A C-terminal mutant of the G protein beta subunit deficient in the activation of phospholipase C-beta.

The molecular mechanism by which the G protein βγ complex modulates multiple mammalian effector pathways is unknown. Homolog-scanning mutagenesis of the G protein β subunit was employed to identify residues critical for the activation of phospholipase C-β2 (PLC-β2). A series of chimeras was made by introducing small segments of the Dictyostelium β subunit into a background of mammalian β1 and tested in COS cell cotransfection assays for their ability to activate PLC-β2 and assemble with mammalian γ2. A chimera that contained four Dictyostelium β substitutions within the C-terminal 14 residues was unable to activate PLC-β2 when cotransfected with γ, despite its demonstrable expression in a γ-dependent manner. Cotransfection of the mutant blocked m2 muscarinic receptor activation of PLC by a pertussis toxin-sensitive pathway. This C-terminal mutant retained the ability, however, to stimulate the mitogen-activated protein kinase pathway. These results imply that activation of different βγ-responsive effectors is mediated by distinct domains.

Heterotrimeric guanine nucleotide-binding proteins (G proteins), 1 composed of ␣, ␤, and ␥ subunits, transmit signals from surface receptors to intracellular effectors (1,2). When receptor is activated by agonist, it catalyzes the exchange of GDP for GTP on the ␣ subunit of G protein resulting in its dissociation from the ␤␥ heterodimer. It is now recognized that both the G␣ subunit and the ␤␥ complex transmit signals to effector molecules (3,4). After the initial observation that ␤␥ could activate potassium channels (5), it has been found that ␤␥ can regulate certain isoforms of adenylyl cyclase (6,7) and phospholipase C-␤ (PLC-␤) (8 -10), activate the mitogen-activated protein kinase (MAPK) (11)(12)(13) and c-Jun N-terminal kinase (14) pathways, and mediate the translocation of the ␤-adrenergic receptor kinase (15,16). The mechanism by which ␤␥ interacts with structurally distinct mammalian effector molecules is unknown although in the budding yeast Saccharomyces cerevisiae studies have identified several mutants in STE4 (the G␤ homolog) deficient in downstream signaling in response to pheromones (17)(18)(19).
To investigate this problem, we employed homolog-scanning mutagenesis (20) to look for G protein ␤ mutants exhibiting loss of function in their ability to stimulate PLC-␤ 2 . To this end a series of chimeras was constructed in which small segments of the Dictyostelium ␤ subunit were introduced into a background of mammalian ␤ 1 and tested for their ability to activate PLC-␤ 2 and assemble with the mammalian ␥ 2 subunit in a cotransfection model. We report that chimeras which contained four or six Dictyostelium ␤ substitutions within the extreme C terminus of ␤ 1 were unable to activate PLC-␤ 2 when cotransfected with ␥, yet retained their ability to activate the MAPK pathway. These results suggest that ␤␥-modulated effector molecules may interact with distinct domains of the ␤␥ complex.
Chimeras between mammalian ␤ 1 and Dictyostelium G␤ were constructed by overlap extension using the polymerase chain reaction (PCR) (29) to generate the coding sequences indicated in Fig. 1. Chimera B, for example, consists of codons 1-19 and 53-340 from mammalian ␤ 1 (21) joined by a segment (shown in box B of Fig. 1) corresponding to codons 25-59 of Dictyostelium G␤ (23). Sequences of the oligonucleotides used for the construction of the chimeras are available upon request.
The PCR reactions used to generate the chimeric ␤ constructs employed Pyrococcus furiosus (Pfu) (Stratagene) or Pyrococcus woesei (Pwo) (Boehringer Mannheim) thermostable DNA polymerases. The finished constructs all contained the sequence GAATTCAAGATG at their 5Ј ends (starting methionine codon underlined) consistent with the Kozak eukaryotic translation initiation sequence (30), and after the stop codon were followed by an XbaI site at their 3Ј end. The ␤ constructs were ligated between the EcoRI and XbaI sites of pCDM8.1. Constructs in pCDM8.1 plasmids were amplified in Escherichia coli MC1061/P3 (Invitrogen), and those in pCDNA3 were amplified in E. coli XL1 Blue (Stratagene). 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 (31) using Sequenase 2.0 (U. S. Biochemical Corp.). Silent mutations were noted in codon 32 of chimera B (ACt) and in codon 143 of chimeras C and D (ACg).
