Role of Dynamic Interactions in Effective Signal Transfer for Gβ Stimulation of Phospholipase C-β2*

Heterotrimeric G protein subunits regulate their effectors by protein-protein interactions. The regions involved in these direct interactions have either signal transfer or general binding functions (Buck, E., Li, J., Chen, Y., Weng, G., Scarlata, S., and Iyengar, R. (1999) Science 283, 1332–1335). Although key determinants of signal transfer regions for G protein subunits have been identified, the mechanisms of signal transfer are not fully understood. We have used a combinatorial peptide approach to analyze one Gβ region, Gβ86–105, involved in signal transfer to the effector phospholipase C (PLC)-β2 to gain a more mechanistic understanding of Gβ/PLC-β2 signaling. Binding and functional studies with the combinatorial peptides on interaction with and stimulation/inhibition of phospholipase Cβ2 indicate that binding affinity can be resolved from EC50 for functional effects, such that peptides that have wild type binding affinities have 15- to 20-fold lower EC50 values. Although more potent, these peptides display a much lower extent of maximal stimulation. These peptides synergize with Gβγ or peptides encoding the second Gβ42–54 signal transfer region in maximally stimulating phospholipase C-β2. Other combinatorial peptides from the Gβ86–105 region that bind to PLC-β2 by themselves submaximally stimulate and extensively inhibit Gβγ stimulation of PLC-β2. The intrinsic stimulation function can be attributed to Arg-96 and Ser-97, the synergy function to Trp-99, and the binding affinity to Thr-87, Val-90, Pro-94, Arg-96, Ser-97, and Val-100. These results indicate that, even within signal transfer regions, residues involved in binding can be resolved from those involved in signal transfer and that signal transfer is likely to be achieved through dynamic rather than steady-state interactions.

Protein-protein interactions represent a major mode by which information is propagated along cell signaling pathways. The heterotrimeric guanine nucleotide binding protein (G protein) 1 regulates the activity of a multitude of different effectors within the cell by direct protein-protein interactions (1). Both the G␣ subunit and the G␤␥ complex of the G protein can interact with effectors (2). Effectors for G␣ include adenylyl cyclases and phospholipase C-␤ (PLC-␤) isoforms. Effectors for G␤␥ subunits include G protein inwardly rectifying K ϩ channels, Ca 2ϩ channels, and PLC-␤ isoforms (3).
We have found that regions of G protein subunits important for signal transfer can be resolved from regions important for binding alone (4,5). For G␤␥ stimulation of the effector PLC-␤2, one G␤ region, G␤86 -105, functions directly in signal transmission. A peptide derived from this region can regulate PLC-␤2 activity on its own in the absence of G␤␥ subunits. Another region of G␤, 115-135, is involved in binding but does not transmit signals, because it does not affect PLC-␤2 activity by itself but inhibits G␤␥ stimulation. Therefore, it is possible to separate general binding domains from signal transfer regions for a protein-protein interaction within intracellular signal flow. Further analysis has shown that G␤ relies on modular collections of these signal transfer and general binding units (6).
We had previously used substituted and truncated peptides to determine the amino acid characteristics of one of the G␤ signal transfer regions, the G␤ 86 -105 region, that render it capable of PLC-␤2 stimulation (4). Residues Lys-89 and Arg-96 are important for its potency (4,7). The six-amino acid region G␤ 96 -101 represents a core signal transfer region, and all contacts contributing to signal transfer for the G␤ 86 -105 signal transfer region likely lie within the G␤ 96 -101 sixamino acid region. However, the G␤ 96 -101 region displays significantly lower EC 50 for PLC-␤2 stimulation, indicating that there may exist important binding contacts within G␤ 86 -105 but outside of G␤ 96 -101.
How the architecture of the G␤ 86 -105 signal transfer region supports effector regulation is still unclear. Would the mechanism for signal transfer be driven largely by complementarity of the interactions between preformed surfaces, including those between charged residues, or by dynamic processes wherein residues on G␤ involved in signal transfer to PLC-␤2 retain conformational flexibility to induce change in activity in PLC-␤2 by transient interactions? To address these issues, herein, we have used a combinatorial peptide library approach to study both binding and functional regulation. Our data suggest that the mechanism of signal transfer to PLC-␤2 is likely to rely on dynamic contacts with this effector that are distinct from those contacts involved in general binding affinity.

EXPERIMENTAL PROCEDURES
Materials-Library vectors and electrocompetent cells were a gift from Affymax, Palo Alto, CA. All oligonucleotides were from Genelink, Syracuse, NY. All peptides were purchased from the Tufts University core facility. The 96-well plates were from Dynatech. Sources of other reagents have been previously described (4,6). For other experiments, all reagents used were of the highest quality available.
