While evidence is accumulating that phosphoinositide signaling plays a crucial role in growth factor and hormone receptor down-regulation, this signaling pathway has also been proposed to regulate endosomal membrane transport and multivesicular endosome biogenesis. Here, we have followed the fate of the down-regulated EGF receptor (EGFR) and bulk transport (fluid phase) markers in the endosomal pathway in vivo and in vitro. We find that bulk transport from early to late endosomes is not affected after inhibition of the phosphatidylinositol-3-phosphate (PI3P) signaling pathway, but that the EGFR then remains trapped in early endosomes. Similarly, we find that hepatocyte growth factor–regulated tyrosine kinase substrate (Hrs) is not directly involved in bulk solute transport, but is required for EGFR sorting. These observations thus show that transport and sorting can be uncoupled in the endosomal pathway. They also show that PI3P signaling does not regulate the core machinery of endosome biogenesis and transport, but controls the sorting of down-regulated receptor molecules in early endosomes via Hrs.
Recent studies have highlighted the role of phosphatidylinositol-3-phosphate (PI3P) signaling in growth factor and hormone receptor down-regulation (Simonsen et al., 2001). Additionally, this signaling pathway has also been proposed to regulate endosomal membrane transport and multivesicular endosome biogenesis (Reaves et al., 1996; Fernandez-Borja et al., 1999; Odorizzi et al., 2000; Futter et al., 2001). In addition to PI 3-kinase, candidate effectors of this PI3P signaling pathway include phosphoinositide-binding proteins, including the PI3P 5-kinase Fab1p (Odorizzi et al., 1998), hepatocyte growth factor–regulated tyrosine kinase substrate (Hrs) in conjunction with other class E proteins (Bishop et al., 2002; Raiborg et al., 2002), and members of the SNX family (Haft et al., 1998; Sato et al., 2001). The precise functions of these proteins in protein trafficking and membrane transport are still unclear. More importantly, it is not known to what extent this signaling pathway as such plays a role in transport, protein sorting, or endosome biogenesis. Here, we report that bulk transport from early to late endosomes, including biogenesis of endosomal intermediates and their subsequent fusion with late endosomes, does not depend on PI3P signaling, but that phosphoinositides tightly control the sorting of down-regulated EGF receptor (EGFR) in early endosomes via Hrs.
Results And Discussion
Bulk transport from early to late endosomes was followed with soluble markers that are incorporated nonselectively within the endosomal content, e.g., fluorescent dextran or HRP, whereas the EGFR was used as a reporter for the trafficking of down-regulated receptors. Since solutes may follow more than one endocytic pathways (Nichols and Lippincott-Schwartz, 2001), the fate of endocytosed dextran was first compared with that of down-regulated EGFR after EGF treatment. After 10 min, both markers colocalized in early endosomes (Fig. 1 A), containing the Rab5 effector early endosomal antigen 1 (EEA1) (Simonsen et al., 1998). After subsequent incubation for 45 min (not shown) or 90 min at 37°C, both EGFR and its ligand EGF reached vesicles that contained the late endosomal-specific lipid, lysobisphosphatidic acid (LBPA) (Kobayashi et al., 1998) (not shown) or Lamp1 (Fig. 1 B), a very abundant protein of late endosomes and lysosomes (Aniento et al., 1993a; Griffiths et al., 1988). Similarly, under the same conditions, dextran reached vesicles that contained Lamp1 (not shown) or LBPA (Fig. 1 C). Hence, both solutes and EGFR follow the same pathway to early endosomes and then to late endocytic compartments.
