Molecular Determinants of Gem Protein Inhibition of P/Q-type Ca2+ Channels*

Background: Gem, an RGK protein, strongly inhibits P/Q-type Ca2+ channels. Results: A 12-amino acid C-terminal fragment and a 3-amino acid motif in the core of Gem play an important role in this inhibition. Conclusion: Gem possesses two candidate inhibitory sites. Significance: This work identifies key molecular determinants of Ca2+ channel inhibition by RGK proteins, offering opportunities to modulate this inhibition. The RGK family of monomeric GTP-binding proteins potently inhibits high voltage-activated Ca2+ channels. The molecular mechanisms of this inhibition are largely unclear. In Xenopus oocytes, Gem suppresses the activity of P/Q-type Ca2+ channels on the plasma membrane. This is presumed to occur through direct interactions of one or more Gem inhibitory sites and the pore-forming Cav2.1 subunit in a manner dependent on the Ca2+ channel subunit β (Cavβ). In this study we investigated the molecular determinants in Gem that are critical for this inhibition. Like other RGK proteins, Gem contains a conserved Ras-like core and extended N and C termini. A 12-amino acid fragment in the C terminus was found to be crucial for and sufficient to produce Cavβ-dependent inhibition, suggesting that this region forms an inhibitory site. A three-amino acid motif in the core was also found to be critical, possibly forming another inhibitory site. Mutating either site individually did not hamper Gem inhibition, but mutating both sites together completely abolished Gem inhibition without affecting Gem protein expression level or disrupting Gem interaction with Cav2.1 or Cavβ. Mutating Gem residues that are crucial for interactions with previously demonstrated RGK modulators such as calmodulin, 14-3-3, and phosphatidylinositol lipids did not significantly affect Gem inhibition. These results suggest that Gem contains two candidate inhibitory sites, each capable of producing full inhibition of P/Q-type Ca2+ channels.

HVA 2 Ca 2ϩ channels, including L-, N-, P/Q-and R-type channels, conduct Ca 2ϩ into excitable cells and are important for diverse physiological processes such as neurotransmitter release, hormone secretion, muscle contraction, and gene transcription. They are typically composed of a pore-forming ␣ 1 subunit (Ca v ␣ 1 ) and auxiliary ␣ 2 ␦ and ␤ (Ca v ␤) subunits. The activity of HVA Ca 2ϩ channels is regulated by protein kinases, trimeric G proteins, membrane lipids, intracellular Ca 2ϩ , and various adaptor and signaling proteins (1,2). Of the latter, the RGK family of Ras-related monomeric small GTPases, including Rad, Rem, Rem2, and Gem/Kir, has emerged as the most potent endogenous protein inhibitors of HVA Ca 2ϩ channels (3)(4)(5)(6).
Two modes of action have been proposed for the inhibition of HVA Ca 2ϩ channels by RGK proteins; they are reducing the number of surface channels (7)(8)(9)(10)(11)(12)(13)(14) or suppressing the activity of channels on the plasma membrane (13,(15)(16)(17)(18)(19)(20). Both forms of inhibition require Ca v ␤ (14,15,21), which interacts directly with RGK proteins (7, 10, 16, 17, 19 -24). This interaction has been generally assumed to be essential for RGK inhibition (15,21). We recently showed, however, that Gem inhibition of surface P/Q-type Ca 2ϩ channels expressed in Xenopus oocytes remained intact after the disruption of the Gem-Ca v ␤ interaction and that this inhibition appeared to be caused by direct interactions between Gem and Ca v ␣ 1 (18). We postulated that Gem and Ca v ␣ 1 interact via an anchoring site in a Ca v ␤-independent manner, that Ca v ␤ induces the formation of an inhibitory site on Ca v ␣ 1 , and that an inhibitory site in Gem binds to this Ca v ␣ 1 site to cause inhibition. The regions involved in these interactions on both Ca v ␣ 1 and Gem remain to be identified.
