Dynamic interactions of Sup35p and PrP prion protein domains modulate aggregate nucleation and seeding

Abstract

Prions are self-propagating infectious protein aggregates of mammals and fungi. The exact mechanism of prion formation is poorly understood. In a recent study, a comparative analysis of the aggregation propensities of chimeric proteins derived from the yeast Sup35p and mouse PrP prion proteins was performed in neuroblastoma cells. The cytosolic expression of the Sup35p domains NM, PrP and fusion proteins thereof revealed that the carboxyterminal domain of PrP (PrP_90–230) mediated aggregate formation, while Sup35p N and M domains modulated aggregate size and frequency when fused to the globular domain of PrP. Here we further present co-aggregation studies of chimeric proteins with cytosolic PrP or a huntingtin fragment with an extended polyglutamine tract. Our studies demonstrate that cross-seeding by heterologous proteins requires sequence similarity with the aggregated protein domain. Taken together, these results demonstrate that nucleation and seeding of prion protein aggregates is strongly influenced by dynamic interactions between the aggregate core forming domain and its flanking regions.

Introduction

In mammals, prions are infectious agents that cause fatal neurodegenerative diseases of humans and animals.¹ Prions appear to be composed primarily or entirely of the abnormally folded aggregated isoform, PrP^Sc, of the host protein PrP^c.^2,3 The cellular prion protein PrP^c is a cell surface anchored protein. Conversion into its abnormal isoform PrP^Sc appears to occur on the cell surface or along the endocytic pathway. PrP^Sc formation proceeds via a nucleated polymerization reaction in which the normal PrP^c adopts an alternative, pathogenic conformation upon binding to a PrP^Sc oligomer.^4,5 Despite intensive research, little is known about the exact prion forming domain, the mechanism of autocatalytic conversion or potential co-factors involved in this process. The NMR structure of PrP^c has recently been solved.⁶ In PrP^c, a region encompassing approximately residues 90–230 forms a globular domain with three α-helices and two short β-strands. The aminoterminus of PrP^c harbors oligopeptide repeats and appears to be flexibly disordered. In PrP^Sc, the globular domain of PrP is partially protected from proteolysis by proteinase K, while the aminoterminus is readily degraded by this treatment. Thus, the globular domain of PrP appears to constitute the core of the PrP^Sc aggregate.

In fungi, several proteins have been identified that act as prions and undergo self-perpetuating changes in conformation that are epigenetically inherited upon cell division or by mating.⁷ Importantly, prions of yeast and fungi do not lead to cell death but rather alter the metabolic phenotype of the host. The Saccharomyces cerevisiae Sup35p is the protein determinant of the [PSI⁺] yeast genetic element. In its normal, soluble form, Sup35p acts as a subunit of the translation termination factor and assists in proper protein translation. In the [PSI⁺] state, Sup35p adopts a β-sheet rich form that increases read-through of nonsense codons, due to sequestration of functional Sup35p monomers into non-functional aggregates.⁸

The yeast Sup35p and the mammalian PrP prion protein share little structural or functional homology except for a repeat region in the aminoterminal domain. In contrast to mammalian prions, regions in the protein involved in yeast prion formation and protein function are well defined. The prion activity of Sup35p is conferred by the aminoterminal N domain,⁸ while the translation termination activity is governed by the carboxyterminal C domain (reviewed in ref. 9). N and C are separated by the M domain that appears to modulate solubility of the protein.¹⁰ In vitro, NM polymerizes into amyloid-like fibrils.^10,11 Overexpression of the N prion domain in yeast increases the frequency of [PSI⁺], potentially by enhancing spontaneous Sup35p aggregation.¹² Furthermore, spontaneous prion formation and maintenance of the prion phenotype are dramatically increased by the presence of several chaperones and other co-factors, such as heterologous proteins in a prion-like state.^12–16