When transport from early to late endosomes is inhibited, HRP is regurgitated into the medium and fails to accumulate intracellularly (Clague et al., 1994; Gu et al., 1997; Mayran et al., 2003). In contrast, when PI3 kinase was inhibited with wortmannin, although HRP internalization was decreased, accumulation was not affected (not shown), suggesting that transport from early to late endosomes does not depend on PI3P signaling. Indeed, PI3 kinase inhibition did not affect transport of dextran to late endosomes containing LBPA (Fig. 1 D, and see quantification in Fig. 3 C) or Lamp1 (see Fig. 3 A). As a control, we confirmed that wortmannin did however lead to the release of EEA1 from early endosomes (Fig. 1 D), and reduced the PI3P content of early endosomal fractions (not shown). To interfere more specifically with PI3P-dependent functions, we used a double FYVE PI3P-binding domain (2xFYVE) that binds PI3P with high specificity (not shown) and inhibits early endosome fusion in vitro (see Fig. 4 B), as expected (Gillooly et al., 2000). When linked to GFP, 2xFYVE colocalized with dextran internalized for 10 min and EEA1 on early endosomes (Fig. 2 A). In agreement with the lack of effect of wortmannin, GFP-2xFYVE did not affect dextran transport to late endosomes (Fig. 2 A, and quantification in Fig. 3 C). Similarly, endocytosed mouse IgGs, a fluid phase marker, were transported to lysosomes and degraded whether or not GFP-2xFYVE was expressed, whereas IgGs accumulated intracellularly when lysosomal degradation was blocked with leupeptin (Fig. 2 B). We thus concluded that PI3P signaling is not involved in bulk endosomal membrane transport.
In marked contrast to our observations with solutes, EGFR endocytosed for 45 min (Fig. 3 C) or even for 90 min after EGF addition failed to reach late endocytic compartments containing LBPA (not shown) or Lamp1 (Fig. 3 A) after inhibition of PI3 kinase with wortmannin. Then, EGFR was trapped in early/recycling endosomes containing the transferrin receptor (TfR) (Fig. 3 A and D, and quantification in E). Solute transport was unaffected in the same cells (Fig. 3 A). Similarly, endocytosed EGF was not transported to LBPA-containing late endosomes in cells expressing GFP-2xFYVE, but was trapped within early endosomes, containing both EEA1 (not shown) and GFP-2xFYVE (Fig. 3 B, and quantification in C). In a previous study, the lysosomal degradation of EGFR preinternalized at 20°C was reported to be insensitive to wortmannin (Futter et al., 2001), perhaps because EGFR had been transported beyond the early endosomal sorting checkpoint before drug addition (Griffiths et al., 1988).
These differences between solute and receptor trafficking prompted us to investigate the role of PI3P signaling in transport from early to late endosomes in more detail. To this end, we used a well-established assay that reconstitutes the formation of the transport intermediates (multivesicular bodies or endosomal carrier vesicles [ECVs]) from donor early endosomes (Aniento et al., 1996). In this assay, the content of early endosomes is labeled with the solute HRP, and early endosomes are prepared. After in vitro incubation, newly formed ECVs exhibit a lower density than donor early endosomal membranes and are conveniently separated by floatation in gradients. The efficiency of vesicle formation is then calculated from the amounts of HRP present in donor endosomes and vesicles formed in vitro. We confirmed with an established fusion assay that ECVs formed in vitro were functional (Aniento et al., 1993a; Gruenberg et al., 1989), since they lost the capacity to fuse with early endosomes (Fig. 4 E, and outline in A) in contrast to donor membranes (Fig. 4 B, and outline in A), but acquired the capacity to fuse with late endosomes (Fig. 4 D, and outline in A).
Using this assay, we found that the in vitro biogenesis of ECVs was not affected by wortmannin or by GST-2xFYVE (Fig. 4 C). Nor was the fusion of ECVs with late endosomes (Fig. 4 D). As a control, we confirmed that both wortmannin and GST-2xFYVE, but not the inactive C215S mutant, inhibited the homotypic fusion of early endosomes (Gruenberg et al., 1989) (Fig. 4 B), as expected, since the Rab5 effector EEA1 was released from endosomes by GST-2xFYVE (Fig. 4 F). These observations thus demonstrate that PI3P signaling is not required for ECV biogenesis in vitro and subsequent docking/fusion with late endosomes.
