Sheep Prion Protein Synthetic Peptide Spanning Helix 1 and (cid:1) - Strand 2 (Residues 142–166) Shows (cid:1) -Hairpin Structure in Solution*

According to the “protein only” hypothesis, a conformational conversion of the non-pathogenic “cellular” prion isoform into a pathogenic “scrapie” isoform is the fundamental event in the onset of prion diseases. During this pathogenic conversion, helix H1 and two

Prion diseases are severe neurodegenerative disorders of genetic, sporadic, and infectious origins (1).According to the "protein only" hypothesis (2,3), a conformational conversion of the non-pathogenic "cellular" isoform of the prion protein (PrP C ), 1 a strongly conserved cell surface glycoprotein ex-pressed in all mammalian species studied so far (4), into a pathogenic "scrapie" isoform (PrP Sc ), is the fundamental event in the pathogenicity.The major structural feature of prion conversion manifests itself as an increase of the ␤-sheet content in PrP Sc .PrP C has 42% of its residues folded in ␣-helices and 3% as ␤-sheets, whereas PrP Sc is composed of 30% ␣-helices and 43% ␤-sheets (5).Transgenic studies argue that infectious PrP Sc acts as a template (6,7) upon which the normal PrP C is refolded into a pathogenic isoform through a process facilitated by a still unknown factor X (8).
Due to a high similarity between PrP of different mammals, their sequences are usually represented as aligned with the human one (9).In its mature form, after removal of N-and C-terminal signal sequences, the mammalian cellular PrP contains about 210 amino acid residues, from 23 to 231 in human numbering.The minimal prion polypeptide fragment required for infectious propagation was mapped to residues 90 -231 (1).
NMR characterization of the recombinant full-length mouse (10,11), hamster (12,13), bovine (14), and human (15) prion proteins showed that all these molecules have very similar three-dimensional structures, including a flexible unstructured N-terminal "tail" composed of residues 23-120 and a mostly ␣-helical globular C-terminal part 121-231 (see Fig. 1 below).This globular PrP domain, which behaves as an independent folding unit (16), is composed of two short antiparallel ␤-strands (S1 and S2) and three ␣-helices (H1-H3) (17).Helices H2 and H3 are stabilized by a disulfide bond and form the C-terminal scaffold, which probably has the same conformation in PrP Sc and PrP C (18).However, structural studies of PrP Sc have been limited because of its aggregated state (19 -22).
Several prion-derived peptides were analyzed in attempts to clarify the molecular basis of the conformational change (23)(24)(25)(26)(27)(28)(29)(30)(31)(32).Most of them belong to the 90 -145 region generally believed to be essential for prion propagation.It was shown that this region as a whole, as well as some of its parts, have intrinsic propensity to produce insoluble intermolecular aggregates of extended ␤-like structure, which might be involved in promoting the PrP C 3 PrP Sc conformational transition.
Here we have examined by CD and NMR spectroscopic analyses the solution structure of a linear 26-mer peptide (hereafter referred to as peptide "n3"), which had not been investigated before.It contains 25 residues corresponding to the domain 145-169 of sheep prion protein (34) (142-166 in human prion protein numbering) and a C-terminal cysteine.Its sequence is 1 GNDYEDRYYRENMYRYPNQVYYRPVC 26 , where the underlined letters represent the segments corresponding to helix H1 (144 -154) and ␤-strand S2 (161-164), respectively, in the PrP C structure.The experimental results obtained in the present work show that this peptide (i) remains soluble in aqueous medium and does not aggregate even in the millimolar concentration range and (ii) exhibits an intrinsic propensity to a ␤-hairpin-like conformation at neutral pH, which contrasts the prion-derived peptides studied earlier.This ␤-propensity can be one of the internal driving forces of the molecular rearrangement responsible for the pathogenic conversion of the prion protein.

