Prion disease susceptibility is affected byβ-structure folding propensity and localside-chain interactions in PrPM. Qasim Khana,1, Braden Sweetingb,1, Vikram Khipple Mulligana, Pharhad Eli Arslanb, Neil R. Cashmanc,Emil F. Paia,b,d, and Avijit Chakrabarttya,b,2

aCampbell Family Institute for Cancer Research, Ontario Cancer Institute/University Health Network, Department of Biochemistry, University of Toronto,Toronto Medical Discovery Tower 4-307, 101 College Street, Toronto, ON, Canada M5G 1L7; bCampbell Family Institute for Cancer Research, OntarioCancer Institute/University Health Network, Department of Medical Biophysics, University of Toronto, Toronto Medical Discovery Tower 4-305, 101College Street, Toronto, ON, Canada M5G 1L7; cBrain Research Centre, Division of Neurology, Department of Medicine, University of British Columbiaand Vancouver Coastal Health, University of British Columbia Hospital, 2211 Wesbrook Mall, Vancouver, BC, Canada V6T 2B5; and dCampbell FamilyInstitute for Cancer Research, Ontario Cancer Institute/University Health Network, Department of Molecular Genetics, University of Toronto, TorontoMedical Discovery Tower 5-358, 101 College Street, Toronto, ON, Canada M5G 1L7

Edited by Robert Baldwin, Stanford University, Stanford, CA, and approved September 24, 2010 (received for review April 19, 2010)

Prion diseases occur when the normally α-helical prion protein (PrP)converts to a pathological β-structured state with prion infectivity(PrPSc). Exposure to PrPSc from other mammals can catalyze thisconversion. Evidence from experimental and accidental transmis-sion of prions suggests that mammals vary in their prion diseasesusceptibility: Hamsters and mice show relatively high susceptibil-ity, whereas rabbits, horses, and dogs show low susceptibility.Using a novel approach to quantify conformational states of PrPby circular dichroism (CD), we find that prion susceptibility trackswith the intrinsic propensity of mammalian PrP to convert fromthe native, α-helical state to a cytotoxic β-structured state, whichexists in a monomer–octamer equilibrium. It has been controversialwhether β-structured monomers exist at acidic pH; sedimentationequilibrium and dual-wavelength CD evidence is presented for anequilibrium between a β-structured monomer and octamer in someacidic pH conditions. Our X-ray crystallographic structure of rabbitPrP has identified a key helix-capping motif implicated in the lowprion disease susceptibility of rabbits. Removal of this cappingmotif increases the β-structure folding propensity of rabbit PrP tomatch that of PrP from mouse, a species more susceptible to priondisease.

protein misfolding ∣ X-ray crystallography ∣ Creutzfeldt–Jakob disease ∣bovine spongiform encephalopathy ∣ folding intermediates

Prion diseases are a group of fatal, neurodegenerative diseasesthat include Creutzfeldt–Jakob disease (CJD), Gerstmann–

Sträussler–Scheinker syndrome, fatal familial insomnia, andkuru in humans, scrapie in goats and sheep, bovine spongiformencephalopathy (BSE) in cattle, and chronic wasting disease inelk and deer (1). According to the “protein-only” hypothesis,prion diseases are caused by the misfolding of the prion protein(PrP) into a conformation that is pathogenic and infectious (2, 3).PrP exists normally in a monomeric, mostly α-helical state (PrPC),and upon prion infection it refolds into an aggregation-prone,mostly β-structured state that is infectious (PrPSc).

Although it is difficult to rank the susceptibility of differentspecies to contract prion disease with the limited data available,certain trends are clear. Previous studies of prion transmission inhamsters indicated that this species is highly susceptible to priondiseases, for hamsters develop prion disease when inoculatedwith various prion isolates from different animal donors, includ-ing humans, cows, sheep, mice, mink, and other hamsters (4–7).Mice show comparable prion disease susceptibility (4, 5, 7–10),but rabbits to date have not succumbed to prion disease despitebeing inoculated with human, sheep, and mouse prions and canresist prion infection for at least 3 y (4, 11). Furthermore, reportsfrom the BSE crisis in the United Kingdom indicate clear differ-ences in BSE susceptibility among larger mammals. Humans and

many feline species, including cheetahs, pumas, and domesticcats, were susceptible to the BSE agent, whereas no cases of priondisease were reported in canine or equine species (12, 13).However, because of a lack of experimental data, it is unclearhow dogs or horses would fare when challenged with a varietyof prion isolates.

Although understanding the determinants of prion susceptibil-ity would yield insights into the mechanism of pathogenesis inprion diseases, the varying susceptibilities of mammalian specieshas yet to be explained mechanistically. It has been known for anumber of years that PrP can form a β-structured state (β-state)under slightly destabilizing conditions (14–16). Hornemann andGlockshuber (15) proposed that the β-state was a monomericfolding intermediate that lay between the native and unfoldedstate on the PrP folding pathway. Later, Baskakov et al. (17) pro-vided evidence that the β-state was composed of oligomeric formsof PrP and suggested that these assemble after the PrP moleculescompletely unfold. Subsequent studies showed the β-structuredoligomers to be predominantly octameric for both mouse andhamster PrP (18, 19). Whether β-structured monomers of PrPexist is controversial. Although some studies have reported theexistence of β-structured monomers (20), other studies suggestthe opposite (17). Results reported in the present study demon-strate that slight changes in solution conditions can greatly influ-ence the degree to which β-structured monomers are populated(vide infra). The proposal that the β-state of PrP in general playsa role in the mechanism underlying prion diseases (21) raisesthe possibility that the propensity to form the β-state is a majordeterminant of prion disease susceptibility. To date, a quantita-tive measure of the propensity of PrPs to populate β-structuredmonomers and octamers has not been available.

We have developed a two-wavelength method of circulardichroism (CD) analysis to quantify the propensity of PrP totransform into the β-state and have applied this, along withanalytical ultracentrifugation experiments reporting on the oligo-meric state of PrP, to compare hamster, mouse, rabbit, horse, and

Author contributions: M.Q.K., B.S., V.K.M., E.F.P., and A.C. designed research; M.Q.K., B.S.,and P.E.A. performed research; M.Q.K., B.S., V.K.M., P.E.A., N.R.C., E.F.P., and A.C. analyzeddata; and M.Q.K., B.S., V.K.M., E.F.P., and A.C. wrote the paper.

Conflict of interest statement: N.R.C. is a founder and Chief Scientific Officer of AmorfixLimited, an early stage theragnostic company.

This article is a PNAS Direct Submission.

Data deposition: Crystallography, atomic coordinates, and structure factors have beendeposited in the Brookhaven Protein Data Bank, www.pdb.org (PDB ID code 3O79).1M.Q.K. and B.S. contributed equally to this work.2To whom correspondence should be addressed. E-mail: chakrab@uhnres.utoronto.ca.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1005267107/-/DCSupplemental.

