Nucleobindin 1 Is a Calcium-regulated Guanine Nucleotide Dissociation Inhibitor of Gαi1*

Nucleobindin 1 (NUCB1) is a widely expressed multidomain calcium-binding protein whose precise physiological and biochemical functions are not well understood. We engineered and heterologously expressed a soluble form of NUCB1 (sNUCB1) and characterized its biophysical and biochemical properties. We show that sNUCB1 exists as a dimer in solution and that each monomer binds two divalent calcium cations. Calcium binding causes conformational changes in sNUCB1 as judged by circular dichroism and fluorescence spectroscopy experiments. Earlier reports suggested that NUCB1 might interact with heterotrimeric G protein α subunits. We show that dimeric calcium-free sNUCB1 binds to expressed Gαi1 and that calcium binding inhibits the interaction. The binding of sNUCB1 to Gαi1 inhibits its basal rate of GDP release and slows its rate and extent of GTPγS uptake. Additionally, our tissue culture experiments show that sNUCB1 prevents receptor-mediated Gαi-dependent inhibition of adenylyl cyclase. Thus, we conclude that sNUCB1 is a calcium-dependent guanine nucleotide dissociation inhibitor (GDI) for Gαi1. To our knowledge, sNUCB1 is the first example of a calcium-dependent GDI for heterotrimeric G proteins. We also show that the mechanism of GDI activity of sNUCB1 is unique and does not arise from the consensus GoLoco motif found in RGS proteins. We propose that cytoplasmic NUCB1 might function to regulate heterotrimeric G protein trafficking and G protein-coupled receptor-mediated signal transduction pathways.

Heterotrimeric guanine nucleotide-binding proteins, G proteins, couple to heptahelical cell surface G protein-coupled receptors (GPCRs) 3 and participate in intracellular signaling events. The G protein heterotrimer is composed of the G␣ subunit and the G␤␥ heterodimer. Upon ligand-mediated activation, GPCRs catalyze the exchange of GDP for GTP on G␣ leading to dissociation of the heterotrimer into G␣⅐GTP and G␤␥ subunits (1)(2)(3). These individual subunits then regulate downstream signaling cascades involving effector systems like adenylyl cyclases, Ca 2ϩ and K ϩ channels, phospholipase C isozymes, and cyclic nucleotide phosphodiesterases (4,5). Thereafter, the intrinsic GTPase activity of G␣ reverts it back to the GDP-bound state, which can reassociate with G␤␥. This inhibits the interaction of G protein subunits with downstream effectors, which results in the turning-off of the signaling pathways. Hence, signaling by heterotrimeric G proteins is directly dependent on the lifetime of the GTP-bound state of G␣. This lifetime is regulated by GTPase-accelerating proteins (GAPs), which catalyze the rapid hydrolysis of the G␣-bound GTP to GDP and by guanine nucleotide dissociation inhibitors (GDIs), which inhibit the exchange of GDP for GTP in the catalytic pocket of G␣ (6).
Together, GAPs and GDIs exert a regulatory control on G protein signaling. In recent years, novel interacting partners of heterotrimeric G proteins called the regulators of G protein signaling or RGS proteins have been discovered that possess GAP activity (7,8). Interestingly, RGS12 and RGS14 in addition to functioning as GAPs can also act as GDIs (9,10). In 1999, Lanier and co-workers (11) used yeast two-hybrid analysis to identify distinct receptor-independent activators of G protein signaling or AGS proteins. Several members of this AGS family (AGS3-6) have been shown to function as GDIs of G␣ i subunits (12). The observed GDI activity of both AGS and RGS proteins toward G␣ i/o is attributed to a 19-amino acid consensus sequence called the GoLoco motif (9,13,14). Earlier in 1995, in an independent yeast two-hybrid screen, Mochizuki et al. (15) established the interaction of a novel Golgi-resident Ca 2ϩbinding protein Nucleobindin 1 or NUCB1, specifically with the heterotrimeric G protein ␣ subunit, G␣ i2 . Subsequently, Lin et al. (16) demonstrated that NUCB1 interacts exclusively with the adenylyl cyclase inhibitory (G␣ i ) and stimulatory (G␣ s ) classes of G␣ subunits. In a recent study, it was shown that overexpression of NUCB1 causes redistribution of only the G␣ i1 subunits and not the G␤␥ subunits to the plasma membrane and regulated secretion granules (17). However, a role for NUCB1 in modulating G␣ i1 activation and the biochemistry of NUCB1-G␣ i1 interaction has not been reported.
NUCB1 is a 55-kDa multidomain Ca 2ϩ -binding protein that was first identified as a novel B cell growth and differentiation factor associated with lupus syndrome (18). NUCB1 derives its trivial name, Calnuc, from its Ca 2ϩ -binding and DNA-binding ability (19). The DNA-binding domain of basic residues (172-218) lies at the N terminus following the signal sequence. The Ca 2ϩ -binding domain is at the core of the protein sequence consisting of two EF hand motifs with an intervening acidic region (residues 253-316). The Ca 2ϩ -binding domain is followed by a leucine zipper domain (residues 347-389), which has been postulated to induce NUCB1 dimerization (Fig. 1A) (20). The C-terminal (CT) region following the leucine zipper domain is predicted to be intrinsically disordered and unstructured. Intriguingly, NUCB1 is strongly conserved from flies to humans (21) and is widely distributed among Golgi (16,22), nucleus (19,23), endoplasmic reticulum (24,25), and cytoplasm (17). The N-terminal signal sequence of NUCB1 targets it to different membrane compartments and its deletion renders NUCB1 cytosolic.
Recently, Brodeur et al. (26) reported the role of NUCB1 in LDL receptor-related protein 9 (LRP9) trafficking where the cytosolic NUCB1 fraction helps in LRP9 endosomal sorting and prevents its delivery to lysosomes. The ubiquitous expression of NUCB1 in various cell and tissue types results in a diverse interactome, including interacting partners like G protein ␣ subunits, cyclooxygenases, and amyloid precursor protein (16,(25)(26)(27). Several classes of G␣ subunits have been shown to interact with NUCB1 (16,(25)(26)(27). Yeast two-hybrid experiments performed with various deletion constructs of either G␣ i3 or NUCB1 mapped the C-terminal ␣5 helix domain of the G protein and the acidic region of NUCB1 (residues 264 -305) to be necessary for interaction (28). Immunofluorescencebased studies showed that NUCB1 and G␣ i protein subunits co-localize on the Golgi lumen and regulated secretion granules (17). Because G␣ subunits are ubiquitously involved in numerous signal transduction pathways in different tissue types, their interaction with NUCB1 might regulate downstream signaling events.
We are interested in understanding the role of the ␣5 helix of the G protein ␣ subunit in regulating nucleotide exchange rates (29 -31). Because the ␣5 helix of G␣ is involved in interaction with NUCB1, we characterized and studied the interaction of NUCB1 with G␣ i1 in detail. In this study, we present the detailed biophysical and biochemical characterization of the Ca 2ϩ -binding ability and the oligomeric state of a heterologously expressed N-terminally truncated soluble form of NUCB1, termed sNUCB1 (Fig. 1A). We show that sNUCB1 binds Ca 2ϩ and exists as a dimer in solution. Ca 2ϩ -free sNUCB1 preferentially binds G␣ i1 and inhibits its GDP release as demonstrated by several independent assays. We have shown that overexpression of sNUCB1 in HEK293 cells markedly decreases ligand-dependent receptor-mediated G␣ i inhibition of adenylyl cyclase. We conclude that sNUCB1 serves as a novel Ca 2ϩ -regulated GDI of G␣ i1 . Our findings suggest that NUCB1 is involved in regulating cytoplasmic G protein signaling pathways.

Reagents
GDP sodium salt and GTP␥S tetralithium salt were purchased from Sigma. The fluorescent GTP␥S analogue, mant(2Ј,3Ј-O-(N-methyl-anthraniloyl)-GTP␥S was purchased from Jena Biosciences (Jena, Germany), and BODIPY-FL-GTP␥S was purchased from Molecular Probes (Invitrogen). The His 6 -tagged mutants of WT G␣ i1 were generated by sitedirected mutagenesis using the high fidelity thermostable DNA polymerase Pfu (Stratagene). The cAMP dynamic 2 kit was purchased from CisBio (Bedford, MA) for cell-based studies. All reagents and chemicals used were of highest available purity.

