A clathrin-dependent pathway leads to KRas signaling on late endosomes en route to lysosomes

Ras proteins are small GTPases involved in the regulation of important cellular functions such as proliferation, differentiation, and apoptosis (Malumbres and Barbacid, 2003). The activity of Ras depends on its association with guanine nucleotides, being inactive when bound to GDP and active when associated with GTP. Ras proteins have an intrinsic low GTPase activity that is increased by GTPase-activating proteins. The activity of Ras is regulated by extracellular factors that activate receptor Tyr kinases and recruit guanine nucleotide exchange factors to the plasma membrane (PM), promoting the Ras-GDP to Ras-GTP conversion, which induces a conformational change that allows association of Ras with effectors, including Raf1, PI3K, or Ral–guanine nucleotide dissociation stimulator, that become activated.

At the PM, Ras isoforms have distinct locations, which depend on their guanine nucleotide status. Thus, in the GDP conformation, Harvey Ras (HRas) resides in cholesterol-rich domains (Roy et al., 1999; Prior et al., 2001), whereas upon GTP loading, it is recruited to cholesterol-poor domains. In contrast, GDP–neuroblastoma Ras (NRas) is resident in cholesterol-poor domains and moves to cholesterol-rich domains when loaded with GTP. Finally, Kirsten Ras (KRas)–GDP normally resides in cholesterol-poor domains, and no clear lateral segregation has been reported to occur after GTP loading. It is becoming evident that the differential Ras partitioning and nanoclustering within the PM influence the generation and transmission of distinct signal outputs (Hancock, 2003; Tian et al., 2007).

The presence of HRas and NRas in the Golgi complex is not transient, and it has been shown that these isoforms are active on this compartment (Chiu et al., 2002; Bivona et al., 2003; Caloca et al., 2003; Perez de Castro et al., 2004; Quatela and Philips, 2006). In the majority of cell types, activation of Ras isoforms at the PM is fast and transient, whereas its activation at the Golgi is delayed and more sustained (Chiu et al., 2002). However, it has been shown that stimulation of primary or Jurkat T cells induced Ras activation exclusively on the Golgi, and there, activation is dependent on Ca²⁺, phospholipase Cγ, and the guanine nucleotide exchange factor RasGPR1 (Bivona et al., 2003; Caloca et al., 2003; Perez de Castro et al., 2004). HRas targeted to the Golgi apparatus or to the ER retained its full transforming activity, indicating that the signaling required for transformation can also be initiated from internal membranes. However, the signaling pathways activated in each case are slightly different (Chiu et al., 2002). Thus, Ras signal outputs are determined to some extent by the intracellular location from which signaling arises.

It has been shown that Ras can activate endocytosis by directly regulating the Rab5 nucleotide exchange activity of RIN1 (Tall et al., 2001). Ras has also been found in the endocytic compartment (Pol et al., 1998; Howe et al., 2001; Jiang and Sorkin, 2002; Roy et al., 2002; Fivaz and Meyer, 2005; Gomez and Daniotti, 2005; Jura et al., 2006).

The transit of Ras from PMs to endosomes has been well documented for the HRas isoform (Jiang and Sorkin, 2002; Roy et al., 2002; Gomez and Daniotti, 2005; Jura et al., 2006); HRas colocalizes with EGF receptor (EGFR) on early endosomes (EEs), where it engages Raf1 and triggers signaling activity. It has also been reported that endocytosis is required for maximal HRas signal output (Roy et al., 2002). In contrast, KRas is less retained on endosomes, probably as a result of a faster recycling to the PM (Jiang and Sorkin, 2002; Roy et al., 2002). HRas and NRas can be ubiquitylated, which stabilizes their interaction with endosomal membranes. KRas is refractory to ubiquitylation (Jura et al., 2006).

KRas specifically interacts with CaM (Villalonga et al., 2001). In rat hippocampal neurons stimulated by glutamate, KRas recruits Ca²⁺/CaM to the PM, inducing its translocation to Golgi and endosomal membranes (Fivaz and Meyer, 2005). Interestingly, KRas can also be translocated from the PM to the ER, Golgi apparatus, and the outer mitochondrial membrane (Bivona et al., 2006) by a PKC-mediated phosphorylation mechanism. On mitochondria, KRas associates with Bcl-Xl and induces apoptosis.

Several crucial aspects of the intracellular trafficking and signaling of KRas remain unclear. In this study, we have analyzed in detail and for the first time in vivo the trafficking of KRas from the PM along the endosomal compartment. The results reported in this study reveal that KRas is internalized through a clathrin-dependent mechanism and is transported along EEs, late endosomes (LEs), and eventually into lysosomes. KRas is active on the LE and together with the p14–MP1 scaffolding complex constitutes an intracellular signaling platform.

