Identification of a Structural Element in Phospholipase C b 2 That Interacts with G Protein bg Subunits*

To delineate the specific regions of phospholipase C b 2 (PLC b 2) involved in binding and activation by G protein bg subunits, we synthesized peptides corresponding to segments of PLC b 2. Two overlapping peptides corresponding to Asn-564–Lys-583 (N20K) and Glu-574–Lys-593 (E20K) inhibited the activation of PLC b 2 by bg subunits (IC 50 50 and 150 m M , respectively), whereas two control peptides did not. N20K and E20K, but not the control peptides, inhibited bg -dependent ADP-ribosylation of G a i1 by pertussis toxin and bg -de-pendent activation of phosphoinositide 3-kinase. To demonstrate direct binding of the peptides to bg subunits, the peptides were chemically cross-linked to purified b 1 g 2 . N20K and E20K cross-linked to both b 1 and g 2 subunits, whereas the control peptides did not. Cross-linking to b and g was inhibited by incubation with excess PLC b 2 or PLC b 3, whereas cross-linking to g but not b was inhibited by r-myr- a i1 . These data together demonstrate specificity of N20K and E20K for G bg binding and inhibition of effector activation by bg subunits. The results suggest that an overlapping region of the two active peptides, Glu-574–Lys-583, mimics a region of PLC b 2 that is involved in binding to bg subunits. Changing a tyrosine to a glutamine in this overlapping region of the run for 1 The of uncross-linked b and molecular weight in experiment was repeated three and

Many transmembrane receptors coupled to heterotrimeric G proteins can initiate the hydrolysis of phosphatidylinositol 4,5bisphosphate (PIP 2 ) 1 to produce inositol trisphosphate (IP 3 ) and diacylglycerol. Multiple experimental paradigms provide convincing evidence that increased PIP 2 hydrolysis occurs as a result of enzymatic activation of phosphatidylinositol-specific phospholipase C (PLC) due to direct interactions of PLC with activated G protein ␣ or ␤␥ subunits (1)(2)(3). G protein ␣ subunits of the Gq class are thought to be responsible for receptormediated activation of PLC that is not inhibited by treatment with the toxin from Bordatella pertussis (PTX). ␤␥ subunits released during activation of Gi/o proteins are thought to activate PLC in a manner that is inhibited by PTX, since this toxin blocks interaction of the Gi/o heterotrimer with receptors.
PLC enzymes consist of at least nine different isoforms that have been classified into three groups; ␤, ␥, and ␦ (2, 4). All of these enzymes require Ca 2ϩ for activity. PLC ␤ isoforms are the primary enzymes that hydrolyze PIP 2 in response to activation by G proteins. Four isozymes of the PLC ␤ class have been identified and designated PLC␤1, ␤2, ␤3, and ␤4. Each isoform is regulated differently by G protein ␤␥ or ␣ subunits. In in vitro enzyme assays and in co-transfection assays there are some conflicting results, but in general PLC ␤1 and ␤4 are regulated primarily by G␣ q , PLC ␤2 is regulated primarily by ␤␥ subunits, and PLC ␤3 is regulated by both ␤␥ and ␣ q subunits.
Based on sequence alignments between PLC isoforms and other proteins, some domain structure has been predicted. The first 100 amino acids are predicted to form a pleckstrin homology domain (5). This domain in PLC ␦1, when expressed in isolation, binds to PIP 2 and IP 3 and when removed from PLC ␦1, inhibits anchoring to PIP 2 -containing membranes but does not inhibit catalysis (6 -8). The role of the pleckstrin homology domain in the other PLC isoforms is unclear but is likely to be distinct from PLC ␦1, since PLC ␤2 membrane binding is unaffected by IP 3 (9). Two highly conserved regions have been identified in all mammalian PLCs and designated X and Y. In PLC ␤ and ␦ isoforms, these regions are adjacent in the primary sequence, whereas in PLC ␥, the X and Y are separated by intervening src-homology SH2 and SH3 domains. PLC ␤ isoforms are unique in that there is an extended (40 kDa) C-terminal domain that extends beyond the Y domain (4).
