Prion subcellular fractionation reveals infectivity spectrum, with a high titre-low PrP^reslevel disparity

Characterisation of the MoRK13-M1000 prion infection model (MoRK13-inf) determined that per detectable 'PrP^res unit', MoRK13-inf cell lysate was approximately 90 times more efficient than crude M1000 brain homogenate at prion transmission to recipient MoRK13 (Additional file 1: Figure S1). This finding, perhaps an example of strain adaptation into a rabbit cell line, highlighted a discrepancy between PrP^res and infectivity levels, and indicated that MoRK13-inf cells may be an appropriate model to study this phenomenon. PrP isoforms have been shown to localise to various subcellular environments, including those of the secretory and endocytic pathways, and at the cell surface [32‐34]. Figure 1 shows the localisation of organelle markers Bcl-2 (mitochondria; MT), Bip (endoplasmic reticulum; ER), and EEA1 (early endosome; EE) in the MoRK13 cells. As expected, the lipid raft marker Flotillin 1 was enriched within fractions at the buoyant end of the gradient, consistent with the high cholesterol:protein ratio resulting in a significantly lower density of lipid rafts compared to other solubilised membrane proteins [35]. This was confirmed by dot blot analysis of the lipid raft marker GM₁, detected by the cholera toxin subunit B (CTB). Figure 1 also indicates that the subcellular localisation of PrP^C in MoRK13, and total PrP (PrP^C and PrP^res, indistinguishable from each other under the conditions of these western blots) in MoRK13-inf is predominantly in the buoyant, lipid raft fractions. There were no substantial differences in organelle or membrane markers or PrP localisation when comparing the MoRK13 and MoRK13-inf cells, and also comparing vector-only transfected RK13 (vecRK13), GT1-7H and GT1-7H-inf cells (Additional file 2: Figure S2); therefore over-expression of PrP^C, prion infection and cell line origin (neuronal/non-neuronal/murine/rabbit) do not appear to alter localisations, or create artefacts in this system of subcellular fractionation. Previous reports indicate PrP^res is also localised in lipid rafts [36]. Figure 2A shows the localisation of MoRK13-inf PrP^res to be principally in lipid raft fractions, with a similar distribution to total PrP (Figure 1) as determined by both western blot and dot blot. Quantification of relative PrP^res levels (Figure 2B) highlights that the lipid raft enriched fractions contain up to eight times the detectable PrP^res levels than the ER/MT marker enriched fractions (eg compare fractions #3 and #8). By way of comparison to the original prion strain, Nycodenz gradient distribution of PrP isoforms and marker proteins from M1000 brain, homogenised in a comparative manner, was also similar to that observed in cell lysates (Additional file 3: Figure S3).

In order to determine whether detectable subcellular PrP^res levels correlate with infectivity, an in vitro cell culture transmission study was carried out of equivalent volumes of fractions obtained after density gradient flotation assays of MoRK13-inf cell lysate using the highly susceptible MoRK13 as recipient cells. Levels of PrP^res produced by recipient cells after exposure to fractions were quantified and used as a surrogate indicator of relative infectivity titres contained within the subcellular fractions. The representative cell blot and quantification (Figure 3A, B) clearly show that infectivity was contained within all fractions derived from MoRK13-inf cells, albeit to varying degrees. The EE enriched fractions (predominantly #9 and #10) contained considerably less infectivity compared with the lipid raft enriched fractions (predominantly #3 and #4), and the ER and MT marker enriched fractions (predominantly #7 and #8), although the comparison of fraction #7 and fraction #9 did not reach significance. Interestingly, there was no significant difference in the amount of infectivity contained within the lipid raft enriched fractions (#3 and #4), and the ER and MT marker enriched fractions (#7 and #8). As shown in Figure 3A, fractions were also diluted in a 1/2 log series, to ensure there was no limiting sensitivity of the in vitro assay truncating the upper end of the dynamic range. Importantly, neat fractions that elicited the highest production of PrP^res by recipient MoRK13 cells (for example fractions #4 and #8), were infectious to the same dilutions (1:100), thereby validating the use of PrP^res produced by recipient cells as a measure of relative infectivity contained within fractions.

