Báo cáo khoa học: Vaccination with prion peptide-displaying papillomavirus-like particles induces autoantibodies to normal prion protein that interfere with pathologic prion protein production in infected cells pptx

As determined by peptide-specific ELISA, rabbit immune sera recognized the inserted murine⁄ rat epitope and also cross-reacted with the homologous rabbit⁄ human epitope differing in one a

papillomavirus-like particles induces autoantibodies

to normal prion protein that interfere with pathologic

prion protein production in infected cells

Alessandra Handisurya1, Sabine Gilch2, Dorian Winter3,4, Saeed Shafti-Keramat1, Dieter Maurer3,4, Hermann M Scha¨tzl2and Reinhard Kirnbauer1

1 Laboratory of Viral Oncology, DIAID, Department of Dermatology, Medical University Vienna, Austria

2 Institute of Virology, Technical University of Munich, Germany

3 Laboratory of Experimental and Clinical Immunology, DIAID, Department of Dermatology, Medical University Vienna, Austria

4 Center of Molecular Medicine, Austrian Academy of Sciences, Vienna, Austria

Prion diseases or transmissible spongiform

encephalo-pathies are untreatable, fatal neurodegenerative

dis-orders caused by proteinaceous infectious pathogens

devoid of nucleic acid, termed prions [1] These

diseases include scrapie of sheep, bovine spongiform encephalopathy in cattle and Creutzfeldt–Jakob disease

in humans [1] Although human prion diseases are rare, the appearance of a new variant of Creutzfeldt–

Keywords

immunotherapy; papillomavirus-like particles;

prion

Correspondence

R Kirnbauer, Laboratory of Viral Oncology,

DIAID, Department of Dermatology, Medical

University Vienna, Wa¨hringer Gu¨rtel 18–20,

A-1090 Vienna, Austria

Fax: +43 1 4030224

Tel: +43 1 40400-7768

E-mail: reinhard.kirnbauer@meduniwien.ac.at

(Received 29 September 2006, revised 21

December 2006, accepted 31 January 2007)

doi:10.1111/j.1742-4658.2007.05721.x

Prion diseases are fatal neurodegenerative disorders caused by proteina-ceous infectious pathogens termed prions (PrPSc) To date, there is no pro-phylaxis or therapy available for these transmissible encephalopathies Passive immunization with monclonal antibodies recognizing the normal host-encoded prion protein (PrPC) has been reported to abolish PrPSc infec-tivity and to delay onset of disease Because of established immunologic tolerance against the widely expressed PrPC, active immunization appears

to be difficult to achieve To overcome this limitation, papillomavirus-like particles were generated that display a nine amino acid B-cell epitope, DWEDRYYRE, of the murine⁄ rat prion protein in an immunogenic cap-sid surface loop, by insertion into the L1 major capcap-sid protein of bovine papillomavirus type 1 The PrP peptide was selected on the basis of its pre-viously suggested central role in prion pathogenesis Immunization with PrP–virus-like particles induced high-titer antibodies to PrP in rabbit and

in rat, without inducing overt adverse effects As determined by peptide-specific ELISA, rabbit immune sera recognized the inserted murine⁄ rat epitope and also cross-reacted with the homologous rabbit⁄ human epitope differing in one amino acid residue In contrast, rat immune sera recog-nized the murine⁄ rat peptide only Sera of both species reacted with PrPC

in its native conformation in mouse brain and on rat pheochromocytoma cells, as determined by immunoprecipitation and fluorescence-activated cell sorting analysis Importantly, rabbit anti-PrP serum contained high-affinity antibody that inhibited de novo synthesis of PrPScin prion-infected cells If also effective in vivo, PrP–virus-like particle vaccination opens a unique possibility for immunologic prevention of currently fatal and incurable pri-on-mediated diseases

Abbreviations

Ab, antibody; BPV-1, bovine papillomavirus type 1; FACS, fluorescence-activated cell sorting; PIPL-C, phosphatidylinositolphospholipase C; PrP C , normal prion protein; PrP Sc , pathologic prion protein; VLP, virus-like particle.

Jakob disease, probably due to consumption of bovine

spongiform encephalopathy prion-contaminated

prod-ucts, has become an important public health issue

[2,3] The possibility of horizontal human transmission

through surgery, organ grafting or blood transfusions

has raised further concern [4,5]

The exact pathogenetic mechanism of prion diseases

has remained uncertain In the most accepted

‘protein-only’ hypothesis, the causative agent, the pathologic

prion protein (PrPSc), is proposed to be a

conforma-tional isoform of the host-encoded normal prion

pro-tein (PrPC) PrPC is expressed ubiquitously on cell

surfaces, in particular by neurons, and also on

leuko-cytes, including T cells and B cells [6,7]

Transforma-tion of normal PrPCinto infectious PrPScis thought to

occur through a template-directed process [1] PrPC

appears to be required for prion infectivity, as mice

deficient in PrPC(Prnp0⁄ 0) do not propagate the

infec-tious agent and fail to develop scrapie following

experimental inoculation [8]

To date, there are no prophylactic or therapeutic

modalities available against prion diseases During

nat-ural infection with PrPSc, PrP-specific immune responses

are not generated [1,9,10] Nevertheless, studies in cell

culture and mice indicate that immunotherapeutic

strat-egies against the cellular form of PrPC can antagonize

prion infectivity and disease Monoclonal antibodies

(mAbs) or recombinant F(ab) fragments recognizing

PrP were effective, in vitro, in preventing prion infection

of susceptible mouse neuroblastoma cells and also in

abrogating PrPScde novoformation in chronically

infec-ted cells [11,12] In addition, transgenic expression or

passive transfer of mAb against PrP into

scrapie-infec-ted mice suppressed peripheral prion replication as well

as prion infectivity, and significantly delayed onset of

disease [13,14] Notably, no obvious side-effects were

observed in these mice

The immune system has evolved to respond

vigor-ously to viral or bacterial antigens, to protect the host

Several groups have demonstrated that papillomavirus

virus-like particles (VLPs) can activate innate immune

responses in dendritic cells to regulate adaptive

immune responses [15–17], and also directly activate

B-cell IgG production via a TLR-4⁄ MyD88-dependent

(and T-helper cell-independent) pathway [17]

