The Same Mutation in Gs and Transducin Reveals Behavioral

Home , Adenylyl cyclase, Signal transduction

The same mutation in Gs␣ and transducin ␣ reveals behavioral differences between these highly homologous G protein ␣-subunits

Adolfo R. Zurita and Lutz Birnbaumer†

Laboratory of Neurobiology, Division of Intramural Research, National Institute of Environmental Health Sciences, National Institutes of Health, Building 101, Room F180, 111 T. W. Alexander Drive, Research Triangle Park, NC 27709

Contributed by Lutz Birnbaumer, December 27, 2007 (sent for review December 17, 2007) Mutating Arg-238 to Glu (R238E) in the switch 3 region of a unit are largely unknown, and even less well, if at all, understood transducin ␣ (*T␣) in which 27 aa of the GTPase domain have been is how all agonist-occupied (activated) GPCRs promote the replaced with those of the ␣-subunit of the inhibitory G protein 1 same nucleotide exchange reaction at their cognate G protein (Gi1␣), was reported to create an ␣-subunit that is resistant to ␣-subunits. Hypotheses as to how nucleotide exchange comes activation by GTP␥S, is devoid of resident nucleotide, and has about have in common the assumption that similar rate-limiting dominant negative (DN) properties. In an attempt to create a DN steps are facilitated, and that while the receptor–G protein stimultory G protein ␣ (Gs␣) with a single mutation we created interaction is determined by their appropriately differing inter- Gs␣–R265E, equivalent to *T␣–R238E. Gs␣–R265E has facilitated action affinities, the kinetic steps that follow productive associ- activation by GTP␥S, a slightly facilitated activation by GTP but ation of receptor to the trimeric G protein are likely to be very much reduced receptor plus GTP stimulated activation, and an similar, if not the same. apparently unaltered ability to interact with receptor as seen in Transducin, originally called ‘‘light-activated GTPase,’’ was ligand binding studies. Further, the activity proﬁle of Gs␣–R265E is not only the first signal-transducing G protein to be recognized that of an ␣-subunit with unaltered or increased GTPase activity. to have an intrinsic GTPase activity, but also the first for which The only change in Gs␣ that is similar to that in *T␣ is that the ␣␤␥

the nature of its subunit composition was uncovered and the BIOCHEMISTRY apparent afﬁnity for guanine nucleotides is decreased in both first for which it was shown that GTP binding and activation was proteins. The molecular basis of the changed properties are dis- accompanied by an ␣␤␥ to GTP␣ plus ␤␥ dimer subunit cussed based on the known crystal structure of Gs␣ and the dissociation reaction (reviewed in ref. 2). In 1993, T␣ became the changes introduced by the same mutation in a *T␣ (Gt␣*) with only first G␣-subunit for which a crystal-based model became avail- 23 aa from Gi1␣.Gt␣*–R238E, with four fewer mutations in switch able (5, 6). Structures for the ␣-subunit of the inhibitory G 3, was reported to show no evidence of DN properties, is activated protein 1 (Gi1␣) (7), the ␤␥ dimer (8), and trimeric ␣␤␥ forms by GTP␥S, and has reduced GTPase activity. The data highlight a of transducin and Gi1 (9, 10) were reported shortly thereafter. critical role for the switch 3 region in setting overall properties of The ␣-subunits were shown to be two-domain structures: an signal-transducing GTPases. Ϸ180-aa GTPase domain, highly homologous to the smaller regulatory GTPase ras, and, inserted into ras’s switch 1 effector adenylyl cyclase ͉ ␤-adrenergic receptor ͉ GTP shift ͉ GTPase ͉ crystal sequence, a Ϸ120-aa six-helix helical domain (␣A through ␣F) connected at each end to Switch 1 by linkers 1 and 2. Comparison eterotrimeric G proteins are molecular machines that trans- of the structures of ␣-subunits in their inactive, GDP-bound Hduce the signal generated by the binding of agonists to forms (e.g., refs. 9–11) to those of their activated forms, occu- seven-transmembrane receptors into changes in the activity of pied by either GTP␥S (5, 12) or GMP-P(NH)P (13) led to the effectors. Seven transmembrane receptors, also known as G identification of three ‘‘switch’’ regions. G␣ switch 1 and 2 are protein-coupled receptors (GPCRs), and G proteins each con- structurally homologous to the switch regions identified previ- stitute a family of structurally and functionally related molecules. ously in ras (14). Switch 3 comprises the loop connecting the The basic mechanism by which all GPCRs act is by catalyzing the GTPase domain’s fourth ␤-strand (␤4) to its third ␣-helix (␣3). exchange of GTP for GDP on the ␣-subunits of the trimeric G The final ␣5-helix of ␣-subunits is somewhat shorter than ras ␣5 proteins. The binding of the nucleoside triphosphate promotes and includes at its C terminus 10 aa involved in receptor the dissociation of the so-far inactive trimer into an ␣GTP recognition (15). The guanine nucleotide binding site is formed complex plus a ␤␥ dimer, both of which are competent to interact by the GTPase domain, but is somewhat occluded by the closely and modulate the activity states of effectors (for recent reviews juxtaposed helical domain. see refs. 1–3). A strict set of specificities exists that defines which Activation of G␣-subunits (as well as of ras and ras-like of the 16 G protein ␣-subunits and which of the G␤␥ dimers GTPases) involves the binding of both GTP and Mg2ϩ (Fig. 1). interact with which effector function. These specificity rules, Binding of Mg2ϩ involves six coordination bonds of which two which are best understood for ␣-subunits, dictate that the are contributed by GTP (one oxygen each of the ␤ and ␥ activated forms of stimultory G protein ␣ (Gs␣) stimulate phosphates, ␤O and ␥O) and two are provided by water oxygens adenylyl cyclases (ACs), the activated forms of Gi␣ inhibit AC, locked in place by hydrogen bonds, one to the ␦O of an aspartic and the rod and cone transducin ␣-subunits (T␣s) activate visual acid (Asp-223 in Gs␣) and the other to an oxygen of the phosphodiesterase (PDE) in rod and cone photoreceptor cells, ␣-phosphate of the GTP. The last two coordination bonds are respectively. Differences in primary amino acid sequence among provided by oxygens of G␣ amino acids: one of a Ser (Ser-54 of G protein ␣-subunits define their effector specificities. In sup- Gs␣) and the other of a Thr (Thr-204 in Gs␣). port, studies of chimeric ␣-subunits have borne out the assumption that effector specificity resides in well defined topologically identified regions of ␣-subunits (cf. ref. 4). Author contributions: A.R.Z. and L.B. designed research; A.R.Z. performed research; A.R.Z. In contrast to the easily understandable differences in effector and L.B. analyzed data; and A.R.Z. and L.B. wrote the paper. specificity, structural features of ␣-subunits that define which of The authors declare no conﬂict of interest. the many highly homologous GPCRs interacts with which ␣-sub- †To whom correspondence should be addressed. E-mail: [email protected].

