Conformational Changes Associated with Receptor-stimulated Guanine Nucleotide Exchange in a Heterotrimeric G-protein α-Subunit

Solution NMR studies of a 15N-labeled G-protein α-subunit (Gα) chimera (15N-ChiT)-reconstituted heterotrimer have shown previously that G-protein βγ-subunit (Gβγ) association induces a “pre-activated” conformation that likely facilitates interaction with the agonist-activated form of a G-protein-coupled receptor (R*) and guanine nucleotide exchange (Abdulaev, N. G., Ngo, T., Zhang, C., Dinh, A., Brabazon, D. M., Ridge, K. D., and Marino, J. P. (2005) J. Biol. Chem. 280, 38071-38080). Here we demonstrated that the 15N-ChiT-reconstituted heterotrimer can form functional complexes under NMR experimental conditions with light-activated, detergent-solubilized rhodopsin (R*), as well as a soluble mimic of R*. NMR methods were used to track R*-triggered guanine nucleotide exchange and release of guanosine 5′-O-3-thiotriphosphate (GTPγS)/Mg2+-bound ChiT. A heteronuclear single quantum correlation (HSQC) spectrum of R*-generated GTPγS/Mg2+-bound ChiT revealed 1HN, 15N chemical shift changes relative to GDP/Mg2+-bound ChiT that were similar, but not identical, to those observed for the \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{GDP}{\cdot}\mathrm{AlF}_{4}^{-}{/}\mathrm{Mg}^{2+}\) \end{document}-bound state. Line widths observed for R*-generated GTPγS/Mg2+-bound 15N-ChiT, however, indicated that it is more conformationally dynamic relative to the GDP/Mg2+- and \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{GDP}{\cdot}\mathrm{AlF}_{4}^{-}{/}\mathrm{Mg}^{2+}\) \end{document}-bound states. The increased dynamics appeared to be correlated with Gβγ and R* interactions because they are not observed for GTPγS/Mg2+-bound ChiT generated independently of R*. In contrast to R*, a soluble mimic that does not catalytically interact with G-protein (Abdulaev, N. G., Ngo, T., Chen, R., Lu, Z., and Ridge, K. D. (2000) J. Biol. Chem. 275, 39354-39363) is found to form a stable complex with the GTPγS/Mg2+-exchanged heterotrimer. The HSQC spectrum of 15N-ChiT in this complex displays a unique chemical shift pattern that nonetheless shares similarities with the heterotrimer and GTPγS/Mg2+-bound ChiT. Overall, these results demonstrated that R*-induced changes in Gα can be followed by NMR and that guanine nucleotide exchange can be uncoupled from heterotrimer dissociation.

G-protein coupled receptors (GPCRs) 3 represent a diverse group of seven transmembrane (TM) helix receptors that require agonist-dependent activation to initiate heterotrimeric (␣␤␥) G-protein-mediated intracellular signaling cascades. GPCR activation of cognate G-proteins are the first steps in cellular communication pathways responsible for signaling cascades that mediate vision, olfaction, taste, and the action of numerous hormones and neurotransmitters (1). Activation of a G-protein by its agonist-stimulated GPCR (R*) requires the propagation of structural signals from the receptor-binding interface to the guanine nucleotide-binding pocket. The structural basis for the interaction of a GPCR with its cognate G-protein and the subsequent activation of the G-protein by R* are not fully understood.
Using signaling of the retinal G-protein transducin (G t ) by rhodopsin as a model system, we are applying solution NMR methods to track changes in the G-protein ␣-subunit (G ␣ ) associated with activated R* interactions. Rhodopsin, the rod cell photoreceptor involved in dimlight vision, represents one of the best studied GPCRs in terms of structure and function (2,3). Photon capture triggers cis 3 trans isomerization of the retinal chromophore, which initiates structural changes in the TM helices resulting in the formation of the light-activated signaling state metarhodopsin II, R*. This is accompanied by functionally significant changes at the cytoplasmic surface that leads to the formation of binding and activation sites for several signaling proteins, including G t (4 -10). Crystal structures for the inactive (dark) state of rhodopsin have provided a detailed view of the retinal binding site and the cytoplasmic region (11)(12)(13)(14)(15). Although remarkably informative, the crystal structures provide few solid insights into the mechanism of signal transfer from R* to G t .
Binding of heterotrimeric G-proteins to activated GPCRs requires the presence of both G ␣ and G-protein ␤␥-subunits (G ␤␥ ). The following three regions on the ␣-subunit of G t (G t␣ ) are known to be important for receptor interactions; the amino-terminal 23 residues, an internal sequence from amino acids 305-315, and the carboxyl-terminal 11 amino acids (16 -18). Upon binding to R*, G t␣ is thought to undergo structural changes in both the amino-and carboxyl-terminal regions. High resolution crystal structures of G ␣ subunits, including G t␣ (19 -24), G ␤␥ (25), and G ␣␤␥ heterotrimeric complexes (Refs. 26 and 27, Fig.  1A), have provided important insights into the structural rearrange-ments accompanying guanine nucleotide exchange and the GTPase cycle, particularly in the conformationally flexible switch regions. In most of the crystal structures, however, the residues at the extreme carboxyl terminus of G ␣ are disordered and/or not visible. This is consistent with findings from transferred nuclear Overhauser enhanced spectroscopy NMR studies (28 -30) on 11 amino acid carboxyl-terminal peptides derived from G t␣ (G t␣ (340 -350) peptides), which show that these peptides are largely unstructured in solution and in the presence of dark state rod outer segment (ROS) rhodopsin, but undergo significant structural changes upon binding to light-activated rhodopsin. Similarly, the helical structure of the amino terminus of G ␣ appears transient and is only ordered in crystal structures of the heterotrimer (26,27). Results from a study in which fluorescent and spin-label probes were introduced at specific positions in the amino-terminal region of G ␣ are consistent with the amino terminus by assuming an ordered helical conformation only in G t␣␤␥ (31).
High resolution structural analysis of R*/G-protein interactions poses many significant challenges given the inherently dynamic nature of this process. The R*/G t interaction can be viewed as taking place in at least five discrete biochemical reaction steps (Fig. 1B). These include R* binding to G t␣␤␥ ⅐GDP to form the R*⅐G t␣␤␥ ⅐GDP complex (steps 1 and 2), GDP dissociation from the R*⅐G t␣␤␥ ⅐GDP complex to form an R*⅐G t␣␤␥ [empty] complex (step 3), GTP uptake by the R*⅐G t␣␤␥ [empty] complex to form R*⅐G t␣␤␥ ⅐GTP (step 4), and dissociation of G t␣␤␥ ⅐GTP from R* followed by G t␣ ⅐GTP from G t␤␥ (step 5), with R* now free for interaction with another G t␣␤␥ ⅐GDP (32). Although the above mentioned crystallographic studies have been instrumental for obtaining static three-dimensional structures of dark state rhodopsin and various guanine nucleotide-bound states of G ␣ , and biochemical and mutational approaches have provided a wealth of information about the nature of R*/G t interactions, a structural representation of the R*-G t complex(es) remains poorly defined. Clearly, a comprehensive description of the structures involved in these reaction steps would provide important insights into the mechanisms governing activated GPCR/Gprotein interactions.
We have shown previously that a G ␣ chimera consisting of sequences from G t␣ and G i1␣ (Chi6; see Ref. 33) can be expressed to high levels in a soluble form by using a subtilisin prodomain (proR8FKAM) fusion construct and milligram quantities of prodomain-released, full-length, isotope-labeled G ␣ (ChiT) purified in a single step by using an immobilized "slow cleaving" mutant form of subtilisin (34). This has allowed us to pursue functional studies under NMR experimental conditions that provide insights into the solution structures of G ␣ in various states. We have also shown that isotope-labeled ChiT can be reconstituted with G t␤␥ subunits to form a functional heterotrimer that is amenable to structural analysis by high resolution NMR (35). This latter work revealed that G ␤␥ binding to ChiT induces structural changes in the guanine nucleotide binding and carboxyl-terminal regions of ChiT, leading to a "preactivated" state that may facilitate interaction with R* and subsequent GDP/GTP exchange. Here we have now applied high resolution NMR to begin to probe the structural basis for the propagation of signals from R* to the G-protein, with the specific goal of developing more robust models for the structural changes in G ␣ that accompany the signal transfer process. Specifically, NMR methods have been used to track the complete cycle of guanine nucleotide exchange in 15 N-ChiT-reconstituted heterotrimer that is triggered by light-activated rhodopsin (R*), thereby providing new insights into G ␣ conformational changes-associated signal propagation from an activated GPCR. Using similar NMR approaches, guanine nucleotide exchange in a 15 N-ChiT-reconstituted heterotrimer stimulated by a soluble mimic of R* has been monitored. In contrast to R*, the soluble mimic remains bound to the nucleotide-exchanged heterotrimer forming a trapped, stable complex akin to the R*⅐G t␣␤␥ ⅐GTP intermediate in the reaction pathway.

