INTRODUCTION

The heterotrimeric G proteins transmit hormonal and sensory signals received by cell surface receptors to effector proteins that regulate processes such as proliferation, secretion, chemotaxis, vision, olfaction, neurotransmitter release, conduction of nerve impulses, and cardiac and smooth muscle contraction (1, 2, 3, 4). The  subunit serves as a molecular switch that determines the activation state of a G protein. Liganded receptors activate G proteins by catalyzing the replacement of GDP bound to the  subunit with GTP, resulting in dissociation of ·GTP from the subunits. Both ·GTP and can transmit signals to effector proteins. ·GTP but not ·GDP modulates the activities of effector proteins. The GTPase activity of the  subunit regulates the timing of deactivation and reassociation of the G protein subunits.

Differences in the amino acid sequences of the G protein  subunits determine the specificity and nature of their interactions with effector proteins. However, the molecular determinants of effector specificity and the mechanisms by which  subunit-effector interactions affect the structure and function of both proteins are poorly understood. Key questions to be answered include the following. How do specific  subunits and effectors recognize their appropriate partners? How do these interactions depend on the guanine nucleotide-binding state of the  subunit? Is there a common mechanism by which the structurally conserved  subunits recognize and modulate the activities of a diverse array of structurally unrelated effector proteins? A structure/function study of the  subunit provides the opportunity to relate the determinants of effector interaction and specificity to the GTP-dependent molecular switch.

Studies of G protein  subunit function can be interpreted in the context of the x-ray crystal structures of _t, which activates cGMP phosphodiesterase in the visual transduction system, and of _i1, which inhibits adenylyl cyclase. The  subunits belong to a structurally conserved family, sharing at least 40% identity at the amino acid level, with 60-90% identity within subfamilies (5). This degree of homology assures (6) that the structures of _t and _i1 (which are closely superimposable) provide an accurate model for the structures of other  subunits. The structures of the GTPS¹-bound (active) and GDP-bound (inactive) forms of _t (7, 8) and _i1 (9, 10) have been solved. These  subunit structures consist of two domains, a GTPase domain that resembles the oncogene protein, p21^ras, and a helical domain consisting of  helices and connecting loops (see Fig. 7). Comparison of the GTPS- and GDP-bound structures reveals three regions (Switches I-III) that change conformation depending on the activation state of the  subunit. Understanding  subunit-effector interactions involves relating effector contact sites to these switch regions.

Mapping of the effector-specifying regions of _i2 and _q onto the x-ray crystal structure of the GTPS-bound form of _t.

As a first step in understanding how G protein  subunits recognize specific effector proteins and modulate their activities, we previously identified residues of _s that specify activation of adenylyl cyclase (11). When these effector-activating residues of _s are mapped onto the x-ray crystal structures of the GTPS-bound forms of _t (7) and _i1 (9), they are localized to three adjacent regions in the GTPase domain. One of these regions is part of Switch II (8, 10), suggesting that this region is directly responsible for the GTP dependence of adenylyl cyclase activation by _s. The  subunit regions that specify activation of adenylyl cyclase by _s are also important for the interaction between _t and its effector, cGMP phosphodiesterase (12, 13, 14, 15).

To investigate whether there is a common mechanism for effector modulation by G protein  subunits, we have localized the residues of _i2 and _q that specify interaction with their respective effectors, adenylyl cyclase and PLC. Adenylyl cyclase and PLC are distinct in structure and cellular localization, so that common features of  subunit interaction with these effectors are likely to be generalizable to the effector interactions of most  subunits. Adenylyl cyclases are a family of proteins that exhibit distinct patterns of regulation by multiple input signals (16). For example, the three isoforms of _i can inhibit the _s-stimulated activity of adenylyl cyclase types V (17) and VI (18). Thus, _s and _i, which are ~40% identical at the amino acid level, both bind to adenylyl cyclase, but have opposite effects on activity. The PLC isoforms are cytosolic proteins that are structurally unrelated to the adenylyl cyclases. Furthermore, the _q-PLC interaction differs from that between _s and adenylyl cyclase in that PLC-1 can stimulate the GTPase activity of the  subunit that activates it (19).

