The N Terminus of GTPγS-activated Transducin α-Subunit Interacts with the C Terminus of the cGMP Phosphodiesterase γ-Subunit*

Dynamic regulation of G-protein signaling in the phototransduction cascade ensures the high temporal resolution of vision. In a key step, the activated α-subunit of transducin (Gαt-GTP) activates the cGMP phosphodiesterase (PDE) by binding the inhibitory γ-subunit (PDEγ). Significant progress in understanding the interaction between Gαt and PDEγ was achieved by solving the crystal structure of the PDEγ C-terminal peptide bound to Gαt in the transition state for GTP hydrolysis (Slep, K. C., Kercher, M. A., He, W., Cowan, C. W., Wensel, T. G., and Sigler, P. B. (2001) Nature 409, 1071–1077). However, some of the structural elements of each molecule were absent in the crystal structure. We have probed the binding surface between the PDEγ C terminus and activated Gαt bound to guanosine 5′-O-(3-thio)-triphosphate (GTPγS) using a series of full-length PDEγ photoprobes generated by intein-mediated expressed protein ligation. For each of seven PDEγ photoprobe species, expressed protein ligation allowed one benzoyl-l-phenylalaine substitution at selected hydrophobic C-terminal positions, and the addition of a biotin affinity tag at the extreme C terminus. We have detected photocross-linking from several PDEγ C-terminal positions to the Gαt-GTPγS N terminus, particularly from PDEγ residue 73. The overall percentage of cross-linking to the Gαt-GTPγSN terminus was analyzed using a far Western method for examining Gαt-GTPγS proteolytic digestion patterns. Furthermore, mass spectrometric analysis of cross-links to Gαt from a benzoyl-phenylalanine replacement at PDEγ position 86 localized the region of photoinsertion to Gαt N-terminal residues Gαt-(22–26). This novel Gαt/PDEγ interaction suggests that the transducin N terminus plays an active role in signal transduction.

To date, x-ray crystallography (1-4) has described many key features of G-protein/effector interactions. However, the majority of x-ray structures of G-protein ␣-subunits, including the ␣-subunit of transducin (G␣ t ), 4 lack information concerning the organization of their N termini. For example, x-ray analysis of the complex formed between the PDE␥ C terminus, G␣ t , and the catalytic domain of the ninth member of the regulator of G-protein signaling family (RGS9), required prior removal of the G␣ t N terminus (1). Intriguingly, the N terminus was present in the x-ray structure of the transducin G␣␤␥ t heterotrimer (4) where it adopts an extended ␣-helix participating in extensive contacts with transducin ␤-subunit. This interaction, called the N-terminal interface, contributes roughly one-third of the total binding energy of G␣ t -GDP toward the ␤-subunit (4). The photocross-linking study reported here extends the role of the G␣ t N terminus to interactions with PDE␥, suggesting a possible regulatory role for the G-protein N terminus in interactions with effectors.
The extreme G␣ t N terminus is myristoylated (5,6). This myristoyl group is required for formation of the resting state G␣␤␥ t heterotrimer (4, 7), but is not primarily responsible for membrane attachment (8 -11). Compared with other G-protein ␣-subunits that are also palmitoylated, G␣ t is uniquely soluble in low-salt buffers or cell cytosol-like buffers, and translocates to the lipid membrane only when in the transition state or its AlF 4 Ϫ analog (8,9,12,13). Myristoylation of the G␣ t N terminus may modulate transducin activity (8,14). Although the precise mechanism is not fully understood, intramolecular binding sites for the G␣ t N terminus and its myristoyl group have been proposed (15,16). The peptidyl portion of the G␣ t N terminus also makes contact with peptidyl sites within G␤␥ t . In this work, we demonstrate that the G␣ t N terminus also makes contact with the C terminus of PDE␥. PDE␥ maintains nanomolar binding affinity to activated G␣ t through two discrete regions. The central portion (residues 24 -46) and the C-terminal region (residues 46 -87) each contribute roughly half of the binding energy toward transducin (17). The PDE␥ C-terminal region also modulates GTPase-activating protein affinity toward transducin, primarily through the action of seven hydrophobic residues (18). The PDE␥ C terminus interacts primarily with the ␣3 helix and conserved switch regions (1,19). It is also thought that key conformational changes in PDE␥ residue Leu-76 and surrounding residues, initiated by G␣ t binding, may displace the PDE␥ C-terminal domain from the P␣␤ catalytic site (20). In the x-ray crystal structure of the RGS9 bound model of PDE␥/G␣ t ⅐AlF 4 Ϫ interactions, a peptide PDE␥-(50 -87) interacts primarily with the ␣3 helix of G␣ t ⅐AlF 4 Ϫ , a binding surface also detected by a peptide truncation study (21) and a photocross-linking experiment utilizing full-length PDE␥ (22). In this crystal structure information concerning the G␣ t N terminus or the myristoyl modification was not available. Fluorescence anisotropy and electron spin resonance measurements of the inhibitory G-protein ␣-subunit, G␣ i , revealed that the conformation of the N terminus is dynamic (15,16), where activation (i.e. dissociation from the G␤␥ i subunits) results in the "melting" of helical structure within the G␣ i N terminus. Considering the high homology and structural similarity of G␣ i to G␣ t (23) it is plausible to speculate that the local structure of the G␣ t N terminus is also modulated by activation. Perhaps a "melted" G␣ t N terminus may interact with PDE␥.
