The Second Intracellular Loop of the (cid:97) 2 -Adrenergic Receptors Determines Subtype-specific Coupling to cAMP Production*

The (cid:97) 2 -adrenergic receptors ( (cid:97) 2 -ARs), which primarily couple to inhibition of cAMP production, have been re-ported to have a stimulating effect on adenylyl cyclase activity in certain cases. When expressed in Spodoptera frugiperda Sf9 cells the (cid:97) 2A subtype showed only inhibi- tion of forskolin-stimulated cAMP production when ac-tivated by norepinephrine (NE), whereas the (cid:97) 2B sub- type displayed a biphasic dose-response curve with inhibition at low concentrations of NE and a potentia- tion at higher concentrations. To further investigate the subtype-specific coupling, we expressed a set of chi- meric (cid:97) 2A -/ (cid:97) 2B -ARs at similar expression levels in Sf9 cells to determine the structural domain responsible for the difference between the two subtypes. When the third intracellular loops were interchanged between (cid:97) 2A and (cid:97) 2B subtypes, the coupling specificity remained un- changed, indicating that this loop does not confer selectivity toward a stimulating response. A biphasic dose- response curve, typical for the (cid:97) 2B subtype, could be seen when the second intracellular loop of the (cid:97) 2B sub- type was inserted into the (cid:97) 2A subtype, suggesting that this loop is important for determining the subtype-spe- cific coupling of (cid:97) 2 -ARs to cAMP production. Site-di-rected mutagenesis of non-conserved amino acids in the second intracellular

The ␣ 2 -adrenergic receptors (␣ 2 -ARs), which primarily couple to inhibition of cAMP production, have been reported to have a stimulating effect on adenylyl cyclase activity in certain cases. When expressed in Spodoptera frugiperda Sf9 cells the ␣ 2A subtype showed only inhibition of forskolin-stimulated cAMP production when activated by norepinephrine (NE), whereas the ␣ 2B subtype displayed a biphasic dose-response curve with inhibition at low concentrations of NE and a potentiation at higher concentrations. To further investigate the subtype-specific coupling, we expressed a set of chimeric ␣ 2A -/␣ 2B -ARs at similar expression levels in Sf9 cells to determine the structural domain responsible for the difference between the two subtypes. When the third intracellular loops were interchanged between ␣ 2A and ␣ 2B subtypes, the coupling specificity remained unchanged, indicating that this loop does not confer selectivity toward a stimulating response. A biphasic doseresponse curve, typical for the ␣ 2B subtype, could be seen when the second intracellular loop of the ␣ 2B subtype was inserted into the ␣ 2A subtype, suggesting that this loop is important for determining the subtype-specific coupling of ␣ 2 -ARs to cAMP production. Site-directed mutagenesis of non-conserved amino acids in the second intracellular loop of the ␣ 2A subtype indicated that several residues are involved in the coupling specificity.
The ␣ 2 -adrenergic receptors (␣ 2 -ARs) 1 are members of a large family of heptahelical receptors mediating the extracellular stimuli to the interior of the cell through G proteins. Three different subtypes of ␣ 2 -ARs, ␣ 2A , ␣ 2B , and ␣ 2C , can be distinguished based on the affinity for selective ligands (1), and this subdivision has been confirmed with the molecular cloning of three separate genes (2)(3)(4). Funtional expression of cloned cDNAs in different cell types has shown that the ␣ 2 -ARs can mediate multiple cellular responses including inhibition or stimulation of adenylyl cyclase activity (5)(6)(7)(8)(9), activation of phospholipase A 2 and D (5,10), stimulation of phosphatidylinositol turnover (6), and mobilization of intracellular Ca 2ϩ (11,12).
In some cells, activation of ␣ 2 -ARs leads to a biphasic regulation of adenylyl cyclase activity with an inhibitory phase at low concentration of agonist and a stimulatory phase at higher concentrations (5,7,13,15), whereas in other cells, the response is exclusively inhibitory (6,8) or stimulatory (9,(13)(14)(15). The stimulation of cAMP production with ␣ 2 -AR agonists has also been shown in physiological systems with endogenous receptors (16,17), suggesting that it is not a mere side effect seen in recombinant systems.
