Convergent Replication of Mouse Synthetic Prion Strains

Abstract

Prion diseases are neurodegenerative disorders characterized by the aberrant folding of endogenous proteins into self-propagating pathogenic conformers. Prion disease can be initiated in animal models by inoculation with amyloid fibrils formed from bacterially derived recombinant prion protein. The synthetic prions that accumulated in infected organisms are structurally distinct from the amyloid preparations used to initiate their formation and change conformationally on repeated passage. To investigate the nature of synthetic prion transformation, we infected mice with a conformationally diverse set of amyloids and serially passaged the resulting prion strains. At each passage, we monitored changes in the biochemical and biological properties of the adapting strain. The physicochemical properties of each synthetic prion strain gradually changed on serial propagation until attaining a common adapted state with shared physicochemical characteristics. These results indicate that synthetic prions can assume multiple intermediate conformations before converging into one conformation optimized for in vivo propagation.

See related Commentary on page 623.

In prion diseases, for example, Creutzfeldt-Jakob disease in humans, scrapie in sheep, and bovine spongiform encephalopathy in cattle, an aberrantly folded conformer of the prion protein (PrP) propagates by catalyzing a posttranslational conversion reaction, using cellular PrP (PrP^C) as substrate.^1,2 This conversion reaction transforms endogenous PrP^C to the pathogenic conformer PrP^Sc.^3–5 Although self-replication of protein conformations was previously thought to be a unique feature of PrP prion diseases, in recent experiments, aggregates of proteins that cause other neurodegenerative diseases, for example, Aβ,^6–8 α-synuclein,^9–11 tau,^12,13 and huntingtin,¹⁴ stimulated the formation of pathogenic conformations in vivo. Thus, it is likely that elucidating the mechanisms of PrP^Sc prion propagation will provide important insights into the pathogenic properties of a broad range of neurodegenerative disorders.

An important advance in prion biology was the creation of infectious synthetic prion strains formed exclusively from bacterially derived recombinant (rec) PrP.^15–18 The ability to create de novo synthetic strains has provided indisputable evidence for the prion hypothesis and has facilitated structural studies with the objective of elucidating the structural basis of protein-based infectivity. However, an enigmatic feature of synthetic prion strains is that the conformations that ultimately accumulate in infected organisms are structurally distinct from recombinant amyloids used to initiate their formation.^19–21 In contrast, specific biochemical features of the amyloid inocula, such as their conformational stabilities, correlate with the prion strains induced in vivo.¹⁶ Thus, the mechanisms by which amyloid preparations are transformed into transmissible prion strains in vivo remain under investigation.^22–24

In an earlier investigation into the transformation of prion strains, we studied the mouse synthetic prion strain MoSP1,²⁵ which was generated by inoculating transgenic (Tg) mice expressing truncated mouse PrP(Δ23–88), which were denoted as Tg9949 mice,²⁶ with recPrP(Δ23–88) refolded into β-rich amyloid fibrils.¹⁵ After >500 days, the inoculated Tg9949 mice amassed protease-resistant infectious prions (rPrP^Sc).¹⁵ The accumulated MoSP1 strain had two biochemical features that differentiated it from naturally occurring prion strains such as the mouse-adapted Rocky Mountain Laboratory (RML; Golden, CO) scrapie strain. First, MoSP1 PrP^Sc had a high conformational stability, requiring >4 mol/L guanidine hydrochloride (GdnHCl) to unfold the prion aggregate and render it susceptible to proteolysis.²⁷ Second, the unglycosylated protease-resistant core had a molecular weight of ∼19 kDa.²⁵ By comparison, RML requires ∼1.5 mol/L GdnHCl to unfold and has a protease-resistant core of ∼21 kDa. We showed that the difference in molecular weights of the protease-resistant cores of MoSP1 and RML was due to conformational differences at their N-termini. When MoSP1 was continually passaged in mice, its physicochemical properties transformed in three ways: i) the unglycosylated protease-resistant core shifted from a molecular weight of ∼19 kDa to ∼21 kDa; ii) MoSP1 became conformationally less stable; and iii) the incubation period of the strain decreased to ∼150 days. We were able to recapitulate these transformations by propagating MoSP1 in cell culture.²⁵ We hypothesized that the mechanism of this transformation involved rare conformational conversion events followed by competitive selection among the resulting pool of conformers.

