Aggregation of cellular prion protein is initiated by proximity-induced dimerization

Abstract

Prion diseases or transmissible spongiform encephalopathies (TSEs) are infectious and fatal neurodegenerative disorders in humans and animals. Pathological features of TSEs include the conversion of cellular prion protein (PrP^C) into an altered disease-associated conformation generally designated PrP^Sc, abnormal deposition of PrP^Sc aggregates, and spongiform degeneration of the brain. The molecular steps leading to PrP^C aggregation are unknown. Here, we have utilized an inducible oligomerization strategy to test if, in the absence of any infectious prion particles, the encounter between PrP^C molecules may trigger its aggregation in neuronal cells. A chimeric PrP^C composed of one (Fv1) or two (Fv2) modified FK506-binding protein (Fv) fused with PrP^C were created, and transfected in N2a cells. Similar to PrP^C, Fv1-PrP and Fv2-PrP were glycosylated, displayed normal localization, and anti-apoptotic function. When cells were treated with the dimeric Fv ligand AP20187, to induce dimerization (Fv1) or oligomerization (Fv2) of PrP^C, both dimerization and oligomerization of PrP^C resulted in the de novo production, release and deposition of extracellular PrP aggregates. Aggregates were insoluble in non-ionic detergents and partially resistant to proteinase K. These findings demonstrate that homologous interactions between PrP^C molecules may constitute a minimal and sufficient molecular event leading to PrP^C aggregation and extracellular deposition.

Abbreviations used

EGFP

enhanced green fluorescent protein

Fv1

FK506 binding domain

GPI

glycosylphosphatidylinositol anchor

HA

influenza hemagglutinin epitope

MCF

human mammary adenocarcinoma

PK

proteinase K

PrP

prion protein

TSEs

transmissible spongiform encephalopathies

The cellular prion protein (PrP^C) is a highly conserved and ubiquitously expressed glycosylphosphatidylinositol (GPI)-anchored protein, and plays a central role in a group of infectious neurodegenerative diseases known as transmissible spongiform encephalopathies (TSEs) or prion diseases (Prusiner 1998; Chesebro 2003; Aguzzi and Polymenidou 2004; Johnson 2005). Ablation of the Prnp gene which encodes PrP^C confers resistance to experimentally-induced TSEs in mice (Bueler et al. 1993), and familial human TSEs exhibit mutations of Prnp (Kong et al. 2004). Prions are infectious agents responsible for the transmission of TSEs and are composed of an abnormal protease-resistant isoform of PrP^C, labelled PrP^Sc. Particles containing 14–28 PrP^Sc molecules are the most infectious (Silveira et al. 2005).

Transmissible spongiform encephalopathies include scrapie in sheep and rodents; spongiform encephalopathy in bovine; sporadic, familial, and iatrogenic Creutzfeldt-Jakob disease in humans (Prusiner 1998). The vast majority of these disorders are characterized by accumulation of disease-associated extracellular deposits of PrP, and spongiform degeneration of the brain (Budka 2003). Deposition patterns of PrP include diffuse, patchy, and plaque-like patterns (Budka et al. 1995). Prion deposits in the brain contain protease-resistant PrP^Sc molecules with the ability to convert the protease-sensitive PrP^C into PrP^Sc, and use this mechanism to propagate. Spongiform changes correspond to the formation of small, round or oval vacuoles in axonal and dendritic processes, as well as in synapses (Jeffrey et al. 1995). In designing strategies to prevent the propagation and the progression of TSEs, it is important to determine the molecular pathway leading to PrP^C aggregation.

Since deposits are composed of aggregated PrP molecules, the first step of their production may consist in the induction of proximity between individual PrP^C molecules. Accordingly, enforcing dimerization or/and oligomerization of PrP^C could result in the formation of such deposits, provided that no other factors are required for conformational rearrangement and aggregation. One way to force PrP^C to dimerize is to use cross-linking monoclonal antibodies. In previous work, cross-linking of PrP^C with monoclonal antibodies resulted in the transduction of intracellular signals in a differentiated neuronal cell line, and in neuronal apoptosis in mice (Mouillet-Richard et al. 2000; Solforosi et al. 2004). The presence of pathological prion deposits in these studies was not reported. However, monoclonal antibodies may interfere with conformational rearrangement of PrP^C, thus preventing prion aggregation (Enari et al. 2001; Peretz et al. 2001). Furthermore, they may introduce structural constraints that prevent the formation of high-order prion oligomerization states.

Here, we further explored the hypothesis that dimerization of PrP^C is a key molecular step in the pathology of TSEs, and we developed another strategy to bring together PrP^C molecules. Fusion proteins between human PrP^C and one or two copies of an FK506 binding domain (Fv) were engineered. In the presence of AP20187, a homodimerizer ligand that binds Fv, proteins containing one or two Fv modules are forced to interact and to dimerize or oligomerize, respectively. This strategy allows a fine regulation of induced dimeric or oligomeric interactions between Fv-containing proteins (Spencer et al. 1993; Clackson et al. 1998; Yang et al. 2000; Gazdoiu et al. 2005). Similar to PrP^C, we report that PrP^C genetically fused to one (Fv1-PrP) or two (Fv2-PrP) Fv domains, displays post-translational glycosylations, localization in the secretory pathway and at the plasma membrane, and anti-apoptotic function. Upon addition of AP20187, Fv1-PrP and Fv2-PrP spontaneously form large amounts of extracellular aggregates insoluble in non-ionic detergents and partially resistant to proteinase K (PK). This work demonstrates that dimerization is a key molecular step in the aggregation of PrP^C.

