|Home | About | Journals | Submit | Contact Us | Français|
Propagation and infectivity of prions in human prionopathies are likely associated with conversion of the mainly a-helical human prion protein, HuPrP, into an aggregated form with amyloid-like properties. Previous reports on efficient conversion of recombinant HuPrP have used mild to harsh denaturing conditions to generate amyloid fibrils in vitro. Herein we report on the in vitro conversion of four forms of truncated HuPrP (sequences 90–231 and 121–231 with and without an N-terminal hexa histidine tag) into amyloid-like fibrils within a few hours by using a protocol (phosphate buffered saline solutions at neutral pH with intense agitation) close to physiological conditions. The conversion process monitored by thioflavin T, ThT, revealed a three stage process with lag, growth and equilibrium phases. Seeding with preformed fibrils shortened the lag phase demonstrating the classic nucleated polymerization mechanism for the reaction. Interestingly, comparing thioflavin T kinetics with solubility and turbidity kinetics it was found that the protein initially formed non- thioflavionophilic, morphologically disordered aggregates that over time matured into amyloid fibrils. By transmission electron microscopy and by fluorescence microscopy of aggregates stained with luminescent conjugated polythiophenes (LCPs); we demonstrated that HuPrP undergoes a conformational conversion where spun and woven fibrils protruded from morphologically disordered aggregates. The initial aggregation functioned as a kinetic trap that decelerated nucleation into a fibrillation competent nucleus, but at the same time without aggregation there was no onset of amyloid fibril formation. The agitation, which was necessary for fibril formation to be induced, transiently exposes the protein to the air-water interface suggests a hitherto largely unexplored denaturing environment for prion conversion.
Human and animal diseases that result from aberrant folding and accumulation of aggregated proteins appear more and more common. In the classical amyloidoses resulting from the accumulation of around 25 different proteins (often specific for each disease) pathology appears closely linked to deposition of ordered aggregates in the form of amyloid fibrils.1 Amyloid fibrils are composed of a tightly packed core where the protein molecules are stacked in a cross beta-sheet conformation with parallel strands that stretch perpendicular to the fibril elongation axis.2 An amyloid fibril has the appearance of a one dimensional crystal with high internal symmetry. Seeding with preformed fibrils shorten the lag-phase and often accelerate the elongation phase.2,3 Hence the amyloid fibrillation process can be considered as a process resembling crystal growth and is described as a nucleation dependent polymerization reaction.
Prion diseases can be sporadic, inherited and infectious diseases. A pathogenic hallmark of the human prionopathies Kuru, Gerstmann-Sträussler-Scheinker disease (GSS) and variant Creutzfeldt-Jakob disease is the deposition of aggregates in the form of amyloid plaque from misfolded prion protein (HuPrP) in various regions of the brain depending on disease.4 The at-the-time heretic notion that prions are infectious particles solely composed of protein5 has gained momentum in the past several years. The initial finding that scrapie associated infections prions formed amyloid-like structures was shown some 25 years ago by Prusiner and Glenner.6 Following the discovery that prions are composed of a normally expressed protein, PrPC, the conformational conversion of PrPC into a misfolded state, PrPSc, has gained acceptance as the causative agent in all prion diseases.7 Likely the conformation of the aggregated PrPSc state dictates disease phenotype, rendering different prion “strains.”8 Hence aggregation and amyloidogenesis of PrP is strongly correlated with prion disease and is an intense research field.7
Normal correctly folded HuPrP is a 209 amino acid protein with a folded domain (residues 121–231) composed of three α-helices and a short double stranded β-sheet (Fig. 1C) and a flexible unstructured N-terminal domain (residues 23–120) comprising five octapeptide repeats.9 HuPrP is anchored to the cell surface via a GPI-anchor and the protein can be N-glycosylated at two different sites. In prion disease the accumulation of PrPSc renders the protein proteolytically stable versus proteinase K in vitro and the stable core fragment which is preserved in PrPSc is composed of residues 90–231, including the folded domain and a part of the dynamic region. There are several interesting findings on the aggregation and fibrillation of mammalian PrPs. The N-terminally truncated version of PrP, i.e., the core fragment of PrPSc, starting around position 89–90 has been subjected to a large number of studies. Acceleration of disease has been reported by injection of amyloid fibrils of a PrP fragment in animals expressing a mutated (P101L) truncated version of mouse PrP, PrP89–230,10 and more recently, albeit showing slow onset, even in vitro generated amyloid fibrils of murine PrP89–230 induced disease in mice expressing the corresponding sequence.11 In vitro studies of amyloid fibrillation and oligomer formation of PrP have so far been performed under conditions where mild to harsh conditions of denaturants of various sorts have been included. This has been performed on human, hamster and mouse PrP variants of different sequence lengths. In this work we show an unusual behavior of recombinant truncated human prion protein sequence 90–231, HuPrP90–231, during protein aggregation and fibrillation when performed under non-denaturing conditions close to physiological.
