|Home | About | Journals | Submit | Contact Us | Français|
GPCRs catalyze nucleotide release in heterotrimeric G proteins, the slow step in G protein activation. Gi/o family proteins are permanently, cotranslationally myristoylated at the extreme amino terminus. While myristoylation of the amino terminus has long been known to participate in anchoring Gi proteins to the membrane, the role of myristoylation in regards to interaction with activated receptors is not known. Previous studies have characterized activation-dependent changes in the amino terminus of Gα proteins in solution [Medkova, M. (2002) Biochemistry 41, 9963-9972; Preininger, A.M. (2003) Biochemistry 42, 7931-7941], but changes in the environment of specific residues within the Gαi1 amino terminus during receptor-mediated Gi activation has not been reported. Using site-specific fluorescent labeling of individual residues along a stretch of the Gαil amino terminus, we found specific changes in the environment of these residues upon interaction with activated receptor and following GTPγS binding. These changes map to a distinct surface of the amino-terminal helix opposite the Gβγ binding interface. The receptor-dependent fluorescent changes are consistent with a myristoylated amino terminus in close proximity to the membrane and/or receptor. Myristoylation affects both the rate and intensity of receptor activation-dependent changes detected at several residues along the amino terminus (with no significant effect on the rate of receptor-mediated GTPγS binding). This work demonstrates that the myristoylated amino terminus of Gαil proteins undergoes receptor-mediated changes during the dynamic process of G protein signaling.
GPCRs act as GEFs for heterotrimeric G proteins, catalyzing nucleotide exchange on Gα subunits, which is the rate-limiting step in G protein activation. Agonist-mediated activation of GPCRs is coupled to GDP release by a mechanism which can be probed using biophysical techniques. While crystal structures for a number of G proteins and their respective subunits (1-7), as well as those of the prototypical GPCR, rhodopsin (8, 9), and more recently the β2-adrenergic receptor (10), reveal much of what we know about the structure and function of these proteins, there is currently no high-resolution structure of an activated receptor in complex with its cognate G protein. Thus, biophysical and biochemical studies provide valuable insights regarding the structural determinants and dynamics of G protein activation. A functional heterotrimer is known to be essential for receptor-mediated G protein activation; the amino-terminal region of Gα directly participates in heterotrimer formation, facilitating high affinity Gα–βγ binding (11, 12).
The amino terminus of Gα (as well as the C terminus (13, 14) and α4-β6 loop (15)) have been implicated in receptor-mediated interactions in a number of biochemical studies. One of the first reports of such an interaction arose from the observation that a 15-residue peptide encompassing amino-terminal residues 8-23 of Gαt competitively inhibited rhodopsin-Gt interaction but alone did not stabilize metarhodopsin II formation (16). More recently, activated rhodopsin was found to crosslink to residues 19-28 spanning the amino-terminal region of native Gαt (17). The amino terminus of Gαq has also been implicated in the selectivity of receptor coupling, as addition of amino-terminal Gαq-specific residues to the amino terminus of Gαi/o proteins conferred the ability of this chimera to couple to Gq-coupled receptors, as measured by phosphatidylinositol hydrolysis (18). These studies indicate a specific interaction between the amino terminus of Gα proteins and activated receptors.
Gαi/o family members (including Gαt) are permanently, cotranslationally modified by myristate (consisting of 14 carbons) at the amino-terminal glycine residue, increasing hydrophobicity of this region. Myristoylation enhances Gα–βγ subunit association (19), but it is not essential for this association. In vitro, myristoylation of Gα and farnesylation of the γ subunit of Gβγ are functionally redundant in facilitating coupling to activated receptors (20), as the ability of an unmyristoylated Gα protein (in complex with Gβγ) to undergo receptor-mediated GTPγS binding was dependent on farnesylation of the Gγ subunit. A recent study on the functional significance of myristoylation which was conducted in transgenic mice revealed a role for myristoylation in deactivation of Gαt subunits. In mice expressing a myristoylation deficient form of Gαt, the fraction of unmyristoylated proteins properly localized to rod outer segments (ROS) were found to be activation competent, with the remainder of Gαt mislocalized to inner compartments (21). The properly localized (unmyristoylated) Gαt fraction in the ROS was able to bind Gβγ and undergo receptor-mediated GTPγS binding similar to native Gαt, but had defects in deactivational processes such as GTPase activity and PDE-mediated GAP activity (21). These data point to a role for myristoylation beyond that of simple membrane localization and Gβγ binding.
Using biochemical and biophysical studies of purified, fluorescently labeled myristoylated (and unmyristoylated Gαil proteins) and membrane bound receptors, we demonstrate here that specific amino-terminal residues of Gαil proteins undergo receptor-mediated changes during the dynamic process of G protein activation. To examine changes at individual positions along the amino terminus, residues of interest in the amino terminus of a Gαil protein lacking solvent-exposed cysteines (Gαil Hexa I) were mutated to cysteine, followed by thiol-directed fluorescent labeling. The receptor-mediated environmental changes in individual amino-terminal Gαil residues reported here, and the effect of myristoylation on those changes, together shed light on receptor-mediated changes that are communicated through the amino terminus of Gαil proteins.
GDP and GTPγS were purchased from Sigma-Aldrich (Milwaukee, WI). BODIPY® 630 methyl bromide (8-Bromomethyl-4,4-difluoro-3,5-bis-(2-thienyl)-4-bora-3a,4a-diaza-s-indacene, MW 449, (BD630) and Alexa Fluor® 594 C5-maleimide, (A1) were purchased from Invitrogen (Madison, WI). All other reagents and chemicals were of the highest available purity.
