β3- and α1-Adrenergic Erk1/2 Activation Is Src- but Not Gi-mediated in Brown Adipocytes*

A novel signaling pathway for mediation of β3-adrenergic activation of the mitogen-activated protein kinases Erk1/2 (associated with proliferation, differentiation, and apoptosis) has recently been proposed, which implies mediation via constitutively coupled Gi-proteins and Gβγ-subunits, distinct from the classical cAMP pathway of β-adrenergic stimulation. To verify the significance of this pathway in cells in primary cultures that entopically express β3-adrenoreceptors, we examined the functionality of this pathway in cultured brown adipocytes. Norepinephrine activated Erk1/2 via both β3 receptors and α1 receptors but not via α2 receptors. Forskolin induced Erk1/2 activation similarly to β3activation, indicating cAMP-mediation; this induction could be inhibited with H89, implying protein kinase A mediation. The Gi-pathway was functional in these cells, as pertussis toxin increased agonist-induced cAMP accumulation. However, pertussis toxin was unable to affect adrenergically induced Erk1/2 activation. Also, wortmannin was without effect, implying that Gβγ activation of the phosphatidylinositol 3-kinase pathway was not involved. PP1/2, which inhibits Src, abolished both β3- and α1-induced Erk1/2 activation. Thus, the proposed novel Gi pathway for β3 mediation is not universal, because it is not functional in the untransformed primary cell culture system with entopically expressed β3 receptors examined here. Here, the β3 signal is mediated classically via cAMP/protein kinase A. β3 and α1 signals converge at Src, which thus mediates Erk1/2 activation in both pathways.

In its general structure, the ␤ 3 -adrenoreceptor distinguishes itself from the ␤ 1 /␤ 2 receptors, which are more similar between themselves (1, 2). This distinction not only reflects different pharmacologies (i.e. differences in the adrenergic ligand binding site) but is also known to be reflected in the signaling function; desensitization via protein kinase A-mediated and G-protein-coupled receptor kinase-mediated receptor phosphorylation only occurs in the ␤ 1 /␤ 2 receptors and not in the ␤ 3 receptors (3). However, the distinction between the receptor subtypes may be anticipated also in other ways to be associated with different intracellular signaling pathways for the ␤ 3 receptor versus the ␤ 1 /␤ 2 receptors.
The signal from ␤ 1 /␤ 2 receptors is unquestionably mediated via the classical pathway leading from activation of G s through activation of adenylyl cyclase to increases in cAMP and activation of protein kinase A. A similar pathway is undoubtedly also activated by the ␤ 3 receptors. However, alternative signaling pathways for the ␤ 3 receptor may also exist. Indeed, based on experiments performed by several groups on the adipocyte-like cell line 3T3-F442A and especially in systems in which ␤ 3 receptors are ectopically expressed (4,5), it has been demonstrated that an alternative, suggested constitutive, primary pathway for ␤ 3 -signaling may exist. According to these investigations, the ␤ 3 receptor is not only coupled to G s but also constitutively to G i (4 -6). In its turn, G i may activate the Erk1/2 1 mitogen-activated protein kinase, as observed in several systems (7)(8)(9). Thus, a very interesting novel pathway has been suggested (4,5) in which ␤ 3 /G i activation, through ␤␥ mediation, would lead to activation of Erk1/2, with implied major effects on cellular proliferation, apoptosis, and cell differentiation. Thus, in this suggested model, ␤ 3 activates Erk1/2 in a G s /cAMP-independent manner.
Due to the potential significance of this novel pathway in several physiological contexts, we found it of importance to examine its role in the paradigmatic physiological system for investigation of ␤ 3 -mediated effects: brown adipocytes. In the cultured brown adipocyte system under investigation here, the ␤-adrenergic signal is virtually exclusively mediated via ␤ 3 receptors. The ␤ 2 receptor gene is practically silent in these cells (10), and the ␤ 1 receptors present do not make a measurable contribution to e.g. cAMP accumulation (11)(12)(13). Furthermore, in these cells, adrenergic stimulation has already been demonstrated to lead to an activation of Erk1/2 (14,15), which has been shown to be of high physiological significance, as it is linked to inhibition of apoptosis (15). We therefore consider this an optimal system in which to elucidate the physiological significance of the proposed novel ␤ 3 /G i pathway for Erk1/2 activation.
