Proc. Natl. Acad. Sci. USA Vol. 95, pp. 12878–12883, October 1998 Biochemistry

The helical domain of a ␣ subunit is a regulator of its effector

WEI LIU AND JOHN K. NORTHUP*

Laboratory of Cellular Biology, National Institute on Deafness and Other Communication Disorders, 5 Research Court, Rockville, MD 20850

Edited by Alfred G. Gilman, University of Texas Southwestern Medical Center, Dallas, TX, and approved August 28, 1998 (received for review April 13, 1998)

ABSTRACT The ␣ subunit (G␣) of heterotrimeric G been mapped to the RasD and the ␣ amino-terminal sequence proteins is a major determinant of signaling selectivity. The (1, 3, 4). Little is known about the corresponding function(s) G␣ structure essentially comprises a GTPase ‘‘Ras-like’’ of the HD. The HD is unique to the heterotrimeric G proteins, domain (RasD) and a unique ␣-helical domain (HD). We used whereas a RasD is present in all of the members of the GTPase the vertebrate phototransduction model to test for potential superfamily. Comparison of the sequences reveals functions of HD and found that the HD of the trans- that diversity in the HD is remarkably greater than in the RasD ducin G␣ (G␣ ) and the closely related gustducin (G␣ ), but among G␣ families (ref. 5; Fig. 1). These observations suggest t g ␣ not G␣i1,G␣s,orG␣q synergistically enhance guanosine some G -specific function(s) for the HD moiety. Identification 5؅-␥[-thio]triphosphate bound G␣t (G␣tGTP␥S) activation of of the role of HD has proved elusive. The divergent sequences bovine rod cGMP (PDE). In addition, both of HD (Fig. 1) have led to the proposal that it may serve as an HDt and HDg, but not HDi1,HDs,orHDq attenuate the effector-recognition domain (8). Various other possible func- trypsin-activated PDE. G␣tGDP and HDt attenuation of tryp- tions for the HD have been postulated, including increasing the sin-activated PDE saturate with similar affinities and to an affinity of GTP binding (9), acting as a tethered intrinsic identical 38% of initial activity. These data suggest that GTPase activating protein (10, 11), participating in effector ␣ recognition (5, 12, 13), participating in the inactive-active interaction of intact G t with the PDE catalytic core may be ␣ ␣ caused by the HD moiety, and they indicate an independent conformational transitions of G (14), and regulating G oligomerization (15, 16). site(s) for the HD moiety of G␣t within the PDE catalytic core ␥ We used the retinal signal-transduction model to investigate in addition to the sites for the inhibitory P subunits. The HD ␣ ␣ the potential function of the HD of subunit (G t). moiety of G␣tGDP is an attenuator of the activated catalytic core, whereas in the presence of activated G␣ GTP␥S the We selected this system because the molecular components of t vertebrate phototransduction have been extensively character- independently expressed HD is a potent synergist. Rhodopsin t ized, as has the in vitro biochemistry for all known G coupled catalysis of G␣ activation enhances the PDE activation t t interactions (reviewed in refs. 17 and 18). The activation of the produced by subsaturating levels of G␣ , suggesting a HD- t retinal rod G␣ by photoreceptor rhodopsin can be reconsti- moiety synergism from a transient conformation of G␣ . t t tuted with preparations of homogenous receptor in native disc These results establish HD-selective regulations of vertebrate ␣ membranes of the rod outer segment (19); G t regulated retinal PDE, and they provide evidence demonstrating that the activation of the unique retinal cGMP phosphodiesterase HD is a modulatory domain. We suggest that the HD works in (PDE) has also been examined extensively (18). The nonac- concert with the RasD, enhancing the efficiency of G protein tivated PDE consists of a catalytic core (P␣␤) and two signaling. ␥ ␣␤␥ inhibitory P subunits (20–23). The P 2 (i.e., holoPDE) ␣ remains inactive until it is activated by GTP-bound G t ␣ Heterotrimeric G proteins play a central role in many cell- (G tGTP). Although much has been learned about signal flow signaling processes (1). These signaling proteins are members from the photoexcited receptor to the sensory synapse (24), of an extensive superfamily of GTP-binding regulatory pro- some mechanistic details of this signaling still remain unelu- ␣ teins characterized by a conserved GTP-binding motif (2). The cidated. It has been shown that the HD of G s (HDs) can fold G proteins differ from the monomeric members of this family independently of the RasD (10, 12). Here we report the in a distinct