G (cid:1)(cid:2) Activation of Src Induces Caveolae-mediated Endocytosis in Endothelial Cells*

Caveolae-mediated endocytosis in endothelial cells is stimulated by the binding of albumin to gp60, a specific albumin-binding protein localized in caveolae. The activation of gp60 induces its cell surface cluster-ing and association with caveolin-1, the caveolar-scaf-folding protein. This interaction leads to G i -induced Src kinase activation, which in turn signals dynamin-2-mediated fission and directed migration of caveolae-derived vesicles from apical to basal membrane. In this study, we investigated the possible role of the G (cid:1)(cid:2) heterodimer in signaling G i -induced Src activation and subsequent caveolae-mediated endocytosis. We observed using rat lung microvascular endothelial cells that expression of the C terminus of (cid:1) -adrenergic receptor kinase (ct- (cid:1) ARK), an inhibitor G (cid:1)(cid:2) signaling, prevented gp60-dependent Src activation as well as caveolae-mediated endocytosis and transcellular transport of albumin and uptake of cholera toxin subunit B, a specific marker of caveolae internalization. Expression of ct- (cid:1) ARK also prevented Src -mediated tyrosine phosphorylation of caveolin-1 and dynamin-2 and

Plasma albumin functions as a carrier protein in delivering the bound fatty acids, hormones, and drugs across the capillary wall. Transcytosis of albumin may also contribute to tissue fluid balance by regulating the transendothelial oncotic pressure gradient (for review, see Ref. 1). The 60-kDa albumin-binding protein, gp60, 1 in endothelial cells functions as a high affinity membrane-associated glycoprotein such that its activation stimulates transcellular albumin transport via caveolae (2)(3)(4)(5)(6)(7)(8)(9)(10). Depletion of cell surface gp60 was shown to reduce transendothelial 125 I-albumin permeability (9,11). Another important characteristic of the endothelial vesicular transport pathway is that, although gp60 activation triggers the transport of the receptor-bound albumin, the bulk of albumin carried in caveolae is in the fluid phase (9).
Transendothelial permeability of albumin is in part regulated by the binding of albumin to a limited number of high affinity albumin binding sites on the endothelial cell surface that activate vesicular trafficking by triggering the release of caveolae from the membrane (7,9,12). Albumin binding studies showed that specific binding sites for albumin on the endothelial cell surface are occupied at physiologic plasma albumin concentration (9), a finding consistent with albumin transcytosis being a continuously active process (13,14). The numerous vesicles (comprising ϳ20% cell volume) seen in situ in microvascular endothelial cells (15) are indicative of the robustness of this transport pathway. Interestingly, albumin uptake and transport were not saturated at the normal plasma albumin concentration (9), suggesting that the system can be up-regulated. We showed that albumin uptake was maximal at ϳ3 mM albumin (9), a concentration 5-fold greater than the plasma albumin concentration.
The structure of caveolae depends on the oligomerization and membrane association of the scaffolding protein, caveolin-1, in the cholesterol-and sphingolipid-rich membrane domains (16 -20). Caveolae are able to form mobile signaling platforms by sequestering signaling proteins that bind to the caveolin-1scaffolding domain (21, for review see Ref. 22) and accumulate in caveolae upon lipid modification (23)(24)(25)(26)(27). The heterotrimeric GTP-binding protein, G i , which also binds to caveolin-1 (7,21,22), appears to play a fundamental role in the mechanism of caveolae-mediated endocytosis (7). We showed that the endocytic response was pertussis toxin-sensitive (7) and also that Src family tyrosine kinases were required for caveolae-mediated endocytosis of albumin subsequent to phosphorylation of caveolin-1 and dynamin-2 in endothelial cells (6,7,12). Others have also reported that endocytosis via caveolae is critically dependent on stimulation of tyrosine kinase signaling (28 -34). Src-induced phosphorylation of caveolin-1 at Tyr 14 (12,29,(35)(36)(37)(38) is believed to be an important signaling "trigger" for caveolae-mediated endocytosis (10,12,29,34). Caveolin-1 phosphorylation is also controlled by the docking of the negative regulator of Src, Csk (39), suggesting a mechanism of negative feedback regulation.
