G Protein Binding Sites on Calnuc (Nucleobindin 1) and NUCB2 (Nucleobindin 2) Define a New Class of Gαi-regulatory Motifs*

Heterotrimeric G proteins are molecular switches modulated by families of structurally and functionally related regulators. GIV (Gα-interacting vesicle-associated protein) is the first non-receptor guanine nucleotide exchange factor (GEF) that activates Gαi subunits via a defined, evolutionarily conserved motif. Here we found that Calnuc and NUCB2, two highly homologous calcium-binding proteins, share a common motif with GIV for Gαi binding and activation. Bioinformatics searches and structural analysis revealed that Calnuc and NUCB2 possess an evolutionarily conserved motif with sequence and structural similarity to the GEF sequence of GIV. Using in vitro pulldown and competition assays, we demonstrate that this motif binds preferentially to the inactive conformation of Gαi1 and Gαi3 over other Gα subunits and, like GIV, docks onto the α3/switch II cleft. Calnuc binding was impaired when Lys-248 in the α3 helix of Gαi3 was replaced with M, the corresponding residue in Gαo, which does not bind to Calnuc. Moreover, mutation of hydrophobic residues in the conserved motif predicted to dock on the α3/switch II cleft of Gαi3 impaired the ability of Calnuc and NUCB2 to bind and activate Gαi3 in vitro. We also provide evidence that calcium binding to Calnuc and NUCB2 abolishes their interaction with Gαi3 in vitro and in cells, probably by inducing a conformational change that renders the Gαi-binding residues inaccessible. Taken together, our results identify a new type of Gαi-regulatory motif named the GBA motif (for Gα-binding and -activating motif), which is conserved across different proteins throughout evolution. These findings provide the structural basis for the properties of Calnuc and NUCB2 binding to Gα subunits and its regulation by calcium ions.

subunits of heterotrimeric G proteins is controlled by accessory proteins that regulate their activity and/or localization (1)(2)(3). The first group of such regulators to be described was the regulators of G protein signaling (RGS) 4 protein family, which serve as GTPase-activating proteins for G␣ i , G␣ q , and G␣ 12 subunits via a 120-aa conserved domain, the "RGS box" (4,5). Subsequent studies revealed another group of regulatory proteins with GDI activity for G␣ i subunits, which have a common signature motif, i.e. the GoLoco or G protein-regulatory (GPR) motif (1,6,7). Both the RGS box (8,9) and the GoLoco/GPR motif (10) have been structurally resolved by X-ray chrytallography, and their critical roles in metabolism, cell division, and cardiovascular function, among others, have made them emerging pharmacological targets (11,12). We recently described another G␣-interacting protein, GIV (13), and showed that it is a GEF that activates G␣ subunits and mediates its biological functions via a defined motif (14) with structural similarity to the synthetic GEF peptide KB-752 (15). GIV is a metastasis-related protein (16) that enhances PI3K-Akt signaling and promotes macrophage, endothelial, epithelial, and tumor cell migration (17)(18)(19).
We identified Calnuc (nucleobindin 1 or NUCB1) as a G␣-binding protein in a yeast two-hybrid screen using G␣ i3 as bait (21). Calnuc, the most abundant protein in the Golgi (24) and the major calcium-binding protein within the Golgi lumen (21)(22)(23), regulates intracellular calcium stores via its two EFhands (22). In addition, we have shown previously that there is a significant soluble pool of Calnuc in the cytosol (21,25), which interacts with G␣ i3 in vivo on the surface of Golgi membranes as demonstrated by FRET and live cell imaging (26). The role of cytosolic Calnuc as a G protein regulator was further substantiated by the finding that it controls the intracellular localization of G␣ i subunits in neuroendocrine cells (27). However, the mechanism by which Calnuc binds or regulates G␣ i subunits remains unknown. Here we identified a conserved motif in Calnuc and the highly homologous protein NUCB2 (nucleobindin 2 or NEFA) (20) with similarity to the GEF motif of GIV and characterized how this motif binds and regulates G␣ i subunits. These findings help define a new class of structurally defined G protein regulatory motifs and provide insights into how the interaction between G␣ i3 and Calnuc is regulated.

EXPERIMENTAL PROCEDURES
Reagents and Antibodies-The peptide corresponding to the Calnuc G␣ i -binding motif, i.e. 309 RLVTLEEFLASTQRKE 324 (Calnuc-(309 -324) peptide, Ͼ95% purity), was synthesized and purified as described (28). The control peptide (EVVTLQQALEESNKLT, Ͼ95% purity) used in our experiments, corresponding to the GEF sequence of GIV with a F1685A mutation that abolishes binding and GEF activity (14), was custom-made by Genway Biotech (San Diego). NUCB2 cDNA was obtained from Open Biosystems. The sources of the remainder of the reagents and antibodies used were described previously (14,17,29).
