Alpha helix content of G protein alpha subunit is decreased upon activation by receptor mimetics.

To elucidate the mechanism whereby liganded receptor molecules enhance nucleotide exchange of GTP-binding regulatory proteins (G proteins), changes in the secondary structure of the recombinant Gi1 alpha subunit (Gi1alpha) upon binding with receptor mimetics, compound 48/80 and mastoparan, were analyzed by circular dichroism spectroscopy. Compound 48/80 enhanced the initial rate of GTPgammaS binding to soluble Gi1alpha 2.6-fold with an EC50 of 30 microg/ml. With the same EC50, the mimetic decreased the magnitude of ellipticity, which is ascribed to a reduction in alpha helix content of the Gi1alpha by 7%. Likewise, mastoparan also enhanced the rate of GTPgammaS binding by 3.0-fold and decreased the magnitude of ellipticity of Gi1alpha similar to compound 48/80. In corresponding experiments using a K349P-Gi1alpha, a Gi1alpha counterpart of the unc mutant in Gsalpha in which Pro was substituted for Lys349, enhancement of the GTPgammaS binding rate by both activators was quite small. In addition, compound 48/80 showed a negligible effect on the circular dichroism spectrum of the mutant. On the other hand, a proteolytic fragment of Gi1alpha lacking the N-terminal 29 residues was activated and showed decreased ellipticity upon interaction with the compound, as did the wild-type Gi1alpha. Taken together, our results strongly suggest that the activator-induced unwinding of the alpha helix of the G protein alpha subunit is mechanically coupled to the enhanced release of bound GDP from the alpha subunit.

The central role played by trimeric GTP-binding regulatory proteins (G proteins) 1 in signal transduction in membranes has received considerable research attention (reviewed in Refs. [1][2][3]. Upon ligand binding, a G protein-coupled receptor promotes the release of GDP from inactive trimeric G␣␤␥, which allows binding of cytosolic GTP to the remaining G␣ subunit, thereby resulting in dissociation of trimeric G␣(GTP)␤␥ complex into active G␣⅐GTP and a ␤␥ subunit complex. In this activation process of G protein, the release of bound GDP is of particular interest as it is the rate-limiting step (4). The analyses of x-ray crystallographic structures of the ␣ subunit of G t and G i1 have indicated the presence of two domains, i.e. a GTPase (or Ras-like) domain comprised of ␣ helices and ␤ strands and a highly ␣ helical domain. In addition, the conformational changes induced in the ␣ subunit by nucleotide exchange (GDP 3 GTP␥S) and the mechanism of GTP hydrolysis have been determined (5)(6)(7)(8)(9). Conformational changes in the ␣ subunit upon binding with a ␤␥ subunit complex have been determined as well (10,11). However, the mechanism whereby liganded receptor molecules enhance the GDP release from the ␣ subunit remains unclear, as pointed out previously (3,12). Likewise, the conformational change of the ␣ subunit upon receptor binding is unknown. The use of physicochemical methods to gain further insight into these key reactions presents difficulties due to the facts that (i) only small amounts of G protein-coupled receptor proteins are expressed in cells, and (ii) no method exists for suitably analyzing the structure of a protein complexed with a large membrane protein.
We considered that mastoparan (MP), a 14-residue peptide discovered in wasp venom as the agent that induces histamine release from mast cells (13), might provide some important clues because it activates G o and G i in a similar manner; its activation is both Mg 2ϩ -dependent and blocked by ADP-ribosylation of G proteins (13,14). In addition, it has an amphiphilic sequence, as do putative G protein-binding sites of many receptors, namely, second and third intracellular loops and C-terminal tail (15). In fact, peptide fragments corresponding to the third intracellular loop of ␤ adrenergic receptors were found to activate G s (Refs. [16][17][18]reviewed in Ref. 19). Also of interest, G o and G i are known to be activated by another histamine releaser that is also amphiphilic, i.e. compound 48/80 (C48/80) (20).
