Observing fibrillar assemblies on scrapie-infected cells
- 499 Downloads
The infectious agent in prion diseases is an aberrant-folded isoform of the cellular prion protein (PrPC). This scrapie-related prion protein (PrPSc) has an increased β-sheet content, is detergent insoluble and proteinase K resistant, and accumulates in prion-infected organisms and cells. In vitro, PrPSc self-aggregates into amyloid fibrils. However, there is no direct experimental proof for the occurrence of PrPSc-containing fibrils in vivo or in cell cultures. Applying atomic force microscopy (AFM) to scrapie-infected mouse neuroblastoma (ScN2a) cells, we discovered growing patch-like assemblies of amyloid-like fibrillar structures on the cell surfaces. Immunofluorescence and AFM images showed heterogeneous accumulation and aggregation of PrPSc in ScN2a cell cultures. The percentage of cells having characteristic fibrils on their surface increased with time after scrapie infection. These endogeneous fibrils had lengths from 0.5 to 3 μm and protruded from the cell surface by 108 ± 30 nm, and thus resembled the heterogeneous shapes and networks of in vitro prepared amyloid fibrils.
KeywordsAtomic force microscopy Immunofluorescence Mouse neuroblastoma (N2a) cells Prion protein Scrapie Fibrils Amyloid
atomic force microscopy
days post plating
mouse neuroblastoma cell line
scrapie-infected mouse neuroblastoma cell line
phosphate buffered saline
cellular prion protein
scrapie prion protein
cellular and scrapie prion protein
rocky mountain laboratory
root mean square
recombinant prion protein
Prion diseases comprise a group of fatal neurodegenerative diseases in humans and other mammals, which are characterized by the accumulation of abnormally folded prion proteins (PrP), designated PrPSc. PrPSc, the constituent of infectious prions, is a self-propagating isoform of the cellular prion protein (PrPC) with distinct biochemical and biophysical properties compared to PrPC (reviewed in [1, 17, 18, 54, 69]). Expression of PrPC is essential for the pathogenesis of prion diseases; knock-out mice with a targeted disruption of the PrP gene are resistant to prion diseases and propagate neither PrPSc nor infectious prions . The biogenesis of PrPC is characterized by a series of co- and post-translation modifications (reviewed in ). In the endoplasmic reticulum, two N-linked glycans (to Asn180 and Asn196 in murine PrP, to Asn181 and Asn197 in human and Syrian hamster PrP)  and a C-terminal glycosylphosphatidylinositol (GPI) anchor  are added. During transit through the secretory pathway, the glycans are processed into complex structures [22, 56, 61], and, finally, PrPC is targeted to the outer leaflet of the plasma membrane. PrPSc, purified from scrapie-infected brains, contains a GPI anchor and sialylated complex N-linked glycans [22, 60], indicating that fully matured PrPC is the precursor for the formation of PrPSc.
To address questions concerning prion infectivity, the cell toxicity of prions, and disease propagation, stable-infected cell lines permanently expressing high amounts of PrPSc have been established [13, 59]. Although many cell lines express PrPC, mouse neuroblastoma (N2a) cells belong to a small group of cell lines susceptible to infection with scrapie-prions and, thus, represent a preferred cell culture model [12, 15, 16]. Interestingly, scrapie-infected (Sc)N2a cells continuously propagate protease K-resistant PrPSc and infectious prions, yet, do not show any growth defects compared to uninfected cells. This cell culture model showed that both protein isoforms, PrPC and PrPSc, are found in detergent-insoluble microdomains (rafts) at the plasma membrane [46, 68]. Based on these and other findings, it appears that PrPSc formation and accumulation occur extracellularly and/or in compartments of the endocytic/lysosomal pathway and depend on the lipid composition of the membrane (reviewed in ).
