The Coiled-coil Region of the G Protein β Subunit

The β and γ subunits of the heterotrimeric G proteins remain tightly associated throughout the signaling cycle as the βγ dimer interacts with Gα, receptors, and effectors. A coiled-coil structure involving α-helical segments at the N termini of the β and γ subunits contributes to the dimerization interface and has been implicated in effector signaling in yeast. Scanning mutagenesis of the coiled-coil region of the mammalian β1 subunit was performed to examine the effect of point mutations on βγ assembly and effector signaling in COS cell cotransfection assays. In addition to the E10K mutation described previously, mutations A11E, L14E, and I18E in β1 were found to block βγ association, as evidenced by the failure of the Gβ mutants to undergo cytosolic translocation with cotransfected nonisoprenylated Gγ. Although none of 14 β1 point mutations prevented the βγ-dependent activation of the c-Jun N-terminal kinase (JNK) effector pathway, the D20K point mutation enhanced JNK but not phospholipase C-β2 activation. These findings implicate the coiled-coil region of Gβ in JNK signaling, provide further evidence that the structural features of the βγ complex mediating effector regulation may differ among effectors, and identify single codons in the mammalian β subunit where mutation might yield a phenotype of defective signal transduction.

Heterotrimeric GTP-binding regulatory proteins (G proteins) contain one each of subunits ␣, ␤, and ␥ and transduce signals from activated plasma membrane receptors to intracellular second messenger-generating effector proteins (1)(2)(3). The ␤ and ␥ subunits exist as a tightly bound complex, which can only be dissociated under denaturing conditions and which functions as an entity throughout the signaling cycle. Both the G protein ␣ subunit and the ␤␥ complex transmit signals to effector molecules (4 -6). The recent determination of the three-dimensional structure of G protein heterotrimers (7,8) and a free ␤␥ heterodimer (9) revealed that, in addition to a seven-bladed ␤-propeller, the ␤␥ complex contains a two-stranded parallel ␣-helical coiled-coil structure involving N-terminal segments of ␤ and ␥.
Coiled coils consist of two or more ␣-helices that wind around each other to form a left-handed supercoil (10,11). They form a key structural element in many proteins, imparting an extended rodlike structure to ␣-fibrous proteins such as myosin, and comprising the dimerization domain in certain regulatory proteins such as the bZIP transcriptional factors c-Fos and c-Jun (12,13) as well as GCN4 in yeast (14). Proteins that form coiled coils contain a repeating pattern of seven amino acids (a heptad repeat, with residues within a heptad designated abcdefg), in which the first (a) and fourth (d) position of the heptad are occupied by hydrophobic residues, and other positions are occupied by polar residues, resulting in a hydrophobic surface along one face of each ␣-helix that is buried within the core of the assembled complex. The complex may be further stabilized by interhelical salt bridges flanking the hydrophobic interface ( Fig. 1, A and B) (10,15). The presence of a coiled coil in the G protein ␤␥ complex had been anticipated, based on analysis of primary sequence data by computer algorithm (16) and on the results of computer modeling and site-directed mutagenesis (17,18). Residues of the yeast G␤ homolog STE4p in a region homologous to the coiled-coil region of G␤ 1 have been implicated in effector signaling ( Fig. 1A) (19,20). To better understand the role of the coiled coil in ␤␥ dimerization and effector signaling, systematic mutagenesis of the coiled-coil region of ␤ 1 was performed and the properties of point mutants examined in transient transfection assays.
The finished G␤ 1 point mutant constructs all contained the sequence GAATTCAAGATG at their 5Ј ends (starting methionine codon underlined), were followed after the stop codon by an XbaI site at their 3Ј end, and were ligated between EcoRI and XbaI sites of pCDM8.1. Constructs in pCDM8.1 were amplified in Escherichia coli MC1061/P3 (Invitrogen). The resulting plasmid preparations were purified by column chromatography (Qiagen Maxiprep kits). The DNA sequence of all inserts was verified by the chain termination method (28) using Sequenase 2.0 (U. S. Biochemical Corp.).