These compounds are particularly useful for analyzing G protein activation when employed as low molecular weight mimetics of receptors. As such, the present study uses circular dichroism (CD) spectra to analyze conformational changes in G i1 ␣ upon interaction with these two compounds. Analysis of CD spectra is an ideal method for determining overall structural changes in proteins. For example, CD measurements of a DNA-binding domain of yeast transcription activator GCN4 estimated that its ␣ helix content increases from 70 -73% to 95-100% upon interaction with DNA containing its binding site (AP-1 site) (21). This estimation was later confirmed by the NMR analysis of the structure in a DNA-free state (22) and the x-ray crystallographic analysis of the structure in a DNAbound state (23). In the present study, the CD analysis of G i1 ␣ allowed us to determine how the interaction affects the ␣ helicity of G i1 ␣ with the resultant conformational changes leading to the enhancement of GDP release.
Preparation of G Protein ␣ Subunits-Because histidine-tagged proteins can easily be purified by affinity purification on Ni 2ϩ -NTA agarose, we prepared G i1 ␣ tagged with 10 histidine residues (His 10 -G i ␣) as well as nontagged full-length G i1 ␣ (FL-G i ␣) as a control. We also prepared a Lys 349 3 Pro mutant in His 10 -G i ␣ (K349P-G i ␣), a mutant corresponding to the unc mutant in G s ␣ (25,26) and is expected to be insensitive to activators, and a 325-amino acid proteolytic fragment of G i1 ␣ lacking the N-terminal 29 residues (⌬N-G i ␣), which allowed us to investigate whether the protein's N-terminal segment is involved in activation.
FL-G i ␣ was expressed in Escherichia coli BL21(DE3) cells harboring the pQE60/G i1 ␣ plasmid (27) and purified as described (28), and His 10 -G i ␣ was prepared as follows. After cloning cDNA of G i1 ␣ using polymerase chain reaction with appropriate primers and QUICK-Clone cDNA (CLONTECH) as a template, the product cDNA was ligated into the NdeI and BamHI sites of pET19b (Novagen), which we modified beforehand by adding an adapter sequence giving a sequence encoding Met-Gly-(His) 10 -(Ser) 2 -Gly-His-Ile-(Asp) 4 -Lys-His at the N terminus of G i1 ␣. Next, the complete coding sequence, the XbaI-BamHI fragment of pET19b-G i1 ␣, was ligated into the XbaI and BamHI sites of pET24a(ϩ) (Novagen) to produce pET24a(ϩ)/His 10 -G i ␣, which was transformed into E. coli BL21(DE3) cells. Finally, His 10 -G i ␣ protein was expressed and purified as described (29). K349P-G i ␣ was prepared by subcloning into the BamHI and SalI sites of pUC19 a BglII-SalI fragment of pET24a(ϩ)/His 10 -G i ␣ corresponding to the 3Ј terminal 490 bp of the entire G i1 ␣ sequence. Then, to substitute Pro for Lys 349 , the generated plasmid was subjected to site-directed mutagenesis (27) using 5Ј-GAGACCACAGTCTGGTAGGTTATT-3Ј as a mutagenic primer; the mutations were confirmed by DNA sequencing. The entire coding sequence was subsequently obtained by ligating the DraI-SalI fragment of pUC19 possessing the mutated sequence of the 3Ј terminal of the G i1 ␣ sequence into the DraI-SalI sites of pET24a(ϩ)/His 10 -G i ␣. The mutant protein was correspondingly expressed and purified like wild-type His 10 -G i ␣.
To prepare ⌬N-G i ␣, His 10 -G i ␣ protein was subjected to limited digestion with endoproteinase Lys-C as described (6). The cleavage site was confirmed by amino acid sequencing (Applied Biosystems 477A Protein Sequencer), and the integrity of the C terminus was determined by Western blot analysis using antiserum specific to the C-terminal 10 residues of G i1 ␣. 3 In addition to these proteins, which are bound by GDP, we also prepared His 10 -G i ␣ in GTP␥S-and GDP⅐AlF 4 Ϫ -bound forms to investigate the effect of bound nucleotide on the secondary structure of the protein. The GTP␥S-bound form was prepared by incubating 10 mg/ml of His 10 -G i ␣ in the GDP-form at 30°C for 5 h in a buffer of 100 mM sodium Hepes (pH 8.0), 1 mM EDTA, 10 mM DTT, 10 mM MgSO4, and 2 mM GTP␥S; the complete conversion was confirmed by demonstrating trypsin resistance (30). His 10 -G i ␣ in the GDP⅐AlF 4 Ϫ -form was correspondingly prepared using a buffer containing 30 M AlCl3 and 10 mM NaF in place of GTP␥S; the complete conversion was confirmed by verifying that this form did not bind [ 35 S]GTP␥S (7). The concentration of all proteins was determined by Amido Black staining using bovine serum albumin as the standard (31).