Heterologous Expression of sNUCB1 and Heterotrimeric G Protein ␣-Subunit G␣ i1
The cDNA clones for human NUCB1 and rat G␣ i1 were obtained from the ATCC. The DNA fragment for the soluble form of NUCB1 or sNUCB1, corresponding to residues 31-461 of the human NUCB1 protein without the N-terminal signal sequence (residues 1-31), was cloned into the pET28a(ϩ) expression vector (Amersham Biosciences). Similarly, the DNA fragment corresponding to full-length WT rat G␣ i1 subunit (residues 1-354) was cloned into the pET28a(ϩ) expression vector. The vector was used to transform BL21 (DE3) cells to express an N-terminal His 6 tag protein containing a PreScission protease site directly after the His 6 tag. The expressed construct for sNUCB1 served as a template for introducing site-specific mutations using the QuikChange system (Stratagene, La Jolla, CA) to generate sNUCB1(W232A/W333A) or for introducing a missense mutation to generate the truncated version of the protein, namely sNUCB1(W333Ter). All proteins were expressed in BL21 (DE3) cells grown in the presence of 50 g/ml kanamycin. Cells were grown at 37°C to A 600 nm of ϳ0.7 and then induced with 500 M isopropyl 1-thio-␤-D-galactopyranoside (United States Biological). Post-induction, the culture was grown overnight at 17°C and subsequently harvested. The resulting pellets were resuspended in a buffer containing 50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 50 mM ␤-mercaptoethanol, bovine lung aprotinin (20 mg/ml), 2 mM phenylmethanesulfonyl fluoride (PMSF), and complete EDTA-free protease inhibitor mixture tablets (Roche Applied Science). Thereafter, the cells were lysed, and the expressed His 6 tag protein was purified from crude extract by affinity chromatography using nickelnitrilotriacetic acid column pre-equilibrated with buffer A (50 mM Tris, pH 8.0, 150 mM NaCl, 50 mM ␤-mercaptoethanol) for sNUCB1 and its variants and buffer AЈ (50 mM Tris, pH 8.0, 150 mM NaCl, 2 mM MgCl 2 ⅐6H 2 O, 10 M GDP, 50 mM ␤-mercaptoethanol) for WT G␣ i1 . The bound His 6 tag protein was eluted from the column by using buffer A or buffer AЈ supplemented with 500 mM imidazole. The His 6 tag was cleaved by PreScission TM protease, which was subsequently removed. To remove Ca 2ϩ from sNUCB1 or its variants, the proteins were extensively dialyzed against buffer A supplemented with 5 mM EGTA and 1 mM EDTA, which are also eventually removed through dialysis. Finally, SEC was done using a Superdex200 26/60 HR column equilibrated with buffer S (50 mM Tris, pH 8.0, 150 mM NaCl) for sNUCB1 and its variants and equilibrated in buffer SЈ (50 sNUCB1 Is a Calcium-dependent GDI of G␣ i1 mM Tris, pH 8.0, 150 mM NaCl, 2 mM MgCl 2 ⅐6H 2 O, 10 M GDP, 1 mM DTT) for WT G␣ i1 to obtain homogeneously pure protein. The typical yield for sNUCB1 and its variants was ϳ5 mg/liter and for WT G␣ i1 ϳ12-15 mg/liter. The purity of the proteins was assessed by Coomassie Brilliant Blue staining after the proteins were separated through SDS-PAGE. The purity of all proteins was greater than 95%.

Isothermal Titration Calorimetry (ITC)
ITC was performed at 25°C (298 K) using a MicroCal VP-ITC (MicroCal, Northhampton, MA) calorimeter. To measure quantitatively the binding of Ca 2ϩ or WT G␣ i1 ⅐GDP to sNUCB1, 2 ml of 50 M sNUCB1 (Ca 2ϩ -free) and 700 l of either 500 M CaCl 2 ⅐6H 2 O (prepared in 50 mM Tris, pH 8.0, 150 mM NaCl) or 600 M WT G␣ i1 ⅐GDP (prepared in 50 mM Tris, pH 8.0, 150 mM NaCl, 2 mM MgCl 2 ⅐6H 2 O, 10 M GDP, 1 mM tris(2-carboxyethyl)phosphine) was used and thoroughly degassed. Contents of the sample cell were stirred continuously at 280 rpm during the experiment. A typical titration of sNUCB1 involved 50 injections each of 5 l of CaCl 2 ⅐6H 2 O (500 M) or WT G␣ i1 ⅐GDP (600 M) into a sample cell containing 1.4 ml of sNUCB1 (50 M). The heat of dilution of the titrant (CaCl 2 ⅐6H 2 O or WT G␣ i1 ⅐GDP) was subtracted from the titration data for base-line correction. The base line-corrected data were analyzed with MicroCal Origin TM 6.0 software to determine the change in enthalpy (⌬H) and association constant (K a ). Thermal titration data fit showed two sets of binding sites for Ca 2ϩ to sNUBC1. The parameters were calculated based on three or more independent titration experiments. Each numerical value reported is the mean Ϯ S.E. among the independent data sets.

Size-exclusion Chromatography
Complex Formation between sNUCB1 and WT G␣ i1 -Complex formation between sNUCB1 and G␣ i1 was monitored using SEC. 150 M sNUCB1 (ϪCa 2ϩ /ϩCa 2ϩ ) was incubated with 256 M G␣ i1 ⅐(GDP/GTP␥S) in 20 mM Tris, pH 8.0, 100 mM NaCl, 10 mM MgCl 2 , 10 M GDP, and 1 mM DTT at 30°C for 20 min with minimal shaking to enable saturation binding. WT G␣ i1 in the GTP␥S-bound state was generated by incubating WT G␣ i1 ⅐GDP with 5 mM GTP␥S for 30 min at 30°C to ensure complete exchange of nucleotide. Furthermore, the complete exchange was confirmed by the AlF 4 Ϫ exchange assay. About 200 l of the reaction mixture was injected onto Superdex200 10/30 GL column attached to AKTA FPLC (GE Healthcare) and run at 4°C. Elution was done with the incubation buffer at a flow rate of 0.5 ml/min with 0.35-ml fractions collected. Peak fractions were analyzed through SDS-PAGE and Coomassiestained to analyze the complex formation.
Estimation of sNUCB1 Molecular Mass Using Heavy Molecular Weight Standards-Purified sNUCB1 or the heavy molecular weight standards (GE Healthcare) were injected onto a Superdex200 10/30 GL column pre-equilibrated with buffer S, and the normalized absorbance of the eluting peak was plotted. A calibration curve of K av (V e Ϫ V o /V c Ϫ V o ) versus log 10 (molecular weight) was plotted, where V e is the elution volume; V o is the column void volume corresponding to the elution of blue dextran, and V c is the geometric column volume.
The data were fit to a straight line curve like Y ϭ A ϩ Bx, where A ϭ 1.915 and B ϭ 0.311 were obtained from the fit. Therefore, for a Y ϭ K av value, the corresponding value of log 10 (molecular weight) was obtained to estimate the molecular mass of sNUCB1 and sNUCB1(W333Ter).

Light Scattering
Multiangle Light Scattering (MALS) and Dynamic Light Scattering (DLS)-The light scattering data were collected using a Superdex200 10/30, HR SEC column (GE Healthcare), connected to an HPLC system (Agilent 1200, Agilent Technologies, Wilmington, DE) equipped with an autosampler. The elution from SEC was monitored by a photodiode array UV-visible detector (Agilent Technologies), differential refractometer (OPT-Lab rEx, Wyatt Corp., Santa Barbara, CA), static and dynamic, multiangle laser light scattering detector (HELEOS II with QELS capability, Wyatt Corp.). The SEC-UV/LS/RI system was equilibrated in buffer (50 mM Tris, pH 8.0, 150 mM NaCl) for sNUCB1 alone and in buffer (50 mM Tris, pH 8.0, 100 mM NaCl, 10 mM MgCl 2 , 10 M GDP, 1 mM DTT) for WT and truncated sNUCB1/G protein complex at the flow rate of 1.0 ml/min. Two software packages were used for data collection and analysis; the Chemstation software (Agilent Technologies) controlled the HPLC operation and data collection from the multiwavelength UV-visible detector, and the ASTRA software (Wyatt Corp., Santa Barbara, CA) collected data from the RI detector and the LS detectors and recorded the UV trace at 280 nm sent from the photodiode array detector. The weight average molecular masses, M w , were determined across the entire elution profile in the intervals of 1 s from static LS measurement using ASTRA software as described previously (32). The approach uses a Rayleigh-Debye-Gans light scattering model (Equation 1), which relates the amount of scattered light to the concentration and weight average molecular weight of solute and second virial coefficient, where R() is the intensity of excess scattered light at angle ; c is the concentration of the solute, M w is the weight average molecular weight of the solute; A 2 is the second virial coefficient; K* is an optical parameter equal to 4 2 n 2 (dn/dc) 2 / ( 0 4 N A ), where n is the refractive index; dn/dc is the refractive index increment for the solute; N A is Avogadro's number, and 0 is the wavelength of the scattered light. DLS measurements were made "on line" at an angle of 100°w ith a 2-s collection time. Time-resolved scatter intensity fluctuations were analyzed using Astra software (Wyatt Corp.), which implements the cumulants method (33) to determine the time dependence of diffusive motion also referred to as the intensity correlation function, G(T), as shown in Equation 2 (34), where B is the average base-line intensity; ␣ is an instrumentspecific correction factor; D T is the concentration-dependent translational diffusion constant of the solute; t is a delay time, and q is the scattering vector equal to (4n)/sin/2, where n is the refractive index of the solvent; is the wavelength of the scattered light, and is the scattering angle. Equation 2 describes the relationship between the time dependence of fluctuation in scatter intensity and the translational diffusion coefficient. The value of D T can be used to estimate the apparent hydrodynamic radius of an equivalent sphere using Stokes-Einstein relationship shown in Equation 3, where k is the Boltzmann constant; T is the absolute temperature, and is the temperature-corrected viscosity of the solution.
Determination of Dimerization Equilibrium Constant from SEC-UV/LS/RI-The concentration-dependent changes in the weight average molecular mass determined for the apex of the eluting peak for sNUBC1 were used to determine the dimerization constant. The weight average molecular weights were plotted as a function of protein concentration for sNUCB1; error bars represent Ϯ3% of M w for concentrations above 100 g/ml, Ϯ5% of M w for concentrations between 1 and 100 g/ml, and Ϯ10% of the M w for concentrations below 1 g/ml. Lines represent the nonlinear least square fits to a monomer-dimer association model for sNUCB1 dimerization.
Determination of K d of Complex Formation from SEC-UV/ LS/RI-The concentration-dependent changes in the M w determined for the apex of the eluting peak were used to determine the complexation of G␣ i1 with sNUCB1/sNUCB1(W333Ter). The M w values were plotted as a function of total protein concentration for complexation; error bars represent Ϯ3% of M w for concentrations above 100 g/ml, Ϯ5% of M w for concentrations between 1 and 100 g/ml, and Ϯ10% of M w for concentrations below 1 g/ml. Lines represent the nonlinear least square fits to a stimulated complexation model using Origin 6.0.