Expression of PLC-␤2-Human PLC-␤2 was expressed in Hi5 insect cells by infection with recombinant baculovirus. Hi5 cells were grown in a 1-liter flask in suspension culture in 200 ml of sf900 media (Invitrogen) with shaking until the cell density reached 0.5 ϫ 10 6 cells/ml. The cells were then infected with 20 ml of PLC-␤2 recombinant baculovirus supernatant. About 3 days post infection the cells were collected. The cells were spun at 1000 rpm for 10 min to produce a pellet. The media was decanted, and the cell pellet was washed with 1ϫ phosphatebuffered saline (PBS). The cells were again spun at 1000 rpm for 10 min. The wash was decanted, and the pellet was resuspended into 20 ml of ice-cold PBS supplemented with protease inhibitors and DTT (10 g/ml aprotinin, 1 g/ml leupeptin, 200 mM phenylmethylsulfonyl fluoride, and 1 mM DTT). The cells were lysed by decompression in a Par bomb after equilibration at 600 p.s.i. for 30 min at 4°C. The lysate was then ultracentrifuged at 35,000 rpm for 1 h at 4°C. The supernatant was removed and distributed into 50-l aliquots. These aliquots were frozen on dry ice and ethylene glycol and then stored at Ϫ70°C.
Purification of PLC-␤2-Human PLC-␤2 was expressed in Hi5 insect cells, harvested, and lysed in 25 ml of lysis buffer, as described previously. Following lysis using the Par bomb, NaCl was added to a final concentration of 500 mM. The lysis mix was allowed to rotate at 4°C for about 30 min. The lysate was then centrifuged at 35,000 rpm for 45 min at 4°C. The supernatant was reserved (ϳ25 ml) and added to 6.25 ml of 50% nickel-nitrilotriacetic acid bead slurry (Qiagen) that had been equilibrated with lysis buffer. The lysis/slurry mix was allowed to rotate for about 2 h at 4°C. The slurry mix was then poured into a Kontex column at 4°C. The column was first washed with 70 ml (about 10 column volumes) of high salt wash buffer (10 mM sodium Hepes, pH 8.0, 0.1 mM EDTA, 0.1 mM EGTA, 800 mM NaCl, 0.5% C 12 E 10 , 15 mM imidazole) supplemented with protease inhibitors (10 g/ml aprotinin, 1 g/ml leupeptin, 200 mM phenylmethylsulfonyl fluoride) and 1 mM DTT. The column was then washed with 70 ml (about 10 column volumes) of low salt wash buffer (10 mM sodium Hepes, pH 8.0, 0.1 mM EDTA, 0.1 mM EGTA, 100 mM NaCl, 15 mM imidazole) supplemented with protease inhibitors and DTT, as before. PLC-␤2 was then eluted by washing the column with six successive 4-ml elutions of elution buffer (10 mM sodium Hepes, pH 8.0, 0.1 mM EDTA, 0.1 mM EGTA, 50 mM NaCl, 125 mM imidazole) supplemented with protease inhibitors and DTT, as before. 25-l aliquots of each of the six fractions were run on an SDS-polyacrylamide gel along with BSA standards to determine purity and the concentration of PLC-␤2. Usually, the second and third fractions that contained highly (Ͼ95%) purified PLC␤2 were aliquoted, frozen on dry ice and ethanol, and stored at Ϫ70°C.
Peptide Synthesis-All peptides were purchased from the Tufts University Core Facility. Peptides were high-performance liquid chromatography-purified, and their identity was verified by mass spectrometry. When needed, peptides were dissolved in HED buffer (10 mM Hepes (pH 7.0) 1 mM EDTA (pH 8.0), and 1 mM DTT).
Expression of G␤␥-G␤␥ was purified from bovine brain as previously described (8) and was a kind gift of Dr. John Hildebrandt. Two different batches of G␤␥ were used in this study. The EC 50 values for PLC-␤2 of the G␤␥ from these two batches were different. For the G␤␥ used in the experiments shown in Fig. 3 the EC 50 for PLC-␤2 stimulation was ϳ200 nM. In contrast, for the experiments in Fig. 4, the second preparation of G␤␥ had an EC 50 of ϳ50 nM for PLC-␤2 stimulation.
Measurement of PLC-␤2 Activity-The phospholipase C assay has been previously described (9). About 10 -15 g of cytosolic fraction of PLC-␤2 was used per 100-l reaction. Phospholipid substrate was a mixture of [ 3 H]phosphatidylinositol 4,5-bisphosphate ([ 3 H]PIP 2 ) and unlabeled phospholipids. [ 3 H]PIP 2 was from PerkinElmer Life Sciences. Unlabeled phospholipids were crude lipids from bovine brain and were from Sigma (P6023). The total diphosphoinositide and triphosphoinositide content was 20 -40%. The remainder was a mixture of phosphatidylinositol and phosphatidylserine. Phospholipids were sonicated in 10 mM Hepes (pH 7.0) to form micelles. A total of 0.01 Ci of [ 3 H]PIP 2 , corresponding to about 7000 cpm, and 5 g of unlabeled mixed phospholipids were used per reaction. Substrate, PLC-␤2, peptide, and G␤␥ subunits were mixed on ice in 100 l of buffer containing 10 mM NaCl, 2 mM EGTA, 1 mM EDTA, and 1 mM MgCl 2 . Reactions were started by the addition of 25 l of 2.5 mM CaCl 2 in 10 mM Hepes (pH 7.0) and incubated at 32°C for 15 min. Reactions were stopped by the addition of 1 ml of chloroform:methanol:HCl (mixed 1:1:1 by volume) and 250 l of EDTA. After extraction, 400 l of the aqueous phase was counted on a Beckman scintillation counter. All experiments were repeated at least three times with very similar results. Typical experiments are shown.
