A Conserved Phenylalanine as a Relay between the α5 Helix and the GDP Binding Region of Heterotrimeric Gi Protein α Subunit*

Background: GPCRs regulate heterotrimeric G protein activation. However, the intermediate steps regulating GDP release are still unknown. Results: Energy analysis pinpoints information flow through Gα-receptor interaction and GDP release. Conclusion: Hydrophobic interactions around α5 helix, β2-β3 strands, and α1 helix are key for GDP stability. Significance: G protein activation defines regulation of high affinity receptor interactions and plays a role defining different cellular responses. G protein activation by G protein-coupled receptors is one of the critical steps for many cellular signal transduction pathways. Previously, we and other groups reported that the α5 helix in the G protein α subunit plays a major role during this activation process. However, the precise signaling pathway between the α5 helix and the guanosine diphosphate (GDP) binding pocket remains elusive. Here, using structural, biochemical, and computational techniques, we probed different residues around the α5 helix for their role in signaling. Our data showed that perturbing the Phe-336 residue disturbs hydrophobic interactions with the β2-β3 strands and α1 helix, leading to high basal nucleotide exchange. However, mutations in β strands β5 and β6 do not perturb G protein activation. We have highlighted critical residues that leverage Phe-336 as a relay. Conformational changes are transmitted starting from Phe-336 via β2-β3/α1 to Switch I and the phosphate binding loop, decreasing the stability of the GDP binding pocket and triggering nucleotide release. When the α1 and α5 helices were cross-linked, inhibiting the receptor-mediated displacement of the C-terminal α5 helix, mutation of Phe-336 still leads to high basal exchange rates. This suggests that unlike receptor-mediated activation, helix 5 rotation and translocation are not necessary for GDP release from the α subunit. Rather, destabilization of the backdoor region of the Gα subunit is sufficient for triggering the activation process.

G protein activation by G protein-coupled receptors is one of the critical steps for many cellular signal transduction pathways. Previously, we and other groups reported that the ␣5 helix in the G protein ␣ subunit plays a major role during this activation process. However, the precise signaling pathway between the ␣5 helix and the guanosine diphosphate (GDP) binding pocket remains elusive. Here, using structural, biochemical, and computational techniques, we probed different residues around the ␣5 helix for their role in signaling. Our data showed that perturbing the Phe-336 residue disturbs hydrophobic interactions with the ␤2-␤3 strands and ␣1 helix, leading to high basal nucleotide exchange. However, mutations in ␤ strands ␤5 and ␤6 do not perturb G protein activation. We have highlighted critical residues that leverage Phe-336 as a relay. Conformational changes are transmitted starting from Phe-336 via ␤2-␤3/␣1 to Switch I and the phosphate binding loop, decreasing the stability of the GDP binding pocket and triggering nucleotide release. When the ␣1 and ␣5 helices were cross-linked, inhibiting the receptor-mediated displacement of the C-terminal ␣5 helix, mutation of Phe-336 still leads to high basal exchange rates. This suggests that unlike receptor-mediated activation, helix 5 rotation and translocation are not necessary for GDP release from the ␣ subunit. Rather, destabilization of the backdoor region of the G␣ subunit is sufficient for triggering the activation process.
Heterotrimeric G proteins play a critical role as molecular switch proteins that couple the activation of cell surface receptors, G protein-coupled receptors (GPCRs), 2 to different intracellular effector proteins mediating intracellular responses. Therefore, G proteins have a crucial role in defining the specificity and temporal characteristics of many different cellular responses (1)(2)(3)(4)(5).
Several structural and biophysical studies have proposed the conformation of the receptor in its active state and have identified potential receptor-mediated mechanisms for G protein activation and GDP release (6 -16). Two well studied receptormediated G protein activation routes have been hypothesized. In the first, the binding of the GPCR to the C terminus of G␣ is thought to trigger conformational changes that can be transmitted via rotation of the ␣5 helix of G␣ to the ␤6-␣5 turn on the purine ring of the GDP ( Fig. 1) (3,9,(17)(18)(19). In the second proposed mechanism, the GPCR is thought to take advantage of G␤␥ as a nucleotide exchange factor to disrupt the phosphate interactions of the nucleotide binding pocket via destabilization of switch I-II regions through perturbing ␣5 interaction with the ␤2-␤3 strands ( Fig. 1) (20 -25).
In 2011, Kobilka and co-workers (18) provided an important missing piece of the puzzle in the receptor-mediated G protein activation cycle by determining the structure of the ␤ 2 -adrenergic receptor-G s heterotrimer complex structure. This groundbreaking study detailed the receptor-G protein interaction and G protein activation. This structure represents the end point in the signal transduction step. The signaling route by which an active receptor interacts with an inactive G protein and causes conformational changes that lead to the final high affinity complex of a receptor with its cognate G protein and GDP release is still unknown.