In order to examine changes in specific residues using thiol-directed labels, we used a parent Gαil in which solvent-exposed cysteines were conservatively replaced by site-directed mutagenesis, and then this construct was used as a basis for site-specific cysteine mutagenesis within the amino terminus. Construction, expression, purification and fluorescent labeling of Gαil and Gαil Hexa I proteins (unmyristoylated and myristoylated) were performed as described previously (22, 23), with minor modifications. Briefly, the parent Gαil Hexa I lacking solvent-exposed cysteines was generated with an expression vector encoding rat Gαi1 with six amino acid substitutions at solvent-exposed cysteines (C3S-C66A-C214S-C305S-C325A-C351I) and a hexahistidine tag between amino acid residues M119 and T120. This construct served as the template for introducing individual cysteine substitutions at sites of interest using the QuickChange system (Stratagene, La Jolla, CA) as previously described; DNA sequencing confirmed all mutations. The mutant constructs are referred to as Gαil Hexa I, followed by the site of cysteine substitution, and followed by M if the protein is expressed as a myristoylated protein. For example, Hexa I-13CM refers to the Gαil Hexa I parent protein containing a cysteine substitution at the 13th residue and expressed with N-myristoyl transferase to obtain myristoylated protein, whereas Hexa I-13C is the same protein expressed in the absence of the N-myristoyl transferase vector to produce the unmyristoylated counterpart. The mutant constructs were expressed in E. coli BL21-Gold (DE3) with or without a N-myristoyl transferase (NMT) vector pbb131, which also encodes kanamycin resistance for selection of myristoylated proteins (generously provided by M. Linder, Washington University), and purified as detailed previously (23). Coomassie staining of urea SDS PAGE gels (24) demonstrate Gαil Hexa I proteins are fully myristoylated (23) when co-expressed with the NMT vector. Purified proteins were stored at -80 °C in buffer A containing 50 mM Tris, 100 mM NaCl, 2 mM MgCl2, 10 μM GDP, pH 7.5, and with 10% (v/v) glycerol; wild-type Gαil containing native, solvent-exposed cysteines was additionally supplemented with 5 mM β-mercaptoethanol or 1 mM DTT prior to storage at -80 °C. After purification, all proteins used in this study were greater than 85% pure, as estimated by Coomassie staining of SDS-polyacrylamide gels.
Gαil Hexa I proteins were labeled as described previously (22, 23) using A1 probe (Invitrogen, WI) with a labeling time of 1 hour. Extensive washing with buffer A in a 10 kDa molecular weight concentrator separated labeled protein from unbound probe, followed by gel filtration/spin purification of labeled proteins, resulting in a labeling efficiency between 0.25-0.6 mollabel/molprotein for all A1-labeled Hexa I proteins (measured by comparing label to protein concentration, ε588nm = 96,000 cm-1•M-1 for the A1 probe and protein concentrations as determined by the Bradford assay). All labeled Gαil Hexa I proteins used in these assays demonstrated ≥ 40% increase in intrinsic tryptophan emission upon AlF4- activation after labeling, as expected for properly folded and functional Gα proteins (25). This is consistent with previous results demonstrating amino-terminally labeled Gαil Hexa I proteins are activation competent and retain the ability to bind to Gβ1γ1 (22, 23). Gβ1γ1 (native, bovine, containing multiple solvent-exposed cysteines) was labeled as described above with the thiol-reactive probe BD630 (Invitrogen, Madison, WI), resulting in a labeling efficiency of >0.85 mollabel/molprotein, using ε635nm = 73,000 cm-1•M-1.
Urea-washed rod outer segment membranes (ROS) containing dark-adapted rhodopsin were prepared as previously described (26) and stored protected from light at -80 °C in a buffer containing 10 mM MOPS, 200 mM NaCl, 2 mM MgCl2, 1 mM dithiothreitol (DTT) and 100 μM phenylmethylsulfonyl fluoride (PMSF). ROS detergent solubilization for supplementary experiments was performed as follows (42): under dim red light, urea-washed dark adpated ROS membranes were solubilized in 50mM Tris, pH 7.5, containing 100 mM NaCl and dodecylmaltoside (DM) to a final concentration of 0.5% at 4 °C with gentle agitation for 15 minutes. Insoluble material was removed by centrifugation at 20,000 × g for 1 hr at 4 °C and solubilized rhodopsin concentration in the supernatant was determined using ε500 = 40,600 cm-1M-1. DM-solubilized dark rhodopsin thus obtained was substituted for ROS membranes in the assay of receptor-mediated fluorescence intensity changes upon light activation in labeled proteins as described below. These supplementary assays (conducted with soluble receptors) employed a final concentration of 750 nM DM-solubilized rhodopsin, 0.01% DM and 200 nM A1-labeled Gαil Hexa I protein in complex with Gβ1γ1 in assay buffer, to determine the ability of solubilized rhodopsin to mediate light dependent changes in emission from amino-terminally labeled Gαil Hexa I proteins in complex with native Gβ1γ1. Gβ1γ1 (hereafter referred to as Gβγ) was purified from bovine retina as described previously (26) and stored in a buffer containing 10 mM Tris, 100 mM NaCl, 5 mM β-mercaptoethanol (BME), pH 7.5, and with 10% (v/v) glycerol.
