Cellular prion protein conformation and function

Abstract

In the otherwise highly conserved NMR structures of cellular prion proteins (PrP^C) from different mammals, species variations in a surface epitope that includes a loop linking a β-strand, β2, with a helix, α2, are associated with NMR manifestations of a dynamic equilibrium between locally different conformations. Here, it is shown that this local dynamic conformational polymorphism in mouse PrP^C is eliminated through exchange of Tyr169 by Ala or Gly, but is preserved after exchange of Tyr 169 with Phe. NMR structure determinations of designed variants of mouse PrP(121–231) at 20 °C and of wild-type mPrP(121–231) at 37 °C together with analysis of exchange effects on NMR signals then resulted in the identification of the two limiting structures involved in this local conformational exchange in wild-type mouse PrP^C, and showed that the two exchanging structures present characteristically different solvent-exposed epitopes near the β2–α2 loop. The structural data presented in this paper provided a platform for currently ongoing, rationally designed experiments with transgenic laboratory animals for renewed attempts to unravel the so far elusive physiological function of the cellular prion protein.

The prion protein (PrP) in its cellular isoform (PrP^C), which is found in healthy mammalian organisms, is among the most extensively studied proteins. Nonetheless, the physiological function of PrP^C and its role in the molecular pathways leading to degeneration of the brain in patients suffering from transmissible spongiform encephalopathies still remain elusive (for example, refs. 1–3).

The NMR solution structure of the recombinant mouse prion protein (4, 5) has the same molecular architecture as the protein part of PrP^C present in healthy organisms (6). It contains a flexibly disordered N-terminal tail of residues 23–124, a globular domain of residues 125–228 with three α-helices and a short two-stranded antiparallel β-sheet, and a short C-terminal tail of residues 229–231 (5) [see Schätzl et al. (7) for the numeration of PrPs from different species]. The globular domain in the NMR structures of recombinant PrPs from different mammalian species is highly similar, except for local structure variability in a surface epitope formed in part by a loop of residues 166–172 that connects the strand β2 with the helix α2 (4, 8–16). This polypeptide segment and its immediate spatial environment also show numerous amino acid exchanges among mammals (17–19), which contrasts with the overall high sequence conservation in the globular domain of mammalian PrPs (7). The β2–α2 loop is structurally disordered in the NMR structures of numerous mammalian prion proteins determined at 20 °C (4, 9–11), but is well defined in the NMR structures of PrP^C from elk, bank vole, horse, wallaby, and rabbit determined under similar conditions (12–16). This structural disorder is associated with the absence of resonances from residues 167–171 from the β2–α2 loop and residue 175 in the 2D [¹⁵N,¹H]-HSQC spectra suggesting the presence of a local conformational exchange process.

The present paper elucidates a structural basis for rationalizing the behavior of the surface epitope formed by the β2–α2 loop and the helix α3 in wild-type mouse PrP^C, based on NMR structure determinations of two newly designed variants of mPrP(121–231) at 20 °C and of wild-type mPrP(121–231) at 37 °C. It is thus shown that the β2–α2 loop structure in mPrP^C manifests exchange between just two locally different conformations, and both limiting structures involved in this conformational exchange are identified by combining structural data on wild-type and variant mPrP(121–231) [Protein Data Bank, www.pdb.org (PDB ID codes 2L39, 2L40 and 2L1D)] with analysis of exchange effects on loop residue NMR signals. The insight gained on the solution conformations available to mouse PrP^C indicate future avenues for defining the physiological functions of the prion protein.

Results and Discussion

NMR Spectra of mPrP[Y169G](121–231) and mPrP[Y169A](121–231) Show No Evidence of Exchange Line Broadening.