PrPSc has distinct biochemical features, such as insolubility in detergent and a partial resistance to proteolytic digestion . In vitro studies using purified PrPSc demonstrate the self-assembly of PrPSc into oligomers and amyloid fibrils [10, 41]. Such PrP assemblies are routinely analyzed, with regard to fibril growth and structure, using circular dichroism [32, 38], Fourier transform infrared spectroscopy [30, 34], electron microscopy [43, 45, 70], and thioflavine assays . However, between these, in vitro results and the observation of in vivo-occurring amyloid plaques in prion diseases [1, 53] remains a large pool of uncertainties about native PrP amyloid formation. Atomic force microscopy (AFM) reveals the topography of cells by a raster scanning the cell surface with a sharp probe . The probe consists of a silicon nitride stylus, typically exhibiting a radius of 10 to 50 nm at its apex, mounted to the end of a small cantilever. In the contact mode, AFM scans the stylus across the cell surface while maintaining the deflection of the cantilever constant by controlling the height of the cantilever. This procedure reveals the height of the sample for every scanned pixel. Thus, raster scanning the cell surface line by line provides the three-dimensional topography of the cell. Of importance for biological applications, AFM allows imaging in physiological relevant buffer solutions and at ambient temperatures. This avoids dehydration artifacts . The signal to noise ratio of AFM is superior to any optical microscopy technique, and imaging does not require labeling or staining of the sample. With limited sample preparation, cells can be imaged directly by AFM to a vertical resolution of ∼5 nm and a lateral resolution ranging between 20 and 100 nm. Since AFM contrasts topographic features of a cell surface, its combination with light microscopy and specific fluorescence labeling of cellular structures is advantageous [24, 31, 51]. While fluorescence microscopy facilitates the identification of cellular structures, the higher resolution provided by AFM allows the structures to be observed in greater detail. In this work, we used both light microscopy and AFM to characterize surface structures exposed on scrapie-infected cells. The AFM topographs provide information about the structural organization of filaments deposited at the cell surface upon scrapie infection.
Materials and methods
Cell culture and sample preparation
N2a cells are immortalized murine neuroblastoma cell lines (ATCC No. CCL 131). ScN2a cells were established by infecting N2a cells with an enriched preparation of prions isolated from the brains of mice infected with rocky mountain laboratory (RML) prions as described . Both cell lines were cultivated in minimum essential medium with Earle's salts (GIBCO), supplemented with 10% fetal bovine serum, 100 units/mL penicillin, and 100 mg/mL streptomycin in 5% CO2 at 37°C. For AFM and immunofluorescence, cells were grown on glass coverslips for 1 to 6 days. For Western blot analysis, cells were grown in parallel and analyzed using the detergent solubility assay.
Cell fixation for AFM and immunofluorescence microscopy
For AFM, cells were washed with phosphate buffered saline (PBS) and fixed with 0.2% glutaraldehyde (GA)/PBS for 45 s followed by 4% paraformaldehyde (PFA)/PBS for 20 min. This ensured a high stiffness of the organic material leading to a good resolution in the subsequent AFM scans. To reduce the autofluorescence background in immunofluorescence microscopy, cells were fixed using only the 4% PFA/PBS treatment for 20 min.
Detergent solubility assay and Western blot analysis
As described previously , cells were washed twice with cold PBS, scraped off the plate, pelleted by centrifugation, and lysed in cold buffer A (0.5% Triton X-100 and 0.5% sodium deoxycholate in PBS). The lysate was centrifuged at 15,000×g for 20 min at 4°C to separate the detergent-soluble (S) from the detergent-insoluble (P) fraction, and PrP was detected by Western blotting as described earlier  using the polyclonal anti-PrP antibody pAb A7 . Loading was controlled using the anti-α-tubulin antibody (Amersham).
N2a and ScN2a cells plated on glass coverslips were fixed using 4% PFA/PBS (for details see “Cell culture and sample preparation”), washed with PBS, and then incubated for 20 min in 4 M guanidinium hydrochloride (GdnHCl)/PBS for antigen retrieval. After washing with PBS and blocking for 1 h with 0.5% goat serum/PBS (Sigma-Aldrich), the polyclonal anti-PrP antibody pAb A7  was applied for 2 h at room temperature. For β-tubulin counter staining, cells were subsequently permeabilized for 3 min in 0.4% Triton X-100/PBS, blocked as before, and incubated in monoclonal mouse anti-β-tubulin antibody (Sigma-Aldrich) diluted 1:1,000 in 0.5% goat serum/PBS. First, antibody recognition was done by FITC-labeled monoclonal goat anti-rabbit-IgG (Jackson ImmunoResearch) and tritc-labeled monoclonal goat anti-mouse-IgG (Jackson ImmunoResearch), respectively, both diluted 1:1,000 in 0.5% goat serum/PBS. Fluorescence images were taken with a Zeiss Axiovert 200 microscope (Carl Zeiss, Goettingen, Germany). Phase contrast and fluorescence microscopy images were taken using a CoolSNAPcf camera (Photometrics) and processed using MetaMorph software (Universal Imaging Corporation).