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0712261105 PNAS ͉ February 19, 2008 ͉ vol. 105 ͉ no. 7 ͉ 2363–2368 Downloaded by guest on October 2, 2021 Fig. 1. Model of the 3D structure, based on coordinates taken from Protein Data Base ID code 1JCV, of the region of Gs␣ thought to be affected in the Gs␣–R265E mutant. Mg and its six coordination bonds are shown in green; oxygen atoms of two water molecules (w), and relevant oxygens contributing to the coordination shell of Mg are shown in red, as are oxygens of Asp-223 and the ␣ phosphate (␣O), which stabilize by hydrogen bonds the coordinat- ing water molecules. Some secondary structure features of Gs␣ are highlighted, as are the distances between atoms of Glu-50, Arg-258, and Arg-265 thought to interact by forming ion pairs.

Using a chimera between T␣ and Gi1␣ (*Ta), which is better suited for expression in bacteria while preserving all known functions of T␣ (4), as a model to explore the effects of mutations that would be predicted by the crystal structures to affect the interaction between the GTPase and the helical domains, Cerione and collaborators (16, 17) have recently tested how these mutations would correlate with changes in function- ality. One such mutant, *T␣–R238E, which caught our attention Fig. 2. Multiple amino acid sequence alignment comparing bovine trans- as it could not be activated by GTP␥S, was devoid of GDP or ducin ␣ (␣t), human Gi1␣ (␣i), the 394 variant of human Gs␣ (␣s), and *T␣ (chim ␣ ␣ ␣ Ϫ GTP, i.e., it was nucleotide-free, and behaved as a dominant t- i1). Amino acid sequences are compared with that of Gs , where denotes that the amino acid is identical to that in Gs␣. A consensus line is negative (DN) when placed in an assay in which it could compete ␣ shown for which an uppercase letter identifies the identity among all com- with activated WT *T for activation of PDE, the effector of pared sequences, and Ϫ denotes that in at least one of the proteins the transducin (17). R238 of *T␣ lies in the Switch 3 domain of *T␣ sequence differs by one or more amino acids. Secondary structure features (Fig. 2). (␤-strands 1–6, ␣-helices 1–5 and A to G, switch 1, 2, and 3 regions, GTPase to There have been previous attempts to create DN forms of helical domain linkers Lk1 and Lk2) are highlighted, as are sites of ADP- Gs␣. The first was the attempt by Hildebrandt et al. (18) who ribosylation by cholera (CTX) and pertussis (PTX) toxins and interactions with introduced into Gs␣ the Ser-to-Asn mutation that confers DN Mg. The alignment was generated with Accelrys GCG software using PILEUP properties to small ras-like GTPases. Although the Gs␣–S54N so to build the nn.msf file and BOXSHADE to render the figure. made did display DN properties [it was not activated by the combined actions of GTP and the ␤2-adrenergic receptor in comparison with the reported propertied of the cognate *T␣ (␤2AR] it exhibited strong spontaneous Gs␣ activity equivalent mutant. to 50% of the activity displayed by the fully activated WT Gs␣. The most successful was Berlot’s step-by-step approach, which Results and Discussion culminated with the creation of a DN Gs␣ with seven mutations in three regions of the molecule (19). Here, we introduced into Parallelism Between Signal Transduction by Transducin and Gs. The ␣ ␣ activation of transducin ␣ by receptor has features in common Gs the mutation that converted *T into a DN, with the ␣ expectation that, with this single change, Gs␣–R265, we might with those of Gs and features that are unique. The progression ␥ create a nucleotide-free ␣-subunit that would promote through of activation of transducin can be assessed by binding of GTP S, its interaction with the ␤2AR the stabilization of the receptor in changes in intrinsic Trp-207 fluorescence, proteolytic break- its high-affinity form for agonist and, because it would not bind down, and development of the ability to activate visual PDE, GTP or GDP, that this interaction would be permanent, allowing which occurs by interaction with and removal of inhibitory PDE␥ ␣␤␥ the mutant Gs␣ to act as a DN element susceptible to be probed from the 2 holoenzyme. On the receptor side, activation of for defined conformations that in systems with transient GPCR transducin is promoted by illuminated MetaII rhodopsin, which interactions could not be observed. has high affinity for nucleotide-free transducin and thereby Unexpectedly, however, the mutant Gs␣, instead of being promotes nucleotide exchange (for a review see ref. 20). De- resistant to activation by GTP␥S, was found to have a facilitated pending on the particular mode of reconstitution of the system activation by guanine nucleotides and to interact well and from its components, one or the other of these parameters is reversibly with receptor. We report the properties of this mutant measured to assess molecular properties of its components. In