A three-dimensional crystal structure has been solved for PLC ␦1 that contains the X and Y domains but is missing the N-terminal pleckstrin homology domain (10). The structure shows that the N-terminal region between the pleckstrin homology domain and the X domain has a structural fold that is very similar to the EF-hand domain found in Ca 2ϩ -binding proteins, including calmodulin. The entire X domain and twothirds of the N terminus of the Y domain fold to form a catalytic core similar in structure to the ␤ barrel found in triose phosphate isomerase (TIM barrel). The C-terminal one-third of the Y domain forms a domain similar to the C2 domains found in PKC and PLA 2 where they function as calcium-dependent phospholipid binding domains. In PLC ␦1, the C2 domain primarily interacts with the N-terminal EF-hand domain, and its function is unclear.
Some progress has been made mapping the regions in the overall sequences of the PLCs that are involved in interaction with G proteins. Removal of the C-terminal third of PLC ␤1 abolishes activation by G␣ q , but Ca 2ϩ -dependent activity remains intact (11,12). Further analysis of this region has served to define important regions for interaction with ␣ subunits (13). Removal of the C-terminal third of PLC ␤2 does not affect its ability to be stimulated by ␤␥ subunits (14,15). In a series of experiments by Kuang et al. (16), segments of the PLC ␤2 X and Y domains when coexpressed with PLC ␤2 in COS-7 cells blocked activation by ␤␥ subunits. When expressed as GST fusion proteins, a 116-amino acid polypeptide from this region bound tightly to G protein ␤␥ subunits, whereas a 60-amino acid sequence of this same region bound weakly to ␤␥ subunits (16).
In this paper we further define the regions of PLC ␤2 that interact with G protein ␤␥ subunits. Synthetic peptides corresponding to sections of the 116-amino acid region previously identified (16) were tested for their ability to inhibit activation of PLC ␤2 by ␤␥ subunits in a reconstituted, purified system. The peptides were designed based on sequence alignment of PLC ␤2 and PLC ␦1 and referral to the crystal structure of PLC ␦1. This allowed us to identify regions within Gln-526-Val-641 that were on the surface and potentially accessible to ␤␥ subunit binding. Based on these studies we propose a model for the structural features of PLC ␤2 involved in ␤␥-PLC interactions.

Materials
Peptides were purchased from Coast Scientific or from Biosynthesis and had a purity of greater than 90% based on high performance liquid chromatography analysis and mass spectrometry. Phosphatidylethanolamine (bovine liver) and phosphatidylinositol (bovine liver) (PI) were from Avanti Polar Lipids. PIP 2 was prepared from bovine brain lipids (Sigma) according to the method of Schacht (17)

Plasmid Construction and Cloning of Recombinant Baculoviruses-
Purified, recombinant PLC ␤2, ␤3, and PI 3-kinase proteins were prepared using a baculovirus expression system. Construction of baculovirus for expression of PLC ␤2 has been previously described (9). 6-Histagged PLC ␤3 was prepared as follows. The cDNA for PLC ␤3 in bluescript SKϪ was cut with BamHI at base pairs 636 and 2556, and the remaining sequence of PLC ␤3 was religated. This removed multiple NcoI sites from the coding region of PLC ␤3. Oligonucleotides encoding an EcoRI site (at the 5Ј-end) followed by the coding sequence for an initiator methionine, an alanine, a histidine tag (6 histidines), and an NcoI site were ligated at the 5Ј end of the PLC ␤3 sequence at the NcoI site at base pair 1 of the cDNA coding sequence and EcoRI of bluescript. A 250-base pair fragment was excised from this construct with EcoRI and BstEII and ligated with the original PLC ␤3 cDNA cut with BstEII and EcoRI after removal of the insert. This fragment was then excised with EcoRI and HindIII and inserted into the EcoRI-HindIII site of Fastbac (Life Technologies, Inc.). Recombinant, clonal baculoviruses were generated according to the protocol described by Life Technologies, Inc.
