Mechanism of scrapie prion precipitation with phosphotungstate anions.

The phosphotungstate anion (PTA) is widely used to facilitate the precipitation of disease-causing prion protein (PrP(Sc)) from infected tissue for applications in structural studies and diagnostic approaches. However, the mechanism of this precipitation is not understood. In order to elucidate the nature of the PTA interaction with PrP(Sc) under physiological conditions, solutions of PTA were characterized by NMR spectroscopy at varying pH. At neutral pH, the parent [PW12O40](3-) ion decomposes to give a lacunary [PW11O39](7-) (PW11) complex and a single orthotungstate anion [WO4](2-) (WO4). To measure the efficacy of each component of PTA, increasing concentrations of PW11, WO4, and mixtures thereof were used to precipitate PrP(Sc) from brain homogenates of scrapie prion-infected mice. The amount of PrP(Sc) isolated, quantified by ELISA and immunoblotting, revealed that both PW11 and WO4 contribute to PrP(Sc) precipitation. Incubation with sarkosyl, PTA, or individual components of PTA resulted in separation of higher-density PrP aggregates from the neuronal lipid monosialotetrahexosylganglioside (GM1), as observed by sucrose gradient centrifugation. These experiments revealed that yield and purity of PrP(Sc) were greater with polyoxometalates (POMs), which substantially supported the separation of lipids from PrP(Sc) in the samples. Interaction of POMs and sarkosyl with brain homogenates promoted the formation of fibrillar PrP(Sc) aggregates prior to centrifugation, likely through the separation of lipids like GM1 from PrP(Sc). We propose that this separation of lipids from PrP is a major factor governing the facile precipitation of PrP(Sc) by PTA from tissue and might be optimized further for the detection of prions.

The fundamental event in prion diseases caused by the prion protein (PrP) is the misfolding of normal PrP (PrPC), which leads to progressive neurodegeneration in many mammals.[1] Such diseases include Creutzfeldt–Jakob disease in humans, scrapie in sheep, bovine spongiform encephalopathy in cattle, and chronic wasting disease in deer and elk.[2] In these prion diseases, endogenous PrPC adopts a self-propagating conformation, designated PrPSc, which can autocatalytically convert PrPC molecules into infectious PrPSc.[3,4] PrP misfolding can be caused by genetic mutations, exposure to PrPSc, or occur spontaneously, resulting in inherited, acquired, or sporadic disease manifestations, respectively.[5] Over time, PrPSc accumulates and spreads throughout the brain, leading to severe neuronal deterioration, loss of cognitive function, and ultimately death. While diseases caused by PrPSc are rare in humans, biochemical insights into PrPSc may provide valuable applications to other, more common neurodegenerative disorders. In recent studies, a self-perpetuating prion mechanism has been described for amyloid β peptides in Alzheimer’s disease, α-synuclein in Parkinson’s disease, and tau in the tauopathies.[6−11]

A detailed knowledge of the structure of PrPSc would greatly facilitate efforts to prevent, diagnose, and develop therapeutics for prion diseases. PrPSc is rich in β-sheet content that renders the large aggregates insoluble and thus difficult to characterize by conventional biophysical techniques. Despite studies employing X-ray fiber diffraction,[12] electron microscopy,[13] and hydrogen–deuterium exchange,[14] the structure of PrPSc remains elusive. Structural models of PrPSc have been proposed from X-ray fiber diffraction and electron microscopy of prions isolated from brain tissue.[12,15] However, such fibril preparations generally contain large amounts of crystalline lipids, which may obscure structural features of the protein aggregates.

Biophysical studies require the purification of a relatively large amount of PrPSc. Historically, prion purification and enrichment required a series of time-intensive gradient or differential centrifugation steps, which resulted in highly purified samples but with a low yield.[16−18] A rapid, alternative method employs an inorganic polyoxometalate (POM) complex known as the phosphotungstate anion (PTA, [PW12O40]3–). Since the discovery that PTA selectively promotes precipitation of PrPSc over PrPC,[19] it has been widely used to isolate PrPSc from various tissues of prion-infected animals.[20−25] The success of PTA has stimulated efforts to vary the structure, charge, and composition of related complexes to optimize PrPSc purification[26] and to influence the resulting type of aggregate assemblies.[15]

PTA precipitation has also been used to concentrate PrPSc for diagnostic purposes.[27,28] Currently, a definitive diagnosis of prion diseases caused by PrPSc can only be made post-mortem. Efforts have been made to develop diagnostic tools for detecting PrPSc in the blood of humans and cattle,[29,30] and PTA has been used to facilitate detection of very low amounts of PrPSc in peripheral tissues of patients with Creutzfeldt–Jakob disease.[31] However, achieving a rapid and reliable method for PrPSc detection has been hampered by difficulties in discriminating infectious PrPSc from endogenous PrPC, low titers of PrPSc in blood, and the presence of lipid and protein contaminants that can complicate the accuracy of determinations.[32]