EXPERIMENTAL PROCEDURES
Peptide Synthesis and Purification-The n3 peptide was synthesized on solid phase on Applied Biosystems 431A peptide synthesizer.The synthesis was carried out using HMP-resin (4-hydroxymethyl phenoxymethyl-copoly(styrene-1% divinylbenzene)) and Fmoc (N-(9-fluorenyl)methoxycarbonyl) and 2-(H-benzotriazole-1-yl)-1,1,3,3-tetramethyluronium hexafluoro-phosphate/N-hydroxybenzotriazole as coupling mediators.After synthesis, peptide products were deprotected and cleaved from the support by treatment with trifluoroacetic acid.The peptide was purified by reverse phase high performance liquid chromatography (Waters) on a semi-preparative column C4 eluted with linear gradient of water/acetonitrile/trifluoroacetic acid (10 -100%).Peptide fractions were pooled and lyophilized.To check the peptide purity, amino acid analysis and mass spectrometry were performed, and the results corresponded well to the expected sequences.The purity exceeded 99%.The mouse prion peptide of 99% grade purity was purchased from Eurogentec (Herstal, Belgium).
Circular Dichroism (CD) Spectroscopy-CD spectra were recorded on a Jasco J-710 instrument.Samples were studied in quartz cells with path lengths 0.5 or 1 mm.Peptide concentrations were determined by ultraviolet absorbance (35).CD signals were expressed as the mean residue weight ellipticity values, [⌰], in deg cm 2 dmol Ϫ1 .Samples were scanned over the wavelength range 190 -260 nm by recording values every 0.1 nm with 100 nm min Ϫ1 scan rate, with an integration time of 0.5 s and a 2-nm bandwidth.Each spectrum was the average of 8 or 16 scans.
Fluorescence-detected Circular Dichroism (FDCD) Spectroscopy-FDCD spectra were measured at room temperature in a Jasco J-720 spectropolarimeter with the fluorescence detector on-line with the incident light to avoid artificial linear dichroic signal.The exciting circularly polarized light was cut off by saturated NaNO 3 solution in 2.5-cm path length cuvette as described by Muto et al. (36).
NMR Spectroscopy-NMR samples were dissolved in H 2 O/D 2 O (9:1, v/v) or D 2 O with 10 mM sodium phosphate-buffered, pH 6.5, solution.A crystal of TSPD 4 (3-(trimethylsilyl) [2,2,3,3-d 4 ] propionic acid), sodium salt, was used as an internal reference for the proton shifts.The experiments were run at 500.13 MHz for 1 H on Bruker AMX 500 spectrometer equipped with a Silicon Graphics workstation.The WATERGATE method (37) was used to eliminate solvent signal in H 2 O/D 2 O (9:1) solution instead of the presaturation method in D 2 O solution.One-and two-dimensional TOCSY, two-dimensional ROESY, and two-dimensional NOESY spectra were recorded at several temperatures within the 5-37 °C range.Mixing times of 35-80 ms were used for twodimensional TOCSY.For two-dimensional ROESY experiments, a spinlock of 200 -400 ms was used.Two-dimensional phase NOESY experiments in the 50-to 800-ms range were carried out using the Statestime-proportional phase incrementation method.The 3 J HN␣ scalar coupling constants were extracted from two-dimensional TOCSY spectra.The 13 C NMR chemical shifts were carefully calibrated using 4,4dimethyl 4-silapentane sodium sulfonate and TSPD 4 .The assignments of 13 C were made using 1 H- 13 C chemical shift correlation (pulse field gradient-heteronuclear multiple quantum correlation (38), and pulse field gradient-heteronuclear multiple bond correlation (39)) spectra.
Molecular Modeling and Data Deposition-Conformational calculations were performed in the torsion angle space with the standard geometry of amino acids and planar peptide groups.The distance restraints obtained from the NOESY experiments were first classified as strong, medium, weak, and very weak, with the upper distances assigned at 2.5, 3.5, 4.5, and 5.5 Å, respectively.The lower distance restraints were unnecessary, because the full atom AMBER94 (40) force field was used throughout.When restraints were applied to pseudoatoms, the distances were increased as usual (41).The calculations involved multiple cycles of simulated annealing starting from arbitrary extended conformations, with pseudoatom restraints and the variable target function approach (42).At the final stage pseudoatoms were replaced by appropriate protons, and all restraints were set at a single upper limit of 5.0 Å.The general purpose internal coordinate molecular dynamics software (43,44) was used with the AMBER94 atom parameters.The polar solvent environment was modeled implicitly by reducing the charge of ionized side chains by one-half and applying the distance-dependent dielectric permitivity D(r) ϭ r.The final refined conformation was additionally verified by a 100-ps molecular dynamics (MD) simulation run with INSIGHT standard software (45).Visualization and analysis of the calculated conformers were performed with MOLMOL (46) and PROCHECK-NMR (47).Upper bound NOE constraints, used in the final round of structure calculation, atomic coordinates for a bundle of 20 MD refined conformers and for the best conformer of the n3 peptide have been deposited in the Protein Data Bank (available at www.rcsb.org,PDB code 1G04).The proton chemical shifts table and the 3 J HN␣ scalar coupling constants of the n3 peptide have been deposited with the BioMagResBank (www.bmrb.wisc.edu,code BMRB-4010).