The four RGK proteins contain a highly conserved Ras-like core and extended variable N and C termini that are absent in Ras (3,4). The core contains binding sites for Ca v ␤ (10, 18) and the binding and catalytic sites for guanine nucleotides (25,26). The N and C termini contain binding sites for calmodulin (CaM), 14-3-3, and phosphatidylinositol lipids, which regulate the subcellular distribution and function of RGK proteins (7)(8)(9)(27)(28)(29)(30)(31). The role of the interactions with these modulators in RGK inhibition of HVA Ca 2ϩ channels is controversial and is not fully characterized (4, 11-13, 15, 32).
In this study we investigated the molecular determinants in Gem that are important for its inhibition of P/Q-type Ca 2ϩ channels expressed in Xenopus oocytes. We identified two distinct regions in Gem, a 12-amino acid stretch in the C terminus and a 3-amino acid motif in the core, that are critical for Gem inhibitory action and may form two separate inhibitory sites.

EXPERIMENTAL PROCEDURES
Constructs and Cloning-For electrophysiology experiments in Xenopus oocytes, cDNAs encoding various constructs were subcloned into a modified oocyte expression vector pGEMHE. The constructs included rabbit brain Ca v 2.1 (GenBank TM accession number X57477), rat skeletal muscle ␣ 2 ␦, rat brain ␤ 3 (GenBank TM accession number M88751), a mutant ␤ 3 named ␤ 3 _Mut2 bearing the M196A/L200A mutation, and WT or mutated human skeletal muscle Gem (GenBank TM accession number BC022010).
For co-immunoprecipitation experiments in HEK 293T cells (gift of Dr. Hiroaki Matsunami at Duke University), an HA (hemagglutinin) tag was added to the N terminus of full-length Gem cDNA (mutant) with a flexible linker of three glycines in between, and the whole construct was cloned into the pCDNA3.1(Ϫ) vector (Invitrogen). An Myc tag was added to the N terminus of full-length rat brain ␤ 3 with three glycines in between, and the whole construct was cloned into the pCDNA3.1(Ϫ) vector (Invitrogen). Rabbit brain Ca v 2.1 was subcloned into the p3xFLAG-CMV-7.1 vector (Invitrogen).
Oocyte Preparation and Expression-Ovarian lobes were obtained from adult Xenopus laevis (Xenopus I) under anesthesia. Stages V-VI oocytes were prepared by treatment with 2.5 mg/ml collagenase A for 1.5-2.5 h under 200 rpm shaking in a solution containing 82.4 mM NaCl, 2.5 mM KCl, 1 mM MgCl 2 , and 5 mM HEPES (pH 7.6) and then rinsed twice (15 min each) with a solution containing 96 mM NaCl, 2.5 mM KCl, 1 mM MgCl 2 , 5 mM HEPES, 1.8 mM CaCl 2 , 100 units/ml penicillin, and 100 g/ml streptomycin (pH 7.6). Single, defolliculated oocytes were individually selected. cRNAs were synthesized in vitro, and varying amounts (0.2-5 ng) were injected into selected oocytes in various combinations. Recordings were performed 3-5 days after injection.
Electrophysiology and Data Analysis-All experiments were performed on Xenopus oocytes prepared 3-5 days after RNA injection. For whole-oocyte recordings by two-electrode voltage-clamp (TEVC), electrodes were filled with 3 mM KCl and had a resistance of 0.5-1 megaohms. The bath solution contained 10 mM BaCl 2 , 5 mM KCl, 60 mM tetraethylammonium hydroxide, 20 mM NaOH, and 5 mM HEPES (pH 7.4 with HCl). The current was evoked every 6 s by a ϩ10-mV pulse for 40 ms from a holding potential of Ϫ80 mV. For inside-out macropatch recordings from oocytes, the electrode had a diameter of 15-30 m and a resistance of 0.2-0.5 megaohms when filled with a solution containing 45 mM BaCl 2 , 80 mM KCl, and 10 mM HEPES (pH 7.3 with KOH). The bath (i.e. cytoplasmic) solution contained 125 mM KCl, 4 mM NaCl, 10 mM HEPES, and 10 mM EGTA (pH 7.3 with KOH). 0.3 M phosphatidylinositol diphosphate and 3 mM Mg-ATP were added freshly to the bath solution to attenuate rundown. Macroscopic currents were evoked from a holding potential at Ϫ80 mV by 10-ms depolarizations ranging from Ϫ30 to ϩ90 mV in 10-mV increments at a 1-s interval.