GFP-KRas was transiently transfected in COS-1 cells, and its presence on endosomes was analyzed by confocal microscopy. For all purposes, we chose only cells with similar moderate levels of KRas expression. Fig. 1 A shows that GFP-KRas colocalized with the early endosomal marker EEA1 in starved cells. Colocalization between both molecules was significantly increased when cells were treated with EGF-TRITC for 20 min (Fig. 1, B and C). Image magnification shows clear-cut colocalization in both individual and clustered endosomes (Fig. 1 D).

In vivo, time-lapse video confocal microscopy revealed colocalization of EGF-TRITC with GFP-KRas on EEs at different times after EGF stimulation (Video 1). Moreover, immunocytochemical results also revealed colocalization among GFP-KRas, EGF-TRITC, and the endogenous EGFR (Fig. 1 E). Consistently, this was also confirmed in NIH3T3 cells (clone Wt8) stably expressing EGFR (Fig. 1 F and Fig. S1, A and B). In NIH3T3-Wt8 cells treated with EGF, we also detected a significant increase of GFP-KRas and endogenous KRas together with the EGFR in the endosomal fractions isolated by sucrose gradients (Fig. 1, G and H, respectively). Endosomal fractions were prepared as described previously (Grewal et al., 2000). Colocalization of endocytosed rhodamine-labeled transferrin with GFP-KRas on EEs was also observed (Fig. S1 C). These results indicate that a certain fraction of KRas localizes on EEs after EGF stimulation.

A recent study argued for a relevant role of CaM in the translocation of PM-bound KRas to Golgi and endosomal membranes of hippocampal neurons stimulated by glutamate (Fivaz and Meyer, 2005). Thus, we aimed to analyze the role of CaM in the regulation of KRas translocation to endosomes in COS-1 cells. We first determined the presence of GFP-KRas on EEs in cells pretreated with the CaM antagonist, W13, for 1 h and subsequently stimulated with EGF (Fig. 2 A). Under CaM-inhibitory conditions, we observed clear colocalization of KRas, EGF-TRITC, and EEA1, suggesting that KRas localization on EEs was a CaM-independent process in this cellular type.

It has been reported that PKC phosphorylation on Ser181 of KRas regulates its translocation to intracellular membranes (Bivona et al., 2006; Plowman et al., 2008). To study the effect of PKC phosphorylation on KRas translocation to endosomes, cells were treated with the general PKC inhibitor bisindolylmaleimide (BIM), and the presence of KRas on endosomes was analyzed. Results revealed a clear colocalization among KRas, EEA1, and EGF on intracellular vesicles, indicating that the translocation of KRas to EEs is a PKC-independent process (Fig. 2 B). The endosomal distribution of the pseudophosphorylated mutant YFP-KRas Ser181D or the nonphosphorylatable mutant YFP-KRas Ser181A after EGF treatment confirmed that translocation is independent of Ser181 phosphorylation (Fig. 2, C and D). Although the KRas Ser181D mutant does not bind CaM, KRas Ser181A still retains this ability (Lopez-Alcala et al., 2008). We observed that KRas Ser181A translocates to endosomes in the presence of W13 (Fig. 2 E), confirming that KRas translocation to endosomes is CaM independent. Interestingly, EGF stimulation increases the relative colocalization of KRas (wild-type and Ser181 mutants) with EEA1 in all of the aforementioned conditions (see quantifications in Fig. 2).

Next, we focused on the mechanisms by which KRas is endocytosed. Therefore, we examined whether the clathrin-dependent endocytic system could be involved in KRas endosomal translocation. Fig. 3 A shows the presence of GFP-KRas in clathrin-coated pits in starved cells. When cells were incubated with EGF-TRITC, the colocalization was increased (Fig. 3 C). Fig. 3 E shows that GFP-KRas colocalized with EGFR after 2.5 min of EGF-TRITC stimulation.

Incubation of EGF for 1 h at 4°C allows binding of the ligand to its receptor clustered in coated pits, but internalization remains blocked (Jiang et al., 2003). Thus, after 1 h at 4°C, clathrin detection was performed on starved COS-1 cells expressing GFP-KRas (Fig. 3 B) or on EGF-TRITC–stimulated cells (Fig. 3 D). Results showed some colocalization of clathrin with KRas in nonstimulated cells and an enhanced colocalization of clathrin with KRas and EGF in small punctate structures, corresponding to coated pits.

The possibility that a clathrin-dependent mechanism could participate in the KRas endosomal translocation was investigated in cells transfected with a dynamin-1 mutant (DynK44A) that blocks clathrin- and caveolae-mediated endocytosis (Damke et al., 1994; Oh et al., 1998). In cells expressing this mutant, EGF endocytosis was greatly reduced (Fig. 3 F), and the colocalization of KRas and EEA1 was significantly diminished (Fig. 3 G). These results together with the clear colocalization of KRas and clathrin revealed that KRas internalization is mainly mediated through a clathrin-dependent mechanism.

