Regulatory Interactions between the Amino Terminus of G-protein βγ Subunits and the Catalytic Domain of Phospholipase Cβ2*

We previously identified a 10-amino acid region from the Y domain of phospholipase Cβ2 (PLCβ2) that associates with G-protein βγ subunits (Sankaran, B., Osterhout, J., Wu, D., and Smrcka, A. V. (1998) J. Biol. Chem. 273, 7148–7154). We mapped the site for cross-linking of a synthetic peptide (N20K) corresponding to this Y domain region to Cys25 within the amino-terminal coiled-coil domain of Gβγ (Yoshikawa, D. M., Bresciano, K., Hatwar, M., and Smrcka, A. V. (2001) J. Biol. Chem. 276, 11246–11251). Here, further experiments with a series of variable length cross-linking agents refined the site of N20K binding to within 4.4–6.7 Å of Cys25. A mutant within the amino terminus of the Gβ subunit, Gβ1(23–27)γ2, activated PLCβ2 more effectively than wild type, with no significant change in the EC50, indicating that this region is directly involved in the catalytic regulation of PLCβ2. This mutant was deficient in cross-linking to N20K, suggesting that a binding site for the peptide had been eliminated. Surprisingly, N20K could still inhibit Gβ1(23–27)γ2-dependent activation of PLC, suggesting a second N20K binding site. Competition analysis with a peptide that binds to the Gα subunit switch II binding surface of Gβγ indicates a second N20K binding site at this surface. Furthermore, mutations to the N20K region within the Y-domain of full-length PLCβ2 inhibited Gβγ-dependent regulation of the enzyme, providing further evidence for aGβγ binding site within the catalytic domain of PLCβ2. The data support a model with two modes of PLC binding to Gβγ through the catalytic domain, where interactions with the amino-terminal coiled-coil domain are inhibitory, and interactions with the Gα subunit switch II binding surface are stimulatory.

Activation of G-protein-coupled receptors releases the G␤␥ subunit from G␣-GTP, leaving G␤␥ free to interact with effector molecules such as enzymes or ion channels (1)(2)(3). A particular group of enzymes regulated by G␤␥ belong to the phosphoinositide-specific phospholipase C␤ (PLC␤) 1 class. Once activated, PLC␤ cleaves phosphatidylinositol 4,5-bisphosphate (PIP 2 ) into the two second-messengers, diacylglycerol and inositol 1,4,5-trisphosphate. All PLC␤ isoforms have an N-terminal pleckstrin homology (PH) domain, four sets of EF-hand domains, X and Y domains comprising the catalytic center, and a C2 domain followed by a C-terminal extension (4). There are four isoforms of PLC␤, ␤1-4; however, only PLC␤2 and PLC␤3 are regulated by ␤␥ subunits (5).
G-protein ␤ subunits have two distinct domains, an N-terminal ␣ helix, which forms a coiled-coil interaction with the ␥ subunit, and a seven-blade ␤-propeller structure composed of seven WD repeating motifs (6,7). Amino acids within blades one, two, and three make extensive contacts with ␣-GDP (7,8).
Since ␣ subunits block G␤␥-dependent regulation of all effectors, effector-binding sites on G␤␥ were predicted to overlap with the ␣ subunit binding site. Indeed, mutagenesis of particular amino acids on G␤␥ important for ␣ subunit binding blocked activation of PLC␤2 (9,10). Furthermore, mutations to the outer strands of blades 2, 6, and 7, which do not make direct contact with G␣, rendered G␤␥ unable to effectively activate PLC␤2 (11).
The mechanism for regulation of PLC␤2 by G␤␥ subunits is not fully understood. Approaches for elucidating the activation mechanism have been to characterize the effects of a variety of mutants, peptides, and fusion proteins on regulation of PLC␤2 by G␤␥. Two major regions on PLC␤2 have been implicated as G␤␥-binding sites, the N-terminal PH domain and a portion of the catalytic Y domain. A large body of evidence supports a role for interactions between the catalytic domain of PLC␤2 and G␤␥ subunits. Overexpression of the Y domain in COS-7 cells blocked receptor-mediated G␤␥-dependent, but not G␣ qdependent activation of PLC. A purified glutathione S-transferase fusion protein comprising amino acids 526 -641 of the Y-domain bound directly to purified G␤␥ subunits (12). Overlapping peptides representing a portion of the Y-domain directly cross-linked to both G␤ and G␥ subunits with the heterobifunctional cross-linker SMCC, and this was blocked by purified PLC␤2 but not ␣ i . The peptides also inhibited G␤␥ regulation of PLC␤2 with an EC 50 of 30 -50 M (13). The crystal structure of PLC␦ reveals that the region implicated by the peptide studies corresponds to the ␣ 5 helix on the surface of the catalytic domain and therefore is accessible to interact with G␤␥ subunits (14). We hypothesized that direct binding of this helix within the catalytic domain to G␤␥ is involved in G␤␥-dependent PLC␤2 activation.
