Src-mediated Tyrosine Phosphorylation of Dynamin Is Required for β2-Adrenergic Receptor Internalization and Mitogen-activated Protein Kinase Signaling*

Some forms of G protein-coupled receptor signaling, such as activation of mitogen-activated protein kinase cascade as well as resensitization of receptors after hormone-induced desensitization, require receptor internalization via dynamin-dependent clathrin-coated pit mechanisms. Here we demonstrate that activation of β2-adrenergic receptors (β2-ARs) leads to c-Src-mediated tyrosine phosphorylation of dynamin, which is required for receptor internalization. Two tyrosine residues, Tyr231 and Tyr597, are identified as the major phosphorylation sites. Mutation of these residues to phenylalanine dramatically decreases the c-Src-mediated phosphorylation of dynamin following β2-AR stimulation. Moreover, expression of Y231F/Y597F dynamin inhibits β2-AR internalization and the isoproterenol-stimulated mitogen-activated protein kinase activation. Thus, agonist-induced, c-Src-mediated tyrosine phosphorylation of dynamin is essential for its function in clathrin mediated G protein-coupled receptor endocytosis.

Agonist stimulation of cell surface receptors results in rapid attenuation of receptor responsiveness, a process termed desensitization. In the case of G protein-coupled receptors (GPCRs), 1 this process is initiated by phosphorylation of agonist-occupied receptors by the G protein-coupled receptor kinase family (1) and the subsequent formation of high affinity complexes with cytosolic proteins termed ␤-arrestins (2). ␤-Arrestin-bound, desensitized GPCRs then undergo ␤-arrestin-me-diated targeting to clathrin-coated pits (3,4), wherein they are sequestered, internalized, and ultimately either recycled to the cell surface or targeted for degradation. Interestingly, the process of GPCR internalization also plays a critical, albeit poorly understood, role in some aspects of GPCR signal transduction, e.g. MAP kinase activation (5,6).
The process of cell surface receptor internalization (also termed endocytosis) is dependent on the invagination and fission of clathrin-coated vesicles from the plasma membrane into the cytosol. Endocytosis of many GPCRs and receptor tyrosine kinases (RTKs) requires the GTPase activity of dynamin. Receptor stimulation leads to recruitment of cytosolic dynamin to coated pits where it induces constriction of the pits and fission of vesicles (7,8). In vitro, dynamin is activated by a variety of mechanisms including binding to microtubules (9), phospholipids (10), and several proteins containing SH3 domains (11). Collectively, these molecules are thought to induce a conformational change in dynamin, inducing homo-oligomerization and subsequent increase in GTPase activity (12). Recently, dynamin was shown to contain a GTPase effector domain that interacts with its N-terminal GTPase domain to stimulate GTP hydrolysis (13). GTPase defective dynamin mutants (e.g. K44A) specifically block endocytic coated vesicle formation and agonist-mediated internalization of GPCRs and RTKs (14,15).
Accumulating evidence suggests that tyrosine protein phosphorylation is critical for the internalization of cell surface receptors. Exposure of cells to tyrosine kinase inhibitors profoundly attenuates cross-linking-induced internalization of B cell receptors (16) and hormone-induced internalization of tyrosine kinase growth factor receptors (17). Recently, overexpression of the nonreceptor tyrosine kinase c-Src was found to cause an increase in the internalization rate constant of the epidermal growth factor (EGF) receptor following EGF treatment (18). Although these results suggest involvement of tyrosine kinases in the process of receptor internalization, their role(s) in this process is not well defined. Therefore, we tested the hypothesis that tyrosine kinase activity might regulate endocytosis by acting on accessory molecules important for receptor internalization such as dynamin. We find not only that Src-mediated tyrosine phosphorylation of dynamin is required for ␤ 2 -AR internalization but that this phosphorylation event itself is regulated by receptor stimulation.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfection-HEK293 cells were maintained and transfected precisely as described (5,6). All assays were performed 40 h after transfection. HEK293 cells stably overexpressing dynamin were generated as described (19).