RESULTS
Secondary Structure Analysis-The overall secondary structure of the n3 peptide was analyzed in different experimental conditions by far-UV CD spectroscopy (Fig. 2).When studied in PB buffer (10 mM sodium phosphate buffer, pH 6.5) at 25 °C, the peptide produced a broad negative Cotton effect at 208 nm with a shoulder at 216 nm, and two smaller signals, namely, a positive one at 233 nm and a negative one at 241 nm (Fig. 2).Because the n3 peptide contains 7 tyrosine residues, for proper interpretation of the CD signal it is important to differentiate the contributions from the backbone and from the aromatic side chains, respectively.Fluorescence-detected circular dichroism (FDCD) spectroscopy was used to probe selectively the CD contribution of the tyrosine residues (48).The separated signal was very small, thus, it was concluded that the total CD signal of the n3 peptide reflected mostly the secondary structure of the backbone.The observed spectrum indicates, therefore, the presence of some secondary structure different from random coil conformation (49).Low acidic pH (less than 2) resulted in typical random coil CD signal (characterized by a prominent negative band around 198 nm) of the n3 peptide in aqueous solution (data not shown) suggesting complete destabilization of the structure observed at pH 6.5.However, the solubility of the peptide remained unaffected by acidic pH.The effects of other environmental factors, notably, 2,2,2-trifluoroethanol, upon the conformation of the n3 peptide were studied by CD and NMR spectroscopic analyses and described elsewhere (50).
To check whether any aggregation of the n3 peptide takes place in PB buffer, the concentration dependence of the normalized CD signals of the peptide samples was studied at two different wavelengths (217 and 225 nm).The mean residue weight ellipticities, [⌰], did not vary in the 5-1000 M range (Fig. 2, inset).Also, no significant changes in chemical shifts or line widths were observed in one-dimensional NMR spectra in the 480 -4800 M concentration range (data not shown).These two tests, which are sensitive to aggregation (51,52), indicate that the peptide does not aggregate in these conditions.It is worth noting that, when stored at room temperature, the peptide samples maintained their CD and NMR spectra unchanged over a 4-month period, demonstrating also the absence of any time-dependent aggregation.
The conformational properties of the 26-mer peptide 1 GND-WEDRYYRENMYRYPNQVYYRPVD 26 , corresponding to the mouse prion protein region 142-167, were also studied.The mouse peptide differs from the n3 peptide by 2 residues (Trp 4 and Asp 26 instead of Tyr 4 and Cys 26 , respectively) and shows the same CD spectra as the n3 peptide in PB buffer in the same range of concentrations (data not shown).Thus, the n3 peptide can be considered as a realistic model for the study of the conformational properties of the 142-166 segment of both sheep and mouse prion proteins.
NMR Solution Structure-The absence of time-dependent aggregation of the n3 peptide at millimolar concentrations at neutral pH allowed us to study conformational features of the peptide by NMR spectroscopy.To obtain the spatial structure of the n3 peptide, the 1 H NMR experiments were carried out at 500.13 MHz with 4.5 mM non-labeled peptide in PB buffer solution.Two-dimensional TOCSY spectra were recorded with 35-to 80-ms mixing times, and the assignment of spin systems thus determined (Fig. 3) was confirmed by NOESY and ROESY experiments.When all spin systems were identified, the sequential assignment of the peptide amino acid residues (Table I) was carried out by using the strategy developed by Wu ¨thrich (41).To resolve the assignment ambiguities resulting from signal overlaps, all experiments were performed at three different temperatures, namely, 5 °C, 20 °C, and 37 °C, exploiting the temperature-induced shift of the amide resonances.