The GTPase dynamin, another important substrate of Src (40,41), is critically involved in the fission of caveolae from the plasma membrane (12,42,43). We showed that Src phosphorylation of dynamin-2 induced its association with caveolin-1 at the plasma membrane and resultant caveolae-mediated endocytosis of albumin and cholera toxin subunit B (12). Interestingly, SV40-induced internalization of caveolae is also dependent on tyrosine kinase activity (30) and recruitment of dynamin to the membrane (32). In addition, studies have shown that Src phosphorylation of dynamin-1 in response epidermal growth factor receptor or ␤ 2 -adrenergic receptor stimulation regulated dynamin self-assembly and its GTPase activity, thereby facilitating endocytosis of clathrin-coated vesicles (40,41,44).
Although gp60 activation of the Src kinase pathway and the subsequent phosphorylation of caveolin-1 and dynamin-2 are known to be important in signaling caveolae-mediated endocytosis (6,7,10,12), the mechanism of activation of Src by G i and its role in signaling caveolae-mediated endocytosis are not understood. Herein, we demonstrate that G␤␥ signals Src kinase activation in endothelial cells and the subsequent release of caveolae from the plasma membrane, thereby inducing the transcytosis of albumin.

MATERIALS AND METHODS
Cell Culture and Adenoviral Infection-Rat lung microvessel endothelial cells (RLMVECs) were grown on tissue culture-treated culture dishes (Corning). For transcytosis experiments (i.e. transcellular permeability assay), cells were grown on Transwell filter inserts with polypropylene membranes (12-mm diameter, 1-cm 2 growth area, 0.4-m pore size, Corning Costar, Cambridge, MA), and for confocal imaging experiments, cells were grown on number 1.5 glass coverslips in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% FBS, 50 units/ml penicillin, and 50 g/ml streptomycin as described previously (9). Adenoviral vector (type 5) encoding the C terminus of ␤-adrenergic receptor kinase 1 (Adv-ct-␤ARK1) or empty vector (control, Adv-EV) were generous gifts from Dr. Walter J. Koch (Jefferson Medical College, Philadelphia, PA) (45). Cells were infected with the Adv-EV or Adv-ct-␤ARK for 36 h in medium containing 10% FBS and then serum-deprived for 12 h.
Reagents-All of the reagents were obtained from Sigma unless otherwise stated. Hanks' balanced salt solution containing NaHCO 3 (4.2 mM) and HEPES (10 mM) was adjusted to pH 7.4. PP2 and PD98059 were purchased from Calbiochem. Bovine serum albumin (fraction V, 99% pure, endotoxin-free, cold alcohol-precipitated) was dissolved in Hanks' balanced salt solution. The cell-permeant ␤␥-activating peptide mSIRK (myristoylated-SIRKALNILGYPDYD) or the control peptide L9A-mSIRK (Leu 9 changed to Ala decreases the binding affinity greater than 100-fold compared with mSIRK) (46) were dissolved in Me 2 SO and stored at Ϫ80°C. Monolayers of RLMVEC were serum-deprived for 12 h to which 0.1, 1.0, or 10.0 M mSIRK or L9A-mSIRK was added for 0 -5 min at 37°C. Hanks' balanced salt solution containing 0.2% Me 2 SO was used as vehicle control for the 10 M dose.
Antibodies and Fluorescent Probes-The antibodies (Abs) and fluorescent probes used in these studies were obtained from the following sources. Monoclonal Abs for dynamin-2, caveolin-1, and PY20 were from Transduction Laboratories. pY416-Src, 42/44, and p42/44 polyclonal Abs were from Cell Signaling. ␣-Tubulin monoclonal Ab was from Sigma. c-Src polyclonal Ab, ␤ARK1 (GRK2) polyclonal Ab, myc monoclonal Ab, and horseradish peroxidase-conjugated goat anti-rabbit or anti-mouse IgG were from Santa Cruz Biotechnology. Goat anti-mouse and anti-rabbit IgG labeled with Alexa 488 or Alexa 568 and Alexa 488or Alexa 594-conjugated albumin or cholera toxin subunit B (CTB) were from Molecular Probes. Anti-gp60 Ab was generated as described previously (37).
Immunoprecipitation and Western Blot Analysis-For Western blot analysis, lysates were prepared as previously described (7). For immunoprecipitations, the lysates were incubated with 1-10 g/ml primary antibodies for 4 h at 4°C followed by incubation with protein A/G-agarose beads (Santa Cruz Biotechnology) overnight. Proteins were resolved by SDS-PAGE as described previously (7), and the protein bands were visualized by incubating the membrane with ECL reagent (Pierce).