Sequestration of ␤ 2 -ARs-The ␤ 2 -AR sequestration was quantified as loss of cell surface receptors determined by immunofluorescence flow cytometry as described (5).
Plasmid Construction and Site-specific Mutagenesis-To generate single tyrosine mutants (Y231F and Y597F), tyrosine 231 or tyrosine 597 of rat dynamin I was mutated to phenylalanine by overlapping polymerase chain reaction (UAC(Y) 3 UUC(F)) (20). The double tyrosine mutant (Y231F/Y597F) dynamin was constructed by recombination of the two single tyrosine mutant constructs.
GST-Grb2 Fusion Proteins-Isolation of total cellular dynamin for analysis of tyrosine phosphorylation was performed by affinity purification with full-length GST-Grb2 fusion protein according to the method of Gout et al. (11). The cDNA encoding GST-Grb2 fusion protein was generously provided by A. R. Saltiel. GST fusion proteins were prepared as glutathione-agarose conjugates as described previously (6,21). Complexed proteins were dissociated from agarose beads with SDS-polyacrylamide gel electrophoresis sample buffer.
Protein Cross-linking-For the detection of transient, agonist-promoted association of dynamin and c-Src, covalent protein cross-linking with Dithiobis(succinimidylpropionate) (DSP, Pierce) was employed. Cells were stimulated in 4.6 ml of phosphate-buffered saline containing 10 mM Hepes, pH 7.4. Stimulation was terminated by the addition of 0.4 ml of 2.5 mM DSP in Me 2 SO, and plates were rocked for 30 min at room temperature. Reactions were quenched by the addition of 0.1 ml of 1 M Tris, pH 7.5, followed by two washes with ice-cold phosphate-buffered saline/Hepes to remove unreacted DSP. Cell pellets were lysed in RIPA buffer (5,6) prior to immunoprecipitation of c-Src-dynamin complexes.
Immunoprecipitation and Immunoblotting-Cell lysates in RIPA buffer were incubated with the specified antibodies and protein A/G-Sepharose beads for 2 h at 4°C. Rabbit polyclonal anti-Src antibody (Src-2, Santa Cruz Biotech. Inc.) was used for c-Src immunoprecipitation after protein cross-linking. Dynamin proteins were immunoprecipitated without cross-linking using either mouse monoclonal anti-dynamin antibody (Hudy-1; Upstate Biotech. Inc.) or rabbit polyclonal anti-dynamin I-specific antibody (kind gift of S. L. Schmid). Immunoprecipitates or GST-Grb2 fusion complexed proteins were resolved on acrylamide gels, transferred to nitrocellulose filters, immunoblotted with anti-dynamin or anti-phosphotyrosine antibodies (RC20H or PY20H, Transduction Laboratories), and visualized by enhanced chemiluminescence (Amersham Pharmacia Biotech).
Metabolic Labeling and Microsequencing-HEK293 cells stably expressing dynamin I were metabolically labeled with [ 32 P]orthophosphate as described (19). Dynamin was purified using GST-Grb2 binding as described above. Partially purified dynamin was resolved by SDSpolyacrylamide gel electrophoresis, and the dynamin band was excised from the gel and subsequently used for microsequencing after tryptic digestion (22).
MAP Kinase Assay-HA-Erk2 was immunoprecipitated from HEK293 lysates using 6 g of 12CA5 antibody. Immunoprecipitates were washed three times with RIPA buffer and twice with kinase buffer (23). In vitro kinase reactions using myelin basic protein (MBP) as substrate were performed as described (23). Labeled MBP was quantified using a Storm PhosphorImager (Molecular Dynamics).