This study utilized chimeric  subunits to localize the effector-specifying regions of _i2 and _q. We found that the effector-specifying regions of both _i2 and _q are localized within the  subunit GTPase domain. The region of _i2 that specifies interaction with adenylyl cyclase was localized to a 78-residue segment within this domain. Replacement of the _q residues that comprise the helical domain with _i2 residues resulted in a chimeric  subunit that activated PLC. Combined with previous studies of the effector-specifying residues of _s and _t, our results suggest that the effector specificity of  subunits, in general, is determined by the GTPase and not the helical domain. Furthermore, although GTP-dependent conformational changes are required for productive effector interactions, the conformational switch regions are not always determinants of effector specificity.

EXPERIMENTAL PROCEDURES

All chimeric  subunits were constructed from mouse _i2 cDNA (20) and mouse _q cDNA (21). Two modifications were made to each of the  subunits to facilitate detection of their activities and expression levels. The arginine at position 179 in _i2 and 183 in _q was mutated to cysteine to inhibit GTPase activity and produce constitutive activation (22, 23). An epitope, referred to as the EE epitope (24) was generated by mutating _i2 residues YPT (166-172) to EEYMPTE and _q residues YPT (171-176) to EYMPTE (single-letter amino acid code; mutated residues are underlined). The resultant constructs were designated _i2RCEE and _qRCEE. The _i2RCEE cDNA was a gift of Ann Pace and Henry Bourne. The _qEE cDNA (gift of Paul Wilson and Henry Bourne) was subcloned into the SphI site of the pcDNA I/Amp expression vector (Invitrogen), and arginine 183 was mutated to cysteine to produce _qRCEE by oligonucleotide-directed in vitro mutagenesis (25) using the Bio-Rad Muta-Gene kit. _oRCEE was generated from the rat _o cDNA (26) by mutating arginine 179 to cysteine and residues YPTE (167-172) to EYMPTE. Subcloning and mutagenesis procedures were verified by restriction enzyme analysis and DNA sequencing.

The sequence of the multiple cloning site in pcDNA I/Amp was mutated to produce two modified constructs. In pcDNA I/AmpA, the EcoRV site was mutated to a BstEII site. In pcDNA I/AmpB, the HindIII site was removed and a BstEII site was introduced in between the BamHI and BstXI sites.

To generate _q/_i2 chimeras, the unique BamHI site in _i2RCEE was removed and the cDNA was subcloned into the EcoRI site of pcDNA I/AmpB. The _qRCEE cDNA was then subcloned upstream of _i2RCEE into the BamHI site of this construct. The plasmid (5 µg) was then linearized at the unique HindIII and BstEII restriction sites in between the two cDNAs and transformed into Library Efficiency HB101 chemically competent cells (Life Technologies, Inc.) according to the manufacturer's procedure. Colonies were initially screened by restriction mapping to distinguish between tandem insert constructs and recombinant  subunits and to roughly localize chimera junctions. The precise cross-over points were then identified by DNA sequencing. The _q/_i2 chimeras were designated as QI followed by the number of the bacterial colony from which the cDNA was isolated.

For the _i2/_q chimeras, the unique SphI site in _i2RCEE was removed and the cDNA was subcloned into the EcoRI site of pcDNA I/AmpA. The _qRCEE cDNA was then subcloned downstream of _i2RCEE into the SphI site of this construct. The plasmid (5 µg) was then linearized at the unique BstEII and SmaI restriction sites in between the two cDNAs and introduced into chemically competent MV1190 cells (Bio-Rad) that were prepared and transformed according to the method of Chung and Miller (27). Colonies were screened and sequenced as described above. The _i2/_q chimeras were designated as IQ followed by the number of the bacterial colony from which the cDNA was isolated.

