Folding of Gα Subunits: Implications for Disease States.

G-proteins play a central role in signal transduction by fluctuating between "on" and "off" phases that are determined by a conformational change. cAMP is a secondary messenger whose formation is inhibited or stimulated by activated Giα1 or Gsα subunit. We used tryptophan fluorescence, UV/vis spectrophotometry, and circular dichroism to probe distinct structural features within active and inactive conformations from wild-type and tryptophan mutants of Giα1 and Gsα. For all proteins studied, we found that the active conformations were more stable than the inactive conformations, and upon refolding from higher temperatures, activated wild-type subunits recovered significantly more native structure. We also observed that the wild-type subunits partially regained the ability to bind nucleotide. The increased compactness observed upon activation was consistent with the calculated decrease in solvent accessible surface area for wild-type Giα1. We found that as the temperature increased, Gα subunits, which are known to be rich in α-helices, converted to proteins with increased content of β-sheets and random coil. For active conformations from wild-type and tryptophan mutants of Giα1, melting temperatures indicated that denaturation starts around hydrophobic tryptophan microenvironments and then radiates toward tyrosine residues at the surface, followed by alteration of the secondary structure. For Gsα, however, disruption of secondary structure preceded unfolding around tyrosine residues. In the active conformations, a π-cation interaction between essential arginine and tryptophan residues, which was characterized by a fluorescence-measured red shift and modeled by molecular dynamics, was also shown to be a contributor to the stability of Gα subunits. The folding properties of Gα subunits reported here are discussed in the context of diseases associated to G-proteins.

Introduction

Guanine nucleotide-binding proteins (G-proteins) represent a family of proteins involved in intricate networks of intercellular signaling. Heterotrimeric G-proteins are comprised of α, β, and γ subunits that interact with transmembrane G-protein-coupled receptors (GPCRs). Upon activation of a receptor by an extracellular stimulus, the α-subunit undergoes a conformational change that allows exchange of guanine diphosphate (GDP) for guanine triphosphate (GTP) with concurrent dissociation from the βγ-dimer and GPCR, and a further relay of a signal via an interaction with an intracellular effector. The signal terminates following hydrolysis of the bound GTP, thereby returning the α-subunit back to its inactive state and its reassociation with the βγ heterodimer and the GPCR.[1−3] Although there are four families of Gα proteins, we limited this study to Giα1 and Gsα, which stimulate or inhibit the production of cAMP by regulating the activity of adenylyl cyclase (AC).

The crystal structures from Giα1 in the inactive GDP-bound conformation, as well as from the active states of both Giα1 and Gsα using GTPγS, a nonhydrolyzable GTP analog, have been solved.[4−6] The crystal structure of Gsα complexed with the target AC is also known.[6] Gα is composed of two domains: the α-helical domain and the GTPase domain. The α-helical domain consists of six α-helices that form a lid over the guanine nucleotide-binding site of the GTPase domain. The GTPase domain is composed of six-stranded β-sheets surrounded by five α-helices and in addition to the nucleotide-binding site, the GTPase domain also contains binding sites for the Gβγ dimer and the GPCR. Also, in the GTPase domain are the switch regions known as switches I–III that are located near the nucleotide-binding site. The switch regions undergo a drastic structural change when going from the inactive GDP-bound conformation to the active GTP-bound conformation.[7] In GDP-bound Giα1, switch II and switch III are disordered in the X-ray structure, but upon activation, they become ordered around the γ-phosphate of GTP.[4,5,8]

Protein folding is a complicated and yet a surprisingly efficient event that is critical for protein viability. Protein folding is driven primarily by noncovalent interactions and proceeds through an energy landscape from its unfolded state to its native conformation.[9−12] The free energy of the native state is lower than that of the unfolded protein, which is in equilibrium with molten globules that have a native-like structure. When a protein denatures, it does not go directly to a random coil, but rather to one of these molten globule states, which resembles the native state and may be able to bind a ligand and retain some activity.[13−15] Improper folding of the molten globules can have devastating consequences and is the cause of many diseases.[16]

Hydrophobic interactions contribute the most toward protein stability, but other interactions, such as hydrogen bonding and electrostatic interactions, are important as well.[6] Tryptophan (W) residues are uncommon and play a key role in protein stability via hydrophobic interactions at the core of the protein. Giα1 contains three W residues, whereas Gsα has four. The W residues in Giα1 are W131, W211, and W258 (depicted in cyan in Figure ), which respectively correspond to W154, W234, and W277 in Gsα. There is an additional W residue in Gsα, W281, that has no corresponding equivalent in Giα1. Gilman and co-workers reported that intrinsic W fluorescence could be used to investigate conformational changes in Gα proteins that occur during activation because the fluorescence intensity increases when individual W residues move toward a more hydrophobic environment.[17,18] Najor et al. built upon this property to quantify the contribution of each W residue toward the overall fluorescence by using phenylalanine (F) mutants of Giα1.[19] We explored this feature to determine the stability at the core of the protein by determining melting temperatures (Tm) from wild-type (WT) and W mutants of Giα1 and Gsα. In addition, a π-cation interaction between W211 and R208 (W234 and R231 in Gsα) is present in the active conformations of WT Gα proteins, which can be detected by red shifts in their fluorescence emission spectra. Disrupting the π-cation interaction may also have consequences for stability.[20]

