A Unique Platform for H-Ras Signaling Involving Clathrin-independent Endocytosis

Cells, Plasmids, and Transfection

HeLa and COS-7 cells were grown in DMEM supplemented with 10% fetal bovine serum, 100 μg/ml streptomycin, and 100 U/ml penicillin at 37°C with 5% CO₂.

Arf6, Arf6T27N, and Arf6Q67L were in pXS or pEGFP plasmid (Radhakrishna and Donaldson, 1997). Green fluorescent protein (GFP)-Rab5 and mutants were from R. Lodge (Laval, QC, Canada). H-Ras and H-RasG12V are in pDCR, pEGFP, or monomeric red fluorescent protein (mRFP) plasmid (Rotblat et al., 2004). The RBD-GFP was a gift from Mark Philips (New York University, New York, NY) (Bivona et al., 2005). Pleckstrin homology (PH)-phospholipase (PL) Cδ was in pEGFP or pECFP and PH-protein kinase B (Akt) in mRFP were from Tamas Balla (National Institutes of Health, Bethesda, MD). Hemagglutinin (HA)-PLD2 was in pCGN and was from Mike Frohman (Stony Brook University, Stony Brook, NY). Akt1 in pEGFP-N3 was provided by Morris Birnbaum (University of Philadelphia, Philadelphia, PA). GFP-tH encoding GFP fused to the double palmitoylated and farnesylated carboxy terminal tail of H-Ras was from Clontech (Mountain View, CA), and tH was also appended onto the monomeric RFP vector from Roger Tsien (University of California-San Diego) forming RFP-tH. Flag-EFA6 was described previously (Brown et al., 2001). For transfection, cells were plated and transfected the next day by using FuGENE (Roche Diagnostics, Indianapolis, IN). Experiments were performed 17–20 h after transfection.

Reagents and Antibodies

Rabbit polyclonal antibody to Arf6 (Song et al., 1998) and mouse monoclonal antibody (mAb) to human MHCI (W6/32) (Naslavsky et al., 2003) were described previously. The mouse monoclonal anti-HA antibody 16b12 was purchased from Covance (Berkeley, CA). Monoclonal anti-early endosomal antigen (EEA)1 was purchased from BD Biosciences (Palo Alto, CA). Monoclonal anti-phospho-Erk and polyclonal anti-phospho-Akt were purchased from Cell Signaling Technology (Danvers, MA). Invitrogen (Carlsbad, CA) was the source for transferrin (Tfn) conjugated to Alexa-633, rabbit polyclonal antibody to GFP, and all secondary antibodies conjugated to Alexa-594, -488, and -647. Cytochalasin D (CD) and EGF were purchased form Sigma-Aldrich (St. Louis, MO).

Immunofluorescence and Live Cell Imaging

For uptake of Tfn and MHCI, cells were serum starved for 30 min at 37°C in DMEM alone, fluorescently labeled Tfn and anti-MHCI (30–50 μg/ml) were added, and cells were incubated for 30 min at 37°C. At the end of incubation, PM-associated ligand and antibodies were removed by rinsing the cells in low pH solution (0.5% acetic acid and 0.5 M NaCl, pH 3.0) for 20–30 s. Cells were then fixed with 2% formaldehyde in phosphate-buffered saline (PBS) at room temperature for 10 min, and internalized MHCI antibody was labeled with 594-goat-anti-mouse immunoglobulin (Ig)G in the presence of 0.2% saponin. For indirect immunofluorescence, cells were fixed as described above and immunostained as described previously (Naslavsky et al., 2003). Briefly, after fixation cells were incubated for 1 h with primary antibody diluted in 10% fetal calf serum in PBS in the presence of 0.2% saponin. After washing, cells were incubated for 1 h with secondary antibody diluted as described above.