Correlation of the percentage of PrP^res present in each fraction with the level of in vitro infectivity produced by each fraction (Figure 4A) revealed a discrepancy in the ER/MT enriched fractions. The relative levels of PrP^res present in the lipid raft enriched fractions (#3 & #4) and the early endosome enriched fractions (#9 and #10) correspond with the relative levels of PrP^res produced by the recipient MoRK13 cells. In contrast, the ER/MT enriched fractions (#7 and #8) that display low detectable PrP^res levels, contained high infectivity levels similar to the lipid raft enriched fractions, with the discrepancy at fraction #8 highly significant (p < 0.001; two way ANOVA). Therefore, there was a reproducible and significant divergence between PrP^res and infectivity levels within ER/MT marker enriched fractions.

To confirm genuine prion infectivity, selected fractions were bioassayed in Tga20 PrP^C over-expressing mice. The fractions were chosen to provide a range of PrP^res and infectivity level combinations; ie high PrP^res and infectivity levels (fraction #4), low PrP^res and infectivity levels (fraction #10), and low PrP^res but high infectivity levels (fraction #8). Controls included inoculating mice with whole cell lysate from naive MoRK13 and MoRK13-inf cells, as well as M1000 brain homogenate, in order to make comparisons with the original prion strain. Figure 4B depicts the incubation periods for the selected fractions and control mice. Mice exposed to uninfected MoRK13 cell lysate were symptom free at 245 days post-inoculation. Mice inoculated with 0.01% M1000 brain homogenate had a significantly shorter incubation period than mice inoculated with MoRK13-inf whole cell lysate. Importantly, concordant with the in vitro cell culture transmissions, mice inoculated with fractions #4 and #8 had indistinguishable incubation periods, despite the significantly different PrP^res levels contained within these fractions. Mice inoculated with fraction #10 had significantly longer incubation periods than mice infected with the other two fractions, again concordant with the relative infectivity levels determined by the cell culture transmissions. For illustrative purposes, this 35 day extension in incubation period approximates a 3 log reduction of infectious titre when modelled on a time interval assay developed in Tga20 mice inoculated with M1000 brain homogenate derived prions (Additional file 4: Figure S4). In summary, the in vivo findings faithfully recapitulate and validate the MoRK13 in vitro model.

Neuropathologically, prion diseases are characterised by the presence of neuronal loss, spongiform change, proliferation of astrocytes and extracellular PrP deposits [37]. Further, for different prion strains the topographical distribution of neuropathological changes is distinctive, creating characteristic lesion profiles [38, 39]. All diseased mice displayed symptoms typical of M1000 prion infection and brains were harvested to assess neuropathological changes. As expected, control mice inoculated with MoRK13 whole cell lysate showed no neuropathological abnormalities (Additional file 5: Figure S5). Figure 5 displays the quantification of the neuropathological features mentioned above within different brain regions, clearly indicating that each inoculum produced very similar lesion profiles (for representative photomicrographs of the histological and immunohistochemical findings see Additional file 5: Figure S5). This distribution pattern of the pathological features, with predominant involvement of the thalamus, midbrain and pons, and minimal involvement of the cerebellum is consistent with previous observations of the M1000 strain [40, 41]. The only significant difference detected in lesion profiles was between MoRK13-inf whole cell lysate and fraction #10 inocula, with the former having significantly more reactive astrocytes in the occipital cortical region. The explanation for this is uncertain, but may suggest an additive effect of the individual fractions, which contribute to the whole-cell lysate profile. Different prion strains often have distinctive PrP^res glycoform ratios and mobilities on western blot analysis [42‐46]. To further evaluate the possibility of different MoRK13-inf subcellular fractions associating with distinct prion strains, the brain PrP^res profiles from Tga20 mice inoculated with selected fractions were examined by western blot. Similar to the lesion profiling, there were no significant differences in PrP^res patterns when comparing MoRK13-inf whole cell lysate with each of the selected fractions, in keeping with them all being the same prion strain (Additional file 6: Figure S6).

