Interaction Sites of the COOH-terminal Region of the (cid:103) Subunit of cGMP Phosphodiesterase with the GTP-bound (cid:97) Subunit of Transducin*

In photoreceptor cells, visual transduction occurs through photoexcitation of rhodopsin, GTP activation of the (cid:97) subunit of transducin, and interaction between GTP-bound transducin (cid:97) subunit and the inhibitory (cid:103) subunit of phosphodiesterase. The (cid:103) subunit of phosphodiesterase, in turn, accelerates the hydrolysis of GTP on the (cid:97) subunit of transducin. Within the COOH-terminal residues (46–87) of the phosphodiesterase (cid:103) subunit, Trp-70 has been implicated in phosphodiesterase activa- tion, transducin (cid:97) subunit-phosphodiesterase (cid:103) subunit interaction, and the GTP hydrolysis accelerating ac- tivity. We have derivatized the phosphodiesterase (cid:103) subunit with a reversible photoactivatable reagent, [ 125 I] N -[(3-iodo-4-azidophenylpropionamido- S -(2-thiopy-ridyl)]cysteine ([ 125 I]ACTP), at cysteine (Cys-68). A light-dependent,cross-linkedcomplexofguanosine5 (cid:42) -( (cid:103) -thio)-triphosphate-bound transducin (cid:97) subunit and ACTP-derivatized phosphodiesterase (cid:103) subunit formed after photolysis of a 1:1 stoichiometic complex of the two pro- teins. The specificity of complex formation between the transducin (cid:97) subunit and the phosphodiesterase (cid:103) subunit was demonstrated by specific protection by the C68A mutant of the phosphodiesterase (cid:103) subunit. The cross-linked complex was treated with (cid:98) -mercaptoetha-nol to transfer the 125 I photomoiety from the phosphodiesterase (cid:103) subunit to the transducin (cid:97) subunit. Com-bined techniques involving electrophoresis, chemical and enzymatic cleavage, and chemical and radiosequencing were used to identify photoinsertion sites on the (cid:97) 3 and (cid:97) 4 / (cid:98) 6 regions of the transducin (cid:97) subunit. Three photo-labeled residues, His-244 ( (cid:97) 3 helix), Met-308, and Arg-310 ( (cid:97) 4 / (cid:98) 6 interface), were specifically iden- tified as photoinsertion sites. Utilizing the crystal structure coordinates of the GTP-bound transducin (cid:97) subunit and molecular modeling, we conclude that Cys-68 of the phosphodiesterase (cid:103) subunit is located at a position between the exposed face of the (cid:97) 3 and (cid:97) 4 helices of the transducin (cid:97) subunit. We propose that the phosphodiesterase (cid:103) subunit interacts with GTP-bound transducin (cid:97) subunit at multiple sites in which the cysteine 68 to tryptophan 70 sequence of the phosphodiesterase (cid:103) subunit, which is critical for GTP hydrolysis accelerating activity, interacts in the (cid:97) 3 / (cid:97) 4 / (cid:98) 6 region of GTP-bound transducin (cid:97) subunit. Automatic amino-terminal protein sequencing was performed using an ABI, model 470A, gas-phase sequencer equipped with an on-line model 120A phenylthiohydantoin analyzer. Microsequencing and radio- sequencing were performed at the Supramolecular Structure Facility, Michigan State University. SDS-PAGE (10, 12, and 15%) was per- formed by the method of Laemmli (33). Tricine 16.5% polyacrylamide electrophoresis was performed, as described by Schagger and von Jagow (34). Protein concentration was determined by the Coomassie Blue binding method (35) using bovine serum albumin as a standard or spectrophotometrically at 280 nm using a molar extinction coefficient of 7100 for P (cid:103) (24). Densitometric scans of the Coomassie Blue-stained gels and autoradiography were performed on a model SL-504XL Zeineh Soft Laser Scanning Densitometer. Computer modeling was performed on Unix Silicon Graphics using the Insight II (version 2.3.0) Program of Biosym Technologies in the laboratory of Dr. Robert Fillingame, De- partment of Biomolecular Chemistry, University of Wisconsin Medical School.