To define the site cross-linking of one of the catalytic domain peptides, N20K (amino acids 565-574 in PLC␤2), to the G␤ subunit and identify a potential Y-domain interaction site on G␤, we used a systematic cysteine mutagenesis approach where each cysteine on G␤ 1 was individually mutated to alanine. The cysteine residue critical for the majority of peptide cross-linking was cysteine 25 within the amino-terminal coiledcoil region of G␤ (15). This region is outside the G␣ subunit-binding site, complementing data demonstrating that G␣ i1 could not block cross-linking of N20K to G␤␥ (13,15). Previous studies have not implicated the amino terminus of G␤␥ as an important effector signaling site in mammals, although it has been implicated to be important in the yeast pheromone pathway (16,17).
These data suggest that the amino-terminal coiled-coil of G␤␥ directly binds to the ␣ 5 helix of the Y-domain to regulate PLC activity. The goals of this study were 3-fold: 1) to further delineate the specific region within the amino terminus of G␤␥ subunits that interact with PLC␤; 2) to demonstrate a functional role for the amino terminus of G␤ in mammalian effector regulation; and 3) to demonstrate the functional importance of the N20K region within full-length PLC for G␤␥ regulation. These results would be the first demonstrating a specific region on PLC interacting with a specific region on G␤␥, leading to a better understanding of how G␤␥ subunits regulate PLC.

EXPERIMENTAL PROCEDURES
Materials-Peptides (N20K, Ac-NRSYVISSFTELKAYDLLSK; SIGK, Ac-SIGKAFKILGYPDYD; L9A, Ac-SIRKALNIAGYPDYD) were purchased from Alpha Diagnostic International (San Antonio, TX) and had purity greater than 90% based on HPLC chromatography analysis. The masses of the peptides were confirmed by mass spectroscopy. All crosslinking reagents used in this study were from Pierce. Ni 2ϩ -nitrilotriacetic acid resin was from Qiagen. [ 3 H]PIP 2 and [ 32 P]NAD were from PerkinElmer Life Sciences. Pertussis toxin was purchased from List Biologicals. All molecular biology reagents were from Invitrogen, unless otherwise indicated. Baculoviruses encoding wild type G␤ 1 , His 6 -G␣ i , and G␥ 2 were obtained from the laboratory of Dr. Alfred Gilman.
Construction of G␤ Subunit and PLC␤2 Mutants-Mutants in rat G␤ 1 and in human PLC␤2 were constructed by overlap-extension PCR with Pfu polymerase (Stratagene) using standard protocols. Final constructs were sequenced to confirm the presence of the mutation. Bacoluviruses were generated as per the manufacturer's instructions (Invitrogen).
Purification of ␤␥ Subunits-G␤ 1 ␥ 2 subunits were purified from 2 liters of Sf-9 cells triply infected with His 6 -G␣ i1 , wild type or mutant G␤ 1 subunits, and G␥ 2 subunits essentially as described (18) with the following changes. Sf-9 cell membranes were prepared as described (18) and then extracted in buffer A containing 50 mM HEPES, pH 8.0, 3 mM MgCl 2 , 10 mM ␤-mercaptoethanol, 10 M GDP, 50 mM NaCl, 1% polyoxyethylene 10 lauryl ether (C 12 E 10 ), and a protease inhibitor mixture for 2 h at 4°C. Detergent-extracted proteins were diluted 5-fold with buffer B, 20 mM HEPES, pH 8.0, 1 mM MgCl 2 , 10 mM ␤-mercaptoethanol, 10 M GDP, 100 mM NaCl, 0.5% C 12 E 10 , and a proteaseinhibitor mixture and loaded onto a 4-ml Ni 2ϩ -nitrilotriacetic acidagarose column at 4°C overnight. The column was washed with 100 ml of buffer B containing 300 mM NaCl and 5 mM imidazole, warmed at room temperature for 15 min, and washed with 12 ml of the same buffer followed by 5 ml of buffer B containing 300 mM NaCl, 5 mM imidazole, and 1% n-octyl-␤-D-glucopyranoside (OG). G␤␥ subunits were eluted with 3 column volumes of buffer C, 20 mM HEPES, pH 8.0, 10 M GDP, 50 mM NaCl, 1% OG, 3 mM imidazole, 50 mM MgCl 2 , 30 M AlCl 3 , and 10 mM NaF followed by 4 volumes of buffer C with 1% cholate instead of OG. With this modified procedure, endogenous Sf-9 G␤␥ elutes in the first fraction, and the expressed G␤␥ elutes in fractions 2-5. Peak elutions, devoid of Sf-9 endogenous G␤␥, were combined, concentrated on a 0.5-ml hydroxyapatite column, and eluted with 20 mM HEPES, pH 8.0, 100 mM NaCl, 1 mM DTT, 1% OG, and 200 mM KP i , pH 8.0. Protein concentrations were determined by an Amido Black assay. To confirm purity, 500 ng of purified wild type or mutated G␤␥ subunits were loaded onto a 12% SDS-polyacrylamide gel and stained with Coomassie Blue G250 stain. To confirm proper assembly of the mutant G␤ subunits with G␥ subunits, equal amounts of wild type and mutant G␤␥ subunits were run on a 17% SDS-PAGE gel, transferred to nitrocellulose, and blotted with G␥ subunit antibody, X263.
