New Short Splice Variants of the Human Cardiac Cavβ2 Subunit

Two new short splice variants of the Ca2+ channel β2 subunit were cloned from human heart poly(A)(+) mRNA. The 410-amino acid β2f subunit is encoded by exons 1A, 2A, 3, 4, 12, 13, and 14 of the human Cavβ2 gene and lacks the protein kinase A phosphorylation site, the β-interaction domain (De Waard, M., Pragnell, M., and Campbell, K. P. (1994) Neuron 13, 495–503), 40% of the β-SH3 domain, and 73% of the guanylate kinase domain of the putative membrane-associated guanylate kinases module (McGee, A. W., Nunziato, D. A., Maltez, J. M., Prehoda, K. E., Pitt, G. S., and Bredt, D. S. (2004) Neuron 42, 89–99), and helix α3 of the α1-subunit binding pocket (Van Petegem F., Clark, K. A., Chatelain, F. C., and Minor, D. L., Jr. (2004) Nature 429, 671–675). The β2g transcript has two potential initiation codons. With the second ATG codon, it generates the 164-amino acid β2Δg subunit encoded essentially by the distal part of exon 14, and thus β2Δg completely lacks any of the above motifs. Immunoprecipitation analysis confirmed stable association of β2f and β2Δg with the α1C subunit. The plasma membrane localization of β2f and β2Δg was substantially increased by co-expression of the α1C,77 and α2δ subunits. In COS1 cells, β2f and β2Δg increased plasma membrane targeting of the pore-forming α1C subunit and differentially facilitated (β2f > β2Δg) the voltage gating of otherwise silent Cav1.2 channels. We conclude that it is unlikely that the β-interaction domain, membrane-associated guanylate kinases module, and the α1-subunit binding pocket helix α3 are essential for the interaction of the α1C and β2 subunits and suggest that in addition to the α1-subunit binding pocket helices α5 and α8, a yet unresolved C-terminal β2 region plays a crucial role.

The L-type high voltage-activated Ca v 1.2 channel is composed of the pore-forming ␣ 1C subunit and the auxiliary ␤ and ␣ 2 ␦ subunits. The Ca v ␤ subunits (1) are essential cytoplasmic modulators of the Ca 2ϩ channel activity that generate a molecular signal necessary for the facilitation of voltage gating as well as for the correct plasma membrane targeting of the functional Ca v 1.2 complex (2). The ␣ 1C and ␤ subunits are tightly associated in the channel complex and can be co-immunoprecipitated in mild non-ionic detergents. A conserved ␤-interac-tion domain (BID) 1 common for genetically different ␤ subunits (␤ 1 -␤ 4 ) was proposed as a binding motif interacting with the conserved ␣-interaction domain (AID) of the I-II cytoplasmic linker between repeats I and II of an ␣ 1 subunit (3,4). The ␣ 1 -␤ interaction is believed to chaperon the channel by inhibiting an endoplasmic reticulum retention signal encoded in the ␣ 1 subunit I-II linker (5). Comparative studies showed that ␤ subunits differentially modulate inactivation kinetics (6) and single-channel properties of the Ca v 1.2 channel (7). We have found that the differential ␤-subunit modulation predominantly influences the slow inactivation of the channel and is associated with distinct voltage-gated rearrangements between the Nterminal regions of the ␣ 1C and ␤ 2 subunits (8). Structural principles underlying the ␤-subunit modulation of Ca 2ϩ channels were also approached by a search of structural homology with known regulatory proteins. It has been found (9) that the vast central conservative region of ␤ 2 subunits shares distant homology with the Src homology 3 (SH3)-guanylate kinase (GK) module of membrane-associated guanylate kinases (MAGUKs). This hypothesis was further elaborated by studying mutations of the rat (N terminus-palmitoylated) ␤ 2a subunit that interfere with interactions between SH3 and GK and affect inactivation of Ba 2ϩ currents (10,11). Intramolecular interactions between the ␤-subunit SH3 and GK homologues were supported by the results of the recent high-resolution crystallography studies of the ␤-subunit "cores" of SH3-GK domains alone or in complex with AID (12)(13)(14). These studies also showed that BID is not available for the binding to AID because it is buried inside the ␤-subunit structure. Instead, the ␣-subunit binding pocket (ABP) distal to the SH3 domain was inferred as a structure engaged in the interaction with AID.