To monitor the ability of Gαi1 Hexa I proteins to undergo activation, both before and after labeling, intrinsic tryptophan (Trp207 in Gαt, analagous to Trp211 in Gαi) fluorescence was measured as previously described (25). Briefly, labeled proteins (100 nM) in buffer A (50 mM Tris, 100 mM NaCl, 1 mM MgCl2, pH 7.5) containing 10 μM GDP were monitored by excitation at 280 nm and emission at 340 nm before and after the addition of AlF4- (10 mM NaF and 50 μM AlCl3) using a Varian Cary Eclipse spectrofluorometer. Properly folded and functional Gα subunits demonstrate a ≥ 40% increase in intrinsic tryptophan emission upon AlF4-activation, in comparison to emission in the basal, GDP-bound state (25).
Subunit binding was measured in buffer A (supplemented with 10 μM GDP) using increasing amounts of Gβγ and 10 nM A1-labeled unmyristoylated Gαil Hexa I-3C (as previously described for fluorescently-labeled Hexa I proteins (22)), with excitation/emission (ex/em) 570/616 nm. Changes in maximal emission (compared to basal emission in the absence of Gβγ) were plotted as a function of increasing Gβγ concentration and fitted using a sigmoidal dose-response model using Prism 4.0 (Graphpad Software). For myristoylated proteins, the binding affinity was determined by FRET between labeled subunits (23), which are relatively solvent-protected and immobile even in the absence of Gβγ proteins (23), preventing efficient detection of binding by the relatively simpler protocol used for unmyristoylated proteins. Binding between A1-labeled myristoylated Gαil Hexa I-3CM (10 nM) and increasing amounts of Gβγ-BD630 in buffer A was measured as a decrease in donor emission as a result of FRET, with ex/em 570/612 nm, less emission changes in the absence of labeled Gα subunits, and fitted to a sigmoidal dose-response curve with Prism 4.0 (GraphPad Software).
Labeled Gαil Hexa I proteins were reconstituted with Gβγ at 4 °C prior to addition of excess dark ROS, which was then transferred to a cuvette containing buffer A, with final concentrations of 400 nM Gα protein, 800 nM Gβγ, and 2 μM rhodopsin. All manipulations involving dark rhodopsin were performed in the dark or under dim red light to prevent its activation. The emission (dark spectra) from the cuvette containing dark ROS and A1-labeled subunits in complex with Gβγ was scanned between 580-750 nm (5 nm bandpass) with excitation at 575 nm (2.5nm bandpass) at 21 °C using a Varian Cary Eclipse. After collecting dark spectrum, cuvettes were subjected to light activation, and light spectra were collected in duplicate 4 minutes after the photoactivation of rhodopsin by a Vivitar electronic flash assembly. Finally, GTPγS spectra were collected in duplicate 4 minutes after addition of GTPγS (40 μM) to the light activated sample in the cuvette. While duplicate readings were averaged for the light and GTPγS spectra, the dark spectrum was collected only once prior to light activation in order to avoid spurious activation of rhodopsin.
The rate of amino-terminal changes in A1-labeled Gαil Hexa I residues upon light activation of rhodopsin was measured by monitoring the increase in emission at 617 nm (excitation 575 nm) of indicated labeled Gαil Hexa I subunits reconstituted with Gβγ (200 nM each) in the presence of dark-adapted rhodopsin (200 nM) at 21 °C both before (basal) and within seconds after activation by light, and monitored over time. The data were plotted as emission increase compared to basal emission prior to receptor activation, and the rate was determined by plotting the data as a percent of maximal and fitting the data to an exponential association curve using Prism 4.0. The rate of changes in A1-labeled Gαil residues upon GTPγS binding was measured in a similar manner. The emission at 617 nm (excitation 575 nm) was monitored at 21 °C both before and after addition of GTPγS (10 μM) to indicated labeled Gαil Hexa I subunits complexed with Gβγ (200 nM each) in the presence of light-activated rhodopsin (200 nM) for 30 minutes. Receptor-free nucleotide exchange (basal exchange) in the absence of rhodopsin in reconstituted heterotrimers is performed essentially as described above to obtain GTPγS spectrum, with emission monitored at 617 nm over a period of 45 minutes, to examine changes in environment of the labeled residue in the absence of ROS membranes. The data (average of 3 or more independent experiments) were normalized to the level of fluorescence before GTPγS addition (set to 1.0), and the rate was determined by fitting the data to an exponential dissociation curve using Prism 4.0.
The rate of receptor-catalyzed nucleotide exchange was measured by monitoring the time-dependent increase in the Trp211 fluorescence (ex/em 290/340 nm) of Gαil proteins reconstituted with Gβγ (200 nM each) in the presence of 100 nM rhodopsin (light activated) at 21 °C following the addition of GTPγS (10 μM). The data (average of 3 or more independent experiments) were normalized to the baseline (0%) and the fluorescence maximum (100%). The exchange rate was determined by fitting the data to an exponential association curve using Prism 4.0 (GraphPad Software).
The ability of labeled myristoylated and unmyristoylated Gαil proteins to bind rhodopsin in urea-washed ROS membranes was determined by adding aliquots of the indicated labeled Gαi Hexa I proteins (5 μM) preincubated with an excess of Gβγ (10 μM) to rhodopsin (50 μM) in a buffer containing 50 mM Tris (pH 8.0), 100 mM NaCl and 1 mM MgCl2 in the dark, after light activation, and with addition of GTPγS (40 μM) to a light activated sample. Following 30 minutes incubation at 4 °C, the membranes in each sample were pelleted by centrifugation at 20,000 × g for 1hour, and supernatants removed from pellets. For the dark samples, reactions were foil-protected from light until removal of supernatant, which was performed under dim red light. The isolated supernatant and pellet fractions were boiled, resolved by SDS-PAGE, visualized with Coomassie blue and quantified by densitometry using BioRad Multimager. Each treatment (dark, light, and light + GTPγS) was quantified by comparison of amount of 40 kDa Gαil proteins in either pellet or supernatant to the sum total of Gα in both fractions, as identified by co-migration with Gαil standard, and expressed as a percentage of the total Gα present in each given sample. Data are average of at least 3 independent experiments.
