Selective activation of effector pathways by brain-specific G protein beta5.

While multiple G protein β and γ subunit isoforms have been identified, the implications of this potential diversity of βγ heterodimers for signaling through βγ-regulated effector pathways remains unclear. Furthermore the molecular mechanism(s) by which the βγ complex modulates diverse mammalian effector molecules is unknown. Effector signaling by the structurally distinct brain-specific β5 subunit was assessed by transient cotransfection with γ2 in COS cells and compared with β1. Transfection of either β1 or β5 with γ2 stimulated the activity of cotransfected phospholipase C-β2 (PLC-β2), as previously reported. In contrast, cotransfection of β1 but not β5 with γ2 stimulated the mitogen-activated protein kinase (MAPK) and c-Jun N-terminal kinase (JNK) pathways even though the expression of β5 in COS cells was evident by immunoblotting. The G protein β5 expressed in transfected COS cells was properly folded as its pattern of stable C-terminal proteolytic fragments was identical to that of native brain β5. The inability of β5 to activate the MAPK and JNK pathways was not overcome by cotransfection with three additional Gγ isoforms. These results suggest it is the Gβ subunit which determines the pattern of downstream signaling by the βγ complex and imply that the structural features of the βγ complex mediating effector regulation may differ among effectors.

The effector specificity of the recently described brain-specific ␤ 5 subunit(3) was examined by transient cotransfection with ␥ 2 in COS cells and compared with ␤ 1 . We report here that while both ␤ 1 and ␤ 5 were found to activate PLC-␤ 2 in a ␥ 2 -dependent fashion consistent with previous reports (3,10), ␤ 5 , unlike ␤ 1 , did not stimulate the MAPK or JNK pathways. Cotransfection of different ␥ isoforms failed to confer MAPK or JNK stimulatory ability to ␤ 5 . These results imply that the G␤ subunit can define the pattern of downstream signaling mediated by the ␤␥ complex and suggest that distinct mechanisms mediate the activation of different ␤␥-responsive effectors.
The cDNA for ␤ 5 was obtained by PCR of mouse brain cDNA (Clontech) using specific primers (3) and the thermostable DNA polymerase Pyrococcus furiosus (Pfu) (Stratagene). The final construct contained two silent base changes (Ser 130 TCt and Phe 136 TTc) relative to the published sequence (3) (GenBank TM accession number L34290). The cDNA for ␥ 4 was obtained by reverse transcription-PCR employing the Pyrococcus woesei (Pwo) (Boehringer Mannheim) thermostable DNA polymerase from human brain total RNA (Clontech) with primers based on the published sequence (4) (GenBank TM accession number U31382). A silent nucleotide substitution in codon Ala 38 (GCt) was found. The construct for ␥ 5 (GenBank TM accession number M95779) was obtained by PCR employing Pwo DNA polymerase and bovine liver ␥ 5 cDNA (kindly provided by Dr. Nathan N. Aronson) as a template in combination with specific primers (21). The cDNA for ␥ 7 (GB M99393) was obtained by PCR employing Pfu DNA polymerase from bovine brain cDNA (Clontech) and specific primers (22). The finished ␤ 5 , ␥ 4 , ␥ 5 , and ␥ 7 constructs all contained the sequence GAATTCAAGATG at their 5Ј ends (starting methionine codon underlined) and after the stop codon were followed by an XbaI site at their 3Ј end and were ligated between EcoRI and XbaI sites of pCDM8.1 (␤ 5 ) or pcDNA3 (␥ 4 , ␥ 5 , and ␥ 7 ). Constructs in pcDNA3 were amplified in Escherichia coli XL1Blue (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 (23) using Sequenase 2.0 (U. S. Biochemical Corp.).