GTP␥S Binding Assay-Because the effects of MP and C48/80 have been studied mostly on trimeric G proteins reconstituted in phospholipid vesicles, their effects on soluble G i1 ␣ were compared against those determined by CD spectra analysis.
Activator enhancement of the initial GTP␥S binding rate was determined as follows. Binding buffer containing 50 mM sodium Hepes (pH 8.0), 1 mM EDTA, 1 mM DTT, 0.1 mM MgCl 2 , 10% glycerol, and 2 M [ 35 S]GTP␥S (6000 cpm/pmol) with or without activators was preincubated at 30°C for 5 min, G proteins (0.2 M) were added, and the solution was incubated further at 30°C as described (32). At the indicated times, 50-l aliquots were diluted with 0.5 ml of ice-cold 20 mM sodium Hepes (pH 8.0), 1 mM EDTA, 160 mM NaCl, 0.2 mM GTP, filtered on BA85 nitrocellulose filter, washed with 25 mM Tris (pH 8.0), 100 mM NaCl, 25 mM MgCl 2 , and the amount of bound [ 35 S]GTP␥S was determined on a liquid scintillation counter. Glycerol (10%), which stabilizes His 10 -G i ␣, was added to the buffer to prevent precipitation. This addition led to approximately a 2-fold decrease in the GTP␥S binding rate for all proteins, in both the absence and presence of the activators.
GTP␥S binding activity of His 10 -G i ␣ under the CD measurement conditions was determined as follows. His 10  Determination of GDP Bound to G i1 ␣-To examine whether His 10 -G i ␣ is denatured in the presence of C48/80, the amount of GDP bound to G i1 ␣ under the CD measurement conditions was determined according to Refs. 32 and 33. His 10 -G i ␣ (0.3 mM) was incubated in 50 mM sodium Hepes (pH 8.0), 1 mM EDTA, 10 mM DTT, 10 mM MgCl 2 , 2 mM [␣-33 P] GTP (87.6 cpm/pmol) at 30°C for 3 h; bound GTP was hydrolyzed to GDP during this incubation. The binding reaction was quenched by adding 12 mM EDTA (pH 8.0) and the free nucleotide was removed by gel filtration. [␣-33 P]GDP-bound His 10 -G i ␣ (3.5 M, 82.3 cpm/pmol) was then incubated in 20 mM Tris (pH 7.4), 0.1 mM EGTA, 0.1 mM DTT, 10% glycerol in the absence or presence of 100 g/ml C48/80 at 25°C. At the indicated times, 50-l aliquots were withdrawn, diluted with 0.5 ml of ice-cold 20 mM sodium Hepes (pH 8.0), 1 mM EDTA, 100 mM NaCl, 0.3 mM AlCl 3 , 10 mM MgCl 2 , 10 mM NaF and filtered on BA85 nitrocellulose filter, and the bound radioactivity was determined on a liquid scintillation counter.
Circular Dichroism-G protein (3.5 M) was incubated in a buffer of 20 mM Tris (pH 7.4), 0.1 mM EGTA, 0.1 mM DTT, 10% glycerol at 25°C for 10 min unless otherwise indicated. CD spectra were then recorded at 25°C on a J-720 spectropolarimeter (JASCO, Tokyo, Japan) using a cuvette with a path length of 2 mm. For each sample, four scans were accumulated in approximately 3 min. To eliminate contributions by the activator or buffer, the CD spectrum of a protein is shown as the difference spectrum: the spectrum recorded in the presence or absence of an activator minus the spectrum of an activator or buffer alone. Measurements were repeated at least twice using freshly prepared solutions, and the results were fully reproducible. The resultant difference spectra are representative of two or three measurements, which varied less than 1% at 210 nm. The dependence of [Ј] 222 on C48/80 concentration was determined in duplicate twice, and an average value for a single run is presented. Corresponding analysis was not performed using MP as described under "Results." The ␣ helix content of G i1 ␣ was estimated by convex constraint analysis (CCA) (34) and a self-consistent method (SELCON) (35). Both techniques utilized CD data at wavelengths from 200 to 240 nm with the assumption of three independent basis curves. In contrast to methods that use ellipticities at a single wavelength (36,37), these methods are known to estimate secondary structure contents accurately (38).
