Tethering of the platelet-derived growth factor beta receptor to G-protein-coupled receptors. A novel platform for integrative signaling by these receptor classes in mammalian cells.

Here we provide evidence to show that the platelet-derived growth factor beta receptor is tethered to endogenous G-protein-coupled receptor(s) in human embryonic kidney 293 cells. The tethered receptor complex provides a platform on which receptor tyrosine kinase and G-protein-coupled receptor signals can be integrated to produce more efficient stimulation of the p42/p44 mitogen-activated protein kinase pathway. This was based on several lines of evidence. First, we have shown that pertussis toxin (which uncouples G-protein-coupled receptors from inhibitory G-proteins) reduced the platelet-derived growth factor stimulation of p42/p44 mitogen-activated protein kinase. Second, transfection of cells with inhibitory G-protein alpha subunit increased the activation of p42/p44 mitogen-activated protein kinase by platelet-derived growth factor. Third, platelet-derived growth factor stimulated the tyrosine phosphorylation of the inhibitory G-protein alpha subunit, which was blocked by the platelet-derived growth factor kinase inhibitor, tyrphostin AG 1296. We have also shown that the platelet-derived growth factor beta receptor forms a tethered complex with Myc-tagged endothelial differentiation gene 1 (a G-protein-coupled receptor whose agonist is sphingosine 1-phosphate) in cells co-transfected with these receptors. This facilitates platelet-derived growth factor-stimulated tyrosine phosphorylation of the inhibitory G-protein alpha subunit and increases p42/p44 mitogen-activated protein kinase activation. In addition, we found that G-protein-coupled receptor kinase 2 and beta-arrestin I can associate with the platelet-derived growth factor beta receptor. These proteins play an important role in regulating endocytosis of G-protein-coupled receptor signal complexes, which is required for activation of p42/p44 mitogen-activated protein kinase. Thus, platelet-derived growth factor beta receptor signaling may be initiated by G-protein-coupled receptor kinase 2/beta-arrestin I that has been recruited to the platelet-derived growth factor beta receptor by its tethering to a G-protein-coupled receptor(s). These results provide a model that may account for the co-mitogenic effect of certain G-protein-coupled receptor agonists with platelet-derived growth factor on DNA synthesis.

Mitogenic stimuli initiate cell proliferation via different classes of cell surface receptors that include growth factor receptor tyrosine kinase receptors and G-protein-coupled receptors (GPCRs). 1 This involves stimulation of the p42/p44 mitogen-activated protein kinase (p42/p44 MAPK) cascade (1). For many years it has been known that certain GPCR agonists can function as co-mitogens with growth factors to stimulate DNA synthesis. However, the molecular mechanism for this interaction has not been fully defined. It is known that both growth factors and GPCR agonists stimulate the tyrosine phosphorylation of Shc (SH2-containing protein) and the sequential activation of Grb-2-mSos (son of sevenless), Ras, c-Raf, MEK1, and p42/p44 MAPK. GPCR agonists also activate non-receptor tyrosine kinases (e.g. c-Src), which function as intermediates between G␤␥ subunits and Ras-dependent p42/p44 MAPK activation (2)(3)(4).
A significant advance in our understanding of co-mitogenicity comes from recent studies showing that certain growth factors can use classical GPCR-mediated signaling pathways to stimulate p42/p44 MAPK in mammalian cells. For instance, the insulin like growth factor-1 (IGF-1) receptor utilizes the G-protein, G i , to stimulate activation of p42/p44 MAPK in fibroblasts (3). This was established using pertussis toxin (which uncouples G-protein-coupled receptors from G i/o ) and the C-terminal domain of ␤-adrenergic kinase (which sequesters G␤␥ subunits), both of which reduced the IGF-1-dependent activation of p42/p44 MAPK. These agents also reduced fibroblast growth factor-dependent activation of p42/p44 MAPK in fibroblasts and promoted differentiation (4). This model has recently been supported by Hallak et al. (5) who demonstrated that the IGF-1 receptor forms a complex with G i ␣␤␥ and that IGF-1 stimulates release of the G-protein ␤␥ subunits, which can initiate activation of the p42/p44 MAPK pathway. There are also other studies that have shown that insulin receptors can interact with G i (6 -8).
