3.1 c-SrcGFP is a bona fide reporter of c-Src
To facilitate the study of c-Src trafficking in cells, a c-Src GFP fusion construct was generated. Since the N-terminus of c-Src contains the myristoylation site that governs its association to membranes [3
], GFP was fused to the C-terminus of the protein. A previous study revealed that fusion of GFP directly to the C-terminus of c-Src generates a constitutively activated c-Src that is no longer capable of undergoing negative regulation [38
]. We therefore created a seven amino acid linker between the C-terminal residue of c-Src and the N-terminal residue of GFP. This c-SrcGFP construct was transfected into COS-1 cells and its subcellular distribution was compared to that of c-Src. Earlier reports demonstrated that c-Src localized to endosomes [14
], and at the cell periphery [39
]. Likewise, c-SrcGFP expressed in COS-1 cells was distributed at the perinuclear region and at the plasma membrane (). COS-1 cells expressing c-Src and c-SrcGFP were lysed, and analyzed by SDS-PAGE and Western blotting with anti-c-Src antibody. As depicted in , expression levels of c-Src and c-SrcGFP were equivalent.
c-SrcGFP is a bona fide reporter of wild-type c-Src
We then compared the stability of c-SrcGFP to that of c-Src. Pulse-chase metabolic labeling using 35S-labeled cysteine and methionine revealed that c-Src and c-SrcGFP have half-lifes of 9.6 and 9.5 hours respectively, indicating that both proteins have similar stabilities (). Lastly, we tested whether fusion of GFP to c-Src abrogated its ability to be regulated during autoactivation. Lysates derived from COS-1 cells expressing c-Src, c-SrcGFP, constitutively activated c-Src (Y527F c-Src) or constitutively activated c-SrcGFP (Y527F c-SrcGFP) were analyzed by immunoblotting with anti-PY416 antibody, which detects activated c-Src phosphorylated at Tyr 416, and total c-Src (). c-Src and c-SrcGFP exhibited similar ratios of activated/total c-Src, indicating that fusing GFP to the C-terminus of c-Src did not result in artificial activation of c-Src kinase activity. In contrast, Y527F c-Src and Y527F c-SrcGFP were both activated to 7- and 8-fold higher levels than their wildtype counterparts, respectively. Taken together, these results demonstrate that c-SrcGFP is localized and regulated in a manner identical to c-Src, and is thus a valid reporter of c-Src localization and function.
3.2 Intracellular c-Src localizes to endocytic compartments and is activated
Earlier studies reported that c-Src was localized to the Golgi apparatus and endosomal membranes, but did not identify the endosomal compartments that contained c-Src. Using indirect immunofluorescence, we observed that c-SrcGFP colocalized with both early endosomal antigen 1 (EEA1), a marker for early endosomes, and CD63, a marker for late endosomes and the multivesicular body (MVB) (). Quantitation using MetaMorph software revealed that the extent of colocalization of the perinuclear pool of c-SrcGFP was 17% and 25% for EEA1 and CD63 respectively (). In contrast, no colocalization of c-SrcGFP with anti-mannosidase II, a cis-medial Golgi marker, was observed (data not shown).
Subcellular localization of c-SrcGFP and activated c-SrcGFP
We next examined the localization of activated c-SrcGFP within cells. Most of the activated c-SrcGFP (67%) was present at the cell surface at sites of membrane ruffling (). In a subset of cells (33%), activated c-SrcGFP was localized both at the plasma membrane and at the membranes of enlarged intracellular vesicles (). Cells expressing a kinase-dead (KD) c-SrcGFP fusion protein were devoid of signal for activated c-Src, confirming the specificity of the anti-pY416 antibody.
3.3 c-Src undergoes constitutive macropinocytosis from the plasma membrane
The presence of distinct populations of activated c-Src and total c-Src in perinuclear vesicles as well as the plasma membrane led us to question how the distribution between the two locations was achieved. A recent report demonstrated that c-Src can traffic from RhoB-positive endosomes to the plasma membrane, and that PDGF stimulation causes c-Src activation during its transit [29
]. However, it is not known how or if c-Src traffics from the plasma membrane to intracellular sites and whether this event occurs via the endocytic pathway. To address this question, time-lapse confocal imaging experiments were performed on live cells expressing c-SrcGFP. COS-7 cells were used for live-cell imaging experiments, since these cells are flatter and more homogeneous than COS-1. c-SrcGFP was distributed at both the plasma membrane and the perinuclear region, as was observed in COS-1 cells. Time-lapse imaging revealed the formation of enlarged vesicles from the plasma membrane at sites resembling membrane ruffles (). With time, c-SrcGFP-enriched plasma membrane ruffles altered shape, circularized, and formed a closed vesicle containing a c-Src-enriched tail that protruded from the vesicle. The tail was then shed, and the vesicle started shrinking and slowly moved towards the perinuclear region of the cell. Vesicles ranged from 2–5 μm in diameter ().