Protein Expression and Immunoblotting-Growth, maintenance, transfection (29), and fractionation of COS-7 cells were as described previously (22). Protein was determined by the method of Bradford (30) using bovine serum albumin as a standard. Crude particulate or cytosolic fractions, or detergent lysates of whole cells, were separated on 11% slab gels by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) (31), and electrotransferred onto polyvinylidene difluoride membranes in Dunn's buffer (32). For analysis of G␥ subunit expression, the Tricine gel system of Schä gger and von Jagow (33) was employed as indicated. Detection of G␤ 1 subunits employed primary antibodies generated in rabbits, antibody SW against a peptide corresponding to residues 330 -340 at the C terminus of ␤ 1 , or antibody RA against an internal sequence corresponding to residues 256 -265 as described previously (34). Detection of G␥ subunits employed the primary antibodies LVKG and EDPL, generated in rabbits against synthetic peptides corresponding to residues 53-70 near the C terminus of ␥ 1 and residues 47-64 near the C terminus of ␥ 2 , respectively, as described previously (24). Detection of HA epitope-tagged JNK in COS cell lysates utilized mouse monoclonal HA.11 (Berkeley Antibody Co.). Secondary detection utilized 125 I-Protein A for rabbit polyclonal antibodies and 125 I-labeled goat anti-mouse IgG for mouse monoclonal antibodies, followed by autoradiography on film or a storage phosphor screen (Molecular Dynamics PhosphorImager). Quantitation of band intensity and digital imaging on storage phosphor screens employed ImageQuant software (version 1.11, Molecular Dynamics).
Phosphoinositol Phospholipase C Activity Asssay-The PI-PLC activity of transfected cells was estimated by a modification of the procedure of Berridge et al. (35) as described previously (36,37).
JNK Activity Assay-The assays for JNK activity were essentially as described by Coso et al. (27). Approximately 2.5 ϫ 10 6 COS-7 cells were plated into 75-cm 2 flasks and incubated at 37°C overnight. On the following day, the cells were transfected by the DEAE-dextran method (29), using a total of 15 g of DNA/cotransfection, typically including 5 g of HA-JNK (27), 5 g of G␥, and 5 g of G␤ or point mutant. Vector DNA was added where necessary to keep the total amount of plasmid DNA per flask constant. The remainder of the assay was as described (27), except that mouse monoclonal HA.11 was used for the immunoprecipitations (3 l of HA.11 ascites fluid/900 l of detergent lysate). GST-c-Jun-(1-79) substrate was from Stratagene.
Computer Analysis of Protein Sequences-The schematic graphical representation of the G␤ and G␥ coiled-coil regions employed the Heli-calWheel program of the University of Wisconsin Genetics Computer Group (GCG) (38) with a specification of 102°rotation/residue (3.5 residues/turn). The Pileup algorithm was also from GCG (38). Analysis of primary protein sequences for regions of predicted coiled coil was performed online, and employed both the COILS algorithm of Lupas et al. (16) 2 and the PAIRCOILS algorithm of Berger and Kim (39). 3

Conservation of the Coiled Coil in G␤␥
Complexes-X-ray crystallographic determination of the structure of two mammalian G protein heterotrimers (an ␣ i1 ⅐␤ 1 ⅐␥ 2 complex (7) and a complex between an ␣ t -␣ i1 chimera and ␤ 1 ⅐␥ 1 (8)) and free G␤ 1 ␥ 1 complex (9) reveals the presence of a two-stranded parallel coiled coil involving N-terminal portions of G␤ and G␥. Analysis of the protein sequences encoded by all vertebrate, invertebrate and plant G␤ cDNAs identified to date employing the COILS (16) or PAIRCOILS (39) computer algorithms predicts an N-terminal coiled coil, with the exception of the Schizosaccharomyces pombe G␤ homolog Gpb1 (40) (Fig. 1A and data not shown). This indicates the G␤␥ coiled coil is a highly conserved structure, which is likely to be of fundamental importance to the function of ␤␥ heterodimers.