EXPERIMENTAL PROCEDURES
Materials-Cyclohexylpentyl-␤-D-maltoside (Cymal-5) and n-dodecyl ␤-D-maltopyranoside (DM) were from Anatrace. GTP␥S was from Roche Applied Science, and Ni 2ϩ -nitrilotriacetic acid-agarose resin was from Qiagen. [ 35 S]GTP␥S was from PerkinElmer Life Sciences. Anti-G t␣ and anti-G t␤ antibodies were from Affinity BioReagents, and the anti-rhodopsin antibody K42-41L (37) was a gift from Prof. Paul Hargrave (University of Florida). Horseradish peroxidase-conjugated goat anti-mouse and anti-rabbit antibodies were from Santa Cruz Biotechnology, and protein G-Sepharose was a gift from Prof. Philip Bryan (University of Maryland Biotechnology Institute). The pG58 expression vector, a fusion vector encoding a modified 77-amino acid prodomain region of subtilisin BPNЈ (proR8FKAM), and the pG58-derived expression vector encoding a G ␣ chimera (Chi6) as a proR8FKAM fusion have been described (34,36). The sources of other materials used in this investigation have been reported (34,35,38).
Expression and Purification of Subtilisin Prodomain/Chi6 Fusions-Methods for the inducible bacterial expression and purification of isotope-labeled wild-type and mutant G ␣ using the proR8FKAM/Chi6 fusion and immobilized S189 subtilisin BPNЈ have been described (34). To generate the GTP␥S/Mg 2ϩ -bound form of ChiT independently of R*, GDP was omitted from the cell lysis and column purification buffers in order to obtain an 'empty pocket' state of G ␣ that could be subsequently reconstituted with GTP␥S. Prior to NMR analysis, the purified and isotope-labeled proteins were concentrated and dialyzed against 25 mM d 11 Tris-HCl, pH 7.5, containing 100 mM NaCl, 5 mM magnesium acetate, 2.5 mM dithiothreitol, and 5% glycerol (Buffer A).
Expression and Purification of HPTRX/CDEF-Detailed protocols for the inducible expression and purification of HPTRX/CDEF, a soluble mimic of R*, have been described (38). Prior to NMR experiments, purified HPTRX/CDEF was concentrated and dialyzed against Buffer A.
Detergent Solubilization and Purification of Rhodopsin-ROS rhodopsin from bovine retina was solubilized and purified in Cymal-5 or DM detergent on rho-1D4-Sepharose essentially as described (39,40). Rhodopsin concentrations were determined by UV-visible spectroscopy at 20°C using a 25 spectrophotometer (PerkinElmer Life Sciences). Prior to NMR experiments, rhodopsin preparations were concentrated and dialyzed against Buffer A containing 0.08% (w/v) Cymal-5.
Filter Binding Assay for Measuring G-protein-mediated Guanine Nucleotide Exchange-The ability of detergent-solubilized rhodopsin preparations to catalyze the uptake of [ 35 S]GTP␥S by G t was determined by initial rate analysis, and at equilibrium, using a nitrocellulose filter binding assay essentially as described (41). For initial rate analyses, the reaction mixtures contained 6.7 nM Cymal-5-or DM-solubilized and -purified ROS rhodopsin and 5 M [ 35 S]GTP␥S in 10 mM Tris-HCl, pH 7.5, containing 100 mM NaCl, 5 mM MgCl 2 , and 2.5 mM dithiothreitol (Buffer B). After illumination (Ͼ495 nm) for 1 min at 20°C, the reactions were initiated by the addition of 4 M G t . The total reaction volume was 250 l, and the final concentrations of Cymal-5, DM, and glycerol in the assay mixtures were 0.08, 0.015, and 5% (w/v), respectively. At various time intervals (5-20 s), a 50-l aliquot was removed, rapidly filtered through nitrocellulose with the aid of a vacuum manifold, and the filters immediately washed three to four times with 5 ml of Buffer B to remove free, unbound [ 35 S]GTP␥S. The filters were dried, and the G-protein-bound [ 35 S]GTP␥S was determined by scintillation counting. For the equilibrium assays, the same reaction mixture containing 4 M G t was illuminated for 1 min, the reaction allowed to proceed for 2 h at 20°C, and the extent of GDP/GTP␥S exchange determined as described above. Identical reactions were performed in the dark for both the initial rate and equilibrium assays. The activity in the dark was subtracted from that in the light for determination of the kinetic and equilibrium values, which are reported as averages Ϯ S.E.
Rate of Metarhodopsin II (R*) Decay in the Absence and Presence of G t -The rate of retinal release upon R* decay in 0.08% Cymal-5 or 0.015% DM, and in the presence G t , was determined by following the decrease in protonated retinyl-Schiff base as measured at 440 nm after acidification essentially as described (42)(43)(44). Briefly, purified rhodopsin (1.17 M) in 25 mM Tris-HCl, pH 7.5, containing 100 mM NaCl, 5 mM magnesium acetate, 2.5 mM dithiothreitol, 5% glycerol (Buffer C), and 0.08% Cymal-5 or 0.015% DM was illuminated (Ͼ495 nm for 1 min) at 20°C in the absence or presence of G t (1.4 M, 1.2-fold excess over rhodopsin). Aliquots (400 l) were removed at specific time points (typically at 1, 15, 30, 60, 120, 240, and 480 min) and acidified to pH ϳ3 by addition of 2 N H 2 SO 4 . After mixing, a UV-visible spectrum (650 nm-250 nm) was recorded, and the amount of the protonated retinyl-Schiff base (440 nm) remaining as a function of time was determined.
Reconstitution of ChiT with G t␤␥ to Form the G ␣␤␥ Heterotrimer-The G-protein heterotrimer was reconstituted from isotope-labeled ChiT and G t␤␥ essentially as described (35). Prior to NMR experiments, heterotrimer preparations were concentrated and dialyzed against Buffer A or Buffer A containing 0.08% Cymal-5.
Immunoprecipitation and Analysis of the G ␣␤␥ -GTP␥S⅐HPTRX/ CDEF Complex-Equimolar concentrations of HPTRX/CDEF and ChiT-reconstituted heterotrimer (500 nM) in Buffer C were mixed with GTP␥S (600 nM) and incubated for 30 min at 20°C. An aliquot of anti-G t␤ antibody (20 l of a 1 mg/ml solution) was added to the mixture, followed by 200 l of pre-cleaned protein G-Sepharose. After gentle mixing for 30 min at 20°C, the beads were allowed to settle, the supernatant removed, and the beads washed five times with 1 ml of Buffer C. The washed beads were resuspended in reducing SDS-PAGE sample buffer (250 l), and aliquots (50 l) were examined by reducing SDS-PAGE (45) using a 5% stacking and a 16% resolving gel. The immunoprecipitated proteins were electroblotted onto poly(vinyl difluoride) membranes (46), detected with the K42-41L (HPTRX/CDEF), anti-G t␣ (ChiT), and anti-G t␤ (G t␤ ) antibodies, or a mixture of these primary antibodies, and horseradish peroxidase-conjugated goat anti-mouse and/or anti-rabbit antibodies. The proteins were visualized by chemiluminescence.