FRET Measurements of Peptide-PLC␤2 Interactions-Recombinant PLC-␤2 was expressed in Sf9 insect cells and purified, and fluorescent studies were done as described (10). To label PLC-␤2 with the aminereactive probe Cascade Blue acetyl azide (Molecular Probes, Eugene OR), the pH was raised to 8.0 and a 4-fold excess of probe was added. The reaction was kept on ice for 30 min before dialysis in 20 mM Hepes (pH 7.2), 0.16 M NaCl, 1 mM DTT, and 2 mM EGTA. Peptides were labeled with 4-dimethulaminophenylazophenyl-4Ј-maleimide, a nonfluorescent energy transfer acceptor, in the presence of an equimolar amount of dye in the absence of reducing agents. The reaction was allowed to proceed for 30 min at room temperature and quenched with 5 mM DTT. The final labeling ratios, as determined by absorption, were 1:1 for Cascade Blue-PLC-␤2 and 0.8 for the two 4-dimethulaminophenylazophenyl-4Ј-maleimide peptides. Fluorescence spectra were taken on an ISS-PC1 (ISS, Champaign, IL) photon-counting spectrofluorometer in a 3-by 3-mm cuvette with excitation at 380 nm and scanning from 400 to 560 nm.
Bacterial Strains, Plasmids, and Oligonucleotides for the Construction of the Combinatorial Peptide on Plasmid Library-Escherichia coli ARI814 electrocompetent cells, the pJS142 library vector, and the pELM3 MBP vector were from Peter Schatz at Affymax Corp., Palo Alto, CA. The use of these reagents to construct combinatorial libraries has been previously described (11). The degenerate library oligonucleotide (5Ј-GA GGT GGT . . . NNN . . . TAA CTA AGT AAA GC), where NNN denotes the nucleotides encoding the 20-amino acid degenerate library, was chemically synthesized, gel-purified, and 5Ј-phosphorylated. Here, N denotes a probability of 70% wild type base and 10% each of the other three bases. This distribution leads to an approximate 50% probability that the wild type residue at each of the 20-amino acid positions will be mutated to another amino acid. The two-linker oligonucleotides, ON-829 and ON-830, were synthesized and 5Ј-phosphorylated. All oligonucleotides were from Genelink, Syracuse, NY.
Library Construction-A G␤ 86 -105 combinatorial peptide library based on the peptides on plasmids method was constructed and expressed as described in detail previously (11,12). Briefly, the library oligonucleotide was annealed with the two linker oligonucleotides and ligated into the pJS142 peptide on plasmid vector. The ligation was electroporated into ARI814 electrocompetent cells, amplified, and frozen in aliquots at Ϫ70°C. A portion of the library was removed prior to amplification to determine the number of individual library clones. The size of the G␤ 86 -105 combinatorial peptide library was greater than 10 9 .
Panning-The methods used for the panning protocol have been described in detail elsewhere (11). 0.25 g of purified PLC-␤2 in HEK buffer (35 mM HEPES, 0.1 mM EDTA, 50 mM KCl, 1 mM DTT) was added to the wells of a 96-well microtiter plate (Dynatech) and allowed to shake gently at 4°C for 1 h. This allowed PLC-␤2 to adhere to the wells of the plates. The wells with PLC-␤2 were designated as (ϩ) PLC wells. For (Ϫ) PLC control wells, 100 l of HEK buffer was added. All the wells were then blocked by the addition of 100 l of a blocking agent. For round 1 of panning 1% BSA in HEKL buffer (35 mM HEPES, 0.1 mM EDTA, 50 mM KCl, 0.2 M ␣-lactose, 1 mM DTT (adjust pH to 7.5 with KOH)) was used to block the wells. For rounds two and three 1% nonfat dry milk in HEKL buffer was used as the blocking agent. After addition of blocking agent the plate was allowed to shake gently for 1 h at 4°C. The wells were then washed four times with HEKL/blocking agent. After the final wash, 200 l of the crude lysed library was added to the wells, and the plate was allowed to shake gently at 4°C for 1 h. For the lysis procedure, please see Ref. 11.
During one set of round 3 panning the native G␤ 86 -105 peptide was added at this step at a final concentration of 40 M to compete with the library peptide clones for binding. The wells were then washed four times with HEKL/blocking agent. After the last wash 200 l of 0.1 mg/ml sonicated salmon sperm DNA in HEKL/blocking agent was added, and the plate was allowed to shake at 4°C for 30 min. The wells were then washed four times with HEKL and two times with HEK. Bound peptides on plasmids were then eluted by the addition of 50 l of elution buffer (1 mM isopropyl-1-thio-␤-D-galactopyranoside and 0.2 M KCl in HE). The plate was allowed to shake vigorously at room temperature for 30 min. All eluates were collected and combined in their respective sets, i.e. (ϩ) and (Ϫ) PLC-␤2 wells.