The Role of Glu-10 in G␤␥ Assembly-Previous computerassisted modeling of the ␤ 1 ␥ 1 coiled coil (18) predicted an electrostatic interhelical interaction between residue Glu-10 of G␤ 1 and Lys-23 of ␥ 1 , an interaction that was subsequently confirmed in the crystal structure (9). An acidic residue in this position in G␤ subunits and a basic residue in G␥ correspond-ing to Lys-23 in ␥ 1 are highly conserved among phylogenetically diverse species. Site-directed mutagenesis of the G␤ 1 cDNA to produce the ␤ 1 -E10K mutant resulted in a construct that failed to assemble with ␥ 2 * in a cotransfection assay, suggesting this electrostatic interaction was critical for ␤␥ heterodimer formation (18). To further explore the effect of mutations at this position, point mutants ␤ 1 -E10R and E10A were constructed. Like Glu and Lys, Arg and Ala residues at this position would be expected to support the ␣-helical secondary structure at this site known from the crystal structure (7,9). For comparison, a ␤ 1 -E12K mutant was generated and its properties studied in parallel. Unlike the Glu at position 10, which occupies the g position of the heptad repeat, Glu-12 maps FIG. 1. G protein ␤ heptad repeats and the G␤␥ coiled-coil region. A, alignment of selected G␤ N-terminal sequences adapted from the Pileup algorithm of the GCG (38) with residue numbers indicated in the margins. Positions in the heptad repeats of G␤ 1 in the ␤ 1 ␥ 1 coiled coil (9) are indicated in italics above its sequence. The asterisks below residues in the S. cerevisiae G␤ (STE4) sequence indicate the positions of single codon mutations found in signaling-defective alleles identified by genetic screening (19,20). The G␤ sequences shown include bovine G␤ 1 (GenBank™ accession number M13236), Drosophila Gbb (Dros. Gbb; GenBank™ accession number M22567), Dictyostelium G␤ (Dicty.; GenBank™ accession number X73641), mouse G␤ 5 (Gen-Bank™ accession number L34290), maize G␤ 1 (GenBank™ accession number U12233), Nicotiana tabacum G␤ 1 (Tobacco; GenBank™ accession number Z84820), Caenorhabditis elegans G␤ (C. eleg.; GenBank™ accession number X17497), squid G␤ (GenBank™ accession number X56757), S. pombe Gpb1 (GenBank™ accession number L28061), and S. cerevisiae G␤ (S. cerev., STE4; GenBank™ accession number M23982). B, schematic representation of the G␤ 1 and G␥ 1 N-terminal regions by the HelicalWheel algorithm of the GCG (38), specifying 102°rotation/ residue. Positions within the heptad repeats are indicated with letters a-g, and the locations of G␤ 1 residues Glu-10, Ala-11, Leu-14, Ile-18, and Asp-20 highlighted in this study are indicated with numbers. Note that, in the G␤ 1 ␥ 1 crystal structure (9), only G␤ residues Glu-3 to Ala-24 and G␥ 1 residues Glu-11 to Glu-25 within this region were found to be in ␣-helical conformation. C, summary of G␤ 1 point mutants employed in this study. Mutations are indicated in single-letter amino acid code above the residues they replace in the native sequence.
to a heptad b position (Fig. 1, A and B).
