DR

KAa

b

c

d

a

ARR2AA

KPPSDV

s

1

aoltmc(

bc

0h

Vaccine 32 (2014) 1988–1997

Contents lists available at ScienceDirect

Vaccine

j o ur na l ho me page: www.elsev ier .com/ locate /vacc ine

evelopment of a Multivalent, PrPSc-Specific Prion Vaccine throughational Optimization of Three Disease-Specific Epitopes

risten Marciniuka,b, Pekka Määttänenb, Ryan Taschukb,c, T. Dean Aireyd,ndrew Potterb, Neil R. Cashmand, Philip Griebelb,c, Scott Nappera,b,∗

Department of Biochemistry, University of Saskatchewan, Saskatoon, Saskatchewan Canada S7 N 5E5, CanadaVaccine and Infectious Disease Organization, University of Saskatchewan, Saskatoon, Saskatchewan S7 N 5E3, CanadaSchool of Public Health, University of Saskatchewan, Saskatoon, Saskatchewan Canada S7 N 5E5, CanadaVancouver Coastal Health Research Institute, University of British Columbia, Vancouver, BC V5Z 3P1, Canada

r t i c l e i n f o

rticle history:eceived 3 September 2013eceived in revised form3 December 2013ccepted 14 January 2014vailable online 29 January 2014

eywords:rion diseasesrotein Misfolding

a b s t r a c t

Prion diseases represent a novel form of infectivity caused by the propagated misfolding of a self-protein(PrPC) into a pathological, infectious conformation (PrPSc). Efforts to develop a prion vaccine have beencomplicated by challenges and potential dangers associated with induction of strong immune responsesto a self protein. There is considerable value in the development of vaccines that are specifically targetedto the misfolded conformation. Conformation specific immunotherapy depends on identification andoptimization of disease-specific epitopes (DSEs)1 that are uniquely exposed upon misfolding. Previously,we reported development of a PrPSc-specific vaccine through empirical expansions of a YYR DSE. Here wedescribe optimization of two additional prion DSEs, YML of �-sheet 1 and a rigid loop (RL) linking �-sheet2 to �-helix 2, through in silico predictions of B cell epitopes and further translation of these epitopes into

Sc

elf-Epitopeisease-Specific Epitopesaccine

PrP -specific vaccines. The optimized YML and RL vaccines retain their properties of immunogenicity,specificity and safety when delivered individually or in a multivalent format. This investigation supportsthe utility of combining DSE prediction models with algorithms to infer logical peptide expansions tooptimize immunogenicity. Incorporation of optimized DSEs into established vaccine formulation anddelivery strategies enables rapid development of peptide-based vaccines for protein misfolding diseases.

© 2014 Elsevier Ltd. All rights reserved.

Published with permission of the Director of VIDO as journaleries number: 674

. Introduction

Transmissible spongiform encephalopathies (TSEs) represent novel form of infectious disease mediated by the misfoldingf a normal cellular protein (PrPC) into an infectious, patho-ogical conformation (PrPSc) [1,2]. PrPSc serves as a templateo convert PrPC into PrPSc in an autocatalytic, self-propagating

anner [3]. Prion diseases affect a number of domestic speciesausing scrapie in sheep, bovine spongiform encephalopathyBSE) in cattle and chronic wasting disease (CWD) in cervids

Abbreviations: DSE, disease specific epitope; RL, rigid loop; TSE, transmissi-le spongiform encephalopathy; BSE, bovine spongiform encephalopathy; CWD,hronic wasting disease; SC, subcutaneaously.∗ Corresponding author. Tel.: +306 966 1546.

E-mail address: scott.napper@usask.ca (S. Napper).

264-410X/$ – see front matter © 2014 Elsevier Ltd. All rights reserved.ttp://dx.doi.org/10.1016/j.vaccine.2014.01.027

[4,5]. Human prion diseases include Creutzfeldt–Jakob dis-ease (CJD), Gerstmann–Straussler–Sheinker syndrome, and fatalfamilial insomnia [6]. Currently prion diseases have a fataloutcome in all species with no effective treatment options[7].

While the threat of BSE was successfully addressed throughimproved management practices, efforts to control CWD throughwildlife intervention strategies have been less effective, highlight-ing the need for novel disease management tools. That antibodiesagainst PrPC offer degrees of protection in in vitro and in vivo mod-els offers proof-of-principle support for the development of a prionvaccine. One of the challenges for prion vaccine development isovercoming tolerance to PrPC. Several investigations have demon-strated induction of immune responses to PrPC using a varietyof carrier systems and adjuvants, most of which are impracticalfor humans or livestock [8–14]. Further, due to the ubiquitous

expression of this cell surface protein, the generation of circu-lating PrPC antibodies may have adverse consequences [15–18].As such a prion vaccine would ideally be specific for PrPSc [19].Specific targeting of PrPSc requires the identification of protein

ccine

rt

etidoaat

pipptbp

YTgoppsutieispf

2

2

2

egastBetcoefo

2c

saLw

K. Marciniuk et al. / Va

egions that are uniquely exposed in the misfolded conforma-ion.

Previously, Cashman et al. reported a YYR motif that wasxposed upon experimental misfolding of PrPC [20,21]. Whilehis region had considerable appeal as a vaccine target, it’s weakmmunogenicity, even with the use of potent formulation andelivery systems, limited its translation to a vaccine [20]. Devel-pment of an immunogenic YYR DSE based PrPSc-vaccine waschieved through optimization of the YYR epitope through trialnd error expansion of this core sequence within the context ofhe prion protein sequence [22].