Despite the widespread use of PTA to isolate PrPSc, the mechanisms by which PTA effectively precipitates PrPSc are still largely unknown. PTA does not appear to induce a conformational change of PrPC to PrPSc, since amyloid seeding assays have shown that PTA does not promote conversion of PrPC to PrPSc in multiple strains of uninfected mice.[28] Using microscopic and immunological techniques, we found that PTA and sarkosyl promoted the formation of fibrils and increased the density of PrPSc aggregates. Na3[PW12O40] (NaPTA) decomposed at neutral pH to give a lacunary [PW11O39]7– (PW11) complex and a single orthotungstate [WO4]2– unit (WO4). When used as a precipitant, the lacunary PW11 complex of PTA resulted in high yields of PrPSc and decreased neuronal lipid monosialotetrahexosylganglioside (GM1) content in the samples, while WO4 was less effective at removing GM1 lipids at equivalent concentrations. These results describe several potential mechanisms for the effective precipitation of PrPSc by PTA.

Results and Discussion

PTA and Sarkosyl Promote Formation of PrPSc Fibrils

In the standard method to precipitate PrPSc using PTA, prion-infected brain homogenates are incubated with PTA and the detergent sodium lauryl sarcosinate (sarkosyl) in phosphate-buffered saline (PBS) prior to a low-speed centrifugation. To test the influence of each precipitant on the aggregation state of PrPSc, we incubated brain homogenates (10% w/v) from FVB mice infected with Rocky Mountain Laboratory (RML) scrapie prions with different concentrations of sarkosyl (0–3.5% w/v), PTA (0–3.4% w/v), and combinations of 0–3.4% PTA with 2% sarkosyl. These scrapie-infected mice generally do not show preformed PrPSc fibrils or amyloid deposits in their brains.[33] The resulting PrPSc fibrillar aggregates were analyzed by electron microscopy of sample aliquots and quantified (Figure 1a). When neither sarkosyl nor PTA was used, very few fibrils were observed (Figure 1b). When used independently, both sarkosyl and PTA induced the formation of large fibrillar structures (Figure 1c,d, respectively). With PTA alone, higher concentrations yielded greater numbers of aggregates. The greatest number of fibrils detectable by electron microscopy was found from using a combination of PTA and sarkosyl, with the highest yield of fibrillar aggregates obtained using 2% PTA and 2% sarkosyl (Figure 1e), in agreement with established optimal concentrations from earlier studies.[12,15,26]

Figure 1

Quantification of electron microscopy images (a) shows how preparative methods with no additives (b), sarkosyl detergent alone (c), PTA alone (d), and PTA in combination with sarkosyl (e) influence the amount of PrPSc fibrils obtained at different reagent concentrations. Error bars in panel a represent the standard error of the mean from 10 representative electron micrographs, and examples of fibrils indicated by the letters in panel a are shown in panels b–e (arrows).

PTA Facilitates PrPSc Separation from Lipids

We further investigated the influence of PTA and sarkosyl on the hydrodynamic properties of aggregates by centrifugation of RML-infected brain homogenate in a three-step sucrose gradient. Brain homogenate (20% w/v) was prepared from RML-infected mice using PBS and 2% sarkosyl. The homogenate was then incubated at 37 °C overnight with or without 2% PTA, and then 1.5 mL of treated or untreated brain homogenate was loaded onto a sucrose gradient consisting of an 8 mL layer of 50% sucrose above a 2 mL layer of 80% sucrose. After centrifugation at 134 000g for 16 h, fractions from lowest to highest density were collected and immunoblotted for PrP levels and for GM1, the main ganglioside in neuronal membranes[34] and a marker for glycolipid raft domains[35,36] (Figure 2).

Figure 2

Sucrose cushion centrifugations of prion-infected brain homogenates reveal levels of PrP and GM1 by immunoblotting. (a) Schematic representation of the three-step gradient centrifugation, where the brain homogenate was layered on top of sucrose layers of increasing density. The resulting fractions were labeled A–P and correspond to lanes in the immunoblots. (b, c) Western blots of brain homogenate incubated without (b) or with (c) PTA. PrP (upper blots) and GM1 (lower blots) colocalized in the absence of PTA, whereas PrP separated from GM1 in the presence of PTA. Molecular weight markers of migrated protein standards are shown in kilodaltons (kDa).

