Clathrin-independent internalization of normal cellular prion protein in neuroblastoma cells is associated with the Arf6 pathway

Normal cellular prion protein (PrP^c) is a glycosylphosphatidylinositol (GPI)-anchored glycoprotein that is expressed predominantly in the brain, as well as in the spinal cord and at lower levels in leukocytes and some peripheral tissues (Pauly and Harris, 1998; Prusiner, 1998; Weissmann and Flechsig, 2003). Prion diseases are caused by the conversion of PrP^c into a protease-resistant misfolded isoform (PrP^sc). The prion diseases include transmissible spongiform encephalopathies such as scrapie in sheep and goats, bovine spongiform encephalopathy, and Creutzfeldt-Jakob disease in humans (Prusiner, 1998). Formation of PrP^sc in all of these disorders occurs when nerve cells are exposed to PrP^sc, which then converts PrP^c to aggregated deposits of PrP^sc. In scrapie, it is not yet clear whether a loss of function of PrP^c or a gain of function by PrP^sc is responsible for the neuronal loss and spongiform changes in the brain. However, the observation that clinical symptoms occur without any obvious scrapie deposits (Collinge et al., 1990; Medori et al., 1992) suggests that it is the loss of normal PrP^c function, not the formation of PrP^sc deposits, that causes prion disease (Aguzzi et al., 1997).

To understand the pathogenesis of scrapie, it is important to know the mode of PrP^c internalization in the cell. The biological role of PrP^c internalization is unknown, although it is stimulated by Cu²⁺ (Pauly and Harris, 1998). Like many GPI-anchored proteins, PrP^c is found clustered in the sphingomyelin- and cholesterol-rich domains of the membrane known as lipid rafts (Taylor and Hooper, 2006). Although PrP^c is present in rafts, it has been widely accepted that, at least in neuronal cells, PrP^c is internalized via clathrin-coated pits. Support for this view has come from multiple studies using different cell models. The clathrin-dependent pathway for PrP^c internalization was first proposed by the Harris laboratory using a mouse N2a cell line stably transfected with chicken PrP^c (Shyng et al., 1994). Their evidence that PrP^c was internalized via clathrin-coated pits came from immunogold labeling of PrP^c in clathrin-coated pits, from blocking PrP^c internalization by hypertonic sucrose, which disrupts clathrin lattices, and from the detection of PrP^c in clathrin-coated-vesicle preparations from chicken brain.

Additional evidence that PrP^c is internalized via clathrin-coated pits has come from kinetic studies using primary neurons. In these studies, the Morris laboratory found that the rate of PrP^c internalization was similar to that of transferrin, a cargo that is known to be internalized by clathrin-mediated endocytosis (Sunyach et al., 2003). They also found, by immunogold labeling, that endogenous PrP^c was localized to clathrin-coated pits in N2a cells. Studies, primarily from the Hooper laboratory, using the human neuronal cell line SHY5Y stably expressing mouse PrP^c have also supported the clathrin-dependent internalization of PrP^c in the presence of Cu²⁺ (Taylor et al., 2005). This group found that the Cu²⁺-stimulated internalization of PrP^c was blocked by the drug tyrphostin A23, which also blocks internalization of the transferrin receptor. More recently, the Taylor laboratory showed that depleting cells of the LRP1 receptor, a scavenger receptor that is internalized via clathrin-mediated endocytosis, blocks Cu²⁺-stimulated internalization of PrP^c (Taylor and Hooper, 2007). However, it is not clear that this effect of LRP1 knockdown is due to a decrease in receptor internalization because LRP1 has also been shown to affect the biosynthetic trafficking of PrP^c (Parkyn et al., 2008).