Purification of Wild Type and Mutant PLC-Wild type His 6 -PLC␤2 and His 6 -PLC␤3 were purified from 1 liter of Sf-9 cells as described (19). These wild type PLC␤s were used for measuring activation by mutated G␤␥ subunits. Mutant PLCs were purified using a protocol similar to the wild type 1-liter protocol from 50-ml cultures of Sf-9 cells, except buffer volumes were scaled down, and 1 ml of nickel resin was used (the heparin column was omitted). For experiments comparing the activities of PLC␤2 mutants, wild type and mutant enzymes were purified simul-taneously and in parallel, snap-frozen in liquid N 2 in single use aliquots, and stored at Ϫ70°C.
Phospholipase C Assays-PLC assays were performed as described (19). The concentrations of PLC, G␤␥, and detergent are indicated in the figure legends. Assays measuring peptide or G␣ subunit inhibition of G␤␥-dependent PLC activation were performed exactly as described (13). Assays measuring G␣ q activation of PLC were performed as described (5). In all experiments, a maximum of 20% of the PIP 2 substrate was hydrolyzed.
Measurement of ADP-ribosylation of G␣ i1 -Pertussis toxin-mediated ADP-ribosylation of 15 pmol of G␣ i1 in the presence of G␤␥ was measured as described (20). Myr-G␣ i was purified from Escherichia coli as described (21).
Chemical Cross-linking-All cross-linkers used in this study, succinimidyl 4-[N-maleimidomethyl]-cyclohexane-1-carboxylate (SMCC), N-[⑀-maleimidocaproyloxy]succinimide ester, N-[␥-maleimidobutyryloxy]succinimide ester, N-[␣-maleimidoacetoxy]succinimide ester, and Nsuccinimidyl iodoacetate, were heterobifunctional, containing both sulfhydryl and amine reactive groups. The functional groups for all crosslinkers except N-succinimidyl iodoacetate had N-hydroxysuccinimideester and maleimide reactive moieties; N-succinimidyl iodoacetate had an iodoacetate group instead of a maleimide. All of the peptides used contained no cysteine residues and were N-terminally acetylated, so cross-linking was through the ⑀-amine of lysine residues. To prevent cross-linking of G␤ to G␥, peptides were preincubated with cross-linker prior to the addition to G␤␥ by use of the following protocol. Peptides at 2 times their final concentration (see figure legends for concentrations) were incubated with 400 M cross-linker in cross-linking buffer containing 50 mM HEPES, pH 7.4, 75 mM NaCl, 1 mM MgCl 2 , and 0.1% C 12 E 10 at room temperature for 10 min. Ethanolamine, pH 8, was then added to 100 mM to quench all unreacted N-hydroxysuccinimide-ester groups. 50 l of the peptide/cross-linker mix was added to 50 l of 60 nM G␤␥ diluted in cross-linking buffer and allowed to incubate at room temperature before quenching with 33 l of 4ϫ sample buffer (4ϫ sample buffer: 125 mM Tris, pH 6.8, 20% glycerol, 4% SDS, 1.4 M ␤-mercaptoethanol, and bromphenol blue) (times for each reaction are listed in the figure legends). The final concentrations were 30 nM G␤␥ and 200 M cross-linker; the peptide concentrations are listed in the figure legends. 10 l of the reactions were loaded onto a 12% SDS-polyacrylamide gel, transferred to nitrocellulose, and probed with a G␤ subunit-specific antibody B-600 followed by an anti-rabbit horseradish peroxidase-conjugated antibody. Proteins were visualized using enhanced chemiluminescence detection reagents (PerkinElmer Life Sciences).

Delineation of the N20K Binding Site on G␤␥-Previous
studies in our laboratory demonstrated two overlapping peptides from the catalytic domain of PLC␤2, N20K, and E20K (amino acids 564 -584 and 574 -594, respectively) could be chemically cross-linked to G␤ subunits (13,15). Based on this, and other studies, we hypothesized the 10-amino acid region of overlap between these peptides, ELKAYDLLSK (amino acids 574 -584), representing an ␣ helix on the surface of PLC␤2, directly bound to G␤␥ subunits. To define this potential PLC␤2 interaction site on G␤␥, we mapped a cross-linking site for the N20K peptide using a cysteine to alanine mutagenesis approach (15). From this data, it was determined the sulfhydryl group of cysteine 25 is within reach of the peptide-binding site. However, since the cross-linker, SMCC, had a spacer arm length of 12 Å, we could only conclude the peptide was binding within a 15-Å radius of cysteine 25 (the length of the spacer arm in the SMCC cross-linker plus the length of a lysine side chain). To further refine the nature of the N20K interaction within G␤␥, we determined which residue(s) in N20K was involved in cross-linking. Since N20K has no cysteine residues and is acetylated at the amino terminus, the cross-linking must be through a one or both of the two ⑀-amine moieties on the lysine residues within the last 10 amino acids of N20K. Two N-terminally acetylated peptides were constructed, R1 and R2, where each lysine residue was individually replaced with arginine (Fig. 1A), abolishing a cross-linkable primary amine but retaining the peptide's charge. Both R1 and R2 cross-linked to ␤␥ with SMCC ( Fig. 1B), indicated by the appearance of a higher molecular weight immunoreactive G␤ subunit band only in the presence of peptide and cross-linker, although neither was as effective as N20K. Cross-linking of both of these peptides to G␤ 1 subunits where Cys 25 was mutated to alanine (G␤ 1 (C25A)) was substantially inhibited.