Coverslips with cultured cells were rinsed with PBS, fixed with 0.2% GA/PBS and 4% paraformaldehyde/PBS (details in “Cell culture and sample preparation”), and mounted onto the AFM (Nanowizard™, JPK Instruments) using a sample holder (Biocell™, JPK Instruments). The AFM was mounted onto a Zeiss Axiovert 200. Contact mode imaging was performed in PBS at room temperature (∼23°C) using non-sharpened V-shaped cantilevers (MLCT, Veeco Instruments, Santa Barbara, CA, USA) with nominal spring constants of 0.01 to 0.03 N/m. The force applied to the cantilever was set to ∼100 pN, while feedback gains were manually adjusted to obtain the best resolution in both height and deflection channels. Images were collected at line-scan rates ranging from 0.1 to 0.5 Hz. AFM images were processed and analyzed using the JPK image processing software. The cell surface roughness was determined by selecting 4 × 4 μm2 areas of the topographs and calculating the root mean square (RMS) of the mean height value. Topographic cross sections of single filaments were generated using ImageJ and IGOR Pro (WaveMetrics Inc.)
For our studies, we have chosen N2a cells and their scrapie-infected counterpart (ScN2a) representing N2a cells infected with RML prions from brains of scrapie-ill mice . The ScN2a cell culture model is commonly used to examine biogenesis of PrPC and propagation of PrPSc . Compared to uninfected N2a cells, ScN2a cells show changes in the levels of several mRNAs , in the maturation of insulin receptors , and in the expression of proteins relevant to cellular iron metabolism . Although intra- and extracellular accumulation of PrPSc is prominent in ScN2a cells, no differences in cell mortality and apoptosis rates have been found . Herein, we characterized this well-studied cellular model system by AFM to screen for morphological changes and extracellular accumulation of PrPSc upon scrapie infection.
Fibrillar networks form on top of murine neuroblastoma cells after scrapie infection
Initially, we labeled the cell surface PrP of both cell lines using the polyclonal PrP pAb A7 antibody . Next, we permeabilized the cell membrane using Triton X-100 and immunolabeled intracellular β-tubulin. Using the immunofluorescence of tubulin as a reference signal allowed us to estimate the ratio of infected vs uninfected cells. As expected, N2a cells showed a homogenous distribution of PrP, while in most ScN2a cells, labeled PrP accumulated in restricted areas (Fig. 1a). These results are in agreement with previous studies .
To see whether this apparent clustering of PrP was accompanied by structural changes to the cell surface, we investigated the cells using AFM. AFM deflection images and height topographs of paraformaldehyde- and glutaraldehyde-fixed monolayers of N2a cells (Fig. 1b–c) showed smooth surfaces lacking any characteristic features. The average roughness of these cells was 73 ± 14 nm (mean RMS ± SD; n = 16). In contrast, ScN2a cell layers that have been cultivated for 5 to 6 days under identical conditions were covered by networks of fibrillar structures (Fig. 1d–e). These networks could extend over several ScN2a cells covering up to ∼100 μm2 of the cell monolayer and differed in thickness, in fibril length, and in the degree of branching. Consequently, the ScN2a cells showed an increased average surface roughness of 104 ± 28 nm (mean RMS±SD; n = 18).
Repeated imaging of the same cell surface area in different scanning directions did not change the structural appearance of single fibrils or the fibrillar network (data not shown). Hence, the possibility that the structures were created or modified by the AFM imaging process can be excluded. Glutaraldehyde fixation of MDCK cells results in fixation artifacts in the form of small spherical structures on the cell surface . Such artificial structures would alter the surface appearance and roughness of both N2a and ScN2a cells. Thus, we exclude the possibility that the fibrillar structures, found only on ScN2a cell surfaces, reflect artifacts of fixation or AFM imaging processes.
Unfortunately, we were unable to directly correlate the fibrillar structures observed by AFM with immunofluorescently labeled PrPSc. After exposure to 4 M GdnHCl, as required for immunolabeling of PrPSc , the ScN2a cell surfaces appeared very compliant. Even at the smallest possible imaging force (<50 pN), AFM images showed the bending of surface structures in the scanning direction. AFM topographs of GdnHCl treated cell surfaces showed an increased surface roughness, but they were marred by AFM tip contamination and scanning artifacts (data not shown). Artifacts may have been caused by loosely bound cell material [57, 67] and could be reduced neither by changing from AFM contact mode to oscillation mode imaging nor by using different AFM tip geometries and materials. Thus, the simultaneous visualization of cell surfaces by fluorescence microscopy and AFM was impeded by the rather harsh chemical treatment required to label PrPSc.