Although many studies have indicated that PrP^c is internalized via clathrin-mediated endocytosis, several studies have suggested that PrP^c is also internalized via cholesterol-enriched raft and/or caveolae-like domains; caveolae are a subset of lipid rafts containing the protein caveolin. Ying et al. found that endogenous PrP^c was localized to caveolae in mouse N2a cells (Ying et al., 1992), although other laboratories have not even detected the presence of caveolin in N2a cells (Marella et al., 2002; Shyng et al., 1994). Immunogold labeling also localized PrP^c to caveolae in a CHO cell line that stably expresses hamster PrP^c (Peters et al., 2003), and endogenous PrP^c has also been detected in lipid rafts, as well as in clathrin-coated pits, in N2a cells (Sunyach et al., 2003). In addition, PrP^c has been found in detergent-insoluble extracts from N2a cells, which is indicative of its presence in raft and/or caveolae-like domains (Sunyach et al., 2003; Vey et al., 1996). These raft and/or caveolae structures seem to be important for PrP^c trafficking, because their disruption by the use of sterol-binding agents, such as filipin and nystatin, inhibits Cu²⁺-stimulated PrP^c internalization in human microglia cells and N2a cells overexpressing mouse PrP^c (Marella et al., 2002; Peters et al., 2003). Although these results indicate the presence of PrP^c in raft and/or caveolae structures, it has been suggested that PrP^c normally occurs in rafts but then exits the rafts to be internalized via clathrin-coated pits (Morris et al., 2006; Sunyach et al., 2003).

In summary, although there is no consensus, it is generally considered that, at least in neuronal cells, PrP^c occurs in rafts, but is then internalized via clathrin-mediated endocytosis. To investigate the importance of this latter pathway for the internalization of PrP^c, we depleted N2a cells of clathrin using RNA interference and measured the internalization of the endogenous PrP^c. Surprisingly, after clathrin depletion, confirmed by marked inhibition of transferrin uptake, we found no significant decrease in internalization of PrP^c either in the presence or absence of Cu²⁺. By contrast, when rafts were disrupted by filipin or nystatin, PrP^c internalization was blocked both in the presence and absence of Cu²⁺, whereas clathrin-mediated uptake of transferrin was unaffected. The internalized PrP^c colocalized with cargo that was associated with the clathrin-independent Arf6-associated pathway and was present in large vacuoles in cells expressing the Arf6 dominant-active mutant. These results show that PrP^c is internalized in a clathrin-independent pathway that is associated with Arf6.

It was previously shown that the rates of PrP^c and transferrin internalization were the same in primary neurons in the absence of Cu²⁺ (Sunyach et al., 2003). This suggested that PrP^c was internalized via clathrin-mediated endocytosis, the pathway for transferrin internalization. Because N2a cells have been used as a model to study PrP^c trafficking and scrapie propagation, we examined whether transferrin and PrP^c are internalized at the same rate in N2a cells. After loading the plasma membrane with either Alexa-Fluor-488–transferrin or anti-PrP Fab at 4°C for 30 minutes, cells were incubated for varying times at 37°C both in the presence and absence of Cu²⁺. The fluorescence intensity of PrP^c and transferrin in the cell was measured to determine the rate of internalization. The data, plotted in Fig. 1 as fraction internalized versus time, shows that Cu²⁺ did not appreciably affect the kinetics of internalization. The major effect of Cu²⁺ was that it changed the distribution of PrP^c in the cell. In the presence of Cu²⁺, the internal and external pools of PrP^c became equal at steady state, which indicates that the in and out rates are equal. Because steady state was reached with a half-life of 5 minutes in the presence of Cu²⁺, half-lives of 10 minutes were calculated for both internalization and externalization of PrP^c. In the absence of Cu²⁺, the internal pool was only one-quarter the size of the external pool at steady state (compared with 50% in the presence of Cu²⁺), which is consistent with the stimulatory role of Cu²⁺ on PrP^c internalization (Pauly and Harris, 1998). Because steady state was reached with a half-life of about 10 minutes in the absence of Cu²⁺, half-lives of 48 and 12 minutes were calculated for the internalization and externalization of PrP^c, respectively. As for transferrin, it was internalized at about the same rate in both the presence and absence of Cu²⁺ (Fig. 1B). Because more than 80% of the transferrin receptor was found in the internal pool, the half-life to reach steady state, 5 minutes, is essentially the half-life of internalization. These results show that, in N2a cells, transferrin is internalized about tenfold faster than PrP^c in the absence of Cu²⁺, unlike in primary sensory neurons in which these rates were similar (Sunyach et al., 2003). In the presence of Cu²⁺, the in rate of transferrin internalization is only twice that of PrP^c in N2a cells, but similar measurements have not been made in primary neurons.

