The N-terminal Region of the Prion Protein Ectodomain Contains a Lipid Raft Targeting Determinant*

The association of the prion protein (PrP) with sphingolipid- and cholesterol-rich lipid rafts is instrumental in the pathogenesis of the neurodegenerative prion diseases. Although the glycosylphosphatidylinositol (GPI) anchor is an exoplasmic determinant of raft association, PrP remained raft-associated in human neuronal cells even when the GPI anchor was deleted or substituted for a transmembrane anchor indicating that the ectodomain contains a raft localization signal. The raft association of transmembrane-anchored PrP occurred independently of Cu(II) binding as it failed to be abolished by either deletion of the octapeptide repeat region (residues 51–90) or treatment of cells with a Cu(II) chelator. Raft association of transmembrane-anchored PrP was only abolished by the deletion of the N-terminal region (residues 23–90) of the ectodomain. This region was sufficient to confer raft localization when fused to the N terminus of a non-raft transmembrane-anchored protein and suppressed the clathrin-coated pit localization signal in the cytoplasmic domain of the amyloid precursor protein. These data indicate that the N-terminal region of PrP acts as a cellular raft targeting determinant and that residues 23–90 of PrP represent the first proteinaceous raft targeting signal within the ectodomain of a GPI-anchored protein.

Lipid rafts represent compositionally and functionally distinct membrane microdomains that serve as the platform for a number of membrane-mediated biological processes including signal transduction and the trafficking and sorting of proteins and lipids (1,2). These domains are enriched in cholesterol and sphingolipids in the exoplasmic leaflet and in phospholipids with saturated acyl chains and cholesterol in the inner leaflet and, as a result, form a liquid-ordered phase within the bilayer (3). Rafts have been defined biochemically by their relative insolubility at low temperature in non-ionic detergents and, hence, are termed detergent-resistant membranes (DRMs) 1 (1,4). Their flotation to low density upon buoyant sucrose density gradient centrifugation allows DRMs to be separated from detergent-soluble membranes and from the detergentinsoluble cytoskeletal fraction (5,6). Based on this criterion of detergent insolubility and flotation in a density gradient, a subset of proteins has been found to associate with rafts, including the glycosylphosphatidylinositol (GPI)-anchored prion protein (PrP) (7). The association of PrP with rafts has been reported to be critical for the conformational conversion of the normal cellular form of the protein (PrP C ) to the pathogenic form (PrP Sc ) (8,9) that is responsible for neurodegenerative diseases such as bovine spongiform encephalopathy and human Creutzfeldt-Jakob disease (10).
Saturated acyl chains covalently linked to the protein appear sufficient for the raft association of both GPI-anchored proteins, where the acyl chains of the anchor interact with lipids in the exoplasmic face of the raft, and doubly acylated proteins, which interact with the cytoplasmic face (1). Proteinaceous raft determinants include the transmembrane (TM) domains of influenza hemagglutinin (11), neuraminidase (12), and CD40 (13). Raft association of hemagglutinin appears to be mediated by interactions between the TM domain and lipids in the exoplasmic face of the bilayer (14). Other proteinaceous raft determinants mediate raft association by interacting with constituent raft proteins and include the caveolin-scaffolding domain on a number of cytosolic and TM proteins that promotes the binding of the proteins to the caveolar coat protein caveolin (15,16), and the sorbin homology domain of the soluble cytosolic c-Cbl-associated protein that binds to the cytoplasmic domain of the constitutive raft protein flotillin (17). That soluble secretory proteins such as thyroglobulin and plasmin are also raftassociated (18,19) raises the possibility of the existence of exoplasmically orientated raft determinants. Evidence supporting such a notion was provided by the recent finding (20) that an extracellularly disposed juxtamembrane region in the epidermal growth factor receptor (EGFR) was sufficient for raft association when fused to the TM and cytoplasmic domains of either EGFR or the low density lipoprotein receptor.
Although PrP is GPI-anchored and, hence, raft-associated, several independent cell-free studies have clearly demonstrated that the interaction of PrP with model membranes, including sphingolipid-cholesterol-rich raft-like liposomes (SCRLs), can occur in a GPI-independent manner (21)(22)(23)(24). Indeed, the conversion of PrP C -like proteinase K-sensitive PrP (PrP-sen) to PrP Sc -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 SCRLs (24). In the present study, we have utilized alternatively anchored forms of PrP expressed in human neuroblastoma cells to identify exoplasmic determinants required for the association of PrP with cellular raft membranes. Our results show that a GPI anchor is not obligatory for the raft association of PrP, which occurred irrespective of both the mode and topology of membrane anchorage. Deletion mutagenesis demonstrated that the N-terminal region of the PrP ectodomain (residues 23-90) was necessary for the raft association of TM forms of the protein. Furthermore, this region was sufficient for raft association when fused to the ectodomain of a non-raft protein and, hence, represents the first example of a raft targeting determinant located within the ectodomain of a GPI-anchored protein.

