Real time conformational changes in the retinal phosphodiesterase gamma subunit monitored by resonance energy transfer.

The γ subunit of the retinal cGMP phosphodiesterase (γPDE) acts as an inhibitor of phosphodiesterase (PDE) catalytic activity and mediates enzyme regulation by the α subunit of the GTP-binding protein transducin (αT). In order to characterize conformational changes in the 87-amino acid γPDE subunit that may accompany the activation of the holoenzyme, γPDE was labeled with the fluorescent probes 5-iodoacetamidofluorescein and eosin-5-isothiocyanate for use in resonance energy transfer measurements. 5-Iodoacetamidofluorescein specifically labeled a cysteine residue at position 68 and served as a resonance energy transfer donor. The site of modification of eosin-5-isothiocyanate, which served as the resonance energy transfer acceptor, was determined to be within the first seven residues of the amino terminus of γPDE. Energy transfer between the labeled sites on free, unbound γPDE indicated that they were separated by a distance of 63 Å, consistent with a random conformation. Upon binding the catalytic αβ subunits of the PDE, the distance between the two probes on γPDE increased to 77 Å. Binding of the labeled γPDE by αT·;guanosine 5′-3-O-(thio)triphosphate did not affect the distance between the probes under conditions where the PDE was activated. These data are consistent with the view that the binding of activated αT to γPDE, which is essential for the stimulation of PDE activity, does not impart significant alterations in the tertiary structure of the γPDE molecule. They also support a model for PDE activation that places active αT in a complex with the holoenzyme.

The vertebrate phototransduction system has served as a paradigm for understanding how receptors containing seven transmembrane helices couple to heterotrimeric GTP-binding proteins (G proteins) and how activated G proteins regulate the activities of their biological effectors. The receptor in this system, rhodopsin (made up of the protein backbone opsin and the chromophore retinal), initiates the signaling pathway following the absorption of light. This leads to the formation of a complex between rhodopsin and the G protein transducin (which consists of a 39-kDa ␣ subunit, designated ␣ T , a 35-kDa ␤ subunit, and an ϳ8-kDa ␥ subunit, designated ␥ T ). Within this complex, rhodopsin stimulates the exchange of GDP for GTP, which in turn causes the dissociation of transducin into an ␣ T ⅐GTP species and intact ␤⅐␥ T complex. The ␣ T ⅐GTP species then stimulates the biological effector, the cyclic GMP phosphodiesterase (PDE), 1 a tetrameric enzyme consisting of two larger subunits (designated ␣ PDE and ␤ PDE , molecular mass ϳ85 kDa) and two identical smaller subunits designated ␥ PDE (ϳ14 kDa). The ␥ PDE subunits serve as the binding sites for the GTP-bound ␣ T subunit, although the specific mechanism by which ␣ T binding to ␥ PDE results in the stimulation of cyclic GMP hydrolysis by the catalytic core of the enzyme (i.e. the ␣ PDE and ␤ PDE subunits) is still not understood. The stimulation of enzyme activity continues until the bound GTP is hydrolyzed to GDP; thus, the GTPase activity returns the signaling system to its starting point.
Recently, a significant amount of information has been reported regarding the tertiary structural features of G protein ␣ subunits, including x-ray crystallographic structures for GDPand GTP␥S-bound forms of ␣ T (Noel et al., 1993;Lambright et al., 1994) and ␣ i1 (Coleman et al., 1994). This structural information has raised a number of possibilities regarding the identity of the regions and amino acid residues on the G protein ␣ subunits that are involved in the regulation of effector activity. However, thus far, no tertiary structural information is available for a G protein-effector complex, and consequently, very little is known regarding the specific mechanisms by which G protein binding is translated into effector regulation. The phototransduction system would seem to be especially amenable to such structure-function characterization, given that the target sites on the effector molecule for the G protein (i.e. the ␥ PDE subunits) are relatively small. However, based on NMR structure studies performed in this laboratory, all indications are that the ␥ PDE subunits do not possess significant secondary structure, at least when these subunits are free in solution (i.e. when dissociated both from the larger PDE subunits and the ␣ T subunit).
In studying the protein-protein interactions important in visual signal transduction, our aim has been to develop fluorescence spectroscopic approaches to examine different aspects of the GTP-binding/GTPase cycle of transducin (Phillips and Cerione, 1988;Guy et al., 1990;Mittal et al., 1994) and to probe the mechanisms underlying the activation of the cyclic GMP PDE Cerione, 1989, 1991;Erickson et al., 1995). In the present study, we have used resonance energy transfer approaches to determine whether the ␥ PDE subunit adopts a unique tertiary structure when it is bound to the ␣ PDE and ␤ PDE subunits (versus when it is free in solution) and when it binds GTP-bound ␣ T . To do this, we developed procedures for generating doubly labeled ␥ PDE subunits, with one label serving as an energy donor and the other as an energy acceptor.
Using these labeled ␥ PDE subunits, we are able to show that the ␥ PDE subunit does change its tertiary conformation upon binding to the ␣ PDE and ␤ PDE subunits; this change extends the distance between the amino terminus and cysteine 68 of the ␥ PDE molecule. However, the binding of the ␣ T subunit does not appear to perturb the relative juxtaposition of these two sites on ␥ PDE . Thus, these results suggest that the changes in the ␥ PDE subunit that accompany the binding of the GTP-bound ␣ T subunit and are responsible for the stimulation of cyclic GMP hydrolysis by the ␣ PDE and ␤ PDE subunits do not occur between residue 68 and the amino terminus of the ␥ PDE molecule. Furthermore, measurements of PDE enzyme activation and inhibition, along with the characterization of corresponding spectroscopic states of the doubly labeled ␥ PDE subunit, support a model in which the activated ␣ T ⅐␥ PDE complexes remain associated with the ␣␤ PDE core of the effector enzyme during the stimulation of cyclic GMP hydrolysis.