These results suggest that, at least in N2a cells in the absence of Cu²⁺, transferrin and PrP^c are not internalized via the same pathway because their internalization rates differ by about one order of magnitude. Therefore, to directly test whether, like transferrin, PrP^c is internalized via clathrin-mediated endocytosis, N2a cells were depleted of clathrin using RNA interference. Knock down of clathrin has been used to identify cargos that are internalized via clathrin-mediated endocytosis while, at the same time, helping to elucidate alternative internalization pathways. N2a cells were depleted of clathrin using two different siRNA oligonucleotides, resulting in a 95% reduction in the clathrin levels of the siRNA-treated cells compared with control cells (supplementary material Fig. S1).

The effect of clathrin depletion on PrP^c internalization was first measured in the absence of Cu²⁺. In these experiments, PrP^c and transferrin internalization were measured concomitantly because transferrin internalization is a way to monitor clathrin-mediated endocytosis. In all of our experiments, internalization of PrP^c was measured using anti-PrP Fab antibodies, whereas fluorescently labeled transferrin was used to measure transferrin uptake. Fluorescent images of PrP^c and transferrin internalization, along with clathrin immunostaining, are shown for control cells in Fig. 2A-C. The clathrin-depleted and control cells were imaged using identical confocal settings to enable direct comparison of fluorescent intensity. Consistent with the western blot analysis, the clathrin-depleted cells (Fig. 2D-F) showed a marked reduction in clathrin staining at the trans-Golgi network and a concomitant decrease in transferrin uptake. These results showed that clathrin-siRNA treatment effectively blocks clathrin-mediated endocytosis. Surprisingly, although transferrin uptake was blocked, the clathrin-depleted cells did not show a significant reduction in PrP^c internalization.

In order to quantify the pool of internalized PrP^c, the staining protocol was modified to enable us to clearly resolve internal from surface-bound PrP^c. The surface bound anti-PrP Fab was stained with a Cy5-conjugated secondary antibody prior to cell permeabilization, whereas the internal pool of PrP^c was labeled with a rhodamine-conjugated secondary antibody after permeabilization. Fig. 3A shows immunostaining of PrP^c in control and clathrin-depleted cells using this method, along with transferrin uptake. The surface PrP^c outlining the cell is now clearly resolved from the internal pool. These images confirm that clathrin depletion caused marked inhibition of transferrin uptake, while not significantly affecting PrP^c internalization.

To analyze the effect of clathrin depletion on PrP^c and transferrin internalization, more than 150 control and clathrin-depleted cells were randomly imaged. The average intensities of PrP^c and transferrin in the recycling endosome were then measured in each cell. In Fig. 3B, the values measured in the clathrin-depleted cells are normalized to the control values. This graph shows that there was no significant difference in the average intensity of PrP^c in the recycling endosome between control and clathrin-depleted cells, whereas transferrin uptake was reduced by 85% in clathrin-depleted cells compared with controls. Cells were also depleted using another clathrin-siRNA oligonucleotide (oligonucleotide #2) to ensure that the particular siRNA oligonucleotide used was not having an effect on PrP^c internalization. With oligonucleotide #2, not only was the data analyzed using the average intensity of the recycling endosome, the internalization of PrP^c and transferrin was also analyzed using MetaMorph software to determine the total internal pools of PrP^c and transferrin from confocal z-stack cell images. The results from the three-dimensional analysis were not significantly different from the two-dimensional analysis, i.e. clathrin depletion did not significantly affect the extent of PrP^c internalization, but markedly inhibited transferrin uptake (Fig. 3B, lanes 5-8). Therefore, constitutive internalization of PrP^c apparently occurs via a clathrin-independent pathway.

Similar experiments were carried out in the presence of Cu²⁺, which has been shown to stimulate the internalization of PrP^c (Pauly and Harris, 1998). Fig. 4A shows the immunostaining of surface and internal pools of PrP^c, along with transferrin uptake, in cells that were incubated in the presence of Cu²⁺. Comparison of PrP^c localization in the presence and absence of Cu²⁺ showed that, in addition to a larger internal pool of PrP^c in the presence of Cu²⁺, the PrP^c on the plasma membrane also had a patchier appearance compared with in the absence of Cu²⁺. Just as we observed in the absence of Cu²⁺, clathrin depletion of the N2a cells did not significantly affect internalization of PrP^c (Fig. 4Ac-f), although, as expected, the clathrin-depleted cells showed a marked reduction in transferrin uptake. Again, we used two different siRNA oligonucleotides to deplete clathrin and two slightly different methods to measure the internalization of proteins. This quantitative analysis, shown in Fig. 4B, confirmed that clathrin depletion did not significantly affect the Cu²⁺-stimulated PrP^c endocytic pathway, although it markedly inhibited transferrin uptake. Therefore, PrP^c is internalized via a clathrin-independent pathway both in the presence and absence of Cu²⁺.