We have employed a library of intein-derived, full-length PDE␥ benzophenone photoprobes, generated by expressed protein ligation (24,25), to screen for photoinsertion into the N terminus of G␣ t -GTP␥S. The limited proteolysis strategy for selective proteolysis of the G␣ t -GTP␥S N terminus (26,27) was used to detect photoinsertion into the G␣ t N terminus by two independent methods. Biotin/streptavidin far Western blot analysis revealed that the W70Z, F73Z, and L81Z and I86Z photoprobes demonstrated significant levels of photoinsertion into the G␣ t -GTP␥S N terminus. Matrix-assisted laser desorption ionization time of flight mass spectrometric (MALDI-TOF MS) analysis of crosslinked peptide fragments localized cross-linking from PDE␥ I86Z to G␣ t N-terminal residues G␣ t - (22)(23)(24)(25)(26).
Synthesis of C-terminal PDE␥ Peptides-The PDE␥ C-terminal peptides and the rhodopsin (Rh) 3rd intracellular loop peptide VKEAAAQQQESATTQKAEKEVTR, residues 230 -252, were synthesized (25 mol scale) by standard Fmoc chemistry (29) on an Applied Biosystems Synergy 432A peptide synthesizer at the University of Wisconsin Peptide Synthesis Center. The seven PDE␥-(62-87ZXG) C-terminal synthetic peptides were created based on the PDE␥ C-terminal sequence CDITVIAPWEAFNHLELHELAQYGIIXG, where the appropriate benzoyl-L-phenylalanine (Z) and biotinyl-lysine/glycine (XG) substitutions were incorporated using the Fmoc derivatives of benzoyl-L-phenylalanine (Bachem) and biotin-L-lysine (Anaspec). The N-termi-nal residue of each peptide was cysteine to accommodate the intein ligation chemistry, and an additional C68A replacement was made to minimize disulfide formation. The C-terminal peptides were deprotected with 0.3 M dithiothreitol in 90% trifluoroacetic acid, precipitated three times in tert-butyl-methyl-ether, resuspended in H 2 O, and lyophilized. The sequence and molecular mass of each PDE␥ C-terminal peptide is reported as follows: Intein-mediated Synthesis and Purification of PDE␥ Photoprobes-The PDE␥-(1-61)/intein fusion protein was overexpressed in E. coli by standard methods, and purified by the method of Evans (24). After cell lysis by sonication, the PDE␥-(1-61)/intein construct was incubated overnight with chitin beads (10 ml/liter of cell extract) at 4°C, in the presence of lysis buffer (50 mM NaH 2 PO 4 , 500 mM NaCl, 1 mM EDTA, 0.1% Triton X-100, 0.1 mM phenylmethylsulfonyl fluoride, pH 7.0). Bead-bound PDE␥ (1-61) was washed with 10 volumes of high-salt buffer (50 mM NaH 2 PO 4 , 500 mM NaCl, 1 mM EDTA, 0.1 mM phenylmethylsulfonyl fluoride, pH 7.0), followed by 10-ml of low-salt buffer (50 mM NaH 2 PO 4 , 50 mM NaCl, 1 mM EDTA, pH 8.0). Bead-bound PDE␥-(1-61) was transferred to a 50-ml Falcon tube and then eluted with an equal volume of 0.5 M ␤-mercaptoethanesulfonic acid in lowsalt buffer. The activated intermediate PDE␥-(1-61)/␤-mercaptoethanesulfonic acid was aliquoted, and frozen at Ϫ80°C until use. Preparations were stable for up to 1 month. To initiate intein ligation, 1 mg of C-terminal PDE␥ peptide was solubilized in 50 l of Me 2 SO, then added to a 2-ml aliquot of PDE␥-(1-61)/␤-mercaptoethanesulfonic acid eluate. The expressed protein ligation mixture was then incubated for 24 h at room temperature (22-24°C). The resultant full-length PDE␥ photoprobes were purified over Ni-NTA resin (Qiagen) in the presence of a urea-containing column wash buffer (20 mM NaH 2 PO 4 , 8 M urea, 20 mM ␤-mercaptoethanol, pH 7.0). After washing with 10 column volumes of wash buffer, photoprobes were eluted in batch using low pH imidazole buffer (100 mM NaOAc, 250 mM imidazole, 8 M urea, 20 mM ␤-mercaptoethanol, pH 4.0). Full-length photoprobes were purified from excess PDE␥-(1-61) by HPLC on a self-packed POROS R2 column, utilizing an acetonitrile gradient from 10 -50% over 40 min, in the presence of 0.1% trifluoroacetic acid. The PDE␥ concentration was determined by the Bradford/Lowry method (30) and also by the level of Coomassie staining when analyzed using SDS-PAGE gels. The full-length PDE␥ photoprobe was lyophilized on a speed vacuum centrifuge (Savant) and stored at Ϫ80°C.
Purification of G␣ t -GTP␥S and G␣ t -(26 -350)-GTP␥S-Rod outer segment membranes were purified from dark-adapted frozen retinas (W. L. Lawson Co). Both the G␣ t -GTP␥S and G␤␥ t subunits were released from the purified rod outer segment membranes then purified over a Blue Sepharose CL-6B column by standard methods (34). Purity of G␣ t -GTP␥S was assessed by SDS-PAGE. To generate G␣ t -(26 -350)-GTP␥S, G␣ t -GTP␥S was first treated with endo Lys-C with 1:100 (w/w, enzyme:protein) for 18 h, and G␣ t -(26 -350)-GTP␥S was isolated over Blue Sepharose as described for intact G␣ t -GTP␥S. Recombinant RGS9/G␤5 for use in GTPase turnover measurements was obtained by published procedures (35).