EXPERIMENTAL PROCEDURES
Materials-SP-Sepharose, phenyl-Sepharose, and blue Sepharose were obtained from Pharmacia Biotech Inc. Factor Xa was purchased from New England Biolabs (Beverly, MA). 5-Iodoacetamidofluorescein and eosin-5-isothiocyanate were purchased from Molecular Probes, Inc. (Eugene, OR). Dark-adapted bovine retina were purchased from Hormel Meat Packers (Austin, MN). All other chemicals and enzymes were purchased from Sigma. The pLCIIFXSG plasmid was a gift from Dr. Heidi Hamm (University of Illinois College of Medicine, Chicago, IL).
␥ PDE Expression and Purification-Recombinant ␥ PDE was expressed in Escherichia coli, and cells were lysed as described by Brown and Stryer (1989). Briefly, E. coli strain AR68 containing the pLCIIFXSG plasmid was grown at 30°C in a Labline fermentor and induced by temperature jump to 42°C at an A 600 of 0.5. After the temperature jump, the cells were grown at 37°C for an additional 2 h. Cells were harvested by centrifugation at 4000 ϫ g and lysed with lysozyme. cII-␥ PDE fusion protein was partially purified based on its insolubility in detergent, and this particulate fraction was subsequently solubilized in 50 mM Tris, 50 mM NaCl, 1 mM EDTA, 6 M urea, pH 8.0.
The cII-␥ PDE present in the urea-solubilized extract was further purified by binding to an SP-Sepharose cation exchange chromatography matrix and eluting the ␥ PDE fusion protein with a linear gradient of NaCl (50 -300 mM), 50 mM Tris (pH 8.0), 1 mM EDTA, 6 M urea. The cII-␥ PDE fusion protein was diluted to a final [NaCl] of 100 mM and bound again to the SP-Sepharose column. While bound to the column matrix, the buffer was changed to 50 mM Tris, 50 mM NaCl, 1 mM EDTA, pH 8.0, with no urea. The protein was again eluted with the same NaCl gradient. At this point, the ␥ PDE was Ͼ90% pure based on Coomassie Blue-stained SDS-polyacrylamide gel electrophoresis.
The purified ␥ PDE was clipped with factor Xa at 4°C overnight. Ammonium sulfate was added to the ␥ PDE to a final concentration of 0.75 M, and the clipped ␥ PDE was then loaded onto a phenyl-Sepharose column. The ␥ PDE eluted from the phenyl-Sepharose column when there was no (NH 4 ) 2 SO 4 in the column buffer. This eluted protein was Ͼ99% pure as judged by Coomassie Blue-stained SDS-polyacrylamide gel electrophoresis. This sample was dialyzed in Spectropore 2,000-Da molecular mass cut-off tubing versus 20 mM HEPES, 150 mM NaCl, 5 mM MgCl 2 , 0.2 mM DTT, pH 7.4, and stored in aliquots at Ϫ80°C until use.
Modification of ␥ PDE by 5-Iodoacetamidofluorescein (IAF) and Eosin-5-isothiocyanate (EITC)-␥ PDE was reacted with 1 mM IAF at pH 7.4 for 1 h. The reaction was quenched with the addition of 30 mM DTT, and the labeled protein was separated from free probe by SDS-polyacrylamide gel electrophoresis. Labeled protein was visualized in the gel using UV trans-illumination and excised (Wensel and Stryer, 1990;Erickson and Cerione, 1991). Gel purification of the labeled ␥ PDE subunit provides a means to obtain only ␥ PDE that contains the IAF (fluorescence donor) molecule. This is due to the fact that the ␥ PDE subunit possesses a single reactive residue at cysteine 68 and that the labeled ␥ PDE subunit undergoes an apparent shift in molecular mass from ϳ11 to ϳ15 kDa after IAF labeling. Thus, pure IAF-␥ PDE subunit can be isolated using a preparative gel; amino acid analysis together with fluorescein absorption measurements at 495 nm (⑀ max ϭ 75,000 M Ϫ1 cm Ϫ1 ) (Carraway et al. 1989) indicate a stoichiometry of IAF incorporation of 1 Ϯ 0.1 mol of IAF/mol of ␥ PDE . The gel slice containing the labeled ␥ PDE was incubated in 4 volumes of distilled water at 4°C overnight, and the gel eluate was dialyzed against a 100 ϫ volume of 20 mM HEPES, 150 mM NaCl, 5 mM MgCl 2 , 0.2 mM DTT, pH 7.4. The fluorescein-labeled ␥ PDE was then reacted in this buffer with 1 mM EITC for 3 h. The reaction was quenched with 15 mM Tris, pH 6.8, and the protein was run on SDS-polyacrylamide gel electrophoresis and again eluted from a gel slice.
The ratio of eosin to fluorescein in the doubly labeled ␥ PDE was determined to be 1:1.08 Ϯ 0.04 (S.E., n ϭ 3) by absorbance spectroscopy using a Hewlett-Packard 8451A spectrophotometer. The fluorescein absorbance was measured at 495 nm and corrected for any contribution from EITC by deconvolution (this represented 20% of the total absorbance). The concentration of EITC in the doubly labeled protein was determined by eosin absorbance at 522 nm, using a molar extinction coefficient of 83,000 M Ϫ1 cm Ϫ1 for EITC (Cherry et al., 1976). There was no detectable contribution of fluorescein to this absorbance.