Heterotrimeric GTP-dependent proteins (G-proteins) mediate transduction of various signals from cell surface seven helical transmembrane receptors to their intracellular effector targets. In vertebrate photoreceptor cells, photoexited rhodopsin activates a visual transduction cascade via the ␣ subunit of the rod G-protein, transducin (␣t), 1 by catalyzing GDP-GTP exchange. The GTP-bound form of transducin, ␣t⅐GTP, dissociates from rhodopsin and the ␤␥ subunit of transducin and activates cGMP phosphodiesterase (PDE) by binding the inhibitory subunit, PDE␥ (P␥), thus releasing the catalytic activity of PDE␣␤ subunits (1)(2)(3). The activated PDE hydrolyzes cGMP, which is crucial in closing cation-specific channels, and the resulting hyperpolarized rod cell regulates neural signal transduction (4 -7). The activated PDE is turned off by activation of the GTPase activity of ␣ subunit of transducin, resulting in the hydrolysis of GTP to GDP. Unlike small G-proteins, such as Ras, which rely on the GTPase-activating protein (GAP) to accelerate the hydrolysis of GTP, heterotrimeric GTP-dependent proteins contain a "built-in" GAP-like domain in the ␣ subunits (8). It is clear, however, that some effectors involved in heterotrimeric G-protein pathways (such as rod cell cGMP phosphodiesterase and phospholipase C ␤ 1 ) accelerate the intrinsic GTPase activity of the interacting ␣ subunits (G␣t and G␣q, respectively) (9,10).
The interaction between the ␣t⅐GTP and PDE␥ subunits is crucial for PDE activation and ␣t⅐GTP hydrolysis. Recent data have indicated that the central region of P␥ (mainly residues P␥ 24 -45) is involved in the interaction between ␣t and P␥, whereas the COOH-terminal region (P␥ 46 -87) is involved in determining both the interaction with the P␣␤ catalytic subunits and the GAP activity of P␥ (11)(12)(13)(14)(15)(16). The COOH-terminal region of P␥ has been shown to bind to the ␣t 293-314 sequence (17). A study of the P␥ mutants, W70F and W70A, indicated that Trp-70 was crucial for P␥ binding to ␣t as well as the GTPase accelerating action of P␥ on ␣t (18,19). Recently, Hamm and colleagues have shown that the 11 COOH-terminal most residues of P␥ (76 -87) are intimately involved in the * This work was supported in part by National Institutes of Health Grant GM33138. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ Supported by National Institutes of Health Grant EY-10336 and by a Jules and Doris Stein Professorship from Research to Prevent Blindness Inc.

EXPERIMENTAL PROCEDURES
Chemical GTP, GTP␥S, soybean trypsin inhibitor, and N-ethylmaleimide (NEM) were products of Sigma; TPCK-treated trypsin and clostripain (sequencing grade) were purchased from Promega; Mono S column and blue Sepharose CL-6B were obtained from Pharmacia Biotech Inc.; BNPS-skatole was obtained from Pierce; hydroxylamine hydrochloride was purchased from Fisher; C4 reversed-phase HPLC column was obtained from Vydac Corp.

Preparation of ␣t⅐GTP␥S and ␣t⅐GDP
Holotransducin was prepared from frozen dark-adapted bovine retinas (from J. A. & W. L. Lawson Co.) by published procedures (21). The holotransducin was then loaded onto a blue Sepharose CL-6B column, and the ␣t⅐GDP and ␤␥t were purified according to Yamazaki et al. (22). The ␣t⅐GTP␥S subunit was released from the bovine rod outer segment membranes using GTP␥S and purified on blue Sepharose CL-6B, as described by the method of Kroll et al. (23). The purity of the ␣t and ␤␥t subunits was determined by SDS-polyacrylamide gel electrophoresis (PAGE). Staining with Coomassie Blue demonstrated that the subunits were more than 95% pure. The purified proteins were stored at Ϫ80°C and used within 1 month.

Preparation of P␥-[ 125 I]ACTP and Assessment of GTPase
Activating Activity Recombinant P␥, which was overexpressed in Escherichia coli strain BL21 DE3, was purified from a Mono S column and C4 reversed-phase HPLC column, as described previously (12,24). Briefly, cells suspended in 50 mM Tris⅐Cl, pH 7.6, containing protease inhibitors (phenylmethylsulfonyl fluoride, TPCK, leupeptin, aprotinin, and pepstatin A) were broken using a French press and centrifuged at 20,000 rpm (ϳ50,000 ϫ g) for 20 min. The supernatant was loaded onto a Mono S column and eluted with 50 mM Tris⅐Cl to 1 M NaCl gradient (ϳ40% NaCl) using an fast protein liquid chromatography system. The P␥-enriched fractions were loaded onto a C4 reversed-phase HPLC column and eluted with a 30 -60% acetonitrile gradient with 0.1% trifluoroacetic acid throughout over a 40-min period (P␥ eluted at approximately 40% acetonitrile). The fractions containing P␥ were pooled and lyophilized with a speed vacuum system and stored at Ϫ80°C for further use.
