The cellular prion protein (PrPC) is essential for the pathogenesis and transmission of prion diseases. Although PrPC is known to be located in detergent-insoluble lipid rafts at the surface of neuronal cells, the mechanism of its internalisation is unclear, with both raft/caveolae-based and clathrin-mediated processes being proposed. We have investigated the mechanism of copper-induced internalisation of PrPC in neuronal cells by immunofluorescence microscopy, surface biotinylation assays and buoyant sucrose density gradient centrifugation in the presence of Triton X-100. Clathrin-mediated endocytosis was selectively blocked with tyrphostin A23, which disrupts the interaction between tyrosine motifs in the cytosolic domains of integral membrane proteins and the adaptor complex AP2, and a dominant-negative mutant of the adaptor protein AP180. Both these agents inhibited the copper-induced endocytosis of PrPC. Copper caused PrPC to move laterally out of detergent-insoluble lipid rafts into detergent-soluble regions of the plasma membrane. Using mutants of PrPC that lack either the octapeptide repeats or the N-terminal polybasic region, and a construct with a transmembrane anchor, we show that copper binding to the octapeptide repeats promotes dissociation of PrPC from lipid rafts, whereas the N-terminal polybasic region mediates its interaction with a transmembrane adaptor protein that engages the clathrin endocytic machinery. Our results provide an experimental basis for reconciling the apparently contradictory observations that the prion protein undergoes clathrin-dependent endocytosis despite being localised in lipid rafts. In addition, we have been able to assign distinct functions in the endocytic process to separate regions of the protein.
The prion protein (PrP) is the causative agent of the transmissible spongiform encephalopathies (TSEs), such as Creutzfeldt-Jakob disease in humans, scrapie in sheep and bovine spongiform encephalopathy (Prusiner, 1998). In these diseases, the normal cellular form of PrP (PrPC) undergoes a conformational conversion to the infectious form PrPSc. Although the physiological function of PrPC is still unclear, the protein has been implicated in copper metabolism and the cellular response to oxidative stress (Roucou et al., 2004; Vassallo and Herms, 2003). PrPC is attached to the cell surface through a glycosylphosphatidylinositol (GPI) anchor and is localised in cholesterol- and glycosphingolipid-rich lipid rafts (Vey et al., 1996). At the plasma membrane, PrPC can interact with PrPSc in the de novo formation of PrPSc aggregates (Lehmann et al., 1999). Accumulating evidence indicates that the conversion of PrPC to PrPSc preferentially occurs within lipid rafts (Baron et al., 2002; Naslavsky et al., 1997; Taraboulos et al., 1995; Vey et al., 1996). A reduction in the level of cell-surface PrPC through enhanced endocytosis or by shedding of the protein from the membrane (Parkin et al., 2004) might reduce PrPSc production by limiting the amount of PrPC substrate available for conversion (Marella et al., 2002). Thus, the elucidation of the mechanisms involved in the endocytosis of PrPC is central to the normal cell biology of the protein and might be of crucial importance in understanding the pathogenesis of TSEs.
The classical and best-characterised mechanism for the endocytosis of cell-surface proteins is through clathrin-coated pits, with sorting at the cell surface achieved through the direct or indirect binding of the cytoplasmic domains of the protein to clathrin-associated proteins (Kirchhausen, 2000). The adaptor proteins AP2 and AP180 are both major components of clathrin coats. The AP2 heterotetrameric complex binds to phosphoinositides in the membrane, as well as to the cytoplasmic domains of membrane proteins destined for internalisation, and interacts with a range of cytoplasmic proteins including AP180 (Gaidarov and Keen, 1999; Owen et al., 2000). Both AP2 and AP180 bind directly to clathrin and stimulate clathrin cage assembly in vitro (Owen et al., 2000; Ye et al., 2000). However, several GPI-anchored proteins, which lack cytoplasmic domains, appear to undergo endocytosis by a non-clathrin-dependent mechanism (Nichols, 2002; Parton et al., 1994; Pelkmans and Helenius, 2002; Rothberg et al., 1990). Such proteins are internalised through caveolae in a raft-dependent mechanism as shown by the ability to block this process with cholesterol-binding drugs, such as filipin and methyl-β-cyclodextrin (MβCD), presumably as a result of extraction of cholesterol from the rafts/caveolae leading to a loss of their morphological and functional integrity (Nichols, 2003; Schnitzer et al., 1994). However, the mechanism of PrPC internalisation is still controversial, as both raft/caveolae-based (Kaneko et al., 1997; Marella et al., 2002; Peters et al., 2003; Vey et al., 1996) and clathrin-dependent processes (Magalhaes et al., 2002; Shyng et al., 1994; Sunyach et al., 2003) have been described (reviewed by Prado et al., 2004). Although other studies have investigated the fate of endocytosed PrPC, the mechanism of its internalisation was not addressed (Brown and Harris, 2003; Hachiya et al., 2004; Lee et al., 2001).