Purification of Amino-terminal Site-directed Mutants of ␤␥ Subunits near Cys 25 -The cross-linking data suggest a PLC binding site in the amino terminus of G␤␥ within 7 Å of cysteine 25. Therefore, surface amino acids in the coiled-coil region of G␤␥ within 10 Å of cysteine 25 were chosen for site-directed mutagenesis to confirm a functional role for the G␤␥ N terminus in PLC regulation. Six amino acids on G␤ and six amino acids on G␥ neighboring cysteine 25 were mutated three amino acids at a time. The mutations were named G␤ 1 (17)(18)(19)(20)(21) and G␤ 1 (23-27) based on the stretches of amino acids where the mutations were located ( Fig. 2A). Similar mutations made in G␥ 2 failed to dimerize with G␤ 1 and were not analyzed further. Wild type and mutant G␤␥ dimers were purified simultaneously and in parallel to near homogeneity in one step based on binding to His 6 -G␣ i and elution by activation of the G␣ subunit with AlF 4 Ϫ (Fig. 2B). This purification strategy insured that the mutants were properly assembled, folded, and functional. Proper assembly with the G␥ subunit was confirmed by immunoblotting for the presence of G␥ subunits in the purified mutant preparations (Fig. 2C). The proteins were Ͼ90% pure based on Coomassie Blue staining. No detectable endogenous Sf-9 G␤ (which runs at a higher molecular weight than overexpressed mammalian G␤ 1 ) was detected on either a Coomas-sie Blue-stained gel or immunoblots with an antibody that recognizes both endogenous Sf-9 and overexpressed G␤ 1 .
Activation of PLC␤ by Wild Type and Mutated G␤␥ Subunits-To demonstrate a functional role for the amino-terminal amino acids of G␤␥ in mammalian PLC␤2 regulation, the purified mutated dimers were tested for activation of PLC␤2. Surprisingly, G␤ 1 (23-27)␥ 2 activated PLC␤2 2-4-fold more effectively than wild type with no significant change in the EC 50 ( Fig. 3A and Table I). There was no significant difference in either the -fold activation or the EC 50 for activation by G␤ 1 (17-21) (Fig. 3A and Table I). Activation of PLC␤2 was abolished by boiling the mutant prior to performing the assay, indicating that the enhancement was not due to a nonprotein contaminant within the buffer (data not shown). The addition of GDP-loaded G␣ i blocked activation of PLC␤2 by G␤ 1 (23-27)␥ 2 , ensuring that the enhancement was due to the addition of G␤␥ and not a contaminating protein (Fig. 3B). The addition of purified PLC was required to observe accumulation of inositol 1,4,5-trisphosphate, indicating that no endogenous Sf-9 phospholipases were co-purified with the mutant G␤␥ (data not shown). These results were repeated with three different preparations of G␤ 1 (23-27)␥ 2 and two different preparations of wild type G␤ 1 ␥ 2 , demonstrating the reproducibility of the results with different batches of protein. As a final confirmation that both proteins were properly folded and functional postpurification, the ability of both G␤ 1 ␥ 2 and G␤ 1 (23-27)␥ 2 to support pertussis toxin-mediated ADP-ribosylation of G␣ i was tested. Both proteins supported ADP-ribosylation of G␣ i with no significant difference in either potency or efficacy (Fig. 3C). These data confirm that Cross-linking reactions were performed as described under "Experimental Procedures." Reactions were performed with SMCC and were carried out for 1.5 min at room temperature before quenching with sample buffer. The final concentration of the peptide was 30 M. C, cross-linking of R1 and R2 to wild type G␤ 1 ␥ 2 using cross-linkers with various lengths between functional groups. Reactions were performed as in B. The length of the spacer arm is given below the cross-linker.

FIG. 2. Construction and purification of N-terminal ␤ subunit mutants.
A, locations of the mutations are highlighted on the crystal structure of G␤␥ as determined by Sondek et al. (6). The sulfhydryl moiety of cysteine 25 is shown in dark blue, G␤ 1 (17)(18)(19)(20)(21) is shown in red, and G␤ 1 (23)(24)(25)(26)(27) is shown in purple. The changes in the amino acid sequences are shown on the right. Dashes indicate no substitution at that position. B, to ensure purity and confirm concentrations, 500 ng of purified wild type or mutated G␤ 1 ␥ 2 dimers were analyzed on a 12% SDS-polyacrylamide gel and stained with Coomassie Blue G250. C, to confirm assembly with ␥ subunits, 10 ng of the G␤ 1 ␥ 2 subunit preparations were resolved on a 17% SDS-polyacrylamide gel, transferred to nitrocellulose, and probed with the ␥ subunit antibody X-263.
both proteins are properly folded and functional and that there are no errors in the protein concentrations.