Derivatization of P␥ was carried out by incubating P␥ with 5 Ci/mmol [ 125 I]ACTP (1:2 molar ratio) in a solution of 120 mM NaCl, 10 mM HEPES, 6 mM MgCl 2 , pH 7.6, at room temperature overnight. The reaction mixture was chromatographed through a C4 reversed-phase HPLC column and eluted with an acetonitrile/H 2 O gradient in 0.1% trifluoroacetic acid (30 -60% acetonitrile over 40 min with a flow rate of 0.5 ml/min). The [ 125 I]ACTP and underivatized P␥ eluted at approximately 35 and 41% acetonitrile, respectively, whereas P␥-ACTP eluted at approximately 45% acetonitrile. The purity of the elution fractions was determined by 15% SDS-PAGE and autoradiography of the wet gel. The P␥-[ 125 I]ACTP was estimated to be greater than 99% radiopure. On occasion, a repeat of this HPLC procedure was necessary. The P␥-[ 125 I]ACTP fractions were collected, lyophilized, stored at Ϫ20°C, and used within 1 week.
Measurement of single-turnover GTPase activity of transducin was performed as described (25). Briefly, photoreceptor membranes free of endogenous PDE but retaining endogenous transducin (20 M rhodopsin) were preincubated with P␥ or P␥-[ 125 I]ACTP (0.3 M), and the reaction was started by adding 0.2 M [␥-32 P]GTP and terminated by addition of perchloric acid. Inorganic phosphate was measured by liquid scintillation. The rates of the single-turnover GTPase reaction were calculated from single exponential fits of the data.

Interaction of ␣t and [ 125 I]ACTP-derivatized P␥ and Photolysis Condition
Purified ␣t⅐GTP␥S and P␥-[ 125 I]ACTP (1:1 molar ratio), at a 5-10 M concentration, were mixed in a solution of 120 mM NaCl, 6 mM MgCl 2 , and 10 mM HEPES, pH 7.6, and incubated at room temperature for 30 min. The mixture was then photolyzed for 6 s in ice water at a distance of 10 cm from a water-jacketed AH-6 1-kilowatt high pressure mercury lamp (20). After addition of 4 mM NEM and 2% SDS, samples were electrophoresed on 12% SDS-PAGE for analysis or purification.

Preparation of 125 I Label-transferred Peptides from ␣t
The cross-linked ␣t-[ 125 I]ACTP-P␥ band was excised from the gel and eluted using a Bio-Rad 422 electric elutor. The cross-linked dimer was cleaved to free P␥ and 125 I-␣t with 2% 2-mercaptoethanol (␤-Me) by reaction at room temperature for 10 min, and the two proteins were separated by 12% PAGE. The 125 I-labeled ␣t was eluted from the gel slice with water by incubation at 4°C overnight. The eluted ␣t was desalted by passing the sample (0.2 ml) through a 3-ml Sephadex G50 column, washing with 0.8 ml of water, and eluting with 1 ml of water.
BNPS-skatole Cleavage-The 125 I-labeled ␣ subunit (10 -50 g) was dissolved in 10 l of water, 30 l of 1.3 mg/ml BNPS-skatole acetic acid solution was added, and the solution was incubated at 47°C for 2 h (26,27). The 15-kDa BNPS-skatole fragment was purified by SDS-PAGE, eluted from the gel, and desalted on Sephadex G25 by the same procedure as described above.
Trypsin and Clostripain Cleavage-Trypsin and clostripain cleavages of native ␣t⅐GTP␥S were performed according to published procedures (31,32).
Following electrophoresis, the proteins or peptides were transferred to PVDF (polyvinylidene difluoride) (Bio-Rad, 0.2 m) in a buffer of 10 mM CAPS, 20% methanol, pH 11, at 5 mA for 60 min using a Hoefer trans-blot apparatus.

Analytical Methods
Automatic amino-terminal protein sequencing was performed using an ABI, model 470A, gas-phase sequencer equipped with an on-line model 120A phenylthiohydantoin analyzer. Microsequencing and radiosequencing were performed at the Supramolecular Structure Facility, Michigan State University. SDS-PAGE (10, 12, and 15%) was performed by the method of Laemmli (33). Tricine 16.5% polyacrylamide electrophoresis was performed, as described by Schagger and von Jagow (34). Protein concentration was determined by the Coomassie Blue binding method (35) using bovine serum albumin as a standard or spectrophotometrically at 280 nm using a molar extinction coefficient of 7100 for P␥ (24). Densitometric scans of the Coomassie Blue-stained gels and autoradiography were performed on a model SL-504XL Zeineh Soft Laser Scanning Densitometer. Computer modeling was performed on Unix Silicon Graphics using the Insight II (version 2.3.0) Program of Biosym Technologies in the laboratory of Dr. Robert Fillingame, Department of Biomolecular Chemistry, University of Wisconsin Medical School.

