Communication between Switch II and Switch III of the Transducin α Subunit Is Essential for Target Activation*

Comparisons of the tertiary structures of the GDP-bound and guanosine 5′-O-(thiotriphosphate) (GTPγS)-bound forms of the α subunit of transducin (αT) indicate that there are three regions that undergo changes in conformation upon αT activation. Two of these regions, Switch I and Switch II, were originally identified in Ras, while Switch III appears to be unique to trimeric GTP-binding proteins (G proteins). We find that replacement of the Switch III region (aspartic acid 227 through asparagine 237) with a single alanine residue yields an αT subunit that fully binds and hydrolyzes GTP but no longer stimulates the activity of the cyclic GMP phosphodiesterase (PDE), the physiological target for transducin. We also show that changing glutamic acid 232 of αT to a leucine (E232L) had no effect on rhodopsin-stimulated GTP-GDP exchange nor on the GTP hydrolytic activity of αT. However, the GTPγS-bound form of the αTE232L mutant was unable to stimulate the activity of the cyclic GMP PDE. The lack of stimulation was not due to an inability of the αTE232L mutant to bind to the target. Taken together, these results indicate that glutamic acid 232 mediates a conformational coupling between Switch II and Switch III, which is essential for converting GTP-dependent G protein-target interactions into a stimulation of target/effector activity.


Comparisons of the tertiary structures of the GDPbound and guanosine 5-O-(thiotriphosphate) (GTP␥S)bound forms of the ␣ subunit of transducin (␣ T ) indicate
that there are three regions that undergo changes in conformation upon ␣ T activation. Two of these regions, Switch I and Switch II, were originally identified in Ras, while Switch III appears to be unique to trimeric GTPbinding proteins (G proteins). We find that replacement of the Switch III region (aspartic acid 227 through asparagine 237) with a single alanine residue yields an ␣ T subunit that fully binds and hydrolyzes GTP but no longer stimulates the activity of the cyclic GMP phosphodiesterase (PDE), the physiological target for transducin. We also show that changing glutamic acid 232 of ␣ T to a leucine (E232L) had no effect on rhodopsin-stimulated GTP-GDP exchange nor on the GTP hydrolytic activity of ␣ T . However, the GTP␥S-bound form of the ␣ T E232L mutant was unable to stimulate the activity of the cyclic GMP PDE. The lack of stimulation was not due to an inability of the ␣ T E232L mutant to bind to the target. Taken together, these results indicate that glutamic acid 232 mediates a conformational coupling between Switch II and Switch III, which is essential for converting GTP-dependent G protein-target interactions into a stimulation of target/effector activity.
GTP-binding proteins (G proteins) serve as molecular switches in a wide variety of biological response systems. Two large families, the Ras-related small G proteins and the trimeric G proteins, have received a great deal of attention because of their central role in signal transduction. The trimeric G proteins serve as intermediate signal transducers for seventransmembrane-spanning (also called heptahelical or serpentine) receptors that are involved in responses to sensory, hormonal and neurotransmitter signals (1,2). These G proteins consist of two functional units, the guanine nucleotide-binding ␣ subunit (G␣) and the ␤␥ complex (G␤␥). The molecular switch capability of a trimeric G protein is mediated through the G␣ subunit, which cycles between the GDP-bound (inactive) and GTP-bound (active) states. A signal received from a receptor promotes the activation of a trimeric G protein by catalyzing the exchange of GTP for GDP on G␣. This results in the dissociation of the GTP-bound G␣ from the G␤␥ complex, thereby enabling these subunits to regulate the activities of downstream target/effectors. GTP hydrolysis on the G␣ subunit promotes its re-association with G␤␥, thus terminating the signal. The vertebrate phototransduction system represents one of the best characterized G protein-coupled signaling cascades. In this system, the photoreceptor rhodopsin activates a trimeric G protein, transducin, generating a GTP-bound ␣ subunit (␣ T -GTP), which stimulates the target/effector enzyme, the cyclic GMP phosphodiesterase (PDE). 1 The three-dimensional structures for ␣ T in different guanine nucleotide-bound states, and more recently for the ␣ T -␤␥ T holotransducin complex, have been solved by x-ray crystallography (3)(4)(5)(6), as have the corresponding structures for the inhibitory GTP-binding protein of the adenylyl cyclase system, G i1 (7)(8)(9). This structural information now provides the foundation for understanding the molecular basis of many aspects of G protein-mediated signaling.