Immuni-zations induced high-titer neutralizing antibodies (Abs)

and potent cell-mediated immune responses to the

viri-on capsid protein [18–20] In large clinical trials,

human papillomavirus VLP vaccinations were safe and

100% effective in preventing persistent infection and

associated genital disease, and polyvalent human

papil-lomavirus vaccines have been approved recently for

human use [21] In further studies, chimeric VLPs have

been developed that display foreign or self-polypep-tides in an ordered and closely packed repetitive array

on the capsid surface [22,23] Immunizations of experi-mental animals induced high-titer and high-avidity IgGs to the surface-displayed (self-)antigens that were long lasting and functionally active in vitro and in vivo Importantly, no adverse effects or induction of auto-immune disease were observed

The aim of this study is to develop a safe and effective vaccine that generates a strong Ab response against the prion self-antigen, but avoids induction of a T-cell response, to minimize the risk of cell-mediated auto-immune disease Development of an effective PrP vac-cine has been hampered by immunotolerance to the ubiquitously expressed endogenous PrPC In the past, this problem has been circumvented experimentally either by generation of anti-PrP immunity in PrP knock-out mice or using PrP of a different species as immuno-gen [24,25] Few groups have succeeded in immuno-generating a humoral immune response to PrP by active immuniza-tion using PrP–PrP polyproteins, PrP-expressing retro-virus particles, bacterially expressed full-length PrP, synthetic PrP peptides or polypeptides as antigen [26– 28] However, synthetic peptides are generally weak immunogens, and may even induce tolerance in the host

In addition, only Abs that recognize PrPCin its native conformation exerted prionostatic effects, whereas Abs

to denatured PrP were not effective [27,29,30]

To gain further advantage over previous full-length

or partial-length PrP vaccines, we decided to restrict the PrP immunogen to a short B-cell epitope comprising a functionally active peptide, to reduce the possibility of inducing cell-mediated autoimmune disease To circum-vent the problem of low immunogenicity and immuno-tolerance to endogenous PrPC, the use of particulate PrP–VLPs was chosen as the vaccine strategy This immunogen is composed of an ordered and closely spaced assembly of capsomer subunits displaying a PrP peptide, to induce a humoral immune response specific

to endogeous PrPC[18,28] A peptide of nine amino acid residues, DWEDRYYRE, of the murine⁄ rat prion pro-tein (amino acids 144–152) was incorporated into an immunogenic surface loop of the L1 major capsid pro-tein of bovine papillomavirus type 1 (BPV-1) and expressed by recombinant baculovirus technology This peptide resides in Helix 1 of PrPC[31,32], and has been extensively characterized previously, as it is the epitope recognized by mAb 6H4 (Prionics, Schlieren, Switzer-land) [33] Helix 1 of the prion protein is suggested to play a central role in the protein-induced

conformation-al changes mAb 6H4 has been well established as one

of the most widely used Abs in prion diseases, and is employed in routine diagnostic tests for transmissible

spongiform encephalopathies The epitope

DWE-DRYYRE is also recognized by other functionally

act-ive Abs, including Ab.Tg [34] On the basis of the

existence of these pivotal Abs, and under the

assump-tion that this epitope may represent an effective B-cell

epitope, in both rabbits and mice, we have chosen this

peptide sequence to generate chimeric PrP–VLPs

Immunization with PrP–VLPs induced high-titer Abs to

PrP in rabbits and in rats, without causing overt adverse

effects Sera specifically recognized PrPC in its native

conformation in mouse brain and on rat

pheochromocy-toma cells, when analyzed by immunoprecipitation and

flow cytometric analysis Importantly, rabbit Abs to PrP

were of high affinity, and effectively inhibited de novo

synthesis of pathogenetic PrPScin prion-infected cells

Results

PrP–L1 fusion protein self-assembles into

virus-like particles (PrP–VLPs) that express the

PrP-DWEDRYYRE epitope

To generate chimeric VLPs that display a PrP epitope

on the particle surface (PrP–VLPs), the murine PrP

peptide DWEDRYYRE (amino acids 144–152) (Fig 1A) was engineered into the L1 major capsid protein of BPV-1 [22] Following expression of the L1– PrP protein by recombinant baculoviruses in Sf9 insect cells, particles were purified by density gradient centrif-ugation Analysis by transmission electron microscopy revealed predominantly spherical structures approxi-mately 50–55 nm in diameter (Fig 1B), indicating self-assembly into VLPs with a morphology similar to that

of wild-type L1–VLPs [35] In addition, incompletely assembled particles and individual capsomers (the pen-tamer subunit of VLPs, consisting of five L1 mole-cules) were observed To verify expression of the inserted PrP epitope DWEDRYYRE, the antigenicity

of PrP–VLPs was analyzed by immunoblotting The polyclonal rabbit serum A7 recognized a predominant band of approximately 55 kDa, corresponding to the expected size of the PrP–L1 fusion protein (Fig 1C, left panel), in the purified PrP–VLP preparation and a crude Sf9 cell lysate The faster-migrating bands prob-ably correspond to proteolytic degradation products Specific immunoreactivity with A7 was absent with parental wild-type L1–VLPs As expected, mAb AU-1 (Fig 1C, right panel), directed against a linear epitope