Specific Interaction between P␥-[ 125 I]ACTP and ␣t⅐GTP␥S-
The ␥ subunit of PDE contains a single cysteine residue, Cys-68. This provides an opportunity to introduce a photoactivatable reversible probe into P␥ in order to determine the interaction between ␣t and the COOH-terminal region of P␥. An additional feature of the Cys-68 position is that it is two residues away from the crucial Trp-70, which has been shown to be involved in P␥ binding to ␣t and for the GAP activity of P␥ (18,19,24). Incubating [ 125 I]ACTP with P␥ at room temperature overnight in the dark in a 1:2 molar ratio (P␥/[ 125 I]ACTP) produced approximately a 50% derivatization of P␥ (data not shown). A C4 reversed-phase HPLC column was used successfully to separate P␥-[ 125 I]ACTP from free [ 125 I]ACTP and P␥ with a final yield of ϳ40%. The purity of P␥-[ 125 I]ACTP, as determined by Coomassie Blue staining and wet gel autoradiography of an SDS-polyacrylamide gel, was found to be more than 99%. A second HPLC was often performed to be certain that there was no contamination from free [ 125 I]ACTP. Analysis by SDS-PAGE of fractions from the HPLC column is shown in Fig. 1a. Autoradiography demonstrates the high purity of the P␥-[ 125 I]ACTP (fractions 3-5) and clear separation from contaminants (fractions 7-10). Following purification of P␥-[ 125 I]ACTP, the ability of the derivatized protein to accelerate ␣t⅐GTP hydrolysis was compared with nonderivatized P␥. The transducin GTPase assay showed that derivatized P␥ has essentially the same ability to accelerate transducin GTPase as nonderivatized P␥. The GTPase assay showed that P␥-[ 125 I]ACTP (ϳ0.372 s Ϫ1 ) had essentially identical activity to P␥ (ϳ0.313 s Ϫ1 ) (Fig. 1b). These data are consistent with previous observations that substitution of Cys-68 on P␥ with alanine or derivatization with a fluorescent probe did not alter the interaction of P␥ with ␣t (17,19).
Based on the previous determination of the stoichiometry of native ␣t⅐GTP and P␥ (36), the ␣t⅐GTP␥S was mixed with P␥-[ 125 I]ACTP at a 1:1 molar ratio. Following photolysis, the sample was treated with 4 mM NEM (to remove free sulfhydryls and prevent disulfide exchange) and 2% SDS (to terminate the reaction). When the reaction mixture was analyzed by 15% SDS-PAGE, a cross-link between ␣t and P␥ produced a band at 49 kDa that could be observed by both Coomassie Blue staining (Fig. 2, lane 2) and autoradiography (Fig. 2, lane 6). This cross-link was not observed in the dark control (Fig. 2, lanes 1  and 5). Light-dependent formation of the cross-linked 49-kDa heterodimer was completely protectable by preincubating ␣t⅐GTP␥S with excess C68A P␥, which has been shown to retain binding and functional GAP activity (Fig. 2, lanes 4 and  8) (19). The use of C68A P␥ as protector (which does not contain cysteine) eliminated the complication of disulfide bond exchange from P␥-[ 125 I]ACTP to C68A P␥. In some experiments, residual ␤␥ subunits of transducin contaminants of the ␣t⅐GTP␥S provided a useful internal control for the specificity of the P␥-[ 125 I]ACTP binding. No cross-linkage between ␤␥ subunits of transducin and P␥-[ 125 I]ACTP was observed (data not shown). Denaturation of ␣t with SDS also completely inhibited the formation of a specific cross-link between ␣t and P␥ (data not shown), indicating a requirement for native protein structure. Taken together, these data indicate that the crosslinkage between ␣t and P␥ is specific. In order to identify the photoinsertion site(s) on ␣t⅐GTP␥S, the cross-linked 49-kDa heterodimer was purified from an SDS gel and treated with ␤-Me in order to cleave the disulfide bond connecting the two proteins, and the radiolabeled ␣t was purified on a 12% SDS gel, as described under "Experimental Procedures." Cleavage of Radioiodinated ␣t with BNPS-skatole-The ␣ subunit of transducin contains two tryptophans at positions 127 and 207. Our laboratory has shown previously (37) that the molecule can be separated into three relatively large fragments of 15 kDa (contains most of the GTPase domain), 14 kDa (contains most of the helical domain), and 9 kDa (contains the central portion of the molecule) (see scheme in Fig. 3). These fragments and the partially cleaved products can be readily separated on SDS gels. The first approach, therefore, to determine the location of the label which was transferred from P␥ to ␣t was to perform a BNPS-skatole cleavage of the label-transferred ␣t and determination of the fragment that had been labeled from P␥. As expected, the photolabeled ␣t⅐GTP␥S, when cleaved with BNPS-skatole, produced three completely cleaved fragments of 15, 14, and 9 kDa and two incompletely cleaved fragments of 24 and 25 kDa following the procedure of Vaillancourt et al. (37). Of the completely cleaved fragments, only the 15-kDa fragment contained radioactivity, as observed by autoradiography (Fig. 4). This 15-kDa fragment encompasses residues 208 -350, which comprises the majority of the GTPase domain. Further cleavage of the 125 I-labeled purified 15-kDa BNPSskatole fragment (as described under "Experimental Procedures") with clostripain produced several cleavage products, including a 12-kDa fragment that could be identified by Coomassie Blue staining (data not shown) and autoradiography (Fig. 5a). Microsequencing (through 10 residues) of the fragment following transfer to PVDF revealed that the NH 2 terminus of this polypeptide was methionine 239 (see Fig. 5, bottom) and, based on apparent size, probably extended to the COOH terminus of ␣t. Radiosequencing showed the release of radioactivity at the 6th cycle, which corresponded to His-244 (Fig.  5b, top). The chemical sequence is an exact match with the sequence through the ␣ 3 helix (38) and a portion of ␤ 4 . These data indicate that one of the regions of ␣t interaction with the Cys-68 region of P␥ is (within the distance of the photoprobe) close to the histidine 244 side chain of the ␣ 3 helix. Additional cleavage strategies were pursued to investigate whether multiple reaction sites on ␣t could be determined. The presence of more than one insertion site would be useful to obtain a threedimensional reconstruction of the interaction of the Cys-68 region of P␥ with ␣t.