To determine whether this enhancement of activity was specific for PLC␤2, activation of PLC␤3 by mutant and wild type G␤␥ was tested. In this case, both WT and G␤ 1 (23-27)␥ 2 activated PLC␤3 to a similar extent, but ␤ 1 (23-27)␥ 2 was nearly 2-fold more potent than wild type ( Fig. 3D and Table I). These data strongly support a function of the amino terminus in the regulation of both PLC␤2 and PLC␤3. In both cases, the role of the amino terminus appears to be inhibitory, although the mechanisms are different. Others have reported differential activation of PLC␤2 and PLC␤3 by G␤␥ subunits (10,11,22).
Interaction between N20K and Mutated ␤␥ Dimers-The amino acids chosen for mutation were amino acids that, due to their close proximity to Cys 25 , were likely candidates to directly interact with the N20K peptide. To determine whether the G␤ subunit mutations eliminated N20K peptide binding, we tested whether N20K could still cross-link to the amino-terminal mutants. G␤ 1 (17-21)␥ 2 did cross-link to N20K, although higher concentrations of peptide were required for cross-linking compared with wild type, whereas G␤ 1 (23-27)␥ 2 substantially lost its ability to cross-link to N20K (Fig. 4A). Normally, N20K cross-linking to G␤␥ generates two higher molecular weight bands. N20K cross-linking to G␤ 1 (23-27)␥ 2 produces only one faint higher molecular weight band. We predicted that if the functional effects of N20K on activation of PLC␤ by G␤␥ subunits were through binding at the amino terminus, then elimination of this binding site would eliminate N20K inhibition of G␤␥-dependent activation of PLC␤2. Unexpectedly, N20K still inhibited G␤ 1 (23-27)␥ 2 activation of PLC␤2 with a similar potency as wild type. G␤ 1 (23-27)␥ 2 -dependent activation of PLC was inhibited by 50%, whereas wild type G␤␥-dependent activation was inhibited 85%. The inability of N20K to fully inhibit PLC activation by G␤ 1 (23-27)␥ 2 may be due to the increased efficacy of G␤ 1 (23-27)␥ 2 . Taken together, the crosslinking data and the peptide inhibition data suggest that there may be two binding sites for N20K on G␤␥, one within the amino terminus of G␤␥ eliminated by the G␤ 1 (23-27)␥ 2 mutation and one remaining intact. The ability of N20K to inhibit G␤ 1 (23-27)␥ 2 activation of PLC would be through binding to this second site. From these experiments, we hypothesize that Y-domain binding to the amino terminus of G␤␥ inhibits PLC activity, whereas Y-domain binding to this second site increases PLC activity.  3. Activation of PLC by wild type and mutated ␤␥ subunits. A, wild type and mutated G␤ 1 ␥ 2 subunits were incubated with 2 ng of PLC␤2 and sonicated lipid vesicles containing 25 M PIP 2 and 100 M liver phosphatidylethanolamine. All reactions contained 0.06% octyl glucoside. Reactions were initiated by the addition of 100 nM free Ca 2ϩ and transfer to a 30°C water bath for 7 min before quenching by the addition of ice-cold 10% trichloroacetic acid. Data shown are mean Ϯ S.E. of duplicate determinations from one of three (G␤ 1 (17)(18)(19)(20)(21)) or seven (G␤ 1 (23)(24)(25)(26)(27)) independent experiments. B, 100 nM G␣ i1 -GDP blocked the activation of PLC␤2 by 100 nM G␤ 1 (23)(24)(25)(26)(27). Data are from one of two independent experiments. C, G␤ 1 (23-27)␥ 2 supported ADP-ribosylation of G␣ i1 . 15 pmol of purified G␣ i1 was incubated with various amounts of wild type or mutated G␤ 1 ␥ 2 subunits in the presence of pertussis toxin and [ 32 P]NAD. Data shown are mean Ϯ S.E. of duplicate determinations. This experiment was repeated three times with similar results. D, wild type and mutated G␤ 1 ␥ 2 subunits were incubated with 5 ng of PLC␤3 as in A, except reactions were at 30°C for 10 min. The data are mean Ϯ S.E. of duplicate determinations from a representative experiment. This experiment was repeated three times with similar results. All curves were fit with a single site binding hyperbolic function using Graph Pad Prism Software. IP 3 , inositol 1,4,5-trisphosphate.
