Virus-like Peptide Vaccines Against Aß N-terminal or C-Terminal Domains Reduce Amyloid Deposition in APP Transgenic Mice without Addition of Adjuvant

Abstract

Immunotherapy against the Aß peptide is increasingly viewed as an effective means of preventing and even decreasing Aß deposition in transgenic mouse models and human cases of Alzheimer’s disease. A prior active immunization trial was halted due to adverse events which occurred subsequent to a change in the adjuvant used in the vaccine preparation. Although widely used in experimental studies, adjuvants available for use in vaccines intended for humans are limited. We compared two vaccine preparations in which an immunogenic bacteriophage was conjugated with either an N-terminal (Aß1-9) or C-terminal (Aß28-40) peptide sequence from the Aß molecule. We found that both produced significant antibody titers without use of additional adjuvants. Surprisingly, the response to the N-terminal sequence was comprised largely of a stable IgM response, while the C-terminal vaccine produced an IgG response with minimal IgM reactivity. Both of these immunogens reduced Aß levels when tissues were examined 8 months after the first inoculation. These data demonstrate that (a) C-terminal specific vaccines can effectively lower Aß and (b) IgM antibodies against Aß may be capable of lowering Aß, possibly through action in the brain rather than the periphery.

Introduction

Accumulation of the amyloid beta peptide (Aβ) has been argued to play a primary role in initiating Alzheimer’s disease (AD; Hardy and Selkoe 2002). One of the more effective amyloid-reducing therapies has been immunotherapy against the Aß peptide (reviewed in Morgan 2006). Initial results using fibrillar Aβ1-42 as a vaccine, along with a strong adjuvant in amyloid precursor protein (APP) transgenic mouse models dramatically reduced Aβ deposition (Schenk et al. 1999) and restored memory function (Janus et al. 2000; Morgan et al. 2000). Subsequently, The Elan and Wyeth pharmaceutical companies applied this approach to AD patients in a phase 2a clinical trial performed in 2001, and in a few patients, beneficial effects were apparent (Hock et al. 2003; Gilman et al. 2005). Unfortunately, this trial was halted due to 6% of the patients developing aseptic meningoencephalitis (Orgogozo et al. 2003) that was considered to be secondary to a T cell response associated with the strong adjuvant or full-length Aβ (Birmingham and Frantz 2002; Nicoll et al. 2003; Sigurdsson et al. 2004). A secondary issue with the phase 2a clinical trial was that the antibody production was variable, with many patients failing to mount an antibody response to the vaccine (Gilman et al. 2005; Patton et al. 2006). Passive immunotherapy using monoclonal antibodies against Aβ also reduced Aβ burden and improved cognitive function in aged APP transgenic mice (Bard et al. 2000; DeMattos et al. 2001; Dodart et al. 2002; Wilcock et al. 2004b) but led to microhemorrhages in association with amyloid-laden vessels (Pfeifer et al. 2002; Wilcock et al. 2004a; Racke et al. 2005).

Therefore, in this study, we investigated the use of a novel vaccine designed to minimize some of the problems encountered in the Elan-Wyeth phase 2a vaccine trial. We conjugated a short fragment of the Aβ peptide from the N-terminal or C-terminal to the surface of virus-like particles (VLPs) that were established to break self-tolerance (Li et al. 2004) and to stimulate a strong antibody response (Chackerian et al. 2006). It was anticipated that this would elicit a strong immune response, with large amounts of antibodies produced, but with minimal T cell activation (to minimize the risk of inflammation). We also wanted to compare whether the N-terminal or the C-terminal specific antibodies had different effects in clearing Aß and perhaps differed in other antibody functions.