Immunostaining and Confocal Microscopy-Confluent RLMEVCs were washed with phosphate-buffered saline, incubated for 12 h in serum-free medium, and incubated with Alexa 488-or Alexa 594-conjugated albumin or CTB in HEPES-buffered Hanks' balanced salt solution for 30 min at 37°C. Subsequently, the cells were washed with pH 2.5 buffer (0.2 M acetic acid and 0.5 M NaCl) to remove non-internalized/ membrane-associated albumin. The cells were then fixed, permeabilized, and stained with ␤ARK1 Ab (1 g/ml) or gp60 Ab (4 g/ml) and the nuclear marker, DAPI (1 g/ml), as previously described (7). A Zeiss LSM 510 microscope was used for confocal microscopy with 488-and 543-nm excitation laser lines. Non-confocal DAPI images were acquired using Hg lamp excitation and UV filter set. Fluorescence emission was detected in optical sections Ͻ1 m in thickness (pinhole set to achieve 1 Airy unit) separately for each fluorophore using a multi-track 4-frame average configuration. Cellular fluorescence intensity (per pixel average) in the acquired confocal images was determined using Zeiss LSM 510 META software.
Uptake and Transendothelial Transport of 125 I-Albumin-Uptake of 125 I-albumin in confluent RLMVEC monolayers grown on plastic 6-well tissue culture plates was measured as previously described (7,9,12). Transendothelial permeability of 125 I-albumin of RLMVEC monolayers grown on clear microporous polyester Transwell membranes was calculated as previously described (7,9,12). Tracer 125 I-albumin transport in the presence of 100 mg/ml unlabeled albumin was subtracted from 125 I-albumin transport in the presence of 0.1 mg/ml unlabeled albumin (9). Tracer albumin permeability in the presence of a 1000-fold excess unlabeled albumin thought to represent the "leakiness" of the monolayer (9) was Ͻ20% of the total 125 I-albumin transport. Thus, this transendothelial permeability assay quantifies receptor-mediated and receptor-associated fluid phase albumin transport through the vesicular pathway (9).

gp60 Stimulation of G␤␥ Induces Src, Dynamin-2, and Caveolin-1 Phosphorylation and Association of Dynamin-2 with
Caveolin-1-We recently demonstrated that Src-mediated phosphorylation of dynamin-2 and resultant association with caveolin-1 are essential for endocytosis of albumin in endothelial cells (12). To determine whether G␤␥ is upstream of Src in this signaling pathway, we investigated whether endothelial cell expression of ct-␤ARK1, the carboxyl-terminal 194 amino acids of ␤-adrenergic receptor kinase 1 that sequesters G␤␥ and thereby prevents activation of its downstream effectors (45), interferes with gp60-activated phosphorylation of Src, dynamin-2, and caveolin-1. Confluent monolayers of RLMVEC were infected with Adv-EV, type 5, or Adv-ct-␤ARK. Cells were infected for 48 h, the last 12 h of which serum was withdrawn from the cells. Fig. 1A shows the expression of ct-␤ARK at a molecular mass of 28 -30 kDa by ␤ARK1 Western blot, whereas this band was not present in EV-infected cells. Endogenous ␤ARK1 (molecular mass of 80 kDa) shown in Fig. 1A, top panel, did not change in cells infected with Adv-ct-␤ARK. As shown in Fig. 1B, ct-␤ARK inhibited gp60 activation-dependent phosphorylation of Src (pY416-Src immunoblot), caveolin-1 (pY14caveolin-1 immunoblot), and dynamin-2 (PY20 immunoblot after dynamin-2 immunoprecipitation).
In other experiments, we examined the effects of inhibition of dynamin-2 phosphorylation by ct-␤ARK on the association between dynamin-2 and caveolin-1 (as assessed by a co-immunoprecipitation study) because dynamin-2 phosphorylation has been shown to facilitate this interaction (12). As shown in Fig.  1, C and D, the infection of cells with Adv-ct-␤ARK prevented phosphorylation-dependent association of dynamin-2 and caveolin-1 as determined by co-immunoprecipitation with dynamin-2 Ab (Fig. 1C) or caveolin-1 Ab (Fig. 1D). In a control experiment, we showed (as evident from confocal images shown in Fig. 1E) that ct-␤ARK expression did not affect the pattern or intensity of anti-gp60 Ab immunostaining (top panels), indicating that a change in gp60 expression could not account for the differences noted. ␤ARK1 Ab immunostaining was prominent in greater than 90% RLMVECs infected with Adv-ct-␤ARK (bottom right panel), whereas endogenous ␤ARK1 expression (bottom left panel; see also Fig. 2A) was weakly detected at the same amplifier gain and offset settings used for non-saturated fluorescence detection of the overexpressed ct-␤ARK (␤ARK1 Ab is directed against the C terminus of ␤ARK1 and thus is able to recognize the expressed ct-␤ARK and endogenous ␤ARK1 (Fig. 2A)).