To address the conformational dynamics underlying nucleotide release from the G␣ subunit, we recently generated a pre-dictive computational model of the energy of receptor activation with the goal of understanding conformational changes and connections between potential key residues during G protein activation (26). In this model of the rhodopsin-G i ␣␤␥ complex, it was suggested that the ␣5 helix is the most critical region for G protein stability and activation and is consistent with previous studies (9,12,14,15,27). The ␣5 helix is protected and surrounded with primarily hydrophobic interactions within six ␤ strands (␤1-␤6) and one ␣ helix (␣1) (Fig. 1, C and D). Energetic analysis predicted that residues Phe-191 and Phe-196 in ␤2-␤3, Ile-265 and Phe-267 in ␤5, Tyr-320 and His-322 in ␤6 strands, and Gln-52 and Met-53 in the ␣1 helix are making critical interactions with the ␣5 helix in both basal and receptor-mediated G protein activation (26). These key residues might either be important for the overall structural integrity of the GTPase domain during the activation process, or they may be directly involved in activation.
To identify the residue-residue interactions that are critical for activation as a part the signaling pathway, we systematically tested the effects of these residue-residue interactions on G protein activation. The residues were examined using biochemical, computational, and structural approaches in both basal and receptor-bound states. In this study, recombinant G␣ i1 was used for all experiments instead of visual G protein, given that G␣ i is a very close homolog of G␣ t yet much more easily expressed in Escherichia coli. Our data showed that single mutations in the ␤5 and ␤6 strands that face the ␣5 helix were not able to break hydrophobic interactions and trigger GDP release from G protein in both receptor-bound and unbound states. In the receptor-bound state, using pairwise coupling energy analysis, we predicted that the ␣5 rotation compensates the effect of ␤5-␤6 mutations on protein activation.
However, the hydrophobic interactions on the opposite side of the ␣5 helix were predicted to directly affect G protein function. Energetic analysis predicted that Phe-336 is the most critical residue in the ␣5 helix; it creates a hydrophobic hot spot of G protein activation, consistent with previous studies (14,28,29). The amplitude of this effect was correlated with decreasing hydrophobicity of the side chain. Experimentally tracing the hydrophobic interactions around the Phe-336 residue together FIGURE 1. Heterotrimeric G protein; localization and function ␣5 helix in G proteins. A, ribbon model of heterotrimeric G protein (G i ␣␤␥, Protein Data Bank entry 1GP2). The G␣ subunit is composed of nucleotide binding (GTPase domain, light blue) and helical domains (green). The ␣5 helix and switch (SW) regions are colored yellow and purple, respectively. GDP is shown as sticks. B, receptor-mediated (orange) G protein activation routes. The binding of the GPCR to the C terminus of G␣ is thought to trigger conformational changes that can be transmitted via rotation of the ␣5 helix (black, arrow 1) of G␣ to the ␤6-␣5 loop (purple, arrow 2) that binds the purine ring of the GDP. In the second route, disruption of the phosphate interactions with the nucleotide binding pocket via destabilization of switch I-II regions through perturbing ␣5 interaction with the ␤2-␤3 strands (arrow 3). The rhodopsin-G i complex model adapted from Alexander et al. (26). C and D, the ␣5 helix is one of the most critical regions for G protein stability and activation. A and B, the ␣5 helix (yellow) is protected by six ␤ strands (␤1-␤6) and one ␣ helix (␣1) (green). The structure is adapted from the crystal structure of the G i heterotrimer (Protein Data Bank entry 1GP2).
with computational analysis provided evidence for a dynamic interplay between Phe-336, the ␤2 and ␤3 strands, and the ␣1 helix on the G protein activation route.

EXPERIMENTAL PROCEDURES
Materials-The TSKgel G2000SW column, GDP, and GTP␥S were purchased from Sigma. All other reagents and chemicals were of the highest available purity.
Rosetta Interface Energy Calculations-Interface energies were computed following the Rosetta ⌬⌬G protocol previously described (26). Briefly, we leveraged the previously published ensembles of 10 structures of the G protein in the basal state and receptor-bound state. Residue-residue interactions across ␣1 helix/GTPase domain interface were evaluated by measuring energetic perturbations when computationally removing the ␣1 helices from the models. The ␣1 helix was defined as residues 45-58. For all analyses, GDP remained fixed within the nucleotide binding pocket. The ⌬⌬G value is reported as an average over the 10 structural models in Rosetta energy units (REU). Absolute values larger than 0.5 REU are considered to be significant. Using the standard deviation over the 10 structures, a Z-score was computed. The total ⌬⌬G value across the interface is calculated as the sum of individual residue contributions.