RESULTS AND DISCUSSION
Spatial distribution and trafficking of receptors control their function. For example, mitogenic signaling emanating from certain GPCRs and RTKs is dependent on receptor internalization (5, 6, 24 -26). To determine whether tyrosine kinase activity is required for G protein-coupled receptor endocytosis, isoproterenol-induced internalization of the ␤ 2 -AR was examined in HEK293 cells pretreated with tyrosine kinase inhibitors. Fig. 1A shows that pretreatment of cells with the generic tyrosine kinase inhibitor herbimycin A reduced isoproterenol-mediated internalization of the ␤ 2 -AR by 60%. More dramatically, acute exposure of cells to the specific c-Src kinase inhibitor PP1 reduced the isoproterenol-induced ␤ 2 -AR sequestration by approximately 80% (Fig. 1A). As shown in Fig. 1B, expressing either a c-Src kinase inhibitor (Csk) or a kinase defective, dominant inhibitory form of c-Src (K298M) impaired the isoproterenol-mediated internalization of the ␤ 2 -AR. Conversely, expression of an activated form of Src (v-Src) increased agonistdependent sequestration of the ␤ 2 -AR (Fig. 1B). These results suggest that Src tyrosine kinase activity is required for agonistdependent internalization of the ␤ 2 -AR.
A direct interaction between c-Src and dynamin in neuronal cells has been described (27). Because c-Src kinase activity was required for the agonist-mediated internalization of ␤ 2 -AR, we examined whether dynamin is a phosphotyrosyl protein and a substrate for c-Src. Cellular dynamin was isolated using a GST-Grb2 affinity purification following agonist exposure. As shown in Fig. 2A, isoproterenol stimulation resulted in 2-2.5fold increase in tyrosine phosphorylation of dynamin (upper immunoblot), which was maximal within 2 min of agonist exposure. These results demonstrate that ␤ 2 -AR activation induces the rapid tyrosine phosphorylation of dynamin.
Next, we examined whether c-Src is the kinase responsible for the tyrosine phosphorylation of dynamin following activa-tion of the ␤ 2 -AR. We expressed the c-Src inhibitors, Csk and K298M c-Src, or the constitutively active v-Src and examined their effect on isoproterenol-induced tyrosine phosphorylation of dynamin in HEK293 cells stably expressing dynamin I. Fig.  2B shows that expression of Csk or K298M c-Src inhibited the agonist-induced tyrosine phosphorylation of dynamin, whereas expression of activated Src (v-Src) further enhanced the level of tyrosine phosphorylation of dynamin (upper immunoblot). Thus, agonist stimulation resulted in a net increase in the phosphotyrosine content of the cellular dynamin pool, which could be reversed by inhibition of Src activity. These data mirror the effects of these reagents on ␤ 2 -AR sequestration (Fig. 1B), suggesting that c-Src-regulated internalization of the ␤ 2 -AR may be mediated by the c-Src-induced tyrosine phosphorylation of dynamin.
Foster-Barber and Bishop (27) have reported that the purified SH3 domain of c-Src binds dynamin in vitro and that dynamin co-precipitates with c-Src from PC12 cell lysates. These authors did not describe tyrosine phosphorylation of dynamin or modulation of c-Src-dynamin complex formation following receptor activation. To test the hypothesis that the agonist-regulated association of dynamin and c-Src might account for the agonist-promoted increase in dynamin tyrosine phosphorylation, we examined the ability of c-Src to form a complex with dynamin following ␤ 2 -AR activation. Protein cross-linking was used to stabilize potentially transient, agonist-mediated, dynamin-c-Src interactions. HEK293 cells stably expressing dynamin I were treated with or without isoproterenol for 2 min followed by covalent protein cross-linking with DSP. As shown in Fig. 2C, treatment of cells with isoproterenol increased the amount of tyrosine phosphorylated dynamin present in c-Src immunoprecipitates.
Reverse phase high pressure liquid chromatography analysis of trypsin-digested phosphodynamin isolated from isoproterenol-stimulated cells revealed the presence of two major phosphopeptides. Microsequencing and amino acid sequence analyses (19,22) of one peptide revealed that Tyr 597 is phosphorylated (Fig. 3A). Sequence analysis of the second phosphopeptide revealed sequences just N-terminal to a GYIG motif (Tyr 231 ) that displays high sequence homology to consensus c-Src phosphorylation sites (28).