Fig 1 (A) Amino acid sequences of the rat ⁄ mouse and rabbit ⁄ human PrP peptides encompassing residues 144–152 of the full-length prion proteins The rodent peptide differs at position 145 (bold) from the sequence of the rabbit ⁄ human peptide [changing tryptophan (W) to tyro-sine (Y)] (B) Transmission electron microscopy of chimeric PrP–VLPs (magnification · 30 000) Scale bar represents 200 nm (C) Immunoblot

of purified wild-type BPV-1 L1–VLPs (lane 1), PrP–VLPs (lane 2), or crude lysate of recombinant baculovirus-infected Sf9 insect cells expres-sing PrP–L1 (lane 3), uexpres-sing polyclonal rabbit anti-PrP serum A7 raised against murine dimeric PrP amino acids 23–231 (left), or mAb AU-1 directed against the linear BPV-1 L1 epitope DTYRYI (right) Molecular weight markers are indicated (D) VLP ELISA under nondenaturing (native) conditions, using conformation-dependent mAbs to L1, which either recognize both pentamers and VLP (mAb 6), or only fully assem-bled VLP (mAb 9), or mAb AU-1, which recognizes a linear BPV-1 L1 epitope Intact VLPs, either PrP–VLPs (black bars) or wild-type BPV-1 L1–VLPs (white bars), were used as the antigens.

of BPV-1 L1, reacted with both wild-type BPV-1 L1

and chimeric PrP–L1 proteins

To determine whether PrP–VLPs retain, at least in

part, the antigenic surface structures of wild-type

L1–VLPs, the immunoreactivity of intact particles was

examined by ELISA, using a conformation-dependent

neutralizing mAb directed against BPV-1 L1 mAb 6

has been shown to bind intact VLPs as well as

penta-meric subunits, whereas mAb 9 requires correctly

assembled VLPs for binding [22,36,37] In contrast, the

non-neutralizing mAb AU-1 recognizes an internal,

linear epitope of L1 (DTYRYI) As shown in Fig 1D,

immunoreactivity of mAb 6 and mAb 9 was observed

with chimeric PrP–VLPs (black bars) and, as a control,

wild-type L1–VLPs (white bars), indicating assembly

of chimeric PrP–L1 protein into a complete VLP

sim-ilar to wild-type L1 mAb AU-1 also reacted with both

protein preparations under nondenaturing conditions,

demonstrating nonassembled L1 protein, as observed

regularly in purified VLP preparations Expression of

the inserted peptide was further tested by ELISA using

rabbit anti-PrP serum A7 The antiserum demonstrated

high absorbance values with PrP–VLPs, but not with

wild-type L1–VLPs, indicating display of the PrP

epitope on the PrP–VLP surface (Fig 1D)

Immunization with PrP–VLPs induces (auto-)Ab

to the PrP peptide DWEDRYYRE

The chosen murine prion epitope DWEDRYYRE is

situated in the central region (Helix 1) of PrP, which is

highly conserved among species [38], and thus generally

is a poor immunogen To determine the

immuno-genicity of the inserted epitope presented in the context

of a VLP, rabbits were chosen for immunization, as

this species displays 100% amino acid sequence identity

in the PrP peptide sequence to that in humans, and dif-fers from the murine sequence in only one amino acid

at position 145 (changing tryptophan W to tyrosine Y) (Fig 1A) Importantly, no adverse clinical effects regarding respiratory function, digestion, weight loss or behavior were observed in the animals over a period of

3 months following PrP–VLP immunization None of the PrP–VLP-immunized animals died until time of sac-rifice To determine the PrP-specific humoral immune response, sera obtained before and after immunization were tested by ELISA, using the synthetic PrP peptide DWEDRYYRE as the antigen The peptide was linked via two alanine spacers to biotin to ensure complete accessibility after attachment to streptavidin-coated microtiter plates Immune sera showed titers > 1 : 400

by endpoint dilution, as compared to the nonspecific ELISA reactivity of preimmune sera Figure 2A (left) shows a representative experiment Specificity was confirmed with rabbit control serum immunized with wild-type BPV-1 L1–VLPs and by testing anti-PrP– VLP immune sera using an irrelevant control peptide (AVLPPVP) as the antigen (data not shown) To fur-ther corroborate the results, rabbit immune sera were preabsorbed with either PrP peptide or control peptide (AVLPPVP), prior to testing by PrP peptide ELISA

As shown in Fig 2A (right), preabsorption of the immune serum with PrP peptide significantly reduced ELISA reactivity by more than 40% as compared to immune serum preabsorbed with the control peptide These results are in agreement with those of Yokoyama

et al., and indicate that DWEDRYYRE represents a B-cell epitope also in the rabbit host [34]

Active immunization strategies have been hampered

by immunotolerance to endogenous PrPC[9,39,40] To

Fig 2 Left: New Zealand White rabbits (A)

or Lewis rats (B) were immunized with PrP– VLPs, and immune sera (squares) were tes-ted for immunoreactivity against the PrP peptide DWEDRYYRE by ELISA Preim-mune sera of the same animal (diamonds) served as the appropriate controls Right: To determine specificity, immune sera were preabsorbed with either PrP peptide DWE-DRYYRE, or control peptide AVLPPVP, prior

to analysis Data obtained using serum from one animal are shown as mean A ± SD of triplicate wells (representative of three inde-pendent experiments).

determine whether tolerance can be overcome with our

chimeric PrP–VLP preparations, Lewis rats were

cho-sen for immunization Rats share 100% amino acid

sequence identity in the inserted DWEDRYYRE

epi-tope, so the PrP peptide represents a complete

self-antigen Serum samples from rats were obtained before

and 2 weeks after the last immunization, and

subse-quently subjected to peptide ELISA with PrP peptide

DWEDRYYRE Immunization with chimeric PrP–

VLPs induced auto-Abs against the PrP peptide with

titers > 400, whereas preimmune serum completely

lacked ELISA reactivity (Fig 2B, left) The specificity

of the results was further confirmed by overnight

pre-absorption of immune sera ELISA reactivity was

reduced by 70% with PrP peptide preabsorption,

com-pared to preabsorption with control peptide (Fig 2B,

right)