Experimental procedures

Antibodies

Monoclonal anti-prion (clone 3F4), anti-influenza hemagglutinin epitope (HA) (clone HA.11), and anti-enhanced green fluorescent protein (EGFP) (clone B-2) were purchased from Serotec (Raleigh, NC, USA), Covance (Quebec, QC, Canada), and SantaCruz Biotechnology (Santa Cruz, CA, USA), respectively. Alexa Fluor 488 F(ab’)2 fragment of goat anti-mouse IgG was purchased from Molecular Probes (Burlington, ON, Canada). Peroxidase-linked anti-mouse IgG from sheep was purchased from Amersham Biosciences (Uppsala, Sweden).

Clones - Fv1-PrP, Fv2-PrP, ΔGPI-PrP, Fv1-ΔGPI-PrP, Fv2-ΔGPI-PrP: Fv1 and Fv2 were amplified from pC4-Fv1E and pC4 M-Fv2E (vectors kindly provided by ARIAD Pharmaceuticals, Cambridge, MA, USA; http://www.ariad.com/regulationkits), respectively. Polymerase chain reaction (PCR) primers for Fv1-PrP were Fv1(PrP)-forward 5′-gggttctagaggagtgcaggtggagactatctcc-3′ and Fv1(PrP)-reverse 5′-gggcgtagtctggtacgtcgtacggtattg-3′. PCR primers for Fv2-PrP were Fv2(PrP)-forward 5′-gggttctagaggcgtccaagtcgaaaccattagtcc-3′ and Fv1(PrP)-reverse. PCR products were inserted in the natural SmaI site of human Prnp at bp113 (amino-acid 38). The resulting Fv1-PrP and Fv2-PrP cDNA constructs were introduced in the HindIII and BamHI sites of pCEP4β (Invitrogen, Burlington, ON, Canada). ΔGPI-PrP and Fv1-ΔGPI-PrP were amplified from pCEP4β-PrP and pCEP4β-Fv1-PrP, respectively. PCR primers were ΔGPI-PrP-forward 5′-cccaagcttgtaatggcgaaccttggctgctgg-3′, and ΔGPI-PrP-reverse 5′-cgcggatcctcacgatcctctctggtaataggcctg-3′. PCR products were introduced in the HindIII and BamHI sites of pCEP4β. Fv1-GPI-EGFP, Fv2-GPI-EGFP, ΔGPI-EGFP, Fv1-ΔGPI-EGFP: Fv1 and Fv2 were amplified from pC4-Fv1E and pC4 M-Fv2E (vectors kindly provided by ARIAD Pharmaceuticals), respectively. PCR primers for Fv1-GPI-EGFP were Fv1(GPI-EGFP)-forward 5′-aactgcagatggcttctagaggagtgcaggtg-3′ and Fv1(GPI-EGFP)-reverse 5′-cgggatcccgtgcgtagtctggtacgtcgtacgg-3′. PCR primers for Fv2-GPI-EGFP were Fv2(GPI-EGFP)-forward 5′-aactgcagatggggagtagcaagagcaagcctaag-3′ and Fv2(GPI-EGFP)-reverse. PCR products were inserted in Pst1 and BamH1 sites located between the N-terminal signal peptide and EGFP encoding sequences of pEGFP-N1-GPI (Nichols et al. 2001). The resulting Fv1-GPI-EGFP and Fv2-GPI-EGFP cDNA constructs were amplified using GPI-EGFPforward 5′-ggggtaccagctagccaccatggagctc-3′ and GPI-EGFPreverse 5′-cccaagcttagttctagaatgaccgctgcttgg-3′. PCR products were introduced in the KpnI and HindIII sites of pCEP4β. ΔGPI-EGFP and Fv1-ΔGPI-EGFP were amplified from pCEP4β-GPI-EGFP and pCEP4β-Fv1-GPI-EGFP, respectively. PCR primers were GPI-EGFPforward, and ΔGPI-EGFP-reverse 5′-cccaagcttagtcttgtacagctcgtccatgcc-3′. PCR products were introduced in the KpnI and HindIII sites of pCEP4β. All constructs were sequenced.