To determine whether tyrosine phosphorylation of dynamin affected its function, each of these tyrosine residues (at positions 231 and 597) was mutated to phenylalanine singly and in combination. As shown in Fig. 3 (B and C), whereas isoproterenol treatment increased the tyrosine phosphorylation content of wild-type dynamin, the single tyrosine mutants (Y231F and Y597F), and the double tyrosine mutant (Y231F/Y597F) of dynamin exhibited diminished tyrosine phosphorylation in both isoproterenol-stimulated and nonstimulated cells. The dramatic decrease in tyrosine phosphorylation of Y231F/Y597F mutant strongly suggests that these residues represent the phosphorylated sites in vivo.
In agreement with previous results (5, 15), overexpression of  4. Effects of tyrosine phosphorylation on the function of dynamin. HEK293 cells were transiently transfected with plasmids expressing the following: empty pRK5 vector (CN), wild-type dynamin I (WT), or each of K44A, Y231F, Y597F, and Y231F/Y597F mutant dynamin I. A, effect on receptor sequestration. Flag-tagged ␤ 2 -AR was co-transfected with the indicated dynamin cDNAs, and the amount of internalized receptor was measured as described for Fig. 1. Values shown are obtained from five independent experiments done in triplicate and expressed as the percentage of loss of agonist-induced cell surface receptor over unstimulated cells. B and C, isoproterenol-induced activation of MAP kinases. HEK293 cells expressing HA-Erk2, c-Src, and dynamin were stimulated with isoproterenol for 5 min after overnight serum starvation. In vitro kinase assays were performed on HA-Erk2 immunoprecipitates using MBP as substrate. Phosphorylated MBP was visualized by autoradiography, and HA-Erk2 was immunoblotted in parallel using anti-Erk2 antibody. Labeled MBP was quantified using a PhosphorImager. Values are shown as agonist-induced fold increase over unstimulated cells. Data represent means Ϯ S.E. from five independent experiments.

FIG. 2. Agonist-induced tyrosine-phosphorylation of dynamin.
A, serum-starved HEK293 cells were stimulated with 10 M isoproterenol at 37°C for the indicated times after serum starvation overnight. Upper panel, tyrosine phosphorylation of dynamin, purified on GST-Grb2 beads was detected by immunoblotting (IB) with anti-phosphotyrosyl antibodies. Lower panel, the filter shown in the upper panel was stripped and blotted with anti-dynamin antibodies to document equivalent loading among the different samples. B, dynamin I-overexpressing HEK293 cells were transiently transfected with plasmids expressing pRK5 empty vector (CN), C-terminal c-Src kinase (Csk), catalytically inactive c-Src (c-Src-K298M), or constitutively activated Src (v-Src) followed by stimulation with isoproterenol for 2 min as indicated. Dynamin immunoprecipitates (IP) were utilized to determine the tyrosine phosphorylation content of dynamin by immunoblotting with anti-phosphotyrosyl antibodies. The lower panel shows an immunoblot of the filter shown in the upper panel to document equivalent expression of dynamin among the different samples. C, HEK293 cells, stably expressing dynamin I, were stimulated with isoproterenol (ISO) followed by exposure to DSP for 30 min at ambient temperature. c-Src proteins were immunoprecipitated, and coprecipitated dynamin was detected by protein immunoblotting (left columns). Dynamin coimmunoprecipitating with c-Src was also tested for its phosphotyrosine content by immunoblotting with anti-phosphotyrosyl antibodies (right columns). NS, not stimulated. A representative immunoblot is shown on the right of each panel. Each data represent means Ϯ S.E. from three independent experiments. wild-type dynamin did not enhance agonist-promoted sequestration of the ␤ 2 -AR (Fig. 4A). However, expression of Y231F or Y597F dynamin resulted in a modest reduction of agonistinduced ␤ 2 -AR internalization. Expression of the Y231F/Y597F dynamin inhibited approximately 70% of the agonist-induced internalization of the ␤ 2 -AR, which was equivalent to the degree of inhibition observed with GTPase-deficient dynamin K44A. These data suggest that tyrosine phosphorylation of dynamin plays an important role in its ability to support endocytosis.