The N-terminal half of PrPC contains four octapeptide repeats (PHGG(G/S)WGQ; residues 59-90) that preferentially bind Cu2+ ions (Stockel et al., 1998; Viles et al., 1999). The physiological importance of Cu2+ binding to the octapeptide repeats is shown by the finding that exposure of neuronal cells to concentrations of Cu2+ similar to that estimated for the extracellular spaces of the brain results in the rapid internalisation of PrPC (Pauly and Harris, 1998; Perera and Hooper, 2001). This metal-dependent endocytosis can be abrogated by deletion of the octapeptide repeats, mutation of the histidine residues of the two central repeats, or an insertional mutation within the repeats that is associated with an inherited form of human prion disease (Perera and Hooper, 2001).
In the present study, we have addressed the mechanism of the Cu2+-induced internalisation of mammalian PrPC in neuronal cells and investigated the role of discrete regions of the N-terminus of the protein in this process. Both tyrphostin A23 and a dominant-negative mutant of the adaptor protein AP180 that selectively inhibit clathrin-mediated endocytosis blocked the endocytosis of PrPC and transferrin but had no effect on the internalisation of ganglioside GM1. Using both buoyant sucrose density gradient centrifugation in the presence of Triton X-100 and immunofluorescence microscopy, we show that cell-surface PrPC is normally present in detergent-insoluble lipid rafts but, following exposure to Cu2+, a significant proportion of the protein re-localises to detergent-soluble regions of the plasma membrane. Furthermore, by using mutants of PrP, we show that Cu2+ binding to the octapeptide repeats is required to dissociate PrPC from lipid rafts whereas the N-terminal polybasic region is required for its clathrin-mediated endocytosis.
Materials and Methods
PrP constructs and cell culture
Insertion of the coding sequence of murine PrP containing a 3F4 epitope tag into pIRESneo (BD Biosciences Clontech) and the generation of the PrP-Δoct, PrP-ΔN and PrP-CTM constructs have been reported previously (Perera and Hooper, 2001; Walmsley et al., 2001) (E. T. Parkin, N.T.W. and N.M.H., unpublished). For stable transfection of the cDNA encoding the PrP constructs, 30 μg DNA was introduced to SH-SY5Y cells by electroporation and selection was performed in normal growth medium containing 500 μg/ml neomycin selection antibiotic (Gibco BRL). The cDNAs encoding AP180-C in pCMVmyc and AP180-N in pcDNA4 were a gift from H. T. McMahon (MRC Laboratory of Molecular Biology, Cambridge, UK). Cells were incubated with the vector in the presence of Lipofectamine 2000 (Invitrogen) for 24 hours. Transfected cells were detected using an anti-Myc antibody (Sigma-Aldrich) for AP180-C, or an anti-AP180 antibody (Santa Cruz Biotechnology). SH-SY5Y and N2a cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% foetal bovine serum, 50 U/ml penicillin and 0.1 mg/ml streptomycin. Cells were maintained in a humidified incubator at 37°C with 5% CO2.
Cell-surface biotinylation endocytosis assay
Cells at confluency were incubated for 1 hour at 4°C with 0.5 mg/ml Biotin sulfo-NHS (Sigma-Aldrich). Where indicated, cells were pre-incubated with tyrphostin A23 or tyrphostin A63 (Merck Biosciences) for 5 minutes at 37°C. Cells were then incubated for 20 minutes at 37°C in OptiMEM (Gibco BRL) in the presence or absence of 100 μM CuSO4 presented as a histidine chelate (Perera and Hooper, 2001). Prior to cell lysis, PrP remaining on the cell surface was removed by digestion with trypsin as described previously (Perera and Hooper, 2001).
Cell lysates were pre-cleared by incubation for 30 minutes with 0.5% (w/v) protein A-Sepharose. The protein A-Sepharose was pelleted by centrifugation for 1 minute at 13,000 g and the supernatant removed and incubated overnight with 0.5% (v/v) 3F4 antibody (Signet Laboratories) for the transfected SH-SY5Y cell lysates or SAF32 antibody (Spi-Bio) for the untransfected N2a cell lysates or anti-transferrin receptor antibody H68.4 (Zymed Laboratories). Protein A-Sepharose was added to 0.5% (w/v) to the samples and incubation continued at 37°C for 1 hour. Immunocomplexes were pelleted at 13,000 g for 1 minute and the pellet washed three times with 50 mM Tris-HCl pH 8.0, 150 mM NaCl, 0.5% (w/v) sodium deoxycholate, 0.1% (w/v) SDS and 1% (v/v) Nonidet P-40.
SDS PAGE and western blot analysis
Immunoprecipitated biotinylated complexes were mixed with dissociation buffer (125 mM Tris-HCl, pH 6.8, 2% (w/v) SDS, 20% (v/v) glycerol, 100 mM dithiothreitol, Bromophenol Blue) and boiled for 5 minutes. Proteins were resolved by electrophoresis through 14.5% or 10% polyacrylamide gels for PrP or the transferrin receptor, respectively, and then transferred to Hybond-P polyvinylidene difluoride membrane. The membrane was blocked for 1 hour in phosphate-buffered saline (PBS; 1.5 mM KH2PO4, 2.7 mM Na2HPO4, 150 mM NaCl, pH 7.4) containing 5% (w/v) dried milk powder and 0.1% (v/v) Tween-20, followed by incubation with peroxidase-conjugated streptavidin [1:1000 dilution in PBS containing 0.1% (v/v) Tween-20] for 1 hour. Bound peroxidase conjugates were visualised using an enhanced chemiluminescence detection system (Amersham Biosciences). Flotillin-1 and transferrin receptor were detected using an anti-flotillin-1 antibody (BD Biosciences Pharmingen) and antibody H68.4, respectively, with a peroxidase-conjugated rabbit anti-mouse secondary antibody (Sigma-Aldrich).