In the absence of PTA, PrP largely remained in fractions above the 50–80% sucrose interface, with most of the PrP settled at the interface of the aqueous buffer fraction and the first sucrose-containing fraction (Figure 2b, lanes C and D). In addition, the highest GM1 content was found to colocalize with PrP in the low-density fractions. This observation is consistent with the presence of a C-terminal glycophosphatidylinositol (GPI) anchor on PrP, which strongly interacts with lipids and increases the buoyancy of both monomeric PrPC and aggregated PrPSc.[37,38] In the presence of PTA, PrP was found in higher density fractions, with the highest levels found below the 50–80% sucrose interface (Figure 2c, lanes K and L) and many aggregates clearly penetrating the 80% layer. With PTA, GM1 was also separated from the PrP aggregates, as evidenced by GM1 remaining in lower-density fractions while PrP migrated into higher-density fractions, even giving rise to a sizable amount of PrP in the pellet fraction. In this way, PTA also provided effective separation of monomeric PrPC from aggregated PrPSc, since the buoyant PrPC remained in the less dense GM1-containing fractions while aggregated PrPSc was more readily precipitated. This effect is further illustrated in samples treated with proteinase K (PK), which degrades PrPC and cleaves PK-resistant PrPSc into PrP 27–30 (Supplementary Figure S1, Supporting Information). Without PTA, the majority of PrP 27–30 was found in the lower density fractions together with GM1 in proteolytically digested samples. Upon addition of PTA, however, PrP 27–30 migrated to the higher density fractions, effectively separating PrP from GM1 lipids, which remained in the low-density fractions.

PTA altered the distribution of PrP and GM1 in sucrose cushion centrifugations, suggesting that PTA may influence the density of PrP aggregates in several different ways. Direct interaction of PrP aggregates with heavy, tungsten-containing compounds like PTA may increase their density and promote the formation of larger PrPSc aggregates. In order to determine the effects of ionic strength on precipitation efficiency, sucrose cushion centrifugations were also performed on both PK-digested and undigested samples treated with sodium chloride (Supplementary Figure S2, Supporting Information). Addition of NaCl facilitated some degree of GM1 separation from PrP in both cases. However, NaCl did not significantly promote precipitation of PrP, suggesting that the efficacy of PTA may involve its direct interaction with PrP rather than simply increasing the ionic strength of the solution.

We speculate that the release of PrP from lipids is also a key factor in promoting fibrillization, as observed in the fibrillization studies using sarkosyl (Figure 1). Indeed, transgenic mice expressing PrP without the GPI anchor showed increased propensity to develop amyloid fibrils in their brains.[39,40] These observations suggest that interaction of PrPSc with lipid membranes may hinder fibril formation, and use of polyanions like PTA may aid in disrupting the membrane–PrPSc interaction to facilitate fibrillization of PrPSc, resulting in denser aggregates that are easier to precipitate than their lipid-bound counterparts.

Decomposition of PTA at Varying pH Studied by 31P NMR

To further elucidate the mechanisms of the PTA precipitation, the POM speciation was determined under a range of conditions, particularly those typically used for PrPSc precipitation. Aqueous solutions of PTA at varying pH were studied by 31P NMR (Figure 3). This technique is very useful in tracking changes in the POM composition because the chemical shift of the central phosphorus atom reflects its unique chemical environment and can be used to determine which species are present. PTA samples were also prepared in complex matrices by dissolving PTA in solutions containing PBS, sarkosyl, and brain tissue at pH 7, which confirmed that PTA speciation did not change in the presence of brain homogenate compared with a water-based solution (Supplementary Figure S3, Supporting Information). Furthermore, the speciation of PTA at neutral pH in water or brain homogenate did not change over the course of eight weeks, as observed by 31P NMR (Supplementary Figure S4, Supporting Information).

Figure 3

31P NMR demonstrates that PTA anion speciation changes dramatically with pH. (a, b) NMR spectra for acidic to neutral pH conditions (a) and for neutral to alkaline pH conditions (b). (c) Up to pH 7.0, the parent [PW12O40]3– ion decomposes to give a 1:1 mixture of [PW11O39]7– and [WO4]2–. (d) Above pH 7.1, the lacunary [PW11O39]7– ion begins to decompose into H[PO4]2– and 11 [WO4]2– complexes. Structural representations for [PW12O40]3–, [PW11O39]7–, and [WO4]2– are from crystallographic information files found in refs (41, 60, and 61), respectively.