In addition to these findings, GRK2 and ␤-arrestin II are implicated in the regulation of IGF-1 and ␤-adrenergic recep-tor-stimulated endocytic signaling and activation of the p42/ p44 MAPK pathway (9 -11). GRK2 is activated in an agonistand G␤␥ subunit-dependent manner. ␤-arrestin II is a clathrin adaptor protein that is recruited to ligand-bound GPCRs that have been phosphorylated by GRK2, and this promotes dynamin II-mediated endocytosis of receptor signal complexes containing Raf-1-MEK1 for subsequent activation of p42/p44 MAPK. Recent studies by Dalle et al. (12) have shown that IGF-1 promotes the binding of ␤-arrestin I to the IGF-1 receptor in adipocytes. Rakhit et al. (13) have also shown that NGF stimulates ␤-arrestin I binding to a GRK2-Trk A receptor complex in PC 12 cells. In this case, GRK2/␤-arrestin I appears to initiate endocytosis of Trk A signal complexes, which leads to the activation of p42/p44 MAPK in response to NGF. In recent studies, we reported that the platelet-derived growth factor (PDGF) receptor stimulates c-Src and p42/p44 MAPK via a pertussis toxin-sensitive pathway in airway smooth muscle cells (14,15). We suggested that G i might function to recruit c-Src close to the PDGF receptor tyrosine kinase for activation. PDGF also stimulates a pertussis toxin-sensitive tyrosine phosphorylation of the Grb-2-associated binding protein, Gab1. This promotes the binding of tyrosine-phosphorylated phosphoinositide 3-kinase (PI3K1a) to Gab1, which is required for dynamin II-mediated endocytic signaling to the p42/p44 MAPK pathway in response to PDGF (15).
The realization that growth factor receptors can use G-proteins to signal to p42/p44 MAPK provides a mechanism that begins to explain the co-mitogenic properties of some GPCR agonists. Thus, GPCR agonists might provide G-protein ␣ and ␤␥ subunits for use by growth factor receptors to produce stronger activation of the p42/p44 MAPK pathway and therefore DNA synthesis. Certain G-protein-coupled receptor agonists have also been shown to stimulate the tyrosine phosphorylation and transactivation of growth factor receptors. The subsequently phosphorylated sites on the receptor act as acceptors for the recruitment of signaling proteins, such as Grb-2, PLC␥ (phospholipase C ␥), and PI3K, and complex assembly to elicit mitogenic responses. For instance, lysophosphatidic acid has been shown to transactivate the EGF receptor and p185 neu to stimulate p42/p44 MAPK activation in Cos-7 cells (16), while angiotensin II promotes PDGF receptor transactivation in vascular smooth muscle (17). However, this model does not adequately explain the co-mitogenicity of certain GPCR agonists with growth factors, since the transactivation of the growth factor receptor by GPCR agonist appears to be growth factor-independent.
In the current paper we have used HEK 293 cells transfected with PDGF␤ receptors to investigate the molecular mechanism for the involvement of G i in regulating the transmission of signals from this receptor. We have focused on the possibility that the PDGF␤ receptor might exist in a functional signaling complex with GPCRs. We have also tested this model by looking at the specific interaction between the PDGF␤ receptor and EDG1 (endothelial differentiation gene 1), whose natural agonist is sphingosine 1-phosphate (S1P). To date, five closely related GPCRs of the EDG family (EDG1, EDG3, EDG5/ AGR16/H218, EDG6, and EDG8/nrg-1) have been identified as high affinity S1P receptors (18 -21). The rationale for looking at interaction between the PDGF␤ receptor and EDG1 was 2-fold. First, S1P is co-mitogenic with PDGF in airway smooth muscle cells, which predominantly express EDG1 (22), and second Hobson et al. (23) showed that PDGF-stimulated cell motility is EDG1-dependent. While these authors reported the existence of cross-talk regulation between these different receptor classes, they did not establish the molecular basis of this interaction.