Immunoblotting-Proteins samples were separated on 10% SDS-PAGE and transferred to PVDF membranes (Millipore, Billerica, MA). In experiments using His-Calnuc, His-Calnuc⌬N, or His-NUCB2 all electrophoretic steps were performed in the presence of 4 M urea, which increased the sensitivity of the immunodetection. Membranes were blocked with PBS supplemented with 5% nonfat milk before sequential incubation with primary and secondary antibodies. Infrared imaging was performed using an Odyssey imaging system (LI-COR Biosciences, Lincoln, NE). Primary antibodies were diluted as follows: anti-His, 1:2000; anti-G␣ i3 , 1:300, anti-G␣ o , 1:500; anti-G␣ s , 1:250.
Steady-state GTPase Assay-This assay was performed as described previously (14,29). Briefly, His-G␣ i3 (100 nM) was preincubated with different concentrations of His-Calnuc⌬N-(171-459) or GST-NUCB2 (173-333) for 15-30 min at 30°C in assay buffer (20 mM Na-HEPES, pH 8, 100 mM NaCl, 1 mM EDTA, 2 mM MgCl 2 , 1 mM DTT, and 0.05% (w:v) C12E10). His-Calnuc⌬N and GST-NUCB2-(173-333) were used instead of full-length His-Calnuc or GST-NUCB2 because the protein concentrations used in these experiments were achievable for only the truncated proteins, which express at higher yields in bacteria. GTPase reactions were initiated at 30°C by adding an equal volume of assay buffer containing 1 M [␥-32 P]GTP (ϳ50 cpm/fmol). Duplicate aliquots (50 l) were removed at different time points, and reactions were stopped with 950 l of ice-cold 5% (w/v) activated charcoal in 20 mM H 3 PO 4 , pH 3. Samples were then centrifuged for 10 min at 10,000 ϫ g, and 500 l of the resultant supernatant was scintillation-counted to quantify released [ 32 P]P i . To determine the specific P i produced, the background [ 32 P]P i detected at 10 min in the absence of G protein was subtracted from each reaction.
Reactions were initiated at 30°C by adding an equal volume of assay buffer containing 1 M [ 35 S] GTP␥S (ϳ50 cpm/fmol). Duplicate aliquots (25 l) were removed at different time points, and binding of radioactive nucleotide was stopped by the addition of 3 ml of ice-cold wash buffer (20 mm Tris-HCl, pH 8.0, 100 mm NaCl, and 25 mm MgCl 2 ). The quenched reactions were rapidly passed through BA-85 nitrocellulose filters (GE Healthcare) and washed with 4 ml of wash buffer. Filters were dried and subjected to liquid scintillation counting. Experiments designed to study the effect of Ca 2ϩ were performed as described above except that no EDTA was used and different concentrations of CaCl 2 were added.
Preparation of Detergent-soluble Extracts from Rat Brain Membranes-Isolation of rat brain membranes was adapted from a fractionation procedure described previously for liver (32). Briefly, rat brains were homogenized in 10 mM HEPES-KOH, pH 7.4, 5 mM EDTA, and 0.5 M sucrose with a Teflonglass homogenizer and spun down at 1,000 ϫ g for 10 min to sediment unbroken tissue and nuclei, and the resulting supernatant (postnuclear supernatant) was collected. Crude membranes were sedimented from the postnuclear supernatant by centrifugation at 100,000 ϫ g, aliquoted, and stored at Ϫ80°C. Rat brain membrane lysates were freshly prepared prior to the pulldown experiments presented in Fig. 4 by solubilizing ϳ750 g of protein/condition in binding buffer (50 mM Tris-HCl, pH 7.4, 100 mM NaCl, 0.4% (v:v) Nonidet P-40, 10 mM MgCl 2 , 5 mM EDTA, 2 mM DTT, and protease inhibitor mixture) for ϳ2 h at 4°C. Lysates were cleared (14,000 ϫ g for 10 min) before use in subsequent experiments.
Other Methods-Protein structure analysis and visualization was performed using ICM-Browser-Pro software (Molsoft Inc., San Diego). Data presented in Figs. 2C, 7C, and 8D and supplemental Fig. S4 were curve-fitted by nonlinear regression using Prism 4.0. (San Diego) to determine the K d , EC 50 , and IC 50 values.