PTA is synthesized by acidification of an aqueous solution containing stoichiometric amounts of tungstate and phosphate to give the product [PW12O40]3– (PW12).[41] When this complex is used in biological contexts, the increase in pH from 1 to 7 leads to notable changes in the structure of the POM in solution.[42−44] By NMR spectroscopy, we confirmed that commercial Na3[PW12O40] (NaPTA) decomposed at neutral pH to give a lacunary [PW11O39]7– (PW11) complex and a single orthotungstate [WO4]2– unit (WO4) in a 1:1 ratio (Figure 3a). As the pH increased above 7.1, however, we observed decomposition of PW11 into its original components, as evidenced by the decreasing signal at −11 ppm and the appearance of free phosphate at 2 ppm (Figure 3b). Decomposition of the parent PW12 anion into PW11 and WO4 at neutral pH is represented in Figure 3c, and subsequent decomposition of PW11 into phosphate and tungstate is shown in Figure 3d. The assignments of 31P NMR shifts were confirmed by electrospray ionization mass spectrometry studies at varying pH (Supplementary Figure S5, Supporting Information). With different sizes and charge densities, PW11 and WO4 may each have distinct interactions with PrP and have different effects on the yield and purity of PrPSc obtained from PTA precipitation.

Effect of PTA-Derived Anions on PrPSc Precipitation

To measure the precipitation efficacy of each PTA component under physiological conditions, increasing concentrations of PW11, WO4, and mixtures thereof were used to precipitate PrPSc from brain homogenate. Brain homogenates from RML-infected FVB mice were incubated at 37 °C for 1 h with varying concentrations of POM solutions, ranging from 0.25 to 8 mM. POM stock solutions were made fresh by dissolving the following tungstates in water and adjusting the pH to 7.1: Na7[PW11O39], Na2WO4, a 1:1 mixture of PW11 and WO4, and commercial NaPTA. Following incubation with the POM solutions, the brain samples were centrifuged at 14 000g for 1 h.[15,19] The supernatants were discarded, and PrP levels were measured in the pellets by enzyme-linked immunosorbent assay (ELISA) (Figure 4) or Western immunoblotting (Figure 5).

Figure 4

ELISA was used to quantify the amount of PrP precipitated from brain homogenates following incubation with varying concentrations of POMs. NaPTA (red), PW11 (green), and PW11 + WO4 (purple) show a similar profile, whereas WO4 (blue) appears to increase linearly in this concentration range. The yellow bar indicates the typical window of PTA concentrations used for PrPSc precipitations. Error bars represent the standard error of the mean across triplicate measurements of three independent sets of brain homogenate (7 brains/set).

Figure 5

Immunoblots show the amounts of PrP and ganglioside GM1 in brain homogenates following incubation with different concentrations (0–8 mM) of POMs, as indicated. From top to bottom, POMs were sodium phosphotungstate (NaPTA), Na7[PW11O39] (PW11), Na2[WO4] (WO4), and PW11 + WO4. Molecular weight markers of migrated protein standards are shown in kilodaltons (kDa).

Three independent sets of POM-precipitated samples were exposed to highly denaturing conditions (4 M guanidine hydrochloride at 100 °C) and then subjected to a sandwich ELISA (Figure 4). Commercial NaPTA resulted in a steep rise in the levels of detectable PrP when the concentration was increased from 0 to 1 mM and decreased gradually for the higher concentrations up to 8 mM. This trend was also observed for PW11 and the 1:1 mixture of PW11 and WO4. By comparison, only a modest increase in detectable PrP was observed for equivalent WO4 concentrations, suggesting that PW11 may have greater contribution in the facile detection of PrP than WO4 by this method. The enhancement of PrP detection using POMs at different concentrations can also be expressed as a ratio of the observed values normalized to control samples where no POMs were used (Table 1). However, lipids have been shown to interfere with the detection of cholesterol, proteins, and other biomolecules by ELISA.[45−47] Indeed, several studies have shown that lipids can obscure the detection of PrPSc by ELISA in blood[32,48] and tissue homogenates.[49]

Table 1

Relative Enrichment of PrP Levels at Different Concentrations of POMsa

POM concn (mM)	PTA	PW11	WO4	PW11 + WO4
0	1.0 ± 0.2	1.0 ± 0.2	1.0 ± 0.2	1.0 ± 0.2
0.25	6.1 ± 0.1	4.1 ± 0.3	2.1 ± 0.1	5.8 ± 0.4
0.5	6.1 ± 1.5	5.3 ± 0.7	2.4 ± 0.2	6.8 ± 0.3
1	9.3 ± 1.5	7.1 ± 0.3	2.5 ± 0.3	8.1 ± 0.6
2	9.0 ± 2.4	7.0 ± 0.3	2.7 ± 0.2	9.0 ± 2.1
3	7.5 ± 1.2	7.7 ± 0.6	3.4 ± 0.3	8.5 ± 1.1
4	5.1 ± 1.0	6.7 ± 0.4	2.9 ± 0.1	8.5 ± 1.2
6	4.7 ± 1.6	6.8 ± 0.1	3.6 ± 0.3	6.9 ± 1.2
8	1.7 ± 0.5	3.6 ± 0.6	3.6 ± 0.1	4.7 ± 0.8

Compared with PrP levels without precipitant, as detected by ELISA.