The seemingly controversial effect of ␣ 2 -AR activation on stimulation of adenylyl cyclase activity has been attributed to relatively high expression levels of the receptors (7), cell-typespecific expression of different types of adenylyl cyclases (14,15), or overexpression of G s protein (18). A direct interaction of ␣ 2 -ARs with G s protein has been shown in Chinese hamster ovary cells (7), while in other cell types, the mechanism for stimulation seems less clear (9,15,19). In many cases, the stimulatory response has shown subtype selectivity, the stimulation being more pronounced with the ␣ 2B subtype compared with the ␣ 2A and ␣ 2C subtypes (9,13,15). This suggests that the G protein coupling domains differ between the subtypes. Different cytoplasmic domains have been implicated in G protein activation and selection both in adrenergic receptors (20,21) and other heptahelical receptors (22,23). The aim of the present study was to investigate the apparent subtype-specific coupling of the mouse ␣ 2 -AR subtypes, ␣ 2A and ␣ 2B , when expressed in Spodoptera frugiperda (Sf9) cells at comparable receptor levels, and to delineate the receptor domain leading to potentiation of cAMP production. Cell Culture-Sf9 cells were maintained as suspension culture at 25-27°C in TNM-FH medium (pH 6.3) supplemented with 10% fetal calf serum (Life Technologies, Inc., Paisley, UK), 100 units/ml penicillin (Nordvacc Media, Skä rholmen, Sweden), and 100 g/ml streptomycin (Nordvacc Media), and 2.5% amphotericin B (Life Technologies, Inc.). For expression, Sf9 cells were subcultured in monolayer and infected with recombinant baculoviruses at a multiplicity of infection of 2-5 for the indicated times.

Materials-[
Recombinant Baculoviruses-The cDNAs for the mouse ␣ 2A -and ␣ 2B -ARs and three mouse chimeric receptors, MCR1, -2, and -3, were a gift from Dr. B. Kobilka (Howard Hughes Medical Institute, Stanford University). All clones contained a hemagglutinin tag fused to the amino terminus of the receptor constructs. For generation of recombinant baculoviruses, the genes were subcloned into the baculovirus transfer vector plucGRBac1 (9) under the transcriptional regulation of the polyhedrin gene promoter. The transfer vectors were then used for cotransfection with wild-type Autographa Californica nuclear polyhedrosis virus DNA, and recombinant viruses were purified essentially as described by Jansson et al. (9).
To generate chimeric receptors MCR4 and MCR5, the cDNA clones for ␣ 2A -and ␣ 2B -ARs were transferred into the SmaI site of pBluescript (Stratagene, La Jolla, CA) in a KS orientation. A SapI restriction site was introduced into the ␣ 2A sequence at nucleotide position 351 of the coding sequence (equivalent to the position of a SapI site in the ␣ 2B cDNA) using standard PCR techniques (24). The sequence from the beginning of the ␣ 2A gene to the novel SapI site was then inserted into pBluescript-␣ 2B cut with the same enzyme to generate pBluescript-MCR4.
MCR5 was constructed by isolating a DraIII-fragment (DraIII cuts the cDNAs of ␣ 2A and ␣ 2B at an equivalent position and the plasmid DNA at one position) from pBluescript-␣ 2A and ligating it into an isolated DraIII-fragment of pBluescript-MCR4 to generate pBluescript-MCR5. The sequences encoding MCR4 and MCR5 were cut out from the pBluescript constructs with EcoRI and XbaI and ligated into pFastBac1 digested with the same enzymes. For production of recombinant baculovirus, a BAC-TO-BAC baculovirus expression system kit (Life Technologies, Inc.) was used.The correctness of all plasmid constructs were checked by restriction enzyme mapping and partial dideoxy sequencing (25).