To generate the QI8+IQ15 chimera, chimera IQ15 was digested with BglII and XhoI to yield a 457-bp fragment of sequence encoding codons 319-330 of _i2 and codons 335-359 and the 3-untranslated region of _q. BglII-XhoI restriction of Chimera QI8 removed 313 bp of sequence encoding codons 319-355 and the 3-untranslated region of _i2 that was replaced by the 457-bp fragment from chimera IQ15. To generate the QI13+IQ1 chimera, chimera IQ1 was digested with MscI and SphI to yield a 1166-bp fragment encoding codons 82-177 of _i2 and codons 182-359 and the 3-untranslated region of _q. Chimera QI13 was digested with MscI and SphI, removing 1024 bp of sequence encoding codons 82-355 and the 3-untranslated region of _i2 which was replaced by the 1166-bp fragment from chimera IQ1.

Recombinant  subunits were transiently expressed in the human embryonic kidney fibroblast line, HEK-293 (American Type Culture Collection, CRL-1573), using DEAE-dextran (28) under the control of the CMV promoter in pcDNA I/Amp. 10⁶ cells/60-mm dish were transfected with 0.1 µg of vector alone or were co-transfected with 0.1 µg of vector containing the constitutively activated _sRC mutant (arginine 201 is mutated to cysteine; Ref. 29), and 0.03, 0.1, or 0.3 µg of vector containing _i2RCEE, _qRCEE, _oRCEE, or a chimeric  subunit. Inhibition of _sRC-stimulated cAMP accumulation by the recombinant  subunits was measured by determining intracellular cAMP levels in cells labeled with [³H]adenine, as described (11). 24 h after transfection, the cells were replated in 24-well plates and labeled with [³H]adenine (5 µCi/ml for 24 h). The cells were then washed with 1 ml of assay medium (20 mM HEPES-buffered minimal essential medium with Earle's salts without bicarbonate) and incubated for 30 min at 37 °C in 0.5 ml of assay medium containing 1 mM 1-methyl-3-isobutylxanthine. The medium was then aspirated, and the cells were lysed by the addition of 5% trichloroacetic acid plus 1 mM each of ATP and cAMP. Nucleotides were separated on ion exchange columns (30). Conversion of ATP to cAMP was expressed as [³H]cAMP/([³H]ATP + [³H]cAMP) × 10³ (31). cAMP values in cells co-transfected with _sRC and other constructs were expressed as the fraction of cAMP accumulation in cells transfected only with _sRC.

Recombinant  subunits were transiently expressed in HEK-293 cells using DEAE-dextran (28). The assay for intracellular inositol phosphates in cells labeled with [³H]inositol was a modification of the one described in Ref. 23. 10⁶ cells/60-mm dish were transfected with 3 µg of vector alone or varying amounts of vector containing an  subunit construct (Figs. 5 and 6). 24 h after transfection, the cells were replated in 24-well plates and labeled with [³H]inositol (5 µCi/ml for 24 h). The cells were then washed with 1 ml of assay medium (20 mM HEPES-buffered minimal essential medium with Earle's salts without bicarbonate containing 10 mM LiCl) and then incubated for 1 h at 37 °C in 0.5 ml of assay medium. The medium was then aspirated, and the cells were lysed by the addition of 0.75 ml of ice-cold 20 mM formic acid. After a 30-min incubation at 4 °C, the lysates were brought to pH 8 with 0.1 ml of 3% ammonium hydroxide. The lysates were loaded onto 1-ml AG1-X8 Dowex columns, followed by the immediate addition of 1 ml of 0.18% ammonium hydroxide to elute the [³H]inositol fraction into scintillation vials containing 4 ml of Ultima Gold scintillation fluid (Packard). The columns were then washed with 4 ml of 40 mM ammonium formate, 0.1 M formic acid. Total [³H]inositol phosphates were eluted with 1 ml of 4 M ammonium formate, 0.2 M formic acid into scintillation vials containing 4 ml of Ultima-Flo AF scintillation fluid (Packard). Inositol phosphate production was expressed as [³H]inositol phosphate/([³H]inositol + [³H]inositol phosphate) × 10³.

The GTPase domain of _q is sufficient to endow an  subunit chimera with the ability to activate PLC.