We performed Western immunoblotting to determine the relative level of PrP enrichment achieved by each PTA component using the D13 antibody conjugated to horseradish peroxidase (HRP);[50] GM1 levels were also probed using the HRP-coupled cholera toxin B subunit, which specifically recognizes the ganglioside GM1 (Figure 5). GM1 is the predominant ganglioside in neuronal membranes and is often used as marker for lipid rafts.[51,52] While other raft lipids such as cholesterol were not measured directly, the detection of GM1 with cholera toxin-based reagents is straightforward, and the presence of GM1 can be used as an indicator for the general distribution of neuronal lipids in relation to PrP.

At concentrations from 0.25 to 8 mM, NaPTA and the 1:1 mixture of PW11 and WO4 showed similar levels of PrP, in agreement with the decomposition of the parent PW12 ion into PW11 and WO4 at physiological pH. Analysis of these POMs alone revealed that both PW11 and WO4 contributed to the precipitation efficiency of PTA solutions. However, the separation of lipids from PrPSc differed substantially for each PTA component at equimolar concentrations. PW11 at concentrations >1 mM resulted in a substantial decrease in GM1 levels, whereas WO4 did not affect GM1 levels. These results suggest that while both PTA components contribute to the precipitation of PrP, PW11 may be more effective in disrupting the interaction between PrP and lipids such as GM1, thereby increasing the relative purity of the precipitates.

Comparing the results from the ELISAs and Western blots, the discrepancy in the levels of PrP measured by these two methods at very low concentrations of POM likely results from the presence of GM1, and perhaps other lipids, associated with PrP. In cases where sensitivity of PrPSc detection is crucial, it is therefore important to use high enough concentrations of PTA (or PW11 in particular) to reduce the amount of coprecipitating lipids and facilitate detection of PrPSc in blood and tissue samples for diagnostic purposes.

Conclusions

Using spectrochemical, biophysical, and microscopic techniques, we observed that the effective precipitation of PrPSc by PTA may occur by some combination of facilitated fibril formation, enhanced separation of lipids from PrPSc, and increased density of PrPSc aggregates. The increase in the amount of large aggregates obtained when PTA and sarkosyl were used together (Figure 1) suggests that the separation of lipids such as GM1 from PrPSc is a major factor that can influence prion precipitation. It is possible that the observed fibrillization is a direct consequence of lipid removal by PTA and sarkosyl, since PrPSc detached from membranes readily forms amyloid.[39] Additionally, ganglioside GM1, the most abundant neuronal ganglioside,[35] was also studied as a marker of total lipid content and an indicator of sample purity. We found decreased levels of GM1 with higher concentrations of NaPTA, PW11 + WO4, and PW11 while WO4 was not as effective at equimolar concentrations (Figure 5).

From sucrose cushion centrifugations, incubation with PTA increased the density of PrP aggregates (Figure 2). PrPSc alone has a density of approximately 1.35 g/mL, while an 80% sucrose solution has a density of ∼1.41 g/mL.[53] The PrP aggregates that permeated the 80% sucrose cushion and formed the pellet must have a density in excess of 1.41 g/mL.

Unlike most proteins, PrP has a high isoelectric point of approximately 9 for the full-length protein,[54] rendering the protein positively charged at neutral pH. This feature of PrP may be used advantageously because manipulation of electrostatic interactions can be achieved using anionic compounds such as PTA and detergents. Our work on the pH-dependent decomposition of polyoxometalates (Figure 3) suggests that PrPSc precipitations using PTA should be carried out on the lower end of the physiological pH range, closer to pH 7.1 rather than 7.4, in order to maintain the integrity of the PW11 ion, particularly when greater sample purity is desired.