19808–19813 ∣ PNAS ∣ November 16, 2010 ∣ vol. 107 ∣ no. 46 www.pnas.org/cgi/doi/10.1073/pnas.1005267107

dog PrP. Using urea and acid treatment to induce structuraltransitions, we find that the PrP proteins from these five speciesunder specific conditions can each populate four distinct confor-mational states: native monomers, unfolded monomers, and aβ-structured state (β-state) that can be monomeric or octameric.We find that the β-state propensity of PrP from these speciesvaries with their prion disease susceptibility. Using this two-wavelength CD method in conjunction with X-ray crystallogra-phy, we have been able to identify a key helix-capping motif thatcontrols the formation of the β-state, which may govern thesusceptibility to prion disease.

ResultsUrea Denaturation of Hamster PrP 90-231 Produces α-Helical, β-Struc-tured, and Unfolded States. Purified, recombinant, hamster PrP(9.5 μM) encompassing residues 90–231 (hamster PrP 90-231)was incubated in urea (0–9 M, pH 4, 25 °C, 5 d) and CDmeasure-ments were made at 220 nm. When plotted as a function of ureaconcentration, the mean residue ellipticity of hamster PrP 90-231shows a biphasic transition (Fig. 1A). The CD spectrum ofhamster PrP 90-231 measured at 0, 2.5, and 7.5 M urea (pH 4),respectively, displays a typical α-helical structure, a β-sheet-richstructure, and an unfolded structure (Fig. 1B). By relating thethree unique CD spectra of hamster PrP 90-231 in Fig. 1B to thebiphasic unfolding curve in Fig. 1A, it is apparent that ureadoes not cause unfolding of the native state to a partially foldedintermediate state, which then completely unfolds. Instead, mod-erate amounts of urea at pH 4 cause the native α-helical state ofhamster PrP 90-231 to convert to a very different, nonnative,β-sheet-rich conformation, and addition of higher amounts ofurea results in complete unfolding of this unique β-sheet-richstate. We denote this β-sheet-rich conformation as β-state PrP.

A Monomer–Octamer Equilibrium best Describes the OligomerizationState of Hamster PrP 90-231 in the β-State. In usual practice, dena-turation curves that display a characteristic biphasic shape arefit to a three-state transition model to determine the thermody-namic stabilities of the three stably populated species. However,these models usually assume that the protein remains monomericat all denaturant concentrations. Therefore, before fitting ourCD data we determined the oligomerization states of native,β-state, and unfolded PrP.

Sedimentation data indicated that native (no urea, pH 4.0)and unfolded PrP (7.5 M urea, pH 4.0) were monomeric witha molecular mass of 16 kDa. However, sedimentation analysis in-dicated that β-state PrP self-associates in a monomer–octamerequilibrium, which was confirmed by analysis of the residualsof the fits to various monomer–oligomer models (SI Appendix,Fig. S1 A and B). Size exclusion chromatography (SEC) (SIAppendix, Fig. S1C) supports the results from sedimentationequilibrium ultracentrifugation. Thus, β-state PrP, unlike nativeor unfolded PrP, exists as an equilibrium mixture of two species:monomers and octamers.

A Novel Two-Wavelength Method of CD Analysis Allows Measurementof Fractional Concentrations of β-State PrP. The finding that β-statePrP exists as a mixture of monomers and octamers indicates that asimple three-state model is inadequate to fit the biphasic, ureaunfolding curves of PrP. Analysis of the data using a four-stateequilibrium model is also not straightforward because varioussequential or branched models are equally probable and cannotbe discerned from the urea unfolding curves. Furthermore, thegreater complexity of the four-state model, with its highernumber of parameters, increases the difficulty in differentiatinglocal goodness-of-fit minima in the parameter landscape from theglobal minimum. For these reasons, we decided to develop anapproach that does not make presumptions about the oligomer-ization state. Our method takes advantage of the fact that β-statePrP possesses a β-sheet CD spectrum that is distinct from boththe α-helical native and random coil-like unfolded PrP spectra(Fig. 1B). Because the CD spectrum is a sum of the CD contribu-tions from all protein molecules present in their particular con-formations, our observed CD signal (θ220;obs) measured from asample can be expressed as follows:

θ220;obs ¼ ½θ�220;Native½Native� þ ½θ�220;β-state½β-state�þ ½θ�220;Unfolded½Unfolded�: [1]

In the above, ½θ�220;Native, ½θ�220;β-state, and ½θ�220;Unfolded repre-sent the molar ellipticities of native state, β-state, and unfoldedstate PrP at 220 nm, respectively. The above expression can besimplified by normalizing the CD data to fraction apparent valueswhere the native state and unfolded state have values of 1 and 0,respectively:

Fapp220 ¼ ðFNativeÞ þ z220ðFβ-stateÞ: [2]

Here, Fapp220 is the observed normalized CD signal, z220 is thenormalized CD signal for β-state PrP at 220 nm, and FNative andFβ-state are fractional concentrations of the native and β-state PrP,respectively.

To solve for Fβ-state, PrP unfolding may be monitored at a sec-ond wavelength, allowing generation of a system of two equa-tions. In order to solve for Fβ-state, the normalized CD signalzλ for the β-state at the second wavelength would need to bedifferent from z220 and also distinct from the native and theunfolded CD signals. On the basis of the spectra in Fig. 1B, awavelength of 229 nm fulfills these criteria, yielding the followingsecond equation:

Fapp229 ¼ ðFNativeÞ þ z229ðFβ-stateÞ: [3]

Combining Eqs. 2 and 3 and solving for Fβ-state yields

Fβ-state ¼ ðFapp220 − Fapp229Þ∕ðz220 − z229Þ: [4]

Fig. 1. Hamster PrP 90-231 adopts three distinct sec-ondary structural states. (A) Urea-unfolding curve ofhamster PrP 90-231 at pH 4 monitored by CD at220 nm. (B) CD spectra of hamster PrP 90-231 inthe native (○), β-state (▪), and unfolded (□) confor-mations at pH 4.

Khan et al. PNAS ∣ November 16, 2010 ∣ vol. 107 ∣ no. 46 ∣ 19809

BIOCH

EMISTR

Y

Thus, by monitoring urea unfolding of PrP at both 220 and229 nm, the relative fraction of β-state PrP can be calculatedat any given urea concentration and pH value. Note that thistwo-wavelength method treats the CD spectrum of the native,β-state, and unfolded PrP as a unique signature of that particularstate. Procedures that deconvolve CD spectra into percent helix,percent β-sheet, and percent random coil were not used.

Urea Treatment of PrP Causes Conversion of PrP to the β-State at lowpH. Fig. 2 shows the normalized urea unfolding curves of hamsterPrP 90-231 at pH values of 7.0, 5.0, 4.5, and 4.0. The CD signalswere monitored at wavelengths of 220 and 229 nm, and the datawere normalized to Fapp. A relationship is observed between theshape of the urea unfolding curves and the difference betweenthe normalized CD signals at the two wavelengths of 220 and229 nm. At pH 7.0, the urea unfolding curves are monophasicin shape and the curves at the two different wavelengths aresuperimposable (Fig. 2A). At pH 5.0, some biphasic shape andsome differences between the curves monitored at two wave-lengths are detectable (Fig. 2B). At pH values of 4.0 and 4.5,the two wavelengths generate strikingly different biphasic curves(Fig. 2 C and D). Because unfolding curves measured at the twowavelengths do not overlay at intermediate urea concentrationsand low pH, this gives a qualitative indication that β-state PrPmust be significantly populated. Quantification of the amount ofβ-state PrP is possible with our two-wavelength method.