2364 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0712261105 Zurita and Birnbaumer Downloaded by guest on October 2, 2021 Fig. 4. The activation of the mutant Gs␣–R265E by GTP is impaired without a signiﬁcant effect on the position of the dose–response relationship with which the agonist–receptor complex increases activation the G protein by the nucleotide. CK7 cells expressing Gs␣ (F) or the mutant Gs␣–R265E (ᮀ) were washed and analyzed for AC activity by incubating 200,000 cells in the presence of 10 ␮M GTP and increasing concentrations of isoproterenol (Left)or Fig. 3. Reconstitution of Gs-regulated AC activities by WT and R265E Gs␣ in PGE (Right). S49 cycϪkinϪ (CK7) cells. CK7 cells expressing the WT Gs␣ or the mutant 1 Gs␣–R265E were washed, suspended at 200,000 cells/10 ␮l in buffer containing 20 mM Na-Hepes (pH. 8.0), 50 mM NaCl, 1 mM DTT, and 1 mM EDTA, and lates AC in membranes from Gs␣–R265E expressing CK7 cells analyzed for AC activity as described in Materials and Methods by incubating ␣ 10 ␮l of the cell suspensions for 20 min at 32°C in the presence of 10 ␮M GTP, less than in membranes of WT Gs -expressing cells, the EC50 Ͻ 10 ␮M GTP plus 10 ␮M isoproterenol (GTP ϩ ISO), 10 mM NaF plus 10 ␮M AlCl3 with which activation is obtained differs by 2-fold (Fig. 4), Ϫ ␮ ␥ (AlF4 ), or 100 M GTP S. suggesting that the mutation does not interfere significantly with