Identification of a Putative G␣-interacting Motif in Calnuc
and NUCB2-By manual examination of the Calnuc sequence, we noticed a region with significant similarity to the GEF motif of GIV and two synthetic peptides with GEF activity, KB-752 (15) and GSP (33) (Fig. 1A). This motif (aa 309 -324) overlaps with the C-terminal part of the second EF-hand of Calnuc and is also found in NUCB2, a protein highly homologous to Calnuc (62% aa sequence identity). Phylogenetic analysis revealed that this motif is evolutionarily conserved in both Calnuc and NUCB2 proteins from invertebrates to humans (supplemental Fig. S1). In addition, the structure of this motif extracted from the NMR coordinates of the calcium-binding domain of Calnuc (34) is very similar to the established crystal structure of the FIGURE 1. Identification of a putative G␣-binding motif in Calnuc and NUCB2. A, a sequence at the end of the Calnuc (aa 309 -324) and NUCB2 (aa 311-326) second EF-hand shows a similarity to the GEF motif of GIV (aa 1678 -1693) and the synthetic KB-752 and GSP peptides. Rat Calnuc, rat NUCB2, and human GIV sequences were obtained from the NCBI and aligned with the KB-752 (15) and GSP (33) sequences using ClustalW. Conserved identical residues are shaded in black and similar residues in gray. A consensus sequence of 7 aa based on this alignment and the phylogenetic analysis of Calnuc (Fig. S1) and GIV (14) is indicated. B, the Calnuc putative G␣ i -binding motif has a structural similarity to the KB-752 peptide and the GEF motif of GIV. The coordinates of the Calnuc putative G␣-binding sequence (aa 309 -320 (red)) were extracted from the Protein Data Bank (ID code: 1SNL) and the coordinates for the GEF motif of GIV (aa 1678 -1689 (green)) from a previously described homology model (14). Both were threaded over the structure of KB-752 (yellow) in complex with G␣ i1 (blue) (Protein Data Bank ID code: 1Y3A) using ICM-Browser-Pro. The Calnuc putative G␣ i -binding motif, the GEF motif of GIV, and KB-752 form a helix in which the side chains of hydrophobic residues corresponding to positions 3, 6, and 7 from the consensus sequence depicted in A dock onto the ␣3/SwII hydrophobic cleft of the G␣ i subunits (cyan surface).
KB-752 peptide bound to G␣ i1 (15) and the homology-based structural prediction for the GEF motif of GIV bound to G␣ i3 (14) (Fig. 1B). These observations indicate that Calnuc and NUCB2 possess an evolutionarily conserved motif with structural similarity to the GEF motif of GIV and GIV-related peptides that display GEF activity.
Calnuc and NUCB2 Bind Preferentially to Inactive G␣ i3 -Based on the sequence and structural similarities described above, we reasoned that Calnuc and NUCB2 might have similar G␣ binding properties to those of the GEF motif of GIV and related peptides that bind specifically to inactive GDP-bound G␣ i subunits. We found that this was indeed the case for both Calnuc ( Fig. 2A) and NUCB2 (Fig. 2B). GST-Calnuc bound robustly to inactive, GDP-loaded His-G␣ i3 but not to the G protein when it was activated by either GDP⅐AlF 4 Ϫ or GTP␥S loading ( Fig. 2A). Similar results (data not shown) were obtained in pulldown assays using GST-G␣ i3 and His-tagged Calnuc or Calnuc⌬N-(171-459), an N-terminally truncated construct containing the putative G␣-binding motif. Similarly, His-NUCB2 showed robust binding to GST-G␣ i3 in the presence of GDP but not in the presence of GDP⅐AlF 4 Ϫ (Fig. 2B). The interaction of GST-Calnuc and GST-NUCB2-(173-333) (containing the putative G␣-binding motif), with GDP-loaded His-G␣ i3 , has a dissociation constant (K d ) of 3.7 Ϯ 1.2 and 1.0 Ϯ 0.3 M, respectively (Fig. 2C). These results indicate that, much like GIV and the GIV-related peptides, binding of Calnuc and NUCB2 to G␣ i3 is state-dependent with a marked preference for the inactive conformation.
Calnuc and NUCB2 Bind to the ␣3/Switch II Cleft on G␣ i ⅐GDP-Next we performed competition assays to determine whether Calnuc and NUCB2 shared a common binding site on G␣ i3 with the synthetic KB-752 peptide and GIV, which bind to the hydrophobic cleft circumscribed by the ␣3 helix and "switch II" (SwII) (14,15). We found that increasing concentrations of KB-752, but not a control peptide (see "Experimental Procedures"), decreased His-G␣ i3 binding to GST-Calnuc (Fig.  3A). Similarly, increasing concentrations of a peptide (aa 309 -324) corresponding to the putative G␣ i binding sequence of Calnuc, but not a control peptide, decreased the amount of His-GIV-CTs (aa 1660 -1870, containing the GEF motif of GIV) that bound to GST-G␣ i3 (Fig. 3B). We also performed similar competition assays with GST-NUCB2-(173-333) and found that it also competed with the KB-752 peptide (Fig. 3C) or His-GIV-CTs ( Fig. 3D) but not with their respective controls for binding to G␣ i3 . Taken together these results suggest that Calnuc and NUCB2 bind specifically to the ␣3/SwII cleft of G␣ i3 ⅐GDP via the newly identified motif.