In order to measure receptor-mediated changes in the amino terminus of Gαil proteins, in the presence and absence of myristoylation, we labeled a number of individual amino-terminal Gαil Hexa I proteins at specific side chain positions (figure 1A). A Gαil Hexa I parent protein lacking solvent-exposed cysteines was used as a template for introduction of cysteine residues at positions indicated in figure 1A, expressed as both myristoylated and unmyristoylated proteins. Both wild-type Gαil and Gαil Hexa I proteins, in complex with Gβ1γ1 (Gβγ), can productively interact with rhodopsin (as seen in receptor-mediated GTPase and 35[S]-GTPγS binding (22, 27, 28)).
Prior to using fluorescently-labeled Gαil proteins as reporters of Gα environmental changes in an experimental protocol involving rhodopsin, it was necessary to select a fluorescent probe with spectral characteristics that do not overlap those of rhodopsin. Rhodopsin and its photoactivated metarhodopsin intermediates MI, MII, and MIII exhibit absorbance spectra which can interfere with many commonly used fluorescent probes. Highly conjugated probes with excitation maxima in the far red range (where rhodopsin and its intermediates are not affected) were therefore sought for this study. We examined three highly conjugated fluorescent probes with spectral characteristics in the far red range: BODIPY595, Atto610, and Alexa595 (A1). Both Atto610 and A1 probes effectively labeled desired residues with an efficiency of labeling between 0.25-0.6 mollabel/molprotein, while BODIPY labeling of Gα subunits was much less efficient (0.1 mollabel/molprotein), likely due to poor solubility of this probe in buffer conditions required to maintain optimal activity of Gα subunits. The Gαil Hexa I proteins successfully labeled with the Atto610 fluorescent probe underwent significant photobleaching under our assay conditions, in contrast to Gαil Hexa I proteins labeled with A1, which were relatively photostable. Although not all fluorescent molecules are environmentally sensitive, the A1 probe reports changes in the polarity of its environment, as free, unreactive A1 probe alone demonstrated an increased emission intensity in methanol as compared to its emission in aqueous buffer (figure 1B).
Labeling amino-terminal residues of Gαil Hexa I proteins with the A1 probe does not perturb the ability of these proteins to undergo activation-dependent changes, nor does it prevent binding to Gβγ subunits, consistent with previous studies using a variety of fluorescent probes (22, 23). For example, in figure 1C, activation-dependent changes induced by the transition state mimetic, AlF4- result in a ≥ 40% increase in intrinsic tryptophan fluorescence, a prerequisite for use of the labeled proteins in these studies. Gβγ binds labeled Gαil Hexa I proteins; the dose-dependent changes in emission from protein labeled at the third residue upon binding to Gβγ (figure 1D) resulted in an apparent Kd of 30 nM and 230 nM for myristoylated (1D, top) and unmyristoylated (1D, bottom) proteins, respectively. These values are consistent with previously published values for the interaction between fluorescently labeled Gαil Hexa I subunits and Gβγ (22,23), and the relatively higher affinity of Gβγ for myristoylated subunits is consistent with the known myristoylation-dependent enhancement of subunit affinity (29).
After confirming that labeled Gαil Hexa I proteins are properly folded and functional, we sought to determine whether these proteins could report interactions with activated, membrane-bound receptors, using an experimental protocol depicted in figure 2A. In vivo, rhodopsin is coupled to Gt activation. Both Gαi (and Gi family member Gαt) functionally couple to rhodopsin in vitro, and both undergo cotranslational myristoylation, but unlike Gαt, Gαi expresses well in E. coli. Both myristoylated and unmyristoylated Gαil subunits, in complex with native, acylated Gβ1γ1 (Gβγ) subunits, are known to bind to rhodopsin upon light activation, and to be released from the membrane upon subsequent GTP binding (14, 20); the goal of these studies was to examine changes in specific amino-terminal residues during this process. Therefore, emission of a fluorescently labeled Gαil Hexa I protein (reconstituted with Gβγ) was first scanned in the presence of ROS membranes containing dark-adapted rhodopsin, resulting in the basal (dark) emission spectrum to which later changes were then compared. Emission was scanned again after light activation (catalyzing GDP release and generating the empty pocket state), and finally after addition of GTPγS to the cuvette, which leads to dissociation from the receptor, enabling both Gα and Gβγ to interact with their respective downstream effectors. Normalizing emission from labeled G protein to that obtained in the basal (dark) states eliminates effects associated with protein-lipid interaction which occur independently of receptor activation.