Because a certain pool of internalized EGFR is sorted from the EE to the late endocytic compartment preceding lysosomal degradation, we aimed to test whether KRas could also follow this late endocytic route. Fig. 4 A confirmed the colocalization between EEA1 and KRas and shows a small fraction of KRas on Rab7-positive LEs.

To improve the detection of KRas on the late endocytic compartment, cells were treated with U18666A, a molecule that inhibits cholesterol transport and alters the trafficking of multivesicular body (MVB)–associated membrane proteins (Liscum and Faust, 1989; de Diego et al., 2002), or with bafilomycin-A1 (B-A1), an inhibitor of vacuolar-type H⁺-ATPase that blocks acidification and protein degradation in lysosomes (van Weert et al., 1995; Recchi and Chavrier, 2006). In COS-1 cells, both drugs greatly enhanced accumulation of GFP-KRas and also the constitutively active mutant KRasG12V (not depicted) on enlarged endosomal structures (1–4 µm in diameter; Fig. 4, B and D). The B-A1–induced phenotype was also studied in living cells expressing GFP-KRas (Fig. S2 A). As KRas has also been found on mitochondria (Bivona et al., 2006; Plowman et al., 2008), we analyzed the possibility that the structures observed in B-A1–treated cells were mitochondria instead of endosomes. In vivo staining with the MitoTracker probe revealed that despite a slight mitochondrial staining of KRas (Fig. S2 B), the fluorescence observed on endosomal membranes was largely higher. Moreover, the presence of two MVB/late endosomal markers (GFP-Rab7 and endogenous lysobisphosphatidic acid [LBPA]) on these enlarged vesicles further confirmed the late endosomal nature of such structures (Fig. 4, C, E, and F). Finally, filipin staining confirmed the accumulation of cholesterol in KRas-positive LEs after U18666A treatment (Fig. S2 C).

Because KRas was detected on LEs, we aimed to study whether it may be targeted to the lysosomal compartment. In COS-1 cells, GFP-KRas colocalized to some extent with EGF-TRITC and the lysosomal marker LAMP1 (Fig. 5 A). We also analyzed KRas intracellular distribution in the presence of B-A1 or the lysosomal protease inhibitor leupeptin. In both cases, the intracellular colocalization of KRas with LAMP1 was significantly increased (Fig. 5, B–D). The same extent of colocalization was also obtained for LAMP2 (not depicted) or cathepsin D (see Fig. 8 D). The accumulation of KRas on the endolysosomal compartment was confirmed in different cell types treated with B-A1 (Fig. S2, D–G). In contrast, GFP-HRas was found to be mainly present on Golgi membranes (Fig. 5, E and F). Similar data were obtained for GFP-NRas (Fig. S3, A and B).

High magnification confocal in vivo imaging enabled a more detailed analysis of the topology of KRas in those LEs. Interestingly, most of the GFP-KRas–expressing cells exhibited a variable number of MVB-like structures (in this study, we do not distinguish between different subpopulations of late endosomal compartments or MVBs) with a KRas staining on intralumenal membranes (Fig. 5 G). Tomographic sectioning allowed 3D reconstruction of such structures (Fig. 5 H and H′). Assuming that the thickness of the optical section is a function of the pinhole diameter, wavelength, refraction index, and numerical aperture for a 488-nm excitation wavelength, the optical section resolution thickness is ∼470–500 nm. Therefore, the size of the MVB-like structures analyzed was higher than 2–3 µm, and the mean intralumenal vesicle diameter was around 500 nm.

All of these results prompted us to study whether KRas targeted to this compartment could undergo lysosomal degradation. For this purpose, we monitored the stability of HA-tagged KRas in control or leupeptin plus B-A1–treated COS-1 cells after EGF stimulation under experimental conditions according to published data (Linares et al., 2007; Fasen et al., 2008; Hill et al., 2008). Results indicated that treated cells showed a significant increase in the stability of KRas as well as for the EGFR, which was used as a control (Fig. 5 J and Fig. S3 G). Similar results were obtained by [³⁵S]Met metabolic labeling experiments (Fig. S3, I–K); data analysis from these experiments shows that after EGF stimulation, the half-life of HA-KRas was 3.3 h, whereas in the absence of stimulation, it was 5.4 h. When cells were treated with lysosomal inhibitors, the half-life in both control and EGF-stimulated cells was 5.1 h and 6.6 h, respectively. In contrast, no changes were detected in KRas stability upon treatment with the proteasomal inhibitor MG132 (Fig. S3 H).