Cell culture, transfections, and treatment

Mouse N2a neuroblastoma cells, human embryonic kidney HEK293, human breast carcinoma human mammary adenocarcinoma (MCF)-7, and human cervical Hela were maintained in Dulbecco’s modified Eagle’s medium plus 10% fetal bovine serum (Wisent, St. Bruno, QC, Canada). Human neuroblastoma BE-(2)-M17 cells were maintained in Optimem medium plus 10% fetal bovine serum (Wisent). Transfections were carried out using lipofectamine according to the manufacturer’s protocol (Invitrogen). When transfecting two constructs, 1 : 3 DNA ratio of pCEP4β-EGFP-Bax:pCEP4β, pCEP4β-EGFP-Bax:pCEP4β-PrP, pCEP4β-EGFP-Bax:pCEP4β-Fv1-PrP, and pCEP4β-EGFP-Bax:pCEP4β-Fv2-PrP were used to transfect MCF-7 cells. Three hours after transfection, medium was changed with fresh medium. In some experiments, fresh medium contained 100 nmol/L AP20187 solubilized in ethanol (kindly provided by ARIAD Pharmaceuticals), or the same volume of ethanol. To test the glycosylation of Fv1-PrP and Fv2-PrP, medium was changed with fresh medium containing 1.5 μmol/L tunicamycin solubilized in dimethylsulfoxide, or the same volume of dimethylsulfoxide. The effect of chlorpromazine and quinacrine on AP20187-induced Fv1- and Fv2-PrP aggregation was tested by adding the desired concentrations to the medium containing AP20187. Cells were incubated 14 h and processed for fluorescence or immunoblot analyses. In some experiments, cell monolayers were dissociated in a versene solution (0.2 g/L EDTA.4Na in phosphate-buffered saline).

Immunofluorescence and microscopy

Cells were fixed and processed for immunofluorescence as previously described (Grenier et al. 2006). Primary antibodies dilutions were as followed: anti-PrP 1/200, anti-HA 1/1000 (Fv1 and Fv2 contain the influenza hemagglutinin HA epitope tag). Secondary antibodies were diluted 1/1000. Cells were examined with an Axioscop 2 phase-contrast/epifluorescence microscope (Carl Zeiss) equipped with band pass filters for fluorescence of Hoechst (Ex. D360/40 : Em. D460/50), FITC (Ex. D480/30 : Em. D535/40), and tetramethylrhodamine isothiocyanate (Ex. D560/40 : Em. D630/60) (Chroma Technology, Rockingham, VT, USA). Photomicrographs of 1315 × 1033 pixels were captured using 63 × oil immersion or dry 40 × objectives and Spot cooled color digital camera (Diagnostic Instruments, Sterling Heights, MI, USA). Images were processed using spot software (Diagnostic Instruments). Within the same figure, all pictures were taken with the same exposure time. For ultrastructural studies, cells were processed as previously described (Grenier et al. 2006).

Protein expression, analyses of PrP aggregation, differential centrifugation

Protein expression was determined in lysates from cells grown in 6-well plates as previously described (Roucou et al. 2003). For PrP aggregation, all cellular and extracellular material was collected by directly lysing cells in 6-well plates. For each well, cells were lysed with 0.2 ml of lysis buffer [50 mmol/L Tris-HCl, pH 7.5, 150 mmol/L NaCl, 2 mmol/L EDTA, 0.5% Triton X-100 (v/v), and 0.5% sodium deoxycholate (w/v)] for 20 min at 4°C. After centrifugation at 10 000 g for 10 min at 4°C, soluble PrP was present in the supernatant and aggregated PrP was present in the pellet. Proteins from the pellet were resuspended in 0.2 mL of lysis buffer. Proteins in both supernatants and pellets were precipitated with four volumes of methanol and analyzed by western blotting.

To determine the aggregation of PrP released in the medium, 6 × 10⁶ cells were detached with a rubber policeman. The cell suspension was centrifuged 5 min at 3000 g for 5 min to collect the cells. The medium was ultracentrifuged at 10 000 g for 30 min and at 100 000 g for 1 h. Pellets 3000 g were lysed in 1 ml of lysis buffer. Pellets 10 000 g and 100 000 g were lysed in 0.1 mL of lysis buffer as described above. The lysates were ultracentrifuged at 100 000 g for 1 h. Proteins from the pellet (insoluble PrP) were resuspended in 0.1 ml of lysis buffer. Supernatants contained soluble PrP. Proteins were precipitated with four volumes of methanol and analyzed by western blotting.

Proteinase K resistance

After transfection and treatment, cells growing on coverslips were rinsed twice with phosphate-buffered saline, and incubated 30 min at 37°C in a 0.5 mL reaction volume. Digestions were terminated with 5 mmol/L phenylmethylsulphonylfluoride. Cells were fixed and processed for immunofluorescence. PK resistance of PrP in whole cell lysates was determined as previously described (Roucou et al. 2003).

Western blotting

Anti-PrP and anti-EGFP primary antibodies were diluted 1/10 000 and 1/1000, respectively. Secondary antibodies were diluted 1/5000.