Figure 1

WT Giα1·GTPγS displaying its 3 tryptophan residues (cyan), 13 tyrosine residues (purple), R208 (green), GTPγS-bound nucleotide (orange), and Mg2+ (green sphere).

Both Giα1 and Gsα have an abundance of tyrosine (Y) residues (13 for Giα1 and 14 in Gsα) (Figure ) from which we can take advantage of the UV absorbance to determine the Tm values at the surface of the protein for WT and W mutants. Although Y as well as W residues absorb light at 280 nm, in both Gα proteins Y residues far outnumber W amino acids resulting in absorbance changes that are dependent on Y and W residues. To obtain a more detailed picture of protein unfolding, we also used circular dichroism (CD) to monitor the secondary structure of the proteins.

Protein stability is an important characteristic of protein function. G-protein signaling must be tightly regulated to ensure appropriate responses to extracellular stimuli. Improperly functioning Gα proteins have been implicated in many disease states, including McCune-Albright syndrome, bipolar disorder, and cancer.[21−24] The focus of this study was to compare the stability of WT Giα1 and WT Gsα from different vantage points: from the inside core of the protein to its surface of the protein and from an overview of the overall secondary structure. Second, we investigated the contribution of each W residue individually and probed the interaction between one of them and the nearby arginine (R) and its effect on protein stability. To elucidate putative folding mechanisms in disease states, we utilized several biophysical techniques to probe the contributions of noncovalent interactions toward the stability of Gα proteins. Computational methods were also used to model the interactions.

Results

Fluorescence Emission Spectra of Gα Subunits

To calculate melting temperatures in both the active and inactive conformations of the WT proteins, we measured the changes in fluorescence intensity, resulting from increases in the solvent exposure of W residues. The amino acid F was chosen as a replacement for W because of its similar structure and size characteristics as well as low quantum yield and distinct λmax values.[19,25]

The fluorescence intensity of WT Giα1·GDP at 50 °C decreased by 53% when compared to that observed at 20 °C (Figure A), and continued declining until 70 °C, at which point there was no change in intensity and the protein was fully unfolded. A transition midpoint (Tm) of 39 °C was calculated for WT Giα1·GDP, and the W mutants in the same conformation were not significantly different from the WT protein (Table ). For Gsα in the GDP conformation, the Tm values for the WT protein were also not significantly different from all W mutants (Table ).

Figure 2

Intrinsic W florescence of WT Giα1 proteins. Emission spectra of 0.4 μM WT Giα1·Mg2+ at 20 °C (blue) and 50 °C (red) in the (A) GDP or (B) GTPγS conformations. Spectra shown were normalized to fluorescence intensities at 450 nm.

Table 1

Estimated Melting Temperature (°C) for Giα1 Proteins Using Three Spectroscopic Methodsa

	fluorescence		UV/vis		CD
Giα1 variant	GDP	GTPγS	GDP	GTPγS	GDP	GTPγS
WT	39	49b	48	67b	44	71b
W211F	35	37c	47	52c	54c	57c
W131F	38	52b	50	54c	44	71b
W258F	42	59b,c	46	63b,c	50c	68b

n ≥ 3; S.E.M. ≤ 3, for all measurements.

p ≤ 0.05 vs GDP-bound conformation.

p ≤ 0.05 vs WT in the same conformation.

Table 2

Estimated Melting Temperature (°C) for Gsα Proteins Using Three Spectroscopic Methodsa

	fluorescence		UV/vis		CD
Gsα variant	GDP	GTPγS	GDP	GTPγS	GDP	GTPγS
WT	41	39	54	64b	52	57b
W154F	45	41	53	60b	50	57b
W234F	40	33c,b	53	57c	51	53c
W277F	45	46c	51	60b	51c	58b
W281F	41	40	53	62b	54	56b

n ≥ 3; S.E.M. ≤ 3, for all measurements.

p ≤ 0.05 vs GDP-bound conformation.

p ≤ 0.05 vs WT in the same conformation.