Cleavage of Radioiodinated ␣t with Hydroxylamine-Since there is a single Asn-Gly bond in ␣t (Asn-287-Gly-288), the radioiodinated ␣t was bisected into two fragments with hydroxylamine. As expected, two major cleavage fragments, a 33-kDa peptide, which contained the majority of the ␣t sequence, including the main portion of the BNPS-skatole cleavable GTPase domain, and a 6-kDa fragment, which contained the entire ␣ 4 helix, the ␤ 6 strand, and the COOH-terminal ␣ 5 helix were obtained (see scheme in Fig. 3 and Fig. 6a, left). A minor fragment at approximately 18 kDa was also observed as a result of alternate cleavage at an asparagine-serine linkage. Both major fragments were found to be radioactive, as seen in the autoradiogram (Fig. 6a, right), indicating that the photoprobe inserted into more than one site on ␣t in the GTPase domain. Since the insertion site on the 33-kDa fragment had been identified previously as His-244, as described in Fig. 5, these data indicated that the additional site(s) of insertion were in the ␣ 4 /␤ 6 /␣ 5 portion of the ␣ subunit. NH 2 -terminal microsequencing of the smaller 6-kDa fragment (Fig. 6) confirmed that the hydroxylamine cleavage had indeed occurred at Asn-287, and based on apparent size, this fragment was likely to extend to the COOH terminus of the molecule. The NH 2 -terminal sequence that was obtained for the 6-kDa fragment is shown through the five residues at the bottom of Fig. 6. This sequence precisely matched the sequence of the ␣ 4 helix/loop leading into the ␤ 6 strand. Radiosequencing of this smaller fragment showed 125 I release at the 21st and 23rd cycle, which corresponded to Met-308 and Arg-310 (Fig. 6b). Further clostripain and trypsin cleavages of label-transferred radioiodi- FIG. 3. Summary of the cleavage schemes for ␣t. Since ␣t contains two tryptophan residues at positions 127 and 207, the cleavage of ␣t with BNPS-skatole yielded three complete peptides with apparent molecular masses of 14, 9, and 15 kDa. Only the 15-kDa peptide, sequence 208 to COOH terminus, was radioiodine-labeled (Fig. 4). This 15-kDa peptide was further cleaved with clostripain to produce three peptides with apparent molecular masses of 12, 10, and 6 kDa. Microsequencing of the 12-kDa peptide indicated that this peptide started with its NH 2 terminus at Met-239 and likely extended to the COOH-terminal end. Radiosequencing of the peptide showed radioiodine release from His-244. Since the ␣ subunit of transducin contains one Asn-Gly bond at position 287-288, cleavage of ␣t with hydroxylamine yielded two major fragments with peptide molecular masses of approximately 33 and 6 kDa. Both peptides were photolabeled. Chemical and radioactive sequencing revealed that the photoinsertion on the 6-kDa peptide, from Gly-288 to COOH terminus, was at Met-308 and Arg-310. The details are described in the text. The black lines and letters in the figure represent the peptides or the residues which were 125 I-labeled, while the dotted gray line and gray letters indicate unlabeled fragments. nated native ␣t⅐GTP␥S showed no radioactivity in the 5-kDa peptide, which comprises the COOH-terminal fragment from residues 311-350 (data not shown) (31,32). This result confirmed that the radioactivity in the 6-kDa hydroxylamine fragment was confined to the region of ␣t between residues 287 and 310.