HEK-293 cells in 150-mm dishes were transfected using DEAE-dextran (28). Membranes were prepared 48 h after transient transfection from cells that had been lysed by 10 passages through a 27-gauge needle in an ice-cold buffer containing 50 mM Tris, pH 8.0, 2.5 mM MgCl₂, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine, and 1 mM dithiothreitol. Nuclei were removed by low speed centrifugation at 4 °C, and the membrane fractions were then isolated by centrifugation for 30 min at 4 °C in a microcentrifuge. 25 µg of membrane proteins were then resolved by SDS-polyacrylamide electrophoresis (10%), transferred to nitrocellulose, and probed with the anti-EE monoclonal antibody (24), which was purified from hybridoma supernatants using E-Z-SEP reagents (Middlesex Sciences, Inc). The antigen-antibody complexes were detected using an anti-mouse horseradish peroxidase-linked antibody according to the ECL Western blotting protocol (Amersham). Quantitation of  subunit expression levels was performed by densitometry using an IS-1000 Digital Imaging System (Alpha Innotech Corp., San Leandro, CA).

RESULTS

Generation of _q/_i2 and _i2/_q Chimeras

To localize the _i2 and _q regions that specify interaction with and modulation of their respective effectors, we produced  subunit chimeras in which portions of _i2 were replaced by the corresponding portions of _q. Since _i2 but not _q inhibits adenylyl cyclase (17, 32) and _q but not _i2 activates PLC (33, 34), the same chimeras could be used to map the effector-specifying regions of both _i2 and _q. In order to characterize these chimeras after transient expression in HEK-293 cells, two features were included to enable measurement of their functions without interference from the activities of the _i and _q proteins endogenous to these cells. First, a conserved arginine was replaced by cysteine (R179C in _i2, R183C in _q). This mutation constitutively activates these  subunits by inhibiting their GTPase activities (22, 23). This mutation made it possible to measure inhibition of adenylyl cyclase or activation of PLC without requiring receptor-mediated activation of the chimeric  subunits. Second, the chimeras include an epitope from an internal region of polyoma virus medium T antigen, referred to as the EE epitope (24). In the x-ray crystal structure of _t (7), the site of this epitope maps onto an exposed loop connecting  helices E and F in the helical domain (Fig. 7). The EE epitope does not interfere with the _i2-adenylyl cyclase interaction (35) or the _q-PLC interaction (36).

We generated  subunit chimeras containing _q/_i2 or _i2/_q junctions using a novel method (37, 38, 39) that utilized Escherichia coli DNA repair enzymes to rapidly generate a set of chimeras containing junctions within regions of amino acid identity between _i2 and _q. Plasmids containing single ``head to tail'' copies of both  subunit genes were linearized in between the two cDNAs and transformed into bacteria. Transformants were derived from either uncut plasmids or plasmids that recircularized in vivo. Using restriction enzyme mapping, we screened 119 colonies generated after transformation of the linearized _q/_i2 plasmid and obtained 73 that were not the original plasmid. The restriction maps of 58 of these 73 constructs were consistent with their being _q/_i2 chimeras. The remaining colonies may have represented recombination events within the 5- or 3-untranslated regions of the cDNAs. Of 131 colonies produced from transformation of the linearized _i2/_q plasmid, 97 were not the original plasmid. The restriction maps of 56 of these 97 colonies were consistent with their being _i2/_q chimeras. The junctions of the chimeras were further localized with additional restriction enzymes and then determined precisely by DNA sequencing. The junctions of these chimeras were always in frame. As shown in Fig. 1, six unique _q/_i2 junctions and three unique _i2/_q junctions were obtained. The amount of sequence identity seen at the chimera junctions ranged from 8 to 20 bases.

Alignment of _i2 and _q sequences and diagrammatic representation of chimera junctions.