For WT Giα1·GTPγS, the fluorescence intensity at 50 °C was 33% of that observed at 20 °C, indicating that the active conformation is more stable than the GDP-bound structure (Figure B). Apart from the W211F mutant, the Tm values for the other W Giα1 mutants in the Giα1·GTPγS conformation were also significantly higher than in the GDP conformation (Table ). Interestingly, the WT Giα1·GTPγS showed only a 10 °C increase, whereas the W131F and W258 mutants in the GTPγS conformation were approximately 14 and 17 °C higher than in their respective GDP conformations. The behavior of WT Gsα·GTPγS and its activated mutants was the opposite of Giα1 proteins in the GTPγS conformation. Alignment of the protein sequences indicates that W234F in Gsα and W211F in Giα1 are both located in the switch II region. The W234F mutant was unique because its Tm value in the GTPγS conformation (33 °C) was significantly lower than in the GDP conformation (40 °C) (Table ), and the analogous mutation in Giα (W211F) has essentially the same Tm in both the GDP- and GTP-bound forms (Table ). The Tm values for WT Gsα and its W154F, W277F, and W281F mutants in the Gsα·GTPγS conformation were not significantly different from their GDP counterparts.

π-Cation Interactions in Gα Subunits

To gain insight into the stability of the switch II region in WT Giα1, which co-ordinates with Mg2+ and the nucleotide-binding pocket, we monitored the π-cation interaction between R208 and W211 that occurs upon activation from the GDP-bound to the GTPγS conformation. At 20 °C, the λmax position exhibited a red shift of 3.5 nm (Figure A), which gradually decreased until 70 °C, at which point the instability of the GDP conformation prevented further measurements (Figure B). Similar changes in the value of the λmax position were observed for the WT Gsα protein until around 53 °C, where it switched from a red to a blue shift (data not shown).

Figure 3

(A) Emission spectra of WT Giα1·GDP·Mg2+ before (blue) and after (red) activation with GTPγS at 20 °C; (B) temperature variation of the difference between the λmax values of the GTPγS and GDP conformations.

UV/Vis Absorption Spectra of Gα Subunits

A useful property of Gα proteins is that W residues move to the hydrophobic core of the protein upon activation.[8] Thus, the spectroscopic and thermal properties of these sites allow for probing the interior of Gα subunits by using fluorescence emission spectroscopy. By contrast, Y residues are predominantly located at the surface of the Gα protein and are therefore useful for determining information on structural changes at or near the exterior of the protein.[8] As the protein unfolds, Y and W residues begin to contribute toward the absorbance. Because W has an absorptivity that is 4 times larger than Y at 280 nm, W would contribute significantly toward the Δabs as a result of the relative number of Y vs W residues in Giα1 (3 vs 13) and in Gsα (4 vs 14). In contrast, F absorptivity is approximately 30-fold lower than that of Tyr and the λmax is 257 nm, resulting in a negligible contribution toward absorbance at 280 nm.

An increase in absorbance intensity at 280 nm, which was associated to Y and W residues becoming more solvent exposed, was observed at temperatures above 44 °C for WT Giα1·GDP. The melting curve for Giα1 in the GTPγS form was shifted to the right of the GDP conformation (Figure a). A Tm value of 48 °C was calculated for WT Giα1·GDP and 54 °C for WT Gsα·GDP, and for the W mutants, the Tm values were not significantly different from their WT GDP counterparts (Tables and 2). For the GTPγS conformations, the Tm values for WT Giα1 and WT Gsα were significantly higher than for the GDP counterparts, but were not significantly different for proteins, in which the W residue involved in a π-cation interaction was mutated to F, i.e., W211F for Giα1 and W234F for Gsα (Tables and 2).

Figure 4

Temperature dependence of the (A) absorption spectra of 2.5 μM WT Giα1·Mg2+ in the GDP (blue) and GTPγS (red) conformations and of the (B) CD spectra of 1.0 μM WT Giα1·GDP·Mg2+.

Temperature Dependence of the Secondary Structure of Gα Subunits

At 20 °C, the CD spectra of WT Giα1·GDP (Table ) and of WT Gsα·GDP (Table ) were indicative of proteins that have secondary structures rich in α-helix (40 and 36%, respectively). The percent of α-helix that we observed for WT Giα1·GDP was in agreement to that also reported by others using CD (43%), which is less than in the reported structure deposited in the PDB (47%).[8,26] As the temperature increased, the CD absorbance intensity at 190 nm decreased, whereas the minima at 205 nm and 222 nm, which are signatures of α-helix, converged to a new minimum at 215 nm (Figure B).

Table 3

Composition of WT Giα1 Secondary Structure at Various Temperaturesa,b,c

	GDP				GTPγS
T (°C)	α	β	RCd	Td	α	β	RCd	Td
20	40	19	26	17	44	12	26	18
40	35	24	24	17	42	14	24	20
52	27	25	27	20	42	13	25	20
64	22	29	29	21	39	16	24	21
80	18	32	28	21	23	26	26	24
92					22	36	51	21

n ≥ 3; S.E.M ≤ 3.