The experimental tractability of yeast prions has greatly helped to elucidate general aspects of prion biogenesis that are also amenable to mammalian prions. In fact, the finding that bacterially expressed yeast prion protein domains in their amyloid form induce the prion phenotype in yeast strongly supports the protein-only hypothesis.^17,18 Aim of this study was to elucidate basic aggregation mechanisms of mammalian and yeast prion proteins. We focused on putative prion domains of the mouse PrP and Sup35p and their intrinsic aggregation propensities in the mammalian cytosol to determine the influence of specific protein domains on nucleation and seeding of prion protein aggregates. In a recent study¹⁹ we demonstrated that the globular domain of PrP encompassing residues 90–230 spontaneously aggregated in the mammalian cytosol. By contrast, the Sup35p NM domain stayed soluble even when expressed stably in murine neuroblastoma N2a cells. Importantly, PrP aggregation frequency and size were modulated by fusion with the yeast N and M domains. Orthogonal co-aggregation experiments performed here show that the NM domain in PrP fusion proteins is unlikely part of the aggregated core, as chimeric protein aggregates failed to recruit soluble epitope tagged NM. Conversely, PrP_cyto readily co-aggregated with NM-PrP, N-PrP and M-PrP, suggesting similar packing in the aggregate cores of recombinant proteins. In summary, our studies reveal important mechanistic details on prion protein aggregation and demonstrate that nucleation and seeding are based on dynamic interactions of the core aggregation domain with flanking sequences.

Results

Fusion of Sup35p N and M domains with PrP_90–230 modulates aggregate size.

To assess sequence and environmental requirements for prion protein aggregation in mammalian cells, comparative analysis of the aggregation behaviors of Sup35p NM, PrP and recombinant fusion proteins comprising fragments of both proteins were performed in a recent study.¹⁹ For that purpose, constructs were generated that code for cytosolic murine PrP (PrP_cyto, PrP_23–230), PrP_90–230 or Sup35p NM domains fused to a HA epitope tag (NM-HA). Furthermore, chimera NM-PrP (NM_1–250-PrP_90–230), N-PrP (N_1–123-PrP_90–230) and M-PrP (M_124–250-PrP_90–230) and PrP-M (PrP_23–120-M_124–250) were also constructed. To discriminate recombinant PrP from endogenous PrP, an epitope tag for the monoclonal antibody 3F4 was inserted into recombinant PrP. Fibrillization studies with bacterially expressed recombinant proteins demonstrated that NM and NM-PrP were capable of forming amyloid-like fibrils in vitro, while PrP-M was not.¹⁹ Constructs were transiently transfected into N2a cells and the aggregation state of the respective proteins was assessed by confocal microscopy (Fig. 1A).¹⁹ NM-HA appeared soluble in the cytosol of mammalian cells although an aggregation in e.g., one per million cells would probably remain undetected. Similarly, PrP-M was evenly distributed throughout the cytoplasm. By contrast, expression of cytosolic PrP (PrP_cyto), PrP_90–230 and N-PrP led to the formation of multiple small aggregates of less than 1 µm that sometimes clustered in one area of the cell. The appearance of small punctate aggregates of PrP_cyto and PrP_90–230 is consistent with previously published results.²³ NM-PrP and M-PrP formed relatively few large aggregates (or foci) that strongly differed from the smaller aggregates observed with PrP_90–230, PrP_cyto or N-PrP. Individual cells usually contained up to four dot- or ring-shaped aggregates, but multiple ring-shaped aggregates were occasionally found. Single aggregates were about 5 µm in diameter, however, in cells transfected with a construct coding for M-PrP, sometimes foci up to 10 µm in diameter were detected. Collection of sequential scans along the vertical (z) axis demonstrated that the giant M- and NM-PrP aggregates were spherical, indicating that the ring-like appearance was likely due to a surface binding of antibodies that were incapable of penetrating the aggregates (data not shown). In conclusion, our previous results argue that spontaneous cytosolic aggregation is an intrinsic property of the globular domain of PrP, PrP_90–230. Importantly, the aggregate appearance could be strongly modulated by the yeast prion domain M that exerted its effect on aggregate size in cis.