All of the chimeric  subunit cDNAs encoded proteins that were expressed in membranes of transiently transfected HEK-293 cells (Fig. 2). Chimeras QI5, QI1, and IQ15 were expressed at higher levels than those of _i2RCEE and _qRCEE, while chimeras QI13, QI23, IQ1, and QI8+IQ15, an _q/_i2/_q chimera derived from QI8 and IQ15 (see Fig. 3), were expressed at somewhat lower levels, and QI4, QI8, and IQ23 were expressed at significantly reduced levels. A commonly used criterion for proper  subunit folding and function is a GTPS-dependent decrease in sensitivity to trypsin at a conserved  subunit cleavage site (11, 40, 41, 42). We were unable to apply this criterion to the chimeric  subunits because _qRCEE expressed in HEK-293 cells exhibited extremely poor GTPS-dependent trypsin resistance relative to _i2RCEE (data not shown), presumably due to inadequate binding of GTPS in the absence of receptor stimulation (32).

Expression of recombinant  subunits in HEK-293 cell membranes.

Localization of the Region of _i2 That Specifies Inhibition of Adenylyl Cyclase

We measured the ability of recombinant  subunits to inhibit adenylyl cyclase in HEK-293 cells by co-expressing them with the constitutively activated _s mutant, _sRC, in which arginine 201 is mutated to cysteine (29). Transfection with 0.1 µg of vector containing _sRC resulted in an approximately 18-fold increase in cAMP production compared to cells transfected with the vector alone. Co-transfection with 0.3 µg of vector containing _i2RCEE resulted in ~60% inhibition of the cAMP response to _sRC, while co-transfection with the same amount of vector containing _oRCEE inhibited the response to _sRC by only ~14% (Fig. 3). _oRCEE was used as a negative control because it does not inhibit adenylyl cyclase types V (17) or VI (18). It is not known which adenylyl cyclases are present in HEK-293 cells, but it seems likely that types V and/or VI are present, since these are the only adenylyl cyclases known to exhibit inhibition by _i of _s-stimulated activity (17, 18). Co-transfection with 0.3 µg of vector containing _qRCEE caused an approximately 2-fold elevation of cAMP levels,² presumably due either to release of Ca⁺² from intracellular stores or to activation of protein kinase C, since _q has no direct effect on adenylyl cyclase activity (32). Since adenylyl cyclase types V and VI are inhibited by Ca⁺² (43, 44, 45), some other adenylyl cyclase isoform must be responsible for the activation seen with _qRCEE.

We tested all of the chimeras shown in Fig. 1 for their ability to inhibit adenylyl cyclase, except QI23, which was the only chimera that activated PLC (Fig. 5). Since activation of PLC indirectly causes activation of adenylyl cyclase in HEK-293 cells, interpretation of the effect of QI23 on adenylyl cyclase activity would have been difficult. Of the IQ chimeras, only IQ15 inhibited adenylyl cyclase, while all of the QI chimeras were able to inhibit adenylyl cyclase (Fig. 3). The inability of IQ1 and IQ23 to inhibit adenylyl cyclase cannot be attributed to their reduced expression levels, which are greater than or equal to those of QI4 and QI8 (Fig. 2), although it is possible that they do not fold properly.

The _q/_i2 and _i2/_q chimera results suggested that the ability to inhibit adenylyl cyclase is specified by a 78-residue segment, _i2 residues 245-322 (_i2 residues 323-330 are identical to _q residues 327-334). Specifically, IQ15 and QI8 showed that _i2 residues 331-355 and 1-244, respectively, could be replaced by the homologous _q residues without affecting _i2 effector specificity. To test whether _i2 residues 245-322 were sufficient to specify inhibition of adenylyl cyclase by an  subunit chimera, we produced an _q/_i2/_q chimera, QI8+IQ15, in which this segment of _i2 replaced the homologous segment of _q (Fig. 3). QI8+IQ15 inhibited adenylyl cyclase activity as effectively as _i2RCEE did (Figs. 3 and 4). Thus, this chimera localizes the _i2 residues that specify inhibition of adenylyl cyclase to residues 245-322.

Substitution of a 79-residue segment of _i2 for its _q homologs endows _q with the ability to inhibit adenylyl cyclase.