heterotrimeric quaternary structure, including a successful expression and purification of the HD proteins from ␤␥ subunit dimer complexed with the GTP-binding ␣ subunit the major G␣ families. With the expressed HD proteins we (1, 3). The ␣ subunit (G␣) is also distinct from other GTP- establish a HD-selective interaction with the transducin effec- binding proteins in the presence of a unique folding motif, the tor, PDE. ␣-helical domain (HD), which is characteristic of G proteins (4, ␣ 5). Although the subunit has received the most attention MATERIALS AND METHODS because most of the known structural determinants of recep- tor–G protein or effector–G protein interactions reside in G␣ Cloning and Expression Constructs. All of the HD con- proteins (1, 3, 6), the G␤␥ also is involved in receptor structs were generated by oligonucleotide-directed mutagen- recognition and effector modulation (7). Solutions for the ␣ crystal structures of the subunits of transducin (Gt) and Gi This paper was submitted directly (Track II) to the Proceedings office. revealed that these proteins fold into two essentially separate Abbreviations: HA, hemagglutinin; HD, ␣-helical domain of G␣ subunits; RasD, Ras-like GTPase domain of G␣ subunits; G␣t, domains, the conserved GTPase or ‘‘Ras-like’’ domain (RasD) ␣ ␣ ␣ ␣ ␣ ␣ transducin subunit; G g, gustducin subunit; G i1, the subunit of and the unique -helical domain (HD) (4, 5). To date, all sites ␣ ␣ ␣ the inhibiting G protein Gil;G s, the subunit of the for G subunit interactions with receptors and effectors have adenylyl cyclase stimulating G protein Gs;G␣q, the ␣ subunit of the ␤ stimulating G protein Gq; PDE, bovine retinal The publication costs of this article were defrayed in part by page charge cGMP phosphodiesterase; tPDE, tryptically activated PDE; NHS, N-hydroxysuccinimide. payment. This article must therefore be hereby marked ‘‘advertisement’’ in *To whom reprint requests should be addressed at: Department of accordance with 18 U.S.C. §1734 solely to indicate this fact. Health and Human Services, Public Health Service, National Insti- 0027-8424͞98͞9512878-6$0.00͞0 tutes of Health, 5 Research Court, Room 2A-11, Rockville, MD PNAS is available online at www.pnas.org. 20850. e-mail: [email protected].

12878 Downloaded by guest on September 25, 2021 Biochemistry: Liu and Northup Proc. Natl. Acad. Sci. USA 95 (1998) 12879

FIG. 1. Sequence alignment for G␣ HDs. The amino acid sequences for the helical domains from G␣t,G␣g,G␣i1,G␣q, and G␣s are compared. The secondary structure (from ␣A to ␣F) corresponding to G␣t/␣i1 (4) is shown below the aligned sequences. Conserved residues at equivalent positions are shaded. Amino acid identities are boxed.

esis with influenza virus hemagglutinin (HA) tags (YPYD- tained from Qiagen (Chatsworth, CA), and HDt was produced VDYA) at their amino termini. All constructs were confirmed in strain BL21(DE3); the cells were harvested according to a by DNA sequencing, and the corresponding expressed proteins modified procedure from ref. 30. The isolation of all of the HD were confirmed by amino-terminal amino acid sequencing. proteins was accomplished by sequential chromatography over The HDg (rat) expression plasmid was the first constructed, DEAE-Sephadex A-25 (Pharmacia) and elution with a linear and its construct intermediate served as a basis for engineering gradient of 50–700 mM NaCl; AcA54 (IBF Biotechnics, the other HD constructs. Briefly, we created versatile HD Columbia, MD) gel exclusion; FPLC monoQ (Pharmacia) expression vectors based on the gustducin construct as follows. eluted with a linear gradient of 100–500 mM NaCl, and Fast To introduce new sites for excising HD from the corresponding Protein Liquid Chromatography (FPLC) Superdex HR-75 cDNA sequence (25), a unique SnaBI site in the gustducin- (Pharmacia) size exclusion. The distribution of the HD protein containing cloning vector located within the region equivalent was determined by immunoblotting using anti-HA monoclonal to transducin linker I (9) was created by using PCR with the antibody 12CA5 (Boehringer Mannheim). The HD proteins primers 5Ј-GGGGTACCGCTGATCAACTGCCCGTCCTC- eluted from DEAE and MonoQ at distinct NaCl concentra- TAACAG-3Ј and 5Ј-GGAATTCGATGCATTCTTGTTTT- tions based on their differing charges. After the final Superdex ACGTAACCATTCTTGTGGATG-3Ј. To introduce a stop HR-75 chromatography, the HD-containing fractions were codon into the 3Ј end of the HD sequence, the HD region was pooled, diluted to a NaCl concentration less than 7 mM with amplified by using