Mapping of the disease-specific YYR motif to �-strand 2,rompted examination of the opposing �-strand, leading to the

dentification of a second prion DSE designated YML [23]. A thirdrobable prion DSE, a loop between �-strand 2 and �-helix 2, wasroposed using an algorithm designed to identify regions of pro-eins most likely to unfold [24]. The loop displays unusual rigidityy nuclear magnetic resonance studies in cervid species when com-ared to other prion susceptible mammals [25,26].

In this report we describe translation of the previously identifiedYR, YML, and RL prion DSEs into highly immunogenic vaccines.he generated YYR, YML and RL vaccines exhibit strong immuno-enicity, specificity and safety profiles when delivered individuallyr in a multivalent format. Using our prion vaccine developmentipeline as a model, we also describe a method for optimization ofeptide epitopes for development of peptide-based vaccines. Ourtrategy involved high throughput in silico analysis of each DSEsing an algorithm that identifies putative B cell epitopes, in ordero identify immunogenic expansions of these core epitopes thatncorporate native sequence patterns consistent with strong B-cellpitopes. Using this approach, the DSEs were rapidly translated intommunogenic peptide-based vaccines capable of inducing PrPSc-pecific immune responses. This method is broadly applicable torotein misfolding diseases, once conformation specific epitopesor a protein of interest have been identified.

. Materials and Methods:

.1. DSE Vaccine Production

.1.1. In Silico Epitope OptimizationPreviously identified DSE sequences were optimized through

xpansion of the core sequence to include endogenous immuno-enic B-cell epitopes. First, a panel of all theoretical expansionsround the DSE core was generated in silico, in the context of theurrounding PrP protein sequence, for up to ten residues in bothhe N and C terminal directions. These sequences were scanned for-cell epitopes by the approach of Larson et al. [27]. Expansions thatxhibited consistently higher immunogenicity prediction scoreshroughout the expansion of the core epitope, while adhering toonformation specific restrictions, were selected for vaccine devel-pment. In the context of the final recombinant carrier proteinpitopes are in a forward-back-back presentation that is repeatedour times [22]. In the analysis for B-cell epitopes, a double repeatf this forward-back-back motif was considered.

.1.2. Construction and purification of Leukotoxin-fusedonstructs

Genes corresponding to the optimized epitopes were synthe-

ized by Genscript (Piscataway, NJ) and sub-cloned to be expresseds C-terminal fusions to the highly immunogenic carrier proteineukotoxin (Lkt) [22,28]. The resulting Lkt recombinant fusionsere produced in E. coli BL21 as described [29].

32 (2014) 1988–1997 1989

2.2. Vaccine formulation and delivery:

2.2.1. MiceC57Bl6 or Balb/c mice (n = 8) were injected subcutaneously

(SC) three times at three week intervals with 10 �g of leukotoxinrecombinant fusion formulated with 30% Emulsigen-D (MVP Tech-nologies, Omaha, NE) for a final injection volume of 100 �l. pervaccine dose. Mice, at 5-6 weeks of age, received 3 injections of thevaccine on days 0, 21 and 42. Injections were administered betweenshoulder blades to mid back (dorsum) using a 25 gauge needle 5/8′′

long or 22 gauge 1′′ needle depending on the viscosity of the injec-tion. Serum samples were obtained on days 0, 21, 28, 42, 49, and70.

2.2.2. SheepSuffolk sheep of mixed sex were randomly assigned to 5 groups

of n = 8, and injected SC with 50 �g of Lkt recombinant fusion pre-pared in PBS and 30% Emulsigen-D in an injection volume of 1 mL.Vaccines were administered 3 times at 6-week intervals. Injectionswere performed at the lateral cervical area in a triangle boundedby the shoulder, dorsum of the neck, and the lateral processes ofthe cervical spine using 20 gauge, 1.5 inch needles with needlesplaced in a tenting manner. All animal work was performed by anindependent research team and experiments were done accordingto the Guide to the Care and Use of Experimental Animals, providedby the Canadian Council on Animal Care. All experimental protocolswere approved by the Saskatchewan Animal Care Committee.

2.3. ELISAs

The epitope, carrier, and PrPC specific antibody titres inserum were quantified by ELISA, as described [22]. To detectpeptide-specific antibody responses, peptides consisting of a sin-gle forward-back-back repeat motif for each DSE sequence weresynthesized as previously described [22].

2.4. Statistical Analysis:

2.4.1. Identification of significant differences between vaccineinduced antibody titres

The antibody titres in animals taken over time did not adhere toa normal distribution. To account for the repeated measures studydesign, data for each animal were first summed over time. The datasums were then ranked to account for their non-normal distribu-tion and a one-way ANOVA analysis was performed on the rankedsums. Where appropriate, Tukey’s test was used to examine the dif-ferences among the groups. P values less than 0.05 were consideredsignificant. All analysis met the assumptions of ANOVA.

2.4.2. Regression analysis comparing in silico predicted andin vivo DSE immunogenicity

The relationship between peak median antibody titres (wk 7) inmice for each epitope and the average immunogenicity predictionscores generated by BepiPred was examined using simple linearregression analysis. The epitopes included in the analysis repre-sent all tested expansions of the YYR, YML, and RL core epitopes,with varying degrees of in silico and in vivo immunogenicity. Priorto performing the analysis the antibody titres were log transformed

to yield a normal distribution. The normality of both variables wasverified using the Shapiro-Wilk test. All analysis met the assump-tions of regression analysis. P values less than 0.05 were consideredsignificant.

