Analysis of the Complex between Ca2+ Channel β-Subunit and the Rem GTPase*

Voltage-gated calcium channels are multiprotein complexes that regulate calcium influx and are important contributors to cardiac excitability and contractility. The auxiliary β-subunit (CaVβ) binds a conserved domain (the α-interaction domain (AID)) of the pore-forming CaVα1 subunit to modulate channel gating properties and promote cell surface trafficking. Recently, members of the RGK family of small GTPases (Rem, Rem2, Rad, Gem/Kir) have been identified as novel contributors to the regulation of L-type calcium channel activity. Here, we describe the Rem-association domain within CaVβ2a. The Rem interaction module is located in a ∼130-residue region within the highly conserved guanylate kinase domain that also directs AID binding. Importantly, CaVβ mutants were identified that lost the ability to bind AID but retained their association with Rem, indicating that the AID and Rem association sites of CaVβ2a are structurally distinct. In vitro binding studies indicate that the affinity of Rem for CaVβ2a interaction is lower than that of AID for CaVβ2a. Furthermore, in vitro binding studies indicate that Rem association does not inhibit the interaction of CaVβ2a with AID. Instead, CaVβ can simultaneously associate with both Rem and CaVα1-AID. Previous studies had suggested that RGK proteins may regulate Ca2+ channel activity by blocking the association of CaVβ subunits with CaVα1 to inhibit plasma membrane trafficking. However, surface biotinylation studies in HIT-T15 cells indicate that Rem can acutely modulate channel function without decreasing the density of L-type channels at the plasma membrane. Together these data suggest that Rem-dependent Ca2+ channel modulation involves formation of a Rem·CaVβ·AID regulatory complex without the need to disrupt CaVα1·CaVβ association or alter CaVα1 expression at the plasma membrane.

tion, secretion, and gene expression (1). L-type voltage-activated calcium channels are comprised of a pore-forming ␣ 1 -subunit and regulatory ␣ 2 ␦ and -␤ subunits. The Ca V ␣ 1 subunit allows calcium entry in response to membrane depolarization, but Ca V ␤ subunits modulate the functional properties of the mature channel complex by regulating both the electrophysiological properties and trafficking of the Ca V ␣ 1 subunit (2). A conserved ␣ 1 -interaction domain (AID) 4 located between the first and second repeats of the Ca V ␣ 1 -subunit (loop I-II) serves as a high affinity binding site for all Ca V ␤ subunits (3). AID association promotes plasma membrane trafficking of the Ca V ␤⅐Ca V ␣ 1 complex and results in higher current densities (2,4). Early biochemical studies defined a conserved ␤-interaction domain (BID) within all Ca V ␤ subunits as directing AID association (2,5). However, recent modeling and crystallography studies have shown that ␤ subunits share homology with the Src homology 3-guanylate kinase (SH3-GK) core of membraneassociated guanylate kinases and have defined a GK domain ␣-subunit binding pocket (ABP) as required for Ca V ␣ 1 association (2, 6 -8). Importantly, although the GK domain alone directs AID binding, both GK and SH3 domains are required for the regulation of Ca 2ϩ channel activity. These studies suggest that ␣ 1 interaction(s) with Ca V ␤ outside of the AID domain is critical for modulating channel function (9 -13).
The role of many membrane-associated guanylate kinase family proteins as scaffolding molecules (14) has led to the suggestion that ␤ subunits may function to recruit additional regulatory factors to the Ca 2ϩ channel complex (2,15). This notion is strengthened by the recent discovery of the RGK GTPase subfamily (16 -20), including Gem, Rem, Rem2, and Rad, as Ca V ␤ binding partners that serve as negative regulators of L-type channel activity (21)(22)(23)(24). In addition to their ability to associate with ␤ subunits , RGK GTPases have several unique characteristics that distinguish them from the majority of the Ras superfamily (25). These include extended N and C termini, low rates of intrinsic GTPase activity, the lack of known lipid anchors, and the potential for distinct modes of regulation including phosphorylation-mediated 14-3-3 binding, calmodulin association, and transcriptional regulation (17, 18, 26 -31). Although all members of the RGK family have been found to inhibit voltage-gated Ca 2ϩ channel activity in a manner that appears to depend upon ␤-subunit interaction (21)(22)(23)(31)(32)(33), many issues remain concerning their mechanism of action. Although several recent studies have indicated that the expression of RGK GTPases inhibits the trafficking of co-transfected epitope-tagged Ca V ␣ 1 subunits to the plasma membrane (21,26,34), we have shown that expression of Rem2 can modulate both endogenous Ca 2ϩ channel activity and glucose-dependent insulin release in insulinoma cells without obviously altering membrane expression of the endogenous channel (23). Moreover, overexpression of Rem2 had no effect on the surface density of N-type Ca 2ϩ channels stably expressed in tsA201 cells while potently inhibiting channel function (32). Thus, questions concerning the nature of the RGK-␤-subunit association and its role in RGK-dependent regulation of Ca 2ϩ channel activity and trafficking remain to be addressed.