PrPSc formation and growth of fibrillar cell surface structures depend on time
AFM topographs highlight surface differences of scrapie-infected N2a cells
To compare the surface structures of N2a and ScN2a cells in greater detail, we recorded AFM images of the top of single cells at higher resolution (Fig. 4a–c). All images were taken at 5 and 6 dpp. N2a cells showed smooth cell membrane surfaces exposing small spikes (Fig. 4a, arrows in top image), with an average height of 107 ± 53 nm (mean ± SD, n = 26). Similar dimensions were previously observed for microvilli on MDCK cell surfaces [29, 50]. N2a cell surfaces resembled landscapes of smooth hills and valleys, although single N2a cells showed minor variations (Fig. 4a, middle and bottom images). In contrast, the increased roughness of ScN2a cell surfaces (Fig. 4b–c) was caused by numerous, sharply contrasted structures protruding from the smooth cell surface. Most of these protrusions had fibrillar shapes. The cell surface background looked alike that of N2a cells (Fig. 4a). The fibrillar structures differed in length and width but exhibited little variance in height (Fig. 4d–e). Their average protruding height above the cell surface was 108 ± 30 nm (mean ± SD, n = 60), while their average length was 1.0 ± 0.6 μm (mean ± SD, n = 35).
Fibrillar networks covering scrapie-infected cells result from PrPSc aggregation
While N2a cells showed a smooth surface lacking any pronounced structural features, ScN2a cultures showed rough and patchy fibrillar networks and exhibited a greater surface roughness. Since N2a and ScN2a cells almost exclusively differ in the high amounts of scrapie protein accumulated by ScN2a cells , we link the observed morphological changes to this protein. Our AFM topographs revealed that the fibrillar structures, found on ScN2a cells after 5–6 days of cultivation, protruded from the cell surface by ∼100 nm. This height is in agreement with the dimensions obtained by transmission electron microscopy of in vitro assembled amyloid bundles (50–250 nm; ). In vitro assembled solitary recombinant mouse prion protein (rPrP) fibrils exhibit a wide range of lengths from ≤0.25 to 4 μm , which increases with assembling time and mechanical breakage of PrP fibrils . Again, these values agree with the wide fibril length distribution and the average length of ∼1.0 μm we observed. Lateral fibril annealing  or other physical rearrangements of the fibrils during growth and aggregation on the cell surface  may also have contributed to the heterogeneity of fibril length and height in the AFM topographs.
Beyond the dimensions of single fibrils, the structure and extension of the observed fibrillar networks resembled that of scrapie-associated fibrils (SAF), which had been either assembled in vitro from recombinant PrP [10, 52] or extracted from solubilized and proteinase K-digested brain homogenates of scrapie-infected animals [19, 20, 43, 44]. However, so far, PrPSc-containing fibrils have not been found in infected brain tissues. The brain derived SAFs may assemble during homogenization or extraction and, thus, be preparation artifacts [42, 65]. To our knowledge, the AFM images presented in this work are the first microscopy images reporting amyloid-like fibrillar structures on the surface of scrapie-infected cultured cells. However, the observed structures may also result from the cellular response to the scrapie infection, e.g. changes in membrane composition or enhanced dendrite outgrowth.
Heterogeneity of ScN2a cell surfaces structures
Our AFM data showed a variety of surface features on scrapie-infected cells, ranging from single isolated protrusions to fibrils of different lengths and large amorphous aggregates. This heterogeneity of structures is in agreement with the assembling of diverse amyloid aggregates upon incubation of PrP fibrils in cell culture medium both in the absence and presence of cells [10, 47]. Furthermore, assembling states and final structure of in vitro assembled oligomers and short uniform fibrils of rPrP were shown to depend on pH, temperature, and electrolyte [4, 11, 33], as well as on the presence of the denaturants GdnHCl and urea [10, 47]. Previous AFM studies of PrP fibrils revealed specific structural characteristics for different prion strains [2, 30]. Thus, the diversity in structures on ScN2a cells can be accredited to multiple cellular and environmental conditions in the cell culture.
Local restriction of fibrillar structures follows heterogeneous PrPSc accumulation
Scrapie-infected cell cultures showed a heterogeneous distribution of patch-like fibrillar networks, and immunofluorescence images of infected cells (Fig. 1a and 3) showed that not every cell exhibited speckled PrP clusters indicative of PrPSc formation. It remains unclear whether cells, lacking fibrillar surface structures or clustered PrP signals, were uninfected, or whether they did not have the sufficient amount of PrPSc to cause these signs of infection. Heterogeneous distribution of PrPSc formation in cultures of ScN2a cells was shown by Klohn et al. . Within one cell culture dish, they found that micro-colonies with high PrPSc levels neighbored uninfected cells. This suggests that the areas of scrapie-infected cell monolayers covered by fibrillar structures coincided with cells exhibiting high PrPSc content. The patch-like appearance of the fibrillar networks can be explained by heterogeneous PrPSc accumulation of individual cells or cell groups in the culture dishes leading to locally restricted PrPSc aggregation, fibril growth, and amyloid formation.