Four HuPrP variants: HisHuPrP90–231, HuPrP90–231, HisHuPrP121–231 and HuPrP121–231 (Fig. 1A) were purified as native monomers using size exclusion chromatography with a nearly physiological buffer (buffer F: 100 mM NaCl, 50 mM KCl, 50 mM phosphate, pH 7.3) at room temperature. The proteins were thereafter distributed into sealable 2 mL plastic vials containing 800 µL samples (12 µM of each protein dissolved in buffer F). The samples were firmly attached in a rack placed horizontally in a shaking incubator and were subjected to intense shaking (350 rpm) at 37°C (Fig. 1C). After 24 h the solution was turbid as deduced by eye. The aggregates were stained with Congo red and were examined using a polarization microscope. The HuPrP variants contained green birefringent protein aggregates under crossed polarizers, a tinctorial hallmark for the presence of amyloid fibrils (Fig. 1D). Fibrillar needle-like aggregates, 3–6 nm in diameter of individual filaments and fibrils respectively, were detected using negative stain transmission electron microscopy (Fig. 1E).
The rate of amyloid fibril formation was analyzed using thioflavin T, ThT fluorescence. A large number of identical samples were subjected to fibrillation conditions in buffer F under intense shaking (350 rpm) at 37°C (as above). Aliquots were withdrawn at different time points and were assayed for ThT fluorescence. All samples were ThT negative at the onset of the reaction, but after a few hours intense ThT fluorescence confirmed formation amyloid fibrils. All kinetic traces followed a lag-growth-equilibrium trajectory for each individual sample (Fig. 2A and B). The kinetic traces exhibited a sharp transition from the lag-phase to initiate the growth phase. The ThT fluorescence spectra after the lag phase showed the diagnostic ThT emission peak for amyloid fibrils centered around 480–490 nm (Fig. 2D). During the lag phase (<2 h) the fluorescence intensities were 3 fold higher than in the presence of native protein (c.f. 0 h) but still 20–30 fold lower than for fibrillated protein (Fig. 2D). The average ThT fluorescence intensity at the equilibrium phase (7 h) of HisHuPrP90–231 and HisHuPrP121–231 was essentially identical but was significantly higher for HuPrP90–231 and for HuPrP121–231 (Fig. 2E). The final ThT fluorescence likely reflected the efficiency of conversion of HuPrP to amyloid fibrils. This was verified by titration of the assay concentration (2 µM) ThT with increasing concentrations of preformed HisHuPrP90–231 fibrils, where a linear correlation between ThT fluorescence and the concentration of fibrils within the dynamic range was observed (Fig. 2F). The augmented ThT fluorescence of the non-his tagged HuPrP90–231 and HuPrP121–231 compared to HisHuPrP90–231 and HisHuPrP121–231 (Fig. 2E) could either reflect more efficient conversion, limited accessibility of ThT binding sites, a conformational difference between the variants or poly-his quenching of ThT. It is likely that the ThT fluorescence intensity difference is due to the quenching effect because the effect was most pronounced for the shorter construct. The 90–120 sequence also contains two intrinsic histidines and four lysine residues, also well known fluorescence quenchers.
Fitting of the kinetic fibrillation data were not satisfactory to a sigmoidal function, but a more reliable fit was obtained employing a cascade function (see materials and methods), hence, we empirically selected this function for the data to obtain a lag-phase for each individual sample. The shortest lag phases were obtained for HisHuPrP121–231 (0.6 ± 0.2 h) and HuPrP121–231 (0.9 ± 0.3 h), followed by HuPrP90–231 (1.2 ± 0.2 h) and HisHuPrP90–231, (2.3 ± 0.4 h). The growth rate (see materials and methods) once fibrillation had initiated was within error for three different variants as shown in the correlation diagram of lag phase versus growth rate in Figure 2G (HisHuPrP121–231: 0.42 ± 0.08 h−1; HuPrP90–231: 0.51 ± 0.08 h−1 and HisHuPrP90–231: 0.54 ± 0.08 h−1). The shortest variant HuPrP121–231 showed the fastest growth rate (1.0 ± 0.24 h−1). As a further control experiment on the experimental conditions necessary for fibril formation, samples of HisHuPrP90–231 were incubated under stagnant conditions in otherwise identical circumstances. Without shaking there was no elevated ThT fluorescence (Fig. 2C).