1990 K. Marciniuk et al. / Vaccine

Table 1Peptide Epitope Sequences. Constructs were generated as C-terminal fusions torecombinant Leukotoxin protein. Each arrow represents a single repeat of the DSEin either a linear (→) or inverted (←) presentation.

Construct Design and Notation Peptide Sequence Epitope Presentation

�2(2 + YYR + 9)I QVYYRPVDQYSNQN (→←←)4

�2(2 + YYR + 8)I QVYYRPVDQYSNQ (→←←)4

�2(2 + YYR + 7)I QVYYRPVDQYSN (→←←)4

�2(2 + YYR + 6)I QVYYRPVDQYS (→←←)4

�2(2 + YYR + 5)I QVYYRPVDQY (→←←)4

�2(2 + YYR + 4)I QVYYRPVDQ (→←←)4

2

wcb

2

Petr

2

npldefaiaa

3

3

soYasstd

tedsTe5er

�2(2 + YYR + 3)I QVYYRPVD (→←←)4

.5. Antibody Purification and Immunoprecipitation of PrPSc

Serum from immunized sheep were evaluated for interactionith PrPSc and PrPC. Immunoglobulin isolated using Protein-A

olumn-affinity purification was conjugated to magnetic beads forrain homogenate immunoprecipitation assays as described [20].

.6. Proteinase K Resistance to Detect PrPSc Formation

The ability of PrPSc specific antibodies to cause misfolding ofrPC into PrPSc, as indicated by proteinase K resistant species, wasvaluated in the brains of vaccinated animals, and by incubation ofhe polyclonal antibodies with brain homogenates using previouslyeported methodologies [21,30].

.7. Immunohistochemical staining for PrPSc

Immunohistochemical staining was conducted at Prairie Diag-ostic Services (Saskatoon, SK) using the Benchmark staininglatform (Ventana Medical Systems, Tuscon, AZ) and an HRP-

abelled multimer detection system (BMK Ultraview DAB Paraffinetection kit, Ventana Medical Systems, Tuscon, AZ). Heat-inducedpitope retrieval consisted of applying CC1 extended incubationollowed by Protease 3 for two minutes (these and other reagentsre proprietary and included in proprietary kit from Ventana Med-cal Systems Inc.). The Mouse anti-TSE clone F99/97.6.1 primaryntibody (VMRD Inc, Pullman, WA) was applied for 32 minutes at

dilution of 1:1500.

. Results

.1. Optimization of the YYR Epitope

Previously, through optimization of the sequence, as well asystems of formulation and delivery, we reported developmentf a PrPSc- specific vaccine based on the weakly immunogenicYR epitope [22]. Development of this PrPSc-specific vaccine wasn empirical process based on consideration of epitope expan-ions that would avoid surface exposure within PrPC [22]. Whileuccessful, a more systematic, high-throughput methodology forranslating multiple DSE targets into peptide-based vaccines isesirable.

As a first step in establishing a pipeline for DSE optimizationhe most immunogenic expansion identified from our previousfforts, 2 + YYR + 9 [22], was used to create a panel of epitope candi-ates based on systematic C-terminal truncations. All the peptideequences used in this investigation are presented in [Table 1].his panel of different vaccine epitopes generated markedly differ-

nt antibody responses in mice with epitope-specific titres ranging0-fold [Figure 1A]. There was no significant correlation betweenpitope length and the magnitude of epitope-specific antibodyesponses, the shortest peptide, 2 + YYR + 3, induced the highest

32 (2014) 1988–1997

peak antibody responses. The magnitude of the response to thisepitope was significantly (p < 0.01) higher than for the longerepitopes [Figure 1B]. Small differences in epitope sequence sig-nificantly influenced immunogenicity, to the extent that thesedifferences defined the presence or absence of a substantial anti-body response. For example, the 2 + YYR + 4 construct elicitedhigher antibody titres than the 2 + YYR + 5 construct (p > 0.01), andan increase in the proportion of responders from 38% to 100%; aconsequence of the addition of a single amino acid [Figure 1A and1B].

3.2. Optimization of the YML and RL Epitopes

Previously, hypothesis-based and thermodynamic computa-tional approaches for predicting protein misfolding suggested twoadditional prion DSEs corresponding to the YML sequence of �-sheet 1 and a rigid loop linking �-sheet 2 to �-helix 2. The sequencesand positioning of the three described prion DSEs (YYR, YML andRL), with respect to the primary and tertiary structure of PrPC, arepresented in [Supplementary Figure 1].

Preliminary iterations of the YML epitope, produced using thescreening method described for YYR, lacked sufficient immuno-genicity to induce an antibody response. As such we adopted a moretargeted approach for epitope optimization based on considerationof expansions to include natural B-cell epitopes. We have adaptedthe publicly available program created by Larsen et al. for opti-mization of peptide epitopes. The program is designed to analyze acomplete protein sequence to identify peptide sequences that cor-relate highly with characteristics of immunogenic B-cell epitopes.Our approach involves analysis of peptide epitopes previously iden-tified as exhibiting conformational specificity. The optimizationinvestigations with the YYR epitope demonstrated the dramaticeffect single amino acid alterations can have on the in vivo immuno-genicity of a peptide epitope. Thus, the basis of our expansionmethod is to analyze all combinations of single amino acid addi-tions to the N and C-terminus of the core epitope, known tobe disease specific. This panel of epitopes is then screened forthe inclusion of native sequence motifs characteristic of B-cellepitopes, indicating the potential for enhanced immunogenicityand greater vaccine utility. Using this approach the expansion(1 + YML + 7) was predicted to be highly immunogenic whereasprevious non-optimized expansions corresponding to 2 + YML + 9and 9 + YML + 2 were predicted to be weakly immunogenic. Con-sistent with these predictions the 1 + YML + 7 vaccine induced thehighest antibody responses, while vaccines predicted to have poorimmunogenicity induced significantly lower antibody responses(p < 0.001) with 100 to 1000-fold differences in antibody responses[Figure 1C]. Notably, optimization of the YML epitope improvedthe proportion of animals that generated an antibody responsefrom 0% to 100%, based on a response titre threshold of 1000[Figure 1D]. In the case of the RL DSE a single peptide epitope,2 + RL + 4, was selected for vaccine production based on its highpredicted immunogenicity. Consistent with the predictive power ofthe B-cell epitope algorithm, this epitope was highly immunogenic[Figure 1E].