Finally, in an effort to detect the endogenous KRas on the endolysosomal compartment, GFP-Rab7–expressing COS-1 cells were treated with B-A1 and then subjected to immunocytochemical analysis using a pan-Ras antibody (because of the low labeling efficiency of the isoform-specific antibodies). Results revealed a considerable colocalization between Rab7 and endogenous Ras (Fig. 5 I). Under the same conditions but using nontransfected cells, we confirmed the presence of endogenous Ras on vesicles containing endocytosed EGF-TRITC or the endogenous lysosomal protease cathepsin D (Fig. S3, E and F). Thus, according to our results, we assume that the major isoform detected by this antibody in this compartment was the endogenous KRas. Therefore, we show for the first time that upon EGF stimulation, KRas is targeted to the lysosomal compartment, where it can be degraded.

We considered the possibility that the endolysosomal compartment acts as a new platform for KRas activation. Therefore, activation of KRas on LEs in B-A1–treated cells was measured by fluorescence recovery energy transfer (FRET) microscopy using the Raichu-KRas probe (Fig. S4 A; Kiyokawa et al., 2006). The probe was properly targeted to late endosomal membranes as seen in B-A1–treated cells cotransfected with RFP-Rab7 (Fig. S4 B). Also, upon EGF treatment, those LEs were labeled with the cotransfected EGFR–monomeric RFP (mRFP; Fig. S4 C). After 20–40 min of EGF stimulation, a high FRET efficiency of Raichu-KRas was observed on late endosomal membranes and also at the PM (Fig. 6 A and Videos 3 and 4). As a negative control, we used a Raichu-Pak-Rho probe (Yoshizaki et al., 2003), which did not show FRET efficiency on endosomal structures or at the PM (Fig. 6 B and Videos 5 and 6). Results were plotted according to the endosomal relative FRET ratio values measured from both Raichu probes only on endosomal membranes (Fig. 6 C). The FRET efficiency of Raichu-KRas reflects the activation state of the molecule, thus concluding that KRas can be activated on these Rab7-positive structures (Fig. S4 D). Importantly, under these conditions endosomal fusion was still functional. Indeed, we observed GFP-KRas–positive vesicles harboring internalized EGF-TRITC undergoing endosomal fusion (Fig. 6 D and Video 2).

It has been reported that Grb2 and Ras coordinately traffic during EGFR endocytosis (Jiang and Sorkin, 2002). Thus, we examined the recruitment of Grb2 to the endolysosomal compartment where KRas is present. Under B-A1 conditions, starved cells overexpressing Grb2-YFP showed a diffuse cytoplasmic distribution (Fig. 7 A). After 45 min of EGF-TRITC internalization, a high proportion of Grb2-YFP was efficiently redistributed together with the EGFR and its ligand to enlarged endocytic structures (Fig. 7 B). Nonstimulated cells cotransfected with CFP-KRas and Grb2-YFP only exhibited endosomal localization of KRas and EGFR (Fig. 7 C). However, EGF stimulation induced a clear recruitment of Grb2 to KRas-positive endosomes harboring both the EGFR and ligand (Fig. 7 D).

Finally, we investigated whether KRas present on the endolysosomal compartment was able to recruit Raf1. In nonstimulated cells coexpressing YFP-Raf1 and CFP-KRas, the distribution of Raf1 was mainly cytosolic, although some degree of colocalization with KRas-labeled vesicular structures could be identified (Fig. 7 G). Importantly, upon EGF-TRITC stimulation, a striking recruitment of Raf1 to late endosomal membranes positive for LAMP1 was observed (Fig. 7 H and Fig. S5 A). Under the same experimental conditions, when YFP-Raf1 was transfected alone, only a certain degree of colocalization of Raf1 with EGF-TRITC could be detected after 45 min of EGF endocytosis (Fig. 7, E and F). These results demonstrated that KRas, on this endolysosomal compartment, is susceptible to being activated and capable for signaling.

To further link KRas localization on LEs to an established signaling pathway, we analyzed the possibility that KRas could be involved in the MAPK activation on LEs. It has been shown (Wunderlich et al., 2001) that p14, a protein specifically associated with the cytoplasmic leaflet of the LE, interacts with the scaffold protein MP1, which assembles a scaffold complex in the extracellular signal-regulated kinase (ERK) cascade, eliciting a MAPK signal output from this intracellular compartment (Wunderlich et al., 2001; Teis et al., 2002, 2006). We found that after 45 min of EGF stimulation, Xpress-tagged p14 or myc-tagged MP1 colocalized with GFP-KRas (Fig. 8, A and C). This colocalization became more evident after B-A1 or leupeptin treatment on ringlike cathepsin D–positive structures (Fig. 8, B and D).