Given the fact that rhodopsin is light sensitive and demonstrates multiple absorbance maxima, the probe selected for these studies must exhibit spectral characteristics in the far red range, to prevent overlap with rhodopsin. The majority of rhodopsin absorbance (and that of its photo-intermediates) occurs at wavelengths below 500 nm. Studies by Imamoto et al. demonstrate that the A1 probe is compatible with assays involving rhodopsin (30). In control experiments to confirm this, rhodopsin was scanned from 580-750 nm (in the absence of labeled protein) in the dark, after light activation, and upon addition of GTPγS, with excitation at 575 nm. Rhodopsin does not substantially contribute to the overall emission signal, as seen in supplementary figure S1A and resulting difference spectra (inset, figure 2B). As a further control, we also measured emission changes from free probe (made unreactive by treatment with DTT) in the presence of ROS membranes, in the absence of G protein. The free, unreactive A1 probe in the presence of dark ROS showed no evidence of light- or GTPγS- dependent changes under our experimental conditions (supplementary figure S1B).
Using the protocol depicted in figure 2A, we examined the emission spectra of myristoylated (figure 2B) and unmyristoylated (figure 2C) Gαil Hexa I proteins labeled at the third residue in complex with Gβγ, first in the presence of inactive rhodopsin (black traces) and again upon receptor activation (red traces). Both myristoylated and unmyristoylated labeled proteins demonstrated robust changes in their environment upon receptor activation, relative to their respective environments in the inactive state. Upon receptor-mediated GTPγS binding, only the unmyristoylated protein demonstrated a decrease in fluorescence, corresponding to a relatively more solvent-exposed (aqueous) environment for the unmyristoylated amino terminus upon GTPγS binding (figure 2C, green). The myristoylation-dependent difference in environment upon GTPγS binding is not due to the presence of membrane lipids in the experiment, as this difference is still evident in a membrane-free, receptor-free environment (figure 2D) as a result of basal (unstimulated) nucleotide exchange. Furthermore, we confirmed the ability of this myristoylated protein to undergo GTPγS-dependent changes in solution by intrinsic Trp211 fluorescence, which was found to increase in a manner similar to wild type upon addition of GTPγS (not shown). Therefore, there is a similar degree of solvent protection in both the heterotrimeric and the GTPγS-bound state for myristoylated protein labeled at the third residue (figures 2B and D). Together these results are consistent with a myristoylation-dependent solvent protection of the extreme amino terminus upon GTPγS binding.
We next examined relative differences in emission of 9 individually labeled amino-terminal residues during the G protein cycle in myristoylated and unmyristoylated proteins, which reflect changes in the probe's environment upon activation and deactivation. Comparison of the emission of each one of the individually labeled proteins upon receptor activation and GTPγS binding (figures 3A-B) paints a picture of the relative changes in environment for each residue as the G protein moves from the inactive, heterotrimeric, dark state (obtained in the presence of dark adapted ROS), to the receptor-bound empty pocket (light) state, and finally to the GTPγS-bound state. Although myristoylation confers a greater degree of membrane association for heterotrimeric Gα subunits in the dark (figure 3C-D), both myristoylated and unmyristoylated subunits efficiently translocate to membrane fractions upon light activation. Because introduction of membranes to buffer a cuvette containing soluble G proteins can induce a small variation in signal due to scatter, we set the basal emission of the labeled proteins in the presence of dark adapted ROS to 1.0; all subsequent changes were normalized to basal. This protocol accounts for contributions to basal emission from membrane association of myristoylated proteins in the dark, as well as small variations in labeling efficiencies at different residues.
Since the A1 probe demonstrates decreased emission intensity in a more aqueous environment and increased intensity in a more hydrophobic environment (figure 1B), the receptor-mediated increases in emission upon light activation from the labeled proteins (figure 3A-B) likely represent increases in the hydrophobicity of the environment detected by the fluorescent probe. The intensities exhibited for the labeled proteins are shown relative to that detected for each one in the inactive, heterotrimeric state (as measured in the presence of ROS membranes containing dark-adapted rhodopsin, figure 3A-B, black bars, and figure 2A, left panel). The light-activated state shown by the red bars in figure 3 effectively represents the receptor-bound, empty pocket conformation depicted in figure 2A, center panel. Light activation is accompanied by increased emission intensity at nearly all positions examined along the amino-terminal helix (figures 3A-B, red bars), as compared to their inactive, heterotrimeric states (black bars). Light-dependent changes in fluorescence were accompanied by enhanced membrane association (where the relative percentage of Gα in the pelleted fraction exceeds the percentage of Gα in the soluble fraction) for both myristoylated and unmyristoylated proteins (figures 3C-D), consistent with activation-mediated binding of labeled proteins to membrane-bound receptors. This localization is reversed by addition of GTPγS (figures 3A-B, green bars) to the light-activated samples, indicating increased solubility upon GTPγS binding.
The light-dependent changes in fluorescence ranged from 25-40% increase in emission over basal upon light activation of receptor in myristoylated proteins labeled at the 3rd, 13th, 17th, and 21st residue (figure 4A). Myristoylation-dependent differences in emission intensity were statistically significant at several positions (figure 4A). These changes (and those related to receptor-mediated GTPγS binding) are eliminated when labeled G proteins are boiled prior to use in the assay (10 minutes at 95 °C). The emission changes in G proteins which occur after flashing light on the system are also ablated when heat denatured ROS membranes (10 minutes at 95 °C) are substituted for ROS membranes kept at 4 °C in the experiment (in conjunction with properly folded G-proteins). The lack of light-dependent changes exhibited using after heat-denaturation of either G proteins or rhodopsin suggest the changes observed are directly related to the proper folding and functioning of these proteins.