For live cell imaging, HeLa or COS-7 cells were plated onto Lab-Tek coverglass chambers (Nalge Nunc International, Rochester, NY) and transfected with the indicated constructs. Eighteen hours after transfection, cells were imaged on a 37°C stage in CO₂-independent media. For single- or double-channel movies, images were acquired every 6 or 10 s, respectively. All images were obtained using a 510 LSM confocal microscope (Carl Zeiss, Thornwood, NY) with 63 × 1.3 numerical aperture PlanApo objective. After acquisition, images were handled using Adobe Photoshop (Adobe Systems, San Jose, CA). All experiments were confirmed at least three times, and a representative image is shown. Videos were generated using MetaMorph (Molecular Devices, Sunnyvale, CA).

Ras Binding Domain (RBD) Assay for GTP-bound Ras

To examine Ras-GTP levels, H-Ras- or H-Ras- and Arf6Q67L-expressing HeLa cells were serum starved for 1 h and then stimulated with 100 ng/ml EGF for 10 min as indicated. Cell were lysed and subjected to the glutathione transferase (GST)-RBD pull-down assay as described previously (Niv et al., 2002). Bound proteins were eluted from the beads by boiling in SDS-polyacrylamide gel electrophoresis (PAGE) sample buffer and resolved by SDS-PAGE, followed by transfer to nitrocellulose membrane. Western blot was carried out using rabbit anti-GFP to detect tagged H-Ras and rabbit anti-Arf6 to detect Arf6Q67L and the appropriate infrared secondary antibodies. The Western blot was visualized and quantified using an Odyssey infrared imager (LI-COR Biosciences, Lincoln, NE). Data from three independent experiments are shown as the average percentage of Ras pelleted on GST-RBD beads with error represented as ±SD.

H-Ras Associates with the CIE Pathway

To determine whether H-Ras was present on endosomal membranes, we followed the endocytosis of two endogenous PM proteins, MHCI and TfR, that are internalized by CIE and CDE mechanisms, respectively (Weigert et al., 2004). We used an antibody to MHCI and fluorescently labeled Tfn to follow MHCI and TfR endocytosis in GFP-H-Ras–expressing HeLa cells. After 30 min of internalization, H-Ras colocalized with MHCI, but not TfR, on peripheral endosomal structures and on distinctive, tubular recycling endosomes (Figure 1A), which we have shown carry MHCI and other CIE cargo back to the PM (Naslavsky et al., 2003; Weigert et al., 2004). MHCI was also observed on converged, juxtanuclear endosomes that contained Tfn as observed previously (Weigert et al., 2004). The colocalization of H-Ras with MHCI, and not with Tfn, on both incoming and recycling tubular endosomes demonstrates that Ras is uniquely localized to this CIE pathway.

This CIE pathway is associated with the Arf6 GTPase, and we have shown that expression of Arf6Q67L causes a block in trafficking of clathrin-independent cargo immediately after internalization, resulting in cargo accumulation in vacuolar structures that are enriched in PIP₂ and actin (Brown et al., 2001; Naslavsky et al., 2003). We examined the effect of expression of Arf6Q67L on Ras trafficking, and we found that H-Ras was at the PM and associated with the Arf6Q67L vacuoles (Figure 1B), similar to what is observed for CIE cargo (Naslavsky et al., 2003). H-Ras association with Arf6Q67L vacuoles was also observed in MCF7, HT1080, and COS-7 cells (our unpublished data). It was recently shown that PLD2 plays a critical role in the activation of H-Ras via the EGF receptor and the T-cell receptor (Mor et al., 2007; Zhao et al., 2007). Intriguingly, in HeLa and COS-7 cells, the recycling of MHCI and other CIE cargo back to the PM is dependent upon the activation of Arf6 and its ability to activate PLD (Jovanovic et al., 2006). We examined the distribution of transfected PLD2 in HeLa cells, and we found that it colocalized with wild-type H-Ras at the PM and on tubular endosomes (Figure 1C).