The first cardiac ␤ 2 subunits cloned from rat (15) and rabbit (16) hearts have over time been co-identified with five N-terminal splice variants (␤ 2a -␤ 2e ; for overview, see Ref. 17). A recent comprehensive study revealed that in the human left ventricle there are nine Ca v ␤ 2 splice variants, including ␤ 2b , ␤ 2c , ␤ 2d , and ␤ 2e ; the exon 7C isoforms of ␤ 2b , ␤ 2c , and ␤ 2d ; the exon 7B isoform of ␤ 2b ; as well as the ␤ 2b transcript lacking exon 7 and truncated in the exon 8 region (18). Here, we report on two new splice variants of the Ca 2ϩ channel ␤ 2 subunit gene cloned from the human normal heart poly(A)(ϩ) mRNA and sequentially named ␤ 2f and ␤ 2g . The ␤ 2f and ␤ 2g transcripts lack exons encoding a single protein kinase A (PKA) phospho-rylation site, BID, as well as a large part (␤ 2f ) or the entire SH3-GK module (␤ 2g ). Despite the wide structural differences between ␤ 2f /␤ 2g and the rest of the ␤ 2 subunits, the new ␤ 2 subunits yielded fully functional Ca v 1.2 channels with different inactivation characteristics when co-expressed in COS1 cells with the ␣ 2 ␦ and the human vascular ␣ 1C,77 subunits. These data indicate the need to redefine the significance of the previously outlined ␤ subunit functional motifs and implicate the C-terminal region encoded by the distal ␤ 2 gene exon in modulation of the channel.

MATERIALS AND METHODS
Cloning of the Short Human Cardiac Ca v ␤ 2 Subunits-The Ca 2ϩ channel ␤ 2f and ␤ 2g subunits were cloned from the human heart poly(A)(ϩ) mRNA pooled from three Caucasian males (ages 35, 55, and 55; cause of death, trauma) (BD Biosciences Clontech, Palo Alto, CA; catalog number 6533-1) using C. therm. Polymerase One-Step reverse transcriptase (RT)-PCR (Roche Applied Science). The RT reaction was carried out for 30 min at 57°C with a common RT oligonucleotide primer, 5Ј-ACATATGATTGCAGTGTAGACC-3Ј, designed from the ␤ 2 subunit 3Ј-untranslated region (nt 804058 -804079 in the Human Genome Contig NT_8705.15 of chromosome 10). In the first round of PCR, a common antisense 5Ј-GCTGTTAGTTATACAAGACTTC-3Ј primer (nt 804014 -804035) was used to amplify ␤ 2f and ␤ 2g with the sense ␤ 2dspecific 5Ј-ATGGTCCAAAGGGACATGTC-3Ј (nt 404991-405010) or the ␤ 2b -specific 5Ј-ATGCTTGACAGACGCCTTATA-3Ј (nt 605181-605201) primer, respectively. PCR mixtures were denatured for 2 min at 94°C followed by 30 PCR cycles, each composed of denaturing for 45 s at 94°C, annealing for 1 min at 51°C, and extension for 2.5 min at 72°C, with the last step of the last cycle extended to 10 min. For the second round of PCR with "nested primers," a common antisense 5Ј-gctctagagTCATTGGCGGATGTAAACATC-3Ј primer (nt 803958 -803978, attached to an XbaI linker shown in small letters) was used with the sense 5Ј-gccaccATGGTCCAAAGGGACATGTCCAAG-3Ј (␤ 2f ) and 5Ј-gccaccATGCTTGACAGACGCCTTATAGCTC-3Ј (␤ 2g ) primers (incorporating Kozak sequence, shown in small letters in front of the ATG start codon) and the same PCR conditions except with an increased annealing temperature (58°C). The RT-PCR products were purified using the QIAquick PCR purification kit (Qiagen, Valencia, CA), ligated into the pCR2.1 TA vector (Invitrogen) and sequenced in both directions. To generate ␤ 2⌬g , the ␤ 2g -coding TA clone was amplified using an antisense primer, as shown above for the second round of PCR, and the sense primer 