PCR with the primers 5Ј-CGGGATCCA- 20 mM Mops (pH 7.5) and 1 mM DTT, and concentrated in TGGCAAACACACTAGAAGATGGT-3Ј and 5Ј-GCTCTA- an Amicon pressure cell on YM10 membranes, then stored at Ϫ80°C or in the same solution containing 35% glycerol GATCAACCAG TGGTTTTCACACGGGAATGTAGAA- Ϫ CGTCT-3Ј and subcloned into pSE420 (Invitrogen). The HD (vol/vol) at 20°C for short-term storage. The chemical iden- constructs were tagged with HA at their amino termini by tities of the purified HD proteins were confirmed by amino- terminal amino acid sequencing and fast atom bombardment/ inserting the 9-aa epitope (YPYDVPDYA) into NcoI/SnaBI mass spectroscopy. The proper folding structures were verified sites of pSE420 by using Klenow DNA polymerase-mediated by circular dichroism spectra. Protein concentrations were 5Ј 3 3Ј polynucleotide synthesis with the oligonucleotides measured by using the Bradford protein assay (Bio-Rad) using 5Ј-CATGCCATGGGATACCCATACGACGTCCCAGAC- BSA as standard. TACG-3Ј and 5Ј-GGAATTCTACGTAAGCGTAGTCTGG- ␥ Ј Purification of PDE, Preparation of tPDE and P Subunits, GACGTCGTATGGGTATC-3 . The resultant plasmid, and PDE Assay. PDE was isolated from the bovine rod outer termed pSEHA, served as the vector for generating the other segment (20). To prepare trypsin-activated PDE (tPDE), 27 ␮l HD constructs as well. Finally, the HD coding sequence was of 10 ␮M purified PDE was incubated with 120 ␮l of agarose- excised with SnaBI and ligated to the SnaBI fragment of immobilized N-tosyl-L-phenylalanine chloromethyl ketone pSEHA to create HA-tagged HDg. (TPCK)–trypsin (Pierce) in a solution containing 50 mM For each of the HDt mouse (26), HDq mouse (27), HDi1 rat Mops (pH 7.5), 5 mM NaCl, and 2 mM DTT for 17 min at 30°C. (28), and HDsϪL rat (29) constructs, the DNA sequence The reaction was terminated by centrifugation at 12,000 ϫ g for corresponding to the HD between linker I and II (9) was 5 min at 4°C to precipitate the agarose beads. When this Ј ␣ ␥ amplified by PCR using primers, respectively, 5 -ATTATCC- method was used, more than 97% of the initial G tGTP S- ACCAGGACGGGTACGTACTGGAGGAATGCCTCGA- activated activity was obtained, and G␣ GTP␥S no longer Ј Ј t GTTC-3 and 5 -GGAATTCTCAACCAGTGGTTTTGAC- activated the PDE. P␥ subunits were prepared from purified Ј Ј ACGAGAACG-3 (HDt); 5 -CACGGGTCGGGCTACGTA- PDE according to the procedure of Hurley and Stryer (21). GACGAAGACAAGCGCG GCTTC-3Ј and 5Ј-GGAATTC- PDE activity was quantified as inorganic phosphate released Ј Ј TTACCCTGTAGTGGGGACTCGAAC-3 (HDq); 5 -GAG- by 5Ј-nucleotidase digestion of GMP as measured by molyb- GCTGGCTACGTAGAGGAAGAGTGTAAGCAG-3Ј and date-dependent absorbence at 790 nm. The assay followed the 5Ј-GGAATTCTTATCCCGTGGTTTTCACTCTAGTTCT- modified procedure from ref. 31. Phosphate production was Ј Ј G-3 (HDi1); and 5 -TCCCCCGGGTTTAACGGAGAGGG- found to be linear with time and with PDE concentration CGGCG-3Ј and 5Ј-GGACTAGTTTATCCAGAGGTCAGG- under all conditions reported in this study and up to 87 nmol Ј ACGCGGCAG-3 (HDs). of cGMP hydrolyzed. ␣ Purification of the HD Proteins. The proteins HDg,HDi1, Other Methods. The G tGDP was isolated from GTP- HDq, and HDs were expressed in strain M15 (pREP4) ob- eluted transducin by chromatography on Blue Sepharose Downloaded by guest on September 25, 2021 12880 Biochemistry: Liu and Northup Proc. Natl. Acad. Sci. USA 95 (1998)

CL-6B (Pharmacia) as modified from Yamazaki et al. (32). their attenuation of tPDE (Fig. 2B). An identical pattern of ␣ ␥ ␣ ␣␤ The G tGTP S was prepared from the purified G tGDP as selectivity was obtained for the HDt attenuation of P as described (19). Immunoblotting for HA-tagged HD was done found for the synergistic activation; the HDs obtained from the with monoclonal antibody 12CA5 (Boehringer Mannheim) members of the other G protein families showed no attenua- and enhanced chemiluminescence detection (Amersham). tion of P␣␤ activity (Fig. 2B). Our data for HD attenuation of ␣␤ ␣ Materials. Immobilized trypsin (beaded agarose) and N- P are similar to those obtained previously for G tGDP hydroxysulfosuccinimide were purchased from Pierce. cGMP, attenuation of the (34). In Fig. 2C, we compare ␣ GTP␥S, and 5Ј-nucleotidase were from Sigma. G tGDP and HDt attenuation of tPDE. At saturating con- ␣ centrations, both G tGDP and HDt only partially inhibit tPDE Ϯ Ϯ ϭ ␣ (62 4%, mean SD, n 3 for both G tGDP and HDt) RESULTS eliminating a competitive mechanism for this attenuation. The Design and Characterization of the Purified HD Proteins. apparent affinities obtained in this experiment