Three distinct regions on trimeric G protein ␣ subunits have been shown to undergo conformational changes in response to GTP/GDP exchange (4,5). Two of these regions, designated Switch I (Ser 173 to Thr 183 in ␣ T ) and Switch II (Phe 195 to Thr 215 ), are structurally analogous to the two conformationally-sensitive regions found in Ras (10) and EF-Tu (11), whereas the third region, designated Switch III (Arg 227 to Arg 238 in ␣ T ), is unique to the ␣ subunits of trimeric G proteins. The various available x-ray crystallographic structures of G proteins show that the conformational changes in Switch I and II are the direct result of GTP binding to residues within these regions. Specifically, the structural changes in Switch I are induced by the interaction of the ␥-phosphate of GTP with Thr 177 , while the changes in Switch II result from a hydrogen bond between Gly 199 and the ␥-phosphate (4). However, Switch III does not directly contact GTP. Rather, it was shown to respond to Switch II through a series of polar interactions that were mediated and/or promoted by GTP-induced conformational changes in Switch II (4). At present, the functional role of the Switch III domain or the importance of its conformational coupling to Switch II is not known. The fact that the residues proposed to be responsible for this conformational coupling are conserved in all trimeric G protein ␣ subunits (4, 7) supports a critical role for Switch II-Switch III communication in some event associated with G protein activation, such as the GTPmediated dissociation of rhodopsin and/or ␤␥ T from ␣ T or the GTP-dependent interaction of ␣ T with the cyclic GMP PDE. In the present work, we have examined the importance of Switch III in G protein function and find that a conserved glutamic acid residue within the Switch III domain of ␣ T is essential for the regulation of target/effector activity.

EXPERIMENTAL PROCEDURES
Purification of Retinal Proteins-Rod outer segments (ROS) were prepared as described previously (12,13). Transducin and the cyclic GMP PDE were purified by exposing ROS to room light and repeated washings with 10 mM Hepes, pH 7.5, 6 mM MgCl 2 , 1 mM dithiothreitol (DTT), and 100 mM NaCl (isotonic buffer) and then with 10 mM Hepes, pH 7.5, 6 mM MgCl 2 , and 1 mM DTT (hypotonic buffer). The cyclic GMP * This work was supported by National Institutes of Health Grant EY06429. 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.
‡ To whom correspondence should be addressed. Tel.: 607-253-3888; Fax: 607-253-3659. PDE is released into the hypotonic wash and is further purified by hydroxyapatite chromatography (13). Transducin is released from ROS by washing with hypotonic buffer in the presence of 0.1 mM GTP or 0.1 mM GTP␥S. The ␣ T and ␤␥ T subunit complexes are then resolved by Blue Sepharose chromatography (14).
Measurement of cGMP PDE Activity-Dark-adapted ROS membranes containing hydroxyapatite-purified cyclic GMP PDE were assayed for cyclic GMP hydrolysis by measuring proton release, as originally described by Yee and Liebman (15). The assays are performed at room temperature in a buffer containing 10 mM Hepes, pH 8.0, 60 mM KCl, 30 mM NaCl, 1 mM DTT, and 5 mM MgCl 2 , together with the protein components described in the legend to Fig. 3. The assays were initiated by the addition of cyclic GMP (5 mM), with the pH being recorded for 1-2 min at one determination/s. The PDE activity (nanomole/s) was calculated as the ratio of the slope of the pH change (millivolts) and the buffering capacity of the medium (millivolt/nmol) (16).