We then used the same assay, but with the EGFR as a marker instead of HRP. EGF-induced endocytosis of the receptor was allowed to proceed for a few minutes to accumulate EGFR in early endosomes. These were then purified (Fig. 5 A) and used as donor membranes. In a highly selective manner, EGFR was incorporated into newly formed ECVs (Fig. 5 C), in contrast to TfR, which remained in donor membranes (Fig. 5 B). Interestingly, incorporation of EGFR into newly formed ECVs was significantly reduced when GST-2xFYVE, but not the inactive C215S mutant, was added to the assay (Fig. 5 C). GST-2xFYVE was recruited by donor membranes (Fig. 5 D) and to a lesser extent by ECVs depleted in EGFR (Fig. 5 C), consistently with the distribution of PI3P on early endosomes and ECVs (Gillooly et al., 2000). Similarly, EGFR incorporation into newly formed ECVs was also inhibited by wortmannin (Fig. 5 C). The selective inhibition of EGFR, but not HRP, incorporation into ECVs by GST-2xFYVE, but not by the C215S mutant, could be reproduced in vitro using donor membranes containing both endocytosed HRP and EGFR (not shown).
In our in vitro assay, EGFR and HRP were incorporated into newly formed ECVs with a similar efficiency (Fig. S3). Similarly, in vivo values for EGFR and dextran were also in the same range. Clearly, however, sorting and transport in vitro occurred with a somewhat reduced efficiency (∼30%), when compared with the in vivo situation, as expected. More importantly, EGFR sorting was inhibited to the same extent by wortmannin and 2xFYVE in vivo as well as in vitro, whereas these treatments did not affect solute transport. Altogether, these data demonstrate that EGFR sorting into ECVs, but not solute transport, is inhibited when interfering with PI3P signaling.
Wortmannin was previously reported to inhibit MVB morphogenesis by interfering with the membrane invagination process, thereby causing a vacuolation of early and late endosomes (Reaves et al., 1996; Fernandez-Borja et al., 1999; Futter et al., 2001). Under our experimental conditions, however, wortmannin did not affect the appearance of endosomes (Fig. 3), nor the amounts and distribution of LBPA, a marker of late endosome invaginations (Kobayashi et al., 1998) (Figs. S1 and S2).
Hrs is likely to mediate the PI3P-dependent sorting of EGFR into ECVs. Indeed, Hrs contains a FYVE domain, is wortmannin sensitive and is involved in EGF down-regulation, presumably by linking ubiquitinated EGFR with endosomal clathrin (Bishop et al., 2002; Lloyd et al., 2002; Raiborg et al., 2002). Hrs overexpression was previously found either to inhibit (Bean et al., 2000) or not inhibit (Raiborg et al., 2001) the recycling pathway, and to interfere with EGFR and dextran transport from early to late endosomes (Raiborg et al., 2001). After low Hrs-myc overexpression to limit problems associated with high overexpression, EGFR remained trapped in early endosomes containing Hrs-myc, and failed to reach late endosomes (Fig. 6 B, and quantification in A), in agreement with Raiborg et al. (2001). Dextran, however, did not accumulate in Hrs-myc–labeled early endosomes and reached late endosomes containing LBPA. This apparent discrepancy with Raiborg et al. (2001), like perhaps the differential effects of Hrs on the recycling pathway (Bean et al., 2000; Raiborg et al., 2001), may well reflect differences in the levels of Hrs overexpression.
We therefore used our in vitro transport assay to further investigate the role of Hrs in ECV biogenesis. Fig. 6 D shows that cytosol depleted of Hrs by immunoprecipitation failed to support EGFR incorporation into newly formed ECVs (Fig. 6 E, and quantification in C). Cytosol depleted of Rab7 (not shown), used as a control, was without effect (Fig. 6, C and E). In contrast, the in vitro biogenesis of ECVs labeled with HRP was only marginally affected by Hrs immunodepletion (Fig. 6 C). This agrees with the findings that, despite defective membrane invagination, solute transport occurs along the degradation pathway of hrs mutant Drosophila cells at a rate similar to wild-type cells (Lloyd et al., 2002). Altogether, our in vivo and in vitro data indicate that Hrs, like PI3P signaling, regulates EGFR, but not solute, trafficking along the degradation pathway, strongly suggesting that Hrs mediates PI3P-dependent EGFR sorting into ECVs.