GTPase Assays-The ability of each PDE␥ photoprobe species to stimulate the RGS9-catalyzed GTPase activity of G␣ t was measured using the single turnover approach (36,37). The assays were conducted at room temperature (22-24°C) in a buffer containing 25 mM Tris-HCl (pH 8.0), 140 mM NaCl, and 8 mM MgCl 2 . Urea-treated rod outer segment membranes, lacking endogenous RGS9 activity, were used as a source for the photoexcited rhodopsin required for transducin activation. The reactions were initiated by the addition of 10-l of 0.6 M [ 32 P]GTP (ϳ10 5 disintegrations/min/sample) to 20 l of urea-treated rod outer segments membranes (20 M final rhodopsin concentration) reconstituted with transducin heterotrimer (1 M) and recombinant RGS9-G␤5 complex (0.5 M). The reactions were performed either in the absence or presence of PDE␥ derivatives (1 M). The reaction was stopped by the addition of 100 l of 6% perchloric acid, then 32 P i formation was measured with activated charcoal as described (37).
Photocross-linking of PDE␥ Benzophenone Photoprobes to G␣ t -GTP␥S-To initiate photocross-linking experiments, 20 g of PDE␥ photoprobe was incubated with 100 g of G␣ t -GTP␥S for 20 min at room temperature. The total reaction volume was 200 l in crosslinking buffer comprised of 20 mM Tris (pH 7.4), 100 mM NaCl, and 8 mM MgCl 2 . Samples were irradiated with 350 nm of UV light in 10-min increments for a total of 30 min in a Rayonet UV irradiator (Southern New England Ultraviolet Company, CT), at 4°C. The percentage of PDE␥ photoinsertion into G␣ t -GTP␥S, relative to the total quantity of G␣ t -GTP␥S present in each cross-linking reaction, was determined from Coomassie-stained gels by measuring the intensity of both the cross-linked and unreacted G␣ t -GTP␥S. Because PDE␥ stains poorly on the 15% SDS-PAGE gels relative to G␣ t -GTP␥S, no adjustment was necessary to account for the PDE␥ content in the cross-linked bands.
Endo Lys-C Digestion of PDE␥/G␣ t -GTP␥S Cross-links and Far Western Analysis-Limited digestion experiments were performed as described previously with minor modifications (26,27). Endo Lys-C digestion of cross-links, which removes the transducin N terminus selectively and the N-terminal half of PDE␥, was performed for 18 h at room temperature (1:100, w/w, enzyme:total protein). Aliquots (2 or 5 l) were electrophoresed on duplicate 15% SDS-PAGE minigels (Hoefer). Protein bands were either stained with Coomassie Blue or transferred to nitrocellulose membranes as described under "Analytical Methods." The extent of PDE␥ photoprobe cross-linking to the G␣ t -GTP␥S protein core ("internal XL") was determined by first measuring the relative level of biotin-positive signal representing intact, cross-linked PDE␥/G␣ t -GTP␥S species (50 kDa) prior to limited digestion, here termed "total XL." Limited endo Lys-C removes cross-links from PDE␥ to the G␣ t N terminus, leaving only unreacted G␣ t at 36 kDa and/or PDE␥-cross-linked G␣ t -(26 -350)-GTP␥S at 41 kDa. The cross-linked band at 41 kDa reflects proteolytic removal of the PDE␥ N-terminal half as well as the G␣ t N terminus. The relative level of biotin-positive signal remaining associated with the 41-kDa band, representing cross-links to G␣ t -(26 -350)-GTP␥S, is denoted as internal XL. The ratio of internal cross-linking to total cross-linking was defined as percent of internal cross-linking. The percent photoinsertion into the G␣ t -GTP␥S N terminus (N-terminal XL) was inversely proportional to the degree of internal photoinsertion.
Endo Glu-C Digestion of PDE␥/G␣ t -GTP␥S Cross-links and HPLC Purification of Fragments-After cross-linking was performed, the unpurified reaction products were treated directly with endo Glu-C protease at a ratio of 1:20, w/w, enzyme:protein. The digest mixture was allowed to incubate 18 h at room temperature, quenched with 100 M TLCK, and frozen at Ϫ80°C. Protein fragments were purified by HPLC over a POROS R2 column, utilizing a steep gradient from 10 to 90% ACN over 20 min. Individual 1-ml fractions were collected, frozen, and dried by vacuum centrifugation.
MALDI-TOF MS Analysis of Cross-linking Reactions-Individual, lyophilized HPLC fractions subjected to the endo Glu-C digestion procedure were resuspended in 5-l of a solution of 20% ACN. A 0.5-l aliquot of each fraction was spotted onto an individual well on the MALDI target. To this, 0.5-l of saturated ␣-cyano-4Ј-hydroxylcinnamic acid matrix solution (70% ACN, 0.1% trifluoroacetic acid) was added. Samples were analyzed on a Bruker REFLEX II MALDI-TOF mass spectrometer (Billerica, MA) using a 337-nm N 2 laser and both positive reflectron and positive linear modes. All spectra (at least 50 shots) were calibrated with an appropriate combination of mass standards including bradykinin (1,060.6), neurotensin (1,672.9), insulin (5,734.6) and cytochrome c (12,360).
Analytical Methods-SDS-PAGE (15%) was performed as described by Laemmli (38), and Tricine SDS-PAGE (16.5%) was performed by the method of Schagger and von Jagow (39). Protein molecular weight standards from Sigma and prestained standards from Bio-Rad were used to approximate molecular weight. The homogeneity of PDE␥ photoprobe preparations was assessed using 16.5% Tricine gels, and all other experiments were performed with 15% SDS-PAGE gels. Protein bands were transferred to 0.2-m nitrocellulose (Pierce) and subjected to far Western analysis according to the manufacturer's instructions. Enhanced chemiluminescence (ECL) detection of biotin was performed using streptavidin-conjugated horseradish peroxidase and ECL reagents obtained from Pierce. Densitometric measurements of Coomassie-stained gel bands and of far Western blots were acquired with a HP ScanJet laser scanner and quantified with NIH Image.