Here we show that the PDGF␤ receptor is tethered to GPCR(s) in a complex. The tethered receptor complex provides a platform on which receptor tyrosine kinase and GPCR signals can be integrated to produce more efficient regulation of downstream effector pathways. This model provides a molecular dynamic that may explain the co-mitogenic effect of GPCR agonists on PDGF-stimulated DNA synthesis.
Cell Culture-HEK 293 cells were maintained in minimum essential medium (MEM) containing 20% (v/v) fetal calf serum. These cells were placed in MEM for 24 h before experimentation. Cultured airway smooth muscle cells were prepared and maintained as described previously (14,15).
Transfection-HEK 293 cells were transiently transfected with PDGF␤ receptor, Myc-tagged EDG1, EDG1, and/or ␤-arrestin I pcDNA3.1 plasmid constructs and/or pRK5-GRK2 plasmid construct. Cells at 90% confluence were placed in MEM containing 2% fetal calf serum and transfected with 2 g of plasmid construct following complex formation with LipofectAMINE TM 2000, according to the manufacturer's instructions. The cDNA-containing medium was then removed following incubation for 24 h at 37°C, and the cells were incubated for a further 24 h in serum-free medium prior to agonist addition.
p42/p44 MAPK Assays-The phosphorylated forms of p42/p44 MAPK were detected by Western blotting cell lysates with anti-phospho-p42/p44 MAPK antibody. An anti-p42 MAPK antibody was also used for Western blotting to establish equal loading of protein in each sample.
Blotting-Immunoblotting was performed as described previously (14,15). Immunoreactive proteins were visualized using the enhanced chemiluminescence detection kit.
Sphingosine 1-Phosphate Formation-Cells (airway smooth muscle cells or PDGF␤ receptor-transfected HEK 293) were quiesced overnight before preincubation for 1 h in serum-free medium supplemented with fatty acid-free bovine serum albumin (2 mg/ml). Incubations, in the presence and absence of PDGF (10 ng/ml) were initiated by replacing the medium with that containing [ 3 H]sphingosine (2.22 ϫ 10 5 dpm/ well). After 3 min, the medium was removed, and the cells harvested into 0.5 ml ice-cold methanol. Lipids were extracted by vortexing in the presence of an equal volume of chloroform. [ 3 H]S1P was isolated by thin layer chromatography on silica gel G60 plates developed with chloroform:methanol:acetic acid:H 2 O (25:10:1:2, v/v). A [ 32 P]S1P standard was run in parallel. [ 3 H]S1P, which co-migrated with the S1P standard, was measured by scintillation counting of appropriate sections of silica that were excised from the plate.

PDGF␤ Receptor Signaling Is G-protein-mediated in HEK 293
Cells-Vector transfected HEK 293 cells exhibit a PDGF␤ receptor null background. Thus, endogenous PDGF␤ receptor was not detected on Western blots probed with anti-PDGF␤ receptor antibodies (Fig. 1a), and PDGF did not stimulate the p42/p44 MAPK pathway in vector-transfected HEK 293 cells (Fig. 1b). This provides an ideal cell model system for transfection with recombinant PDGF␤ receptors to study the involvement of GPCR(s) in regulating PDGF␤ receptor signaling. Overexpression of recombinant PDGF␤ receptor (R m ϭ 180 kDa) in PDGF␤ receptor-transfected cells was confirmed by Western blotting with anti-PDGF␤ receptor antibody (Fig. 1a). Moreover, PDGF induced a robust activation of p42/p44 MAPK in these cells, a response that was abolished by the PDGF receptor kinase inhibitor, tyrphostin AG 1296 (Fig. 1b). In general we found that the extent to which p42 MAPK was phosphorylated in response to PDGF exceeded that of p44 MAPK, which on some Western blots was barely detectable.