In order to explore whether subcellular fraction specific co-factors might contribute to the observed transmission efficiency spectrum, M1000 brain homogenate was spiked into fractions derived from uninfected MoRK13 or vecRK13 cells, prior to use as inocula in vitro. Figure 6A shows that even at the lowest dilutions of M1000 in MoRK13 fractions, PrP^res was produced by recipient MoRK13 cells. However, quantification (Figure 6B) shows there were no significant differences in amounts of PrP^res produced when comparing the various MoRK13 subcellular fraction diluents. Notably, there was a general trend for MoRK13 subcellular fractions to enhance M1000 transmission efficiency to MoRK13 cells compared to lysis buffer alone. A similar result was observed using fractions obtained from vecRK13 cells (Additional file 7: Figure S7), suggesting this trend is not PrP^C specific or dependent.

To further evaluate the possible contribution of cellular co-factors on the efficiency of prion infection, M1000 brain homogenate was diluted in PBS or selected subcellular fractions from uninfected MoRK13 cells (fractions #4, #8 and #10), and bioassayed in Tga20 indicator mice. M1000 was also diluted in 'empty' Nycodenz fractions (#4, #8 and #10), in order to control for any affect the Nycodenz gradient material itself may have on incubation period or the neuropathology. Neither the exogenous cellular co-factors in the selected fractions, nor Nycodenz alone, had any significant affect on the incubation time (Figure 6C) or overtly affected the neuropathology (Addtional file 8: Figure S8) of M1000 in Tga20 mice.

There is experimental evidence that some disease associated [47‐50], and synthetic [51] prions, may be protease sensitive. To test whether the infectious prions within the MoRK13-inf fractions may differentially contain significant amounts of protease sensitive species, selected fractions were digested mildly with PK prior to using them as a source of infectivity. There were no significant differences in PrP^res levels produced by MoRK13 infected with PK digested fractions #4, #8 or #10 compared to cells infected with the untreated fraction counterpart (Figure 7A, B). However there was a modest general trend towards reduced PrP^res production by recipient cells after PK treatment of each fraction inoculum.

Evidence suggests misfolded prion protein aggregate size correlates with efficiency of conversion or prion infection [30, 31]. As a consequence, in vitro conversion assays such as PMCA may be most efficient when incorporating sonication steps [52, 53]. To assess this possibility, selected fractions (#4, #8 and #10) were subjected to sonication (equivalent to one 'round' of PMCA) prior to using them as a source of infectivity in the MoRK13 cell culture model. As demonstrated by the representative cell blot in Figure 7C, quantified in Figure 7D, only subcellular fraction #4 had significantly increased levels of infectivity compared to its untreated fraction.

Rabbit kidney epithelial cells which have no detectable endogenous PrP^C protein, were stably transfected to over-express murine PrP^C (MoRK13) or the empty vector (vecRK13) [78] and mouse hypothalamic GT1-7H cells were maintained as described previously [79], in a humidified incubator at 37°C with 5% CO₂.

