The Effect of Disease-associated Mutations on the Folding Pathway of Human Prion Protein

Propagation of transmissible spongiform encephalopathies is believed to involve the conversion of cellular prion protein, PrP(C), into a misfolded oligomeric form, PrP(Sc). An important step toward understanding the mechanism of this conversion is to elucidate the folding pathway(s) of the prion protein. We reported recently (Apetri, A. C., and Surewicz, W. K. (2002) J. Biol. Chem. 277, 44589-44592) that the folding of wild-type prion protein can best be described by a three-state sequential model involving a partially folded intermediate. Here we have performed kinetic stopped-flow studies for a number of recombinant prion protein variants carrying mutations associated with familial forms of prion disease. Analysis of kinetic data clearly demonstrates the presence of partially structured intermediates on the refolding pathway of each PrP variant studied. In each case, the partially folded state is at least one order of magnitude more populated than the fully unfolded state. The present study also reveals that, for the majority of PrP variants tested, mutations linked to familial prion diseases result in a pronounced increase in the thermodynamic stability, and thus the population, of the folding intermediate. These data strongly suggest that partially structured intermediates of PrP may play a crucial role in prion protein conversion, serving as direct precursors of the pathogenic PrP(Sc) isoform.

characterized by an oligomeric β-sheet rich structure, partial resistance to proteinase K digestion and a pronounced tendency to aggregate into insoluble plaques (1-4, 7, 16-19). In some cases, these insoluble aggregates have characteristics of amyloid fibrils. Given the conformational differences between the normal and pathogenic prion protein isoforms, transmissible spongiform encephalopathies are often classified as diseases of protein misfolding.
One of the key arguments in support of the protein only hypothesis is the evidence linking familial prion diseases with mutations in the gene coding for human prion protein (1)(2)(3)20). Over 20 mutations in this gene have been shown to segregate with inherited forms of Creutzfeldt-Jakob disease, Gerstmann-Straussler-Scheinker disease or fatal familial insomnia.
Since the pathogenic conversion process in individuals carrying these mutations appears to develop spontaneously, the mutant proteins provide an invaluable model for studying molecular mechanisms of the PrP C →PrP Sc conversion in vitro. An important step towards understanding this mechanism is to dissect the folding pathway of the prion protein and identify the monomeric precursor of the oligomeric PrP Sc species. In a recent kinetic stopped-flow study, we demonstrated that the wild-type human prion protein folds by a three-state mechanism (excitation wavelength of 296 nm). The unfolding curves were analyzed using a two-state transition model as described previously (25). In the control experiments, it was verified that an incubation time of less than 1 sec is sufficient for the system to reach equilibrium [the unfolding/refolding of the prion protein is very fast, occurring on the millisecond time scale (21,26)], and that the unfolding and refolding reactions are fully reversible. Urea used in this study was deionized by treatment with a mixture of anion exchange