In the present study we define a region located within the conserved GK domain as being critically involved in the interaction of ␤ subunits with Rem. Biochemical characterization indicates that the Rem-␤-subunit interaction is of relatively low affinity, with competitive binding studies demonstrating that Rem fails to effectively compete with AID for ␤-subunit association. Moreover, Rem can enter into a Ca V ␤ 2a ⅐AID complex, establishing that Ca V ␤⅐Rem binding is mechanistically and structurally distinct from Ca V ␤⅐AID binding. Taken together, these data indicate that Rem does not inhibit channel activity by competing with Ca V ␣ 1 subunits for a limiting intracellular pool of uncomplexed Ca V ␤ subunits. In support of this hypothesis, surface biotinylation studies indicate no change in the number of surface-exposed Ca 2ϩ channels after Rem co-expression at a time when channel activity is greatly inhibited. These studies suggest that Rem-mediated regulation of Ca 2ϩ channel activity involves direct regulation of the plasma membrane-located channel complex and can occur without the need to disrupt the steady-state levels of surface-expressed Ca 2ϩ channels.

EXPERIMENTAL PROCEDURES
Plasmids-Mammalian expression vectors for Ca V 1.2␣-subunit, ␤ 2a -subunit, FLAG epitope-tagged ␤ 2a -subunit, and HA epitope-tagged Rem have been described previously (22). Ca V ␤ 2a was subcloned into pCite 4 (Novagen) for production of in vitro translated protein. Truncation mutations were generated by PCR using Ca V ␤ 2a as the template and fully sequenced. The Ca V ␣1.2 loop I-II and loop II-III domains were generated by PCR and subcloned into pGex KG for production of recombinant proteins.
Protein Production-GST Rem and thrombin-cleaved Rem were produced as described previously (17). GST Ca V ␣1.2 loop I-II (AID) was made in BL21DE3 and affinity-purified on glutathione beads (Sigma). The protein was then dialyzed into 50 mM Tris, 150 mM NaCl, 1 mM DTT, 10% glycerol and used in binding assays. When needed, thrombin was used to cleave the Ca V ␣1.2 loop I-II from GST by washing GST-Ca V ␣1.2AID immobilized on glutathione beads with thrombin cleavage buffer (17) and then incubating with thrombin for 1 h with rotation. Thrombin was then removed with benzamidine-Sepharose (17), and the protein concentration was determined by the Bradford assay. To ensure that thrombin was removed by benzamidine-Sepharose, GST immobilized on glutathione beads was incubated with thrombin and then subjected to treatment with benzamidine-Sepharose. This treatment labeled Buffer in Fig. 2B indicates that thrombin was efficiently removed.
In Vitro Pulldown Assays-Radiolabeled full-length Ca V ␤ 2asubunit or the indicated Ca V ␤ 2a truncations were prepared by in vitro transcription and translation in the presence of [ 35 S]methionine using the Single Tube Protein System 3 (STP3) kit (Novagen) according to the manufacturers' instructions. Ca V ␤ 2a or the indicated Ca V ␤ 2a truncation mutants cloned into the pCITE vector were used as the template. Binding of radiolabeled ␤ 2a to Rem was assessed as follows. All manipulations were carried out at 4°C. Ten l glutathione-Sepharose beads (GE Healthcare) were washed 2 times with 500 l of EDTA buffer (50 mM Tris, pH 7.5, 100 mM NaCl, 0.05% Tween 20, 0.1 mM DTT,1 mM EDTA). The beads were resuspended in 1 ml of EDTA buffer, and either GST (10 g) or GST⅐Rem (10 g) was added. The beads were incubated for 5 min with end-over-end rotation at 4°C. The beads were washed 2 times with 1 ml of EDTA buffer and then with 1 ml of EDTA buffer or 1 ml of GDP buffer (50 mM Tris, pH 7.5, 100 mM NaCl, 0.05% Tween 20, 0.1 mM DTT, 10 mM MgCl 2 , 20 M GDP) or 1 ml GTP buffer (50 mM Tris pH 7.5, 100 mM NaCl, 0.05% Tween 20, 0.1 mM DTT, 10 mM MgCl 2, 20 M GTP␥S) as indicated to facilitate nucleotide exchange. The buffer was aspirated so that 20 l of a 50% slurry was allowed to remain in the tube. EDTA, GDP, or GTP buffer (76 l) was added as indicated, and binding was initiated by the addition of 35 S-labeled ␤ 2a -subunit (4 l). The reaction was incubated for 3 h with end-over-end rotation. The beads were then washed three times with the same buffer used in the assay. GST or GST Rem was eluted from the beads with two 20-l washes of assay buffer supplemented with 25 mM glutathione. The eluted proteins were resolved on 10% SDS-PAGE gels, which were dried and exposed to film for 16 -72 h.
Competition Assays-The ability of Rem to compete with Ca V ␣ subunit intracellular loop I-II (AID) was assayed as follows. First, the minimal amount of GST⅐AID protein required to bind 35 S-labeled ␤ 2a was determined by adding the indicated amount of GST⅐AID or unfused GST to 10 l of pre-equilibrated glutathione-Sepharose beads (GE Healthcare). GTP buffer (76 l) was then added, and binding was initiated by the addition of 4 l of 35 S-labeled ␤ 2a subunit. The reaction was incubated for 3 h with end-over-end rotation. The beads were then washed 3 times with 500 l of GTP buffer. Bound protein was eluted from the beads with two 20-l washes of assay buffer containing 25 mM glutathione. The eluted proteins were resolved on 10% SDS-PAGE gels, which were dried and exposed to film for 16 -72 h to examine ␤ 2a /AID association. The ability of Rem to compete with this interaction was determined by adding the indicated amount of thrombin cleaved Rem to the reaction. As a positive control, I-II loop that had been thrombin-cleaved from GST was used. Also, a mock cleavage buffer treatment was used to confirm that thrombin had been completely removed from the cleaved proteins.