Several lines of evidence suggest that the PDGF␤ receptor uses a classical GPCR signaling pathway to stimulate p42/p44 MAPK in PDGF␤ receptor-transfected HEK 293 cells. Thus, pretreatment of cells with pertussis toxin, which uncouples GPCRs from G i/o, substantially reduced the PDGF-dependent activation of p42/p44 MAPK (Fig. 2a). The specificity of action of pertussis toxin was established by results showing that the stimulation of p42/p44 MAPK by the G i/o receptor agonist, thrombin, was also reduced by pretreating cells with this toxin (Fig. 2b), while the response to EGF, which does not use G i/o to signal, was unaffected (Fig. 2b). Pertussis toxin did not increase basal cAMP (data not shown), which in some cases can block (via protein kinase A) growth factor-stimulated p42/p44 MAPK activation. Because pertussis toxin is an inhibitor of G i ␣ signaling, we tested the effect of transfecting cells with recombinant G i ␣ on the PDGF-dependent stimulation of the p42/p44 MAPK pathway. Fig. 2c (upper panels) shows that the overexpression of recombinant G i ␣ markedly increased the stimulation of p42/p44 MAPK by PDGF. This reflects more efficient downstream signaling from the PDGF␤ receptor because recombinant G i ␣ was without effect on PDGF␤ receptor expression (Fig. 2d) and reduced PDGF␤ receptor auto-phosphorylation (Fig. 2d). Studies to delineate the mechanism by which G i ␣ potentiates the PDGF-dependent activation of p42/p44 MAPK were performed. These results show that both endogenous and recombinant G i ␣ were basally tyrosine-phosphorylated in PDGF␤-transfected cells and that PDGF induced an increase in their tyrosine phosphorylation (Fig. 2c, lower panels). Tyrosine-phosphorylated recombinant G i ␣ migrated on SDS-polyacrylamide gel electrophoresis as a 40-kDa protein. Two lines of evidence were used to identify G i ␣. First, endogenous G i ␣ co-migrated with recombinant protein on SDS-polyacrylamide gel electrophoresis. Second, tyrosine-phosphorylated endogenous G i ␣ was immunoprecipitated by anti-G i ␣ antibodies (Fig.  2e). These results show that only a small fraction of the recombinant G i ␣ is tyrosine-phosphorylated in response to PDGF but that this is sufficient to increase p42/p44 MAPK activation in response to this growth factor. The pretreatment of cells with the PDGF receptor kinase inhibitor, tyrphostin AG 1296, reduced the PDGF-stimulated tyrosine phosphorylation of endogenous G i ␣ (Fig. 2f), suggesting that the PDGF␤ receptor kinase might catalyze the phosphorylation of this G-protein. The concentration range of tyrphostin AG 1296 that was inhibitory (1-5 M) was the same as that which inhibits PDGF receptor kinase activity and PDGF-stimulated p42/p44 MAPK activation (14).
Interaction of the PDGF␤ Receptor and EDG1 Receptors-These results are compatible with a model in which an endogenous ligand-bound GPCR(s) releases G i ␣ from the ␣␤␥ heterotrimeric G-protein complex for subsequent tyrosine phosphorylation by the PDGF␤ receptor kinase. The efficiency of this process could be greatly increased by proximity-induced effects if the PDGF␤ receptor was tethered in a complex with the GPCR. To test the model, we investigated whether the recombinant forms of the PDGF␤ receptor and the GPCR EDG1 (whose natural agonist is S1P) form a functional signaling complex in HEK 293 cells.