the signal transduction process. A similar result was obtained by BIOCHEMISTRY assessing the stimulation of AC by PGE1 (Fig. 4). contrast, the progress of activation of the signal-transducing Gs Ϫ Activation of Gs by AlF4 requires GDP instead of GTP. To protein is commonly followed by assessing its stimulation of AC Ϫ Ϫ Ϫ Ϫ test whether the reduced stimulation of AC by AlF4 in mem- in membranes of S49 cyc or cyc kin cells (CK7 cells), as seen branes with the mutant Gs␣ might be casued by a reduction in in response to added guanine nucleotide or aluminum fluoride affinity of the system for guanine nucleotides as suggested by the or to activation of the ␤2AR in the presence of GTP. data obtained with *T␣, membranes from R265E-expressing In the present work we generated CK7 cells expressing cells were incubated at 4°C for 10 min with up to 5 mM GDP at Gs␣–R265E, a mutation that is equivalent to the *T␣–R238E a constant concentration of free Mg2ϩ of 5 mM and assayed at mutation found by Pereira and Cerione (17) to be resistant to the same final concentration of Mg2ϩ after a 5-fold dilution into ␥ Ϫ activation by rhodopsin and GTP and studied its susceptibility AC assay reagents. Even at 5 mM/1 mM GDP, the AlF4 - to be activated by GTP␥S, both without the cooperation of stimulated activity was still reduced by 70–80% when compared receptor stimulation (hormone-independent activation) and with activity obtained under the same conditions in membranes with simultaneous activation by receptor (hormone-stimulated with WT Gs␣ (data not shown). The reason for this decrease in activation of AC). the ability of the mutant Gs to be activated by the combination Ϫ of AlF4 and GDP is unknown and was not explored further. Gs␣–R265E Is Susceptible to Hormone-Independent Activation by The data reported so far indicate that Gs␣ and *T␣ differ GTP␥S and Hormone-Stimulated Activation by GTP. Although the markedly in their ability to be activated by guanine nucleotides parental CK7 cells have no measurable AC activity under the and in their response to activated GPCR, i.e., light-activated conditions of our assay (data not shown), CK7 cells expressing rhodopsin vs. ligand-activated ␤2AR. Thus, a mutation that human Gs␣ display a classical hormone-stimulated AC activity essentially inactivates *T␣ as a signal-transducing molecule has caused by reconstitution of the receptor–G protein–AC signal very different effects on Gs, which when viewed from the transduction pathway (Fig. 3). The reconstituted AC activity guanine nucleotide point of view, results in facilitated activation resembles that of the WT parental S49 cells (which express the by GTP␥S instead of refractoriness to activation. native murine Gs␣ instead of the human protein) in that receptor-mediated stimulation almost absolutely depends on Mutant Gs–R265E Interacts With and Stabilizes the ␤2AR in a State of addition of a guanine nucleotide, and basal activities are very low High Affinity for Agonist That is Indistinguishable from That Stabilized (21). Compared with WT Gs␣, the R265E mutant reconstituted by WT Gs. In addition to monitoring the interaction of receptor an activity that in the presence of isoproterenol was reduced by with the G protein by determining activation of the effector Ϫ 80% and with AlF4 was reduced by 70%. In contrast, activities (PDE activity for T␣ and AC activity for Gs␣), the ability of a measured in the presence of GTP were Ͻ2-fold higher, and with G protein to interact productively with receptor can be assessed GTP␥S were 2-fold higher than activities reconstituted with WT by monitoring changes in affinity for agonist of its cognate Gs␣. Even so, isoproterenol still stimulated the AC activity receptor as seen by displacement of antagonist binding. As mediated by the mutant Gs␣ between 2- and 3-fold. These results illustrated in Fig. 5, competitive inhibition curves of antagonist indicated that the mutant Gs␣, unlike *T␣–R238E, is susceptible binding to the ␤2AR obtained by increasing concentrations of to activation by guanine nucleotides and is capable of transduc- agonist are monophasic and of low affinity for agonist when ing receptor signals into effector activation. The expression of membranes in which the receptor is embedded are devoid of Gs␣ the WT and mutant Gs␣ in the transformed cells appeared to be (Fig. 5A), but are biphasic when membranes also contain Gs similar, as seen by ADP-ribosylation of a Ϸ45-kDa band in (Fig. 5 B and C). The biphasic nature of the displacement curves membranes of CK7 cells expressing the WT and the R265E is the result of the coexistence of complexes of nucleotide-free forms of Gs␣ (data not shown). Although isoproterenol stimu- Gs and agonist-occupied receptor, which display high affinity for

Zurita and Birnbaumer PNAS ͉ February 19, 2008 ͉ vol. 105 ͉ no. 7 ͉ 2365 Downloaded by guest on October 2, 2021 Fig. 6. Mutant Gs–R265E requires higher concentrations of nucleotides than the WT Gs to shift the CK7 cell ␤2AR from its state of high affinity for agonist to its state of low affinity for agonist. Membranes (15–20 ␮g protein per incubation) from CK7 cells expressing Gs␣ (A)orGs␣–R265E (B) were incubated in the absence of a nucleoside triphosphate regenerating system with 0.1 nM of 125I-CYP, 50 nM isoproterenol, and the indicated concentrations of GDP (open symbols) or GTP␥S (closed symbols). Note that the binding of 125I-CYP to the membrane increases with the concentration of nucleotides because of the gradual transition of the ␤2AR from its high agonist affinity state to its low agonist affinity state. The EC50 values reveal that the mutation causes marked a decrease in the affinity for nucleotide. The results are representative of two independent experiments.