Next, we examined the effect of EGF treatment on Ras activation in live COS-7 cells cotransfected with a RBD-GFP chimera, which binds Ras-GTP specifically (Bivona et al., 2005). Within 30 s of EGF addition, RBD-GFP was recruited from the cytosol onto the PM colocalizing with H-Ras in PM ruffles (Figure 1D and Supplemental Video 1). Shortly after this, these cells were actively ruffling, and, as a consequence, forming macropinosomes (Figure 1D and Supplemental Video 1) as has been reported previously (Bar-Sagi and Feramisco, 1986). H-Ras and RBD were both present at the PM and on the macropinosomes (Figure 1D and Supplemental Video 1), indicating that EGF stimulated Ras-GTP persisted on the macropinosome. In parallel experiments, we monitored biochemically the activation of H-Ras in HeLa cells upon addition of EGF by using the RBD pull-down assay. We found that very little of the overexpressed H-Ras was GTP-bound even when expressed alone, but EGF treatment led to a 10-fold increase in Ras-GTP (Figure 1E). Coexpression of H-Ras with Arf6Q67L did not lead to activation of H-Ras (Figure 1E). Similar results were also obtained with COS-7 cells (data not shown).

H-RasG12V Induces Macropinocytosis, a Stimulated Form of CIE

Our observations that H-Ras was trafficking with the CIE pathway and that EGF treatment activated H-Ras and led to PM ruffling and macropinocytosis led us to examine whether we could also observe this with expression of the active Ras mutant (G12V). Expression of H-RasG12V induced membrane ruffling and macropinocytosis as observed in both fixed and live HeLa cells (Figure 2, A and B, and Supplemental Movie 2). This was also observed in COS-7, MCF7, and Chinese hamster ovary cells (unpublished data). We examined whether H-RasG12V–induced macropinosomes contained CIE cargo such as MHCI, and we found that the macropinosomes contained MHCI but not Tfn (Figure 2A). It was notable that MHCI and H-RasG12V colocalized extensively in the periphery and on incoming macropinosomes (Figure 2A, inset), whereas MHCI also colocalized with Tfn in more centrally located endosomes. Live cell imaging of cells expressing GFP-RasG12V revealed that these incoming macropinosomes were dynamic and turned over with time (Supplemental Video 2). The formation and turnover of macropinosomes and the absence of tubular endosomes is similar to what we had observed previously in cells expressing EFA6A (Brown et al., 2001). These enlarged structures represent macropinosomes as they contained fluid phase markers and an increase in fluid uptake was observed in cells expressing RasG12V compared with cells expressing H-Ras (data not shown).

In resting cells, recycling of CIE membranes back to the PM is dependent on Arf6 activation of PLD (Jovanovic et al., 2006) and actin polymerization (Weigert et al., 2004). Macropinocytosis, in this case, is a form of stimulated CIE. This can be demonstrated by treatment of H-RasG12V–expressing cells with CD to inhibit actin polymerization (Figure 2C) or coexpression of dominant-negative Arf6T27N (Figure 2D), which resulted in the cessation of ruffling and macropinocytosis, and the accumulation of H-RasG12V on tubular, recycling endosomes. Quantification of cells with tubular endosomes revealed that ∼30% of untransfected cells display MHCI tubular endosomes (data not shown); the same proportion of cells with tubular endosomes can be observed in cells expressing GFP-tH, which labels the tubular endosome, and in cells expressing GFP-H-Ras (Figure 2F). Few cells expressing H-RasG12V exhibit tubes; however, CD treatment resulted in the reappearance of Ras on tubular endosomes (Figure 2F). Interestingly, cells coexpressing Arf6Q67L continued to ruffle and form macropinosomes, and H-RasG12V was present both at the PM and on the accumulated vacuoles (Figure 2E). These results suggest that H-RasG12V, like the expression of EFA6 (Brown et al., 2001), changes the membrane architecture of the CIE pathway and shifts the cargo flow from a constitutive to a stimulated form. At the same time, Arf6 activities can influence the cellular distribution of H-RasG12V.