Materials and methods

Vaccine preparation

As previously described (Chackerian et al. 2006), Qß bacteriophage were prepared by infecting a 500-mL culture of Escherichia coli strain A/lambda at A600 = 0.25 with Qß bacteriophage at a multiplicity of infection of 0.5. After lysis (approximately 3 h), 25 mg of hen egg lysozyme (Sigma) were added, and the solution was rocked for an additional 30 min at 37°C. The solution was incubated for an additional 30 min following addition of 1 mL of 0.5 M ethylenediaminetetraacetic acid (EDTA) and 1 mL chloroform. Unlysed cells and cellular debris were removed by centrifugation at 8,000 rpm for 30 min at room temperature using a Beckman JA10 rotor. Phage were precipitated from the supernatant at 4°C overnight following addition of NaCl to 0.5 M and polyethylene glycol 8000 to 10% (w/v). Precipitate was collected after centrifugation (at 10,000 rpm in a Beckman JA14 rotor for 30 min at 4°C) and then dissolved in approximately 5 mL TNME buffer (10 mM Tris–HCl, pH 7.4, 100 mM NaCl, 0.1 mM MgSO4, 0.01 mM EDTA). CsCl was added to a density of 1.4 g/mL, and the solution was centrifuged to equilibrium (22–24 h) at 40,000 rpm in a SW50.1 rotor at 20°C. The phage band was collected using a 20-gauge needle, and then the phage was dialyzed against phosphate buffered saline (PBS), pH 7.4.

A 12-amino acid peptide that contains the N-terminal nine amino acids of Aβ linked to a C-terminal gly-gly-cys (Aβ(1–9)-gly-gly-cys; DAEFRHDSGGGC) and a 16-amino acid peptide that contains the C-terminal 13 amino acid of Aβ linked to a C-terminal gly-gly-cys (Aβ(28-40)-gly-gly-cys; KGAIIGLMVGGVVGGC) were directly linked to Qβ phage by using a bifunctional cross-linker with amine- and sulfhydryl-reactive arms (SMPH, Pierce Endogen, Rockford, IL, USA). The amine-reactive arm of SMPH was linked to surface-exposed lysines on Qβ by reacting phage with SMPH at a 1:10-M ratio. Qβ-SMPH conjugates were purified by centrifugation using an Amicon Ultra-4 (100 kD molecular weight cut-off) centrifugal filtration device. Qβ-SMPH was linked to the Aβ (1–9 and 28-40)-cys peptides by virtue the exposed sulfhydryl residue on the C-terminal cysteine residue of the peptide. Qβ-SMPH was reacted with Aβ(1–9 and 28-40)-cys at a 1:10-M ratio, and Qβ-Aβ(1–9 and 28-40) conjugated particles were purified by centrifugation using the Amicon unit as described above. The extent of modification was assessed through gel electrophoresis and by determining the percentage of peptide-conjugated coat protein with decreased mobility. This protocol resulted in phage particles conjugated with an average of 270 copies of peptide (data not shown).

Vaccine administration

Eighteen APP Tg2576 transgenic mice aged 9 months, in which the first Aβ histological deposits can be discerned, were divided into three equal groups. Mice in group 1 received the N-terminal Qβ- Aβ1-9 -gly-gly-cys vaccine (Qβ-Aβ1-9; 10 μg in 50 μl PBS); group 2 received the C-terminal Qβ-Aβ 28-40-gly-gly-cys vaccine (Qβ-Aβ 28-40; 10 μg in 50 μl PBS), and group 3 received the PBS vehicle control (50 μl). Mice were injected intramuscularly with vaccines or vehicle and received two boosts at biweekly intervals, then monthly boosts for the next 7 months.

Serum and tissue preparation

Approximately 100 μl of blood was collected by tail bleeding 14 days after each injection, starting with the second inoculation. The blood was centrifuged for 15 min at 1,500×g, and then the serum was subsequently frozen at −80°C. Fourteen days following the final inoculation, mice were sacrificed by overdose of pentobarbital (200 mg/kg (Nembutal sodium solution; Abbott Laboratories, North Chicago, IL, USA). Blood was collected by ocular enucleation, and the mice were intracardially perfused with 25 ml of 0.9% saline. The brain was removed, the right hemisphere was rapidly dissected over ice, frozen on dry ice, and stored at −80°C for Aβ enzyme-linked immunosorbent assay (ELISA) assay. The left hemisphere was immersion fixed in 4% neutral buffered paraformaldehyde for 24 h, and the hemispheres were cryoprotected by sequential incubations in 10%, 20%, and 30% sucrose solutions prior to sectioning. Horizontal sections of 25 μm thickness were collected using a sliding microtome and stored at 4°C in Dulbecco’s PBS (pH 7.4) containing 1 mM sodium azide, until histopathologic staining.