The Fractional Concentration of β-State PrP from Hamster, Mouse,Rabbit, Horse, and Dog PrP 90-231 Correlates with Prion Susceptibility.Mouse, rabbit, horse, and dog PrP 90-231 formed monomer–octamer equilibrium mixtures at pH 3.5–4.0 and 2–6 M urea thatwere similar to hamster PrP 90-231 (SI Appendix, Fig. S2). Urea-induced unfolding of PrP 90-231 constructs of each of these spe-cies at pH values of 7.0, 5.0, 4.5, and 4.0 was monitored by CDmeasurements at 220 and 229 nm (SI Appendix, Figs. S3 and S4).A clear gradation is observed between the five species. Differ-ences between the two wavelength curves are greatest with ham-ster followed by mouse and then by rabbit, horse, and dog, thusproviding an index of β-state concentrations in the five species.Importantly, refolding experiments of these PrP proteins demon-strate that, at pH values of 7.0 and 4.0, unfolding and refolding ofthe proteins are completely reversible, indicating that equilibriumhas been achieved (SI Appendix, Fig. S5).

The value of z220–z229 was calculated at pH values thatrendered a biphasic curve for the PrP constructs; this value

was 0.144� 0.020 and was the same, within error, for all PrPproteins examined.

By using a value of 0.144 for z220–z229, the fractional concen-trations of β-state PrP were calculated for each of the five speciesat various urea concentrations and pH values (Fig. 3). At pH 7.0,none of the five PrP proteins populates the β-state (Fig. 3A).At pH 4.0, all five species populate the β-state fraction to somedegree (Fig. 3D). At pH 5.0 (Fig. 3B) and 4.5 (Fig. 3C), they showvarying fractional concentrations of the β-state. The propensity toform the β-state is greatest for hamster PrP, followed by mousePrP, and then by rabbit, horse, and dog PrP. This propensity cor-relates with prion disease susceptibility in these mammals.

Comparison of the Fraction Octamer with the β-State Fraction: Detec-tion of a Monomeric, β-State PrP Species. The fractional concentra-tions of octamer in the various samples can be determined fromsedimentation equilibrium data from hamster, mouse, and rabbitPrP 90-231 by fitting the sedimentation data to a double-exponential equation (SI Appendix, Fig. S6). The fraction octa-mer was calculated at urea concentrations of 2.5, 3.2, 4.0, and4.5 M for hamster PrP, at 3.7, 4.3, and 5.0 M urea for mousePrP, and at 3.0, 3.7, and 4.4 M urea for rabbit PrP 90-231, allat pH 4.0 (SI Appendix, Fig. S7). It is clear that the fractionalconcentration of β-state PrP is significantly higher than thefractional concentration of the octamer. The maximum fractionoctamer reaches a value of 70% for hamster PrP (SI Appendix,Fig. S7A), 60% for mouse PrP (SI Appendix, Fig. S7B), and50% for rabbit PrP (SI Appendix, Fig. S7C). Under these condi-tions the fractional concentration of β-state PrP is 100% for eachof these species. Because sedimentation equilibrium and SECanalysis indicated that only monomers and octamers are present,we conclude that the β-state PrP fraction is a mixture of β-struc-ture-containing monomers and octamers for hamster, mouse, andrabbit PrP 90-231.

To further demonstrate the existence of β-structured mono-mers, hamster PrP 90-231 was diluted into 3.6 M urea, pH 4.0and secondary structure and oligomerization status were exam-ined by CD and SEC, respectively. The data show that the reduc-tion in the levels of monomers and the appearance of octamersoccur over a period of hours (SI Appendix, Fig. S8A); however,formation of β-sheet structure is complete at the starting point of

Fig. 2. Urea-unfolding curves of hamster PrP 90-231 monitored by CD at 220(▴) and 229 nm (○) at pH (A) 7.0, (B) 5.0, (C) 4.5, and (D) 4.0. The lines areintended as a guide to the eye.

Fig. 3. The propensity to populate β-state PrP correlates with species sus-ceptibility to prion disease. Comparison of the β-state PrP fraction betweenhamster (red line), mouse (green line), rabbit (blue line), horse (dark yellowline), and dog PrP 90-231 (purple line) as a function of urea concentrationat pH (A) 7.0, (B) 5.0, (C) 4.5, and (D) 4.0. The lines are intended as a guideto the eye.

19810 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1005267107 Khan et al.

the experiment (SI Appendix, Fig. S8 B and C). Thus the mono-mers that are well-populated at early time points are β-structured.Furthermore, the monomeric peak of hamster PrP 90-231was collected from the size exclusion column, and the CD spectraof the sample indicated that monomers of hamster PrP 90-231 in4.1 M urea, pH 4.5 were β-structured (SI Appendix, Fig. S8 D–F).SEC of the protein sample after performing the CD scan indi-cated that hamster PrP 90-231 remained monomeric during thistime period, thus providing direct evidence that the monomersare β-structured (SI Appendix, Fig. S8E).

β-State PrP from Mouse PrP 90-231 Is Cytotoxic. We performedcytotoxicity studies to confirm that β-state PrP has similar toxicityto oligomeric β-structured PrP examined previously (22, 23). Wehave observed that, once formed, β-state PrP remains stable insolution for at least one week, when the conditions are changedfrom 4.0 M urea, pH 4.0 to phosphate buffered saline, pH 7.4.This stability of β-state PrP allowed investigation of its cytotoxicproperties using differentiated PC12 cells, a common cell culturemodel of neurons. Native-state or β-state mouse PrP 90-231 wereadded to differentiated PC12 cells, and the effects of each on cellsurvival were monitored (Fig. 4). Bee venom mellitin, a knowncytotoxin, was used as a positive control. Whereas native-statePrP did not display any toxic properties, β-state PrP was as cyto-toxic as bee venom mellitin on a molar basis. These data indicatethat these β-state PrP preparations possess potent cytotoxic activ-ity under physiological conditions.

X-Ray Crystal Structure of Rabbit PrP 121-230. To find out whetherthe structure of natively folded rabbit PrP contains any uniquestructural elements that might contribute to its resistance to con-version into the β-state, we solved the X-ray crystal structure ofrabbit PrP 121-230 to 1.6-Å resolution (see SI Appendix, Table S1for data collection and refinement statistics) (Fig. 5). Althoughnumerous structures of PrP from various species have beensolved by nuclear magnetic resonance spectroscopy, to date X-raycrystal structures are available only for sheep PrP, human PrP (asa domain-swapped dimer), and human PrP bound to an antibody(24–26). These crystal structures are all determined to resolutionsof 2.0 Å or poorer.