After our previous study, it remained uncertain whether the observed MoSP1 transition constituted a strain-specific transformation or a common pathway for the in vivo adaptation of all mouse synthetic prion strains. To address this, we created a conformationally diverse set of amyloids by refolding recPrP under various denaturing conditions.¹⁶ These preparations were inoculated intracerebrally into Tg mice overexpressing full-length PrP, denoted as Tg4053 mice.²⁸ After the initially infected mice developed prion disease, we serially passaged the resulting PrP^Sc strains (MoSP5, MoSP6, MoSP7, and MoSP9) and monitored changes in their biological and biochemical characteristics including incubation period, rPrP^Sc banding patterns, conformational stability, and ability to seed amyloid formation in vitro. On initial infection, the synthetic prion strains accumulated as a diverse set of conformations. However, on repeated passage, all four synthetic prion strains acquired the same set of physicochemical characteristics including short incubation periods, low conformational stabilities, 21-kDa banding patterns for unglycosylated rPrP^Sc, and diminished abilities to seed amyloid formation in vitro. In addition, we tested the infectivity of MoSP7 in three different cell lines and found an increased ability to induce the cellular formation of rPrP^Sc on passage. Together, these results show that synthetic prions can assume different intermediate conformations before achieving a conformation that is optimized for in vivo propagation.

Materials and Methods

Ethics Statement

All animal experiments were performed according to The Guide for the Care and Use of Laboratory Animals (National Research Council, 1996). All operations and procedures were approved by the Institutional Animal Care and Use Committee at the University of California San Francisco.

Production of Amyloid Fibers

The expression and purification of truncated recMoPrP(89–230) and full-length recMoPrP(23–230) and the formation of amyloid fibers have been previously described.^15,29,30 In brief, lyophilized purified protein was dissolved in 10 mol/L urea at 10 mg/mL. Fibers were formed in buffers (Table 1), with the addition of 30 μmol/L thioflavin T (ThT). Protein concentrations ranged from 0.2 to 1.0 mg/mL. Solution (200 μL per well) was added to 96-well plates and continuously shaken at 37°C in a plate reader. ThT fluorescence was detected at 442 nm excitation and 485 nm emission. Fibers were dialyzed against PBS to remove traces of urea and other solution components before inoculation.

Table 1.

Conditions Used for Amyloid Preparation

Amyloid	PrP sequence	Urea concentration (mol/L)	Acetate buffer		GdnHCl1/2 (mol/L)
			pH	NaCl concentration (mol/L)
MoSP5	23−230	4	5	0.4	4.2
MoSP6	89−230	4	5	0.4	3.5
MoSP7	89−230	0.2	5	0.2	3.1
MoSP9	89−230	0.5	5	0.2	2.9

Mouse Bioassays

The creation and characterization of Tg4053 mice have been previously described.³¹ The concentration of recMoPrP in the inocula was ∼0.5 mg/mL. Serial passage was conducted by inoculation with brain homogenates in sterile PBS without calcium or magnesium. Brain homogenates were prepared by repeated extrusion through syringe needles of successively smaller size, from 18 to 22 gauge. All work was performed in laminar flow hoods to prevent cross-contamination. Mice of either sex, aged 7 to 10 weeks, were inoculated intracerebrally with 30 μL amyloid fibrils or 1% brain homogenate for passaging. Inoculation was performed using a 27-gauge disposable hypodermic needle inserted into the right parietal lobe. After inoculation, mice were examined daily for neurologic dysfunction. Standard diagnostic criteria were used to identify animals exhibiting signs of prion disease.^28,32 The indicated mice were sacrificed, and their brains were removed for biochemical analysis.

Cell Culture

The N2a cell line was purchased from ATCC (Manassas, VA). Cath.a-differentiated (CAD) and murine fibroblast cloned cell line L929 (LD9) were a gift from Charles Weissmann. Cloning was performed by limiting dilution, as previously described.³³ Cells were infected via exposure to the indicated brain homogenates, as previously described.³³ N2a and CAD cells were maintained at 37°C in 10 mL Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum (FBS), 1% penicillin-streptomycin, and 1% l-glutamine (GlutaMAX; Invitrogen Corp., Carlsbad, CA). LD9 cells were maintained in minimal essential medium with 10% FBS, 1% penicillin-streptomycin, and 1% l-glutamine. Media were refreshed every 2 days. Cells were propagated in 100-mm plates and allowed to grow to 95% confluence before dissociation with 1 mL enzyme-free cell-dissociation buffer. Cells were then replated at 10% confluence for further propagation. To collect cell lysates, cells were rinsed three times with PBS (×10 mL each) and lysed with 1 mL cold lysis buffer [10 mmol/L Tris HCl (pH 8.0), 150 mmol/L NaCl, 0.5% Nonidet P-40, and 0.5% sodium deoxychloate]. Lysates were centrifuged for 3 minutes at 10,000 × g to remove cell debris, and the total protein concentration was measured in the supernatant using the bicinchoninic acid assay (Pierce Biotechnology, Inc., Rockford, IL). Aliquots containing 500 μg total protein were titrated by adding lysis buffer to achieve a final protein concentration of 1 mg/mL and stored at −20°C until further analysis.