of 500 nM for ␣ We designed HDs based on the crystallographic data of G tGDP and 330 nM for HDt attenuation are nearly equal. G␣ /G␣ chimera complexed with GTP␥S⅐Mg2ϩ and with Furthermore, the coaddition of saturating concentrations of t i1 ␣ GDP (4, 9). The G␣ crystal structures show three distinct G tGDP and HDt produced no additional inhibition over that structural components: a HD consisting of a long central helix found for each independently (Fig. 2C). These data suggest ␣␤ ␣ (␣A) surrounded by 5 shorter helices (␣B–␣F) (4); a RasD that the attenuation of P by G tGDP may be conferred by consisting of a six-stranded ␤-sheet surrounded by six helices its HD moiety. (␣1–␣5) and ␣G; and an amino-terminal segment. Connection To gain insight into the molecular basis of the synergy of the HD to the other two components is achieved by linker conferred by the HDt, we examined the HDt influence on the ␣ ␥ 1 and linker 2 polypeptides. Therefore, we engineered HDs in saturation of PDE activation by G tGTP S. As shown in Fig. such a way that they consisted of ␣A–␣F with their amino 3A, the HDt remarkably increases the apparent affinity of PDE ␣ ␥ termini capped by the linker 1 polypeptides and carboxy for G tGTP S, decreasing the apparent Kd values from 460 † nM to 3 nM. In separate experiments using higher concentra- termini by the linker 2 polypeptides . The amino acid se- ␣ ␥ ␮ tions of G tGTP S (up to 3 M) to reach saturation, we quences of the HD proteins used in these investigations are Ϯ determined the apparent Kd for PDE activation to be 500 70 shown in Fig. 1. Consistent with previous reports of the HD Ϯ ϭ being an independently folded domain (10, 12), the HDs nM (mean SD, n 5), and the Kd in the presence of HDt ␣ ␣ ␣ ␣ ␣ to be 2.5 Ϯ 1.2 nM (n ϭ 5). This Ͼ150-fold enhancement of obtained from G s,G q,G i1,G t, and G g were expressed, although with different yields, as stable and soluble proteins affinity suggests an of PDE by the HDt. ␣ These results, along with the data showing a direct interaction with predominantly -helical secondary structures confirmed ␣␤ by circular dichroism spectral analyses (unpublished data). of the HDt with P (Fig. 2 B and C) strongly argue the existence of at least two distinct sites in PDE for binding of Selective HD Interactions with PDE. Our basic approach to ␣ ␣␤ G t—a HD site on the P as well as the previously defined identify potential interaction(s) of the HD proteins was to test ␥ ␣ binding sites for P . We suggest that the HD moiety of G t may for the competitive interference in the cascade of rhodopsin- ␥ catalyzed activation of the PDE. We predicted that the recep- function as an allosteric modulator of the P interaction with the G␣ RasD. The principal mechanism described for PDE tor or effector contacts with the HD moiety of G␣ (if any) t activation involves the G␣ -mediated release of inhibition would be insufficient to elicit function independent of the t imposed by two inhibitory P␥ subunits on the P␣␤ catalytic RasD. Therefore, HD should competitively inhibit the inter- core (17, 35). This G␣ –P␥ interaction has been attributed to actions of intact G␣ with these signaling components. Our t ␣ binding of the RasD in the active GTP-bound conformation to initial investigation of rhodopsin–G t interaction found no ␥ ␣ ␥ ␤␥ P (4, 36, 37). The HDt synergy with the G tGTP Sisnot evidence for HDt binding to rhodopsin or G (data not ␥ ␮ prevented by P . Fig. 3B shows that 1 MHDt enhances the shown). Therefore, we set out to investigate the potential ␣ ␥ ␣ affinity for G tGTP S activation of PDE even in the presence involvement of HD in G –effector interaction. of 750 pM P␥, a concentration more than 30-fold greater than The rod cGMP-PDE can be stimulated in vitro by the ␥ ␣␤ ␣ ␥ the Kd of P for P . These data argue that the HD acts as a activator G tGTP S (33). When we examined the HD reac- modulator rather than a competitor for the P␥–P␣␤ interac- tivity toward the PDE, to our surprise, we found that rather ␣ ␥ tion. than competing with G tGTP S, the HDt caused an enhanced ␣ ␣ The HD Moiety Function in Intact Subunit. A correlation PDE activation (Fig. 2A). HDi1 (G i family), which is closely between our data from the independently expressed HDt and related to HDt;HDq,orHDs elicited no enhanced activation ␣ its function within the intact G t subunit has been provided by under the same conditions (Fig. 2A), indicating a selective ␣ examination of the interaction of the HDt and G tGDP with interaction between HDt and the PDE enzyme. In addition, ␣␤ ␣ the PDE