All of the above PrP^c-internalization studies examining the effect of clathrin depletion were performed using the D13 anti-PrP Fab. To validate these results further, PrP^c internalization was remeasured using a different anti-PrP Fab, one made from the AH6 anti-PrP mAb. This latter antibody recognizes residues 159-174, whereas the D13 Fab recognizes residues 95-105 (Novitskaya et al., 2006). Data obtained with the AH6 Fab confirms that neither constitutive nor Cu²⁺-stimulated internalization of PrP^c was significantly affected by blocking clathrin-mediated endocytosis (Table 1).

To better characterize the clathrin-independent pathway for internalizing PrP^c, we examined the effect of disrupting lipid rafts on PrP^c internalization. Lipid rafts were disrupted by treating cells with either filipin or nystatin, drugs that disrupt rafts by sequestering cholesterol while only having a limited effect on clathrin-mediated endocytosis (Orlandi and Fishman, 1998; Ricci et al., 2000). When the cells were incubated with nystatin, PrP^c internalization was blocked both in the presence and absence of Cu²⁺ (Fig. 5A). By contrast, transferrin uptake was not significantly affected by nystatin treatment. Quantitative analysis of PrP^c and transferrin internalization in nystatin-treated cells showed a marked inhibition of PrP^c internalization and only a slight decrease in transferrin endocytosis (Fig. 5B). Similar results were obtained by treating the cells with filipin (Fig. 5B). In agreement with these data, Marella et al. found that, in the presence of Cu²⁺, nystatin and filipin blocked internalization of PrP^c, but did not block transferrin uptake (Marella et al., 2002). Therefore, disruption of raft organization inhibited PrP^c internalization both in the presence and absence of Cu²⁺. In addition to being dependent on rafts, PrP^c internalization was found to be dynamin dependent both in the presence and absence of Cu²⁺. When cells were transfected with GFP constructs of either wild-type dynamin-1 or the dominant-negative mutant of dynamin-1, dynamin-1 (K44A), the latter inhibited internalization of both PrP^c and transferrin in the presence and absence of Cu²⁺ (Table 2, supplementary material Fig. S2A). Consistent with these results, incubation of the cells with the dynamin inhibitor dynasore also inhibited PrP^c internalization both in the presence and absence of Cu²⁺ (Table 2, supplemental material Fig. S2B). Therefore, whether or not Cu²⁺ is present, the internalization of PrP^c in N2a cells is raft- and dynamin-dependent, but clathrin-independent.

A raft-dependent dynamin-dependent pathway would typically suggest that PrP^c is internalized via a caveolae-raft internalization pathway. However, the Harris laboratory reported that N2a cells contain trace amounts of caveolin-1 (Shyng et al., 1994) and we confirmed their finding using western blot analysis to show that the level of caveolin-1 in N2a cells was barely detectable compared with the level in HeLa cells (supplemental material Fig. S3A). To confirm that the internalization of PrP^c does not occur via caveolae, cells were transiently transfected with GFP-labeled caveolin-1 with a mutation in tyrosine 14 (Y14F). Because phosphorylation of tyrosine 14 in caveolin is essential for caveolae internalization (del Pozo et al., 2005; Orlichenko et al., 2006), mutation of this tyrosine residue inhibits internalization of caveolin-associated rafts. In fact, we found that expression of this mutant caveolin-1 had no effect on PrP^c internalization either in the presence or absence of Cu²⁺ (supplemental material Fig. S3B). By contrast, PrP^c internalization was reduced by about 50% when actin was depolymerized by cytochalasin D.