Western Blot Analysis, Conformational-Stability Assays, and Densitometry

Nuclei and debris were removed from brain homogenates and cell lysates via centrifugation at 1000 × g for 10 minutes. Cleared extracts were adjusted to 1 mg/mL protein in 100 mmol/L NaCl, 1 mmol/L EDTA, 2% sarkosyl, and 50 mmol/L Tris HCl (pH 7. 5). To achieve a protein/enzyme ratio of 50:1, 20 μg/mL proteinase K (PK; Boehringer Mannheim Corp., a division of Roche, Indianapolis, IN) was added to 0.5 mL adjusted homogenate. This relative concentration of PK and time of digestion have previously been shown to result in complete digestion of PrP^C and steady-state levels of protease-resistant PrP^Sc.³³ After incubation at 37°C for 1 hour, proteolytic digestion was terminated via addition of 8 μL 0.5 mol/L phenylmethylsulfonyl fluoride in absolute ethanol. Both PK-digested and undigested samples were prepared for 12% SDS–PAGE by mixing equal volumes of adjusted homogenate and 2× sample buffer. After electrophoresis, Western blot analysis was performed as previously described. Membranes were blocked with 5% nonfat milk protein in calcium- and magnesium-free PBS plus 0.1% Tween 20 (PBST) for 1 hour at room temperature. Blocked membranes were incubated with primary PrP-specific recombinant Fab human-mouse-D18 at 1 μg/mL in PBST for 1 hour at 4°C. After incubation with primary Fab, membranes were washed 3 times for 10 minutes in PBST, incubated with horseradish peroxidase–labeled anti-human Fab secondary antibody (ICN Pharmaceuticals, Inc., Biomedical Division, Aurora, OH), diluted 1:5000 in PBST for 25 minutes at room temperature, and washed again 3 times for 10 minutes in PBST. After chemiluminescent development using an ECL reagent (Amersham Biosciences Corp., Piscataway, NJ) for 1 to 5 minutes, blots were sealed in plastic covers and exposed to ECL Hypermax film (Amersham).

The conformational-stability assay has been previously described.³⁴ In brief, 50 μL 10% brain homogenate or 1 mg/mL cell lysate was mixed with 50 μL GdnHCl, varying from 0 to 6 mol/L, and incubated for 1 hour at room temperature. For GdnHCl concentrations >4 mol/L, less volume of extract was used. Before PK digestion, all samples were diluted in lysis buffer to obtain GdnHCl concentrations of 0.4 mol/L. Densitometry on the appropriate bands was performed using a CCD camera (FluorChem 8800; Alpha Innotech Corp., San Leandro, CA). Measurements were normalized with respect to the highest intensity band in the denaturation curve. The half-maximal concentration of GdnHCl (GdnHCl_1/2) values were obtained by least-square fitting to the following sigmoidal equation: Y = 1/[1 + (X/^GdnHCl1/2)ˆ^HC], where Y is the measured intensities, X is the GdnHCl concentrations, and HC is the Hill coefficient.