Subcloning into the MBP Vector-pELM3 was digested with Age1 (New England BioLabs) followed by ScaI. The digest was run on a 1% agarose gel to resolve the 5.6-kb MBP vector fragment from a 1-kb fragment. The vector band was excised and gel-purified. Plasmid DNA from round 3 of panning was digested with BspE1 and ScaI. The digest was resolved on a 1% agarose gel, and the 0.9-kb peptide-encoding fragment was excised, gel-purified, and ligated into the pELM3 MBP vector at a vector-to-insert ratio of 1:2. The MBP-library peptide fusions were expressed under an isopropyl-1-thio-␤-D-galactopyranoside-inducible promoter in the pELM3 vector in ARI-814-competent cells. Lysates from these cells were frozen and stored at Ϫ70°C .
Detection of MBP by ELISA-The procedure for the MBP ELISA has been previously described. 0.25 g of purified PLC-␤2 was added to the wells of a microtiter 96-well plate (Dynatech) and allowed to shake gently at 4°C for 1 h. In no-PLC control wells 100 l of HEK buffer was added. All wells were then blocked by adding 100 l of 2% BSA in HEK with 1 mM DTT. Blocking was carried out shaking at 4°C for 1 h. The MBP lysates were then thawed and diluted 1:100 in HEK with 1 mM DTT. Following blocking, the wells of the plate were washed 4ϫ with HEK, 1 mM DTT. 100 l of the diluted MBP lysates was then added to the wells, and the plate was allowed to shake for 1 h at 4°C. The plate was washed 4ϫ with PBS/0.05% Tween 20. The primary antibody, rabbit anti-MBP (New England BioLabs), was diluted 1/1000 in PBS. 100 l of the diluted primary antibody was added to each well, and the plate was allowed to shake for 30 min at 4°C. The plate was washed 4ϫ with PBS/0.05% Tween. The secondary antibody, goat anti-rabbit conjugated to horseradish peroxidase (Roche Molecular Biochemicals), was diluted 1/7500 in PBS. 100 l of the diluted secondary antibody was added to each well, and the plate was allowed to shake at 4°C for 30 min. The plate was then washed 4ϫ with PBS/0.05% Tween. 100 l of the True Blue horseradish peroxidase substrate (KPL, Gaithersburg, MD) was added to the wells at room temperature. Color formed in about 10 -20 min. Reactions were stopped by the addition of 100 l of 2 N H 2 SO 4 . The plate was read at 450-nm wavelength in a Spectracount plate reader, Packard Instruments. A given peptide was scored as a positive binder if it generated an ELISA signal that was greater than two standard deviations above the blank.
Replication of Results-All of the experiments shown in Figs. 1-4 were repeated at least three times with qualitatively similar results. Typical experiments are shown.

RESULTS
Screening of the Combinatorial Peptide Library-We screened a combinatorial peptide library based on the sequence derived from the G␤ 86 -105 region for binding to PLC-␤2 using the peptides on plasmids method (11). The library had greater than 10 9 individual members, and we completed three rounds of panning on PLC-␤2 immobilized in microtiter plate wells. For one group of round 3 panning we used the wild type G␤ 86 -105 peptide to compete away any peptide sequences that bound with affinities less than that of the wild type. Following panning, the DNA encoding captured peptides was subcloned into the pELM3 vector so that peptides could be expressed monovalently as a chimera with the maltose binding protein (MBP), thus enabling us to score the selected library peptides individually for binding in an MBP ELISA assay. We scored a peptide as a positive binder if it generated an ELISA signal that was greater than two standard deviations above the background signal. 25 peptides tested positively in the ELISA, and we sequenced the DNA that encoded each of these peptide clones.
The sequences of the ELISA positive clones for round 3 of panning in the absence of and in presence of the wild type peptide are shown in Tables I and II, respectively. An amino acid position was considered as "selected for" if, in the pool of selected peptides, the homology at that position was greater than 70%. The amino acid positions in each group that we considered to be part of the consensus sequence are highlighted. Wild type amino acids Lys-89 (86% homology), Val-90 (93% homology), Ser-97 (71% homology), and Val-100 (75% homology) are selected for in binding interactions with PLC. Both our studies with substituted peptides from the G␤ 86 -105 region and site-directed mutagenesis experiments of G␤ (13) predicted that amino acid positions Lys-89 and Ser-97 might be key contributors to PLC-␤2 affinity. Given the homologies of 93% for position 90 and 75% for position 100 in the pool of selected peptides from the library screen, these two residues are also likely to be important contributors to PLC-␤2 binding affinity.
When the wild type G␤ 86 -105 peptide is allowed to compete with library peptides during the third round of panning for interactions with PLC, the resulting consensus sequence is more extensive and includes the following amino acid positions: Thr-87 (82% homology), Val-90 (94% homology), Pro-94 (73% homology), Arg-96 (82% homology), Ser-97 (73% homology), and Val-100 (82% homology). Three of these positions lie within G␤ 96 -101, the region we had previously found to be the core signal transfer region for G␤ 86 -105. A number of truncated peptides were selected, suggesting that the last one or two amino acid positions of this region, G␤104 and G␤105, probably do not substantially contribute to binding affinity for PLC-␤2. We did not find any strong consensus, i.e. greater than 70% homology, for a switch of one amino acid for another at any position within G␤ 86 -105. However, we did select some peptides where the lack of homology at a consensus position is accompanied by a switch at another amino acid position. An example is seen with the T14 peptide. This peptide lacks homology at consensus position 96. It has the substitution R96C. However, this mutation is accompanied by a mutation at position 99, W99R. This second mutation might compensate for the R96C mutation. Few selected peptides lack consensus at more than one amino acid position. Among the two groups of peptides, i.e. 25 total sequences, a non-conservative mutation of the basic amino acids at positions 89 and 96 is selected for only once.