In search of a structural basis for possible local conformational exchange on the chemical-shift timescale involving the β2–α2 loop, we designed variants of mPrP(121–231) containing single amino acid exchanges in this region. We thus found that 2D [¹⁵N,¹H]-HSQC NMR spectra of mPrP[Y169G](121–231) recorded at a ¹H frequency of 750 MHz in 5° intervals over the temperature range 5–40 °C showed no effects of conformational exchange on the NMR line shapes for residues in the β2–α2 loop (Fig. 1A) or residue 175. The same observation was made for mPrP[Y169A](121–231), whereas the line shapes of signals of the loop residues show pronounced temperature variation in mPrP[Y169F](121–231) (Fig. 1B), similar to wild-type mPrP(121–231) (Fig. 1C). Hence, either tyrosine or phenylalanine in position 169 maintains an NMR-observable dynamic local polymorphism in mouse PrP^C, with conformational exchange at intermediate rates on the chemical-shift frequency timescale, whereas there is no evidence in the NMR data for this conformational polymorphism after substitution of the aromatic residue in position 169 by glycine or alanine. Based on these observations, we determined NMR structures for the two variants of mPrP(121–231) containing replacements of Y169 with Gly or Ala, respectively, as well as the NMR structure of mPrP(121–231) at 37 °C.

Fig. 1.

Temperature dependence over the range 5–40 °C of the NMR signals of the β2–α2 loop residues 167–171 in two designed variants of mPrP(121–231) generated by exchange of residue Y169 and in wild-type mPrP(121–231). Cross-sections along ω₂(¹H) from 750 MHz 2D [¹⁵N,¹H]-HSQC spectra are shown. (A) mPrP[Y169G](121–231). (B) mPrP[Y169F](121–231). (C) mPrP(121–231). The signal of A133 represents the behavior of those residues that are not affected by intramolecular conformational exchange.

NMR Structures of mPrP[Y169G](121–231) and mPrP[Y169A](121–231).

For both proteins, nearly complete backbone and side-chain NMR assignments were obtained (for details, see deposits 17081 and 17087 at www.bmrb.wisc.edu), using standard techniques for uniformly ¹³C,¹⁵N-labeled proteins (20). Statistics of the structure determinations, which used the softwares ATNOS/CANDID (21, 22), DYANA (23), and OPALp (24, 25), are given in Table 1. The results are visualized in Fig. 2 A and B and Fig. S1, with the amino acid side chains in Fig. S1 color-coded according to their global displacement values, D (26). The architecture of the two proteins coincides with the structures of other mammalian PrPs (4, 8–16), consisting of three α-helices spanning residues 144–153, 172–190, and 200–226, and a short antiparallel β-sheet of residues 128–131 and 161–164, where the helices α1 and α2 terminate with 3₁₀-helical turns of three to four residues. Within this conserved structural scaffold, the β2–α2 loop (located below the number 172 in the two structures shown in Fig. 2 A and B) adopts a different structure from that seen in all other PrP^Cs studied so far, with the residues 167–170 forming a type I β-turn (Fig. 3) (27, 28).

Table 1.

Input for the structure calculation and characterization of the energy-minimized NMR structures of mPrP[Y169G](121–231) and mPrP[Y169A](121–231) at 20 °C and mPrP(121–231) at 37 °C (“WT at 37 °C”)

	Y169G	Y169A	WT at 37 °C
Constraints
NOE distance limits*	3,107	2,901	3,298
Intraresidual	616	644	608
Sequential	832	762	846
Medium-range	893	817	938
Long-range	766	678	908
Dihedral angles	128	126	116
Target function, Å2	1.60 ± 0.39	2.31 ± 0.20	2.43 ± 0.59
Residual violations
NOEs
Number > 0.1 Å	34 ± 5	31 ± 5	32 ± 6
Maximum (Å)	0.15 ± 0.01	0.14 ± 0.02	0.14 ± 0.01
Dihedral angles
Number > 2.0°	0 ± 0	0 ± 0	0 ± 0
Maximum, °	1.17 ± 0.48	1.54 ± 0.61	1.56 ± 1.83
Amber energies, kcal·mol-1
Total	−4,892 ± 49	−4,922 ± 105	−4,975 ± 92
van der Waals	−350 ± 11	−322 ± 12	−338 ± 16
Electrostatic	−5,469 ± 42	−5,554 ± 94	−5,577 ± 82
Rmsd, ņ
bb (125–226)	0.39 ± 0.06	0.49 ± 0.06	0.39 ± 0.05
ha (125–226)	0.72 ± 0.05	0.89 ± 0.05	0.70 ± 0.05
Ramachandran statistics, %‡
Most favored	80.5	81.4	80.8
Additional allowed	16.7	16.6	16.9
Generously allowed	1.5	1.9	2.2
Disallowed	1.4	0.3	0.1

Except for the entries describing the input for the structure calculations, the average values for the 20 energy-minimized conformers with the lowest residual DYANA target function values and the standard deviations among them are given.