In Vitro Amyloid Assay

The amyloid formation assay was based on the real-time quaking-induced conversion assay as described by Wilham et al.³⁵ Ten percent brain homogenates were obtained as described (see Western Blot Analysis, Conformational-Stability Assays, and Densitometry). To 90 μL brain homogenate, 10 μL 10× clearance buffer [10% Triton X-100 (Sigma-Aldrich Corp., St. Louis, MO), 1.5 mol/L NaCl, and 40 mmol/L EDTA] was added, and the solution was centrifuged for 30 seconds at 1500 × g on a tabletop centrifuge. The supernatant was collected and frozen at 4°C until further use. The cleared homogenate was diluted in PBS containing 0.1% SDS. The extent of dilution ranged from 50× to 300× based on the intensity of the rPrP^Sc band on the Western blot. RecMoPrP(89–230) to a final concentration of 20 μg/mL was added to the reaction buffer, which consisted of 130 mmol/L NaCl, 10 mmol/L sodium phosphate (pH 7.4), 10 μmol/L EDTA, and 10 μmol/L ThT. The reaction buffer plus protein, in 100-μL aliquots, were added to each well of a 96-well microtiter plate, and 1 μL diluted brain homogenate was added to each well. The plates were shaken at 42°C for 48 hours, and readings were taken every 2 minutes at 442 nm excitation and 485 nm emission. For each seed, six different wells were averaged to produce resulting traces plotted. The “lag time” was calculated as the time by which the fluorescence surpassed 3× SD of the initial baseline.

Neuropathologic Analysis

Brains were fixed in 10% buffered formalin and embedded in paraffin. Staining with H&E was performed on sections 8 μm thick. Peroxidase immunohistochemistry with antibodies to glial fibrillary acidic protein was used to evaluate the degree of reactive astrocytic gliosis. Immunohistochemistry of PrP^Sc was performed via the hydrated autoclaving method using the PrP-specific human-mouse P and R2 recombinant monoclonal antibody fragments (Fab).

Results

Characterization of Amyloid Fibrils and Serial Passage in Mice

Creation of PrP amyloids used to generate the synthetic strains analyzed in the present study (MoSP5, MoSP6, MoSP7, and MoSP9) has been previously described.¹⁶ PrP amyloids were created in buffers with varying pH, denaturant, and salt concentrations, and the polymerization reaction was performed via continuous shaking at 37°C (Table 1). The resulting fibrils had a range of conformational stabilities, as determined by the concentration of GdnHCl required to unmask antibody epitopes that are differentially exposed in PrP^Sc and PrP^C conformations (Table 1). The fibril preparations also had varying morphologic features, as observed by electron microscopy.¹⁶

The PrP preparations were inoculated intracerebrally into Tg4053 mice overexpressing full-length mouse (Mo) PrP at 4 to 8 times the levels found in wild-type FVB mice.³¹ Mice infected with the amyloid preparations developed neurologic disease at 450 to 800 days after inoculation¹⁶ (Figure 1A). The onset of disease was characterized by the presence of vacuoles and astrocytic gliosis in the brain (Figure 1B).¹⁶ Immunohistochemical analysis of infected brains indicated the presence of rPrP^{Sc 16} (Figure 1B). Conversely, Tg4053 mice inoculated with PBS or bovine serum albumin did not develop neurologic disease during their lifespan and did not accumulate infectious prions in their brain.¹⁷

Figure 1.

A: Kaplan-Meier survival curves of Tg4053 mice after inoculation with MoSP5, MoSP6, MoSP7, and MoSP9. Colored curves indicate first (P1, red), second (P2, green), third (P3, blue), and fourth (P4, gray) passages. Three different MoSP7 isolates were passaged: predominantly type 1 (MoSP7b), type 2 (MoSP7a), and mixed type 1/2 (MoSP7c). Open circles indicate terminal mice that were biochemically analyzed, and solid circles signify mice that were biochemically analyzed and subsequently used as the inoculum for the next round of passage. B: Neuropathologic profile for passage of MoSP7c. H&E staining of the hippocampus (top panel) shows vacuolation. Immunohistochemistry with anti-PrP antibodies (middle panel) indicates the presence of granular PrP^Sc deposits. Staining with antibodies to GFAP (bottom panel) shows a mild level of astrocytic gliosis. Scale bar = 50 μm (all micrographs).

Brain homogenates from infected mice were serially passaged in Tg4053 mice for three or four rounds (Figure 1A). For MoSP7, extracts from multiple initially infected brains were used as inocula for additional rounds of injection. On serial passaging, the incubation period for all strains decreased stochastically. By the third or fourth passage, the incubation periods of the synthetic prions strains ranged from 80 to 300 days.

Neuropathologic analysis was consistent with prion disease for all passages¹⁷ and indicated the presence of punctate PrP^Sc deposits, varying degrees of vacuolation in the neuropil, and different levels of reactive astrocytic gliosis. No clear trends were evident between the degree of these neuropathologic changes and passage number (data not shown).