EXPERIMENTAL PROCEDURES
Generation of the PrP Constructs-The construction of PrP, PrP-CTM, PrP-NTM, and PrP⌬ GPI in pIRESneo, ACE, and MDP-CTM in pIREShyg have been described previously (25,26). DNA encoding PrP-CTM2 was constructed by PCR using the following primers: primer 1, 5Ј-TACGACGGGAGAAGATCCGGTGCAATCATTGGACTC-3Ј; primer 2, 5Ј-CGGGATCCCTAGTTCTGCATCTGCTC-3Ј; and primer 3, 5Ј-ATAAGAAGGCGGCCGCATGGCGAACCTTGGCTAC-3Ј. Primers 1 and 2 were used to generate from APP:pBluescript a primary PCR product comprising a 3Ј BamHI site followed by a sequence encoding the TM and cytoplasmic domains of APP and the last six residues of the PrP ectodomain. The primary PCR product and primer 3 were used to generate from PrP:pBC12/CMV (a gift from Dr. D. Harris, Washington University, St. Louis, MO) the PrP-CTM2 sequence flanked by NotI and BamHI sites which was inserted into the respective sites of pIRESneo (Clontech). The following constructs were cloned into the NotI and BamHI sites of pIRESneo. PrP-CTM⌬23-90 was generated from PrP-CTM in pIRESneo (26), using the following primers: primer 4, 5Ј-ATTATGGGTACCCCCTCCGCAGAGGCCGACATCAGT-3Ј; primer 5, the antisense of primer 4; and primer 6, 5Ј-CGGGATCCTCAGGAGT-GTCTCAGCTC-3Ј. Primers 3 and 4 were used to amplify a 5Ј fragment encoding residues 1-22 of PrP fused to the first six residues following the octapeptide repeat region. Primers 5 and 6 were used to amplify a 3Ј fragment encoding residues 16 -22 of PrP fused to residues 91-254. Both DNA fragments were fused by PCR and amplified using the terminal primers 3 and 6. PrP-CTM⌬51-90 was generated by the same protocol using primer 7, 5Ј-ATTATGGGTACCCCCTCCTGGGTAACG-GTTGCCTCC-3Ј; and primer 8, the antisense of primer 7. Primers 3 and 7 were used to amplify a 5Ј fragment encoding residues 1-50 of PrP fused to the first six residues following the octapeptide repeat region. Primers 6 and 8 were used to amplify a 3Ј fragment encoding residues 44 -50 of PrP fused to residues 91-254. For PrP-MDP-CTM, a 5Ј fragment encoding residues 1-90 of PrP fused to the first six residues of the MDP ectodomain was amplified from PrP:pBC12/CMV using primer 3 and primer 9, 5Ј-CAGGTCCCGGAATTGGTCTTGGCCCCATCCACCG-CC-3Ј. Primers 6 and 10, the antisense of primer 9, were used to amplify from MDP-CTM/pIREShyg a 3Ј fragment encoding residues 17-431 of MDP-CTM fused to the last six residues of the octapeptide repeat region of PrP. Both DNA fragments were fused by PCR and amplified using the terminal primers 3 and 6.
Cell Culture and Transfection-The human neuroblastoma SH-SY5Y cell line was cultured and transfected with the various constructs as described previously (26). To generate cells co-expressing PrP-CTM and MDP-CTM, G418-resistant cells expressing PrP-CTM were transfected with MDP-CTM in pIREShyg and subsequently selected with 0.5 mg/ml hygromycin.
Sucrose Density Gradient Centrifugation-All procedures were carried out at 4°C. As described previously (27), cells at confluency in a 175-cm 2 flask were scraped into Mes-buffered saline (MBS; 25 mM Mes, 150 mM NaCl, pH 6.5) and harvested by centrifugation. Cells were resuspended in 0.5 ml of MBS containing 1% (v/v) Triton X-100 and homogenized by 20 strokes in a Dounce homogenizer. The homogenate was made 40% (w/v) with respect to sucrose by the addition of 0.5 ml of 80% (w/v) sucrose in MBS and loaded at the bottom of a 5/30% sucrose gradient. Gradients were centrifuged at 100,000 ϫ g for 18 h in an SW50.1 rotor and subsequently separated into nine 0.5-ml fractions, with fraction 1 containing the highest density of sucrose.