NMR Spectroscopy-One-dimensional 1 H-and 15 N-filtered 1 H water flip-back, water gate, 15 N-decoupled spectra and two-dimensional 15 N-HSQC water flip-back, water gate spectra (47) were acquired at 30°C using a Bruker AVANCE 600-MHz spectrometer (Bruker Instruments, Billerica, MA) equipped with a triple-resonance 1 H, 13 C, 15 N z axis gradient cryoprobe and linear amplifiers on all three channels. Spectra were collected on uniformly 15 N-labeled samples ( 15 N-ChiT) dissolved in Buffer A at concentrations of 150 -300 M. The nitrogen frequency was centered at 118 ppm and the proton frequency on H 2 O (ϳ7.5 ppm). One-dimensional spectra were collected using a sweep width of 7,200 Hz and 2,048 complex points, and two-dimensional data were acquired using sweep widths of 7,200 Hz in 2 and 2,000 Hz in 1 , 2,048 by 64 complex data points in t 2 and t 1 , respectively, (t 1(max) ϭ 293 ms and t 2(max) ϭ 64 ms) and 128 scans per increment. NMR samples containing rhodopsin were placed in the spectrometer under dim-red light conditions. To initiate the guanine nucleotide exchange reaction, rhodopsin was illuminated with Ͼ495 nm light for 1 min prior to spectral acquisition. For NMR samples containing HPTRX/CDEF, the protein was added directly to the reaction mixture to stimulate guanine nucleotide exchange. All spectra were processed and analyzed on a SGI UNIX work station using NMRPipe (48). Trp indole and Phe-350 amide 1 NH and 15 N resonances were assigned using ChiT mutants as described previously (34,35). The aluminum fluoride (AlF 4 Ϫ ) adduct of GDP/Mg 2ϩbound ChiT was formed by addition of NaF (10 mM) and AlCl 3 (300 M) directly to the NMR tube.
Other Methods-A fluorescence assay for monitoring G t activation by HPTRX/CDEF was performed essentially as described (38). Protein determinations were done as described previously (49).

Experimental Design and General
Considerations-Having previously demonstrated that 15 N-ChiT can be functionally expressed as a subtilisin BPNЈ prodomain fusion and purified on immobilized S189 subtilisin BPNЈ, reconstituted with unlabeled G t␤␥ subunits to form a heterotrimer (ϳ85 kDa), and characterized by high resolution NMR (34,35), it was of keen interest to now investigate whether NMR could also be used to follow some of the reaction steps involved in R*-catalyzed guanine nucleotide exchange (Fig. 1B). Because these solution NMR studies would necessitate the use of rhodopsin, an ϳ40-kDa lightsensitive integral membrane protein, it was necessary to identify a detergent that could not only effectively solubilize rhodopsin but also support the formation of R* at concentrations below the critical micelle concentration (CMC). The rationale for choosing such a detergent for these NMR studies was as follows. At detergent concentrations below the CMC, an R*-G-protein complex would be expected to behave essentially as a complex composed of the constituent components (ϳ125 kDa plus detergent molecules). Although the size of this complex represented a significant challenge, if well behaved and with selective isotope labeling, it was reasoned that it would still be possible to analyze using solution NMR methods. In contrast, at detergent concentrations above the CMC, rhodopsin would be expected to be in micelles, which could contain several randomly oriented molecules and approach several hundred kDa. Such micelles would tumble extremely slowly and result in significant additional line broadening and relaxation effects. Interpretation of the NMR spectra for this case would represent an additional technical challenge. Therefore, we expended a considerable effort in identifying such a detergent, and we found that Cymal-5 (5-cyclohexyl-1-pentyl-␤-D-maltoside), a relative of the commonly used rhodopsinsolubilizing detergent DM, appears to fulfill the necessary criteria (see below). The CMC of Cymal-5 has been determined to be 2.4 mM (ϳ0.12%) in aqueous buffers (50), and its readily dialyzable nature enables facile manipulation of sample detergent concentration for NMR measurements.
Properties of Rhodopsin and R* in Cymal-5 Detergent-Representative UV-visible absorption spectra of immunoaffinity-purified ROS rhodopsin in 0.08% Cymal-5 are shown in Fig. 2A. The Cymal-5-solubilized and -purified rhodopsin exhibits the characteristic 500 nm dark state chromophore and shows a shift in max to ϳ380 nm upon illumination. Subsequent acidification of this photoproduct yields a 440-nm absorbing species, which is characteristic of a protonated retinyl-Schiff base, and suggests that the 380-nm absorbing species most likely represents the metarhodopsin II photointermediate R*. In general, these spectral transitions are qualitatively similar to those observed upon similar treatment of purified ROS rhodopsin in DM (51), suggesting that Cymal-5 does not adversely affect the structural integrity of the pigment.
Previous guanine nucleotide exchange assays have shown that purified rhodopsin in 0.08% Cymal-5 also catalyzes the light-dependent uptake of GTP␥S by G t and the ChiT-reconstituted heterotrimer (35). A more detailed analysis of the kinetics of G t activation by Cymal-5 purified light-activated rhodopsin showed that the rate of GDP/GTP␥S exchange was nearly indistinguishable from that obtained for DM-purified rhodopsin (Fig. 2B). In both cases, ϳ3 mol of GTP␥S were exchanged per second at 4 M G t (2.9 Ϯ 0.23 (n ϭ 3) and 3.1 Ϯ 0.32 (n ϭ 3) for Cymal-5 and DM-solubilized ROS rhodopsin, respectively). In contrast, equilibrium assays showed that the level of GTP␥S exchanged after exhaustive binding was greater for R* in Cymal-5 than in DM, each catalyzing on average the exchange of ϳ324 Ϯ 21.4 (n ϭ 3) and ϳ262 Ϯ 17.3 (n ϭ 3) mol at 4 M G t , respectively.
The rate of R* decay in 0.08% Cymal-5 was also examined in the absence and presence of G t and compared with that for R* in 0.015% DM, a concentration just above the CMC of this detergent in aqueous buffer (50). As shown in Fig. 2C, purified rhodopsin in 0.08% Cymal-5 shows the characteristic loss of 440 nm absorbance as a function of post-illumination time. After 8 h at 20°C, only ϳ40% of the protonated retinyl-Schiff base remained, indicating that the majority of R* had decayed to opsin and free all-trans-retinal. The t1 ⁄ 2 of this process was FIGURE 2. Properties of Cymal-5 detergent-solubilized and -purified rhodopsin. A, UV-visible absorption spectra of purified rhodopsin in 0.08% Cymal-5 detergent. ROS rhodopsin was solubilizedin1%Cymal-5,purifiedonimmobilizedrho-1D4,andelutedfromtheantibodymatrix in 0.08% Cymal-5. Spectra were recorded in the dark (labeled dark), after illumination (Ͼ495 nm for 10 s) to form R* (labeled light), and after acidification to pH ϳ3 (labeled acid) to form a protonated retinyl-Schiff base. B, initial rate analysis of GDP/GTP␥S exchange reactions of G t catalyzed by Cymal-5 and DM-solubilized and -purified rhodopsin. Purified ROS rhodopsin in 0.08% Cymal-5 (circles) or 0.015% DM (squares) were preincubated with [ 35 S]GTP␥S and allowed to remain in darkness (open symbols) or illuminated (filled symbols) prior to the initiation of guanine nucleotide exchange by the addition of G t . The amount of bound [ 35 S]GTP␥S was determined using a filter binding assay as described under "Experimental Procedures." Representative data from three independent experiments are shown. C, rate of R* decay in the absence and presence of G t . Purified rhodopsin in 0.08% Cymal-5 or 0.015% DM was illuminated (Ͼ495 nm for 1 min) in theabsenceorpresenceofG t ,andaliquotswereremovedatindicatedtimepoints,acidifiedtopH ϳ3, and examined by UV-visible spectroscopy as described under "Experimental Procedures." The data are plotted as a percentage of 440 nm remaining absorbance as a function of time after illuminationofpurifiedrhodopsinin0.08%Cymal-5detergentintheabsence(triangles)andpresence of G t (circles), or purified rhodopsin in 0.015% DM (squares). Representative data from two or three independent determinations are shown.
ϳ28 min. For rhodopsin in 0.015% DM, ϳ70% of the R* decayed to opsin and free all-trans-retinal over the 8-h time course with a t1 ⁄ 2 of ϳ15 min, a half-life in agreement with the previous findings of Farrens and Khorana (42). These results suggest that the lifetime of R* is prolonged in 0.08% Cymal-5 when compared with R* in 0.015% DM. Moreover, the rate of R* decay in 0.08% Cymal-5 was significantly reduced in the presence of G t , with less than 20% of the R* being converted to opsin and free all-trans-retinal. Taken together, these findings clearly show that R* in 0.08% Cymal-5 interacts with G t to form a complex that triggers guanine nucleotide exchange, thereby facilitating analysis of R*/G-protein interactions under suitable NMR experimental conditions.