H-RasG12V–induced Macropinosomes Undergo Changes in Phosphoinositide Composition

We previously showed that PIP₂ is lost from incoming EFA6-induced macropinosomes and that this loss is necessary for recycling of macropinosomal membrane back to the PM (Brown et al., 2001). Because H-RasGV12 induces macropinocytosis (Bar-Sagi and Feramisco, 1986) (this study) and activates PI 3-kinase to generate phosphatiylinositol 3,4,5-trisphosphate (PIP₃) (Shields et al., 2000), we examined the distribution of PIP₂ and PIP₃ in living cells. To do this, we expressed chimeric proteins of the PH domain of PLCδ fused to the green fluorescent protein (PH-PLCδ-GFP) for PIP₂ and the PH domain of Akt fused to RFP (PH-Akt-RFP) for PIP₃ in cells expressing H-RasG12V. The PH domain of Akt can also recognize phosphatidylinositol (3,4)-bisphosphate [PI(3,4)P₂], but here, for simplicity, we will refer to it binding to PIP₃. We found that the two PH domains colocalized at the PM, on membrane ruffles and on newly formed macropinosomes (Figure 3 and Supplemental Video 3). Although both PH domains were subsequently lost from the macropinosome, PH-PLCδ-GFP was always released before the release of PH-Akt-RFP (Figure 3 and Supplemental Video 3), suggesting that PIP₂ was lost from the macropinosome before PIP₃. In both cases the “loss” of PIP₂ and PIP₃ could be due to the conversion to another PIP form or destruction by hydrolysis. We quantified the frequency with which this sequential release of PH domains was observed, and we found that this change was observed in 75% of incoming macropinosomes that were observed forming within 2 μm from the PM (60 of the 79 macropinosomes that began with both PH domains, lost PIP₂ followed by the loss of PIP₃; scored in five cells over a 5–10-min period). Although some macropinosomes did not change over the course of imaging, none of the observed macropinosomes lost PIP₃ before PIP₂ or simultaneously released both PH domains.

Rab5 Is Recruited onto Macropinosomes after the loss of PIP₂

Rab5 is a known effector of Ras proteins and both have been implicated in the regulation of endocytosis (Tall et al., 2001). Ras proteins can induce Rab5 activation through direct binding of Rin1, a Rab5 GEF (Tall et al., 2001). To examine whether Rab5 was associated with the CIE pathway, we expressed wild type GFP-Rab5 with H-Ras or H-RasG12V in HeLa cells. When expressed with H-Ras, Rab5-positive endosomes were distributed throughout the cell with some localized to the juxtanuclear region (Figure 4A). The expression of H-Ras had no effect on Rab5 distribution, and the two GTP binding proteins did not colocalize. By contrast, in cells expressing H-RasG12V, Rab5 was recruited onto newly formed macropinosomes in the periphery (Figure 4A). However, we found that the Rab5 effector, EEA1 was not recruited onto peripheral macropinosomes suggesting that phosphatidylinositol-3-phosphate (PI3P) was absent from nascent macropinosomal membranes (Supplemental Figure 1). Given that the macropinosomes lacked Tfn and other clathrin-dependent cargo, these results suggest that H-RasG12V recruits Rab5 onto the CIE pathway. Having observed Rab5 recruitment onto macropinosomes in cells expressing H-RasG12V, we examined whether this could be observed in live H-Ras–expressing cells treated with EGF. Before EGF, GFP-Rab5 localized to perinuclear structures as seen in Figure 4A. Addition of EGF led to the recruitment of GFP-Rab5 onto newly formed macropinosomes in H-Ras–expressing cells (see Supplemental Figure 2 and Supplemental Video 10).