Secondary Structure Estimation-To predict which ␣ helix is unwound upon interaction with C48/80, secondary structure preference of rat G i1 ␣ was estimated by two independent methods: Garnier-Osguthorpe-Robson (GOR) (39) and neural network (NNPREDICT) (40). Those amino acid sequences in the GTPase-domain that are ␣ helical in the crystal structure of G i1 ␣ in the GDP form (Protein Data Bank accession code, 1GDD) (9) with five residues extended to both the Nand C-terminal directions were analyzed by these methods. The prediction was not performed of the very short ␣N2 helix (residues 20 -23).

RESULTS
Enhancement of GDP Release from G i1 ␣-The effects of MP and C48/80 on GTP␥S binding to FL-and K349P-G i ␣ are illustrated in Fig. 1, A and B, respectively, and Table I summarizes the initial binding rates and fold enhancement of each G protein examined. Note that (i) all four proteins show similar initial GTP␥S binding rates in the absence of the activators; (ii) MP and C48/80 similarly enhance the initial binding rate of FL-, His 10 -, and ⌬N-G i ␣ by 2.5-3.0-fold; and (iii) enhancement is very weak for K349P-G i ␣. Accordingly, the affinity of a GDP molecule for G i1 ␣ was not altered by the addition of the His 10 tag, substitution of Pro for Lys 349 , or the deletion of the Nterminal 29 residues. However, the substitution of Pro for Lys 349 substantially weakened activation by MP and C48/80.
Intactness of His 10 -G i ␣ during CD Measurements-To confirm that His 10 -G i ␣ is not denatured in the presence of C48/80, the amount of GDP bound to His 10 -G i ␣ was determined under the CD measurement conditions. In the CD buffer that did not contain guanine nucleotides, dissociation of GDP did not occur in either the absence or the presence of 100 g/ml C48/80 up to 50 min (Fig. 2). When free GDP was included in the buffer, marked release of GDP was observed, and its release rate was increased in the presence of C48/80 (data not shown). GTP␥S binding activity of His 10 -G i ␣ in the presence of 100 g/ml C48/80 was also determined with different preincubation times with C48/80. The GTP␥S binding activity did not change significantly (Ͻ5%) up to 50 min (data not shown).
␣ Helix Content of G i1 ␣ in the Absence of Activators and Effect of the Bound Nucleotide on the Secondary Structure of G i1 ␣- Fig. 3A illustrates the CD spectrum of FL-G i ␣ in the GDPbound form. CCA and SELCON analysis of the spectrum gave ␣ helix values of 50.6 and 55.9%, respectively; these values were in good agreement with that obtained by x-ray crystallographic analysis of the G i1 ␣⅐GDP structure (9), which should be expected because both CCA and SELCON are known to show high accuracy in estimating ␣ helix content (38). Fig. 3A also illustrates the CD spectra of His 10 -G i ␣ in the GDP-, GDP⅐AlF 4 Ϫ -, and GTP␥S-bound forms. The magnitude of ellipticity is greater in FL-G i ␣⅐GDP than in His 10 -G i ␣⅐GDP. This is presumably due to the absence of an ordered structure in the His 10 tag segment. There were few spectral differences among the three forms of His 10 -G i ␣, indicating that the secondary structure of G i1 ␣ does not substantially change irrespective of the chemical structure of the phosphate moiety of the guanine nucleotides bound (see Fig. 3, B and C, for expansion around 210 and 220 nm, respectively). This is consistent with the x-ray analysis results indicating the presence of few differences among the secondary structure contents of G i1 ␣ in the GDP-, GDP⅐AlF 4 Ϫ -, and GTP␥S-bound forms and among corresponding   forms with Gt␣ (5-9). 4 Effects of Activators on the CD Spectrum of FL-G i ␣-The difference spectra of FL-G i ␣ in the presence of MP (100 M) or C48/80 (100 g/ml) are illustrated in Fig. 3D. Their presence decreased the magnitude of the ellipticity at 205-235 nm, which indicates changes in its secondary structure. The difference spectrum in the presence of MP, however, is not considered to accurately reflect the structure of the protein for the following reason. The MP molecule is known to adopt an ␣ helical conformation when bound to G i1 ␣ (41), although taking no ordered conformation in an aqueous solution (42). When it is considered that the magnitude of negative ellipticity of the ␣ helix is larger than that of random coil from 205 to 240 nm (38) and that the fraction of G i1 ␣-bound MP molecules is only 3.5% at most, this indicates that MP shows negative ellipticity slightly larger in magnitude in the presence of G i1 ␣ than in its absence; accordingly, the magnitude of ellipticity contributed by G i1 ␣ in the presence of MP should be smaller than that shown in Fig. 3D. The conclusion still holds, however, that the magnitude of ellipticity of G i1 ␣ is decreased upon interaction with MP. On the other hand, the difference spectra in the presence of C48/80 are accurate because C48/80 shows no ellipticity from 200 to 250 nm (data not shown); only the difference spectra obtained with this activator were subsequently considered.