^*The different numbers of experimental upper distance limits are primarily due to somewhat different protein concentrations used to record the NMR data for the individual PrPs.

^†The rmsds were calculated relative to the mean coordinates for the bundle of 20 conformers. bb indicates the backbone atoms N, C^α, C′; ha stands for “all heavy atoms.” The numbers in parentheses indicate the residues for which the rmsd values were calculated.

^‡As determined by PROCHECK (40).

Fig. 2.

NMR structures at 20 °C and pH 4.5 of designed variants of mPrP(121–231) generated by exchange of Y169 and of mPrP(121–231) at 37 °C. (A) mPrP[Y169G](121–231). (B) mPrP[Y169A](121–231). (C) mPrP(121–231) at 37 °C. Shown are bundles of 20 energy-refined DYANA conformers representing the polypeptide backbone of residues 125–228 by a spline function through the C^α atoms. The sequence locations at the start and end of the three α-helices are indicated. Color code: β-strands, green; 3₁₀-helical turns, cyan; α-helices, red (A), orange (B), and yellow (C). All-heavy-atom stereo views of structures from Fig. 2 A and B are displayed in Fig. S1.

Fig. 3.

NMR structures and chemical shifts of the polypeptide segment 165–175, which includes the β2–α2 loop, in eight PrP(121–231) constructs. All data were collected at pH 4.5. The temperature was 20 °C, except for B, D, and F. (A) Wild-type mPrP. (B) Wild-type mPrP at 37 °C. (C) Elk PrP. (D) Horse PrP at 25 °C. (E) mPrP[S170N]. (F) mPrP[D167S] at 25 °C. (G) mPrP[Y169G]. (H) mPrP[Y169A]. Shown are the polypeptide backbone as a spline function through the C^α atoms, with the radius of the tubes proportional to the mean global backbone displacement per residue among a bundle of 20 energy-minimized conformers used to represent the NMR structures, the hydrogen bonds as dashed yellow lines, and the chemical-shift deviations from the random coil values, , where red color highlights data associated with the 3₁₀-helical conformation observed in B–F, and blue color those for the β-hairpin observed in G and H. The hydrogen bonds are H^N168–O165 and H^N169–O166 in B–F, and H^N170–O167 and H^N167–O^γ170 in G and H. In A, the chemical-shift data are incomplete because of a missing assignment for residue 167, and because the loop is structurally disordered, no hydrogen bonds are indicated (see text). The atom coordinates were taken from the following PDB deposits: (A) PDB ID code 2L1H, (B) PDB ID code 2L39, (C) PDB ID code 1XYW, (D) PDB ID code 2KU4, (E) PDB ID code 2K50, (F) PDB ID code 2KU5, (G) PDB ID code 2L1D, and (H) PDB ID code 2L40.

NMR Structure of Wild-Type mPrP(121–231) at 37 °C.

An NMR structure determination of wild-type mPrP(121–231) was pursued under conditions where the amide proton NMR signals of all β2–α2 loop residues are observable (see Fig. 1C). Tests of the stability of a 1.5 mM NMR sample of [u-¹⁵N]-mPrP (121–231) in 10 mM sodium acetate buffer, pH 4.5, at variable temperature by monitoring the signal intensities in 1D ¹H NMR spectra showed that at 37 °C there was a loss due to deterioration of the protein solution of only about 5% after 4 d. We therefore recorded each of three 3D heteronuclear-resolved [¹H,¹H]-NOESY spectra with a freshly prepared 1.5 mM solution of uniformly ¹³C,¹⁵N-labeled mPrP(121–231) at 37 °C. A 3D ¹⁵N-resolved [¹H,¹H]-NOESY dataset recorded at 500 MHz, where the lower ¹H resonance frequency was chosen to obtain narrower line widths, yielded NOE cross-peaks with the amide protons of all the β2–α2 loop residues. Based on nearly complete resonance assignments (deposit 17174 at www.bmrb.wisc.edu), the NMR structure was determined (Table 1).