To obtain the ␤-less channels described in Fig. 3F, a fast perfusion speed (ϳ1.5 ml/min) was used. Wash-off of ␤ 3 _Mut2 was determined by monitoring the positive shift of the activation curve. To obtain the activation curve, macroscopic currents were evoked by 20-ms depolarizations ranging from Ϫ40 to ϩ100 mV in 10-mV increments at a 6-s interval. Tail currents were recorded by repolarization to Ϫ40 mV regardless of the preceding test pulse, normalized by that after the depolarization to ϩ100 mV, and plotted against the test potentials.
Currents were sampled at 10 kHz and filtered at 2.5 kHz. All data were analyzed with Clampfit and were represented as the mean Ϯ S.D. (number of observations). Statistical significance was determined using two-tailed Student's t tests.
Protein Synthesis in E. coli-Transformed DE3 bacteria were cultured at 37°C until A 600 reached 0.6 and then induced at room temperature by 0.5ϳ1 mM isopropyl 1-thio-␤-D-galactopyranoside for overnight. Cells were collected at 4000 rpm for 15 min and resuspended in a lysis solution containing 50 mM Tris-HCl, 250 mM NaCl, 2.5% glycerol, and 7 mM ␤mercaptoethanol (pH 7.8). Resuspended bacteria were sonicated with a Branson digital sonifier followed by centrifugation at 14,000 rpm for 40 min. For Gem-(Ser-68 -Lys-276) and Gem-(Ser-68 -Arg-264) protein purification, the supernatant was collected and incubated with nickel-nitrilotriacetic acid His-Bind beads (Novagen) in the presence of 20 mM imidazole at 4°C for 1 h. Proteins were eluted from the beads with 200 mM imidazole in the lysis solution. For CaM purification, most bacterial proteins were denatured by incubation at 70°C for 3 min and removed by centrifugation at 13,000 rpm for 30 min. The supernatant was supplemented with 2 mM CaCl 2 and 1 mM MgCl 2 and then was mixed with phenyl-Superose resin for 1 h. After washing the resin with a buffer containing 2 mM CaCl 2 and 1 mM MgCl 2 , calmodulin protein was eluted with a buffer containing 2 mM EGTA and 1 mM MgCl 2 . Proteins were concentrated and further purified by a Superdex 75 gel-filtration chromatography (GE Healthcare).
Pulldown Assay-MBP fusion peptides were expressed in DE3 and purified by incubation with amylose resin (New England Biolabs). After elution from the resin with 20 mM maltose, MBP fusion peptides were immobilized on nickelnitrilotriacetic acid His-Bind beads in PBS buffer with or without Ca 2ϩ (4 mM). The CaM protein bound to peptides was eluted from the beads and denatured at 100°C for 15 min. The elution was detected with Coomassie Blue staining on SDS-PAGE gel.
Cell Culture and Transfection-HEK 293T cells were maintained at 37°C in a DMEM medium containing 10% FBS, 100 units/ml penicillin, and 100 g/ml streptomycin. Lipofectamine 2000 (Invitrogen) was used for all transfections.
Oocyte Preparation for Western Blot-Oocytes were collected after TEVC recordings, washed in cold PBS, and lysed with 15-20 l/oocyte of lysis buffer (PBS, 10% glycerol, 1% Triton X-100, 1-2% protease inhibitor) using trituration through a syringe with a 25-gauge needle (10 times). Lysates were incubated at 4°C for 1 h while rotating, with brief vortexing every 15-20 min. Aliquots of the lysates were then mixed with 5ϫ SDS and boiled at 55°C for 15-20 min. 15-l lysates were loaded onto an acrylamide gel for Western blot.