In Vivo GST Pulldown Assays-HEK293 cells were transiently transfected with the indicated plasmids as described (35). 48 h post-transfection cells were harvested and lysed in 20 Defining the Rem-Interaction Domain of Ca 2؉ Channel ␤ Subunits mM Tris, pH 7.5, 250 mM NaCl, 1% Triton X-100, 0.5 mM DTT, 1ϫ protease inhibitor mixture (Calbiochem), 10 mM MgCl 2, and 10 M GTP␥S and centrifuged at 100,000 ϫ g for 10 min. The supernatant was harvested, and protein concentration was determined using the Bradford assay. Cleared supernatant (1 mg) was incubated with GST (control) or GST⅐AID and 10 l of glutathione-agarose (GE Healthcare) as indicated for 3 h at 4°C with gentle end-over-end rotation. The beads were then isolated by centrifugation for 10 s at 14000 rpm in a microcentrifuge, and 5 l of the supernatant was retained for analysis. The beads were washed 3 times with 1 ml of lysis buffer and then eluted with assay buffer containing 25 mM glutathione. The eluted fraction was then analyzed with the supernatant for the presence of FLAG-Ca V ␤ 2a as follows. The supernatant and pellet were boiled in SDS-PAGE buffer and resolved on 10% SDS-PAGE gels. The gels were transferred to nitrocellulose and immunoblotted with either anti-FLAG or anti-HA monoclonal antibodies as indicated.
Surface Biotinylation Studies-HIT-T15 cells were obtained from ATCC and maintained in F12-K media (Invitrogen) supplemented with 50 g/ml gentamicin, 2.5% fetal bovine serum, and 10% dialyzed horse serum (prepared by dialyzing horse serum in 14,000-kDa cutoff dialysis bags extensively versus 0.15 M NaCl at 4°C). Ten-cm dishes were seeded with cells the day before infection. Cells were either cultured alone (uninfected control) or incubated for 24 h with CsCl-purified adenovirusexpressing GFP (control) or co-expressing GFP and Rem or GFP and Rem 1-265 (10 7 plaque-forming units/ml) (22). This resulted in near complete HIT-T15 cell infection based on the analysis of GFP-positive cells (24 h post-infection). Monolayers were washed three times with ice-cold phosphate-buffered saline (PBS), and surface proteins were biotinylated using 1 mg/ml membrane-impermeant sulfo-NHS-LC-biotin (Pierce) in PBS for 1 h at 4°C with gentle rocking. Cells were then washed three times with ice-cold phosphate-buffered saline, harvested on ice in 1 ml of Versene (Invitrogen), and pelleted by gentle centrifugation, and the Versene was aspirated. Radioimmune precipitation assay buffer (1% Triton X-100, 1% deoxycholic acid, 0.1% SDS, 50 mM Tris-HCl, pH 7.4, 1ϫ protease inhibitor mixture (Calbiochem)) was added to the cell pellet, which was sonicated twice for 10 s (Kontes). The soluble fraction was isolated after centrifugation at 100,000 ϫ g for 10 min. Protein concentrations were determined using the Bio-Rad assay kit with bovine serum albumin as a standard. Biotinylated proteins were isolated by adding cleared cell lysates (500 g) to 100 l (50% slurry) of streptavidin beads (Pierce) in a total volume of 1 ml of radioimmune precipitation assay buffer. The reaction was gently rotated end over end at 4°C for 1.5 h, and resin was pelleted by centrifugation, washed once with radioimmune precipitation assay buffer (RIPA) containing 0.3 M NaCl (two times), twice with RIPA containing 0.15 M NaCl, and finally twice with wash buffer containing no detergent (150 mM NaCl, 50 mM Tris-HCl, pH 7.4, 2.5 mM EDTA). The beads were resuspended in 30 l of 2ϫ SDS loading buffer and boiled for 5 min. The released protein as well as 10 g of the input was resolved using 6% SDS-PAGE gel electrophoresis, transferred to nitrocellulose, and subjected to immunoblot analysis with affinity-purified L-type calcium channel ␣-subunit polyclonal antibody at 2 g/ml and horseradish peroxidate goat anti-rabbit (Zymed Laboratories Inc.) secondary antibody at a 1:20,000 dilution. Super signal (Pierce) was used as the enhanced chemiluminescent reagent. For inhibition studies, immunoblotting was performed as above, but the ␣-subunit polyclonal antibody was preincubated for 1 h at 20°C with 10 g/ml GST-fused Ca V 1.2 II-III loop protein. To assure that the cells remained intact throughout the surface labeling, biotinylation of glyceraldehyde-3-phosphate dehydrogenase (GAPDH), a cytosolic protein, was analyzed by immunoblotting with GAPDH monoclonal antibody (Ambion) at 1:2000 dilution.