Rosetta Pairwise Binding Energy Calculations-Average energies between pairwise interacting residues were computed using Rosetta's per residue energy breakdown protocol. The energy between all possible pairs of interacting amino acid residues within the G protein were calculated across the previously published ensembles of 10 structures (26). These energies between all residues pairs were then averaged across the 10 models in both the receptor-bound and basal state. Predicted energy values are reported in REU and considered significant if greater than 0.5 REU.
Preparation of Urea Washed Rod Outer Segment Membranes and G␤ 1 ␥ 1 -Urea washed rod outer segment membranes and G␤ 1 ␥ 1 were prepared from bovine retina as described previously (30,31).
Construction, Expression, and Purification of Proteins-Briefly, the pSV277 expression vector encoding G␣ i1 with N-terminal His tag served as the template for introducing individual mutant substitutions using the QuikChange system (Stratagene). All mutations were confirmed by DNA sequencing (DNA Sequencing Facility, Vanderbilt University). The mutant constructs were then expressed and purified as previously described (32). The purified proteins were cleaved with thrombin (Sigma; final concentration, 0.5 unit/mg) for 16 h at 4°C to remove the N-terminal His tag. The sample was then loaded onto a nickel-nitrilotriacetic acid column to separate the protein from the cleaved His tag and any uncleaved fraction. For further purification, the protein solution was loaded onto a size exclusion column (TSKgel G2000SW) that was equilibrated in buffer A (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 2 mM MgCl 2 , 40 M GDP (or 1 M GTP␥S), 2 mM DTT, and 100 M PMSF). SDS-PAGE was used to test the purity of the proteins. Protein concentrations were determined by Bradford assay (33).
Nucleotide Exchange Assay-The basal rate of GTP␥S binding was determined by monitoring the relative increase in the intrinsic tryptophan (Trp-211) fluorescence (ex 290 nm, em 340 nm) of G␣ i1 (200 nM) in buffer containing 50 mM Tris (pH 7.2), 100 mM NaCl, and different amounts of MgCl 2 for 60 min at 25°C after the addition of 10 mM GTP␥S. Receptor-mediated nucleotide exchange was determined with G␤1␥1 (400 nM) in the presence of 50 nM rhodopsin at 21°C for 60 min after the addition of GTP␥S. The data were normalized to the baseline and maximum fluorescence and then fit to the exponential association equation (Y ϭ Y max * (1 Ϫ e Ϫkt )), to calculate the rate constant (k) as previously described (9).
Intrinsic Trp Fluorescence Assay with AlF-Intrinsic tryptophan (Trp-211) fluorescence upon AlF 4 Ϫ activation, relative to emission in the GDP bound state of G protein ␣ subunit, was monitored as previously described (34). The data represent the averages from six to eight experiments.
Trypsin Digestion and Analysis-2 g of G␣ i1 were incubated in buffer containing 50 mM Tris (pH 7.5), 100 mM NaCl, 20 M GDP, and different amounts of MgCl 2 (0.5, 1, and 2 mM). 10 mM NaF and 50 M AlCl 3 were added to samples and then incubated for 2 min at 25°C. One microliter of a 1 mg/ml TPCK trypsin solution was added and incubated on ice for 25 min. The reaction was stopped by adding 2.5 l of termination solution (10 mg/ml aprotinin, 10 mM PMSF). Subsequently, samples were boiled with Laemmli sample buffer for 5 min and run on a 12.5% SDS-polyacrylamide gel, stained with Coomassie Blue, and quantified by densitometry (Multimager; Bio-Rad) (30,35,36).
Cross-linking-An expression vector encoding G␣ i1 with six amino acid substitutions at solvent-exposed cysteines (G␣ i1 HI) and an internal His 6 tag between residues Met-119 and Thr-120 served as the template for introducing individual cysteine substitutions using the QuikChange system (Stratagene) as describe above. The bifunctional cross-linking reagent bismaleimidoethane (Pierce) was incubated in a 2:1 molar ratio with G␣ i1 HI as previously described (37). The concentrated, cross-linked monomeric protein was then purified by size exclusion chromatography on a calibrated G2000SW column. Calibration was performed under the same conditions as purification, using a broad range of molecular weight standards (Bio-Rad) (37).
Membrane Binding Assay-The ability of mutant G␣ subunits to bind rhodopsin in urea-washed rod outer segment membranes was determined as previously described (9). Each sample was evaluated by comparison of the amount of G␣ i1 subunit within the pellet or supernatant to the total amount of G␣ i1 subunit in both treatments expressed as a percentage of the total G␣ i1 protein. The data represent the averages of three experiments.