Inactivation of G␤␥ Decreases Endocytosis of Albumin and Cholera Toxin Subunit B and Albumin Permeability in Endo-
thelial Cells-To investigate whether G␤␥ is required for caveolae-mediated endocytosis, we infected confluent monolayers of RLMVEC with Adv-ct-␤ARK or EV. Transfected cells were then incubated with Alexa 488-albumin or Alexa 488-CTB, permeabilized, and stained with ␤ARK1 Ab to confirm that Ͼ90% of the cells expressed ct-␤ARK. We achieved high transfection efficiency and ct-␤ARK expression as evident by immunostaining of cells transfected with Adv-ct-␤ARK. Confocal images in Fig. 2A show abundant fluorescent-tagged albumin or CTB internalized in vesicles in EV-infected cells, whereas albumin and CTB endocytosis were markedly reduced in ct-␤ARK-infected cells. Infection of cells with empty vector had no effect on the uptake of albumin or CTB compared with nontransfected cells (data not shown). The ct-␤ARK-expressing cells were easily distinguished from non-expressing cells by the intensity of the ␤ARK1 Ab staining. ct-␤ARK-expressing cells showed a more intense red-staining pattern due to the cumulative staining of endogenous ␤ARK1 and ct-␤ARK1 ( Fig. 2A). Quantification of the fluorescence intensity (per pixel average) in 12 cells/treatment group from two separate experiments is shown in Fig. 2B. Albumin and CTB uptake in cells expressing ct-␤ARK were reduced by 42 and 54%, respectively, compared with cells infected with EV.
To quantify the effects of ct-␤ARK on endocytosis and transcytosis of albumin, we measured 125 I-albumin uptake as well as transendothelial transport in confluent RLMVEC monolayers infected with EV or ct-␤ARK. As shown in Fig. 2C, 125 Ialbumin uptake was reduced by 55% compared with EV-infected cells, similar to that seen with the fluorescence uptake assay. The measurement of luminal-to-abluminal transport of 125 I-albumin in confluent RLMVEC monolayers grown on Transwell filters showed that Adv-ct-␤ARK infection reduced 125 I-albumin permeability by 52% (Fig. 2D). From these data, we conclude that G␤␥ signaling is important in the mechanism FIG. 1. Confluent monolayers of RLMVEC were infected with Adv-EV or Adv-ct-␤ARK for 48 h, and the cells were maintained in serum-free conditions during the last 12 h. A, cell lysates were applied to 10% gels and analyzed by Western blotting for endogenous ␤ARK1 (80 kDa, top band) and overexpressed ct-␤ARK (28 -30 kDa, bottom band). B, cells were treated with 20 g/ml anti-gp60 cross-linking antibody for 5 min to activate albumin-binding protein, gp60 (similar results were obtained with 30 mg/ml albumin, the gp60 ligand; thus, albumin was used in all of the subsequent experiments to activate gp60). Cell lysates were subjected to immunoprecipitation, and samples were applied to 10% gels and analyzed by Western blotting for the proteins as indicated. Phosphorylation of dynamin-2, Src, and caveolin-1 induced by gp60 activation was reduced or prevented in cells expressing ct-␤ARK. C and D, expression of ct-␤ARK prevented the gp60 activation-dependent phosphorylation of dynamin-2 and the association between dynamin-2 and caveolin-1 as determined by immunoprecipitation with anti-dynamin-2 Ab (C, Dyn-2) or anti-caveolin-1 Ab (D, Cav-1). E, confocal merged images of DAPI (blue) and anti-gp60 Ab plus Alexa 488-anti-rabbit secondary Ab immunostaining (green; top panels) in cells infected with Adv-EV or Adv-ct-␤ARK indicate that ct-␤ARK expression did not affect albumin-binding protein gp60 expression. Adv-ct-␤ARK-infected cells showed overexpression of the C-terminal end of ␤ARK1 (green; bottom right) relative to endogenous ␤ARK1 (bottom left) as detected by anti-␤ARK1 Ab plus Alexa 488-anti-rabbit secondary Ab and DAPI staining. All of the experiments were repeated at least three times with similar results. IP, immunoprecipitation; IB, immunoblotting. of caveolae-mediated endocytosis and transcytosis of albumin in endothelial cells.