To assess the ability of the mutants to assemble with G␥, a cotransfection assay in COS cells employing the nonisoprenylated ␥ 1 -C71S mutant, ␥ 1 *, was used (Fig. 2). As shown previously (18,(22)(23)(24), cotransfection of wild-type ␤ 1 with ␥ 1 * produced a large increase in cytosolic ␤ and ␥ immunoreactivity above that seen with transfection of either construct alone (Fig.  2, cf. lanes 2 and 3 with lane 4). This results both from the failure of ␤␥ complexes containing nonisoprenylated ␥ subunits to undergo membrane targeting (22,41) and from the mutual stabilizing effect of heterodimerization on G␤ and G␥ (42)(43)(44). This assay can be thus used to assess the ability of G␤ and G␥ mutants and chimeras to interact (18,23,24). As seen in Fig. 2, although the ␤ 1 -E12K mutant demonstrates ␥-dependent expression nearly to the level of wild-type ␤ 1 and comparably promotes ␥ 1 * expression in the soluble fraction (Fig. 2, cf. lanes 4 and 12; see also Figs. 3 and 4), neither ␤ 1 -E10K nor ␤ 1 -E10R exhibits significant ␥-dependent expression or enhances the steady-state expression of ␥ 1 * (Fig. 2, lanes 5-8). The ␤ 1 -E10A mutant showed an intermediate level of expression, suggesting neutralization of negative charge was less deleterious to ␤␥ assembly than charge reversal at this position (Fig. 2, lanes 9  and 10). As can be seen in Fig. 2, analysis of the particulate fractions in these experiments is much less informative due to the presence of endogenous G␤ immunoreactivity and the predominant localization of G␤␥* complexes to the cytosolic fraction (22). Such analyses were therefore not included in evaluation of the G␥* compatibility of the G␤ coiled-coil point mutants described below.
Scanning Mutagenesis of the Coiled-coil Region of G␤ 1 -To more thoroughly evaluate the role of individual residues in the coiled-coil region of G␤ in the interaction with G␥, scanning mutagenesis was performed on 16 consecutive residues from the N-terminal ␣-helix of G␤ 1 forming the coiled coil with ␥ (7, 9) (Gln-6 through Ala-21; see Fig. 1C). Neutral and basic residues were mutated to Glu, whereas acidic residues were mutated to Lys, both amino acids compatible with ␣-helices. It was felt that these charged substituents might be potentially more disruptive to the G␤␥ coiled coil than Ala, which is widely represented among all seven positions of repeats in known coiled coils (16,45).
The ability of the 16 point mutants to assemble with ␥ 1 * was assessed in COS cells by the cotransfection assay. Blots of the soluble fractions of COS cells cotransfected with wild-type or mutant G␤ cDNAs and ␥ 1 * were probed with G␤ or ␥ 1 -directed antibodies, and further developed with 125 I-Protein A (Fig. 3). This allowed quantitation of the immunoreactive bands and estimation of the ␥-dependent G␤ expression levels and the ␤-dependent increment in ␥ expression for the point mutants relative to wild-type G␤ 1 (Fig. 4). The strong correlation between the steady-state ␤ and ␥ expression associated with any particular G␤ mutant-␥ 1 * combination in this cotransfection assay is readily seen from the data in Figs. 3 and 4. This likely reflects the much greater stability of the two subunits when complexed in heterodimeric form (42)(43)(44).
In addition to the ␤ 1 -E10K mutant, three other G␤ 1 point mutations were found to largely abrogate interaction with ␥ 1 * in this assay: A11E, L14E, and I18E (Figs. 3, 4). This trio of G␤ 1 point mutants also interacted poorly with the ␥ 2 * isoform in parallel experiments (data not shown). These three residues map to heptad a and d positions, where Glu is uncommon in coiled coils (16,45), tending to disrupt the interhelical hydrophobic interface. Indeed, in the G␤ 1 ␥ 1 crystal structure, all three residues are seen to form hydrophobic interactions with residues in ␥ 1 (9). Interestingly, two other heptad d position mutants in this series, L7E and A21E, were found to be more compatible with ␥ assembly. Among the mutations involving residues in the heptad e and g positions of the G␤ 1 coil, the E10K (g) mutation had a profound effect on ␥ assembly as noted previously (18), whereas the R8E (e) and K15E (e) mutants showed intermediate expression levels. The ␤ 1 -Q17E mutant (g) had a phenotype indistinguishable from wild-type G␤ 1 in this assay ( Fig. 3 and 4). Three other G␤ 1 point mutants employed in functional studies described below (A26E, D27N, and A28E), from a short non-helical region immediately following the coiled coil ( Fig. 1C) (9), demonstrated ␥-dependent expression levels in the range of 70 -100% of wild-type G␤ 1 in the same assay (data not shown).