The globular fold of mPrP(121–231) at 37 °C (Fig. 2C) is closely similar to that of mPrP(121–231) at 20 °C (4, 12) and the structures in Fig. 2 A and B, with an antiparallel two-stranded β-sheet of residues 128–131 and 161–164, three α-helices comprising the residues 144–153, 172–190, and 200–227, and 3₁₀-helical turns formed by the residues following the helices α1 and α2 (Fig. 2C). The β2–α2 loop structure includes a well-defined 3₁₀-helical turn for residues 166–169, which is different from both the loop structure in the variant proteins of Fig. 2 A and B, and the structurally disordered loop in the mPrP(121–231) structure at 20 °C (4, 12).

Survey of β2–α2 Loop Conformations in NMR Structures of Mammalian PrP^Cs.

In this section, we compare the conformations of the polypeptide segment 165–175 in NMR structures of recombinant mammalian PrP^Cs and designed variants of mPrP^C (Fig. 3 and Fig. S2). Using the standard criteria implemented in the program MOLMOL (29), we searched these structures for regular secondary structure elements and the presence of hydrogen bonds, and we evaluated the deviations of the chemical shifts from the random coil values, , for the individual residues.

In all NOE-based NMR structures of PrP^Cs that contain Tyr in position 169 and show a well-defined β2–α2 loop, the residues 166–169 form a 3₁₀-helical turn, with hydrogen bonds H^N/H^ε168–O165 and H^N169–O166 (Fig. 3 B–F). The 3₁₀-helical turn is also manifested by a typical pattern of values, with a large positive value for Val166 and smaller positive values for the residues 167 and 168. This loop conformation is observed in the wild-type mPrP(121–231) structure at 37 °C (Fig. 3B) and in all PrP^Cs that show a well-structured β2–α2 loop at 20 °C, including the wild-type proteins of elk, bank vole, wallaby, and horse (Fig. 3 C–F and Fig. S2) (12–15). Because the NMR signals of all β2–α2 loop residues except Asp167 are observable in mPrP(121–231) also at 20 °C, and the measured values are the weighted average over the values of all the conformers involved in the exchange process that leads to line broadening of amide group signals (Fig. 1 B and C), the pattern of values in Fig. 3A implies that a 3₁₀-helical turn is the dominantly populated β2–α2 loop conformation also in mPrP(121–231) at 20 °C.

The type I β-turn conformation of the β2–α2 loop observed in the NMR structures of mPrP[Y169A](121–231) and mPrP[Y169G](121–231) (Fig. 2 A and B), with hydrogen bonds H^N167–O^γ170 and H^N170–O167 (Fig. 3 G and H), has so far only been observed in mPrP^C variants with glycine or alanine in position 169. The values show a different pattern from that observed for the 3₁₀-helical turn, with residue 168 having a large positive value, residue 166 a somewhat smaller positive value, and residue 167 a value near zero.

Conformational Exchange Between 3₁₀-Helical and Type I β-Turn Conformations of the β2–α2 Loop Causes NMR Line Broadening in Wild-Type Mouse PrP^C.

The line narrowing observed in loop signals of mPrP(121–231) upon raising the temperature suggest that each amide resonance is the average of multiple conformers that interconvert rapidly relative to their chemical-shift difference. The line shapes of NMR signals averaged by conformational exchange depend on the exchange rate constant, k; the populations of the exchanging states; and the difference between the chemical shifts (in frequency units) in the exchanging structures, Δν. The results described in the following reveal that the NMR data for wild-type mouse PrP^C can be explained by exchange between two structurally characterized conformers, A and B, with populations p_A and p_B, and k≫Δν. The contribution from the conformational exchange to the observed line widths at half height, Δν_1/2, then is

[1]

where Δν_1/2 and Δν are given in Hz, and k is in s^-1 (30, 31). For a common set of k, p_A and p_B values, the NMR signals of the individual exchanging residues may be differently broadened, due to different Δν values (32). Consistent with the prediction of Eq. 1 that the line broadening is enhanced at increasing spectrometer field because of increasing Δν, broader lines were observed for the loop residues at 750 MHz than at 500 MHz (Fig. S3A). For this reason, we measured the ¹⁵N-resolved [¹H,¹H]-NOESY used for the structure determination of mPrP(121–231) at 500 MHz (Fig. S3B).