c-SrcGFP traffics from the plasma membrane to intracellular sites via macropinocytosis
Both the size of the vesicles and their formation from the plasma membrane at sites resembling membrane ruffles suggested that c-SrcGFP was trafficking via macropinocytosis, a clathrin-independent endocytic pathway. Three methods were then used to assess macropinocytosis: uptake of the fluid phase marker, Dextran; co-localization with markers for membrane ruffles; and co-localization with F-actin, since membrane ruffling and macropinosome formation are actin-dependent processes. First, c-SrcGFP expressing COS-1 cells were subjected to a thirty-minute pulse of fluorescently-labeled Dextran. The cells were fixed and visualized by confocal imaging. Dextran uptake was clearly observed in c-SrcGFP vesicles emanating from plasma membrane ruffles (). Second, co-staining with antibody to Rac1 revealed that c-SrcGFP colocalized with Rac-enriched membrane ruffles. Third, significant colocalization between c-SrcGFP and Texas-Red phalloidin, which stains F-actin, was detected at plasma membrane ruffles (). In some cells, c-Src and actin also co-localized at the membranes of enlarged vesicles. To determine whether c-Src tyrosine kinase activity was required for macropinocytosis, KD c-SrcGFP was expressed in cells and visualized in time-lapse confocal imaging experiments. KD c-SrcGFP localized to sites of plasma membrane ruffling. However, with time, the ruffles receded into the cytoplasm without forming vesicles (). Taken together, these data strongly suggest that c-SrcGFP traffics from the plasma membrane to internal membranes via macropinocytosis and that this movement is dependent on c-Src catalytic activity.
3.4 c-Src traffics with activated EGFR
To date, there has been little characterization of the underlying mechanisms governing macropinocytosis. It is known that EGF induces both Rac-mediated membrane ruffling and macropinocytosis, and that EGFR internalizes via macropinocytosis in addition to classical clathrin-dependent pathways. In addition, several lines of evidence point to functional synergism between c-Src and EGFR. c-Src modulates EGFR internalization from the cell surface, and potentiates EGFR-driven signaling. We therefore hypothesized that c-Src and EGFR would traffic and signal together.
To address this question, GFP, c-SrcGFP, and KDc-SrcGFP were expressed in COS-1 cells, which contain abundant amounts of endogenous EGFR. Serum starved cells were incubated for 10 minutes with EGF, chased for various timepoints in media lacking EGF, fixed, and stained with an antibody directed against activated EGFR (anti-pEGFR, Y1173). At the zero chase timepoint, activated EGFR was observed at the cell surface and intracellularly in GFP and KDc-SrcGFP cells. In c-SrcGFP-expressing cells, strong colocalization between activated EGFR and c-SrcGFP was evident at cell surface ruffles and inside the cell. In the perinuclear region, we observed that EGFR and c-SrcGFP colocalized at punctae as well as at enlarged vesicles reminiscent of macropinosomes ().
c-SrcGFP traffics with activated EGFR
EGFR down-regulation requires approximately 2–3 hours for internalized EGFR to be targeted to the lysosome for degradation [12
]. In cells expressing GFP or KD c-SrcGFP, the fluorescence intensity for activated EGFR was clearly diminished at the 2 hour chase point, consistent with EGFR downregulation. However, c-SrcGFP expressing cells still retained a prominent pEGFR signal (). Both c-Src and activated EGFR colocalized at the plasma membrane and in intracellular locations. Neighboring cells that did not get transfected with c-SrcGFP served as a built-in control; these cells had no pEGFR signal remaining at the 2 hour timepoint. (). The specificity of the anti-pEGFR antibody was confirmed by including a blocking peptide targeted against the pEGFR epitope during primary antibody incubation. No pEGFR signal was detected (). These findings indicate that c-Src and EGFR co-localize and traffic together during EGF stimulation, and that the presence of c-Src prolongs EGFR activation in a c-Src kinase dependent manner.