Defining a Second Site for N20K Interaction-Cross-linking data presented here and in our previous publications support the notion of two N20K binding sites. Cross-linking of N20K to G␤␥ produces two higher molecular weight bands (Figs. 1B and 4A). The first higher molecular weight band is G␤ cross-linked to one N20K peptide. The nature of the second band had been unclear, but its appearance was prevented by preincubation with purified PLC␤ or mutation of Tyr 15 to Gln in the peptide (13), indicating that it was not nonspecific. Moreover, replacement of Cys 25 with alanine still produced one faint higher molecular weight band (15). We propose that this second higher molecular weight band is G␤ cross-linked to two N20K peptides, one at the amino terminus and one elsewhere. We believe the residual cross-linking of peptide to the G␤ 1 (23-27) mutant and G␤ 1 (C25A) is through binding at the second site.
In order to define the second N20K binding site, we utilized the G␤ 1 (C25A) mutant. Our previous data showed that N20K cross-linked weakly to G␤ 1 (C25A)␥ 2 . Since there are no cysteine residues near Cys 25 capable of interacting with the crosslinker, G␤1(C25A) should have abolished all cross-linking if the N terminus was the sole site of binding. This finding indicates that there is a second N20K binding site outside of the amino terminus, and binding to this site results in the residual crosslinking to G␤ 1 (C25A). N20K has sequence similarity to G␤␥ binding peptides identified in a random peptide phage display screen conducted in our laboratory (23). This sequence similarity led us to hypothesize that these peptides were binding to a similar site on G␤␥ as the Y-domain of PLC (23). Recently, a binding site for one of these peptides was clearly defined in the x-ray crystal structure of one of these peptides, SIGKAF-KILGYPDYD (SIGK), bound to G␤ 1 ␥ 2 . 2 SIGK was identified in a doping mutagenesis screen to identify derivatives of one of the original peptides obtained in the phage display screen, SIRKALNILGYPDYD (SIRK). SIGK binds to G␤␥ with higher affinity than SIRK but retains most of the characteristics of SIRK. 3 The SIGK peptide bound to the top surface of G␤␥ overlapping with the switch II binding site of G␣. To determine whether this site was the second N20K binding site, we tested the ability of SIGK to compete away cross-linking of N20K to G␤␥ (SIGK itself does not cross-link to G␤␥ under these conditions). Fig. 5A shows that replacement of Cys 25 with alanine greatly reduces, but does not completely eliminate, cross-linking of N20K, confirming our previous studies (due to poor resolution of the second higher molecular weight band, we are focusing only on the lower band, which is a combination of G␤ cross-linked to N20K at either one of the two sites). The addition of SIGK at concentrations above the K d (ϳ1-5 M) partially reduces N20K cross-linking to wild type G␤␥, suggesting that binding to one of the two sites has been blocked by SIGK, but one remains intact. However, incubation with SIGK completely eliminated cross-linking of N20K to ␤ 1 (C25A)␥ 2 (Fig. 5,  A and B). The addition of a control peptide that does not bind ␤␥, L9A (23), has no effect on N20K cross-linking (Fig. 5A). These data demonstrate that N20K has two distinct binding sites on G␤, one within the amino terminus near Cys 25 and one at the G␣ interface, at the SIGK binding site.
We hypothesized that N20K blocked G␤ 1 (23-27)␥ 2 -dependent activation of PLC␤2 by interacting with the site outside of the amino-terminal binding site. Our cross-linking studies presented in Fig. 5, A and B, show that this second site overlaps with the SIGK peptide binding site; therefore, SIGK should also block PLC␤ activation by both wild type G␤␥ and G␤ 1 (23-27)␥ 2 subunits. Fig. 5C demonstrates that the SIGK peptide blocked G␤␥-dependent activation of PLC␤2 by both wild type and G␤ 1 (23-27)␥ 2 with no significant change in either the maximal inhibition or the IC 50 . From this finding, it is reasonable to conclude that the ability of N20K to block activation of PLC␤2 by G␤ 1 (23-27)␥ 2 was through binding to this site.
Regulation of Catalytic Domain Mutants of PLC␤2 by G␤␥ Subunits-This study and previous data suggesting that this peptide from the Y-domain of PLC binds to two sites are taken as evidence that the ␣ 5 helix on PLC represented by this peptide can bind to these two sites. If N20K is truly a model of the Y-domain ␣ 5 helix in full-length PLC, then mutations of the region corresponding to N20K within full-length PLC should have altered G␤␥-dependent activation. We constructed three triple alanine mutants (Fig. 6A) corresponding to the last 10 amino acids of N20K and the ␣ 5 helix in full-length PLC␤2. The basal Ca 2ϩ -dependent activities of all three PLC␤2 mutants were similar or slightly less than wild type (Fig. 6B), indicating that they were properly folded and active. The three PLC␤2 mutants were then tested for their ability to be activated by saturating concentrations of G␤ 1 ␥ 2 . PLC␤2(ELK/ AAA) had a significant decrease in its G␤␥-dependent activation as compared with wild type. PLC␤2(YDL/AAA) was also significantly deficient, whereas PLC␤2(LSK/AAA) was not significantly different from wild type. Full titration curves for G␤␥-dependent activation of PLC␤2(ELK/AAA) indicate that this mutant was activated very poorly at all concentrations tested (Fig. 6D). To confirm that the G␤␥-dependent activation resulted from specific alteration of G␤␥-dependent PLC activation, activation by G␣ q was tested. Previous studies have shown that the regions in PLC␤ necessary for G␣ q -dependent activation are different from those required for G␤␥ (25). The data in Fig. 6E show that PLC␤2(ELK/AAA) was activated to a greater extent than wild type PLC␤2, indicating that this mutant has no global deficiency in the catalytic machinery or in the ability to be activated in general. The nature of the enhanced activation by G␣ q is unknown and will be the subject of further investigation. Overall, these data demonstrate that there is a functional interaction between G␤␥ subunits and the ␣ 5 helix of PLC␤2.