While both PW11 and WO4 components of PTA contributed to the yield of PrP during precipitation, PW11 is primarily responsible for a reduction in coprecipitating GM1 lipids, as evidenced by the levels of ganglioside GM1 by Western blot (Figure 5). PW11 at concentrations >1 mM substantially reduced the level of coprecipitating lipids, while WO4 did not at equivalent concentrations. This lipid reduction might explain why precipitation with PW11 facilitated detection of PrP by ELISA more effectively than the WO4 component at concentrations between 0.25 and 6 mM (Figure 4). The lipids remaining in the WO4-incubated samples may mask epitopes on PrP and hinder its binding to the detection antibodies in ELISA. The difference in lipid levels following PW11 and WO4 precipitations may be attributed to the size and charge differences between the two tungstates, because the WO4 ion has a single heavy tungsten atom and a delocalized −2 charge while PW11 is a much larger, heavier anion with a −7 charge. These differences between PW11 and WO4 as well as our experiments with NaCl indicate that electrostatic interactions may be involved in promoting PrPSc precipitation by their disruption of interactions between PrPSc and buoyant anionic lipids like GM1. Direct association of dense tungstates with positively charged PrPSc may also contribute to the precipitation efficiency.

Reducing the amount of coprecipitating lipids such as GM1 may have implications for various areas of prion research, including biophysical studies of protein structure and diagnostic assay development. Lipids have been shown to interfere with protein detection assays, particularly those relying on the specificity of a protein–antibody interaction.[32,48] By reducing the lipid content of biological samples with typically low prion titers, high-throughput detection assays for cattle and human blood may be developed to reduce the risk of contracting variant Creutzfeldt–Jakob disease from meat products or blood transfusions.

By elucidating the mechanisms of PTA precipitation of PrPSc, we can improve our methods for sample preparation, which may play a crucial role in studies to probe the structure of PrPSc and to develop robust diagnostic tools. The principles governing the precipitation of PrPSc may also prove useful in studying other protein-misfolding diseases, in which the hydrophobic nature of protein aggregates promotes undesired binding of lipids to the aggregates.

Methods

Chemicals

NaPTA, sodium tungstate dihydrate, sodium phosphate monobasic, sodium carbonate monobasic, N-laurylsarcosine sodium salt (sarkosyl), sucrose, cholera toxin subunit-B peroxidase conjugate, β-mercaptoethanol, trizma hydrochloride, sodium chloride, tween-20, sodium hydroxide, nitric acid, and hydrochloric acid were purchased from Sigma-Aldrich and used without further purification. Deuterium oxide was obtained from Cambridge Isotopes and used as received. Bovine serum albumin (BSA), ABTS substrate, and 8 M GdnHCl were obtained from Thermo Scientific. SDS, 10× TGS running buffer, and 10× Ca2+/Mg2+-free PBS were obtained from Gibco and diluted to the appropriate strength with purified water (18.2 MΩ·cm resistivity). Instant nonfat dry milk was from SACO Foods. Amersham enhanced chemiluminscence (ECL) Western blotting detection reagents were from GE Healthcare. Recombinant mouse PrP(23–230),[55] HRP-conjugated anti-PrP Fab antibody P,[56] and anti-PrP Fab antibodies D18 and HRP-D13[50] were prepared as described.

Prion-Infected Brain Samples

FVB weanling mice (Charles River Laboratories) were inoculated intracerebrally with RML prions prepared from infected CD1 mice. Inoculated mice were sacrificed when symptoms of neurologic dysfunction presented in accordance with protocols approved by the UCSF Institutional Animal Care and Use Committee. The brains were collected and stored at −80 °C. All manipulations of mice, brain tissue, and purified prions were performed in a Biosafety Level 2 facility in accordance with established guidelines from the NIH.

PrP Fibrillization Assays

Brain homogenates (10% w/v) were prepared in Ca2+/Mg2+-free PBS and clarified by centrifugation at 500g for 5 min in a tabletop centrifuge. The resulting supernatant was separated into 1-mL aliquots and centrifuged at 100 000g for 1 h (Beckman, TLA55 at 48 000 rpm). The high-speed supernatant was discarded, and the pellets remaining in the centrifuge tubes were frozen on ice and stored at −80 °C until use.

For the fibrillization assays, brain homogenate samples were resuspended with 200 μL of Ca2+/Mg2+-free PBS with Complete Protease Inhibitor Cocktail (Roche Applied Science, Indianapolis, IN) and then pooled together. The pooled samples were supplemented with increasing concentrations of PTA (0–3.4% w/v), sarkosyl (0–3.5% w/v), or a combination of PTA (0–3.4% w/v) in the presence of 2% w/v sarkosyl. The samples were incubated overnight (16–18 h) at 37 °C under constant shaking (1200 rpm). After incubation, the samples were place on ice, and 50-μL aliquots were taken and analyzed by electron microscopy.