Results

Characterization of Fv1-PrP and Fv2-PrP

Fv modules were inserted in the unstructured N-terminal domain of PrP^C to generate Fv1- and Fv2-tagged PrP^C (Fig. 1a). PrP constructs were expressed in murine N2a cells, since they are the most intensively described and used in vitro model to study the cell biology and propagation of prions. PrP constructs are from human origin and endogenous mouse PrP is not detected with 3F4 monoclonal antibodies. Both Fv1-PrP and Fv2-PrP migrated on a sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) as several bands, indicating the presence of glycosylations similar to PrP^C. Sensitivity to tunicamycin, an inhibitor of N-linked glycosylation, confirmed that fusion proteins were glycosylated (Fig. 1b). Next, we determined whether addition of Fv1 modules interfered with the localization of PrP^C (Fig. 1c). Fv1-PrP and Fv2-PrP were detected at the plasma membrane in non-permeabilized cells. In permeabilized cells, Fv1-PrP and Fv2-PrP were also detected in the secretory pathway, mainly in the Golgi (Fig. 1c). This distribution was similar to that of PrP^C (Fig. 1c). Finally, we determined if the activity of PrP^C is retained after the addition of Fv domains. PrP^C inhibits Bax-mediated cytochrome c release and cell death in breast carcinoma MCF-7 cell line (Roucou et al. 2005). To determine if Fv1- and Fv2-PrP retain anti-apoptotic activity, MCF-7 cells were transfected with EGFP-Bax cDNA, or EGFP-Bax cDNA in combination with PrP^C, Fv1-PrP, or Fv2-PrP cDNAs. Cells were immunostained with anti-cytochrome c antibodies. EGFP-Bax localized mainly at the mitochondria and induced the loss of mitochondrial cytochrome c immunostaining. Cells expressing EGFP-Bax also displayed condensed nuclei typical of cells undergoing apotosis. In contrast, coexpression of EGFP-Bax with PrP^C, Fv1- or Fv2-PrP inhibited cytochrome c release and nuclear condensation. Thus, introduction of one or two Fv domains in the N-terminal region does not interfere with expression, post-translational modification, localization, and anti-apoptotic function of PrP^C.

Figure 1.  Expression and subcellular localization of cellular prion protein (PrP^C), Fv1-PrP, and Fv2-PrP. (a) Representation of PrP^C, Fv1-PrP, and Fv2-PrP. Black boxes between residues 1–22 and 231–254 indicate N-terminal and C-terminal signal peptides, respectively. Grey and white boxes indicate unstructured and structured domains, respectively. (b) Glycosylation of Fv1-PrP and Fv2-PrP. Transfected N2a cells were grown in the absence or in the presence of 1.5 μg/mL tunicamycin. Cells were lysed and processed for immunoblotting using anti-PrP antibodies (3F4 mAb). Molecular weight standards are shown on the right. (c) Cellular distribution of Fv1-PrP and Fv2-PrP. Cells expressing PrP^C, Fv1- and Fv2-PrP were fixed and processed for immunofluorescence before (−TX-100) or after (+TX-100) permeabilization with Triton X-100 (3F4 mAb, Green). Nuclei were stained with Hoechst (blue). Green and blue channels are shown merged. Scale bar: 10 μm. Original magnification ×63.

Enforced dimerization of PrP^C in cultured cells induces its aggregation

To assess if addition of AP20187 to cells expressing Fv1-PrP and Fv2-PrP could trigger the formation of aggregates, the solubility of Fv1-PrP and Fv2-PrP in mild detergents was determined. Cells were directly lysed in the culture wells to collect any extracellular aggregates. In mock-treated cells, Fv1-PrP and Fv2-PrP, like PrP^C, were mostly present in the soluble fraction of cell lysates (Fig. 2a). In contrast, Fv1-PrP and Fv2-PrP became insoluble upon addition of 100 nmol/L AP20187 (Fig. 2a), demonstrating that dimerization/oligomerization of PrP^C spontaneously causes its aggregation. The localization and molecular morphology of PrP aggregates were determined by immunofluorescence. In the absence of AP20187, PrP^C, Fv1-PrP and Fv2-PrP were detected at the plasma membrane (Fig. 1c). In AP20187-treated cell cultures expressing Fv1- or Fv2-PrP, both Fv1-PrP and Fv2-PrP were detected in extracellular deposits (Fig. 2b). In contrast, PrP^C remained associated with the plasma membrane of AP20187-treated PrP^C-expressing cells (Fig. 2b). Fv1-PrP and Fv2-PrP deposits were produced in massive amount, and exhibited diffuse, patchy-like and dense-like patterns. Some deposits were located in areas where no cells were visible, indicating that Fv1-PrP and Fv2-PrP had been released from the plasma membrane and had spread to extracellular spaces. The conversion of Fv1-PrP and Fv2-PrP into deposits is extremely efficient as no detectable protein remains associated with the plasma membrane. In control experiments, a mono-functional ligand of Fv1, FK506, did not induce the formation of extracellular Fv1-PrP and Fv2-PrP aggregates (not shown). The extracellular nature of PrP aggregates was confirmed after treatment with a versene solution. While versene-treated cells detached from their support, PrP deposits remained intact (Fig. 2c). Deposits did not co-localise with three proteins of the extracellular matrix, laminin, fibronectin, and collagen (data not shown).