Monitoring Light-activated R*-catalyzed Guanine Nucleotide Exchange in the 15 N-ChiT-reconstituted Heterotrimer by Solution NMR-To simulate the reaction scheme of the R*/G-protein interaction, which leads to the formation of G ␣ (GTP) (Fig. 1B), 15 N-ChiT-reconstituted heterotrimer was first dialyzed into Buffer A to which 0.08% (w/v) Cymal-5 had been added. One-dimensional 15 N-filtered 1 H and two-dimensional HSQC spectra, which allow selective monitoring of amide and side chain 1 HN, 15 N resonance signals from 15 N-ChiT (ϳ150 M), were then acquired (data not shown) and compared with spectra of the heterotrimer in Buffer A alone. Comparisons of the spectra showed no differences in either amide or side chain chemical shifts or line widths, indicating that addition of 0.08% Cymal-5 does not affect either the conformation or stability of 15 N-ChiT. The nonhydrolyzable GTP analog, GTP␥S (ϳ175 M), was then added to the 15 N-ChiT-reconstituted heterotrimer sample (ϳ150 M), and one-dimensional 1 H-and 15 N-filtered 1 H spectra were acquired (Fig. 3, A and B (red traces)). The 15 N-filtered 1 H spectrum again showed no change in the 15 N-ChiT signals with respect to the spectrum acquired for the 15 N-ChiTreconstituted heterotrimer, indicating that the added GTP␥S remains free in solution and is not exchanged with bound GDP. The one-dimensional 1 H spectra further supports this conclusion as two new relatively sharp signals are observed to appear after addition of GTP␥S, indicated by arrows in Fig. 3B, that are attributed to the anomeric (ϳ5.8 ppm) and aromatic H8 protons (ϳ8.1 ppm) of the added GTP␥S free in solution.
In contrast, addition of purified, dark state, rhodopsin that had been dialyzed into Buffer A containing 0.08% Cymal-5, to 15 N-ChiT reconstituted heterotrimer (ϳ150 M) in Buffer A with 0.08% (w/v) Cymal-5 and GTP␥S (175 M) to a final concentration of 15 M, resulted in approximately a 50% reduction in the observed 15 N-ChiT amide and side chain proton signal intensity (Fig. 3A, light blue trace). This reduction in the signals was more pronounced than could be explained by simple dilution of the sample (addition of 75 l of rhodopsin to 250 l of reconstituted heterotrimer) and suggests that the heterotrimer may interact with Cymal-5-solubilized dark state rhodopsin. As would be expected, however, the one-dimensional 1 H spectrum (Fig. 3B, light blue trace) showed no apparent changes in the signals of the free GTP␥S, indicating that Cymal-5-solubilized dark state rhodopsin does not stimulate guanine nucleotide exchange.
After exposure of the NMR sample mixture ( 15 N-ChiT⅐Gt ␤␥ -GDP/ Mg 2ϩ ϩ GTP␥S ϩ solubilized rhodopsin) to light (Ͼ495 nm) for 1 min, there was an increase in peak intensities (Fig. 3A, green trace) that continued until a final state was reached (Fig. 3A, dark blue trace), suggesting formation of GTP␥S/Mg 2ϩ -bound ChiT dissociated from R* and Gt ␤␥ . It should be noted based on the measured rate for R*-catalyzed guanine nucleotide exchange (Fig. 2B) that the exchange reaction in the NMR sample would also be rapidly completed within minutes after illumination. Most surprisingly, changes in the intensity of the ChiT signals detected in the one-dimensional NMR experiments (Fig. 3A) were still observed after several hours and even after changes in the intensity of the proton signals from the added GTP␥S were observed to be complete. Given this significantly slower "rate" observed for the change in the intensity of the ChiT signals in the NMR experiment, the associated change in the state of ChiT does not appear to be associated with R*-catalyzed guanine nucleotide exchange. In addition, because Cymal-5-solubilized rhodopsin was added to the NMR sample in substoichiometric amounts relative to G ␣␤␥ , it also appears that R* catalytically interacts with the heterotrimer under our NMR experimental conditions. We therefore came to the conclusion that the slow rate observed in the NMR experiment reflects neither guanine nucleotide exchange nor release of the exchanged heterotrimer from R*, but rather is a measure of the slow dissociation of G␣(GTP␥S) from G ␤␥ . In this respect, the observed increase in the intensity of the ChiT signal is  15 N-filtered spectra of 15 N-ChiT-reconstituted heterotrimer at various points in the R*-catalyzed guanine nucleotide exchange reaction. B, one-dimensional 1 H spectra acquired at the same points in the guanine nucleotide exchange reaction. Note that GTP␥S uptake by ChiT in the R*-catalyzed reaction is directly observed as a time-dependent reduction in intensity for the H8 (ϳ8.1 ppm) and H1Ј (ϳ5.8 ppm) signals (indicated by arrows) of added GTP␥S, as would be expected upon association of free nucleotide with a large protein complex. Additional sharp signals (ϳ7.8 and 8.4 ppm) are also observed to increase in intensity and may correspond to "free" GDP, which should be released during the course of the reaction. Other sharp signals are observed in these spectra that are attributed to buffer and that do not change over the course of the reaction. All spectra were acquired at pH 7.5 and 30°C using a Bruker 600 MHz NMR Cryoprobe system as described under "Experimental Procedures." consistent with ChiT assuming a faster correlation time, which results in a narrowing of the NMR line widths, as would be expected upon heterotrimer dissociation. Over the time course of subunit dissociation, the observed NMR signals would be contributed from a smaller protein with a faster correlation time, i.e. ChiT (ϳ40 kDa) versus the heterotrimer (ϳ85 kDa). Alternatively, the observed changes in ChiT signal intensity could be the result of a slow alteration in the conformation of G ␣ after it dissociates from G ␤␥ . Although this is also a possible explanation, a lack of observed associated changes in 1 H chemical shifts makes it less likely. R*-catalyzed uptake and release of guanine nucleotides is also apparent from changes in the 1 H signals (H8/H1Ј) arising from added GTP␥S after illumination. As shown in Fig. 3B, GTP␥S uptake by ChiT in the R*-catalyzed reaction is directly observed as a reduction in the 1 H signals arising from added GTP␥S, as would be expected upon association of free nucleotide with a large protein or protein complex. Additional sharp signals are also observed in the final state spectra of the guanine nucleotide exchange reaction (Fig. 3B). Although these signals may correspond to free GDP, which would be expected to be released from the 15 N-ChiT-reconstituted heterotrimer in the exchange reaction, their chemical shifts do not correspond to those measured for H1Ј and H8 proton signals of free GDP in Buffer A (data not shown). Further experimentation will therefore be necessary to establish a definitive assignment and ascertain the reason(s) for this apparent difference. Ϫ /Mg 2ϩ ), and activated (GTP␥S/ Mg 2ϩ ) are indicated by arrows. All spectra were acquired at pH 7.5 and 30°C using a Bruker 600 MHz NMR Cryoprobe system as described under "Experimental Procedures." Given the observed slow rate of change in the ChiT spectrum after illumination, the NMR sample was allowed to remain overnight before further NMR analysis (total time ϳ16 h). Additional one-dimensional 15 N-filtered and 1 H spectra of the NMR sample ( 15 N-ChiT⅐Gt ␤␥ -GDP/ Mg 2ϩ ϩ GTP␥S ϩ R*) collected after this time period revealed no further changes in the spectra. Upon observation of no additional changes in the one-dimensional spectra, an HSQC was then acquired in situ to probe the conformation of GTP␥S/Mg 2ϩ -bound ChiT. The HSQC spectrum reveals clear changes in the chemical shifts for the amide and side chain 1 HN, 15 N resonances when compared with the GDP/Mg 2ϩ -bound form of ChiT (Fig. 4A). Most interestingly, the changes observed in the R*-generated GTP␥S/Mg 2ϩ -bound form relative to the GDP/Mg 2ϩ -bound form are similar, but not identical, to those observed upon formation of the GDP⅐AlF 4 Ϫ /Mg 2ϩ adduct (Fig. 4B) and the heterotrimer (34,35). For example, two of the three assigned cross-peaks for the tryptophan indoles (Fig. 4C, left panel) shift to the same position in these states. However, one tryptophan indole (Trp-207), located in the functionally important switch II region of G ␣ , appears to broaden beyond detection in the R*-generated GTP␥S/ Mg 2ϩ -bound state, and the carboxyl-terminal Phe-350 residue (Fig. 4C, right panel) is completely shifted into the "activated" conformation. In addition, and in contrast to the HSQC spectra acquired for the GDP/ Mg 2ϩ -and GDP⅐AlF 4 Ϫ /Mg 2ϩ -bound states of 15 N-ChiT and the GDP/ Mg 2ϩ -bound form of the 15 N-ChiT-reconstituted heterotrimer, the spectrum of the R*-generated GTP␥S/Mg 2ϩ -bound ChiT shows nonuniform line widths, with a number of resonances, like those associated with Trp-207 indole ring, appearing to be significantly exchange broadened, and a second subset of resonances observed to be considerably sharper than on average. These NMR line width observations suggest that certain parts of the structure of ChiT are dynamic and exchanging between distinct conformations in the R*-generated GTP␥S/Mg 2ϩbound state in a way that affects both the local and global structure. 