Expression of Recombinant ␣ T Subunits-The coding region of the bovine ␣ T was amplified by the polymerase chain reaction using primers that create a 5Ј-end NdeI site and a 3Ј-end BamHI site. The polymerase chain reaction product was digested with NdeI and BamHI and ligated into pET15b (Novagen). The resultant vector was digested with NcoI and PstI to release the coding region of ␣ T with its 5Ј-end fused in-frame to the hexa-His tag present in the vector pET15b. This released fragment was then blunt-ended using T4 DNA polymerase and ligated into pVL1393. To generate the ␣ T ⌬I 2 deletion mutant (Asp 227 through Asn 237 replaced by a single Ala), and the ␣ T E232L point mutant (Glu 232 replaced by Leu), we employed a single-stranded DNAbased mutagenesis strategy (17), using synthetic oligonucleotide primers containing the indicated deletion or point mutation (␣ T ⌬I 2 : 5Ј-CAGGCTCTCGTGCATTCGAGCGTAGGCGCTCAG-3Ј; and ␣ T E232L: 5Ј-CACTTCGTCCTCGAGCACCAGC-3Ј). The pVL1393 vector carrying either the wild type, ␣ T ⌬I 2 , or ␣ T E232L gene was introduced into Sf9 insect cells using the Baculogold transfection kit (PharMingen). The recombinant extracellular virus (rECV) was purified by a limiting dilution procedure (18). For production of the recombinant proteins, Sf9 insect cells were infected at 80% confluence with the purified rECVs at a multiplicity of infection of 5 and harvested typically 60 h post-infection. The His-tagged ␣ T proteins were purified through Ni 2ϩ -nitrilotriacetic acid affinity chromatography following a protocol provided by Qiagen. The purified proteins were finally dialyzed against HMDN buffer (20 mM Hepes, pH 7.4, 5 mM MgCl 2 , 150 mM NaCl, and 1 mM DTT) containing 40% of glycerol and stored at Ϫ20°C.

RESULTS AND DISCUSSION
The original finding that a third region on heterotrimeric G protein ␣ subunits undergoes structural changes upon GTP-GDP exchange (i.e. in addition to the Switch I and Switch II regions originally identified in Ras and EF-Tu (10, 11)) suggests that it may play a critical role in a GTP-dependent G protein function. To obtain experimental support for this suggestion, we examined the properties of an ␣ T deletion mutant in which the entire Switch III domain (residues Asp 227 through Asn 237 ) was replaced by a single alanine residue. The deletion mutant, designated ␣ T ⌬I 2 , was expressed in Spodoptera frugiperda (Sf9) cells as a hexahistidine (His)-tagged fusion protein and purified by Ni 2ϩ affinity chromatography. This results in a rapid and highly effective purification of the recombinant ␣ T subunit, as shown in Fig. 1. The first three lanes in A show the Coomassie Blue-stained profiles for the ␣ T subunit purified from bovine retina, the recombinant wild type His-tagged ␣ T purified from Sf9 cells (which has a slightly slower mobility on SDS gels because of the His-tag), and the His-tagged ␣ T ⌬I 2 mutant purified from Sf9 cells. B shows the corresponding Western blots that were obtained using a specific antibody raised against the carboxyl-terminal 10 amino acids of ␣ T (16).
We first examined whether the deletion of the Switch III domain from ␣ T affected rhodopsin-and ␤␥ T -promoted [ 35 S]GTP␥S/GDP exchange. Fig. 2A shows that as has been documented previously (19,20), when ␣ T purified from bovine retina was added to urea-stripped ROS containing light-activated rhodopsin, there was a marked increase in [ 35 S]GTP␥S binding that was strongly stimulated by the addition of puri-fied retinal ␤␥ T . Virtually identical results were obtained with the Sf9-expressed, His-tagged wild type ␣ T and the His-tagged ␣ T ⌬I 2 deletion mutant. Likewise, the ␣ T ⌬I 2 mutant was able to fully hydrolyze [␥-32 P]GTP (Fig. 2B). Taken together, the results presented in Fig. 2, A and B, indicated that the deletion of the Switch III domain did not impair the ability of ␣ T to interact with rhodopsin nor with the ␤␥ T subunit complex and that removal of Switch III did not interfere with the GTP-binding/ GTP hydrolytic cycle of the G protein.