Protein Crystallization, Data Collection, and Structure Determination-Purified GDP bound G␣ subunits were exchanged into crystallization buffer (50 mM EPPS, pH 8.0, 1 mM EDTA, 2 mM MgCl 2 , 5 mM DTT, 1 mM GDP) using a size exclusion chromatography column. The appropriate fractions were pooled as described above, and SDS-PAGE was used to assess to test the purity of the proteins. Crystals were grown by the hanging drop vapor diffusion method at 18°C by equilibration against a reservoir solution containing 2.0 -2.3 M (NH 4 ) 2 SO 3 and 100 mM sodium acetate (pH 5.9 -6.4). Proteins (10 mg/ml) were mixed 1:2.5 ratio with reservoir solution and crystals appeared after 14 -18 days with in the space group I4. A similar strategy was used to grow crystals in GTP␥S bound form of G␣ i1 proteins. Proteins were incubated with 10 M GTP␥S for 30 min on ice and then storage buffer replaced the crystallization solution containing 50 M GTP␥S instead of GDP. G␣ i1 -GTP␥S samples crystallized in the space group P3 2 21. Crystals were cryo-protected prior to data collection by briefly soaking in stabilization solution containing 18% glycerol and 2.4 M (NH 4 ) 2 SO 3 for ϳ30 s and cryo-cooled by immersion in liquid nitrogen.
Data sets were collected at the Life Sciences Collaborative Access Team (21-ID-G) of the Advanced Photon Source at Argonne National Laboratory at Ϫ180°C using a wavelength of 0.98 Å on a MAR CCD detector. The data were processed and scaled using the HKL2000, CCP4, and Phenix suites (38 -40). Crystallographic data processing and refinement statistics are reported in Table 1. Criteria for data cutoff were a combination of R sym and I/, which both rose to unacceptable levels if the resolution were extended by G i ␣. The structures of the G␣ i1 -GDP and G␣ i1 -GTP␥S complexes were determined by molecular replacement using 1GDD (WT G␣ i1 -GDP) (41) and 1GIA (G␣ i1 -GTP␥S⅐Mg 2ϩ ) (42) as search models for Phaser-MR in the Phenix suite (40). Because 1GDD and 1GIA preceded the requirement for deposition of structural factors R free reflections were randomly selected for F336C variant and was the same as F336Y. As a result, the free R is of limited utility. Model building was performed in Coot (43)

RESULTS
In this study, our strategy was to test residues around the ␣5 helix that were previously identified as critical for the function of this helix during G protein activation. Residues were examined using biochemical, computational, and structural approaches in both basal and receptor-bound states.
The Effects of ␤5-␤6 Strand Mutants on G Protein Activation-In our previous study, we proposed four residues that face the ␣5 helix in ␤5 (Ile-265 ands Phe-267) and ␤6 (Tyr-320 and His-322) (26). Any one of these might be critical for ␣5 helix stability and therefore the G protein activation ( Fig. 2A and Ref. 26). To test the effect of these residues on G protein function, we evaluated nucleotide exchange rates after introduction of site-directed mutations. Basal and receptor-mediated nucleotide exchange rates of mutants were determined by monitoring the relative increase in the intrinsic tryptophan (Trp-211) fluorescence of G␣ i1 . All of the mutants showed similar nucleotide exchange rates compare with WT G␣ i1 in both receptorbound and unbound states (Fig. 2B). The simplest way to explain these data would be that those residues do not play a major role in G protein activation or that a single mutation is not enough to disturb the ␣5 helix for GDP release. However, when we computed pairwise residue interactions, we identified interesting details for receptor-mediated activation. In the basal state, Ile-265, Phe-267, Tyr-320, and His-322 were inter- , where I is running over the number of independent observations of reflection hkl, and N is representing the number of replicate observations.
where F o and F c are the observed and calculated structure factor amplitudes. R free is the same as R work for a set of data omitted from the refinement. e Ramachandran analysis from MOLPROBITY (53).
acting hydrophobically with Val-339, Val-335, Val-342, and Val-335, respectively, within the ␣5 helix. After receptor interaction and ␣5 helix rotation, the same residues in ␤5 and ␤6 were predicted to hydrophobically interact with new sets of residues in the ␣5 helix that were previously pointing toward solvent and not involved in binding in the basal state. Specifically Ile-265, Phe-267, Tyr-320, and His-322 started to interact with Ala-338, Asn-331, Ala-338, and Phe-334, respectively (Fig.  2, C and D; see supplemental Table S1 and supplemental Movies S1-S3 for full data). The ␣5 helix can glide along this hydrophobic surface during its rotation. These calculations thus suggested how new interactions on the rotated the ␣5 helix can possibly compensate for the effect of single mutations in ␤5 and ␤6 strands during receptor-mediated G protein activation.