Sf9 Cell Culture and Purification of Phospholipase C and PI 3-Kinase-Sf9 cells were grown at 27°C in IPL-41 medium containing 10% fetal bovine serum, 0.1% pluronic acid, and 50 mg/ml gentamycin. For large scale cultures (1 liter and above), the cells were switched into medium containing 1% fetal bovine serum and 1% lipid concentrate (Life Technologies, Inc.). Baculoviruses directing expression of recombinant 6-His PLC ␤2 and 6-His PLC ␤3 were used to infect 1 liter of Sf9 cells at a density of 1.5 ϫ 10 6 cells/ml. The proteins were purified according to (9).
For PI 3-kinase, the Sf9 cells were coinfected with viruses coding for expression of 6-His-tagged p110␥ with EE-tagged p101, and the protein was purified on a column containing immobilized anti-EE antibody (18). The protein prepared in this way is predominantly heterodimeric p110␥ and p101.
Phospholipase C and PI 3-Kinase Assays-Phospholipase C assays were conducted as described previously (9) and in the figure legends. PI 3-kinase activity was assayed using sonicated micelles containing 600 M phosphatidylethanolamine and 300 M bovine liver PI as in Parish et al. (22).
ADP-ribosylation of G␣ i1 -ADP-ribosylation assays were performed as described by Casey et al. (23). Briefly, 0.4 pmol of ␤ 1 ␥ 2 was mixed with 20 pmol of ␣ i1 followed by the addition of various concentrations of peptide. Reactions were initiated by the addition of pertussis toxin (5 g/ml final concentration), NAD (2.5 M final concentration), and [ 32 P]NAD (750,000 cpm/assay) in a total reaction volume of 40 l. No phospholipids were used in the reaction. Reactions were terminated after 20 min, and proteins were precipitated in 15% trichloroacetic acid and collected on nitrocellulose filters. After extensive washing with 6% trichloroacetic acid, the filters were dried and analyzed by liquid scintillation counting.
Cross-linking and Immunoblot Analysis-Peptides (30 or 150 M) were added to ␤ 1 ␥ 2 (100 nM) at room temperature in a solution containing 1 mM MgCl 2 , 80 mM KCl, 50 mM HEPES, pH 7.4, and 0.1% C 12 E 10 (decaethylene glycol dodecyl ether) in a total volume of 100 l. The cross-linker, SMCC, was dissolved in Me 2 SO and diluted to 2 mM in 50 mM P i buffer, pH 5.0, before dilution into the reaction buffer (final concentration 200 M). The cross-linking reaction was carried out as described in the figure legends and quenched with 10 mM Tris, pH 8.6, and 10 mM 2-mercaptoethanol. When PLC ␤2, PLC␤3, or ␣ i1 (250 or 500 nM) were added, they were incubated with ␤␥ subunits at room temperature for 10 min before the addition of peptides and cross-linker. The cross-linked proteins were resolved on a 12% SDS-acrylamide gel or a 17% SDS-acrylamide gel containing 4 M urea. The proteins were then transferred to nitrocellulose and analyzed by Western blotting with antibodies against ␤ 1 or ␥ 2 . ␥ subunit antibody X-263 has been previously described and recognize ␥ 2 , ␥ 3 , and ␥ 7 (24). The ␤ subunit antibody B600 was raised against a synthetic peptide corresponding to the C terminus of ␤ 1 , ␤ 2 , and ␤ 3 (MAVATGSWDSFLKIWN) and could recognize ␤ 4 as well. Enhanced chemiluminescence reagents ECL (Amersham Pharmacia Biotech) were used to visualize the proteins.

RESULTS
Peptide Design-There is significant homology between PLC ␤2 and PLC ␦1 in the 116-amino acid region of PLC ␤2 that was found to bind to ␤␥ subunits (16): 33% identity and 47% similarity overall with a 70-amino acid region having 54% identity and 67% similarity (Fig. 1). Based on this high level of homology, we predicted that PLC ␤2 might have a very similar fold as PLC ␦l in this region. Using this assumption, we examined the structure of PLC ␦1 to determine which regions of PLC ␤2 would be likely to be on the surface of the protein and accessible to G protein ␤␥ subunits. Three peptides were chosen for our initial studies, corresponding to amino acids 564 -583 (N20K), 584 -609 (A20G), and 575-594 (E20K) of PLC ␤2 shown in (Fig.  1B). Also tested as a control was a peptide composed of the same amino acids as N20K, except the primary sequence was randomized (YVLSKNRSDLFTKAYISSEL).