mSIRK Induced Phosphorylation of Src, Dynamin-2, and Caveolin-1-To test the hypothesis that Src activation and downstream signaling of endocytosis is triggered by G␤␥, cells were stimulated with the permeable peptide mSIRK. This peptide promotes the dissociation of G␣ i and G␤␥ in the absence of receptor stimulation or nucleotide exchange by binding to ␤␥, thereby increasing the concentration of free ␤␥ available to stimulate effectors (46,47). Confluent monolayers of RLMVEC were serum-deprived for 12 h and then treated with 0.1, 1.0, or 10.0 M mSIRK or L9A-mSIRK (control peptide) for 5 min at 37°C. As shown in Fig. 3A, mSIRK caused concentration-dependent activation of Src as indicated by the phospho-Src immunoblot (pY416-Src Ab). Equivalent concentrations of L9A-mSIRK were less effective (Fig. 3A, n ϭ 3), and there was no effect of the vehicle alone (Ϫ). Stimulation of cells with 10 M mSIRK for 0 -5 min activated Src within 30 s (Fig. 3B). Caveolin-1, a substrate for Src (36), was also phosphorylated within 30 s after the addition of mSIRK, suggesting that activation of G␤␥ and subsequent tyrosine phosphorylation of caveolin-1 occur rapidly, perhaps secondary to coincident cellular localization (Fig. 3B).
We next determined whether mSIRK induced Src activation and phosphorylation of caveolin-1 as well as dynamin-2, both of which are important components of the endocytic machinery in endothelial cells (6,7,12). RLMVECs were exposed to 10 M mSIRK or L9A-mSIRK, 30 mg/ml albumin, or vehicle for 5 min at 37°C, and total cell lysates were subjected to Western blot analysis to evaluate the phosphorylation of Src, dynamin-2, and caveolin-1. As shown in Fig. 4A, both mSIRK and gp60activation increased Src, caveolin-1, and dynamin-2 phosphorylation. The specificity of pY416 Src blot as an indicator of Src activation was assessed by pretreating cells with PP2 (15 M pretreatment for 15 min), an Src family tyrosine kinase inhibitor. PP2 blocked Src phosphorylation induced by mSIRK (Fig.  4B). Thus, caveolin-1 and dynamin-2 are both phosphorylated by Src following the activation of G␤␥ with the peptide mSIRK.
We next determined whether the induction of Src, caveolin-1, and dynamin-2 phosphorylation by mSIRK also led to increased association of dynamin-2 and caveolin-1 as shown with gp60 activation-dependent phosphorylation of dynamin-2 (12).

FIG. 2.
A, confluent monolayers of RLMVEC were infected with Adv-ct-␤ARK or EV for 36 h, serum-deprived for 12 h, and incubated with Alexa 488-labeled albumin or Alexa 488-CTB for 30 min (green). The cells were then fixed, permeabilized with 0.1% Triton X-100, and stained with DAPI (blue) and ␤ARK1 Ab plus anti-rabbit Alexa 568 secondary Ab (red), which detects both endogenous ␤ARK1 and ct-␤ARK. The dramatic increase in ␤ARK1 staining indicates that Adv-ct-␤ARK infection was highly efficient and that ct-␤ARK was overexpressed. B, quantification of fluorescence data shows that endocytosis of Alexa 488-albumin and Alexa 488-CTB (a marker of caveolae in endothelial cells) was significantly reduced (ϳ50%) in cells expressing ct-␤ARK compared with EV-infected cells (*, p Ͻ 0.05). Results are mean Ϯ S.E., n ϭ 12. C and D, uptake and transendothelial permeability of 125 I-albumin were reduced ϳ50% in cells expressing ct-␤ARK compared with cells expressing EV. Confluent monolayers of RLMVEC on 6-well tissue culture plates (C) or 1-cm 2 Transwell filter inserts (D) were infected with Adv-EV or Adv-ct-␤ARK for 48 h. Cells were serum-deprived for the last 12 h, and then endocytosis (C) or permeability (D) of 125 I-albumin was determined from the tracer internalized or accumulating in the lower chamber as described under "Materials and Methods." *, ct-␤ARK reduced albumin uptake (p Ͻ 0.05, n ϭ 5) and transendothelial transport (p Ͻ 0.05; n ϭ 5).