A method of non-toxic/non-detergent cell lysis and subcellular separation was necessary to allow subsequent use of fractions for infecting recipient cells or mice. Also, due to the possible involvement of lipid rafts in prion conversion, a lysis method was chosen in order to maintain lipid raft integrity and buoyancy [35], with minor modifications. Briefly two confluent T175 cm² flasks (approximately 4 x10⁷ cells) were washed twice with 20 mls ice cold lysis buffer (20 mM Tris-Cl pH 7.8, 250 mM sucrose, 1 μM CaCl₂, 1 μM MgCl₂) and harvested by scraping into a further 20 ml of lysis buffer and pelleting at 700 × g for 3 minutes. The cell pellet was re-suspended in 500 μl cold lysis buffer. Cells were lysed on ice, by passing the cell suspension through a 22 g needle exactly twenty times, and the crude lysate was centrifuged at 1000 × g, 10 minutes at 4°C. The post-nuclear supernatant was retained on ice and the extraction repeated on the pellet. The lysate was then assayed for total protein content by performing a bicinchoninic acid (BCA) assay (Pierce, Thermo Scientific, Scoresby, VIC, AUS) as per the manufacturer's instructions, and adjusted with lysis buffer to 1.8 mg/ml. The Nycodenz (HistoDenz™, Sigma-Aldrich, Castle Hill, NSW, AUS) density gradient fractionation method was adapted from a published protocol [80], to suit a Beckman Optima Max-E Benchtop Ultracentrifuge and MLS-50 rotor. Nycodenz solutions were prepared in TNE (25 mM Tris-Cl pH 7.5, 150 mM NaCl, 5 mM EDTA) and an 8-35% linear step Nycodenz gradient was poured (400 μl of each of 8%, 12%, 15%, 18%, 20%, 22.5% and 25%) with 1.1 ml of a 35% Nycodenz cushion consisting of equal parts ice cold 70% Nycodenz and cell lysate (1 mg total protein) pipetted to the bottom of the gradient. In some cases, an 'empty Nycodenz' gradient was poured, whereby the 35% Nycodenz/lysate cushion mixture was substituted for 35% Nycodenz alone. Nycodenz gradients were centrifuged at 200,000 × g (average) for 342 minutes at 4°C. Following centrifugation, 10 equal volume fractions of 390 μl were collected and stored at -80°C or kept on ice for immediate use.

The M1000 prion strain used in this study was derived from a well characterised stock of pooled mouse brain homogenate [81]. Recipient MoRK13 or GT1-7H cells were infected using an overlay technique as described previously [79]. For comparisons of cell lysate and brain homogenate M1000 infectivity, cell lysates were prepared by harvesting and lysing in sterile phosphate buffered saline (PBS; Invitrogen, Mulgrave, VIC, AUS) by three cycles of freezing (10 minutes at -80°C) and thawing (3 minutes at 37°C), and centrifugation at 1000 × g for 3 minutes at 4°C to obtain a post-nuclear supernatant. The total protein content of the PBS supernatant and M1000 brain homogenate were determined by BCA assay, and the lysates and homogenate were balanced to the same protein concentration with PBS. Following this 100 μl of lysate or homogenate was mixed with 400 μl of complete medium and this was used to infect recipient cells. For fraction infections 100 μl fraction ('neat') or, where indicated, fraction which had been serially diluted in medium, was mixed with 400 μl complete medium, and used to infect recipient MoRK13 cells. For 'spiking' experiments, M1000 brain homogenate was diluted in 400 μl media and mixed with 100 μl MoRK13 or vecRK13 fraction (or as a control the lysis buffer used to prepare cells prior to fractionation) to give a final concentration of 0.05%, 0.025% and 0.0125% M1000, prior to being used to infect recipient MoRK13 cells.

For PK digestion, 100 μl of a fraction was treated with a final concentration of 1 μg/ml PK for 8 hours at 37°C, conditions found to reduce PrP^C by approximately 80% when tested on control fractions (data not shown). For sonication pre-treatment, 100 μl of a fraction in a 1.5 ml microfuge tube was subjected to 60 seconds at amplitude 70 in a S4000 sonicator (Misonix, Farmingdale, NY, USA) with microplate horn adapter in 300 mL water maintained at 37°C. The PK digested or sonicated fractions (and untreated controls) were mixed with 400 μl of complete medium and used for in vitro infections as described above.