monophasic nature. Expression of WT Gs␣ or R265E–Gs␣ changed agonist binding from monophasic to biphasic in agreement with a ternary complex model in which the ␤2AR exists in two states: one not associated with a G protein, identical to that found in Gs-deficient CK7 cell membranes, the other with a Gs-induced high affinity for agonist (21–24). The Kd values obtained for low and high agonist affinities did not differ significantly among membranes expressing one or the other Gs (WT or R265E) and were averaged, Fig. 5. WT Gs␣ and mutant R265E Gs␣ have similar ability to stabilize the ␤2AR Ϯ Ϯ of CK7 cells in its high agonist affinity state. Membranes (5–10 ␮g per incubation) yielding values of 279 55 and 3.0 0.3 nM for the low agonist from CK7/mock cells (A) or from CK7 cells expressing Gs␣ (B and C) or Gs–R265E affinity Kd (KL) and the high agonist affinity Kd (KH), respectively. (D and E) were incubated as described in Materials and Methods with 0.1 nM of At the levels at which WT Gs␣ and mutant Gs␣ were expressed, the 125I-CYP and the indicated concentrations of unlabeled isoproterenol in the effectiveness with which they stabilized the receptor in its high- ␮ absence or presence of 100 or 500 M GTP or GDP. Bound and free ligands were affinity state (RH) was indistiguishable (47%; Fig. 5 B and D vs. C separated by the polyethylene glycol precipitation procedure described in Ma- and E). Thus, WT and mutant Gs induce qualitatively and quan- terials and Methods. A global fit of a two-state receptor model (lines) to the titatively the same high agonist affinity state of the receptor. experimental results (symbols) was obtained by using an equilibrium dissociation However, they differed markedly in the concentration of nucleotide constant of 0.16 nM for 125I-CYP for the receptor–probe interaction, and the indicated equilibrium dissociation constants of the receptor states of high affinity required to disengage from the receptor. This difference is best seen (KH) and low affinity (KL) for the agonist isoproterenol. The resulting partition of in the experiment of Fig. 6, which shows EC50 values with which the receptor between high-and low-affinity states, as percent of total in the high GTP␥S and GDP disengaged the Gs from the receptor to be shifted affinity state, is depicted for each experimental condition. The KL and KH values by 35- and 52-fold when Gs␣ is compared with Gs␣–R265E. Thus, used for the global fit are the averages of values obtained by fitting each of the the R265E mutation causes a significant reduction in the effective- displacement data independently of the other displacement data. The results are ness with which guanine nucleotides act to cause the dissociation of representative of three independent such experiments with similar results. Val- the Gs–receptor complex. ues are means Ϯ SD for the data set shown. A time-course study of the rate of activation of AC in membranes of CK7 cells expressing WT Gs as compared with Gs–R265E the agonist, and the Gs-free receptor, which displays the same showed that the typical lag in activation of Gs in the absence of ␣ low affinity for agonist seen in for the bulk of the receptor in hormonal stimulation is shortened in membranes of Gs –R265E- membranes without Gs. expressing cells (Fig. 7). The lag in AC activation by nonhydrolyz- As seen by Scatchard analysis of specific 125I-CYP binding to able GTP analogues correlates with slow binding of the analogues membranes of mock-infected and membranes from cells expressing to G protein ␣-subunits, which has been shown to reflect a slow rate Gs␣ or Gs␣–R265E, the receptor abundance was between 0.8 and of dissociation of resident GDP from G␣-subunits (25). Taken 1.3 pmol/mg membrane protein and had an equilibrium dissociation together, our results are consistent with Gs–R265E having between constant that varied between 0.14 and 0.18 nM in the presence of 35- and 50-fold lowered affinities for GTP and GDP. In this respect ␣ 5 mM MgCl2. In agreement with previous studies, binding of the the mutant Gs resembles the cognate mutant *T found by Pereira antagonist was unaffected by presence of guanine nucleotides (cf. and Cerione (17) to be devoid of resident guanine nucleotide. It refs. 22 and 23 and data not shown) and differed by Ͻ20% from one differs, however, in that the Gs mutant does interact productively membrane preparation to another (data not shown). At 0.16 nM, with receptor and is activated by GTP␥S, whereas the *T␣ mutant equilibrium binding was reached within 5 min of incubation (data does not. In agreement with a lower affinity for nucleotides, not shown). The data shown in Figs. 5 and 6 were obtained by activation of the Gs–R265E–AC complex by GTP␥S, in the pres- incubating the reaction mixtures of CK7 membranes for 30 min at ence of hormone required higher concentrations of the nucleotide 32°C. Binding of isoproterenol to the ␤2AR was of an essentially for half-maximal effects than the WT counterpart (Fig. 8).