Characterization of Calnuc Specificity for G␣ Subunits-We have previously shown that Calnuc can interact with G␣ i , G␣ o , and G␣ s but not G␣ 12/13 or G␣ q in yeast two-hybrid assays (21). Next we investigated the relative strength of the interaction of Calnuc with G␣ i , G␣ o , and G␣ s using in vitro protein-protein binding assays. We found that GST-Calnuc bound GDP-loaded G␣ i3 from rat brain membrane lysates, whereas binding of G␣ o was very weak and binding of G␣ s was undetectable (Fig. 4A). We further performed pulldown assays using purified GST-G␣ i1 , GST-G␣ i2 , and GST-G␣ i3 and His-Calnuc⌬N to investigate whether Calnuc binds to other G␣ i subunits. We used Calnuc⌬N instead of full-length Calnuc because initial experiments indicated that they shared virtually identical G␣ i binding properties (data not shown). We found that His-Calnuc⌬N binds strongly to GDP-loaded GST-G␣ i1 and GST-G␣ i3 but showed significantly less binding to G␣ i2 (Fig. 4B). These results indicate that the binding preference of Calnuc for G␣ subunits His-G␣ i3 binding was determined by quantitative immunoblotting using an Odyssey infrared imaging system, and data were fitted to a nonlinear, one-site binding hyperbola (solid lines) using Prism 4.0.

Lys-248 in G␣ i3
Determines the Preferential Binding of Calnuc to G␣ i3 versus G␣ o -Recently we found that a single residue differing between the G␣ i and G␣ o subunits, i.e. Trp-258 in G␣ i3 and Phe-259 in G␣ o (Fig. 5A), accounts for the preferential binding of GIV to G␣ i versus G␣ o (29). To determine whether this is the case for Calnuc, we investigated the effect of mutating Trp-258 in G␣ i3 and Phe-259 in G␣ o on the binding of these G proteins to Calnuc. GST-Calnuc bound robustly to purified wild-type His-G␣ i3 but not to purified wild-type His-G␣ o (Fig.  5B); this striking preference remained unchanged when Trp-258 in G␣ i3 was mutated to Phe or when Phe-259 in G␣ o was mutated to Trp (Fig. 5B). This indicates that Trp-258 in G␣ i is not responsible for the preferential binding of Calnuc to G␣ i versus G␣ o .
We reasoned that Lys-248 in G␣ i could be responsible for the preferential binding to G␣ i versus G␣ o because it is the only amino acid that is not conserved between the two G␣ subunits in the Calnuc binding site (Fig. 5A), i.e. the ␣3/SwII cleft (Fig. 3). Mutation of Lys-248 in GST-G␣ i3 to Met, the corresponding residue in G␣ o , dramatically decreased His-Calnuc⌬N binding (Fig. 5C). Importantly, GST-G␣ i3 K248M did bind to GIV, AGS3 (activator of G protein signaling 3), and G␤␥ (Ref. 29 and data not shown), indicating that this mutation specifically affects Calnuc binding. Conversely, when Met-249 in His-G␣ o was mutated to Lys, it enhanced its binding to GST-Calnuc compared with wild-type His-G␣ o (Fig. 5D). Furthermore, structural analysis (supplemental Fig. S2A) revealed that the positively charged Lys-248 of G␣ i3 might establish an electrostatic interaction with negatively charged Glu-314 of Calnuc. We hypothesized that inverting the charge of the G␣ i3 Lys-248 alone would impair the G␣ i3 -Calnuc interaction by electrostatic repulsion of the Calnuc Glu-314 but that inverting the charges of these two residues simultaneously would restore the interaction by establishing an electrostatic attraction analogous to that found in the native situation. In fact, inversion of the charge of the G␣ i3 Lys-248 by mutation to Glu abolished G␣ i3 binding to wild-type Calnuc, and binding was restored if the charge of Calnuc Glu-314 was simultaneously inverted by mutation to Lys (supplemental Fig. S2B), suggesting that the electrostatic interaction between G␣ i3 aa 248 and Calnuc aa 314 stabilizes G␣ i3 -Calnuc binding. These results demonstrate that although GIV and Calnuc have an overlapping binding site on G␣ subunits, i.e. the ␣3/SwII cleft, and display preference for G␣ i versus G␣ o , the critical residues in the G␣ subunit that determine binding specificity are different.
Identification of Residues in Calnuc Required for Binding and Regulating G␣ i -Calnuc, NUCB2, GIV and GIV-related peptides share hydrophobic residues in positions 3, 6 and 7 of the consensus sequence depicted in Fig. 1A. In the structure of the KB-752⅐G␣ i1 complex these residues are packed against the hydrophobic cleft formed by the ␣3 helix and switch II to stabilize the interaction (Ref. 15 and Fig. 1B). We reasoned that residues in the same position might also be required for Calnuc and NUCB2 to bind G␣ i3 . We found that mutation of each of the corresponding residues in Calnuc, i.e. Leu-313, Phe-316, and Leu-317 to Ala dramatically reduced His-G␣ i3 binding to GST-Calnuc (Fig. 6A). The double mutant L313A/L317A decreased the interaction even further than the single mutations (Fig. 6A, see high exposure blot). Similar findings were obtained for NUCB2 (Fig. 6B), indicating that these hydrophobic residues are required for both Calnuc and NUCB2 to interact with G␣ i3 . In addition, mutation of G␣ i3 Trp-211 or Phe-215 in the predicted binding site for Calnuc (supplemental Fig.  S3A) also disrupted the interaction (supplemental Fig. S3B), suggesting that they mediate a hydrophobic interaction with the Calnuc Leu-313, Phe-316, and Leu-317. These results indicate that the interaction of Calnuc and NUCB2 with G␣ i3 requires the hydrophobic residues found in their conserved motif shared with GIV and GIV-related peptides.