mSIRK Activation of CTB Endocytosis-We determined whether mSIRK activation of G␤␥ in endothelial cells stimulates caveolae-mediated endocytosis. RLMVECs were incubated with Alexa 488-CTB for 30 min at 4°C to induce the binding of CTB to cell surface, washed with ice-cold serum-free medium to remove excess CTB, and then incubated at 37°C for 5 min in the presence of 10 M mSIRK or L9A-mSIRK. Control cells were stimulated with vehicle (0.2% Me 2 SO). Confocal images showed increased internalization of Alexa 488-CTB in plasmalemmal vesicles (Fig. 5A). Fluorescence intensity of the internalized Alexa 488-CTB increased by 75% in cells stimulated with mSIRK compared with vehicle alone. In cells infected with ct-␤ARK, the mSIRK-induced increase in CTB uptake was abolished (Fig. 5B), suggesting that the effect was dependent on the activation of G␤␥. The treatment of cells with 10 M control peptide, L9A-mSIRK, increased CTB uptake by 26% relative to the vehicle control. A bar graph of the cellular fluorescence intensity of CTB internalized in caveolae under control and stimulated conditions is shown in Fig. 5B (mean Ϯ S.E.; n ϭ 12 cells/group from two separate experiments).
mSIRK Activation of Src Signaling-To reinforce the concept that mSIRK activates G␤␥, we measured mSIRK and gp60mediated Src (pY416 immunoblot) and caveolin-1 phosphorylation (pY14 immunoblot) in cells expressing ct-␤ARK. As shown in Fig. 6A, ct-␤ARK expression prevented mSIRK-mediated activation of Src as well as phosphorylation of the Src substrate, caveolin-1. The gp60 activation-dependent caveolin-1 phosphorylation was also blocked by ct-␤ARK. This observation indicates that gp60 activation induces G␤␥-mediated signaling in RLMVEC similar to mSIRK.
Previous studies showed that mSIRK stimulates ERK1/2 activation in rat arterial smooth muscle cells (46). Therefore, in a control experiment, we determined whether mSIRK also induced ERK1/2 phosphorylation in RLMVEC (Fig. 6B) and whether this was blocked by the mitogen-activated protein  4. A, confluent RLMVECs were serum-deprived and treated with 10 M mSIRK or L9A-mSIRK, 30 mg/ml albumin to activate gp60, or 0.2% Me 2 SO (vehicle control) for 5 min, and cell lysates were subjected to Western blotting as described. Phosphorylation of Src, caveolin-1, and dynamin-2 induced by mSIRK was comparable to that seen with the activation of gp60. Blots were reprobed for the detection of tubulin levels in each lane to ensure equal loading. B, PP2, a Src family kinase inhibitor, prevented mSIRK-induced Src phosphorylation as detected with the phospho-specific pY416-Src Ab. C, mSIRK stimulates Src-mediated phosphorylation of dynamin-2 and caveolin-1 as well as anti-dynamin-2 Ab co-immunoprecipitation of caveolin-1. L9A mSIRK was much less effective at inducing dynamin-2 phosphorylation and the phosphorylation-dependent association between dynamin-2 and caveolin-1. Results are representative of three experiments. IP, immunoprecipitation; IB, immunoblotting. kinase (MAPK) and ERK inhibitor PD98059 (Fig. 6B). We observed no increase in ERK1/2 phosphorylation as determined by phospho-ERK1 and -ERK2 Western blot analysis in vehicletreated cells (control), whereas ERK1/2 activation was seen upon the addition of FBS (used as a positive control) (48) and albumin (through gp60 activation), although the latter effect was the less pronounced. The increases in pY416-Src and pY14-caveolin-1 were not inhibited by the addition of PD98059, indicating that G␤␥ activation of the Src pathway is independent of MAPK/ERK activation by G␤␥. In support of this finding, we observed that the level of Alexa-CTB uptake upon mSIRK addition was also unaffected by PD98059 pretreatment (data  6. A, expression of ct-␤ARK prevents mSIRK-mediated activation of Src. Confluent RLMVECs were infected with Adv-ct-␤ARK or EV for 36 h and serum-deprived for 12 h. Cells were then incubated with vehicle, 10 M mSIRK, or 30 mg/ml albumin (to induce gp60 activation) for 5 min at 37°C, lysed, and subjected to Western blotting for phospho-Y416-Src, phospho-Y14-caveolin-1, or ␣-tubulin. Note that increased Src and caveolin-1 phosphorylation induced by gp60 activation or mSIRK were blocked by ct-␤ARK. B, MAPK inhibitor PD98059 failed to prevent Src phosphorylation induced by FBS, mSIRK, and albumin. RLMVECs were serum-deprived for 12 h, treated for 20 min with 10 M PD98059, and stimulated with vehicle, 10 M mSIRK, 30 mg/ml albumin, or 10% FBS