Next, we tried to study the kinetics of MAPK activation on the endolysosomal compartment by performing pulse-chase experiments with EGF. To determine the subcellular distribution of endogenous activated MAPK, we used an antibody that recognizes only the phosphorylated form of ERK1/2. Using confocal imaging, we analyzed the subcellular distribution of endogenous activated ERK1/2 in relation to the GFP-KRas/p14-positive endosomes after 5, 10, 30, or 60 min of EGF stimulation. Starved nonstimulated cells showed no significant phospho-ERK1/2 (pERK1/2) staining (Fig. 9 A). However, after 5 min of EGF stimulation, pERK1/2 staining could be detected at certain domains of the PM and some nuclei (Fig. 9 B). At this time point, no colocalization of pERK1/2 and KRas/p14-positive endosomes was observed. After 10 min of stimulation, the pERK1/2 signal was still observed at the PM, but partial staining could also be readily detected on GFP-KRas/p14-positive structures (Fig. 9 C). Such colocalization became more evident after 30 min of stimulation (Fig. 9 D). Even after the 60-min chase period, some cells still maintained the endosomal colocalization of pERK1/2, GFP-KRas, and p14 (Fig. 9 E).

We also performed biochemical analysis of ERK1/2 activation in control and leupeptin plus B-A1–treated COS-1 and NIH3T3-Wt8 cells expressing HA-KRas (Fig. 9, F and G). The relative activation of ERK1/2 was analyzed by densitometry (Fig. 9, F and G, graphs), revealing a sustained signaling in cells treated with the lysosomal inhibitors. However, in nontransfected cells, no significant sustained MAPK activation was observed (Fig. S5, B and C). Finally, we observed that the KRas-positive endosomal structures, in which endogenous pERK was detected, colocalized with endocytosed EGF-TRITC and the LysoTracker probe (Fig. 9, H and I). Thus, the major stability of KRas and the more sustained activation of ERK1/2 in cells treated with lysosomal inhibitors suggest that endosomal KRas-mediated MAPK signaling can be down-regulated in later stages of the endocytic pathway.

The results presented in this study indicate that KRas translocates from the PM to the early and late endocytic compartment through a clathrin-dependent pathway and independently of CaM and PKC phosphorylation. Then, from LEs, KRas is eventually targeted to lysosomes. FRET analysis using the Raichu-KRas probe shows that KRas is active on LEs and recruits Raf1 to elicit a MAPK signal output.

The localization of KRas on clathrin-coated pits and the abrogation of its internalization in cells expressing the dynamin K44 mutant indicate that internalization of KRas takes place via a clathrin-dependent mechanism. It possibly occurs as a complex of signaling-related molecules in a similar mode as proposed by Jiang and Sorkin (2002), in which YFP-HRas and CFP-Grb2 together with the EGFR converge on endosomes. GFP-KRas was found on EEs in both resting and EGF-stimulated COS-1 (Fig. 1, A, B, and D–F), NIH3T3 (Fig. S1, A and B), or porcine aortic endothelial cells (not depicted), which is in agreement with previous studies (Jiang and Sorkin, 2002; Fivaz and Meyer, 2005; Gomez and Daniotti, 2007). On this compartment, GFP-KRas was found together with internalized EGF and its cognate receptor.

In this study, we report that a fraction of KRas is continuously endocytosed and sorted to LEs. Upon EGF stimulation, EGFR in complex with Grb2 is endocytosed and transported to LEs together with Grb2-binding proteins such as Sos, resulting in the activation of KRas on LEs. FRET detection kinetics of KRas activation on the late endosomal compartment is similar to that of the EGF-TRITC internalization.

Several studies indicated that KRas may be transported from the PM to intracellular compartments by diffusion mechanisms that rely on electrostatic gradients driving the polybasic domain within the KRas membrane anchor region (Fivaz and Meyer, 2005; Bivona et al., 2006; Gomez and Daniotti, 2007). This mechanism is similar to that described for myristoylated Ala-rich C kinase substrate, a protein that traffics from the PM to lysosomes (Allen and Aderem, 1995). In this study, we demonstrate for the first time in vivo endosomal fusion of GFP-KRas–labeled endosomes, providing new evidence that a KRas membrane–based trafficking also exists. Altogether, all of this evidence indicates that KRas translocation from the PM to endomembranes may proceed through different pathways, by diffusion or vesicular traffic (Fig. 10). The selection of one of these pathways probably depends on the cellular type but also on the specific stimuli and/or the intracellular destination (Ashery et al., 2006).

In this study, we show that enlarged LEs of cells treated with lysosomal perturbing agents allowed the identification of KRas on the perimeter membrane as well as on intralumenal membranes morphologically resembling MVB structures. These results are in agreement with previous data showing that taxol-treated cells redistribute GFP–truncated KRas (a GFP-tagged KRas-specific membrane anchor probe) to MVB-like structures (Apolloni et al., 2000). Moreover, a recent study demonstrates that proteins with cationic domains (like the KRas C-terminal polybasic motif) are preferentially recruited to phosphatidylserine-enriched organelles (such as endosomes or lysosomes; Yeung et al., 2008).