Cells were seeded onto coverslips and grown to 50% confluency. The fate of cell-surface PrPC was monitored by pre-labelling cells with antibody 3F4 for 30 minutes at 4°C. For endocytosis experiments, cells were pre-incubated with 500 μM tyrphostin A23 and then incubated for 20 minutes at 37°C in OptiMEM in the presence or absence of 100 μM CuSO4 and tyrphostin A23. During this time, cells were also incubated where indicated with 5 μl of a 5 mg/ml solution of Texas-Red-conjugated transferrin (Molecular Probes) or 10 μl of a 10 mg/ml tetramethylrhodamine-labelled 10,000 MW fixable dextran (Molecular Probes). Where required, cells were permeablised in PBS containing 0.1% Triton X-100, fixed with 4% (v/v) paraformaldehyde/0.1% (v/v) glutaraldehyde in PBS for 15 minutes, and blocked overnight in PBS containing 3% (v/v) goat serum (Sigma-Aldrich). Finally, coverslips were incubated with the appropriate fluorescent-probe-conjugated secondary antibodies (Molecular Probes) for 1 hour and mounted on slides using fluoromount G mounting medium (SouthernBiotech). Individual cells were visualised using a DeltaVision Optical Restoration Microscopy System (Applied Precision). Data were collected from 30-40 0.1 μm thick optical sections, and 3D datasets were deconvolved using the softWoRx programme (Applied Precision). The presented images represent individual Z-slices with only one Z-slice per image used for quantitation purposes. Analysis of cell-surface PrPC staining was performed using ImageJ (http://rsb.info.nih.gov/ij/) to measure the intensity of fluorescence around the cell membrane. This was plotted as Pixel Intensity versus Distance around the cell using Microsoft Excel, and then the percentage of cell surface with detectable staining was calculated from multiple images.
Triton X-100 and MβCD treatments
In the Triton X-100 extraction experiments, antibody 3F4-labelled cells were incubated at 4°C for 10 minutes with PBS containing 1% Triton X-100 prior to paraformaldehyde and then incubated with Alexa488-conjugated rabbit anti-mouse antibody. For disruption of rafts by cholesterol extraction, cells at confluency were incubated with 1 mM MβCD for 1 hour at 37°C.
BODIPY FL C5-ganglioside GM1 endocytosis
Cells, unlabelled with antibody 3F4 but pre-incubated with tyrphostin A23 or MβCD, were incubated with 5 μM BODIPY FL C5-ganglioside GM1 (Molecular Probes) in the presence or absence of 500 μM tyrphostin A23 or 1 mM MβCD for 20 minutes at 37°C.
Lipid raft isolation
Harvested cells were resuspended in 2 ml MES-buffered saline (MBS; 25 mM Mes, 150 mM NaCl, pH 6.5) containing 1% Triton X-100 and homogenised by passing 15 times through a Luer 21-gauge needle. After centrifugation at 500 g for 5 minutes, the supernatant was made up to 40% sucrose by adding an equal volume of 80% sucrose in MBS. A 1 ml aliquot of the sample was placed beneath a discontinuous gradient of sucrose consisting of 3 ml of 30% sucrose and 1 ml of 5% sucrose, both in MBS. The samples were then centrifuged at 140,000 g in an SW-55 rotor (Beckman Coulter) for 18 hours at 4°C. The sucrose gradients were harvested in 0.5 ml fractions from the base of the gradient and the distribution of proteins monitored by western blot analysis of the individual fractions.
Tyrphostin A23 blocks the Cu2+-induced endocytosis of PrPC
Previously we have shown that PrPC in the human neuroblastoma SH-SY5Y cell line undergoes rapid Cu2+-mediated endocytosis (Perera and Hooper, 2001). In order to investigate the mechanism of this endocytosis, SH-SY5Y cells stably transfected with the cDNA encoding murine PrPC were incubated in the presence of tyrphostin A23 that has been shown to inhibit selectively clathrin-mediated endocytosis (Banbury et al., 2003; Crump et al., 1998). In order to distinguish surface protein that was endocytosed from protein either already endocytosed or in the secretory pathway, surface proteins were biotinylated with a cell-impermeant reagent prior to incubation of the cells in serum-free medium that contained Cu2+ presented as a histidine chelate (Perera and Hooper, 2001). After incubation of the cells at 37°C for 20 minutes, residual surface PrPC was removed by trypsin treatment prior to lysis of the cells. Any surface biotinylated PrPC that was endocytosed during the course of the experiment was protected from trypsin digestion. In the absence of Cu2+, no endocytosis of PrPC was observed (Fig. 1A, lane 2), whereas 100 μM Cu2+ caused the internalisation of the biotinylated protein (Fig. 1A, lane 4), consistent with our previous results (Perera and Hooper, 2001). Tyrphostin A23 blocked this Cu2+-mediated endocytosis of cell-surface PrPC in a dose-dependent manner (Fig. 1A, lanes 5-7). Tyrphostin A23 also blocked the Zn2+-stimulated endocytosis of PrPC (data not shown).
