n ϭ 5, see Fig. 11B, panel b) indicating that it is the LA determinant of CDI that is the functional target of ␤ 2 CED modulation of the channel. Under the same conditions, ␤ 2d supported I Ca through both ␣ 1C,77K (Fig. 11D, panel a; average maximal amplitude 231 Ϯ 13 pA, n ϭ 14) and ␣ 1C,77L channels (Fig. 11D, panel b; average maximal amplitude 389 Ϯ 24 pA, n ϭ 8). However, mutation of AID combined with the deletion of LA/IQ from the ␣ 1C subunit (␣ 1C,77AIDM/⌬LK ) completely inhibited modulation of the channel by ␤ 2d (Fig. 11E).

DISCUSSION
Our study revealed that modulation of Ca v 1.2 channels by large Ca v ␤ 2 subunits is mediated by inputs from multiple binding sites. There are at least three interactions between ␣ 1C and Ca v ␤ 2 subunits (Fig. 12, A-C) that induce activity of the channel not only jointly, but also when any two of the interactions are disrupted by mutations of ␣ 1C . The Ca v 1.2 channel modulation common to all large Ca v ␤ subunits is supported by the binding of the central MAGUK domain to the AID site of the ␣ 1C subunit I-II linker (4) (Fig.  12A). This interaction stabilizes the functional conformation of AID (and, respectively, the I-II linker), as well as provides specific orientation of the rigid core of Ca v ␤ important for multiple isoform-specific interactions leading to differential modulation of the channel (15,17). However, disruption of this interaction by the mutation of AID (Fig.  8B) does not prevent activation of the channel by Ca v ␤ that relies on two other ␣ 1C determinants located in the LA-IQ region (Figs. 5 and 8, C, D, and F). Deletion of these LA/IQ  2 and 4). Proteins were co-IP with anti-FLAG Ab and resolved by SDS-PAGE. Upper panels, identification of CaM and ␤ 2 CED on Western blot by monoclonal anti-LC Ab (lanes 1-4). Lower panels, identification of LA/IQ by anti-FLAG Ab. The amount of plasmid DNA (g) per transfection reaction is shown in parentheses.  (panel b). B and C, ␤ 2 CED supports gating of the IQ-mutated ␣ 1C,77L channel but not the LA-mutated ␣ 1C,77K channel. ␣ 1C,77L (B) or ␣ 1C,77K (C) was co-expressed in COS1 cells with ␣ 2 ␦ in the absence (panels a) or presence (panels b) of mVenus N -␤ 2 CED. D, evidence that ␤ 2d supported I Ca in the ␣ 1C,77K (panel a) and ␣ 1C,77L (panel b) channels. E, simultaneous mutation of AID and deletion of LA/IQ abolished modulation of the ␣ 1C,77AIDM/⌬LK /␣ 2 ␦ channel by ␤ 2d . Shown are representative traces of maximal I Ca recorded in response to a stepwise depolarization to ϩ30 (A and D) or ϩ20 mV (B, C, and E) applied for 600 ms from V h ϭ Ϫ90 mV.
determinants partially inhibited activity of the channel in the presence of ␤ 2d as can be seen from the smaller I Ca amplitude (Fig. 11A, panel a). A dynamic Ca 2ϩ -dependent interaction between the N-terminal SH3 region of MAGUK and the IQ domain of the ␣ 1C subunit (36) (Fig. 12B) appears to also be common to all large Ca v ␤ subunits. This interaction alone is sufficient to support I Ca (Fig. 8D).
