Normal prion protein trafficking in cultured human erythroblasts

Normal prion protein (PrP^c) is a glycosylphosphoinositol (GPI)–linked membrane protein with a very wide tissue distribution. Its function is not clearly understood, although recent evidence suggests a role in self-renewal of hematopoietic stem cells.¹  In contrast, abundant evidence supports a pivotal role for PrP^c in the propagation of transmissible spongiform encephalopathies (TSEs) through interaction with misfolded prion protein (PrP^sc).²  In particular, mice genetically modified to lack PrP^c expression in all tissues do not succumb to disease when exposed to PrP^sc.^3,4  The process whereby PrP^sc interacts with host PrP^c, converting it to PrP^sc and causing disease, is not clear, but a requirement for the GPI membrane anchor of PrP^c has been demonstrated in mice engineered to express PrP^c without a GPI anchor. When these animals are exposed to PrP^sc, there is an extracellular proliferation of the abnormal protein but no disease develops, presumably because of a failure to internalize PrP^sc within neural cells.⁵ 

Bovine spongiform encephalopathy acquired by oral ingestion of PrP^sc present in meat is the most likely cause of variant Creutzfeldt-Jakob disease (vCJD) in humans.^6,–8  Recent evidence demonstrates it is highly probable vCJD can be transmitted by transfusion of red cell preparations.⁹  These data highlight the need for a greater understanding of the distribution and trafficking of PrP^c in human hematopoietic cells so that knowledge of the extent to which PrP^sc binding and internalization can occur in blood and blood-forming tissues can be used to inform decisions regarding the safety of blood components used for therapy.

In this study, we have examined the expression and trafficking of PrP^c in human erythroid cells derived from in vitro culture of CD34⁺ cells from peripheral blood. The results show that PrP^c in erythroblasts behaves differently from other GPI-linked proteins (CD59 and CD55) in that it cycles rapidly from the plasma membrane to internal membrane compartments by a process likely to be clathrin-mediated and involving tetraspanin-enriched microdomains (TEMs).

Waste buffy-coat material from anonymous blood donors were made available from the National Blood Service (Bristol, United Kingdom). This provision complies with the Nuffield Council on Bioethics Guidance on Human Tissue Ethical and Legal Issues 1954, the Medical Research Council's Operational and Ethical Guidance on Human Tissue and Biological Samples for use in Research 2005, and the Royal College of Pathologists Transitional Guidelines to facilitate changes in procedures for handling “surplus” and archival material from human biological samples.

Monoclonal antibodies to CD63 (H5C6) and CD81 (JS-81) were from BD Biosciences (Oxford, United Kingdom). Anti-CD82 (B-L2) was from Serotec (Oxford, United Kingdom). Monoclonal antibodies to glycophorin A (BRIC163) and CD59 (BRIC229) were from IBGRL (Bristol, United Kingdom). Anti-PrP (ICSM 18) was from D-Gen Ltd (London, United Kingdom). Rabbit polyclonal anti-calnexin was from Stressgen Bioreagents (Ann Arbor, MI), anti-giantin from CRP (San Diego, CA), and anti–LAMP-1 from Affinity Bioreagents (Golden, CO). AffiniPure Fab fragment rabbit anti–mouse IgG and rhodamine (TRITC)–conjugated AffiniPure goat anti–rabbit IgG were from Jackson ImmunoResearch (Newmarket, United Kingdom). FITC-conjugated F(ab′)2 fragment of goat anti–mouse Igs was from DAKO (Ely, United Kingdom). Alexa Fluor 488 goat anti–mouse IgG1, Alexa Fluor 546 goat anti–rabbit IgG, and Alexa Fluor 546 conjugate of anti-transferrin were from Invitrogen (Paisley, United Kingdom).

Human CD34⁺ cells were isolated from buffy coats (National Blood Service). Buffy coats were diluted 1:1 with Hanks balanced salt solution (HANKS; Sigma, Poole, United Kingdom) then layered on top of Histopaque 1077 (Sigma). Cells were harvested from the 1077 gradient, washed with HANKS, and then resuspended in red cell lysis buffer (ammonium chloride, 4.15 g/L; EDTA, 0.02 g/L; and potassium bicarbonate, 0.5 g/L). CD34⁺ cells were isolated by positive selection using the Direct CD34⁺ progenitor cell isolation kit (Miltenyi Biotech, Bisley, United Kindgom).