We found that in these non-transformed cells, the adrenergic signal that activated Erk1/2 was mediated via both ␣ 1 and ␤ 3 receptors. There was, however, no contribution from ␣ 2 stimulation; thus, ␣ 2 -induced G i activation could not activate Erk1/2. The ␣ 1 activation was mediated via Src activation. The ␤ 3 stimulation was also mediated via Src activation, but the pathway from the ␤ 3 receptor to this Src activation was classical in the sense that it was mediated via G s , adenylyl cyclase activation, increase in cAMP levels, and activation of protein kinase A rather than through the ␤␥-subunits of the G-proteins. It was especially noteworthy that inhibition of the G i system did not in any way affect the ability of ␤ 3 stimulation to activate the Erk1/2 cascade.
We conclude that the proposed novel ␤ 3 /G i /G␤␥ pathway for Erk1/2 activation is not universal, because it is not responsible for Erk1/2 activation in the paradigmatic ␤ 3 system of the brown adipocytes. To which extent the proposed G i /Erk1/2 pathway is of significance in other physiologically relevant systems requires identification of other entopic ␤ 3 systems coupled to Erk1/2 activation in cell types different from brown adipocytes.

EXPERIMENTAL PROCEDURES
Cell Culture-Three-week-old male mice (NMRI strain; Eklunds, Stockholm, Sweden) were kept at 22°C with free access to food and water for 2-3 days before sacrifice by CO 2 anesthesia. Brown adipocyte precursor cells were isolated from pooled interscapular, axillary, and cervical brown adipose tissue, as described earlier (16,17). In brief, tissue was minced in a Hepes-buffered Ringer solution containing 0.1-0.2% (w/v) collagenase type II (Sigma) and digested for 30 min at 37°C. The digest was successively filtered through a 250-m and a 25-m nylon filter to remove undigested material and mature cells. The precursor cells were pelleted by centrifugation (10 min, 2000 rpm), washed in Dulbecco's modified Eagle's medium (Life Technologies, Inc.), pelleted again, and resuspended in 0.5 ml of culture medium per mouse. The precursor cells were seeded into multi-well culture dishes (Falcon; Corning) at an inoculation density corresponding to 2.5 mice/6-well plate (growth area 9.4 cm 2 /well). Cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% newborn calf serum (Life Technologies, Inc.), 2.4 nM insulin, 25 g/ml sodium ascorbate, 10 mM Hepes, 4 mM glutamine, 50 IU/ml penicillin, and 50 g/ml streptomycin at 37°C in an atmosphere of 8% CO 2 in air. Cells were washed with Dulbecco's modified Eagle's medium on day 1, and medium was changed on days 1 and 3 of culture. On day 4, the medium was changed to one without serum, consisting of Dulbecco's modified Eagle's medium/Ham's F12 (1:1 mixture, Life Technologies, Inc.) supplemented with 0.5% (w/v) fatty acid-free bovine serum albumin (fraction V, Roche Molecular Biochemicals), 2.4 nM insulin, 25 g/ml sodium ascorbate, 10 mM Hepes, 4 mM L-glutamine, 50 IU/ml penicilin, and 50 g/ml streptomycin. The cells were incubated in this medium for 24 h to reduce basal Erk1/2 activity before experiments (15).