Anti-Aβ antibody ELISA

Anti-Aβ antibody titers in mouse sera were measured as previously described (Li et al. 2004). Monoclonal antibody 6E10 (Signet, Dedham, MA, USA) was used as a standard, and the results are calculated as microgram antibody per milliliter serum. IgG isotype-specific antibodies: IgG1, IgG2b, IgG2a, and IgG2c (Jackson ImmunoResearch Laboratories, West Grove, PA, USA) were compared as detection antibodies for anti-Aβ IgG isotypes ELISA assay. The μ chain specific anti-mouse IgM was used for detecting anti-Aβ IgM (Jackson ImmunoResearch Laboratories, West Grove, PA, USA). ELISAs for antibody epitope mapping were performed using the following human Aβ peptide fragments: Aβ1-9, Aβ1-16, Aβ12-28, Aβ28-40, and Aβ1-40 (rPeptide, Bogart, GA, USA) preincubated mouse antisera overnight at 4°C, and then incubated in an ELISA plate coated with human Aβ peptide 1-40.

Brain tissue and serum Aβ ELISA

Aβ in the frontal cortex was extracted following the protocol described by Kawarabayashi et al. (2001) with minor modifications. Briefly, frozen cortical tissues (150 mg/ml wet weight) were homogenized in Tris-buffer saline (TBS), which contained protease inhibitors (complete protease inhibitor cocktail, one tablet in 10 ml solution; Boehringer Mannheim, Mannheim, Germany) and then centrifuged at 100,000×g for 1 h at 4°C. The supernatant was then removed as TBS-extractable Aβ, and the pellet was homogenized in 2% sodium dodecyl sulfate (SDS) in water with the same protease inhibitors and centrifuged by the same procedure, yielding the SDS-soluble Aβ. Finally, the remaining pellet was homogenized in 70% formic acid (FA) in water. The three extractions and sera samples were appropriately diluted and incubated overnight at 4°C in 96-well Immulon 4HBX Microtiters plates (USA), which were precoated with monoclonal antibody 6E10, which recognizes the Aβ 1-16 domain (Signet, Dedham, MA, USA) at 5 μg/ml in PBS buffer, pH 7.4. A biotinylated 4G8, which recognize Aβ17-21 domain (Signet, Dedham, MA, USA) at 1:2,000 dilution was used as secondary to detect serum total Aβ. Alternatively, biotin labeled monoclonal antibodies against Aβ 1-40 (11A50-B10, specific for the isoform ending at 40th amino acid of Aβ) or Aβ1-42 (12F4, specific for the isoform ending at 42th amino acid of Aβ, Signet, Dedham, MA, USA) were used as secondary antibodies for detecting Aβ40 and Aβ42, respectively. Detection was performed using a streptavidin-horseradish peroxidase conjugate at 1:1,000 (Vector Laboratories, Burlingame, CA, USA), followed by 3′,3′,5′,5′-tetramethylbenzidine substrate (TMB; Sigma). The reaction was stopped with 2 M sulfuric acid. Standard curves of Aβ 40 or 42 (rPeptide; USA) scaling from 125 pg to 10,000 pg/ml were used for quantifying Aβ levels. The standards were dissolved with buffer containing as the same concentration of SDS or FA as each of 2% SDS and 70% FA extractions for the assay of SDS and FA extraction samples, respectively. The ELISA plates were analyzed spectrophotometrically at 450 nm. Protein content of the cortex homogenates were measured using bicinchoninic acid assays following the manufacturer’s instructions (Pierce, Rockford, IL, USA). Using the amount of protein in the homogenate for each extraction, the final values of Aß in the brain were expressed as nanograms per milligram of protein.