In conclusion, our data show that bulk transport from early to late endosomes in vivo and biogenesis of functional endosomal intermediates in vitro are not affected, when interfering with the PI3P signaling pathway or with Hrs functions. In marked contrast, EGFR then remains trapped within early/recycling endosomes, and fails to be transported to late endosomes in vivo or to be incorporated into newly formed endosomes in vitro. Similarly, the ubiquitination machinery, which regulates EGFR sorting, does not play a role in solute transport along the degradation pathway (van Kerkhof et al., 2001). Our data thus show that membrane transport and receptor sorting can be uncoupled in the endosomal pathway. They also demonstrate that, although PI3P signaling does not regulate the core machinery of endosome transport, it controls the sorting of down-regulated receptor molecules in early endosomes, presumably via Hrs.
Materials And Methods
Wortmannin, EGF, and mouse IgG were from Sigma-Aldrich; fugene and leupeptin were from Roche Diagnostic; anti-TfR antibody was from Zymed Laboratories; anti-EEA1 antibody was from BD Bioscience; EGF-biotin and 10-kD rhodamine-dextran were from Molecular Probes; and rhodamine-labeled or AMCA-labeled anti–mouse IgGs were from Jackson ImmunoResearch Laboratories. pGEX-2xFYVE and pGEX-2xFYVEC215S have been described (Gillooly et al., 2000), as well as anti-LBPA antibodies (Kobayashi et al., 1998), and anti-Hrs antibodies and Hrs-myc (Raiborg et al., 2001). EGFR-GFP was from Alexander Sorkin (University of Colorado, Denver, Colorado).
In vivo experiments
When indicated, BHK-21 (Gruenberg et al., 1989) and HeLa cells (Rojo et al., 1997) were transfected with fugene 36 h before the experiment. To label early or late endosomes, cells were incubated for 10 min at 37°C with 3 mg/ml rhodamine-dextran (pulse) or further incubated without the marker (chase) for the indicated time period, with or without 100 nM wortmannin (for long incubations, the drug was always added freshly after 30 min). Alternatively, cells were first incubated for 1 h at 4°C with EGF-biotin and streptavidin-phycoerythrin. The marker was endocytosed with the same pulse–chase protocole as mentioned here. For quantification, cells expressing or not expressing GFP-2xFYVE were separately incubated with EGF-biotin/streptavidin-phycoerythrin (as described here), or with rhodamine-dextran for 10 min, followed by a 45- or 60-min chase, and processed for immunofluorescence. When indicated, cells were incubated for 16 h at 37°C with 40 µg/ml mouse IgG with or without 50 µg/ml leupeptin. An EGFR-GFP cell line was obtained after transfection of BHK cells using a lentiviral vector (Trono, 2002) with a construct encoding the EGFR fused to GFP (Carter and Sorkin, 1998) under the control of a tetracycline promoter. The clonal population was obtained by limited dilution after a first selection by FACS.