The asymmetric unit of the rabbit PrP 121-231 crystal structurecontains two molecules. The two molecules associate closelyalong a twofold-symmetry axis that is almost parallel to helix 2(SI Appendix, Fig. S9). The rmsd between all backbone atomsin the two molecules is 0.38 Å, indicating that the folds of thetwo molecules are practically identical. They are also very similar

to those of other structurally known PrP proteins; for example,the rmsd between the backbone atoms of rabbit PrP 121-230and sheep PrPC is 0.80 Å (24).

The loop between β-strand 2 and helix 2 (residues 166–174),sometimes referred to as the β2–α2 loop or rigid loop (27), hasbeen implicated as a possible amyloidogenic motif that forms asteric zipper (28). In the rabbit PrP 121-230 structure, electrondensity for residues 165–175 is well defined (SI Appendix,Fig. S10), indicating that the region is clearly ordered. The main-chain carbonyls of P165 and V166 form hydrogen bonds withthe amides of Q168 and Y169, respectively, generating a short310 helix, as seen in previous structures (27, 29–31). The side chainsof D167 and Q168 are surface exposed, but that of V166 formshydrophobic interactions with Y218 of helix 3, possibly contribut-ing to the stability of this loop (SI Appendix, Fig. S11). The 310helix preceding the N terminus of helix 2 produces a small kinkin the loop before the 171-NQNS-174 sequence.

Side-Chain Interactions Between Residues 171 and 174 Form a Helix-Capping Motif That Affects β-State Propensity of Rabbit PrP. Theamino acid at position 174 (serine in rabbit PrP and asparaginein mouse PrP) is of particular interest because it is known to af-

Fig. 4. β-state PrP from mouse PrP 90-231 is cytotoxic. Toxicity of β-state PrPwhen exposed to neuronal-differentiated, PC-12 cells. Absorbance readingswere taken at 550 nm after performing the sulforhodamine B (SRB) cytotoxi-city assay to measure cell viability of cells exposed to no protein (*), β-statePrP of mouse PrP 90-231 (▴), native PrP (▪), and melittin (○). The SRB assaywas performed after 4 d cultivation of cells at 37 °C.

Fig. 5. The crystal structure of rabbit PrPC 126-230. The observedmonomericfold is similar to previously observed structures of the ordered domain ofPrPC. The β2–α2 loop is highlighted in the Inset box.

Fig. 6. The helix-capping motif in rabbit PrPC. Comparison of residues170–174 of the rigid loop from rabbit PrPC structures and the lowest energystructures from the hamster and mouse PrPC NMR structure ensembles.

Khan et al. PNAS ∣ November 16, 2010 ∣ vol. 107 ∣ no. 46 ∣ 19811

BIOCH

EMISTR

Y

fect prion susceptibility (32). In rabbit PrP 121-231, the side-chaincarbonyl of N171 forms a hydrogen bond with the backboneamide of S174 and, in turn, the side-chain hydroxyl of S174 bindsto the backbone carbonyl of N171 (Fig. 6). These reciprocal side-chain-to-backbone-hydrogen bonds, together with the flankinghydrophobic residues Y169 and F175, form a helix-capping motifsimilar to the so-called “hydrophobic staple” (33, 34), therebybestowing additional stability upon the N terminus of helix 2. TheS174N mutation in this helix-capping motif causes rabbit PrP121-231 to populate the β-state fraction in 4 M urea, pH 4.5,whereas the wild-type rabbit PrP 121-231 does not populate theβ-state under these conditions (SI Appendix, Fig. S12). Disruptionof this helix-capping motif increases the β-state propensity of rab-bit PrP to match that of PrP from mouse, a species more suscep-tible to prion disease.

Discussion.Previous studies have reported that PrP can populate β-struc-tured octameric states (18, 19), though there are conflicting re-ports on the existence of β-structured monomeric states (17, 20)(SI Appendix, Fig. S8). We believe that this discrepancy is causedby slight differences in solution conditions of the differentstudies. It is well-established that the conformation of PrP is ex-quisitely sensitive to pH, ionic strength, and denaturant concen-tration (17–20). A study by Baskakov et al. (17), which did notdetect a β-structured monomer, used the relatively strongly desta-bilizing condition of 200 mM NaCl, 5 M urea, and pH 3.6. An-other study by Gerber et al. (20), which did detect a β-structuredmonomer, used the less destabilizing condition of 200 mM NaCl,1 M urea, and pH 3.6. We have found that the β-structured mono-mer can be populated in 50 mM sodium acetate, 74 mM NaCl,3.6 M urea, and pH 4.0 (SI Appendix, Fig. S8). It appears that theβ-structured monomer can be populated only under moderatelydestabilizing conditions. We suggest that the β-structured mono-mer is transiently populated to much lower levels under otherconditions.

We have investigated whether the propensity to form theseβ-structured monomeric and octameric states could underlie sus-ceptibility to prion disease. The work presented here provides aunique method to quantify the populations of β-state equilibriumspecies under a broad range of conditions and to assess thepropensity for PrP from a given animal species to populate theβ-state, allowing interspecies comparisons. By using a two-wave-length CD method, combined with analytical ultracentrifugation,we were able to measure the fractional concentrations of allconformational states present at equilibrium for several animalPrP proteins. We found that there was a gradation of propensitiesto populate the monomeric and octameric β-states, with a des-cending rank order of hamster, mouse, rabbit, horse, and dog PrP.Because PrP from susceptible species (hamsters and mice) hadsignificantly higher β-state propensity than PrP from apparentlyresistant species (rabbits, horses, and dogs), this shows thatβ-state propensity varies with prion disease susceptibility.

The fact that PrP can populate a β-structured, nonnative statehas been known for a number of years (14–16), and some havesuggested that β-structured, oligomeric forms of PrP representthe more toxic and/or infectious form underlying prion diseases(21, 35). Much has been done to map the pathway of conversion,but historically, it has been difficult to quantify the relative abun-dances of potentially disease-relevant folding intermediates. Ithas been shown by stopped-flow techniques that prion disease-causing variants of human PrP have larger populations ofpartially folded, kinetic intermediates than the wild-type protein(36). Our finding that the β-state propensity at equilibrium trackswith prion disease susceptibility across several animal specieslends support to the notion that β-state monomers and/or octa-mers play a prominent role in the mechanism of prion disease.

Contributions of Sequence Versus Structure to Prion Disease Suscept-ibility. Early experimental findings on prion susceptibility sug-gested that primary sequence similarity between the prionsfrom the donor animal and the PrP from the recipient animalplays an important role in determining the outcome of priontransmission (37). For example, wild-type mice show low suscept-ibility to hamster prions, yet transgenic mice expressing hamsterPrP are much more susceptible (38). However, there have beencases where primary sequence similarity does not determine amammal’s susceptibility to prion strains. Bank voles are moresusceptible to human prions than to hamster or mouse prions,despite the fact that there is greater sequence similarity betweenbank vole PrP and hamster or mouse PrP than between bank volePrP and human PrP (39). Also, variant CJD isolates from humansinfect wild-type mice more readily than the same isolates infecttransgenic mice expressing human PrP (40). Furthermore, thefact that many prion strains exist for a single sequence of PrP,and exhibit varying biochemical, histopathological, and neuro-pathological characteristics when introduced in the same speciesof mammal, suggests that prion susceptibility involves more thanjust primary sequence similarity between donor prions and hostPrP (41).