SDS-PAGE and Western Blot-SDS-PAGE was performed in a Tris-glycine-SDS buffer with 8 or 12% acrylamide gel. Precision Plus Protein All Blue Standards (Bio-Rad) were used as molecular weight markers. For the Western blot, after electrophoresis, the protein gel was transferred to the PVDF membrane and processed with the Odyssey Western-blot kit (Li-Cor). The monoclonal mouse anti-HA antibody HA.11 (Covance), the monoclonal mouse anti-Myc antibody (Sigma), or the monoclonal mouse anti-FLAG antibody M2 (Sigma) was used as the primary antibody (usually overnight incubation at 4°C). After triple washes with PBS, 1% Tween, membranes were incubated with a secondary antibody (Alexa Fluor 680conjugated goat anti-mouse IgG; Invitrogen). Images were scanned and analyzed with the Odyssey Infrared Imaging System (Li-COR). The intensity of the protein bands was analyzed with the software ImageJ (NIH).
Co-immunoprecipitation-About 30 -48 h later, after transfection, cells were collected in the PBS buffer (Invitrogen (pH 7.2)) at 4°C. Cell lysates were obtained by adding 1%  (V/V) Triton X-100 (Sigma) and 1/50 (V/V) Protease Inhibitor Mixture (Sigma) and rotating at 4°C for 1 h. Supernatant was obtained by centrifuging at 14,000 rpm for 20 min and was mixed with monoclonal anti-HA antibody (HA.11) coated beads (Covance). After incubation at 4°C overnight, the beads were spun down and washed 3 times with a solu- Currents (recorded at ϩ20 mV) were obtained from inside-out membrane patches excised from oocytes expressing Cav2.1, ␣ 2 ␦, and ␤ 3 _Mut2. Before time 0, the patch had been perfused for 5 min such that the channels had lost ␤ 3 _Mut2 and become ␤-less. n ϭ 4. B-F, right panels, shown are current traces selected from a representative patch for each condition. Currents were evoked by a depolarization to ϩ20 mV from a holding potential of Ϫ80 mV and were obtained immediately before (red), during (blue), or after (green) perfusion of the indicated Gem peptides. The dashed line indicates zero current level. G, left panel, whole oocyte currents were recorded at ϩ10 mV by TEVC from oocytes expressing Ca v 2.1, ␣ 2 ␦, and ␤ 3 , without Gem (control), with WT Gem (Gem_WT), or with the indicated mutant Gem. *** p Ͻ 0.01 (compared with control). Middle panel, a Western blot shows protein expression of WT Gem and the indicated mutant Gem in the lysates of the oocytes recorded in the left panel. ␤ 3 expression was shown as the loading control. Right panel, the bar graph shows the normalized intensity of the corresponding Gem bands in the middle panel.
tion containing 1 ϫ PBS with a total NaCl concentration of 250 mM and 1/50 (V/V) protease inhibitor mixture (Sigma). The bound proteins were then eluted by adding 0.4 mg/ml HA peptide (Genscript), boiled with an SDS loading buffer for 10 min, and analyzed by SDS-PAGE and Western blot.

12-Amino Acid C-terminal Region in Gem Is Critical for Gem
Inhibition-We first carried out systematic deletions to roughly delineate the structural elements of Gem critical for its inhibition of P/Q-type Ca 2ϩ channels expressed in Xenopus oocytes (Fig. 1A). WT or mutant Gem was coexpressed with Ca v 2.1, ␣ 2 ␦, and ␤ 3 . Ba 2ϩ was used as the charge carrier, and wholeoocyte currents were recorded by TEVC. WT Gem completely abolished P/Q channel currents (Fig. 1, B-D). The core region of Gem (Gem-(68 -247)) was totally ineffective (Fig. 1B, left  panel) even though its protein level was much higher than that of WT Gem (Fig. 1B, middle and right panels), indicating that the N and/or C termini were required. Deleting the entire N terminus, however, did not affect Gem inhibition (Gem-(68 -296) in Fig. 1C), indicating that the N terminus is not necessary, consistent with previous work showing that the N terminus is not required for Rem2 inhibitory effect (15). In contrast, Gem-(1-264), in which the last 32 amino acids in the C terminus was deleted, was completely inactive (Fig. 1D, left panel) despite its much higher protein level (Fig.  1D, middle and right panels), indicating that this 32-amino acid region is critical for Gem inhibitory action. This result agrees with previous work showing that the corresponding last 32-34 C-terminal residues of Rem and Rad are crucial for their inhibitory effects (20,21,23). Another deletion construct, Gem-(68 -276), which lacks the last 20 amino acids in the C terminus as well as the entire N terminus, was still fully active (Fig. 1C). The complete lack of effect of Gem-(1-264) and the potent effect of Gem-(68 -276) demonstrate that the 12-amino acid region from Lys-265 to Lys-276 in the C terminus is indispensable for Gem inhibition.