2366 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0712261105 Zurita and Birnbaumer Downloaded by guest on October 2, 2021 protein may be a general property of GPCRs and underlies the mechanism by which they promote nucleotide exchange. Not explained by our experiments is why, while GTP␥S activates mutant Gs very well and the mutant Gs interacts productively with receptor, activation by GTP is minimal regardless of a cooperating receptor stimulation (Fig. 8B). One possibility not explored here is that the mutant may have a much higher intrinsic GTPase activity. The low abundance of Gs␣ in the transformed CK7 cells does not permit addressing this question at this time. After many of the experiments shown here had been performed we became aware of an article by Barren at al. (28), in which the R238E mutation was introduced into a T␣–Gi1␣ chimera (Gt␣*, chimera 8 in ref. 4) with four fewer amino acids from Gi1␣, leaving switch 3 of T␣ unaltered. R265E–Gt␣* binds nucleotides, interacts productively with photoactivated rhodop- ␥ Fig. 7. Kinetic analysis of AC stimulation reveals much faster activation of the sin exchanging GTP S for GDP, and has a much lower GTPase Gs–R265E by GTP␥S than that of WT Gs. Membranes from Gs␣ (A) and Gs␣–R265E activity. There are no indications that Gs␣–R265E has an (B) expressing CK7 cells were tested for the rate of activation of AC by 100 ␮M impaired GTPase (also see above). Such a feature would predict GTP␥ in the absence (E) and the presence of 100 ␮M isoproterenol (F). Assay an increased activation effectiveness for GTP, instead of the mixtures (1.5 ml each) were incubated at 32°C, and formation of [32P]cAMP was decreased effectiveness seen in our experiments (Figs. 1 and 6B). monitored by removing 50-␮l aliquot at the indicted times. Note that at steady It is clear that in comparison with Gs␣ the highly homologous state the receptor-independent activation of Gs is similar to that of the mutant Gs. transducin ␣ is functionally very different. Moreover, the differing properties of R265E–*T␣ and R265E–Gt␣* highlight a critical role for the switch 3 region of G protein ␣-subunits. The fact that not only GTP but also GDP is competent in promoting the disengagement of Gs from the receptor is in On the Mechanism of the Changes Caused by the R265E Mutation. agreement with the original 1970 finding that both GTP and Upon inspection of the 3D model of Gs␣ deduced from its crystal GDP affect binding of glucagon to its receptor (26), which were structure (e.g., Protein Data Base ID code 1JCV; Fig. 1), it becomes BIOCHEMISTRY extended to the turkey and S49 cell ␤-adrenergic receptors in apparent that at a distance of 3.00 Å, Arg-265 and Glu-50 form an 1980 (27). It would thus appear that the high agonist affinity ionic pair. By inverting the polarity of the charge at the side chain receptor–G protein complex represents the receptor complexed of amino acid 265, the R265E mutation can be expected to disturb with the nucleotide-free G protein, most likely trimeric because not only the positions of the side-chain atoms of Glu-50 but also the subunit dissociation requires occupancy of the Gs␣ by GTP. The backbone peptide chain of the highly conserved ␣-subunit P-loop fact that the ␤2AR forms a stable complex with nucleotide-free (sequence GAGESGK). The P-loop forms the pocket in which the Gs and therefore has highest affinity for this form of the G phosphates of the guanine nucleotides are held. The negatively charged ␦-carboxyl group of Glu-50 also interacts with the positively charged guanidino group of Arg-258 (distance of Arg-258 guanidino NHϩ to Glu-50 O␧Ϫ ϭ 4.03 Å and Arg-258 N␧ to Glu-50 O␧ ϭ 3.02 Å). In transducin ␣ and Gi1␣, Arg-258 is replaced with Val (T␣–V231) and Ala (Gi1␣–Ala-231), respectively (see Fig. 2). It is likely that Gs␣–R258 contributes to the stability of the P-loop. This stabilizing effect would be absent in the chimeric *T␣ (T␣–Gi1␣) protein and may be at the root of the differing impacts that mutating *T␣ R238E and Gs␣ R265E have on the properties of the resulting mutant proteins.

Materials and Methods Radioisotopes [␣-32P]ATP, [3H]cAMP, and [32P]NADϩ were purchased from GE Healthcare-Amersham. [125I]Iodocyanopindolol (125I-CYP) was from NEN– PerkinElmer. Bovine gamma globulin, (Ϫ)isoproterenol, and PGE1 were from Sigma–Aldrich; polyethyleneglycol 6000 was from Calbiochem, and guanine nucleotides were from Boehringer-Ingelheim. All other chemicals and biochemicals were of the highest purity commer- cially available and used without further purification. Standard laboratory and recombinant DNA techniques were used to gen- erate the R265E mutation in the cDNA coding for the 394-aa splice variant of the human Gs␣ (29). cDNAs were mutated by using the Quikchange II XL Site-Directed Mutagenesis kit from Stratagene. The WT and R265E Gs␣ subunits were stably expressed CK7 cells transformed by retroviral-mediated Fig. 8. Compared with WT Gs, the Gs␣–R265E mutation requires a higher infection as described by Berlot (30). CK7 cells are derivatives of S49 thymoma Ϫ Ϫ concentration of GTP␥S for half-maximal activation in the presence of agonist- cells that are cyc and kin , i.e., deficient in Gs␣ activity and cAMP-dependent activated receptor and is very poorly activated by GTP. Membranes (10 ␮g protein kinase activity. CK7 cell clones we grown in suspension, harvested, and protein per assay) from cells expressing Gs␣ (A and C) and Gs␣–R265E (B and used for preparation of purified membranes as described by Ross et al. (21), D) were tested for stimulation of AC activity by increasing concentration of except that the membrane purification was stopped after obtaining the GTP␥S(C and D) in the presence (closed symbols) or absence (open symbols) of 43,000 ϫ g pellet. Membranes were suspended in 20 mM Hepes-Na (pH 8.0), 100 ␮M isoproterenol. Note that the reduction in the shift in EC50 for GTP␥S 50 mM NaCl, 1 mM DTT, 1 mM EDTA, and 2 mM MgCl2. activation of the mutant Gs in the presence of isoproterenol is in agreement Cholera toxin was from List Biological Laboratories. Activation of the toxin with the data of Fig. 6, showing that activation of Gs proceeds with a marked and treatment of CK7 cell membranes were as described by Birnbaumer et al. lag, whereas that of the mutant Gs is much faster. (31). [32P]ADP-ribosylation of membranes with cholera toxin followed by