We next sought to determine whether

PrP–VLP-induced immune sera were qualitatively different,

whether induced in the rat, for which the PrP peptide

comprised a ‘complete’ prion self-antigen (tryptophan

at position 145), or in the rabbit, for which it

com-prised an ‘incomplete’ self-antigen [one amino acid

dif-ference (tyrosine) at position 145] Thus immune sera

of rabbits and rats were compared by ELISA for

reac-tivity against synthetic peptides representing either the

murine⁄ rat amino acid sequence (DWEDRYYRE) or

the rabbit⁄ human sequence (DYEDRYYRE) of the

PrP peptide Purified IgG from rabbit antiserum

dem-onstrated comparable immunoreactivity to both the

murine⁄ rat sequence and the rabbit ⁄ human sequence

(Fig 3A) In contrast, reactivity of IgG isolated from

rat antiserum was directed specifically to the

murine⁄ rat peptide (Fig 3B), but was absent with the

rabbit peptide To further examine the Ab affinity of

rabbit and rat anti-PrP immune sera, a modified

pep-tide ELISA was employed The chaotropic agent

ammonium thiocyanate was added at increasing

con-centrations to dissociate antigen–Ab complexes

Toler-ance to thiocyanate elution is proportional to the

relative strength of antigen–Ab interactions, thus

rep-resenting a measure of Ab affinity [41] As shown in

Fig 3C, addition of approximately 1.35 m and 0.35 m

thiocyanate, respectively, was required to achieve a

50% reduction of rabbit or rat Ab binding

Ab-bind-ing curves (Fig 3C) for the rabbit immune sera

revealed an almost identical affinity distribution, when

tested against either the murine⁄ rat or the

rab-bit⁄ human peptide In contrast, the rat immune serum

showed a significantly (up to four-fold) lower affinity

for the murine⁄ rat peptide over a large molar range of

thiocyanate concentrations, demonstrating a

qualitat-ive difference in the nature of the evoked immune

response As shown previously (Fig 3B), rat sera lacked specific reactivity to the rabbit PrP peptide

Immune sera recognize native PrPCin mouse brain and on rat pheochromocytoma cells Recent studies have found an association between the ability of Abs to recognize native PrPC expressed on the cell surface and their ability to cure prion-infected cells in vitro and inhibit prion pathogenesis in vivo

Fig 3 Comparison of ELISA reactivity of purified IgG from PrP– VLP-immunized rabbit (A) and rat (B) to murine ⁄ rat PrP peptide DWEDRYYRE (white bar), or rabbit ⁄ human PrP peptide DYED-RYYRE (black bar) IgG isolated from preimmune serum of the same animals served as control (C) Antibody-binding curves of rab-bit and rat immune sera in the presence of increasing concentra-tions of ammonium thiocyanate (NH 4 SCN) Rabbit immune sera were tested for affinity against the murine ⁄ rat DWEDRYYRE pep-tide (diamonds) or the rabbit ⁄ human DYEDRYYRE peptide (squares) Similarly, rat immune sera (triangles) were tested for their binding affinity to the murine peptide As rat sera did not bind

to the rabbit ⁄ human peptide (B), they were not further tested All experiments were conducted at a serum dilution of 1 : 100 The results for one representative animal each are shown.

[11–14,27,29,30] To determine whether evoked

anti-sera recognize native PrPC, brains of healthy,

uninfect-ed wild-type C57Bl⁄ 6 mice (Prnp genotype a ⁄ a) were

homogenized and immunoprecipitated using purified

IgG from PrP–VLP-immunized rabbit or rats

Pre-immune IgG of the same animals or anti-PrP rabbit

serum A7 served as appropriate negative and positive

controls, respectively Equal volumes of brain

homo-genate were used as starting material, as well as equal

amounts of IgG, to allow semiquantitative comparison

of recovered immune complexes Immunoprecipitates

were analyzed by western blot with mAb 4H11 As

shown in Fig 4, IgG purified from rabbit PrP–VLP

immune serum specifically detected PrPC, whereas

reactivity with IgG purified from preimmune serum of

the same animal was basically not detectable As

expected, rabbit immune serum A7 also recognized

PrPCin mouse brain In contrast, no specific

immuno-reactivity could be detected with IgG from immunized

rats

As a more sensitive assay for the detection of IgG

binding to native PrPC, fluorescence-activated cell

sort-ing (FACS) analysis of PC-12 rat pheochromocytoma

cells, which express low levels of PrPCon their surface

[42], was used An increase in surface binding was

observed with PrP–VLP-induced rabbit immune IgG

(Fig 5A, lower panel) as compared to preimmune IgG

of the same rabbit, demonstrating specific binding to

PrPCon the PC-12 cell surface Similarly, rat immune

IgG (Fig 5B, lower panel), but not preimmune IgG,

reacted with PC-12 cells, albeit to a lesser degree than

rabbit IgG

PrPCis linked to the plasma membrane via a glyco-sylphosphatidylinositol anchor, which is sensitive to phosphatidylinositolphospholipase C (PIPL-C) treat-ment To confirm that IgG binding to cell surface PrPC was specific, PC-12 cells were incubated with PIPL-C prior to FACS analysis with IgG from immune sera or the positive control mAb 4H11 Compared to fluores-cence obtained in the absence of PIPL-C treatment, removal of PrPCby PIPL-C digestion induced a signi-ficant reduction in binding of rabbit and rat IgG, respectively (data not shown), indicating specificity of the FACS results Furthermore, immune IgG was pre-absorbed overnight with either PrP–VLPs or wild-type

Fig 4 Immunoprecipitation of native PrPCfrom healthy,

uninfect-ed, wild-type mouse brain by anti-PrP sera IgG purified from rabbit

PrP–VLP immune serum (lane 1), rabbit preimmune IgG (lane 2),

PrP–VLP-immunized rat immune IgG (lane 3) or rat preimmune IgG

(lane 4) were tested Rabbit immune serum A7 (lane 5) raised

against dimeric murine PrP amino acids 23–231 served as positive

control A molecular size marker is indicated.