Our finding that PrP proteins from hamsters, mice, rabbits,horses, and dogs display a gradation of resistance to adoptingthe β-state, which correlates with prion susceptibility in thesemammals, suggests that the conformational malleability of PrP,which is encoded in the primary sequence, is a determinant ofprion susceptibility.

Structural Features of Native Rabbit PrP That May Impede Conforma-tional Transition into the β-State. Studies have shown that by expos-ing scrapie-infected mouse neuroblastoma cells to molecules thatstabilize PrPC, such as antibodies (26) or chemical chaperoneslike trimethylamine N-oxide and dimethylsulfoxide (42), one hin-ders the conformational transition into the PrPSc form, enablingimproved survival when cultured cells are exposed to infectiousscrapie material. Thus, stabilizing certain aspects of the nativestructure of PrPC hinders the conformational transition intothe PrPSc form. From our studies it appears that the differencein the β-state propensity of the five species studied resides withinthe covalent structure of PrP. Therefore, certain atomic interac-tions in the natively folded rabbit PrP structure may stabilizesegments of secondary structure that reduce conversion to theβ-state.

Structural studies on PrPC from a variety of mammalianspecies indicate that single amino acid changes do not have dras-tic structural effects on the overall fold. Rather, variation in localinteractions in the PrPC monomer may affect the mechanism ofconversion to the infectious form. For example, whereas mouseneuroblastoma cells expressing wild-type PrP are susceptible toscrapie infection, an N174S mutant of mouse PrP (the analogousresidue in rabbit PrP) is not (32). In our high-resolution crystalstructure of rabbit PrPC 121-231, N171 and S174 are part of ahydrophobic staple-like helix-capping interaction at the N termi-nus of α-helix 2 (Fig. 6). Such an arrangement is not present inany other known PrP structure. Our demonstration that disrup-tion of this motif through the S174N mutation increases theβ-state propensity of rabbit PrP (SI Appendix, Fig. S12) to matchthat of mouse PrP implicates this motif as a determinant of prioninfectivity.

Concluding Remarks.Our current findings demonstrate that the propensity to formβ-state PrP is a valid marker of prion disease susceptibility inhamsters, mice, rabbits, horses, and dogs. This relationship islikely to apply to other species. β-state propensity measurementswere used in conjunction with high-resolution structural techni-ques to identify key amino acid side-chain interactions that affect

19812 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1005267107 Khan et al.

the conformational transition between helical and β-structuredstates of PrP—namely the hydrophobic staple capping motif.The methods employed here have the potential to be used to testthe efficacy of compounds that may reduce the effective β-statepropensity of PrP, and which could therefore be of therapeuticbenefit in prion disease.

Materials and MethodsDetails of the protein preparation, CD spectroscopy, analytical ultracentrifu-gation, cytotoxicity assay, and protein crystallization and structure determi-nation are provided in the SI Appendix.

Urea Unfolding of PrP 90-231 Monitored by CD at 220 and 229 nm. CD measure-ments were obtained at wavelengths of both 220 and 229 nm; 100 measure-ments at a rate of 1 measurement per second were averaged. For each pH,equilibrium curves were normalized according to Santoro and Bolen (43). Thefraction β-state PrP at each urea concentration was determined by applyingEq. 4. The z parameters in Eq. 4 represent the normalized CD signal of β-statePrP at wavelengths 220 and 229 nm; the values of these critical points weredetermined graphically (44).

Determination of the Fraction Octamer. Sedimentation data collected fromhamster, mouse, and rabbit PrP 90-231 at pH 4.0 in urea concentrationsranging from 2.5 to 5.0 M were fit to a double-exponential model, wherethe absorbance of the PrP sample equals the summation of the absorbanceof the monomer and the absorbance of the octamer, as a function of radius:

AbsorbanceðPrP sampleÞ ¼ AbsorbanceðMonomerÞþAbsorbanceðOctamerÞ; [5]

AbsorbanceðPrP sampleÞ ¼ A � expðk � radius2Þ þ B

� expð8 � k � radius2Þ; [6]

where A and B represent proportionality constants and k ¼ ½ω2∕ð2RTÞ��½M � ð1 − νρÞ�; here, ω represents the angular velocity of the rotor (in radiansper second), R represents the gas constant (8.314 � 107 erg∕mol � K), T repre-sents the temperature in kelvin, M represents the gram molecular weight ofthe protein, ν represents the partial specific volume of the protein, and ρ

represents the density of the solvent. For each sample, the fractionoctamer was determined at the radius value at which the sedimentation ab-sorbance readings at speeds ranging from 10,000 to 15,000 rpm intersectedwith the absorbance readings at 3,000 rpm. Because the absorbance readingsat 3,000 rpm represent the absorbance of the sample before protein sedi-mentation has occurred, the total PrP concentration at this radius equalsthe concentration prior to any sedimentation. The fraction octamer was cal-culated by using 3–6 separate datasets.

ACKNOWLEDGMENTS. B.S. was the recipient of an Ontario Graduate Scholar-ship. V.K.M. was the recipient of a Canadian Institutes of Health Research(CIHR) Frederick Banting and Charles Best Canada Graduate ScholarshipDoctoral Award. Part of the research described in this paper was performedat the Canadian Light Source, which is supported by Natural Sciences andEngineering Research Council of Canada, National Research Council, CIHR,and the University of Saskatchewan. This study was supported in part by agrant from PrioNet Canada and by the Ontario Ministry of Health and LongTerm Care (MOHLTC). The views expressed do not necessarily reflect those ofthe Ontario MOHLTC.