The CSI protocol, involving 1 H ␣ , 13 CЈ, 13 C ␣ , and 13 C ␤ chemical shifts, allows four independent approaches to be used in identifying and locating secondary structures in proteins (53).In the case of the n3 peptide, no residues in regular secondary structures was found by applying the 1 H ␣ CSI method (data not shown), whereas the other CSI techniques revealed that residues 20 -23 appeared to show ␤-strand propensity (Fig. 4).Additionally, a hairpin structure was reflected in the 3 J HN␣ scalar coupling constants (Table I).Among 19 values measured, the 15 ones from residues 2-7, 10, 12, 13, 15, 18 -20, 25, and 26 are larger than their random coil values (54,55), which means that these residues preferentially populate ␤-strand angles.
The amide signal shift temperature coefficients (⌬(␦NH)/⌬T, in ppb K Ϫ1 ) were derived for all backbone amide protons of the n3 peptide (Table I).For the residues involved in intramolecular hydrogen-bonding network and/or hidden in a structural core-protected from solvent, one can expect low absolute values of this coefficient as compared with the range of values measured for random-coil peptides (56).Indeed, residues 6, 12, 13, 15, and 19 give values between Ϫ5 and Ϫ3.4 ppb K Ϫ1 and can be considered as at least partially protected from external water molecules.
The TOCSY spectral analysis shows the presence of duplicate cross-peaks for the spin systems of the two C-terminal residues: Val 25 and Cys 26 (Fig. 3).The corresponding ratio of peak intensities was estimated as 2:1.The strong d ␣-␦ sequential NOE connectivity between Arg 23 and Pro 24 indicates that the major form corresponds to the peptide structure in which the peptide bond between Arg 23 and Pro 24 is in the trans conformation.For the same bond in the minor form, the strong d ␣-␣ agrees with the cis conformation.This trans:cis equilibrium at Pro 24 points to a higher mobility of the C-terminal part of the backbone suggesting that the 24 -26 segment is not crucial for the overall peptide structure.
NOESY spectra were obtained at a variety of mixing times from 50 to 800 ms.All distance restraints used in the structure refinement were derived from NOEs observed at 5 °C in PB buffer during NOESY experiments at 500-ms mixing time (Fig. 5).The sequential NOE pattern observed was typical of extended backbone structure: weak d N-N (i, iϩ1) and strong d ␣-N (i, iϩ1) (Fig. 6).Weaker d ␣-N (i, iϩ1) registered for residues 13-15 are indicative of a turn region.A total of 169 unambiguous sequential and long and medium range NOE restraints were used in the n3 peptide structure calculation (Table II).The refinement procedure involved multiple cycles of simulated annealing (see "Experimental Procedures").At the end of each cycle the structure was quenched by energy minimization, and these conformations were ranked according to the residual weight of NOE restraints.The n3 peptide may have in solution a few conformations with the same backbone fold, but slightly different orientations of side chains, as commonly observed for short linear polypeptides (57).The NOE-derived contacts that involve side-chain protons may, therefore, contradict one another, and this presents the main source of experimental error.Fig. 7 shows backbone conformations of the five top ranked conformers with the residual NOE scores within 20% variation.This bundle is characterized by the r.m.s.d. of 3.64 Å for all non-hydrogen atoms, which may be considered as an estimate of the experimental error in the best structure shown in red.For this structure, the NOE upper limit violation was 0.17 Ϯ 0.15 Å, and its stereochemical quality (Table II) was considered as "satisfactory" by the PROCHECK software, which means that there are no strong violations of chemical geometry and no bad atom contacts.
The 20 low energy conformers shown in Fig. 8 have been obtained from the best structure by a 100-ps MD simulation with NOE restraints.The conformers are characterized by a low r.m.s.d. of 0.29 Å (all backbone atoms) and 0.54 Å (all non-hydrogen atoms), which indicates that restraints leave only a small valley around the minimum in conformational space of the peptide, and that, due to a sufficient number of long range NOEs, there is no uncertainty in the relative orientations of distant parts of the structure.