Inhibition of ␤␥ Stimulated PLC ␤2 Activity-The peptides were tested for their ability to inhibit the stimulation of PLC ␤2 activity by ␤␥ subunits (Fig. 2). In the absence of peptide, ␤␥ subunits stimulated PLC ␤2 approximately 10-fold over basal, Ca 2ϩ -dependent activity. The addition of N20K inhibited ␤␥stimulated activity with an IC 50 of 50 M with 95% inhibition of activity occurring at 200 M. The adjacent peptide, A20G, and the scrambled peptide had no effect on stimulation of PLC activity by ␤␥ subunits at concentrations up to 300 M. The peptide that overlaps the C terminus of N20K and the N terminus of A20G-E20K (E20K) inhibited ␤␥-stimulated activ-ity with an IC 50 of 150 M. Both peptides inhibited the basal Ca 2ϩ -stimulated activity of PLC ␤2, with N20K inhibiting activity by 58 Ϯ 10% and E20K inhibiting activity by 43 Ϯ 10% (n ϭ 4 experiments). Importantly, ␤␥-stimulated activity was inhibited to a greater extent than was basal activity. To quantitate this effect, data were normalized by determining the fold stimulation of activity over basal activity by ␤␥ subunits (activity with ␤␥ divided by activity without ␤␥). The percent inhibition of the fold stimulation by ␤␥ subunits in the presence of 200 M peptides was then calculated. N20K inhibited fold stimulation over basal by 67 Ϯ 2%, whereas E20K inhibited fold stimulation by 42 Ϯ 4%. We also tested a peptide corresponding to N20K where Tyr-15 in the region that overlaps with E20K was changed to Gln (N20K(Y15Q)). This peptide stimulated PLC ␤2 activity in the absence of ␤␥ subunits (data not shown). The significance of this is unclear, but it did not allow us to measure effects of N20K(Y15Q) peptide on ␤␥stimulated PLC ␤2 activity. To prove that the inhibition of ␤␥-stimulated PLC activity was attributable, at least in part, to binding of the peptides to ␤␥ subunits, several other assays were performed to demonstrate binding of the peptides to ␤␥ subunits.
Inhibition of ADP-ribosylation-G protein ␣ subunits block the ability of ␤␥ subunits to activate PLC by sequestering the ␤␥ subunits in the heterotrimeric form. This is thought to work because the ␣ subunits sterically hinder interaction between the ␤␥ subunits and the PLC. This predicts that peptides mimicking PLC ␤2 binding to ␤␥ subunits could block interaction between ␣ and ␤␥ subunits. One way to measure this is to measure the ␤␥-dependent enhancement of ADP-ribosylation of ␣ subunits by pertussis toxin. Since ␤␥ binding to ␣ subunits is required for ADP-ribosylation of ␣, peptides that block interaction between ␣ and ␤␥ subunits will inhibit ADP-ribosylation.
N20K or E20K inhibited incorporation of ADP-ribose into purified recombinant myristoylated ␣ i1 in the presence of ␤␥ subunits. ␤␥-stimulated ADP-ribosylation was inhibited by close to 100% by 100 M N20K or E20K with an IC 50 of 8 M (Fig. 3A). Neither A20G nor the scrambled peptide (100 M each) had an effect on ADP-ribosylation. This indicates that N20K and E20K are binding to the ␤␥ subunits at a site that may overlap the binding sites for ␣ i1 or that the peptides are binding at a site on ␤␥ that prevents binding of PTX. We also tested the N20K(Y15Q) peptide. This peptide was much less potent in inhibiting ␤␥-dependent ADP-ribosylation (Fig. 3B)  units-To further confirm that these peptides inhibit various effectors by binding to ␤␥ subunits, we tested the effects of these peptides on the stimulation of PI 3-kinase by ␤␥ subunits. We utilized a p110␥/p101 heterodimer purified from sf9 cells as described by Stephens et al. (18). The activity of the p110/p101 heterodimer was increased 5-fold by 150 nM ␤␥ subunits. N20K or E20K inhibited the stimulation of this enzyme by 150 nM ␤␥ subunits, with an IC 50 of 50 -100 M (Fig. 4). A20G and scrambled peptide had no effect. The peptides did not have a significant effect on basal PI 3-kinase activity.