Our previous analysis of substituted peptides from the G␤ 86 -105 region led us to predict that Lys-89, Arg-96, Ser-97, Ser-98, and Met-101 would likely to be important for binding and signal transfer for PLC (4). For round 3 of panning, when the wild type peptide was allowed to compete with the binding of the library peptides, we found consensus at positions Arg-96 and Ser-97 but failed to find good consensus at position 89 and very little consensus at position 101. Interestingly, the homology at position 89 dropped from 86% to 64% when the wild type peptide was used to compete during panning. Our interpretation of this result is that the contribution of position Lys-89 to binding affinity is less than that of Arg-96 and Ser-97. In selecting Arg-96 and Ser-97 for binding, changes at position Lys-89 might be tolerated. In our substituted peptide studies the peptide G␤ 86 -105 M101N showed no measurable binding or activity. We did not find much of a consensus at position 101 in the library, but, as expected, we did not find any selected peptides with the M101N substitution.
Combinatorial Peptides That Display Synergism-During the library screening we selected peptides that bound PLC-␤2 with affinities greater than or equal to the wild type G␤ 86 -105 region, and we tested whether the binding would translate into stimulation of PLC-␤2 activity. To test if amino acid residues that make contacts important for tight binding also make contacts that directly generate signal transmission, we synthesized representative peptides that display the consensus sequence and measured the effect of these peptides on the activity of basal and G␤␥-simulated PLC-␤2. We selected peptides from both panning groups on the basis of how well conserved they were at the six consensus positions for round 3 of panning in the presence of the wild type peptide. Our criterion was that they would have consensus at 5 or more of these 6 positions. Most of the peptides that fit the cutoff criteria were from round 3 of panning in the presence of the wild type peptide (Table II). Of these 11 clones 8 showed consensus at 5 or more of the consensus positions. However, 5 of the 14 clones from round 3 of panning in the absence of the wild type peptide had consensus at 5 or more of the 6 sites. We, therefore, also included these peptides in our activity measurements.
The P9 peptide variant displays 5 of the 6 elements of the consensus sequence. The effect of the P9 peptide on PLC-␤2 basal activity is shown in Fig. 1A. This peptide stimulates PLC-␤2 with a much better EC 50 value than the wild type peptide, about 3 M as compared with 45 M, but with a maximal stimulation much lower than that of the wild type G␤ 86 -105 sequence, about 1.3-fold as compared with greater than 3-fold. A comparison of the relative EC 50s and extents of stimulation is shown in Fig. 1B.
To determine if the decreased EC 50 value for the P9 peptide is accompanied by a decrease in the K d value for PLC-␤2, we measured the binding affinity of the P9 peptide to PLC-␤2 by FRET analysis and compared it to the binding affinity of the G␤ 86 -105 wild type peptide. The binding affinity of the P9 variant is roughly the same as that of the wild type peptide (Fig. 1C). It is possible that the P9 peptide may have a lower affinity, because it does not appear to become fully saturating at 10 Ϫ5 M peptide concentration. To test for binding specificity, we measured the binding of two control peptides, G␤ 300 -313 and G␤ 64 -84. These peptides have no effect on PLC-␤2 activity, and our data indicate that they are from regions of G␤ that are important structurally but do not play a direct role in protein-protein interactions with PLC-␤2. We found that they show no measurable binding to PLC-␤2 (Fig. 1D).
These results indicate that the better EC 50 value for the P9 peptide is not accompanied by a decrease in the K d value, suggesting that there are distinct determinants within the G␤ 86 -105 signal transfer region contributing to its EC 50 and K d values for PLC-␤2. We wondered what molecular characteristic of the P9 peptide rendered it with a decrease in maximal stimulation even though it had a significantly better EC 50 value and no change in K d value as compared with the wild type G␤ 86 -105 peptide. Comparison of the amino acid sequence of the P9 peptide with the sequence of the wild type peptide shows it to differ in homology at only two amino acid positions within the core G␤ 96 -101 region, the six-amino acid core signal transfer region. The two amino acid changes are S98P and M101F. Because we have previously found that contacts within G␤ 86 -105 involved in signal transfer to PLC lie within the G␤ 96 -101 region, it is possible that one or both of these amino acids are directly involved in signal transfer. The changes at positions Ser-98 and/or Met-101 may result in a decrease in maximal observed stimulation even though there is no change in binding affinity. To determine which of these amino acids, or both of them, are responsible for the better EC 50 value and decreased maximal stimulation, we tested another library clone, the P3 peptide. The P3 peptide also has the amino acid change S98P but has the wild type residue Met at position 101. The effect of the P3 peptide clone on PLC-␤2 basal activity is shown in Fig. 2A. Like the P9 peptide, P3 also has a decreased EC 50 value and decreased maximal stimulation as compared with the wild type region peptide. The binding affinity of the P3 peptide, as measured by FRET, is about 0.5 M, not significantly different from that of the P9 peptide variant or the G␤ 86 -105 wild type peptide (Fig. 2B). Because the P9 and P3 peptides behave very similarly in binding and stimulating PLC-␤2, it is likely that the common amino acid sequence shared by these two peptides, RSPW, G␤ 96 -99 S98P, renders the P3 and P9 peptide clones with a decreased maximal stimulation and better EC 50 values even though there is not a significant change in binding affinity from the wild type peptide. This sequence motif is highlighted in Fig. 2C.