Structure Description-The NMR-derived data obtained in the present study indicate that in physiologically relevant conditions the n3 peptide is folded in a conformation with a ␤-hairpin overall topology and a sharp turn (Fig. 8).Part 8 -17 of the structure apparently is very stable, because most of the NOE signals detected belong to this region.Fragment Asn 12 -Arg 15 involves a gamma-turn hydrogen bond O 13 -HN 15 , but its overall geometry is very close to a type I ␤-turn, which could be stabilized by the O 12 -HN 15 hydrogen bond.A genuine ␤-turn conformation is easily obtained by minimization with one additional restraint upon the O 12 -HN 15 distance so that no extra  violations of NOE restraints are introduced, which means that these two similar conformations are equally possible according to experimental data.
The two strands protruding from the center are nearly antiparallel and close to each other in regions 2-11 and 16 -22, although we did not obtain the classic ␤-sheet backbonebackbone hydrogen bonding network.This does not contradict our experimental data, notably the CD spectra and the long distance NOE patterns, and may be attributed to the extremely high content of polar side chains in the sequence.As many as 20 residues of 26 are either charged or polar, and they all can compete for the backbone hydrogen bonding valences.Also, the lack of interstrand backbone contacts may be indicative of the dynamic plasticity of the peptide structure.However, segments 3-5 and 20 -22 have linear geometry very close to that of typical ␤-sheet strands.Interestingly, segment 20 -22 of the n3 peptide corresponds to residues 161-163 of the entire prion where they are involved in ␤-strand S2.
The C-terminal residues Pro 24 -Val 25 -Cys 26 form a helical stretch loosely bound to the rest of the structure suggesting its relatively high mobility and weak influence on the peptide overall conformation.The TOCSY experiments mentioned above revealed the trans:cis isomerization of Pro 24 , which, therefore, may occur without gross conformational perturbations.The C-terminal Cys residue is absent in the natural prion   sequence.In the refined conformation, it makes no external contacts, therefore, its structural role is negligible.DISCUSSION Perhaps, the most remarkable feature in the sequence of the n3 peptide is the very high content of polar and charged residues.At neutral pH, both termini should be ionized as well as eight side chains, namely, Asp 3 , Glu 5 , Asp 6 , Arg 7 , Arg 10 , Glu 11 , Arg 15 , and Arg 23 .The number of positively and negatively charged groups is equal, therefore, the sequence itself is perfectly balanced electrostatically.Moreover, according to the three-dimensional structure obtained, the relative positions of the charged groups in space are such that they tend to neutralize one another.Notably, local charges at Asp 3 and Arg 23 are partially compensated by those at N and C termini.These four groups do not form salt bridges, but they are close and interact apparently as two macro-dipoles.The remaining six charged groups form salt bridges, namely, Glu 5 -Arg 7 , Asp 6 -Arg 10 , and Glu 11 -Arg 15 .These observations suggest strongly that the delicate balance of electrostatic interactions is crucially important for the overall stability of the solution structure of this peptide.Earlier it has been reported that a longer peptide derived from mouse prion sequence 142-170 aggregates at neutral pH (33).This peptide has 4 additional residues at its C terminus, namely Asp 167 -Gln 168 -Tyr 169 -Ser 170 .If added to the three-dimensional structure reported here, the Asp 167 would replace the C-terminal carboxyl, but an additional non-compensated negative charge appears at the Ser 170 terminus.This slight modification can disturb the electrostatic balance so that the overall stability of the fold is reduced and the thermodynamic equilibrium is shifted from intra-to inter-molecular salt bridges, which may prompt the aggregation.
The sequence of the reverse turn 12-15 in the n3 peptide corresponds to residues 153-156 of the mouse prion protein, where they form a ␤-turn that connects helix H1 and coiled loop L3.A superposition of those two backbone segments from the mouse prion protein and the n3 peptide structures, fitted along the turn region, is shown in Fig. 9.The C-terminal parts of both structures are quite similar and seem to be virtually "invariant."In contrast, the N-terminal fragments are very different as regards their secondary structure and the global positioning with respect to the C terminus.To adopt the n3 peptide conformation starting from the corresponding native prion protein structure, helix H1 would be "unwound" into an extended form and rotate around the turn region toward the C-terminal region (58).