The longest lag phase in the spontaneous fibrillation kinetics was obtained for HisHuPrP90–231, hence this variant was selected for seeding experiments. Samples were run using sonicated preformed fibrils as seeds at two different concentrations (1 and 5% of the total amount of protein). As expected from nucleated growth of amyloid fibrils, the seeded reactions showed shorter lag phases compared to the unseeded protein [0.9 ± 0.2 h (1%) and 0.8 ± 0.1 h (5%)], however the growth rate was unaffected (Fig. 2C and G). Furthermore the seeded samples (with identical sequences) showed higher final ThT fluorescence signal than the unseeded samples, suggesting a somewhat more efficient conversion into the amyloid fibrillar state (Fig. 2E).
The HuPrP fibrillation kinetics (ThT fluorescence) showed a sharp transition from the lag-phase which indicated a cooperative assembly of protein molecules into the fibrillar structure. As expected at the equilibrium phase the loss of free monomer halts the progression. In the majority of samples the ThT signal dropped following the process >10 h, and up to 24 h (often more significantly than in Fig. 2D), which could indicate either re-solvation of fibrils or flocculation that reduced the fluorescence signal. We therefore precipitated the material stained with ThT and subjected these to fluorescence microscopy (Fig. 3A). Evidently the ThT positive material was highly thioflavinophilic even at longer incubation times (c.f. 5 h and 24 h in Fig. 3A), which was expected from electron microscopy and Congo red birefringence obtained on 24 h protein aggregates (Fig. 1D and E). Hence flocculation was the most likely reason for ThT fluorescence decease over time.
During sampling of the fibrillation reaction we noted that samples were turbid already after a few minutes of fibrillation conditions (agitation at 37°C), long before any ThT fluorescence appeared. To measure the aggregation rate we subjected identical samples of HisHuPrP90–231 to turbidity measurements in parallel with ThT fluorescence. Within 15 minutes a plateau of turbid protein aggregates was reached in each sample (Fig. 3C). Comparing the trace of turbidity with the onset of ThT fluorescence revealed dramatically different kinetics (Fig. 3C). We verified that the turbidity mirrored the loss of soluble protein by centrifugation (14,000 rpm for 10 minutes) at different time points and analyzed the supernatant by UV-spectroscopy (Fig. 3C), showing that 80% of the protein was insoluble already after 15 minutes of agitation.
The kinetic traces of ThT fluorescence showed some quite variable reaction profiles of identical samples (e.g., Fig. 2A). As the initial aggregation during fibrillation in buffer F was fast this results in an initial condensation of HuPrP molecules which induced intermolecular interactions but could also become a kinetic trap into disordered aggregates which limit fibrillation. We therefore subjected HisHuPrP90–231 to fibrillation at half the protein concentration (6 µM, 0.1 mg/l). Interestingly, although the variations of lag phase and growth rates increased, the average lag-phase (2.6 ± 0.5 h), was within error comparable to the lag-phase obtained at 12 µM (0.2 mg/ml) but the fibril growth rate of was significantly faster (0.82 ± 0.17 h−1), (Fig. 2G). The increased growth rate is likely to reflect more efficient fragmentation during fibril formation at the low protein concentration.