Immunoprecipitation and Cell Surface Biotinylation-Immunoprecipitation with antibody 3F4 (Signe) was carried out as described previously (26). Cell surface biotinylation was performed as described previously (26). Briefly, cells at confluency were incubated for 30 min at 4°C with PBS containing 0.5 mg/ml biotin sulfo-NHS (Sigma), rinsed three times with PBS containing 50 mM glycine, and lysed in lysis buffer.

A GPI Anchor Is Not Essential for the Raft Association of
PrP-To investigate determinants required for the raft association of PrP, alternatively anchored forms of the protein were stably expressed in the human neuronal cell line SH-SY5Y, which lacks detectable levels of endogenous PrP (33) (Fig. 1A). Cells were homogenized in the presence of Triton X-100 and subjected to buoyant sucrose density gradient centrifugation (Fig. 1B). Clathrin, a marker of detergent-soluble membranes (27), was detected in fractions 1-2 of the gradient, corresponding to the 40% sucrose layer (Fig. 1B). Conversely, flotillin, a characteristic DRM marker protein (34), was detected in fractions 4 -7 of the gradient, corresponding to the 5/30% interface (Fig. 1B). Similarly, wtPrP was predominantly localized in the DRM-enriched fractions 4 -7 of the gradient, consistent with the protein having a GPI anchor and, hence, associating with DRMs ( Fig. 1B).
Surprisingly, the majority of PrP-CTM, in which the GPI anchor had been replaced with the TM and cytoplasmic domains of the type I integral membrane protein ACE (26), was also detected in the DRM-enriched fractions (Fig. 1B). The DRM association of PrP-CTM cannot be attributed to the TM and cytoplasmic domains of ACE, as ACE itself failed to associate with DRMs and remained in the clathrin-containing fractions when stably expressed in SH-SY5Y cells (Fig. 1B), consistent with earlier observations that this TM protein is excluded from rafts (35,36). In addition, PrP-NTM, which lacks a GPI anchor but has an inverse topology to PrP-CTM due to the provision of an uncleaved N-terminal signal anchor, and PrP⌬ GPI, which lacks any type of anchoring domain, (26), were primarily located in the DRM-enriched fractions (Fig. 1B). Flotation to fractions of low density was not an intrinsic property of the PrP ectodomain as secreted PrP⌬GPI failed to float to the gradient interface (Fig. 1B). These results demonstrate that the GPI anchor of PrP was not an absolute requirement for the cellular raft association of the protein, which occurred irrespective of both the mode and topology of membrane anchorage.
Under the conditions used for the isolation of the DRMs, the non-raft proteins clathrin and ACE were excluded from the DRM fractions indicating that association with DRMs is not due to incomplete solubilization of membrane fragments. To confirm that the presence of PrP-CTM in the DRM fractions was due to association with cholesterol-rich rafts and not due to incomplete solubilization of membrane fragments, the cholesterol-depleting agent methyl-␤-cyclodextrin was used to disrupt the rafts prior to isolation of the DRMs (Fig. 1C). Following methyl-␤-cyclodextrin treatment, the majority of PrP-CTM was present in fractions 1-3 of the sucrose gradient indicating that the flotation of this construct was due to its interaction with cholesterol-rich rafts and not due to incomplete solubilization of the membranes.