3.3. Sequence Determinants of Immunogenicity

A number of parameters can be used to predict the surfaceaccessibility, and immunogenicity, of an epitope [27]. Considera-tion of sequences anticipated to be surface exposed is complicatedin the context of conformation specific targets, where epitope

selection is highly restricted. Instead, a more appropriate methodfor selection of highly immunogenic epitopes, in this context,involves an evaluation of potential DSE expansions for sequencesconsistent with the introduction of B-cell epitopes, indicative of

K. Marciniuk et al. / Vaccine 32 (2014) 1988–1997 1991

Figure 1. Immunogenicity of DSE vaccines in mice. C57BL6 or Balb/c mice (n = 8) were vaccinated three times at three week intervals with 10 �g of the indicated vaccinein 30% Emulsigen-D. Serum antibody titres were determined by peptide-capture ELISAs. Data is presented as median values. The dashed line indicates the threshold fora positive titre as greater than 1000. A) Comparison of median epitope-specific antibody titres for vaccines described in Table 1. B) Individual peak animal responses arepresented (week 7 from 1A) and significant differences (p < 0.01) among the treatment groups are indicated. (C-D) In silico optimization improves the immunogenicity of theYML construct. Comparison of C) median serum antibody titres throughout the 10-week time-course and D) peak serum antibody titres for each vaccine construct. E) Mediantitres for the optimized YYR, YML, and RL vaccines. F) Regression analysis of the correlation between in silico B-cell epitope prediction and in vivo immunogenicity. Averageprediction scores for all YYR, YML, and RL epitope expansions. The median antibody titres represent the peak titres at week 7 for each DSE vaccine. (R2 = 0.7426, P < 0.0002,P

eccsmsms0

earson r = 0.8617).

nhanced immunogenicity. We evaluated the utility of the B-ell epitope prediction algorithm in the optimization of DSEs byomparing the predicted immunogenicity of several DSE expan-ions, with varying predicted immunogenicity scores, to theagnitude of induced antibody responses in vivo. There is a

trong relationship between the predicted immunogenicity and theagnitude of the induced antibody responses [Figure 1F]. Regres-

ion analysis yielded an R2 value of 0.7426 with a P-value of.0002.

3.4. Univalent vs. Multivalent Vaccination

The immunogenicity of the optimized Lkt-YML, Lkt-YYR, andLkt-RL vaccines was assessed in a relevant large animal species.Groups of sheep (n = 8) received 3 injections at 6 week intervals

of either the Lkt-YML, Lkt-YYR, or Lkt-RL vaccines. Additionally,a multivalent vaccine, that includes all three optimized DSEs,was investigated. Two groups of sheep (n = 8) were vaccinatedwith the three DSE vaccines, either individually at separate sites

1992 K. Marciniuk et al. / Vaccine 32 (2014) 1988–1997

Figure 2. Immunogenicity of Various DSE-based Vaccines Administered Alone, Co-Administered and Co-Formulated. Sheep (n = 8) were immunized with 50 �g of Lkt-YYR,Lkt-YML, or Lkt-RL formulated in 30% Emulsigen-D and delivered as either a univalent, multivalent co-administered, or multivalent co-formulated vaccine, at the indicatedtime-points (↑). Peptide-specific serum antibody titres were determined by peptide-capture ELISAs. A) Individual animal titres. B) Comparison of median titres specific foreach peptide antigen when using different vaccine delivery and formulation strategies. C) Effect of vaccine formulation and delivery on peptide-specific antibody titresspecific for each DSE.

ccine

ocpi

seDheYtciecm

m

FrahtiE

K. Marciniuk et al. / Va

r co-formulated and injected at a single site. The separate ando-formulated groups were included to determine if there wereossible interactions among the three DSEs that may influence

mmunogenicity in a multivalent format.With each vaccination strategy consistent epitope-specific

erum antibody responses were observed in all animals butpitope-specific antibody titres varied significantly among theSEs [Figure 2A]. The RL-based vaccine consistently exhibited theighest immunogenicity. In the univalent vaccine format, the RLpitope induced stronger epitope-specific antibody responses thanYR (p < 0.05), that also appeared to be stronger than for YML,hough not with statistical significance. For both multivalent vac-ine immunization protocols, the RL epitope was statistically moremmunogenic than both YYR and YML (p < 0.001). The YML and YYRpitope-specific serum antibody titres were similar except when

omparing the co-formulated multivalent vaccine, where YYR wasore immunogenic than YML (p < 0.05) [Figure 2B].Differences in vaccine formulation and delivery (univalent vs

ultivalent) significantly altered antibody responses to each DSE

igure 3. Specificity of Immune Responses. A) ELISAs against PrPC. Serum samples colleecombinant sheep PrPC in ELISAs. PrPC specific sheep serum and week 0 pre-immune sernimals in the multivalent groups are highlighted. B) Antibodies generated against the oomogenate. Serum from sheep immunized with the YYR, YML, and RL univalent vaccines

o magnetic beads and incubated with non-infected and infected 10% brain homogenate. Mncluded as positive and negative controls, respectively. C) Multivalent DSE immune seruLISA PrPC positive and negative serum samples were included as controls.