Fluorescence Spectroscopy
Steady-state Trp Fluorescence upon Ca 2ϩ Binding-Steadystate fluorescence measurements for sNUCB1 were carried out in a SPEX 3 spectrofluorimeter (Jobin-Yvon Instruments) at 25°C. The excitation wavelength was set at 295 nm, although emission spectra were collected from 310 to 460 nm. The excitation and emission slit widths were 2.5 and 5.0 nm, respectively. 178 nM Ca 2ϩ -free sNUCB1 equilibrated in buffer containing 50 mM Tris, pH 8.0, 100 mM NaCl was incubated with increasing concentrations of Ca 2ϩ with constant stirring for 5 min. Thereafter, Trp emission fluorescence spectra were collected for each concentration of Ca 2ϩ added.
Quenching of Tryptophan Fluorescence by Small Molecule Quenchers-The excitation wavelength was set at 295 nm, and the emission spectra were collected from 315 to 460 nm with the excitation slit and emission slit width being 2.5 and 5 nm, respectively. 333 nM of protein was equilibrated in buffer containing 50 mM Tris, pH 8.0, and 100 mM NaCl. To study the interaction between the protein and the quenchers, small aliquots of the quenchers were added incrementally from a con-centrated stock solution. In the case of potassium iodide, a small quantity of Na 2 S 2 O 3 (0.1 mM) was added to prevent it from forming free triiodide (I 3 Ϫ ). The accessibility of Trp was monitored by analyzing the quenching data using the Stern-Volmer Equation 4, where F 0 and F are the fluorescence intensities in the absence and presence of quenchers, respectively; [Q] is the concentration of the quenching molecules, and K SV is the Stern-Volmer quenching constant. In the case of purely collisional quenching mechanism, the Stern-Volmer plot of F 0 /F versus [Q] is linear with a slope equal to K SV .
Fluorescence-based mant-GTP␥S (mGTP␥S) Binding Assays-Steady-state fluorescence measurements experiments were carried out in a SPEX 3 spectrofluorimeter (Jobin-Yvon Instruments) at 25°C. 15 M WT G␣ i1 was incubated with 50 M sNUCB1(W232A/W333A) in the reaction buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 2 mM MgCl 2 , and 1 mM DTT) for 5 min with constant stirring. Thereafter, mGTP␥S (15 M) was added to the reaction mixture. FRET was monitored by exciting the intrinsic Trp fluorescence at 295 nm and measuring the fluorescence emission from 310 to 540 nm. In all cases, blank spectra containing buffer with sNUCB1(W232A/ W333A) alone were subtracted from the final spectra.

Time-based Fluorescence Assays
Time-based fluorescence measurements were performed in a SPEX Fluorolog-2 spectrofluorometer (SPEX Industries, Edison, NJ) at 25°C. The excitation wavelength was set at 295 nm specifically for Trp residues, although emission wavelength was set at 340 nm. The excitation and emission slit width were 5 nm each.
AlF 4 Ϫ Uptake Assay-WT G␣ i1 was incubated at 25°C in reaction buffer K (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 2 mM MgCl 2 , and 1 mM DTT) with increasing amounts of sNUCB1(W232A/W333A) for 10 min with constant stirring. Thereafter, for AlF 4 Ϫ -dependent activation of G␣ i1 , a premixed solution of AlCl 3 Ϫ (20 mM) and NaF (1 M), respectively, was injected at 180 s. A time course of enhancement in Trp fluorescence coupled to activation was monitored in the presence of increasing concentrations of sNUCB1(W232A/W333A) in the reaction mixture.
GTP␥S Nucleotide Exchange Assay-WT G␣ i1 was incubated at 25°C with increasing concentrations of sNUCB1(W232A/ W333A) in reaction buffer K for 10 min with constant stirring. The base-line fluorescence was monitored for 200 s after which GTP␥S (20 M) was injected into the reaction mixture, and the relative increase in intrinsic Trp fluorescence ( ex ϭ 295 nm, em ϭ 340 nm) was measured as a function of time for each concentration of sNUCB1(W232A/W333A).
BODIPY-FL-GTP␥S Nucleotide Exchange Assay-WT G␣ i1 (20 M) was incubated at 25°C in reaction buffer K with increasing concentrations of Ca 2ϩ -free sNUCB1 for 10 min with constant stirring. BODIPY-FL-GTP␥S (20 M) was added to the reaction mixture, and the relative increase in intrinsic fluorescence ( ex ϭ 485 nm, em ϭ 512 nm) was measured for each concentration of Ca 2ϩ -free sNUCB1 as a function of time.

sNUCB1 Is a Calcium-dependent GDI of G␣ i1
BODIPY-FL-GTP␥S Nucleotide Binding Assay-WT G␣ i1 (20 M) was incubated alone and with Ca 2ϩ -free sNUCB1 (100 M) at room temperature in reaction buffer K. 10-l samples were withdrawn from each reaction mixture at various time points and loaded onto a pre-equilibrated Zeba TM micro-desalt spin column to remove the unbound nucleotide. The columns were spun at 1000 ϫ g for 1 min, and the flow-through was collected. The amount of bound nucleotide in each flowthrough was measured by using BODIPY absorbance at 504 nm (⑀ ϭ 80,000 M Ϫ1 cm Ϫ1 ). The absorbance at 280 nm was also monitored to ensure that the same amount of protein was being collected in each flow-through sample.

Radioligand [ 35 S]GTP␥S-based Nucleotide Exchange Assay
The effect of Ca 2ϩ -free sNUCB1 or Ca 2ϩ -free sNUCB1-(W333Ter) binding to WT G␣ i1 ⅐GDP on the nucleotide exchange was monitored using [ 35 S]GTP␥S exchange assay. 20 M WT G␣ i1 ⅐GDP alone or with 100 M Ca 2ϩ -free sNUCB1/ sNUCB1(W333Ter) was incubated in buffer (10 mM Tris, pH 7.5, 100 mM NaCl, 2 mM MgCl 2 , 2 mM DTT) for 20 min at 30°C. After incubation, [ 35 S]GTP␥S solution diluted with GTP␥S was added to the reaction mixtures such that the final GTP␥S concentration in each reaction mixture was 100 M with an associated radioactivity of 50 nCi/8 l of reaction volume. Thereafter, 8-l samples (50 nCi) were withdrawn from each reaction mixture and added to buffer-equilibrated nitrocellulose filters. Subsequently, the filters were extensively washed with Wash buffer (50 mM Tris, pH 7.5, 5 mM MgCl 2 ) to remove excess radioligand. For data analysis, the counts were recorded, and the value for buffer alone was subtracted from each reading. Each individual data point was scaled to the value of radioactivity within 8 l of reaction volume to get the picomoles of radioligand bound to WT G␣ i1 in each reaction. As a positive control, the C-terminal GoLoco motif peptide from RGS14, namely R14GL(496 -531), was used in an identical experimental setup.

Circular Dichroism Spectroscopy
Wavelength Scan-All CD experiments were performed using an Aviv 62A DS CD spectrophotometer. Far-UV CD spectra were recorded at the end of each kinetic run. Spectra were recorded over the wavelength range of 190 -250 nm at 1-nm intervals with an averaging time of 3 s using a 0.1-cm path length cell. Background spectrum was subtracted from each of the collected data sets. Each spectrum obtained was an average of three scans. The recorded spectra in millidegrees of ellipticity was converted to mean residue ellipticity in degrees⅐cm 2 dmol Ϫ1 using Equation 5, where [] is the mean residue ellipticity in degrees cm 2 dmol Ϫ1 ; is the ellipticity in millidegrees; c is the concentration in M; l is the path length in cm, and n is the number of peptide bonds.
Thermal Unfolding-The unfolding of the protein (8 M) with temperature was monitored using CD at a wavelength of 222 nm, which is characteristic of an ␣-helix. The data points were averaged over 30 s for every unit increment in tempera-ture. A plot of CD signal versus temperature was fit to Equations 6 and 7, where, where f(T) is the signal as a function of temperature; T is the temperature; R is the gas constant; ␣ N defines the intercept; ␤ N is the slope of the post-transition region of the curve; ␣ D defines the intercept; ␤ D is the slope of the pre-transition region of the curve; ⌬G 0 is the free energy change for the unfolding reaction; is the change in enthalpy for unfolding at the T m , and ⌬C p o is the change in heat capacity. Using the above expressions, T m , the mid-point transition temperature, was estimated along with the change in enthalpy ⌬H D-N o , for the unfolding reaction at that T m .