The Ectodomain of PrP but Not of Another GPI-anchored Protein Mediates Raft Association-The GPI-independent raft localization of PrP-CTM, PrP-NTM, and PrP⌬ GPI, but not ACE (Fig. 1B), strongly implied that the ectodomain of PrP contained determinants sufficient for the raft association of both TM anchored and non-anchored forms of PrP. To determine whether the ectodomain of another GPI-anchored protein could mediate raft localization, the ectodomain of membrane dipeptidase (MDP), a glycosylated, GPI-anchored, raft-associated protein of similar size to PrP (37,38), was fused to the TM and cytoplasmic domains of ACE to create MDP-CTM ( Fig. 2A). SH-SY5Y cells stably co-expressing PrP-CTM and MDP-CTM were subjected to buoyant sucrose density gradient centrifugation (Fig. 2B). PrP-CTM co-localized with flotillin in the DRMenriched fractions (Fig. 2B). Co-expressed MDP-CTM, however, even though it had the same TM and cytoplasmic domains as PrP-CTM and was trafficked to the cell surface (data not shown), failed to incorporate into DRMs to any significant extent and remained in the clathrin-containing fractions of the sucrose gradient (Fig. 2B). These results verify that the PrP ectodomain was responsible for the raft association of PrP-CTM and, in addition, demonstrate that not all GPI-anchored proteins contain raft targeting determinants within their ectodomains, although it should be noted that in a recent study (24) several other phosphatidylinositol-phospholipase C-released proteins, in addition to PrP, were found to bind to SCRLs.
The Raft Targeting Determinants in the Ectodomain of PrP Suppress the Coated Pit Localization Signal in the Cytoplasmic Domain of the Amyloid Precursor Protein-To ascertain the strength of raft targeting determinants in the ectodomain of PrP relative to an antagonistic targeting signal, the TM and cytoplasmic domains of ACE in PrP-CTM were replaced with those of the APP, to generate PrP-CTM2 (Fig. 3A). The cytoplasmic domain of APP contains an endocytosis signal (GY-  (26). In PrP-CTM, the GPI anchor signal sequence of PrP (checkered box) was replaced with the TM (squared box) and cytoplasmic domains (horizontally striped box) of ACE. In PrP-NTM, the GPI anchor signal sequence was deleted, and the N-terminal signal sequence (diagonally striped box) was replaced with the signal anchor (vertically striped box) and stalk region (horizontally dashed box) of aminopeptidase A. In PrP⌬ GPI, residues 229 -254 of PrP, comprising the C-terminal GPI signal sequence, were deleted. All the PrP constructs contained a two-point mutation (L109M and V112M) to generate the 3F4 epitope. The ACE construct used in this study comprises the C-terminal active site domain and the TM and cytoplasmic domains (25). B, SH-SY5Y cells expressing either wtPrP, PrP-CTM, ACE, PrP-NTM, or PrP⌬GPI were homogenized in the presence of 1% (v/v) Triton X-100 and subjected to buoyant sucrose density gradient centrifugation. Conditioned medium from a 48-h incubation of cells expressing PrP⌬GPI (secreted PrP⌬GPI) was also subjected to sucrose density gradient centrifugation in the presence of Triton X-100. Following centrifugation, the gradient was separated into 9 fractions. ENPTY) that causes the internalization of the protein via clathrin-coated pits and vesicles (39,40). Following buoyant sucrose density gradient centrifugation of SH-SY5Y cells stably expressing PrP-CTM2 (Fig. 3B), the majority of the endogenous APP was detected in the clathrin-containing fractions, although a minor proportion of the protein (ϳ5%) co-localized with flotillin in the DRM-enriched fractions, consistent with previous observations (27). In contrast, PrP-CTM2 was localized solely in the DRM-containing fractions (Fig. 3B). The DRM localization of PrP-CTM2 demonstrates that the raft targeting determinants in the ectodomain of PrP were dominant to the clathrin-coated pit targeting motif in the cytoplasmic domain of APP.