for 5 min. Whole cell lysates were separated by SDS-PAGE, transferred to nitrocellulose, and immunoblotted for phospho-ERK1/2 and total ERK1/2 (p44, p42) and phospho-Src (pY416-Src). Although PD98059 prevented ERK1/2 phosphorylation induced by mSIRK, gp60 activation, or FBS, it did not block Src phosphorylation. To some extent, PD98059 alone also increased Src phosphorylation. Results are representative of three experiments. IB, immunoblotting. not shown). Thus, mSIRK induces the activation of Src kinase, which in turn phosphorylates dynamin-2 and caveolin-1, and thereby regulates caveolae-mediated endocytosis. mSIRK also activates ERK1/2 phosphorylation in endothelial cells. However, these results suggest that G␤␥-dependent Src activation and the signaling of caveolae-mediated endocytosis are independent of the Gb␥-mediated activation of ERK. DISCUSSION We have investigated key steps in signaling caveolae-mediated endocytosis and transcellular transport of albumin in endothelial cells. Our results demonstrate that G␤␥ activation of Src is a critical signal mediating endocytosis and transendothelial albumin transport via transcytosis. Our observations are important because they (i) link ␤␥ signaling to the regulation of caveolae-mediated endocytosis and transcytosis in endothelial cells and (ii) demonstrate that Src is the downstream effector of the ␤␥ heterodimer in this process. Our results are consistent with reports that G␤␥ is a key signaling intermediate responsible for the activation of Src (49 -51).
Stimulation of G i -coupled seven-transmembrane-spanning receptors is known to activate Src in a G␤␥-dependent manner (49). However, less conventional means of stimulating heterotrimeric G-proteins independent of G i -coupled seven-transmembrane-spanning receptors have also been reported such as the activation of G i via the epidermal growth factor receptor (52,53). Peptides have recently been synthesized based on G protein regulatory motifs (52,54) that compete with G␣ for the interaction with ␤␥ subunits (46). These peptides promote subunit dissociation in a receptor and nucleotide exchange-independent manner (47). The activation of G i signaling by albumin-binding protein, gp60, which is responsible for mediating caveolae-induced endocytosis (7), is yet another example of engagement of a heterotrimeric G protein by a G i -coupled seven-transmembrane-spanning receptor-independent mechanism. The activation of gp60-dependent Src kinase signaling of endocytosis (6) was blocked by pertussis toxin and G i antagonist peptide (7), indicating that Src activation was the result of G i signaling. In this study, we demonstrated that G␤␥ mediates Src activation and thereby signals G i -linked Src kinase activation of caveolae-mediated endocytosis.
The caveolae-associated protein, caveolin-1, in endothelial cells serves to bind signaling molecules such as various G proteins, Src family tyrosine kinases, endothelial nitric-oxide synthase, and protein kinase C␣ (for review see Ref. 55) as well as several G protein-coupled seven-transmembrane-spanning receptors and other transmembrane signaling molecules (23,24,57,58). In endothelial cells, caveolae are the primary endocytic vesicle internalized by plasma membrane fission (9,12,24,59,60), which upon release from the membrane mediate transcellular transport of albumin and other macromolecules from blood to the underlying tissues (5,8,13,15,59,61). G␣ i and Src kinase have been shown to compete for the same binding site on caveolin-1 (7), presumably in the caveolin scaffold binding domain (21,22). The activation of the 60-kDa albumin-binding protein, gp60, induced the activation of G i and the Src kinase pathway responsible for caveolae-mediated endocytosis in microvessel endothelial cells (6,7). Whereas the mechanisms regulating the trafficking of caveolae to the basal membrane remain unclear, it is known that the activation of Src kinase and downstream phosphorylation events mediated by Src, i.e. phosphorylation of caveolin-1 and dynamin-2, are required for caveolar fission from the membrane (4,6,7,10,12,32,33). Other studies have implicated protein kinase C-dependent signaling as being necessary for caveolae-mediated endocytosis of glycosphingolipids (26,28,34). However, we have observed that the pan-protein kinase C inhibitor calphos-tin C had no effect on the gp60-activated endocytosis of the vesicle marking dye FM1-43 in endothelial cells (62).