Analytical Ultracentrifugation (AUC)
Sedimentation equilibrium studies were carried out at different centrifugal speeds and protein concentrations for sNUCB1 and sNUCB1(W333Ter). AUC experiments for sNUCB1 (Ϫ/ϩCa 2ϩ ) were done at 25°C with protein concentrations of 50, 100, and 150 M each at speeds of 20,000, 30,000, and 40,000 rpm. Similarly for sNUCB1(W333Ter), AUC experiments were done with 32, 70, and 150 M protein concentrations each at speeds of 25,000, 30,000, and 40,000 rpm. The individual protein samples were run for sufficiently long periods of time to allow for the equilibrium to generate a time invariant concentration gradient balanced by diffusion of the macromolecular species. The speeds were decided based on the molecular weight of the protein. Under no net transport conditions, the following correlation (Equations 8 and 9) should be observed between concentration and the radial distance (35), Cr is the concentration of macrosolute at any radial distance r; Cr o is the concentration of the macrosolute at the reference radial distance r o ; is the partial specific volume; is the angular velocity; is the density; R is the gas constant; T is the absolute temperature; and M is the molecular mass. The data were analyzed using Optima TM XL-A/XL-I data analysis software (Beckman, 2001). HEK293 cells were maintained in Dulbecco's modified Eagle medium (Invitrogen) with 10% (v/v) fetal bovine serum (FBS) at 37°C and 5% CO 2 . For Western blot analysis, cells were plated in 6-well plates and transiently transfected with 4 g of total DNA (3.0 g of pcDNA3.1(ϩ) ϩ 1.0 g of WT G␣ i1 /sNUCB1/ CXCR4), and cells were harvested after 48 h. Thereafter, cells were lysed, and the heat-denatured supernatant was subjected to SDS-PAGE analysis. Subsequently, the separated protein fractions were transferred to a PVDF membrane using semi-dry transfer procedures and probed with either 1D4 antibody to detect CXCR4, which has a C-terminal 1D4 tag, or with the respective primary antibodies for NUCB1 or G␣ i1 . Finally, the excess antibody was washed away, and the membrane was probed with a horseradish peroxidase-conjugated secondary antibody, and the chemiluminescence upon addition of the substrate was recorded.
cAMP Assay cAMP dynamic2 assay kit was purchased from CisBio. HEK293 cells (ATCC) were maintained in Dulbecco's modified Eagle's medium (Invitrogen) with 10% (v/v) fetal calf serum (FCS) at 37°C and 5% CO 2 . For cAMP assay, 800,000 cells were plated in a 6-well plate and transiently transfected with pcDNA3.1(ϩ) control, WT G␣ i1 , sNUCB1, or CXCR4 expression constructs using Lipofectamine 2000 (Invitrogen). Briefly, 4.0 g of total DNA (3.0 g of pcDNA3.1(ϩ) ϩ 1.0 g of WT G␣ i1 /sNUCB1/CXCR4) was complexed with 12 l of Lipofectamine 2000 reagent and incubated for 20 min at room temperature before addition of the DNA/lipid mixture to the cells. Cells were plated 24 h after transfection at a density of 8000 cells/well in a poly-D-lysine-coated 384-well plate. 48 h after transfection, cells were left untreated or stimulated with forskolin alone or forskolin (10 M) ϩ SDF1␣ in media containing 100 M inhibitor of phosphodiesterase, namely isobutylmethylxanthine (3,7-dihydro-1-methyl-3-(2-methylpropyl)-1H-purine-2,6-dione), for 15 min at 37°C and 45 min at room temperature. After an hour of stimulation, lysis buffer alone was added to the untreated cells and lysis buffer with cAMP-d 2 was added to the stimulated cells. Thereafter, cryptate-conjugated cAMP antibody was added to all cells and incubated for 1-2 h. Homogeneous time-resolved fluorescence was recorded with excitation at 320 nm and emission at 620 and 665 nm. A standard curve was prepared by varying the amount of cAMP-d 2 and probing it with cryptate-conjugated cAMP antibody using homogeneous time-resolved fluorescence. The normalized signal from treated cells with respect to untreated cells was converted to the cAMP concentration using the standard curve.

RESULTS
Ca 2ϩ Binding to EF Hand Domains of sNUCB1 Induces Conformational Changes-A soluble form of NUCB1 was engineered, expressed in Escherichia coli, and purified to homogeneity ( Fig. 1A and supplemental Fig. S1A). The domain architecture of sNUCB1 shows two EF hand Ca 2ϩ -binding regions sandwiched between an N-terminal DNA binding domain and a C-terminal leucine zipper (20). The thermodynamics of Ca 2ϩ binding to sNUCB1 were studied using ITC. for Ca 2ϩ binding to the second EF hand, suggesting a spontaneous uptake of Ca 2ϩ by Ca 2ϩ -free sNUCB1. The recorded data were best fit to a "two-set binding sites" model giving K d values of 6.3 and 73.5 M for Ca 2ϩ binding (Fig. 1B). The difference in binding affinities arises due to the presence of a noncanonical Arg residue substituting for the canonical Gly at the sixth position of the 12-residue Ca 2ϩ binding loop of the second EF hand. Our results show that the two EF hands in sNUCB1 bind to Ca 2ϩ differentially with an order of magnitude difference in their K d values. In 2004, de Alba and Tjandra (36) heterologously expressed and purified only the EF hand domain (residues 228 -326) of NUCB1 and also reported two Ca 2ϩ -binding sites with different affinities. The EF hand regions undergo conformational changes upon Ca 2ϩ binding that can be monitored through change in sNUCB1 Trp fluorescence. sNUCB1 has two Trp residues at positions 232 and 333 lying sequentially before and after the EF hand domain, respectively. Steady-state fluorescence measurements show a blue shift of 15 nm and an enhancement in fluorescence intensity as more Ca 2ϩ was added to Ca 2ϩ -free sNUCB1 (supplemental Fig. S1B). Thus, Ca 2ϩ binding causes a conformational change in sNUCB1 causing reorientation of the Trp residues from a solvent-exposed polar to a hydrophobic environment. Fluorescence quenching experiments with acrylamide and iodide were performed to probe the conformational changes around the Trp residues upon Ca 2ϩ binding. Acrylamide is a diffusible neutral quencher of Trp fluorescence unaffected by the charges surrounding the fluorophore. I Ϫ , on the other hand, is an anionic quencher for surface-exposed fluorophores only (37). Quenching experiments show that Trp residues are more accessible to acrylamide than I Ϫ for quenching. On binding to Ca 2ϩ , acrylamide quenching decreases much more than I Ϫ quenching (supplemental Fig. S1C). F 0 /F versus quencher concentration gave a straight line indicative of a dynamic quenching rather than static quenching. These observations suggest that conformational changes associated with Ca 2ϩ binding make Trp residues less accessible by reorienting them into a hydrophobic environment.
The binding of Ca 2ϩ to an EF hand motif forms an octahedral coordination sphere, which can cause structural rearrangement in the protein (38). This secondary structural change can in principle be followed by CD spectroscopy. Ca 2ϩ binding to sNUCB1 causes only a slight increase in helicity. Intriguingly, the far-UV CD spectra of sNUCB1 show a smooth structural transition from an ␣-helical secondary structure at lower concentrations to a predominant ␤-sheet structure at higher concentrations. These concentration-dependent structural changes were observed both in the absence and presence of Ca 2ϩ (supplemental Fig. S2A). Furthermore, the thermal unfolding data for sNUCB1 showed an increase in apparent T m of 4°C on binding to Ca 2ϩ . The folding of sNUCB1 was observed to be noncooperative and irreversible both in the absence and presence of Ca 2ϩ (supplemental Fig. S2B). The increase in stabilization on binding to Ca 2ϩ suggests a structural ordering of sNUCB1.