Electron Microscopy

Sample aliquots of 5 μL were adsorbed for 30 s onto Formvar/carbon-coated, 200-mesh copper grids (Ted Pella, Redding, CA) that were glow-discharged prior to use. The grids were then washed briefly with 0.1 and 0.01 M ammonium acetate buffer, pH 7.4, and stained with two 50 μL drops of freshly filtered 2% w/v ammonium molybdate[57] or dried without additional staining. The contrast provided by the bound PTA was sufficient to identify fibrillization products at low magnification.[15] After drying, the samples were viewed with a FEI Tecnai F20 electron microscope (FEI Company, Hillsborough, OR) at an acceleration voltage of 80 kV. Electron micrographs were recorded on a Gatan (Pleasanton, CA) UltraScan 4000 CCD camera.

Fibril Quantification

The quantification of the PrPSc fibrillization was performed essentially as described.[15] In brief, 10 electron micrographs were taken at random positions on the grid. Micrographs that contained obvious film defects, dirt particles, and other artifacts were discarded and replaced. All images were taken at a low magnification of 7575× (corresponding to 19.8 Å/pixel) and a higher-than-usual underfocus (5–10 μm) to facilitate the quantification. The micrographs were processed with Adobe Photoshop (San Jose, CA). The number of pixels (representing the area) for each visible fibril in the electron micrographs was determined. Individual fibrils and fibril clusters were selected by using the magic wand tool with a setting of 50–100; the Fresnel fringes caused by the strong underfocus made it easier to delineate each fibril/fibril cluster separately. The histogram menu provided the number of pixels for each particle. Fibrils were classified as such if the length was >40 pixels, equivalent to ∼78 nm, and the length/width ratio was at least 2:1. The amount of fibrils was averaged over all 10 micrographs. The stochastic distribution of the fibrils on the support film is reflected in the relatively large standard errors of the mean (SEM).

Sucrose Cushion Centrifugations

Brain homogenate was prepared from RML-infected FVB mice at 20% w/v in Ca2+/Mg2+-free PBS. The homogenate was centrifuged for 5 min at 500g at 4 °C, and the pellet was discarded to remove cell debris from the sample. The supernatant was then split into two aliquots, and sarkosyl was added to each aliquot to give a final concentration of 2% w/v sarkosyl. To one sample, 10% PTA solution, pH 7.2, was added (2% final w/v), while the other was left as a control without PTA. The samples were then incubated overnight at 37 °C, and the aliquots were each pipetted onto a two-step sucrose cushion containing 50% sucrose (8 mL) and 80% sucrose layers (2 mL). All sucrose solutions contain 0.5% sarkosyl, 50 mM sodium HEPES, and 1 mM sodium azide. The gradients were centrifuged at 134 000g (20 °C) for 16 h in a SW41 Ti swinging bucket rotor (Beckman-Coulter). Fifteen aliquots were taken from each gradient and labeled A–O from low to high density. For fractions A–F, 1 mL samples were collected; for fractions G–O, 500 μL collected. The pellets were resuspended in 1 mL of a 2% sarkosyl solution and labeled as the pellet fraction. All aliquots and gradient pellet samples were analyzed by Western blot using the P-HRP antibody and developed using ECL reagents.

31P Nuclear Magnetic Resonance

Samples were prepared at 10% w/v PTA in 90:10 H2O/D2O mixture and run on a Bruker Avance III NMR spectrometer equipped with a BBO broadband probe operating at 600.13 MHz for 1H and 242.93 MHz for 31P. Sample pH was adjusted with stepwise addition of NaOH and allowed to equilibrate for 1 h before spectra were collected. Additional spectra were collected in more complex buffer solutions containing PBS, sarkosyl, and brain homogenate (Supporting Information). While the phosphate concentration increased due to the use of PBS as the solvent, the NMR signals corresponding to PTA did not change in these complex buffer solutions.

Preparation of POM Stock Solutions

The 10% w/v (31.6 mM) PTA solutions were prepared by dissolution of NaPTA in deionized water (18.2 MΩ·cm resistivity); the pH of the solution was adjusted to pH 7.1–7.2 with NaOH. Na7[PW11O39] and Na2WO4·2H2O solutions of equivalent concentrations (31.6 mM) were prepared in deionized water, and the pH was adjusted by NaOH or glacial acetic acid to pH 7.1–7.2. The solutions were prepared fresh prior to each experiment.

Synthesis of Na7[PW11O39]·12H2O

Procedure was adapted from Brevard et al.[58] Sodium tungstate dihydrate (36.3 g, 0.110 mol) and anhydrous sodium phosphate monobasic (1.38 g, 0.010 mol) were dissolved in 75 mL of deionized water (18.2 MΩ·cm resistivity). The solution was heated to 85–90 °C with vigorous stirring, and concentrated nitric acid was added dropwise to reach pH 4.8. The solution volume was reduced to half by evaporation, and extraction with 3 × 50 mL of acetone afforded the POM in the aqueous layer as a white powder after evaporation of the water.