Importantly, although KRas accumulates on/in Rab7/LAMP1/cathepsin D/LysoTracker-positive structures, HRas and NRas remained excluded from this compartment, being mostly located on Golgi and the PM. Interestingly, a recent study showed that Ras levels are modulated after leupeptin treatment in an axin–β-catenin–overexpressing system, suggesting that lysosomal function may participate in the regulation of Ras protein levels (Jeon et al., 2007). In this study, we show that HA-KRas stability is significantly decreased when cells are stimulated with EGF. Interestingly, in EGF-stimulated cells exposed to lysosomal inhibitors, the half-life is increased. These results further support our model that EGF-induced trafficking of KRas leads to lysosomal degradation.

Signaling at the late endocytic compartment has been demonstrated. The EGFR as well as other receptor Tyr kinases colocalize with signaling modules on endocytic compartments and can dictate different signal outputs from distinct subcellular locations (Jiang and Sorkin, 2002; Jiang et al., 2003; Huang et al., 2006; Hisata et al., 2007; Taub et al., 2007; Valdez et al., 2007). Scaffold and adapter proteins organize these signaling modules. Three scaffold proteins, the kinase suppressor of Ras-1 (KSR1), similar expression to fgf genes (Sef), and MP1, are known to assemble ERK complexes on different intracellular compartments. KSR1 regulates the formation of a MAPK signaling unit at the PM upon EGF stimulation (Mor and Philips, 2006). Sef selectively localizes MAPK/ERK kinase (MEK)–ERK complexes to the cytosolic face of the Golgi (Torii et al., 2004). Finally, the MP1 scaffold forms a stable heterodimeric complex with the adapter protein p14 and assembles an ERK cascade on the LE (Wunderlich et al., 2001). Endosomal localization of the p14–MP1 complex is required for the activation of ERK on these structures (Lunin et al., 2004). However, Jiang and Sorkin (2002) showed FRET between CFP-HRas and YFP–Ras-binding domain on endosomal membranes in vivo. Altogether, these studies support the idea that traffic from the PM to the LE is connected with signaling events in the cell.

In this study, we report the targeting of KRas to late endosomal membranes, where it colocalizes with the p14–MP1 scaffolding complex and activates the ERK signaling cascade. In addition, by means of the Raichu-KRas probe, we report activation of KRas on LEs. Because the membrane-anchoring motifs of the intramolecular FRET probes in this study consist of the native KRas polybasic sequence and CAAX moiety, it is likely that these domains represent the minimal membrane-interacting requirement that targets KRas to late endosomal membranes.

The recruitment of Raf1 by KRas on LEs indicates that in this compartment, KRas actively mediates intracellular signal outputs through specific effectors. Interestingly, Corey and Kelley (2007) recently reported an elevated Ras activity in Niemann-Pick disease type C cells. Because Niemann-Pick disease type C cells contain aberrant LEs with morphological features similar to those shown in this study, we speculate that such Ras activation could be explained by an increased KRas accumulation/activity on this aberrant late endosomal compartment.

Whereas a previous study (Roy et al., 2002) showed that endocytosis was not necessary for KRas signaling, in this study, we demonstrate that in addition to a PM signal output, the LE can be envisaged as a novel KRas signaling platform. However, further studies will be required to evidence the physiological role of KRas signaling on LEs as well as the cross talk between other signaling molecules (i.e., p14 and MP1) on this compartment.

Overall, this study sheds new light on the spatiotemporal control of Ras signaling, providing new clues on the distinctive behavior of Ras proteins in the cell and how endosomal traffic orchestrates this behavior. Thus, we propose the LE as a novel KRas signaling platform. From this compartment, KRas triggers functional outputs while en route to lysosomes, where signaling will be attenuated/down-regulated.

HRas, KRas, and Raf1 cDNAs were provided by R. Marais (Institute of Cancer Research, London, England, UK) and subcloned into living color vectors (Clontech Laboratories, Inc.). The GFP-NRas vector was supplied by J. Miyazaki (Osaka University, Osaka, Japan; Niwa et al., 1991). The HA-HRas vector (pCEFLHA) was provided by P. Crespo (Consejo Superior de Investigaciones Científicas, Santander, Spain). Plasmids for mRFP-Rab7 and EGFR-mRFP have been described previously (Itoh et al., 2005; Kawase et al., 2006). The EGFR-CFP, Grb2-YFP, and pEGFP-Rab7 vectors were supplied by A. Sorkin (University of Colorado, Denver, Denver, CO). The pEF–HA-KRas plasmid was obtained from R. Marais. KRas (Ser181A) and KRas (Ser181D) mutants were obtained by PCR with reverse or forward oligonucleotides carrying the appropriate mutations. The Raichu-KRas and Raichu-Pak-Rho probes used in this study have been developed previously (Mochizuki et al., 2001; Yoshizaki et al., 2003). The Xpress-p14 and myc-MP1 constructs were a gift from L.A. Huber (Innsbruck Medical University, Innsbruck, Austria).