Site-directed mutagenesis of the ␣ 2A sequence was performed using PCR as described (24). Mutated ␣ 2A sequences were subcloned into pBluescript, and the mutations were verified by sequencing. The mutated sequences were subsequently transferred to pFastBac1 with EcoRI and NotI. Recombinant baculoviruses were generated using the BAC-TO-BAC baculovirus expression system kit.
Measurement of Cellular cAMP-Sf9 cells were plated on tissue culture dishes and allowed to attach for 1 h before infection with respective recombinant baculovirus for the indicated times. The cells were incubated with 5 Ci/ml Receptor Binding Assay-Infected cells from monolayer cultures were harvested in phosphate-buffered saline solution and centrifuged 1500 ϫ g for 5 min. The cell pellet was resuspended in cold potassium phosphate buffer (40 mM K 2 HPO 4 , 10 mM KH 2 PO 4 , pH7.4) and homogenized with an Ultra-Turrax homogenizer (Janke and Kunkel, Germany). 100 -200 g of protein of the homogenate was incubated with 10 nM [ 3 H]RX821002 in a volume of 0.3 ml potassium phosphate buffer at 25°C for 30 min. 10 M phentolamine was used to determine nonspecific binding. The reactions were terminated by filtration through printed filtermat B filters (Wallac) using a Harvester 96 (Tomtec Inc., Orange, CO). After the filters had dried, a MeltiLex B/HS scintillator sheet was melted on them, and radioactivity was determined in a Microbeta scintillation counter (Wallac).

RESULTS
Expression of Mouse ␣ 2 -AR Subtypes-Infection of Sf9 cells with baculovirus harboring the genes for the mouse ␣ 2A -or ␣ 2B -AR resulted in a time-dependent increase in receptor density as determined by specific binding of [ 3 H]RX821002 (data not shown). Cells infected with wild-type virus did not show specific binding of [ 3 H]RX821002. When assayed for functional coupling to regulation of cAMP production, the ␣ 2A subtype mainly inhibited the forskolin stimulation (Fig. 1A), whereas the ␣ 2B subtype showed a biphasic response with inhibition at low concentration (1 M) of norepinephrine (NE) and a potentiating effect at higher concentrations (100 M) (Fig. 1B). The potentiation appeared to reach a maximum around 38 h postinfection (p.i.) and then to decline with longer infection times. For further characterization, 48 h p.i. was chosen because the magnitudes of inhibition versus potentiation of the forskolin response between the two subtypes were similar at this time point.
In Fig. 2, the dose-response curves for the two receptor subtypes with two different agonists, NE and UK14,304, are shown. UK14,304 was used as a control agonist for inhibition since this compound has been shown to be very weak in potentiating forskolin stimulation with the ␣ 2B subtype while being full agonist for the inhibition (27). Both NE and UK14,304 inhibited cAMP production in ␣ 2A -expressing cells to the same extent and with similar potencies. In cells expressing the ␣ 2B subtype, NE displayed a biphasic response with an inhibition at low concentrations, that reached a maximum at 1 M, and that potentiated the forskolin stimulation at higher concentrations. UK14,304 elicited mainly inhibition with the ␣ 2B subtype, confirming that UK14,304 is much less effective in elic- iting a stimulation of cAMP production compared with the endogenous agonist NE.
Pertussis toxin (PTX) has been used in Chinese hamster ovary cells to abolish the inhibition of cAMP production through G i proteins and thus reveal a less efficient coupling of ␣ 2 -ARs to a stimulatory component, presumably G s (7,19,27). When Sf9 cells were treated with PTX for 48 h, the inhibition was abolished in both ␣ 2A -and ␣ 2B -expressing cells, and the stimulatory response with the ␣ 2B subtype was enhanced over 2-fold (Fig. 3). Treatment of Sf9 cells with cholera toxin (CTX), which is known to activate G s proteins persistently, drastically reduced the inhibitory response with both subtypes (Fig. 3).