Protein Expression, Assembly with ␥ 2 -C68S and Tryptic Analysis-Growth, maintenance, and transfection of COS-7 cells by the DEAEdextran method (32) was as described previously (27). Briefly, 5 g of the nonisoprenylated ␥ 2 mutant (␥ 2 -C68S; ␥ 2 *) (27) with or without 10 g of wild-type ␤ 1 or one of the ␤ chimeras were used for each transfection in a 75-cm 2 culture flask. The total amount of DNA per flask was kept constant by supplementing with vector DNA where necessary. The * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Upon thawing, the cells were lysed and fractionated into cytosolic and crude membrane fractions as described previously (27), and protein was determined by the method of Bradford (33) using bovine serum albumin as a standard. Cytosolic proteins were separated on 11% gels by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) (34) and electrotransferred onto polyvinylidene difluoride membranes in Dunn's buffer (35). Detection of G␤ subunits employed primary antibodies generated against synthetic peptides corresponding to ␤ 1 residues 127-136 (KT) or residues 256 -265 (RA) (36). Chimeras with a heterologous sequence replacing a portion of ␤ 1 corresponding to an immunogenic peptide were necessarily evaluated by the other antiserum. Secondary detection employed 125 I-Protein A followed by autoradiography on film (28) or a storage phosphor screen (Molecular Dynamics PhosphorImager), or else enhanced chemiluminescence using goat anti-rabbit antibody coupled to horseradish peroxidase (Boehringer Mannheim). Limited tryptic digestion of cytosolic fractions was performed as described previously (28).
Phosphoinositol Phospholipase C Activity Assay-The PI-PLC activity of transfected cells was estimated by a modification of the procedure of Berridge et al. (37) as described previously (9). Subconfluent COS cells were cotransfected with combinations of 1.5 g of human PLC-␤ 2 (in vector pMT2) (24), 2-20 g of wild type ␤ 1 or one of the chimeras, and 5 g of wild-type ␥ 2 per 75-cm 2 flask as indicated in the figure legends. In experiments like that shown in Fig. 3, 3 g of m2 mAChR cDNA (25) was included, and ␥-cDNA was omitted to assess the impact of wild-type and mutant G protein ␤ cDNAs on receptor-mediated activation of PLC-␤ 2 via heterotrimeric G proteins composed of endogenous ␣ and ␥ subunits. On the second day of transfection, the cells were trypsinized and replated in 12-well plates (4.5 cm 2 /well) at a cell density of ϳ4.0 ϫ 10 5 cells per well, and the cells were allowed to reattach. Approximately 20 h before the assay, the medium was replaced with DMEM without inositol (Life Technologies, Inc.) supplemented with 10% dialyzed fetal bovine serum and 10 Ci/ml of myo-[2- MAPK Activity Assay-The assay for MAPK activity was essentially as described by Crespo et al. (11). Approximately 1.6 ϫ 10 6 COS-7 cells were plated on 100-mm plates and then incubated at 37°C overnight. On the following day, the cells were transfected by the DEAE-dextran method using a total of 12 g of DNA per cotransfection, including 2 g of HA (hemagglutinin) epitope-tagged ERK2-pCDNA3 (11,38), 5 g of ␥ 2 -pCDM8.1, and 5 g of either wild-type ␤ 1 , or one of the ␤ chimeras, in pCDM8.1. The remainder of the assay was as described (11) except that [␥-33 P]ATP (DuPont NEN) instead of [␥-32 P]ATP was employed.