Purification of Component Proteins-Purification of transducin and holo-PDE from bovine retina were performed as described previously (Kroll et al., 1989). Rod outer segments were purified as described by Gierschik et al. (1984) and washed several times with isotonic buffer (10 mM HEPES, 5 mM MgCl 2 , 1 mM DTT, 0.1 mM EDTA, 100 mM NaCl, 0.3 mM phenylmethylsulfonyl fluoride, pH 7.5). The rod outer cell membranes were then resuspended in hypotonic buffer (10 mM HEPES, 1 mM DTT, 0.1 mM EDTA, 0.3 mM phenylmethylsulfonyl fluoride, pH 7.5) to release PDE from the membranes and centrifuged at 39,000 ϫ g. The supernatant from the hypotonic wash (containing the holo-PDE) was concentrated using an Amicon 30,000-Da molecular mass cut-off membrane.
Transducin was recovered from the hypotonically washed pellet by resuspending the membranes in hypotonic buffer supplemented with 100 mM GTP or GTP␥S (for inactive GDP-bound ␣ T or active GTP␥Sbound ␣ T , respectively). The membranes were washed several times with nucleotide-containing buffer, and the supernatants were pooled. The pooled extract containing crude transducin was purified by blue Sepharose chromatography as described by Pines et al. (1985).
Assay of PDE-inhibitory Activity-Trypsin-activated PDE (tPDE) was prepared by limited tryptic digest of purified PDE (Kroll et al., 1989). Trypsin at 65 g/ml was added to 1 M PDE and incubated at room temperature for 2 min. The reaction was quenched by the addition of 260 g/ml soybean trypsin inhibitor. PDE activity was determined using a pH microelectrode as described by Yee and Liebman, 1978. Activity was measured in 5 mM HEPES, 100 mM NaCl, 2 mM MgCl 2 , 5 mM cGMP, pH 7.5, at room temperature. Proton release from cGMP hydrolysis due to PDE activity was recorded in mV at one determination per second.
Assay of PDE Stimulation by Activated ␣-Transducin-Phospholipid vesicles were prepared by sonication of 17 mg of lecithin in 1.0 ml of deionized and distilled water. PDE purified from ROS was incubated with the phospholipid vesicles to allow binding, and the vesicles were pelleted by centrifugation in a Beckman Airfuge (10 min, 30 p.s.i.). Membrane-bound PDE was activated as described by Brown (1992). Briefly, PDE was digested with ArgC protease at a concentration of 0.1 units/l for 6 h at room temperature. The PDE that remained in the membrane was then purified by recentrifugation in the Beckman Airfuge. Approximately 15-20% of the total PDE was recovered after binding and ArgC activation, and the ArgC-proteolyzed PDE had an activity of greater than 75% of that achieved by trypsinization. The PDE was inhibited by the addition of labeled recombinant ␥ PDE or unlabeled, purified recombinant ␥ PDE while activity was monitored to ensure that the minimum amount of ␥ PDE required for full inhibition of PDE activity was added. The reconstituted, lipid vesicle-bound PDE was stimulated by the addition of the ␣ T ⅐GTP␥S subunit.
Fluorescence Spectroscopy-In most cases, relative fluorescence was measured on an SLM 8000C spectrofluorimeter using a 1 ϫ 0.3-cm quartz cuvette. Samples were diluted into 5 mM HEPES, 100 mM NaCl, 2 mM MgCl 2 , pH 7.5, and were stirred continuously. Excitation from a xenon lamp passed through a monochrometer set at 465 nm, and orthogonal emission was monitored continuously at 520 nm. Emission was corrected for changes in the lamp intensity by recording in ratio mode. The amplitude of the fluorescence changes (for example, following the addition of ␣ PDE and ␤ PDE to labeled ␥ PDE ) were measured by subtracting the peak height of the fluorescence change (i.e. after it has leveled off, typically after 800 s) from the fluorescence that is measured after unlabeled ␥ PDE is added (i.e. conditions where the effect has been reversed by competition). Contributions to the fluorescence signals due to intrinsic protein fluorescence and light scattering were negligible (Ͻ1%) in all experiments presented here.
The distance between donor and acceptor fluorophores was calculated according to Förster energy transfer theory (Lakowicz, 1983).
When making resonance energy transfer measurements for the doubly labeled ␥ PDE , we measured the change in IAF (donor) fluorescence that occurred after trypsin treatment (since this treatment effectively separates the donor (IAF) and acceptor (EITC) probes and eliminates energy transfer) and subtracted any changes caused by trypsin treatment of singly labeled IAF-␥ PDE . The efficiency of energy transfer (E) was calculated as follows, where F FE-␥ is the relative fluorescein fluorescence of the intact doubly labeled ␥ PDE , and F F-␥ is the relative fluorescein fluorescence of the ␥ PDE control following complete trypsinization of the singly labeled fluorescein species. Using the efficiency of energy transfer, the distance between the two probes (R) can be estimated using the following equation (Lakowicz, 1983), and the R 0 for fluorescein and eosin was calculated to be 56.8 Å, using the integral overlap method (Carraway et al., 1989;Epe et al., 1983). In calculating R 0 , the orientation factor 2 was assumed to be 2 ⁄3, indicating equal probability for all dipole-dipole orientations. Polarization values of the free, labeled ␥ PDE were determined to be Յ0.1, which indicates errors of Ͻ10% if 2 is set to 2 ⁄3 (data not shown; see Tables II  and III in Hass et al. (1978)).