Because nystatin and filipin have been shown to block internalization of cargo that traffic via the Arf6-regulated pathway (Naslavsky et al., 2003), we examined whether PrP^c colocalizes with other cargo that traffic via this pathway. After internalizing PrP^c, the cells were immunostained to detect CD98, a glycoprotein that traffics via the Arf6-associated pathway (Eyster et al., 2009). Both in the presence and absence of Cu²⁺, internalized PrP^c was found to colocalize with CD98 in vacuolar structures that are frequently found in N2a cells (Fig. 6A). As expected, no internalized transferrin was associated with PrP^c in these decorated vacuolar structures (supplemental material Fig. S4). We next determined whether cargo that traffics via the Arf6 pathway and is internalized at the same time as PrP^c also colocalizes with PrP^c. For this experiment, we followed the internalization of MHC-I, another protein whose trafficking via the Arf6 pathway has been well characterized (Naslavsky et al., 2003). Cells were preloaded with anti-PrP and anti-MHC-I antibodies, followed by incubation at 37°C for 30 minutes. Following staining of these proteins with secondary antibodies, MHC-I and PrP^c were found to be colocalized on the same vacuolar structures both in the presence and absence of Cu²⁺ (Fig. 6B).

To further validate that PrP^c trafficking is associated with the Arf6 pathway, N2a cells were transfected with the dominant-positive Arf6 mutant Arf6Q67L. This mutant has been shown to drive the formation of vacuoles that trap cargo that enters cells via a clathrin-independent pathway prior to the cargo merging with the clathrin pathway (Donaldson et al., 2009). If trafficking of PrP^c was regulated by Arf6, then PrP^c would be found on the large vacuoles that develop in cells expressing Arf6Q67L (Brown et al., 2001). As shown in Fig. 7A, both in the presence and absence of Cu²⁺, internalized PrP^c indeed decorated large vacuoles that co-stained for CD98. Moreover, when antibodies against PrP^c and MHC-I were used to simultaneously examine the uptake of these proteins, PrP^c and MHC-I were colocalized on large vacuoles both in the presence and absence of Cu²⁺ (Fig. 7B). PrP^c was found localized to these large vacuoles even in cells incubated with the dynamin inhibitor dynasore.

To determine whether other cargos are found on these large vacuolar structures, the internalization of cholera toxin B and transferrin was examined in cells expressing Arf6Q67L. Both in the presence and absence of Cu²⁺, cholera toxin B was found to be colocalized with CD98 on these large vacuolar structures (Fig. 8). These results indicate that the trafficking of cholera toxin B is associated with the Arf6 pathway in N2a cells; this result has also been reported for the trafficking of cholera toxin B in HeLa cells (Naslavsky et al., 2004). As expected, there was no significant colocalization of internalized transferrin with CD98-immunostained vacuoles (Fig. 8). Therefore, the Arf6Q67L vacuoles trap PrP^c and cholera toxin B, but not transferrin. These results show that PrP^c is internalized via a clathrin-independent pathway that is associated with Arf6 in N2a cells.

To determine whether clathrin-mediated endocytosis is essential for internalizing PrP^c, RNA interference was used to deplete clathrin in N2a cells. This technique has been extensively used to identify the role of clathrin in the internalization of various cargos. Several studies have indicated that PrP^c is internalized via clathrin-mediated endocytosis in neuronal cells (Shyng et al., 1994; Sunyach et al., 2003; Taylor et al., 2005). Therefore, we expected that clathrin depletion would block PrP^c internalization. Surprisingly, both in the presence and absence of Cu²⁺, the same amount of PrP^c was internalized in cells treated with clathrin siRNA as in control cells. Although blocking the clathrin-dependent pathway did not significantly affect PrP^c internalization, disrupting lipid rafts with nystatin or filipin markedly inhibited the internalization of PrP^c both in the presence and absence of Cu²⁺ without significantly affecting transferrin uptake. These results suggest that, at least in N2a cells, both in the presence and absence of Cu²⁺, endogenous PrP^c is internalized via a clathrin-independent pathway.

Our results are in agreement with the study from the Chabry laboratory showing inhibition of PrP^c internalization by filipin, which disrupts rafts, but not by chloroperazine, which blocks clathrin-mediated endocytosis (Marella et al., 2002). Conversely, our findings disagree with a number of other studies that suggest that PrP^c is internalized via clathrin-mediated endocytosis (Shyng et al., 1994; Taylor et al., 2005; Sunyach et al., 2003). This disagreement might partly be due to differences in cell types, as well as to differences in trafficking of endogenous PrP^c and stably expressed PrP^c. In addition, instead of using pharmacological agents with ill-defined targets to block clathrin, our study blocked clathrin internalization using siRNA, which specifically inhibits clathrin-mediated endocytosis. Finally, support for the view that PrP^c is internalized via clathrin-mediated endocytosis has come from electron-microscopic images showing immunogold-labeled PrP^c in clathrin-coated pits (Shyng et al., 1994; Sunyach et al., 2003), but these images do not provide a measure of the extent to which PrP^c traffics via this pathway. In fact, electron microscopy of the immunolabeled PrP^c found that greater than 80% of the labeled PrP^c was in detergent-resistant membranes (Morris et al., 2006).