15 N-ChiT Generated Independently of R*-To dissect the relative contributions of changes in the conformation and dynamics of GTP␥S/Mg 2ϩ -bound 15 N-ChiT that are related to interaction with R* (and G ␤␥ ) from those associated with guanine nucleotide exchange, GTP␥S/Mg 2ϩ -bound 15 N-ChiT was also generated in an "R*independent" manner by directly reconstituting the "empty pocket" state of G ␣ with GTP␥S. The HSQC spectrum of GTP␥S/Mg 2ϩ -bound 15 N-ChiT generated in this way reveals changes in the chemical shifts for the amide and side chain 1 HN,15 N resonances that are again indicative of a shift in the protein conformation to an activated conformation (supplemental Fig. S1). As with R*-generated GTP␥S/Mg 2ϩ -bound 15 N-ChiT, changes observed in the R*-independent generated GTP␥S/ Mg 2ϩ -bound form relative to the GDP/Mg 2ϩ -bound form are similar, but not identical, to those observed upon formation of the GDP⅐AlF 4 Ϫ / Mg 2ϩ adduct and the heterotrimer (34,35). For example, two of the three assigned cross-peaks for the tryptophan indoles (Fig. 4D, left  panel) shift to the same position in both the R* and R*-independent generated GTP␥S/Mg 2ϩ -bound forms of 15 N-ChiT. In contrast, Trp-207, which appears to broaden beyond detection in the R*-generated GTP␥S/Mg 2ϩ -bound state, is found to be shifted to a unique downfield position, relative to the GDP⅐AlF 4 Ϫ /Mg 2ϩ adduct and heterotrimer states of ChiT, in the GTP␥S-bound state of ChiT generated independently of R*. Moreover, the carboxyl-terminal Phe-350 residue in this GTP␥S-bound state (Fig. 4D, right panel) is not observed to shift significantly to the activated position. In contrast to R*-generated GTP␥S/Mg 2ϩ -bound ChiT, the spectrum of GTP␥S/Mg 2ϩ -bound ChiT generated independently of R* shows relatively uniform line widths, as observed for the GDP/Mg 2ϩand GDP⅐AlF 4 Ϫ /Mg 2ϩ -bound states of 15   Structural Changes in G ␣ Accompanying GPCR Activation MARCH 17, 2006 • VOLUME 281 • NUMBER 11 that the differential conformational dynamics within GTP␥S/Mg 2ϩbound ChiT is a direct result of interactions with G ␤␥ and R*. Biochemical Analysis of a Complex between a Soluble Mimic of R* and G ␣␤␥ -Previously, we have designed, constructed, and expressed soluble cytoplasmic surface polypeptides of rhodopsin by grafting different combinations of the cytoplasmic loops of rhodopsin onto a thioredoxin scaffold. Biochemical and structural studies (38,52) have shown that some of these rhodopsin cytoplasmic fragment/thioredoxin fusion proteins exhibit properties similar to those of R*. In particular, the HPTRX/ CDEF fusion protein, which contains tandemly linked segments from the CD (amino acids 132-154) and EF (amino acids 231-252) loops of rhodopsin (Fig. 5A), was found to exhibit constitutive activity toward G t . Specifically, HPTRX/CDEF displayed virtually identical kinetics of G t activation when compared with light-activated ROS rhodopsin (38), suggesting that the CD and EF loops in the fusion protein can adopt a conformation that is a close approximation to that present in R*. It was also shown that HPTRX/CDEF bound G t␣ (340 -350) peptides (52) inducing structural changes in these peptides analogous to those observed previously in studies with native, light-activated rhodopsin in ROS membranes (28 -30).

Comparisons of GTP␥S/Mg 2ϩ -bound 15 N-ChiT Generated by R* with GTP␥S/Mg 2ϩ -bound
To characterize further the interaction between HPTRX/CDEF and G t , a fluorescence assay for G-protein activation has been used to monitor guanine nucleotide exchange as a function of HPTRX/CDEF concentration. As shown in Fig. 5B, addition of sub-stoichiometric amounts of HPTRX/CDEF relative to G t elicits only partial GTP␥S uptake. Subsequent additions of HPTRX/CDEF result in increased levels of guanine nucleotide exchange, ultimately saturating the response at a concentration nearly equal to that of G t . These findings suggest that HPTRX/ CDEF, in contrast to light-activated rhodopsin, lacks the ability to catalytically interact with G t and raised the possibility that these proteins may form a stable complex. To obtain biochemical evidence for such a stable interaction, an equimolar mixture of HPTRX/CDEF and ChiTreconstituted heterotrimer was incubated in the presence of GTP␥S and then immunoprecipitated with an anti-G t␤ antibody followed by protein G-Sepharose. The bound proteins were analyzed by SDS-PAGE and immunoblotting using an antibody mixture containing the antirhodopsin antibody K42-41L (37), which recognizes the EF loop region of rhodopsin, and anti-G t␣ and anti-G t␤ antibodies. The results of this analysis (Fig. 5C) clearly show that HPTRX/CDEF, ChiT, and G t␤ (presumably G t␤␥ ) co-precipitate and support the hypothesis that the noncatalytic nature of HPTRX/CDEF is likely because of stable complex formation between HPTRX/CDEF and the GTP␥S/Mg 2ϩ -bound heterotrimer.
NMR Analysis of a "Trapped" R*⅐G ␣␤␥ ⅐GTP Complex-The results of our earlier work with HPTRX/CDEF (38,52) combined with the results of the G t activation and immunoprecipitation assays (Fig. 5, B and C) prompted us to attempt to apply NMR methods to follow HPTRX/ CDEF-stimulated guanine nucleotide exchange in 15 N-ChiT-reconsti-  (38). A Gly/Pro linker is present between the CD and EF loops, and between these loops and the thioredoxin scaffold (represented by zigzag lines). The gray cylinders show the positions of the TM helices in rhodopsin. B, activation of G t by HPTRX/CDEF. The assay mixture (100 l) initially contained 500 nM G t and 750 nM GTP␥S. After obtaining a stable base line, the reaction was initiated by the addition of 200 nM HPTRX/CDEF and followed for 1000 s. A second aliquot of HPTRX/CDEF was then added bringing the concentration to 400 nM, and the reaction was followed for another 1000 s. This same procedure was repeated until saturation of the fluorescence response (an additional two times). The fluorescence measurements, which allow the rate of GTP␥S uptake by G t␣ to be determined by monitoring the increase in intrinsic fluorescence of Trp-207, suggest that HPTRX/CDEF does not catalytically interact with G t . Representative data from two independent determinations are shown. C, immunoprecipitation of the HPTRX/CDEF⅐G ␣␤␥ complex. HPTRX/CDEF (lane 1) was detected with the anti-rhodopsin K42-41L antibody; ChiT (lane 2) was detected with an anti-G t␣ antibody; and G t␤ (lane 3), from isolated G t␤␥ , was detected with an anti-G t␤ antibody. The HPTRX/CDEF⅐G ␣␤␥ ⅐GTP␥S/Mg 2ϩ exchanged complex (lane 4) was immunoprecipitated with the anti-G t␤ antibody followed by protein G-Sepharose, and the individual components detected with a mixture of K42-41L, anti-G t␣ , and anti-G t␤ antibodies as described under "Experimental Procedures." Note that the intensities of the signals in each of the lanes most likely reflect differences in the affinities of the antibodies used to detect the various components as equimolar amounts of protein were analyzed. The positions of molecular mass standards are shown