Fig 5 Demonstration of anti-PrP serum binding to cell surface PrP C on PC-12 rat pheochromocytoma cells by flow cytometric ana-lysis (A) Upper panel: Fluorescence obtained with negative con-trols, PC-12 cells alone (open histogram), or PC-12 cells reacted with second-step anti-rabbit serum (closed histogram) Lower panel: Comparison of fluorescence intensity between PC-12 cells incubated with PrP–VLP-induced rabbit immune IgG (closed histo-gram) or preimmune IgG (open histohisto-gram) of the same rabbit (IgG dilution 1 : 300) (B) Upper panel: Negative controls with PC-12 cells (open histogram) or with second-step anti-rat serum (closed histograms) Lower panel: Fluorescence signal obtained with PC-12 cells after addition of rat immune IgG (closed histogram), as com-pared to rat preimmune IgG (open histogram) (IgG dilution 1 : 100).

BPV-1 L1–VLPs prior to FACS analysis A decrease

of IgG binding was consistently observed with PrP–

VLP preabsorption of immune sera, as compared to

wild-type VLP preabsorption Reactivity was

specific-ally reduced by 36% following overnight

preabsorp-tion of rabbit immune IgG with PrP–VLPs The

reactivity of PrP–VLP-preabsorbed rat IgG was

dimin-ished by 82%, as compared to prior incubation with

wild-type L1–VLPs (data not shown) Taken together,

these results confirmed specific recognition of native

cell surface PrPCby IgG induced by PrP–VLP

immun-ization

PrP–VLP-induced rabbit immune sera are

functionally active and inhibit PrPSc biogenesis in

living cells

To determine the biological relevance of the anti-PrPC

humoral immune response, antisera were analyzed for

their ability to inhibit PrPSc de novo synthesis in

per-sistently prion-infected mouse ScN2a neuroblastoma

cells [43] As shown in Fig 6, ScN2a cells were

meta-bolically labeled with [35S]l-cysteine⁄ methionine for

16 h, and purified rabbit or rat immune IgG was

added in parallel to the culture medium As

appropri-ate negative and positive controls, rabbit and rat

pre-immune IgG or mAb 4H11 were included in the

analysis, or cells were left untreated Cells were then lysed, and subjected to proteinase K digestion and ultracentrifugation to separate PrPSc from PrPC; the PrPSc-containing fraction was further subjected to immunoprecipitation, followed by a deglycosylation step with N-glycosidase F (PNGase F) PrPSc was readily detectable in untreated ScN2a cells, demonstra-ting consistent production of PrPSc by infected ScN2a cells Addition of the positive control mAb 4H11 resul-ted in lack of detectable immunoreactivity against PrPSc in the cells Incubation of the PrPSc-producing cells with IgG, purified from PrP–VLP-immunized rab-bits, completely blocked PrPSc reactivity, demonstra-ting successful inhibition of de novo synthesis of pathologic PrPSc in persistently prion-infected ScN2a cells In contrast, a strong band at the expected size of PrPSc was observed in cells treated with preimmune IgG Thus, inhibition of PrPSc biogenesis was not attributable to nonspecific serum components, but to the presence of anti-PrP IgG induced by immunization with chimeric PrP–VLPs Addition of rat immune IgG did not completely abolish PrPSc production in infec-ted cells This negative result for rat immune IgG cor-related with weaker binding to PrPCby FACS (Fig 5) and lower affinity to PrP peptide by thiocyanate ELISA (Fig 3C), as compared to rabbit sera

Discussion

This study shows that high-density display of a short (nine amino acid residues) PrP B-cell epitope on the papillomavirus VLP surface is sufficient to effectively induce Abs to PrP These Abs recognize PrPC in its native cell-associated conformation, thus overcoming the reported immunotolerance for endogenous, host-encoded PrPC[9,39,40] The murine neuroblastoma cell line ScN2a represents a persistently prion-infected cell culture model, which produces considerable amounts

of PrPSc Importantly, immunized rabbit IgGs did not merely bind PrP, but effectively inhibited PrPScde novo synthesis in living cells This strongly suggests a func-tionally relevant protective role of the induced Ab to PrP (Fig 6), indicating that even previously infected cells may become cured of infection Considering the long latency phase and limited diagnostic methods available, immunization with PrP–VLPs may offer ave-nues for further experimentation that could suggest an effective strategy for prevention and perhaps even ther-apy of the as yet uncurable prion diseases

When immunologic antiprion strategies are employed, the issue of safety is critical Whereas our strategy has similarities with a recent study employing retroviral particles for display of prion protein [28],

Fig 6 Effects of anti-PrP sera on PrP Sc conversion in persistently

prion-infected mouse ScN2a neuroblastoma cells ScN2a cells were

left untreated or incubated with the indicated preimmune or

immune IgG, or mAb 4H11 as positive control To detect PrP Sc

con-version, cells were lysed, and then subjected to proteinase K

diges-tion and immunoprecipitadiges-tion A deglycosyladiges-tion step was included

to reduce the PrP Sc and PrP C glycoforms to one single band, as

described in detail in Experimental procedures.