Figure 2.  Extracellular prion deposits induced by enforced dimerization/oligomerization of cellular prion protein (PrP^C). (a) AP20187-mediated aggregation. Supernatant (S) and pellet (P) fractions of detergent lysates from PrP^C-, Fv1-PrP, and Fv2-PrP-expressing N2a cells incubated in the absence or in the presence of 100 nmol/L AP20187, were analyzed by immunoblotting. Molecular weight standards are shown on the right. (b) AP20187-mediated deposition and vacuolation. Immunofluorescence of AP20187-treated N2a cells expressing PrP^C, Fv1-PrP, or Fv2-PrP (3F4 mAb, green). Cells were permeabilized with Triton X-100. Nuclei were stained with Hoechst (blue). Top panels, green and blue channels are shown merged. Bottom panels, phase-contrast views of top panel images. Asterisks indicate prion deposits in areas where no cells are visible. Scale bar: 10 μm. Original magnification ×63. (c) Extracellular deposits. N2a cells expressing PrP^C, Fv1-PrP, or Fv2-PrP were grown on coverslips and treated with 100 nmol/L AP20187 or the same volume of vehicle. After 24 h, cells were incubated with a versene solution for 10 min. Detached cells were removed and coverslips were processed for immunofluorescence with 3F4 mAb. Green and visible channels are shown merged. Although no cells are visible by phase-contrast, prion deposits are still present on coverslips when cells have been incubated with AP20187 before versene treatment. (d) Detergent lysates of N2a cells expressing PrP^C, Fv1-PrP, or Fv2-PrP were analyzed by immunoblotting (3F4 mAb). Lysates contain extracellular and intracellular material (see Experimental Procedures). Equal loading was verified with anti-actin antibodies. Molecular weight standards are shown on the right. (e) The cell suspension was centrifuged at 3000 g to collect the cells. The culture medium was ultracentrifuged at 10 000 g for 30 min to remove cell debris, and the supernatant was ultracentrifuged at 100 000 g for 1 h. The resulting pellets of each centrifugation step were solubilized, centrifuged to separate soluble (supernatant, S) and insoluble (pellet, P) fractions, and analyzed by western blot for PrP.

Levels of Fv1- and Fv2-PrP in cells treated with AP20187 were dramatically increased compared with mock-treated cells (Fig. 2d). These observations were not unexpected since extracellular Fv1- and Fv2-PrP deposits would not be degraded by intracellular proteolytic enzymes that are normally involved in the catabolism of PrP^C. PrP^C levels remained unchanged in cells incubated with AP20187 (Fig. 2d). No species with altered electrophoretic mobility were detected in AP20187-treated cultures, suggesting that PrP molecules in extracellular deposits are neither proteolytically cleaved, nor do they lack the GPI anchor.

It was important to determine if the aggregation of PrP occurred at the cellular level or resulted from the binding of released dimers/multimers to the plastic of the culture wells. To address this issue, cells were collected and the cell culture medium was submitted to sequential centrifugation steps with increasing centrifugal forces. Each pellet was solubilized, and soluble and insoluble fractions analyzed by western blotting with 3F4 monoclonal antibodies (Fig. 2e). PrP^C was soluble independently of the presence of AP20187. PrP^C was detected in very small amount in the soluble fraction of the 100 000 g pellet, confirming that N2a cells spontaneaously release some PrP^C molecules in the culture medium in association with exosomes (Fevrier et al., 2004). In contrast, Fv1- and Fv2-PrP were insoluble after AP20187 treatment. Furthermore, the amount of Fv1-PrP and Fv2-PrP collected in the 100 000 g pellet after AP20187 treatment was largely increased compared with untreated cells. Some Fv1- and Fv2-PrP was collected at 10 000 g after treatment with AP20187, indicating the release of PrP aggregates in a relatively large complex. These results demonstrate that AP20187-induced aggregation of PrP takes place at the cellular level.

To test if any GPI-anchored protein would aggregate upon enforced oligomerization, we engineered a GPI-anchored EGFP protein bearing one (Fv1-GPI-EGFP) or two (Fv2-GPI-EGFP) Fv modules (Fig. 3a). Both Fv1-GPI-EGFP and Fv2-GPI-EGFP did not form extracellular deposits upon addition of AP20187 (Fig. 3b). Thus, the information responsible for aggregation, release, and extracellular deposition of Fv1- and Fv2-PrP is encrypted only in the amino acids sequence of the PrP^C moiety, and neither in the Fv module nor in the GPI anchor.

Figure 3.  Forced dimerization or oligomerization of GPI-EGFP does not result in the formation of EGFP deposits. (a) Representation and expression of GPI-EGFP, Fv1-GPI-EGFP, and Fv2-GPI-EGFP. Black boxes between residues 1–24 and 239–260 indicate N-terminal and C-terminal signal peptides, respectively. Grey box indicates EGFP protein. Fv modules contain 120 residues. Note that domains are not in scale. Expression of GPI-EGFP, Fv1-GPI-EGFP, and Fv2-GPI-EGFP was analyzed by immunoblotting with anti-EGFP antibodies. Molecular weight standards are shown on the right. (b) Fluorescence of mock- and AP20187-treated N2a cells expressing GPI-EGFP, Fv1-GPI-EGFP, or Fv2-GPI-EGFP. Nuclei were stained with Hoechst (blue). Green and blue channels are shown merged. Scale bar: 10 μm. Original magnification ×40.