For determination of subcellular localisation of PrP and other proteins, fractions were mixed with 4X sample buffer and subject to PAGE (using either 4-20% or 10-20% Tris-glycine SDS or 4-12% Bis-tris NuPAGE pre-cast gels (Invitrogen), depending on the size of the proteins to be detected, and then transferred to PVDF membrane for western blotting of PrP as described previously [79] and organelle marker proteins using antibody dilutions as outlined in the manufacturer's instructions (BD Pharmingen™ Organelle Sampler Kit, BD Biosciences, North Ryde, NSW, AUS). For detection of the ganglioside GM₁ (lipid raft marker), 3 μl of each fraction was spotted onto nitrocellulose membrane, and allowed to dry for 20 minutes at 37°C. The membrane was blocked for a minimum of 1 hour in 5% skim milk powder in PBS containing 0.05% (v/v) Tween-20 (PBST) and then incubated in 1:100,000 cholera toxin B subunit (CTB)-horseradish peroxidise (HRP) conjugate (Sigma, stock concentration 0.45 mg/ml CTB and 1 mg/ml HRP in H₂O) solution in block for 1.5 hours at room temperature prior to chemiluminescent detection (ECL Plus, GE Healthcare, Rydalmere, NSW, AUS). For PrP^res detection, fractions, cell lysates or brain homogenate were digested with a final concentration of 50 μg/ml PK, 1 hour at 37°C before SDS-PAGE and western blotting. Dot blots of fractions were also carried out for PrP^res detection, whereby 5 μl of fraction was spotted onto nitrocellulose membrane, which was dried 30-60 minutes at 37°C and then treated in exactly the same manner as the cell blot assay nitrocellulose membrane. Cell blots for the detection of PrP^res in recipient cells, at four passages (P4) post-infection were carried out as described previously [79]. All immunoblotting (western blots and cell blots) for PrP species used the monoclonal antibody ICSM18, (D-Gen, London, UK). All chemiluminescent images were captured by a Fujifilm LAS-3000 (Berthold Australia, Bundoora, VIC, AUS).

All animal experiments were carried out in strict accordance with the 'Australian Code of Practice for the Care and Use of Animals for Scientific Purposes (NHMRC)', with approval from the University of Melbourne Animal Ethics Committee (AEC #04154). Tga20 PrP^C over-expressing mice [82] were anesthetised using methoxyfluorane and inoculated intracerebrally with 30 μl of 0.01% M1000 brain homogenate diluted in PBS or the appropriate fractions as indicated, or with 30 μl of 'neat' fraction or whole cell PBS lysate. Mice were provided food and water ad libitum and housed following routine animal husbandry practices. Mice were examined daily for symptoms of prion disease. Once mice developed persisting features of advanced prion disease, including impaired righting reflexes, hunched posture and hind limb paresis, they were culled by cervical dislocation under anaesthesia and the number of days post-inoculation was recorded. Brains were removed and sagittally hemi-sectioned, with half the brain fixed and stained to allow scoring of vacuolation, astrocytic gliosis and PrP deposition as described previously [40], and the other half made to 10% (w/v) homogenates in PBS, with homogenates stored at -80°C until required. Neuropathological scoring was performed on two separate occasions, blinded as to the inoculum group, to provide a semi-quantitative comparison of lesion profiles between the groups of animals. Stained sections were visualised using a Zeiss Axioskop 50 microscope with images captured using a Zeiss AxioCam HRC camera (Carl Zeiss, North Ryde, NSW, AUS).

All densitometric analyses used the public domain ImageJ software (National Institutes of Health, USA). For determining relative levels of PrP^res in fractions, each fraction PrP^res dot blot signal intensity was measured, with the sum of the 10 individual PrP^res levels providing the 'total PrP^res'; each individual fraction was expressed as a percentage of the total PrP^res. For determining relative levels of infectivity contained within MoRK13-infectious fractions, PrP^res produced by MoRK13 cells exposed to each 'neat' fraction was measured at passage 4 (P4) post-exposure by cell blot signal intensity, with the sum of the 10 individual PrP^res levels providing 'total PrP^res'. Once again, each individual fraction was expressed as a percentage of the calculated total PrP^res. All statistical analyses were performed in GraphPad Prism 4, with one way ANOVA and Tukey's multiple comparisons or two way ANOVA and Bonferonni post-tests used as indicated, unless stated otherwise.