Thus, we have defined a region of P␥ Cys 68 interaction with ␣t⅐GTP␥S which is between the exposed face of the ␣ 4 and ␣ 3 helices, as determined by computer modeling (Fig. 7). The distance from P␥ Cys-68 to His-244, Met-308, and Arg-310 of ␣t is approximately 9 -12 Å, which is estimated to be the distance from the azido moiety to the disulfide bond. Identification of three insertion sites on ␣t provided the opportunity to utilize molecular modeling and triangulation to locate the Cys-68 -Trp-70 sequence of P␥ on the ␣t structure using the crystal structure coordinates of the ␣ subunit of transducin (38). DISCUSSION The interaction between ␣t and P␥ is important for both cGMP PDE activation and ␣t⅐GTP hydrolysis. Several investigators have shown that ␣t⅐GTP␥S is physically bound to P␥, thus activating PDE by removing the inhibitory constraint of P␥ subunit (39 -43). P␥ has been shown to have binding sites for both ␣t and PDE␣␤ (15,17,44,45). Mutational experiments have revealed that there are two major functional regions in P␥; that is, a central polycationic region encompassing residues 24 -45 containing a binding site for ␣t and PDE␣␤ and a COOH-terminal region, 46 -87, which contains both the inhibitory region of P␥ that binds to P␣␤ and the functional region that accelerates the GTPase activity of ␣ subunit of transducin (11-15, 24, 44, 46). Among the P␥ 24 -45 residues, Lys-41, Lys-44, Lys-45, Arg-24, and Arg-33, as well as other hydrophilic amino acids, appear to be involved in binding to ␣t (15,47). Considerable evidence has been obtained supporting the thesis that the COOH-terminal region of P␥ also contains binding sites for ␣t, especially within residues 53-76 (15,16). The COOH-terminal 11 residues of P␥, 77-87, are crucial for interaction with P␣␤ or GTPase accelerating activity but less important for ␣t binding (16,24).
Strategies to investigate protein-protein interactions involving transfer of a radiolabeled moiety from a "donor" protein (in this case P␥) to a "receiver" protein (in this case ␣t⅐GTP␥S) has been used successfully in our laboratory (20,37) ]␣t⅐GTP␥S was separated on a 12% PAGE gel, eluted with water, and cleaved with BNPS-skatole, as described under "Experimental Procedures." The 15-kDa BNPS-skatole cleavage fragment (208 -350 sequence of ␣t) was further eluted from the gel with water and then cleaved with clostripain, as described under "Experimental Procedures." a, autoradiogram of PVDF-transferred clostripain cleavage residues from the 15-kDa BNPS-skatole fragment. A major radiolabeled polypeptide from clostripain cleavage (12-kDa band) was obtained. b, chemical and radiochemical sequencing of the 12-kDa polypeptide. The 12-kDa peptide was chemically sequenced and shown to contain NH 2terminal Met-239 and would likely extend to the COOH-terminal end of the molecule. The first 10 letters represent the actual sequence obtained from NH 2 -terminal microsequencing. Radiosequencing indicated radioiodine release on the 6th cycle, which corresponds to His-244.
FIG. 6. Identification of radiolabeled photoinsertion sites on ␣t using hydroxylamine cleavage. a, PVDF-transferred hydroxylamine cleaved peptides from radiolabeled ␣t. On the left is the Coomassie Blue staining and on the right is the resulting autoradiogram. Hydroxylamine cleavage of ␣t⅐GTP␥S produced two major bands (33 and 6 kDa). Both polypeptides were radioactive. b, chemical and radiochemical sequencing of the 6-kDa peptide. NH 2 -terminal microsequencing of the smaller 6-kDa fragment confirmed that the peptide started at glycine and likely extended to the COOH-terminal end. The first five letters shown in bold represent the actual microsequencing result. Radiosequencing indicated radioiodine release at the 21st and 23rd cycle, which corresponds to Met-308 and Arg-310, respectively. "label transfer" from the donor protein to the receiver protein.
As discussed previously, it was possible to use the entire P␥ subunit in these experiments since only one cysteine exists (Cys-68) in the molecule. Another important requirement of the [ 125 I]ACTP-derivatized P␥ is that this molecule must be functionally active. This property is clearly shown in Fig. 1b, with comparison with native P␥. The ACTP probe was further useful in these experiments, since this reagent provided multiple insertion sites at His-244, Met-308, and Arg-310, thus demonstrating that the Cys-68 sulfhydryl is located in a bridging region between the ␣ 4 /␤ 6 and the ␣ 3 region of ␣t.
The specificity of interaction between the derivatized P␥ and ␣t was demonstrated by the protection of covalent cross-linking by the C68A mutant of P␥ (Fig. 2), as well as the need for native ␣t⅐GTP␥S. The ␣t⅐GDP under these same experimental conditions was able to interact weakly with P␥, producing approximately 25% of the light-dependent cross-link that was observed with the ␣t⅐GTP␥S (data not shown). This is in agreement with the observation of Hamm and colleagues for a photoprobederivatized 24 -45 fragment of P␥ and its interaction with ␣t⅐GDP (48). Taken together, these data are consistent with that of Kutuzov and Pfister (49), which indicated an interaction between the GDP-bound form of ␣t and the P␥ effector subunit.