Histology

A series of eight horizontal tissue sections spaced 300 μm apart were used for free-floating immunohistochemical staining to determine total Aβ using a rabbit antiserum at a concentration 1:10,000 (generously provided by Paul E. Gottschall, University of Arkansas for Medical Sciences, Little Rock, AK, USA). This antiserum is blocked in competition studies on by peptide Aß1-16, but not other peptides (10-20, 20-30, or 28-40) arguing that the antigen is within the first ten amino acids (as is true for most antisera raised against full-length Aß; see Dickey et al. 2001). Immunohistochemical procedural methods were described by Gordon et al. (2002). Briefly, eight sections from each animal were placed in multisample staining tray, and endogenous peroxidase was inhibited (10% methanol, 30% H₂O₂, in PBS). Tissue samples were then permeabilized (with 1% Triton X-100 in PBS containing 0.2% lysine) and incubated overnight in appropriate primary antibody. Sections were washed in PBS, then incubated in corresponding biotinylated secondary antibody (Vector Laboratories, Burlingame, CA, USA). The tissue was again washed after a 2-h incubation period and incubated with Vectastin^® Elite^® ABC kit (Vector Laboratories, Burlingame, CA, USA) for enzyme conjugation. Finally, sections were stained using 0.05% diaminobenzidine and 0.3% H₂O₂. Tissue sections were mounted onto slides, dehydrated, and coverslipped. Each immunochemical assay omitted some sections from primary antibody incubation to evaluate nonspecific reaction of the secondary antibody. Another series of eight sections were stained with 0.2% Congo red solution in NaCl saturated 80% ethanol, which detects the compact amyloid plaques made up of fibrillar Aß. Tissue amyloid plaque immunoreactivity (TAPIR) assay was performed as described previously (Hock et al. 2003) with minor modifications. Briefly, the tissue sections from unvaccinated APP-Tg mice with human Aβ plaques were incubated overnight with antisera raised against Qβ-Aβ 1-9 and Qβ-Aβ 28-40 peptides. Sera from the PBS treatment group were used as negative controls. Anti-mouse IgG and IgM were applied to detect IgG and IgM that bound Aβ plaques on the sections.

Quantification of the histological markers was performed using the Image-Pro Plus version 5.1 software (Media Cybernetics, Silver Spring, MD, USA) to analyze the percent area occupied by positive stain. We sample every eighth 25-μm section from the mouse brain, starting with a section between one and eight identified by a random number generator (West et al. 1991). Frontal cortical measurements were performed on laminae II-VI of the most anterior portion of the cortex (×100 magnification). Hippocampal measurements were performed by centering the field over the CA1, CA3, or dentate gyrus subfields at ×100 magnification on four sections in which the subfields could be reliably identified (usually sections 3–7). On the sections in which the hippocampal subfields could be reliably distinguished, a measurement of the cortical regions immediately posterior to the hippocampus was also performed centering the field between the surface of the brain and the white matter (laminae II-V). These hippocampal fields were combined for each mouse, and the cortical fields were similarly combined to yield a minimum of 12 measured fields per animal for each region. To assess possible treatment-related differences, the values for each treatment group were analyzed by one-way analysis of variance (ANOVA) followed by Fisher’s least significant difference (LSD) means comparisons.

Statistical analyses

Aβ ELISA and histochemical measurements were analyzed by ANOVA followed by Fischer’s LSD means comparisons using Stat View software version 5.0 (SAS Institute Inc, Cary, NC, USA).

Results

Antibody titers

The time course for IgG antibody development against the Aß peptide by N-terminal and C-terminal specific vaccines is shown in Fig. 1. Although the IgG response using the N-terminal vaccine was apparent after two inoculations (4.7 μg/ml), it declined to approximately 1 μg/ml over the next 2 months of the experiment, in spite of repeated monthly boosts with the vaccine preparation. On the other hand, the anti-Aß IgG antibodies raised against the C-terminal vaccine developed a high antibody titer of 12 μg/ml, which peaked about 2 months after the first inoculation and was largely maintained over the course of the experiment. At no time point did we detect anti-Aß signals in the mice injected with PBS vehicle, arguing that auto-antibodies to Aß are not common in these mice. IgG isotyping revealed that the predominant IgG isotype for the C-terminal vaccine was IgG2b, and for the N-terminal vaccine IgG 2b and IgG1 (data not shown).

Fig. 1.