The overexpression of recombinant EDG1 was demonstrated by a more robust stimulation of p42/p44 MAPK in response to S1P compared with that observed in vector-transfected cells (Fig. 3a). The evidence that the PDGF␤ receptors and EDG1 form a complex was obtained from results showing that PDGF␤ receptor and Myc-tagged EDG1 were co-immunoprecipitated from cell lysates using anti-Myc antibodies (Fig. 3b) or anti-PDGF␤ receptor antibodies (Fig. 3c). Neither PDGF nor S1P stimulated an increase in the amount of each receptor in the complex. Controls were included to show that the PDGF␤ receptor was not co-immunoprecipitated using anti-Myc tag antibodies when PDGF␤ receptor plasmid construct was omitted from the co-transfection (Fig. 3b), while Myc-tagged EDG1 was not co-immunoprecipitated using anti-PDGF␤ receptor antibodies when Myc-tagged EDG1 plasmid construct was omitted from the co-transfection (Fig. 3c). Consistent with a model in which GPCR(s) release G i ␣ for tyrosine phosphorylation by the PDGF␤ receptor kinase, we found that PDGF induced a stronger tyrosine phosphorylation of G i ␣ (Fig. 2c, lower panels) and a more robust activation of p42/p44 MAPK (Fig. 2, a and c, upper panels, Fig. 3d) in cells co-transfected with PDGF␤ receptor and EDG1 versus PDGF␤ receptor alone. This response remained sensitive to pertussis toxin (Fig. 2a). We have excluded the possibility that the improved effect of PDGF was due to an EDG1-induced increase in the expression (Fig. 1a) or tyrosine phosphorylation (data not shown) of recombinant PDGF␤ receptor. Furthermore, PDGF did not stimulate p42/p44 MAPK in cells transfected with EDG1 alone (data not shown). Thus, our results suggest that the formation of the EDG1-PDGF␤ receptor improves the efficiency of PDGF␤ receptor signaling to p42/p44 MAPK. It is also important to note that PDGF did not increase S1P synthesis/ release in these cells (Fig. 4), thereby excluding the possibility that S1P might function as an autocrine to stimulate p42/p44 MAPK activation by binding to recombinant EDG1 receptors. However, we do not exclude this as a possible mechanism in other cell types. As a positive control, we have shown that PDGF stimulated S1P formation in airway smooth muscle cells, measured after 3 min cell stimulation (Fig. 4). Furthermore, the pretreatment with DL-threo-dihydrosphingosine at a concentration (1 M) that inhibits sphingosine kinase (23), an enzyme that catalyzes S1P formation, did not reduce PDGFstimulated p42/p44 MAPK activation in PDGF␤ receptor/ EDG1-transfected cells or exogenous S1P stimulation of p42/ p44 MAPK in cells overexpressing EDG1 alone (Fig. 5). We confirmed that the concentration of DL-threo-dihydrosphingosine was effective at inhibiting sphingosine kinase by showing that this inhibitor prevented PDGF-stimulated S1P formation in airway smooth muscle cells (Fig. 4).
GRK2/␤-Arrestin I Signaling-Several growth factors can also use G-protein ␤␥ subunits to initiate a GRK2-and ␤-arrestin-mediated stimulation of the p42/p44 MAPK pathway (9 -11). The involvement of GRK2 and ␤-arrestin I might be explained by the binding of these proteins to a growth factor receptor-GPCR complex. Several lines of evidence support this model. Using cells transfected with recombinant forms of PDGF␤ receptor and GRK2 (R m ϭ 85 kDa), we show that both proteins were co-immunoprecipitated from cell lysates using anti-PDGF␤ receptor antibodies (Fig. 6a), suggesting that these proteins form a complex. The stimulation of cells with PDGF or S1P did not increase the amount of GRK2 associated with the PDGF␤ receptor (Fig. 6a). Less than 1% of the total GRK2 present in the anti-PDGF␤ receptor immunoprecipitates was endogenous GRK2. This was evidenced by residual binding of GRK2 in anti-PDGF␤ receptor immunoprecipitates from cells where recombinant GRK2 plasmid construct had been omitted from the co-transfection (Fig. 6b). As with recombinant GRK2, the stimulation of cells with PDGF did not increase the association between endogenous GRK2 and the PDGF␤ receptor. Using cells transfected with recombinant forms of PDGF␤ receptor and ␤-arrestin I (R m ϭ 55 kDa), we show that ␤-arrestin I can also be co-immunoprecipitated from cell lysates using anti-PDGF␤ receptor antibodies (Fig. 6c). In common with GRK2, we found that stimulation of cells with PDGF did not induce further association of recombinant ␤-arrestin I with the PDGF␤ receptor (Fig. 6c). We also detected residual endogenous ␤-arrestin I (Ͻ1% of recombinant ␤-arrestin I) in the anti-PDGF immunoprecipitates. Treatment of cells with PDGF did not increase ␤-arrestin I association with the PDGF␤ receptor (Fig. 6c).