Figure 1.

Aggregate size and frequency of aggregate appearance are influenced by cis-acting N and M domains of Sup35p. (A) Confocal microscopy images of transiently transfected N2a cells subjected to immunofluorescence analysis. Ectopically expressed proteins were stained with 3F4 or anti-HA antibodies (green), nuclei were stained with Hoechst (blue). Schematic representations of proteins are shown below the images. Scale bars: 10 µm. (B) Determination of the ratio of cells that exhibited aggregates to cells that appeared to express soluble recombinant proteins. Aggregation was assessed for each recombinant protein in at least 300 transfected cells in three independent experiments. Statistical analysis were performed using the χ²-test (n.s. = not significant; * = significant with p < 0.05; ** and *** = highly significant, with p < 0.01 or p < 0.001 respectively). The standard deviation of mean is shown (S.D.). Adapted from reference ¹⁹.

Fusion of Sup35p N and M domains to PrP_90–230 modulates aggregation frequency.

The foregoing results demonstrated that the aggregate phenotype could be influenced by protein sequences that aminoterminally flanked the globular domain of PrP. Thus, we further assessed the influence of PrP_23–89, N or M on the PrP aggregate nucleation rate.¹⁹ Therefore, recombinant proteins were transiently expressed in N2a cells and the rate of cells displaying visible aggregates to transfected cells was determined (Fig. 1B). PrP_23–89 appeared to exert a negative effect on nucleation, as significantly more cells exhibited visible PrP_90–230 aggregates compared to cells expressing PrP_cyto (45 ± 5% vs. 24 ± 5% p < 0.01). The M domain of Sup35p appeared to exert a similar effect on nucleation, as fusion of M to PrP_90–230 led to a significant decrease in transfected cells with aggregates (30 ± 3% vs. PrP_90–230: 45 ± 5%, p < 0.05). By contrast, fusion of N to M-PrP_90–230 significantly increased the amount of cells displaying visible aggregates (55 ± 9% vs. M-PrP: 30 ± 3%, p < 0.001). Thus, the inhibitory effect of M on nucleation could be compensated by fusion with N in NM-PrP. However, fusion of N to PrP_90–230 did not lead to an enhanced nucleation rate (50 ± 2% vs. PrP_90–230: 45 ± 5%, p > 0.05). In conclusion, our previous results demonstrated that PrP_90–230 aggregate induction was influenced by prion protein regions in cis.¹⁹ The fact that N counteracted the effect of M argues that aggregate size and appearance are modulated by a dynamic interaction of individual prion protein domains. Thus, the globular domain of PrP has aggregate induction activity whereas N and M domains of Sup35p influence nucleation and seeding.

Sequence similarity in the aggregate core domain is required for co-aggregation.

Our previous results clearly demonstrated that the globular domain of PrP was sufficient for spontaneous cytosolic PrP aggregation, suggesting that this domain was also the aggregated domain in chimeric Sup35p-PrP fusion proteins. To assess if the N, M or NM domains in chimeric Sup35p-PrP fusion proteins were also present in the aggregate core, we performed orthogonal co-aggregation experiments of NM-HA and NM-PrP, N-PrP or M-PrP in N2a cells. Orthogonal cross-seeding has recently been used as an assay to monitor the nature of the aggregate core.²⁴ This method is based on the concept that recruitment of monomers into growing aggregates is due to equivalent molecular structures within aggregation-prone proteins or fragments thereof.²⁴ Sequestration of a given protein is usually dependent on sequence similarity with the aggregated core of the heterologous protein. We first studied if NM-PrP, M-PrP or N-PrP could recruit NM-HA into aggregates. Recombinant proteins were transiently expressed and co-aggregation was studied by confocal microscopy (Fig. 2A). Similar results were obtained with all chimeric Sup35p-PrP molecules, so only results of the co-expression of NM-HA and NM-PrP are shown. None of the Sup35p-PrP chimeric protein aggregates was able to seed NM-HA aggregation. These results suggest that the aminoterminal NM domain in NM-PrP is not in an aggregated state that can cross-seed NM-HA.