Protein Expression, Immunoblotting, and Proteolytic Analysis-Growth, maintenance, transfection (24), and fractionation of COS-7 cells was as described previously (18). Protein was determined by the method of Bradford (25) using bovine serum albumin as a standard. Membrane proteins or crude lysates were separated on 11% slab gels by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) (26) and electrotransferred onto polyvinylidene difluoride membranes in Dunn's buffer (27). For analysis of low molecular weight proteolytic fragments, the Tricine gel system of Schä gger and von Jagow (28) was employed as indicated. Detection of G␤ 5 subunits employed the primary antibody SGS generated in rabbits against the synthetic dodecapeptide SG-SWDHTLRVWA (conjugated to keyhole limpet hemocyanin (29)) corresponding to residues 342-353 at the C terminus of ␤ 5 (3). The antibody RA generated against a peptide corresponding to ␤ 1 residues 256 -265 has been described (30). Secondary detection employed 125 I-Protein A followed by autoradiography on film (19) or a storage phosphor screen * 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. (Molecular Dynamics PhosphorImager), or else enhanced chemiluminescence using goat anti-rabbit or anti-mouse antibodies coupled to horseradish peroxidase (Boehringer Mannheim).
Cleared cholate extracts of crude membrane preparations of transfected COS cells or BALB/c mouse brains were prepared by extraction with 1% (w/v) cholate, 10 mM EDTA, 50 mM Tris-HCl (pH 8.0) (buffer A) on ice for 30 min followed by centrifugation at 16,000 ϫ g for 10 min. The preparation of bovine brain membrane cholate extract was as described (31,32).
Phosphoinositol Phospholipase C Activity Assay-The PI-PLC activity of transfected cells was estimated by a modification of the procedure of Berridge et al. (33) as described previously (10,34).
MAPK and JNK Activity Assays-The assays for MAPK and JNK activity were essentially as described by Crespo et al. (12) and Coso et al. (20), respectively. 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 (24) using a total of 15 g of DNA per cotransfection, typically including 5 g of HA-ERK2 or HA-JNK (12,20), 5 g of G␥, and 5 g of G␤. 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 (12,20) except that mouse monoclonal HA.11 (Berkeley Antibody Co.) was used for the immunoprecipitations (3 l of HA.11 ascites fluid per 900 l of detergent lysate).

RESULTS AND DISCUSSION
Functional Properties of ␤ 5 Cotransfected with ␥ 2 -The low degree of sequence homology between the brain-specific ␤ 5 subunit recently described by Simon and co-workers (3) and other G␤ subunits (e.g. only 53% amino acid identity with ␤ 1 ) suggests a possible specialization of ␤ 5 for interactions with G␣, G␥, receptor, and/or effector molecules. The effector specificity of the ␤ 5 subunit was examined by cotransfection with ␥ 2 in COS cells and compared with ␤ 1 . As previously reported (3), transfection of the combination of ␤ 5 and ␥ 2 produced a robust stimulation of cotransfected PLC-␤ 2 (Fig. 1A). This stimulation was comparable with that of the ␤ 1 /␥ 2 combination and was not seen with transfection of ␤ 5 or ␥ 2 alone (Fig. 1A).
G protein ␤ 5 and ␤ 1 were then compared in their ability to activate the MAPK pathway in a ␥-dependent fashion using a cotransfection paradigm employing HA-epitope-tagged ERK2 (12). The ␤␥ complex of heterotrimeric G proteins activates the MAPK cascade in metazoans (12)(13)(14) as it does in yeast (35). Previous analysis by Hawes et al. (36) of G␤ subtypes 1-4 and multiple ␥ isoforms in transfected COS-7 cells showed a strong correlation between the ability of a specific ␤␥ combination to activate PLC-␤ 2 and its ability to activate the MAPK pathway. In contrast, we found that ␤ 5 was unable to stimulate MAPK activity in combination with ␥ 2 despite the ability of ␤ 1 to do so under the same conditions ( Fig. 1B and see below) and despite the efficacy of ␤ 5 /␥ 2 in the PLC assay (Fig. 1A). The inability of ␤ 5 to activate the MAPK pathway was not due to failure of expression of the ␤ 5 or HA-ERK2 proteins under the conditions of the MAPK assay as documented by immunoblots of the Triton lysates of transfected COS cells with specific antibodies (Fig. 1C and see below).