catalytic subunit (P ) (Fig. 2 B and C). This kinetic the HDg (the HD of G g) stimulated to the same extent as the ␣ similarity between HDt and G tGDP strongly argues that the HDt (Fig. 2A). To exclude the possibility that the HD effect on ␣ G tGDP reactivity toward the core enzyme is a result of its HD augmentation of the enzyme activation may be attributed to ␣ moiety. However, unlike HDt protein, G tGDP is unable to denatured or unfolded proteins, we tested chemically modified enhance the PDE activity under the same conditions for HDt HD . Modification of the HD with N-hydroxysuccinimide ␣ ␥ t t synergy with G tGTP S. This suggests a conformational (NHS) abrogated all enhancement of the PDE activation (Fig. dependence of the HD moiety for synergy of the PDE acti- ␣ ␥ 2 A and B). This result indicates that the HDt–PDE interaction vation by the RasD moiety of G tGTP S. This hypothesis is ␣ ␥ is conformation-dependent. difficult to assess because the RasD of G tGTP Sisinits Furthermore, both the HDt and the HDg directly interact activated conformation for binding to P␥. We have estimated ␣␤ with the PDE catalytic subunits (P ), as demonstrated by the half-saturating concentration of the isolated HDt at 300– ␣ ␥ 500 nM, which is near the half-saturation of G tGTP S †The linker 1 and linker 2 sequences of the designed helical domains activation in the absence of HDt. The contribution of the HDt ␣ are the first two amino acid residues (except for the HD of G g that moiety as a synergist of PDE activation is difficult to segregate ␣ possesses only one residue) at the amino terminus of A helix, and from the RasD interaction with P␥ subunit at these concen- the last seven residues at the carboxyl terminus of ␣F helix. ‡ ␣ trations. To examine the possibility of conformational depen- We have noted that G t, which was isolated by aluminum fluoride ␣ activation of transducin (47), although displaying a ‘‘basal’’ activity dence of the HD moiety of intact G t subunits, we designed state as assessed by inability to activate holoPDE, nevertheless fails experiments using rhodopsin to catalyze the conformational ␣ ␣ to reproduce G tGDP attenuation of tPDE. transitions of G t. The experiment in Fig. 4 compares the PDE Downloaded by guest on September 25, 2021 Biochemistry: Liu and Northup Proc. Natl. Acad. Sci. USA 95 (1998) 12881

FIG. 3. Enhancement of the G␣tGTP␥S-elicited PDE activation by HDt.(A)HDt enhancement of the G␣tGTP␥S affinity for the PDE. The activity of PDE (0.5 nM) was determined in the presence of the indicated concentrations of G␣tGTP␥S with (F) or without (E) addition of 1 ␮MHDt. The data are representative of three indepen- dent experiments. The basal activity of the PDE (4.7 nmol of cGMP) was subtracted for calculation of the curve fits. Curve fitting was performed with nonlinear least squares criteria by using Graphpad PRISM software. (B) Independence of the HDt and P␥ regulation of PDE. The activity of PDE (0.5 nM) was determined with the indicated concentrations of G␣tGTP␥S either alone (■) or in the presence of 750 pM P␥ (ᮀ), 1.0 ␮MHDt (F), or 750 pM P␥ with 1.0 ␮MHDt (E). The data are representative of three independent experiments. The inhib- itory P␥ subunit was prepared from the purified PDE according to the procedure described by Hurley and Stryer (21). The basal activity of the PDE (3.4 nmol of hydrolyzed cGMP) was subtracted from all values. ␣ ␥ activation produced by 125 nM G tGTP S to that produced by ␣ ␥ 125 nM G tGDP in the presence of rhodopsin and GTP S. If the only product of the rhodopsin-catalyzed reaction regulat- ␣ ␥ ing PDE activity was G tGTP S, the activity produced by ␣ ␥ rhodopsin with 125 nM G tGDP and GTP S must be less than ␣ ␥ or equal to that of the preformed 125 nM G tGTP S activity. As shown in Fig. 4, in the presence of rhodopsin and GTP␥S ␣ ␣ to catalyze the G t activation, 125 nM G tGDP is a more FIG. 2. Selective HD regulation of PDE. (A) Selectivity of HD potent activator of PDE than the same concentration of ␣ ␥ ␣ ␥ synergy with G tGTP S. The enzymatic activity of PDE (0.5 nM) was G tGTP S. The rhodopsin alone does not enhance the determined with the indicated additions to the reactions. The concen- ␣ ␥ ␣ ␮ G tGTP S-mediated activation, nor does G tGDP alone, but trations for HDi1,HDq,HDs, and NHS-HDt were 2.0 M, HDt and HDg HD does, demonstrating that its synergy is independent of were 1.0 ␮M, and G␣ GTP␥S was 0.45 ␮M. Error bars indicate ϮSEM t t rhodopsin catalysis. We found no activation of PDE by derived from values obtained in three independent