Zurita and Birnbaumer PNAS ͉ February 19, 2008 ͉ vol. 105 ͉ no. 7 ͉ 2367 Downloaded by guest on October 2, 2021 SDS/PAGE and visualization by autoradiography were as described by Hilde- where RP* is the concentration of 125I-CYP specifically bound; P*isthe 125 brandt et al. (32). concentration of the probe, I-CYP; KP is the equilibrium dissociation con- AC activities were measured as described by Iyengar et al. (33). stant for the interaction of 125I-CYP with the H and L states of the receptor; and Specific binding of 125I-CYP to the ␤2AR of CK7 cell membranes and the concentration of total receptor in the incubation ϭ sum of concentrations separation of free ligand from bound ligand by the polyethyleneglycol pre- of the receptor in the high-affinity, RH, and low-affinity, RL states that bind cipitation method were as described by Abramowitz et al. (34). Specific agonist H with high- and low-affinity equilibrium dissociation constants KH 125 binding was defined as the difference between I-CYP bound in the absence and K , respectively. Fits of the model to the data were calculated by using the ␮ L and that bound in the presence of 0.4 M unlabeled isoproterenol. A two- Solver routine of Microsoft Excel. state model was used to fit by the least sum of the squares of differences All experiments were repeated between two and four times. Figures show method to the data according to the equation: data that are representative of the data obtained in the repeats. R R ϭ H ϩ L RP* , ACKNOWLEDGMENTS. This work was supported by the Intramural Research KP H KP H 1 ϩ ͩ 1 ϩ ͪ 1 ϩ ͩ 1 ϩ ͪ Program of the National Institutes of Health, National Institute of Environ- P* KH P* KL mental Health Sciences.