The Raft Association of PrP-CTM Requires the N-terminal Region of the PrP Ectodomain-To define the region of the PrP ectodomain responsible for the raft association of PrP-CTM, a deletion construct was created, PrP-CTM⌬23-90, that lacked the sequence immediately following the signal peptide up to residue 90 including the highly conserved octapeptide repeat region involved in binding copper (41) (Fig. 4A). Following cell surface biotinylation, the 3F4-immunoprecipitated construct was effectively recognized by streptavidin conjugated to horseradish peroxidase, indicating that the deletion construct was trafficked to the cell surface (Fig. 4B). When subjected to buoyant sucrose gradient centrifugation in the presence of Triton X-100, the majority of PrP-CTM⌬23-90 failed to associate with DRMs and remained within the clathrin-containing fractions (Fig. 4C), indicating that the targeting determinant necessary for the raft association of PrP-CTM was located within residues 23-90. As PrP-CTM⌬23-90 lacked the copper-binding octarepeat region, we investigated if binding of copper may be in-volved in the raft localization of PrP-CTM. However, deletion of the octapeptide repeat region (residues 51-90) to generate PrP-CTM⌬51-90 or incubation of cells expressing PrP-CTM with the Cu(II) chelator diethylenetriaminepentaacetic acid (DTPA) (42) both failed to abolish the association of the protein with DRMs (Fig. 4D). These data indicate that neither binding of copper to the octapeptide repeat region nor the octapeptide region itself is responsible for the raft localization of PrP-CTM. Although the expression level of PrP-CTM and PrP⌬23-90 both in the cell lysates and at the cell surface following surface biotinylation were similar (Fig. 4B), only the former construct was associated with DRMs (Fig. 4C), indicating that differences in expression level had not influenced the extent of raft The N-terminal Region of PrP Functions as a Raft Targeting Determinant Independently of the C-terminal Globular Domain-To demonstrate that residues 23-90 of the PrP ectodomain were sufficient for the raft association of another protein, this region was fused to the ectodomain of the non-raft TM protein MDP-CTM (Fig. 2) to generate PrP-MDP-CTM (Fig.  5A). SH-SY5Y cells stably expressing PrP-MDP-CTM were subjected to buoyant sucrose density gradient centrifugation (Fig.  5B). As well as being recognized by the anti-PrP antibody P45-66, PrP-MDP-CTM migrated with a larger apparent molecular weight than MDP-CTM (54 kDa as compared with 45 kDa, respectively) consistent with it containing the 23-90 region of PrP. In contrast to the non-raft localization of MDP-CTM (Fig. 2B), the majority of PrP-MDP-CTM co-localized with flotillin in the DRM-enriched fractions (Fig. 5B). Thus, the 23-90 region of the PrP ectodomain was sufficient to confer raft association to a non-raft TM protein and, hence, functions as a raft targeting determinant independently of the C-terminal globular domain of PrP.

DISCUSSION
In this study we have shown that PrP contains a determinant within the N-terminal region of the ectodomain that can mediate lipid raft association in neuronal cells independently of either the GPI anchor or the C-terminal globular domain. The major raft determinant so far identified for exoplasmically oriented proteins is the C-terminal GPI anchor, whose saturated acyl chains appear to intrinsically associate with glycosphingolipids and cholesterol in the outer leaflet of the bilayer (43). Our results, along with the recent finding that raft targeting information is present in the juxtamembrane region of the EGFR ectodomain (20), provide a further example of an exoplasmically orientated raft targeting determinant. Although our findings corroborate a recent cell-free study demonstrating that PrP can bind to SCRLs in a GPI-independent manner (24), they are contrary to previous reports showing that the raft association of PrP was abolished by substitution of the GPI moiety with the TM and cytoplasmic domains of either CD4 or Qa (8,44). One explanation for this discrepancy is that the TM and cytoplasmic domains of CD4, unlike those of ACE and APP used in this study, may contain non-raft targeting determinants that are dominant to the raft targeting determinants in the ectodomain of PrP. In this respect, we have found that rafts containing TM and non-anchored forms of PrP are solubilized by a lower ratio of Triton X-100 to total protein than rafts containing GPI-anchored PrP (data not shown), suggesting that the raft targeting determinant in the ectodomain of PrP may be relatively weak and, in some cases, be superseded by non-raft targeting signals.
The 23-90 region of the PrP ectodomain may function as a raft targeting determinant by directly interacting with raftassociated lipids in a similar manner to the cholesterol binding domain of caveolin (45). Such an interaction is supported by the recent report (24) that hamster PrP lacking a GPI anchor can bind to SCRLs, and this binding is markedly reduced by deletion of the 34 -94 region of the protein. However, other cell-free studies have demonstrated lipid binding domains C-terminal to the 23-90 region of PrP (21)(22)(23). For example, a sphingolipidbinding domain peptide comprising residues 179 -214 of human PrP was identified on the basis of its structural similarity to the V3 domain of human immunodeficiency virus, type 1, gp120 and subsequently shown to interact with monomolecular films of galactosylceramide and sphingomyelin (22). In contrast, that PrP-CTM⌬23-90 was not raft-associated demonstrates that binding sites in the 91-230 region of murine PrP are insufficient for the cellular raft localization of a TM form of the protein. This may be attributable to the lipid binding domains in the C-terminal globular domain of PrP being sterically hindered from interacting with raft lipids by the membrane anchoring domain of PrP-CTM⌬23-90. On the other hand, the high flexibility of the N-terminal region of PrP (46,47) may allow the interaction of the raft targeting determinant in this region with raft lipids irrespective of whether the protein is N-terminally or C-terminally anchored to the membrane, as indicated by the raft association of both PrP-NTM and PrP-CTM, respectively. Raft association is unlikely to require the full integration of the 23-90 region into the bilayer, as in PrP-NTM the N-terminal signal anchor that tethers it to the membrane would prevent insertion of the N-terminal residues of mature PrP (i.e. residues 23 onwards) into the bilayer.