Electrophysiology-HEK293 cells were transiently transfected with plasmids 48 h before recordings using Effectene (Qiagen) according to the manufacturer's instructions. tsA201 cells were transiently transfected 48 h before recordings using the calcium phosphate method as described previously (35).  (22). Because RGK proteins are capable of associating with all four ␤-subunit gene products (18,21,22) and Ca V ␤-subunit binding appears critical to the ability of RGK proteins to disrupt Ca V ␣ 1 function (21-23, 26), we reasoned that RGK binding occurs via a conserved Ca V ␤-subunit sequence. Although previous work suggested that Ca V ␤ binding to RGK GTPases may be GTP-dependent (21), in vitro Ca V ␤ 2a binding to Rem was found to be nucleotide-independent, with both GDP-bound and GTP␥S-bound Rem displaying equivalent Ca V ␤-subunit binding (Fig. 1B). To identify the structural domain of Ca V ␤ 2a responsible for Rem association, we used 35 S-labeled, in vivo translated wild-type and truncated Ca V ␤ 2a -subunit probes to examine the interaction between the various structural domains of Ca V ␤ 2a and both wild-type Rem and the loop I-II (AID) domain of Ca V 1.2 (Fig. 1A). GST⅐Rem and GST⅐AID both associated with 35 S-labeled in vitro trans-Defining the Rem-Interaction Domain of Ca 2؉ Channel ␤ Subunits AUGUST 18, 2006 • VOLUME 281 • NUMBER 33 lated wild-type Ca V ␤ 2a and, more importantly, with the Ca V ␤ 2a 226 -604 truncation mutant (Fig. 1C). This localized the Rem interaction site within the highly conserved GK domain (GK) found in all Ca V ␤ subunit genes and known to harbor the ABP binding domain primarily responsible for Ca V ␤subunit/AID binding (2). No interaction could be detected with Ca V ␤ 2a 1-225 , which comprises the first conserved SH3-related structural domain, or Ca V ␤ 2a 356 -604 , which included the C-terminal portion of Ca V ␤ 2a beyond the GK domain. Recent crystallographic studies have resolved the structure of the ␤-subunit in association with the AID sequence domain of Ca V ␣1 and allowed us to generate mutants predicted to disrupt  35 S-labeled ␤ 2a subunits were resolved on 10% SDS-PAGE gels, which were then dried and exposed to film. B, recombinant GST or GST⅐Rem proteins were preloaded with the indicated nucleotide, incubated with 35 S-labeled ␤ 2a or 35 S-labeled ␤ 2a 1-355 , and the reactions were then pelleted and washed as described under "Experimental Procedures." The bound fractions were subjected to SDS-polyacrylamide gel electrophoresis, and the dried gel was exposed to film. C, the ability of GST⅐Rem or GST-Ca V ␣ loop I-II (AID) fusion proteins to interact with 35 S-labeled Ca V ␤ 2a was determined as described under "Experimental Procedures." D, a schematic diagram of the Ca V ␤ 2a used in the corresponding interaction assay (C) is presented. The conserved SH3-and GK-related structural domains are indicated.

Identification of the Rem Binding Domain in
the ABP domain (6 -8). Consistent with these studies, Ca V ␤ 2a 1-355 (lacking the ␣8 helix), Ca V ␤ 2a 1-343 (deletion of both ␣5 and ␣8 helices), and Ca V ␤ 2a 261-604 (deletion of the ␣3 helix) truncation mutants (8), which delete critical portions of the ABP domain, fail to demonstrate binding to the Ca V ␣1-AID domain, but importantly, all three were capable of binding WT-Rem. Taken together, this analysis indicates that both Rem and Ca V ␣1-association domains falls within the well conserved GK module found in all ␤ subunits and suggest that these domains are separate.
Rem Does Not Disrupt the Interaction between the AID of Ca v ␣ 1 and ␤ 2a -Although the binding studies indicate that the Rem and AID association domains of ␤ 2a are structurally distinct, we next examined whether these binding domains functionally overlap. As seen in Fig. 1C, more 35 S-labeled ␤ 2a was bound to the AID domain of Ca V 1.2 than to GST⅐Rem (see the top panel). To examine the relative binding affinity in more detail, we examined the interaction of in vitro translated 35 Slabeled ␤ 2a with increasing concentrations of GST-Ca V ␣ 1 -AID or GST⅐Rem. As shown in Fig. 2A, Ca V ␤ 2a demonstrates greater interaction with the isolated AID domain than with Rem, with similar amounts of 35 S-labeled Ca V ␤ 2a found to associate with 300 pg of GST⅐AID as with 10 g of GST⅐Rem. As expected, no interaction between radiolabeled ␤ 2a and unfused GST protein was seen in control GST pull-down reactions. This affinity difference suggested that Rem may not be capable of competing with the Ca V ␣ 1 -AID for Ca V ␤ 2a binding. To test this notion, we next examined the ability of increasing concentrations of untagged wild-type Rem to disrupt the association of GST⅐AID with 35 S-labeled Ca V ␤ 2a . As shown in Fig. 2B, more than a 100-fold molar excess of recombinant Rem protein failed to disrupt GST⅐AID-35 S-labeled Ca V ␤ 2a association, although the addition of only 25 ng of the unfused Ca V ␣ 1 -AID effectively inhibited the interaction of Ca V ␤ 2a with GST⅐AID. Although Rem failed to efficiently disrupt ␤-subunit association with AID, it inhibited the interaction of GST⅐Rem with ␤ 2a (Fig. 2C). Taken together, these results indicate that Rem displays much lower binding affinity for Ca V ␤ subunits than the previously described high affinity interaction between Ca V ␣ 1 -loopI-II and Ca V ␤ subunits and indicate that Rem does not disrupt AID/ Ca V ␤-subunit binding.