All numbers reported as percentages.

Hyphens denote temperatures at which proteins denatured.

RC and T stand for random coil and turns.

Table 4

Composition of WT Gsα Secondary Structure at Various Temperaturesa,b,c

	GDP				GTPγS
T (°C)	α	β	RCd	Td	α	β	RCd	Td
20	36	18	33	13	37	16	33	13
40	30	22	34	14	33	20	34	13
52	29	24	34	13	31	20	35	14
64	28	25	34	13	26	24	36	14
80	25	27	35	13	20	27	39	14

n ≥ 3; S.E.M ≤ 3 for all measurements.

All numbers reported as percentages.

Hyphens denote temperatures at which proteins denatured.

RC and T stand for random coil and turns.

The data in Table indicated that regardless of the conformation, WT Giα1 initially was predominantly α-helical, but at higher temperatures, it became increasingly dominated by β-strands and to a lesser extent by random coil. By comparison, WT Gsα in both conformations had less α-helical and turn content, but more random coil and had a less dramatic α/β temperature-induced conversion (Table ). A CD-determined Tm value of 44 °C was calculated for WT Giα1·GDP, while the W211F mutant afforded the highest Tm value (Table ). Experiments with WT Giα1·GDP at temperatures greater than 64 °C did not exhibit significant changes in the CD spectra, with the protein eventually precipitating out of solution at 84 °C. Apart from the W211F mutant, WT and W mutants of Giα1 in the GTPγS conformation withstood temperatures near 100 °C without precipitation.

At 80 °C, the secondary structure of WT Giα1 protein in the active conformation had at least an additional 5% of α-helix content compared to the GDP conformation (Table ). Except for the Gsα W234F and W211F Giα1 mutants, the Tm values for the active conformations of WT Gsα and the remaining W mutants are significantly higher than for the inactive forms. The CD-determined Tm values for the inactive and active conformations of W234F Gsα and W211F Giα1 are not significantly different, and the Tm values for the active conformations are significantly lower when compared to the WT proteins (Table ).

Refolding

We have also investigated the ability of Gα subunits to refold after completion of the denaturation process. A decrease in temperature was accompanied by an increase in fluorescence intensity indicating that the W residues were refolding into hydrophobic environments, as demonstrated for WT Giα1·GTPγS (Figure A). Refolding WT Giα1·GDP from 96 to 4 °C exhibited no significant increase in fluorescence, however, upon renaturation from 48 °C, the observed increase in the fluorescence intensity indicated a refolding recovery of 21% (Figure B). When refolding from 32 °C, which is less than the fluorescence-determined Tm value of 39 °C (Table ), WT Giα1·GDP exhibited the largest recovery (72%). Unlike WT Giα1·GDP, the GTPγS conformation experienced increases in fluorescence intensity even when refolding was initiated from 96 °C, i.e., at temperatures larger than the Tm (Figure B and Table ). These observations demonstrate that the ability of Gα subunits to refold is conformation-dependent. Although this is the case for both Gα proteins, WT Giα1 was able to recover the most folded structure compared to WT Gsα (spectra not shown). Such traits were drawn out by fluorescence spectra of WT Giα1·GTPγS that revealed a 76% recovery after denaturation at temperatures up to 70 °C. By contrast, we found that WT Gsα·GTPγS only recovered 30% of its folded structure after denaturation at temperatures ≤ 84 °C. In addition, WT Gsα·GDP precipitated at temperatures less than 80 °C during renaturation.

Figure 5

(A) Refolding of WT Giα1·GTPγS as monitored via emission spectroscopy. Spectra shown were scaled to fluorescence intensities at 450 nm. (B) Percent fluorescence recovered after refolding of WT Giα1. Temperatures denote the maximum temperatures to which protein solutions were exposed before cooling. (C) Probing of denaturation and refolding of WT Giα1·GDP by circular dichroism. R represents refolded Giα1.

CD was also used to monitor the reversibility of protein unfolding. As shown in Figure C, when WT Giα1·GDP was cooled from 76 to 20 °C, there was a concomitant increase in the spectral intensity at 190 nm and a decrease at 222 nm. Spectral deconvolution showed that at 80 °C, WT Giα1·GDP consisted primarily of 18% α-helices and 32% β-sheets (Table ), but protein refolding back to 20 °C increased the α-helical content to 31%, whereas the percentage of β-sheets decreased to 18%. Terminating the denaturation process at 52 °C rather than at 76 °C resulted in recovery of 88% of the original α-helical structure. Similar effects were observed with WT Giα1·GTPγS. Although this conformation was more resistant to unfolding as evidenced by an initial 44% α-helical content at 20 °C (Table ), 93% of which was recovered when refolding from 76 to 20 °C.