RESULTS
Chimera Construction-In an effort to identify residues of the G protein ␤ subunit critical for activation of phospholipase C-␤, we introduced segments of phylogenetically remote ␤ sequence into a background of mammalian ␤ 1 by homolog-scanning mutagenesis with the aim of creating loss of function mutants. After unsuccessful trials employing segments of STE4, the G␤ homolog from S. cerevisiae, we turned to Dictyostelium ␤, which is 67% identical to mammalian ␤ 1 (39), as the source of heterologous ␤ sequence. In the slime mold Dictyostelium, there is a single form of phospholipase C which is most homologous to the mammalian PLC-␦ isoform (40,41). Furthermore, Dictyostelium PLC appears to be under the control of a G protein ␣ and not ␤␥ (42,43).
An initial series of 15 Dictyostelium ␤/mammalian ␤ 1 chimeras was constructed as shown in Fig. 1. The series of chimeras incorporates all the residues which are nonconserved between the two parental subunits, in clusters of 2 to 27 per chimera arranged according to primary sequence. The boundaries between the Dictyostelium ␤ and ␤ 1 sequences in chimeras A through O were chosen prior to the elucidation of the three-dimensional structure of the ␤␥ complex (44 -46).
Ability of ␤ Chimeras to Activate PLC-␤-Screening for possible loss of function mutants among the 15 chimeras shown in Fig. 1 involved two stages: (i) an initial activity screen to evaluate ␤␥-dependent PLC activation following cotransfection of wild-type or chimeric ␤ in combination with wild-type ␥ 2 and human PLC-␤ 2 and (ii) assessment of ␥-dependent protein expression by cotransfection of chimeric ␤ constructs with nonisoprenylated ␥ (27,28). Cotransfection of G protein ␤ and ␥ cDNAs has been used by others to evaluate the ␤␥-dependent activation of PLC-␤ (9, 47). As shown in Fig. 2A, the transfection of COS cells with PLC-␤ 2 and either ␥ 2 or ␤ 1 alone had little effect on PLC activity, whereas cotransfection of both ␥ 2 and ␤ 1 cDNAs produced a nearly 10-fold enhancement of PLC activity consistent with previous reports (9). In transfection experiments employing increasing amounts of wild-type ␤ 1 cDNA with constant amounts of ␥ 2 and PLC-␤ 2 cDNA, it was found that 10 g of plasmid gave a near-saturating signal in this assay ( Fig. 2A, inset). Thus, for screening of chimeric ␤ constructs, 10 g of plasmid was used per transfection to make the assay most sensitive for detection of loss of function.
The results with the 15 Dictyostelium ␤/mammalian ␤ 1 chimeras are shown in Fig. 2A. A wide range of activities is evident among the 15 chimeras with the lowest PLC activation demonstrated by chimera O. The level of PLC activity seen with chimera O transfection was most often at or below the activity seen with transfection of ␥ 2 (and PLC-␤ 2 ) alone ( Fig.  2A). Chimera A also demonstrated consistently low activity, but was always more active than chimera O. The levels of PLC activation seen with chimeras B, E, F, G, I, J, and L approached or exceeded that of wild-type (␥ 2 *) was employed (27). The observation that nonisoprenylated ␥ subunits direct expression of cotransfected ␤ subunits to the cytosol has been used extensively to estimate the ability of wild-type and mutant subunits to assemble (27,28,48).