While enhanced membrane localization of myristoylated proteins might be predicted to mediate the increases in intensity upon light activation, it is important to keep in mind that these changes are relative to their environment in the dark state. Both myristoylated and unmyristoylated proteins efficiently translocate to membrane fractions upon light activation, with the myristoylated proteins demonstrating a significantly higher degree of membrane association in the dark state than unmyristoylated (figure 3C-D), which results in a smaller relative difference between the dark versus light membrane localization for myristoylated proteins, and likewise a larger difference between these two states for the unmyristoylated protein. If membrane localization was solely responsible for light-dependent changes in emission, then the myristoylated proteins would be predicted to show a smaller difference in emission upon light activation than the unmyristoylated protein. However, this is not the case. This argues against membrane translocation as the cause of the myristoylation-dependent enhancement of emission we observe upon receptor activation.
Although this study is focused on the relationship between membrane-bound receptors and amino terminal residues of Gαil proteins, and although detergents would be expected to diminish interactions between membrane bound receptors and myristoylated proteins, we investigated the role of membrane lipids in this system using a detergent solublized form of rhodopsin and labeled, myristoylated Gαi1 Hexa I proteins. The labeled proteins retained the ability to report changes in emission upon light activation of receptors, albiet to a slightly lesser extent to that seen using intact ROS membranes, which were eliminated by heating the soluble rhodopsin preparation for 10 minutes at 95 °C prior to use in these experiments (supplemental figure 2).
We further characterized the time-dependence of the increase in emission upon light activation from myristoylated and unmyristoylated proteins labeled at the 21st residue (figure 4C), which differed significantly upon receptor activation (figure 4A). The rate of the light-dependent change is considerably faster for the myristoylated protein (0.023 sec-1), in comparison to its unmyristoylated counterpart (0.012 sec-1). These results suggest myristoylation can enhance both the rate and intensity of receptor-catalyzed changes in fluorescence emission seen upon light activation.
Finally, binding of GTPγS increased the solvent exposure of nearly all of the residues along the amino terminus (figures 3A-B, green bars versus red bars) as compared to their environment in the light activated state. Myristoylation dramatically inhibited the GTPγS-mediated change in solvent exposure at position 3, and had a more subtle effect at residues 10 and 13 (figure 4B). We characterized the time-dependence of the decrease in emission intensity upon GTPγS binding at the 13th residue (figure 4D), which was significantly altered by myristoylation as seen in figure 4B. The rate of change in emission which accompanies GTPγS binding in the presence of light-activated receptor shown in figure 4D was nearly the same for myristoylated and unmyristoylated labeled proteins (0.018 sec-1 and 0.022 sec-1, respectively) a similarity which is confirmed using unlabeled proteins, described below. Therefore, myristoylation-dependent differences in emission intensity which accompanies GTPγS binding does not necessarily result in differences in the rates of GTPγS binding from the labeled proteins.
Since we detected no significant differences in the rates of GTPγS binding in labeled proteins (despite some differences in intensity), we confirmed this finding by comparison of these rates in unlabeled proteins (figure 5). This was accomplished by taking advantage of the fact that the intrinsic tryptophan fluorescence emission (reported by Trp211 located in the Switch II region) increases upon GTPγS binding, and the time dependence of this increase can be measured in the presence of light-activated rhodopsin (14) which is able to catalyze nucleotide exchange in Gi (as well as Gt) proteins. Gαil Hexa I-βγ complex is known to undergo a slightly higher rate of nucleotide exchange than wild-type Gαilβγ (14), shown in figure 5B for comparison. This effect is due to conservative replacement of native solvent-exposed cysteines, a prerequisite to generation of the cysteine-depleted Gαil Hexa I parent protein, previously shown to have functional activities similar to wild-type Gαil (22, 23). Receptor-mediated nucleotide exchange rates for myristoylated and unmyristoylated proteins reveal no significant differences in the rates of nucleotide exchange for myristoylated proteins and unmyristoylated Gαil Hexa I proteins (figures 5A-B), similar to results obtained with labeled Gαil Hexa I proteins. Furthermore, the slightly higher (but statistically indistinct) rate of receptor-mediated nucleotide exchange in unmyristoylated proteins is not likely a result of increased basal exchange, as the exchange rate of unmyristoylated Gαil Hexa I -3C (figure 2D) is 0.007 sec-1, which is one-third the receptor-catalyzed rate (figure 5B). Therefore it is unlikely that receptor-independent activation of Gα is a major contributor to exchange rates in these unmyristoylated proteins. The ability of unmyristoylated proteins to undergo efficient receptor-mediated nucleotide exchange also confirms that myristoylation is not a requirement for productive interaction with membrane-bound receptors, consistent with results from these and other published studies (both in vitro and in vivo) (20, 21).
Closer examination of the emission maxima (emmax) of the free A1 probe in aqueous buffer solution versus emission in a more hydrophobic environment such as methanol (figure 1B) reveals a small but reliable shift of the emmax to lower wavelengths (blue-shifted emission) in a less aqueous environment, when plotted as percent of each of their respective maximum intensities (figure 6, top panel). The emmax of an equimolar amount of free A1 probe shifted from 617 nm in aqueous buffer to 609 nm in methanol, a change of 8 nm. The emission of A1-labeled proteins also demonstrated a small but detectable blue shift upon receptor activation, on the order of 1-4 nm, which are reversed by GTPγS binding (figure 6). Closer examination of the magnitude of these shifts reveals a relatively greater blue shift for proteins labeled at the 3rd, 13th, 17th, and 21st residues. While this probe reports more robust changes in overall emission intensity than in the relative smaller shifts seen in emission maxima, the intensity changes (figure 3A-B) are generally consistent with blue shifts seen upon light activation, which are reversed by GTPγS binding (figure 6).