3.5 c-Src induces EEA1 clustering in EGF-dependent manner
EGFR signals both from the plasma membrane and also from endosomes [13
]. Additionally, EGFR and the downstream signaling adaptors Grb2 and Shc have been shown to colocalize both at the plasma membrane, and in endosomes [42
]. Given our observation that c-Src and EGFR traffic together, we asked what influence c-Src had on the endocytic pathway during EGFR activation.
COS-1 cells were pulsed for 10 minutes with EGF and chased for either zero, 30, and 120 minutes, and the distributions of EEA and CD63 were monitored. In cells expressing GFP, EEA1 staining was evident in a punctate distribution spread throughout the entire cell. In contrast, the majority of c-SrcGFP expressing cells exhibited clustering of EEA1-positive endosomes into the perinuclear region. When c-Src kinase activity was abolished, EEA1 clustering was no longer observed, indicating that c-Src catalytic activity is important for early endosome localization during EGF stimulation (). Quantitation revealed that approximately 42% of c-SrcGFP expressing cells had induced clustering of EEA1 positive endosomes after a 10 minute EGF pulse, compared with 18% and 25% in GFP and KDc-SrcGFP expressing cells. After two hours, EEA1 endosomes remained clustered in 58% of c-SrcGFP expressing cells, compared to 29% and 26% for GFP and KDc-SrcGFP expressing cells, respectively (). Similar experiments were performed using indirect immunofluorescence against CD63 (). No c-Src-dependent endosomal clustering was observed for CD63-positive compartments upon EGF stimulation. These data suggest that clusters of c-SrcGFP and EEA1-positive endosomes serve as hubs for EGF-mediated signaling.
c-Src induces EEA1 clustering during EGF stimulation and colocalizes with activated EGFR in a ligand-independent manner
3.6 c-Src colocalizes with activated EGFR in a ligand-independent manner
To further study the synergy between c-Src and EGFR, we examined cells expressing GFP or c-SrcGFP in the absence of EGF stimulation. As expected, no signal for pEGFR was detected in GFP expressing cells. In sharp contrast, a strong pEGFR signal was detected specifically in cells expressing c-SrcGFP, even in the absence of EGF addition. c-SrcGFP and activated EGFR remained together throughout the cell both at the plasma membrane and intracellularly (). Cells expressing KDc-SrcGFP did not possess any activated EGFR in the absence of EGF addition. Two lines of evidence support the contention that the signal for pEGFR detected in the c-SrcGFP expressing cells is specific. First, neighboring nuclei representing untransfected cells did not have activated EGFR. Second, pre-treatment of cells with a blocking peptide directed against the anti-pEGFR antibody completely blocked the pEGFR signal ().
The results depicted in suggested that c-Src sustained EGFR activation in a ligand-independent manner. To test this hypothesis, COS-1 transfected cells were pre-incubated with an excess of LA1, a neutralizing anti-EGFR antibody that binds to the ligand-binding domain on the extracellular side of EGFR, thus preventing EGF binding to EGFR. We first verified that LA1 addition neutralized the anti-EGFR signal under our conditions. GFP-transfected cells were pre-incubated with LA1, pulsed for ten-minutes with EGF, and then fixed and stained for activated EGFR. Little to no activated EGFR was detected in cells that were pre-incubated with the neutralizing antibody () compared to control cells (). Next, the experiment was performed in the absence of EGF in c-SrcGFP expressing cells. Activated EGFR colocalized with c-SrcGFP at the cell surface and in intracellular vesicles in both the absence or presence of LA1 (). Thus, the presence of the anti-EGFR neutralizing antibody failed to prevent EGFR activation. These data suggest that EGFR activation by c-Src is independent of EGF ligand.
To assess whether the presence of a c-Src-dependent EGFR signal was representative of a functionally active and signaling-competent receptor, we tested activation of downstream effectors in the EGFR signaling pathway. For these experiments, we used NIH3T3 fibroblasts that were stably expressing either c-Src or control vector. Phosphorylation levels of ERK and Shc – two known effectors in EGFR-mediated signaling – were evaluated by Western immunoblotting of whole cell lysates. As depicted in , activation of ERK and Shc was enhanced in c-Src expressing cells in the presence of EGF (lanes 3 and 4). Moreover, a 2-fold increase in phosphorylated ERK and Shc was also observed in c-Src-expressing cells compared to control, even in the absence of EGF stimulation (, lanes 1 and 2 of each blot). Taken together, these results point to a c-Src-mediated enhancement of EGFR activation and downstream signaling in the absence of exogenously added EGF ligand.