Estimation of the ␥-dependent Expression Level of the ␤ Chimeras A and O-The results of experiments in which wild-type ␤ or chimeras A or O were cotransfected with ␥ 2 * in COS cells are shown as immunoblots of the cytosolic fractions in Fig. 2B. Transfection of vector or ␤ 1 alone produced no significant ␤ immunoreactivity in the cytosol, whereas transfection of ␥ 2 * alone resulted in faint ␤ immunoreactivity representing translocation of endogenous ␤ subunits, as described previously (27,28,48) (Fig. 2B). In contrast, cotransfection of both wild-type ␤ 1 and ␥ 2 * resulted in a large increase in cytosolic ␤ immunoreactivity as seen previously (27,28,48) (Fig. 2B). Cotransfection of both chimera A with ␥ 2 * and chimera O with ␥ 2 * produced weak ␤-immunoreactive signals (Fig. 2B). Chimera O was consistently expressed to a higher level than A, however (Fig. 2B), even though its PLC-␤ stimulatory activity was lower ( Fig. 2A). Chimera O was therefore used as a starting point to look for a more highly expressed variant with the same loss of function phenotype. Construction and testing of 14 additional variants of chimera O (data not shown) identified chimera O 2,3,4 , containing only four heterologous Dictyostelium residues at codons 327, 328, 335, and 340 of ␤ 1 , which was expressed to a higher level than chimera O (Fig. 2B). Like chimera O, chimera O 2,3,4 was severely impaired in its ability to activate PLC-␤, exhibit-ing Ͻ10% of wild-type activity at 20 g of transfected plasmid (Fig. 2D). The corresponding maximal expression of chimera O 2,3,4 was ϳ40% of wild-type ␤, as estimated by quantitation of immunoblots of CHAPS extracts of crude particulate fractions (data not shown). Limited tryptic digestion of COS cell cytosolic fractions containing ␤ 1 /␥ 2 * and chimera O 2,3,4 /␥ 2 * produced stable ϳ26-kDa C-terminal ␤ fragments indicative of properly folded ␤ polypeptide chains (49 -51) (Fig. 2C). Chimeras B through N also demonstrated variable levels of cytosolic ␤ immunoreactivity when cotransfected with ␥ 2 *, in a rank order roughly paralleling their relative activity in the PLC assays of Fig. 2A (data not shown).

Dominant Negative Effects of Chimeras O and O 2,3,4 on
Receptor-stimulated PLC Activation-G protein ␤␥-mediated stimulation of PLC by a pertussis toxin-sensitive pathway is seen upon activation of a variety of G i -coupled receptors, including the m2 muscarinic receptor (9). We reasoned that if chimeras O and O 2,3,4 , while deficient in their ability to activate PLC-␤, retained their ability to interact with endogenous elements of this receptor-driven pathway they might exhibit dominant inhibitory effects on agonist signaling. Experiments were therefore performed in which carbachol-induced PLC activation was assessed in COS cells cotransfected with cDNAs for PLC-␤ 2 and m2 mAChR without or with wild-type ␤ 1 or chimeras O or O 2,3,4 . As seen in Fig. 3, carbachol induces an activation of PLC activity in m2 mAChR-transfected cells which is largely abolished by pretreatment with pertussis toxin. Co-

FIG. 2. Identification and characterization of C-terminal chimeras O and O 2,3,4 deficient in PLC-␤ stimulation.
A, PI-PLC activity of wild-type ␤ 1 and chimeras A through O determined in a cotransfection assay. COS cells in 75-cm 2 flasks were transfected with PLC-␤ 2 (1.5 g) either alone (Control) or in combination with ␥ 2 (5 g) alone, ␤ 1 (10 g) alone, or ␥ 2 in combination with ␤ 1 or one of the chimeras A through O (10 g) as indicated, and PI-PLC activity was assayed as described under "Experimental Procedures." Inset, PI-PLC activity in response to increasing amounts of wild-type ␤ 1 cDNA (0 -20 g as indicated) cotransfected with constant PLC-␤ 2 (1.5 g) and ␥ 2 (5 g) cDNA. B, G␥ 2 *-dependent cytosolic expression of wild-type ␤ 1 and chimeras A, O, and O 2,3,4 . RA antibody immunoblot