5Ј-cgggatccgccaccATGTATCTCTGGAG-GAGGACCGGG-3Ј, which starts at the second ATG codon and has a Kozak sequence preceded by a BamHI linker at the 5Ј-end. For expression in eukaryotic cells, the ␤ 2f -and ␤ 2⌬g -coding clones were ligated into the pcDNA3 vector at the XbaI (filled)-NotI and BamHI-XhoI sites, respectively. To systematically investigate subcellular localization of the ␣ 1C and ␤ subunits in situ, their N termini were genetically fused to EYFP and ECFP, respectively. Experimentally, such labeling did not interfere with the functional expression of the channel in our previous (8,19) and current investigations as compared with the unlabeled subunits. To generate (ECFP) N -␤ 2f , the ␤ 2f coding sequence was supplemented with the 5Ј-NcoI (Klenow-blunted) and 3Ј-ApaI linkers and ligated into the 5Ј-ECFP-pcDNA3 vector at the XhoI (filled) and ApaI sites. To generate (ECFP) N -␤ 2⌬g , the ␤ 2⌬g -coding sequence was supplemented with the 5Ј-BamHI and 3Ј-XhoI linkers and ligated into the 5Ј-ECFP-pcDNA3 vector at the respective sites.
Immunoprecipitation Analysis-The (ECFP) N -labeled ␤ 2f or ␤ 2⌬g subunits (replaced by EGFP in the negative control) were expressed in HEK293 cells in the presence or in the absence of ␣ 1C,77 and ␣ 2 ␦. The labeled proteins were co-immunoprecipitated with Living Color fulllength A.V. polyclonal antibody (BD Biosciences Clontech) as described in detail earlier (8). The (ECFP) N -labeled ␤ subunits were identified on Western blots by Living Color monoclonal antibody JL-8 (1:8,000 dilution; BD Biosciences Clontech). The co-precipitated ␣ 1C subunits were detected with the affinity-purified rabbit anti-␣ 1C calcium channel polyclonal antibody (1:10,000; Chemicon International) using an ECL Plus Western blotting detection system (Amersham Pharmacia Biotech).
Fluorescent Microscopy-All cell images were obtained on an epifluorescent Nikon microscope TE200 equipped with a 10-bit digital Hamamatsu CCD camera C4742-95 (Hamamatsu, Bridgewater, NJ). Image processing and analysis were performed with the SimplePCI 5.3 software (Compix, Pittsburgh, PA).

Molecular Cloning and Structural Properties of the Short
Human Cardiac ␤ 2f and ␤ 2g Subunits-Two new short splice variants of the Ca 2ϩ channel ␤ 2 subunit gene, named ␤ 2f and ␤ 2g (Fig. 1A), were cloned by RT-PCR from the poly(A)(ϩ) fraction of the human normal heart mRNA. The RT and PCR primers for the cloning were deduced from the Human Genome Project draft sequence NT_008705.15 of chromosome 10. The ␤ 2f subunit has a theoretical molecular mass of 45.8 kDa, is composed of 410 aa, and has the N terminus encoded by exons 1A and 2A analogous to ␤ 2d (18). However, missing in the ␤ 2f transcript are exons 5-11 coding for the 250-aa central region of the ␤ 2 protein, including BID, as well as 55 of 136 aa (i.e. 40%) of the C-terminal part of the SH3 domain and 192 of 262 aa (i.e. 73%) of the N-terminal part of GK domain. Overall, the following crystallographically resolved structures (13) are absent from the ␤ 2f protein: antiparallel ␤-strands ␤3-␤5 and the second ␣-helix (␣2) of the first conserved domain, the entire variable domain V2, as well as parallel ␤-sheets ␤6 -␤9, two ␣-helices (␣3-␣4), and two 3 10 helices (2-3) of the second conserved domain. Thus, the entire linker between SH3 and GK ("HOOK domain") present in MAGUKs is deleted from the ␤ 2f subunit.