Biochemical Characterization

Brain extracts from selected infected mice (Figure 1A) were digested with PK, and rPrP^Sc bands were detected on Western blots using the D18 antibody³⁶ (Figure 2). The epitope of D18 lies between residues 132 and 157 of MoPrP. Detected rPrP^Sc had variable banding patterns. These patterns could be divided into three categories, with the unglycosylated band having apparent molecular weights of 21 kDa, 19 kDa, or a mixture of the two types; we refer to these banding patterns as types 1, 2, and 1/2, respectively, analogous to the nomenclature used for Creutzfeldt-Jakob disease strains in humans.³⁷ It is likely that each set of bands represents a spectrum of protein conformations, given the limited resolution of the Western blots. For comparison, RML and first-passage MoSP1,²⁵ which have type 1 and type 1/2 patterns, respectively, are shown in the immunoblot. On first passage, MoSP5 and MoSP6 appeared as type 2 strains, MoSP9 appeared as type 1/2, and MoSP7 showed all three strain types (Figure 2). By second passage, MoSP5, MoSP6, and MoSP9 all transformed to a type 1 strain. On second passage of MoSP7 type 2 (MoSP7a), a transformation to type 1 occurred, similar to that observed for MoSP5, MoSP6, and MoSP9. However, when predominantly type 1 (MoSP7b) and type 1/2 (MoSP7c) strains were injected for a second passage, some mice showed type 2 PrP^Sc. After subsequent passages, a type 1 strain emerged (fourth passage of MoSP7b and third passage of MoSP7c). Thus, despite traversing different intermediate states, the four strains ultimately converged on a type 1 banding pattern.

Figure 2.

Western blot analysis of selected Tg4053 mice (circles in Figure 1) infected with synthetic mouse prions at different passages. Brain homogenates were subjected to immunoblotting after limited digestion with PK. Blots were probed with D18 antibody. RML and MoSP1 at passage 1²⁵ are shown as controls. Arrows denote the unglycosylated protease-resistant bands of PrP^Sc migrating to 19 kDa (type 2) and 21 kDa (type 1). For passage of MoSP7c, two different mice are shown from second passage, indicated as P2-1 and P2-2. Molecular masses of protein standards are given in kilodaltons.

We next measured the conformational stability of passaged synthetic prion strains. The GdnHCl_1/2 of PrP^Sc that accumulated on initial passage ranged from 2.8 to 3.4 mol/L (Figure 3). On continuous passage, the conformational stabilities generally decreased for all mouse synthetic prions, reaching ∼1.6 mol/L GdnHCl by third or fourth passage.

Figure 3.

Conformational stability of synthetic prion strains at different passages. A: As an example of a typical conformational stability assay, Western blot of PK-digested brain homogenate after incubation with increasing concentrations of GdnHCl (left panel) and the corresponding densitometry analysis (right panel) of MoSP7 at second and third passages are shown. Blots were probed with D18 antibody. Molecular masses of protein standards are given in kDa. B: Conformational stabilities of all analyzed passaged synthetic mouse prion strains. GdnHCl_1/2 values were measured for each brain homogenate as shown in A.

Quantitative changes in incubation periods, banding patterns, and conformational stabilities of the mouse synthetic prions on passage in Tg4053 mice are graphed in Figure 4. For MoSP5, MoSP6, MoSP7a, and MoSP9 (Figure 4), the data indicate that serial passage results in i) shortening of the incubation period, ii) a switch from type 2 or 1/2 to a type 1 banding pattern, and iii) a decrease in conformational stability. These physicochemical changes were also observed for MoSP1 in our earlier study. However, for MoSP7 that emerged as predominantly type 1 and type 1/2 on first passage, subsequent passages showed more complex results. Some Tg4053 mice inoculated with MoSP7b or MoSP7c for a second passage accumulated type 2 PrP^Sc with relatively high conformational stability (GdnHCl_1/2 >3.5 mol/L) and relatively short incubation periods (110 days). We consider this combination of properties to be unusual because in our previous studies, short incubation periods correlated with low conformational stabilities and type 1 banding patterns.^16,38 On subsequent passages, MoSP7b and MoSP7c showed physicochemical characteristics similar to the other mouse synthetic prions: incubation periods of ∼100 days, type 1 PrP^Sc, and GdnHCl_1/2 values of ∼1.6 mol/L. Together, these data argue that synthetic mouse prion strains attain a similar set of physicochemical characteristics on continuous propagation.

Figure 4.