Erk1/2 Phosphorylation-On day 5, cultures were treated as indicated in the individual experiments, medium was aspirated, and cells were lysed directly in the well by the addition of 80 l of 65°C SDS sample buffer (62.5 mM Tris-HCl (pH 6.8), 2% SDS, 10% glycerol, 50 mM dithiothreitol, and 0.1% bromphenol blue). Cells were scraped, transferred to an Eppendorf tube on ice, and sonicated for 10 s followed by heating to 100°C for 5 min. Aliquots of the samples were separated on a 12% polyacrylamide gel and electrotransferred to a Hybond-C Extra nitrocellulose membrane (pore size 0.45 m; Amersham Pharmacia Biotech) with a semidry electroblotter. After transfer, the membranes were allowed to soak in Tris-buffered saline for 5 min, followed by quenching of nonspecific binding (1 h at room temperature in 5% nonfat dry milk, 0.1% Tween 20 in Tris-buffered saline). After quenching, the membranes were incubated with primary antibody (1:1000 dilution of phospho-p44/42 mitogen-activated protein kinase (Thr-202/Tyr-204) (New England BioLabs)) overnight at 4°C. The primary antibody was detected with a secondary antibody (horseradish peroxidase-conjugated goat anti-rabbit (New England BioLabs)) and enhanced chemiluminescence (ECL; Amersham Pharmacia Biotech). The membranes were then stripped with 10 M urea, 50 mM sodium phosphate, 100 mM ␤-mercaptoethanol for 30 min at 50°C and reprobed with a p44/42 mitogenactivated protein kinase antibody (New England BioLabs), detected with the same secondary antibody. The blots were exposed to Kodak X-Omat AR films and quantified on a Molecular Dynamics densitometer. The ratio between phosphorylated and total Erk protein p-Erk/Erk is shown in the results; the ratio was normalized in each experiment to that observed with 1 M NE.
cAMP Assay-Brown adipocytes were stimulated as indicated and treated as described by Bronnikov et al. (13). cAMP levels were determined with the cyclic AMP assay system (Amersham Pharmacia Biotech) according to the manufacturer's instructions. In the pertussis toxin (PTX) experiment, 10 M yohimbine and 3 g/ml adenosine deaminase (Sigma) were added 10 min before agonist stimulation.

Adrenergic Receptors
Involved in the Activation of Erk1/2 in Brown Adipocytes-Norepinephrine activates Erk1/2 in brown adipocytes (14,15). To delineate in detail which adrenergic receptor subtype(s) mediates the norepinephrine effect, we stimulated brown adipocytes with different adrenergic agonists. Erk1/2 activation was measured as the level of phosphorylation of the two isoforms Erk1 and Erk2, which correlates with activation (15).
As expected, norepinephrine led to a very marked increase in Erk1/2 phosphorylation (Fig. 1A), which was transient (Fig. 1B). The ␣ 1 -adrenergic agonist cirazoline and the ␤-adrenergic agonist isoprenaline were also able to activate Erk1/2 in a transient manner ( Fig. 1, A-B), although to a lower extent than norepinephrine, indicating that the norepinephrine response was of a composite nature. In confirmation of this, the addition of cirazoline and BRL-37344 together gave 94 Ϯ 26% of the norepinephrine-induced level (mean of 2 experiments).
The ␣ 2 receptor agonist UK-14304 was without effect (Fig. 1A), as was the ␣ 2 agonist clonidine (10 M, 5 min, data not shown). As the ␣ 2 receptor antagonist yohimbine markedly increases norepinephrine-induced cAMP in these cells (13), the absence of effect was not due to an absence of G i -coupled ␣ 2 receptors in these cells. In the light of the suggested ability of a G i -coupled receptor to activate Erk1/2 (4,5,7,19), the absence of effect of ␣ 2 -adrenoreceptor stimulation is therefore noteworthy. These results imply that ␣ 1 -and ␤ 3 -coupled receptors, but not ␣ 2 -coupled receptors, may activate Erk1/2 activation in brown adipocytes.
To substantiate that activation of adenylyl cyclase is responsible for the ␤ 3 -adrenoreceptor activation of Erk1/2, we investigated the time course of forskolin-induced Erk1/2 phosphorylation. As seen in Fig. 1C, forskolin was able to induce Erk1/2 phosphorylation (in agreement with earlier findings (15)). The transient activation was correlated both temporarily and in extent to the isoprenaline-induced response (Fig. 1B).
Even in the presence of the ␤-antagonist propranolol (10 M, 5 min) during the experiment, forskolin (10 M, 5 min) was still able to induce Erk1/2 phosphorylation, implying that the presence of the ␤-adrenoreceptor in its activated form is not necessary for the forskolin effect (data not shown).