Rab5 recruitment on to membranes derived from CIE was intriguing to us, because Rab5 has been described as a component present on classical early endosomes that represent the convergence point for clathrin-dependent and clathrin-independent cargo (Naslavsky et al., 2003, 2004). Furthermore, we do not observe Rab5 colocalizing with Arf6 or recruited to the Arf6Q67L vacuoles under any condition (Naslavsky et al., 2003, 2004). Therefore, we suspected that Rab5 recruitment onto macropinosomes by H-RasG12V would be dependent on the lipid composition and that PIP₂ loss would be required before Rab5 recruitment. We coexpressed PH-PLCδ-CFP, GFP-Rab5, and an untagged H-RasG12V in COS-7 cells and, using live cell imaging, followed the dynamics of PIP₂ and Rab5 recruitment to newly formed macropinosomes. PH-PLCδ-CFP labeled the PM and incoming macropinosomes, and then it was subsequently released (Figure 4B and Supplemental Video 4), as we observed previously (Figure 3 and Supplemental Video 3). Between 10 and 20 s after PIP₂ release, Rab5 was recruited onto the macropinosome (Figure 4B and Supplemental Video 4). We found that PIP₂ loss was followed by Rab5 recruitment for 63% of the incoming macropinosomes (40 of 63 macropinosomes observed in 7 cells). Only newly formed (<2 μm from PM) and PIP₂- but not Rab5-associated macropinosomes were counted. None of the observed macropinosomes demonstrated the opposite trend (i.e., Rab5 recruitment before PIP₂ loss or Rab5 and PIP₂ simultaneously present on incoming macropinosomes). These results demonstrate the Rab5 recruitment to the macropinosome by H-RasG12V is dependent on PIP₂ loss. We next examined whether PIP₃ was also released from macropinosomes before Rab5 recruitment by using GFP-Rab5 and PH-Akt-RFP. We found that Rab5 recruitment occurred before PH-Akt loss, suggesting that PIP₃ or PI(3,4)P₂, unlike PIP₂, was permissive for the association of Rab5 with macropinosomes (Figure 4C and Supplemental Video 5).

Macropinosome Maturation Provides Three Distinct Signaling Platforms for H-Ras

Our observations suggest that macropinosomes undergo distinct changes in protein and lipid composition after they enter cells. We can detect three separate stages of H-Ras–associated macropinosomes: PIP₂/PIP₃, PIP₃/Rab5, and Rab5 (Figure 5A). Because the incoming macropinosome contains PIP₂, and H-Ras traffics with this Arf6-associated, CIE pathway, we wondered whether expression of Arf6Q67L would capture this early signaling platform of H-Ras G12V (stage I). H-RasG12V localized to the PM and vacuoles in cells expressing Arf6Q67L (Figure 2E), and RBD was recruited onto both the PM and the vacuoles (Figure 5B), indicating that Arf6Q67L provides a suitable environment for active H-Ras to recruit its effectors.

Activation of Ras also leads to phosphorylation and activation of kinases including Erk and Akt. Antibodies detecting endogenous phospho-Erk and phospho-Akt also colocalized with H-RasG12V on these structures. Rab5, however, was not observed on these structures (Figure 5B) probably due to the presence of PIP₂ (data not shown). By contrast, although wild-type H-Ras localized to the Arf6Q67L vacuoles, it was not activated there, because the RBD remained cytosolic (Figure 5C), and H-Ras-GTP was barely detectable by RBD pull-down assay (Figure 1E). Consistent with this, neither endogenous phospho-Erk nor phospho-Akt was observed recruited to H-Ras present on the vacuoles (Figure 5C). However, we detected an increase in phospho-Erk staining in cells expressing Arf6Q67L and wild-type H-Ras (compare phosporylated [p] Erk staining in Figure 5C to pErk staining of untransfected cell marked with an arrow in Figure 5B) that was also seen in cells expressing Arf6Q67L alone (data not shown). This ability of Arf6-GTP to activate Erk has been observed by others (Tague et al., 2004; Robertson et al., 2006). However, in the presence of H-RasG12V phospho-Erk staining was enhanced and localized to both the PM and the Arf6Q67L vacuoles, which suggests that H-RasG12V activates Erk on the vacuoles (Figure 5B). These results show that Arf6Q67L captures macropinosomes at a maturation stage where active H-Ras recruits certain effectors, such as Erk and Akt, but not others, such as Rab5 that require further maturation of the macropinosome.