Effects of Peptides on Stimulation of PLC ␤3 by ␤␥ Subunits-PLC ␤3 is activated by G protein ␤␥ subunits at a similar potency as PLC ␤2 (25). Activation of PLC ␤3 by ␤␥ subunits and Ca 2ϩ was measured in the presence of E20K and N20K. Surprisingly, E20K had little effect on the activation of PLC ␤3 by G protein ␤␥ subunits, whereas N20K inhibited activity but not to the same extent as for PLC ␤2 (Fig. 5). Basal, Ca 2ϩ -dependent activity of PLC ␤3 was not measurable in these assays. We have reported previously that basal activity of PLC ␤3 is much lower that for PLC ␤2 under these assay conditions (25).
Cross-linking of Peptides to ␤␥ Subunits-To directly demonstrate that these peptides bind to ␤␥ subunits, we developed a cross-linking assay that uses a heterobifunctional crosslinker to covalently link the peptides to purified ␤␥ subunits. The cross-linker (SMCC) has a succinimide ester group that reacts with primary amines and a maleimido group that reacts with SH moieties. Since the peptides have no cysteine residues, the only way to cross-link the peptides to ␤ or ␥ is via a primary amine on the peptide and SH groups on either ␤ or ␥ subunits. Peptide cross-linking was monitored by immunoblotting for ␤ or ␥ subunits after electrophoresis to resolve cross-linked subunits from the unmodified subunits.
The specificity for peptide cross-linking to ␤␥ was tested by incubating purified ␤␥ subunits with N20K, E20K, N20K(Y15Q), A20G, or scrambled peptides and cross-linker. Only in the presence of N20K or E20K (30 or 150 M each) was there an increase in the apparent molecular weight of ␥ after incubating with cross-linker (Fig. 6A). Some cross-linking of N20K(Y15Q) to ␥ is observed at 150 M peptide, consistent with the data demonstrating that N20K(Y15Q) is less potent in inhibiting ADP-ribosylation of ␣ i1 .
N20K and E20K also cross-linked to the ␤ subunit (Fig. 7A). Only in the presence of N20K or E20K and cross-linker does a prominent cross-linked species appear. A number of minor higher molecular weight species are also seen only in the presence of cross-linker and peptide. It is unclear what the higher molecular weight bands correspond to, although all can be visualized with ␤ subunit antibodies, and some are visualized with ␥ subunit antibodies. These could represent more than one peptide cross-linked to each ␤ subunit or peptide crosslinked to ␤ that is cross-linked to ␥. The higher molecular weight band observed in the absence of peptides (Fig. 7B, lane 1) corresponds to ␤ cross-linked to ␥, because the upper band could be visualized with a ␥ subunit antibody (data not shown).
To further demonstrate the relevance of the cross-linking, the ␤␥ subunits were incubated with excess PLC ␤2, PLC ␤3, or ␣ i1 before incubation with peptide and cross-linker to determine if they could compete for peptide cross-linking. PLC ␤2, PLC ␤3, and ␣ i1 all inhibited cross-linking of N20K and E20K to the ␥ subunit, although PLC ␤3 was less effective than PLC ␤2 (Fig. 6B). PLC ␤2 and PLC ␤3 both inhibited cross-linking of the peptides to the ␤ subunit, although PLC ␤3 was again consistently less effective than PLC ␤2 (Fig. 7B). Interestingly, GDP-liganded ␣ i1 was unable to inhibit cross-linking of peptides to the ␤ subunit. DISCUSSION Two synthetic 20-amino acid overlapping peptides that mimic a region of PLC ␤2 bind specifically to G protein ␤␥ subunits and prevent interaction with different biochemical partners. An adjacent peptide and scrambled N20K had no effect in any of the assays we have examined. All the peptides are very similar with respect to amino acid chemistry. Al-though the peptides have some effect on the basal activities of PLC ␤2, this cannot explain the extent of the inhibition of ␤␥ stimulation of PLC ␤2 activity. It is clear based on the other assays involving ␤␥ subunits that these peptides bind to the ␤␥ subunits and that this is responsible, at least in part, for the observed inhibition of ␤␥ stimulated PLC ␤2.