AP20187-mediated aggregates are partially resistant to proteinase K and sensitive to chlorpromazine and quinacrine

The PrP^Sc in pathological deposits is generally resistant to PK treatment; the first approximately 90 residues are very sensitive to PK, whereas the rest of the protein is resistant (Parchi et al. 2000). To evaluate whether deposits resulting from enforced dimerization/oligomerization of PrP^C were also PK-resistant, mock- and AP20187-treated cells expressing Fv1-PrP or Fv2-PrP were incubated with increasing concentrations of PK (Fig. 4). In the absence of AP20187, Fv1-PrP and Fv2-PrP were completely degraded in the presence of concentrations of PK as low as 0.25 μg/mL. In contrast, PrP deposits in cell cultures treated with AP20187 were resistant to doses of PK up to 10 times higher (Fig. 4b). PrP aggregates were remarkably stable and resistant to degradation. After 42 days of culture without changing medium, AP20187-treated cells expressing PrP^C, Fv1-PrP, and Fv2-PrP had degenerated and died. While PrP^C was already undetectable after 10 days of culture, dense deposits of Fv1-PrP and Fv2-PrP were still detected after 42 days. The resistance of PrP aggregates to PK was also determined in whole cell lysates (Fig. 4b). In untreated cells, Fv1- and Fv2-PrP were degraded by low concentrations of PK. In contrast, PK-resistant fragments were detected in AP20187-treated Fv1- and Fv2-PrP expressing cells. PK resistant fragments had similar electrophoretic mobility, suggesting that Fv1- and Fv2-PrP aggregates resulting from enforced dimerization or oligomerization of PrP are similar.

Figure 4.  Resistance of AP20187-induced prion protein (PrP) aggregates to proteinase K (PK), and sensitivity to chlorpromazine and quinacrine. (a) N2a cells expressing Fv1-PrP or Fv2-PrP were mock-treated or treated with 100 nmol/L AP20187, and incubated with various concentrations of PK. Cells were processed for immunofluorescence (3F4 mAb, green). Nuclei were stained with Hoechst (blue). All panels show merged pictures. PK concentrations are indicated in μg/mL. Scale bar: 10 μm. Original magnification ×63. (b) Western blot analysis (3F4) of PrP^C, Fv1-PrP, and Fv2-PrP in cell lysates from N2a cells incubated in the absence or in the presence of 100 nmol/L AP20187, treated with increasing concentrations of PK. Molecular weight standards are shown on the right. (c) Inhibition of AP20187-mediated aggregation of Fv1-PrP by chlorpromazine and quinacrine. Soluble (supernatant, S) and insoluble (pellet, P) fractions of detergent lysates from Fv1- and Fv2-PrP-expressing N2a cells incubated in the presence of 100 nmol/L AP20187 and increasing concentrations of chlorpromazine and quinacrine, were analyzed by immunoblotting (3F4). Molecular weight standards are shown on the right.

Chlorpromazine and quinacrine are potent inhibitors of PrP^Sc formation in ScN2a cells (Korth et al. 2001). We determined if these compounds were also effective in preventing AP20187-mediated aggregation of Fv1- and Fv2-PrP. At concentrations in the range of 2–6 μmol/L, chlorpromazine treatment led to the inhibition of Fv1- and Fv2-PrP aggregation (Fig. 4c). Whilst Fv1- and Fv2PrP were largely insoluble in AP20187-treated cells, the proteins were completely soluble in lysates from cells treated with AP20187 in the presence of 6 μmol/L chlorpromazine (Fig. 4c). Quinacrine was slightly more potent in preventing AP20187-induced Fv1- and Fv2-PrP aggregation. A concentration of 1.2 μmol/L was sufficient to prevent the aggregation of both Fv1- and Fv2-PrP (Fig. 4c).

Since the patterns and biochemical characteristics of Fv1-PrP aggregates are undistinguishable from those of Fv2-PrP aggregates, all together the results indicate that AP20187-induced PrP dimers spontaneously combine to form high-order aggregates.