Early and late endosomes were separated by floatation in a step sucrose gradient (Aniento et al., 1993a; Gorvel et al., 1991), and recovered at d ≈ 1.1031g/cm3 (35%/25% sucrose interface) and d ≈ 1.0772g/cm3 (25%/8.5% sucrose), respectively. Cytosol was prepared from rat liver (Aniento et al., 1993b), aliquoted, frozen, and stored in liquid nitrogen. When indicated, the cytosol was depleted of Hrs by immunoprecipitation, or mock-treated with anti-Rab7 as a control (Aniento et al., 1993a). The homotypic fusion activity of early endosomes was measured after mixing in the assay two early endosome populations, which had been separately labeled with endocytosed biotinylated HRP and avidin, respectively (Aniento et al., 1993a; Gruenberg et al., 1989). The mixture was incubated with ATP and cytosol for 45 min at 37°C with or without PI3 kinase inhibitors or 3 μg recombinant protein/100 μg endosomal protein. The avidin–biotinylated HRP (bHRP) complex formed upon fusion was immunoprecipitated with anti-avidin antibodies, and the enzymatic activity of bHRP was quantified and expressed as a percentage of the maximal amount of avidin–bHRP complex formed after detergent solubilization (efficiency).
We measured the formation of ECVs from donor early endosomal membranes, using endocytosed HRP as a marker of the endosomal content, as described (Aniento et al., 1996; Gu and Gruenberg, 2000). In the assay, donor early endosomes were incubated for 30 min at 37°C with ATP and cytosol, with or without drugs or recombinant proteins, as above. ECVs formed in vitro were separated by floatation in a sucrose gradient, and the HRP content of both ECVs and donor membranes was quantified. The efficiency of the reaction was calculated as the percentage of total HRP present in ECVs formed in vitro. To measure the fusion activity of ECVs formed in vitro with late endosomes, ECVs were prepared using biotinylated HRP instead of HRP (Aniento et al., 1996; 1993b). The fusion assay was carried out as described above for early endosomes, except that avidin was endocytosed for 5 min followed by a 40-min chase, and then late endosomal fractions were used in the assay. To study transport of EGFR-GFP in vitro, cells were pretreated with 20 μg/ml brefeldin A for 1 h at 37°C, to deplete early endosomes from Golgi markers (Gu et al., 1997; Rojo et al., 1997). Cells were incubated for 1 h at 4°C with 0.5 μg/ml of EGF and 20 μg/ml brefeldin A, and then for 10 min at 37°C, and fractionated as above. In vitro formation of ECVs was as described here, except that cytosol and donor membranes were reduced 10-fold and the volume of the reaction was adjusted. Donor membranes and ECVs formed in vitro were analyzed by SDS gel electrophoresis and Western blotting.
The 2x FYVE motif was isolated from the pGEMEGFP 2xFYVE (Gillooly et al., 2000) after an EcoRI-SalI digestion and introduced into the peGFPC2. Recombinant GST-2xFYVE and C215S mutant were produced in bacteria and purified (Cavalli et al., 2001). ELISA of LBPA in fractions was as described using Immulon I 96-well plates (Kobayashi et al., 1998). Quantification of protein was according to Bradford (1976). Western blot analysis was carried out using peroxidase-conjugated sheep anti–mouse or goat anti–rabbit IgG as secondary antibodies and detected by chemiluminescence using the ECL reagent (Amersham Biosciences). Immunofluorescence was as described (Rojo et al., 1997).
Online supplemental material
Figs. S1 and S2 illustrate our observations that LBPA is not affected by PI3-kinase inhibition, and Fig. S3 includes detailed quantification of the efficiency of transport measured in vivo and in vitro. All supplemental material is available.
We thank Marie-Hélène Beuchat, Monique Beulet, and Marie–Claire Velluz for expert technical assistance, and Gisou van der Goot for critical reading of the manuscript.
This study was supported by grants from the Swiss National Science Foundation and the Human Frontier Science Programme Organization (to J. Gruenberg). A. Petiot was a recipient of Roche Foundation and ARC (Association pour la Recherche sur le Cancer) fellowships.
The online version of this article includes supplemental material.
Abbreviations used in this paper: bHRP, biotinylated HRP; ECV, endosomal carrier vesicle; EEA1, early endosomal antigen 1; EGFR, EGF receptor; Hrs, hepatocyte growth factor–regulated tyrosine kinase substrate; LBPA, lysobisphosphatidic acid; PI3P, phosphatidylinositol-3-phosphate; TfR, transferrin receptor.