Isolated CD34⁺ cells were cultured under conditions based on the method of Southcott et al¹⁰  using serum-free StemSpan Expansion Medium (Stemcell Technologies, London, United Kingdom) supplemented with stem cell factor (SCF; 10 μL/mL; R&D Systems, Abingdon, United Kingdom), erythropoietin (EPO; 3 μL/mL, Roche, Lewes, United Kingdom), low-density lipoprotein (LDL; 8 μL/mL; Calbiochem, Nottingham, United Kingdom), and interleukin-3 (IL-3; 1 μL/mL; R&D Systems). Cells were seeded at 0.5 × 10⁵/mL on day 0 and were maintained at a concentration of 1 × 10⁵/mL thereafter in vented T25 Falcon flasks (Becton Dickinson, Oxford, United Kingdom) in 5% CO₂ at 37°C.

Cells were seeded on 0.01% (wt/vol) poly-L-lysine (Sigma) coated coverslips (1 × 10⁶ cells per coverslip) and incubated overnight at 37°C in 5% CO₂.

All steps were carried out in 6-well plates, with cells on the top surface of the coverslips for all washes, and with cells facing down on Parafilm M (Lab 3, Bristol, United Kingdom) for all antibody incubations.

Cells were fixed with 3% formaldehyde (TAAB, Aldermaston, United Kingdom) for 20 minutes and permeabilized with 0.05% (wt/vol) digitonin (Sigma) for 5 minutes. Cells were incubated for 15 minutes in 4% bovine serum albumin (BSA; Park Scientific, Northampton, United Kingdom) and then incubated with primary antibodies. Antibodies to glycophorin A or CD59 were used as positive controls, and 4% BSA was used as a negative control. For dual labeling, antibodies were incubated separately with 2 × 5-minute washes in phosphate-buffered saline (PBS) between incubations.

When dual labeling with 2 monoclonal primary antibodies, the weaker antibody was incubated first and then subjected to an extra conversion step by the addition of AffiniPure Fab Fragment Rabbit Anti-Mouse IgG (Jackson ImmunoResearch) at a concentration of 1:10 in 4% BSA. After this conversion step and 2 additional PBS washes, the other monoclonal antibody was added. Goat anti-mouse FITC/Alexa Fluor 488 or goat anti-rabbit TRITC/Alexa Fluor 546 secondary antibodies were diluted in 4% normal goat serum and incubated with the cells for 30 minutes at room temperature in the dark. Coverslips were mounted on Vectashield Mounting Medium (Vector Laboratories, Burlingame, CA) on microscope slides and sealed with nail varnish.

Samples were imaged at 22°C using 63× oil-immersion lenses (magnification = 158.7 μm at zoom 1, 1.32 NA) on a Leica DM RBE upright epifluorescence microscope with phase contrast connected to a Leica TCS-NT confocal laser scanning microscope (Leica, Wetzlar, Germany) at the Medical Research Council (MRC) Imaging Facility, University of Bristol. Images were obtained using Leica software and subsequently processed using Adobe Photoshop (Adobe, San Jose, CA).

For protein internalization experiments, cells were seeded overnight as described and washed once in PBS before incubation with the primary antibody for 20 minutes at 10°C to allow antibody binding. The negative control used for internalization studies was normal mouse IgG (DAKO). Cells were incubated at 37°C for up to 1 hour to allow internalization of the antibody to occur and then fixed and permeabilized as described. After blocking with 4% BSA, either another primary antibody or the secondary antibody was added. All subsequent steps were carried out as described here.

Erythroid cells were seeded overnight as described, washed once in PBS before incubation with primary antibodies (20 minutes at 10°C), washed twice in PBS, and incubated with secondary antibody (room temperature for 40 minutes) to allow capping to occur. Cells were then washed twice and fixed with 3% formaldehyde for 20 minutes, washed 3 times, and then mounted on slides. For dual labeling using mouse monoclonals, the rabbit conversion antibody was used before capping with secondary antibodies. The second primary antibody was then used after fixation as described.