The Effects of Phe-336 Mutants on G Protein Activation-To test the role of interactions with the opposite site of the ␣5 helix post-rotation, we focused on one specific residue in the ␣5 helix, Phe-336. Phe-336 is one of the highly conserved residues in the G␣ protein family as well as the small GTPases. The side chain faces the ␤1, ␤2, and ␤3 strands as well as the ␣1 helix, which creates one of the conserved hydrophobic clusters in the G␣ subunit. Our previous energetic study predicted that Phe-336 is the most critical residue for both basal and receptormediated G protein activation within the ␣5 helix (Fig. 3, A and   B) (26). To test the effect of mutating this residue, we substituted Phe-336 with residues with decreasing hydrophobicity. All of the Phe-336 mutants displayed increased basal exchange rates compared with WT (Fig. 3C). Furthermore, a strong correlation was identified between the hydrophobicity of this residue and basal activity (Fig. 3E). The fastest nucleotide exchange rate was detected for F336Y. However, in receptormediated activation, the nucleotide exchange rates were decreased compared with WT without any correlation with hydrophobicity (Fig. 3, D and F; supplemental Movies S2 and S4). This result is consistent with a rotation of ␣5, leading to a new surface-exposed location of Phe-336 during ␣5 helix rotation and translation caused by interaction with the receptor (26). Overall, these data suggest that Phe-336 is one of the critical control points that regulate GDP release during G protein activation.
The Effects of Phe-336 Mutations on ␤6-␣5 Loop; Cross-linking ␣1 and ␣5 Helices-The most obvious connection between the ␣5 helix and the nucleotide binding pocket is the ␤6-␣5 loop. Perturbation of the ␣5 helix during receptor-mediated activation would disturb the interaction between the ␤6-␣5 loop and the guanine ring of the nucleotide, leading to destabilization of the GDP in its binding pocket and domain opening of the ␣ subunit. To test the effect of Phe-336 mutations on this  (52). C, most favorable interactions between the ␣5 helix (Val-335, Val-339, and Val-342), ␤5 strand (Ile-265 and Phe-267), and ␤6 strand (Thr-320 and His-322) interface in the basal state. D, after receptor interaction and ␣5 helix rotation (arrow), the same residues in ␤5 and ␤6 were hydrophobically interacting with new residues in the ␣5 helix (red). Please see supplemental Table S1 and supplemental Movies S1-S3 for full interactions in both receptor-bound and unbound states. loop, we cross-linked ␣1 to ␣5 to minimize the disruption of its interactions with the guanine ring by translocation toward the receptor. Cross-linking was performed between I56C/T329C residues on a cysteine-depleted G␣ i1 (G␣ i1 HI) protein (Fig.  4A). Without cross-linking, G␣ i1 HI I56C/T329C showed higher basal nucleotide exchange rates compared with the G␣ i1 HI protein (Fig. 4B, black bars). Moreover, as expected, substitution of Phe-336 for Cys on G␣ i HI I56C/T329C further increased the protein activity. After cross-linking, the nucleotide exchange rate of cross-linked G␣ i1 cross-linking HI I56C/ T329C was decreased as compared with un-cross-linked proteins, demonstrating the stabilizing effect of the cross-linking. Substitution of F336C on cross-linked G␣ i1 HI I56C/T329C increased basal protein activation as compared with the un-cross-linked G␣ i1 HI I56C/T329C-F336C mutant (Fig. 4B,  black bars). This indicates that perturbation of Phe-336 can trigger the activation mechanism without translocation of ␣5 toward the receptor and disruption of ␤6-␣5 loop region.
Because receptor-mediated activation causes both a rotation of the ␣5 helix and an uncoiling of one turn of helix, we expected the cross-linked G␣ would be resistant to receptormediated activation. This is indeed what was found in both cross-linked proteins (Fig. 4B, gray bars). This result might be caused by the reduced capability of cross-linked G␣ to interact with either G␤␥ subunits or the receptor. To test the first possibility, we measured the basal nucleotide exchange rates of G␣ mutants in the presence or absence of G␤␥ subunits (Fig. 4C). The results showed that basal nucleotide exchange rates  (26). C and D, basal (C) and receptor-mediated (D) nucleotide (Nuc.) exchange (Exc.) rates of G␣ i1 Phe-336 mutants. The data were normalized to the baseline and maximum fluorescence and then fit to the exponential association equation (Y ϭ Y max * (1 Ϫ e Ϫkt )) to calculate the rate constant (k). The data were collected at 21°C for 60 min. The results represent the means Ϯ S.E. values of at least three independent experiments. E and F, correlation between nucleotide exchange rates and hydrophobicity identity of the amino acids in basal (E) and receptor-bound (F) state. The Engelman scale was used during comparison and correlation coefficients were calculated with or without F336Y mutant data. The Pearson correlation in the basal state with F336Y is 0.9358; without F336Y, it is 0.9945. In the receptor-mediated state, the Pearson correlation with F336Y is 0.6992; without F336Y, it is 0.4861. decreased on both cross-linked and un-cross-linked mutant G␣ proteins in the presence of the G␤␥ subunit, like the WT protein. This suggested that cross-linked G␣ subunits were still capable of interacting with G␤␥ subunits. To test the receptor binding capability of mutant G␣ i1 subunits, we determined the effect of cross-linking on the membrane association of the G protein with light-activated rhodopsin, a measure of the formation of the high affinity receptor-G protein complex. As expected, cross-linking between ␣1 and ␣5 impaired this membrane binding (Fig. 4, D and E), consistent with a lack of ability of the cross-linked ␣5 helix to translocate toward the receptor and the decreased nucleotide exchange rates. Overall, the cross-linking data suggest that perturbation of Phe-336 triggers GDP release through destabilization of switch I-II regions via perturbing the ␣5 helix interactions along the ␣1 helix and ␤2-␤3 strands rather than disrupting the ␤6-␣5 loop region.