Preparation of Brain Homogenates for Tungstate Precipitations

Triplicate sets of brain tissue (7 brains/set) from RML-infected FVB mice were thawed and homogenized with Ca2+/Mg2+-free PBS to give a 10% w/v crude brain homogenate. Nuclei and debris were removed by centrifugation at 1500g for 5 min, and the supernatant was diluted 1:1 with a sarkosyl detergent solution to a final concentration of 5% brain homogenate in 2% sarkosyl by weight. Samples were stored at −80 °C in 500-μL aliquots.

Tungstate Precipitations from Brain Homogenate

Aliquots (500-μL) of 5% brain homogenate in 2% sarkosyl (w/v) were treated with varying concentrations of tungstates by addition of 0, 8, 16, 32, 64, 96, 128, 192, or 256 μL of the stock solutions at pH 7.1 and a corresponding amount of water to adjust the final volume of each sample to 1012 μL. The samples were incubated at 37 °C with shaking for 1 h and then centrifuged at 14 000g for 1 h at room temperature (RT) in a 5415D Eppendorf microcentrifuge equipped with the fixed-angle rotor F-45-24-11. The supernatants were promptly decanted, and the 1.5 mL Eppendorf tubes were left upside-down to facilitate removal of liquid residue from the pellet, which was subsequently removed by cotton swabs. The pellets were stored at −20 °C.

ELISA Plate Preparation

Thermo Immulon 2HB flat-bottom, 96-well microtiter plates were used for ELISA experiments. To each well, 100 μL of 4 μg/mL solution of D18 antibody in 0.1 M carbonate (pH 8.6) was added, and the plates were sealed and incubated with shaking at 4 °C overnight. The plates were washed 5 times with Tris-buffered saline with Tween-20 (TBST) and then blocked with 300 μL/well of 1% BSA in TBST for 1 h at RT. After aspiration of the blocking solution, the plates were sealed and stored at 4 °C until use.

Sandwich ELISA

Three independently prepared sets of POM-precipitated pellets were resuspended in 500 μL of 4 M GdnHCl, heated at 100 °C for 10 min, and cooled on ice for 5 min. The samples were diluted 40-fold with 1% BSA in TBST to prevent antibody denaturation, and 100 μL of each sample was loaded onto D18-coated plates in triplicate. Each plate also contained a ladder of recombinant mouse PrP(23–230) loaded in known quantities in duplicate or triplicate to establish a standard concentration curve with a four-parameter logistic regression.[59] The plates were incubated overnight at 4 °C and subsequently washed 3× with TBST. The plates were then incubated for 1 h with a 1 μg/mL D13-HRP antibody solution in 1% BSA in PBS, rinsed 3× with TBST, and then developed with ABTS HRP substrate. Plate absorbance was read at 405 nm, and the sample optical densities were correlated to the known calibration curve to determine the level of PrP in each sample. To convert PrP levels from nanograms per well to nanograms per milligram of brain tissue, the well concentrations were multiplied by a factor of 8. The mean level of PrP precipitated at each POM concentration was calculated by taking the average of all samples at the same POM concentration, and the standard error of the mean was determined by taking the standard deviation of the independent averages and dividing by the square-root of three. To calculate the enrichment factor (eq 1) from each POM concentration relative to controls (Table 1), the values were normalized to the amount of PrP detected when no POMs were used in the precipitation.where x̅ = average ng of PrP/mg of brain tissue for POM concentration i, x̅c = average ng of PrP/mg of brain tissue for control (i = 0 mM POM), s = standard error of the mean for POM concentration i, and sc = standard error of the mean for control (i = 0 mM POM).

Immunoblotting of POM-Precipitated Samples

POM-precipitated pellets were resuspended in 200 μL of SDS sample buffer, diluted 10-fold, heated to 100 °C for 10 min, and allowed to cool to RT. The samples were run on a BioRad Criterion 4–20% polyacrylamide gel, transferred to a PVDF membrane using the iBlot wet transfer system, blocked with a 5% milk solution in TBST, and incubated with 1 mg mL–1 D13-HRP antibody overnight at RT. To detect GM1, HRP-conjugated cholera toxin B subunit (Sigma-Aldrich) was added to the D13-HRP-containing milk solution the following day and incubated for 1 h at RT. The membrane was washed 3 times with TBST, incubated with ECL reagents for 1 min, and then developed on film.