Given the low extent of stimulation, we wondered if the P9 peptide would act as a partial antagonist to G␤␥ stimulation; therefore, we measured the effect of the P9 peptide in the presence of subsaturating stimulation by G␤␥ subunits. The result was surprising. The presence of subsaturating concentrations of G␤␥ subunits increased the observed maximal stim-ulation by the P9 peptide, while not having much effect on the EC 50 value of the P9 peptide (Fig. 3A). P9 stimulates PLC-␤2 to a maximal extent of greater than 3-fold in the presence of G␤␥, whereas it only modestly stimulates about 1.3-fold by itself. The effect of varying concentrations of G␤␥ on PLC-␤2 activity in the presence and absence of maximal stimulation by the P9 peptide is shown in Fig. 3B, and this data indicate that the effect of P9 is synergistic with the effect of G␤␥.
One explanation for these results is that the P9 peptide induces structural changes in PLC so that other signal transfer regions on G␤ can form more productive interactions with PLC, resulting in a greater extent of stimulation. We have previously identified another signal transfer region on G␤ for stimulation of PLC-␤2, the G␤ 42-54 signal transfer region. We wondered if the P9 peptide, by presumably inducing key structural changes in PLC-␤2, would enable the G␤ 42-54 signal transfer region to be a better stimulator of PLC-␤2 activity, and so we tested the effect of varying concentrations of the G␤ 42-54 peptide on PLC-␤2 activity in the presence of close to saturating concentrations of the P9 peptide, 8 M. The G␤ 42-54 peptide stimulates basal activity by about 20% by itself, however, in the presence of 8 M P9 peptide its maximal extent of stimulation increases by ϳ50% (Fig. 3C). The inverse of this experiment yields similar results (Fig. 3D). Here we measured the effect of varying concentrations of the P9 peptide in the presence of subsaturating concentrations of the G␤ 42-54 peptide. The P9 peptide stimulates basal activity by about 20% by itself; however, in the presence of 1 M G␤ 42-54 its maximum extent of stimulation increases by greater than 2-fold, to about 50% (Fig. 3D). These data would suggest that the effects of the P9 signal transfer region variant peptides are likely more than additive with the G␤ 42-54 signal transfer region peptide. The cooperative effect of the P9 peptide with G␤ 42-54 is not as pronounced as that observed with G␤␥ subunits. This might indicate there are other regions of G␤, not yet characterized, that function in signal transfer or the general binding domains within G␤ contribute to the synergy and that binding determinants of G␤ 42-54 are not sufficiently strong by themselves to allow for effective signal transfer. These alternatives will have to be experimentally resolved and will require combinatorial analysis of the G␤ 42-54 region as well. We tested whether the G␤ 86 -105 wild type peptide had the same effect as the P9 variant on G␤␥ and G␤ 42-54 stimulation of PLC-␤2. G␤ 86 -105 stimulates PLC-␤2 to a higher extent, greater than 3-fold, than the P9 variant. The presence of saturating concentrations of G␤ 86 -105 peptide, 80 M, had little effect on the observed maximal stimulation by G␤␥ (Fig. 3E). We next measured the effect of 5 M G␤ 42-54 peptide on G␤ 86 -105 stimulation. Subsaturating concentrations of G␤ 42-54 increase the maximal observed stimulation attained by the G␤ 86 -105 (Fig. 3F). These data indicate that the G␤ 86 -105 and G␤ 42-54 signal transfer regions have the capability to act in a synergistic manner to stimulate PLC-␤2.
Combinatorial Peptides That Behave as Antagonists/Partial Agonists-Two other peptides from the library, T7 and T8, display all six elements of the consensus sequence but differ within the G␤ 96 -101 region from both the wild type sequence and the P9/P3 sequences. We measured the effect of these variant peptides on PLC-␤2 activity. The T8 peptide has an EC 50 value of about 50 M, in the same range as the G␤86 -105 wild type peptide (Fig. 4A). However, it only stimulates PLC activity about 60%. However, in contrast to the P9 and P3 peptides, the T8 peptide is capable of inhibiting most of G␤␥ stimulation (Fig. 4B). Thus, the T8 variant of the G␤ 86 -105 signal transfer region is an antagonist as well as a partial agonist. Measurement of the binding affinity of the T8 peptide for PLC-␤2 by FRET shows that the K d value in the 1-2 M range is similar to the wild type G␤ 86 -105 peptide (data not shown).