RESULTS
Characterization of PDE␥ Photoprobes-All seven full-length PDE␥ photoprobes (Fig. 1A) were purified to homogeneity. The progress of the two-step purification by Ni-NTA chromatography, followed by HPLC, is illustrated for the I86Z photoprobe (Fig. 1, B and C). Purified photoprobe preparations were homogeneous as judged by SDS-PAGE analysis (Fig. 1D, lane 1). Analysis of two PDE␥-based standards (Fig.  1D, lane 2) demonstrated that the apparent molecular mass of the fulllength PDE␥ photoprobes on the SDS-PAGE gel was greater than both wild type PDE␥ (molecular mass Ϸ 11 kDa) and the PDE␥-(1-61) fragment (M r Ϸ 7 kDa). The anomalous apparent migration at approximately 14 kDa was presumably because of the His 6 and biotin/glycine additions. MALDI-TOF mass spectrometry (see "Experimental Proce-dures") indicated, however, that these photoprobes attained the correct molecular mass.
The maximal percentage of photoinsertion of the seven PDE␥ C-terminal photoprobes into G␣ t -GTP␥S was evaluated at saturating concentrations of G␣ t -GTP␥S (Fig. 2, A and B). The F73Z photoprobe demonstrated the highest level of cross-linking to G␣ t -GTP␥S. In contrast, a low level (ϳ10%) of the V66Z photoprobe to G␣ t -GTP␥S was detected on Coomassie-stained SDS-PAGE gels, even at saturating concentrations of PDE␥. The W70Z, L76Z, L78Z, L81Z, and I86Z photoprobes showed greater than 30% cross-linking.
The functional activity of each PDE␥ photoprobe was determined by its ability to stimulate the RGS9-catalyzed GTPase activity of G␣ t in reconstituted membrane preparations (Fig. 3A). The role of PDE␥ in this reaction is to enhance the affinity between activated G␣ t and RGS9 (40), and it has been demonstrated that the His 6 addition to the PDE␥ N terminus has no effect on functional activity toward the G␣ t -GTPase or the cGMP phosphodiesterase catalytic core (37). All PDE␥ photoprobes except W70Z were functionally active and retained between 50 and 80% of the wild type PDE␥ activity. The activity of the W70Z PDE␥ photoprobe was severely impaired and amounted to no more than 5% of the wild type PDE␥ activity, consistent with a previous report that alanine replacement at this position abrogated the PDE␥ GTPase-activating protein activity (18). A species of the F73Z photoprobe lacking the biotin-lysine/glycine modification retained wild type activity levels (data not shown). Although the F73Z photoprobe containing the biotin modification retained only 70% activity in the turnover assay, percent crosslinking for both the F73Z with biotin and the F73Z photoprobe lacking biotin were identical (data not shown).
The specificity of cross-linking of PDE␥ photoprobes to G␣ t -GTP␥S was addressed in protection experiments (Fig. 3B). Protection of PDE␥/ G␣ t -GTP␥S cross-linking by the prior addition of wild type PDE␥ was complete, demonstrating the specific nature of the light-dependent photoinsertion. The dark control confirmed that cross-links failed to form in the absence of UV irradiation. Irradiation with 350-nm UV light as described under "Experimental Procedures" resulted in the formation of the 50-kDa band representing PDE␥/G␣ t -GTP␥S cross-linked spe-FIGURE 1. Synthesis, purification, and characterization of full-length PDE␥ photoprobes. A, benzoyl-L-phenylalanine (Z) replacements were made throughout the PDE␥ C terminus (Val-66, Trp-70, Phe-73, Leu-76, Leu-78, Leu-81, and Ile-86), at hydrophobic positions known to influence PDE␥ GTPase-activating protein activity toward G␣ t -GTP␥S. Each photoprobe contained both hexylhistidine (6HIS) and biotin-lysine/glycine (XG) motifs. B, full-length PDE␥ photoprobes were purified from excess synthetic peptide by chromatography over Ni-NTA resin as described under "Experimental Procedures." Lane 1, urea-solubilized synthetic mixture. Lane 2, elution of PDE␥ photoprobe, excess PDE␥-(1-61), and gyrA intein fragment from Ni-NTA resin. C, full-length PDE␥ photoprobes (e.g. PDE␥ I86Z) were readily purified from excess PDE␥-(1-61) and the intein fragment by reverse-phase HPLC as described under "Experimental Procedures," and analyzed by spectrophotometric analysis of the HPLC trace at 215 nm. D, a Coomassie-stained SDS-PAGE gel revealed that the purified PDE␥ I86Z photoprobe (lane 1) was Ͼ95% pure and migrated at a higher molecular weight compared with either wild-type PDE␥ or the PDE␥-(1-61) precursor (lane 2).
cies. Furthermore, PDE␥ photoinsertion into G␣ t -GTP␥S was blocked by the addition of 20 M wild type PDE␥. In contrast, addition of a large excess (350 M) of a peptide representing the 3 rd intracellular loop of rhodopsin had minimal effect on the extent of G␣ t -GTP␥S cross-linking.
To determine the extent of photoinsertion of the F73Z and I86Z PDE␥ photoprobes into the G␣ t (GDP)␤␥ heterotrimer, cross-linking experiments were performed with 10 M of either PDE␥ photoprobe in the presence of 12 M G␣ t (GDP)␤␥. Photoinsertion of the F73Z and I86Z photoprobes into the heterotrimer was minimal (Fig. 3C). The F73Z photoprobe cross-linked to G␣ t to less than 5%, and the I86Z photoprobe did not exhibit a measurable level of photoinsertion into G␣ t (GDP). Cross-linking to the ␤-subunit was not observed in either case. These experiments confirmed that the PDE␥ photoprobes crosslinked primarily to activated G␣ t -GTP␥S.