1. Wettschureck N, Offermanns S (2005) Mammalian G proteins and their cell type- 21. Ross EM, Maguire ME, Sturgill TW, Biltonen RL, Gil-man AG (1977) Relationship specific functions. Physiol Rev 85:1159–1204. between the ␤-adrenergic receptor and adenylate cyclase: Studies of ligand binding 2. Birnbaumer L (2007) The discovery of signal transduction by G proteins: A personal and enzyme activity in purified membranes of S49 lymphoma cells. J Biol Chem account and an overview of the initial findings and contributions that led to our 252:5761–5775. present understanding. Biochim Biophys Acta Biomembranes 1768:756–771. 22. Kent RS, DeLean A, Lefkowitz RJ (1980) A quantitative analysis of ␤-adrenergic recep- ␣ 3. Birnbaumer L (2007) Expansion of signal transduction by G proteins from 3 to 16 tor interactions: Resolution of high and low affinity states of the receptor by computer ␤␥ subunits plus dimers: The second 15 years or so. Biochim Biophys Acta Biomem- modeling of ligand binding data. Mol Pharmacol 17:14–23. branes 1768:772–793. 23. DeLean A, Stadel JM, Lefkowitz RJ (1980) A ternary complex model explains the 4. Skiba NP, Bae H, Hamm HE (1996) Mapping of effector binding sites of transducin agonist-specific binding properties of the adenylate cyclase-coupled ␤-adrenergic ␣-subunit using Gt␣/Gi1␣ chimeras. J Biol Chem 271:413–424. receptor. J Biol Chem 255:7108–7117. 5. Noel JP, Hamm HE, Sigler PB (1993) The 22-Å crystal structure of transducin ␣-GTP␥S. 24. Cerione RA, et al. (1984) The mammalian ␤2-adrenergic receptor: Reconstitution of the Nature 366:654–663. pure receptor with the pure stimulatory nucleotide binding protein (N ) of the ade- 6. Lambright DG: Noel JP, Hamm HE, Sigler PB (1994) Structurel determinants for the s nylate cyclase system. Biochemistry 23:4519–4525. activation of the ␣ subunit of a heterotrimeric G protein. Nature 239:621–628. 25. Ferguson KM, Higashijima T, Smigel M, Gilman AG (1986) The influence of bound GDP 7. Coleman DE, et al. (1994) Structure of active conformations of Gi␣1 and the mechanism of GTP hydrolysis. Science 265:1405–1412. on the kinetics of guanine nucleotide binding to G proteins. J Biol Chem 261:7393– 8. Sondek J, Bohm A, Lambright DG, Hamm HE, Sigler PB (1996) Crystal structure of a 7399. G-protein ␤␥ dimer at 2.1-Å resolution. Nature 379:369–374. 26. Rodbell M, Krans HMJ, Pohl SL, Birnbaumer L (1971) The glucagon-sensitive adenyl 9. Wall MA, et al. (1995) The structure of the G protein heterotrimer Gi␣1 ␤1 ␥2. Cell cyclase system in plasma membranes of rat liver. IV. Binding of glucagon: Effect of 83:1047–1058. guanyl nucleotides. J Biol Chem 246:1872–1876. 10. Lambright DG, et al. (1996) The 2.0-Å crystal structure of a heterotrimeric G protein. 27. Iyengar R, Abramowitz J, Bordelon-Riser ME, Birnbaumer L (1980) Hormone receptor- Nature 379:311–319. mediated stimulation of adenylyl cyclase systems: Nucleotide effects and analysis in 11. Coleman DE, Sprang SR (1998) Crystal structures of the G protein Gi␣1 complexed with terms of a two-state model for the basic receptor-affected enzyme. J Biol Chem GDP, Mg2ϩ: A crystallographic titration experiment. Biochemistry 37:14376–14385. 255:3558–3564. 12. Tesmer JJ, Sunahara RK, Gilman AG, Sprang SR (1997) Crystal structure of the catalytic 28. Barren B, Natochin M, Artemyev NO (2006) R238E in transducin-␣ yields a GTPase and domains of adenylyl cyclase in a complex with Gs␣.GTP␥S. Science 278:1907–1916. effector-deficient, but not dominant-negative, G-protein ␣-subunit. Mol Vis 12:492– 13. Coleman DE, Sprang SR (1999) Structure of Gi␣1.GppNHp, autoinhibition in a galpha 498. protein-substrate complex. J Biol Chem 274:16669–16672. 29. Mattera R, et al. (1986) Identification by molecular cloning of two forms of the ␣- 14. Milburn MV, et al. (1990) Molecular switch for signal transduction: Structural differ- subunit of the human liver stimulatory (Gs) regulatory component of adenylyl cyclase. ences between active and inactive forms of protooncogenic ras proteins. Science FEBS Lett 206:36–42. 247:939–945. 30. Berlot CH (2002) Expression and functional analysis of G protein ␣ subunits in S49 15. Conklin BR, Farfel Z, Lustig KD, Julius D, Bourne HR (1993) Substitution of three amino lymphoma cells. Methods Enzymol 344:261–277. acids switches receptor specificity. Nature 363:274–276. 31. Birnbaumer L, Swartz TL, Abramowitz J, Mintz PW, Iyengar R (1980) Transient and 16. Majumdar S, Ramachandran S, Cerione RA (2004) Perturbing the linker regions of the steady-state kinetics of the interaction of guanyl nucleotides with the adenylyl cyclase ␣-subunit of transducin: A new class of constitutively active GTP-binding proteins. J Biol system from rat liver plasma membranes: Interpretation in terms of a simple two-state Chem 279:40137–40145. model. J Biol Chem 255:3542–3551. 17. Pereira R, Cerione RA (2005) A switch three-point mutation in the ␣-subunit of 32. Hildebrandt JD, et al. (1983) Stimulation and inhibition of adenylyl cyclases is mediated transducin yields a unique dominant-negative inhibitor. J Biol Chem 280:35696–35703. 18. Hildebrandt JD, Day R, Farnsworth CL, Feig LA (1991) A mutation in the putative by distinct proteins. Nature 302:706–709. Mg2ϩ-binding site of Gs␣ prevents its activation by receptors. Mol Cell Biol 11:4830– 33. Iyengar R, Bhat MK, Riser ME, Birnbaumer L (1981) Receptor-specific desensitization of 4838. the S49 lymphoma cell adenylyl cyclase: Unaltered behavior of the regulatory compo- 19. Berlot CH (2002) A highly effective dominant negative ␣s construct containing muta- nent. J Biol Chem 256:4810–4815. tions that affect distinct functions inhibits multiple G-coupled receptor signaling 34. Abramowitz J, Iyengar R, Birnbaumer L (1982) Guanine nucleotide and Mg ion regu- pathways. J Biol Chem 277:21080–21085. lation of the interaction of gonadotropic and ␤-adrenergic receptors with their 20. Sakmar TP, Menon ST, Marin EP, Awad ES (2002) Rhodopsin: Insights from recent hormones: A comparative study using a single membrane system. Endocrinology structural studies. Annu Rev Biophys Biomol Struct 31:443–484. 110:336–346.

2368 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0712261105 Zurita and Birnbaumer Downloaded by guest on October 2, 2021