sNUCB1 Is a Calcium-dependent GDI of G␣ i1
NUCB1 Is a Dimer-The domain diagram of NUCB1 reveals the presence of a C-terminal domain leucine zipper, which is a canonical structural motif known to cause dimerization (39). The oligomeric state of NUCB1 was evaluated through sedimentation equilibrium experiments using AUC. This exper-iment is insensitive to the shape of the molecule and directly reports on the molar mass of the sedimenting species characterized by the Boltzmann distribution. Sedimentation equilibrium measurements were performed with Ca 2ϩfree sNUCB1 for concentrations varying from 50 to 150 M. The data could be well fit to a monomer-n-mer model, which gave a molecular mass of ϳ100 kDa corresponding to a dimer of sNUCB1 in solution ( Fig. 2A). AUC analysis of Ca 2ϩ -bound sNUCB1 indicated that it is also primarily dimeric in solution ( Fig. 2A, inset). To confirm that the dimerization of sNUCB1 is caused by the leucine zipper domain, a C-terminal truncation mutant, sNUCB1(W333Ter), lacking the leucine zipper was designed and heterologously expressed. The far-UV CD spectrum of sNUCB1(W333Ter) showed an ␣-helical secondary structure even at high protein concentrations in contrast to the concentration-dependent transition observed for sNUCB1 (supplemental Fig. S2C). AUC analysis of sNUCB1(W333Ter) revealed a monomeric species in solution over the concentration range of 32 to 150 M (Fig. 2B). Similar to sNUCB1, the association state of sNUCB1(W333Ter) was unaffected upon binding to Ca 2ϩ (data not shown). The experiments with the truncation mutant confirm that the leucine zipper domain of sNUCB1 is essential for the dimeric state of the protein in solution.
The thermodynamics of the dimerization of sNUCB1 was further investigated using MALS. In a series of SEC experiments, Ca 2ϩfree sNUCB1 was analyzed by gel filtration using a Superdex200 10/30 HR column. The eluted protein peak was fractionated and subjected to MALS analysis at seven different angles to estimate the molecular mass of the eluting species. As shown in Fig. 2C, a gradual increase in the weight-average molecular weight of the eluting species was observed with increasing concentrations of the protein. At concentrations around 10 nM, sNUCB1 exists as a monomer. As the concentration is gradually increased, a monomer-dimer equilibrium was FIGURE 1. Domain architecture of human NUCB1 and engineered NUCB1 variants and Ca 2؉ -binding activity of a soluble NUCB1 variant. A, modular nature of the NUCB1 protein structure is depicted schematically with its N-terminal signal sequence (gray square), putative DNA binding domain (green square), two EF hand loops (magenta rectangles), acidic region (blue oval), and leucine zipper domain (yellow hexagon). The amino acid residues at domain borders are numbered. We engineered a construct for expression in E. coli of a soluble form of NUCB1 that included an N-terminal His 6 tag and a PreScission protease recognition sequence with an intervening spacer sequence as shown. Endoprotease cleavage yields a soluble form of NUCB1 that begins with the amino acid sequence GPHMAS and continues with the remainder of the native sequence beginning at Gly-32. We refer to this expressed protein construct as soluble NUCB1 or sNUCB1 throughout the text. We also constructed a double replacement mutant sNUCB1(W232A/W333A) and a truncation mutant sNUCB1(W333Ter) as shown. B, we used isothermal titration calorimetry (ITC) to measure the Ca 2ϩ -binding affinity of sNUCB1. sNUCB1 presumably binds one Ca 2ϩ cation at each of its two EF hand domains, and a nonlinear least squares fit of the calculated values using the two-set of sites model resulted in an excellent fit as shown in the inset. The calculated dissociation constants for Ca 2ϩ binding to the two EF hand domains of sNUCB1 are 6.3 Ϯ 0.18 and 73.5 Ϯ 0.19 M, respectively.

sNUCB1 Is a Calcium-dependent GDI of G␣ i1
OCTOBER 8, 2010 • VOLUME 285 • NUMBER 41 observed with dimeric sNUCB1 dominating at concentrations greater than 1 M. The data collected at different angles for each run were fit to a monomer-dimer equilibrium model giving a dissociation constant, K d of dimerization, for sNUCB1 of 0.26 Ϯ 0.12 M. sNUCB1(W333Ter) in MALS measurements continued to exist as a monomer even at high protein concentrations. Thus, the leucine zipper region is responsible for the dimeric state of sNUCB1 in solution under physiological conditions. sNUCB1 Is Structurally Asymmetric with an Elongated C Terminus-Ca 2ϩ -free sNUCB1 in SEC experiments elutes at a much higher volume than a globular protein with a molecular mass similar to that of a sNUCB1 monomer or a dimer (supplemental Fig. S3A). A plot of the elution parameter, K av , versus log(molecular mass) for a number of globular protein standards was utilized to estimate an apparent molecular mass of 300 kDa for sNUCB1 (supplemental Fig. S3B). This implies that either sNUCB1 exists as a trimer of dimers (3ϫ 100 kDa) or is asymmetric in its shape. Because AUC and MALS analysis convincingly ruled out any association of sNUCB1 dimers into higher order structures, we proceeded to derive structural information on sNUCB1 through DLS. DLS analysis of Ca 2ϩfree sNUCB1 yields a hydrodynamic radius (R h ) of 6.2 nm and a frictional coefficient (f/f 0 ) of 2.03 (Fig. 2D). The frictional coefficient of globular proteins usually lies in the range of 1.1-1.25 (40). A value of 2.03 for sNUCB1 indicates an asymmetric shape for the dimer, deviating considerably from a sphere. Interestingly, Ca 2ϩ -free sNUCB1(W333Ter) has an R h of 3.3 nm and a FIGURE 2. sNUCB1 forms a stable dimer mediated by its leucine zipper domain. A, we carried out AUC experiments to determine the oligomeric state of sNUCB1 in solution. The sedimentation equilibrium plot for Ca 2ϩ -free sNUCB1 (50 M monomeric concentration) shows the evolution of a sample concentration curve resulting from the applied centrifugal force. The data were fit to a monomer-n-mer equilibrium model, which gave a measured molecular mass of 98.9 Ϯ 0.41 kDa, consistent with a dimeric structure for sNUCB1 in solution (the theoretical molecular mass of monomeric sNUCB1 is ϳ51 kDa). The inset shows the sedimentation equilibrium plot for Ca 2ϩ -bound sNUCB1, which gave essentially an identical molecular mass of 99.4 Ϯ 0.78 kDa. B, AUC experiments were carried out on the truncation mutant, sNUCB1(W333Ter), which lacked the leucine zipper domain in the C-terminal region of the protein. The AUC sedimentation equilibrium plot for Ca 2ϩ -free sNUCB1(W333Ter) (70 M monomeric concentration) gave a molecular mass of 35.2 Ϯ 0.05 kDa, consistent with a monomeric species in solution (the theoretical molecular mass of sNUCB1(W333Ter) is ϳ36.8 kDa). C, we studied the monomer-dimer equilibrium of sNUCB1 with MALS of fractions with increasing protein concentration collected from Superdex200 10/30 HR size-exclusion chromatography in 50 mM Tris-HCl, pH 8.0, 150 mM NaCl. The weight average molecular weight of the complex at a given protein concentration was determined from a nonlinear least square fit of a collection of values determined for the apex fractions of each eluting peak. A monomer-dimer association model of the values as a function of sNUCB1 concentration gave an apparent dissociation constant (K d ) for dimerization of 0.26 Ϯ 0.12 M. The error bars indicate the extent of variation in molecular mass determination originating from the light scattering measurement. We next measured DLS chromatograms for sNUCB1 (D) and sNUCB1(W333Ter) (E). Each protein sample was injected onto a Superdex200 10/30 HR column, and the refractive index detector was used to analyze the protein peak. The change in refractive index as a function of protein concentration was used to compute the molecular mass as shown in green for sNUCB1 and red for sNUCB1(W333Ter). The molecular mass corresponds to a dimer for sNUCB1 and a monomer for sNUCB1(W333Ter). The insets show the correlation functions as each protein diffuses through the solution in the DLS experiment. The diffusion coefficient from the correlation function gives the hydrodynamic radius (R h ) using Stokes equation. The computed R h values were 6.2 and 3.03 nm for sNUCB1 and sNUCB1(W333Ter), respectively.