Graph showing changes in the incubation periods (abscissa), banding patterns, and conformational stabilities (ordinate) of MoSP5, MoSP6, MoSP7, and MoSP9 passaged in Tg4053 mice. Open, solid, and half-solid circles represent type 1, 2, and 1/2 strains, respectively, for the indicated synthetic prion inoculum and passage. Sold gray lines represent the range of incubation periods observed for the bioassay from which the analyzed mouse was selected.

Amyloid Seeding Ability of Passaged Synthetic Mouse Prion Strains

We next assessed the ability of passaged synthetic mouse prion strains to seed the formation of amyloid fibrils in vitro using the real-time quaking-induced conversion assay.³⁵ For each synthetic strain, brain homogenates collected at each passage were mixed with purified recPrP and incubated at 42°C with continuous shaking. Fibril formation was quantified by monitoring the change in fluorescence intensity of the dye ThT (Figure 5). The concentrations of brain homogenates added to incubation buffers were normalized on the basis of the intensities of rPrP^Sc bands on the Western blots. All other buffer parameters were kept constant among different kinetic experiments. For MoSP5, MoSP6, MoSP7, and MoSP9, subsequent passage in Tg4053 mice decreased the ability of the prions to seed amyloid formation, as demonstrated by increased lag times for ThT kinetic traces.

Figure 5.

In vitro amyloid seeding ability of passaged synthetic mouse prions. A: Brain homogenates of Tg4053 mice after inoculation with MoSP5, MoSP6, MoSP9, and MoSP7a, as indicated, at different passages (indicated in different colors) were used to seed in vitro amyloid formation using the real-time quaking-induced conversion assay. The concentration of brain homogenate used as seed was normalized with respect to the total PrP^Sc present in each extract. Homogenate from an uninfected brain (black line) was used as a negative control. The kinetic traces show ThT fluorescence averaged for six wells in a microtiter plate and normalized with respect to the final signal. B: Lag time, or hours elapsed before ThT fluorescence attained 3× SD of the initial baseline, for each averaged kinetic trace.

Infectivity of MoSP7 in Cultured Cells

We next assessed the ability of MoSP7, collected after different passages, to infect a panel of three murine cell lines: N2a, CAD, and LD9 (Figure 6). This panel of cell lines had been previously used to differentiate a number of prion strains.³⁹ Each cell line was freshly cloned by limiting dilution, and up to 10 clones were expanded for subsequent analysis. For each cell line, we selected two clones that had the highest level of PrP^C expression, as determined by using Western blot analysis (data not shown). These clones were exposed to brain homogenates from P1, P2, and P4 of MoSP7 lineage that was initiated by the accumulation of a predominantly type 1 strain (MoSP7b). Cells were also exposed to RML as a positive control. Infected cell lines were passaged for eight rounds. Subsequently, cells were lysed, and the resulting extract was analyzed for the presence of rPrP^Sc via Western blot analysis after limited PK digestion. The results indicated that MoSP7 preparations after P1 and P2 were able to infect LD9 cells only, whereas MoSP7 after P4 was able to infect all three cell lines. Thus, the host specificity of MoSP7 changed during the course of its propagation in mice. The ability of MoSP7 to infect a wider array of cell lines after P4 compared with the earlier passages may be indicative of an increased level of infectivity and/or a faster rate of replication attained during serial passages in mice.

Figure 6.

Infection of CAD, LD9, and N2a cells with MoSP7b at different passages. Two different subclones of each cell line were incubated with brain homogenates of Tg4053 mice infected with MoSP7b at different passages. Western blots of PK-digested lysates indicate the presence of protease-resistant PrP^Sc. For each indicated subclone, the two gel lanes contain lysates taken after 8 and 9 weekly passages. Cells infected with RML are shown as positive control. Blots were probed using D18. Molecular masses of protein standards are given in kilodaltons.

Discussion

In previous studies, prion disease could be induced in mice by inoculation with amyloid preparations composed of recPrP.^15–18 The prion conformers that accumulated in the brains of infected mice had some physicochemical properties that differed from the amyloid used for inoculation, as evidenced by increases in resistance to proteolysis, changes in seeding specificity, and structural transformations characterized by X-ray diffraction^19,20 as well as atomic force and electron microscopy.²¹ Remarkably, amyloids created under different in vitro conditions resulted in the accumulation of conformationally distinct prion strains.¹⁶ In the present study, for a diverse set of mouse synthetic prions, we have shown that on continuous in vivo propagation, the physiochemical properties of the strains changed until a consensus state was reached. For the four synthetic mouse prion strains analyzed, the strain changes that occurred on repeated passage in vivo led to conformations that had shorter incubation periods, lower conformational stabilities, type 1 banding patterns, and decreased amyloid seeding capability. For MoSP7, these changes also resulted in the ability to infect a broader range of cell lines.