Contribution of ␤-Subunit to Rem-mediated Ca 2ϩ Channel Regulation-Although it is now well established that RGK GTPases associate with a range of Ca V ␤ subunits (21)(22)(23), the question of whether the nature of the auxiliary ␤-subunit modulates Rem-mediated Ca 2ϩ channel regulation has not been thoroughly examined. Because the mapping data indicate that Rem binds ␤ subunits near the junction of the highly conserved GK domain and a variable linker domain positioned between the SH3 and GK domains (Fig. 1C), we next examined whether structural differences within distinct ␤ subunits may functionally contribute to Rem-dependent channel regulation. To examine this issue, we assessed the ability of Rem to regulate Ca V 1.2 channel complexes containing either Ca V ␤ 1b or Ca V ␤ 2a . These ␤ subunits differ greatly within their variable domains (Fig. 3C), including the domain between the conserved SH3 and GK modules, but demonstrate similar in vitro binding to Rem and Rem2 (22,23).
As seen in Fig. 3B, HEK293 cells transiently co-transfected with Ca V 1.2 and Ca V ␤ 2a express a Ͼ9 pA/pF peak inward current (9.33 Ϯ 1.91 pA/pF; n ϭ 7). As we have described previously (22), co-expression of wild-type Rem with Ca V 1.2 and ␤ 2a results in a complete absence of detectable ionic current expression (Fig. 3B). Co-expression of Ca V 1.2 and Ca V ␤ 1b in HEK293 cells results in a Ͼ5 pA/pF peak inward current (5.85 Ϯ 2.02 pA/pF; n ϭ 10), and importantly, expression of Rem results in a complete blockade of ionic current expression (Fig. 3A). These data appear to rule out a major contribution of the variable regions of different ␤ subunits to Rem-mediated Ca 2ϩ channel activity. Taken together these results suggest that Rem association occurs within the well conserved GK domain found within all ␤-subunit proteins but that sequence differences within the variable regions of ␤ subunits are unlikely to significantly modulate Rem-mediated channel inhibition.
Rem Does Not Disrupt ␤-Subunit/AID Domain Association-It is known that Ca V ␤ subunits regulate Ca 2ϩ channel activity in part through interaction with the AID domain in Ca V ␣ subunits (2). In addition, it has been suggested that the Rem-related Gem GTPase regulates Ca 2ϩ channel function by inhibiting the trafficking of newly synthesized Ca V ␣ 1 subunits to the plasma membrane by sequestering cellular ␤ subunits (21,34). The in vitro binding data presented in Fig. 2 indicate that Rem cannot effectively compete with Ca V ␣ 1 for cellular Ca V ␤ subunits; however, these studies may lack endogenous cellular factors that promote Rem-␤-subunit complex formation. To further examine the ability of Rem to interfere with Ca V ␣/␤-subunit association in a cellular context, HEK293 cells were transfected with FLAG-tagged wild-type Ca V ␤ 2a , and empty vector or Flag- with the indicated amount of inhibitor, and binding assays were performed as described under "Experimental Procedures." ␤ 2a was co-transfected with either HA-tagged wild-type Rem or HA-Rem 1-265 , a previously characterized C-terminal Rem truncation mutant that does not bind Ca V ␤ subunits (22). Cell lysates were then incubated with recombinant GST⅐AID, the bound complexes were isolated with glutathione-Sepharose, and Ca V ␤ association was analyzed by anti-FLAG immunoblotting. FLAG-␤ 2a was found in the pellet fraction in a GST⅐AIDdependent manner (Fig. 4A). The specificity of the binding to GST⅐AID was confirmed by the finding that Ca V ␤ 2a failed to bind to glutathione-Sepharose in the absence of GST-loop I-II.
To determine whether additional cellular factors stabilize the Rem-Ca V ␤ interaction, we explored the ability of Rem to disrupt the association of Flag-␤ 2a with GST⅐AID when co-expressed in HEK293 cells. As seen in Fig. 4A, FLAG-␤ 2a associated with GST⅐AID but not with GST alone, whereas wild-type Rem failed to bind either GST⅐AID or GST. Co-expression of either wild-type Rem or Rem 1-265 did not result in a detectable disruption of Ca V ␤ 2a ⅐AID complex formation. Thus, although Rem expression in HEK293 cells completely inhibits de novo Ca 2ϩ current expression, the in vitro and in vivo binding data indicate that Rem modulates channel function without necessarily disrupting the association between Ca V ␣1 and Ca V ␤ subunits.
Formation of a Rem, Ca V ␤ 2a , AID Domain Complex-Because the minimal Rem and AID interaction domains on ␤ subunits do not overlap and Rem does not disrupt Ca V ␣ 1 /␤subunit interaction, we next investigated whether Rem might modulate Ca 2ϩ channel activity through direct association with the Ca V ␣ 1 ⅐␤ channel complex. To explore this possibility, HEK293 cells were co-transfected with FLAG-tagged Ca V ␤ 2a and empty vector or with HA-tagged wild-type Rem. Cell lysates were then incubated with recombinant GST⅐AID, the bound complexes were isolated with glutathione-Sepharose resin, and Rem association was analyzed by anti-HA immunoblotting. As seen in Fig. 4B, HA-Rem was found associated with GST⅐AID, but not with unfused GST, in a FLAG-Ca V ␤ 2a -dependent manner. These findings indicate that ␤ 2a bound to Rem is still capable of associating with the AID domain. To confirm that ␤-subunit association was necessary for Rem association with the GST⅐AID complex, the ability of Rem 1-265 to enter the complex was examined. The binding of HA-Rem 1-265 to the GST⅐AID complex was significantly reduced compared with that of wild-type Rem (Fig. 4B), consistent with its reduced Ca V ␤binding affinity (22). These results suggest that Rem does not compete with Ca V ␣ 1 -AID for ␤-subunit association but, instead, that Ca V ␤ functions as a necessary scaffold, mediating independent interactions with both Rem and Ca V ␣ 1 to promote a higher order complex.