To ascertain whether the partial recovery of structural refolding described above translated into a gain in protein activity, we investigated the kinetics of GTPγS binding at several temperatures (Figure ). Because of differences in protein stability, the decrease in fluorescence intensity for the protein in the inactive conformation is larger than in the active conformation (Figure ). Consequently, normalizing the initial fluorescence intensities of WT Giα1 in the GDP conformation at 40 and 30 °C to the same value accounts for the maximal fluorescence intensity observed upon GTPγS binding being the largest at 40 °C (Figure ). Heating WT Giα1 to 30 °C followed by cooling to 20 °C resulted in approximately 45% recovery of GTPγS binding, but when the protein was heated to 40 °C and then cooled to 20 °C, no GTPγS binding was found. As shown in Table , the melting temperature of WT Giα1 in the GDP conformation as measured by fluorescence is 39 °C, indicating that WT Giα1·GDP is unstable at 40 °C for GDP → GTPγS exchange to occur. In summary, these findings indicate that Gα subunits have the ability to partially regain GTPγS binding activity (Figure ) and that to some extent, refold the structure as demonstrated by our data obtained with two independent spectroscopic methods (fluorescence and circular dichroism; Figure ). To the best of our knowledge, this is the first time that a regain of function after refolding was reported for Gα subunits.

Figure 6

Temperature dependence of GTPγS binding to WT Giα1·GDP as monitored by time-based tryptophan fluorescence emission assays. R denotes traces from protein solutions that were heated to the higher temperature shown and then cooled to 20 °C. % fluorescence = ((Fo – Fi)/Fi) × 100, where Fi and Fo are the fluorescence intensities in arbitrary units at the start of GTPγS activation and at time t °C.

Discussion

Protein stability is critical for biological function. Our study focused on characterizing the noncovalent interactions that contribute to the stability of Gα proteins and to the reformation of the protein structure after unfolding. Surprisingly, given the importance of Gα proteins, there have been few studies of their stabilities.[27−29] In vivo, chaperones contribute toward protein stability. With respect to Gα subunits, the Ric-8A and Ric-8B chaperones play a part in the folding of nascent Giα1 and Gsα.[30]

A comparison of the WT Giα1 crystal structures in the GDP and GTPγS conformations reveals that the GDP-bound structure has a larger surface area than the active GTPγS conformation.[5,8] One would predict that compared to the GDP form, a denser folding profile for the GTPγS conformation of WT Giα1 would result in a more stable structure, as evidenced by the higher Tm values calculated from fluorescence emission, combined Y and W absorption, and CD spectra as well as from the larger interaction energies calculated for the GTP-bound protein (Tables and 5). This conclusion is also supported by solvent accessible surface area (SASA) calculations for WT Giα1 indicating that protein activation resulted in a 2.6% decrease in overall solvent exposure (19 520 Å2 for GDP-bound protein vs 19 010 Å2 for the active conformation). Therefore, WT Giα1·GTPγS is more stable, thus requiring more energy to unfold.

Table 5

Interaction Energies between R208 and W211 for Giα1 WTa,b

	GDP		GTP		Δ (GTP-GDP)
temperature	electrostatic	vdW	electrostatic	vdW	electrostatic	vdW
37 °C	–0.96	–3.38	–2.85	–4.38	–1.89	–1.00
50 °C	0.33	–2.16	–1.38	–4.61	–1.71	–2.45
Δ (50–37 °C)	1.29	1.22	1.47	–0.23	0.18	–1.45

S.E.M. ≤ 3.0.

Values are in kcal/mol.

Utilizing W → F single-point mutations, we followed the unfolding by measuring the temperature dependence of the fluorescence emission spectra of nine Gα proteins (WT and three W mutants of Giα1 and WT and four W mutants of Gsα) in the inactive GDP and active GTPγS conformations. Because burial of W residues in hydrophobic pockets is known to result in an increase in ΔFmax, protein unfolding is accompanied by a decrease in fluorescence intensity.[31] In the GDP forms (Table ), the fluorescence-measured Tm values for WT Giα1 were not significantly different (p < 0.1) from its W mutants. Except for the W211F mutant, the Tm values were, however, significantly smaller than for the active WT Giα1, W131F, and W258F proteins (p < 0.01). The GTPγS conformation of the W211F mutant proved to be the least stable of all of the active Giα1 proteins and displayed a fluorescence-derived Tm value similar to its GDP conformation (Table ), which is the opposite of the general trend of higher melting temperatures observed for the GTPγS conformations. The difference in the interaction energies for the GDP and GTP found during the molecular dynamics simulations was smaller for the W211F variant than for the WT, which might contribute to the active conformation of this mutant being less stable.