At different time points aliquots were withdrawn and analyzed by transmission electron microscopy to assess the morphology of the formed HisHuPrP90–231 aggregates (Fig. 3B). From the micrographs it was evident that early formed aggregates <2 h was composed of heavily clustered material which over time (>2 h) transformed into fibrillar material (Fig. 3B). A proteinase K (PK) digestion procedure followed by SDS-PAGE analysis of the aggregates formed after 15 min and after 7 h of fibrillation showed that both types of aggregates were partially resistant to PK as deduced by the remaining full length protein (at 18 kD). In comparison the native protein was cleaved into shorter fragments. Comparing the PK resistance of 15 min aggregates and 7 h fibrils did not reveal any significant differences in resistance (Fig. 4A). Longer incubation times with PK decreased the intensity of the band at 18 kD but with a concomitant increase of protein trapped in the well of the gel (not shown). To obtain more information on the transformation process and the conformation of the aggregated protein we analyzed the aggregates using the conformation sensitive luminescent conjugated polythiophenes (LCPs) PTAA8,12,13 and tPTT.14,15 We thereafter analyzed the LCP stained aggregates using microspectroscopy. The material pelleted after 15 min of agitation showed a green shifted PTAA fluorescence compared to HisHuPrP90–231 agitated for 24 h (Fig. 4B) as revealed by plotting spectral ratios in a correlation diagram (Fig. 4B). These data demonstrate that the LCP binds in a more planar and stacked conformation to 24 h fibrils compared to the early formed aggregates of HisHuPrP90–231.8,12 Even though the 24 h fibrillar aggregates were red shifted compared to the 15 min aggregates, both types of aggregates displayed a variety of fluorescence spectral distribution indicating conformational heterogeneity within each population. From the ThT fluorescence data it appeared evident that conversion into the amyloid form was not complete based on increased ThT fluorescence in the seeded samples compared to the unseeded samples and the presence of thick aggregates in the electron micrographs (24 h micrograph, Fig. 3B). We therefore dialyzed 24 h HisHuPrP90–231 fibrils against acetate buffer (50 mM pH 4.0) and stained the fibrils with the LCP tPTT followed by spectral analysis by microspectroscopy. The pseudo color image (“filter view”) revealed the presence of at least two coexisting conformations of the LCP when bound to the fibrils, colored green and red (Fig. 4C). Spectral analysis of these images verify these dramatically different spectral species fluorescing with a peak at 575 nm (green) indicating solitary twisted LCP molecules and at 606 nm (red) are planar stacked LCP molecules.14,15 Interestingly both conformations were also frequently proximal in space (Fig. 4C, highlighted with arrows). This corroborated that aggregates/fibrils of HisHuPrP90–231 were composed of a mixture of conformations.
It was also possible to generate amyloid fibrils of HisHuPrP90–231 under otherwise identical conditions in the absence of physiological concentrations of salt. The process was then somewhat slower (data not shown). In analysis of aged fibrils, formed in the absence of physiological salt (four day incubation), by transmission electron microscopy we identified border regions where there were thin layers of morphologically disordered aggregates which were transmissible to the electron beam. It appeared that rigid fibrils protruded from these regions and formed a fibrillar network (Fig. 4D). This together with the kinetic analysis of aggregate morphologies (Fig. 3B) indicated that the fibrils were spun and woven from the morphologically disordered aggregates.
A huge number of diseases are associated with protein misfolding. In many cases these are loss of function diseases where compromised protein stability renders misfolding of e.g., an important enzyme.16 In other diseases the misfolded protein is the culprit and accumulates in diseased organs and impairs cellular function. Amyloid formation is a common process to a range of degenerative diseases where approximately 25 different proteins accumulate in human diseases, each specific for a given disease.1 A mysterious subtype of these diseases is the prion diseases, which also frequently display deposition of PrP in aggregated form and likely are transmissible through a nucleated polymerization mechanism. Amyloidogenesis of folded globular proteins in vitro often require conditions where the protein is subjected to partial unfolding. This can be achieved either through introduction of destabilizing mutations or through environmental stress e.g., altered pH, or temperature or added chemical denaturants. Studies of PrP amyloidogenesis in vitro have been subject to intense investigations, and as expected denaturing conditions have always been employed to render misfolding of PrP. The closest to native conditions reported so far is from the Riesner group working at low SDS concentrations at neutral pH, but the conversion required seeding with PrPSc from hamster brain to be efficient.17 A protocol developed by the Baskakov group include initiation of fibrillation from either the urea unfolded state to final concentration of urea 4 M,18 or from the guanidine hydrochloride, GuHCl (6 M) unfolded state into fibril formation conditions at 1 M GuHCl, 2–3 M urea at varying acidic pH.19 The Wüthrich group have been using detergents forming lipid bicelles at pH 5.0 to convert native PrP to the misfolded beta-sheet form of the protein.20 The Surewicz group used 1 M GuHCl21 and recently mildly acidic pH 4.0 in the absence of denaturants22 for conversion of HuPrP90–231 into amyloid fibrils.