Mouse EGF, W13, U18666A, B-A1, and leupeptin were obtained from Sigma-Aldrich; BIM-I and anti–pan-Ras (ab-3; Ras10) antibody were obtained from EMD. EGF-TRITC, rhodamine-transferrin, and the MitoTracker red and LysoTracker DND-99 probes were obtained from Invitrogen. Mouse monoclonal anti-EGFR (clone 225) and anti–clathrin heavy chain (×22) antibodies were obtained from A. Sorkin and F.M. Brodsky (University of California, San Francisco, San Francisco, CA), respectively. The GMAP210 polyclonal antibody was supplied by R.M. Rios (Universidad de Sevilla, Sevilla, Spain). Monoclonal antibodies to EEA1 and LAMP1 were obtained from Transduction Laboratories. The monoclonal antiactin antibody was obtained from ICN Pharmaceuticals, and the polyclonal antibodies against EGFR (clone 1,005), KRas (F234), Rab5, and cathepsin D were obtained from Santa Cruz Biotechnology, Inc. The pERK1/2 and ERK1/2 antibodies were purchased from Cell Signaling Technology. Monoclonal anti-GFP antibody was obtained from Assay Designs. Polyclonal and monoclonal anti-HA antibodies were obtained from Sigma-Aldrich and Roche, respectively. Peroxidase-labeled antibodies and SDS-PAGE molecular weight markers were purchased from Bio-Rad Laboratories. Monoclonal anti-Xpress antibody was obtained from Invitrogen, and monoclonal anti–myc tag clone 9E10 was purchased from Millipore.

COS-1, HeLa, and VERO cells were grown in DME (Biological Industries) containing 10% FBS (Biological Industries), pyruvic acid, antibiotics, and Gln. Porcine aortic endothelial cells were grown in Ham's F12 (Biological Industries) containing 10% FBS, pyruvic acid, antibiotics, and Gln. NIH3T3 cells (clone Wt8) stably expressing 4 × 105 human EGFRs/cell were provided by A. Sorkin. Transient expressions were performed using Effectene (QIAGEN) according to the manufacturer's specifications. Cells were used 12–36 h after transfection and starved at least 12 h before each experiment.

Cells grown on coverslips were fixed with 4% paraformaldehyde, permeabilized with 0.1% Triton X-100, and blocked with 1% BSA in PBS for 15 min at 37°C. Cells were incubated with the indicated primary antibodies and subsequently with secondary antibodies labeled with Alexa Fluor 488, 594, and 647 (Invitrogen) or Cy5 (Jackson ImmunoResearch Laboratories); finally, coverslips were mounted with Mowiol (EMD). The immunofluorescence protocol for endogenous pERK1/2 detection has been previously described (Teis et al., 2002). Confocal images were acquired using a laser-scanning confocal spectral microscope (TCS SL; Leica) with argon and HeNe lasers attached to an inverted microscope (DMIRE2; Leica). Colocalization of GFP-KRas with EEA1 on endosomal structures was quantified using ImageJ software (National Institutes of Health) and the Colocalization Highlighter plugin (P. Bourdoncle, Institute Jacques Monod, Service Imagerie, Paris, France). The Colocalization Highlighter plugin generated an image of colocalized pixels (binary). Then, with the ImageJ process Image Calculator plugin (minimum operation), the values of the colocalized points were converted to the real value of green (green colocalization image). After this, all images were converted into 32 bits, and original thresholds were set with the background as NaN (not a number). Finally, a total cell region of interest was defined, and the integrated density of both the green colocalization image and the 32-bit threshold green image was measured for the specified area. The ratio of these two resulting values was defined as the percentage of green colocalization. Results were normalized to the control condition, and statistical analysis was conducted by a Student's t test.