In agreement with the observation that GDP molecules are not released from His 10 -G i ␣ in the absence of free guanine nucleotides even in the presence of 100 g/ml C48/80 (Fig. 2), the addition of 50 M GDP did not affect the CD spectra in either the absence or the presence of C48/80 (data not shown). These data confirm that the difference spectrum in the presence of this activator (Fig. 3D) reflects the secondary structure of G i1 ␣⅐GDP in the activated state rather than the structure of guanine nucleotide-free G i1 ␣. Effect of C48/80 on CD of Modified G i1 ␣- Fig. 4, A-C, illustrates difference spectra for His 10 -, ⌬N-, and K349P-G i ␣ with C48/80 (100 g/ml). The CD spectra of His 10 -and ⌬N-G i ␣ show a marked and similar decrease in the magnitude of ellipticity as for FL-G i ␣. This indicates that the His 10 tag segment of His 10 -G i ␣ and the N-terminal 29 residues of FL-G i ␣ do not change their conformations upon interaction with C48/80. This also suggests that His 10 -G i ␣ can be conveniently used as a substitute for FL-G i ␣ for conformation analyses, taking advantage of the fact that the His 10 -G i ␣ can easily be purified in large amounts. To further confirm that the spectral change upon addition of C48/80 is not due to the release of GDP and resultant denaturation of G i1 ␣, the dependence of CD spectrum of His 10 -G i ␣ on the preincubation time with 100 g/ml C48/80 was examined. The [Ј] 222 value did not change significantly up to 50 min (Fig. 4D).
In marked contrast to these wild-type proteins, K349P-G i ␣ shows only a small decrease upon addition of C48/80 (Fig. 4C). This observation indicates that the ␣ helical structure in K349P-G i ␣ is not unwound upon interaction with C48/80. Secondary Structure Prediction of G i1 ␣- Table II shows the predicted secondary structure preferences of those sequences that are ␣ helical in the GTPase domain of G i1 ␣ in the GDP form (9). Among six helices, the ␣5 helix is predicted by both the Garnier-Osguthorpe-Robson and NNPREDICT methods to possess the lowest propensity to form helices.

DISCUSSION
Enhancement of Initial GTP␥S Binding Rate of Soluble G i1 ␣ by MP and C48/80 -The enhancement of GTP␥S binding to soluble G i1 ␣ by MP and C48/80 was 3.0-fold and 2.6-fold, respectively, being markedly lower in comparison with 11-fold (13) and 6-fold (20) enhancements of the release of bound GDP from trimeric G i ␣␤␥ reconstituted in phospholipid vesicles at the same activator concentrations. The observed values, however, are nevertheless significant because the enhancement was saturable (Fig. 1C) and quite small for the K349P, mutant which corresponds to the unc mutant in G s ␣ (Fig. 1B). The greater enhancement for reconstituted G i ␣␤␥ can be attributed to a higher concentration of activators, which have a high affinity for phospholipid membranes (42,43) near the vesicle surface and to an appropriate conformation of the activators induced by lipidic environment (41,44).