PKA-mediated Phosphorylation Is Required for Erk1/2 Activation by the ␤-Adrenergic
Receptor-Although the above experiments demonstrate that cAMP is able to mimic the ␤ 3adrenergic activation of Erk1/2, this may not necessarily be the pathway used for ␤ 3 -adrenergic mediation. To investigate whether the cAMP/PKA pathway is the one that mediates the ␤ 3 effect, we pretreated brown adipocytes with the PKA inhibitor H89 before stimulation with different adrenergic agonists (Fig. 2). Norepinephrine stimulation led to a reduction in the level of phosphorylation, but the reduction was not total, only about half that of the norepinephrine effect (Fig. 2). However, as norepinephrine stimulated Erk1/2 activation through both the ␣ 1 and the ␤ pathway, it would not be anticipated that the ␣ 1 component should be sensitive to H89. Indeed, the ␣ 1 component (i.e. the cirazoline-induced) was not inhibited at all by H89, explaining the only partial inhibition of the norepinephrine effect.
In contrast, isoprenaline-, CGP-12177-and BRL-37344-induced Erk1/2 phosphorylations were all totally abolished by H89 pretreatment. Also, the forskolin-induced Erk1/2 phosphorylation could be abolished by H89 pretreatment. This is of particular interest as it indicates that the inhibition is indeed of a post-cyclase step and is not due to receptor antagonism by H89 of the type demonstrated for the ␤ 1 -and ␤ 2 -adrenoreceptor (20) (furthermore, we have found no antagonism by H89 on ␤ 3 receptors in brown adipocytes 2 ). Thus, these results indicate that not only forskolin/cAMP-induced Erk1/2 activation is mediated via the PKA pathway, but that this is also the case for the ␤ 3 -induced activation. These results are therefore in contrast to earlier reports in other systems, indicating that ␤ 3adrenoreceptor-induced Erk1/2 activation was mediated by cAMP-independent pathways (4,5).
Erk1/2 Activation Is Independent of Pertussis Toxin-sensitive G i Protein in Brown Adipocytes-Concerning ␤-adrenergic activation of Erk1/2, two pathways have been suggested, both involving G i . Thus, in ␤ 2 -transfected HEK-293 cells, PKA is suggested to mediate a switch in receptor coupling from G s to G i , and it is further suggested that it is through this G i -coupling that Erk1/2 is activated (21). Similarly, as noted above, the ␤ 3 -adrenoreceptor has been suggested to couple to G i directly and thus activate Erk1/2 (4,5). Because both these suggestions ascribe an essential role to G i in the mediation, we investigated if G i was necessary for Erk1/2 activation in the brown adipocyte system. For this, we used the G i inhibitor PTX. To demonstrate that PTX was effective in these cells, we first examined the effect of PTX on agonist-induced cAMP accumulation (Fig. 3A). In the presence of PTX, the ␤ 3 -adrenoreceptor and the adrenocorticotropic hormone-induced cAMP levels were much elevated, indicating a PTX-sensitive G i -inhibitory effect. The experiment also demonstrated that G i is active during ␤ 3 stimulation, in agreement with earlier observations (5,6), and could therefore mediate the ␤ 3 effect on Erk1/2 activation.
Based on the above experiment, it was clear that PTX was able to inhibit G i in our system. Therefore, to investigate the G i involvement in ␤-adrenergic stimulation of Erk1/2 activation, brown adipocytes pretreated with PTX as above were stimulated with different adrenergic agonists. PTX pretreatment had no effect in itself on Erk1/2 phosphorylation (Fig. 3B). However, unexpectedly, PTX was also fully unable to inhibit norepinephrine-induced Erk1/2 phosphorylation. Whereas an inability to inhibit the ␣ 1 -(cirazoline)-induced phosphorylation was expected, it was clear that it was also equally unable to inhibit ␤ 3 -induced activation. Also, a long pre-incubation with PTX (100 ng/ml, 16 h) was fully unable to inhibit BRL-induced Erk1/2 phosphorylation (not shown). PTX-sensitive G i -proteins were therefore not required for activation of Erk1/2 in response to ␤ 3 -adrenoreceptor stimulation in brown adipocytes.