RESULTS
Our recent kinetic stopped-flow experiments revealed that human prion protein fragment 90-231 (huPrP90-23) refolds from urea according to a three-state transition model with an intermediate that accumulates at low denaturant concentrations (21). Here, we have extended the folding studies to a number of PrP variants containing amino acid substitutions associated with familial prion diseases. The 90-231 fragment of prion protein is especially suitable for such studies since, in addition to containing the folded domain, it encompasses the entire proteinase K resistant sequence of PrP Sc , contains all point mutations that are known to segregate with the familial forms of prion disease, and is sufficient for the propagation of the disease. The specific mutations considered in the present study, along with their location within the structure of PrP C , are shown in Fig. 1. All these mutations were introduced on the background of an engineered single Trp variant Y218W/W99F of huPrP90-231. The latter protein was previously identified as especially useful for stopped-flow experiments since, unlike Trp99, the fluorescence of Trp218 is very sensitive to protein conformation. Furthermore, the Y218W substitution does not significantly perturb the structure or thermodynamic stability of PrP (21).
Equilibrium Unfolding Studies -Prior to kinetic stopped-flow experiments, the effect of individual disease-associated point mutations on the global thermodynamic stability of Y218W/W99F huPrP90-231 was probed by equilibrium unfolding in urea. Essentially identical curves were obtained when equilibrium unfolding was monitored by circular dichroism spectroscopy (ellipticity at 222 nm) or Trp218 fluorescence. Figure 2 shows representative equilibrium unfolding curves for the wild type huPrP90-231 and the F198S variant. The free energies of unfolding, ∆G UN O,eq , obtained from the best fits of equilibrium unfolding data for individual mutant proteins are shown in Table I. These data show that while some of the disease-associated mutations produce a pronounced decrease in the global stability of Y218W/W99F huPrP90-231, the effect of others (e.g. P102L, E200K) is essentially negligible.
Importantly, the destabilization induced by familial mutations found here for Y218W/W99F huPrP90-231 corresponds closely to the values reported in similar studies using unmodified mouse (28) or human prion proteins (29). This indicates that the introduction of  (21,26). For all PrP variants studied, both the refolding and unfolding kinetic curves at each denaturant concentration could be fit by a single exponential function, yielding the apparent rate constants. Figure 3 shows representative kinetic traces for the refolding and unfolding of the wild type huPrP90-231 and the V210I variant. These data were used to construct the "chevron plots", in which the logarithm of the rate constant is plotted as a function of urea concentration. Consistent with our previous report (21), the refolding branch of this plot for the "wild-type" protein 2 sharply deviates from linearity at low urea concentrations (Fig. 4 A). Similar inflections in the refolding sections of the chevron plots were found for all huPrP90-231 mutants studied (Fig. 4). The unfolding (high urea concentration) branches showed no well-defined curvature. However, for most huPrP90-231 variants, the urea concentration range available for unfolding measurements was too narrow for any conclusions to be drawn.
Downward curvatures ("rollovers") in the refolding branches of chevron plots have been reported for a number of other proteins. These effects are usually attributed to accumulation of partially folded intermediates that form during the dead time of the instrument and become increasingly populated under stabilizing conditions (i.e. at low denaturant concentrations) (30,31). Recent studies suggest, however, that deviations from linearity in chevron plots could also be caused by other factors (32)(33)(34). Therefore, we performed a series of control experiments to test alternative interpretations of curvatures in chevron plots for huPrP90-231 variants.
The most common experimental problem that could lead to nonlinearities in chevron plots is transient aggregation of protein during the refolding reaction (32). In a previous study, we showed that aggregation does not occur upon refolding of the wild-type huPrP90-231 (21).   Table II). Since in the absence of the denaturant N is always much more populated than either U or I, these ratios provide a good approximation of the population of the unfolded and intermediate states relative to total protein concentration.
Inspection of the thermodynamic data shows that, for all proteins studied, the partially structured intermediate is much more stable as compared with the fully unfolded state (Table I). (Table II) interactions (50). For the wild-type PrP, the concentration of these partially folded monomeric species is very low. Therefore, except for extremely rare sporadic cases, these species may be recruited into the aggregated state -and converted to intermolecular β-sheet structure -only in the presence of externally provided PrP Sc seeds. In contrast, for mutant proteins the concentration of partially structured intermediates may become sufficient to initiate the aggregation process even in the absence of preexisting seeds, leading to de novo formation of the pathogenic PrP Sc isoform. Folding intermediates have also been implicated as crucial monomeric precursors of fibrils formed by classical amyloidogenic proteins such as transthyretin or human lysozyme variants (51)(52)(53). In a recent study with the amyloidogenic lysozyme variant D67H, it was estimated that approximately one molecule in 1,500 is in the partially folded state (53). Interestingly, this number is of the same order of magnitude as the population of the intermediate state found here for the majority of disease-associated prion protein variants.

When translated into the [U]/[N] and [I]/[N] ratios
While our model based on mutation-induced stabilization of the partially folded state is applicable to a majority of inherited prion diseases, this model is clearly not universal. Data of Table II indicate that at least for two PrP variants -P102L and E200K -the population of an intermediate is similar to that for the wild-type prion protein, even though the intermediate state is always much more populated than the fully unfolded state. Clearly, in these cases additional factor(s) must be involved to facilitate a spontaneous PrP C →PrP Sc conversion. For the E200K variant, these factors may include an alteration of surface electrostatic potential, as observed in a recent NMR structural study (54). Mutation-dependent changes in surface properties could   Thermodynamic and kinetic parameters for the folding of huPrP variants associated with inherited prion diseases ∆G UN 0,eq represents the free energy obtained from the best fit of the equilibrium unfolding data to a two-state model (25). The remaining parameters were obtained from the best fit of the kinetic data from Fig. 4 to a three state sequential model (31)