JNK Effector Signaling of ␥-competent G␤ Coiled-coil Mutants-Recently ␤␥ subunits were found to activate the MAPK/ ERK (46 -48) and JNK (49) pathways in vertebrates, signaling cascades with many homologies to the ␤␥-driven pheromone response pathway in the yeast Saccharomyces cerevisiae (50). Genetic screening in yeast has identified several pheromone signaling-defective G␤(STE4) mutants with point mutations in the vicinity of the putative coiled-coil region (Fig. 1A) (19,20). Because the yeast G␤ coiled-coil region has been implicated in effector signaling, the ability of the ␥-competent G␤ 1 coiled-coil point mutants to activate the JNK pathway was evaluated in a cotransfection assay in COS cells involving an HA epitopetagged JNK reporter (49). The ␥-competent mutations from the G␤ 1 N-terminal ␣-helical segment included two corresponding to yeast residues mutated in dominant negative alleles. The ␤ 1 D20K mutation maps to STE4 residue Lys-55, where a K55E mutation produces a signaling defect (19); and the G␤ 1 A21E mutation involves the residue homologous to Ala-56 in STE4, where an A56P point mutation abrogates effector signaling (20) (Fig. 1A). In addition, three ␥-competent mutants were assayed that involved residues from a short interhelical region (9) of the G␤ chain immediately following the coiled coil: A26E, D27N, and A28E (Fig. 1A). The D27N mutation corresponds to the dominant-negative D62N STE4 mutation (19,20) and Ala-28 in G␤ 1 corresponds to Ala-63 in STE4, another site of point mutation in yeast causing defective effector signaling (20). The A26E mutation was made as well for control purposes.
Cotransfection with G␥ 2 was used instead of G␥ 1 for these assays, as the former supported higher levels of G␤-dependent JNK activity in pilot experiments. Previous work has shown that G␤ 1 interacts readily with both G␥ 1 and G␥ 2 (51)(52)(53) and has excluded the coiled-coil region of G␤ 1 ⅐␤ 2 chimeras as a major site of discrimination between these two G␥ isoforms (23,51,54).
All of the ␥-competent G␤ point mutants were able to activate the JNK signaling pathway in the COS cell cotransfection assay to at least 50% of the wild-type level (Fig. 5). This was in contrast to a negative control construct encoding wild-type G␤ 5 , recently found to lack MAPK/ERK and JNK effector signaling ability (37) (Fig. 5). The A21E, D27N, and A28E mutants were found to be JNK-competent (Fig. 5). Taken together with evidence of their G␥*-dependent expression in the cytosol (Figs. 3 and 4 and data not shown), the ability of these G␤ 1 mutants to activate the JNK effector pathway provides evidence of their ability to form functional G␤␥ complexes. An unexpected finding was that the D20K point mutant produced consistently greater JNK activation than wild-type G␤ under these assay conditions (Fig. 5). This was true despite the slightly lower expression level of the D20K mutant relative to wild-type G␤ 1 in the G␥* cotransfection assay (Figs. 3 and 4). This phenomenon was further explored in experiments described below.