Macropinosomes Generated through Activation of H-Ras or Arf6 Share Similarities in Phosphoinositide Composition and Differences in Effector Recruitment

The ability of Arf6Q67L to trap H-Ras-stimulated macropinosomes in the first stage of maturation (Figure 5A) led us to reexamine the phosphoinositide changes that occur when we activate Arf6 by expression of EFA6. We previously showed that expression of EFA6 stimulates ruffling and macropinocytosis and that loss of PIP₂ from the incoming macropinosome was required to recycle the membrane back to the PM (Brown et al., 2001). In cells expressing EFA6, both PH-PLCδ-GFP and PH-Akt-RFP were observed at the PM and on incoming macropinosomes and the PLCδ PH domain was lost from the macropinosome before loss of the Akt PH domain (Figure 6A and Supplemental Video 6), again suggesting that loss of PIP₂ preceded the loss of PIP₃ similar to what we observed with activated H-Ras. We quantified the frequency with which this sequential release of PH domains was observed, and we found that this change was observed in 73% of incoming macropinosomes that were observed forming within 2 μm from the PM (13 of the 17 macropinosomes that began with both PH domains lost PIP₂, followed by the loss of PIP₃; scored in 9 cells). Although most macropinosomes showed this trend in lipid maturation over the course of imaging, some macropinosomes lost PIP₃ and maintained PIP₂. These macropinosomes were not dynamic and might reflect that under these conditions (EFA6 activation of Arf6), PIP 5-kinase may be continually activated. Indeed, under conditions of high EFA6 expression or coexpression of EFA6 plus Arf6, cells form vacuoles, mimicking the phenotype of cells expressing Arf6Q67L (unpublished observations) demonstrating once again that Arf6Q67L vacuoles are trapped macropinosomes.

Remarkably, in EFA6-expressing cells coexpressing PH-PLCδ-CFP and Rab5-GFP, we also saw that Rab5 was recruited onto the macropinosome subsequent to the loss of PIP₂ (Figure 6B and Supplemental Video 7). We found that PIP₂ loss was followed by Rab5 recruitment for 72% of the incoming macropinosomes (16 of 22 macropinosomes observed in 3 cells). Only newly formed (<2 μm from PM) and PIP₂- but not Rab5-associated macropinosomes were counted. None of the observed macropinosomes demonstrated the opposite trend (i.e., Rab5 recruitment before PIP₂ loss or Rab5 and PIP₂ simultaneously present on incoming macropinosomes).

The similarity of the Ras-GTP and Arf6-GTP-induced macropinosome phosphoinositide composition was unexpected given that Ras activates PI 3-kinase and Arf6 activates PIP 5-kinase. This suggests that macropinocytosis induced through different mechanisms involves changes in both PIP₂ and PIP₃ during maturation and the subsequent recruitment of Rab5. In contrast, one would expect that H-Ras would recruit some specific effectors to macropinosomes that Arf6-GTP would not. Having used the PH domain from Akt, an effector of H-Ras, for detecting PIP₃, we next examined whether full length Akt could be observed on incoming macropinosomes generated by activation of H-Ras or Arf6. In cells expressing RFP-H-RasG12V, Akt1-GFP colocalized with Ras at the PM, and it was present for a time on the incoming macropinosome and then lost (Figure 7A and Supplemental Video 8) similar to the observed loss of the PIP₃ from the macropinosome. In cells expressing EFA6, in contrast, Akt1-GFP was present on the ruffling PM but not on the incoming macropinosome labeled with RFP-tH (Figure 7B and Supplemental Video 9). Together, this demonstrates that the stimulated macropinocytosis induced by activation of Arf6 or Ras involves similar stages of maturation in terms of phosphoinositide composition and Rab5 recruitment but differences in effector recruitment and activation.