The endocytosis of the transferrin receptor in the SH-SY5Y cells was also inhibited in a dose-dependent manner by tyrphostin A23 (Fig. 1B), consistent with this protein being endocytosed through clathrin-coated pits (Dautry-Varsat et al., 1983). The specificity of the blockade of clathrin-mediated endocytosis by tyrphostin A23 was shown by the observation that the structurally related tyrphostin A63, which does not inhibit transferrin receptor internalisation (Banbury et al., 2003), failed to block the Cu2+-mediated endocytosis of PrPC (Fig. 1C). To ascertain that endocytosis of PrPC by a clathrin-mediated mechanism was not an artefact of the expression of PrP in the SH-SY5Y cells, the effect of tyrphostin A23 on the internalisation of endogenous PrPC in the mouse neuroblastoma N2a cell line was examined (Fig. 1D). As in the SH-SY5Y cells, tyrphostin A23 blocked the Cu2+-mediated endocytosis of cell-surface PrPC in the N2a cells.
These results were confirmed using immunofluorescence microscopy (Fig. 2A). To monitor directly the fate of cell-surface PrPC during the course of the experiment, SH-SY5Y cells were prelabelled with antibody 3F4 at 4°C in the absence of Cu2+. Following incubation with Cu2+ at 37°C, the cells were fixed, permeabilised and reacted with fluorescently labelled secondary antibody to visualise the distribution of antibody-tagged PrPC. Prior to incubation of the cells with Cu2+, PrPC was localised predominantly at the cell surface. Following incubation of the cells with Cu2+, a significant proportion of intracellular staining for PrPC was observed. However, in the presence of tyrphostin A23, very little intracellular staining was observed upon incubation of the cells with Cu2+; the majority of the PrPC was localised at the cell surface, indicating that tyrphostin A23 was blocking the internalisation of PrPC. Similarly, tyrphostin A23 blocked the constitutive endocytosis of transferrin in the SH-SY5Y cells (Fig. 2A).
In order to ensure that tyrphostin A23 had no effect on raft-based endocytosis, we examined the endocytosis of a fluorescently labelled ganglioside GM1 (Fig. 2B) that is internalised by a raft-based mechanism (Cobbold et al., 2003). Although tyrphostin A23 had no effect on the endocytosis of BODIPY FL C5-ganglioside GM1, the cholesterol-binding drug MβCD completely blocked its endocytosis, consistent with it being internalised by a cholesterol-dependent raft-based mechanism in the SH-SY5Y cells.
A dominant-negative form of AP180 blocks the Cu2+-induced endocytosis of PrP
To confirm that PrPC was being endocytosed by a clathrin-mediated mechanism, SH-SY5Y cells were cotransfected with a construct (AP180-C) of the adaptor protein AP180. This dominant-negative form of AP180 has been shown to inhibit the uptake of epidermal growth factor and transferrin through disrupting the formation of clathrin-coated pits (Ford et al., 2001). Efficient transient transfection of AP180-C into the SH-SY5Y cells was confirmed by immunofluorescence microscopy using an antibody against the Myc tag on the C-terminus of the protein (Fig. 3A). AP180-C completely blocked the Cu2+-mediated endocytosis of PrPC and the internalisation of transferrin (Fig. 3A). By contrast, when cells were cotransfected with the N-terminal domain of AP180 (AP180-N), no such inhibition of endocytosis was observed (Fig. 3A), in agreement with previous results (Ford et al., 2001). To provide a quantitative measure of these observations in a population of cells, the percentage of AP180-C- and AP180-N-transfected cells showing endocytosis of transferrin and PrPC relative to untransfected control cells is given (Fig. 3B). In order to guard against the possibility that AP180-C-transfected cells could not endocytose PrPC because they were senescent and thus not endocytosing by any mechanism at all, we looked at the uptake of dextran in these cells. As can clearly be seen (Fig. 3C), the AP180-C-transfected cells (n=30) in which PrPC internalisation is blocked are still able to take up dextran from the medium. Taken together, these data unequivocally show that, in two neuronal cell lines, PrPC is endocytosed through a clathrin-dependent mechanism and not through a raft-based mechanism.
Cu2+ causes cell-surface PrPC to exit detergent-insoluble rafts and translocate to detergent-soluble regions of the plasma membrane
PrPC in the SH-SY5Y cells is localised in detergent-insoluble lipid rafts (Perera and Hooper, 1999; Walmsley et al., 2003) and yet is endocytosed by a clathrin-dependent mechanism (see above). Either the whole detergent-insoluble raft containing PrPC could be internalised or PrPC moves out of the raft into non-raft regions of the membrane prior to endocytosis. To distinguish between these two possibilities, two complementary approaches were used: (1) cell-surface immunofluorescence and (2) buoyant sucrose density gradient centrifugation in the presence of Triton X-100.