Analysis of PDE␥/G␣ t -GTP␥S Cross-links by Limited Proteolysis and Far Western
Blot-After photoactivation, cross-linking of intact PDE␥ photoprobes to intact G␣ t yielded a 50-kDa cross-linked band, which contained the biotin-labeled cross-links. The biotin-positive PDE␥/G␣ t cross-linked complexes were detected by far Western analysis (Fig. 4A), utilizing streptavidin-conjugated horseradish peroxidase to visualize the biotin incorporated into the PDE␥ C terminus. Limited proteolysis of the PDE␥/G␣ t cross-linked species with endo Lys-C selectively removed the G␣ t N terminus. If PDE␥ were cross-linked only to the G␣ t N terminus, all PDE␥ cross-links would have been removed concomitantly by limited digestion. In this case the trimmed G␣ t -(26 -350) molecule lacking the biotin-labeled cross-link would form a 36-kDa gel band that would remain silent in far Western analyses with streptavidin.
In contrast, post cross-linking proteolytic removal of the G␣ t N terminus with endo Lys-C would not remove PDE␥ cross-links that inserted into G␣ t -(26 -350), termed here internal cross-links. The PDE␥ portion of the cross-link is split into a 5-kDa C-terminal fragment (residues 46 -87XG), as well as several smaller PDE␥ N-terminal fragments. Because all the benzoyl-L-phenylalanine replacements reported in this work occur beyond position Val-66, the biotinylated PDE␥ C terminus does not become separated from PDE␥/G␣ t internal crosslinks. In this case, the species representing PDE␥-(46 -87XG) crosslinked to G␣ t -(26 -350) migrated as a 41-kDa (36 kDa ϩ 5 kDa) band on SDS-PAGE gels, which was visible in streptavidin-based far Western experiments.

FIGURE 2. Percent photoinsertion of each full-length PDE␥ photoprobe into G␣ t -GTP␥S was assessed by SDS-PAGE analysis.
The percentage of PDE␥/G␣ t cross-link formation was determined as described under "Experimental Procedures." Cross-linking reactions were performed at saturating concentrations of PDE␥, at a molar ratio of at least 1.5:1. Individual aliquots were electrophoresed on 15% SDS-PAGE gels. A, Coomassie-stained gel demonstrating cross-linking yield for each of the seven photoprobes, at saturating concentrations of photoprobe. B, percent photoinsertion was determined by quantifying the Coomassie staining of both the cross-linked and non-cross-linked transducin bands. Data represent the average of triplicate measurements.

FIGURE 3. Characterization of PDE␥ C-terminal photoprobe functional activity and cross-linking potential.
A, the ability of each PDE␥ photoprobe to promote GTP hydrolysis by transducin was tested in single turnover GTPase assays as described under "Experimental Procedures." The experimentally determined rate constants of GTP hydrolysis were plotted as percent of that measured for wild type PDE␥ after subtracting the rate constant value obtained in the absence of PDE␥. All of the PDE␥ photoprobes demonstrated GAP activity similar to that of unmodified PDE␥ except the W70Z photoprobe, which was severely compromised. B, cross-linking from PDE␥ to G␣ t -GTP␥S is specific, as demonstrated for the F73Z photoprobe. Cross-linking reactions were performed as described under "Experimental Procedures," and analyzed by a Coomassiestained gel: dark control (lane 1), photoinsertion as a result of irradiation at 350 nm (lane 2), and protection by the addition of wild type PDE␥ at a molar ratio of 2:1 (lane 3). In contrast, the presence of an unrelated peptide from the rhodopsin (Rh) 3rd loop did not interfere with cross-linking of PDE␥ photoprobes to G␣ t -GTP␥S (lane 4). C, the ability of the F73Z and I86Z PDE␥ photoprobes to photoinsert into the G␣ t (GDP)␤␥ heterotrimer was examined. The control lane containing irradiated heterotrimer (but no PDE␥ photoprobe) demonstrated some faint Coomassie-stained protein bands in the 49 -50-kDa range (lane 1). The F73Z photoprobe cross-linked the heterotrimer ␣-subunit to an extent of less than 5% (lane 2), whereas the I86Z photoprobe did not demonstrate crosslinking to the transducin heterotrimer (lane 3).
The degree of insertion of the PDE␥ benzophenone photoprobes into the N terminus of G␣ t -GTP␥S was assessed by first measuring the percentage of biotin-labeled internal cross-links remaining on the G␣ t -(26 -350) after limited endo Lys-C proteolysis. These represent crosslinks to the G␣ t "core." Six of the seven photoprobes were screened for photoinsertion into the G␣ t -GTP␥S N terminus, with the exception of the V66Z photoprobe, which did not appreciably form cross-links.
The photoinsertion level of the W70Z photoprobe (Fig. 4A) was moderate, and roughly equally proportioned between N-terminal and internal cross-linking. Although this replacement had a deleterious effect on GTP turnover ( Fig. 2A), the unsaturated aryl rings of the benzophenone made functionally non-productive contacts with G␣ t -GTP␥S. The primary site of photoinsertion for the F73Z photoprobe was the G␣ t -GTP␥S N terminus (Fig. 4A), because limited proteolysis of the G␣ t N terminus removed the majority (Ͼ90%) of the biotin-labeled PDE␥ cross-links from G␣ t . The L76Z and L78Z photoprobes cross-linked primarily into internal G␣ t regions, as indicated by the higher percentage of biotin label remaining covalently bound to G␣ t -(26 -350) after limited proteolysis. In contrast, the L81Z and I86Z photoprobes crosslinked substantially to both the G␣ t -GTP␥S N terminus as well as to internal regions. These data demonstrate position-dependent crosslinking from several positions within the PDE␥ C terminus to the N terminus of G␣ t -GTP␥S.