A further possibility is that the 23-90 region of PrP mediates raft association by interacting with raft proteins, similar to the interaction of the sorbin homology domain of the soluble cytosolic c-Cbl-associated protein with flotillin (17). Recently identified binding partners of PrP that are raft-localized include neuronal cell adhesion molecules (48) and plasminogen (49), the product of which, plasmin, is associated with hippocampal neuronal rafts (18). However, the interaction domains of PrP with either neuronal cell adhesion molecules or plasminogen include sites within the C-terminal globular domain of PrP, and this is not concurrent with our finding that the 23-90 region of the PrP ectodomain was solely sufficient for the raft association of another TM protein. Although PrP has also been shown to interact with the 37/67-kDa laminin receptor (50) and stress-inducible protein 1 (51), it is not known if these proteins localize to rafts, and the major binding site on PrP involved residues 144 -179 (52) and residues 113-128 (51), respectively.
Although PrP is N-glycosylated on Asn-180 and Asn-196, the N-glycans are not responsible for the raft targeting of the ectodomain of PrP for two reasons. First, the N-terminally anchored construct PrP-NTM and the anchorless PrP⌬GPI which are predominantly unglycosylated (26) were localized almost exclusively in DRMs; and second, PrP-CTM⌬23-90, which still retains the N-glycan addition sites and was glycosylated, failed to localize in DRMs. Certain TM receptors, including the Fc⑀RI receptor and the epidermal growth factor receptor (53,54), are recruited into plasma membrane rafts by ligand-mediated oligomerization that subsequently instigates signal transduction (1). The raft association of PrP-CTM, however, occurred in the presence of a chelator of Cu(II), a proposed ligand for PrP. Alternatively, raft incorporation may result from the self-oligomerization of PrP-CTM in a manner analogous to the yeast prion-like protein Sup35p. Here the N-terminal oligopeptide repeat region of Sup35p is critically required for oligomerization and can be functionally replaced with the homologous octapeptide repeat region of mammalian PrP (55). However, deletion of the octapeptide region (residues 51-90) from PrP-CTM failed to abolish the raft association of the protein. Rather, our results clearly demonstrate that the raft association of PrP-CTM was constitutive.
A physiological function of the 23-90 region of murine PrP has been highlighted by the finding that deletions within this region severely compromise the rate of trafficking of the protein to the surface of neuroblastoma cells, leading to the suggestion that this region contains cell surface targeting determinants (56). Furthermore, substitution of the 23-90 region with the homologous region of Xenopus laevis PrP, which lacks the Cu(II)-binding octarepeats of murine PrP, did not significantly affect the kinetics of transport to the cell surface. Accordingly, our findings show that the 23-90 region of murine PrP is necessary and sufficient for the raft association of a TM protein and that this association fails to be abolished by deletion of the octapeptide repeat region (residues 51-90). As rafts appear to be platforms for the sorting and trafficking of specific proteins and lipids from the trans-Golgi network to the cell surface (1), the possibility exists that the raft and cell surface targeting determinants contained within the 23-90 region of PrP may be congruent. Another possible function of the raft targeting determinant in the ectodomain of PrP may be to facilitate the retrieval of the protein, or C-terminally truncated fragments of the protein lacking the GPI anchor, to rafts. For example, the normal function of PrP may require the protein to translocate out of rafts. In this context PrP is seen to translocate out of rafts upon exposure of cells to copper prior to its clathrin-dependent endocytosis. 2 This is consistent with PrP being located in the peripheral semi-ordered subdomains of neuronal rafts that border the disordered, and detergent-soluble, phase of cellular membranes (57). The subsequent reincorporation of PrP into rafts may require the raft targeting determinant in the ectodomain. Apart from mediating the retrieval of PrP C to rafts, the raft targeting determinant in the ectodomain of the protein may play a critical role during the process of prion infection by targeting exogenous PrP Sc specifically to neuronal rafts where its substrate, host PrP C , is concentrated (8). Thus, the raft targeting determinant in the ectodomain of PrP may not only contribute to the normal function of the protein but also to its dysfunction in neurodegenerative diseases.