Recent data suggest that agonist-promoted endocytosis plays a dual regulatory role in signaling pathways emanating from GPCRs or RTKs. Activation of the Erk1/2 MAP kinase pathway by GPCRs, such as those for isoproterenol, lysophosphatidic acid, thrombin, and bombesin (5, 6), as well as RTKs, such as those for EGF (25), nerve growth factor (29,30), and insulinlike growth factor-1 (26), is dependent on receptor internalization. In the case of the ␤ 2 -AR, protein kinase A-mediated phosphorylation of the receptor confers receptor-G i coupling, with subsequent Ras-dependent Erk1/2 activation mediated by G␤␥subunits derived from pertussis toxin-sensitive G proteins (31). In addition, expression of dominant negative mutants of ␤-arrestin 1 or dynamin, which attenuate agonist-mediated endocytosis, blocks ␤ 2 -AR-mediated Erk1/2 activation (5). Considering the role of dynamin in endocytosis, we investigated the effects of expressing wild-type or mutant dynamin on the ␤ 2 -AR-mediated activation of the Erk2 MAP kinase. Expression of the Y231F or Y597F mutated forms of dynamin reduced the isoproterenol-mediated activation of Erk2 approximately 60% (Fig. 4, B and C). However, expression of the Y231F/Y597F mutant reduced the agonist-induced Erk2 activation by 80 -90%, similar to the effects observed using the K44A dynamin (Fig. 4, B and C). These data suggest that c-Src-mediated tyrosine phosphorylation of dynamin is required for Ras-mediated activation of MAP kinase by ␤ 2 -AR by virtue of its involvement in receptor sequestration.
Our results establish that agonist-induced tyrosine phosphorylation of dynamin is required for ␤ 2 -AR internalization and internalization-dependent signaling to MAP kinase. Clathrincoated vesicle-mediated receptor internalization is regulated by the enzymatic (GTPase) activity of dynamin. As shown in Fig. 3A, Tyr 231 and Tyr 597 reside in the GTPase and pleckstrin homology (PH) domains of dynamin, respectively. Intermolecular interaction between GTPase and GED domains of adjacent dynamins has been reported to regulate the GTPase activity of the enzyme (13). Thus, phosphorylation of Tyr 231 might regulate the GTPase activity of dynamin by controlling these intermolecular interactions. Another mutation that disables the GTPase activity of dynamin (K44A) also ablates the ability of dynamin to function in internalization of receptors (14,15). PH domains in proteins such as phospholipase C-␦ 1 (32) have been shown to mediate interactions with acidic phospholipids. Although there is as yet no direct evidence for dynamin PH domain-phospholipid interaction, acidic phospholipids such as phosphatidylinositol 4,5-bisphosphate have been shown to stimulate GTPase activity of dynamin in vitro (10,33). Additionally, G protein ␤␥ subunits, which interact with PH domains of many proteins (34), have been shown to control the GTPase activity of dynamin (33). Tyr 597 is located in one of three variable loops of the dynamin PH domain that possess positive electrostatic potential (35). Phosphorylation of Tyr 597 may alter the ability of dynamin to interact with effector mol-ecules such as phosphatidylinositol 4,5-bisphosphate and G protein ␤␥ subunits, resulting in modulation of dynamin GTPase activity. Although the molecular mechanism(s) whereby tyrosine phosphorylation of dynamin regulates its function remain unclear, the present results establish that the c-Src-mediated tyrosine phosphorylation of dynamin provides one mechanism by which GPCRs regulate their own internalization and MAP kinase signaling.