To visualize the 78-residue _i2 effector-specifying segment in three dimensions, we mapped it onto the x-ray crystal structure of the GTPS-bound form of _t (7). 67% and 49% of the _t amino acids can be aligned with identical residues in the sequences of _i2 and _q, respectively. For comparison, 67% of the _i1 residues can be aligned with identical residues in _t and the structures of the GTPS-bound forms of _i1 and _t are closely superimposable. The _i2 effector-specifying segment defines a region that extends from 3 to 6, shown in magenta on the _t·GTPS structure (Fig. 7A). The segment includes three  helices (3, G, and 4), two  strands (5 and 6), and the loops that connect these elements of secondary structure. This _i2 segment includes regions corresponding to two of the three regions of _s that specify activation of adenylyl cyclase, which are located in the 3/5 and 4/6 loops (11). However, the _i2 region corresponding to the third effector-activating region of _s, which is part of Switch II (8), and the other two switch regions are not included.

Localization of the Region of _q That Specifies Activation of PLC

The abilities of the _q/_i2 and _i2/_q chimeras to activate PLC were tested by transiently expressing them in HEK-293 cells. In cells transfected with 3 µg of vector containing _qRCEE, inositol phosphate levels were elevated ~30-fold over levels in cells transfected with vector alone. Of the _q/_i2 chimeras, only QI23, in which the extreme carboxyl-terminal region of _q (residues 335-359) was replaced by _i2 sequence, exhibited any PLC-activating ability, elevating inositol phosphate levels ~8-fold above levels in vector-transfected cells (Fig. 5). This chimera was expressed at approximately half the level that _qRCEE was (Fig. 2). Thus, the decreased activity of QI23, relative to _qRCEE, was not entirely due to its decreased expression level. _q/_i2 chimeras in which larger carboxyl-terminal segments of _q sequence were replaced by _i2 sequence (QI8, QI4, QI1, QI5, and QI13) exhibited no ability to activate PLC. Furthermore, the _q/_i2/_q chimera, QI8+IQ15, in which only 86 _q residues in the carboxyl-terminal portion of the molecule (amino acids 249-334) were replaced by _i2 homologs, had no effect on PLC. All of the chimeras that did not activate PLC were expressed in HEK-293 cell membranes (Fig. 2) and all except IQ23 and IQ1 could inhibit adenylyl cyclase (Fig. 3), indicating that they were functional  subunits.

The above results suggested that a region in the carboxyl-terminal portion of _q that includes residues 249-334 is necessary for PLC activation. However, results with the _i2/_q chimeras demonstrated that the carboxyl-terminal region of _q is not sufficient to specify PLC activation. Chimeras IQ23 and IQ1, which contain _q residues 222-359 and 182-359, respectively, did not activate PLC (Fig. 5). Reasoning that in addition to _q residues 182-359, amino-terminal _q residues are required for PLC activation, we generated an _q/_i2/_q chimera using fragments from IQ1 and QI13. In the resultant chimera, QI13+IQ1 (Fig. 5), _i2 sequence was substituted for _q residues 57-181 in the helical domain (_q residues 57-59 are identical to _i2 residues 51-53).

QI13+IQ1 activated PLC, although both its activity (Figs. 5 and 6A) and expression level (Fig. 6B) were reduced relative to those of _qRCEE. Inositol phosphate production in cells transfected with 3 µg of vector containing QI13+IQ1 was ~5-fold elevated over that of vector-transfected cells (Fig. 5). For both _qRCEE and QI13+IQ1, the relationship between amount of transfected plasmid and inositol phosphate production was approximately linear when cells were transfected with 0.5 µg of plasmid. Within this range, _qRCEE elevated inositol phosphate production above that in vector-transfected cells ~9 times as much as QI13+IQ1 did (Fig. 6A). Similarly, the amount of immunoreactive  subunit expressed in membranes varied linearly with amount of transfected plasmid, and _qRCEE was expressed at ~8-fold the level of QI13+IQ1 (Fig. 6, B and C). Therefore, the PLC-activating ability of QI13+IQ1 is comparable to that of _qRCEE when expression level is controlled for. The fact that QI13+IQ1 but not IQ1 activates PLC indicates that _q residues 1-56 are either directly involved in effector interaction or are required for proper  subunit folding. We cannot distinguish between these possibilities since we were unable to determine whether IQ1 adopts a native  subunit conformation.