Fluorescence emission anisotropy was measured on an SLM 8000C fluorimeter. Emission in the horizontal and vertical orientations was measured simultaneously using a monochrometer set at 520 nm and a band pass filter of 520 nm. The G factor was set to 1.0 by adjusting the photomultiplier gain prior to beginning acquisition. All measurements were made at 20°C.
Data Analysis for ␥ PDE Binding to the ␣ PDE and ␤ PDE Subunits-Activity data in Fig. 2 were fit to the binding equation, where A is the activity, A 0 is the initial activity value, A f is the final (inhibited) activity, L T is the total ␥ PDE concentration, R T is the total tPDE concentration, and K I is the inhibition constant.
Anisotropy data in Fig. 2 were fit to the binding equation, where r is the anisotropy value, r 0 is the initial anisotropy value, r f is the maximum anisotropy, L T is the total tPDE concentration, R T is the total ␥ PDE concentration, and K D is the dissociation constant. All fits were generated using IgorPro software using a least squares fit criterion (WaveMetrics, Inc., Lake Oswego, OR). For the comparison of the kinetics of relative fluorescence and fluorescence anisotropy changes (e.g. Fig. 5B), relative fluorescence was determined using the mathematical relationship between fluorescence intensity and anisotropy (Lakowicz, 1983), where r is the anisotropy, I is the calculated emission intensity, I ʈ is the intensity of emission polarized in parallel with the excitation beam, and I Ќ is the intensity of emission polarized perpendicular to the excitation beam. This calculated intensity was corrected for changes in the lamp intensity to give relative fluorescence. Calculated relative fluorescence was identical when using either I ʈ or I Ќ and agreed very closely with data obtained when using the conventional measurement of relative fluorescence described above. In order to compare the rate of reversal by (unlabeled) ␥ PDE of the ␣ PDE /␤ PDE -induced enhancement in the fluorescein fluorescence of the doubly labeled ␥ PDE and the rate of reversal of the increase in anisotropy (Fig. 3B), the fluorescence intensity differences for doubly labeled ␥ PDE that is bound to tPDE and for doubly labeled (unbound) ␥ PDE that is unbound must be mathematically cor-rected in the apparent anisotropy plot. This was performed using the equation, where B is the fraction of ␥ PDE that is bound to the trypsin-treated PDE, A u is the anisotropy of unbound (free) ␥ PDE (0.176 for the data in Fig.  3B), A B is the anisotropy of ␥ PDE bound to trypsin-treated PDE (0.232 in Fig. 3B), and A is the measured (uncorrected) anisotropy. The 0.34 and 1.34 factors are derived from the 34% higher fluorescence of the bound, doubly labeled ␥ PDE (to ␣ PDE and ␤ PDE ) relative to the unbound doubly labeled ␥ PDE species.

RESULTS
The primary aim of these studies was to generate a ␥ PDE subunit that was labeled at two distinct sites so that resonance energy transfer approaches could be used to monitor changes in the juxtaposition of these sites caused by the binding of other signaling molecules to the ␥ PDE . In particular, we were interested in determining if the structure of ␥ PDE changes upon its binding to the core of the PDE molecule and/or upon its binding to an activated ␣ T subunit.
Characterization of the Doubly Labeled ␥ PDE Subunit-We used the following strategy to generate a doubly labeled ␥ PDE subunit. One site of labeling (with IAF) was the single cysteine residue at position 68, both because this site is located near one end of the ␥ PDE molecule and because the conditions for its selective and stoichiometric modification have been well established (e.g. Erickson et al. (1995)). We then set out to label the primary amino group of the amino-terminal methionine residue, because previous studies have demonstrated that the Nterminal primary amino group of a protein can be selectively labeled when performing the modification at pH values below 7.5 (Carraway et al., 1990).
However, given that the ␥ PDE subunit contains a number of lysine residues, we first examined whether the labeling with EITC was in fact occurring at the amino-terminal residue rather than at one or more of the lysine residues on ␥ PDE . When mass spectrometry was performed on the full-length recombinant ␥ PDE labeled with both EITC and IAF, a single predominant peak was determined with a size of 10,880 (Fig. 1A). The position of this peak is within the experimental error of the calculated size for a full-length ␥ PDE molecule containing just one IAF moiety and one EITC moiety (10,762 Da).
We performed an additional experiment where the fulllength ␥ PDE , labeled with IAF and EITC, was trypsin-treated and then subjected to reverse-phase HPLC in order to resolve the resulting peptides. This was followed by mass spectrometry of the eosin-labeled species. The trypsin treatment resulted in the generation of only three peaks showing (EITC) absorbance at 525 nm. Two of these peaks showed lower relative levels of 525 nm absorbance and were subsequently found not to contain peptides as revealed by mass spectral analysis. It seems most likely that these peaks represented fluorescent dye aggregates, and we therefore concentrated on the third peak of 525-nm absorbance for further analysis. When this peak was subjected to mass spectrometry, two observable mass spectrum peaks were obtained (Fig. 1B). One peak of 1555 Da corresponds to the first seven amino acids from the amino terminus (ending with lysine residue 7) plus a single eosin moiety. The second peak of 2025 Da represents the first 11 amino acids (ending with arginine residue 11). Comparison of these experimentally determined labeled peptide masses revealed no other close matches when compared with hypothetical tryptic fragments in the ␥ PDE sequence (not shown). These results indicate that the EITC label was attached either at the amino terminus as originally intended or at the lysine located seven residues from the terminus. Because the modification of lysine residues with isothiocyanate reagents typically requires several hours of incubation at pH Ͼ 8.5 (compared with the 3-h modification at pH 7.4 used here), we feel it is most likely that the EITC is attached to the amino-terminal methionine of ␥ PDE . However, even if the EITC were placed at position 7, this would not affect the general aim of the study, which was to determine whether the position of the amino terminus of ␥ PDE relative to cysteine 68 changed when binding to the larger subunits of the PDE and/or to the ␣ T subunit of transducin.