We next investigated the effect of Calnuc and NUCB2 on G protein activation. For this we measured the steady-state GTPase activity of His-G␣ i3 (which depends directly on the rate of nucleotide exchange (29,35)) in the presence of wild-type His-Calnuc⌬N or His-Calnuc⌬N L313A/L317A, which has dramatically impaired binding to G␣ i3 (Fig. 6A) (negative control). His-Calnuc⌬N was used instead of full-length His-Calnuc because the protein concentrations used in these experiments were achievable only for the truncated protein, which expresses at higher yields in bacteria (see "Experimental Procedures"). Wild-type Calnuc⌬N but not Calnuc⌬N L313A/L317A increased the rate of steady-state GTP hydrolysis (Fig. 7A). Similarly, wild-type GST-NUCB2 but not its binding-deficient mutant, L315A/L319A, also increased the steady-state GTPase activity of G␣ i3 (Fig. 7B). Other mutants of the G␣-binding motif such as Calnuc F316A and NUCB2 F318A also impaired G␣ i3 activation when compared with their respective wild-type controls (supplemental Fig. S4), suggesting that Calnuc and NUCB2 are GEFs for G␣ i3 and that they work via their conserved G␣-binding motif. Wild-type Calnuc⌬N and NUCB2 increased the GTPase activity of His-G␣ i3 1.74-Ϯ 0.18-fold (n ϭ 6) and 1.76-Ϯ 0.13-fold (n ϭ 3), respectively, at the maximal concentration tested (Fig. 7C). The EC 50 values were 7.3 Ϯ 1.0 M and 4.0 Ϯ 1.0 M (Fig. 7C), which are in good agreement with their respective K d values for G␣ i3 binding (Fig. 2C). To further validate the role of Calnuc as a GEF, we performed GTP␥S binding experiments, which are a direct measure of GDP for GTP exchange activity. Incubation of His-G␣ i3 in the presence of Calnuc⌬N increased the initial rate of GTP␥S bind-  4, 8, and 12). Solubilized proteins from 750 g of rat brain membranes were incubated with ϳ20 g of purified GST (lanes 2, 6, and 10) or GST-Calnuc (lanes 3, 4, 7, 8, 11, 12, and 13) immobilized on glutathione beads in the presence of GDP (30 M; lanes 2, 3, 6, 7, 10, and 11) or GDP and AlF 4 Ϫ (AlCl 3 , 30 M; NaF, 10 mM; lanes 4, 8, and 12). An additional control without rat brain membrane lysate was performed to validate G␣ s antibody specificity (lane 13). Input (lanes 1, 5, and 9), 10% of the membrane lysate. No binding of G␣ i3 , G␣ o , or G␣ s to the negative control GST was detected (lanes 2, 6, and 10). The arrows (lanes 1, 3, 5, and 9) denote the specific bands corresponding to the different G␣ subunits (including the long and short splice forms of G␣ s , lane 9), and the star (lanes 11, 12, and 13) 6), GST-G␣ i2 (lanes 4 and 7), or GST-G␣ i3 (lanes 5 and 8) preloaded with GDP (lanes 2-5) or GDP⅐AlF 4 Ϫ (lanes 6 -8), immobilized on glutathione beads, and analyzed as described for Fig. 2A. Input (lane 1), 1 g of His-Calnuc⌬N. ing 1.70-Ϯ 0.10-fold (n ϭ 4) compared with the G protein alone (Fig. 7D). Taken together, these results indicate that the G␣ ibinding motif in Calnuc and NUCB2 has GEF activity toward G␣ i3 .