on the left. tuted heterotrimer in a manner analogous to that described above for native, light-activated rhodopsin. In this series of experiments, HPTRX/ CDEF (ϳ150 M), an ϳ15-kDa soluble protein, was first added to an equimolar amount of 15 N-ChiT-reconstituted heterotrimer (ϳ150 M) in the absence of GTP␥S. The 15 N-filtered 1 H spectrum of 15 N-ChiT in this protein mixture shows less intensity (ϳ30% reduction) with respect to the spectrum of the 15 N-ChiT-reconstituted heterotrimer (Fig. 6A, red trace) and is attributed to broadening of 15 N-ChiT resonances that occur upon formation of a complex between HPTRX/CDEF, a constitutively active soluble mimic of R*, and the 15 N-ChiT reconstituted heterotrimer. In contrast to the experiments using native, light-activated rhodopsin (Fig. 3A), subsequent addition of GTP␥S (ϳ175 M) resulted in only a very small initial increase in the peak intensities for signals associated with 15 N-ChiT (Fig. 6A, red trace). Nonetheless, as with R*, one-dimensional 1 H NMR spectra clearly indicate that HPTRX/CDEF stimulates GTP␥S uptake (compare the green and red traces in Fig. 6B) and changes in the chemical shifts of some 1 HN, 15 N resonances of 15 N-ChiT are apparent (Fig. 6A). Furthermore, distinct upfield-shifted methyl 1 H resonances, assigned to ChiT and G t␤␥ in the heterotrimeric state, are also observed not to shift or change in intensity during guanine nucleotide exchange (data not shown). Collectively, these data are consistent with a nondissociated state of a GTP␥S-bound heterotrimer in complex with HPTRX/CDEF (akin to the R*⅐G t␣␤␥ ⅐GTP complex shown in Fig. 1B, step 4), rather than dissociated constituent components as follows: HPTRX/CDEF, GTP␥S/Mg 2ϩ -bound ChiT, and G t␤␥ . Moreover, given these NMR observations, together with previous studies which show that the GTP␥S-bound heterotrimer normally dissociates into activated G ␣ and G ␤␥ components, the GTP␥Sbound, nondissociated state of the heterotrimer is likely maintained through the stable interaction with HPTRX/CDEF. Fig. 7A shows an overlay of the HSQC spectra of the HPTRX/ CDEF⅐G ␣␤␥ ⅐GTP␥S/Mg 2ϩ -exchanged complex and the GDP/Mg 2ϩbound form of the heterotrimer. Although there are many similarities among these two states of the heterotrimer as might be expected, the spectrum of the HPTRX/CDEF⅐G ␣␤␥ ⅐GTP␥S/Mg 2ϩ -exchanged complex reveals a few unique cross-peaks suggestive of a distinct conformational state for this form of ChiT. Furthermore, although an overlay of the HSQC spectra of the HPTRX/CDEF⅐G ␣␤␥ ⅐GTP␥S/Mg 2ϩ -exchanged complex and GTP␥S/Mg 2ϩ -bound ChiT also shows similarities between these two states, a number of unique cross-peaks, as well as differences in resonance line widths, are again observed suggesting a distinct conformation for this GTP␥S-bound form of ChiT (Fig. 7B). For example, although the carboxyl-terminal Phe-350 residue is completely shifted to the activated position in both R*-generated GTP␥S/Mg 2ϩbound ChiT and the HPTRX/CDEF⅐G ␣␤␥ ⅐GTP␥S/Mg 2ϩ -exchanged complex (Fig. 7C, right panel), the 1 HN, 15 N cross-peak of the Trp-207 indole, which appears broadened beyond detection in R*-generated form of GTP␥S/Mg 2ϩ -bound ChiT, shows a weak, yet discernible cross-peak that resonates in a new position that is shifted downfield from the cross-peaks observed previously (34) for the major conformation of this residue in the GDP/Mg 2ϩ -bound reconstituted heterotrimer and GDP⅐AlF 4 Ϫ /Mg 2ϩ -bound ChiT (Fig. 7C, left panel). The position of the cross-peak observed for the Trp-207 indole in the HPTRX/ CDEF⅐G ␣␤␥ ⅐ GTP␥S/Mg 2ϩ -exchanged complex spectrum, however, is close to the chemical shift position observed for a minor conformational form of the Trp-207 indole ring (Fig. 7C, left panel, denoted by an asterisk) observed in these spectra. Taken together, these results suggest that R* populates this alternative, likely intermediate, conformation of switch II in the exchanged heterotrimer. In summary, the present results show that solution NMR methods can be used to track the cycle of guanine nucleotide exchange in an isotope-labeled G ␣ -reconstituted heterotrimer that is triggered by light activation of detergent-solubilized rhodopsin. Moreover, these same methods can be used to track guanine nucleotide exchange in an isotope-labeled G ␣ -reconstituted heterotrimer stimulated by a soluble mimic of R*, which remains bound forming a trapped, stable intermediate complex. Overall, our study demonstrates how NMR approaches can allow simultaneous detection of both local and global changes in G ␣ associated with R*-stimulated guanine nucleotide exchange. The observations made using the soluble mimic of R* also indicate that GDP/GTP exchange can be uncoupled from the process of heterotrimer release and dissociation, suggesting that distinct R*-induced changes in G ␣ may facilitate these events.

DISCUSSION
The mechanism(s) by which R* catalyzes rapid and tightly regulated guanine nucleotide exchange by heterotrimeric G-proteins is not fully  1 H spectra acquired at the same points in the guanine nucleotide exchange reaction. HPTRX/CDEFstimulated GTP␥S uptake by ChiT is directly observed as a time-dependent broadening of the H8 (ϳ8.1 ppm) and H1Ј (ϳ5.8 ppm) signals (indicated by arrows) of added GTP␥S, as observed with R* (see Fig. 3B). In addition, sharp signals (ϳ7.8 and 8.4 ppm) are again observed, as in Fig. 3B, which may correspond to released GDP in the same reaction, but require further experimentation for definitive assignment. Note also that as in Fig. 3B, the other sharp signals observed are attributed to buffer. All spectra were acquired at pH 7.5 and 30°C using a Bruker 600 MHz NMR Cryoprobe system as described under "Experimental Procedures." understood. We are applying solution NMR methods to probe the structural basis for the propagation of signals from R* to the G-protein, with the specific goal of developing more robust models for changes in G ␣ that accompany signal transfer from R*. For our studies, we are using the well characterized rhodopsin/transducin interaction of the vertebrate visual system, as it serves as the paradigm for understanding activated GPCR/G-protein interactions (2,3). Moreover, knowledge of the biochemical steps in this reaction (Fig. 1B), combined with the availability of crystal structures, provides a solid starting foundation on which to build our understanding of the structural changes in both rhodopsin and G t that accompany complex formation and R*-catalyzed guanine nucleotide exchange.
We have shown previously that a full-length isotope-labeled G ␣ chimera (ChiT) can be prepared in milligram amounts and that the expressed protein shares comparable properties to native G t␣ (34). Furthermore, ChiT can be reconstituted with G t␤␥ to form functional heterotrimers that are amenable to analysis using solution NMR methods (35). These studies have not only provided new insights into the role of Note that as seen for the R*-generated GTP␥S/Mg 2ϩbound form of ChiT, the carboxyl-terminal Phe-350 amide 1 HN, 15 N cross-peak is observed to completely shift to the activated position in the HPTRX/CDEF⅐G ␣␤␥ ⅐GTP␥S/Mg 2ϩ -exchanged complex, although the Trp-207 1 HN, 15 N indole cross-peak is broadened and shifted to a position close to that observed for a minor conformational form of Trp-207 (denoted by an asterisk) in both the GDP/Mg 2ϩ -bound reconstituted heterotrimer and the GDP⅐AlF 4 Ϫ /Mg 2ϩ -bound form of 15 N-ChiT. All spectra were acquired at pH 7.5 and 30°C using a Bruker 600 MHz NMR Cryoprobe system as described under "Experimental Procedures." G ␤␥ in remodeling the conformation of G ␣ to facilitate R* binding and GDP/GTP exchange, but these studies formed the basis for pursuing further studies focused on R*-induced changes in G ␣ structure. As such, we have set out in this study to demonstrate a systematic NMR approach for tracking the complete cycle of guanine nucleotide exchange in a 15 N-ChiT-reconstituted heterotrimer using light-activated, detergent-solubilized rhodopsin and a previously characterized soluble mimic of R*.