Fluorescence Read-out for the Binding of ␥ PDE to the ␣ PDE and ␤ PDE Subunits-It has been well documented that trypsin treatment of the cyclic GMP PDE selectively degrades the ␥ PDE subunits, resulting in a constitutively active PDE complex (composed of the ␣ PDE and ␤ PDE subunits) (Wensel and Stryer, 1990). Readdition of ␥ PDE to the trypsin-treated enzyme reverses the constitutive activation and returns the enzyme to its inactive state. Fig. 2A shows that changes in the fluorescence anisotropy of IAF-␥ PDE can be used to monitor the interaction between the IAF-and EITC-labeled ␥ PDE molecule and the ␣ PDE and ␤ PDE subunits. This anisotropy change directly reflects the association of the IAF-labeled ␥ PDE subunit with the significantly larger core subunits (␣ PDE , ␤ PDE ) of the effector enzyme. The resultant titration profile is consistent with a single class of binding sites between the labeled ␥ PDE and the trypsin-treated PDE and yields an apparent K D value of 6 nM. Fig. 2B shows the results of an experiment where the ability of the doubly labeled ␥ PDE subunit to inhibit the trypsin-activated PDE was assayed. The data again yield a titration profile that can be fit by a single class of sites with an apparent K I value of 7 nM. The close agreement between these two observed K D values argues that the doubly labeled ␥ PDE is fully functional in terms of its ability to bind to the core of the effector enzyme and inhibit cyclic GMP hydrolysis.
We next determined whether the position of the aminoterminal domain (labeled with EITC) relative to cysteine 68 (labeled with IAF) changed upon the binding of the labeled ␥ PDE to the trypsin-treated PDE. The rationale here was that any change in the proximity of these sites on the ␥ PDE mol-ecule would be reflected by a change in resonance energy transfer between the donor IAF and the acceptor EITC moiety, as monitored by a change in (donor) IAF fluorescence. The results of this experiment are shown in Fig. 3A. The lower trace represents a control where ␥ PDE subunits, labeled at just a single site (cysteine 68) with IAF, were added to the trypsin-treated PDE. A slight increase (Յ5%) in the IAF fluorescence was detected, presumably reflecting a change in the microenvironment of cysteine 68. This small change can be reversed by the addition of excess, unlabeled ␥ PDE . The subsequent addition of trypsin also resulted in a minor increase in the singly IAF-labeled ␥ PDE fluorescence. This suggests that the microenvironment of cysteine 68 within the full-length ␥ PDE molecule is altered relative to the environment of cysteine 68 within the trypsinized fragment that contains this residue. A parallel set of controls was performed with a singly labeled eosin-␥ PDE species binding to FIG. 1. Mass spectroscopy of the doubly labeled ␥ PDE . Panel A shows the results of a MALDI-TOF mass spectroscopy experiment performed using approximately 100 pmol of ␥ PDE in 25% acetonitrile and 0.1% trifluoroacetic acid. The molecular ion mass of 10,880 is within the experimental error for the expected molecular weight of 10,762, corresponding to 1 ␥ PDE :1 IAF:1 EITC. Note that the predicted molecular mass for unlabeled ␥ PDE (9660 Da) does not appear to be present. There also do not appear to be sufficient amounts of heterogeneously labeled ␥ PDE to produce other predicted molecular weight peaks in the spectrum. Panel B shows a MALDI TOF mass spectroscopy trace for a ␥ PDE peptide that was produced from tryptic digestion of the doubly labeled ␥ PDE . 150 pmol of protein was trypsinized, and an eosincontaining peptide species was purified by reverse-phase HPLC based upon optical absorbance at 523 nm. Mass spectroscopy of this peptide shows two components with molecular weights of 1554 and 2025. One representative spectrum is shown. The experiment was repeated one additional time with MALDI and twice with ESI mass spectroscopy techniques.

FIG. 2. Effect of doubly labeled ␥ PDE on trypsinized PDE. Panel
A shows fluorescein fluorescence anisotropy of the doubly labeled ␥ PDE . 12 pmol (15 nM) of doubly labeled ␥ PDE (as determined by absorbance at 520 nm) was bound to successive additions of 2 pmol (2.5 nM) trypsinized PDE. The curve is fit to a quadratic binding equation for total tPDE binding to doubly labeled ␥ PDE . The apparent K D for this binding is 7.3 nM. This experiment was repeated one additional time, which resulted in an apparent K D of 6.0 nM. Panel B shows the inhibition of tPDE by doubly labeled ␥ PDE . 1 pmol (5 nM) of tPDE was inhibited with successive additions of 0.9 pmol (4.5 nM) of the doubly labeled ␥ PDE species. The rate of cGMP hydrolysis was monitored using the pH microelectrode assay (see "Experimental Procedures"; Yee and Liebman (1978)). These data are plotted as an inhibition curve, fit to a quadratic binding equation for total ␥ PDE binding to tPDE. The concentration of labeled ␥ PDE was determined by eosin absorbance at 520 nm. The apparent K I for the inhibition is 6 nM. The experiment was repeated one additional time, which resulted in an apparent K I of 7.8 nM.
trypsinized ␣␤ PDE . No detectable change in either eosin emission (at 545 nm) or absorbance (at 525) was observed when eosin-␥ PDE bound to the PDE enzyme core (data not shown).