For the immunofluorescence experiments, cells expressing PrPC were pre-incubated with antibody 3F4 at 4°C to label cell-surface PrPC, then incubated at 37°C in the absence or presence of Cu2+. These experiments were performed in the presence of tyrphostin A23 to block endocytosis. To determine the distribution of PrPC between detergent-insoluble rafts and the detergent-soluble regions of the membrane, the cells were incubated with 1% Triton X-100 prior to fixation and immunofluorescence microscopy (Fig. 4A). In the absence of Cu2+, PrPC had a punctate appearance on the cell surface, consistent with that observed for other raft-associated proteins (Ledesma et al., 1999; Parkin et al., 2003; Zeng et al., 2003b). This punctate appearance remained unaltered when cells incubated in the absence of copper were subsequently treated with Triton X-100 prior to fixation, as the detergent only solubilised the non-raft regions of the membrane. Following incubation of the cells with Cu2+, PrPC had a more even distribution over the cell surface. When the Cu2+-treated cells were incubated with Triton X-100 prior to fixation, the surface distribution of PrPC returned to a more punctate appearance, as the PrPC outside of raft regions was solubilised by the detergent. To provide a quantitative measure of this, the percentage of the surface area covered by PrPC was determined as described in the Materials and Methods from the fluorescence images (Fig. 4B). The percentage of the cell surface with PrPC staining in the absence of Cu2+ with or without subsequent Triton X-100 treatment, and in the presence of Cu2+ with or without Triton X-100 treatment, respectively, was then represented graphically (Fig. 4C). In the absence of Cu2+, 49±13% of the cell surface was covered by PrPC and this remained essentially unchanged at 43±7% following treatment with Triton X-100. By contrast, the percentage of the cell surface with PrPC staining increased to 79±7% upon incubation of the cells with Cu2+ and this decreased to 33±9% upon treatment of the cells with Triton X-100.
For the approach using buoyant sucrose density gradient centrifugation in the presence of detergent, cells were first surface biotinylated and then incubated with Cu2+ in the presence of tyrphostin A23 to block endocytosis. The cells were homogenised in the presence of Triton X-100 and subjected to buoyant sucrose density gradient centrifugation (Fig. 5). Rafts, as shown by the position of the raft marker protein flotillin-1, migrated from the 40% sucrose layer to the 5%/30% interface of the gradient owing to their high ratio of lipid to protein, whereas non-raft proteins, as shown by the position of the transferrin receptor, remained at the base of the centrifuge tube. In cells not incubated with Cu2+, biotinylated PrPC was located exclusively at the 5%/30% sucrose interface (Fig. 5A), consistent with the protein residing in detergent-insoluble rafts at the cell surface under basal conditions. However, following incubation of the cells with Cu2+, the majority of the biotinylated PrPC was detected in the detergent-soluble region of the sucrose gradient (Fig. 5B), indicating that it was no longer resident in the detergent-insoluble rafts. This relocation of PrPC to detergent-soluble regions of the membrane in the presence of Cu2+ was not due to non-specific disruption of the rafts, as the raft-associated protein flotillin-1 was detected exclusively at the 5%/30% sucrose interface in both the absence or presence of Cu2+ (Fig. 5A,B). Together, these data suggest that, on exposure of the cells to Cu2+, PrPC moves out of the detergent-insoluble lipid rafts into detergent-soluble regions of the plasma membrane prior to clathrin-mediated endocytosis.
Binding of Cu2+ to the octapeptide repeats dissociates PrPC from lipid rafts, whereas the N-terminal polybasic region mediates its internalisation
Previously, we have shown that a mutant of PrP lacking the octapeptide repeats (PrP-Δoct; Fig. 6A) fails to be endocytosed upon exposure of cells to Cu2+ (Perera and Hooper, 2001). Another mutant, PrP-ΔN (Fig. 6A), which lacks the four residues (KKRP) at the N-terminus of the mature protein, also fails to undergo Cu2+-mediated endocytosis (Fig. 6B). In an attempt to understand the mechanism by which PrPC moves laterally out of rafts prior to clathrin-mediated endocytosis, we examined the surface distribution of PrP-Δoct and PrP-ΔN by immunofluorescence microscopy before Cu2+ treatment, after Cu2+ treatment and following incubation with Triton X-100. Both mutants had a punctate appearance on the cell surface in the absence of Cu2+ (Fig. 6C,D) consistent with their localisation in rafts. However, upon incubation of the cells with Cu2+, only PrP-ΔN and not PrP-Δoct displayed a more even distribution on the cell surface that was reversed upon treatment with Triton X-100 (Fig. 6C,D). Quantification of the amount of cell-surface staining for PrP-ΔN and PrP-Δoct clearly indicated that Cu2+ increased the amount of cell-surface staining for PrP-ΔN but had no significant effect on the amount of cell-surface staining of PrP-Δoct (Fig. 6E).