The even smaller ␤ 2g -subunit transcript is composed of exon 2c followed by the 72-nt upstream portion of exon 3 and the 463-nt distal part of exon 14. Although exon 2c encodes the N terminus of the ␤ 2b subunit, the reading frame is interrupted just 150 nt downstream of the first initiation codon. The second ATG codon occurs 91 nt downstream from the first one, or 34 nt downstream from the acceptor splice site of exon 3. Its reading frame includes the distal 463-nt sequence of the ␤ 2 3Ј-terminal exon 14. Thus, both exons 3 and 14 are alternatively spliced in ␤ 2g , but the respective utilized donor and acceptor splice sites do not conform to the consensus sequences. With the second initiation codon, the ␤ 2g -subunit transcript encodes a 164-aa protein with the calculated molecular mass of 19.5 kDa. This splice variant, referred further as ␤ 2⌬g , has been generated by a PCR deletion of the 91-nt upstream region including the first ATG codon. Both the ␤ 2f and ␤ 2⌬g subunits lack the PKA phosphorylation site (RKST in position 216 of the human ␤ 2d ) that was shown to be involved in the PKA-mediated stimulation of cardiac L-type Ca 2ϩ currents (20). In addition, ␤ 2f retains only 6 of 12 potential protein kinase C phosphorylation sites, (S/T)X(R/K), in positions 78, 198, 302, 343, 382, and 441. The last three sites are also present in the ␤ 2⌬g subunit.
Although ␤ 2f and ␤ 2⌬g were cloned from the polyadenylated fraction of mRNA, none of the studied short ␤ 2 splice variants has yet been shown to be expressed in the human heart as functional proteins. Because of this and as the usage of initia-tion codons in ␤ 2g remains unknown, ␤ 2⌬g is considered a putative splice variant of the ␤ 2g subunit. Nevertheless, because ␤ 2⌬g lacks not only BID and the entire SH3-GK domains but also the variable ␤ 2 -subunit N termini-coding exons, ␤ 2⌬g is particularly interesting for the study of the functional significance of these genetically deleted motifs. To investigate whether the structural deletions discovered in the new ␤ 2 splice variants affect their interaction with the ␣ 1C subunits, co-immunoprecipitation of the short ␤ 2 subunits with the wildtype ␣ 1C,77 was analyzed by Western blot assay (Fig. 1B). ECFP was genetically fused to the N termini of the new ␤ 2 splice variants by in-frame ligation of the ECFP and ␤ 2f /␤ 2⌬g coding sequences, and the (ECFP) N -labeled ␤ 2 subunits were co-expressed in HEK293 cells with the ␣ 2 ␦ and ␣ 1C,77 subunits. Similar to the ␤ 1a and the ␤ 2(r) subunits (8), the (ECFP) N labeling did not compromise their functional activity (data not shown). Use of antibody against GFP to co-immunoprecipitate (ECFP) N -␤ 2 variants and ␣ 1C,77 from the solubilized membrane particulate fraction of HEK293 cells helped to avoid a contamination with the nonassociated cytosolic (ECFP) N -␤ 2 and the membrane-bound orphan ␣ 1C subunits. Both ␤ 2f (Fig. 1B, lane  1) and ␤ 2⌬g (lane 2) pulled down the ␣ 1C,77 protein from the solubilized membrane preparations suggesting stable association between the subunits. A similar observation was made with ␤ 1a , ␤ 2(r) (8), and other ␤ proteins. In controls, no immunoprecipitation of the ␣ 1C subunit by anti-GFP antibody was found in the absence of (ECFP) N -␤ 2 variants in COS1 cell with (Fig. 1B, lane 3) or without (lane 4) EGFP co-expressed. These data indicate that the ␣ 2 ␦ and ␣ 1C subunits are necessary for the plasma membrane targeting by the ␤ 2f and ␤ 2⌬g subunits.