When mapped onto the _t·GTPS structure, the _q segment, residues 57-181, which was replaced by homologous _i2 residues in an  subunit chimera that activates PLC, localizes to the entire  subunit helical domain (shown in magenta in Fig. 7B), beginning within the second half of 1 and extending to the end of F immediately before Switch I. Thus, the GTPase domain of _q (shown in green in Fig. 7B) is sufficient to specify activation of PLC. Furthermore, our results suggest that residues within both segments of primary structure that comprise the GTPase domain, residues 1-56 and residues 182-359, are important for specifying this effector interaction.

DISCUSSION

In the study described here, the regions of _i2 and _q that specify interaction with and modulation of adenylyl cyclase and PLC, respectively, were localized using chimeric  subunits composed of portions of the two  subunits. Of the two  subunit domains, the GTPase domain and not the helical domain specifies the effector interactions of both _i2 and _q. The region of _i2 that specifies inhibition of adenylyl cyclase was localized to a 78-residue segment (amino acids 245-322) that extends from 3 to 6 within the GTPase domain. The effector-specifying region of _q was also localized to the GTPase domain, by means of a chimeric  subunit that substituted _i2 homologs for an _q segment (residues 57-181) extending from the second half of 1 to the end of F in the helical domain.

Previous experiments localized the effector-specifying residues of _s to three adjacent regions within the GTPase domain: the carboxyl-terminal end of 2 and the 2/4 loop, the 3/5 loop, and the 4/6 loop (11). In _t, the following regions within the GTPase domain are involved in activation of cGMP phosphodiesterase: 2 (14), 3 and the 3/5 loop (15), and 4 and the 4/6 loop (12, 13). Thus, the effector-specifying segment of _i2 includes residues that correspond to residues in two of the three effector-specifying regions of _s and _t, while that of _q includes all of these regions.

Although the function of the helical domain is still under investigation, current evidence suggests that it serves as a regulator of guanine nucleotide handling. When the helical and GTPase domains of _s are individually expressed and reconstituted in vitro, the helical domain dramatically stimulates the GTPase activity of the GTPase domain and also promotes binding of GTPS (46). Furthermore, interactions between the helical and GTPase domains are most likely involved in assuring that guanine nucleotide exchange is tightly regulated by activated receptors, since the two domains effectively bury the bound GDP (8, 10, 47, 48). The two domains, which are connected by two short stretches of sequence, presumably shift relative to one another to allow receptor-dependent nucleotide exchange.

Others have suggested that residues within the helical domains of _s and _i might be important for effector interactions. A chimera in which human _s residues extending from A to C in the helical domain were replaced by the corresponding residues from Xenopus _s, which does not activate adenylyl cyclase, had a greatly reduced ability to activate adenylyl cyclase. Conversely, a chimera in which the corresponding human _s residues were substituted for Xenopus _s residues effectively activated adenylyl cyclase (49). However, 17 of the 19 residues that differ between human and Xenopus _s in this region also differ between mammalian _s and _i2, and substitution of _i2 residues for these _s residues does not prevent activation of adenylyl cyclase (11, 50, 51). The two residues within this region that are the same in mammalian _s and _i2, but different in Xenopus _s, are also different from mammalian _s in either Drosophila _s (52) or _olf (53), both of which stimulate adenylyl cyclase. These results suggest that the inability of Xenopus _s to activate adenylyl cyclase is not due to a lack of effector-specifying residues, but instead is the result of a more general difference in  subunit function such as nucleotide handling and/or ability to become activated in response to GTP binding. Interestingly, a segment of _i1 within this region (residues 110-120) exhibits the largest deviation between the structures of the active forms of _i1 and _t, which led to the suggestion that this region might be an effector contact site for _i (9). However, this region does not specify interaction with adenylyl cyclase, because  subunit chimeras in which entirely divergent _q sequence replaced the _i2 sequence in this region effectively inhibited adenylyl cyclase.