The sequence of the T8 variant differs from the wild type sequence at only six amino acid positions. Three of these changes occur within G␤ 96 -101, the six-amino acid region that we have previously found to be the core signal transfer region of the G␤ 86 -105 peptide for stimulation of PLC-␤2. These substitutions are S98P, W99R, and V100F. It is possible that one or all of these three amino acids, residues 98, 99, and 100 of G␤ 86 -105, directly contribute to signal transfer for PLC-␤2. Substituting these residues within G␤ 86 -105 may render this signal transfer region with a lower maximal stimulation even though it has EC 50 and K d values that are similar to those of the wild type. We have already found that positions 98 and 99 are likely to be directly involved in contacts for signal transfer to PLC-␤2, and the G␤ 96 -99 sequence motif RSPW likely renders a higher potency but lower maximal stimulation of PLC-␤2 compared with the wild type sequence of RSSW. The RSPW sequence motif also appears to be responsible for synergizing with other signal transfer regions on G␤ to stimulate PLC-␤2. To better understand how the changes of the T8 peptide result in a decrease in maximal effect on PLC-␤2 and loss of synergism with G␤␥ subunits, we tested the effect of another library peptide, the T7 peptide, on PLC-␤2 activity. The T7 peptide is very similar in sequence to the T8 peptide in the G␤ 96 -101 region. It also has the amino acid substitutions W99R and V100F. However, it has the wild type amino acid at position 98 and has the substitution M101L. The effect of the T7 peptide PLC-␤2 basal activity is shown in Fig. 4C. This peptide, like the T8 peptide, has an EC 50 value in the same range as the wild type sequence peptide and also only minimally stimulates PLC-␤2, around 1.5-fold. The T7 peptide clone affects G␤␥ stimulation of PLC-␤2 in a manner similar to that of the T8 peptide (Fig. 4D) and has a K d value for PLC-␤2 that is similar to the wild type sequence (data not shown). Thus, the T7 peptide is also a partial agonist and antagonist of PLC-␤2. We noted, however, that the T7 peptide fully inhibits G␤␥ stimulation of PLC-␤2, suggesting that in the presence of G␤␥ it is unable to stimulate PLC-␤2. These results indicate that it is likely the shared amino acid sequence between these two peptides in the G␤ 96 -101 region, highlighted in Fig. 4E, that renders them as antagonists/partial agonists. Specifically, these residues are Arg-96, Ser-97, Arg-99, and Phe-100. It is interesting that the substitutions W99R and V100F affect the maximal extent of stimulation for this signal transfer region peptide but not its EC 50 value. This indicates for residues 99 and 100 of G␤ that the EC 50 and K d values are independent of the maximal extent of stimulation.

DISCUSSION
Through a combinatorial screen we have identified variant peptides of the G␤86 -105 signal transfer region that have binding affinities for PLC-␤2, are very similar to that of the wild type G␤86 -105 peptide, but have very different signal transfer properties. The range of signaling behaviors we observe provides clues toward initial understanding of the underlying mechanisms of signal transfer. From these analyses we have identified two key features. First, for the wild type G␤86 -105 peptide we find that the EC 50 for the stimulation of PLC-␤2 is over 20-fold to the right of the binding affinity and that the position of the EC 50 is inversely related to the extent of stimulation. Second, intrinsic stimulation as evidenced by partial agonist activity could co-exist with two very opposite behaviors: synergistic stimulation with G␤␥ or antagonist activity in the presence of G␤␥. Both these features provide mechanistic insights into how signal transfer might occur. The residues involved in the distinct signaling functions are summarized in Table III Relationship between Binding Affinity, EC 50 , and Maximal Stimulation-The FRET experiments showed that all of the variant peptides had binding affinities similar to that of the wild type peptide. Our lack of success in identifying any variant of G␤86 -105 region that had a significant increase in binding affinity suggests that the G␤86 -105 region may have been engineered for relatively modest binding affinity. A number of positively charged residues were identified in the library screen as important for binding affinity, and it is likely that electrostatics contribute to at least part of the interaction affinity of the G␤ 86 -105 region for PLC-␤2. We have also found this to be true for the other signal transfer region G␤ 42-54. One reason we did not find any higher affinity peptide is that it might be energetically more costly to reengineer a protein-protein interaction for higher affinity when most of the binding affinity is contributed by electrostatics rather than hydrophobic interactions.
The 86 -105 region of G␤ has been shown to interact with a number of other G␤␥ effectors, including adenylyl cyclases. One requirement for this diverse interaction capability might be flexibility that results in an inherently low affinity interaction surface. A plausible model for signal transfer from G␤␥ to PLC-␤2 from inherently low affinity binding regions, such as the G␤ 86 -105 region, would involve initial interactions driven by electrostatic forces using induced fit mechanisms for additional dynamic contacts to affect activity changes in the effector (PLC-␤2). These signal transfer regions might be interspersed with higher affinity binding regions that utilize hydrophobic forces and are more selective for specific effectors. Such a model  would also explain the role of G␥ subunits in stimulation of effectors (14,15). Of the library peptide variants selected and tested, all show both conserved and variant residues within the G␤ 96 -101 region, the core signal transfer region. Residues Arg-96 and Ser-97 are part of the consensus sequence for selected library clones. These residues likely contribute to binding affinity. Other residues within the G␤ 96 -101 sequence are somewhat varied, suggesting that amino acid substitutions in this region might be tolerated with no significant changes in binding affinity. These changes, however, affect signal transfer as evidenced by the change in maximal stimulation. As indicated in Table III, these data suggest that contacts made by positions 98 -101 of G␤ that are not involved in binding affinity are involved in signal transfer. An example is observed with the T8 peptide variant. Here the amino acid substitutions W99R, V100F, and M101L render this peptide with a better EC 50 but decreased maximal stimulation even though there is no significant change in overall binding affinity. Contacts made by one or all of these positions in the wild type peptide might be important for signal transmission but not for binding affinity.