Using cells transfected with Myc-tagged EDG1 and recombinant GRK2, we show that both proteins can be co-immunoprecipitated from cell lysates using anti-Myc tag antibodies (Fig.  6d), suggesting that these proteins also form a complex. The stimulation of cells with S1P did not increase the amount of recombinant GRK2 associated with Myc-tagged EDG1 (Fig.  6d). However, the treatment of cells with S1P stimulated an increase in the association of recombinant ␤-arrestin I with Myc-tagged EDG1 (Fig. 6e). The presence of endogenous GRK2 and ␤-arrestin I in anti-Myc-tagged immunoprecipitates could not be detected, possibly as the amounts of these proteins may be below the level of detection (Fig. 6, d and e). DISCUSSION Certain growth factor receptor tyrosine kinases can use classical GPCR-mediated signaling pathways to stimulate p42/p44 MAPK in mammalian cells. It is therefore possible that these growth factor receptors might exist in functional signaling complexes with GPCRs. Here, we provide evidence to show that the PDGF␤ receptor is tethered to an endogenous GPCR(s) and to recombinant EDG1 in HEK 293 cells. The tethered receptor complex provides a platform on which receptor tyrosine kinase and GPCR signals can be integrated to produce more efficient regulation of the p42/p44 MAPK pathway. This involves (i) tyrosine phosphorylation of G i ␣, possibly catalyzed by the PDGF␤ receptor kinase and (ii) the association of GRK2/␤arrestin I with the PDGF␤ receptor. There are reports that show that G i ␣ is also a substrate for tyrosine phosphorylation by another growth factor receptor, the insulin receptor in vitro (6,24).
Several lines of evidence suggest that the PDGF␤ receptor can use G i ␣ to signal to the p42/p44 MAPK pathway in HEK 293 cells. First, pertussis toxin reduced the PDGF stimulation of p42/p44 MAPK. Second, transfection of cells with recombinant G i ␣ increased the activation of p42/p44 MAPK by PDGF. Third, PDGF stimulated the tyrosine phosphorylation of G i ␣. These results are compatible with a model in which an unidentified ligand-bound GPCR(s) releases G i ␣ from the ␣␤␥ heterotrimeric G-protein complex for tyrosine phosphorylation by the PDGF␤ receptor kinase. Indeed, the inhibitory effect of pertussis toxin on the PDGF stimulation of p42/p44 MAPK is consistent with the established effect of this toxin on GPCR signaling. This is to prevent ligand-bound GPCRs from activating inhibitory G-proteins. Therefore, the mode of action of pertussis toxin suggests involvement of a GPCR, particularly as there is no evidence that the PDGF␤ receptor can directly activate G-protein ␣␤␥ complexes. The supply of G i ␣ to the PDGF␤ receptor kinase suggests that the GPCR is either tethered to or is in close association with the PDGF␤ receptor. The results showing the presence of Myc-tagged EDG1-PDGF␤ receptor complexes in transfected HEK 293 cells suggest that the former possibility is more likely.