of the soluble fraction (10 g of protein/lane) of COS cells transfected with vector alone (V), ␥ 2 *, ␤ 1 or chimera alone, or ␥ 2 * in combination with ␤ 1 or chimera as indicated. Antibody detection via chemiluminescence as under "Experimental Procedures." Arrow indicates molecular mass of bovine brain G␤ standard (not shown). C, limited tryptic digestion of cytosolic fractions of COS cells cotransfected with G␥ 2 * and ␤ 1 or chimeras A, O, or O 2,3,4 . RA antibody immunoblot of the soluble fraction (25 g of protein per lane) of COS cells transfected with the indicated constructs and either maintained on ice or incubated with trypsin (1 g) for 30 min at 37°C prior to SDS-PAGE on 11% polyacrylamide gel as described previously (28). Antibody detection was by 125 I-Protein A as described under "Experimental Procedures." Indicated to the left are the molecular masses of intact G␤ 1 subunit (36 kDa) or its major C-terminal tryptic fragment (26 kDa). D, PI-PLC activity as a function of transfected wild-type ␤ 1 or chimera O 2,3,4 cDNA. COS cells were transfected with a fixed amount of PLC-␤ 2 (1.5 g) and ␥ 2 (5 g) cDNA, without or with increasing amounts of plasmid cDNA for wild-type ␤ 1 (open squares) or chimera O 2,3,4 (solid circles) (0 -20 g as indicated), and PI-PLC activity was measured as described under "Experimental Procedures." transfection of wild-type ␤ 1 enhances this effect, perhaps by promoting the formation of m2-coupled wild-type G protein heterotrimers. It should be noted that the transfected ␤ subunit must assemble with endogenous COS cell G protein ␣ and ␥ subunits for its effect since neither cDNA is being supplied exogenously. In contrast to the effect of wild-type ␤ 1 , cotransfection of chimeras O and O 2,3,4 abolished the carbachol-induced activation of PLC, producing an effect resembling that of pertussis toxin (Fig. 3). These two mutant ␤ subunits thus function as dominant negative inhibitors of the m2 muscarinic-PLC stimulatory pathway in this system.

Effects of Chimera O 2,3,4 on Mitogen-activated Protein
Kinase-It has recently been demonstrated that, as in budding yeast, the mammalian MAPK pathway can be activated by the ␤␥ complex of heterotrimeric G proteins (11)(12)(13). We therefore questioned whether the loss of function of chimeras O and O 2,3,4 with respect to PLC-␤ activation would extend to the MAPK pathway. To this end, COS cells were cotransfected with cDNAs encoding HA epitope-tagged ERK2, wild-type ␥ 2 , and either wild-type ␤ or one of chimeras A, O, or O 2,3,4 , and ERK2 kinase activity was measured in the HA immunoprecipitates as described (11). As found previously for wild-type ␤ 1 (11), both chimeras O and chimera O 2,3,4 were capable of MAPK activation and required ␥ cotransfection for this activity (Fig. 4A). Chimera A, however, was inactive, as in the PLC assay ( Fig.  2A), suggesting a broader deficiency of function in this mutant. In contrast to its inactivity over a wide range of transfected cDNA in the PLC assay (Fig. 2D), chimera O 2,3,4 was comparable to wild-type ␤ 1 in its ability to activate the kinase activity of HA epitope-tagged ERK2 (Fig. 4B). DISCUSSION Homolog-scanning mutagenesis has been used to generate loss-of-function mutants of the G␤ 1 subunit to investigate domains critical for the activation of PLC-␤ 2 . Dictyostelium ␤ was used as the source of heterologous sequence because the single Dictyostelium PLC isoform (homologous to the mammalian PLC-␦ subclass) appears to be under the control of a G protein ␣ and not ␤␥ (42,43). Of the 15 initial G␤ chimeras outlined in Fig. 1, it is remarkable that so many were capable of PLC-␤ 2 stimulation at or near wild-type levels ( Fig. 2A). This suggests that the interaction of ␤␥ with PLC-␤ 2 may involve a limited domain or domains and is therefore insensitive to mutations in many parts of G␤.