Aggregation of PrP during prion disease in humans most likely does not involve severely denaturing conditions although there are destabilizing conditions within cells including lipid bilayers or the acidic compartments of the cellular lysozome which both are possible conversion sites. In humans suffering from Kuru, GSS or vCJD large prion deposits are found extracellularly. The brain interstitial fluid has a pH of 7.3,23 and is certainly a non-denaturing milieu. Neutral pH has been reported as conditions of poor solubility of PrP24 and has precluded structure determination of the unstructured N-terminal domain containing the octapeptide repeats. It has been reported that this domain becomes structured during pH values close to neutral.24 The NMR structure of HuPrP121–231 has been determined at pH 7.0 and showed minor changes compared to the structure at pH 4.5,9 the structure of HuPrP90–231 has only been solved at acidic pH 4.5.9 Interestingly, the crystal structure of HuPrP90–231 solved at pH 8.0 showed a dimeric domain swapped structure,25 strongly suggesting intermolecular affinity of the protein under native conditions.
In this work we investigated if it was possible to induce amyloid fibril formation of recombinant truncated HuPrP (HuPrP90–231 and HuPrP121–231) as well as truncated HuPrP carrying an N-terminal his-tag (HisHuPrP90–231 and HisHuPrP121–231) starting from native folded protein in the absence of denaturants under physiological salt, pH and temperature. Intense shaking of the folded protein under these gentle conditions in partially filled sealed plastic tubes induced irreversible conversion of native recombinant HuPrP variants into aggregates composed of amyloid fibrils within hours. The agitation proved to be crucial for the conversion process as identical samples under quiescent conditions showed completely ThT negative fluorescence (Fig. 2C). Agitation confers continuous exposure to the air-water interface known to be destabilizing conditions for proteins,26 suggesting that also our conditions are in fact transiently denaturing, when the protein resides in that phase. Nevertheless the amyloid formation process appears to be specific as at least four other well folded proteins (GroEL, GroES, BiP and transthyretin) tested in the same setup did not form ThT positive aggregates (unpublished results). The HuPrP aggregates were positive using thioflavin T (ThT) fluorescence and showed Congo red birefringence under polarized light. Fibrillar morphology was also verified by transmission electron microscopy. These methods show that the initial native largely helical HuPrP90–231 had misfolded into the aggregated state of cross-β-sheet structure of amyloid. The amyloid fibrillation kinetics followed a trajectory with a lag phase, a growth phase and an equilibrium phase. Interestingly under these conditions the fastest protein to form ThT positive fibrils in terms of lag-phases was the HisHuPrP121–231 variant containing the folded domain and an N-terminal hexa histidine-tag, followed by the HuPrP121–231 variant which was slightly slower but essentially within error. In comparison, in terms of lag-phases, HuPrP90–231 converted faster than HisHuPrP90–231, which nevertheless spontaneously fibrillated rapidly with an average lag phase of 2.3 h. The HuPrP121–231 variant also revealed an increased fibrillation growth rate compared to the other variants of the study. The common denominator between these variants is the folded domain (121–231) (Fig. 1B) which hence is most likely to form the core of the amyloid fibrils during fibrillation during near physiological conditions. Surprisingly data presented by the Surewicz group reported that the rate of HuPrP90–231 D178N, a mutant known to be more amyloidogenic than HuPrP90–231 showed a lag-phase of >48 h (50 mM acetate pH 4.0, 37°C, 0.5–1.0 mg/ml),22 which is >20 fold slower than the kinetics shown in our work (Fig. 2A), unfortunately the kinetics of the wild-type sequence was not presented in the paper, but was likely even slower. Hence, it appears that the neutral pH protocol used by us where the protein initially aggregates prior to fibril formation significantly accelerated the conversion.