Based on Eq. 1, the experimental evidence supports that the NMR line broadening of β2–α2 loop residues in wild-type mPrP^C is due to exchange between two conformations that contain, respectively, the 3₁₀-helix or the type I β-turn loop structure (Fig. 3). Thus, the residues 170, 171, and 175, which show the largest amide proton chemical-shift differences, Δν(¹H), between the two limiting conformations, which are represented here by the chemical shifts at 40 °C of wild-type mPrP(121–231) and the designed variant mPrP[Y169A](121–231), respectively, also show the largest line-broadening effects along ω₂(¹H), Δν_1/2(¹H), in mPrP(121–231) (Fig. 4 A and B), whereas there is no exchange line broadening for these residues in mPrP[Y169A](121–231) (Fig. 4C). A similar qualitative correlation is seen for residues 167 and 171, which show large ¹⁵N chemical-shift differences, Δν(¹⁵N), and large line broadening along ω₁(¹⁵N)) in mPrP(121–231) (Fig. S4) where as no line broadening is observed in mPrP[Y169A](121–231). Furthermore, there is no loop residue with small Δν values along ω₁(¹⁵N) or ω₂(¹H) that shows visible exchange broadening of its NMR signals. These qualitative correlations between corresponding Δν and Δν_1/2 values have been quantitatively substantiated through verification of the linear relation between Δν_1/2 and Δν² predicted by Eq. 1 as described in Materials and Methods (Fig. 4D). The data points for residues 166, 168, and 172–174 (Fig. 4 A–C and Fig. S3) would also fall on the linear regression curve of Fig. 4D, near its origin at Δν² = 0 and Δν_1/2 = 0.

Fig. 4.

Proton chemical-shift differences between corresponding 750 MHz 2D [¹⁵N,¹H]-correlation NMR signals in mPrP(121–231) and mPrP[Y169A](121–231) at 40 °C, and exchange broadening of the NMR signals. (A) Histogram-type plot versus the mPrP(121–231) amino acid sequence of the square of the amide proton shift differences between the two proteins in hertz, . (B) Plot of the linewidths at half-height in mPrP(121–231) along ω₂(¹H), . (C) Same as B for mPrP[Y169A](121–231). (D) Quantitative assessment, using Eq. 1, of the hypothesis that the NMR line broadening for β2–α2 loop residues in mPrP(121–231) is due to exchange between the 3₁₀-helix and β-hairpin conformations of the β2–α2 loop. ¹H and ¹⁵N NMR line broadening is plotted along the horizontal axis, and the square of chemical-shift differences between corresponding NMR signals in mPrP(121–231) and mPrP[Y169A](121–231) is plotted along the vertical axis. Data are shown only for those signals of residues 166–175 that show large line broadening, with data (Fig. 4 A and B) indicated by squares and ¹⁵N data (Fig. S4 A and B) by crosses. The horizontal axis is in units of hertz and the vertical axis in squared hertz. The slope of the plot representing in Eq. 1 is equal to 9,000 s^-1. The Pierson correlation coefficient is 0.95. The chemical shifts used here were obtained by transferring the resonance positions from the available assignments for mPrP(121–2321) at 37 °C (BMRB accession no. 17174) and mPrP[Y169A](121–231) at 20 °C (BMRB accession no. 17213) to those at 40 °C by experiments of the type shown in Fig. 1.

Estimates for the values of the parameters p_A, p_B and k in Eq. 1 resulted from the following considerations: The slope of the linear regression curve in Fig. 4D corresponds to a value of 9,000 s^-1 for the product k/(2π·p_A·p_B) in Eq. 1. Using some simplifying assumptions for the experimental determination of Δν and considering that the largest possible value of the product p_A·p_B is 0.25, k ≤ 14,000 s^-1 is obtained as an estimate for the upper limit of the exchange rate constant. A more conservative treatment indicates a limiting value of k ≤ 56,000 s^-1 (see SI Text). A lower limit of k = 800 s^-1 has been derived from the fact that the largest Δν values affecting the exchange-averaged signals of the two conformations were about 800 Hz at the ¹H resonance frequency of 750 MHz used for these measurements (Fig. 4A). The corresponding lower limit for the population of the minor species is p_B = 0.014.