Figure 2.

Co-expressed NM-HA is not incorporated into NM-PrP aggregates. (A) N2a cells were transiently transfected with Sup35p NM-HA and NM-PrP. 48 h post transfection cells were stained for NM-HA (anti-HA) and NM-PrP (3F4), respectively. Nuclei were visualized using Hoechst staining (blue). Scale bars: 10 µm. (B) Induction of NM-HA aggregation in mammalian cells by expression of HD72Q-GFP. N2a cells stably expressing NM-HA were transiently transfected with a construct coding for HD72Q-GFP. 48 h post transfection cells were stained for NM-HA (anti-HA). NM-HA appeared to aggregate around polyQ aggregates, forming ring-like structures. (C) Loss of NM-HA aggregation upon continuous culture in the absence of HD72Q-GFP. HD72Q-GFP was transiently expressed in N2a cells stably expressing NM-HA. One week post transfection cells were stained for NM-HA (anti-HA).

As a positive control for NM-HA aggregate induction in the mammalian cytosol, we tested if NM-HA was capable of co-aggregating with aggregates of a protein that shared sequence similarities such as glutamine/asparagine (Q/N)-rich tracts in the aggregated domain. N2a cells stably expressing NM-HA (N2a_NM-HA cells) were transiently transfected with pEGFP-HD72Q, a construct coding for exon 1 of huntingtin protein with a polyQ tract of 72 glutamines.²⁰ Cells were fixed and stained for NM-HA using an anti-HA antibody (Fig. 2B). In cells not transfected with a construct coding for HD72Q-GFP, stably expressed NM-HA stayed soluble and was evenly distributed throughout the cytosol. As previously reported, transiently expressed HD72Q-GFP formed visible aggregates in the cytosol.²⁰ In these cells, NM-HA co-localized with HD72Q-GFP aggregates and appeared to cluster around the rims of the HD72Q-GFP aggregates (Fig. 2B). However, continuous passage of cells revealed that the aggregated state of NM-HA was not maintained. One week post transfection, NM-HA aggregation was no longer detectable (Fig. 2C). Similar results were also obtained with neuronal hippocampal HpL3-4_NM-HA cells (data not shown). Thus, NM-HA aggregation can be induced by protein aggregates in which a protein tract with a similar sequence is in its aggregated state. The fact that NM-PrP was unable to seed NM-HA aggregation but HD72Q-GFP was suggests that in NM-PrP the NM domain is not in an aggregated state that is capable of promoting co-aggregation of NM-HA.

Cytosolic co-aggregation of proteins harboring the globular domain of PrP.

We next tested if PrP was capable of promoting co-aggregation of Sup35p-PrP chimeras. Due to the lack of a PrP specific antibody epitope tag that could discriminate PrP_90–230 and Sup35p-PrP chimeras, visualization of co-aggregation could only be assessed with cytosolic PrP. PrP_cyto was stained with an antibody against the octarepeat region, POM2,²² while NM-PrP, N-PrP and M-PrP were visualized with the N- or M-specific antibodies 7H5 and 4A5, respectively (Fig. 3, left). The construct coding for cytosolic PrP was co-transfected with vectors coding for N-PrP, M-PrP or NM-PrP and co-aggregation was visualized by the use of the respective antibodies. Interestingly, PrP_cyto and N-PrP, M-PrP or NM-PrP localized in the same aggregates (Fig. 3, right). Surprisingly, only small heterologous aggregates were formed and the large foci typical for M- and NM-PrP disappeared (compare Figs. 1A and 3). Thus, M appeared to be less efficient in seeding soluble proteins into one aggregate, as chimeric NM-PrP and M-PrP molecules were unable to form large aggregates in the presence of PrP_cyto (Fig. 3).