Internalized PrP was measured by drawing regions of interest (ROIs) both inside the cell under the plasma membrane and around the outside of the cell using Leica quantitation software. The percentage difference in pixels was then calculated between these 2 ROIs.

Cells from day 6 to day 8 were harvested from culture, washed once in cold PBS, and the cell pellet was put on ice. Lysis buffer (25 mM HEPES buffer, 1 complete protease inhibitor cocktail tablet [Roche, Sussex, United Kingdom], plus 1% detergent: either Triton X-100 [Sigma], Nonidet P40 [NP-40] substitute [USB, Cleveland, OH], Brij97 [Sigma] or CHAPS [Calbiochem] with 0.5% sodium deoxycholate [Sigma] supplementing some 1% detergents) was added to the cell pellet and incubated on ice for 20 minutes. Lysed cells were then centrifuged at maximum speed (11 000g) for 15 minutes at 4°C to pellet debris.

Lysates were precleared with 30 μL protein G–sepharose beads (Amersham Biosciences, Bucks, United Kingdom) for 1 hour at 4°C and subsequently incubated with specific mAbs bound to protein G–sepharose beads overnight at 4°C, rotating. For immunoprecipitations from the cell surface, intact cells were incubated with antibody at 4°C for 1 hour before lysis and then incubated with protein G–sepharose beads overnight. For internalization immunoprecipitations, cells were incubated at 4°C for 30 minutes, then at 37°C for 1 hour to allow endocytosis to occur. After these incubation steps, cells were subject to lysis and then incubated with beads overnight.

Immunoprecipitated proteins were eluted from beads by boiling with sodium dodecyl sulfate (SDS) nonreducing sample buffer, subjected to SDS-PAGE on a 12.5% polyacrylamide gel, and transferred to a nitrocellulose membrane. After blocking in PBS containing 5% (wt/vol) BSA and 0.2% (wt/vol) Tween-20, membranes were probed with either primary mAbs followed by anti–mouse Igs conjugated to horseradish peroxidase (HRP; DAKO) or biotinylated primary mAbs followed by streptavidin peroxidase polymer (Sigma). Primary mAbs were biotinylated using EZ-Link sulfo-NHS-biotin Reagents (Pierce, Rockford, IL).

The presence of secondary protein was detected using the ECL Western Lightning Chemiluminescence Kit (Perkin Elmer, Waltham, MA) and photography film (Kodak BioMax MR film; Kodak, Rochester, NY) or the Kodak 4000R Image station, in accordance with the manufacturer's instructions.

PrP^c expression in cultured human erythroblasts was examined using immunofluorescence confocal microscopy. CD34⁺ cells isolated from peripheral blood were cultured in the presence of SCF, IL-3, and EPO and erythroid cells examined at various stages of maturation. After 6 to 8 days in culture, most cells are at the proerythroblast stage, and there is a large amount of intracellular PrP^c with faint cell-surface staining. By day 11, when the cultures contain a heterogeneous population of basophilic erythroblasts and pronormoblasts, PrP^c expression is lower and more diffuse with minimal cell-surface staining (Figure 1A).

The intracellular localization of PrP^c in the cultured human erythroblasts was examined by dual staining with markers for organelles using cells from days 6 to 8 of culture (Figure 1B). Only a small proportion of PrP^c staining colocalized with the endoplasmic reticulum (ER) marker calnexin, the Golgi marker giantin, and the lysosomal marker LAMP-1, with most PrP^c that is present outside these organelles in intracellular membranes most likely to be endosomal (perinuclear recycling compartment) because it shows complete colocalization with the endosomal marker CD63. CD63, a member of the tetraspanin family, is known as a resident of late endosomes and lysosomes, but is also present in secretory vesicles and on the plasma membrane, and it cycles between these compartments.¹¹ 

PrP^c localization in erythroblasts was compared with that of another GPI-linked protein, CD59. Dual staining of PrP^c and CD59 revealed substantial intracellular colocalization in the perinuclear area (Figure 2Ai,ii). However PrP^c and CD59 were present in different regions of the plasma membrane (Figure 2Aiii,iv). PrP^c was also present in vesicle-like structures just below the plasma membrane, whereas CD59 is not observed in these vesicle-like structures either alone or colocalized with PrP^c (Figure 2Aii; arrows highlight PrP^c). Another GPI-linked protein, CD55, showed almost complete colocalization with CD59 (data not shown). These results suggest trafficking of PrP^c is not typical for GPI-linked proteins.