The data from experiments shown in Figs. 5 and 6 define the location of the transferred photoprobe from P␥ to ␣t⅐GTP␥S. NH 2 -terminal microsequencing and radiosequencing identified the location of insertion into the 15-kDa BNPS-skatole fragment, which yields a smaller clostripain fragment of 12 kDa at His-244 in the middle of the ␣ 3 helix (see Fig. 5b). The hydroxylamine cleavage result, as shown in Fig. 6 and illustrated in the scheme in Fig. 3, demonstrated that photoprobe insertion occurred in ␣t on both "sides" of the single Asn-Gly bond at Asn-287 and Gly-288 at the entry into the ␣ 4 helix, since both the 33-kDa and the 6-kDa fragments were radiolabeled (see Fig. 6a). Labeling of the 33-kDa segment was consistent with location of label in the 15-kDa BNPS-skatole fragment and the 12-kDa clostripain fragment (Fig. 5a). Labeling of the 6-kDa hydroxylamine fragment (Fig. 6) was shown to be split between Met-308 and Arg-310, which further located specific labeling at the end of ␣ 4 helix and the loop leading to ␤ 6 on ␣t (Fig. 6b). Interestingly, the 306 -310 region, which encompasses the residues that we have identified, were also identified by Hamm and colleagues (48) using the central 24 -45 sequence of P␥ as a probe. A possible explanation of these results is summarized below with the use of molecular modeling (see Figs. 7 and 8). We proposed that when the COOH-terminal region of P␥ binds to ␣t, Cys-68 is placed within the exposed face of ␣t between the ␣ 4 helix and the ␣ 3 helix in such a way as to allow a 9 -12-Å radius of interaction of the ACTP probe, which reacts with His-244, Met-308, or Arg-310. This was strongly supported by the recent data from Erickson et al. (50). Using a resonance transfer method, Erickson et al. (50) found the distance be- FIG. 8. Speculation concerning the interaction between ␣t⅐GTP␥S and the central and COOH-terminal region of P␥. The ribbon drawing depicting the ␣ 2 , ␣ 3 , and ␣ 4 helices, as derived from the crystal structure of ␣t⅐GTP␥S (38), is shown in purple. The secondary structure of the P␥ central region (24 -45) and COOH-terminal region (63-87), as predicted by the Chou and Fasman method (53), is shown in green. We speculate that the central region of P␥ may possibly be an ␣ helix, which binds to the ␣ 3 /␤ 5 , ␣ 4 /␤ 6 , and possibly the ␣ 2 /␤ 4 loops. The COOH-terminal region of P␥ has Cys-68 situated in an exposed face between the ␣ 3 and ␣ 4 helix, and the predicted ␣ helix of P␥ 72-82 interacts with the ␣ 4 helix of ␣t⅐GTP␥S. The P␥ 64 -68 may interact with side chain from the ␣ 3 helix and possibly the Switch III region of ␣t⅐GTP␥S. Side chains are colored as follows: red, His-244, Met-308, and Arg-310 of ␣t⅐GTP␥S are photo-insertion sites; pink, Asn-297, Val-301, and Glu-305 are residues on ␣t⅐GTP␥S identified to be involved in the interaction with P␥ by Spickofsky et al. (54); yellow, Cys-210 and Cys-250 of ␣t⅐GTP␥S and Cys-68 of P␥; white, Trp-70 of P␥; green, Pro-69 of P␥. White indicates the residues from ␣t, whereas yellow indicates the residues from P␥. White indicates the residues from ␣t, whereas yellow indicates the residues from P␥. The estimated distances from Cys-68 of P␥ to His-244, Met-308, and Arg-310 assume a 9 -12-Å probe length.
tween Lys-267 (␣G helix) of the GTP form of ␣t and Cys-68 of P␥ to be approximately 45 Å. Based on the crystal structure, the distance between Lys-267 of ␣t and the 106 -116 region of ␣t (the interaction region proposed by Erickson et al. (50) for Cys-68 of P␥) is approximately 30 -39 Å. Similarly, the distance between Lys-267 of ␣t and the interaction site that we have identified for the Cys-68 of P␥ on ␣t, which is located between ␣ 3 and ␣ 4 (see Figs. 8 and 9), is approximately 29 -36 Å. Therefore the position of P␥ Cys-68 on ␣t according to our results is completely consistent with the measurements made by Erickson et al. (50).
