Activated EGFR (epidermal growth factor receptor) undergoes ESCRT (endosomal sorting complex required for transport)-mediated sorting on to ILVs (intraluminal vesicles) of endosomes before degradation in the lysosome. Sorting of endocytosed EGFR on to ILVs removes the catalytic domain of the EGFR from the cytoplasm, resulting in termination of receptor signalling. EGFR signalling is also subject to down-regulation through receptor dephosphorylation by the ER (endoplasmic reticulum)-localized PTP1B (protein tyrosine phosphatase 1B). PTP1B on the cytoplasmic face of the ER interacts with endocytosed EGFR via direct membrane contacts sites between the ER and endosomes. In the present paper, we review the relationship between ER–endosome membrane contact sites and ILV formation, and their potential role in the regulation of EGFR sorting on to ILVs, through PTP1B-mediated dephosphorylation of both EGFR and components of the ESCRT machinery.
- endoplasmic reticulum (ER)
- endosomal sorting complex required for transport machinery (ESCRT machinery)
- epidermal growth factor receptor (EGFR)
- intraluminal vesicle (ILV)
- multivesicular endosome/body (MVB)
- protein tyrosine phosphatase 1B (PTP1B)
Following EGF (epidermal growth factor)-induced dimerization, activated EGFRs (EGF receptors) are endocytosed. Signalling from the endocytosed EGFR is subject to down-regulation by at least two mechanisms: ESCRT (endosomal sorting complex required for transport)-mediated sequestration on to the ILVs (intraluminal vesicles) of endosomes before lysosomal delivery and degradation and dephosphorylation by protein tyrosine phosphatases, such as PTP1B (protein tyrosine phosphatase 1B). The ESCRT machinery is a series of protein complexes that are thought to act sequentially to irreversibly concentrate ubiquitinated cargo into domains on the perimeter membrane of the MVB (multivesicular endosome/body). The later components of the machinery then initiate the formation of the inwardly budding ILV . EGFR ubiquitination  and the ESCRT machinery have been shown to be necessary for sorting of EGFRs on to ILVs and for EGF-stimulated ILV formation [3,4]. The ESCRT components are soluble proteins/complexes that can be recruited from the cytoplasm. In contrast, PTP1B is anchored in the ER (endoplasmic reticulum) membrane and so interacts with EGFRs on the perimeter membrane of the endosome via direct membrane contacts that form between the ER and endosomes . What is the temporal relationship between interaction of EGFR with the ESCRT machinery and with PTP1B? The ultimate result of ESCRT activity is sequestration of EGFR on ILVs. Once on an ILV, the EGFR would no longer be available to interact with PTP1B at membrane contact sites between the perimeter membrane of the MVB and the ER, and so interaction with PTP1B must occur before ILV formation. However, a number of recent observations suggest that the EGFR–PTP1B interaction and the EGFR–ESCRT interaction may not operate in a simple chronological sequence.
Coated domains are present on MVBs that have membrane contact sites with the ER
Clathrin-containing coats were identified on flattened domains of endosomes more than 10 years ago  and these domains were shown to contain Hrs (hepatocyte growth factor-regulated tyrosine kinase substrate) [7,8], a critical ubiquitinated protein-binding component of ESCRT0. These coats are readily identified by conventional EM (electron microscopy) and are frequently observed on MVBs that have a clearly identifiable contact site with the ER (Figures 1A and 1B). Sometimes the coat is observed immediately adjacent to the contact site (Figure 1A). Although we have not co-stained these coats with Hrs, we have shown that expression of a substrate-trapping mutant of PTP1B, which stabilizes contact sites and causes them to increase in size, does not affect Hrs recruitment to MVBs . Since the clathrin coat and membrane contact sites can be found on the same MVB, together with our previous demonstration that these contacts can form within 15 min of EGF stimulation, and that the interaction between EGFR and PTP1B must occur while EGFRs are still on the perimeter membrane, this all suggests that these contacts form on immature ‘early’ MVBs. The clathrin coat would probably be a physical barrier to the formation of contact sites where the apposing membranes of the ER and MVB are less than 20 nm apart, suggesting either that the clathrin coat forms and disassembles before contact site formation or the clathrin coat forms after contact site disassembly. The clathrin coat may mask components on the MVB membrane required for contact site formation. Disassembly of the clathrin coat may thus allow the contact site to form as depicted in the possible model of the relationship between contact sites and the ESCRT machinery shown in Figure 2, or disassembly of the contact site may allow the clathrin coat to be recruited.