Analysis of Recombinant ␤ 5 by Immunoblotting and Limited Proteolysis-The ability of ␤ 5 to activate PLC-␤ 2 in a ␥-dependent fashion strongly implies its proper expression and assembly with G␥. To further exclude the possibility, however, that the failure of transfected ␤ 5 to activate the MAPK pathway was due to abnormal expression or folding of the recombinant polypeptide, additional analysis was performed using the anti-FIG. 1. Functional properties and expression of G protein ␤ 5 in COS cells. A, PI-PLC stimulatory activity of G protein ␤ 1 and ␤ 5 compared in a cotransfection assay. COS cells in 75-cm 2 flasks were transfected with vector alone or with PLC-␤ 2 (1.5 g) either alone or in combination with ␤ 1 (10 g), ␤ 5 (10 g), ␥ 2 (5 g), and PI-PLC activity assayed as described under "Experimental Procedures." The results shown represent the mean (Ϯ S.E.) value of triplicate determinations in a single experiment. Three additional experiments produced comparable results. B, MAPK stimulatory activity of G protein ␤ 1 and ␤ 5 compared in a cotransfection assay. COS cells in 75-cm 2 flasks were transfected with HA-ERK2 in combination with vector alone, or with ␤ 1 (5 g), ␤ 5 (5 g), ␥ 2 (5 g) either alone or in combinations as indicated. An additional HA-ERK2-transfected flask was treated with 100 ng/ml epidermal growth factor (EGF) for 5 min at 37°C prior to cell lysis as a positive control. MAPK activity was measured as described previously (12) by quantification of 32 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 results shown are from a single experiment and are representative of three experiments giving an identical pattern of results. C, expression of HA-ERK2 (HA-MAPK) and G␤ 5 in cotransfected COS cells. Aliquots (20 g of protein) of the Triton detergent lysates of COS cells transfected under the conditions indicated in Fig. 1B were subjected to SDS-PAGE on 11% polyacrylamide gels and then immunoblotted with antibodies to the HA epitope (HA.11 mouse monoclonal) or the G␤ 5 C-terminal dodecapeptide (SGS rabbit polyclonal). Subsequent detection employed horseradish peroxidase-coupled secondary antibody and enhanced chemiluminescence as described under "Experimental Procedures." body SGS generated against the C-terminal dodecapeptide of mouse ␤ 5 . Like the N-terminal ␤ 5 antibody described by Watson et al. (3), SGS identified a major ϳ39-kDa band in immunoblots of detergent extracts of membranes prepared from ␤ 5 /␥ 2 cotransfected but not control COS cells (Figs. 1C and  2A, left panel). This immunoreactive band comigrated with the major band in both mouse and bovine brain membrane extracts ( Fig. 2A, left panel) consistent with previous reports (3). A faint upper SGS-reactive band of M r ϳ90,000 was also noted in blots of mouse brain extract. The pattern of C-terminal fragments generated by limited proteolysis of bovine brain ␤ 5 was then compared with that of brain ␤ 1 by immunoblotting with specific antibodies (Fig. 2A, right panel). Such biochemical analysis has proven very useful in assessing the folding of native and recombinant G protein ␤␥ complexes (19,(37)(38)(39). Whereas a stable ϳ26-kDa C-terminal fragment of ␤ 1 is identified by RA antibody (30) in a tryptic digest of bovine brain proteins, analysis of the same sample with the ␤ 5 C-terminal antibody SGS revealed no stable immunoreactive fragments ( Fig. 2A, right  panel). Limited digestion of the bovine brain samples with endoproteinase Lys-C produced an SGS-reactive C-terminal fragment visible at the dye front, while digestion with V8 protease yielded a diffuse band centered at ϳ35 kDa. The RA immunoreactivity at ϳ36 kDa in the endoproteinase Lys-C and V8 digests remained unchanged from control ( Fig. 2A, right  panel). Further analysis of the low molecular weight products on Tricine gels (28) revealed a major C-terminal SGS-reactive band of 12 kDa and a faint band of ϳ22 kDa in the endoproteinase Lys-C digests of both mouse and bovine brains (Fig.  2B). In addition to the major 35-kDa band product, a minor product of ϳ11 kDa was noted in the Tricine gel blots of V8 protease digests of both brain samples (Fig. 2B). An identical pattern of SGS-reactive C-terminal fragments was seen upon limited proteolysis of detergent extracts of membranes prepared from COS cells cotransfected with ␤ 5 ϩ ␥ 2 (Fig. 2B). Taken together, these data demonstrate that the state of folding of the ␤ 5 polypeptide expressed in transfected COS cells inferred from selective proteolysis is indistinguishable from that of native ␤ 5 found in brain.