1. Collins SJ, Lawson VA, Masters CL (2004) Transmissible spongiform encephalopathies.Lancet 363:51–61.

2. Griffith JS (1967) Self-replication and scrapie. Nature 215:1043–1044.3. McKinley MP, Bolton DC, Prusiner SB (1983) A protease-resistant protein is a structural

component of the scrapie prion. Cell 35:57–62.4. Gibbs CJ, Jr, Gajdusek DC (1973) Experimental subacute spongiform virus encephalo-

pathies in primates and other laboratory animals. Science 182:67–68.5. Thomzig A, et al. (2006) Pathological prion protein in muscles of hamsters and mice

infected with rodent-adapted BSE or vCJD. J Gen Virol 87:251–254.6. Kimberlin RH, Walker C (1977) Characteristics of a short incubation model of scrapie in

the golden hamster. J Gen Virol 34:295–304.7. Bessen RA, Marsh RF (1992) Identification of two biologically distinct strains of

transmissible mink encephalopathy in hamsters. J Gen Virol 73:329–334.8. Lasmezas CI, et al. (1997) Transmission of the BSE agent to mice in the absence of

detectable abnormal prion protein. Science 275:402–405.9. Chandler RL (1961) Encephalopathy in mice produced by inoculation with scrapie

brain material. Lancet 1:1378–1379.10. Hill AF, et al. (2000) Species-barrier-independent prion replication in apparently

resistant species. Proc Natl Acad Sci USA 97:10248–10253.11. Barlow RM, Rennie JC (1976) The fate of ME7 scrapie infection in rats, guinea-pigs and

rabbits. Res Vet Sci 21:110–111.12. Aldhous P (1990) BSE: Spongiform encephalopathy found in cat. Nature 345:194.13. Kirkwood JK, Cunningham AA (1994) Epidemiological observations on spongiform

encephalopathies in captive wild animals in the British Isles. Vet Rec 135:296–303.14. Zhang H, et al. (1997) Physical studies of conformational plasticity in a recombinant

prion protein. Biochemistry 36:3543–3553.15. Hornemann S, Glockshuber R (1998) A scrapie-like unfolding intermediate of the

prion protein domain PrP(121-231) induced by acidic pH. Proc Natl Acad Sci USA95:6010–6014.

16. Swietnicki W, Morillas M, Chen SG, Gambetti P, Surewicz WK (2000) Aggregation andfibrillization of the recombinant human prion protein huPrP90-231. Biochemistry39:424–431.

17. Baskakov IV, Legname G, Prusiner SB, Cohen FE (2001) Folding of prion proteinto its native alpha-helical conformation is under kinetic control. J Biol Chem276:19687–19690.

18. Baskakov IV, et al. (2002) Pathway complexity of prion protein assembly into amyloid.J Biol Chem 277:21140–21148.

19. Sokolowski F, et al. (2003) Formation of critical oligomers is a key event duringconformational transition of recombinant syrian hamster prion protein. J Biol Chem278:40481–40492.

20. Gerber R, Tahiri-Alaoui A, Hore PJ, James W (2008) Conformational pH dependence ofintermediate states during oligomerization of the human prion protein. Protein Sci17:537–544.

21. Collinge J, Clarke AR (2007) A general model of prion strains and their pathogenicity.Science 318:930–936.

22. Novitskaya V, Bocharova OV, Bronstein I, Baskakov IV (2006) Amyloid fibrils ofmammalian prion protein are highly toxic to cultured cells and primary neurons. J BiolChem 281:13828–13836.

23. Simoneau S, et al. (2007) In vitro and in vivo neurotoxicity of prion protein oligomers.PLoS Pathog 3:e125.

24. Haire LF, et al. (2004) The crystal structure of the globular domain of sheep prionprotein. J Mol Biol 336:1175–1183.

25. Knaus KJ, et al. (2001) Crystal structure of the human prion protein reveals amechanism for oligomerization. Nat Struct Biol 8:770–774.

26. Antonyuk SV, et al. (2009) Crystal structure of human prion protein bound to atherapeutic antibody. Proc Natl Acad Sci USA 106:2554–2558.

27. Gossert AD, et al. (2005) Prion protein NMR structures of elk and of mouse/elk hybrids.Proc Natl Acad Sci USA 102:646–650.

28. Sawaya MR, et al. (2007) Atomic structures of amyloid cross-β spines reveal variedsteric zippers. Nature 447:453–457.

29. Liu H, et al. (1999) Solution structure of Syrian hamster prion protein rPrP(90-231).Biochemistry 38:5362–5377.

30. Christen B, Perez DR, Hornemann S, Wüthrich K (2008) NMR structure of the bank voleprion protein at 20 degrees C contains a structured loop of residues 165-171. J Mol Biol383:306–312.

31. Christen B, Hornemann S, Damberger FF, Wuthrich K (2009) Prion protein NMRstructure from tammar wallaby (Macropus eugenii) shows that the beta2-alpha2 loopis modulated by long-range sequence effects. J Mol Biol 389:833–845.

32. Vorberg I, Groschup MH, Pfaff E, Priola SA (2003) Multiple amino acid residueswithin the rabbit prion protein inhibit formation of its abnormal isoform. J Virol77:2003–2009.

33. Seale JW, Srinivasan R, Rose GD (1994) Sequence determinants of the capping box, astabilizing motif at the N-termini of alpha-helices. Protein Sci 3:1741–1755.

34. Aurora R, Rose GD (1998) Helix-capping. Protein Sci 7:21–38.35. Silvera JR, et al. (2005) The most infectious prion protein particles. Nature

437:257–261.36. Apetri AC, Surewicz K, Surewicz WK (2004) The effect of disease-associated mutations

on the folding pathway of human prion protein. J Biol Chem 279:18008–18014.37. Prusiner SB, et al. (1990) Transgenetic studies implicate interactions between

homologous PrP isoforms in scrapie prion replication. Cell 63:673–686.38. Scott M, et al. (1989) Transgenic mice expressing hamster prion protein produce

species-specific scrapie infectivity and amyloid plaques. Cell 59:847–857.39. Nonno R, et al. (2006) Efficient transmission and characterization of Creutzfeldt-Jakob

disease strains in bank voles. PLoS Pathog 2:e12.40. Hill AF, et al. (1997) The same prion strain causes vCJD and BSE. Nature 389:448–450.41. Beringue V, Vilotte JL, Laude H (2008) Prion agent diversity and species barrier.

Vet Res 39:47.42. Tatzelt J, Prusiner SB, Welch WJ (1996) Chemical chaperones interfere with the

formation of scrapie prion protein. EMBO J 15:6363–6373.43. Santoro MM, Bolen DW (1988) Unfolding free energy changes determined by the

linear extrapolation method. 1. Unfolding of phenylmethanesulfonyl alpha-chymo-trypsin using different denaturants. Biochemistry 27:8063–8068.

44. Barrick D, Baldwin RL (1993) Three-state analysis of sperm whale apomyoglobinfolding. Biochemistry 32:3790–3796.

Khan et al. PNAS ∣ November 16, 2010 ∣ vol. 107 ∣ no. 46 ∣ 19813

BIOCH

EMISTR

Y

Prion disease susceptibility is affected by β-structure folding propensity and local side-chain interactions in PrP

M. Qasim Khan, Braden Sweeting, Vikram Khipple Mulligan, Pharhad Eli Arslan, Neil R. Cashman, Emil F. Pai and Avijit Chakrabartty

Supporting Information

Fig. S1. The β-State of hamster PrP 90-231 exists as an equilibrium between monomers and octamers (a) Goodness-of-fit (variance) of sedimentation equilibrium data of hamster PrP 90-231 analyzed by various self-association models in 2.5 M (▲) and 3.2 M (○) urea at pH 4. (b) Sedimentation equilibrium data of hamster PrP 90-231 in 3.2 M, pH 4.0 at 15,000 rpm, fit to a monomer-octamer equilibrium model (lower panel). Residuals to the fit of the data to a monomer-octamer model (upper panel) (c) Size exclusion chromatography of hamster PrP 90-231 in 3.6 M urea at pH 4.0.