Evidence that G i ␣ is involved in the PDGF-dependent activation of p42/p44 MAPK was based upon the fact that transfection of cells with recombinant G i ␣ potentiated p42/p44 MAPK activation by this growth factor. Moreover, this could be directly correlated with increased tyrosine phosphorylation of G i ␣. We also noted that recombinant and endogenous G i ␣ were basally tyrosine-phosphorylated in PDGF␤ receptor-transfected HEK 293 cells. This is not unexpected, as there is substantial PDGF␤ receptor kinase activity in unstimulated cells (indicated by the basal tyrosine phosphorylation of the PDGF␤ receptor) to support basal tyrosine phosphorylation of G i ␣. However, it is important to note that the basally tyrosinephosphorylated recombinant G i ␣ does not induce activation of p42/p44 MAPK on its own. Therefore, the potentiating effect of recombinant G i ␣ on the PDGF-stimulated p42/p44 MAPK activation may require the PDGF-induced recruitment to tyrosine-phosphorylated G i ␣ of other intermediates involved in the p42/p44 MAPK pathway. We also found that overexpression of recombinant G i ␣ reduced PDGF receptor auto-phosphorylation consistent with substrate competition between specific phosphotyrosine sites on the PDGF␤ receptor and G i ␣ for the kinase. The fact that overexpression of recombinant G i ␣ reduced PDGF␤ receptor tyrosine phosphorylation provides further evidence to support our proposal that G i ␣ interacts with the PDGF␤ receptor kinase. It also is significant that the stimulatory effect of overexpressing recombinant G i ␣ on the PDGFdependent activation of p42/p44 MAPK occurs in the face of reduced PDGF␤ receptor autophosphorylation. This suggests that the PDGF␤ receptor uses the G i signaling pathway as a major route for regulating p42/p44 MAPK in HEK 293 cells. Indeed, this is supported by other results reported here, which show that inactivation of endogenous G i ␣ signaling by pertus- sis toxin reduced the PDGF stimulation of p42/p44 MAPK by ϳ90% in these cells.
A major finding confirming our proposal was obtained by data showing that the PDGF␤ receptor can exist in a tethered complex with Myc-tagged EDG1 in cells co-transfected with these receptors. There is one other example of growth factor receptor tethering to a GPCR. This was shown for the insulin and ␤-adrenergic receptors in adipocytes (25,26). These authors showed that insulin stimulates the phosphorylation of the ␤-adrenergic receptor on Tyr-350 and this promotes the binding of the insulin receptor via Grb-2, which serves to tether the two receptors. Moreover, the integrity of the insulin receptor-Grb-2-␤-adrenergic receptor complex is critical for ␤-adrenergic agonist amplification of insulin-dependent activation of p42/p44 MAPK. Therefore, we suggest that the purpose of tethering the PDGF␤ receptor with a GPCR is to improve the efficiency of PDGF␤ receptor signaling, possibly via proximityinduced effects. This is borne out experimentally where we show that PDGF induced a stronger tyrosine phosphorylation of G i ␣ and a more robust activation of p42/p44 MAPK in cells transfected with both PDGF␤ receptor and EDG1 compared with PDGF␤ receptor alone. We consider these studies to be very important in identifying novel signaling platforms. It is also possible that endogenous PDGF receptors form complexes with endogenous GPCRs in other cells types, such as airway smooth muscle cells. This is based upon data showing that pertussis toxin abolished PDGF-induced Gab1 tyrosine phosphorylation, PI3K-Gab1 association, and dynamin II binding to Grb-2 (15) and reduced the PDGF-dependent activation of p42/ p44 MAPK by ϳ50% (14,15).
The effect of recombinant EDG1 on PDGF␤ receptor signaling was not a consequence of its binding S1P released from cells in response to PDGF but is instead probably due to constitutive activation of EDG1. Indeed, it is a common feature of recombinant GPCRs that they are often constitutively activated when expressed in cells. Moreover, others have shown this to be the case for EDG1, EDG3, and EDG5. These authors showed that when these receptors are overexpressed they form active complexes with G-proteins ␣ subunits (27,28). We have excluded a role for released S1P based on the following. First, we were unable to detect S1P formation in response to PDGF in PDGF␤ receptor-transfected HEK 293 cells. Second, the sphingosine kinase inhibitor, DL-threo-dihydrosphingosine, had no effect on PDGF-stimulated p42/p44 MAPK activation in PDGF␤ receptor/EDG1 co-transfected cells. However, we do not exclude the possibility that released S1P could act via tethered EDG1-PDGF␤ receptors in other cells types.