Time course of the serum IgG response to the N-terminal, C-terminal, and vehicle immunizations. Amyloid precursor protein Tg mice were immunized with Qβ-Aβ 1-9 (squares, solid lines), Qβ-Aβ 28-40 (triangles, dotted lines), or phosphate buffer solution vehicle (circles, dashed lines). Blood was taken 2 weeks after the each inoculation (except the first). Arrows beneath the graph indicate time of each inoculation. 6E10 was used as the standard anti-Aß antibody for quantifying the titers. Data are presented as μg IgG/ml. Data are mean±sem. Sample size is five to six per group

The two vaccines produced IgG antibodies with antigen specificities expected for the Aß peptide sequences used (Fig. 2). The ELISA signal against Aß40 from sera obtained from mice administered the N-terminal vaccine was inhibited by Aß peptide 1-9, 1-16, and by full-length Aß1-40, but not Aß12-29 or Aß28-40. Conversely, the ELISA signal in the sera from mice administered the C-terminal vaccine was only inhibited by preincubation with Aß28-40 or full-length Aß1-40, but not by Aβ 1-9, Aß1-16, or Aß12-28.

Fig. 2.

Epitope mapping of antibodies. Aβ epitope mapping of anti-Aβ IgG induced by Qβ-Aβ 1-9 (open bars) and Qβ-Aβ 28-40 (solid bars) was performed by preincubating mouse sera at 1:250 dilution overnight at 4°C with or without Aβ peptides (10 μg/ml each). The remaining ability of the antiserum to bind ELISA plates coated with Aβ1-40 peptide was measured by ELISA using anti-mouse IgG as reporter. Values are normalized to the condition with no competitor added (Aβ free) for each serum sample (100%)

Because the IgG titers against the N-terminal vaccine were unexpectedly low, we evaluated if there might be IgM antibodies raised against the N-terminal vaccine. Using limiting titers to estimate antibody levels (we could not obtain an IgM anti-Aß antibody to use as a standard), we found an average titer of 1:35,000 in the IgM response against Aß using the N-terminal vaccine (Fig. 3), but very little IgM antibody in the mice immunized with the C-terminal vaccine (1:200 median titer), or the vehicle injected mice (1:100 median titer).

Fig. 3.

Time course of the serum IgM response to the N-terminal, C-terminal, and vehicle immunizations. Amyloid precursor protein Tg mice were immunized with Qβ-Aβ 1-9 (squares, solid lines), Qβ-Aβ 28-40 (triangles, dotted lines), or phosphate buffer solution vehicle (circles, dashed lines). Blood was taken after 2 weeks after the second inoculation and 2 weeks after subsequent monthly inoculations. Arrows beneath the graph indicate the time of each inoculation. The greatest antiserum dilution producing a significant signal (2.5 times background) were averaged and plotted on a log y-axis. Data presented are mean±SEM. Sample size is five to six per group

We also examined if these vaccines raised serum levels of the Aß peptide as evaluated by ELISA. We detected a roughly fourfold elevation of serum Aß content in the mice vaccinated with the C-terminal vaccine, but no apparent increase in the mice injected with the N-terminal vaccine (Fig. 4).

Fig. 4.

Serum Aß content in mice vaccinated against Aß. Sandwich ELISA was used to measure circulating Aβ levels. Sera were collected from final bleeding, 6E10 was used as capture antibody, biotinylated 4G8 as detection antibody, and the reaction was developed by streptavidin-horseradish peroxidase conjugate and TMB substrate. Data are presented as mean±SEM (n=4–5 per group). **P<0.01 compared to phosphate buffer solution control

Tissue amyloid plaque immunoreactivity (TAPIR) assay

In order to test whether the antisera bind to Aβ plaques in the brain of APP mice, we performed a TAPIR assay using antisera from the PBS, N-terminal, and C-terminal vaccinated mice (Hock et al. 2003). When an anti-mouse IgM (anti-μ chain) was used as the secondary antibody (Fig. 5a), Aβ plaques were only detected by the antisera induced by N-terminal vaccine. Many dense plaques were observed throughout the cerebral cortex (Fig. 5a panel A) and hippocampus (Fig. 5a panel B) in patterns typical for Tg2576 mice. In contrast, neither treatment with the C-terminal directed vaccine (Fig. 5a panel C and D) nor PBS control (Fig. 5a panel E and F) resulted in positive staining of plaques. However, when anti-mouse IgG (anti-γ chain) was applied as a secondary antibody, the plaques were stained with antisera induced by both N-terminal (Fig. 5b panels A and B) and C-terminal (Fig. 5b panels C and D) vaccines but not PBS controls (Fig. 5b panels E and F). Our data indicated that the anti-Aβ antibodies induced by the vaccines not only bind the synthetic human Aβ peptide in vitro but also react with endogenous Aβ plaques in vivo.