The influence of Ca 2ϩ on the cAMP production was tested by chelating extracellular Ca 2ϩ with EGTA. The chelation of extracellular Ca 2ϩ did not effect the responses to any larger extent in cells expressing either the ␣ 2A subtype or the ␣ 2B subtype (Fig. 3).
Interchange of Third Intracellular Loop Does Not Affect Coupling Specificity-To try to delineate the domain in the ␣ 2B subtype responsible for a potentiation of cAMP production, we expressed a set of chimeric ␣ 2A -/␣ 2B -receptors (Fig. 4).
MCR1 and MCR2, ␣ 2A and ␣ 2B with interchange of the third intracellular loop (i3-loop), respectively, show similar doseresponse curves with NE as the parent receptor (Fig. 5). MCR1 showed a reduced maximal inhibition compared with the ␣ 2A subtype, but no biphasic response mode could be seen. MCR2 displayed a typical biphasic response but with a reduced potentiating effect. Employing UK14,304 as the agonist increased the maximal inhibition with MCR2 compared with the parent ␣ 2B subtype, whereas the response with MCR1 did not deviate from the parent ␣ 2A subtype (data not shown).
MCR3, in which the carboxyl-terminal part from the end of the i3-loop of the ␣ 2A subtype was exchanged with ␣ 2B sequence, responded in the same way as the native ␣ 2A subtype (Fig. 6). MCR4, with ␣ 2B -sequence from the beginning of the second intracellular loop (i2-loop) to the C terminus, behaved as the intact ␣ 2B subtype in displaying a biphasic dose-response curve with NE (Fig. 6), indicating that the second intracellular loop might be involved in the potentiation seen with the ␣ 2B subtype.
Effect of Second Intracellular Loop on Coupling Specificity-To test the involvement of the i2-loop in the different coupling modes, we expressed a chimeric ␣ 2A receptor in which the i2-loop had been exchanged to ␣ 2B receptor sequence, MCR5. This construct exhibited a similar biphasic dose-response curve with NE as the ␣ 2B subtype 48 h p.i. (Fig. 7). Since this viral construct expressed receptors at somewhat higher density compared with the other constructs, we measured the dose-response relationship at 38 h p.i. when the receptor level was comparable to the parent ␣ 2A subtype (see Fig. 4.). At 38 h p.i., this construct stimulated cAMP production very potently, showing almost no inhibition with NE (Fig. 7), which indicated that the change in coupling specificity could not be attributed to differences in the expression levels of the chimeric receptor.
Site-directed Mutagenesis of the Second Intracellular Loop-There are six amino acid residues that differ between the ␣ 2A and ␣ 2B subtypes in the second intracellular loop (Fig. 4.). Three of these residues are essentially similar (Ile-135 in ␣ 2A versus Val at the corresponding position in ␣ 2B , Thr-136 versus Ser, and Ile-139 versus Leu). The three other non-conserved residues differ in terms of polarity (Ser-134 in ␣ 2A versus Ala at the corresponding position in ␣ 2B , Gln-137 versus Arg, and Leu-143 versus Ser). Site-directed mutagenesis of each of these three amino acids of the ␣ 2A subtype to corresponding residues of the ␣ 2B subtype did not give clear indications which residues might be responsible for the coupling specificity (Fig. 8.). These mutants all showed a lower degree of inhibition with NE compared with the ␣ 2A subtype, and no marked biphasic response could be seen. On the contrary, a double mutant, S137A, L143S, exhibited a biphasic response with NE similar to the ␣ 2B subtype although with a smaller magnitude of stimulation. All of the mutated constructs were expressed at receptor levels comparable with the ␣ 2A subtype, and the inhibition with UK14,304 was 40 -50% of forskolin stimulation with all four constructs (not shown). DISCUSSION During the last few years, it has become evident that G protein-coupled receptors can couple to multiple G proteins to elicit different cellular responses (7, 28 -30). The ␣ 2 -ARs have been shown to couple to both negative and positive regulation of adenylyl cyclase activity (5,7,9,(13)(14)(15). If coupling to both pathways occurs simultaneously but with different potencies, one could expect to obtain a biphasic dose-response curve. With the ␣ 2 -ARs, this is often the case; an inhibition of forskolinstimulated cAMP production is seen with low concentrations of agonist and