Effect of Calcium Binding to Calnuc and NUCB2 on Their Interaction with G␣ i3 -The domain of Calnuc containing both EF-hands is known to be disorganized in the absence of Ca 2ϩ and to adopt a globular conformation upon binding of the divalent cation (34). Analysis of the structure of calcium-bound Calnuc revealed that the residues that interact with G␣ i3 , i.e. Leu-313, Phe-316, and Leu-317 (Fig. 6A), are also utilized to make an intramolecular contact in the calcium-bound conformation (Fig. 8A), suggesting that Calnuc would not be able to interact with G␣ i3 upon binding of Ca 2ϩ . We found this to be the case, because neither Calnuc nor NUCB2 bound G␣ i3 in the presence of excess CaCl 2 (Fig. 8, B and C). No difference was observed when the experiment was performed in the presence of excess MgCl 2 or LiCl (data not shown), indicating that the observed effect is specific for Ca 2ϩ . We also investigated the effect of Ca 2ϩ on the GEF activity of Calnuc and found that increasing concentrations of Ca 2ϩ inhibited the increase in GTP␥S binding to G␣ i3 promoted by Calnuc with an IC 50 of 1.2 Ϯ 0.12 M (n ϭ 3) (Fig. 8D), a value similar to the previously reported K d of Calnuc for Ca 2ϩ . Finally, we investigated the effect of elevating intracellular Ca 2ϩ levels on the interaction between Calnuc and G␣ i3 in cultured cells. COS-7 cells were co-transfected with G␣ i3 -FLAG and a truncated Calnuc (⌬SS-Calnuc-CFP), lacking the signal sequence which is present pre-FIGURE 5. Lys-248 but not Trp-258 is responsible for preferential binding of Calnuc to G␣ i3 versus G␣ o . A, sequence alignment of G␣ o , G␣ i1 , G␣ i2 , and G␣ i3 indicating the G␣ i3 and G␣ o mutants studied. Rat G␣ o , G␣ i1 , G␣ i2 , and G␣ i3 sequences corresponding to the SwII and the ␣3 helix were obtained from the NCBI database and aligned using ClustalW. Conserved identical residues are shaded in black and similar residues in gray. The secondary structure elements (␣ ϭ ␣-helix, ␤ ϭ ␤-sheet) indicated below the alignment are named according to their crystal structures. Residues within this region that are conserved among G␣ i1 , G␣ i2 , and G␣ i3 but are different in G␣ o were mutated in G␣ i3 to the corresponding residues in G␣ o (indicated below with arrows) or in G␣ o to the corresponding residues in G␣ i3 (indicated above with arrows). B, wild-type His-G␣ i3 ⅐GDP (lane 1) and His-G␣ i3 ⅐GDP W258F (lane 5) but not wild-type His-G␣ o ⅐GDP (lane 3) or His-G␣ o ⅐GDP F259W (lane 7) bind to GST-Calnuc. No binding of any of the G␣ subunits loaded with GDP⅐AlF 4 Ϫ to GST-Calnuc was detected (lanes 2, 4, 6, and 8). 6 g of His-G␣ i3 (lanes 1 and 2), His-G␣ o (lanes 3 and 4), His-G␣ i3 W258F (lanes 5 and 6), or His-G␣ o F259W (lanes 7 and 8) preloaded with GDP (lanes 1, 3, 5, and 7) or GDP and AlF 4 Ϫ (lanes 2, 4, 6, and 8) was incubated with ϳ20 g of purified GST-Calnuc immobilized on glutathione beads and analyzed as described in the legend for Fig. 2A. C, His-Calnuc⌬N binding to GST-G␣ i3 K248M⅐GDP (G␣ i3 K m (lane 4)) is reduced ϳ80% compared with wild-type GST-G␣ i3 ⅐GDP (G␣ i3 WT, lane 3). No binding of His-Calnuc⌬N to GST (lane 2) or GDP⅐AlF 4 Ϫ -loaded G␣ i3 (lanes 5 and 6) is detected. 10 g of His-Calnuc⌬N was incubated with purified GST (lane 2), wild-type GST-G␣ i3 (lanes 3 and 5), or GST-G␣ i3 K248M (lanes 4 and 6) preloaded with GDP (lanes 2-4) or GDP⅐AlF 4 Ϫ (lanes 5 and 6) immobilized on glutathione beads and analyzed as described in the legend for Fig. 2A. Input (lane 1), 1 1-3). 6 g of each His-G␣ subunit preloaded with GDP was incubated with ϳ20 g of purified GST (lanes 1-3) or GST-Calnuc (lanes 4 -6) immobilized on glutathione beads and analyzed as in the legend for Fig. 2A. dominantly in the cytosol (26). Cells were stimulated with thapsigargin (which elevates the cytosolic levels of Ca 2ϩ by blocking the endoplasmic reticulum Ca 2ϩ pump) or ATP (which activates purinergic receptors at the cell surface) (22), and immunoprecipitation was carried out using anti-FLAG IgG followed by immunoblotting for Calnuc. We found that ⌬SS-Calnuc-CFP co-immunoprecipitated with G␣ i3 exclusively in nonstimulated cells, but it was virtually abolished after stimulation with either thapsigargin or ATP (Fig. 8E). Co-immunoprecipitation of G␤␥ with G␣ i3 -FLAG was not affected by thapsigargin or ATP, indicating that elevation of the intracellular levels of Ca 2ϩ specifically affects the interaction of G␣ i3 with Calnuc but not other G␣-binding proteins. These results suggest that elevation of intracellular Ca 2ϩ levels can efficiently disrupt the interaction between Calnuc and G␣ i3 in living cells, corroborating our observations in vitro. Taken together, these data indicate that calcium binding can promote conformational changes in Calnuc (and presumably also in NUCB2) that block its interaction with G␣ subunits in vitro and in living cells (Fig.  8F), subsequently inhibiting its GEF activity.