In conclusion, the data presented in the preceding sections lead to the following description of the conformational state of wild-type mouse PrP(121–231) in solution: mPrP^C forms two locally different and widely unequally populated structures. The more abundant species contains a 3₁₀-helix structure of the β2–α2 loop and its population is above 90%, and the lesser populated species includes a type I β-turn structure of the β2–α2 loop. Over the temperature range of 20–40 °C the two structures exchange at intermediate rates on the chemical-shift timescale, as manifested in the signal line shapes of the NMR spectrum (Fig. 1C). The major structure involved in this exchange process on the millisecond to microsecond timescale represents an ensemble of rapidly exchanging conformers that all contain the 3₁₀-helix loop structure (Fig. 3B), and the less populated structure represents a similar ensemble of conformers that all contain the type I β-turn structure of the loop (Fig. 3H).

Biological Implications.

A possible link between the data of Figs. 1–4 and physiological functions of the cellular prion protein resulted from a search of the database of all available prion protein amino acid sequences (see Materials and Methods), which showed that Tyr169 is strictly conserved in mammalian species. Although it maintains both the local exchange process in the β2–α2 loop of mPrP[Y169F](121–231) (Fig. 1B) and the overall stability of the PrP^C fold, Phe169 is not encountered in mammalian PrPs. There is thus an indication that the hydroxyl group of tyrosine in position 169 has an essential role for the physiological function of mammalian prion proteins.

Comparison of mPrP(121–231) (Fig. 5A) and mPrP[Y169A](121–231) (Fig. 5B) shows that different amino acid side chains are exposed to the solvent in the two structures. Firstly, although the hydroxyl function of Ser170 is part of the surface epitope in the 3₁₀-helix loop structure, it is buried in the protein interior in the type I β-turn conformation. Secondly, the side chain of residue 169 changes from a location in the 3₁₀-helical loop structure, where it points toward the protein core, to a surface-exposed orientation in the β-turn conformation of the β2–α2 loop. Modeling studies based on replacing Ala169 in the structure of Fig. 5B by Tyr indicate that if the β2–α2 loop in mammalian mPrP^Cs takes on the type I β-turn conformation, the side chain of Tyr169 is exposed on the protein surface (Fig. 5 C and D). From combined analysis of the exchange broadening of the NMR signals (Figs. 1B and 4 A–D) and the chemical shifts, and in particular from the fact that the patterns of chemical shifts in the two forms of the β2–α2 loop are characteristically different (Fig. 3 and Fig. S2), we conclude that in wild-type mPrP^C the 3₁₀-helix conformation of the loop is the more abundant form, with estimated populations of ≥0.9 and ≤ 0.1 for the two locally different conformations (see also the preceding section and SI Text).

Fig. 5.

Protein surface epitopes formed in the NMR structures by residues 164–175 in mPrP(121–231) at 37 °C and mPrP[Y169A](121–231) at 20 °C, which represent the two limiting structures connected by intermediate-rate conformational exchange in wild-type mammalian PrP^Cs (see text). The polypeptide backbone from the end of strand β2 at residue 163 to the second turn of helix α2 at residue 178 is shown, with space-filling presentations of the side chains 164 and 171 in blue, 168–170 in functional colors, and 175 in green. (A) NMR structure of mPrP(121–231) at 37 °C. (B) NMR structure of mPrP[Y169A](121–231) at 20 °C. (C and D) Models of the less-abundant conformation of mPrP(121–231) (see text) generated by replacement of A169 in the experimental structure (B) by Y169; only two of the three staggered rotamers about the C^α–C^β bond are shown, with X¹ = +60° (C) and X¹ = -60° (D), because the third rotamer would involve extensive steric crowding.