Figure 3.

Co-aggregation of cytosolic PrP with Sup35p NM-PrP chimera. N2a cells were transiently transfected with constructs coding for PrP_cyto and N-, M- or NM-PrP. 48 h post transfection cells were stained for PrP_cyto (POM2) and N-PrP (7H5), M- or NM-PrP (4A5), respectively, and analyzed by confocal microscopy. Nuclei were visualized using Hoechst staining (blue). Yellow colour in the merged pictures indicates co-localization of both recombinant proteins. Scale bars: 10 µm.

Discussion

Nucleation and seeding are modulated by flanking regions of the aggregation domain.

In vitro, most proteins are able to form amyloids.²⁵ Still, why some proteins also aggregate in vivo is poorly understood. We have established a model system to compare Sup35p and mouse PrP in mammalian cell cultures to elucidate potential cis and trans acting elements involved in cytosolic prion protein aggregate formation. Although both PrP^2,26 and NM^10,11,19 are able to form amyloid-like fibrils in vitro, they exhibit very different aggregation behaviors when expressed in mammalian cells. We¹⁹ and others²³ have recently demonstrated that the globular domain of PrP, PrP_90–230, has aggregate inducing activity when expressed in the mammalian cytosol. Our studies revealed that prion protein domains aminoterminally flanking this aggregation domain strongly influenced aggregate induction and seeding.¹⁹ While Sup35p NM was not able to aggregate on its own, fusion of Sup35p domains to PrP_90–230 modulated aggregate size and frequency of chimeric proteins.¹⁹ Interestingly, a negative effect on nucleation could also be reported for the aminoterminal part of PrP, PrP_23–89. Figure 4 illustrates the influence of the different Sup35p and PrP domains on PrP_90–230 aggregate formation.¹⁹ In this model, PrP_90–230 is the driving force for aggregation. When fused to PrP_90–230, PrP_23–89 and M decrease nucleation resulting in fewer cells with visible foci, arguing for some functional similarity of PrP_23–89 and the M domain of Sup35p. The negative effect of M could be counteracted by N. The M domain of Sup35p further promotes seeding of soluble chimeric protein into the growing aggregate, resulting in large aggregates typical for M- and NM-PrP. Thus, prion protein aggregation is strongly modulated by amino acid sequences flanking the aggregation domain, a finding that has also been reported for other aggregation-prone molecules, such as polyQ proteins.^27,28

Figure 4.

Model explaining the cis-effect of PrP_23–89 and the Sup35p domains N and M on PrP_90–230 aggregation. The globular domain of PrP spontaneously forms small and numerous aggregates in the cytosol of mammalian cells. Fusion of both PrP_23–89 and Sup35p-M to PrP_90–230 decreases the nucleation rate, leading to fewer cells displaying visible aggregates. N counter-acts the influence of M. The PrP_90–230 aggregate size is drastically increased by the fusion with M, potentially due to an enhanced seeding capacity. Adapted from reference ¹⁹.

Co-aggregation and seeding of recombinant proteins is a specific event.