PrP^c belongs to the family of GPI-anchored proteins, which are typically internalized via clathrin-independent pathways including a caveolin-pathway of uptake, flotillin-1 pathway, a Cdc42-regulated pathway and an Arf6-regulated pathway (Donaldson et al., 2009; Glebov et al., 2006; Mayor and Riezman, 2004). Our results using the dominant-active Arf6 mutant, in combination with our results using clathrin siRNA, show that PrP^c traffics via a clathrin-independent Arf6-associated pathway. In addition to the presence of PrP^c in Arf6Q67L vacuoles, the internalized PrP^c colocalized with CD98 and MHC-I, proteins that traffic via the Arf6-regulated pathway (Eyster et al., 2009; Naslavsky et al., 2003). Typically, cargo that traffics via the Arf6 pathway is dynamin independent (Donaldson et al., 2009), but perhaps this is different in neuronal cells, which express both neuronal-specific dynamin-1 as well as dynamin-2.

It is important to recognize that many cargos are internalized via multiple pathways. For example, similar to PrP^c in our study, the β-amyloid peptide is internalized in a clathrin-independent, dynamin-dependent and raft-dependent pathway in the absence of apoliproprotein E (Saavedra et al., 2007). By contrast, in the presence of apoliproprotein E, β-amyloid peptide is internalized via clathrin-mediated endocytosis (Kang et al., 2000). Furthermore, even though there is disagreement on whether PrP^c is internalized via a clathrin-dependent or -independent pathway, it is important to note that both of these pathways converge either directly or indirectly on the early endosome (Donaldson et al., 2009; Kirkham et al., 2005). It has previously been shown that PrP^c first traffics to the early endosome (Magalhaes et al., 2002; Pimpinelli et al., 2005) and then to the late endosome (Peters et al., 2003; Pimpinelli et al., 2005). Ultimately, it is then either degraded in the lysosome or recycled back to the plasma membrane (Borchelt et al., 1992).

Although the simplest way of interpreting our studies that used clathrin siRNA is that the majority of PrP^c does not enter the cells via clathrin-mediated endocytosis, it is possible that, in N2a cells, PrP^c is internalized via multiple pathways. Perhaps when clathrin is depleted, a clathrin-independent raft pathway is upregulated. Compensatory mechanisms of trafficking were previously observed when a dominant-negative dynamin-1 mutant, dynamin-1 K44A, was expressed in cells (Damke et al., 1994). Therefore, it is possible that a clathrin-independent pathway compensates for the clathrin-mediated endocytosis when clathrin is depleted. As for the inhibition of PrP^c internalization that occurs in the presence of clathrin when the raft pathway is disrupted, this might occur if PrP^c must be present in intact rafts before it can shift over to clathrin-coated pits to be internalized via clathrin-mediated endocytosis (Taylor and Hooper, 2006). One way to test this rather complicated model would be to isolate the carrier protein that PrP^c binds to during internalization. A recent report from the Hooper laboratory suggested that the LRP1 receptor was the carrier for PrP^c (Taylor and Hooper, 2007), but a more recent study from the Morris laboratory suggested that the primary role of this receptor was to transport PrP^c to the plasma membrane rather than act as a carrier for internalization of PrP^c (Parkyn et al., 2008). Isolation of the PrP^c carrier protein might finally provide a definitive answer as to how PrP^c is internalized in N2a cells and primary neurons.

The N2a mouse neuroblastoma cell line was purchased from ATCC (American Type Culture Collection). N2a cells were grown at 37°C in a humidified incubator under 95% air, 5% CO₂ in a MEM (Mediatech, Herndon, VA) supplemented with 10% FBS, 100 IU/ml penicillin, 200 μg/ml streptomycin and nonessential amino acids (Invitrogen, Carlsbad, CA). Cells, cultured for 24 hours in complete medium without antibiotics, were transfected with the following constructs using Lipofectamine 2000 (Invitrogen, Carlsbad, CA): GFP-labeled wild-type dynamin-1 and K44A dynamin-1 (gift from Pietro De Camilli, Yale University, New Haven, CT), Arf6Q67L (gift from Julie Donaldson, NHLBI, NIH, Bethesda, MD), and GFP-labeled caveolin-1 and GFP-labeled mutant Y14F caveolin-1 (gift from Mark A. McNiven, Mayo Clinic, Rochester, MN).