sNUCB1 Is a Calcium-dependent GDI of G␣ i1
f/f 0 of 1.59 showing that the deletion of the C-terminal domain reduces the radius of hydration by almost 50% (Fig. 2E). This suggests that although the N terminus of sNUCB1 is relatively more compact and globular, the C-terminal domain of the protein is elongated, contributing to the observed asymmetry in the shape of sNUCB1.
sNUCB1 Preferentially Interacts with GDP-bound WT G␣ i1 in the Absence of Ca 2ϩ -NUCB1 is a Ca 2ϩ -binding protein that has been shown to interact with G i and G s classes of ␣ subunits (16). Physiologically, G␣ subunits can individually exist both in the "inactive" or GDP-bound state and the "active" or GTPbound state. To investigate the interaction of sNUCB1 with WT G␣ i1 , sNUCB1 was incubated with WT G␣ i1 ⅐GDP (Fig. 3A) and WT G␣ i1 ⅐GTP␥S (Fig. 3B), respectively, and the complex formation was analyzed by SEC using a Superdex200 10/30 HR column. As shown in the top panels of Fig. 3, the Coomassiestained gels of the fractionated peaks show elution of the complex containing both sNUCB1 and G␣ i1 with excess G␣ i1 fractionating at higher elution volume. The amount of G␣ i1 eluting with sNUCB1 was significantly higher when G␣ i1 was present in the GDP-bound state rather than in the GTP␥S-bound state. Thus, Ca 2ϩ -free sNUCB1 preferentially interacts with G␣ i1 in the GDP-bound form. Interestingly, the interaction with G␣ i1 in both the GDP-and GTP␥Sbound states is absent when sNUCB1 is bound to Ca 2ϩ as shown in the lower panels of Fig. 3. The two protein subunits elute out as individual components as the complex is disrupted in the presence of Ca 2ϩ . The inhibition of the complex formation between sNUCB1 and G␣ i1 in the presence of Ca 2ϩ can be explained by the structural rearrangement of EF hands upon Ca 2ϩ binding. Yeast two-hybrid analysis for interaction of G␣ with deletion mutants of NUCB1 as bait proposed the EF hand intervening acidic region of NUCB1 to be involved in G protein binding (16). The conformational change induced upon Ca 2ϩ -binding can possibly result in the masking of the acidic region sandwiched between the two EF hands, thereby making it inaccessible for G protein binding.
The thermodynamics of Ca 2ϩfree sNUCB1 complex formation with G␣ i1 ⅐GDP was investigated using ITC. The exothermic nature of the reaction shows a thermodynamically favorable binding event between the two protein subunits with ⌬H 0 ϭ Ϫ1.45 Ϯ 0.17 kcal mol Ϫ1 and ⌬S 0 ϭ 16.9 cal mol Ϫ1 K Ϫ1 . The data were best fit to a "one-set binding sites" model with dimeric sNUCB1 and monomeric G protein as the interacting partners. The fit provides a K d value of 18.3 Ϯ 1.45 M suggesting that sNUCB1 interacts relatively weakly with GDP-bound G␣ i1 (Fig. 4A). The association between Ca 2ϩ -free sNUCB1 and G␣ i1 ⅐GDP was also monitored using light scattering. Protein samples with increasing concentrations of sNUCB1-G␣ i1 ⅐GDP complex were injected onto a Superdex200 10/30 HR column, and MALS was done on the complex elution peak. The analysis shows that binding of G␣ i1 ⅐GDP to sNUCB1 initiates only after dimerization of Ca 2ϩ -free sNUCB1. At lower concentrations, dimerization of sNUCB1 from molecular mass of 50 -00 kDa was traced. Once the dimer was formed, binding of one G␣ i1 ⅐GDP subunit to sNUCB1 was evident from an increase in molecular mass to 140 kDa (Fig. 4B). The trace observed for sNUCB1-G␣ i1 ⅐GDP complex formation can be understood from the experimental K d values. The K d value for sNUCB1 dimerization is much smaller than the K d value for binding to G␣ i1 ⅐GDP, resulting in sNUCB1 dimerization prior to G␣ i1 ⅐GDP binding. However, sNUCB1-G␣ i1 ⅐GDP association curve does not rule out the ability of sNUCB1 monomer to interact with G␣ i1 ⅐GDP.
MALS experiments were also conducted with the monomeric truncation mutant sNUCB1(W333Ter). A plot of the FIGURE 3. sNUCB1 forms a stable complex with G␣ i1 ⅐GDP only in the absence of Ca 2؉ . We used sizeexclusion chromatography to determine the nature of the interaction between sNUCB1 and G␣ i1 in solution under several conditions. sNUCB1 and G␣ i1 ⅐GDP were incubated together in the absence (upper left) or presence (lower left) of Ca 2ϩ , and the mixture was subjected to size-exclusion chromatography using a Super-dex200 10/30 HR column. Peak fractions were analyzed by SDS-PAGE with Coomassie Brilliant Blue staining as shown in the insets. In the absence of Ca 2ϩ , sNUCB1 and G␣ i1 ⅐GDP formed a complex as judged by the following: (i) the slight shift (from ϳ11.2 to 10.6 ml) and enhancement of the peak containing only sNUCB1 in the presence of Ca 2ϩ and both sNUCB1 and G␣ i1 in the absence of Ca 2ϩ , and (ii) the depletion, in the absence of Ca 2ϩ , of the peak eluting at about 14.6 ml, which contained only G␣ i1 . The same experiment was repeated with sNUCB1 and G␣ i1 ⅐GTP␥S in the absence (upper right) or presence (lower right) of Ca 2ϩ . Although the same general trend and Ca 2ϩ dependence were observed, the apparent ability of G␣ i1 ⅐GTP␥S to form a stable complex with sNUCB1 was markedly diminished compared with that of G␣ i1 ⅐GDP. sNUCB1 Is a Calcium-dependent GDI of G␣ i1 OCTOBER 8, 2010 • VOLUME 285 • NUMBER 41 weight average molecular weight obtained from light scattering plotted versus the protein concentration shows that monomeric Ca 2ϩ -free sNUCB1(W333Ter) forms a 1:1 complex with G␣ i1 ⅐GDP. The fit yields a K d value of 1.99 Ϯ 0.26 M showing that the interaction of sNUCB1 with WT G␣ i1 ⅐GDP becomes stronger on deletion of the C-terminal region of sNUCB1 (Fig.  4C). This suggests that the C-terminal region of sNUCB1 possibly acts as a regulatory domain modulating the interaction of sNUCB1 with the G protein ␣ subunit. Alternatively, the structural rearrangement associated with dimerization can also result in the reduced affinity of sNUCB1 for G␣ i1 in comparison with sNUCB1(W333Ter).
sNUCB1 Is a GDI-GPCRs activate heterotrimeric G proteins by catalyzing the exchange of GTP for GDP on G␣ subunits. The receptor facilitates the nucleotide exchange by using the C-terminally linked ␣5 helix of G␣ subunits. The ␣5 helix region is postulated to be a molecular microdomain on G␣ that allosterically couples the activation signal from the receptor to the catalytic pocket of G proteins (29,30,41,42). Farquhar and co-workers (28) have shown that NUCB1 possibly interacts with the C-terminal ␣5 helix region of G␣ i3 . We evaluated the effect of this interaction on G protein activation. We measured the nucleotide exchange on G␣ in the presence of Ca 2ϩ -free sNUCB1. The concentrations of both sNUCB1 and G␣ i1 were kept above the determined K d value for sNUCB1-G␣ i1 complex formation. Steady-state and time-based fluorescence experiments were performed by monitoring the enhancement in Trp fluorescence, which is coupled to activation of G␣ i1 . Any potential contribution from the Trp residues of sNUCB1 was eliminated by mutating both Trp-232 and Trp-333 to Ala, thereby generating sNUCB1(W232A/W333A) (supplemental Fig. S4A). The binding of sNUCB1(W232A/W333A) to G␣ i1 ⅐GDP was confirmed using SEC and SDS-PAGE analysis. The interaction of sNUCB1(W232A/W333A) with G␣ i1 ⅐GDP was present only in the absence of Ca 2ϩ (supplemental Fig. S4B). Both AlF 4 Ϫ uptake (supplemental Fig. S4C) and GTP␥S exchange (supplemental Fig. S4D) by G␣ i1 ⅐GDP alone shows an increase in Trp fluorescence as G␣ i1 molecules were acti- integrated. The heat of dilution for the addition of G␣ i1 ⅐GDP to buffer alone was subtracted. A nonlinear least squares fit of the calculated values using the one-set of sites model for dimeric sNUCB1 resulted in a satisfactory fit with a dissociation constant of 18.3 Ϯ 1.45 M as shown in the inset. We used MALS to measure the formation of a complex between sNUCB1 and G␣ i1 ⅐GDP (B) or the truncation mutant sNUCB1(W333Ter) and G␣ i1 ⅐GDP (C). MALS data were collected on protein complex peaks eluting from Superdex200 10/30 HR sizeexclusion chromatography connected to a high performance liquid chromatography system equipped with an autosampler. Samples with increasing protein concentrations were successively injected onto the column, and the eluate was monitored using a photo-diode UV-visible detector, a differential refractometer, and a static multiangle laser light scattering detector. The weight average molecular weight of the complex was determined as a function of increasing protein concentrations. The magenta curve in B, inset, and the red curve in C indicate nonlinear least square fits for a complex formation model of weight average molecular weight values determined for the apex of each eluting peak at various concentrations. The error bars indicate extent of variation in molecular mass determination originating from the light-scattering measurement. The data for sNUCB1-G␣ i1 ⅐GDP complex formation shows that association between the protein subunits occurs only after dimerization of sNUCB1 (blue curve). The data suggest that one G␣ i1 subunit binds to a dimer of sNUCB1 and a monomer of sNUCB1(W333Ter), respectively, and that the binding site for G␣ i1 lies in the sNUCB1(W333Ter) sequence.