experiments, with ␣ each determination performed in triplicate. PDE activity assay was G tGDP alone or in the presence of GDP and rhodopsin. This ␣ ␥ performed as described in Materials and Methods. The G␣tGTP␥S was result indicates that the formation of the activated G tGTP S, prepared from the purified G␣tGDP as described (47). (B) Selective interaction of the HDs with PDE catalytic core. Reactivity of the HD proteins isolated from different G␣ families toward PDE catalytic core concentrations of either HDt (F)orG␣tGDP (E). The activity for each was determined with 0.2 nM tPDE in the presence of the indicated HD condition is expressed as the fractional inhibition of the activity concentrations. Limited trypsinization to produce tPDE was performed assayed for the tPDE alone (34.2 Ϯ 1 nmol of cGMP hydrolyzed). The as described in Materials and Methods. Each value is the mean Ϯ SEM bar presents the fractional inhibition obtained in the presence of 4.5 of data obtained from three independent experiments. (C) Attenuation ␮MG␣tGDP and 5 ␮MHDt. The error bars are ϮSEM for values of the catalytic core enzyme by HDt and G␣tGDP. Attenuation of obtained in three independent experiments. The curves are represen- 0.2 nM tPDE was determined in the presence of the indicated tative of the saturations from three independent experiments. Downloaded by guest on September 25, 2021 12882 Biochemistry: Liu and Northup Proc. Natl. Acad. Sci. USA 95 (1998)

The attenuation of tPDE activity by HDt is, in essence, ␣ identical to the inhibition of tPDE by G tGDP, which has been ␣ reported (34). Neither HDt nor G tGDP inhibits the tPDE activity completely. In our experiments, these both saturated at 62% inhibition, whereas the prior reports found 60–70% inhibition. These data, showing partial inhibition of the tPDE ␣ activity, exclude the possibility that the HDt and G tGDP inhibit by competing with cGMP. Rather, these data suggest an allosteric site(s) on the catalytic core enzyme noncompetitively regulating PDE activity. Furthermore, we found nearly iden- ␣ tical apparent affinities for both HDt and G tGDP attenuation of the PDE catalytic core, and the two inhibitors were not additive. At saturation, the coaddition of both HDt and ␣ G tGDP attenuated tPDE activity to the same extent as either alone. Together, these data suggest that the attenuation of ␣ tPDE by G tGDP may be conferred by an interaction of its HD ␣ moiety. The attenuation of tPDE by G tGDP is conforma- ␣ ␥ tionally specific. G tGTP S does not inhibit tPDE (34). ␣ Rather, this conformation of G t is a competent form for the activation of holoPDE. The influence of HDt on the interac- ␣ tion of holoPDE with the activated conformation of G t is a profound enhancement of apparent affinity—over 150-fold. ␣ ␥ Because the extent of activation is identical by G tGTP S either alone or in the presence of HDt, we interpret this as an allosteric regulation of the PDE by HDt. The activation of PDE ␣ FIG. 4. Influence of rhodopsin on PDE activation by G t and HDt. via binding of the inhibitory P␥ to the RasD moiety of The enzymatic activity of 0.2 nM PDE was determined with the ␣ ␥ G tGTP S has been well established (36, 37). Our data indicated additions to the reactions. When present, rhodopsin was ␮ ␣ ␣ ␥ suggest that the efficiency of this process is improved by the added at 1 M, G tGDP and G tGTP S were 125 nM, and HDt was allosteric interaction of HD with P␣␤. Because the HD also 0.5 ␮M. For nucleotide exchange, 5 ␮MGTP␥Sor5␮M GDP were t t functions as an inhibitor of the catalytic core of PDE, and this added with rhodopsin as indicated. The PDE reactions were conducted ␣ for 15 min. Error bars indicate ϮSEM for values obtained in three reproduces identically the attenuation by G tGDP, we suggest ␣ independent experiments. The G␣tGDP was purified as described (32), that distinct conformations of the HD moiety in G t may be and its concentration was determined by rhodopsin-catalyzed GDP/ involved in the two functions. Indeed, it has been noted that [35S]GTP␥S exchange according to the modified method (19, 47). the so-called ‘‘switch IV’’ region within the HD undergoes conformational transition between GDP and GTP␥Sstatesof as expected, is required for the activation of PDE. The extent G␣ (15). Our examination of the conformational dependence of activation in the condition of rhodopsin-catalyzed activation of HD synergy suggests an additional conformation(s) pro- ␣ t of 125 nM G tGDP is nearly identical to the activation by 125 duced during the rhodopsin-catalyzed