In order to compare the trafficking of PrP^c and CD59 in erythroid cells, the rates of internalization were compared. After binding of the relevant antibody to the cell surface at 10°C, cells were transferred to 37°C and incubated for up to 60 minutes to allow endocytosis to occur, and the localization of the proteins were determined. The results showed that PrP^c is visible inside erythroblasts after 10 minutes of incubation (Figure 2Bi). In contrast, internalized CD59 is not detected after 60 minutes of incubation (Figure 2Bii). These data suggest that PrP^c is not internalized by the slower clathrin-independent mechanism used by other GPI-linked proteins.¹²  PrP^c and the transferrin receptor (TfR) were internalized after 10 minutes and colocalized in vesicle-like structures under the plasma membrane (Figure 2Biii arrow). TfR is a prototypical nonraft protein and endocytoses in a clathrin-dependent manner.¹³  These results indicate that PrP^c is subject to clathrin-mediated endocytosis. However, as a GPI-linked protein, PrP^c lacks a cytoplasmic domain and so cannot bind directly to clathrin adaptor proteins. PrP colocalizes with the tetraspanin CD63 (Figure 1Biv) which, through its interaction with the adaptor proteins, has a role in protein trafficking.¹⁴  Our experiments showed that it internalizes over a similar time-scale to PrP (Figure 2Biv). Very little CD63 is evident at the plasma membrane of fixed and permeabilized erythroblasts. A similar distribution is seen in other cell types, where it is known to cycle to the cell surface.¹⁵ 

When erythroblasts were subjected to the internalization protocol with anti-PrP^c, fixed, permeabilized, and stained for CD63, PrP^c and CD63 were colocalized in vesicles just under the plasma membrane after 5 minutes (Figure 2Ci arrow). The profile in Figure 2Cii shows that PrP^c internalized at 5 minutes is colocalized with CD63. After 20 minutes, PrP^c-containing vesicles (corresponding to 22.7% ± 12.5% total PrP^c staining; n = 30, results from 4 independent experiments) were colocalized with CD63 deep inside the erythroblasts at the perinuclear recycling area (Figure 2Ciii arrows). The profile in Figure 2Civ shows that all PrP^c internalized at 20 minutes is colocalized with CD63.

Tetraspanins associate specifically with partner proteins and other members of the tetraspanin family in TEMs. TEMs are distinct from lipid rafts and are thought to organize the plasma membrane and intracellular membranes by selectively concentrating specific membrane proteins.¹⁶  Nydegger et al¹⁵  demonstrate that in HeLa cells, the small fraction of CD63 found in the plasma membrane is located in microdomains together with other tetraspanins, including CD81 and CD82. In order to address the possible involvement of tetraspanin microdomains in cycling PrP^c from the plasma membrane to the endosomal compartment, dual staining of PrP^c with tetraspanins CD81 and CD82 was performed, since these are expressed in the plasma membrane of erythroblasts (Figure 3). PrP^c and the tetraspanin CD151 were also dual stained but showed only a small amount of intracellular colocalization in the perinuclear area (data not shown).

At the plasma membrane, PrP^c and CD81 are present both separately and colocalized in discrete areas distributed evenly over the cell surface (Figure 3Ai,ii). The profile of red and green fluorescence from the plasma membrane (Figure 3Aiii) shows some colocalized peaks of fluorescence. Colocalization was also observed in vesicles just under the plasma membrane and in the perinuclear area (Figure 3Ai,ii). The profile of red and green fluorescence from the cell cross-section (Figure 3Aiii) shows colocalization of PrP^c and CD81 at the cell surface and internally.

On the plasma membrane there are small areas of colocalization between PrP^c and CD82 where cells are in contact with one another (Figure 3Bi,ii). Internally, PrP^c and CD82 show colocalization around the perinuclear area, but none was observed in vesicles under the plasma membrane (Figure 3Bi,iii).