Fig. S2. (a) Goodness-of-fit (variance) of equilibrium sedimentation data of mouse (▲) and rabbit (○) PrP 90-231 to various self-association models in 5.0M and 4.0 M urea, respectively, at pH 4. Sedimentation data of mouse (b) and rabbit (c) PrP 90-231 at 15,000 rpm fitted to a monomer-octamer equilibrium model. (d) Goodness-of-fit (variance) of equilibrium sedimentation data of horse (▲) and dog (○) PrP 90-231 to various self-association models in 2.33 M urea at pH 3.5. Sedimentation data of horse (e) and dog (f) PrP 90-231 at 15,000 rpm fitted to a monomer-octamer equilibrium model.

Fig. S3. Urea-unfolding curves monitored by circular dichroism at 220 nm (▲), and 229 nm (○) at pH 7.0, 5.0, 4.5, and 4.0 of mouse PrP 90-231 (top panels) and rabbit PrP 90-231 (bottom panels). The lines are intended as a guide to the eye.

Fig. S4. Urea-unfolding curves monitored by circular dichroism at 220 nm (▲), and 229 nm (○) at pH 7.0, 5.0, 4.5, and 4.0 of horse PrP 90-231 (top panels) and dog PrP 90-231 (bottom panels). The lines are intended as a guide to the eye.

Fig. S5. Urea-unfolding (○) and refolding (●) at pH 7.0, and urea-unfolding (□) and refolding (■) at pH 4.0 for (a) hamster, (b) mouse and (c) rabbit PrP 90-231. Circular dichroism data was collected at 220 nm.

Fig. S6. Determining octamer fraction: (a) Sedimentation data of hamster PrP 90-231 (2.5M Urea, pH 4.0) collected at 10,000 rpm (○) fitted to a monomer-octamer equilibrium model (solid curve). The dashed line represents the radius at which the data collected at 3,000 rpm (■) intersect with the fitted data at 10 000 rpm. (b) The monomer (□) and octamer (▲) absorbance fractions can be plotted as a function of radius from the fitted data in (a). The dashed line represents the radius at which the octamer fraction is calculated. At this radius the total PrP concentration equals the initial concentration in the cell before sedimentation.

Fig. S7. The β-State PrP fraction is a mixture of β-structure-containing monomers and octamers. Comparison of the β-State PrP fraction (●) with the octamer fraction (black bars) as a function of urea concentration at pH 4.0 for (a) hamster, (b) mouse and (c) rabbit PrP 90-231. The lines are intended as a guide to the eye. The error bars represent the standard error based on 3 to 6 measurements of the fraction octamer.

Fig. S8. Direct evidence of a β-structured monomer (a) The percentage of the area of both monomer (○) and octamer (■) peaks were plotted as a function of time after diluting folded hamster PrP 90-231 into 3.6M urea, pH 4.0 buffer. The lines are intended as a guide to the eye. (b) The CD signal at a wavelength of 218 nm was used to monitor the CD changes over the course of 5 hours. (c) CD spectra of hamster PrP 90-231 diluted in 3.6 M urea, pH 4.0 at time zero (■) and after 48 hours (○). (d) Elution profile of hamster PrP 90-231 showing octamer and monomer peaks at equilibrium in 4.1 M urea, pH 4.5 buffer. (e) Size exclusion chromatrography of the monomer after collecting and re-applying the monomer peak in (d) to the size exclusion column. (f) CD spectra of monomeric hamster PrP 90-231 (○) from the same protein sample that was applied to the column in (e), and CD spectra of the monomer-octamer mixture (■) from the same protein sample that was applied to the column in (d).

Fig. S9. Observed dimeric arrangement in the asymmetric unit of wild-type rabbit PrP 121-231 crystal lattice. The two molecules are not domain swapped and closely associate between the helix 2 – loop – helix 3 region of the C-terminal half of the ordered domain.

Fig. S10. Omit FO-FC electron density (at 2.0σ) for resides V166-N168 (left) and Y169-S174 (right) of the β2-α2 loop. Residues 166-168 are well ordered forming a 310 helical loop and show good electron density over the main and side chains. Residues 169-174 also show good electron density except for S170 for which no side chain electron density was observed. Electron density of S174 shows that the side-chain hydroxyl points inward toward the backbone, making a hydrogen bond with backbone amide.

Fig. S11. Long rang contacts between the β2-α2 loop and Helix-3. V166 and F175 make hydrophobic contacts with Y218. Polar contacts are also formed between the backbone of helix 3 and N172.

Fig. S12. Comparison of the β-State PrP fraction between rabbit PrP 121-231 (○) and the S174N mutant (▲) as a function of urea concentration at pH (a) 5.5 and (b) 4.0. The lines are intended as a guide to the eye.

Supplementary Table 1Summary of crystallographic data collection and model refinement statistics.

Data StatisticsSpace group P212121

Cell constants a (Å) 29.5b (Å) 86.2c (Å) 87.1

Resolution (Å) 30 – 1.6 (1.7 - 1.6)Overall reflections 214 933 (35 458)Unique reflections 30 148 (4 929)Redundancy 7.1 (7.2)Completeness (%) 99.6 ( 99.6)R merge 7.0 (24.6)<I>/σI 18.35 (7.71)

Refinement StatisticsFinal R Cryst (%) 16.1R-free (%) 21.8Solvent (%) 44.7No. of molecules 2No. of all atoms 1902No. of water molecules 192No. of sodium ions 4No. of Chloride ions 2Average B-factor (Å2) 19.60Ramachandran plot

Favored 99.5%Allowed 0.5%

R.M.S from ideal valuesLengths 0.022Angles 1.790

SI MethodsProtein Expression and Purification.

DNA constructs of golden Syrian hamster (hamster), mouse, and rabbit PrP 23-231 were generous gifts from M. Coulthart (Health Canada, Winnipeg, MA, Canada), D. Westaway and J. Watts (University of Alberta, Edmonton, AB, Canada), and S. Priola (Rocky Mountain Laboratories, Hamilton, MT, USA), respectively. Horse and dog PrP 23-231 were generous gifts from J. Castilla (The Scripps Research Institute, Jupiter, FL, USA). The DNA sequences coding for hamster, mouse , rabbit, horse and dog PrP 90-231 were cloned and expressed using the pProEX-Htb plasmid (Invitrogen, Burlington, ON, Canada) with an N-terminal 6xHis-tag and a Tobacco Etch Virus (TEV) protease cleavage site. Rabbit PrP121-231 was cloned and expressed using pET28a (Novagen, Gibbstown, NJ, USA) with an N-terminal 6xHis-tag and a thrombin protease cleavage site.