32 (2014) 1988–1997 1993

(YYR, p < 0.0001; YML, p < 0.0001; RL, p < 0.04). Co-formulationresulted in a significant decrease in the antibody response gen-erated, when compared to the univalent (p < 0.01) and multivalentco-immunized (p < 0.001) vaccines. There was no significant dif-ference between antibody responses induced by either univalentor multivalent co-immunization. The magnitude of antibodyresponses induced by the RL vaccine were less sensitive tomanipulations in vaccine formulation or delivery. The multiva-lent co-immunization format generated an antibody responsesignificantly greater than the multivalent co-formulated vaccine(p < 0.05). However, this response was not significantly greater thanthe antibody response to the univalent vaccine and there was nosignificant difference between the univalent and the multivalentco-immunization protocols [Figure 2 C]. When comparing the twoapproaches to multivalent vaccine delivery, co-delivery at separate

sites induced significantly greater antibody responses for all threeDSEs than co-formulation. No significant difference was observedwhen vaccines were delivered individually, either through univa-lent or co-immunization strategies [Figure 2 C].

cted at the peak in antibody titre for each group were assessed for reactivity withum were included as positive and negative controls, respectively. The PrPC reactiveptimized YYR, YML, and RL epitopes immunoprecipitate PrPSc from infected brainwas assessed for reactivity with PrPSc and PrPC. Serum antibodies were cross-linked

agnetic beads coated with 6D11 monoclonal antibody or naïve sheep serum werem from animals showing PrPC reactivity in ELISAs (A) were tested in IPs as in (B).

1 ccine 32 (2014) 1988–1997

3

agptrvwmvTSoBp

eai[iiPo

3

tmftsgattavtboaPTdifawH2wppt

4

s[i

Figure 4. DSE immunization does not induce template-directed misfolding of PrPC

to PK resistant PrPSc. A) Antibody induced misfolding of PrPC in vivo. Proteinase K(PK) digests and western blots of obex (OB) and cerebellum (CER) samples fromsheep that received a 50 �g dose of each vaccine (SN25, SN15 and RLM) in separateformulations administered to separate sites 3 times following a 6 week vaccinationschedule, with an additional boost 2 weeks prior to tissue collection. Brain sampleswere collected 23 weeks after the first vaccination. -, no PK, + 20 �g/ml PK digestfor 1 hr at 37◦ C. PrP was detected with the monoclonal antibody 6H4 (Prionics,Switzerland). B) Antibody induced misfolding of PrPC in brain homogenates in vitro.PK digestion and western blots of 10% sheep brain homogenate following incuba-tion with pooled prebleed serum and group-based pooled post-vaccination serum

994 K. Marciniuk et al. / Va

.5. Antibody Specificity

It was important to verify that gains in immunogenicity were nott the expense of specificity. Antibody specificity was first investi-ated using ELISA to compare the reactivity of pre-immune andeak immune sera with recombinant PrPC. All animals receivinghe univalent DSE vaccines demonstrated no difference in PrPC

ecognition between the pre-immune and post-vaccination serumalidating that these targets retain conformational-specificityhen delivered individually [Figure 3A]. Interestingly, one ani-al in each of the co-immunized and co-formulated multivalent

accine groups displayed low-level responses to PrPC [Figure 3A].hese titres are above the pre-immune levels and are reproducible.ubsequently, samples representing the full time-course for eachf the positive animals were assessed for reactivity against PrPC.oth animals displayed PrPC reactivity throughout the trial, whicheaked following each immunization (data not shown).

The conformational specificity of the antibodies was furtherxamined by immunoprecipitations. For all three DSEs the serumntibodies generated in response to the univalent vaccines exhib-ted strong reactivity with PrPSc and no reactivity with PrPC

Figure 3B]. In addition, the two PrPC positive animals from the co-mmunized and co-formulated multivalent groups were screenedn IPs and serum samples from these positive animals did not bindrPC in brain homogenates, whereas strong PrPSc reactivity wasbserved [Figure 3 C].

.6. Antibody-induced Misfolding of PrPC

One concern in the generation of antibodies against PrPSc ishe possibility that these antibodies may initiate template-directed

isfolding of endogenous PrPC through stabilization of the mis-olded structure. The ability of the DSE based vaccines to facilitateemplate-directed misfolding was examined in vitro and in vivo. Noymptoms of scrapie were observed in the sheep in all vaccinatedroups up to collection, 23 weeks after their first vaccination. Obexnd cerebellum samples from three sheep co-immunized with allhree DSEs were assayed for the presence of Proteinase K (PK) resis-ant PrPSc. This vaccine group was selected as the animals received

higher total vaccine dose (50 �g x3/dose) compared to the uni-alent vaccine group (50 �g/dose), and also generated the highestotal PrPSc specific antibody responses. No PK resistant PrPSc coulde detected after 1 hr digestion at a relatively low concentrationf PK (20 �g/ml) [Figure 4A]. IHC examination of obex, cerebellum,nd rectal lymphoid follicles [Figure 5] coupled with ELISA tests forK resistant PrPSc in obex and cerebellum were confirmed negative.hese results indicate that an antibody response to all three DSEsid not induce formation of PrPSc in vivo. To determine if antibod-

es generated in response to the DSE vaccines could act as catalystsor the misfolding of PrPC to PrPSc in vitro [31,40,41], pre-bleednd post-vaccination serum samples from all vaccinated animalsere pooled separately and applied to ovine brain homogenates.omogenates and sera were incubated with shaking at 37 ◦C for4 hrs in an attempt to convert PrPC to PrPSc [41]. PK resistant PrPSc

as undetectable in homogenates treated with either the pooledre-bleed or post-vaccination sera [Figure 4B]. These results sup-ort the conclusion that PrPSc DSE-specific antibodies are unableo induce misfolding of PrPC in vivo or in vitro.