It will now be of keen interest to follow up on these structural studies with experiments using transgenic mice expressing variant mouse prion proteins that are devoid of the tyrosine hydroxyl group in position 169. In this context, one should also recall that earlier work advanced the suggestion, which has so far not been substantiated, that a surface epitope formed in part by the β2–α2 loop in PrP^C might be a recognition site for effector molecules that would affect the transition from the cellular form of the prion protein to the disease-related, aggregated PrP^Sc form (33, 34).

Materials and Methods

Protein Preparation.

Clones for mPrP[Y169A](121–231), mPrP[Y169F](121–231), and mPrP[Y169G](121–231) were obtained by single-amino acid substitutions in the mPrP(121–231) gene, using the QuikChange® site-directed mutagenesis kit (Stratagene). Uniformly ¹⁵N- and ¹³C,¹⁵N-labeled proteins were prepared as described (9, 35, 36). For the NMR experiments, concentrated solutions containing 1 to 2 mM protein were prepared in H₂O containing 10 mM [d₄]-sodium acetate buffer at pH 4.5, 10% D₂O, 0.02% sodium azide and a protease inhibitor cocktail (Roche).

NMR Experiments.

NMR data for the structure determinations of mPrP[Y169A](121–231) and mPrP[Y169G](121–231) were measured at 20 °C, those for mPrP(121–231) at 37 °C. The presence of the PrP^C fold in mPrP[Y169F](121–231) was verified by the close similarity of the 2D [¹⁵N,¹H]-HSQC spectrum to that of mPrP(121–231). Resonance assignments were obtained with standard triple resonance experiments (20) recorded on a Bruker DRX500 spectrometer with cryogenic probehead. Three-dimensional ¹⁵N-resolved [¹H,¹H]-NOESY and ¹³C-resolved [¹H,¹H]-NOESY spectra were recorded on Bruker Avance900 and DRX750 spectrometers, using a mixing time of 60 ms, except that the 3D ¹⁵N-resolved [¹H,¹H]-NOESY data for mPrP(121–231) at 37 °C was recorded at 500 MHz in order to reduce the exchange line broadening. Chemical shifts for mPrP(121–231) were referenced relative to internal 2,2-dimethyl-2-silapentane-5-sulfonic acid (DSS), and for the other two proteins with a coaxial insert (Norell Inc.) containing DSS in NMR buffer, because of protein precipitation upon addition of DSS. The program CARA (37) (www.nmr.ch) was used for the analysis of the NMR spectra.

The temperature-dependence of 2D [¹⁵N,¹H]-HSQC spectra was measured at 750 MHz with 1 mM solutions of the ¹⁵N-labeled proteins. Prior to Fourier transformation, the datasets were zero-filled to 16 k and 1 k points in the ¹H and ¹⁵N dimensions, respectively, and linewidths were measured at half the height of the maximum signal intensity. The exchange contribution to the observed linewidths was estimated by subtracting the average linewidth determined from all signals excluding those deviating by one standard deviation from the mean value. These experiments were also used to measure the chemical shifts of mPrP(121–231) and mPrP[Y169A](121–231) at 40 °C, which have been used in Fig. 4 and Fig. S4.

NMR Structure Calculation.

The standard protocol of the stand-alone ATNOS/CANDID program package (21, 22), version 1.2, was used together with DYANA (23) for automatic peak picking, automatic NOE assignment, and structure calculation. The final cycle of the calculation was started with between 80 and 120 randomized conformers, and the 20 conformers with the lowest residual target function values were energy-minimized in a water shell with the program OPALp (24, 25), using the AMBER force field (38). The program MOLMOL (29) was used to analyze the results of the protein structure calculations, including regular secondary structure identification with the method of Kabsch and Sander (39).

Database Search of Residue Types in the Sequence Position 169 of Mammalian Prion Proteins.

Mammalian PrP sequences deposited as of December 2010 were analyzed to determine residue variations in position 169, using the UniProtKB (Protein Knowledgebase; www.uniprot.org) with the sequence of mPrP(1–254) as input. Nonmammalian PrP sequences as well as mammalian PrP sequence fragments that do not contain information on position 169 [according to the numeration used by Schätzl et al. (7)] were removed from the list, which finally contained 524 PrP sequences from 213 different mammalian species.