DISCUSSION
In this work we identify a new class of G protein-binding motif with defined structural features. This motif is found in two closely related proteins, Calnuc and NUCB2, and was previously found in another unrelated protein, GIV, and in the synthetic peptides KB-752 and GSP, shown previously to have GEF activity for G␣ i (15,33). It consists of a relatively disordered N-terminal region followed by an ␣-helix that docks onto the ␣3/SwII cleft of the G␣ i subunits only in the inactive conformation to enable GEF activity in vitro. We named this signature sequence the GBA motif (for G␣-binding and -activating motif). We propose that the conserved GBA motif found in native proteins is a signature structure that defines a new family of G protein regulators with GEF activity, in the same fashion as the GoLoco/GPR motif or the RGS box define families of pro-teins with GDI or GTPase-activating protein activity, respectively. An important observation is that the GBA motif in Calnuc is evolutionarily conserved across species from sponges to man (supplemental Fig. S1), and the Caenorhabditis elegans orthologues of Calnuc and the G␣ subunits have been shown to interact (36), suggesting that its function as a G␣-binding motif is also evolutionarily conserved. This evolutionary conservation suggests a selection imposed by a crucial biological function associated with the interaction with G␣ i . It is interesting that a similar consensus motif was found in two synthetic peptides, KB-752 and GSP, which were identified by two different in vitro approaches, phage display of random sequence peptides (15) and iterative optimization of in vitro mRNA-translated peptides (33). In both cases the selection is determined solely by the chemical properties of the peptides and not their biological function. These observations suggest that the sequences found in vivo in the GBA motif of Calnuc, NUCB2, and GIV have highly optimized chemical properties for G␣ i binding.
Based on the sequences of Calnuc, NUCB2, and GIV in different species (supplemental Fig. S1 and Ref. 14) and related synthetic peptides (15,33), we propose that the GBA motif can be defined by a conserved core sequence of seven amino acids (Fig. 1A), ⌿-[S/T]-[⌽/⌿]-X-[D/E]-F-⌿, in which ⌿ are aliphatic residues and ⌽ are aromatic residues. Residues in positions 3, 6, and 7 in this consensus motif, i.e. Leu-313, Phe-316, and Leu-317 in Calnuc, correspond to hydrophobic residues aligned on one side of the ␣-helical part of the motif, which are used to stabilize the interaction with G␣ i by packing against the ␣3/SwII hydrophobic cleft. Residues in positions 2 and 5 form a hydrogen bond in the structures of KB-752 and Calnuc, which is required for the motif to adopt its helical conformation. This design for molecular coupling resembles that observed for other signaling interfaces. For example, A-kinase-anchoring proteins (AKAPs) are characterized by a signature motif that forms an aliphatic helix that docks onto a hydrophobic pocket on the regulatory subunit of cAMP-dependent protein kinase (PKA) (37), and the N-terminal region of the GoLoco/GPR motif, also binds to the ␣3/SwII hydrophobic cleft of G␣ i subunits via an aliphatic helix (10).
Our results also provide the structural basis for the regulation of Calnuc and NUCB2 binding to and activation of G␣ i subunits by calcium. Calnuc is the major calcium-binding protein in the lumen of Golgi cisternae, where it regulates the intracellular calcium stores (21,22). On the other hand, there is a cytosolic pool of Calnuc that interacts directly with G␣ i3 (26). NUCB2 has been described as sharing a similar subcellular distribution (20). The regulation of the Calnuc/NUCB2-G␣ i interaction by Ca 2ϩ described in this work is not surprising considering that the G␣-binding motif on Calnuc overlaps with the EF-hands responsible for calcium binding. Our finding that G␣ i binding to Calnuc and NUCB2 is impaired by calcium binding is compatible with previous observations by NMR indicating that when calcium-bound, the Calnuc EF-hands fold into a globular domain that hides the G␣ i -binding residues, whereas in the absence of calcium this domain is disordered and probably exposes the G␣ i -binding motif (Fig. 8F). Thus, we propose that the conformational changes in Calnuc and NUCB2 that occur upon Ca 2ϩ binding regulate their interaction with the G protein and its subsequent activation. This mode of regulation by Ca 2ϩ is consistent with our results presented here indicating that Calnuc and G␣ i3 interact in the cytosol of cells under resting conditions (in accord with our own previous observation using FRET and live cell microscopy (26)) but not upon stimulation with thapsigargin or ATP (Fig. 8E). This is probably because in resting conditions the cytosolic concentration of free Ca 2ϩ (50 -100 nM) is significantly lower than the K d value of Calnuc for Ca 2ϩ binding (ϳ7 M, (22)), thereby allowing the interaction of calcium-free Calnuc with G␣ i3 , whereas upon stimulation with thapsigargin or ATP the intracellular levels of Ca 2ϩ are increased and calcium-bound Calnuc cannot interact with G␣ i3 . It will be important in the future to ask whether the regulation of the Calnuc-G␣ i3 interaction by Ca 2ϩ might influence the interplay between G protein-and calcium-dependent signaling, two major signaling events that regulate a multitude of cellular functions.