The raft association of PrP-Δoct and PrP-ΔN following incubation of the cells with Cu2+ was also examined by buoyant sucrose density gradient centrifugation in the presence of Triton X-100 (Fig. 5). In the absence of Cu2+, both mutants were localised exclusively in the raft fractions of the sucrose gradient (Fig. 5A). However, following incubation of the cells with Cu2+, PrP-ΔN, like PrPC, was present predominantly in the soluble region at the bottom of the sucrose gradient (Fig. 5B), whereas PrP-Δoct was still localised in the raft fractions near the top of the gradient (Fig. 5B). Together, these data indicate that binding of Cu2+ to the octapeptide repeats is required to displace PrPC from the detergent-insoluble rafts into the detergent-soluble region of the cell membrane and that the basic residues at the extreme N-terminus of PrPC are required to mediate its internalisation.
Displacement of PrPC from lipid rafts triggers its endocytosis
The above data suggest that Cu2+ actually promotes the endocytosis of PrPC by dissociating the protein from another component within rafts, thereby enabling it to interact through its N-terminal polybasic region with the clathrin endocytosis machinery. We explored this hypothesis in two ways. First, we reasoned that disrupting rafts would result in the endocytosis of PrPC even in the absence of Cu2+. To this end, cells were incubated with MβCD to disrupt the rafts and the endocytosis of PrPC was examined (Fig. 7A). As predicted, in the absence of Cu2+, MβCD treatment caused an increase in the endocytosis of PrPC, a process that was still inhibited by tyrphostin A23. This increased endocytosis of PrPC was relatively modest, as MβCD also promotes the shedding of cell-surface PrPC (Parkin et al., 2004). MβCD had no significant effect on the endocytosis of the transferrin receptor under these conditions (Fig. 7A).
Second, we utilised a construct, PrP-CTM (Fig. 5A), in which the GPI anchor attachment signal in PrPC is replaced with the transmembrane and cytosolic domains from angiotensin-converting enzyme, which lacks a known endocytosis signal (Walmsley et al., 2001). This transmembrane anchored form of PrP when expressed in SH-SY5Y cells displayed a less-punctate surface distribution than PrPC (Fig. 7B), suggesting that it was not completely localised in rafts at the cell surface. This was confirmed following incubation of the cells with either Triton X-100 or MβCD. Whereas solubilisation of the non-raft regions of the plasma membrane with Triton X-100 had little effect on the punctate appearance of PrPC, the surface staining of PrP-CTM was significantly reduced (Fig. 7B). When the cholesterol-binding agent MβCD was used to disrupt rafts, the punctuate distribution of PrPC changed to a more diffuse staining pattern, whereas the diffuse staining pattern of PrP-CTM at the cell surface did not change (Fig. 7B). These data indicate that PrP-CTM is not localised in lipid rafts at the cell surface. Consistent with our hypothesis, biotinylated PrP-CTM was rapidly endocytosed from the cell surface in the absence of Cu2+ (Fig. 7C). This Cu2+-independent internalisation of PrP-CTM was blocked by tyrphostin A23 (Fig. 7C), indicating that it was being endocytosed by a clathrin-mediated mechanism.
In this study, we show that PrPC is present in detergent-insoluble lipid rafts at the surface of neuronal cells but that, on exposure to Cu2+, the protein moves laterally out of the rafts into detergent-soluble regions of the plasma membrane prior to internalisation by clathrin-mediated endocytosis. Furthermore, we show that binding of Cu2+ to the octapeptide repeats is required to dissociate PrPC from the lipid rafts, whereas the polybasic region at the N-terminus of the mature protein is required to mediate its clathrin-dependent internalisation. Our data thus provide an experimental basis for reconciling the apparently contradictory observations that, although localised in lipid rafts, PrPC undergoes clathrin-dependent endocytosis and, in addition, assigns separate functions to distinct regions of the protein.
We investigated the mechanism of PrPC endocytosis by the novel approach of selectively blocking clathrin-mediated endocytosis. Tyrphostin A23 has been shown to disrupt specifically the interaction between tyrosine motifs (YXXΦ where Φ represents a bulky hydrophobic residue) in the cytosolic domains of integral membrane proteins and the medium chain (μ) subunits of adaptor complex AP2 that links to the clathrin coat (Banbury et al., 2003). The dominant-negative C-terminal fragment AP180-C blocks clathrin-coated pit formation and inhibited uptake of epidermal growth factor and transferrin, whereas the N-terminal fragment AP180-N had no such inhibitory effect (Ford et al., 2001). The observation that both tyrphostin A23 and AP180-C block the endocytosis of PrPC provides unequivocal evidence that clathrin-coated pits are involved in the Cu2+-stimulated internalisation of PrPC. During the course of the present study, Sunyach et al. used electron microscopy to show that PrPC was constitutively endocytosed in the absence of Cu2+ by clathrin-coated pits in primary neurons and N2a cells as a result of its predominant colocalisation with the transferrin receptor rather than with the GPI-anchored Thy-1 (Sunyach et al., 2003). Previously, chicken PrP had been localised to clathrin-coated pits and vesicles by electron microscopy (Shyng et al., 1994).