papillomavirus VLP appears to be advantageous as a

vaccine carrier, comprising nonreplicating subunit

vac-cines that have proven to be safe and highly effective

in thousands of vaccinees [21] Papillomavirus L1 has

the intrinsic capacity to self-assemble into pentamers

that further multimerize into small or full-size VLPs

[35] PrP–VLPs accommodate epitope insertion and

display the PrP peptide at high density (up to 360

cop-ies per VLP) in a repetitive, closely spaced and highly

immunogenic manner on the particle surface

Import-antly, VLP-based vaccines that induce continuously

present auto-Ab to PrPC could represent a safe and

effective alternative to passive administration of Abs

or peptide vaccination Possible advantages are a more

constant Ab level over time, the need for less frequent

administration, and the absence of an inactivating Ab

response to therapeutic Ab Auto-Ab against PrPC,

which is ubiquitously expressed throughout the

organ-ism, in particular in neuronal and immune cells, may

induce Ab-mediated autoimmune encephalitis or

com-plement-dependent lysis of immune cells However, we

have not observed overt adverse effects in animals

fol-lowing PrP–VLP immunizations The relatively low

serum titers induced by our PrP–VLP immunizations

may have contributed to this favorable result

Immunization of rats did not induce antisera that

effectively inhibit PrPSc conversion The weaker

bind-ing of rat antiprion Ab to pheochromocytoma cells

and the four-fold lower affinity as compared to rabbit

Ab may account for the observed difference in efficacy

This observation may indicate the existence of a

threshold in Ab titer or affinity that dictates

therapeu-tic efficacy Modifications of the PrP peptide

immuno-gen, e.g single amino acid differences from the

endogenous PrP sequence (as in the rabbit

vaccin-ation), may offer an experimental approach to further

evaluate the putative efficacy of PrP–VLP vaccines

A study has demonstrated apoptosis of hippocampal

and cerebellar neurons following intracerebral injection

of mAb to PrPC[44] Neurodegeneration was thought

to result from cross-linking and clustering of PrPC,

trig-gering an abnormal signaling pathway However,

neur-onal loss occurred only when extremely high

concentrations of Abs to epitopes encompassing amino

acids 95–105 of the PrP were used Injection of mAbs

that recognize the central region of the PrPC (residues

133–157) and bind efficiently to cell surface PrPC did

not induce neuronal apoptosis Given the absence of

overt neurologic abnormalities in our vaccinated

animals, we argue that auto-Ab to PrP amino acids

144–152 induced by PrP–VLPs either do not cross-link

PrPC or sterically mask a region of PrPC that

inter-acts with a putative signaling partner Moreover, to

minimize the risk of induction of T-cell-mediated auto-immune disease, including encephalitis, we have employed a short PrP peptide that is unlikely to com-prise a T-cell epitope, as compared to the larger or full-length PrP proteins that have been used in previous reports Nevertheless, larger animal studies are required

to further evaluate the safety of PrP–VLP vaccination Several potential mechanisms may underlie the effic-acy of antiprion Ab induced by PrP–VLPs [45] Ab to PrP may bind to and mask PrPC or induce redistribu-tion of PrPC from the cell surface into cellular com-partments, where it is not available for conversion into pathogenic PrPSc In vivo, anti-PrP IgG may even bind

to the surface of B-lymphocytes and thus directly inhi-bit prion replication and infectivity On the basis of previously published studies [27,33,34], a peptide from central Helix 1 of the prion protein was chosen as immunogen, as this region has been implicated in

pri-on replicatipri-on and transmissipri-on, and to allow func-tional testing in mouse models [31] Our results are in agreement with those obtained with mAb to Helix 1, which has been shown to be effective in inhibiting

pri-on pathogenesis [12,14,29]

Taken together, the results that we have presented here show that a VLP-based PrP vaccine can circum-vent immunotolerance to the widely expressed self-anti-gen PrPC, and induce a robust Ab response that can protect against PrP infection in vitro Further studies are required to evaluate the potential of the PrP–VLP vaccine to prevent or even cure prion infectivity and disease in experimental animal models If proven effect-ive, PrP–VLP vaccination may open a unique possibil-ity for medically applicable immunologic prevention of currently fatal and uncurable prion-mediated diseases

Experimental procedures

Generation of recombinant baculoviruses expressing the PrP epitope DWEDRYYRE-BPV-1 L1 fusion protein and purification of chimeric PrP–VLPs

To generate recombinant baculoviruses expressing the PrP epitope DWEDRYYRE (corresponding to amino acids 144–152 of the murine PrP) on a predicted surface loop of L1–VLP as a fusion protein, oligonucleotides encoding the epitope were inserted by inverse-touchdown PCR into an immunogenic region (between amino acids 133 and 134) of the BPV-1 L1 major capsid protein, using the pEVmod transfer vector as previously described [22] The forward and reverse primer sequences were 5¢-GCTACTACCGTG AAACCCAAACAACAGATGAC-3¢ and 5¢-GGTCCTCC CAGTCGGTGACTTTTCTATTCAC-3¢, respectively Final

clones were verified by restriction enzyme digestion and by

nucleotide sequencing of the inserted sequence and the

junc-tional L1 region By cotransfection of Sf9 insect cells with

transfer vector and linearized baculovirus DNA

(Baculo-Gold; BD Biosciences Pharmingen, San Diego, CA),

recom-binant baculoviruses were generated by standard methods

[35] Sf9 insect cells were infected with baculovirus stocks

and lysed, and high molecular mass structures were

separ-ated by density gradient ultracentrifugation VLP-containing

bands were collected and dialyzed [22,46] BPV-1 L1–VLPs

consisting of wild-type L1 major capsid protein were

gener-ated in a similar manner [35]