We also found that GRK2/␤-arrestin I can associate with the PDGF␤-receptor. The stimulation of cells with PDGF did not increase the amount of either protein in the receptor complex, suggesting that the PDGF␤ receptor-GRK2-␤-arrestin I complex is preformed. These data again support the possibility that GRK2 and ␤-arrestin I are bound to an endogenous ligand-GPCR(s) that is tethered with the PDGF␤ receptor. This is a plausible explanation given that the PDGF␤ receptor does not have binding sites for GRK2 and ␤-arrestin I. We surmise that the putative GPCR is ligand-bound based upon the fact that GRK2 and ␤-arrestin I association with GPCRs is dependent upon free G␤␥ activation released upon ligand binding to the GPCR. In addition, the finding that PDGF did not stimulate further association of these proteins with the PDGF␤ receptor suggest to us that this growth factor receptor cannot itself induce G␣␤␥ dissociation and that PDGF does not stimulate release of the putative GPCR ligand from these cells. GRK2 and ␤-arrestin I play an important role in regulating endocytosis of GPCR signal complexes, which is required for activa-tion of p42/p44 MAPK. In this regard, we have previously shown that endocytosis of PDGF receptor signal complexes is also required for p42/p44 MAPK activation and that this is G-protein-regulated in airway smooth muscle cells (14). It is therefore possible that endocytosis of PDGF␤ receptor-signal complexes may be initiated by GRK2/␤-arrestin I that have been recruited to the PDGF␤ receptor by its tethering to GPCR(s). This possibility is currently under investigation in our laboratory.
The model proposed here for the interaction of the PDGF␤ receptor with GRK2/␤-arrestin I appears to be slightly different from that reported for two other growth factor receptors where binding of ␤-arrestin I is apparently growth factor-dependent. First, Dalle et al. (12) showed that IGF-1 uses G␤␥ subunits to stimulate the binding of ␤-arrestin I to the IGF-1 receptor in adipocytes. These authors also showed that IGF-1 induced G␤␥ subunits dissociation and stimulated binding of G i ␣ to the IGF-1 receptor. These workers therefore proposed that the IGF-1 receptor uses both G-protein subunits to promote signal transduction. Second, we have shown that the NGF-dependent activation of p42/p44 MAPK can be potentiated in PC 12 cells transfected with GRK2 or ␤-arrestin I (13). GRK2 is preassociated with the Trk A receptor, while NGF stimulates the pertussis toxin-sensitive binding of ␤-arrestin I to the Trk A receptor-GRK2 complex. However, in the light of the current findings in this article, there is a need to establish whether these growth factor receptors are also tethered to GPCRs and whether they elicit release of the corresponding GPCR ligand, which may in turn trigger binding of ␤-arrestin I to the GPCR within the putative complex with growth factor receptor.
We also found that GRK2 and ␤-arrestin I are associated with recombinant EDG1, thereby providing further evidence that this receptor may be constitutively active when overexpressed. However, recombinant EDG1 appears to be capable of further activation based upon the fact that S1P induces an increase in the binding of ␤-arrestin I and can stimulate activation of p42/p44 MAPK. This is in line with current thinking where it is proposed that receptors can exist in different receptor-G-protein conformations and that ligands induce a more productive conformation in terms of efficacy. Thus, S1P may convert recombinant EDG1 from a partially constitutively activated conformation to one that is capable of stimulating further G␤␥ release and binding of ␤-arrestin I.
In conclusion, growth factor receptor-GPCR complexes provide a platform for integrating signals from these different receptor classes. This represents a mechanistic model that may account for the co-mitogenic effect of GPCR agonists with growth factors. More specifically, our proposed model provides a mechanism that may account for the recent findings of Hobson et al. (23) who showed that PDGF-stimulated cell motility is EDG1-dependent. Thus, S1P released from cells (not HEK 293 cells) in response to PDGF could act back on EDG1-PDGF␤ receptor complexes to induce more efficient downstream stimulation of effector pathways in response to PDGF. This might be specific to certain cell types where S1P functions as an autocrine with PDGF.
The findings in the current study break the conventional paradigm for growth factor receptor signaling and strengthen an emerging model that such receptors can use GPCR-mediated pathways to stimulate p42/p44 MAPK. Future goals will be to identify the structural determinants that govern the interaction between the PDGF␤ receptor and GPCRs.