By contrast, it has been reported that PrPC is endocytosed by a caveolin-dependent pathway based on the observation that the cholesterol-binding agent filipin prevented its Cu2+-stimulated endocytosis (Marella et al., 2002). However, this conclusion is confused by the observation that filipin also promotes the shedding of PrPC (Marella et al., 2002; Parkin et al., 2004). Another study concluded that PrPC is endocytosed through a caveolae-mediated pathway (Peters et al., 2003), although in this case non-neuronal CHO cells were used. As neuronal cells lack caveolin (Gorodinsky and Harris, 1995; Parkin et al., 1997) and morphologically distinguishable caveolae (Shyng et al., 1994), the relevance of these observations to the in vivo situation remains unclear.
In addition to PrPC, a few other proteins are localised in detergent-insoluble lipid rafts at the cell surface and yet are internalised by a clathrin-dependent mechanism. These include the epidermal growth factor receptor (Mineo et al., 1999) and the anthrax toxin receptor (Abrami et al., 2003). In the latter case, the authors concluded that `further studies will be required to determine whether the anthrax toxin receptor moves laterally out of rafts before endocytosis' (Abrami et al., 2003). Similarly, both the ganglioside-binding cholera toxin and the glycosphingolipid-binding Shiga toxin are found in detergent-insoluble domains but are internalised by clathrin-coated pits (Sandvig et al., 1989; Shogomori and Futerman, 2001). To rationalise their data, Shogomori and Futerman put forward two models for the mechanism of internalisation of cholera toxin in neurons (Shogomori and Futerman, 2001). In the first model, cholera toxin moves out of the detergent-insoluble rafts into clathrin-coated pits. In the second model, the whole raft domain is internalised en bloc by a clathrin-dependent mechanism. By examining the detergent solubility of PrPC following exposure to Cu2+, we have been able to distinguish between these two models. The observation that PrPC becomes detergent soluble prior to endocytosis clearly fits with the first model, but not the second, in which the protein would be expected to remain detergent insoluble during endocytosis. Our observation that PrPC moves out of rafts in the process of being endocytosed is consistent with the observation of Sunyach et al., who used surface biotinylation and sucrose density gradient fractionation, along with electron microscopy colocalisation of PrPC with the transferrin receptor (Sunyach et al., 2003). Indeed, the ability of PrPC to move laterally out of lipid rafts upon exposure of the cells to Cu2+ is consistent with the model of raft structure proposed by Madore et al. (Madore et al., 1999). On the basis of its differential insolubility in non-ionic detergents, PrPC was proposed to occupy a position on the outer edges of rafts in a semi-ordered lipid domain from which it would more readily be able to move into the surrounding detergent-soluble regions of the membrane than if it was in the centre of the raft.
What is the mechanism by which Cu2+ causes PrPC to exit detergent-insoluble rafts and move laterally to detergent-soluble regions of the plasma membrane prior to clathrin-mediated endocytosis? Cu2+ binding to PrPC might depress the affinity of the protein for other raft components or Cu2+ might promote the interaction of PrPC with non-raft proteins that can either directly or indirectly associate with clathrin on the cytosolic face of the plasma membrane, thus facilitating internalisation. The latter possibility was suggested by Pauly and Harris (Pauly and Harris, 1998). However, our observations using the octapeptide deletion mutant and a transmembrane anchored form of PrP argue for the former possibility. As shown by immunofluorescence microscopy and buoyant sucrose density gradient centrifugation in the presence of detergent: (1) PrP-Δoct does not leave the rafts on incubation of the cells with Cu2+; (2) disruption of rafts with MβCD promotes the endocytosis of PrPC; and (3) PrP-CTM, which is not localised in rafts in the plasma membrane, constitutively endocytoses in the absence of Cu2+. This latter observation in particular would not be expected if Cu2+ was required to promote the interaction of PrPC with a protein that engages the clathrin endocytic machinery. It should be noted that the transmembrane and cytosolic domains from angiotensin-converting enzyme used in the PrP-CTM construct (Walmsley et al., 2001) lack recognisable endocytosis signals and that angiotensin-converting enzyme itself does not undergo constitutive endocytosis from the cell surface in the time scale measured here (Warner et al., 2005). Together, these data indicate that binding of Cu2+ to the octapeptide repeats is required to promote the dissociation of PrPC from rafts (summarised in Fig. 8).