The N20K and E20K peptides inhibit the ␤␥-dependent activation of PLC ␤2 with IC 50 s of 50 and 150 M, respectively. These IC 50 s are similar to what has been reported for a 27amino acid peptide derived from adenylyl cyclase type II (QEHA peptide) for inhibiting a number of ␤␥ subunit-regulated effectors (26). N20K and E20K do not contain the ␤␥ binding consensus sequence identified in the QEHA peptide. However, in the common 10 amino acids of the N20K and E20K there is a short stretch of seven amino acids where 4 amino acids are identical to a sequence at the C terminus of QEHA and a 5th amino acid is conservatively substituted. When we changed one of these amino acids in N20K, binding to ␤␥ subunits was dramatically inhibited (Figs. 3B, 6A, and 7A).
The N20K and E20K peptides are more potent in inhibiting ␤␥-dependent ADP-ribosylation of ␣ i1 (IC 50 8 M) than for inhibition of PLC ␤2. This is surprising if we assume that the peptides block ADP-ribosylation of ␣ i1 by blocking ␣ i1 interactions with ␤␥ because the affinity of ␤␥ for ␣ i1 (K d ϳ 1 nM) is much greater than the affinity for PLC ␤2 (K d ϳ 100 nM). The exact mechanism for how ␤␥ subunits stimulate ADP-ribosylation of ␣ subunits is unclear but is known to involve a process where ␤␥ subunits must cycle catalytically among a stoichiometric excess of ␣ subunits. Since the mechanism is not entirely understood, it is possible that the peptides interfere with this catalytic ADP-ribosylation in a way that is not directly  2-7). Cross-linking reactions were performed as in Fig.  6 except only for 5 min. Proteins were then resolved by SDS-polyacrylamide gel electrophoresis in 12% acrylamide mini-gels. The gels were run at a constant voltage of 150 volts for 1 h. The proteins were transferred to a nitrocellulose membrane and blotted with anti-␤ 1 antibody. Positions of uncross-linked ␤ 1 , and molecular weight markers in kDa are shown. Each experiment was repeated three times and yielded similar results. related to the known high affinity of ␤␥ subunits for ␣ subunits. One possibility is that the peptides are not blocking ␣ binding to ␤␥ but are occupying a binding site on ␤␥ required for PTX interaction. This possibility is supported by the cross-linking data that shows that cross-linking of the peptides to the ␤ subunit is not blocked by ␣ i1 -GDP.
Both N20K and E20K must cross-link directly to cysteine residues in either the ␤ and ␥ subunits due to the nature of the cross-linker. The site of ␥ cross-linking is clearly cysteine 41, since this is the only cysteine in ␥ 2 . The site(s) of cross-linking of the peptide to the ␤ subunit is unclear, but there are 14 cysteine residues in the ␤ subunit where cross-linking could have occurred. Cross-linking of E20K and N20K to both ␤ and ␥ was prevented by incubation with PLC ␤2. PLC ␤3 also prevented cross-linking of N20K and E20K but was not as effective as PLC ␤2, suggesting that the binding sites for these two enzymes on ␤␥ subunits do not entirely overlap. This is consistent with the observation that the peptides were not as effective or as potent at inhibiting PLC ␤3 activation by ␤␥ subunits.