None of the variants of G␤ 86 -105 that we identified in this screen had a maximal stimulation as great as that of the wild type sequence. For each variant peptide tested, its maximal stimulation was less than 2-fold, whereas the maximal stimulation of the wild type G␤ 86 -105 peptide is generally greater than 3-fold. Thus, the G␤ 86 -105 signal transfer region appears to be optimized in terms of the efficacy with which it can regulate PLC-␤2 basal activity. This is to be true, at least, for variants of G␤ 86 -105 with binding affinities equal to or greater than the wild type sequence. It is possible that there might be variants of G␤ 86 -105 that have weaker binding affinities for PLC-␤2 but are more efficacious in stimulating PLC-␤2 activity. An example of this scenario was seen with a substituted peptide from the G␤ 42-54 region. The substituted peptide G␤ 42-54 R48A displayed a weaker EC 50 that was accompanied by an increase in maximal stimulation (6). That none of the peptide variants could transfer signals to the same extent as the wild type region also indicates that positions within the core signaling region G␤ 96 -101 that are not important for binding affinity can be important for signaling; therefore, the roles of binding and signal transfer for this region can be resolved at an amino acid level.
Synergism and Antagonism-The P9 peptide was found to act in a synergistic fashion with G␤␥ and the second G␤ signal transfer region, G␤ 42-54 and displayed no antagonist properties, whereas the other set of library peptides, T7 and T8, acted as antagonists to G␤␥ simulation. However, by themselves, all the peptides stimulated PLC-␤2 only 30 -60%. These data would suggest that the set of contacts leading to the small amount of stimulation is distinct from those interactions that generate synergism with other signal transfer regions on G␤, including the G␤ 42-54 region to yield extensive (severalfold) stimulation. As summarized in Table III, amino acids Arg-96 and Ser-97 of G␤, residues common to both sets of peptides are important for binding and for generating low efficacy signal transmission. A distinct contact from the amino acid at position 99 appears to be required for generating additional efficacy for signal transmission by synergizing with other signal transfer regions on G␤ such as the G␤ 42-54 region. The synergistic effect of the wild type G␤86 -105 peptide with the G␤ 42-54 signal transfer region is not as pronounced as that observed with G␤␥ subunit. This observation suggests that there might be additional regions in G␤ that play a role in synergistic signal transfer. Whether these are additional signal transfer regions or general binding domains need to be experimentally determined.
Overall we propose that synergism occurs because the binding of one signal transfer region may enhance the efficiency of signal transfer from the second region, because it appears that there are multiple domains in PLC-␤2 that are capable of receiving signals (16,17). Such a model would be based on the idea that interaction of the G␤ 86 -105 region with PLC-␤2 may induce several discrete conformational changes in this effector. These conformational changes can be functionally resolved from one another, and this is illustrated by the T7 and T8 variants where the small extent of activation is not accompanied by synergism with G␤␥ subunits. Both of these peptides have the amino acid substitution W99R. This substitution, although not affecting contacts involved in binding affinity or contacts for generating a low efficacy signal transmission, might prevent contacts with PLC-␤2 important for synergism with other signal transfer regions. Thus the resultant functional effect is antagonism in the presence of G␤␥ subunits.
Although parts of our model are speculative, our data clearly indicate that binding does not directly translate into signal transmission for the G␤86 -105 signal transfer region. Signal transfer from this G␤ region to PLC-␤2 must involve at least a partially non-overlapping set of protein-protein contacts as compared with binding. This conclusion is supported by two sets of observations: first, individual amino acids important for binding can be resolved from those important for signaling, and second, some of these signaling residues appear to play little or no direct role in contributing to the overall binding affinity. There have been similar findings for the resolution of binding from signal transfer in other signaling systems as well. Hamm and coworkers (18) have shown that at least two distinct regions are involved in interactions between Gt␣ and ␥ subunits of cGMP phosphodiesterase and that only one of these appears to be important for stimulation of cGMP phosphodiesterase. Similarly for the Rho family member Cdc42, interactions with its effector phospholipase D1 (PLD1), a region from Cdc42 that is important for stimulating PLD1 activity is not involved in binding affinity (19). The data from Gt␣/phosphodiesterase ␥ (18), Cdc42/PLD (19), Gs␣/adenylyl cyclase (5), and the G␤␥/ PLC-␤2 systems indicate that the design of separating signal transfer regions from general binding domains may be widely used for signal transfer in G protein systems.
In summary, the studies presented here provide an initial mechanistic model for signal transfer where initial contact within signal transfer regions and possibly concurrent interactions between general binding domains induces the effector to attain a state of high receptivity such that signals from multiple signal transfer regions can be effectively transmitted. Such a model predicts a coordinated set of induced fit interactions between the signal transfer regions on the G protein subunits and signal-receiving regions of the effectors. Future experiments that provide direct information about dynamics of the interacting structures involved in signal transfer from G␤ to PLC-␤2 will be needed to test this prediction.