Ability of ␤ 5 to Assemble with Different G␥ Isoforms and Activate PLC-␤-Previous studies of ␤␥-responsive effectors comparing purified recombinant G␤␥ complexes have found significant differences due to the ␥ component of the ␤␥ heterodimer (40,41). We therefore wondered if the inactivity of ␤ 5 in the MAPK assay was specific to the ␤ 5 /␥ 2 combination and might be overcome by employing other ␥ isoforms. Because certain combinations of ␤ and ␥ subunits fail to assemble as heterodimers (39,42), we first compared three additional ␥ subunits with ␥ 2 in their ability to assemble with ␤ 5 in cotransfected COS cells. Immunoblots of cholate extracts of COS cell membranes revealed no endogenous ␤ 5 immunoreactivity and no ␤ 5 signal resulting from transfection of any ␥ subunit alone (Fig. 3, upper panel), consistent with the results of Fig. 1C. Transfection of ␤ 5 alone or in combination with ␥ 1 (transducin ␥) produced only a faint SGS-reactive band at ϳ39-kDa which may reflect assembly of ␤ 5 with the endogenous pool of ␥ subunits (Figs. 1C and 3, upper panel). In contrast, cotransfection of ␤ 5 with ␥ 2 , ␥ 4 , or ␥ 7 gave a reproducible increment of ␤ 5 immunoreactivity over that seen with ␤ 5 alone, with a rank order ␥ 2 Ͼ Ͼ ␥ 4 ϭ ␥ 7 (Fig. 3). This increase in the steady state levels of membrane ␤ 5 seen with ␥ cotransfection presumably reflects increased stability and/or membrane targetting of the ␤ subunit engendered by assembly with the exogenous ␥. The reciprocal ability of cotransfected ␤ to stabilize ␥ has been previously documented (43).
The functional association of ␤ 5 with ␥ 4 and ␥ 7 was also demonstrated by the ability of ␤ 5 /␥ 4 and ␤ 5 /␥ 7 to stimulate PLC-␤ 2 activity above control levels in parallel functional assays (Fig. 3, lower right). In contrast to results with ␥ 2 , ␥ 4 , and ␥ 7 , cotransfection of ␥ 1 with ␤ 5 produced no increment in PLC activity (data not shown). The ability of the different ␥ isoforms to promote the ␤ 5 -dependent activation of PLC-␤ 2 correlated strongly with their ability to increase the steady-state expression of ␤ 5 documented by quantitation of the SGS immunoblots (Fig. 3, lower panels). Simon and co-workers (3) also reported that the ␤ 5 /␥ 2 combination stimulated PLC-␤ 2 activity much more strongly in COS cells than combinations employing other ␥ isoforms, although ␤ 5 expression levels were not compared.