. Discussion

A number of groups have provided proof-of-principle evidenceuggesting immunotherapy may be of value for prion diseases8,13,31–33]. While encouraging, these investigations typicallynduce immune responses that do not discriminate the healthy and

obtained at 15 weeks after receiving 3 injections of the DSE vaccines in the univalent,multivalent co-administered, and multivalent co-formulated formats.

infectious conformations of the prion protein. While no adversereactions have yet to be reported as a consequence of PrPC reac-tive antibodies, there is evidence from in vitro investigations, aswell as other examples in which induction of immune responsesto a self-protein, in the context of therapeutic efforts, has hadadverse consequences [34]. Further, due to the inherent lack ofPrP immunogenicity, these efforts often require reagents that limitreal-world applicability. An ideal vaccine would induce immuneresponses specific for the pathological conformation employingstrategies of formulation and delivery that are consistent with costsand regulations associated with a livestock vaccine.

Previously, we translated a prion DSE (YYR) into a PrPSc-specificvaccine [22]. While achieving the primary objective, this investi-gation did not provide a defined protocol for translating DSEs intovaccines. Relative to the previous efforts of our group to optimizethe YYR epitope through trial-and-error the application of a definedstrategy of in silico analysis of potential epitope expansions rep-resents a significant advancement. Merging of computational andwet lab approaches is emerging as a powerful approach towardsrational development and optimization of vaccines [35,36]. Suchapproaches have the potential to accelerate the development ofvaccines for structurally complex proteins, in particular those that

serve as the basis of protein misfolding diseases or those wherestructural diversity is an inherent aspect of their function, such asintrinsically disordered proteins that are emerging as key factorsin many regulatory, signalling and control processes [37].

K. Marciniuk et al. / Vaccine 32 (2014) 1988–1997 1995

Figure 5. Immunohistochemical staining of obex (OB), cerebellum (CRBL), and rectal 2 lymphoid follicles (RLF) to detect PrPSc. Staining of tissues collected from sheep (n = 3)at week 23, following injections of the multivalent co-administered vaccine at weeks 0, 6, 12, and a 4 boost at week 21. Tissues were screened for the presence of PrPSc usingIHC by Prairie 5 Diagnostic Services (Saskatoon, SK). All tissues assayed were negative for PrPSc, the tissues 6 from one negative animal are shown (right panel). Staining ofPrPSc positive tissues from a 7 scrapie infected animal are included for comparison (left panel). Photos were taken through an 8 Olympus BX41 microscope with a flip-outcondenser and 20x or 40x UPlan Fluorite lens using a, 12 Megapixel Olympus DP71 Camera with DP controller acquisition and managing software, (Olympus).

olvrtfihema

ecvvce

In the current study, we demonstrate, through rational selectionf epitopes based on predicted immunogenicity, efficient trans-ation of two additional predicted prion DSEs into peptide-basedaccines. Using this approach, highly immunogenic sequences wereapidly identified and a truncated, more immunogenic version ofhe previously reported YYR optimized epitope was also identi-ed. It is clear that the immunogenicity of peptide epitopes isighly sensitive to minor sequence manipulations. As such, anffective, high-throughput system for the rational selection of opti-ized epitopes represents a significant advance over empirical

pproaches.While it is difficult to anticipate which, if any, of these DSE

pitopes has the greatest therapeutic potential as a prion vac-ine, a multivalent vaccine may be advantageous over a univalent

accine. Here we demonstrate the generation of a multivalentaccine capable of inducing responses to each target. Althougho-formulation generates a diminished antibody response toach individual DSE, the overall PrPSc specific antibody titres

are amplified relative to the univalent formulations. In addi-tion, the combination of antibodies reactive with three separateepitopes in a multivalent vaccine may lead to increased PrPSc

neutralization, despite the reduction in individual DSE antibodytitres. To avoid compromising immunogenicity, separate injectionsare recommended until co-formulation methods can be opti-mized.

The PrPSc specificity of the DSE antibodies was examined usingELISA and immunoprecipitation methods. We observed PrPC anti-body reactivity in two animals, one from the co-immunized, andone from the co-formulated group. PrPC reactivity could resulteither from one of the expanded epitopes, or a portion of one ofthe expanded epitopes, being surface exposed on PrPC, or throughepitope spreading from the PrPC-DSE regions to other portions of

the protein. In either scenario there is also the possibility that thevaccines, and the induced antibodies, may not function indepen-dently of each other at either a structural or immunological level.For instance, one antibody binding to its epitope may influence

1 ccine

tnmf

vwvPtsatTbwv

oecsaiiotm

C

L

A

glCa

A

f2

R

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

996 K. Marciniuk et al. / Va

he structure of the protein to reveal protein regions that wouldormally not be surface exposed [21]. Similarly, the presence ofultiple epitopes, representing a large portion of the protein, may

acilitate epitope spreading.The DSE antibodies generated following immunization with uni-

alent and multivalent formulations exhibited strong reactivityith PrPSc. In ELISA, DSE antibodies generated through univalent

accine immunization remain non-reactive with PrPC (n = 24), andrPC reactivity was only observed in one animal in each of the mul-ivalent DSE vaccine groups. However, this multivalent immuneerum PrPC reactivity was not observed in immunoprecipitationssays. Thus, it is unclear if co-immunization or co-formulation ofhe three DSE vaccines truly compromises PrPSc specificity in vivo.he reactivity against PrPC is confined to ELISA analysis with recom-inant protein and the antibodies remain non-reactive against PrPC

ithin the context of brain homogenate, a more biologically rele-ant protein sample.