Hydrophobic Interactions around Phe-336: ␣1 Helix Interface Binding Energy and G Protein Activation-Previous data suggested that the interaction of Phe-336 with the ␣1 helix and ␤2-␤3 strands might be crucial for domain opening because the ␣1 helix is positioned at the interface of the G␣i-GTPase domain and the helical domain (25). In addition, the ␣1 helix and ␤2-␤3 strands interact with the phosphate binding loop and switch I-II, respectively. To probe the effects of hydrophobic interactions around Phe-336 with the ␣1 helix, we computed interaction energies for all residues within the ␣1 helix in both basal and receptor-bound states of the heterotrimeric Gi␣␤␥ using our established protocol (26). These ⌬⌬G values probed for a potential network of intramolecular interactions that could propagate the conformational changes necessary for G protein activation and nucleotide exchange. ⌬⌬G calculations predicted the importance ofinteractions between the  Table 2). This pairwise interaction couples with Phe-336 on the ␣5 helix. Other predicted stabilizing interactions between ␣1 (Gln-52 and Ile-56) keep the ␣5 helix (Thr-329) fixed in the receptor unbound state; receptor interaction triggers unwinding of a turn of the ␣5 helix, disturbing this interaction ( Fig. 5B and Table 2). On the face of ␣1, in contact with the helical domain, residues (Lys-51, Lys-54, Ile-55, Tyr-61, and Leu-175) on both the ␣1 and ␣F helices assist to secure the helical domain in a "closed" GDPbound conformation. The total interaction energy was ϳ25.4 REU. In the basal state, the ␣1 helix was predicted to interact favorably with ␤2-␤3 (Phe-189, Met-198, and Asp-200; 3.59 REU), ␣5 (Val-332 and Phe-336; REU 2.44), and helical domain (Glu-65 and Leu-175; 1.84 REU) ( Table 2). In the receptorbound state, the ␣1 helix was predicted to interact favorably with ␣5 (Asn-331 and Val-332; 2.1 REU), and as expected, the overall interaction was calculated as lower than the unbound state (Table 2).
To test our computational results, we mutated two residues that are predicted to stabilize the ␣1-␣5 interaction (Phe-189 and Phe-191). In the basal state, F189C increased nucleotide exchange 5-fold, whereas F191C showed no change relative to WT G␣ i1 (Fig. 5C). We prepared double and triple mutants with M53C and F196C mutants, which we had previously tested (26). Double mutants (M53C/F189C and F189C/F196C) exhibited similar basal activation and a triple mutant (M53C/F189C/ F196C) showed an even higher basal exchange rate compared with the F336C G␣ i1 mutant protein (Fig. 5C). In receptormediated activation of exchange, there was again a pattern of only modest inhibition, with F191C showing the largest decrease (Fig. 5D) consistent with previously predicted ␣5 (26) and ␣1 interface binding energy calculations.
Perturbation of Phosphate Site of Nucleotide Binding Region with Phe-336 Mutants-To determine whether the hydrophobic pocket around Phe-336 was necessary to control the local order of the phosphate binding region of GDP, we used the sensitive monitor of Mg 2ϩ binding into this region. Three different strategies were used to investigate the influence of G␣ i1 mutants on Mg 2ϩ binding to this region: (a) [Mg 2ϩ ] effects on the kinetics of nucleotide exchange, (b) AIF 4 Ϫ binding, and (c) trypsin digestion of G␣ i1 in the presence of different concentrations of Mg 2ϩ . The results showed that the high nucleotide exchange rates of the mutants could be decreased in elevated Mg 2ϩ concentrations (Fig. 6, A and B), suggesting that these mutations had allosteric effects on the phosphate binding region that could be overcome with higher Mg 2ϩ concentration. The highest decrease in the rate of exchange, as a function of increasing concentrations of Mg 2ϩ , was observed for the F336Y mutant, which showed the fastest exchange rate in the presence of low Mg 2ϩ concentrations (Fig. 3C).