Failure of Additional G␥ Isoforms to Support MAPK and JNK Activation by ␤ 5 -Despite biochemical and functional evidence of their assembly with ␤ 5 (Fig. 3), neither cotransfected ␥ 4 nor ␥ 7 conferred to ␤ 5 the ability to activate the MAPK pathway (Fig. 4A), resembling ␥ 2 in this regard (Figs. 1B and 4A). On the other hand, ␥ 4 and ␥ 7 both supported MAPK stimulation by cotransfected ␤ 1 (Fig. 4A). The analysis of the effector signaling functions of ␤ 5 was extended to look at potential JNK stimulatory activity. The ability of cotransfected mammalian ␤ 1 /␥ 2 to activate JNK through a p21 ras -and p21 rac1dependent pathway bearing many similarities to the ␤␥-driven pheromone response pathway in Saccharomyces cerevisiae was described recently (15). We compared the ability of ␤ 1 and ␤ 5 to activate HA-JNK when transfected alone or in combination with ␥ 2 , ␥ 4 , or ␥ 7 . The kinase activity of the cotransfected HA-JNK reporter in ␤ 5 ϩ ␥ transfected cells was consistently the same as or lower than the activity in ␥-only transfected cells, regardless of which ␥ was employed (Fig. 4B). The inactivity of ␤ 5 in this assay was in stark contrast to the activity of ␤ 1 which produced a robust increment in JNK activity when cotransfected with each of the three ␥ isoforms (Fig. 4B). Immunoblotting of cell lysates from such experiments demonstrated that the expression of the HA-JNK reporter enzyme was comparable whether ␤ 1 or ␤ 5 was employed in the cotransfection (data not shown). Additional experiments with the ␥ 5 isoform, shown previously by Watson et al. (3) to promote ␤ 5 stimulation of PLC-␤ 2 , revealed that it too supported MAPK and JNK activation by ␤ 1 , but not by ␤ 5 (data not shown). 5 -While signaling through particular G␣ subunits is characteristically confined to a restricted set of effector targets, ␤␥ heterodimers have been found to modulate a wide range of structurally diverse effector molecules, an observation true even when a defined pair of ␤ and ␥ isoforms has been studied. Reconstitution experiments have shown, for example, that purified recombinant ␤ 1 ␥ 2 can regulate adenylyl cyclase (40), phospholipase C-␤ (40,44), inwardly rectifying potassium channels (41), and ␤-adrenergic receptor kinase activities (45). Differences in effector regulation ascribed to the G␤ component of defined heterodimers noted previously in cotransfection and reconstitution experiments have been only modest (45,46). The present results offer the strongest evidence to date that G protein ␤ subunits are sufficient to determine the effector selectivity of ␤␥ and suggest that the structural features of the ␤␥ complex mediating effector regulation may differ depending on the effector.

Implications of the Effector Selectivity of G␤
The G protein ␤ 5 fails to activate critical intermediates in the MAPK and JNK pathways in COS cells when complexed with ␥ 2 , ␥ 4 , ␥ 5 , or ␥ 7 in ␤␥ heterodimers. In vertebrates, the G protein ␤␥ complex triggers the MAPK cascade by a p21 ras -dependent mechanism (12,14) involving an unknown number of upstream intermediaries which may include nonreceptor tyrosine kinases (47), phosphatidylinositol 3-kinase (48), and Shc (49) acting through Ras guanine nucleotide exchange factors including Sos (50) and Ras-guanine nucleotide releasing factor (51). The upstream mediators of the Ras-and Rac1-dependent activation of JNK by ␤␥ subunits (15) remain unknown, but activation of a homologous pathway in S. cerevisiae appears to involve the direct interaction of ␤␥ with the scaffolding protein Ste5p (52) and the guanine nucleotide exchange factor Cdc24p (53). It is tempting to speculate that the structure of ␤ 5 precludes its interaction with the mammalian counterparts of these or perhaps other critical intermediates in the kinase pathways. Such a loss of function might represent an evolutionary goal of ␤ 5 . It is possible to speculate, for example, that expression of ␤ 5 might promote the survival of certain neurons by failing to activate JNK, a pathway linked to apoptosis in neuronal cell culture (54). An alternative speculation is that the inability of G␤ 5 to activate the kinase cascades is an incidental consequence of evolution to gain a function such as an ability to regulate novel effector molecules in brain.