Deposition patterns of PrP aggregates

Deposition patterns of aggregates are important aspects of the pathogenic process, and we tested if enforced dimerization of PrP^C in cell cultures could reproduce specific deposition patterns observed in animal models. In infected transgenic mice expressing a prion protein lacking the GPI membrane anchor (ΔGPI-PrP), aggregates accumulate in dense amyloid plaques, unlike the diffuse and patchy-like patterns observed in non-transgenic mice (Chesebro et al. 2005). cDNA constructs encoding ΔGPI-PrP and Fv1-ΔGPI-PrP were engineered (Fig. 5a). In mock-treated cells, ΔGPI-PrP and Fv1-ΔGPI-PrP were detected in the secretory pathway, mainly in the Golgi (Fig. 5b). The distribution of ΔGPI-PrP remained unchanged in AP20187-treated cells. In contrast, AP20187-treated cells expressing Fv1-ΔGPI-PrP produced deposits with exclusive dense-like deposits (Fig. 5b), unlike the diffuse and patchy-like Fv1-PrP deposits (Fig. 2b). In control experiments, AP20187-treated cell cultures expressing Fv1-ΔGPI-EGFP did not produce extracellular deposits (Figs 5c and d). ΔGPI-PrP forms amyloid plaques in infected transgenic mice (Chesebro et al. 2005). However, AP20187-mediated Fv1-ΔGPI-PrP deposits where not stained with thioflavinS, indicating the absence of amyloid character of these deposits in cell culture (data not shown).

Figure 5.  Forced ΔGPI-PrP dimerization results exclusively in the formation of dense deposits. (a) Representation and expression of ΔGPI-PrP and Fv1-ΔGPI-PrP. Black boxes between residues 1–22 indicate N-terminal signal peptide. Note the absence of C-terminal signal peptide (residues 232–254). Grey and white boxes indicate unstructured and structured domains, respectively. Transfected N2a cells were lysed and processed for immunoblotting using anti-PrP antibodies (3F4 mAb). Molecular weight standards are shown on the right. (b) Immunofluorescence of mock- (−) or AP20187-treated (+) cells expressing ΔGPI-PrP or Fv1-ΔGPI-PrP (3F4 mAb, green). Nuclei were stained with Hoechst (blue). Green and blue channels are shown merged. (c) Representation and expression of ΔGPI-EGFP and Fv1-ΔGPI-EGFP. Black box between residues 1–26 indicates N-terminal signal peptide. Note the absence of C-terminal signal peptide (residues 240–260). Grey box indicates EGFP. Transfected N2a cells were lysed and processed for immunoblotting using anti-EGFP antibodies. Molecular weight standards are shown on the right. (d) Fluorescence of mock- (−) or AP20187-treated (+) cells expressing ΔGPI-EGFP or Fv1-ΔGPI-EGFP (green). Nuclei were stained with Hoechst (blue). Green and blue channels are shown merged. Scale bar: 10 μm. Original magnification ×40.

Vacuolation in cells producing aggregates

We sought to examine if besides the formation of aggregates, cells would display other pathological features. Strikingly, AP20187-treated cells expressing Fv1-PrP or Fv2-PrP exhibited a spongiform aspect with vacuole-like structures visible by phase-contrast microscopy (Fig. 6a). Morphological changes often included swollen cytoplasm in addition to large and smaller vacuoles. Vacuoles were never observed in the absence of AP20187 (Figs 6a and b) or in PrP^C-transfected cells treated with AP20187 (Figs 2b and 6b). The presence of membrane-bound rounded vacuoles was confirmed by electron microscopy (Fig. 6c). Despite the presence of PrP deposits and moderate to severe vacuolation, we did not observe abnormal levels of apoptosis as shown by normal nuclear morphology (Figs 2–6).

Figure 6.  Vacuolation following enforced oligomerization of cellular prion protein (PrP^C). (a) Vacuolation and swelling. N2a cells expressing Fv1-PrP or Fv2-PrP were mock-treated or treated with 100 nmol/L AP20187. Cells were fixed and the nuclei stained with Hoechst. Blue (Hoechst) and visible channels are shown merged. Note the formation of intracellular vacuoles and enlargement of the cytoplasm. Scale bar: 10 μm. Original magnification ×63. (b) The percentage of cells with vacuoles was determined by examining cells expressing PrP^C, Fv1-PrP, and Fv2-PrP incubated in the absence (filled columns) or in the presence (empty columns) of 100 nmol/L AP20187. Values are mean ± SD of three independent experiments. More than 200 cells were counted for each condition. (c) Transmission electron microscopy of vacuoles. N2a cells expressing Fv1-PrP were untreated or treated with 100 nmol/L AP20187, and processed for electron microscopy. Right panel represents a higher magnification of the box in the middle panel. n, nucleus; m, mitochondria, v, vacuole. Bar is 500 nm.

De novo aggregation of PrP^C in various cell lines

Besides N2a cells, other neuronal and non-neuronal cells also displayed similar pathological features upon enforced dimerization/oligomerization of PrP^C. Human neuronal BE-(2)-M17, like murine neuronal N2a, cells produced PrP deposits and vacuoles. Embryonic kidney 293 and cervical Hela cells produced extracellular PrP deposits, but did not display vacuolar lesions. Thus, all tested cell lines are able to produce AP20187-mediated extracellular PrP deposits, whilst vacuolation seems to be restricted to neuronal cells.

Discussion

In the present study, we demonstrate that induction of proximity between PrP^C molecules in a cellular context with a bivalent synthetic ligand, AP20187, is sufficient for de novo production of extracellular deposits. This result strongly suggests that the encounter between two PrP^C molecules is the limiting step during the aggregation process of PrP^C. Once individual PrP^C molecules have interacted, a cascade of events encoded in the PrP^C amino-acid sequence is initiated and results in aggregation, and extracellular deposition.