An accumulation of misfolded proteins is the key event in various proteinopathies. It is commonly anticipated that the toxicity of these proteinaceous deposits is at least partially mediated by the sequestration of unrelated cellular proteins which subsequently perturbs their normal cellular functions.²⁹ Evidence that aggregation-prone proteins can co-aggregate with heterologous proteins in mammalian cells has previously been demonstrated.³⁰ Orthogonal cross-seeding appears to occur with extraordinary specificity amongst proteins with equivalent molecular structures, for example amongst Q/N-rich stretches.^31,32 Our data are consistent with this finding and demonstrate that the globular domain of PrP promotes co-aggregation of proteins that share this region but differ in their aminoterminal flanking regions (Fig. 5A). It is likely that the globular domain of PrP allows correct intermolecular interactions between heterologous proteins that are necessary for aggregate formation. However, co-aggregation did not solely depend on sequence similarities in common aggregation-promoting motifs. Our co-aggregation experiments nicely demonstrate that co-aggregation depends on sequence similarities in the protein domain that assembles into ordered aggregates. A good example is the co-aggregation behavior of NM-PrP. When co-expressed with PrP_cyto, both proteins are sequestered into the same aggregates, suggesting that the globular domain of PrP in PrP_cyto and NM-PrP promotes co-aggregation. Interestingly NM-PrP was unable to form large aggregates in the presence of PrP_cyto, potentially due to the increase in aggregate nuclei. NM-PrP was unable to seed NM-HA aggregation, indicating that the NM domain is not in an aggregated state that can support seeding of NM-HA aggregation (Fig. 5B).

Figure 5.

Model explaining the co-aggregation behaviors of recombinant proteins with Sup35p and PrP domains. (A) The shared aggregation domain PrP_90–230 allows co-aggregation of PrP_cyto with N-PrP, M-PrP and NM-PrP. Nucleation of PrP_cyto provides numerous nuclei that efficiently cross-seed M-PrP and NM-PrP, thereby preventing large aggregate formation typical for M- and NM-PrP. (B) In NM-PrP, PrP_90–230 acts as the aggregation domain. The NM domain is unlikely part of the aggregate core, as NM-PrP is unable to seed NM-HA aggregation. By contrast, sequence similarities in the aggregated domain of HD72Q-GFP enable NM-HA monomers to sediment around these aggregates.

The N domain of Sup35p is rich in glutamines and asparagines, a feature reminiscent to the expanded Q/N-rich tracts of several pathogenic aggregation-prone proteins such as huntingtin. In yeast, aggregation-prone proteins harboring Q/N-rich regions similar to the Q/N-rich region in Sup35p N or with extended polyQ stretches can act as heterologous seeds for the aggregation of NM.^33,34 Interestingly, NM-HA molecules were also recruited into inclusions containing a huntingtin fragment with an expanded polyQ tract in mammalian cells, indicating that cytosolic NM-HA is capable of co-aggregating, given that a heterologous protein with a similar domain in its aggregated state is provided as a seed (Fig. 5A). However, aggregation of NM-HA was transient and crucially depended on the expression of HD72Q-GFP. This result indicates that HD72Q-GFP induced NM-HA aggregates were not able to faithfully propagate in the absence of seed-forming HD72Q-GFP. The fact that heterologous co-aggregation failed to induce propagation of NM-HA aggregates as prions also indicates that prion formation is controlled at the level of conformational transition, as has been previously suggested for mammalian and yeast prions.^35,36 Remarkably, our results are in contrast to studies in yeast where heterologous polyQ seeds can induce the [PSI⁺] phenotype.³⁴ In Saccharomyces cerevisiae, however, NM prion aggregate inheritance is strongly dependent on the expression of several chaperones, most importantly Hsp104 (reviewed in ref. 9). Hsp104 acts as a disaggregase that disassembles Sup35p aggregates into smaller seeds that are transmitted to daughter cells. As no Hsp104 orthologs have been identified in the mammalian cytosol, alternative chaperones or factors that enable disaggregation of HD72Q-GFP/NM-HA inclusions might be absent. Alternatively, NM-HA molecules in HD72Q-GFP inclusions in the mammalian cytosol are in an aggregated state that differs from the usual NM fold in the prion state, thus precluding HD72Q-GFP independent seeding. In summary, the formation of prion protein co-aggregates is a highly selective process that strongly depends on sequence similarities in the protein domains that form the core of the protein aggregates.

Materials and Methods

Description of expression vectors.