Lys-265-Lys-276 C-terminal Region in Gem Functions as Candidate Inhibitory Site-How is the Lys-265-Lys-276 region involved in Gem inhibition? One possibility is that it forms an inhibitory site that interacts directly with P/Q channels. To test this hypothesis, a peptide corresponding to this region named GCP1 (Gem C-terminal Peptide 1, Fig. 3A) was synthesized and applied to inside-out membrane patches excised from oocytes expressing Ca v 2.1, ␣ 2 ␦, and ␤ 3 . GCP1 at 100 M greatly reduced the currents (Fig. 3B). In contrast, a peptide corresponding to the last 20 amino acids (Asn-277-Leu-296) of Gem (named GCP2, Fig. 3A) had no effect even at a doubled concentration (Fig. 3C). These results suggest that the inhibitory effect of GCP1 is specific and that the Lys-265-Lys-276 region indeed forms an inhibitory site. The inhibitory effect of GCP1, however, became much weaker when its concentration was decreased to 5 M (Fig. 3D), a concentration at which purified Gem-(Ser-68 -Lys-276) was fully effective (Fig. 2A). An explanation of this discrepancy is that Gem-(Ser-68 -Lys-276) can associate with surface P/Q channels through Gem-Ca v ␤ and Gem-Ca v ␣ 1 interactions, which would greatly increase the effective concentration of the Lys-265-Lys-276 inhibitory site near the channels.
In our previous study we showed that although purified Gem-(Ser-68 -Lys-276) protein inhibited surface P/Q channels containing Ca v ␤, it did not affect channels lacking Ca v ␤ (18). The latter, termed ␤-less channels, were created by washing off a mutant Ca v ␤ that was able to traffic Ca v 2.1 to the plasma membrane but dissociated quickly from the surface channel upon vigorous perfusion (18). Using this approach, we found that GCP1, like Gem-(Ser-68 -Lys-276), did not inhibit surface ␤-less P/Q channels (Fig. 3E). This result further demonstrates the specificity of GCP1 and suggests that GCP1 binds to a Ca v ␤induced inhibitory site on Ca v 2.1.
Remarkably, when GCP1 was scrambled into a new peptide named GCP1_S with the same amino acid content but a different sequence, it retained its inhibitory effect (Fig. 3F). This result suggests that the amino acid composition of this region rather than its sequence per se is important for its inhibitory effect.
To examine whether the inhibitory action of GCP1_S was a fortuitous effect of the peptide in isolation, we linked GCP1_S to Gem-(1-264), which as shown in Fig. 1D was inactive. The resulting construct, named Gem-(1-264)ϩGCP1_S, was able to inhibit P/Q channels, albeit it was slightly less potent than WT Gem was (Fig. 3G, left panel). In contrast, Gem-(1-264)ϩGCP2, in which the inactive GCP2 peptide was linked to Gem-(1-264), failed to cause inhibition at a similar protein level (Fig. 3G). These results further support GCP1 as a genuine inhibitory site in the full-length Gem.
CaM Binding to Gem Is Not Involved in Gem Inhibition-The identified Lys-265-Lys-276 inhibitory site has been reported to bind CaM (7-9, 14, 27). We confirmed this finding by using purified MBP-tagged GCP1 (MBP_GCP1) to pull down purified CaM in the presence of Ca 2ϩ (Fig. 4A). A fragment of the C terminus of Ca v 2.1 containing a canonical CaM binding IQ motif, denoted as Ca v 2.1-(Val-1899 -Glu-1989), served as a positive control, whereas MBP-tagged GCP2 (MBP_GCP2) served as a negative control (Fig. 4A). CaM binding to Gem is severely attenuated by mutations of Trp-270 (14). Thus, to examine whether CaM binding to the Lys-265-Lys-276 site plays a role in Gem inhibition, we mutated four large hydrophobic residues (Phe-269, Trp-270, Ile-273, and Val-274) in the Lys-265-Lys-276 region to alanines in the full-length Gem. The resultant Gem mutant (Gem_M1) showed complete inhibition of P/Q channels (Fig. 4B), indicating that CaM binding to the Lys-265-Lys-276 site is not required for Gem inhibition.