No Phosphatidylinositol 3-Kinase (PI3K) Involvement in Erk1/2 Activation-In certain systems, it has been reported that adrenergic receptors stimulate the Erk1/2 pathway through PI3K (4). To examine if this was also the case in the system here investigated, brown adipocytes were pretreated with the PI3K inhibitor wortmannin and then stimulated with norepinephrine or BRL-37344. The wortmannin pretreatment had no effect in itself on Erk1/2 phosphorylation, but it did inhibit insulin-induced Erk1/2 phosphorylation. However, wortmannin did not alter the norepinephrine-or BRL-37344induced Erk1/2 phosphorylation (Fig. 4) (in agreement with Shimizu et al. (14)). As PI3K has been discussed as a further mediator of G i -coupled stimulation, the inability of wortmannin to inhibit adrenergically induced Erk1/2 activation may be considered compatible with the demonstration above that G i is not the mediator of adrenergic stimulation.
Tyrosine Kinase Involvement in Norepinephrine-mediated Activation of Erk1/2-As the above experiments indicated that ␤ 3 -adrenergic activation of Erk1/2 was not mediated via the suggested G i /PI3K pathway, but rather via a classical cAMP/ PKA pathway, the nature of the steps downstream of PKA activation required elucidation. In general, activation of mitogen-activated protein kinase/Erk1/2 pathways has often been demonstrated to proceed via tyrosine kinases. To investigate whether also norepinephrine-induced Erk1/2 phosphorylation was dependent on protein-tyrosine kinase activity, we used the non-selective tyrosine kinase inhibitor genistein. As shown in Fig. 5, genistein fully inhibited norepinephrine-induced Erk1/2 phosphorylation. This implies that activation of Erk1/2 by norepinephrine requires tyrosine kinase activity, independent of whether the activation was mediated via ␤ 3 -or ␣ 1 -adrenoreceptor (although the tyrosine kinase involved, of course, does not have to be the same for the two receptors).
Both ␤ 3 -and ␣ 1 -Adrenergic Receptor-mediated Erk1/2 Activation Is Src-dependent in Brown Adipocytes-To identify the tyrosine kinase(s) involved in Erk1/2 activation in response to adrenergic receptors, we specifically analyzed the possible involvement of Src tyrosine kinases. This was performed because Src has been considered a key protein in G-protein-coupled receptor-mediated Erk1/2 activation in several systems (19,22).
To investigate the involvement of Src, we used the inhibitor PP2. To demonstrate an intact cAMP elevation even in the presence of the Src inhibitor PP2, brown adipocytes were first pretreated with PP2 and then stimulated with norepinephrine. The ability of norepinephrine to increase cAMP levels was unchanged in the presence of PP2 (Fig. 6A).
In contrast, PP2 (or PP1) markedly reduced the level of Erk1/2 phosphorylation in response to norepinephrine, practically down to control levels ( Fig. 6B) (in other experiments, 50 M PP2 was used, which totally abolished the norepinephrineinduced Erk1/2 phosphorylation (61)). This in itself implied that both the ␤ 3 and the ␣ 1 pathways were Src-dependent. In agreement with this, activation via ␣ 1 -adrenoreceptor stimulation and Ca 2ϩ elevation (15) was fully inhibited by PP1/2, as was activation via ␤-adrenoreceptor and cAMP.