Unlike WT Giα1, for which crystal structures are known for the inactive and active conformations, only the structure of WT Gsα·GTPγS has been published, precluding an explanation of protein stability based on compactness or differences in GTP and GDP interaction energies with the protein.[5,8,32] The fluorescence-derived data in Table indicate that WT Gsα and its mutants do not follow the same folding pattern as for Giα1. For Tm values calculated from fluorescence spectra, there is no significant difference between the active and inactive conformations of WT Gsα and of its W154F, W277F, and W281F mutants suggesting that with the exception of W234F Gsα, stability of the protein structure around the W residues in Gsα is different from Giα1. Figure shows that at room temperature, the ΔFmax values were significantly lower for WT Gsα relative to WT Giα1. Since ΔFmax is a result of W movement, this trend suggests that after activation a smaller displacement of the W residues occurs in Gsα compared to Giα1. Therefore, unlike WT Giα1, the W residues in the GDP conformation of WT Gsα are relatively protected in hydrophobic environments, presumably accounting for the insignificant difference between the Tm values from WT Gsα·GTPγS and WT Gsα·GDP (Table ). The insignificant differences between the Tm values from the active and inactive conformations of the W154F, W277F, and W281F mutants of Gsα are likely to have the same origin.

Figure 7

Time-based emission assay monitoring percent change in intrinsic tryptophan fluorescence of (A) WT Gsα and (B) WT Giα1 and their respective Trp mutants after the addition of GTPγS. % fluorescence was calculated in the same manner as for Figure .

The W211F mutant of Giα1 and the W234F mutant of Gsα do not show detectable changes in ΔFmax (Figure , panels A and B). The W211 residue in WT Giα1 has been shown to have the largest difference in solvent accessibility between the inactive and active conformations and therefore contributes the most toward ΔFmax.[19] Not surprisingly, for the W211F mutant of Giα1, no ΔFmax is observed (Figure B). Similarly, the W234 residue in Gsα likely undergoes a similar large decrease in solvent accessibility during the course of the conformational change, as evidenced by the negligible ΔFmax observed in the W234F mutant (Figure A). The fluorescence-derived Tm values for the W211F mutant of Giα1 are not statistically different in the two conformations (Table ) presumably because of the absence of the W211-R208 cation−π interaction. Interestingly, the Tm value for the W234F mutant is significantly lower than for WT Gsα (Table ).

The secondary structure of WT Giα1 proved to be the most stable in its GTPγS form relative to the GDP conformation (Table ). At 20 °C and upon binding of GTPγS, we identified a 4% increase in the α-helical content of WT Giα1 (Table ), but not for WT Gsα (Table ). Activation of WT Giα1 creates a hydrophobic pocket via folding of the switch regions, resulting in a protein that has an ordered secondary structure with an increased α-helical content.[4,8] The smaller ΔFmax observed for activation of WT Gsα relative to WT Giα1 (Figure ) may be related to a smaller change in the secondary structure of WT Gsα. In either conformation, as the temperature increased, the α-helical content of both WT Gα proteins was reduced and the subunits became richer in β-sheet while the random coil and turn structures were not altered significantly from the native form. We have done molecular dynamics simulations of the thermal unfolding of the monomeric Gα proteins and have not observed an increase in β-sheet, although the amount of α-helix decreased. These simulations may indicate that the β-sheet increase is due to aggregation.

A shift in secondary structure from primarily α-helices to β-sheets poses an increased risk for protein aggregation that may lead to amyloidogenesis.[16] Amyloid fibril formation occurs when unfolded, native-like proteins aggregate into long filaments of packed β-sheets.[33−35] Many debilitating neurodegenerative diseases, such as Parkinson’s, Creutzfeldt-Jakob’s, and Alzheimer’s, have been proposed to arise from the accumulation of amyloid fibrils in the brain or in the central nervous system.[16] In vitro studies have shown that it is not uncommon for proteins to form amyloid fibrils under denaturing conditions.[36,37] Furthermore, fibril formation has been shown to inhibit refolding into the native conformation.[38]

The absorbance assays helped visualize the global unfolding of Gα subunits from another perspective. The Tm values for WT Giα1 that were calculated from the absorbance of Y and W residues correlate to the unfolding process (Figure A). UV/vis experiments with Giα1 showed that the protein surface in the GTPγS conformation to be significantly more stable than the W microenvironments, whereas the CD-determined values indicated that the surface unfolded before the secondary structure (Table ). In the case of WT Gsα·GTPγS, the UV/vis-calculated Tm value was the highest compared to those derived from the other measurements, indicating that the surface of the WT Gsα is the last to unfold (Table ). In the W211F mutant of Giα1 and in the W234F mutant of Gsα, no significant difference between the Tm values was observed upon activation. One possibility is that π–cation interactions, involving W211 in Giα1 and W234 in Gsα, affect unfolding proximal to Y and W residues. π–cation interactions are found in many proteins.[39,40] They are known to contribute significantly to thermal stability.[41,42] The average energy for W–cation interactions is −2.9 ± 1.4 kcal/mol.[41,42] For the W154F, W277F, and W281F mutants of Gsα, the UV/vis-determined Tm values were significantly higher for the active conformations. For Giα1, however, only the W258F mutant was stabilized, suggesting a distinct folding pattern for the two Gα subunits in each conformation.