sNUCB1 Is a Calcium-dependent GDI of G␣ i1
vated with time. However, supplemental Fig. S4, C and D, shows that binding to sNUCB1(W232A/W333A) reduces the relative enhancement of Trp fluorescence coupled to G␣ i1 activation. Furthermore, to simultaneously monitor the activation-associated conformational change in G␣ i1 and the uptake of nucleotide, a mGTP␥S-based FRET assay was performed. The binding of mGTP␥S in the catalytic pocket of G␣ i1 brings it in close vicinity of Trp-211 on the switch II region of G␣ i1 . Hence, FRET can be observed between the Trp and mant fluorophores upon G␣ i1 activation (43). As shown in supplemental Fig. S4E, the FRET intensity for G␣ i1 alone at each time point is significantly higher in comparison with the FRET intensity for G␣ i1 associated with Ca 2ϩ -free sNUCB1(W232A/W333A). These experiments suggest that sNUCB1(W232A/W333A) might inhibit nucleotide exchange by G␣ i1 .
G protein activation was also monitored using a fluorescent nucleotide analogue BODIPY FL-GTP␥S, which absorbs at 504 nm on binding to G␣ i1 . We isolated the BODIPY FL-GTP␥Sbound fraction of G␣ i1 in the absence and presence of sNUCB1 and calculated the amount of bound BODIPY FL-GTP␥S in each case (Fig. 5). The supplemental Fig. S5 shows that binding of sNUCB1 to G␣ i1 does not affect the max of bound BODIPY FL-GTP␥S. However, binding of sNUCB1 to G␣ i1 reduces the amount of nucleotide exchange by G␣ i1 as seen by a decrease in the BODIPY absorbance. Thus, measuring absorbance of bound BODIPY FL-GTP␥S provides a direct monitor of the number of bound nucleotides on G␣ i1 in the presence or absence of sNUCB1. Fig. 5A shows BODIPY FL-GTP␥S nucleotide exchange by G␣ i1 alone (blue) and by G␣ i1 in complex with sNUCB1 (orange). The results show a significant reduction in the number of bound BODIPY FL-GTP␥S nucleotides to G␣ i1 when complexed with Ca 2ϩ -free sNUCB1. This shows that binding of sNUCB1 to G␣ i1 causes inhibition of nucleotide exchange.
To further substantiate the effect of sNUCB1 on nucleotide exchange by WT G␣ i1 , we monitored the exchange of radioligand [ 35 S]GTP␥S by G␣ i1 in the presence and absence of sNUCB1. Fig. 5B shows the kinetics of exchange of GDP for [ 35 S]GTP␥S by G␣ i1 alone (blue). However, in the presence of sNUCB1 (Fig. 5B, green), the rate and extent of [ 35 S]GTP␥S exchange are significantly reduced. Thus, sNUCB1 on binding to G␣ i1 influences the nucleotide exchange rate by substantially inhibiting GDP release and subsequently the GTP uptake.
GDI Activity of sNUCB1 Does Not Arise from a Canonical GoLoco Motif-Earlier work by Siderovski and co-workers (9, 14, 44 -46) has unraveled several interacting partners of G␣ i1 ⅐GDP, namely RGS12, RGS14, AGS3, GPSM2, PCP-2, and G18, all of which possess GDI activity. The GDI activity was attributed to the presence of a single or tandem repeats of a 19-amino acid-long conserved GoLoco motif present in each of these proteins. The crystal structure of G␣ i1 ⅐GDP complexed to RGS14-derived GoLoco peptide showed that the conserved C terminus of the 19-amino acid GoLoco motif makes contacts with both the switch I region and the bound nucleotide. Each GoLoco motif ends in a conserved triad of (Asp/Glu)-Gln-Arg. On binding to G␣, the Arg side chain extends into the catalytic pocket to interact directly with the bound GDP and stabilize it.
In addition, the highly conserved Asp/Glu and Gln residues are absolutely essential for binding to G protein ␣ subunit as they are involved in both intra-and intermolecular H-bonding interactions (47). A sequence alignment of C-terminal residues of NUCB1(381-399) with GoLoco motifs of several GDIs of G␣ i subunits reveals that NUCB1 also possesses the highly conserved triad of Glu-Gln-Arg and the Gln conserved at the 11th position of the GoLoco motifs (supplemental Fig. S6, A and B). The other residues in the aligned sequence of NUCB1 show relatively poor conservation across the GoLoco motif.
We synthesized the NUCB1 peptide corresponding to residue 381-419, namely NUCB1(381-419) (supplemental Fig. S6A). We first performed an ITC experiment to evaluate the interaction of the NUCB1(381-419) with G␣ i1 . The supplemental Fig. S6C shows that NUCB1(381-419) does not interact with G␣ i1 , suggesting that this C-terminal region of NUCB1 alone is not responsible for the observed GDI activity of sNUCB1. To identify further the region of sNUCB1 responsible for its GDI activity, we tested the effect of C-terminal truncation mutant sNUCB1(W333Ter) on the nucleotide exchange by G␣ i1 using [ 35 S]GTP␥S exchange assay. Fig. 5B illustrates the ability of sNUCB1(W333Ter) to inhibit nucleotide exchange on G␣ i1 in a manner similar to sNUCB1. This suggests that the GDI activity of sNUCB1 lies in the N-terminal part of the protein containing the two EF hands. Thus, the GDI activity of sNUCB1 may not depend on a peptide sequence that is similar to a canonical GoLoco sequence.
As a positive control, we also monitored the [ 35 S]GTP␥S uptake by G␣ i1 in the presence of RGS14 GoLoco peptide RGS14(496 -531). RGS14(496 -531) completely inhibits the nucleotide exchange by G␣ i1 (Fig. 5B) as reported by Siderovski et al. (54). Furthermore, using ITC, we tested the binding of RGS14(496 -531) to G␣ i1 both in the absence and presence of sNUCB1. As shown in supplemental Fig. S7A, RGS14(496 -531) (200 M) binds to G␣ i1 (20 M) alone with an affinity of 268.8 nM. However, when G␣ i1 was complexed to sNUCB1, RGS14(496 -531) did not show any binding to G␣ i1 (supplemental Fig. S7B) even at a concentration of 500 M. Similar results were obtained when G␣ i1 was complexed to sNUCB1(W333Ter) (supplemental Fig. S7C). RGS14(496 -531) has been postulated to compete off G␤␥ for G␣ i binding (47). In our experiments, RGS14(496 -531) was not able to bind to G␣ i1 in the presence of sNUCB1 suggesting that both RGS14(496 -531) and sNUCB1 may bind to the same site on G␣ i1 . These results demonstrate that the GDI activity of sNUCB1 does not originate from the conventional GoLoco motif but that it competes for the same general binding site on G␣ i1 .
Physiological Relevance of GDI Activity of sNUCB1-To test the physiological relevance of the observed GDI activity of sNUCB1, we expressed the human WT G␣ i1 in the HEK293 cells in the absence and presence of sNUCB1. We first measured forskolin-stimulated intracellular cAMP levels for each transfection. Forskolin directly activates adenylyl cyclase (AC), which converts adenosine monophosphate to cAMP. Thus the amount of cAMP production can evaluate the ability of G␣ s or G␣ i to stimulate or inhibit cellular AC activity. We used the cAMP assay to measure the effect of GDI activity of sNUCB1 on sNUCB1 Is a Calcium-dependent GDI of G␣ i1 OCTOBER 8, 2010 • VOLUME 285 • NUMBER 41 G␣ i signaling (Fig. 6, inset). Intriguingly, the results show that sNUCB1 decreases G␣ i -mediated AC inhibition after forskolin stimulation.
We further evaluated the effect of sNUCB1 on GPCR-mediated signaling. We transfected HEK293 cells with CXCR4 and monitored cAMP production as a function of SDF-1␣ concentration. SDF-1␣ is an agonist ligand of CXCR4, which activates the receptor to facilitate nucleotide exchange on G␣ i . CXCR4transfected cells show a dose-response curve for cAMP production upon SDF-1␣ stimulation with an EC 50 value of 1 nM.
However, when sNUCB1 was overexpressed along with CXCR4 and cells were stimulated with SDF-1␣, no dose-dependent inhibition was observed. The amount of cAMP produced does not diminish upon stimulation with increasing concentrations of SDF-1␣ (Fig. 6.).