The misfolding of self-proteins is also the basis for a varietyf other neurodegenerative diseases including Alzheimer’s Dis-ase, Parkinson’s Disease, and Amyotrophic Lateral Sclerosis. Theurrent strategy for inducing immune responses against crypticelf-epitopes may have far reaching applications for immunother-py of protein misfolding diseases. In particular, a rational pipelinen which predicted disease-specific epitopes can be optimized formmunogenicity and rapidly translated into established strategiesf formulation and delivery has significant potential to advancehe field of conformation-specific immunotherapy for protein-

isfolding diseases.

onflict of Interest

Neil Cashman is the Founder, CSO and Board Chair of Amorfixife Sciences.

cknowledgements

We would like to thank VIDO animal care, Hugh Townsend foruidance of statistical analysis, Chris Stuart from the Western Col-ege of Veterinary Medicine for help with microscopy, and Gordonrockford for technical assistance. This work was funded by PrioNetnd Alberta Prion Research Institute.

ppendix A. Supplementary data

Supplementary data associated with this article can beound, in the online version, at http://dx.doi.org/10.1016/j.vaccine.014.01.027.

eferences

[1] Griffith. J. Self-replication and scrapie. Nature 1967;215:1043–4.[2] Prusiner S. Novel proteinaceous infectious particles cause scrapie. Science

1982;216:136–44.[3] Aguzzi A, Sigurdson C, Heikenwaelder M. Molecular mechanisms of prion

pathogenesis. Annu Rev Pathol 2008;3:11–40.[4] Silveira JR, Caughey B, Baron GS. Prion protein and the molecular features of

transmissible spongiform encephalopathy agents. Curr Top Microbiol Immunol2004;284:1–50.

[5] Baeten LA, Powers BE, Jewell JE, Spraker TR, Miller MW. A Natural Case ofChronic Wasting Disease in a Free-ranging Moose (Alces alces shirasi). J Vet

Dis 2008;43:309–14.

[6] Collinge J. Prion diseases of humans and animals: Their causes and molecularbasis. Ann Rev Neurosci 2001;24:519–50.

[7] Geshwind MD. Clinical trials for prion disease: difficult challenges, but hopefor the future. Lancet Neurol 2009;8(4):304–6.

[8] Sigurdsson EM, Brown DR, Daniels M, Kascsak RJ, Kascsak R, Carp R, MeekerHC, Frangione B, Wisniewski T. Immunization delays the onset of prion diseasein mice. Am J Pathol 2002;161:13–7.

[

[

32 (2014) 1988–1997

[9] Hanan E, Priola SA, Solomon B. Antiaggregating antibody raised againsthuman PrP 106-126 recognizes pathological and normal isoforms of thewhole prion protein. Cellular and Molecular Neurobiology 2001;21(6):693–703.

10] Koller MF, Grau T, Christen P. Induction of antibodies against murine full-length prion protein in wildtype mice. J Neuroimmunol 2002;132(1-2):113–6.

11] Rosset MB, Ballerini C, Gregoire S, Metharom P, Carnaud C, AucouturierP. Breaking immune tolerance to the prion protein using prion proteinpeptides plus oligodeoxynucleotide-CpG in mice. J Immunol 2004;;172(9):5168–74.

12] Polymenidou M, Heppner FL, Pellicioli EC, Urich E, Miele G, Braun N, Wopfner F,Schatzl HM, Becher B, Aguzzi A. Humoral immune response to native eukaryoticprion protein correlates with anti-prion protection. Proc Natl Acad Sci U S A2004;101:14670–6.

13] Schwarz A, Kratke O, Burwinkel M, Riemer C, Schultz J, Henklein P, BammeT, Baier M. Immunization with a synthetic prion protein-derived peptide pro-longs survival times of mice orally exposed to the scrapie agent. Neurosci Lett2003;350:187–9.

14] Gilch S, Wopfner F, Renner-Muller I, Kremmer E, Bauer C, Wolf E, Brem G,Groschup MH, Schatzl HM. Polyclonal anti-PrP auto-antibodies induced withdimeric PrP interfere efficiently with PrPSc propagation in prion-infected cells.J Biol Chem 2003;278(20):18524–31.

15] Cashman NR, Loertscher R, Nalbantoglu J, Shaw I, Kascsak RJ, Bolton DC,Bendheim PE. Cellular isoform of the scrapie agent protein participates in lym-phocyte activation. Cell 1990;61(1):185–92.

16] Schneider B, Mutel V, Pietri M, Ermonval M, Mouillet-Richard S, KellermannO. NADPH oxidase and extracellular regulated kinases 1/2 are targets of prionprotein signaling in neuronal and nonneuronal cells.