R*-mediated Guanine Nucleotide Exchange in the ChiT-reconstituted Heterotrimer-A primary concern from the outset of these experiments was to identify a suitable detergent that would support the formation of metarhodopsin II or R*, the rhodopsin signaling photointermediate, and also facilitate R*/G-protein interactions that trigger guanine nucleotide exchange at concentrations below the CMC of the detergent. Cymal-5 appeared to effectively serve this purpose as rhodopsin samples prepared in this detergent, like those in the commonly used DM, showed the characteristic ground state chromophore ( max ϳ500 nm) that was converted to R* upon photolysis ( Fig. 2A). Moreover, initial rate analysis of the GDP/GTP␥S exchange reaction of G t catalyzed by Cymal-5-purified ROS rhodopsin revealed virtually indistinguishable kinetics of activation from that observed for DM-purified ROS rhodopsin (Fig. 2B). At equilibrium, however, the amount of GTP␥S exchanged was significantly higher for R* in Cymal-5 when compared with R* in DM. This latter finding agrees with the observation that the lifetime of Cymal-5purified R* is prolonged when compared with R* in DM, and could be stabilized further in the presence of G t (Fig. 2C). Collectively, these findings show that Cymal-5-solubilized and -purified ROS rhodopsin exhibits functional properties that are comparable, and even more desirable, with those of rhodopsin in DM.
The tracking of R*-induced changes in the structure of G ␣ over the course of R*/G-protein interactions and correlating these changes with the uptake of GTP␥S could be achieved using a series of one-dimensional NMR spectra (Figs. 3 and 6). These spectra, which could be acquired in a relatively quick fashion (Ͻ5 min), allowed us to observe time-dependent shifting and changes in signal intensity of protein amide/side chain-associated 1 HN, 15 N resonances and guanine nucleotide-associated 1 H resonances, over the course of the light-activated rhodopsin-catalyzed nucleotide exchange reaction (Fig. 3). The uptake of GTP␥S in the reaction could be monitored directly (Fig. 3B), and the conformation of the resulting product of the reaction, dissociated GTP␥S/Mg 2ϩ -bound 15 N-ChiT, could be analyzed in situ using NMR methods and compared with other functional states of ChiT previously generated and analyzed using similar NMR approaches (Fig. 4). It is worth noting that the addition of dark state rhodopsin to the NMR sample containing the 15 N-ChiT-reconstituted heterotrimer, and GTP␥S resulted in a significant reduction in the observed 15 N-ChiT amide and side chain proton signal intensity (Fig. 3A, light blue trace). This reduction in the signals, which was more pronounced than could be explained by dilution of the sample, suggests G t associates with the dark state of rhodopsin and is consistent with the results of plasmonwaveguide resonance experiments (53,54).
The apparent rate of change in ChiT signal intensity under our NMR experimental conditions suggests that heterotrimer dissociation proceeds with very slow kinetics. It has been fairly well established that G t␣␤␥ dissociates in the presence of GTP (55), although recent in vivo studies suggest that certain other G ␣ subtypes may not dissociate from G ␤␥ following R*-catalyzed guanine nucleotide exchange (56). Based on the differences in experimental variables and conditions used in studies examining heterotrimer dissociation (55,57,58), the measured rate of subunit dissociation may vary widely depending on the presence or concentrations of detergents, membranes, guanine nucleotides, activating compounds (AlF 4 Ϫ ), as well as receptors and downstream binding proteins/effectors. As with ROS rhodopsin, the HPTRX/CDEF-stimulated guanine nucleotide exchange reaction could be monitored using one-dimensional NMR methods and showed GTP␥S uptake as well as distinct signal changes in the amide resonances of 15 N-ChiT associated with guanine nucleotide exchange. However, in contrast to the observation of the release of GTP␥S/Mg 2ϩ -bound 15 N-ChiT in the native R*-catalyzed reaction, the NMR data collected for the HPTRX/CDEF-stimulated guanine nucleotide exchange reaction did not show similar evidence of heterotrimer dissociation from this soluble mimic of R* and subsequent release of activated 15 N-ChiT from G t␤␥ (Fig. 6). These observations, in combination with the results of G t activation and immunoprecipitation assays (Fig. 5, B and C), suggest that this GTP␥S/ Mg 2ϩ -bound form of ChiT persists as part of a nondissociated heterotrimer to which HPTRX/CDEF remains bound. Such a complex is analogous to the R*⅐G ␣␤␥ ⅐GTP intermediate of the R*/G t reaction pathway (Fig. 1B, step 4) and affords the opportunity to study the structure of G ␣ in this R*-bound state. Although it is not currently known as to why HPTRX/CDEF fails to dissociate from the GTP␥S/Mg 2ϩ -exchanged heterotrimer, it is possible that additional, key amino acids not present in this soluble mimic of R*, such as those found in the H8 region of rhodopsin, or a conformation not adopted by the grafted CD and EF loops of rhodopsin in the thioredoxin scaffold (38), are required to initiate heterotrimer release and dissociation.
Changes in the Carboxyl Terminus of G ␣ Accompanying R* Interactions-From the available G-protein crystal structures, particularly that for the heterotrimer (Fig. 1A), and the results of several biochemical experiments focused on R*/G-protein interactions, a number of different proposals (32,59,60,62) have been put forth to explain how structural changes in the receptor-interacting regions of the G-protein may lead to GDP release and GTP uptake. Each of the proposals suggest mechanisms that involve long range global changes in the structure of G ␣␤␥ upon interaction with R* to account for the fact that the guanine nucleotide-binding site in G ␣ is located ϳ30 Å from the proposed R*-interacting surfaces (see Fig. 1A), the so-called "action at a distance" hypothesis of heterotrimeric G-protein activation. Although the details of each of the proposed mechanisms varies considerably, a common element in all of them is a central role for the extreme carboxyl terminus of G ␣ in functioning as a recognition element in R*/G-protein interactions. Evidence in support of its involvement has been provided by experiments showing that ADP-ribosylation of Cys-347 on G t␣ by pertussis toxin uncouples G t from R* (16,17), mutational studies of G t␣ that have shown amino acid residues in the extreme carboxyl terminus are essential for R*-G t binding and R*-catalyzed activation (64 -66), synthetic peptide studies with various G t␣ -(340 -350) peptides that mimic the conformational effect of G t by stabilizing R* and/or competitively blocking R*/G t interactions (18,32,(67)(68)(69)(70)(71)(72), and chemical cross-linking of R* to G t followed by mass spectrometry of the labeled G t␣ peptide(s) (73). A potential route of communication between R* and the guanine nucleotide pocket involves alterations in the ␤ 6 /␣ 5 loop that are transmitted from the carboxyl terminus via the ␣ 5 helix (74,75). Mutations in the ␣ 5 helix have been shown to dramatically increase basal guanine nucleotide exchange rates and reduce R*-catalyzed activation rates. Furthermore, a mutation in the ␤ 6 /␣ 5 loop has been shown to mimic the action of R* by causing rapid guanine nucleotide exchange (76,77).
In this study, chemical shift perturbation of Phe-350 resonances observed to be associated with the interaction of light-activated rhodopsin and 15 N-ChiT-reconstituted heterotrimer indicate that the confor-mation of the carboxyl terminus of G ␣ completely shifts to an activated form in the GTP␥S/Mg 2ϩ -bound state, relative to the "ground" state conformation observed for GDP/Mg 2ϩ -bound ChiT. In comparison, the conformation of the carboxyl terminus in GDP/Mg 2ϩ -bound ChiT in the presence of AlF 4 Ϫ (Fig. 4, A and B) and in the reconstituted heterotrimer shows evidence of a slow exchange equilibrium between ground and activated conformations, with each state roughly equally populated in the AlF 4 Ϫ state and shifted more toward the activated conformation in the heterotrimer. Interestingly, the carboxyl terminus of ChiT in the HPTRX/CDEF⅐G ␣␤␥ ⅐GTP␥S/Mg 2ϩ -exchanged complex also appears to adopt a fully activated conformational form (Fig. 7A). In contrast to our observation of a discrete shift in the conformation of the carboxyl terminus of ChiT from a ground to an activated state as followed by shifting of the amide 1 H, 15 N cross-peak of Phe-350, solution NMR studies utilizing a segmental isotope-labeled G ␣ prepared through expressed protein ligation techniques have shown that certain side chains in the carboxyl-terminal region of the GDP/Mg 2ϩ -bound form of G ␣ lose intensity upon AlF 4 Ϫ adduct formation (78), suggesting an ordering of these side chains. Although a change in conformation of the carboxyl terminus of G ␣ upon AlF 4 Ϫ activation has also been shown by monitoring a fluorescent reporter group attached at position Cys-347 of G ␣ (61), the reason for the apparent differences in the backbone and side chain dynamics of the carboxyl terminus in the GDP⅐AlF 4 Ϫ /Mg 2ϩbound state is unclear and will require further NMR experiments to be resolved.