PTP1B is required for EGF-stimulated ILV formation
We have shown that depletion of PTP1B reduced the number of identifiable membrane contacts between the ER and EGFR-containing MVBs, suggesting that the PTP1B–EGFR interaction plays a role in formation of the contact . Membrane contacts between MVBs and the ER were not completely abolished, suggesting that another component(s) may be required. Depletion of PTP1B not only affected contact site formation, but also inhibited EGF-stimulated ILV formation, suggesting a possible relationship between the two processes. This ILV formation is dependent on the ESCRT machinery and the EGFR substrate annexin 1 . The mechanism of action of annexin 1 in EGF-stimulated ILV formation and its relationship with the ESCRT machinery is not clear. How PTP1B promotes ILV formation is also not clear, but it could modulate the activity of components of the ILV formation machinery through dephosphorylation (see below). An additional and untested possibility is that it is the contact site itself that is required for EGF-stimulated ILV formation, possibly providing extra membrane lipid from the ER for the formation of the additional ILVs that are formed upon EGF stimulation. Either possibility suggests a close relationship between the contact site and ESCRT-dependent ILV formation.
Components of the ESCRT machinery are phosphorylated in response to EGF and are then dephosphorylated by PTP1B
A number of ESCRT components become phosphorylated following EGF stimulation. The endosomal Hrs–STAM (signal-transducing adaptor molecule) complex (ESCRT0) becomes tyrosine-phosphorylated following EGF stimulation , although is not phosphorylated directly by the EGFR kinase . Recent data have shown that Hrs and STAM can be dephosphorylated by PTP1B [5,12]. Although it is possible that PTP1B could interact with soluble ESCRT components without the requirement for a membrane contact site, given that early ESCRT components and contact sites are present on the same MVB, it seems likely that the interaction between PTP1B and the ESCRT machinery occurs at the contact site. Hrs phosphorylation depends on Cbl ubiquitin ligase activity , and Cbl-mediated ubiquitination of Hrs at the level of the endosome, which would inhibit binding of Hrs to ubiquitinated substrates , may facilitate the handing on of ubiquitinated cargoes to ESCRTI. Hrs phosphorylation promotes the release of the phosphorylated ubiquitinated protein from the perimeter membrane of the MVB and promotes Hrs degradation . This has been proposed to allow non-ubiquitintated non-phosphorylated Hrs to bind the endosome and facilitate further rounds of cargo sorting. An alternative/additional way to maintain a pool of Hrs at the perimeter membrane of MVB that is competent for cargo sorting may be dephosphorylation by PTP1B and deubiquitination by a deubuitinating enzyme, several of which have been shown to play a role in EGFR sorting at MVBs . If phosphorylation of Hrs–STAM occurs within the clathrin-coated domain and dephosphorylation of Hrs–STAM occurs at the membrane contact site, then clathrin-coated domain assembly/disassembly must precede membrane contact site formation, as suggested in our proposed model in Figure 2. It does, however, remain possible that phosphorylation of Hrs–STAM and their subsequent dephosphorylation at the contact site occur before formation of the clathrin-coated domain. A better understanding of the relationship between phosphorylation/dephosphorylation and ubiquitination/deubiquitination of Hrs–STAM and how these modifications affect recruitment of clathrin or recruitment of the subsequent ESCRT machinery would greatly aid the elucidation of the order of events with respect to the ESCRT machinery and membrane contact sites.