Comparison of the JNK and PLC Signaling Ability of the ␤ 1 -D20K Mutant-To ascertain whether the ␤ 1 -D20K mutant was enhanced in its ability to signal via more than one effector FIG. 4. Summary of relative expression levels of G␤ coiled-coil point mutants and G␥ 1 * in the soluble fraction of cotransfected COS cells. G␤ or G␥ immunoreactive band intensity from three or four independent transfections was quantitated on storage phosphor screens (one such experiment is shown in each panel of Fig. 3), and expressed as a percentage relative to the G␤ or G␥ band intensity upon wild-type (wt) G␤ 1 and G␥ 1 * cotransfection in the same experiment (100%). Within each experiment, the G␤ or G␥ band intensity under conditions of G␥ 1 * transfection alone was set as 0% and subtracted from all values. Although only trace G␤ immunoreactivity was seen in the soluble fraction upon transfection of G␥ 1 * alone, G␥ 1 * immunoreactivity was always detectable under these conditions (usually ϳ10% of peak G␥ intensity; see Fig. 3) so that the G␥ expression summarized here represents the increment in G␥ signal seen upon addition of wild-type or mutant G␤ cDNA to the transfection mix. Error bars indicate S.E. pathway, its JNK and PLC signaling ability was compared with wild-type G␤ at different doses of transfected cDNA (Fig.  6). Whereas the JNK stimulation of the ␤ 1 -D20K mutant was greater than that of wild-type over a range of doses of transfected DNA (Fig. 6A, left panel), its PLC-stimulating ability was at a level near or below that of the wild-type under two different PLC assay conditions (Fig. 6B). The relative PLCstimulating ability of the ␤ 1 -D20K mutant compared with wildtype paralleled their relative G␥-dependent expression levels (Fig. 6A, right panel; see also Figs. 3 and 4). The enhanced JNK-stimulatory activity of the ␤ 1 -D20K mutant relative to wild-type G␤ 1 was not due to increased expression of cotransfected HA-JNK reporter or G␥ 2 (Fig. 6A, right panel). This finding implied that the enhanced signaling of the D20K mutant was pathway-selective and that the G␤␥ coiled-coil region was involved in JNK signaling.
This study was undertaken, therefore, to better understand the protein-protein interactions of the G␤ coiled coil by the use of point mutations in this region. The ␥ 1 *-compatibility of the series of G␤ 1 Glu-10 point mutants characterized in the present work is compatible with the G␤ 1 ␥ 1 crystal structure demonstrating a salt bridge between this Glu and Lys-23 in ␥ 1 (9); the mutations inducing charge reversal, E10K and E10R, were tolerated more poorly than the E10A substitution. It is nevertheless of interest that charge reversal at this heptad g position was more deleterious than at position e, where R8E and K15E mutations were compatible with ␥ 1 * association and ␤␥ effector signaling function, even though G␤ Lys-15 is known to interact electrostatically with ␥ 1 Glu-18 (9). It may be that these latter FIG. 5. c-Jun N-terminal kinase activity in COS cells cotransfected with wild-type or ␥-competent G␤ coiledcoil point mutants and wild-type G␥ 2 . A, COS cells were cotransfected with HA-JNK reporter and the G␤ and G␥ constructs indicated, and assayed for JNK activity in HA immunoprecipitates of cellular lysates as described under "Experimental Procedures." Shown are autoradiograms of the phosphorylated GST-c-Jun-(1-79) substrate (mass ϳ35 kDa) and immunoblots of the corresponding cell lysates showing HA-JNK expression (mass ϳ44 kDa) after electrophoretic separation by 11% SDS-PAGE. Wild-type G␤ 5 lacks JNK stimulatory activity and was used as a negative control (37). B, summary of relative JNK stimulation in COS cells from four independent cotransfections such as those shown in A. In each experiment, the phosphorylation of GST-c-Jun-(1-79) substrate under each transfection condition was quantitated by exposure to storage phosphor screens and expressed as a percentage of that seen with wildtype G␤ 1 and G␥ 2 cotransfection (100%). The basal level of substrate phosphorylation (vector (Vec) only ϩ HA-JNK transfection; e.g. leftmost lanes in A) within each experiment (0%) was subtracted from all values. Error bars indicate S.E. Comparison of the JNK activity of G␤-D20K ϩ ␥ 2 with wild-type G␤ ϩ ␥ 2 transfected cells by a paired two-tailed t test yielded a p value Ͻ 0.001. residues are more important in orienting the G␤ coil during assembly or ensuring the specificity of ␤␥ pairing than contributing to interhelical stability (59).