Binding of Cu2+ to the octapeptide repeats in PrPC is known to cause a conformational change in the protein (Stockel et al., 1998; Zahn, 2003) and, recently, increasing the concentration of Cu2+ has been shown to cause a transition from His-Cu-His intermolecular interactions to Cu-His intramolecular interactions within the octapeptide repeats (Morante et al., 2004). Thus, one possibility is that PrPC is maintained in rafts through interaction between low levels of Cu2+ bound to the octapeptide repeats and another raft-resident molecule. Upon increasing the Cu2+ concentration, this intermolecular interaction is disrupted and the octapeptide repeats become fully saturated with Cu2+. Alternatively, another region outside of the octapeptide repeats could interact with a raft-resident molecule and, on binding of Cu2+ to the octapeptide repeats, a conformational change in PrPC disrupts this interaction. This second alternative is supported by the observations that PrP-Δoct, which lacks the octapeptide repeats, is localised in rafts and that a raft-targeting determinant is present in the N-terminal region (residues 23-90) of PrPC (Walmsley et al., 2003). This N-terminal region of PrPC was sufficient to mediate raft association when fused to the ectodomain of a non-raft protein (Walmsley et al., 2003). The 23-90 region of the PrPC ectodomain might function as a raft-targeting determinant either by association with another raft protein or by directly interacting with raft-associated lipids. Although PrPC has been reported to bind to several proteins, including neuronal cell adhesion molecules [NCAMs (Schmitt-Ulms et al., 2001)], plasminogen (Fischer et al., 2000), the 37 kDa/67 kDa laminin receptor (Hundt et al., 2001) and stress-inducible protein 1 (Zanata et al., 2002), binding in all cases involves regions C-terminal to residue 90. Recently, it has been reported that PrPC is involved in both cis and trans interactions with NCAM at the neuronal cell surface and that these interactions promote the recruitment of NCAM to lipid rafts (Santuccione et al., 2005). An interaction of PrPC with raft lipids is supported by the report that PrP lacking a GPI anchor can bind to sphingolipid-cholesterol-rich raft-like liposomes and that this binding is markedly reduced by deletion of the 34-94 region of the protein (Baron and Caughey, 2003). The interaction between the N-terminal region of PrPC and a raft resident molecule may play a critical role during the process of prion infection. In support of this, the conversion of PrPC-like proteinase K-sensitive PrP (PrP-sen) to PrPSc-like proteinase K-resistant PrP (PrP-res) by exogenous PrP-res required a GPI-independent, rather than a GPI-directed, interaction of PrP-sen with sphingolipid-cholesterol-rich raft-like liposomes (Baron and Caughey, 2003).
Although other studies have clearly shown the importance of the N-terminal region of PrP in its endocytosis (Pauly and Harris, 1998; Perera and Hooper, 2001; Shyng et al., 1995), using different mutants of PrP we have shown that the region of the protein that is responsible for dissociating PrPC from rafts (the octapeptide repeats) is distinct from the region within the protein required for endocytosis (the N-terminal polybasic region). Deletion of just four residues from the N-terminal polybasic region in PrP-ΔN did not prevent the protein moving laterally out of rafts on exposure of the cells to Cu2+ but did prevent its endocytosis. Deletions of large regions of the unstructured N-terminus of PrPC [residues 27-89 (Kiachopoulos et al., 2004) and residues 23-90, 48-93 and 23-51 (Nunziante et al., 2003)] have been reported to disrupt the internalisation of PrPC, and point mutations within the N-terminal polybasic region disrupted the constitutive endocytosis of PrPC (Sunyach et al., 2003). However, none of these studies discriminated between the function of the polybasic region and that of the octapeptide repeats. As tyrphostin A23 disrupts the interaction between tyrosine motifs in the cytosolic domains of integral membrane proteins and the adaptor complex AP2 (Banbury et al., 2003), and PrPC lacks a cytoplasmic domain (as it is GPI anchored), these data provide evidence that the internalisation of PrPC requires an integral transmembrane protein (Fig. 8). The N-terminal polybasic residues of PrPC bind to the extracellular domain of this putative transmembrane protein, whereas a YXXΦ motif in its cytosolic domain binds to clathrin-adaptor complexes inside the cell. In addition to this crucial role in the clathrin-mediated endocytosis of PrPC, the N-terminal polybasic region has been implicated in dynein-mediated retrograde axonal transport (Hachiya et al., 2004), in nuclear targeting of truncated forms of PrP (Gu et al., 2003), in cellular resistance to oxidative stress (Zeng et al., 2003a), in neuroprotective activities of PrPC (Atarashi et al., 2003; Drisaldi et al., 2004) and in contributing to the action of dominant-negative PrPC alleles that inhibit conversion of PrPC to PrPSc (Zulianello et al., 2000). The identity of the putative transmembrane protein and its role in these other processes await determination.
In conclusion, we have shown that Cu2+ promotes the internalisation of PrPC on neuronal cells through a clathrin-mediated mechanism. Furthermore, we have shown that binding of Cu2+ to the octapeptide repeats is required to promote the dissociation of PrPC from detergent-insoluble lipid rafts and cause it to translocate to detergent-soluble regions of the plasma membrane, whereas the N-terminal polybasic region is required to interact with a transmembrane adaptor protein that couples to the clathrin endocytic machinery.
This work was supported by grants from the Medical Research Council (MRC), the European Union (QLG3-CT-2001-02353) and the Wellcome Trust (Bioimaging Facility, University of Leeds). D.R.T. was in receipt of a studentship from the MRC. We thank H. T. McMahon (MRC Laboratory of Molecular Biology, Cambridge, UK) for the kind gift of the plasmids containing AP180-C and AP180-N, G. Banting (University of Bristol, UK) for discussion on the use of tyrphostins, G. Howell for assistance with the cell imaging, and members of our laboratory for helpful discussions.
- Accepted August 9, 2005.
- © The Company of Biologists Limited 2005