transitions of G␣ as the ␣ ␥ t nM G tGTP S with HDt. The fact that these activities are not basis for the synergistic activation. These data may reconcile limited by the capacity of the PDE is shown by the addition of the seemingly low affinity we report, as found by others ␮ ␣ ␥ ␣ 2 MHDt with 125 nM G tGTP S. Together, these data (40–42) for the PDE activation by G tGTP in solution exper- suggest that the process of rhodopsin-catalyzed GDP/GTP iments, as opposed to the affinities obtained in the presence of exchange produces a synergized activation of PDE by intact rod outer segment discs (43). Rhodopsin dynamically catalyzes ␣ ␣ G t subunits. the formation not only of activated G tGTP but also of all of ␣ the intermediate states of the G t, including those in which the DISCUSSION HD moiety may attain distinct conformation(s). We offer the possibility that a transient intermediate, such as the ‘‘empty ␣ In this work we have confirmed the earlier report that the HD state’’ of G t, provides a synergistic HD moiety, enhancing the ␣ ␣ portion of the G s could be expressed as a properly folded activation by the RasD moiety of the G tGTP, and that this structure autonomous of the amino terminus and Ras-domain conformation is more readily attained in the isolated HDt sequences of the intact protein structure (10, 12). Although the protein that we have tested. In this way, the two separate former report suggested that the HD may serve as an internal folding domains may cooperate in the overall efficiency of GTPase-activating protein for the G␣-Ras domain, we have effector regulation, the RasD being the activator and the HD identified an effector regulation by the isolated HD protein. a modulator of the RasD affinity for effector interaction. ␣ Our data demonstrate two HD-selective interactions with Furthermore, the conformational switch of the G t on GTP PDE by using the isolated proteins in solution. HDt synergis- hydrolysis produces a HD that acts as an attenuator of the tically activates the holoPDE enzyme in conjunction with activated PDE. ␣ activated Gt , and HDt attenuates the activity of the tryptically Our data clearly establish the interaction of the isolated HD activated catalytic core enzyme. Both regulatory effects are protein with the PDE, and suggest a synergism of the two selective for HDt and HDg; neither effect is elicited by the HDs folding domains in the regulation of the downstream effector. ␣ ␣ ␣ from s, q, or the closely related i1. These data reiterate the These data provide evidence for a function of the HD of a G well established selectivity in the G protein regulation of this protein, i.e., modulation of the effector regulation by the enzyme. The similarity between HDt and HDg agrees with the activated Ras domain. Because the overall folding motif is ␣ ␣ ␣ previous report that G g is functionally indistinguishable from essentially identical for the G t and G i proteins from the ␣ G t in biochemical assays (38). It has been noted that among solutions of their three-dimensional structures in crystals, and ␣ ␣ all of the known G protein subunits, the G g shares highest all G␣ proteins contain the HD, it is tempting to speculate that ␣ amino acid identity to rod and cone G t subunits (39). This a similar function has been conserved for each of the proteins. selectivity compellingly argues that the HD regulation is The recent determinations of the three-dimensional structures biologically appropriate rather than an artifact of the recom- for the soluble catalytic domains of adenylyl cyclases exhibit binant expression of these sequences. intrachain dimerization of a catalytic fold (44, 45). We have Downloaded by guest on September 25, 2021 Biochemistry: Liu and Northup Proc. Natl. Acad. Sci. USA 95 (1998) 12883

obtained hydrodynamic evidence that the retinal PL-C␤ ho- 19. Fawzi, A. B. & Northup, J. K. (1990) Biochemistry 29, 3804–3812. mologue from squid retina (46) is a dimer of the 140-kDa 20. Baehr, W., Devlin, M. J. & Applebury, M. L. (1979) J. Biol. Chem. catalytic chains (J.K.N., unpublished data). A common allo- 254, 11669–11677. steric regulation similar to the retinal PDE may well apply to 21. Hurley, J. B. & Stryer, L. (1982) J. Biol. Chem. 257, 11094–11099. 22. Deterre, P., Bigay, J., Forquet, F., Robert, M. & Chabre, M. these major target for G protein signaling. We would (1988) Proc. Natl. Acad. Sci. USA 85, 2424–2428. expect that the HD modulation mechanism would represent a 23. Chabre, M. & Deterre, P. (1989) Eur. J. Biochem. 