Potential deleterious conformational effects of the biotinyl-lysine/ glycine addition to the C terminus of the intein-derived photoprobes were evaluated in an independent experiment using two PDE␥ photoprobe species incorporating a maleimido-benzophenone moiety at either position 73 (F73CysMBP) or 87 (I87CysMBP). In this experiment, the full-length PDE␥ photoprobe contained a maleimido-benzophenone addition to a cysteine engineered into the appropriate position but retained the intact, unmodified extreme PDE␥ C terminus (i.e. no biotin addition). The F73CysMBP photoprobe (Fig. 4B, lane 1) cross-linked to G␣ t at a level similar to the F73Z intein-based photoprobe. Furthermore, limited endo Lys-C digestion completely reduced the PDE␥/G␣ t cross-linked molecule (50 kDa) to the 36-kDa G␣ t -(26 -350) species (Fig. 4B, lane 2), demonstrating photoinsertion predominantly into the G␣ t N terminus. Although the PDE␥ I87CysMBP photoprobe crosslinked to G␣ t at levels higher than that of the PDE␥ I86Z intein-derived photoprobe (approximately 50% ; Fig. 4B, lane 3), limited endo Lys-C digestion resulted in the formation of both the 36-kDa G␣ t -(26 -350) species and the 40 -41-kDa trimmed, cross-linked PDE␥/G␣ t -(26 -350) species (Fig. 4B, lane 4). Therefore, the PDE␥ I87CysMBP photoprobe cross-linked to both G␣ t -(26 -350) and the G␣ t N terminus. Because position-dependent photoinsertion into the G␣ t N terminus by the F73CysMBP and I87CysMBP PDE␥ photoprobes confirmed the results obtained with the intein-derived PDE␥ photoprobes, we con-

. Cross-linking of PDE␥ C-terminal photoprobes to the G␣ t -GTP␥S N terminus. A, screening for biotin by far Western analysis of cross-linking reactions before (Total XL,
Ϫ lanes) and after (ϩ lanes) removal of the G␣ t -GTP␥S N terminus by limited endo Lys-C proteolysis revealed the extent of N-terminal cross-linking for the W70Z, F73Z, L76Z, L78Z, L81Z, and I86Z PDE␥ C-terminal photoprobes. The percentage of PDE␥ cross-links remaining covalently cross-linked with the G␣ t -(26 -350) core (Internal XL) was compared with the total extent of cross-linking before digestion (Total XL). Percent photo-incorporation into the G␣ t -GTP␥S N terminus was measured as described under "Experimental Procedures" and estimated from at least two experiments, where the percent of N-terminal cross-linking (NTERM) was calculated as (Total-Internal XL). B, intact cross-links between G␣ t -GTP␥S and the F73CysMBP (lane 1) and I87CysMBP (lane 3) PDE␥ photoprobes were treated to limited Endo-Lys C proteolysis as described under "Experimental Procedures" (lanes 2 and 4,  respectively). The presence of cross-links to either G␣ t at 50 kDa or G␣ t -(26 -350) at 41 kDa was evaluated on a 15% SDS-PAGE gel stained with Coomassie Blue. C, cross-linking to G␣ t -(26 -350)-GTP␥S. Photoinsertion of the F73Z and I86Z intein-based PDE␥ photoprobes to G␣ t -(26 -350) was evaluated. The G␣ t -GTP␥S N terminus was removed by limited endo Lys-C proteolysis to generate G␣ t -(26 -350)-GTP␥S (⌬NT), and this truncated species was purified by gel filtration as described under "Experimental Procedures." Photoinsertion of the F73Z and I86Z PDE␥ photoprobes into both intact G␣ t -GTP␥S (lanes 1 and 3) and G␣ t -(26 -350)-GTP␥S (lanes 2 and 4) was assessed by SDS-PAGE. Photoinsertion into the G␣ t -GTP␥S N terminus is reported as a percentage, as described under "Experimental Procedures." clude that the presence of the biotinyl-lysine/glycine addition to the PDE␥ C terminus did not significantly affect the conformation of the PDE␥ C terminus when bound to G␣ t -GTP␥S.
To confirm the importance of the G␣ t N terminus as a photoinsertion site, cross-linking experiments were carried out directly on G␣ t -(26 -350)-GTP␥S, termed ⌬NT, where the N terminus was removed prior to cross-linking (Fig. 4C). In contrast to the robust cross-linking to intact, wild-type G␣ t -GTP␥S (lane 1), the F73Z photoprobe did not cross-link appreciably to G␣ t -(26 -350)-GTP␥S (lane 2), emphasizing the role of the G␣ t N terminus as the predominant photoinsertion site. Compared with intact G␣ t -GTP␥S (lane 3), the I86Z photoprobe retained a substantial degree of cross-linking into G␣ t -(26 -350)-GTP␥S (lane 4) despite prior removal of the G␣ t N terminus. This indicated that crosslinking to the G␣ t protein core was significant for this photoprobe. These data support the results of the far Western/proteolysis experiments identifying the G␣ t N terminus as the primary target of the F73Z photoprobe, whereas the PDE␥ I86Z photoprobe cross-linked to both the G␣ t N terminus and the protease-resistant G␣ t protein core.

Direct Detection of PDE␥/G␣ t -GTP␥S Cross-linked Peptides by MALDI-TOF MS-
The PDE␥ I86Z photoprobe was selected for further analysis of the photoinsertion site within the G␣ t N terminus because of the small size of its PDE␥ contribution to endo Glu-C-digested crosslinks. After cross-linking of the I86Z photoprobe to G␣ t -GTP␥S, the reaction mixture was subjected to limited endo Glu-C digestion. Then, the crude digest was chromatographed over a POROS R2 HPLC column. MALDI-TOF MS analysis of individual HPLC fractions was performed to identify PDE␥ peptide fragments that were cross-linked to the G␣ t N terminus.