The Site of Unwinding-Although MP cross-links to the Cys 3 of G o ␣ (45), the observation that the ⌬N-G i ␣ lacking the Nterminal 29 residues can be activated by MP, as well as by C48/80 (Table I), indicates that the N-terminal residues of the protein are not essential for activation or interaction. In fact, polyclonal antibody directed against the C-terminal nine residues of G i ␣ was able to block MP-stimulated GTPase activity (46), which indicates that MP interacts with the C-terminal portion of G i ␣. When it is considered that receptors interact with the C-terminal portion of G protein ␣ subunits (reviewed in Ref. 47) and that the presumed receptorbinding domain (residues 314 -354 in G i1 ␣) (48) contains the C-terminal ␣ helix (residues 329 -350 in G i1 ␣⅐GDP and 328 -343 in G i1 ␣⅐GTP␥S⅐Mg 2ϩ ), our observations suggest that MP and C48/80 unwind some portion of the C-terminal helix. In agreement with this postulation, the secondary structure prediction of the sequences that are ␣ helical in the GTPase domain of G i1 ␣⅐GDP (Table II) indicates that the C-terminal ␣5 helix possesses considerably lower helix-forming propensities than other helices. Furthermore, the 11-amino acid peptide from the C terminus of G t ␣ is known to adopt an ␣ helical conformation when bound to unexcited rhodopsin, whereas it adopts an extended conformation when bound to photoexcited rhodopsin (49).
Mechanism of Helix Unwinding-Although speculative, the following mechanism is conceivably that of the helix unwinding. On the same side of the ␣5 helix of G i1 ␣⅐GDP (9), Asp 337 and Asp 341 are adjacently located and are solvent accessible. The electrostatic repulsion between the two negative charges on these residues would destabilize the helix structure if they were not neutralized by some positive charges. Actually, the two carboxylate groups of these residues form a bidentate salt bridge with the side chain amino group of Lys 192 in the ␤2/␤3 loop (9). This bidentate salt bridge seems to stabilize the potentially unstable ␣5 helix structure and, by bridging the ␣5 helix and the ␤2/␤3 loop, the whole tertiary structure as well. Such a salt bridge is formed also in G t ␣⅐GDP (6) and is expected to occur in G s ␣ (50), G o ␣ (51), and G z ␣ (52), as well. Because both MP and C48/80 have multiple positive charges, interactions of each positive charge with either Asp 337 or Asp 341 may result in the destabilization and subsequent unwinding of the ␣5 helix. Such an interaction is compatible with the observation that [Tyr 3 ,Cys 11 ]MP is cross-linked with Cys 3 of G o ␣ (45). Although the N-terminal eight residues are disordered and are not observed in the crystal structure of G i1 ␣⅐GDP (9), Asp 9 is located very close to Asp 341 and, therefore, to Asp 337 .
Coupling of Helix Unwinding and Enhanced GDP Release-C48/80 enhanced GTP␥S binding to G i1 ␣ with nearly the same EC 50 as it decreased [Ј] 222 of G i1 ␣, i.e. 33.0 Ϯ 1.5 g/ml (Fig.  1C) and 31.8 Ϯ 2.6 g/ml (Fig. 3E), respectively. In addition, the K349P mutant in G i1 ␣, which is activated minimally by C48/80 and MP (Fig. 1B), demonstrated an insignificant decrease in the magnitude of ellipticity in the presence of C48/80 (Fig. 4C). Taken together, these results strongly suggest that the decrease in the ␣ helix content of G i1 ␣ is coupled to the enhanced GTP␥S binding, i.e. enhanced GDP release. In other words, the unwinding of ␣ helical residues (presumably in the ␣5 helix) upon binding with C48/80 (or MP) is considered to be propagated to the guanine nucleotide binding sites of G i1 ␣ such that the affinity of the protein to a bound GDP molecule is lowered. The disruption of the bidentate salt bridge would release the ␤2 sheet (residues 185-191) from the ␣5 helix and result in the dislocation of guanosine-binding residues (Leu 175 and Arg 176 ) positioned to the N terminus of the ␤2 sheet; ultimately, this would decrease the affinity to the bound GDP. In support of this postulation, the affinity of GDP to G o ␣ is known to decrease remarkably by the removal of the C-terminal 14 residues of the protein (including Asp 341 ) (53), which adopt an ␣ helical conformation in G i1 ␣⅐GDP (9).
In summary, the present study reports for the first time the conformational change of G protein ␣ subunit upon activation by receptor mimetics. Additional studies should prove MP and C48/80 useful in elucidating the interaction between receptor and G protein ␣ subunits in more detail. II Secondary structure prediction of the sequences that are ␣-helical in the crystal structure of G i1 ␣ ⅐ GDP Those sequences that are ␣-helical in the crystal structure of the GTPase domain of rat G i1 ␣ in the GDP form (9) with five amino acid residues extended both to the N-and C-terminal directions were subjected to secondary structure predictions by GOR and NNPREDICT methods (see "Experimental Procedures").