Three G␤ mutations within the hydrophobic core were found to block ␥* interaction in the COS cell assay: A11E, L14E, and I18E. This does not exclude the possibility that other G␤ mutations might also prevent ␤␥ pairing, nor should one infer that any mutation in these codons would be equally disruptive. Similar core heptad a and d mutations in yeast G␤ homolog STE4p, including L49E and I53E, which correspond to G␤ 1 mutants L14E and I18E, were found to partially inhibit interaction with yeast G␥ STE18p in two-hybrid assays (Fig. 1A). 4 These results suggest that the coiled-coil interactions between G␤ and G␥ make a critical contribution to heterodimerization across a range of structurally diverse G␤␥ complexes.
Mutations in heptad repeat segments that impair coiled-coil interactions of structural proteins such as keratin and spectrin have been linked to inherited human diseases (60 -62). As shown in this study, assembly of the ␤␥ heterodimer is also vulnerable to point mutations in the coiled-coil domain. Such coiled-coil mutations may presage identification of similar lossof-function G␤ mutations in humans. Such G␤ loss-of-function might impart a disease phenotype at developmental stages and/or in specific cells where redundancy among G␤ isoforms is minimal (e.g. in retinal rod or cone cells where a single G␤ isoform predominates; Ref. 5).
Given the homologies between multiple components of the yeast S. cerevisiae pheromone-response pathway and mammalian MAPK and JNK pathways (63), it is perhaps surprising that no loss-of-function mutations of G␤ 1 were identified among the ␥-competent coiled-coil mutants in the JNK screening assay. Among the G␤ 1 mutants were several involving residues (Asp-20, Ala-21, Asp-27, and Ala-28) homologous to yeast STE4 codons where loss-of-function mutations have been identified in genetic screens (19,20) (Fig. 1A). It is possible that the yeast genetic screening methods are more sensitive to mutations than the overexpression paradigm used for JNK assay, and that an alternative assay method might reveal loss-offunction among the pool of G␤ 1 mutants. Inasmuch as the molecular components of neither the yeast pheromone-response nor the mammalian G protein JNK pathways are fully resolved, it may also be that mechanistic differences account for failure to uncover JNK-signaling defective G␤ 1 mutants in this study.
Facilitation of JNK signaling by the ␤ 1 -D20K mutant implicates the coiled-coil region of the ␤␥ complex in the mechanism of JNK activation. In the G␤ 1 ␥ 1 crystal structure G␤ Asp-20, in a heptad c position (Fig. 1, A and B), is seen to project from the face of the coiled coil opposite the ␤-propeller, where it might be free to interact with other proteins (9). It is interesting to note that the residue corresponding to G␤ 1 Asp-20 is Lys-55 in STE4, and that its alteration to an acidic Glu is sufficient to produce a signaling-defective dominant-negative (19) (Fig. 1A). Thus, there may be some evolutionary or ontogenetic advantage in restraining the JNK signaling potential of mammalian G␤s at this locus. The apparent lack of effect of the ␤ 1 -D20K mutation on PLC-␤ signaling adds to previous evidence that the structural features of the G␤␥ complex important for effector signaling may differ among effectors (37). Although it remains to be seen if the D20K mutation will alter any other 4 S. Pellegrino and W. F. Simonds, manuscript in preparation. ␤␥-responsive effector pathways, it is possible that this or a similar mutation in G␤ might result in selective gain-of-function with an associated clinical phenotype.