Characterization of purified PrP–VLPs

To analyze the self-assembly of L1 and the integrity of

chimeric PrP–VLPs by transmission electron microscopy

[47], purified VLP preparations were absorbed onto

glow-discharged carbon-coated copper grids, fixed, negatively

stained with 1% uranyl acetate, and analyzed in a JEOL

(Eching, Germany) 1010 electron microscope at 80 kV and

· 30 000 magnification Purified PrP–VLPs and wild-type

BPV-1 L1–VLPs were further analyzed for the presence of

PrP or BPV-1 L1 epitopes For western blot analysis,

PrP–VLPs, wild-type BPV-1 L1-VLPs or lysates of Sf9

cells infected with recombinant baculoviruses were

dena-tured in SDS sample buffer containing 2%

b-mercaptoeth-anol, electrophoresed on 10% SDS polyacrylamide gel,

immunoblotted, and probed with Ab A7 to PrP or mouse

mAb AU-1 (BabCo, Berkeley Antibody, Richmond, CA),

recognizing a linear BPV-1 epitope DTYRYI Polyclonal

Ab A7 specific to PrP was obtained in our laboratories

after immunization of rabbits with recombinant dimeric

mouse PrP amino acids 23–231 [26,48] Finally, blots were

incubated with second-step peroxidase-labeled Ab, either

goat anti-rabbit (Kierkegaard and Perry Laboratory Inc.,

Gaithersburg, MD) or goat anti-mouse (Jackson

Immuno-Research Laboratory Inc., West Grove, PA), and

develo-ped using the ECL system Purified PrP–VLPs or

wild-type BPV-1 L1–VLPs were used as antigen in an

ELISA Native VLPs were coated overnight at 4C onto

96-well microtiter plates mAb 6 and mAb 9 are

BPV-1-neutralizing mouse mAbs directed against

conformation-dependent and type-specific L1 epitopes [36]; mAb AU-1

against L1 and serum A7 against murine PrP are

des-cribed above Serial dilutions of mAb or antiserum were

added to triplicate wells, and developed by

peroxidase-labeled goat anti-mouse serum or goat anti-rabbit serum

and the peroxidase substrate di-ammonium

2,2¢-azino-bis(3-ethylbenzo-6-thiazolinesulfonic acid) Absorbances

were determined at 405 nm Specific absorbances were

cal-culated by subtracting mean values obtained in wells

with-out antigen (NaCl⁄ Pi) from mean values obtained in

antigen-coated wells The replication variation in these

assays was less than 5%

Immunization of animals with PrP–VLPs and induction of a humoral immune response

New Zealand White rabbits and Lewis rats (Charles River Laboratory, Kisslegg, Germany) were inoculated with

100 lg or 50 lg of PrP–VLPs, respectively, each four times

at 2–4-week interval At each injection, VLPs were adminis-tered to New Zealand White rabbits with CpG 2006 (80 nmol) plus incomplete Freund’s adjuvant, and to Lewis rats with CpG 2006 (40 nmol) only, to avoid eventual development of granuloma due to incomplete Freund’s adjuvant in the small-sized animals Serum samples were collected from animals prior to immunization, 10 days after the third injection, and 2 weeks after the final boost When indicated, total IgG was purified using the ImmunoPure IgG Purification Kit A Plus or G, respectively (Pierce Bio-technologies, Rockford, IL)

Sera were tested by ELISA for reactivity to synthetic PrP peptide DWEDRYYRE (amino acids 144–152) or con-trol peptide AVLPPVP as a specificity concon-trol Peptide AVLPPVP is derived from murine choriongonadotropin-b Streptavidin-coated 96-well plates (Nunc A⁄ S, Roskilde, Denmark) were coated with 1 lg of peptide linked at the N-terminus via two alanine residues to biotin (JPT Peptide Technologies, Berlin, Germany) in 100 mm Tris⁄ HCl (pH 7.5)⁄ 150 mm Nacl ⁄ 0.1% Tween-20 coating buffer over-night at 4C, and blocked with NaCl ⁄ Pi⁄ 1% nonfat dry milk Serial dilutions of sera were added in triplicate and, after incubation, plates were developed as described above Mean A405 values of uncoated (NaCl⁄ Pi) wells were sub-tracted to determine specific absorbance values of the sam-ples Mean A405values of immune sera greater than twice the mean A value (± SD) of preimmune sera (at the same serum dilution) were considered positive To further exam-ine specificity, sera of immunized animals were preabsorbed with either PrP peptide DWEDRYYRE or the irrelevant control peptide AVLPPVP by overnight incubation in streptavidin-coated plates coated with 5 lg of the respective biotinylated peptide Subsequently, absorbed sera were ana-lyzed for remaining immunoreactivity to PrP peptide by ELISA For determination of eventual cross-reactivity, immune sera were tested against the murine⁄ rat sequence

of the PrP peptide, DWEDRYYRE, as well as against the rabbit⁄ human sequence DYEDRYYRE by peptide ELISA

as described above To assess the Ab affinity distribution, immune sera were analyzed by peptide ELISA as described above, with the modification of a thiocyanate elution step

to dissociate antigen–Ab complexes [41] After Ab incuba-tion (all at a diluincuba-tion of 1 : 100), ammonium thiocyanate (NH4SCN) in 0.1 m sodium phosphate (pH 6.0) was added

to triplicate wells at molarities ranging from 0.0625 to 5 m for 15 min at room temperature Control wells were incuba-ted with 0.1 m sodium phosphate without NH4SCN The