We then examined the ability of the ␣ T ⌬I 2 mutant to functionally couple to the cyclic GMP PDE, by first loading the ␣ T ⌬I2 mutant with GTP␥S (by incubation with ROS and bovine retinal ␤␥ T ) and then assaying cyclic GMP PDE activity, by measuring the H ϩ release that accompanies cyclic GMP hydrolysis. The results presented in Fig. 3A illustrate that the bovine retinal ␣ T subunit and the recombinant wild type ␣ T were essentially equivalent in their abilities to stimulate cyclic GMP hydrolysis. However, the ␣ T ⌬I2 deletion mutant was unable to stimulate PDE activity (relative to the basal activity measured in the absence of added ␣ T ). Thus, these results suggested that the integrity of the Switch III domain was essential for ␣ Tmediated regulation of its target/effector enzyme.
An interesting possibility that was originally proposed following an examination of the x-ray crystallographic structure of the ␣ T ⅐GTP␥S complex (3) was that the acidic amino acid residues, Asp 233 , Asp 234 , and Glu 235 formed a potential binding site for a basic stretch of amino acids on the ␥ PDE subunit. If this were the case, it would then explain why the deletion of the Switch III domain yields an ␣ T subunit that is unable to stimulate effector activity. However, we have expressed and purified an ␣ T mutant from Sf9 cells in which the three acidic amino acids were replaced by alanine residues and found that this triple mutant was fully active, not only in its ability to bind and hydrolyze GTP, but also in its ability to stimulate cyclic GMP PDE activity (data not shown).
This then led us to examine another possibility, namely that a conserved glutamic acid residue in the Switch III region, Glu 232 , is responsible for mediating the conformational communication between the Switch II and Switch III domains (4). To test this, we generated a mutant of ␣ T in which a leucine residue was substituted for Glu 232 (␣ T E232L) and expressed it in S. frugiperda (Sf9) insect cells as a hexahistidine (His)tagged protein (see lane 4 in Fig. 1, A and B). Like the ␣ T ⌬I2 deletion mutant, we found that the ␣ T E232L mutant was able to functionally couple to rhodopsin and/or ␤␥ T , as read-out by its ability to undergo [ 35 S]GTP␥S/GDP exchange and GTP hydrolysis in a rhodopsin-and ␤␥ T -dependent manner (Fig. 2, A  and B). Moreover, just as was the case for the Switch III deletion mutant, the ␣ T E232L mutant was unable to stimulate target/effector (PDE) activity (Fig. 3A), even when using amounts of the mutant that were in 10-fold excess relative to the retinal or recombinant wild type ␣ T proteins (data not shown). Thus, the mutation of the single conserved Glu 232 residue appeared to fully mimick the effects obtained upon the removal of the entire Switch III domain.