Rem Inhibits Ca 2ϩ Channel Function in HIT-T15 Cells without Reduction in Membrane Localization-Previous studies
have suggested that the association of Gem and Rem2 with ␤ subunits prevents the trafficking of Ca V ␣ 1 subunits to the plasma membrane, resulting in down-regulation of channel activity (21,26,34). In contrast, two recent reports indicate that Rem2 expression inhibits voltage-gated Ca 2ϩ channel current without affecting channel surface density (23,32). Thus, the nature of RGK-mediated Ca 2ϩ channel regulation remains controversial. To begin to explore the mechanism of Rem-mediated channel regulation, we next determined whether the number of surface-exposed Ca 2ϩ channels in HIT-T15 cells was altered after adenoviral-mediated Rem expression. Surface proteins were biotinylated with membrane-impermanent sulfo-NHS-LC-biotin 24 h after adenoviral infection, the same time that patch clamp analysis demonstrated Rem-mediated inhibition of Ca 2ϩ channel function in parallel cultures (Fig.  5A). Biotinylated proteins were isolated by incubation with streptavidin resin and subjected to Western blot analysis using an anti-Ca V ␣ 1 antibody. As seen in Fig. 5B, the surface expression of endogenous Ca V ␣ 1 subunits was not detectably altered by Rem expression when compared with either uninfected or control Ad-GFP-infected HIT-T15 cell cultures. As an added control, expression of the Rem 1-265 , which displays diminished Ca V ␤-subunit binding and fails to regulate Ca 2ϩ channel function (22), did not alter Ca 2ϩ channel surface expression. Endogenous Ca V ␣ 1 protein expression was unaffected by Rem (Fig.  , n ϭ 8). Rem co-expression completely inhibits current through this channel complex. C, sequence identity between ␤ 2a and ␤ 1b is high in the conserved SH3 and GK domains but much lower in the variable N-terminal, C-terminal, and linker regions. 5B), indicating that Rem signaling does not markedly reduce overall channel stability. Taken together, these data indicate that Rem regulates channel activity in HIT-T15 cells without reducing the surface density of L-type Ca 2ϩ channels.
Ca V ␤ Association Is Important for Rem-mediated Channel Regulation-To explore the significance of ␤-subunit association, we examined the effect of ␤ 2a 260 -604 mutant expression on Rem-dependent channel regulation. Because the ␤ 2a 260 -604 mutant lacks an SH3 domain and does not bind Ca V ␣ 1 -AID but retains the ability to bind Rem (Fig. 1C), we reasoned that the mutant would not modulate Ca V ␣ 1 function but might compete with wild-type Ca V ␤ 2a for Rem association. In agreement with our initial studies, FLAG-tagged ␤ 2a 260 -604 retained the ability to associate with Rem ( Fig. 6C) but failed to bind GST⅐AID (Fig. 6A). Cells transiently co-transfected Ca V 1.2 and Ca V ␤ 2a expressed Ͼ16 pA/pF peak inward current (Fig. 6B, open circles, 16.9 Ϯ 3.76 pA/pF; n ϭ 8). As expected for a ␤-subunit mutant unable to bind AID and lacking its SH3-related domain, ␤ 2a 260 -604 was unable to modulate Ca V ␣ 1 activity, having no effect on current amplitude or inactivation kinetics when co-expressed with either Ca V 1.2 alone (Fig. 6B, closed squares, 0.953 Ϯ 0.970 pA/pF, n ϭ 8) or with a wild-type Ca V ␣ 1 /Ca V ␤ channel (Fig. 6D, open triangles, 18.2 Ϯ 5.44 pA/pF, n ϭ 16). As we have described previously (22), co-expression of wild-type Rem with Ca V 1.2 and ␤ 2a results in a complete absence of detectable ionic current expression (Fig. 6D, closed squares, 0.136 Ϯ 0.157 pA/pF, n ϭ 16), and importantly, co-expression of ␤ 2a 260 -604 was able to partially relieve the Rem blockade of ionic current expression (Fig. 6D). In these experiments, ␤ 2a 260 -604 expression increased peak current amplitude to Ͼ3 pA/pF peak inward current (3.84 Ϯ 1.77 pA/pF; n ϭ 15). Taken together, these data indicate that Ca V ␤ association plays a critical role in Remmediated channel regulation and suggests that formation of a Ca V ␣ 1 ⅐Ca V ␤⅐Rem scaffolding complex may be required Rem-dependent channel regulation.  , adenoviral-infected HIT-T15 cells were surface-biotinylated with sulfo-NHS-LC-biotin, and biotinylated proteins were isolated using streptavidin resin. The entire biotinylated eluted protein fraction (500 g of whole cell lysate) (Pulldown) and 10 g of input lysate (Input) were resolved on 6% SDS-PAGE gels, transferred to nitrocellulose, and immunoblotted with anti-Ca V ␣ 1 antibody to determine relative surface expression of Ca V ␣ 1 channels. Membranes were subsequently re-blotted using anti-glyceraldehyde-3-phosphate dehydrogenase (GAPDH) antibody as a control for cell integrity. Input lysate (10 g) was resolved on a 10% SDS-PAGE gel, transferred to nitrocellulose, and immunoblotted with anti-Rem antibody as a control for Rem protein expression.