Our data also unveiled the structural basis for the state-dependent binding of Calnuc and NUCB2 to G␣ i subunits. Based on the structures of G␣ i1 and other G␣ subunits bound to GTP␥S and GDP-AlF 4 Ϫ (38, 39), the hydrophobic cleft circumscribed by the ␣3 helix and the switch II is occluded when the G protein adopts the active conformation, thereby hindering its interaction with Calnuc and NUCB2 by steric clashes. From this we concluded that conformational changes of G␣ i3 upon activation determine its interaction with Calnuc and NUCB2. We previously reported that Calnuc binds to a site different from the ␣3/SwII cleft (i.e. the ␣5 helix) of G␣ i3 by using C-terminal truncations of the G protein (25). Although we cannot rule out the presence of two binding sites for Calnuc on G␣ i subunits, one likely explanation for the previous results is that truncation of the G␣ i C terminus promotes constitutive activation of the G protein (40) which in light of the data presented here would abolish its interaction with Calnuc and NUCB2.
The studies presented here also provide insights into the specific features of the Calnuc-G␣ subunit interface and its differences from another GBA motif-containing protein, i.e. GIV. Both Calnuc and GIV bind preferentially to G␣ i subunits over G␣ o or G␣ s , but Calnuc shows preference for G␣ i1 and G␣ i3 over G␣ i2 , whereas GIV binds equally to G␣ i1 , G␣ i2 , and G␣ i3 in vitro (14). The basis for the preference of Calnuc for G␣ i1 and G␣ i3 over G␣ i2 is still unclear, because all G␣ i subunits have identical residues in the switch II and the ␣3 helix (Fig. 5A), which based on our results presented here is a major binding site for Calnuc on the G protein. It is possible that other residues of the G protein outside of this major binding surface, i.e. the ␣3/SwII cleft, may also determine the specificity of binding by making additional contacts, as reported for other G protein regulators with preference for G␣ i1 and G␣ i3 over G␣ i2 such as ificity in the pharmacological targeting of these interfaces for therapeutic purposes, as proposed for GIV (14,29).
Our results suggest a role for Calnuc and NUCB2 as regulators of G protein activity. The affinity of Calnuc and NUCB2 for G␣ i3 (K d ϳ4 and ϳ1 M, respectively) is lower than that of GIV (K d ϳ300 nM, data not shown) but similar to the GEF peptide KB-752 (K d ϳ4 M (15)) and other G protein regulators such as the GDI proteins LGN (K d ϳ6 M (45)) and G18/AGS4 (K d ϳ2.5 M (46)). Like GIV and related synthetic peptides, Calnuc and NUCB2 possess GEF activity in vitro, indicating that this is a common feature associated with the conserved GBA motif. While this manuscript was in preparation, Kapoor et al. (47) reported that Calnuc possesses GDI activity toward G␣ i1 . Although the reason for the discrepancy between their work and ours is not clear, one possible explanation is the different experimental conditions used; specifically, Kapoor et al. (47) used ϳ200 -400-fold higher concentrations of G protein and nucleotide in their G protein activity assays than those used here and in most previous studies of this type (6,7,10,14,15,29,33,40,48) which might affect the enzymatic properties of the reaction studied. In addition, some of the evidence presented by Kapoor et al. (47) is based on spectroscopic studies with fluorescent nucleotide analogs, and this type of analysis has been reported to generate artifactual readouts (48) for peptides such as Calnuc that bind to the ␣3/SwII cleft (Fig. 3). By using the G␣ i binding-deficient mutants L313A/L317A and F316A of Calnuc as a negative controls (Figs. 6 and S4A), we demonstrated that the observed increase in G protein activation can be attributed specifically to the GBA motif. The Calnuc interaction with and activation of G␣ i3 occurs at relatively high concentrations in vitro, and its GEF activity is weaker than that observed for GIV or G protein-coupled receptors. However, our previously published data demonstrate that the interaction occurs in vivo because cytosolic Calnuc and G␣ i3 bind to each other in living cells as determined by FRET and live cell imaging (26) and that Calnuc regulates the subcellular localization of G␣ i3 (27), suggesting a functional role for the Calnuc-G␣ i coupling in the physiological setting. Further investigations will be required to elucidate whether G␣ i3 activation by Calnuc occurs in vivo and to determine the functional consequences of the interaction.