In this study, we localized the third molecular determinant of the Ca v 1.2 channel modulation, ␤ 2 CED (Fig. 12C), which is specific only to Ca v ␤ 2 (Fig. 4) and resides in the C termini of ␤ 2a , ␤ 2b , ␤ 2c , ␤ 2d , ␤ 2e , ␤ 2f , and ␤ 2⌬g subunits. Thus, ␤ 2 CED represents a functional element of the Ca v 1.2 modulation that is conserved in primary cardiac Ca v ␤ 2 subunits. In the case of full size Ca v ␤ 2 (␤ 2a -␤ 2e ) (20), ␤ 2 CED acts in synergy with other determinants, as seen from the ability of ␤ 2 CED and ␤ 2d⌬CED to support I Ca with different properties (Figs. 5 and 7). However, ␤ 2 CED may induce activity of the channel independently on MAGUK, i.e. either alone (Fig. 5) or in the context of "short" ␤ 2 subunits, as in the case of the naturally occurring ␤ 2⌬g subunit (19) (Fig. 8E). Similar to other Ca v ␤ subunits, ␤ 2 CED activates the channels by binding to ␣ 1C (Fig. 3F) and targeting to PM (Fig. 3E), but does not support CDI (Fig. 5D). It appears that ␤ 2 CED affects voltage gating of the channel. Indeed, similar to ␤ 2d (Fig. 7C), the ␤ 2 CED-modulated channel has the maximum I-V curve at ϩ40 mV (Fig. 5D), whereas deletion of ␤ 2 CED from ␤ 2d shifted the voltage dependence of I Ca by 20 mV to lower potentials (Fig. 7D). We find that ␤ 2 CED binds to the LA/IQ region of the ␣ 1C subunit C-terminal tail in a Ca 2ϩ -and CaMindependent manner and needs the LA, but not IQ, motif to activate the channel (Fig. 10). This is the first observation of the Ca v ␤ 2 subunit regulation of the Ca v 1.2 calcium channel that does not rely on Ca v ␤/AID interaction.
Although ␤ 2 CED did not bind to AID (Fig. 8A), mutation of AID interfered with the interaction of the channel with ␤ 2 CED that was reflected in the smaller amplitude of I Ca (cf. Figs. 5A  and 8F). This result points to high sensitivity of all three ␣ 1C / Ca v ␤ 2 interactions (Fig. 12) to the conformation of AID that is probably a key component of mutually dependent determinants of channel regulation (7,24).
Another important conclusion from our study is that CDI does not depend solely on the ␣ 1C subunit. Indeed, co-expression of intact ␣ 1C and ␣ 2 ␦ subunits with ␤ 2 CED generates the channel lacking CDI (Figs. 5 and 6). It is known that CDI is mediated by interactions of CaM with two adjacent sites (LA and IQ) of the ␣ 1C subunit C terminus (for review, see Ref. 38). We found that CDI ultimately requires both MAGUK/AID and SH3/IQ interactions (Fig. 12, A and B). Thus the role of Ca v ␤, AID, and LA/IQ interactions in the ensemble of mutually coordinated determinants of CDI is essential.
The exact number of Ca v ␤ subunits (of the same or different type) that bind to an individual ␣ 1C subunit is unknown. Results in Fig. 9 show that ␤ 2d is not involved in a simultaneous binding to LK and AID, although this subunit can be engaged in all three types of interactions shown in Fig. 12. Therefore, it is possible that there is more than one Ca v ␤ 2 subunit interacting with the same ␣ 1C . On the other hand, if Ca v ␤ does dissociate from the AID of the functional channel (39), then it is possible that modulation of the channel may be mediated by a single Ca v ␤ 2 molecule alternating between AID and LA/IQ sites. An additional complexity (36) to this general picture may be brought about by the N-terminal palmitoylation site known to anchor the ␤ 2a subunit in PM (40).
Taken together, our results provide new insight into potential role(s) of ␤ 2 CED in modulation of Ca v 1.2 channels. Ca v ␤ 2 is a major cardiac ␤ subunit and its splice variation is an important correlate of the Ca v 1.2 calcium channel regulation (20,41). One of the most puzzling questions raised by our study is why Ca v ␤ 2 contains more than one ␣ 1C interaction motif. One possible reason for this complexity may be associated with the role of ␤ 2 CED in additional tuning of the voltage dependence of the current (20). Indeed, results of electrophysiological measurements (Fig. 7) show that deletion of ␤ 2 CED from ␤ 2d significantly changed kinetics of inactivation of I Ca and shifted the peak of the I-V curve by 20 mV toward more negative potentials. Another reason may be that differential, tissue-specific splicing of the Ca v ␤ 2 gene (18) may generate subsets of the Ca v 1.2 calcium channel modulated only through ␤ 2 CED. These channels do not support CDI and generate small, but longlasting Ca 2ϩ currents. It is usually assumed that I Ca is rapidly and fully inactivated, but our results raise the hypothesis that Ca 2ϩ signaling in human cardiac cells expressing small Ca v ␤ 2 subunits (19) may involve Ca 2ϩ -insensitive Ca v 1.2 channels in addition to L-type channels regulated by CDI. One possibility is that Ca 2ϩ channels that rely on ␤ 2 CED-dependent, MAGUKindependent modulation in cardiac muscle cells may account for the prolongation of L-type I Ca and therefore contribute to the balance that controls the shape of the action potential plateau. Whichever role of ␤ 2 CED is predominant, it may be a new potential pharmacological target.