To investigate the order of the Mg 2ϩ binding region in the presence of GDP, the AIF 4 Ϫ binding assay was used. In this assay, changes in intrinsic tryptophan fluorescence rates of G␣ i1 were measured upon AIF 4 Ϫ addition in the presence of different MgCl 2 concentrations. Mg 2ϩ is necessary for AIF 4 Ϫ binding and generation of the active or transition state. Thus, this assay reflects both AIF 4 Ϫ and Mg 2ϩ coordination in that region without nucleotide exchange. All mutations showed destabilization effects that were overcome with increasing Mg 2ϩ concentration. The EC 50 for Mg 2ϩ stabilization of AIF 4 Ϫ binding for F336M, F336C, F336A, and F336Y was increased by 1.4-, 2.1-, 2.8-, and 3.1-fold, respectively, over the WT G␣ i1 under the same experimental conditions (Fig. 6C). In addition to the ␣5 helix mutants, the M53C/F189C/F196C mutant also exhibited statistically significant increased EC 50 (Fig. 6C).
The sensitivity of the G␣ i1 mutants to the trypsin digestion assay is a complementary assay to show the subtle changes in local order at the trypsin digestion site at Arg-208 in the presence of varying Mg 2ϩ concentrations. After activation by either GDP-AIF 4 Ϫ or GTP␥S, G␣ i1 yields a ϳ34-kDa fragment following trypsin digestion. All high nucleotide exchange mutants had reduced stability as assayed by decreased 34-kDa fragment in the presence of low Mg 2ϩ concentrations compared with the WT G␣ i1 subunit (Fig. 6D).
Structural Features of the X-ray Structures of the F336C and F336Y Mutants-To probe the structural basis for the increased rates of nucleotide exchange observed in the Phe-336 mutants, the crystal structures of the F336C and F336Y variants of the G␣ i1 subunit were determined in both the GDP-and the GTP␥S-bound states. The data collection and refinement statistics are summarized in Table 1. The mutations in the protein were confirmed by the crystal structure, where electron density at position 336 corresponded to either cysteine or tyrosine (Fig.  7, A and B). The structures of the GDP-bound form of F336C and F336Y G␣ i1 were refined to 2.0 and 2.4 Å resolution, respectively. Both GTP␥S-bound structures were refined to a resolution of 2.0 Å. The GDP-and GTP␥S-bound structures for F336C and F336Y were determined in space groups identical to those of the WT G␣ i1 structures. Neither mutant showed significant structural differences compared with WT G␣ i1 . Even with the Phe-336 mutations in the ␣5 helix, the crystal structures showed the same localization and similar average B (temperature) factors around Phe-336 region relative to those of WT G␣ i1 structures (Fig. 7C). The effects of Phe-336 mutations on the ␤2-␤3 strands and ␤2-␤3 loop were minimal (Fig. 7, D  and E). Overall, the root mean square deviations between WT G␣ i1 -GDP with F336C and F336Y G␣ i1 -GDP were 0.42 and 0.36 Å (310 C␣ atoms aligned of 324 total), respectively, whereas they were 0.31 and 0.29 Å (304 C␣ atoms aligned of 315) for their GTP␥S-bound structures.

DISCUSSION
The ␣5 helix of the G␣ subunit is a critical region for both the receptor-mediated and basal activity (1,9,14,15,18). It is encircled by hydrophobic interactions from six ␤ strands (␤1-␤6) and the ␣1 helix (␣1). In the current study, we tested residues around the ␣5 helix that we predicted as critical for the function of this helix during G protein activation in our previous studies. We highlight information flow within the G protein, starting from the ␣5 helix to the GDP binding site of G␣ using biochemical, structural, and computational approaches.
Our previous study predicted that Phe-336 within the ␣5 helix is an important amino acid for both the active and inactive states (26), a finding consistent with other studies (14,28). Mutation of this residue resulted in constitutive activity in both monomeric and heterotrimeric G proteins (14,16,29,47). It is also known that in small GTPases, structural perturbation of that region through mutation causes increased guanine nucleotide turnover that can lead to several diseases, including Noonan, cardiofaciocutaneous, and Costello syndromes (47)(48)(49). In contrast to strong constitutive G protein activation, in this study, we did not observe drastic differences in the crystal structures of either GDP-or GTP␥S-bound Phe-336 mutants. Like another highly constitutively active G protein mutant, G␣ i1 A326S (50), Phe-336 mutants showed similar structural features compared with WT G␣ i1 . The guanine nucleotide provides a number of stabilizing interactions to the protein, perhaps inhibiting our ability to visualize subtle allosteric changes in the protein. In addition, other residues in the ␣5 helix and ␤-strands may contribute in holding this region intact during the crystallization process.