Biophysical studies have broadened our understanding of receptor-mediated G protein activation, and in combination with crystal structures, provide powerful tools to probe G protein signaling. The most complete crystallographic views of Gα amino-terminal structure have been obtained with unmyristoylated Gα proteins in complex with Gβγ, which shows the Gα amino-terminal α-helix making extensive contacts with Gβγ (figure 1A). Since isolated Gα proteins are unstable in the nucleotide-free state, and since there is no high resolution structure of an activated receptor in complex with a G protein, biophysical studies offer unique insights into the conformation of receptor-bound, empty pocket state of Gα subunits. Our results demonstrate that residues all along the amino-terminal helix of Gαil Hexa I proteins undergo specific receptor-mediated changes upon activation. These changes map to a distinct surface of the amino-terminal helix, suggesting intimate involvement of the amino terminus of Gα in changes that accompany receptor-mediated G protein activation processes. While these data do not rule out contributions from membrane lipids in receptor-mediated amino-terminal environmental changes observed in Gαil Hexa I proteins, such contributions are secondary to receptor activation, as heat denaturation of rhodopsin (as well as heat denaturation of soluble rhodopsin) eliminates these changes, and because these changes persist in the presence of detergent-solubilized receptors.
The fluorescent probe linked to individual amino-terminal Gαil Hexa I residues reports changes in its environment during signaling. The current data, including observations that the amino terminus can crosslink to activated receptors (17) and the ability of amino-terminal peptides to compete with Gαt for stabilization of activated rhodopsin (16) are altogether consistent with a specific interaction between amino terminal residues of Gα with activated receptors. These interactions result in changes in the environment of individually labeled residues.
In the heterotrimeric (basal) state, this environment is depicted in dark spectra, which are obtained in the presence of membrane-bound, inactive receptor. The light spectra, obtained after light activation of receptor and before GTPγS addition, generates the receptor-bound, empty pocket conformation. The light-activated emission spectra reflect changes in the environment of the probe linked to individual amino-terminal residues that occur upon receptor-mediated G protein activation. Myristoylation enhanced not only the intensity, but also the rate of the light-mediated changes in emission upon light activation. The myristoylation-dependent increase in the rate of light-dependent emission changes may be due to enhanced G protein binding to ROS in the dark state, as increased G protein membrane localization prior to light activation would reduce the factors affecting the rate of light activation from a 3-dimensional to a 2-dimensional diffusion issue.
The increases in intensities upon light activation, accompanied by blue-shifted emission maxima, together indicate a less aqueous-accessible environment upon receptor activation. Furthermore, several of the residues which demonstrate both the greatest emission shifts and intensity changes upon receptor activation (residues 13, 17 and 21) map to a distinct face of the Gα amino-terminal α-helix (figure 7A) located opposite the Gβγ binding interface (figure 7B). The native residues at positions 13, 17, and 21 (V13, K17, and R21) in Gαi proteins would be predicted to promote membrane interactions through hydrophobic and electrostatic interactions, which may augment interactions of G proteins with membrane bound receptors in vivo. Despite the fact that mutation to cysteine and addition of the A1 probe did not perturb the ability of the labeled proteins to undergo activation-dependent changes, to bind Gβγ, to exchange nucleotide or to report activation-dependent changes, it is not unreasonable to expect that unlabeled, native proteins may interact even better with membrane bound receptors than the labeled proteins used in this study. Nevertheless, since heat denaturation of the labeled proteins eliminated their ability to detect changes in their environment, as did heat denaturation of the rhodopsin preparations, this suggests the changes we observe are a result of G protein-receptor coupling between functional proteins.
Receptor activation induces release of GDP and binding of GTP, thus lowering the affinity of activated Gα subunits for receptors. This is reflected by a decrease in emission for nearly all of the A1-labeled proteins upon GTPγS binding relative to the light-activated state, and this decrease is generally more pronounced for unmyristoylated proteins, suggesting myristoylation reduces the aqueous exposure of amino-terminal residues in the GTPγS-bound state. This was unrelated to the presence of membrane lipids, as a similar decrease in emission was noted after basal (uncatalyzed) nucleotide exchange occurred in solution (figure 2D). The relatively greater degree of aqueous exposure for several of the unmyristoylated proteins compared to their myristoylated counterparts upon GTPγS binding is consistent with previous EPR studies of unmyristoylated Gαil Hexa I proteins, which show a dramatic transition towards disorder at the amino terminus upon addition of GTPγS to the receptor:G protein complex (22). Solution studies of GDP-bound myristoylated Gαi Hexa I proteins reveal a relatively immobilized EPR spectra for individual residues all along the amino terminus, even in the absence of Gβγ (23), unlike the more mobile amino-terminal residues seen in unmyristoylated Gαil Hexa I proteins (22). Previous fluorescence studies of these isolated, labeled, myristoylated proteins in solution indicated that the amino terminus in the activated, GDP-AlF4- bound state is in a more hydrophobic environment than the inactive, GDP-bound state (23). A solvent protection of the myristoylated amino terminus via an (unidentified) intramolecular binding site would explain the relatively similar recovery we observe of myristoylated and unmyristoylated GTPγS-bound protein in the soluble fraction after centrifugation.