The amyloid fibrils may be the result of in situ PrPSc self-assembling or of the deposition of pre-assembled amyloid from solution. Our AFM images of ScN2a cells recorded after different days of cultivation (Fig. 3) point towards a stepwise increase in surface amyloid formation, thereby excluding the deposition of preformed amyloid meshes.
Amyloid formation and PrPSc accumulation are progressive processes
While most ScN2a cells did not reveal any morphological changes within the first 2 days of cultivation, large fibril networks had fully developed on numerous areas of the cell monolayers after 4 days. Longer cultivation times of 5–6 days resulted only in minor rearrangements of amyloid-like fibrils. These structures may reflect different stages in the progressive assembling and maturing process of amyloid fibrils [4, 40] on the cell surface. In all the three independently performed assays, Western blot, immunofluorescence, and AFM topographs, the first signs for PrPSc in scrapie-infected cells were detected 4 days after cell plating. The parallel occurrence of PrPSc formation, PrP aggregation, and morphological changes to the cell surface suggests the direct involvement of PrPSc in the progressive growth of amyloidic fibril networks in scrapie-infected cell cultures.
Mice and hamster animal models represent much more complex systems than the herein-studied ScN2a cell line. Accordingly, incubation times of 7 weeks are required in mice until proteinase K-resistant PrP is found at brain sites of inoculation with RML prions . Afterwards, a linear increase of PrPSc was observed until 19 weeks post inoculation. Elsewhere, a rise in prion titer was reported to plateau before observing the actual onset of neuropathological changes [3, 8, 36]. In contrast, in vitro PrP fibril formation occurs within a few hours. Such assays represent a crude and oversimplified model for in vivo amyloid formation. However, in vitro studies revealed a strong dependence of the fibril assembly on environmental conditions such as pH, temperature, and electrolytes . In this context, ScN2a cells reflect a more natural tool to study complex in vivo processes associated with scrapie infection, because they provide a cellular context for PrPSc formation and aggregation. Comparing the complexity and amyloid formation characteristics of the three model systems – in vivo, in vitro, and cell culture – the time scale of days for progressive PrPSc accumulation and fibril formation in ScN2a cells appears to be reasonable.
The accumulation of PrPSc was shown using the solubility assay of whole cell lysates (Fig. 2), as well as by the clustered signals occurring in the immunofluorescence analysis of ScN2a cells (Fig. 3). Although the immunofluorescence labeling of PrPC+Sc was initially performed without permeabilizing the cells, perforation of the cell membrane during GdnHCl incubation could have provided access for the antibody to intracellular PrP. Therefore, neither of these assays allows for a strict discrimination of extra- and intracellular PrP species. For different disease-associated mutations in the PrPC gene, changes in protein trafficking and rerouting lead to cytosolic and endoplasmatic reticulum accumulation of misfolded PrP in N2a cells [27, 35, 39] (reviewed in ). Assuming that PrPSc is processed similar to misfolded PrPC, the PrPSc signals observed by immunofluorescence may rather result from intracellular aggregates than from extracellular deposits.
Using AFM, we discovered differences in the cell surface morphology of uninfected and scrapie-infected cells. Cellular formation and accumulation of PrPSc was shown to occur in parallel to fibril formation and growth on the cell surface. Therefore, we conclude that the single and complex branched fibrillar structures found on scrapie-infected cells are PrPSc-containing amyloid fibrils. The patch-like appearance of the fibrillar networks can be explained by heterogeneous PrPSc accumulation of individual cells or cell groups in the culture dishes. Consequently, questions regarding the cellular characteristics and environmental conditions that make a cell susceptible to scrapie infection and induce high level PrPSc expression remain to be answered.
We thank Jonne Helenius and Angelika Rambold for helpful discussions, and Isabel Richter and Eren Yuskel for experimental assistance. This work was supported by the Deutsche Forschungsgemeinschaft (SFB 596, GK 1401/1), the AO-Davos Foundation, and by the Max-Planck-Society.
- 41.Martins SM, Frosoni DJ, Martinez AM, De Felice FG, Ferreira ST (2006) Formation of soluble oligomers and amyloid fibrils with physical properties of the scrapie isoform of the prion protein from the C-terminal domain of recombinant murine prion protein mPrP-(121–231). J Biol Chem 281:26121–26128PubMedCrossRefGoogle Scholar
- 57.Schwarz UD, Haefke H, Reimann P, Guntherodt HJ (1994) Tip artifacts in scanning force microscopy. J Microsc 173:183–197Google Scholar