We used the ␣ T E232L mutant to further examine the importance of Switch II domain-Switch III domain coupling in the stimulation of target/effector activity. We found that the inability of the ␣ T E232L mutant to stimulate PDE activity cannot be attributed to its inability to bind to its PDE target. This was determined through competition experiments. Fig. 3B shows that like the GDP-bound wild type ␣ T subunit (open bar in column 2), the GDP-bound form of the ␣ T E232L mutant (open bars in columns 3 and 4) did not competitively inhibit the PDE stimulatory activity of the GTP␥S-bound retinal ␣ T subunit (shown as the solid bar in column 1). This was as expected, since the GDP-bound form of ␣ T has only a weak affinity for the ␥ PDE subunit. However, the GTP␥S-bound form of ␣ T E232L showed a dose-dependent inhibition (hatched bars in columns 3 and 4 in Fig. 3B), thus indicating that the ␣ T E232L mutant can bind to ␥ PDE in a GTP␥S-dependent manner. The fact that the GTP␥S-bound wild type ␣ T subunit did not competitively inhibit the stimulatory activity of the retinal GTP␥S-bound ␣ T (column 2, hatched bar, in Fig. 3B) illustrates that the activated wild type ␣ T subunit can fully substitute for the activated retinal ␣ T . Moreover, these results demonstrate that the inhibitory effects are specific for the GTP␥S-bound ␣ T E232L mutant, such that the ␣ T E232L molecule can act as a dominantnegative mutant.
Thus, GTP␥S can both bind and induce the appropriate conformational changes within the ␣ T E232L mutant that enable it to specifically interact with the target/effector molecule. This is further indicated by the results of limited trypsin treatment (Fig. 4). It has been well documented that trypsin treatment of the retinal ␣ T subunit gives rise to defined proteolytic patterns that are absolutely dependent on the guanine nucleotide-bound state of ␣ T (21,22). Trypsin treatment of the GDP- bound wild type ␣ T yields two stable fragments, an ϳ23-kDa fragment (shown in lane 2 under ␣ T wt in Fig. 4) and an ϳ9-kDa fragment (not shown), whereas trypsin treatment of the GTP␥S-bound wild type ␣ T yields a stable 32-kDa (precursor) fragment (lane 3 under ␣ T wt in Fig. 4). Based on the information provided from the tertiary structures for the different nucleotide forms of ␣ T (4, 5), it is now clear that the protection afforded by GTP␥S directly reflects a GTP␥S-dependent conformational change that occurs within the Switch II domain and effectively moves the trypsin-sensitive Arg 204 residue from a solvent-exposed environment to a less accessible position (by virtue of its interaction with Glu 241 ). Thus, the protection against trypsin proteolysis afforded by GTP␥S serves as a highly sensitive read-out for GTP␥S-induced conformational changes within the Switch II domain and has frequently been used as a monitor for ␣ T activation (23). The results presented in Fig. 4 (lanes 2 and 3 under ␣ T E232L) show that GTP␥S binding to the ␣ T E232L mutant provides a similar protection against trypsin proteolysis, as observed with the wild type ␣ T subunit. Therefore, the mutation of Glu 232 neither perturbs GTP␥S binding nor the GTP␥S-induced conformational alteration of Switch II. However, mutation of Glu 232 , while preserving the GTP-dependent binding of the ␣ T subunit to the cyclic GMP PDE, completely uncouples this binding from target/effector stimulation.
The location of Glu 232 in the loop connecting the ␤ 4 strand and ␣ 3 helix of ␣ T places it in a prime position to couple conformational transitions between Switch II and Switch III. In particular, x-ray crystallographic analysis shows that upon GTP␥S binding, Glu 232 is engaged in direct interactions with Arg 201 and Arg 204 of Switch II and in a water-mediated interaction with Gly 199 of Switch II (5). Given our findings, we conclude that the conformational coupling between Switch II and Switch III is responsible for converting a primary binding interaction between activated ␣ T and its target (␥ PDE ), perhaps involving residues in Switch II (24) or in other regions of ␣ T (25)(26)(27), into a secondary stimulatory interaction between the ␥ PDE subunit and the ␣ 4 -␤ 6 residues 305-314 of ␣ T (28). Moreover, these results indicate that Glu 232 plays an essential role in mediating this conformational coupling, thereby translating ␣ T -target (PDE) interactions into a specific regulatory event. The fact that this glutamic acid residue is conserved in all trimeric G␣ subunits further suggests that it plays a fundamental role in converting target binding into target/effector regulation in a wide variety of G protein-coupled signaling pathways.