The ␤ 2f Subunit Stimulates the Plasma Membrane Targeting and Facilitates the Ca v 1.2 Channel Voltage Gating-To better characterize the cellular location of the expressed ␣ 1C and ␤ subunits, the (ECFP) N -labeled ␤ 2f and ␤ 2⌬g were expressed with ␣ 1C,77 and ␣ 2 ␦ subunits in COS1 cells lacking the endogenous Ca 2ϩ channels (21). In contrast to the data obtained with Xenopus oocytes (10,14) and HEK293 cells (11), heterologous expression of the ␣ 1C and ␣ 2 ␦ subunits in COS1 cells did not induce an appreciable Ca 2ϩ channel activity ( Fig. 2A) unless a ␤ subunit was co-expressed (Fig. 2B). Thus, selection of the COS1 cells expression system allowed us to avoid ambiguity in the assessment of the ␤-subunit modulation of the Ca 2ϩ channel typical for the cited studies (10,11,14) and to define the ␤-subunit modulation here as a facilitation of the Ca 2ϩ channel voltage gating by a ␤ subunit. Essential prerequisites for a ␤-subunit modulation of the channel are binding of Ca v ␤ to the ␣ 1C subunit and targeting of the oligomeric channel complex to the plasma membrane.
In the absence of Ca v ␤, the (EYFP) N -labeled ␣ 1C,77 is essentially retained in intracellular compartments of the cell (Fig.  2C, a). Similar to ␤ 2(r) (Fig. 2E, c) (8), co-expression of the ␤ 2f subunit increased surface membrane targeting of (EYFP) N -␣ 1C,77 , which is evident from a comparison of the subcellular Note that the ␤ 2g transcript is initiated from exon 2c similar to ␤ 2b . Because the reading frame of the ␤ 2g transcript opened at the first initiation codon is interrupted at the position 153, the second ATG (shown above exon 3) at the position 98 was utilized to generate the ␤ 2⌬g transcript. B, co-immunoprecipitation with the short ␤ 2 splice variants. The (ECFP) N -␤ 2f (1), (ECFP) N -␤ 2⌬g (2) or no ␤ subunit (3,4) was co-expressed in HEK293 cells with (1-3) or without (4) the ␣ 1C,77 and ␣ 2 ␦ subunits. EGFP was co-expressed in control sample 3. Cells were lysed by freeze-thaw. To remove free ␤ subunits and other cytosolic proteins, the crude membrane particulate fractions were prepared from the lysed cells as described earlier (8). Membranes were solubilized with 0.5% Nonidet P-40, and the channel proteins were immunoprecipitated with the antibody against green fluorescent protein, separated by SDS-polyacrylamide gel electrophoresis, and analyzed by immunoblotting. The (ECFP) N -labeled ␤ 2 splice variants (1,2) and EGFP (3) were detected on the blot by the monoclonal antibody against green fluorescent protein, whereas the co-immunoprecipitated ␣ 1C subunit was detected using the anti-␣ 1C antibody. distribution of fluorescence in the absence (Fig. 2C, a) and in the presence of ␤ 2f (Fig. 2C, b). Conversely, ␣ 1C increased plasma membrane targeting of the ␤ 2f subunit. In the absence of ␣ 1C , the (ECFP) N -labeled ␤ 2f subunit was diffusely distributed over the cytoplasm (Fig. 2C, c). A similar intracellular localization, typical for cytosolic proteins, was