The upper trace in Fig. 3A shows the results obtained when ␥ PDE labeled with both IAF and EITC was added to the ␣ PDE and ␤ PDE subunits, generated by trypsin treatment of the PDE. There was a significant increase in the IAF fluorescence of the doubly labeled ␥ PDE upon binding to ␣ PDE and ␤ PDE . We have found that this increase, which ranged from 20 to 40% in different experiments (34 Ϯ 5%, S.E., n ϭ 4), could be completely eliminated by adding excess ␥ PDE , consistent with the interaction between ␥ PDE and the ␣ PDE and ␤ PDE subunits being fully reversible. Significantly, the addition of trypsin to the free, unbound doubly labeled ␥ PDE caused a significant enhancement in the IAF fluorescence. After correction for the small changes caused by the addition of trypsin to the control IAF-labeled ␥ PDE subunit (shown in the lower trace of Fig. 3A), the average value for trypsin-induced enhancement of the free IAF-␥ PDE fluorescence was 36 Ϯ 2% (S.E., n ϭ 8). In control experiments, the magnitude of the enhancement was not influenced by the presence of other nonfluorescent proteins (e.g. bovine serum albumin; data not shown). This level of IAF fluorescence represents the donor fluorescence in the absence of resonance energy transfer, since tryptic digestion of the ␥ PDE molecule results in the IAF moiety being attached to a (tryptic) peptide that is distinct from the peptide containing the EITC moiety. Thus, when this level of IAF fluorescence is compared with the IAF fluorescence for the doubly labeled ␥ PDE molecule when it is free in solution, this yields an efficiency of energy transfer of 36%. When that value is used to calculate the distance separating the IAF and EITC moieties, using an R 0 value of 56.8 Å (see "Experimental Procedures"), an apparent distance between the IAF and EITC moieties of 62.7 Ϯ 1.05 Å (S.E., n ϭ 8) is obtained. The same measurement made for the doubly labeled ␥ PDE subunit when it is bound to the ␣ PDE and ␤ PDE subunits, generated by treatment of the PDE with trypsin or with ArgC (see below), yielded an efficiency of energy transfer of 15 Ϯ 3% (S.E., n ϭ 4) and an effective distance of 77.3 Ϯ 3.3 Å (S.E., n ϭ 4). Thus, these results indicate that the ␥ PDE molecule becomes significantly more extended when it binds to the ␣ PDE and ␤ PDE subunits.
The results shown in Fig. 3B show the corresponding real time fluorescence assay for doubly labeled ␥ PDE interactions with the ␣ PDE and ␤ PDE using the changes in fluorescence anisotropy for the IAF moiety bound to ␥ PDE . The fluorescence anisotropy change that occurs upon the addition of trypsintreated PDE to the doubly labeled ␥ PDE subunit can be rapidly reversed by the addition of excess ␥ PDE (as the change in IAF fluorescence can also be rapidly reversed; Fig. 3A). The addition of trypsin to this sample then results in an immediate decrease in the anisotropy for IAF due to the degradation of the labeled ␥ PDE subunit and the dissociation of labeled ␥ PDE peptide fragments. The raw anisotropy data shown in Fig. 4B can be further analyzed taking into account the disproportionate contribution of the ␣␤ PDE -bound state of the doubly labeled ␥ PDE to the overall 520 nm emission (see "Experimental Procedures"). When the corrected data for the decrease in the fluorescence anisotropy, upon the addition of unlabeled ␥ PDE , and the corresponding decrease in the fluorescence enhancement, are fit to a single exponential decay, the rate of the decay in the IAF fluorescence (0.066 s Ϫ1 ) is similar to the decrease in doubly labeled ␥ PDE anisotropy (0.042 s Ϫ1 ).