The ESCRT machinery is not required for contact site formation
Whereas ESCRT0 has been identified on clathrin-coated domains, recruitment of ESCRTs I–III most likely occurs after disassembly of the clathrin coat, as these components have not been localized to clathrin-coated domains, and the coat may prevent recruitment of later components. If clathrin coat assembly/disassembly precedes contact site formation, then recruitment of ESCRTs I–III could occur before or concomitantly with contact site formation. A simple way to couple clathrin disassembly to contact site formation would be for the ESCRT machinery itself to play a role in contact site formation. However, two lines of evidence suggest that ESCRT recruitment is not required for contact site formation. We have used a mutant EGFR (15KR-EGFR) in which 15 lysine residues in the kinase domain have been mutated so that, when expressed in porcine aortic endothelial cells that lack endogenous EGFR, this receptor is negligibly ubiquitinated . This receptor is severely impaired in its ability to recruit the ESCRT machinery, is not efficiently sorted on to ILVs and shows enhanced recycling compared with the wild-type receptor . However, MVBs containing 15KR-EGFR clearly form contacts with the ER (Figure 3A), suggesting that ESCRT engagement by the EGFR is not required for contact site formation. Furthermore, in EGF-stimulated cells depleted of Tsg101 (tumour-susceptibility gene 101, a component of ESCRTI), although MVBs are enlarged and contain very few ILVs, contacts with the ER are clearly present (Figure 3B and see Figure 2A in ).
ESCRT activity cannot be complete before contact site formation because the EGFR must be on the perimeter membrane of the MVB to interact with PTP1B at the contact site and ILV formation cannot occur at a contact site, otherwise the ER membrane would be pulled in to the inwardly invaginating bud. At least, localized disassembly of the contact site or movement of the EGFR/ESCRT domain out of the zone of contact must occur before the final ESCRT activity, i.e. ILV formation, is complete.
Many questions remain concerning contact sites between the ER and EGFR-containing MVBs. What is the order of events with respect to PTP1B-mediated dephosphorylation of EGFR and ESCRT components, formation of ESCRT0-containing clathrin-coated domains and ESCRT-mediated ILV formation? What regulates membrane contact site disassembly? Is the role of membrane contact sites between EGFR-containing MVBs and the ER restricted to PTP1B-mediated dephosphorylation of endosome-localized proteins or could they have other functions? For instance, could lipids be transferred via the contact site from the ER to the MVB to provide extra membrane lipid from the ER to make the additional ILVs that we have shown previously occur after EGF stimulation? What is the relationship between the contacts between the ER and EGFR-containing MVBs and those identified by Rocha et al.  that involve the interaction between the cholesterol-sensing protein ORP1L (oxysterol-binding protein-related 1L) and VapA? ORP1L is recruited to late endosomes/lysosomes via Rab7 and, as Rab7 is a marker of mature MVBs and lysosomes, it is unlikely that the formation of the membrane contact sites that we have identified on early MVBs is initiated through ORP1L. How then is the formation of early MVB contact sites regulated? That PTP1B promotes, but is not required for, contact site formation is consistent with the presence of another, as yet unidentified, protein complex at the membrane contact site (protein complex? in Figure 2) that anchors the ER to the EGFR-containing domains on the MVB allowing PTP1B-mediated dephosphorylation of EGFR and ESCRT0 before inward vesiculation .
This work was funded in part by the Medical Research Council [grant number G0801878].
We thank the Institute of Ophthalmology EM unit members for technical help and advice.
Cellular Traffic of Lipids and Calcium at Membrane Contact Sites: A Biochemical Society held at the Snowbird Ski and Summer Resort, Snowbird, UT, U.S.A., 6–9 October 2011. Organized and Edited by Tim Levine (Institute of Ophthalmology, London, U.K.) and William Prinz (National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, MD, U.S.A.).
Abbreviations: EGF, epidermal growth factor; EGFR, EGF receptor; EM, electron microscopy; ER, endoplasmic reticulum; ESCRT, endosomal sorting complex required for transport; Hrs, hepatocyte growth factor-regulated tyrosine kinase substrate; ILV, intraluminal vesicle; MVB, multivesicular endosome/body; ORP1L, oxysterol-binding protein-related 1L; PTP1B, protein tyrosine phosphatase 1B; STAM, signal-transducing adaptor molecule; Tsg101, tumour-susceptibility gene 101
- © The Authors Journal compilation © 2012 Biochemical Society