Inhibitory Site Exists in Core Region of Gem-To investigate the role of the Lys-265-Lys-276 site in the inhibition of P/Q channels by full-length Gem, we created two Gem mutants named Gem_10G and Gem_11A, in which 9 of the 12 residues in this region were simultaneously mutated to glycine or alanine, resulting in 10 glycines or 11 alanines in the region (Fig.  5A). Surprisingly, however, both Gem mutants were still capable of inhibiting P/Q channels (Fig. 5, B and C), albeit the magnitude of inhibition by Gem_10G varied in different experiments, ranging from 60 to 100%. The underlying cause for the variation was unclear, but a potential factor was the variation in the effective concentrations of the Gem_10G protein in different batches of oocytes. Regardless of the varying inhibition, the results in Fig. 5, B and C, suggest the existence of an additional inhibitory site in Gem. To map this site, further deletions and point mutations were made in Gem_10G or Gem_11A. Deleting the N terminus of Gem_10G or Gem_11A did not abolish the inhibition (Fig. 5D) nor did simultaneously mutating 4 or 5 amino acids in the C terminus of Gem_10G to alanine either upstream or downstream of Lys-265-Lys-276 (Fig. 6). These results suggest that the additional inhibitory site resides in the core region of Gem.
To pinpoint this site, we selectively mutated residues in the core region that are conserved among RGK proteins but are distinct from the corresponding Ras residues (labeled red and 1, 2, 3, and 4 in Fig. 7A). To minimize undesirable structural changes, these residues were mutated to their counterparts in Ras in the Gem_10G background (Fig. 7B). Three of the resulting Gem mutants (Gem_10G_CM1, Gem_10G_CM2, and Gem_10G_CM3) were still able to cause inhibition, albeit less potently than WT Gem but comparable to Gem_10G itself (Fig. 7C), but one mutant, Gem_10G_CM4 (carrying the L241K/R242H/R243K mutation), completely lost its activity (Fig. 7D), indicating that residues Leu-241, Arg-242, and Arg-243 play an important role in Gem inhibition. However, when these residues were mutated in WT Gem (producing a mutant named Gem_CM4), the inhibitory effect was fully preserved (Fig. 7, D and E, left panel). This result suggests that the inhibitory site in the core region and the Lys-265-Lys-276 site in the C terminus can act indepen- B, shown are whole-oocyte currents recorded at ϩ10 mV in oocytes expressing Ca v 2.1, ␣ 2 ␦, and ␤ 3 without Gem (control), with WT Gem, or with Gem_M1, in which four hydrophobic residues (asterisks) in the indicated region were mutated to alanine.  . The C terminus of Gem does not contain a second inhibitory site. A, shown is the amino acid sequence of the last 41 residues (Gln-256 -Leu-296) of Gem and Gem_10G. The indicated full-length Gem mutants were all made from Gem_10G, with the asterisks indicating residues that were mutated to alanine. B and C, shown are whole-oocyte currents recorded at ϩ10 mV in oocytes expressing Ca v 2.1, ␣ 2 ␦, and ␤ 3 without Gem (control) or with the indicated Gem mutants. *, p Ͻ 0.05; ***, p Ͻ 0.01 (compared with Gem_10G). dently and that each site functioning alone is sufficient to produce full-fledged inhibition.