DISCUSSION
In this work, we have reinforced and extended previous data suggesting that the catalytic domain of PLC␤2 can bind to the amino-terminal coiled-coil region of G␤␥ subunits near cysteine 25. Importantly, the mutagenic analysis demonstrates a functional role for interactions between the catalytic domain of PLC␤2 and both the N terminus and propeller regions of G␤␥. The amino-terminal region we identified in mammalian G␤␥ is the same region previously identified as an effector binding site in yeast G␤␥ subunits involved in the pheromone response pathway (17,26). Another study with single point mutations in the N terminus of mammalian ␤ failed to produce a mutant deficient in G␤␥-dependent effector activation when transfected into COS-7 cells, but interestingly, one mutant, K20A, showed a slight gain of function with respect to c-Jun N-termi- nal kinase activation but not activation of PLC␤2 (22). This study did not include mutations to Lys 23 or Ala 24 , which were mutated in our analysis. Other mutagenesis studies of G␤ subunits did not test for effects of mutations in the aminoterminal coiled-coil region, focusing on regions at the G␣ subunit interface and other amino acids within the propeller region (9 -11, 27).
Despite apparently eliminating a binding site for N20K in the G␤ 1 (23)(24)(25)(26)(27) mutant, N20K still inhibited the ability of this mutant to activate PLC␤2 with a similar potency as wild type (Table I, Fig. 4B), indicating that N20K must still be associating with the mutant. To explain this result, we hypothesize that there are two binding sites for N20K on the G␤ subunit, one eliminated by the G␤ 1 (23-27)␥ 2 mutation and one still intact. This hypothesis is supported by our cross-linking studies in which both G␤(C25A)␥ 2 and G␤ 1 (23-27)␥ 2 still displayed some cross-linking to N20K (Figs. 5A and 4A, respectively). Any cross-linking to G␤(C25A) must be through binding of the peptide to a second binding site on G␤␥. Since the cross-linking of G␤ 1 (23-27)␥ 2 is similar to G␤(C25A), we suspected that this residual cross-linking is due to N20K binding at the site outside of the amino terminus. Our previous work (13,15) and Fig. 1B support this idea, showing that cross-linking of wild type G␤␥ to N20K leads to two cross-linked species on a Western blot. The lowermost band is uncross-linked G␤. We believe that the first cross-linking band corresponds to G␤ cross-linked to one N20K peptide at one of two binding sites, and the upper band is G␤ cross-linked to two N20K peptides. This band is not G␤ cross-linked to G␥ due to the nature of the cross-linking protocol and confirmation by immunoblotting (data not shown; see Ref. 13).
We recently identified a family of peptides in a random peptide phage display screen that have homology to N20K. The binding site for a peptide derived from this screen, SIGK, has recently been defined at the G␣ subunit-switch II binding interface on the ␤-propeller. 2 Since SIGK competes for the residual cross-linking of N20K to G␤ 1 (C25A), the second N20K binding site must overlap with the SIGK binding site. This surface on ␤␥ subunits has been shown to be important for PLC regulation (9,10). There are two surface cysteine residues in proximity to this site, Cys 204 and Cys 271 , which could be participating in the cross-linking reaction. This interface is ϳ30 Å from Cys 25 . With SMCC as a cross-linker with a 12-Å spacer arm, it is not possible for the peptide to be binding at the amino terminus and use either of these two cysteine residues for cross-linking; nor is it possible for the peptide to be solely binding at this interface and use cysteine 25 for cross-linking. Binding of SIGK to this surface is able to block G␤␥-dependent activation of PLC, implying that it is a functional site for PLC binding.