As deduced from site directed spin labeling and electron paramagnetic resonance measurements it was recently shown by the same group that the structure of the HuPrP90–231 amyloid fibril core comprising residues 160–220, formed parallel β-strands comprising the cross β-sheet both under conditions where HuPrP90–231 was folded in the absence of denaturants (pH 4.0) as compared to amyloid fibrils made from HuPrP90–231 generated under denaturing conditions (2 M GuHCl, pH 7.0).22 The most evident difference between fibrils formed under denaturing conditions (pH 7.0) and in the absence of denaturants (at pH 4.0) resided in the N-terminal sequence 103–144 suggesting that the difference between different amyloid fibrils reside from interaction within the sequence 90–160.22 We noted a massive amount of lateral stacking of the fibrils which appeared as a natural consequence of the fibrils being spun from the initial formed aggregates. Limited proteinase K digestion have shown to induce a similar type of lateral assembly as the networks detected by us (Figs. 1E and 4D) of murine PrP amyloid fibrils at neutral pH, while preserving the amyloid core.19 We also found that the lag phase of HisHuPrP90–231 appeared to be due to nucleation because seeding with preformed fibrils significantly shortened the lag phase (Fig. 2C), without affecting the fibril growth rate. Interestingly lowering the protein concentration to half did not significantly affect the lag phase but increased the rate of fibril growth. Taken together this suggests that the initial aggregation operates as a kinetic trap that decelerates initial nucleation into a fibrillation competent nucleus, but at the same time without aggregation there was no onset of amyloid fibril formation. The morphologically disordered aggregates can undergo a conformational conversion process into rigid well organized amyloid fibrils, resulting in a distinct phase transition, which was demonstrated in the kinetic experiments (Fig. 3B). Using partial PK digestion, both types of aggregates were more resistant than the native protein, but comparing early aggregates and fibrillar aggregates did not reveal a significant difference in PK resistance (Fig. 4A). Nevertheless, from microspectroscopy of LCP stained aggregates at least two conformations (disordered and rigid) of aggregates were identified. This method was previously employed to enable mapping of amyloid beta, A-beta, conformational states in a mouse model of Alzheimer disease.14 The mechanism suggested for PrP conversion in this paper closely resemble the suggested conformational conversion of A-beta in morphologically diverse senile plaque, where we hypothesized that rigid fibrils protrude from a disordered plaque center.14 In contrast, murine PrP aggregates identified using LCP staining of brains of terminally ill prion infected mice showed specific fluorescence profiles for each distinct prion strain.8,27 Similar to the findings in this paper on recombinant human PrP in vitro, the prion infected mice showed PrP deposits positive for PTAA albeit with different fluorescence spectral properties and all showed PK resistant PrP. In addition, in mice, non-thioflavinophilic PrP aggregates were also identified with PTAA.8 Hence, PTAA has been shown to give distinct emission profiles for non-thioflavinophilic PK-resistant PrP deposits. On the other hand, it has also been shown that PTAA can show identical fluorescence spectra of a serially transmitted prion strain although the PK resistance only appeared upon maturation of the prion strain.27 Taken together it seems that PK resistance does not simply correlate with elevated ThT fluorescence or to a specific PTAA fluorescence profile.
In conclusion we have demonstrated that amyloid fibril formation of truncated HuPrP (sequence 90–231) under near physiological conditions in vitro that are initiated through intense agitation, that expose the protein to the air-water interface, proceeds through initial aggregation that produce non-thioflavinophilic morphologically disordered aggregates. Amyloid fibrils are thereafter nucleated from these initial aggregates. The fibrillation process can be accelerated if seeded with preformed amyloid fibrils. The folded domain (sequence 121–231) comprises the core region in the amyloid fibril whereas unstructured N-terminal extensions appear to slow down fibrillation. Our data show that morphologically disordered aggregated HuPrP is conformationally plastic and is rather rapidly convertible until the amyloid fibrillar state is reached. It is tempting to hypothesize that the aggregation-fibrillation route of conformational conversion also occurs in vivo.
Ultra Pure GuHCl was obtained from ICN Biomedicals and the concentration of all GuHCl solutions was determined by refractometrical measurements.28 Isopropyl-β,D-thiolgalactopyranoside (IPTG) was obtained from Saveen, thioflavin T (ThT) was obtained from Sigma, congo red was obtained from Kodak.
Plasmids (pREST A) containing HuPrP(129M) genes of desired lengths (coding for protein sequences HisHuPrP90–231 and HisHuPrP121–231, [Fig. 1A]) were obtained from Prof. K. Wüthrich.29 E. coli BL21/DE3 was transformed with the pREST A plasmids. The HuPrP variant proteins were purified using Ni-NTA agarose followed by gel filtration as previously described in detail30 where Buffer F (50 mM phosphate, 100 mM NaCl, 50 mM KCl, pH 7.3) containing 0.01% sodium azide was used for running buffer in the final step.