179, 255–266. common signaling feature among the members of G protein 24. Stryer, L. (1991) J. Biol. Chem. 266, 10711–10714. families. Now that we know that a selectivity is encoded in the 25. McLaughlin, S. K., Mckinnon, P. J. & Margolskee, R. F. (1992) G␣-HD, we next want to know how it cooperates intramo- Nature (London) 357, 563–569. lecularly with G␣-RasD and intermolecularly with G␤␥ and 26. Raport, C. J., Dere, B. & Hurley, J. B. (1989) J. Biol. Chem. 264, other factors to regulate G protein signaling. 7122–7128. 27. Strathmann, M. & Simon, M. I. (1990) Proc. Natl. Acad. Sci. USA Note Added in Proof. The mechanistic implications of this work are the 87, 9113–9117. subject of an additional report (48). 28. Jones, D. T. & Reed, R. R. (1987) J. Biol. Chem. 262, 14241– 14249. 29. Itoh, H., Kozasa, T., Nagata, S., Nakamura, S., Katada, T., Ui, M., We thank Dr. Gerson H. Cohen for his help and access to his protein Iwai, S., Ohtsuka, E., Kawasaki, H., Suzuki, K. ,et al. (1986) Proc. database terminal; Dr. Howard Jaffe (National Institute of Neuro- Natl. Acad. Sci. USA 83, 3776–3780. logical Disorders and Stroke) for performing amino acid sequencing; 30. Lee, E., Linder, M. E. & Gilman, A. G. (1994) Methods Enzymol. and Dr. Pushkar Sharma for operation of circular dichroism spec- 237, 146–164. trometer. We thank Loren Chen for his able technical assistance and 31. Gillespie, P. G. & Beavo, J. A. (1989) Mol. Pharmacol. 36, Drs. William Clark, Xiaoying Jian, Paul Randazzo, and Nicholas Ryba 773–781. for numerous helpful discussions of this work. 32. Yamazaki, A., Hayashi, F., Tatsumi, M., Bitensky, M. W. & George, J. S. (1987) J. Biol. Chem. 265, 11539–11548. 1. Hamm, H. E. (1998) Annu. Rev. Biochem. 66, 669–672. 33. Wensel, T. G. & Stryer, L. (1986) Proteins Struct. Funct. Genet. 2. Bourne, H. R., Sanders, D. A. & McCormick, F. (1990) Nature 1, 90–99. (London) 348, 125–132. 34. Kroll, S., Phillips, W. J. & Cerione, R. A. (1989) J. Biol. Chem. 3. Sprang, S. R. (1997) Annu. Rev. Biochem. 66, 639–678. 264, 4490–4497. 4. Lambright, D. G., Noel, J. P., Hamm, H. E. & Sigler, P. B. (1994) 35. Fung, B. K.-K., Uoung, J. H., Yamane, H. K. & Griswold- Nature (London) 369, 621–628. Prenner, I. (1990) Biochemistry 29, 2657–2664. 5. Coleman, D. E., Berghuis, A. M., Lee, E., Linder, M. E., Gilman, 36. Skiba, N. P., Bae, H. & Hamm, H. E. (1996) J. Biol. Chem. 271, A. G. & Sprang, S. R. (1994) Science 265, 1405–1412. 413–424. 6. Conklin, B. R. & Bourne, H. R. (1993) Cell 73, 631–641. 37. Liu, Y., Arshavsky, V. Y. & Ruoho, A. E. (1996) J. Biol. Chem. 7. Clapham, D. E. & Neer, E. J. (1997) Annu. Rev. Pharmacol. 271, 26900–26907. Toxicol. 20, 167–203. 38. Hoon, M. A., Northup, J. K., Margolskee, R. F. & Ryba, N. J. P. 8. Masters, S. B., Stroud, R. M. & Bourne, H. R. (1986) Protein Eng. (1995) Biochem. J. 309, 629–636. 1, 47–54. 39. Margolskee, R. F. (1993) BioEssays 15, 645–650. 9. Noel, J. P., Hamm, H. E. & Sigler, P. B. (1993) Nature (London) 40. Malinski, J. A. & Wenesl, T. G. (1992) Biochemistry 31, 9502– 366, 654–663. 9512. 10. Markby, D. W., Onrust, R. & Bourne, H. R. (1993) Science 262, 41. Artemyev, N. O., Rarick, H. M. Mills, J. S. Skiba, N. P. & Hamm, 1895–1901. H. E. (1992) J. Biol. Chem. 267, 25067–25072. 11. Landis, C. A., Masters, S. B., Spada, A., Pace, A. M., Bourne, 42. Bennett, N. & Clerc, A. (1989) Biochemistry 28, 7418–7424. H. R. & Vallar, L. (1989) Nature (London) 340, 692–696. 43. Fung, B. K.-K. & Stryer, L. (1980) Proc. Natl. Acad. Sci. USA 77, 12. Benjamin, D. R., Markby, D. W., Bourne, H. R. & Kuntz, I. D. 2500–2504. (1995) J. Mol. Biol. 254, 681–691. 44. Tesmer, J. J. G., Sunahara, R. K., Gilman, A. G. & Sprang, S. R. 13. Antonelli, M., Birnbaumer, L., Allende, J. E. & Olate, J. (1994) (1997) Science 278, 1907–1916. FEBS Lett. 340, 249–254. 45. Zhang, G., Liu, Y., Ruoho, A. E. & Hurley, J. H. (1997) Nature 14. Codina, J. & Birnbaumer, L. (1994) J. Biol. Chem. 269, 29339– (London) 386, 247–253. 29342. 46. Mitchell, J., Gutierrez, J. & Northup, J. K. (1995) J. Biol. Chem. 15. Mixon, M. B., Lee, E., Coleman, D. E., Berghuis, A. M., Gilman, 270, 854–859. A. G. & Sprang, S. R. (1995) Science 270, 954–960. 47. Fawzi, A., Fay, D. S., Murphy, E. A., Tamir, H., Erdos, J. J. & 16. Rodbell, M. (1992) Top. Cell. Regul. 32, 1–47. Northup, J. K. (1991) J. Biol. Chem. 266, 12194–12200. 17. Yarfitz, S. & Hurley, J. B. (1994) J. Biol. Chem. 269, 14329–14332. 48. Liu, W., Clark, W. A., Sharma, P. & Northrup, J. K. (1998) J. Biol. 18. Stryer, L. (1986) Annu. Rev. Neurosci. 9, 87–119. Chem., in press. Downloaded by guest on September 25, 2021