COS-1 cells were plated on 35-mm glass-base dishes (Asahi Techno Glass Co.) and then transfected with plasmid Raichu–derived vectors (Raichu-KRas and Raichu-Pak-Rho). Starved cells were treated with 20 nM B-A1 overnight and then with 100 ng/ml EGF for 50 min. Images were obtained every 5 min, starting 10 min before and ending 50 min after EGF addition, on an inverted epifluorescent microscope (IX81; Olympus) that was equipped with a cooled charge-coupled device camera (CoolSNAP HQ; Roper Scientific) and controlled by MetaMorph software (MDS Analytical Technologies). For dual-emission ratio imaging of Raichu probes, we used a 440AF21 excitation filter, a 455DRLP dichroic mirror, and two emission filters, 480AF30 for CFP and 535AF26 for YFP (all from Omega Optical, Inc.), as described previously (Mochizuki et al., 2001). Cells were illuminated with a 75-W xenon lamp through a 12% neutral density filter (Olympus) and a 100× oil immersion objective lens. The exposure time was 400 ms when the binning of the charge-coupled device camera was set to 4 × 4 binning. After background subtraction, the ratio image of YFP/CFP was created with MetaMorph software and used to represent the FRET efficiency. The relative FRET values from specified regions of interest corresponding to vesicular structures were averaged for each time point and then plotted on a graph. In the cotransfection experiments, mRFP-Rab7 or EGFR-mRFP was introduced along with plasmid Raichu–KRas. A dichroic mirror (86006bs; Chroma Technology Corp.) was used when RFP fluorescence was monitored with FRET probes. An XF1207 (580AF20) excitation filter and an XF3015 (636DF55) emission filter were used for RFP imaging. After background subtraction, FRET/CFP ratio images were created using MetaMorph software, and the images were used to represent FRET efficiency.

Starved COS-1 cells transfected with HA-KRas were pooled and then plated in 60-mm dishes at 80% confluence. Then, an equal number of dishes were preincubated for 6 h either with 100 µg/ml cycloheximide (CHX; Sigma-Aldrich) alone or in combination with the lysosomal inhibitors leupeptin (15 µM) and B-A1 (20 nM). Immediately after this, a 0-h time point lysate was obtained, and 100 ng/ml EGF was added in all of the remaining dishes. At 3, 6, 15, and 21 h after EGF, additional medium was removed, and cells were washed twice with cold PBS and then frozen at −80°C. Cells were lysed with lysis buffer (20 mM Tris-HCl, pH 7.5, 2 mM EDTA, 100 mM NaCl, 5 mM MgCl₂, 1% Triton X-100, 5 mM NaF, 10% glycerol, and 0.5% 2-mercaptoethanol) containing protease and phosphatase inhibitors. Cleared lysates (10,000 g) were assayed for protein concentration, and 15 µg of protein was subjected to Western blot analysis using specific HA and actin antibodies. HA-KRas signal was determined by densitometry and normalized to the actin levels. Data were analyzed by a Student's t test. Alternatively, in some experiments, metabolic labeling using [³⁵S]Met was performed.

Pulse and chase experiments were performed using starved COS-1 cells nontransfected or transfected with HA-KRas. After 15 h of transfection, cells were stimulated by pulse chase with 100 ng/ml EGF at 37°C for the indicated times. Then, medium was removed, and cells were washed twice with cold PBS and frozen at −80°C. Cells were lysed with lysis buffer (see previous section) containing protease and phosphatase inhibitors. Cleared lysates (10,000 g) were assayed for protein concentration, and 15 µg of protein was subjected to Western blot analysis using anti-MEK1/2, anti–phospho-MEK1/2, anti-ERK1/2, and anti-pERK1/2 antibodies. The phospho-MEK1/2 and pERK1/2 signals were normalized against the MEK1/2 and ERK1/2 signals, respectively. Data were analyzed by a Student's t test.

Fig. S1 shows early endosomal localization of KRas. Fig. S2 shows that the LE to lysosome transition inhibitors induce accumulation of KRas in the late endosomal compartment. Fig. S3 provides further data on the lysosomal targeting and degradation of KRas. Fig. S4 depicts a schematic representation of the Raichu-KRas probe and shows in vivo localization of this probe in B-A1–treated cells together with data on the activation dynamics of KRas on LEs. Fig. S5 demonstrates that KRas recruits Raf1 to late endosomal membranes and also shows further data on the effect of lysosomal inhibitors on MAPK signaling. Video 1 shows in vivo visualization of endosomal GFP-KRas after EGF stimulation. Video 2 shows live imaging of endosomal fusion of GFP-KRas–positive endosomes. Videos 3 and 4 show FRET imaging of Raichu-KRas on LEs. Videos 5 and 6 show FRET imaging of Raichu-Pak-Rho on LEs.

We thank Dr. A. Sorkin for critical comments on the manuscript and for supplying vectors. We are very grateful to Dr. L.A. Huber for revision of the manuscript and for supplying important constructs used in this study. We would also like to thank Drs. M. Kitano and K. Aoki for valuable comments on the paper and discussions.

This study was supported by grants from Ministerio de Educación y Ciencia of Spain (BFU2006-01151/BMC and V-2006-RET2008-0 to C. Enrich, SAF2006-5212 and SAF2003-08339 to O. Bachs, SAF2007-60491 to N. Agell, and BFU2007-67652 to F. Tebar). A. Lu is thankful to Ministerio de Educación y Ciencia of Spain (Formación Profesorado Universitario mobility program) for a short-term fellowship at the laboratory of M. Matsuda.