Previous data suggesting that the Y-domain of PLC␤2 interacts with G␤␥ subunits has been inferred through indirect experiments from peptide and small domain binding and inhibition assays. Other data have suggested involvement of the PH domain in ␤␥-dependent regulation of PLC␤ (28). To conclusively demonstrate that the Y-domain is indeed important for G␤␥ regulation of PLC␤2, we made mutations in the ␣ 5 helix within the Y-domain on the surface of PLC␤ corresponding to the last 10 amino acids N20K. PLC␤2(ELK/AAA) was significantly impaired with respect to G␤␥-dependent activation relative to wild type PLC␤2 (Fig. 6, C and D). However, none of the three mutations tested completely blocked G␤␥-dependent activation. Previous work by Illenberger et al. (29) demonstrated that replacement of the PH domain of PLC␤2 with the PH domain of PLC␤1 (a phospholipase poorly regulated by G␤␥ subunits) blocked some but not all of the activa-tion by G␤␥ subunits, leading the authors to believe that interactions between G␤␥ subunits and the catalytic domain of PLC␤2 were both necessary and sufficient for G␤␥-dependent activation. Thus, data suggesting the PH domain is important for PLC regulation by G␤␥ subunits, taken together with the data presented here and previously, suggest that both the PH domain and the Y-domain of PLC␤2 play important roles in regulation of the enzyme by G␤␥ subunits. We also found that PLC(ELK/AAA) was activated to a greater extent by G␣ q relative to wild type PLC␤2. This is an interesting observation that needs to be investigated in further detail but clearly demonstrates that these mutants are selectively impaired with respect to G␤␥-dependent activation of PLC.
The results of this study suggest that the Y-domain of PLC␤2 can bind two distinct regions on the G␤␥ subunit, the N-terminal ␣ helix and the G␣ subunit switch II binding interface. Fig. 7 outlines a simple model for PLC regulation through binding at these two distinct sites. In the model, binding of the Y-domain of PLC␤2 (shown as a triangle, with the N20K region highlighted) to the amino terminus of G␤␥ is inhibitory. This complex is in equilibrium with a G␤␥-PLC complex where the Y-domain is bound to the propeller region. This contact is stimulatory and necessary for enzymatic activation of PLC. During interactions between wild type PLC and wild type G␤␥, these two processes are in equilibrium, with the balance resulting in a net activation of the enzyme. For the G␤ 1 (23)(24)(25)(26)(27) mutant, disruption of the interaction between the amino terminus of G␤␥ and PLC relieved this inhibitory constraint, resulting in the enhanced activation. The N20K peptide was still able to block this mutant's activation of PLC, because the stimulatory binding contact between the Y-domain and the blade region remained intact. Mutations to the N20K region within full-length PLC␤2 would be expected to reduce the ability of the enzyme to be activated by G␤␥, since binding to both the stimulatory and the inhibitory region would be equally affected. This was confirmed with the PLC␤2(ELK/ AAA) mutant.
There are two simple mechanisms by which binding to the amino terminus of G␤␥ could result in an inhibition in the activation of the enzyme. First, binding of the Y-domain to this region would sequester it from binding to the propeller region, which is necessary to stimulate PLC activation, as shown by our N20K and SIGK competition data. Second, the N20K region within the Y-domain of PLC␤ is within close proximity to the substrate binding site of the enzyme (14). Recent low resolution structures of G␤␥ bound to membrane tubules place the N-terminal region of G␤␥ far from the membrane surface (30). Therefore, if the Y-domain of PLC␤ is bound to the N terminus, it would be sequestered from its substrate. Some possible physiological implications for an inhibitory binding constraint within the amino terminus of G␤␥ for PLC␤2 can be imagined. First, inhibitory binding to the amino terminus could be a mechanism for regulating the activation of the enzyme. There are data to suggest that G-proteins can negatively regulate PLC activity. It had previously been demonstrated that the addition of a nonhydrolyzable version of GTP, Gpp(NH)p, to solubilized membranes from cortical neurons, decreased inositol phosphate production at low concentrations, whereas it stimulated production at high concentrations (31,32). Moreover, activation of G␣ i/ o-coupled receptors, such as the D2 dopamine receptor and the A1 adenosine receptors, could antagonize inositol phosphate production resulting from activation of the G␣ q -linked thyrotropin-releasing hormone receptor (24,33,34). These experiments suggest G␤␥ could negatively regulate PLC activation in vivo. Alternatively, the amino terminus could be functioning as a scaffold, anchoring the PLC near its activator and its substrate, while the G-protein is in its inactive heterotrimer and the propeller region is blocked by G␣. Upon activation of the G-protein signaling cascade and release of ␣-GTP, the Y-domain could dissociate from the amino terminus, and both the Y-domain and the PH domain of PLC could now make contact with the stimulatory areas within the propeller region, leading to enzyme activation.
Three novel conclusions have evolved from the course of this study. First, we clearly demonstrate that the amino-terminal coiled-coil region of G␤␥ is directly involved in effector regulation in mammalian isoforms of G␤␥ subunits, the first such data to map a mammalian effector binding site outside of the propeller region. Second, we show that mutations within the catalytic domain of PLC␤2 had reduced activation by G␤␥ subunits, supporting previous data with synthetic peptides and GST fusion proteins suggesting that the Y-domain of PLC was important for regulation by G␤␥ subunits. Finally, we show evidence suggesting that the N20K peptide may bind two distinct regions on the G␤␥ subunit, leading to the possibility the Y-domain of full-length PLC may also bind G␤␥ at two distinct sites as well. What is becoming clearly apparent with G␤␥ regulation of effectors is that no single binding interaction defines all of the contact interfaces. Thus, the mechanisms for G␤␥ binding and effector activation appear to be significantly more complex than had been previously imagined.