All PrP proteins were expressed as inclusion bodies using E. coli BL21 AI cells (Invitrogen) and refolded and purified according to Zahn et al. (45), with slight modifications. For PrP 90-231, the His-tag was cleaved with His-tagged TEV protease (Invitrogen) and both the His-tag and TEV were removed with Ni-NTA (Qiagen, Mississauga, ON, Canada). For PrP 121-231, anion exchange chromatography was performed using a HiTrap Q-sepharose 5/5 anion exchange column on an ÄKTA FPLC system (GE Bioscience, Mississauga, ON, Canada).CD Spectroscopy

CD spectroscopy was performed on an Aviv circular dichroism spectrometer model 62DS. All CD measurements were taken using a 1 mm quartz cuvette and a spectral bandwidth of 1 nm at 25 °C. Far-UV CD spectra were averaged from three wavelength scans from 250 nm to 190 nm with a 5 second averaging time, and blanked with respective buffers.Unfolding and refolding of PrP 90-231 monitored by CD

Stocks (95 μM) of folded PrP 90-231 were dialyzed into the following buffers that were of identical ionic strength, each with 5 mM EDTA; pH 7.0 (50 mM sodium phosphate), pH 5.0 (50 mM sodium acetate, 67 mM NaCl), pH 4.5 (50 mM sodium acetate, 70 mM NaCl), pH 4.0 (50 mM sodium acetate, 74 mM NaCl). Equilibrium samples were prepared by diluting the PrP stock 10-fold to 9.5 μM into buffered solutions containing urea at concentrations ranging from 0 M to 9 M and incubated for 5 days at room temperature before making CD measurements.

Unfolded PrP stocks (95 μM) were prepared by denaturing folded PrP in 9.9 M urea in pH 7.0 and pH 4.0 buffers using 3,500 MWCO dialysis cups (Pierce, Rockford, Illinois). Urea-refolding samples were prepared by diluting the PrP stock 10-fold (to a final concentration of 9.5 μM) into pH-buffered solutions containing urea to final concentrations ranging from 1 M to 9 M urea. Samples were incubated for 5 days at room temperature. The CD signal for each sample was the mean of 100 separate 1s readings at a wavelength of 220 nm. Analytical Ultracentrifugation

Equilibrium sedimentation experiments were performed on hamster, mouse and rabbit PrP 90-231 at pH 4.0 in urea concentrations ranging from 2.5

M to 5.0 M, and horse and dog PrP 90-231 at pH 3.5 in urea concentrations ranging from 1.33M to 3.67M, at protein concentrations of 9.5 µM. Samples were loaded into six-sectored cells with quartz windows using rotor speeds of 3,000, 10,000, 15,000, 20,000, 25,000, 30,000, 35,000 and 40,000 rpm at 25 °C. For each speed, 5 replicates of data were collected every 0.001 cm at a wavelength of 280 nm after 18 to 34 hours. Molecular weight determinations involved global analysis of data acquired at 5 different speeds using Beckman XL-I software, where absorbance versus radial position data were fitted to the sedimentation equilibrium equation using non-linear least squares techniques. The partial specific volume and density of the sample were calculated using the program SEDNTERP from the amino acid sequence and buffer composition, respectively.Size Exclusion Chromatography.

Hamster PrP 90-231 at 9.5 μM in 3.6 M Urea, 50 mM sodium acetate pH 4.0, 80 mM NaCl was incubated at room temperature for a minimum of 72 hours. An aliquot of 300 μL was then injected onto a Superdex 200 10/30 column (GE Bioscience) equilibrated in the same buffer as above. The flow rate was 0.25 mL/min and the column elution was monitored by absorbance at 280 nm using an ÄKTA FPLC system at room temperature.Sulforhodamine B Cytotoxicity Assay.

Pheochromocytoma (PC-12) cells were plated at 1,000 cells per well in a 96-well plate and incubated in 60 ng/mL NGF diluted in N2 supplement/DMEM (Gibco BRL). Cells were differentiated for four days. Samples (0 to 9 µM) of β-state mouse PrP90-231, native state mouse PrP90-231, and bee venom mellitin were prepared by diluting concentrated stock solutions into cell culture medium and then dialyzing the samples against large volumes of cell culture media in a 3,500 MWCO dialysis cup to remove urea from the samples. After differentiation, the medium was exchanged with the PrP/mellitin samples and the cells were incubated for 4 days at 37 ºC. Toxicity was assayed using the sulforhodamine B (SRB) assay. Briefly, cells were fixed with 8.5% trichloroacetic acid for 30 minutes. The plates were washed with H20 and air-dried. Protein was stained with 0.4% SRB (Molecular Probes Inc.) in 1% acetic acid for 30 minutes. Plates were washed with 1% acetic acid and air-dried. The dye was extracted in 10 mM Tris (pH 9.0) and absorbance was assayed at 550 nm on a plate reader.Crystallization of rabbit PrP 121-231 and Data Collection

Initial screens using Crystal Screens I & II (Hampton Research, Aliso Viejo, CA, USA) yielded showers of crystalline plates in 100 mM Hepes pH 7.5, 4.3 M NaCl. Further optimization and microseeding gave single, plate-like crystals (300 μm x 300 μm x 30 μm) that grew overnight in 100 mM Tris pH 7.0, 2.5 M NaCl and diffracted to a resolution of 1.6 Å. Diffraction data were collected from a single crystal flash-frozen at 100 K in 100 mM Tris, pH 6.5, 2.5 M NaCl, and 30 % v/v glycerol as cryoprotectant at a wavelength λ = 0.97934 Å using a Marmosaic CCD300 at the CMCF-I beamline at the Canadian Light Source Synchrotron (Saskatoon, SK, Canada).Data Processing, Refinement and Model Building

Data were processed with the program package XDS (46). Statistics for data collection and unit cell parameters are summarized in Table 1. The structure

was solved applying molecular replacement techniques contained in the program PHASER (47) with sheep PrP (1UW3) as a search model. The model was refined without NCS with the ML-method using REFMAC (48) and TLS refinement (49) assisted by TLSMD (50). Model building was done using Coot (51). Summary of refinement statistics are given in Table 1. Atomic coordinates have been deposited in the Brookhaven Protein Data Bank (PDB ID: 3O79).

SI References

45. Zahn R, von Schroetter C, Wüthrich K (1997) Human prion proteins expressed in Escherichia coli and purified by high-affinity column refolding. FEBS Lett 417:400-404.

46. Kabsch W (1993) Automatic processing of rotation diffraction data from crystals of initially unknown symmetry and cell constants. J Appl Cryst 26: 795-800.

47. McCoy AJ, et al. (2007) Phaser crystallographic software. J Appl Cryst 40: 658-674.

48. Murshudov GN, Vagin AA, Dodson EJ (1997) Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr D Biol Crystallogr 53:240-255.

49. Winn MD, Murshudov GN, Papiz MZ (2003) Macromolecular TLS refinement in REFMAC at moderate resolutions. Methods in Enzymology 374:300-321.

50. Painter J, Merritt EA (2006) TLSMD web server for the generation of multi-group TLS models. Acta Crystallogr D Biol Crystallogr 62:439-450.

51. Emsley P, Cowtan K (2004) Coot: model-building tools for molecular graphics. Acta Crystallogr D Biol Crystallogr 60:2126-2132.