DISCUSSION
It is now clear that all members of the RGK family of GTPbinding proteins interact with auxiliary ␤ subunits to regulate L-type Ca 2ϩ channel activity (18,21,23,(31)(32)(33). In the present study we provide new insight into the mechanism of RGK-mediated Ca 2ϩ channel regulation through the identification of the Rem interaction site on Ca V ␤ 2a . The most interesting result of this study is that although both the Rem and AID association motifs are located within the conserved Ca V ␤subunit GK domain, Rem binding does not inhibit AID binding. Instead, these studies suggest that Ca V ␤ subunits serve as molecular scaffolds to promote a novel regulatory interaction between Rem and the AID domain of Ca V ␣ 1 . In addition, surface biotinylation studies indicate that this Ca V ␤-subunitmediated regulatory complex inhibits Ca 2ϩ channel activity without down-regulating surface-expressed channel. Based upon these findings we propose that Rem-dependent inhibition of Ca 2ϩ channel activity involves ␤-subunit-dependent regulation of the membrane-localized channel complex rather than disruption of plasma membrane channel trafficking.
Auxiliary Ca V ␤ subunits share a conserved domain structure with three variable regions separated by conserved SH3-like and GK-like domains (Fig. 3C) (2). In agreement with recent crystallographic studies (6 -8), in vitro deletion analysis demonstrates that the high affinity Ca V ␣ 1 subunit interaction region is located within the ABP region of the GK domain. Although we did not directly examine the effect of truncation on the folding of the SH3/GK core of Ca V ␤ 2a , deletion of helixes ␣8, ␣5, and ␣3, which contribute to the ABP domain (6 -8), disrupted association with Ca V ␣ 1 -AID (Fig.  1). More importantly, ␤ 2a 1-355 (lacking ␣8), Ca V ␤ 2a 261-604 (deletion of the ␣3 helix), and ␤ 2a 1-343 (deletion of both ␣5 and ␣8) (8) GK domain mutants that fail to bind AID are each capable of associating with Rem (Fig. 1C). Moreover, deletion of the entire N-terminal domain of Ca V ␤ 2a , including the conserved SH3-like domain, failed to disrupt either Rem or AID association. These results suggest that Rem and perhaps all other RGK GTPases will demonstrate uniform binding with all Ca V ␤ subunits containing the conserved GK module. Indeed, extensive in vitro binding analysis supports this conclusion (21-23, 26, 34). Moreover, these data indicate that the nature of the auxiliary Ca V ␤-subunit, as long as it con-tains an intact GK module, is unlikely to significantly modulate RGK-dependent channel regulation. This is in agreement with the ability of Rem to completely inhibit ionic current expression from heterologously expressed Ca V 1.2␤ 2a and Ca V 1.2␤ 1b channels (Fig.  3) and data from a variety of native and heterologous expression studies (21-23, 26, 31-34). Finally, these studies suggest a novel role for a series of exotic Ca V ␤-subunit splice variants that have recently been described (9, 36 -38). These Ca V ␤ subunits fail to express a GK module due to alternative splicing and are unable to promote channel trafficking but are capable of modulating single channel gating properties (9,10). It is likely that these short Ca V ␤ subunits will escape Rem-mediated Ca 2ϩ channel regulation and may, therefore, play a particularly prominent role in shaping the functional properties of channels in the presence of active Remmediated regulation. Studies to address these questions are currently under way.
The AID domain represents a high affinity binding site for all Ca V ␤ subunits, with the affinity of Ca V ␤⅐AID binding reported to be in the low nanomolar range (2). Although recent work has suggested that RGK proteins compete with Ca V ␣ 1 for available Ca V ␤, the in vitro binding studies presented here indicate that Rem binds ␤ subunits with an apparent affinity much lower than that for the association of Ca V ␤ and the AID domain (Fig.  2). More importantly, these assays show that Rem and AID can simultaneously associate with Ca V ␤ 2a (Fig. 4), indicating that Rem and AID interaction sites are distinct, without obvious structural or functional overlap. Although the lack of effective Rem antibody reagents has thwarted analysis of the putative interaction of endogenous Rem with the native Ca 2ϩ channel complex, these data suggest that Rem association will depend upon the formation of the AID-␤-subunit interaction, with Ca V ␤ serving as a scaffold to recruit Rem to the channel complex. Support for this model comes from studies using the ␤ 2a 260 -604 mutant (Fig. 6). As expected, overexpression studies found that this mutant was unable to modulate channel activity when co-expressed with Ca V ␣ 1 or alter wild-type Ca V ␣ 1 /␤ 2a channel activity. Interestingly, ␤ 2a 260 -604 co-expression can partially relieve Rem-mediated channel inhibition (Fig. 6). These data further support a crucial role for ␤-subunit association in Rem-mediated channel regulation. Also, Rem, in the absence of ␤-subunit overexpression, partially blocks Ca V ␣ 1 current, but ␤ 2a co-expression yields a complete blockade (41). Thus, it will be important to determine whether association with the Ca V ␣ 1 subunit or other components of the larger membrane-associated channel complex may influence Rem function.
Our results differ from those of Sasaki et al. (34), who recently reported that the BID domain, first reported to serve as the Ca V ␤ region involved in direct interaction with the Ca V ␣ 1 -AID, is required for association of the Rem-related Gem protein with ␤ subunits (34). However, the notion that the BID region is required for AID association has been challenged by recent structural studies that indicate that the BID domain is located within the ␤-subunit core (6 -8). Thus, deletion of the BID domain is likely to result in structural changes to the ␤-subunit, suggesting that the effects of BID mutants on Gem binding may result from alterations to protein folding or stability rather than any specific disruption to a Gem interaction site. Further studies are needed to fully define the structural interactions between RGK proteins and Ca V ␤ subunits and to examine the possibility that RGK proteins may directly interact with Ca V ␣1 subunits.