Using synthetic peptides as probes, Rarick et al. (51) found that a 22-amino acid peptide, corresponding to residues 293-314 of ␣t, mediated PDE activation. This 22-amino acid peptide bound to the COOH-terminal region of P␥ (P␥ 46 -87), but not the central region, since P␥ 24 -45 had no effect on P␥ and ␣t 293-314 interaction (17). Solid-phase binding assays have also identified an additional region of ␣t (residues 250 -275), which showed a high affinity for P␥ and P␥ 24 -45 and the capability to compete with ␣t for binding to PDE␥. More importantly, this peptide activated a fully inhibited P␣␤␥2 complex (52), although it was not clear if all 25 amino acids were required for the binding and functional activity. The 293-314 sequence of ␣t consists of the ␣ 4 helix and the ␣ 4 /␤ 6 loop, and residues 250 -275 of ␣t extend from ␣ 3 , the ␣ 3 /␤ 5 loop, to ␤ 5 , as judged from the crystal structure of ␣t⅐GTP␥S (38). The ␣ 3 /␤ 5 loop and ␣ 4 /␤ 6 loop are very close to each other structurally. The secondary structure of P␥ 24 -45, as predicted by Chou and Fasman analysis (11,53), may contain an ␣ helix beginning from residues 29 to 42, with a ␤-turn or loop at COOH-terminal residues 42 and 43. It is possible that the P␥ 24 -45 central sequence binds to the ␣ 4 /␤ 6 and ␣ 3 /␤ 5 loop with the P␥ 45 in close proximity to the ␣ 4 /␤ 6 loop of ␣t and the P␥ 28 -42 sequence binding to the ␣ 3 /␤ 5 loop of ␣t (see Fig. 8). It is generally agreed that there are at least two critical binding sites between ␣t and P␥ (17,48).
A particularly relevant series of experiments reported by Spickofsky et al. (54) has demonstrated that five residues on one exposed face of the ␣ 4 helix of rod transducin interact with P␥, but not the ␣ 4 /␤ 6 loop, as suggested by Artemyev et al. (48). Our data identify the binding region for Cys-68 of P␥ to be between the ␣ 3 and ␣ 4 helix of ␣t⅐GTP␥S. These data, taken together, indicate that the ␣ 4 helix and possibly part of the ␣ 4 /␤ 6 loop of ␣t interact with the COOH-terminal region of P␥. Based on the Chou-Fasman analysis, the secondary structure of P␥ 69 -82 is likely to be an ␣ helix, with a turn occurring at Pro-69. This is also supported by the alanine substitution experiment of Slepak et al. (19). These investigators found that replacement of hydrophobic residues with alanine at a two or three residue frequency decreased the ability of P␥ to accelerate transducin GTPase.
Closely relevant data have also been found by Arkinstall et al. (55), using multiple overlapping synthetic peptides mapping regions of G␣q interaction with phospholipase C ␤ 1 . It was shown that the ␣ 3 -␣ 3 /␤ 5 and the ␣ 4 -␣ 4 /␤ 6 regions of G␣q are involved in binding to phospholipase C ␤ 1 . In addition, the recent work by Skiba et al. (56) also implicated that the ␣ 3 and ␣ 4 helices of ␣t are directly involved in binding to P␥. Our data are consistent with the results discussed above and further indicate that it is very likely that the COOH-terminal region of P␥ binds to ␣ 3 and the ␣ 4 helices of ␣t.
The use of molecular modeling based on data reported in this paper and by other investigators provides a useful platform upon which to consider the binding interactions between P␥ and the GTP-bound form of ␣t. These speculations are summarized in Figs. 7 and 8. Taken together, we propose one possible interaction is that the central region of P␥ (residues 24 -45 of P␥) interacts with the ␣ 3 /␤ 4 , ␣ 4 /␤ 6 , and possibly ␣ 2 /␤ 3 loops of ␣t, whereas residues 69 -77 of P␥ interact with the ␣ 4 helix, ␣ 4 /␤ 6 loop and that P␥ residues 63-68 may interact with the ␣ 3 helix (Fig. 8).
Previous suggestions have been made that the interaction mechanism of ␣t with P␥ occurs with the initial binding of ␣t 293-314 at the COOH-terminal of P␥, resulting in a lowered binding affinity at a second site, ␣t 250 -275, with P␥ 24 -45 (17,52). Based on our experiments, an alternative mechanism that is appealing is that when the binding of the central basic region (24 -45) of P␥ to ␣t (␣t ␣ 4 /␤ 6 , ␣ 3 /␤ 5 , and possibly ␣ 2 /␤ 3 ) occurs, the interaction of the COOH-terminal of P␥ to ␣t is reduced, which was supported by the recent study from Hamm and colleagues (56). Upon receptor activation of GTP binding to ␣t and dissociation of the ␤␥ subunits of transducin, the COOH terminus of P␥ binds tightly to the ␣ 3 /␣ 4 helix regions of ␣t and accelerates GTP hydrolysis by a mechanism that is yet to be determined.