DISCUSSION
NUCB1 is a ubiquitously expressed multidomain Ca 2ϩ -binding protein whose physiological role is not well understood. Our results provide the first direct evidence that sNUCB1 is a Ca 2ϩ -regulated GDI for G␣ i1 . SEC experiments showed that Ca 2ϩ -free sNUCB1 interacts primarily with the GDP-bound state of G␣ i1 and that no interaction occurs when sNUCB1 is bound to Ca 2ϩ (Fig. 3). The role of Ca 2ϩ binding as a regulatory switch for the association of sNUCB1 with G␣ i1 can be explained structurally through the involvement of the acidic region of sNUCB1. The acidic region has been previously shown to be a G protein interaction site (28). Thus binding of Ca 2ϩ may cause structural rearrangement that can occlude the G protein interaction site on sNUCB1. In addition, yeast two-hybrid experiments have suggested that the C-terminally linked ␣5 helix of G protein interacts with NUCB1 (28). Interestingly, the ␣5 helix is conformationally identical in both the G␣ i1 ⅐GDP and G␣ i1 ⅐GTP␥S structures. However, as G␣ undergoes transition from the inactive to the active state, the structurally dynamic switch regions (switch I to IV) of G␣ subunits change conformation (48). Thus, the preferential binding of Ca 2ϩ -free sNUCB1 to G␣ i1 ⅐GDP, which can modulate G protein activation, points toward a possible participation of switch regions in this interaction.
Analysis of the dimer of sNUCB1 using SEC (supplemental Fig. S3) and dynamic light scattering (Fig. 2D) revealed that it is considerably asymmetric in its shape with a long axis of 12.4 nm. This elongated shape may facilitate simultaneous interaction of sNUCB1 with both the ␣5 helix and the switch regions of G␣ i1 ⅐GDP. Interestingly, deletion of the leucine zipper domain along with the C-terminal region of sNUCB1 increased the binding affinity of the monomeric truncation FIGURE 5. Evidence that sNUCB1 acts as a GDI of G␣ i1 . A, we estimated the number of available nucleotidebinding sites on G␣ i1 in the absence or presence of Ca 2ϩ -free sNUCB1 by monitoring the absorbance of G␣ i1 -bound BODIPY FL-GTP␥S at 504 nm. G␣ i1 ⅐GDP alone (20 M, black) or in complex with Ca 2ϩ -free sNUCB1 (100 M, red) was allowed to undergo nucleotide exchange with 100 M BODIPY FL-GTP␥S in the reaction mixture. Samples were withdrawn at different time points, and unbound BODIPY FL-GTP␥S was removed using a micro-spin desalting column. The amount of bound BODIPY FL-GTP␥S was measured using the UV-visible absorbance spectroscopy at each time point. A relative decrease in the number of nucleotide-binding sites on G␣ i1 was observed on association with sNUCB1. This shows that Ca 2ϩ -free sNUCB1 binds to G␣ i1 in its GDPbound form and inhibits nucleotide exchange. We further investigated the guanine nucleotide inhibitory activity of Ca 2ϩ -free sNUCB1 using the radioligand-based binding assay (B). WT G␣ i1 ⅐GDP (20 M) was incubated alone (blue) or with 100 M Ca 2ϩ -free sNUCB1 (green)/sNUCB1(W333Ter) (red) in the presence of [ 35 S]GTP␥S, and samples were withdrawn at various time points to estimate the amount of radioligand bound upon nucleotide exchange. As a positive control, 100 M of a C-terminal GoLoco motif peptide from RGS14, namely RGS14(496 -531), with established GDI activity was used as a positive control (magenta). The data show that both Ca 2ϩ -free sNUCB1 or Ca 2ϩ -free sNUCB1(W333Ter) considerably inhibits nucleotide exchange with similar potency. The GDI peptide RGS14(496 -531) shows complete inhibition of nucleotide exchange. Each experiment was repeated at least three times, and the data were plotted as mean Ϯ S.E. among the independent measurements.
sNUCB1 Is a Calcium-dependent GDI of G␣ i1 mutant sNUCB1(W333Ter) for G␣ i1 ⅐GDP and decreased the R h to 3.3 nm. This suggests that the deleted C terminus of sNUCB1 contributes significantly to the shape asymmetry of sNUCB1 and may play a role in regulating sNUCB1-G␣ i1 ⅐GDP interaction.
NUCB1 is a major Ca 2ϩ -binding protein of Golgi and assists in the uptake, storage, and release of Ca 2ϩ from the Golgi apparatus (49). Optimal Golgi Ca 2ϩ levels are required for anterograde (endoplasmic reticulum to Golgi), retrograde (Golgi to endoplasmic reticulum), and intra-Golgi transport (50). Thus, NUCB1 along with Cab45 (51), the only other known Ca 2ϩbinding Golgi resident protein, may be involved in maintaining the essential Ca 2ϩ levels and gradient from cis-to trans-Golgi. The Ca 2ϩ concentration in cis-Golgi is around 0.3 mM under unstimulated conditions (50). Thus, the NUCB1 pool in cis-Golgi should largely be Ca 2ϩ -bound given the K d values for Ca 2ϩ binding. However, at cytosolic Ca 2ϩ concentrations of 100 -125 nM (52), NUCB1 should essentially be in a Ca 2ϩ -free state. Similar to G␣ i1 , NUCB1 also exists both in the cytosolic and membrane fractions (17), and this creates an opportunity for these two protein subunits to interact and perform physiologically relevant functions. Our ITC experiments along with results from MALS analysis establish the interaction of the G␣ i1 ⅐GDP subunit with a dimer of Ca 2ϩ -free sNUCB1 (Fig. 4), which is a variant of NUCB1 that lacks the N-terminal signal sequence and represents the cytosolic pool of NUCB1.
Our nucleotide-exchange assays with GTP␥S and its fluorescent analogues (supplemental Fig. S4 and supplemental Fig. 5) together show that Ca 2ϩ -free sNUCB1 significantly inhibits nucleotide exchange by G␣ i1 ⅐GDP. The GDI activity of sNUCB1 was established by monitoring exchange of radioligand [ 35 S]GTP␥S. Our results clearly show that sNUCB1 significantly inhibits G␣ i1 activation. Interestingly, this ability was unaffected upon deletion of the C-terminal domain of sNUCB1, showing that the GDI ability of sNUCB1 does not depend on a region of the proteins that resembles a GoLoco motif. Thus, the mechanism of GDI activity of sNUCB1 is likely to be different from the GoLoco motif peptides derived from RGS proteins. The RGS peptides generally have nanomolar binding affinity for G␣ i ⅐GDP in comparison with micromolar binding affinities observed for GPSM2-, PCP-2-, or G18-derived GoLoco motif peptides (44 -46). Unlike these peptide-based studies, our in vitro assays report the binding affinity of 18.3 M for a 99-kDa sNUCB1 dimer with G␣ i1 . Yeast two-hybrid analysis by Lin et al. (16) showed that NUCB1 preferentially interacts with G␣ i1 and G␣ i3 with higher affinity than with G␣ i2 and G␣ s . Further structural studies now underway should reveal the mechanism of GDI activity of sNUCB1 and its relevance to other classes of G protein subunits.
G proteins are widely expressed in different tissue types and are involved in various signal transduction pathways interacting with a number of binding partners. The interaction of NUCB1 with G␣ i1 has been previous established, but no guanine nucleotide exchange factor or GDI activity was observed because the in vitro experimental concentrations used were below the K d values of association. Our results show that sNUCB1 inhibits G protein activation and that Ca 2ϩ binding regulates interaction of G␣ i1 with sNUCB1. Furthermore, to investigate the physiological importance of this interaction, we performed tissue culture-based experiments. Our results show that upon expression in HEK293 cells, sNUCB1 significantly decreases G␣ i -mediated AC inhibition. In addition, even receptor-mediated inhibition of AC was prevented by overexpression of sNUCB1. This can be explained through the possible sequestration of G␣ i subunits by sNUCB1, thereby alleviating inhibition of AC activity. Together with the biophysical studies, this result suggests that sNUCB1 and possibly NUCB1 can affect and regulate downstream signaling events involving heterotrimeric G proteins. In 2009, Lin et al. (17) showed that NUCB1 affects the dynamic distribution of the G␣ subunits to the plasma membrane. It is possible that the association of the two proteins due to colocalization on the secretory granules may facilitate inhibition of G protein activation, which would tend to promote delivery of the G␣ protein subunit primarily in the GDP-bound state to the plasma membrane for heterotrimer formation with G␤␥. Thus the cytosolic pool of NUCB1 may exclusively regulate G protein activation, whereas the membrane-bound pool would regulate both the activation and trafficking to facilitate receptor association upon heterotrimer formation.
In addition, NUCB1 also has a nuclear localization signal sequence within the DNA binding region. Recently, its role in FIGURE 6. sNUCB1 overexpression decreases apparent G␣ i1 -dependent inhibition of AC. The physiological role of interaction of sNUCB1 with G␣ i1 was investigated by monitoring its effect on cellular cAMP production. We transfected HEK293 cells with CXCR4 alone (blue) or with sNUCB1 and CXCR4 (red) cDNA and monitored cAMP production upon ligand (SDF1␣)-mediated receptor activation. To increase basal cAMP levels through AC activation at each ligand concentration, cells were stimulated with 10 M forskolin. For CXCR4 alone (blue), we observe significant inhibition of cAMP production upon stimulation with increasing concentrations of receptor agonist ligand. However, upon co-expression of sNUCB1 with CXCR4, ligand-dependent receptor-mediated inhibition of cAMP production is markedly reduced (red). In the inset, we transfected HEK293 cells with mock (yellow), G␣ i1 alone (orange), and with sNUCB1 and G␣ i1 (green) cDNA, respectively, and then measured cAMP production. The bar in yellow shows the concentration of cAMP produced with mock transfection, and the bar in orange shows cAMP production with overexpression of G␣ i1 alone, and the bar in green shows cAMP production when sNUCB1 was transfected with G␣ i1 . The expression of sNUCB1 results in significant enhancement of cAMP production. The results suggest that sNUCB1 affects the ability of G␣ i1 to promote forskolin-stimulated cAMP production by AC; *, p Ͻ 0.05. sNUCB1 Is a Calcium-dependent GDI of G␣ i1 OCTOBER 8, 2010 • VOLUME 285 • NUMBER 41 affecting the biogenesis of amyloid precursor protein was reported (27). Such a potential for diverse functionality makes the comprehensive understanding of the physiological role of NUCB1 difficult. Based on our current knowledge, NUCB1 seems to be neither an RGS nor an AGS protein. Future experiments, including structural studies, will unravel the functional relevance of the interaction of NUCB1 with other G␣ subunits. Understanding the role of NUCB1 in the context of different G␣ subunits will highlight the signaling pathways modulated by NUCB1.
In summary, our results show that sNUCB1 is a physiological dimer that interacts with the GDP-bound form of G␣ i1 . The binding of sNUCB1 to G␣ i1 inhibits GDP release and suggests that NUCB1 is a GDI. Ca 2ϩ binding to sNUCB1 regulates sNUCB1-G␣ i1 ⅐GDP complex formation and possibly modulates this interaction in different cellular compartments. We conclude that NUCB1 is a novel Ca 2ϩ -binding GDI. The elucidation of the Ca 2ϩ -regulated GDI activity of NUCB1 adds a new level of understanding to the role of NUCB1 and to the complexity of G protein-mediated signal regulation.