A unique, cross-linked peptide was identified at m/z 1900 (Fig. 5A), reflecting the presence of contributions from both PDE␥ and G␣ t . The PDE␥ contribution of 1325 Da reflects complete endo Glu-C digestion of the PDE␥ C terminus to yield PDE␥-(81-87XG), termed Fragment I. The mass of the G␣ t peptide (574 Da) identified this peptide as DAEKD (Table 1). This signal in the mass spectrum could not be attributed to unreacted G␣ t -GTP␥S fragments, excess unreacted PDE␥ fragments, or intramolecularly cross-linked PDE␥ fragments. A second peptide signal at m/z 1585 was assigned to either G␣ t - (16 -17) or G␣ t -(25-26), but could not be assigned to one unique photoinsertion site within G␣ t . To further validate the light-activated cross-linking approach, control experiments were performed in which identical cross-linking mixtures were treated similarly, but were not photoactivated (Fig. 5B). No analyte signals were detected at m/z 1585 and 1900, which confirmed the uniqueness of the cross-linked peptides to genuine photolabeling experiments.
In the same experiment, a third cross-linked peptide was identified at m/z 1944 (data not shown) from the ninth fraction after chromatography over POROS resin. This peptide eluted at 70% ACN, and reflected cross-linking of an incompletely digested PDE␥ C-terminal fragment to G␣ t - (23)(24). In this case, one missed endo Glu-C cleavage event resulted in a slightly larger contribution from the PDE␥ peptide fragment PDE␥-(78 -87XG), termed PDE␥ Fragment II (Fig. 5C) that contributed 1705 Da to the cross-linked species. A background signal because of the presence of incompletely digested, non-cross-linked PDE␥ Fragment II was present in the mass spectrum of the dark control that was not subjected to UV irradiation (data not shown).
The digestion experiment was repeated several times in duplicate, and each time the cross-linked peptides were purified over the POROS column. This approach revealed variation in the position of proteolytic cleavage, as well as heterogeneous photoinsertion into the G␣ t N terminus. Overlapping cross-linked peptide fragments indicate the clustering of photoinsertion sites into the distal portions of the G␣ t -GTP␥S N-terminal region, residues 22-26 (Table 1). Four unique cross-linked peptides were identified. The smallest sequence identified in these experiments was AE (23 to 24) at 217 Da. The peptide AEKD was also identified in these experiments. Overall, the overlapping nature of peaks identified in this manner confirmed cross-linking of PDE␥ photoprobes to the G␣ t -GTP␥S N terminus, and localized them primarily to the region G␣ t - (22)(23)(24)(25)(26).

DISCUSSION
The G␣ t interactions with PDE␥ are critical for both effector activation and rapid inactivation of G␣ t by RGS9. Multiple studies based on x-ray crystallography (1), alanine scanning mutagenesis (18), sequential truncation of PDE␥ C-terminal residues in a C-terminal peptide study FIGURE 5. MALDI-TOF MS identification of PDE␥ C-terminal photoprobe insertion into the G␣ t -GTP␥S N terminus. PDE␥/G␣ t -GTP␥S cross-links were subjected to limited endo Glu-C proteolysis, and peptide fragments were fractionated by HPLC as described under "Experimental Procedures." MALDI-TOF MS analysis of the chromatographed fractions from photocross-linking experiments identified several cross-linked peptides. A, photocross-linking experiment. As an example, the mass spectrum from the seventh HPLC fraction identified two cross-linked peptides at m/z 1585 and 1900. Subtraction of the mass of the completely digested PDE␥ C-terminal contribution (PDE␥-(81-87)) from the mass of XL A NTERM identified two putative photoinsertion sites, into either G␣ t - (15)(16) or G␣ t - (25)(26). The cross-linked peptide at m/z 1900 (XL B NTERM ) identified cross-linking between G␣ t - (22)(23)(24)(25)(26) and PDE␥-(81-87XG). B, dark control. These peptide signals were completely absent from the mass spectrum of dark control fractions treated similarly. C, PDE␥ contributions to cross-links. Complete digestion of the PDE␥ C terminus resulted in the formation of PDE␥-(81-87XG), the smallest possible PDE␥ contribution to PDE␥/G␣ t cross-links, termed PDE␥ Fragment I. In the case of incomplete endo Glu-C digestion of the PDE␥ C terminus, a larger PDE␥ contribution to PDE␥/G␣ t cross-links was present. This fragment spanned PDE␥ residues PDE␥-(78 -87XG), and is identified as PDE␥ Fragment II. (21), and a cross-linking study (22) revealed that the effector activating region of PDE␥ is located within the residues 46 -87 and that it interacts with the switch II/␣3 region of G␣ t . However, some aspects of the molecular mechanism of effector activation remain unknown, particularly regarding the role of the G␣ t N terminus, which is missing in the G␣ t /PDE␥/RGS9 crystal structure.
The N terminus of G␣ t plays an important role in G␣ t function. First, it serves as the site of G␣ t myristoylation. Second, it is one of the interacting sites between G␣ t and G␤ t . Third, and most intriguingly, it was suggested to adopt different conformations in the GDP-and GTPbound forms of G␣ t thus serving as an additional, "fourth" switch region in the molecule (15,16). The latter is evident from electron paramagnetic resonance studies indicating that in the activated G␣ i -GTP␥S species devoid of myristate, the G␣ i N terminus adopts a random coil (15). Furthermore, these authors predicted that the N terminus of an activated G␣ i molecule binds intramolecularly to a hydrophobic binding site within G␣ i internal regions. It is therefore reasonable to speculate that the PDE␥ C terminus, being strongly hydrophobic, would also provide a binding surface for the lipophilic N terminus of G␣ t . A second EPR investigation (16) indicated that in the properly myristoylated, activated G␣ i species, myristate confers some degree of structure onto N-terminal residues, although it remains to be determined whether the N terminus folds onto itself or onto G␣ i internal regions. The strong homology of G␣ t to G␣ i suggests that myristate might also influence the conformation of the G␣ t N terminus, and may also influence interactions with PDE␥.