a stimulation or potentiation of cAMP production with higher concentrations (5,7,13,15). Earlier studies have indicated that the stimulatory effect of ␣ 2 -ARs is cell-typespecific (15) and/or dependent on the expression level of the receptors (7). This response also seems to be subtype-specific, the ␣ 2B subtype having a more pronounced stimulatory effect than the ␣ 2A and ␣ 2C subtypes when expressed in the same cells (8,9,13,15,31). In the present study, we expressed the mouse ␣ 2A -and ␣ 2B -AR subtypes in Sf9 cells at comparable expression levels and obtained a marked difference in the coupling specificity between the subtypes. The ␣ 2B subtype showed a biphasic dose-response curve with a potentiation of the forskolin stimulation at high concentration of NE, whereas the ␣ 2A subtype displayed a monophasic inhibitory dose-response curve. With prolonged infection times, which parallel an increase in receptor density, the maximal inhibition with ␣ 2A was increased. This is in contrast to the finding by Eason et al. (7) that an increase in receptor density promotes the stimulatory pathway with the human ␣ 2A subtype (␣ 2 -C10). The reason for this discrepancy is unclear but might involve interspecies variation of the receptor subtypes or differences in the types of G proteins expressed by the two different cell lines used. A reduction in the maximal stimulation with longer infection times for the ␣ 2B subtype was also seen, which indicates a more efficient coupling to the inhibitory component at higher expression levels. The imidazoline-like agonist UK14,304 was very weak in eliciting a biphasic response with the ␣ 2B subtype. The ability of UK14,304 to promote coupling to a stimulatory pathway with the ␣ 2B subtype has been shown to be very weak in Chinese hamster ovary cells (27).
PTX treatment, which is known to reveal a stimulatory pathway to cAMP production with the ␣ 2 -ARs (7,19,27), increased the stimulation over 2-fold with the ␣ 2B subtype. This supports the hypothesis that coupling to both pathways occur simultaneously, and thereby, an elimination of either pathway would result in an enhancement of the other. PTX treatment of cells expressing the ␣ 2A subtype did not reveal any significant coupling to a stimulatory pathway although the inhibition was abolished. This confirms that the ability of the receptors to potentiate cAMP production is, apart from being related to expression levels, also subtype-specific.
Treatment of the cells with CTX, which would abolish the stimulation if it was G s -mediated, drastically reduced the inhibition with both subtypes but did not seem to affect the stimulatory component with the ␣ 2B subtype. These data are difficult to interpret, however, since CTX stimulated the adenylyl cyclase activity several-fold over the forskolin-stimulated activity, and this activity may be difficult to inhibit (9,32).
␣ 2 -AR-mediated stimulation of adenylyl cyclase activity has been suggested to occur through a rise in intracellular calcium concentration in PC12 cells (15). Since we observed a small but significant elevation of intracellular Ca 2ϩ in Sf9 cells expressing the ␣ 2B subtype when assayed with fura-2 fluorescence (data not shown), we measured the change in cAMP production in the prescence of EGTA. The potentiating response did not differ significantly from the control experiment while EGTA largely prevented the Ca 2ϩ elevation in the fura-2 assay. This suggests that Ca 2ϩ elevation is not the stimulating factor in the ␣ 2B -mediated potentiation of cAMP production in these cells although Ca 2ϩ may enhance the stimulatory response.
The third intracellular loop of G protein-coupled receptors has been implicated in G protein selectivity and activation (for review, see Ref. 33). When we expressed chimeric ␣ 2A -/␣ 2B -ARs with interchange of the i3-loops, the responses with NE were very similar to the parent receptor subtypes. A reduction of the potentiating response of ␣ 2B was seen with MCR2, and a reduction of the inhibition of ␣ 2A could be seen with MCR1. This may be related to a more efficient coupling to G i proteins through the i3-loops of the receptors. Another possibility is that the change of the i3-loop might alter the general conformation of the coupling device leading to slightly altered responses.