Our findings indicate that shorter incubation periods in vivo do not correlate with amyloid seeding ability in vitro. For a prion, the conformational properties that are optimal for robust in vivo propagation are likely multifaceted. For example, a robustly infectious prion strain must not only propagate nascent PrP^Sc molecules but also be able to survive cellular degradation processes and to horizontally infect neighboring cells. It is likely that conformations optimal for such in vivo requirements are not selectively amplified in in vitro amyloid fibril formation assays. Thus, when synthetic prion strains undergo in vivo changes, the ability of the prion to propagate in vivo may be optimized at the expense of its amyloid seeding kinetics. This observation is consistent with amyloid accumulation as a nonobligatory feature of prion diseases.⁴⁰

What is the molecular mechanism of in vivo synthetic prion transformation? On the basis of our previous analysis of MoSP1 in cell culture, we proposed that the transformation of synthetic prions in vivo occurs through a process of competitive selection.²⁵ Prions typically have high replication specificities and breed true on serial passage.^41–43 However, numerous studies have indicated that conformational changes can occur in replicating prions.^23,44–49 Mechanistically, strain transformations may occur either by rare spontaneous structural changes to the PrP aggregate or by infrequent lack of fidelity during the replication process.^20,24 These low-frequency conversion events result in formation of prion conformations that structurally differ from their template. If a mutated conformation is more proficient at replicating in its biological host, it is selectively amplified. In a heterogeneous prion population, the presence of optimized strains will cause the depletion of unoptimized strains by outcompeting them for limited resources required for replication. As a mixture, strains are competing for the same pool of PrP^C substrate and auxiliary factors required for de novo PrP^Sc formation. Thus, faster replicating strains will deny slower strains of resources required for their propagation. This process provides a mechanism by which an oligomeric conformer can stochastically survey a conformational landscape and transform into a structure that is most proficient at replicating in the biological host (Figure 7). Thus, in this landscape view of prion transformation, nonoptimized prion states can be initiated by a wide array of PrP^Sc structures and gradually transform into strains with short incubation periods.

Figure 7.

Schema of a putative conformational landscape of a self-propagating prion. Prions are capable of surveying a structural landscape through rare stochastic conformational mutations. Initiating from an amyloid fibril with a relatively slow rate of in vivo propagation (A), prions can traverse different intermediate states (B) and transform to a conformation with a faster rate of replication (C).

Our data not only show that a diverse set of synthetic prions can gradually acquire the same set of physicochemical properties on passage but also confirm that parallel folding pathways involving different intermediate states can be traversed before reaching an optimized state. It is noteworthy that some of the intermediate MoSP7 conformations characterized in the present study have a combination of physicochemical properties not previously observed for natural prion strains (type 2 banding patterns, high conformational stability, and relatively short incubation periods), which suggests that this conformation represents a metastable strain that cannot propagate with fidelity in vivo.

Not all prion strains are transient, and most natural strains can propagate with a high degree of conformational fidelity over many rounds of passage. Why do not all strains gradually transform to a single optimized state on perpetual passage? As has been noted previously by Weissmann et al,²⁴ the conformational landscape of prions likely encompasses a large number of local energy minima that can indefinitely trap structures not at the global energy minimum. Thus, localization of the prion conformation on the energy landscape, ie, whether a prion conformation is at a local energy minimum, will determine whether the strain is biologically stable or transient. The prion energy landscape is influenced by the sequence of PrP, and, thus, the process of strain selection is expected to be host-dependent. Future studies of heterologous synthetic prion propagation in different transgenic mice might provide insight into the relationship between the host PrP sequence and the prion adaptation process.

Although it is now apparent that prions are dynamic pathogens capable of altering their conformation to adapt to new hosts and environments, the mechanistic details of this evolutionary process remain under investigation.²⁴ The results of the present study demonstrate that diverse prions can probe a structural landscape and traverse different intermediate states to transform to an optimized conformation. That prions are capable of convergent evolution suggests that an array of distinct conformers can initiate the formation of a single biologically stable prion strain.