Truncation of the N-terminal his-tag was performed using the biotinylated thrombin cleavage kit from Novagen, and was used as described by the manufacturers, followed by gel filtration purification. Elution of monomeric protein was assessed through the final gel-filtration step, using buffer F as running buffer (Superdex 75, GE-Healthcare) and purity was confirmed using coomassie stained SDS-PAGE gels. The protein concentration was determined by measurement of absorbance at 280 nm, A280, using the extinction coefficients ϵ280 = 16240 M−1cm−1 for HisHuPrP121–231 and HuPrP121–231; and ϵ280 = 21640 M−1cm−1 for HisHuPrP90–231 and HuPrP90–231.29
Samples (800 µl) were prepared containing HisHuPrP90–231, HuPrP90–231 and HisHuPrP121–231, (Fig. 1A) (12 µM) in buffer F that were incubated in 2 ml sealable plastic vials (Sarstedt) in a shaking incubator (Innova 4230, New Brunswick), at 37°C with orbital rotation at 350 rpm. After 24 h the samples were supplemented with a final concentration of 5 µM congo red (from a 1 mM stock in EtOH). Congo red stained HuPrP aggregates were allowed to sediment by gravitation over night at 4°C. The supernatant was colorless in the incubated HuPrP fibril samples since the protein aggregates had precipitated and displayed a pinkish hue. No precipitation of congo red was observed in the absence of protein. Precipitated aggregates were applied on glass slides and were visualized using a Nikon H550S optical microscope equipped with double polarizers and a digital camera. Images were obtained in bright field mode with open and crossed polarizers.
For TEM analysis 5 µl aliquots at different time points of samples subjected to fibrillation conditions were applied to carbon coated copper mesh grids (Ted Pella, inc.,) for 2 min followed by 5 µl of 25% electron microscopy grade glutaraldehyde (Sigma) and were blotted dry with a filter paper. The samples were counter stained with 2% (w/v) uranyl acetate dissolved in dH2O for 30 s. Micrographs were collected using a Jeol 1230 transmission electron microscope equipped with a CCD camera operating at 100 kV and with a Philips CM200 transmission electron microscope operating at 120 kV.
Aliquots (32 µl) HisHuPrP90–231 at different time points were mixed with 168 µl of assay buffer containing 1.5 µM the luminescent conjugated polythiophene (LCP) polythiophene acetic acid (PTAA)31 (to a final ratio PrP:PTAA of 1:8) dissolved in 50 mM sodium carbonate, pH 10. The LCP stained aggregates were pelleted at 10,000 g for 30 min and were placed on glass slides where the samples were covered Dako mounting solution for fluorescence (Dako, Glostrup, Danmark) and were thereafter imaged with an epifluorescence microscope (Zeiss Axiovert inverted microscope A200 Mot) equipped with a SpectraCube® (Optical head) module, using the 470/20 nm bandpass filter (LP515). The analysis of images was achieved with the standard software (SpectraView™) for spectral resolution of 500–800 nm emission spectra. The fluorescence spectra of aggregates were analyzed by collecting five regions of interest for each image. The procedure was repeated for four images from each time point. The procedure was repeated using two independent protein preparations and staining sessions. Fully formed amyloid fibrils of HisHuPrP90–231 (24 h) were dialyzed against 50 mM acetic acid pH 4 and were stained with a final concentration of 2 µM of the LCP (poly [5,5″]tertiophene-[2S,3R]″2-Amino-3-[2-(3″-[2-([1R,2S]-2-amino-2-carboxy-1-methyl-ethoxy)-ethyl]-[2,2′;5′,2″]terthiophen-3-yl)-ethoxy]-butyric acid), tPTT.14 The fibrils were allowed to self sediment at 4°C and were thereafter imaged with an epifluorescence microscope (Zeiss Axiovert A200 Mot inverted microscope) equipped with a SpectraCube® (Optical head) module (ASI, Israel), through a 470/40 nm bandpass filter (LP515).
Preformed amyloid fibrils (7 h) of HisHuPrP90–231 in buffer F were mixed in different ratios with a final concentration of 2 µM ThT. Full spectra were obtained as above to assay the fluorescence intensity at different fibril concentrations. References in the absence of HuPrP and with native HuPrP were also obtained and showed 50–100 fold lower fluorescence than fully formed fibrils.