We have examined the thermal denaturation of the Gα proteins using three different optical probes: absorbance, fluorescence, and CD. These probes primarily measure changes in the environments of Y residues or W residues or the secondary structure, respectively. Since they give different Tm values for the same protein (Tables and 2), the denaturation of both Gα proteins appears to be multistate rather than two state.[43] The differences in Tm values in Gα that were observed by different methods may be rationalized via an analysis of the hydrophobic interactions, which are fundamental folding determinants for all proteins. Noncovalent interactions underpin the driving forces in protein folding. The observed Tm values suggest that denaturation of the active conformation of Giα1 starts near W131 and W258 microenvironments, and then propagates outward through the protein surface where the Y residue proximal to W258 is located, and at this point of unfolding leaving the secondary structure intact. Additional heating results in the conversion of α-helixes into β-sheets and random coil, possibly involving aggregation until precipitation occurs. In contrast, denaturation of the active conformation of Gsα initiates equally around all W residues, continuing to the secondary structure and is completed near the Y residues.

The robustness and resistance of a protein to misfolding minimize the chances for disease. The reversibility of folding observed with WT Giα1 via fluorescence emission and CD (Figure a,c) can therefore shed important light on the misfolding of Gα subunits. During the course of denaturation, a protein may develop multiple intermediate conformations, or molten globule states, which are reflected by the different Tm values obtained by the three techniques.[44] The fluorescence spectra monitored, to a significant degree, the polarity changes surrounding the W sites. Oscillations of the nonpolar side chains at these sites would generate molten globules with relatively low thermal energies. These movements would account for the lower Tm values calculated from fluorescence measurements, compared to those obtained with the other two spectroscopic probes. Multiple Y residues, which may be involved in hydrogen bonding, are distributed throughout Gα. Once protein unfolding is initiated, molten globule states that are populated will exhibit diminished secondary structure, which is determined by hydrogen bonding. The additional contribution of hydrogen bonding associated with Y microenvironments and secondary structure relative to primarily hydrophobic interactions present in the vicinities of W residues may explain the higher Tm values measured from Y absorption and CD spectra.

Previous work by Najor et al. and Hamm and co-workers showed that W211 forms a π–cation interaction with R208 in WT Giα1·GTPγS, as evidenced by a red shift of 2.5 nm in the λmax value (Figure a).[19,20] Molecular dynamic simulations predict that the conformational change from the inactive to the active conformation results in an increase in the electrostatic interaction between W211 and R208 from −0.96 to −2.85 kcal/mol, which is consistent to the higher stability seen in the active conformation (Table ). Thus, stronger ligand–protein interactions would help stabilize the GTPγS-bound structure. Molecular dynamics studies showed that the interaction energy between GTP and Giα1 at 323 K (−621.7 kcal/mol) indicated that GTP binds more tightly than GDP (−494.4 kcal/mol). This binding energy partially may explain why the GTP-bound structure refolds better.

An increase in temperature at which the simulation was conducted (37 → 50 °C) resulted in weakening of the W211-R208 π–cation interaction, which is supported by the observed decrease in the Δλmax (Figure b). The increased van der Waals interactions calculated at higher temperatures may be associated with these residues swinging into more hydrophilic environments upon unfolding. This conclusion is supported by a blue (rather than red) shift observed upon the GTPγS activation of Gsα at temperatures higher than 53 °C. For the W211F mutant of Giα1, there was no significant difference between the Tm values from the active and inactive conformations further suggesting that the π–cation interaction is important for the structural integrity of Giα1.

This study underscores the importance of π–cation interactions toward protein stability. The disruption of these noncovalent interactions may lead to significant decreases in the stabilities for the active conformations of Gα subunits and could promote improper folding. Mutations of the arginine residue involved in the π–cation interaction have been identified in the R208Q Giα1 and in the R231H Gsα oncogenes and are thought to have similar characteristics as the W mutants.[23] The loss of the π–cation interaction could translate into changes in structure–function relationships by disrupting the signaling cascade for cAMP. Future studies will focus on the effect of these mutations on the structure and function of oncogenic Gα subunits.