In this study, we investigated the involvement of G␤␥ in signaling Src activation and in inducing caveolae-mediated endocytosis and transcytosis. We observed that ct-␤ARK expression, known to sequester and thus inactivate G␤␥ (46), prevented the gp60-induced activation of Src kinase and subsequent phosphorylation of caveolin-1 and dynamin-2. In addition, ct-␤ARK expression blocked the Src phosphorylation-dependent association between dynamin-2 and caveolin-1 at the membrane. As Src kinase phosphorylation of dynamin-2 Y597 in the pleckstrin homology domain and association of dynamin-2 with caveolin-1 are both required for the gp60-induced endocytosis of albumin in endothelial cells (12), our results support a model in which G␤␥ is essential for the activation of Src and hence caveolae-mediated endocytosis.
In a series of experiments, we used the cell-permeant peptide mSIRK, which promotes G␤␥-dependent signaling and effector stimulation in the absence of receptor stimulation or nucleotide exchange (46,47). The treatment of cells with mSIRK activated Src within 30 s and led to the phosphorylation of dynamin-2 and caveolin-1. PP2, the Src kinase inhibitor, prevented the mSIRK-induced activation of Src. Consistent with the results using Adv-ct-␤ARK, the activation of Src by the G␤␥ activator mSIRK also promoted the association of caveolin-1 and dynamin-2. To address whether G␤␥ stimulation by mSIRK was capable of activating endocytosis, we assessed the internalization of fluorescent CTB (Alexa 488-CTB) in endothelial cells treated with mSIRK. The uptake of CTB increased by 75% in these cells compared with vehicle-treated cells. To prove that G␤␥ was activated by mSIRK, we examined whether the expression of ct-␤ARK in endothelial cells would block the mSIRK-induced activation of Src as assessed by Western blot using the phospho-specific anti-Src antibody, pY416-Src. We observed that mSIRK-induced Src activation and phosphorylation of caveolin-1 were prevented by ct-␤ARK, indicating that G␤␥ induced the activation of Src and the subsequent phosphorylation of caveolin-1. Although the present results are definitive in their own right, it is clear that additional studies will be required to characterize fully the molecular mechanisms of G protein ␤␥ heterodimer stimulation and the downstream signaling of Src activation in response to mSIRK peptide and the implications of this on endothelial transport. mSIRK activation of G␤␥ has also been shown to activate MAPK signaling in certain cell types (46). As shown herein in endothelial cells, mSIRK increased the phosphorylation of ERK1/2 compared with control cells. The ERK1/2 inhibitor PD98059 abolished the mSIRK activation of ERK1/2 phosphorylation. However, PD98059 had no affect on mSIRK-induced phosphorylation of Src or caveolin-1 or on endocytosis of CTB in endothelial cells. Thus, although G␤␥ is upstream of both Src and MAPK signaling pathways, MAPK does not appear in the G␤␥ signaling cascade regulating caveolae-mediated endocytosis.
Sequestration of the G␤␥ heterodimer has been shown to inhibit endocytosis via clathrin-coated vesicles (63,64), in part by interfering with actin polymerization (63). Although the role of actin in caveolae-mediated endocytosis remains unclear, both Src and dynamin are known to participate in the changes in actin cytoskeleton by regulating cortactin (65)(66)(67). Therefore, it is possible that Src may control the function of actin or associated binding proteins and regulate caveolae movement along the actin filaments or microtubule "tracks" (56,67) in addition to G␤␥-dependent Src activation and the phosphorylation of caveolin-1 and dynamin-2. Thus, G␤␥-mediated activation of Src kinase may be a general requirement for caveolaemediated endocytosis and transcytosis on the basis of both phosphorylation of caveolin-1 and dynamin-2 and engagement of the cytoskeletal protein machinery required for transcellular vesicle trafficking.
In conclusion, we have shown that G␤␥ signaling of Src activation is an important mechanism regulating caveolaemediated endocytosis and transcytosis of albumin. ct-␤ARK expression prevented albumin-binding protein gp60-induced activation of Src, phosphorylation-dependent association of caveolin-1 and dynamin-2, and caveolae-mediated endocytosis and transcytosis of albumin and cholera toxin subunit B. The permeable peptide, mSIRK, which activates G␤␥ independent of receptor stimulation or nucleotide exchange (47), also induced Src activation, phosphorylation of caveolin-1 and dynamin-2, the phosphorylation-dependent association of dynamin-2 and caveolin-1 at the membrane, and activated caveolae-mediated endocytosis of cholera toxin subunit B. Taken together, these data demonstrate that the activation of G␤␥ in endothelial cells induces the fission of caveolae by triggering Src phosphorylation of dynamin-2 and caveolin-1 and that this signaling mechanism is an important regulator of transcytosis across the vascular barrier.