To examine whether Src mediation was specific for the adrenergic pathway, we also tested whether protein kinase Cmediated Erk1/2 activation (15) was Src-dependent. However, Erk1/2 phosphorylation induced by the protein kinase C-activating phorbol ester 12-O-tetradecanoylphorbol-13-acetate (50 nM, 5 min) could not be inhibited by PP1/PP2, demonstrating a mechanism for activation different from the adrenergic activation (not shown). This is in agreement with findings from HEK-293 cells where phorbol ester-induced Erk1/2 activation could not be abolished by Src inhibition (19). Thus, norepinephrine-induced Erk1/2 activation is dependent upon Src in brown adipocytes, irrespective of whether the adrenergic effect is via ␣ 1 -or ␤ 3 -adrenoreceptors or whether the intracellular mediator is Ca 2ϩ or cAMP. DISCUSSION In the present study, we have investigated norepinephrineinduced activation of the Erk1/2 cascade in brown adipocytes. The activation could be induced via both ␤ 3 and ␣ 1 receptors but not via ␣ 2 receptors. The ␤ 3 pathway was mediated via an increase in cAMP levels and an activation of protein kinase A, but in contrast to what was expected from studies in cell lines and in cells with ectopically expressed ␤ 3 receptors, it did not proceed via G i proteins. Both ␤ 3 -and ␣ 1 -induced Erk1/2 activation was mediated via Src activation (Fig. 7). Thus, in this physiologically relevant system for ␤ 3 mechanisms, we have not been able to find support for the existence of the ␤ 3 /G i pathway for Erk1/2 activation demonstrated in studies in other experimental systems.
Norepinephrine Can Activate Erk1/2 through Both ␣ 1 and ␤ 3 Receptors-The ability of norepinephrine to activate Erk1/2 in brown adipocytes is, as such, in agreement with earlier findings (14,15). An ability of norepinephrine to activate this pathway has also been demonstrated in several other cell types besides brown adipocytes, both in primary cell culture systems (23)(24)(25) and in cell lines (26,27). Classically, Erk1/2 activation has been considered to be induced by peptide growth factors and mediated via receptor tyrosine kinases (28,29). The generally accepted association between peptide growth factors and Erk1/2 activation is, however, probably not an inherent feature of this pathway but rather a consequence of the fact that most earlier studies were performed with fibroblast-like cell cultures, which are probably physiologically controlled by peptide growth factors. The ability of, for example, norepinephrine to activate Erk1/2 in brown adipocytes thus indicates that activation of this pathway is not agent/receptor-specific but rather occurs as a result of stimulation by the relevant external factor, i.e. the external factor that physiologically influences the processes of proliferation, differentiation, and survival in a specific cell type. As these features are under adrenergic control in brown adipocytes (11,15,17,30,31), it is consequential that in these cells, Erk1/2 activation is also under adrenergic control. That Erk1/2 activation is thus induced via distinct, tissue-specific pathways in cells from different tissues underlines the generality of the Erk1/2 pathway for control of proliferation, differentiation, and survival.
No G i Involvement in ␤ 3 -induced Erk1/2 Activation-In the present system, we were unable to observe any involvement of the G i pathway in the mediation of the ␤ 3 signal that activates Erk1/2. In this respect, the system responded distinctly differently from what would be predicted based on observations in ␤ 3 -transfected Chinese hamster ovary and HEK-293 cells and in 3T3-F442A cells (4,5). The non-involvement of G i was not due to a functional absence of G i in this system, as G i exhibited its classical inhibitory effects on adenylyl cyclase activity in these cells (Fig. 3A). Also ␣ 1 -induced Erk1/2 activation was mediated in a G i -independent manner, confirming the nonessentially of this mediator protein for Erk1/2 activation (although such an ␣ 1 -pathway has been described in primary hepatocytes (32)). An ability of G i to activate Erk1/2 has been observed in several other systems (7-9, 32), although a noninvolvement of G i in the mediation has also been reported in other G-protein-coupled systems (33,34).
When G i is actively involved in Erk1/2 activation, the signal has been discussed to be further mediated via the G␤␥-subunits through their interaction with the lipid kinase PI3K (4,35,36). This kinase has also been shown to mediate Erk1/2 activation by other agents such as insulin (37,38). However, in the present system, the PI3K inhibitor wortmannin did not influence norepinephrine-or ␤3-induced Erk1/2 activity. This absence is thus congruent with the non-involvement of G i and also indicates that PI3K activation is not an obligatory step in Erk1/2 activation. ␣ 2 -Adrenoreceptors, although Coupled to G i , do Not Activate Erk1/2 in Brown Adipocytes-In the brown adipocyte system studied here, ␣ 2 stimulation did not lead to Erk1/2 activation. This was not due to an absence of ␣ 2 receptors or a lack of coupling of the ␣ 2 receptors to internal pathways in these cells (13). However, considering the observation above that in these cells the ␤ 3 pathway does not proceed via G i , the absence of an ␣ 2 effect is not unexpected.