Experimental Section

Expression and Protein Purification

Gαi1 and Gsα were obtained and purified as previously described.[45] Single-point W mutants of Gαi1 and Gsα were prepared by site-directed mutagenesis using a kit provided by Stratagene (La Jolla, CA). After purification on a Ni2+ affinity column followed by a Superdex 200 pg size exclusion column, the purity of GDP-bound Gα proteins was found to be greater than 95% as estimated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis. Protein was stored at −80 °C in 20 mM Tris, pH 8.0 buffer containing 10% (v/v) glycerol, and 1 mM dithiothreitol (DTT).

Fluorescence Measurements of Protein Activation

Experiments were performed with a PTI QuantaMaster fluorimeter (Photon Technologies, Inc., Mirmingham, NJ). Indirect activity assays were conducted with excitation and emission wavelengths set at 280 and 340 nm, respectively. Assays were initiated after 60 s by addition of 20 μM of GTPγS to preincubated 400 nM Gα·GDP protein samples in buffer containing 50 mM N-(2-hydroxyethyl)piperazine-N′-ethanesulfonic acid, pH 7.5, 2 mM MgSO4, and 1 mM DTT, and was monitored for 3 h at 25 °C. The GDP- and GTPγS-bound proteins that were characterized by the activity assays were used in the following denaturation studies.

Fluorescence-Measured Protein Denaturation

Emission spectra for both GDP- and GTPγS-bound proteins were recorded over the wavelength range of 300–400 nm with the excitation wavelength set at 280 nm. Signal integration time was 0.2 s with the bandpass for excitation and for emission set at 5 nm. The denaturation experiments started at a temperature of 4 °C followed by 4 °C increments and concluding at the highest temperature before precipitation occurred. There was a 2 min equilibration period at each set temperature. All Tm values were calculated from fluorescence intensities at the spectral λmax positions for the selected temperatures, using methods adapted from those previously described.[46]

UV/Vis-Measured Protein Denaturation

The environments of Y (and to a lesser extent W) residues in Gα proteins were monitored on a Hewlett Packard UV/vis spectrophotometer. All samples contained 50 mM Tris, pH 7.5, 1 μM Gα·GDP protein, 1 mM DTT, and 2 mM MgSO4. Prior to initiating the experiments, samples were incubated with their respective nucleotide, 2.5 μM Gα·GDP or 20 μM GTPγS, at room temperature for 1 h. The temperature was increased from 20 to 80 °C, at 0.3 °C/min over 180 min. For each temperature studied, samples were equilibrated for 1 min, and the absorbance was monitored in the wavelength range of 220–300 nm. All melting temperatures were calculated from the absorbance values at 280 nm for the different temperatures, using methods previously described.[47]

CD-Measured Protein Denaturation

Experiments were performed using an Olis DSM 20 circular dichroism spectrophotometer. All samples were measured in a cylindrical quartz cuvette with a 1 mm pathlength and contained either 3 μM Gα·GDP or 24 μM Gα·GTPγS, in 10 mM phosphate, pH 7.5 buffer, 1 mM DTT, and 2 mM MgSO4. Data were collected at 150 V every 1 nm in the wavelength range of 190–260 nm. The temperature was increased from 20 to 100 °C at 4 °C increments with an incubation time of 3 min at each temperature studied. The CONTIN LL algorithm was used to deconvolute the spectra using reference sets with denatured proteins to calculate the percent of each type of secondary structure and Tm values for each protein studied.[48−50]

Refolding of Gα Subunits

To test whether unfolding of Gα proteins was reversible, fluorescence emission scans and CD spectrophotometry were used. Once spectra from the final temperature of an unfolding experiment were obtained, Gα samples were cooled down in 8 °C increments and incubation times remained the same as indicated above for each respective technique. Final temperatures varied depending on aggregation and ability to refold. All renaturation experiments were stopped at 4 °C for fluorescence measurements and at 20 °C for CD experiments.

Molecular Modeling

The co-ordinates of GDP (1BOF[8]) and GTPγS (1GIA[5]) derivatives of Giα1 and GTPγS of Gsα (1AZT[32]) were downloaded from the Protein Data Bank (PDB[51]). Missing loops in the Giα1 structures were modeled using Swiss Model[52] and the corresponding transducin structures (1TAG,[53] 1TAD,[54] and 1TND[55]). The simulations were done using procedures previously described.[19] Unrestrained dynamics was run for 14 ns before the data were acquired for an additional 1 ns. The simulations were done at 37 °C (310 K) and 50 °C (328 K). These data were then used in the analyses. The initial W point mutation models were generated using VMD[56] and then subjected to the same equilibration procedure as the wild-type structures. All molecular graphic diagrams were generated using VMD.[56] Pairwise van der Waals and electrostatic interaction energies were calculated using NAMD.[57] The solvent accessible surface area (SASA) was measured with the SASA routine in VMD.[56] The SASA values and the van der Waals and electrostatic energy values presented in Table were calculated for the final 1 ns in each simulation and then averaged.