It has been proposed that the amino terminus of _i binds to adenylyl cyclase (48), since _i must be myristoylated in order to inhibit this effector enzyme (17). We have shown that the amino-terminal sequence of _i2 does not specify inhibition of adenylyl cyclase, since chimeric  subunits in which the amino terminus of _i2 is replaced by the highly divergent sequence of _q can effectively inhibit adenylyl cyclase. However, since _q is palmitoylated at the amino terminus (54), it is possible that a lipid modification in this region is required for the interaction between _i and the membrane-bound adenylyl cyclase.

The effector-specifying region of _i2, as defined by QI8+IQ15, does not include any of the GTP-dependent conformational switch regions identified by comparison of the x-ray crystal structures of the GTPS-bound forms of _t (7) and _i1 (9) with the GDP-bound forms of these  subunits (8, 10). Since the GTP-bound form of _i is much more effective than the GDP-bound form is at inhibiting adenylyl cyclase (18), we expected that one or more of these conformational switch regions would be involved in this effector interaction. However, since the switch regions are highly conserved among  subunits, their importance as effector-interacting sites could have been missed using a chimeric  subunit approach. Although conserved residues within the switch regions might not confer specificity for a particular effector, they could be contact sites that allow effectors to discriminate between the GTP- and GDP-bound forms of  subunits. In fact, alanine substitution studies have shown that residues within a conserved part of Switch II are important for the effector interactions of both _s and _i2.³

Our study with  subunit chimeras demonstrates that the GTPase domain of _q is both necessary and sufficient to specify interaction with PLC. The GTPase domain consists of two noncontiguous segments of sequence, a short amino-terminal region that extends from the amino terminus to the end of 1 and a larger carboxyl-terminal region that extends from 2 to the carboxyl terminus. Our results and those of others implicate both segments as being important for specifying interaction with PLC. Within the amino-terminal segment, mutational replacement of _q residues 9 and 10, the sites of palmitoylation, greatly reduces the ability of _q to activate PLC, although enzymatic removal of palmitate does not affect PLC activation (55). Peptides corresponding to two regions of _q sequence in the carboxyl-terminal segment, residues 251-265 and 306-319, block _q-stimulated PLC activity (56). These peptides map onto 3 and the 3/5 loop, and 4 and the 4/6 loop. Using _q/_s chimeras, Venkatakrishnan and Exton (57) found that _q residues 277-359 could be replaced by _s residues without reducing the ability of _q to activate PLC. However, replacement of _q residues 217-276 with _s homologs did cause a loss of function. Alanine scanning in this region identified two regions, _q residues 243-245 and 256-257, which, when mutated, caused a loss of PLC-activating ability without affecting the ability of _q to attain the GTP-dependent activated conformation. These residues are located in the 4/3 loop, which includes Switch III (8, 10), and in 3. The other two conformational switch regions are also localized within the GTPase domain and may be important for PLC activation.

The method used here to generate chimeric  subunits in vivo in E. coli should be applicable to any pair of genes that encode proteins with a similar amount of amino acid identity as _i2 and _q (~50%) and has already been used to make chimeric -amylases (39), neurotransmitter transporters (38), and adenylyl cyclases (37). The exact mechanism of recombination is not understood, but does not require the RecA protein since it occurs in RecA E. coli strains. The most likely recombination mechanism is that the linearized DNA is exonucleolytically digested upon entering bacteria, and homologous regions on opposite strands then base pair. The resultant molecule is repaired and ligated closed. The amount of sequence identity seen at the chimera junctions ranged from 8 to 20 bases. However, there are other regions in the sequences of _i2 and _q that share at least 8 bases of identical sequence where chimera junctions were not obtained. Although additional junctions might have been found if more colonies had been screened, junction formation may have been nonrandom since the same chimeras were isolated multiple times.

Localization of the effector-specifying residues of _i2 and _q at higher resolution as well as investigation of the roles of the conformational switch regions in effector interaction will provide further insight into the mechanism by which  subunits modulate effectors in a GTP-dependent manner. A complete understanding of  subunit-effector interactions will require determination of the structures of the effectors and characterization of the regions with which they interact with  subunits.