Our strategy succeeded in the identification of a segment in the extreme C terminus of G␤ 1 , which when substituted with a minimum of four homologous residues from Dictyostelium G␤, produced mutant ␤␥ complexes deficient in their ability to stimulate phospholipase C-␤ 2 in a transient transfection model system. Because our strategy relied on differences in the primary sequences of G␤ 1 and Dictyostelium G␤, it would be expected to miss any residues critical for PLC-␤ activation which happened to be conserved between the divergent ␤ subunits. The C-terminal chimera O 2,3,4 identified in this study was nevertheless competent to assemble with ␥ as evidenced by its ability to stimulate the MAPK pathway in a ␥-dependent manner, translocate to the cytosol upon cotransfection with nonisoprenylated ␥, and resist tryptic proteolysis with generation of a stable ϳ26-kDa C-terminal fragment. These latter findings exclude the possibility that chimera O 2,3,4 is globally misfolded and stand in contrast to results with another chi- A, COS cells were transfected with HA-ERK2 (2 g) and vector alone (Control) or the indicated constructs (5 g each) and MAPK activity measured as described previously (11) by quantification of 33 P-phosphorylated myelin basic protein substrate in dried SDS-PAGE gels by PhosphorImager analysis. Data are expressed as fold stimulation relative to the activity of ␥ 2 alone. The activity induced in untransfected control cells assayed with 100 ng/ml epidermal growth factor (EGF) is shown. B, MAPK activity in COS cells cotransfected with HA-ERK2 (2 g) and ␥ 2 (5 g) cDNA alone or with increasing amounts of wild-type ␤ 1 (open symbols) or chimera O 2,3,4 (solid symbols) plasmid cDNA as indicated. Data are expressed as fold stimulation relative to the activity of ␥ 2 alone. mera failing to stimulate PLC activity in the cotransfection assay, chimera A. Chimera A has multiple substitutions in the coiled-coil domain of G␤ (44 -46), a region where point mutations blocking ␥ assembly have been demonstrated previously (52), and its generalized lack of function appears to result from failure to form a stable ␤␥ complex. In light of our findings with chimera O 2,3,4 , it is interesting to note that a six-amino acid deletion in the extreme C-terminal region of the STE4 is reported to block ␤␥-mediated effector signaling in yeast (19).
In addition to the loss-of-function exhibited by chimeras O and O 2,3,4 with respect to PLC-␤ stimulation, these two mutants demonstrated a dominant inhibitory activity when coexpressed with wild-type signaling elements. Chimeras O and O 2,3,4 functioned as dominant inhibitors of carbachol-stimulated PLC-␤ 2 activation when cotransfected with the m2 muscarinic receptor (Fig. 3). These inhibitory effects of chimeras O and O 2,3,4 may result from competition with wild-type ␤␥ complexes at the level of the effector (PLC-␤), the receptor, or the G␣ subunit. Definitive analysis of these possibilities must await reconstitution of purified wild-type and mutant G␤␥ with PLC-␤, G␣, and receptor.
The molecular basis for the phenotype exhibited by chimera O 2,3,4 remains unclear. One possibility would be that the Cterminal ␤ 1 residues mutated in chimera O 2,3,4 (Val 327 , Ala 328 , Phe 335 , and Asn 340 ) comprise all or part of a site of physical contact with PLC-␤ 2 . Alternatively, the effect of the mutations in chimera O 2,3,4 may result from indirect alteration of a PLC-␤ 2 contact site involving other residues in the ␤ or ␥ subunits. For example, Trp 332 of G␤ 1 , positioned between two residues altered in chimera O 2,3,4 , contacts His 209 of G t ␣ in the heterotrimer, within its switch II region, and might be available for effector interaction upon G␣ dissociation (45). The lower level of ␥-dependent expression seen with mutant O 2,3,4 compared to wild-type ␤ 1 may result from an inherent instability which may also contribute to its phenotype.
The ␥-dependent activation of the MAPK pathway by chimera O 2,3,4 was comparable to that of wild-type ␤ 1 , despite its severe impairment with respect to PLC-␤ activation. This finding suggests that the activation of different ␤␥-responsive effectors is mediated by distinct domains. Indeed, activation of the homologous MAPK pathway in yeast is impaired by mutations in the two regions mapping to the N-terminal half of STE4 (17,18), although it remains unclear if the same regions of mammalian G␤ are involved in MAPK activation.
Whether the mutations present in chimera O 2,3,4 affect other ␤␥-regulated effector pathways such as those involving adenylyl cyclase and inwardly-rectifying potassium channels is yet to be determined. Nevertheless, the ability to dissect ␤␥ effector pathways with the C-terminal ␤ mutants described here drives the hypothesis that the ␤␥ complex, instead of providing a single effector interface, may present different facets to different targets.