The cross-linking of the peptides to the ␤ subunit was not prevented by preincubation of the ␣ subunit with the ␤ subunit. This suggests that the peptide binding site on ␤␥ does not entirely overlap with the ␣ subunit. Although it is known that ␣-GDP blocks activation of PLC ␤ by ␤␥ subunits, it is probable that the binding sites for PLC ␤ and ␣ subunits overlap but do not match. Thus the binding site represented by the peptide may lie outside the overlap region. This idea is supported by a recent study by Bluml et al. (27) where it was demonstrated that a peptide from phosducin binds to ␤␥ subunits. Visualization of the location of this peptide in the crystal structure of ␤␥ complexed with phosducin (28) shows that this peptide would bind at a region on the ␤ subunit that does not overlap the ␣ subunit binding site, whereas the N-terminal domain of phosducin does overlap the ␣ subunit binding site. That ␣-GDP inhibits cross-linking to the ␥ subunit may result from steric interference with the cross-linking reaction without interfering directly with peptide binding.
The data we have presented allows us to narrow down in the primary sequence of PLC ␤2 a region of 20 amino acids that may be involved in ␤␥ subunit binding. The N terminus of the E20K peptide overlaps the C terminus of N20K peptide by 10 amino acids. The C-terminal 10 amino acids of E20K overlaps with the N-terminal 10 amino acids of A20G, which does not inhibit ␤␥ activation in any assay. This suggests that the potential region that is binding to ␤␥ subunits can be narrowed down to 10 amino acids, Glu-574 -Lys-583. We have tested a 10-amino acid peptide corresponding to this region and found no evidence for inhibition. A possible explanation is that this 10-amino acid peptide may be too short to adopt the appropriate secondary structure. This has been seen for other peptides including the peptide from adenylyl cyclase type II (26).
Visualization of the regions homologous to our peptides in the PLC ␦1 structure supports the idea that the 10-amino acid overlap region is the critical region of the peptides involved in binding to ␤␥ subunits. The catalytic domain of PLC ␦ is composed of parts of the conserved X and Y domains that form a TIM barrel constructed from of a series of ␤ sheet ␣ helix repeats. The ␤ strands line the inside of the barrel, and side chains from these strands project into the core of the structure to provide the chemistry for substrate binding and catalysis. The ␣ helices are on the outside of this barrel structure and in some cases are exposed to solvent. When the sequence of the N20K is aligned with the PLC ␦1 sequence, the N-terminal 10 amino acids align with a small amount of linker sequence and the T␤6 strand of the barrel (nomenclature of Essen et al. (10)).
The sequence in T␤6 is very conserved between the various PLC isoforms with two amino acids, serine 571 and phenylalanine 572 (␤2 sequence), being conserved in all known PLC isoforms. For this reason we predict that this sequence in PLC ␤2 will occupy a similar position to that observed in the PLC ␦ sequence. Since this region is on the inside of the TIM barrel in PLC ␦ and is directly involved in substrate binding, it would be unlikely to be accessible to interaction with ␤␥ subunits. The region homologous to Glu-574 -Lys-583 of PLC ␤2 (overlap region between N20K and E20K) forms an ␣ helix on the surface of PLC ␦1, and because of the significant sequence homology in this region, would likely form the same type of structure on PLC ␤2. The surface exposure of just this 10amino acid region of N20K further suggests that this overlap region corresponds to a portion of PLC ␤2 that is important for interaction with and regulation by G protein ␤␥ subunits. The location of this binding site directly adjacent to structures that contribute amino acids to substrate binding and catalysis suggests a mechanism by which ␤␥ subunits could activate PLC ␤. If the binding of ␤␥ subunits to the helix on the surface caused movement of the adjacent ␤ strands, this could position amino acids so they are more favorable for catalysis there by increasing enzymatic activity.
The region we have defined only overlaps the portion of PLC ␤2 (Leu-580 -Val-641) originally defined by Kuang et al. (16) as a ␤␥ binding domain by 4 amino acids (580 -583). Using twohybrid analysis, Yan and Gautam (29) used the N-terminal 100-amino acid region of the ␤ subunit to demonstrate interactions with this 62-amino acid domain. One peptide (A584-G604) overlaps extensively with this domain yet had no effect in any of our assays for ␤␥ subunit interactions. This suggests that either only a very short 4-amino acid region is necessary for some ␤␥ binding or that other regions within a 62-amino acid region are also involved in PLC ␤2-␤␥ interactions.