observed with the full-size (ECFP) N -␤ 2(r) (Fig. 2E, a). Co-expression of ␣ 1C,77 increased the plasma membrane localization of the (ECFP) N -␤ 2f subunit (Fig. 2C, d) analogously to the effect of ␣ 1C,77 on the (ECFP) N -␤ 2(r) in positive control (Fig. 2E, b). These data corroborate the results of immunoprecipitation analysis (Fig. 1B,  lane 1) and provide strong evidence that ␤ 2f binds to ␣ 1C,77 and facilitates the channel formation and its plasma membrane targeting. Thus, the ␤ 2f subunit displayed both the chaperon function and binding to ␣ 1C , which are features characteristic for ␤ 2(r) and other ␤ subunits (2). Despite the large structural differences between ␤ 2f and the other known ␤ 2 subunits, when co-expressed in COS1 cells with the human vascular ␣ 1C,77 subunit and ␣ 2 ␦, the new ␤ 2f subunit yields a fully functional Ca v 1.2 channel. Fig. 2B (panel c) shows representative traces of the Ba 2ϩ current (I Ba ) evoked by the 600-ms test pulses to the indicated voltages applied from V h ϭ Ϫ90 mV. The corresponding averaged normalized I-V relationship is presented in Fig. 2B, d. The values for the half-maximal activation (V 0.5 ϭ 1.8 Ϯ 1.7 mV) and the slope factor (k I-V ϭ Ϫ7.9 Ϯ 0.9; n ϭ 9) were essentially similar to those reported earlier for I Ba through the ␣ 1C,77 /␣ 2 ␦/␤ 2(r) channel expressed in Xenopus oocytes (22). Respectively, a replacement of Ba 2ϩ for Ca 2ϩ as the charge carrier (Fig. 2D, a) produced a typical strong acceleration of the current decay characteristic for Ca 2ϩinduced inactivation of the Ca v 1.2 channel. Contrasting with the rabbit ␤ 2(r) subunit, the new short ␤ 2f renders notably faster inactivation of I Ba . Fig. 2D (b-d) shows maximum I Ba elicited by test pulses of different durations to ϩ20 mV. A two-exponential fitting of the 30-s I Ba through the (EYFP) N -␣ 1C,77 /␣ 2 ␦/␤ 2f channel (n ϭ 5) showed the inactivation time constants () and fractions (I) of the fast and slow components to be f ϭ 465 Ϯ 20 ms (I f ϭ 46.4 Ϯ 1.4%) and s ϭ 4.31 Ϯ 0.68 s (I s ϭ 35.2 Ϯ 2.1%), respectively. Although complete inactivation of the current was not reached even with the 45-s pulse, the fraction of the current that remains by the end of the 30-s pulse (18.0 Ϯ 2.0%) was, on average, significantly (p Ͻ 0.05) smaller that those we have found (27.5 Ϯ 3.3%) with the ␤ 2(r) subunit. The fast component of I Ba through the (EYFP) N -␣ 1C,77 / ␣ 2 ␦/␤ 2(r) channel (Fig. 2E, d) ( f ϭ 499 Ϯ 68 ms; I f ϭ 53.4 Ϯ 1.6%; n ϭ 5) was essentially similar to those of the ␤ 2f channel current. However, the slow component showed a delayed decay ( s ϭ 12.88 Ϯ 2.40 s; I s ϭ 19.4 Ϯ 4.7%; p Ͻ 0.01). Overall, our study revealed that ␤ 2f exhibits properties characteristic of the ␤ 2 subunits and yields a fully functional Ca v 1.2 channel. These data suggest that a requirement of the ␤ subunit for functional conformation of the channel (6, 23) is conserved in the regions of ␤ 2 other than BID and the missing essential parts of SH3-GK.