There are at least four possible explanations for the complete lack of inhibition by Gem_10G_CM4. (i) It cannot fold properly. (ii) It expresses poorly in oocytes. (iii) It cannot interact with Ca v 2.1 or Ca v ␤. (iv) It no longer possesses a functional inhibitory site. To test the first three possibilities, we examined Gem_10G_CM4 expression in oocytes and its interaction with Ca v 2.1 or ␤ 3 by co-immunoprecipitation in HEK 293T cells. Gem_10G_CM4 protein was expressed abundantly in oocytes (Fig. 7E, middle and right panels), and as shown in Fig. 8, Gem_10G_CM4 was able to associate with Ca v 2.1 (Fig. 8A) or ␤ 3 (Fig. 8B). This result suggests that the first three scenarios are less likely and that Gem_10G_CM4 may not contain a functional inhibitory site.

DISCUSSION
In our previous study we proposed a "Ca v ␤-priming" model for Gem inhibition of surface P/Q-type Ca 2ϩ channels (18). This model postulates that Gem associates directly and independently of Ca v ␤ with Ca v 2.1 through an anchoring site on both Gem and Ca v 2.1 and that Ca v ␤ induces the formation of an inhibitory site in Ca v 2.1 where a corresponding inhibitory site in Gem binds to cause inhibition. In this study, through systematic deletions and targeted point mutations, we identify two putative inhibitory sites in Gem (key constructs, their functional effects, and the functional states of the two putative inhibitory sites are summarized in supplemental Fig. 1). One site is formed by a 12-amino acid region (Lys-265-Lys-276) in the C terminus, as evidenced by its ability to directly, acutely, specifically, and in a Ca v ␤-dependent manner inhibit surface P/Q-type channels on its own as a peptide (Fig. 3). That this 12-amino acid region can inhibit P/Q channels on its own is in line with the report that a Gem fragment containing the last ϳ80 residues of the C terminus was sufficient to inhibit P/Q channels (19). Eight of the 12 residues in this 12-amino acid region are fully conserved or highly similar among RGK proteins (Fig. 7A), and this region overlaps with a 13-amino acid region in Rem (corresponding to Phe-271-Arg-283 of human Rem) that was found to be critical for Rem inhibition of L-type channels (23). Thus, it is highly likely that this region also forms an inhibitory site in other RGK proteins.
The Lys-265-Lys-276 region falls within the polybasic motif that has been postulated to be involved in membrane targeting, which has been shown to be critical for Ca 2ϩ channel inhibition by Rem and Rem2 (15,20,23). However, the observation that Gem_10G and Gem_11A, in which all the positively charged residues were mutated to either glycine or alanine, were still capable of inhibiting P/Q-type channels (Fig. 5) indicate that either intrinsic membrane targeting is not essential for Gem inhibition or another membrane targeting signal is present elsewhere in Gem.
How the Lys-265-Lys-276 region produces Ca 2ϩ channel inhibition remains to be elucidated. It is notable that the inhibition of surface P/Q channels by the 12-amino acid peptide GCP1 is dependent on Ca v ␤, just as is the inhibition by the purified Gem-(Ser-68 -Lys-276) protein (Fig. 3E). This observation indicates that the Lys-265-Lys-276 region is one of the Gem inhibitory sites that binds to the Ca v ␤-induced inhibitory pocket in Ca v 2.1. Because this region contains five positively charged amino acids and because a scrambled peptide (GCP1_S) with the same number of positive charges was still effective in inhibiting P/Q channels either on its own (Fig. 3F) or when linked to Gem-(1-264), an inactive Gem deletion mutant (Fig. 3G), it is likely that the Lys-265-Lys-276 region exerts its inhibitory effect mainly through charge-charge interactions with P/Q channels rather than sequence-specific interactions. In the context of full-length Gem, specificity of the Lys-265-Lys-276 region is probably conferred by specific Gem/ Ca v 2.1 interactions through the anchoring site on both proteins.