Changes in the Switch II Region of G ␣ Accompanying R* Interactions-In contrast to the observation of a similar behavior for the carboxyl terminus in these two GTP␥S-bound forms of ChiT, differences are observed for the behavior of resonances associated with the indole ring of Trp-207, which is located in the functionally important switch II region of G ␣ . In the R*-generated GTP␥S/Mg 2ϩ -bound form of ChiT, the Trp-207 indole resonances appear to be broadened beyond detec-tion, although resonances assigned to Trp-207, while also exchange broadened, are observed in a new chemical shift position in the spectrum of the HPTRX/CDEF⅐G ␣␤␥ ⅐GTP␥S/Mg 2ϩ -exchanged complex (Fig. 7B), relative to the position observed previously for the major conformations of this residue in other states of ChiT (34,35). Interestingly, however, the position of the cross-peak observed for Trp-207 in the HPTRX/CDEF⅐G ␣␤␥ ⅐GTP␥S/Mg 2ϩ -exchanged complex spectrum is close to the chemical shift position observed for a minor conformational form of the Trp-207 indole (Fig. 7C, denoted by an asterisk) in the spectra of both the GDP/Mg 2ϩ -bound 15 (22) does not provide any indication that the switch II region and/or other portions of G ␣ would display a range of conformational dynamics. However, the GTP␥S/Mg 2ϩ -bound G t␣ that was crystallized lacked residues 1-25 in the amino terminus, which were removed prior to crystallization by proteolysis. Because the G ␣ chimera used in our solution NMR experiments contained the full G t␣ amino-and carboxyl-terminal sequences, it is likely that the dynamics observed for the GTP␥S/Mg 2ϩbound form of ChiT is a consequence of the presence of these additional residues. In particular, the extreme amino terminus likely adopts a dynamically disordered form after G ␣ (GTP) is released from the heterotrimer (31), which may result in fast exchange dynamics locally, as well as cause more global intermediate to slow exchange dynamic effects in G ␣ . In this respect, it is interesting to note that the GTP␥S/Mg 2ϩ -bound form of ChiT generated independent of R* did not display a similar dynamic character, nor was the amide cross-peak of Phe-350 in this FIGURE 8. A schematic representation of the proposed conformational changes in G ␣ coupled to R*-catalyzed guanine nucleotide exchange. The model has been extended from a previous study examining changes in G ␣ accompanying heterotrimer formation (35). The GTPase and helical domains of G ␣ are shown in different shades of green, G ␤ is shown in yellow, and G ␥ in blue. The region representing the guanine nucleotide binding pocket in G ␣ is shown in blue in the 'ground' state, and in red or orange (see below) in the 'activated' states. GDP and GTP are shown in salmon and purple, respectively. Upon association of G ␣ (GDP) with G ␤␥ , G ␣ undergoes structural changes in the R* interacting amino-and carboxyl-terminal regions, as well as the switch regions surrounding the guanine nucleotide binding pocket, and suggest that although G ␤␥ binding changes the G ␣ structure to a 'preactivated' form, it displays at least two conformational states for the carboxyl terminus. These changes in the G ␣ structure may both potentiate R* interactions and pre-organize the guanine nucleotide binding pocket. In the 'empty pocket' state following R* catalyzed release of GDP, pre-organization of the guanine nucleotide binding pocket would facilitate GTP binding to form G ␣ (GTP)G ␤␥ , which ultimately dissociates into G ␣ (GTP) and G ␤␥ . The results of this NMR study suggest further changes in the carboxyl-terminal and switch II regions of G ␣ upon formation and dissociation of the R*⅐G ␣ (GTP)G ␤␥ complex. In particular, the carboxyl terminus appears to adopt a fully 'activated' state conformation, although switch II, as well as other regions of the G ␣ (GTP) structure, appear to become more dynamic upon heterotrimer dissociation. In this schematic, these changes in released G ␣ (GTP) are highlighted through different green shading for the GTPase domain that also has an altered conformation for the carboxyl terminus and lacks an ordered amino terminus, and the orange color for the guanine nucleotide binding pocket. Such changes in the structure of the amino-terminal region of G ␣ are consistent with observations from the crystal structures of the heterotrimer (26,27) and fluorescent and site-directed spin labeling experiments (31).
GTP␥S/Mg 2ϩ -bound state observed to shift completely to the activated position as seen with R*. These results further support the idea that it is specific changes in the conformation of the carboxyl and amino termini of ChiT associated with heterotrimer reconstitution and subsequent interaction of the heterotrimer with R* that result in the differences in the conformational dynamics observed for these GTP␥S/Mg 2ϩ -bound forms of ChiT.
Guanine Nucleotide Exchange Can Be Uncoupled from G ␣␤␥ Dissociation-Given that the carboxyl terminus and switch II regions of G ␣ have already been observed using NMR methods to adopt a preactivated conformation that may facilitate R* binding and subsequent GDP/GTP exchange in the heterotrimer (35), it would seem reasonable that further structural transitions in these two regions of G ␣ would be associated with R*-catalyzed GDP release and GTP uptake. Such structural coupling between distant portions of G ␣ is the basis for the proposed action at a distance hypothesis, which as mentioned above has been considered in most structure-based mechanisms of R*-catalyzed guanine nucleotide exchange. However, the observation of a trapped intermediate GDP/GTP exchanged heterotrimer state in complex with the soluble mimic of R* (Fig. 7) suggests that guanine nucleotide exchange is not sufficient for heterotrimer release from R* and subsequent dissociation of activated G ␣ from G ␤␥ . In particular, the observation that the switch II region of ChiT is in an activated conformation in the exchanged, complexed heterotrimer, together with the previous observation that the switch II region of ChiT already adopts a preactivated conformation in the GDP/Mg 2ϩ -bound heterotrimer, strongly suggests that changes in the switch II region of G ␣ associated with activation are not sufficient to drive heterotrimer dissociation as has been proposed previously (63).
Functional Implications of R*-induced Changes in G ␣ -We have developed a working model based on our previous NMR observations that also incorporates other fundamental results focused on elucidating changes in the structure of G ␣ that accompany heterotrimer formation and R* interactions (35). Based on the findings reported here, and elsewhere (31), we have extended this model (Fig. 8) to highlight the additional observed perturbations in the conformation of the receptor interacting amino-and carboxyl-terminal regions as well as switch II that arise upon formation and dissociation of the R*⅐G ␣␤␥ ⅐GTP complex (Fig. 1B, steps 4 and 5). Specifically, the carboxyl terminus, which displays at least two conformational states in the heterotrimer, is as noted above found exclusively in an activated state in both the R*⅐G ␣␤␥ ⅐GTP complex (the HPTRX/CDEF⅐G ␣␤␥ ⅐GTP␥S/Mg 2ϩ -exchanged complex) and in G ␣ (GTP) (the GTP␥S/Mg 2ϩ -bound form of ChiT), although Trp-207 is detected in only the R*⅐G ␣␤␥ ⅐GTP complex, suggesting an altered conformation for switch II in G ␣ (GTP). This latter apparent change in the conformational dynamics of G ␣ could be a consequence of the R*-induced changes in the structure of the amino-and carboxylterminal regions of G ␣ and/or subsequent guanine nucleotide exchange and dissociation of G t␤␥ . In either case, the increase in the apparent conformational flexibility of G ␣ , and switch II in particular, may have important functional ramifications for both interaction of G ␣ with downstream effectors, as well as GTP hydrolysis.
Conclusions-We have demonstrated previously the ability to express and isolate milligram quantities of an isotope-labeled G ␣ chimera ( 15 N-ChiT) that can be reconstituted with G t␤␥ to form functional heterotrimers (34,35). We have now shown that 15 N-ChiT-reconstituted heterotrimer forms functional complexes under NMR experimental conditions with light-activated, detergent-solubilized rhodopsin and a soluble mimic of R*, both of which trigger guanine nucleotide exchange. Collectively, the studies carried out to date demonstrate that solution NMR can be used to describe at atomic resolution changes in the structure and dynamics of G ␣ that accompany heterotrimer formation and signal transfer from R* to the G-protein. A key aspect of our approach is the ability to generate and interrogate trapped R*-bound conformations of G ␣ , which have so far proven refractory to analysis by other high resolution structural methods. Overall, our work demonstrates that new insights into the structure of G ␣ can be gained from these methods, which will augment our understanding of the structural mechanism(s) underlying signal transduction that are currently based on available high resolution, static crystal structures for G ␣ in various states. Future efforts aimed at NMR analysis of other R*-bound conformations of G ␣ , in particular the R*⅐G ␣␤␥ [empty] complex, should provide additional information about the conformational changes in G ␣ accompanying R*/G-protein interactions.