The ␤ 2⌬g Subunit Narrows the Functional Correlates of the ␤ Subunit to the C-terminal 153-Amino Acid Region-The immunoprecipitation analysis (Fig. 1B, lane 2) showed that the ␤ 2⌬g subunit binds to ␣ 1C,77 despite the lack of BID, the entire MAGUK module, and the ABP. The ␤ 2⌬g subunit is composed of 164 aa of which the N-terminal 11 residues are new to ␤ 2 subunits. The last 153 aa are common for the C-terminal region of ␤ 2 subunits encoded by the distal part of exon 14. Our data indicate that this C-terminal region is sufficient to confer the assembly of the functional channel. Fig. 3 shows cellular distribution and functional expression of the Ca 2ϩ channel assembled with ␤ 2⌬g . When co-expressed with ␣ 2 ␦ but in the absence of ␣ 1C , the (ECFP) N -␤ 2⌬g subunit was diffusely distributed over the cytoplasm without selectively targeting the plasma membrane (Fig. 3A, a). The ␣ 1C,77 subunit strongly enhanced accumulation of (ECFP) N -␤ 2⌬g in the plasma membrane (Fig. 3A,  panel b). Conversely, ␤ 2⌬g effectively increased membrane targeting of the (EYFP) N -labeled ␣ 1C,77 (Fig. 3A, compare c and d). Thus, ␤ 2⌬g retains the molecular signals necessary for binding to the ␣ 1C subunit and plasma membrane targeting of the channel.
Co-expression of the ␣ 1C,77 , ␣ 2 ␦, and ␤ 2⌬g subunits gave rise to a functional Ca 2ϩ channel that exhibited somewhat unusual properties. First, with both Ca 2ϩ and Ba 2ϩ as the charge carrier, the maximum amplitude of the current was ϳ10 times smaller than through the ␣ 1C,77 /␣ 2 ␦/␤ 2f channel. Fig. 3B shows the 600-ms traces of the Ca 2ϩ and Ba 2ϩ currents through the ␤ 2⌬g channel evoked by depolarization to ϩ20 mV from V h ϭ Ϫ90 mV. Both the activation and inactivation properties of the ␤ 2⌬g channel are significantly different from those of the ␤ 2f channel. The Ca 2ϩ current did not show accelerated inactivation (Fig. 3B, left trace), which would be characteristic for the Ca 2ϩ -conducting L-type channels (e.g. compare traces a and b in Fig. 2D), thus suggesting that Ca 2ϩ -induced inactivation may be impaired by the ␤ 2⌬g subunit. The activation of the Ba 2ϩ current is delayed for Ն1 s (Fig. 3, B and C). A twoexponential fitting of inactivation kinetics of the 30-s I Ba through the (EYFP) N -␣ 1C,77 /␣ 2 ␦/␤ 2⌬g channel (n ϭ 6) showed a substantial decrease of the fast component ( f ϭ 350 Ϯ 150 ms; I f ϭ 16.5 Ϯ 4.0%), a prolongation of the slow decay ( s ϭ 13.9 Ϯ 5.3 s; I s ϭ 47.5 Ϯ 3.0%), and an almost 2-fold increase of the sustained current component by the end of the 30-s pulse (36.0 Ϯ 5.5%). Thus, the ␤ 2⌬g and ␤ 2f channels clearly exhibit differential modulation of inactivation despite the lack of MAGUK and BID structures. Overall, ␤ 2⌬g demonstrates typical properties of the Ca 2ϩ channel ␤ subunits, including binding to the ␣ 1C subunit and stimulation of the surface membrane targeting by the channel, but provides an altered and weak facilitation of channel gating. DISCUSSION The most interesting result of this study is that vast deletions in the central region of the ␤ 2 subunit that include most of or the entire SH3-GK region do not compromise the ␤-subunit modulation of the Ca 2ϩ channel. The ␤ 2f subunit exhibits an array of properties that resemble those of the ␤ 2(r) and other ␤ 2 subunits, whereas the shortest known ␤ 2⌬g isoform shows more unusual features in modulation of the channel. Our results with the ␤ 2f subunit confirm the main structural implications of the crystallographic study of the Ca v ␤ 2a -AID complex (12)(13)(14) by highlighting the fact that it is the C-terminal region encoded by exons 12-14 that appears to bear essential structures involved in interaction with an ␣ 1 subunit. Although it is not clear to what extent the identified structural features of the ␤ 2 "conserved core" are preserved in its remnants in the ␤ 2f subunit, they may include the C-terminal fragment of the helix ␣5 (153-159) and the entire helix ␣8 (196 -215) that jointly form a part of ABP (13). However, ␤ 2f lacks the helix ␣3 coidentified with the ABP. It is unlikely that the small parts of the SH3 and GK domains remaining in the ␤ 2f subunit are sufficient to support their cross-interaction, because the distal loop and strand E of the conserved domain I implemented in this interaction (12) are both absent from ␤ 2f .