In order to assess whether or not PrP^c and CD81 share the same microdomains on the plasma membrane, PrP^c was cross-linked with antibody, and the PrP^c clusters were examined for the presence of CD81 and vice versa.

Analysis of confocal fluorescence images showed that cross-linking of PrP^c led to its redistribution from small, fairly evenly spread dots of fluorescence over the cell surface to larger coalesced patches, and that CD81 redistributed into the PrP^c patches. This is demonstrated by the profiles of red and green fluorescence from the plasma membrane of noncapped and capped cells (Figure 4Ai,ii). Similarly, PrP^c staining colocalized with CD81-induced patches (Figure 4Bi,ii).

In contrast, CD59 did not redistribute at all after anti-PrP^c or anti-CD81 induction (Figure 4Ci,ii). After anti-CD59–induced, patching partial copatching of PrP^c and CD81 was observed (illustrated by the profiles of red and green fluorescence; Figure 4Di,ii).

Evidence for colocalization of PrP^c with CD81 and CD63 raised the possibility that PrP^c might bind directly to the tetraspanins. PrP^c could be immunoprecipitated from cultured erythroblasts (day 7) after lysis with the zwitterionic detergent CHAPS and nonionic detergents Brij97, NP-40, or TX-100 (all used at 1%). In contrast, internalized PrP^c could be immunoprecipitated with NP-40 or Brij97, weakly with CHAPS, but not at all with Brij99, suggesting internalized PrP^c is present in a compartment needing more stringent detergent lysis conditions in order to solubilize PrP^c (data not shown).

Western blotting of PrP^c immunoprecipitates with anti-CD81 or anti-CD63 did not demonstrate coimmunoprecipitation under any of the detergent conditions. We could not demonstrate direct binding of either CD63 or CD81 to PrP^c immunoprecipitated from whole cell lysates, the cell surface or internalized PrP^c. Neither could we demonstrate the presence of PrP^c in immunoprecipitates obtained with anti-CD81 or anti-CD63 (data not shown).

Normal prion protein (PrP^c) must be present for prion disease to occur.^3,4  PrP^c is a widely expressed cell-surface GPI-anchored protein that can be converted to a misfolded infectious isoform by interaction with another molecule of misfolded prion protein (PrP^sc), thereby initiating an autocatalytic process which ultimately results in prion disease. Compelling evidence that vCJD can be transmitted by transfusion of red cell preparations in man⁹  has highlighted the need for a greater understanding of the distribution and biology of normal prion protein in human blood and blood-forming tissues so that the risk of disease transmission by individual blood components can be evaluated. In the present study, we examined the distribution and trafficking of PrP^c in cultured human erythroblasts. PrP^c is found on the cell surface, but most is in the perinuclear region of pronormoblasts, and the amount of PrP^c declines as erythroid cells mature. Colocalization experiments using the endosomal marker CD63 suggest most of the PrP^c is in the endosomal compartment. The distribution of PrP^c in the plasma membrane of pronormoblasts was markedly different from that of other GPI-anchored proteins (CD59, CD55), and the rate of internalization of PrP^c was much faster than that of CD59. PrP^c colocalized with the transferrin receptor, which is known to be endocytosed in clathrin-coated vesicles. GPI-linked proteins in general occupy “lipid rafts” and are internalized slowly through clathrin-independent mechanisms.¹²  These data suggest that a significant proportion of PrP^c on the plasma membrane of erythroblasts is not located in “lipid rafts.”

These findings are in agreement with observations of PrP^c internalization kinetics in primary neurons where PrP^c internalizes much faster than Thy-1 and almost as quickly as TfR.¹³ 

Sunyach et al¹³  concluded that PrP^c leaves its lipid raft environment while still on the cell surface and moves to a nonraft environment where it enters coated pits and is rapidly endocytosed.