Fluorescence Read-out for the Binding of ␥ PDE to an Activated ␣ T ⅐GTP␥S Complex-We next examined whether the position of cysteine 68 relative to the amino terminus of ␥ PDE changed upon binding to an activated ␣ T subunit. This was done employing the same fluorescence read-outs as those used to examine changes in ␥ PDE that accompanied its binding to the ␣ PDE and ␤ PDE subunits. Fig. 4A shows the results from a control experiment where the ␣ T ⅐GTP␥S complex was added to an IAF-labeled ␥ PDE subunit. We found that the ␣ T ⅐GTP␥S/IAF-␥ PDE interaction resulted in an ϳ10% enhancement in the IAF fluorescence, presumably reflecting an ␣ T -induced change in the microenvironment of cysteine 68, as previously reported (Erickson et al., 1995). This fluorescence change was immediately reversed upon the addition of excess (unlabeled) ␥ PDE , due to its competition with the IAF-labeled ␥ PDE for the activated ␣ T subunit. As shown earlier (Fig. 3A), trypsin treatment of the IAF-␥ PDE then results in a minor increase in the IAF fluorescence. Fig. 4B shows the results obtained when the same type of experiment was performed using the doubly labeled ␥ PDE . Essentially, the same results were obtained as those seen with the singly (IAF-) labeled ␥ PDE . Specifically, there was an immediate enhancement (ϳ10%) in the IAF fluorescence of the doubly FIG. 3. Interaction of ␥ PDE with tPDE. Panel A shows the emission at 520 nm of IAF-and EITC-labeled ␥ PDE and IAF-labeled ␥ PDE . The upper curve shows the emission of 12 pmol (15 nM) of doubly labeled ␥ PDE . The addition of 10 pmol of (12.5 nM) trypsinized PDE caused an increase in fluorescence emission, and the further addition of another 10 pmol of tPDE did not cause an additional increase in the emission. This increased fluorescence was reversed with 155 pmol (1.95 M) of unlabeled ␥ PDE . The lower curve shows the results for the corresponding control experiment with IAF-labeled ␥ PDE . Inset, single exponential fit of the fluorescence changes that accompany the dissociation of doubly labeled ␥ PDE upon the addition of unlabeled ␥ PDE . Panel B shows anisotropy data recorded during the relative fluorescence measurements of the doubly labeled ␥ PDE shown in panel A. Trypsinized PDE caused an increase in the anisotropy. This increase is saturable and reverses with the addition of the excess unlabeled ␥ PDE . Inset, single exponential fit of the changes in the corrected fluorescence anisotropy ("Experimental Procedures") that accompany the dissociation of doubly labeled ␥ PDE upon the addition of unlabeled ␥ PDE . labeled ␥ PDE molecule upon the addition of ␣ T ⅐GTP␥S, and the ␣ T -induced enhancement was rapidly reversed by the addition of excess unlabeled ␥ PDE . However, in this case, trypsin treatment of the IAF and EITC-labeled ␥ PDE subunit resulted in a significant enhancement of the IAF fluorescence (as shown earlier in Fig. 3B), because the IAF and EITC labels end up on different tryptic fragments, and thus the resonance energy transfer between these labels is eliminated. Consequently, an increase in IAF fluorescence occurs. Thus, although ␥ PDE binding to the ␣ PDE and ␤ PDE subunits effectively results in an extension in the distance between the amino terminus and cysteine 68 of the ␥ PDE molecule, the results in Fig. 4 argue that there is no such change in the juxtaposition of these sites when ␥ PDE binds to the activated ␣ T subunit.
An important question was whether any change in the ␥ PDE structure occurred when an activated ␣ T subunit bound to an intact PDE molecule (i.e. an ␣ PDE ⅐␤ PDE ⅐␥ PDE complex). The results shown in Fig. 5 suggest that this is not the case. In this experiment, the IAF fluorescence of an IAF and EITC-labeled ␥ PDE subunit was monitored after protease treatment of the holo-PDE molecule. The protease ArgC was used because it has been shown to have some selectivity in degrading the ␥ PDE subunits of the holo-PDE complex, leaving intact the ability of an activated ␣ T subunit to stimulate cyclic GMP hydrolysis following the addition of this G protein subunit together with fluorescently labeled ␥ PDE subunits (Brown, 1992). The addition of the doubly labeled ␥ PDE subunits to holo-PDE molecules that were pretreated with ArgC resulted in an enhancement in IAF fluorescence, consistent with our previous findings that a reduction in energy transfer occurs between the IAF and EITC labels when the doubly labeled ␥ PDE subunit binds to the ␣ PDE and ␤ PDE subunits. Subsequently, no additional change in the IAF fluorescence was observed upon the addition of an ␣ T ⅐GTP␥S complex under conditions where we found that the addition of the G protein to this mixture caused a significant stimulation of cyclic GMP hydrolysis (Fig. 5, inset). Thus, under conditions where the activated ␣ T subunits were clearly binding to the IAF and EITC-labeled ␥ PDE subunits (which in turn are part of a holo-PDE complex), there was no detectable change in ␥ PDE structure as read-out by changes in resonance energy transfer between the two labels.

DISCUSSION
In the present study, we set out to use fluorescence spectroscopic approaches to determine whether structural changes occur within the ␥ PDE molecule when it binds to the larger ␣ PDE and ␤ PDE subunits of the effector enzyme. We also were interested in determining whether the tertiary structure of ␥ PDE is The addition of approximately 25 pmol of (31 nM) ArgC-treated PDE caused an increase in fluorescence. The subsequent addition of 12.5 pmol (15.6 nM) of purified ␣ T ⅐GTP␥S caused a slight further increase in the fluorescence. The addition of another the 12.5 pmol of ␣ T ⅐GTP␥S then caused no further increase. The inset shows that ␣ T ⅐GTP␥S can stimulate ArgC-treated PDE under the same conditions used to measure the fluorescence changes described above. Approximately 12.5 pmol (62 nM) of ArgC-treated PDE was inhibited with 12 pmol of (60 nM) doubly labeled ␥ PDE . The addition of 12.5 pmol (62.4 nM) ␣ T ⅐GTP␥S caused a 2.5-fold increase in the rate of cGMP hydrolysis. The addition of trypsin caused a further 5.8-fold increase in activity. affected by the binding of an activated ␣ T subunit, because such information could provide insight into the molecular mechanism by which the retinal G protein stimulates PDE activity (i.e. cyclic GMP hydrolysis). Our general strategy for examining tertiary structural changes within ␥ PDE was to attach a fluorescence donor moiety at cysteine 68 (IAF) and an acceptor chromophore (EITC) at the amino-terminal end of the ␥ PDE molecule and then to use resonance energy transfer to monitor the juxtaposition of these labels under different experimental conditions. By using this approach, we found that the distance between the fluorescent probes (i.e. the distance between the EITC-labeled amino group and the IAF-labeled cysteine) was increased from 63 Å when ␥ PDE was free in solution to 77 Å when it was bound to the larger subunits of the PDE molecule. However, there was no significant change in the proximity of these two sites on the ␥ PDE molecule upon the binding of an activated ␣ T subunit.