The remnant of the first conserved domain of ␤ 2f , encoded in part by exons 3 and 4 (aa 73-152), shows 72-80% homology with the other cloned human ␤ subunits, including ␤ 1B-1 and ␤ 1B-2 (24), ␤ 3 (25), and ␤ 4 (26). The remnant sequence (aa 153-217) of the second conserved region of ␤ 2f , which includes helices ␣5 and ␣8 of the ABP encoded in part by exons 12 and 13, shares 83-88% of the overall homology with other ␤ subunits. Experimentally established conventional modulation of the channel by the ␤ 2f subunit suggests that the combination of ␣5 and ␣8 helices is an important structural requirement for ABP and/or that helix ␣3 might be replaced in ABP by an unidentified motif(s) in ␤ 2f . In contrast, ␤ 2⌬g lacks any structural bases for ABP or MAGUK, and ␤ 2⌬g does not share a substantial (Ͼ30%) homology with the ␤ 1 , ␤ 3, and ␤ 4 subunits. Nevertheless, ␤ 2⌬g binds to the ␣ 1C,77 subunit (Fig. 1B, lane 2), stimulates membrane targeting of the channel (Fig. 3), and shows properties of a weak modulator of the Ca v 1.2 channel voltage gating. Specifically, ␤ 2⌬g supports slowly inactivating Ba 2ϩ currents at prolonged depolarization, which is a characteristic feature of other ␤ 2 subunits. There are no structural data available for the ␤ 2 -subunit C-terminal region to explain the results of the functional evaluation of the ␤ 2⌬g except to suggest that within the C-terminal 153-amino acid sequence, an essential structure of the ␤ 2 subunit is present that endows a modulation of the channel via functional interaction with ␣ 1C .
The ␤ subunit is an important component of a molecular complex that determines cytoplasmic helical packing stabilizing channel gating. We hypothesize that it is the Ca v ␤ 2 Nterminal half that finely tunes the gating facilitation and affects differential modulation of the Ca 2ϩ channel current inactivation and the voltage-gated rearrangements of the ␣ 1C and ␤ subunit N-tails (8). The MAGUK module may contribute to this regulation, but its partial or even complete deletion does not impede the channel modulation as defined in this paper.
Several Ca v ␤ variants with truncated C-terminals were previously identified (18). Our work has established the fact that the main splice isoforms (␤ 2a -␤ 2e ) of the ␤ 2 subunit gene may generate smaller functional variants through the genetically encoded deletions of the central regions. These data interject significant complications in the reassessment of tissue and cellular distribution of Ca v ␤ 2 variants. Development of specific immunohistochemical tools for ␤ 2f and ␤ 2g is problematic because they share the same aa sequences with other ␤ 2 subunits. Given the large size of untranslated regions of the ␤ 2 -subunit transcripts (15), relatively small structural deletions encoded in ␤ 2f and ␤ 2g may be difficult to assess by Northern blot analysis.
Our results also show that alternative splicing of the ␤ 2 subunit gene may affect regulation of the Ca v 1.2 channels in the human heart by PKA. Activation of PKA through the ␤-adrenergic receptor pathway is crucial for an up-regulation of the cardiac L-type Ca 2ϩ currents (27). This effect was found to be in part because of PKA phosphorylation of a single specific site of the ␤ 2 subunit (20). Because this site is genetically deleted from the naturally occurring short splice variants of the human ␤ 2 subunit described in this paper, these variants may have a distinct role (or no role) in human cardiac electrophysiology. In any case, the discovery of the new functional short ␤ 2 -subunit isoforms lacking many of the predicted functional motifs adds understanding to the molecular bases for the ␤-subunit modulation of Ca 2ϩ channels and may give rise to new molecular tools of the study of mechanisms of Ca 2ϩ signal transduction.