17] Arsenault RJ, Li Y, Potter A, Griebel PJ, Kusalik A, Napper S. Induction ofligand-specific PrP (C) signaling in human neuronal cells. Prion 2012;6(5):477–88.

18] Solforosi L, Criado JR, McGavern DB, Wirz S, Sánchez-Alavez M, Sugama S,DeGiorgio LA, Volpe BT, Wiseman E, Abalos G, Masliah E, Gilden D, Oldstone MB,Conti B, Williamson RA. Cross-linking cellular prion protein triggers neuronalapoptosis in vivo. Science 2004;303(5663):1514–6.

19] Li L, Napper S, Cashman NR. Immunotherapy for Prion Disease: Opportunitiesand Obstacles. Immunotherapy 2010;2:269–82.

20] Paramithiotis E, Pinard M, Lawton T, LaBoissiere S, Leathers VL, Zou WQ, EsteyLA, et al. A prion protein epitope selective for the pathologically misfoldedconformation. Nature Medicine 2003;9(7):893–9.

21] Li L, Guest W, Plotkin SS, Cashman NR. Immunological mimicry of PrPC-PrPSC interactions: antibody-induced PrP misfolding. Protein Eng Des Sel2009;22(8):523–9.

22] Hedlin PD, Cashman NR, Li L, Gupta J, Babiuk LA, Potter AA, Griebel P, DesignNapper S. Delivery of a cryptic PrPC epitope for induction of PrPSc-specificantibody responses. Vaccine 2010;28:981–8.

23] Cashman NR. Antibodies and epitopes specific to misfolded prion protein EPPatent 2012, 2403871.

24] Cashman NR, Plotkin SS, Guest WC. Methods and systems for predicting mis-folded protein epitopes EP Patent 2011, 2342220.

25] Sigurdson CJ, Aguzzi A. Chronic wasting disease. Biochim Biophys Acta2007;1772:610–8.

26] Gossert AD, Bonjour S, Lysek DA, Fiorito F, Wuthrich K. Prion protein NMRstructures of elk and of mouse/elk hybrids. Proc Natl Acad Sci U S A 2005;102:646–50.

27] Larsen JEP, Lund O, Nielsen M. Improved method for predicting linear B-cellepitopes. Immunome Research 2006;2:2.

28] Manns J, Barker C, Attah-Poku S. The design, production, purification, and test-ing of a chimeric antigen protein to be used as an immunosterilant in domesticanimals. Can J Chem 1997;75:829–33.

29] Harland RJ, Potter AA, van. Drunen-Little-van den Hurk S, Van Donkersgoed J,Parker MD, Zamb TJ, Janzen ED. The effect of subunit or modified live bovineherpesvirus-1 vaccines on the efficacy of a recombinant Pasteurella haemolyt-ica vaccine for the prevention of respiratory disease in feedlot calves. Can VetJ 1992;33:734–41.

30] Li L, Coulthart MB, Balachandran A, Chakrabartty A, Cashman NR. Species barri-ers for chronic wasting disease by in vitro conversion of prion protein. BiochemBiophys Res Commun 2007;364(4):796–800.

31] White AR, Enever P, Tayebi M, Mushens R, Linehan J, Brandner S, Anstee D,Collinge J, Hawke S. Monoclonal antibodies inhibit prion replication and delaythe development of prion disease. Nature 2003;422:80–3.

32] Sigurdsson EM, Sy MS, Li R, Scholtzova H, Kascsak RJ, Kascsak R, Carp R, MeekerHC, Frangione B, Wisniewski T. Anti-prion antibodies for prophylaxis followingprion exposure in mice. Neurosci Lett 2003;336:185–7.

33] Goni F, Prelli F, Schreiber F, Scholtzova H, Chung E, Kascsak R, Brown DR, Sig-urdsson EM, Chabalgoity JA, Wisniewski T. High titers of mucosal and systemicanti-PrP antibodies abrogate oral prion infection in mucosal-vaccinated mice.Neuroscience 2008;153:679–86.

34] Sonati T, Reimann RR, Falsig J, Baral PK, O’Connor T, Hornemann S, YaganogluS, Li B, Herrmann US, Wieland B, Swayampakula M, Rahman MH, Das D, Kav

N, Riek R, Liberski PP, James MN, Aguzzi A. The toxicity of antiprion anti-bodies is mediated by the flexible tail of the prion protein. Nature 2013;Jul 31.

35] Jardine J, Julien JP, Menis S, Ota T, Kalyuzhniy O, McGuire A, Sok D, HuangPS, MacPherson S, Jones M, Nieusma T, Mathison J, Baker D, Ward AB,

ccine

[

[

K. Marciniuk et al. / Va

Burton DR, Stamatatos L, Nemazee D, Wilson IA, Schief WR. Rational HIV

immunogen design to target specific germline B cell receptors. Science2013;340(6133):711–6.

36] Joshi H, Lewis K, Singharoy A, Ortoleva PJ. Epitope engineering and molecularmetrics of immunogenicity: A computational approach to VLP-based vaccinedesign. Vaccine 2013;31:4841–7.

[

32 (2014) 1988–1997 1997

37] Uversky VN. Targetting intrinsically disordered proteins in neurodegenera-

tive and protein dysfunction diseases: another illustration of the D(2) concept.Expert Rev Proteomics 2010;7(4):543–64.

38] James TL, Liu H, Ulyanov NB, Farr-Jones S, Zhang H, Donne DG, et al. Solutionstructure of a 142-residue recombinant prion protein corresponding to theinfectious fragment of the scrapie isoform. PNAS USA 1997;94(19):10086–91.