The finding that Gem_10G and Gem_10A, in which the Lys-265-Lys-276 inhibitory site was completely disrupted, was able to fully inhibit P/Q channels (Fig. 5) indicates that the core region of Gem contains another inhibitory site. The existence of an inhibitory site in the core region of other RGK proteins is indicated by previous findings that Rem and Rem2 truncation mutants missing most of the C terminus (including the equivalent Lys-265-Lys-276 inhibitory site) can inhibit Ca 2ϩ channels when they are fused to an exogenous membrane-targeting sequence (15,20,23). Our results suggest, but do not prove, that in Gem residues Leu-241/Arg-242/Arg-243 may contribute to form this inhibitory site (Fig. 7). These residues are pivoted at the C-terminal end of an ␣-helix, with their side chains exposed and ample room to possibly engage in interactions with another protein, as revealed by the crystal structure of a GDP-bound Gem protein fragment (Met-73-Arg-264) containing most of the core (26) (Fig. 9). These residues are completely conserved among RGK proteins but are substituted by other residues in Ras that do not directly regulate HVA Ca 2ϩ channels. Together, these observations suggest that residues Leu-241/Arg-242/Arg-243 form a candidate inhibitory site not only in Gem but also in other RGK proteins. However, more direct evidence is needed to firmly establish this conclusion.
Our results further show that the core inhibitory site requires the distal C terminus to function properly. Deleting the last 32 C-terminal residues (including Lys-265-Lys-276) rendered this site totally ineffective (Fig. 1D), and mutating various clusters of residues among the last 20 amino acids attenuated its potency (Fig. 6). Furthermore, although the core inhibitory site was capable of producing full inhibition in Gem_10G, this site became completely ineffective in  Gem_10G⌬C, which lacks the last 20 amino acids but was expressed strongly in oocytes (Fig. 10). These results suggest that the last 20 amino acids are necessary for the proper function of the core inhibitory site. How this distal C-terminal region modulates the function of the core inhibitory site remains to be elucidated. The distance between this region and the core inhibitory site seems to be an important factor, as linking the last 20 amino acids to the inactive Gem-(1-264) did not restore inhibition (Fig. 3G, left panel), although this mutant construct (Gem-(1-264)ϩGCP2) expressed robustly in oocytes (Fig. 3G, middle and right panels).
The notion that RGK proteins contain more than one inhibitory site is in harmony with a report that Rem inhibits L-type Ca 2ϩ channels through three distinct mechanisms, including reducing the number of surface channels, decreasing channel open probability, and immobilizing channel voltage sensors (13). It remains to be investigated how different inhibitory sites are functionally coordinated and regulated.
Various regions of RGK proteins have been found to play a role in the inhibition of HVA Ca 2ϩ channels (15,19,21,23,24). The N and C termini of RGK proteins contain binding sites for CaM and 14-3-3, and interactions with CaM and 14-3-3 have been shown to regulate the subcellular distribution of RGK proteins (7-9, 28 -31). In Gem, these sites include Trp-269 (for CaM) and Ser-22 and Ser-288 (for 14-3-3) (corresponding to Trp-270, Ser-23, and Trp-289 in human Gem) (9,14,30,33). A Gem mutant, W269G (Trp-270 in our studies), which is defective in binding CaM (33), has been reported to weaken Gem regulation of Ca 2ϩ channels (9,14). Overexpression of CaM in tsA201 cells attenuates Rem inhibition of Ca v 1.2 currents (34). However, we found that Gem_10G, whose mutations include W270G and are much more severe than W269G, was fully capable of inhibiting P/Q channels (Fig. 5B), indicating that CaM binding to Gem is not required for Gem inhibitory effect. A possible explanation for the discrepancy lies in the observation that the magnitude of RGK regulation of Ca 2ϩ channels depends on the concentration of the RGK protein (35). Our observation that Gem-(Ser-68 -Lys-276) , which lacks both the N-and C-terminal binding sites for 14-3-3, was fully active in inhibiting P/Q channels (Figs. 1 and 2) indicates that interac-tion with 14-3-3 is not required for Gem inhibition, in agreement with earlier studies (9,14).
The Lys-265-Lys-276 region and the Leu-241/Arg-242/ Arg-243 tri-amino acid motif do not appear to be involved in the anchoring interaction between Gem and Ca v 2.1, as Gem_10G_CM4, which contains mutations in both sites, was still able to associate with Ca v 2.1 in the absence of Ca v ␤ (Fig. 8A). Thus, the anchoring site in Gem remains to be identified. Also remaining to be found are the anchoring site and the Ca v ␤-induced inhibitory site in Ca v 2.1.