Cytosolic PrP, Sup35p NM and fusion proteins thereof were cloned as described previously.¹⁹ All PrP sequences contain a 3F4 epitope tag allowing discrimination from endogenous PrP using the monoclonal antibodies (mAb) 3F4. Sup35p NM was tagged with a carboxyterminal HA epitope tag. The vector pEGFP-HD72Q²⁰ coding for the huntingtin (htt) exon 1 protein with a polyQ tract of 72 glutamines was a kind gift of Dr. Erich Wanker (Max Delbrück Center of Molecular Medicine, Berlin). For stable transduction, NM-HA coding sequences were subcloned into the lentiviral vector LV-PGK-EGFP, containing a phosphoglycerate kinase (PGK) promoter driven EGFP expression cassette,²¹ by replacing the EGFP coding sequence. All constructs were confirmed by DNA sequencing (GATC, Konstanz).

Transgene expression.

The mouse neuroblastoma cell line N2a (ATCC CCL-131) was maintained in Optimem supplemented with 10% fetal bovine serum and penicillin/streptomycin. Transfections and co-transfections were performed using Fugene according to the manufacturer's recommendations. Expression of transgenes was assessed 48 h post transfection by confocal microscopy analysis. For stable transduction of N2a cells, 2.5 × 10⁵ cells were plated in 6-well plates. Stable expression of NM-HA was achieved by transduction with recombinant lentivirus particles coding for NM-HA (kindly provided by Dr. Andreas Hofmann; Laboratory of Dr. Alexander Pfeifer; Institute of Pharmacology and Toxicology, University of Bonn) for 24 h. Upon transduction, close to 100% of the cells stably expressed NM-HA.

Indirect immunofluorescence analysis.

The indirect immunofluorescence technique was used to visualize antigens in situ on a single cell level. Cells were plated on 6 cm dishes with glass cover slips. In some experiments, cells were transfected with the respective constructs 24 h later. The next day or 48 h post transfection, cover slips were transferred to 12-well culture plates, rinsed 3 × with PBS, and fixed with 500 µl Roti-Histofix for 30 min at room temperature. Cells were quenched with 50 mM NH₄Cl, 20 mM glycine, permeabilized with 0.1% Triton X-100 in PBS and blocked with 0.2% gelatine in PBS. Each step was performed for 10 min and terminated by 3 × rinsing steps with PBS. Eventually, the primary antibody diluted in blocking solution was added for 45 min at room temperature in a humid chamber. Samples were rinsed three times with PBS prior to addition of the Cy2- or Cy3-conjugated secondary antibody diluted in blocking solution for 30 min at room temperature in the dark. Again, cells were rinsed three times and nuclei were stained by incubation with 2 µg/ml Hoechst DNA staining solution in PBS for 10 min at room temperature in the dark. After rinsing cover slips were mounted on microscope slides in anti-fading solution Permafluor and kept dry at −20°C. Confocal laser scanning microscopy was carried out using a LSM510 confocal laser microscope (Zeiss). The following primary antibodies were used: mouse mAb anti-PrP 3F4 (Signet Pathology, Dedham, MA, USA), anti-HA F7 (Santa Cruz Biotechnology, Santa Cruz, CA, USA), and POM2²² (a kind gift of Dr. Polymenidou and Dr. Aguzzi; Institute of Neuropathology; University Hospital of Zürich) and rat mAb anti-Sup35p N 7H5 of IgG2a subclass (raised against a peptide comprising aa 8–27 of Sup35p), and anti-Sup35p M 4A5 of IgG2a subclass (using a peptide antigen comprising aa 229–247 of Sup35p) generated by our laboratories. For determination of the ratio of transfected cells displaying visible aggregates to cells expressing soluble protein, at least 300 cells were scored for each construct in three independent experiments. The Chi-Square (χ²) test was used for statistical analysis.

Abbreviations

PrP^c

cellular prion protein

PrP^Sc

disease-associated prion protein

NMR

nuclear magnetic resonance

htt

huntingtin

mAb

monoclonal antibodies

PGK

phosphoglycerate kinase

polyQ

polyglutamine

Q/N-rich

glutamine/asparagine-rich