A solution structure of a shorter mouse prion peptide mPrP (143-158), NDWEDRYYRENMYRYP, spanning helix H1 and 4 of 6 residues of loop L3, was previously published (59).It was shown that the residues involved in PrP C helix H1 also formed a helix in the peptide, whereas the loop region was less structured, suggesting no specific intramolecular interactions exist between these two peptide segments.Contrariwise, in the n3 peptide, which contains, in addition to the mouse peptide, the C-terminal sequence NQVYYRPVC, the segment corresponding to PrP C helix H1 demonstrates an explicit ␤-extended conformation, whereas the conformation of loop L3 and ␤-strand S2 regions seems to be very similar to that observed in PrP C .As stated above (see "Results"), the impact of N-terminal segment PVC was minimal on the n3 peptide overall structure.So, it seems that the intramolecular interactions between amino acid residues of ␤-strand S2 and helix H1 regions play a critical role in the n3 peptide ␤-hairpin conformation.
In intact PrP C , any contacts between ␤-strand S2 and helix H1 are hindered mostly by surface loop L2 and by ␤-strand S1.One might propose that these contacts could be switched on by shifting loop L2 and ␤-strand S1 from their native positions.This rearrangement is sterically possible due to the surface exposure of the entire segment 121-143 and may occur upon partial unfolding of the prion protein, which would result in refolding of helix H1 into a ␤-extended conformation and should change significantly the fold of this region.
Sharman et al. (33) already studied three linear synthetic peptides spanning the ␤-sheet and helix H1 region of PrP C (residues 125-170).Unfortunately, their structural analysis could * The work was supported in part by the funding of the Institut National de la Recherche Agronomique (INRA) and INSERM.The costs of publication of this article were defrayed in part by the payment of page charges.This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C.Section 1734 solely to indicate this fact.§ Supported by an INRA fellowship.

FIG. 1 .
FIG. 1. Relative positioning of secondary structure elements of the minimal infectious part of PrP C (residues 90 -231 in human numbering).The coiled N-terminal tail is shown by a dotted line, surface loops L1-L6 are indicated by solid curves, ␤-strands S1 and S2 are represented as arrows, and helices H1-H3 are shown as cylinders.Sequential numbers of the amino acid residues constituting the elements are shown under the elements' names.

FIG. 2 .FIG. 3 .
FIG. 2. Far-UV CD spectrum of the n3 peptide (151 M) in PB buffer at 25 °C.The inset shows concentration dependence of the mean residue weight ellipticity, [⌰], measured in PB buffer at 25 °C at two different wavelengths: 217 nm (Ⅲ) and 225 nm (q).FIG. 3. Expansion of the fingerprint region of a 500-MHz twodimensional [ 1 H, 1 H]-TOCSY NMR spectrum of the n3 peptide in PB buffer at 5 °C.Assignments of cross-peaks are denoted with the sequence numbers.The asterisks indicate minor signals of residues Val 25 and Cys 26 raised because of cis isomerization of Arg 23 -Pro 24 peptide bond.

FIG. 4 .
FIG. 4. Deviations of 13 C chemical shift values (⌬␦, in ppm), determined for the n3 peptide in PB buffer at 37 °C, from those characteristic of the random coil.A, for carbonyl CЈ carbons; B, for C ␣ carbons; C, for C ␤ carbons.Any group of three or more consequent residues is considered to form a ␤-strand structure, if the corresponding ⌬␦ 13 CЈ and ⌬␦ 13 C ␣ values are negative and the ⌬␦ 13 C ␤ is positive for each residue within the group.The significant values indicative of ␤-strand conformation are shown within the shadowed box.

FIG. 5 .FIG. 6 .
FIG. 5. Selected regions of the 500-MHz NOESY spectrum of the n3 peptide (4.5 mM in PB buffer at 5 °C).Some important medium and long range NOEs are boxed, and the C ␣ , C ␤ , C ␥ , C ␦ , and C ⑀ protons involved in these NOEs are labeled as ␣, ␤, ␥, ␦, and ⑀, respectively, preceded by the corresponding residue one-letter code and number.
FIG. 7. A stereopair of backbone conformations of the five top ranked n3 peptide structures obtained by simulated annealing with NOE restraints.The structures are superimposed in the central region corresponding to residues 8 -17.The structure with the lowest NOE violations is shown in red.