Several elements indicate that AP20187-induced aggregation of PrP^C is not an artefactual aggregation irrelevant to TSEs. First, AP20187-mediated aggregation does not apply to any GPI-anchored protein but is specific to PrP^C. Second, if dimerization induces spontaneous aggregation of PrP^C in a simple cell culture system, dimerization is also likely to trigger aggregation in vivo. In acquired TSEs, PrP^Sc physically interacts with PrP^C, and may force PrP^C molecules to dimerize/oligomerize (Prusiner et al. 1990; Horiuchi and Caughey 1999; Meier et al. 2003; Chesebro et al. 2005). How PrP^C molecules are brought together in sporadic and familial TSEs is still unknown. Third, chlorpromazine and quinacrine, two compounds able to inhibit the formation of PrP^Sc in infected N2a cells were also able to prevent the formation of AP20187-induced PrP aggregates. Fourth, dimerization of PrP^C resulted in the formation of PrP aggregates in all tested cell lines, raising the possibility that in vivo, all cells expressing PrP^C are potential producers of PrP aggregates. In contrast, vacuolation occurred only in neuronal cells. Interestingly, an increasing number of reports indicate the presence of extraneural PrP deposits in human patients and in animals, while vacuolation is restricted to neurons (Bosque et al. 2002; Glatzel et al. 2003; Andreoletti et al. 2004; Thomzig et al. 2004).

Some biochemical characteristics of AP20187-induced PrP aggregates were different from those of disease-associated aggregates. PrP^Sc is generally resistant to very high concentrations of PK (above 50 μg/mL), whereas Fv1-PrP and Fv2-PrP aggregates were destroyed above 10 μg/mL of PK. However, the development of alternative assays, including conformation-dependent immunoassays, has lead to the discovery that disease-associated PrP^Sc is composed of PK-sensitive (sPrP^Sc) and PK-resistant (rPrP^Sc) forms (Safar et al. 1998, 2005). Up to 90% of PrP^Sc might be composed of PK-sensitive material (Thackray et al. 2006). AP20187-mediated PrP aggregates are mainly composed of partially resistant PK-sensitive aggregates that are evidently different from rPrP^Sc. Fv1-ΔGPI-PrP did not form amyloid deposits whereas infected transgenic animals expressing ΔGPI-PrP develop amyloid plaques in the brain (Chesebro et al. 2005). Absence of Fv1-ΔGPI-PrP amyloids indicates that Fv1 modules may introduce structural constraints that do not prevent aggregation, but do interfere with the process of amyloidosis. Alternatively, amyloidosis likely involves a mechanism more complex than simple homologous interactions between PrP^C molecules.

N2a cells infected with PrP^Sc and producing aggregates do not undergo vacuolation (Butler et al. 1988), whereas we showed here that the same cells producing Fv1- or Fv2-PrP aggregates also produce intracellular vacuoles. A possible explanation for the dissociation between aggregation and vacuolation in PrP^Sc-infected N2a cells resides in the clonal selection of these cells. N2a cultures exposed to PrP^Sc produce only low levels of infectious prions, apparently because only a small percentage of the cells become infected (Race et al. 1987, 1988). Therefore, to obtain cultures that produce sufficient quantities of PrP^Sc for biochemical analysis, prion-exposed cell cultures must be subcloned and the most highly infected sublines must be selected (Butler et al. 1988; Race et al. 1988). These PrP^Sc infected subclones may be incidentally selected for their ability to resolve vacuoles as this would likely confer a growth advantage compared with cells with vacuolar lesions. This hypothesis is consistent with a recent report describing that in a mouse model of TSEs, prion deposition can be dissociated from neuronal vacuolation (Mallucci et al. 2003). The origin of the vacuoles is not clear. Interestingly, analysis of the vacuoles by electron microscopy showed that they have a size similar to that of mitochondria, raising the possibility that vacuoles might originate from mitochondria. Further investigation is underway.

In summary, inducible dimerization of PrP^C in cultured cells has revealed an initial molecular step in the pathway of PrP^C aggregation and deposition. This system should facilitate the decoding of cellular and molecular mechanisms involved in the pathogenesis of TSEs, including aggregation, extracellular deposition of PrP^C, and formation of intracellular vacuoles.

Acknowledgements

We thank Victor M. Rivera (ARIAD Pharmaceuticals, Cambridge, MA, USA) for providing the pC₄-F_v1E and pC4 M-Fv2E plasmids, and the AP20187 chemical dimerizer; Dr Ben Nichols (MRC Laboratory of Molecular Biology, Cambridge, UK) for providing the plasmid encoding GPI-EGFP; the people from the electron microscopy facility of the Université de Sherbrooke; Drs Jean-Pierre Perreault and Simon Labbé for comments on the manuscript. This work was supported by a grant from Canadian Institutes for Health Research to XR. XR is a Junior 2 research scholar from the Fonds de la Recherche en Santé du Québec.