The present cross-linking study employing full-length PDE␥ photoprobes provides direct evidence in support of the interaction of the PDE␥ C terminus with the flexible N terminus of G␣ t . This interaction is contingent on G␣ t activation. The PDE␥ F73Z photoprobe demonstrated the greatest degree of photoinsertion into the G␣ t -GTP␥S N-terminal region. In contrast, the L76Z and L78Z photoprobes crosslinked primarily into the G␣ t -GTP␥S protein core, whereas the V66Z photoprobe did not substantially photoinsert into G␣ t -GTP␥S. Photoinsertion into the region G␣ t - (22)(23)(24)(25)(26) was demonstrated for the PDE␥ I86Z photoprobe when cross-linked PDE␥/G␣ t molecules were proteolytically fragmented using endo Glu-C proteinase, then analyzed by MALDI-TOF MS.
Molecular modeling with energy minimization using the Sybyl version 6.8 software package (Tripos Associates, St. Louis, MO) determined that the reactive benzophenone carbonyl (cross-linking radius 3.2 Å) could photoinsert into a carbon source 9.8 Å distant from the PDE␥ polypeptide backbone. This also includes the length of the phenylalanine-based side chain (6.6 Å). Either there is direct contact between G␣ t and residues in positions 73 and 86 of PDE␥, or the N terminus of G␣ t -GTP␥S and C terminus of PDE␥ come within 9.8 Å of each other upon binding without making discrete contacts. Both possibilities indicate association of the G␣ t N terminus with PDE␥ upon G␣ t activation. This re-arrangement may be important for providing a high affinity interface for the interaction of G␣ t with PDE␥.
The position-dependent cross-linking of the PDE␥ C-terminal photoprobes to G␣ t -GTP␥S did not conform to predictions derived from the G␣ t -GDP⅐AlF 4 Ϫ transition state crystal structure of Slep et al. (1).
According to this structure, the L76Z and I86Z photoprobes are unlikely to generate cross-links to G␣ t , because Leu-76 and Ile-86 of PDE␥ form instead intramolecular PDE␥/PDE␥ contacts. However, the L76Z and I86Z cross-linking data demonstrate both N-terminal and internal photoinsertion into G␣ t . Surprisingly, the F73Z photoprobe cross-linked primarily to the G␣ t N terminus, in contradiction to predictions from the crystal structure (1) that the F73Z photoprobe should form cross-links only to the switch II/␣3 region of G␣ t -(26 -350). Finally, the V66Z photoprobe failed to cross-link to G␣ t , whereas the crystal structure predicted robust cross-linking to G␣ t -(26 -350). The cross-linking experiments described here offer several unique advantages over x-ray crystallographic analysis of PDE␥ complexes with G␣ t . First, use of full-length PDE␥ provides greater affinity for G␣ t , much stronger than micromolar K d values available for C-terminal peptides (19), such as the PDE␥-(46 -87) peptide employed in crystallographic analysis. Indeed, this higher affinity of full-length PDE␥ allowed study of its interactions with G␣ t -GTP␥S, whereas the crystal structure analysis required inclusion of a RGS9 fragment to act as a "molecular glue." Furthermore, in terms of diffractible data within the x-ray structure, PDE␥ and RGS9 were represented by fragments spanning no more than half of their sequence, which could introduce some artifacts during crystallization.
Although it is possible that incorporation of benzophenone and biotinyl-lysine components may perturb the interaction of PDE␥ with G␣ t , we have tried to mitigate this potential by replacing only hydrophobic amino acids with benzoyl-L-phenylalanine, and incorporating the biotinyl-lysine/glycine affinity motif at the extreme PDE␥ C terminus. Furthermore, position-dependent photo-insertion into the G␣ t -GTP␥S N terminus was confirmed using the F73ZCysMBP and I87ZCysMBP PDE␥ photoprobes, which did not require modification of the extreme PDE␥ C terminus.
Overall, the cross-linking data suggest new interactions between the PDE␥ C terminus and G␣ t in the "signaling state," which involves the first 25 N-terminal amino acids of G␣ t and particularly residues 22-26. This finding advances our knowledge about the nature of interaction between transducin and its effector, cGMP PDE. It appears that the C terminus of PDE␥ interacts simultaneously with the previously identi-

TABLE 1 Identification of cross-links from PDE␥ I86Z to the G␣ t N terminus
Tabulation of cross-linking results, showing the masses of cross-linked peptides observed in the MALDI-TOF mass spectrum, the PDE␥ contribution to cross-links, and the G␣ t contribution to cross-links. fied switch II/␣3 region of G␣ t and a novel switch region, the N terminus of G␣ t . We speculate that this additional interaction may increase the affinity of G␣ t for PDE␥ with a consequent increase in the efficiency of effector activation by transducin.
Although it is too early to speculate on potential regulatory effects of interactions between the G␣ t N terminus and PDE␥, a physical interaction has been detected in this study. Furthermore, the overall pattern of the cross-linking from the seven full-length PDE␥ photoprobes to G␣ t -GTP␥S suggests that the pre-catalysis GTP␥S-modeled signaling state is distinct from that of the transition state for GTP hydrolysis in the presence of RGS9.