N2a cells, grown on eight-well-chamber glass slides (Nalge Nunc International, Rochester, NY), were washed twice with phosphate buffer saline and then incubated at 4°C for 30 minutes with 2.5 μg/ml anti-PrP Fab antibody. Transferrin (10 μg/ml), either Alexa-Fluor-488-conjugated transferrin (Invitrogen, Carlsbad, CA) or Cy5-conjugated transferrin (Jackson ImmunoResearch, West Grove, PA), was routinely added to the cells, followed by incubation of the cells for 30 minutes at 37°C prior to fixation. To follow cholera-toxin-B internalization, rhodamine-labeled cholera toxin B (Sigma) was added at 1 μg/ml. In the experiments in the presence of Cu²⁺, 400 μM CuSO₄ was added to the cells when they were incubated at 37°C. In the lipid-raft-disruption experiments, cells were first treated nystatin (50 μg/ml) and filipin (5 μg/ml) (Sigma-Aldrich, St Louis, MO) in complete medium at 37°C for 30 minutes. Cells continued to be incubated with the drugs during the preincubation at 4°C and incubation at 37°C. To distinguish between internal and surface-bound PrP^c, fixed cells were first immunostained with Cy5-conjugated secondary anti-Fab antibody and then, following permeabilization, were immunostained with a secondary anti-Fab antibody conjugated to a different fluorophore to detect internalized PrP^c. To depolymerize actin, 1 μM cytochlasin D (Sigma) was added to N2a cells for 30 minutes at 37°C. To inhibit dynamin, 10 μM dynasore (Sigma) was added to cells at 37°C for 30 minutes.

The anti-PrP Fab antibody used was either D13 Fab (InPro, San Francisco, CA) or AH6 Fab, made from the AH6 antibody (TSE Resource Center, Berkshire, UK) using the Fab preparation kit (Pierce, Rockford, IL). Cells were immunostained for clathrin with X22 antibody (Affinity BioReagents, Golden, CO), GFP antibody (Abcam) and Alexa-Fluor-488-conjugated CD98 antibody (AbD Serotec, Raleigh, NC). Immunoblots were probed with anti-clathrin-heavy-chain mAb clone 23 (BD Biosciences, San Jose, CA), β-actin (Abcam, Cambridge MA), caveolin-1 (Santa Cruz Biotechnology, Santa Cruz, CA), glycerol 3-phosphate dehydrogenase (Novus Biologicals, Littleton, CO) and major histocompatibility protein class I (MHC-I) (from Natalie Porat-Shoram, NHLBI, NIH, Bethesda, MD). Fluorescent secondary antibodies were from Jackson ImmunoResearch and Invitrogen.

Images were obtained using the LSM510 confocal microscope (Zeiss, Thornwood, NY) using Plan-Apochromat 63×/1.4 objective. In comparing control and either siRNA- or drug-treated cells, identical confocal settings were used to enable direct comparison between images. Analysis of the internalized PrP^c and transferrin was performed using MetaMorph software (Molecular Devices, Sunnyvale, CA).

RNA interference of clathrin heavy chain was performed using siRNAs (Dharmacon, Chicago, IL) to the following sequence: 5′-UCAGAAGAAUUGCUGCUUAUU-3′ (oligonucleotide #1) and 5′-UAAUCCAAUUCGAAGACCAAUUU-3′ (oligonucleotide #2). Lipofectamine RNAiMAX (Invitrogen) was used to transfect the siRNA oligonucleotides using both the forward and reverse method to transfect cells as described in the manufacturer's transfection protocol. To determine cellular levels of clathrin, cell lysates were run on 4-12% gels (Invitrogen), followed by transferring the proteins to nitrocellulose for immunoblotting. Protein bands were detected using the Odyssey infrared detection system (Li-Cor Bioscience, Lincoln, NE) or chemiluminescent substrate (Pierce, cat. no. 34080) followed by densitometer imaging (ChemiImager, Alpha Innotech).

We thank Julie Donaldson and members of the Donaldson laboratory for helpful discussions.