Ab content in control wells without thiocyanate represents the total bound anti-PrP; Ab contents in wells incubated

with increasing molar concentrations of thiocyanate were

expressed as proportions of this total

Determination of PrPC-specific IgG binding by

immunoprecipitation and western blot

Wild-type C57Bl⁄ 6 mice (Prnp genotype a ⁄ a) were killed, and

their brains were removed and homogenized in lysis buffer

(150 mm NaCl, 50 mm Tris⁄ HCl, pH 8.0, 1 mm EDTA,

pH 8.0, 1% Nonidet P-40, and protease inhibitors); this was

followed by incubation on ice for 2 h Homogenates were

cen-trifuged at 10 000 g for 20 min at 4C using a Hettich

Univer-sal 30 RF centrifuge with 1412 rotor; supernatants were then

collected and stored at) 20 C until further use As polyclonal

anti-PrP serum displayed a high background in a direct

west-ern blot assay of mouse brain extracts, we decided to use a

sequential immunoprecipitation–western blot approach to

determine PrPC-specific anti-PrP IgG binding PrPCof brain

homogenate supernatants was immunoprecipitated with

puri-fied IgG from either PrP–VLP-immunized New Zealand White

rabbits or Lewis rats with protein G beads (Pierce

Biotechno-logies) IgG purified from preimmune sera of the same animals

or polyclonal rabbit antiserum A7 served as controls After

denaturation in SDS sample buffer containing 2%

b-merca-ptoethanol, samples were electrophoresed on 15% SDS

polya-cryamide gels, and western blot was performed using mAb

4H11 (dilution 1 : 1000) as described above mAb 4H11 was

generated using a dimeric murine PrP as an immunogen [49]

All animal experiments were performed in accordance

with the Austrian and EU ethical guidelines and approved

by the Federal Ministry of Education, Science and Culture

(GZ66.009/124-BrGt/2003)

Determination of PrPC-specific IgG binding by

flow cytometric (FACS) analysis on rat

pheochromocytoma cells

The rat pheochromocytoma cell line PC-12 (ATCC

CRL-1721), which has been reported to express PrPC [42], was

maintained in RPMI 1640 supplemented with 5% fetal

bovine serum, 10% horse serum, antibiotics, and glutamine

PC-12 cells were washed with NaCl⁄ Piand detached from

the culture flasks by keeping cells at 4C in NaCl ⁄ Pi for

10 min After detachment, cells were washed twice with cold

2 mm EDTA⁄ NaCl ⁄ Pi containing 0.5% BSA As primary

Ab, purified IgG from PrP–VLP-immunized rabbits (diluted

1 : 300) or rats (diluted 1 : 100) was added Purified IgG

from preimmune sera at the same dilution served as a

negat-ive control, and mAb 4H11 (dilution 1 : 100) as a positnegat-ive

control Second-step Abs used were either Alexa-488-labeled

goat anti-rabbit serum (Molecular Probes, Eugene, OR),

phycoerythrin-labeled donkey anti-rat serum (BD

Bioscien-ces Pharmingen) or Alexa-488-labeled goat anti-mouse serum

(Molecular Probes); all Abs were diluted 1 : 150

To confirm the specificity of the results, cells were treated with PIPL-C (Sigma Chemicals, St Louis, MO) to remove cell surface PrPC After two washes with NaCl⁄ Pi, cells were incubated in serum-free medium for 30 min at 37C with or without 1 UÆmL)1 PIPL-C prior to analysis Fur-thermore, IgG from rabbit or rat immune sera were preab-sorbed overnight with either PrP–VLPs or parental wild-type BPV-1 L1–VLPs before being subjected to FACS ana-lysis After exclusion of dead cells by 7-aminoactinomycin

D staining, analysis was performed using a FACScalibur flow cytometer (Becton Dickinson, Franklin Lakes, NJ) Results are expressed in fluorescence units (log scale)

Analysis of PrPScde novo synthesis

The mouse neuroblastoma cell lines N2a (ATCC CCL 131) and prion-infected ScN2a have been described previously [43] Cells were maintained in Opti-MEM medium contain-ing 10% fetal bovine serum, antibiotics, and glutamine For analysis of PrPSc de novo synthesis, ScN2a cells were washed twice with NaCl⁄ Pi at about 80% confluency, and starved for 1 h in RPMI medium without cysteine⁄ methi-onine (Sigma) supplemented with 1% fetal bovine serum Then, 400 lCiÆmL)1 [35S]l-cysteine⁄ methionine (Amersham Biosciences, Piscataway, NJ) was added and incubated for

16 h without a chase period to allow PrPSc synthesis, either in the presence or the absence of immune sera as indicated at a dilution of 1 : 50 As a control for the inhi-bition of de novo synthesis, the mAb 4H11 (10 lgÆmL)1) was used Lysis, proteinase K digestion and immunopre-cipitation of PrP was performed as described [26,50], using the rabbit polyclonal Ab A7 In order to separate PrPC and PrPSc, lysates were subjected to ultracentrifugation at

100 000 g for 1 h (Beckman TL100; TLA-45 rotor; Beck-man Coulter, Fullerton, CA) in the presence of 1% N-lau-rylsarcosine, and the pellet fraction containing PrPSc was used for immunoprecipitation To facilitate quantification,

a deglycosylation step with PNGase F was included to reduce the PrPC and PrPSc glycoforms to one single band [26,48] Both Ab A7 and mAb 4H11 recognize each of the PrP glycoforms The samples were analyzed on 12.5% SDS polyacrylamide gel, and this was followed by autora-diography on X-OMAT AR films (Kodak, Amersham Bio-sciences)

Acknowledgements

This work was supported by the SFB-576 (project B12) and the BMBF (KO0108) (H Scha¨tzl), and the Austrian Science Foundation, FWF (P18990-B13) (R Kirnbauer), and was performed within the frame-work of the EU FP6 ‘Viraskin’ (R Kirnbauer) and Network of Excellence Neuroprion (H Scha¨tzl)