A number of proteins are known to utilize myristoyl-switching mechanisms to regulate signaling, including recoverin, PKA catalytic subunit, MARCKS and ARF, with ARF's myristoyl switch triggered by GTP binding. Membrane lipids affect the nucleotide binding of a full-length ARF, but not an amino-terminally truncated (unmyristoylated) Arf protein, and a myristoylated amino-terminal peptide derived from the ARF protein also inhibited ARF activities (44). Myristoylation of Gα proteins may serve an analogous switching function in heterotrimeric Gi proteins, mediating association with membrane-bound receptors (in concert with acylation of Gγ). Myristoylation may regulate solubility of individual Gα-GTP subunits as they dissociate from membrane-bound receptors by protecting the myristoylated amino terminus of the GTP-bound subunit from the aqueous solvent through an intramolecular binding site on the surface of the protein. The current study reveals a myristoylation-dependent solvent protection for the extreme amino terminus of Gαil after receptor-mediated GTPγS binding. Further work to unambiguously identify a potential intramolecular binding site for the myristoylated amino terminus of Gαil is currently underway.
There is a vast amount of data supporting GPCR dimerization/oligomerization, both in vitro and in vivo (reviewed in (31), (32). Although the current work is not aimed at addressing the question of receptor:G protein stoichiometry, it is interesting to consider a role for the myristoylated amino terminus in receptor dimerization. A dimeric receptor provides a plausible explanation for the ability of an activated receptor to crosslink to spans of residues in both the carboxy- and amino-terminus of Gαt (17, 33) with chemical and photo-activatible crosslinkers. These regions are separated by as much as 28 Å, as seen in the crystal structure of inactive Gt heterotrimer (PDB file 1GOT). The carboxy-terminal eighth helix of rhodopsin (often referred to as the fourth cytoplasmic loop of rhodopsin) is also known to be involved in transducin coupling through both its amino and carboxyl terminus, as seen in peptide studies and site-directed mutagenesis of this region (34-36).
Structural studies have confirmed that the length of an entire heterotrimeric G protein is sufficient to span a receptor dimer, using crystallographic constraints from G proteins, rhodopsin and the recently solved β-adrenergic receptor (2, 10, 33). Results from atomic force microscopy support a dimeric model of rhodopsin (33), and are also consistent with higher order structures, which is not unlikely in vivo, given the density of receptors in the rod outer segment. The existence of GPCR dimers is supported by results from small angle neutron scattering which suggested that the leukotriene B4 receptor-1, BLT1 (a member of the rhodopsin family of GPCRs) assembles with Gi proteins in a 2:1 receptor:G protein ratio upon agonist activation, using receptors stabilized in detergent solution (37). Recent reports suggest rhodopsin dimerization may play roles distinct from activation in vivo, as monomeric rhodopsin is sufficient for G protein activation in vitro (43).
While little has been published on the functional coupling of specific residues in the amino terminus of Gα proteins to activated receptors, the carboxy-terminus of Gα proteins has been more extensively characterized in terms of GPCR interactions. Recent work shows that GPCR activation is coupled to a rotation and translation of the α-5 helix in the carboxy terminus of Gαil Hexa I proteins, which may aid in GDP release, as elimination of this movement by a flexible linker significantly impairs receptor-mediated nucleotide exchange (14). Crystal structures of isolated Gα subunits (unmyristoylated) place residues within the carboxy terminus (residue 345) and the last resolved residues in the amino terminus (residue 33) into close proximity with each other. The amino-terminal portion of Gα (and its proximity to the carboxy terminus) has also been implicated in the specificity of receptor coupling in functional assays using a chimeric approach. The region which links the amino-terminal α-helix to the first β-sheet of the GTPase domain (residues 31-33 in Gαi) is proximal to the carboxy terminus in Gα crystal structures, and this linker was shown to be of particular importance to the specificity of receptor coupling, using a Gα15/16 chimera (38). This is consistent with the idea that the amino and carboxy termini work together in concert to affect nucleotide release upon receptor activation.
The lever arm model of receptor-mediated nucleotide release proposed by Bourne et al. (39) implicates a receptor-mediated amino-terminal conformational change in Gα which rotates Gα away from Gβγ, thus distorting the nucleotide binding site and opening an exit route for GDP release. However, the functional redundancy of farnesylation of Gγ and myristoylation Gα in receptor-mediated G protein activation does not support this lever-arm model of G-protein activation. Furthermore, although our data are consistent with a role for the amino terminus of Gαil in receptor-mediated G protein activation, the data do not directly address the contribution from Gβγ. Nevertheless, the current work provides convincing evidence for the intimate contact between the amino terminus of Gαil and activated, membrane bound receptors, indicated by the receptor-mediated change in the environment of the amino terminus. More intensive studies detailing changes in the spatial relationship between Gα, Gβγ and receptor (and the order of such changes) leading to nucleotide release must be undertaken to fully elucidate the complete mechanism of receptor-mediated nucleotide release leading to G protein activation.
Altogether this study identifies specific receptor-mediated changes occurring along the amino terminus, many of which are distinctly affected by myristoylation, and a myristoylation-dependent solvent protection of the extreme amino terminus of Gαil proteins after receptor-mediated G protein activation. This work paints a picture of a conformationally adaptible myristoylated amino-terminal region of Gα proteins. Myristoylation-dependent variations in the environment and conformation of amino-terminal residues of Gαil proteins upon G protein activation may play a role regulating interactions with Gβγ, membrane-bound receptors and soluble effectors.
Molecular graphics images were produced using the UCSF Chimera package from the Computer Graphics Laboratory, University of California, San Francisco (supported by NIH P41 RR-01081). We acknowledge Dr. Songhai Chen for insightful comments and R. Beavins for careful reading of this manuscript.