How does the perturbation at Phe-336 connect to the GDP binding region that is ϳ16 Å removed? Phe-336 is a part of a highly conserved hydrophobic core in the G␣ subunit. The effect of Phe-336 mutations on basal G protein activation is correlated with the hydrophobicity of this region (Fig. 3, C and  E). Once the receptor contacts the ␣5 helix and causes its rotation and displacement into the receptor binding site, this Phe-336 is now in a hydrophilic environment. We propose that breaking the hydrophobic core is a key event in perturbing GDP binding (26). Interestingly, we did not observe any effects of the hydrophilic mutants on receptor-mediated activation; this is likely due to the new solvent-exposed site, which prevents these side chains from contacting anything other than solvent upon receptor binding (Fig. 3, D and F).
To trace the hydrophobic interactions and to discern a possible interaction network from the Phe-336 residue to the GDP binding site, we computed binding energies of different regions in the G␣ subunit by using different Rosetta algorithms. Adding to our previous calculations (␣5 helix-G i ␣ interface binding energy (26)), we predicted that the Phe-336 side chain is mostly coupled with Met-53 (␣1), Ile-56 (␣1), Phe-189 (␤2), Phe-191 (␤2), Phe-196 (␤3), Val-332 (␣5), Gln-333 (␣5), Val-339 (␣5), and Thr-340 (␣5). Thus, the effects of Phe-336 are not solely local and not coupled to a single residue but rather might be part of a distributed network of interactions in which the activation is coupled to changes in regions dispersed across both domains of the G␣ subunit. Phe-336 is likely making direct hydrophobic contacts with Phe-191 and Met-53. It potentially communicates with Phe-189 via two paths.
The second path begins from the direction of Phe-196, which interacts with Phe-336 via the Phe-191 and Thr-340 residues. These observations indicate that the perturbation effects of Phe-336 spread with complex interactions via the ␣1 helix and ␤2-␤3 strands. These interactions also spread to the Mg 2ϩ ion and the nucleotide binding region (Figs. 5 and 6) as evidenced by our nucleotide exchange data combined with the perturba- Ϫ addition in the presence of different MgCl 2 concentrations (0.1-2 mM). Intensity of tryptophan signal were monitored (excitation, 290 nm; emission, 340 nm) at 21°C for 10 min before and after the addition of AIF 4 Ϫ (10 mM NaF and 50 M AlCl 3 ). The data were calculated as described above and rate constants plotted against MgCl 2 concentrations. D, trypsin digestion and analysis of G␣ i1 protein subunit. The densitometric measurement of proteolytic fragments in the presence of GDP-AIF 4 Ϫ ϩ 0.5 mM MgCl 2 . The results were normalized with WT G␣ i1 data, and fragments were quantified by densitometry (Multimager; Bio-Rad). The results represent the means Ϯ S.E. values of at least six to eight independent experiments.
We also tested the effects of residues within the ␤5-␤6 strands (Ile-265 (␤5), Phe-267 (␤5), Tyr-320 (␤6), and His-322 (␤6)) interacting with the other side of the ␣5 helix on G protein activation. We observed no major effects from the mutations either in the basal or receptor-mediated exchange assays. These data suggest how new interactions on the rotated ␣5 helix can compensate for the effects of single mutations in the ␤5 and ␤6 strands during receptor-mediated G protein activation. It also strongly suggests that the activation route goes through the other side of the protein (␤1-␤3/␣1 to Switch I, phosphate binding loop, Mg 2ϩ binding site, and GDP binding site), consistent with previously published findings (28). In addition, after restricting the C-terminal rotation and translocation by cross-linking the ␣1 and ␣5 helices, Phe-336 mutants can still induce increased basal nucleotide exchange (Fig. 4). This observation indicates that G proteins do not need a large displacement of ␣5 for basal state activation; rather, perturbing the ␤2-␤3 and ␣1 regions are sufficient.
In summary, our study used a predictive energetic analysis to pinpoint information flow through G␣ from receptor interaction to triggering of GDP release. We highlighted the hydrophobic interactions around Phe-336 as a key for stability of GDP binding, as well as removal of these hydrophobic interactions by receptor-mediated helical rotation to trigger GDP release. We suggested the route of information triggers through the ␣5 helix, ␤2-␤3 strands, and the ␣1 helix using energetic analysis and mutagenesis. We also showed that the dynamics of the Mg 2ϩ and ␤-phosphate binding area of GDP are perturbed by mutagenesis of this conserved residue. The ␤5-␤6 residues that face the ␣5 helix are likely important structurally rather than functionally according to our analysis. Thus, our data suggest that after the initial interaction of the G protein with the receptor and C-terminal rotation, disruption of a conserved hydrophobic network around Phe-336 engages both ␤1-␤3 and ␣1 to Switch I and the phosphate binding loop, which decreases the stability of the GDP binding pocket and triggers nucleotide release.