The ␤ 3 -induced Erk1/2 Activation Is cAMP-mediated-In agreement with observations in rat brown adipocytes (14), in ␤ 3 -transfected Chinese hamster ovary (4) and HEK-293 cells and in 3T3-F442A adipocytes (5) ␤ 3 receptor agonists were found to be able to activate Erk1/2 in the present experimental system. Also the classical ␤ receptor messenger cAMP was able to mimic this ␤ 3 activation. The ability of cAMP to activate Erk1/2 was in contrast to observations in other ␤ 3 -activated systems (4,5). However, the ability of cAMP, induced by forskolin or mimicked by analogues, to activate Erk1/2 is in agreement with earlier reports in brown adipocytes (14,15) in the 3T3-F442A adipocyte cell line (41) and in several cell types of neuronal origin (42,43). The close similarity between the response to forskolin and to ␤ 3 stimulation (Fig. 1) indicates a crucial positive role for cAMP in mediating the ␤-adrenergic response.
There is thus clearly no universality in the role of cAMP in regulation of the Erk1/2 pathway. Often, elevation of intracellular cAMP has inhibitory effects (44 -48), even in white adipocytes (49), but this is apparently paralleled by inhibitory effects of cAMP on cell proliferation in these systems (46,50). In contrast, in systems where cAMP activates proliferation, it also activates Erk1/2 (50). Concerning brown adipocytes, there is thus a clear parallelism between the ability of cAMP to promote proliferation, activation of Erk1/2, and inhibition of apoptosis (11,(13)(14)(15).
The ␤ 3 -induced Erk1/2 Activation Is Mediated via Protein Kinase A-The ␤ 3 signal was further mediated via protein kinase A activation, as evidenced by the ability of the protein kinase A inhibitor H89 to eliminate the activation. This pathway is thus not identical to that observed in ␤ 3 -transfected Chinese hamster ovary cells (4), in ␤ 3 -transfected HEK-293 cells, and in ␤ 3 -stimulated 3T3-F442A cells (5). In all these systems, Erk1/2 activation was independent of cAMP/PKA; in other systems, PKA may even be inhibitory (43,47,48).
The observation reported here concerning positive PKA involvement in Erk1/2 activation induced by ␤ 3 receptors must principally be considered to be of another nature than the positive involvement of PKA in the mediation of ␤ 2 stimulation earlier described in several systems (␤ 2 -transfected HEK-293 cells (5,21) and ␤ 2 -transfected cardiac myocytes (51)). In the model proposed for the involvement of PKA in mediation of this ␤ 2 effect (21,51,52), phosphorylation of the ␤ 2 receptor itself by the activated PKA mediates a G s to G i switch in the coupling of the ␤ 2 receptor, leading to Ras-dependent Erk1/2 activation by the G␤␥ subunit of G i . In this model, ␤ 2 -induced Erk1/2 activation is thus mediated successively via both G s /PKA and G i (although other models have been proposed (53)). As we found no evidence in brown adipocytes for an obligatory involvement of a G s to G i switch, the positive effect of cAMP in the ␤ 3 mediation described here is therefore not of the same nature as for the ␤ 2 receptor. The further mediation in the cAMP/PKAdependent Erk1/2 activation pathway may involve Rap-1/B-Raf (42,43), which are present in brown adipocytes (15).