In an earlier study, where part of the carboxyl-terminal tail of the ␣ 2A subtype was introduced into the ␤ 2 -AR, a small reduction in isoproterenol-stimulated adenylyl cyclase activity was seen, suggesting an involvement of the carboxyl-terminal tail in G protein coupling (21). In this study, the exchange of the carboxyl-terminal tail in ␣ 2A subtype for ␣ 2B sequence (MCR3) did not alter the coupling mode for the ␣ 2A subtype (Fig. 6), indicating that this domain, if involved in coupling to G proteins in the ␣ 2 -ARs, probably interacts with a G i protein.
The chimeric receptor MCR4 displayed a typical biphasic doseresponse curve. This was also expected since this chimera contains ␣ 2B sequences in all the proposed G protein-coupling domains. We also constructed a chimeric ␣ 2A receptor with ␣ 2B sequence from the amino terminus to the distal end of the i2-loop, but this construct was not expressed at such a density that a functional characterization could be accomplished.
When the i2-loop from ␣ 2B was introduced into the ␣ 2A subtype, the chimeric receptor displayed a biphasic dose-response curve with NE similar to the ␣ 2B subtype when assayed 48 h p.i. At an earlier time point (38 h p.i.), the chimeric receptor stimulated cAMP production even more potently than did the native ␣ 2B subtype at the same time postinfection. There are two possible explanations for this. First, the chimeric construct was expressed at higher receptor levels than the ␣ 2B construct, about 3.8 pmol/mg of protein for MCR5 compared with about 1.5 pmol/mg of protein for the ␣ 2B subtype. A higher expression could increase coupling to the stimulatory component and thereby mask a coupling to the inhibitory component. Second, a G s -coupling domain in the third intracellular loop of the human ␣ 2A subtype has been identified based on studies with chimeric ␣ 2 /5-HT 1A receptors (34). Exchange of the i2-loop from the ␣ 2B subtype might result in a receptor that couples more tightly to a stimulating G protein.
The i2-loop has been implemented in G protein selectivity in a study using muscarinic m1:␤-adrenergic receptor chimeras (22). The role of the i2-loop in G protein activation has also been studied using synthetic peptides derived from ␣ 2 -ARs and muscarinic receptors (35,36). In the study by Okamoto and Nishimoto (35), the peptide derived from the i2-loop of the human ␣ 2A subtype (␣ 2 -C10) potently stimulated GTP␥S binding to G s proteins. Although this finding is in contrast to our results that the ␣ 2A receptor is strictly inhibiting, other regions of the receptor are probably also involved in governing the selectivity.
The i2-loop sequences of the ␣ 2A and ␣ 2B -ARs differ at six positions when predicted from the cDNA sequences. There are three non-conserved amino acid substitutions between the two subtypes that will change the polarity of the i2-loop, Ser-134, Gln-137, and Leu-143, in the ␣ 2A subtype. Single point-mutated receptors with each of these non-conserved residues substituted with corresponding ␣ 2B residues showed no clear biphasic responses with NE. In contrast, a double mutant with S134A and L143S responded in a similar way to the ␣ 2B subtype. The interpretation of this is that at least these two residues are needed to maintain the integrity of the coupling domain. Interestingly, in the ␤ 2 -adrenergic receptor, the equivalent residues (Ser-134 and Leu-143) are the same as in the ␣ 2B subtype.
In conclusion, we have presented evidence that the coupling of ␣ 2 -ARs to cAMP production is subtype-specific and that the second intracellular loop of the ␣ 2B subtype determines the different coupling specificity leading to stimulation of cAMP production. Site-directed mutagenesis of non-conserved amino acid residues indicates that the whole structure of the second intracellular loop is important for efficient coupling.