Although these studies have begun to define the motifs involved in Rem-␤-subunit association, the question of how the binding of Rem to ␤-subunit results in the inhibition of functional Ca 2ϩ channel activity remains to be fully characterized. Previous studies have suggested that co-expressed Gem and Rem2 GTPases regulate channel function at least in part by reducing the density of membrane-localized L-type Ca 2ϩ channels through selective blockade of Ca V ␣ 1 -subunit membrane trafficking (21,26,34,39). Indeed, Gem overexpression in cardiomyocytes results in a reduction in L-type Ca 2ϩ channel gating currents, consistent with inhibition of functional channel expression at the plasma membrane (33). However, we observe potent Ca 2ϩ channel inhibition within 24 h of adenoviral-mediated Rem expression in HIT-T15 cells without a corresponding reduction in the density of plasma membrane Ca V a 1 subunits (Fig. 5). In support of this finding, we reported similar properties for the Rem2 protein (23), and Chen et al. (32), using a ligand binding assay, have shown that Ca V ␣2.2-subunit surface expression is not reduced by Rem2 overexpression at a time when N-type channel activity is potently inhibited. These data indicate that both Rem and Rem2 allow for acute channel regulation without disrupting channel trafficking or significantly reducing Ca V ␣ 1 -subunit stability at the cell surface. Taken together, these studies suggest that Rem, Rem2, and perhaps other RGK proteins may regulate Ca 2ϩ channel activity using two complementary mechanisms. The first signaling pathway, which has been proposed to operate for Gem and Rem2, would allow for chronic regulation of the surface density of Ca 2ϩ channels by disrupting Ca V ␣ 1 -subunit trafficking. The second regulatory mechanism, now described for both Rem and Rem2, allows for the acute regulation of Ca 2ϩ channels found within the plasma membrane.
Although these studies contradict the hypothesis that RGK binding disrupts the interaction of Ca V ␤ with Ca V ␣, thereby altering the trafficking of newly synthesized channels or generating a population of Ca V ␤-less channels at the cell surface, the data presented here do not address how the Rem/Ca V ␤ complex might regulate channel function. Within the Ca V ␤ proteins a structural interaction is known to exist between the SH3 and GK domains (2). Structural studies and biochemical studies have found that the GK domain is responsible for high affinity Ca V ␣ 1 -subunit binding (2). However, the isolated SH3 domain can modulate channel inactivation kinetics, and recent work (2,9 -11,13) suggests that low affinity interactions involving both the SH3 domain and the larger SH3/GK core are critical for the regulation of channel activity. In this context the recruitment of Rem to the domain interface between SH3 and GK is particularly intriguing. Whether the association of Rem with ␤-subunit may affect the intramolecular interaction or orientation of the SH3 and GK domains, promote or disrupt interactions with the Ca V ␣ 1 -subunit, or serve to recruit additional regulatory factors to the channel complex has not been addressed.
Recent studies have suggested that RGK-mediated Ca 2ϩ channel regulation and ␤-subunit binding are GTP-dependent, indicating that the effector domain of RGK proteins contributes critical contacts to ␤-subunit association (21,31). This result is surprising, since the effector domain of Ras-related GTP-binding proteins is predicted to be the primary interaction site for cellular binding proteins, and this region within the RGK proteins is highly variable (25). Thus, sequence comparisons suggest that RGK proteins are unlikely to share common effector domain binding targets, although both Ca V ␤ association and regulation of L-type Ca 2ϩ channel currents are common functions of all members of this GTPase subfamily (17,18). Indeed, at least using in vitro binding assays, Rem association with Ca V ␤ subunits is nucleotide-independent (Fig. 1B). Alternatively, it is possible that RGK proteins adopt a unique interaction face with ␤ subunits that does not involve the classical "effector loop" region (25). Recently, we have found that the conserved and extended C-terminal domain of RGK proteins might contribute to in vivo ␤ 2a -subunit association, since deletion of the terminal 32 residues from either Rem or Rad proteins destabilizes their ␤ 2a -subunit binding (22). Finally, it is possible that GTP binding might result in a conformational change that recruits additional cellular factors that promote in vivo ␤-subunit association. Resolution of these questions is likely to require structural analysis of a RGK family member-␤subunit complex. The recent report of preliminary crystallographic data for the Rad GTPase (40) represents an important first step in this process.
In summary, it is well established that the association of ion channel subunits with intracellular binding proteins is involved in the regulation of channel activity (1). The present studies provide insights into the mechanism of Rem-mediated control of Ca 2ϩ channel activity by defining the Rem interaction domain on Ca V ␤ subunits. Our results suggest a new context for considering the association of Rem and ␤ subunits because they indicate that Ca V ␤ provides a scaffold for the recruitment of the Rem GTPase into the plasma membrane Ca 2ϩ channel complex to allow for acute regulation of calcium currents. This novel control mechanism is likely integrated with other Ca 2ϩ channel regulatory processes. Indeed, we have recently found that Rem-mediated channel regulation can be modulated by cellular kinase pathways (41). Clearly, additional studies will be needed to clarify the cellular signaling pathways that control Rem activity, the molecular mechanism of Rem-Ca V ␤mediated regulation of Ca 2ϩ channel function, and whether Ca V ␤ scaffolding is a general feature of RGK protein channel control including the long-term modulation of Ca 2ϩ channel surface density.