These data raise the question as to the nature of the nonraft environment occupied by PrP^c and the mechanism by which it is targeted to coated pits. Our data show that the tetraspanin CD63 colocalizes with PrP^c within the cell and is internalized at a similar rate to PrP^c. The role of CD63 in protein trafficking has been demonstrated by the interaction of the ion pump H,K-ATPase β-subunit and CD63 at the cell surface.¹⁴  Enhanced internalization of the H,K-ATPase β-subunit was shown when it was CD63 associated. This internalization appeared to be mediated in part by μ2 and μ3.¹⁴  The μ subunits of the adaptor protein complexes 2 and 3 interact with the tyrosine-based motif in CD63.¹⁷  AP-2 participates in clathrin-mediated endocytosis, while AP-3 is involved in lysosomal targeting.^17,18  In order to internalize within clathrin-coated pits, PrP^c may need a transmembrane protein with which to piggyback since it lacks a cytoplasmic domain. The GPI-linked urokinase receptor piggybacks with the LDL receptor–related protein.¹⁹ 

TEMs form preferentially at and around clathrin-coated pits.¹⁵  Therefore, we considered the possible involvement of tetraspanin-enriched domains (TEMs) in the process of PrP^c endocytosis and compared the distribution of PrP^c with that of the tetraspanins CD81 and CD82, which colocalize with CD63 in HeLa cells.¹⁵ 

Confocal microscopy revealed that a significant proportion of PrP^c at the cell surface is associated with CD81. Furthermore patching with anti-CD81 or anti-PrP^c increases the degree of association between the proteins, indicating that they can share a similar microenvironment. The role of CD81 as a facilitator of membrane organization has been demonstrated in other systems. It plays a central role as a scaffolding protein for G-protein–coupled receptor transduction,²⁰  and is essential for the raft-stabilizing function of the B-cell coreceptor complex CD19/CD20/CD81.²¹  However, we were unable to demonstrate a direct interaction between PrP^c and CD81 or CD63 by immunoprecipitation.

Endosomal-mediated recycling of proteins to the plasma membrane may also result in the release of exosomes. Exosomes are known to contain PrP^c, CD63, and CD81.²²  Exosomes bearing PrP^sc are infectious.²³ 

Based on these data, we propose a model (Figure 5) in which PrP^c is assembled at the cell surface in tetraspanin microdomains containing CD81 and then internalized in endosomal vesicles containing CD63. This model is analogous to that proposed by Vogt et al²⁴  for MHC class II antigen presentation.

Several lines of evidence suggest that PrP^c is resistant to conversion to PrP^sc when it is located in lipid rafts (reviewed in Morris et al²⁵ ). It is therefore more likely PrP^c that is outside of lipid rafts, perhaps in TEMs, is the focus for PrP^sc binding and conversion. Cell-surface TEMs are involved in the pathogenesis of many viral diseases. They are the site at which HIV-1 proteins gather for potential particle exit.¹⁵  CD9 and CD81 play an important role in membrane fusion induced by the HIV-1 envelope.²⁶  The HTLV-1 Gag protein is found enriched at TEMs, with the inner loop of CD81 and CD82 mediating the interactions.²⁷  CD81 is a receptor for hepatitis C virus (HCV). The HCV envelope protein E2 binds to the major extracellular loop of CD81²⁸ ; in concert with an unidentified partner, it mediates hepatitis C entry into cells.²⁹ 

Recent evidence that MoMuLV (Moloney murine leukemia virus) infection strongly enhances the release of scrapie infectivity into the supernatant of coinfected cells led LeBlanc et al³⁰  to propose that retroviruses are cofactors involved in distributing PrP^sc to different tissues.

Clearly, further work is necessary to elucidate whether or not TEMs are directly involved in the pathogenesis of prion diseases, but our data suggest this approach represents a promising area for future research. Our observations also show that the mechanism of PrP^c cycling in erythroid cells is similar to that in neuronal cells, raising the question as to whether or not erythroid cells can be infected by PrP^sc.

We thank the Medical Research Council for providing an Infrastructure Award to establish the School of Medical Sciences Cell Imaging Facility, University of Bristol.

This work was supported by the United Kingdom Department of Health.

Contribution: R.G. designed and performed research, analyzed data, and wrote the paper. K.H. performed research. D.A. designed research, analyzed data, and wrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: David J. Anstee, Bristol Institute for Transfusion Sciences, National Blood Service, Southmead Road, Bristol, BS10 5ND United Kingdom; e-mail:david.anstee@nbs.nhs.uk.