G i -mediated Coupling of ␤ 3 Receptors in Other Experimental Systems-The distinctive absence of G i -mediated ␤ 3 -induced Erk1/2 activation in brown adipocytes, as compared with its experimental prominence in other systems (4,5), obviously raises the question of whether there are underlying principal differences between these systems that may explain their different signaling patterns. However, the ␤ 3 G i /Erk1/2 pathway has so far only been observed in two types of experimental systems: ␤ 3 -transfected cell lines and the 3T3-F442A cell line. In the transfected cells (4,5), the ␤ 3 receptors are ectopically expressed. The level of expression is probably high, and it is possible that an enhanced level of expression may cause an interaction between the receptors and intracellular pathways that would not occur if only physiological levels of receptors were present. Both the transfected cell lines and the 3T3-F442A cells (5) may also, as they are immortalized cells, be in a state concerning their Erk1/2 pathway activation level, that is more active than in non-transformed cells. Thus, also in this respect, it is possible that signaling connections may be induced between receptors and Erk1/2 pathways that would not occur in non-transformed cells, where these pathways may be in a state less prone to activation. ␣ 1 -Induced Erk1/2 Activation-Not only ␤ 3 receptors, but also ␣ 1 receptors, activated Erk1/2 in brown adipocytes. An ability of this pathway to activate Erk1/2 has also been observed in other experimental systems, both with ectopically expressed ␣ 1 receptors (19,27) and in cells endogenously expressing these receptors (25,32,54). In the present system, the mediation is probably initially through elevated Ca 2ϩ levels, as demonstrated in other ␣ 1 -induced systems (19,25,26). The further mediation has, at least in HEK-293 cells, been attributed to an interaction between Ca 2ϩ -calmodulin and the protein tyrosine kinase Pyk2 (19).
␤ 3 -and ␣ 1 Pathways Converge at Src-Both ␤ 3 -induced and ␣ 1 -induced Erk1/2 activation were fully inhibited by the Src inhibitor PP1/2, implying that the ␤ 3 and the ␣ 1 pathways, although clearly initially distinct, converge at this mediation point. This finding is in contrast to other suggestions concerning ␤ 3 -mediated Erk1/2 activation in Chinese hamster ovary cells, where Erk1/2 activation was independent of Src or other tyrosine kinases (4).
The further pathway from Src activation to Erk1/2 activation has not yet been clarified. A pathway involving G-proteincoupled receptor kinase-mediated receptor phosphorylation that recruits ␤-arrestin and Src to the membrane to form a complex with the adrenergic receptor has been suggested. Assembly of this protein complex, consisting of the receptor, ␤-arrestin, and Src, is supposed to trigger internalization of the complex and, through this, activation of the Erk1/2 pathway (22,52,55,56) (the mechanism by which this receptor internalization actually activates Erk1/2 has not been detailed). However, internalization is apparently not a universal phenomenon for Erk1/2 activation (57). To which extent receptor internalization is necessary for Erk1/2 activation in brown adipocytes is presently not known.
Conclusions-In several experimental systems, evidence for ␤ 3 -adrenergic activation of Erk1/2 through G i -coupled receptors has been demonstrated (4,5). This novel pathway has been investigated here in a primary cell culture system entopically expressing ␤ 3 receptors. In this system, we found no evidence for G i -mediation of ␤ 3 receptor-induced Erk1/2 activation.
It should be remembered that, concerning ␤ 3 effects, brown adipocytes are not a subsidiary system. Rather, this system is one of very few with entopically expressed ␤ 3 receptors where an essential function of these receptors, in vivo as well as in cell culture, has been established (13, 17, 58 -60). It was therefore of importance to test the suggested hypothesis for ␤ 3 -induced Erk1/2 activation in this physiologically relevant system. The absence of G i involvement in the brown adipocyte system must imply that its functioning in other systems needs to be demonstrated before the suggested pathway can be accepted as a physiologically relevant alternative for ␤ 3 -induced Erk1/2 activation. It is, however, clear already from the present experi-ments that this pathway cannot be considered a universal system for ␤ 3 -induced Erk1/2 activation.
Our data instead support a model in which ␤ 3 receptors mediate Erk1/2 activation through cAMP and protein kinase A. The signal converges with the ␣ 1 signal at the level of the tyrosine kinase Src and is independent of pertussis-toxin-sensitive G i -proteins.