Previous resonance energy transfer measurements suggested that upon the formation of an ␣ T ⅐␥ PDE complex, cysteine 68 of ␥ PDE (which is thought to be close to the ␣ T -binding site (Faurobert et al., 1993)) is a significant distance (ϳ40 Å) from the guanine nucleotide binding site on ␣ T . Various studies have implicated residues 300 -310 of ␣ T as being involved in the stimulatory interaction with the PDE molecule (Rarick et al., 1992;Artemyev et al., 1993); however, the coordinates from the x-ray crystallographic structure of ␣ T indicate that this region of the ␣ T subunit is only 20 -25 Å away from lysine 267 at the guanine nucleotide-binding site. Thus, one possible explanation for the resonance energy transfer measurements that suggest a longer distance (than 20 -25 Å) between the nucleotide binding site and the ␥ PDE -binding site on ␣ T is that ␣ T makes at least two contacts with ␥ PDE , one involving residues 300 -310 and another involving a site within the large helical domain of ␣ T (e.g. residues 106 -116). If the contact region in the helical domain of ␣ T bound to ␥ PDE in the vicinity of cysteine 68, this would be consistent with the measured distance of ϳ40 Å between cysteine 68 of ␥ PDE and lysine 267 of ␣ T . In this view, residues 300 -310 of ␣ T would bind in the vicinity of lysine residues 41, 44, and 45 of ␥ PDE , consistent with the results obtained from chemical cross-linking studies (Artemyev et al., 1993). Our present results would suggest that the distance between residues 41-45 and cysteine 68 (i.e. the two proposed ␣ T -binding sites on the ␥ PDE molecule) does not change when ␥ PDE binds to an activated ␣ T subunit, either when ␥ PDE is free in solution or bound to the ␣ PDE and ␤ PDE subunits. We would expect that even subtle ␣ T -induced conformational changes occurring upstream from cysteine 68 would have been detected by a change in the effective distance between cysteine 68 and the amino-terminal end of the ␥ PDE molecule. However, we (Erickson et al., 1995) and others (Faurobert et al., 1993) have found that ␣ T binding perturbs the microenvironment surrounding cysteine 68. Thus, in light of our present findings, we hypothesize that these ␣ T -induced changes are communicated to the carboxyl-terminal domain of the ␥ PDE molecule, resulting in the activation of cyclic GMP hydrolysis by the ␣ PDE and ␤ PDE subunits.
An important point raised by the energy transfer experiments is that the ␣ T -mediated stimulation of PDE activity occurs while the ␣ T ⅐␥ PDE complex is still associated with the core of the effector enzyme (i.e. the ␣ PDE and ␤ PDE subunits). The latter suggestion has been a subject of controversy over the years, since various lines of data have argued that ␣ T -mediated stimulation is accompanied by the dissociation of an ␣ T ⅐␥ PDE complex from the core of the enzyme (Wensel and Stryer, 1990), while other lines of evidence have suggested that ␥ PDE dissociation is not necessary for the stimulation of PDE activity (Erickson and Cerione, 1993;Catty et al., 1992;Clerc and Bennett, 1993). However, the fact that the changes in the ␥ PDE subunit that are specifically induced by the ␣ PDE and ␤ PDE subunits (i.e. the increase in the distance between the amino terminus and cysteine 68 of ␥ PDE ) are not reversed upon the binding of an activated ␣ T subunit (under conditions where ␣ T causes a stimulation of PDE activity) argues that the ␥ PDE ⅐␣ PDE ⅐␤ PDE complex remains intact during the ␣ T ⅐GTP␥Smediated stimulation of cyclic GMP hydrolysis (see Fig. 5). The activated ␣ T subunit does not induce similar changes within the ␥ PDE subunit (as those induced by the ␣ PDE and ␤ PDE subunits) when ␥ PDE is free in solution (i.e. dissociated from the ␣ PDE and ␤ PDE subunits), and ␣ T appears to bind to ␥ PDE independently of the ␣ PDE and ␤ PDE subunits. Specifically, the K D values that we have measured for the interaction of ␣ T ⅐GTP␥S with ␥ PDE are similar for ␥ PDE that is free in solution (ϳ30 nM; Erickson et al. (1995)) or bound to ␣ PDE and ␤ PDE (21 nM; Erickson and Cerione (1989)). Thus, based on a consideration of microscopic reversibility, these data argue against the possibility that the activated ␣ T subunit binds to ␥ PDE and stimulates its dissociation from the ␣ PDE and ␤ PDE subunits (during PDE activation) but still maintains the free ␥ PDE subunit in a specific conformation that was originally induced by the larger PDE subunits.
Future work will be directed toward further examining both of the suggestions raised by the present study. In particular, we will set out to directly demonstrate that ␣ T binding to ␥ PDE results in specific conformational changes within the carboxylterminal domain of this subunit that are directly translated into the stimulation of PDE activity. We also will try to determine the types of changes that must occur in the orientation of the ␥ PDE subunits relative to the larger subunits of the effector enzyme. This remains an important issue, because although our data argue that the ␥ PDE subunits remain associated with ␣ PDE and ␤ PDE during ␣ T -mediated stimulation of enzyme activity, it has been well established that the addition of excess free ␥ PDE molecules will inhibit the ␣ T -stimulated PDE activity (Wensel and Stryer, 1986). Taken together, these findings imply that a positional change in the binding interaction of the ␥ PDE subunits with the ␣ PDE and ␤ PDE subunits occurs during PDE enzyme activation following the binding of the activated ␣ T subunit.