For kinetic analysis of the fibril formation reaction, a number of identical samples (800 µl) were prepared containing HisHuPrP90–231, HuPrP90–231, HisHuPrP121–231 and HuPrP121–231 (Fig. 1A) (12 µM) in buffer F that were incubated in 2 ml sealable plastic vials (Sarstedt) in a shaking incubator (Innova 4230, New Brunswick), at 37°C with orbital rotation at 350 rpm. Samples (32 µl) were withdrawn at different time points and were placed in empty 96 well plates (Corning non-treated, black) placed in a refrigerator at 4°C. The following day the samples were diluted with 168 µl assay buffer (Buffer F, supplemented with ThT) resulting in a final concentration of ThT and HuPrP of 2 µM. Following incubation at room temperature for 1 h the samples were analyzed in a Tecan Saphire2 plate reader (Tecan cooperation) and full spectra were obtained (465 nm–650 nm) following excitation at 440 nm using top down fluorescence. Seeding experiments were performed using preformed fibrils of HisHuPrP90–231 which were sonicated in a water bath using a Branson sonicator horn for 30 min.
Fitting of the kinetics of the HuPrP data to regular sigmoidal growth kinetics were not satisfactory. A much more reliable fit was obtained employing a cascade function from the software TableCurve 2D (Jandel Scientific). Hence, we empirically selected this function for our data to obtain a fitted lag-phase for each individual sample.
The cascade function was used to model the two step reaction:
It is mathematically described by the following function:
Where I(t) is the ThT fluorescence at time t (in hours), Imax is the maximum ThT fluorescence, k1 and k2 are rate constants (which are interchangeable). The reaction begins at time tL, i.e., the end of the lag phase. Fitting of the raw data to the cascade function (1), with TableCurve 2D, reliably provided the lag phase, where the sharp transition was initiated. However, the rate constants from the cascade fit frequently showed large deviations from seemingly indistinguishable initial curve slopes due to variations at the end of the growth phase. This stems from the interchangeability of the rate constants in the equation which makes them interdependent and hence proved unreliable. The fibril growth rate (h−1) was therefore calculated using the initial slope of the cascade transition up to half the ThT fluorescence intensity (normalized for each individual trajectory), according to:
Where I0 is the initial and Imax is the maximum ThT fluorescence, is the time at half the ThT maximum fluorescence and tL, is end of the lag phase.
Samples of HisHuPrP90–231 subjected to fibril formation conditions as above were also assayed at different time points for turbidity by transfer to a quartz cuvette. Light scattering (optical density) was assessed used to determine the formation of large aggregates using a Libra 22 (Biochrom) absorbance spectrophotometer set at 330 nm. After 7 h the samples were assayed with ThT to confirm the presence of amyloid (as described above).
Partial proteinase K (PK)-digestion was performed by additon of PK to prefibrillized protein samples. For this purpose HisHuPrP90–231 was prepared as previously described and aliquots were subjected to agitation at 37°C for 0 min, 15 min and 7 h before stagnant cooling at 4°C. PK was added to a final concentration of 2 µg/ml (dilution 1:100) and the samples were incubated without agitation at room temperature. Aliquots were withdrawn before PK addition (0 min) and 15 min and 60 min post PK addition, and were immediately mixed with SDS loading buffer (reaching a final concentration of 2% SDS and 100 mM DTT) and followed by boiling for 5 min. The samples were run on an 18% Tris-HCl gel (Biorad) and protein bands were visualized using BioSafe Coomassie (BioRad).
We thank Andreas Éslund and Peter Konradsson for synthesis of tPTT, we also thank Mikael Lindgren for helpful discussions on fluorescence spectroscopy and Adriano Aguzzi for discussions of prion protein aggregation and disease. The pRSET A plasmids containing the HuPrP variants were a kind gift from Kurt Wüthrich.
This work was supported by the Swedish Research Council (P.H.), Knut and Alice Wallenberg Foundation (P.H. and P.N.). The Swedish Foundation for Strategic Research (P.H. and P.N.). A generous gift from Astrid and Georg Olsson is gratefully acknowledged. P.H. is a Swedish Royal Academy of Science Research Fellows sponsored by a grant from the Knut and Alice Wallenberg Foundation.
Previously published online: www.landesbioscience.com/journals/prion/article/10112