Fig. 5.

Antiserum reactivity with tissue amyloid plaques. a Immunoreactivity when anti-IgM was used as a secondary antibody. b Immunoreactivity when anti-IgG secondary antibody was used. Aβ plaques on the sections from unvaccinated amyloid precursor protein Tg mice were stained by antiserum from mice vaccinated with Qβ-Aβ1-9 (A and B), Qβ-Aβ 28-40 (C and D), or phosphate buffered saline vehicle control (E and F) in the frontal cortex (A, C, and E) and hippocampus (B, D, and F). Note that background staining is high and somewhat variable for these mouse antisera against mouse tissue. Results are representative of all mice in each group

Effects of vaccines on Aß deposition

We collected anterior cortical and hippocampal tissues for chemistry and histology at 17 months of age in these mice, which was 8 months after the first immunization. Aβ immunostaining (Fig. 6) in control mice was widespread throughout the cerebral cortex (panel a) and hippocampus (panel b) as expected. Numerous focal deposits were observed, along with some more diffuse accumulation such as along the hippocampal fissure. By visual inspection, there appears to be less immunostaining in the mice given the vaccines than in the mice given the PBS vehicle injections. Qβ Aβ 1-9 appeared to reduce immunostaining in both cerebral cortex (panel c) and hippocampus (panel d), as did Qβ Aβ 28-40 (panels e and f). When quantified by image analysis (Fig. 6g), this visual impression is confirmed for the changes in the cerebral cortex, with a significant reduction of Aß stained area to 30% of control in the cortical tissue after Qβ Aβ 1-9. The reduction after Qβ Aβ 28-40 in the cortex was not significant, but the mean was 50% lower than in the PBS-treated mice. In the hippocampus, the means were again less than the vehicle, but these were not significant statistically.

Fig. 6.

Total Aβ immunohistochemistry is reduced after both Qβ-Aβ1-9 and Qβ-Aβ28-40 vaccine administrations. Aβ immunostaining in the frontal cortex (a, c, e) hippocampus (b, d, and f) is shown after reaction with a rabbit polyclonal anti-Aß serum. Mice received treatment with phosphate buffered saline vehicle (a, b), Qβ-Aβ1-9 (c, d), and Qβ-Aβ28-40 (e, f). Scale bar=120 μm. Measurement of total Aβ immunohistochemistry by image analysis is shown in g. Data are presented as mean±SEM. Sample size is four to five per group. *P<0.05 compare to phosphate buffered saline vehicle control

Aß immunohistochemistry measures both diffuse and compacted amyloid deposits, while Congo red stains only the compacted amyloid plaques (Fig. 7). In the Tg2576 mice, many fewer amyloid deposits are positive for Congo red than for Aβ immunostaining. Nevertheless, treatment with Qβ Aβ 1-9 appeared to reduce Congo red staining in the cerebral cortex (panel c) to about 30% of control values (panel g); Qβ Aβ 28-40 also reduced cortical Congo red staining (panels e, g). In the hippocampus, the measurements did not demonstrate statistically significant reductions in Congo red staining with either vaccine (panels d, f, g).

Fig. 7.

Congo red stain is reduced after both Qβ-Aβ 1-9 and Qβ-Aβ 28-40 vaccines administration. Congo red stain in the frontal cortex (a, c, and e) hippocampus (b, d, and f); mice received treatment with phosphate buffered saline (PBS) vehicle (a, b), Qβ-Aβ 1-9 (c, d), and Qβ-Aβ 28-40 (e, f). Scale bar=120 μm. Measurement of Congo red staining by image analysis is shown in g. Data are presented as mean±SEM. Sample size is four to five per group. *P<0.05 compared to PBS vehicle control

Aß40 and Aβ42 levels were measured by ELISA also (Fig. 8). We found significant reductions in all three fractions (soluble, SDS extractable, and FA extractable) caused by both vaccine preparations. These reductions ranged between 40% and 60% of the PBS vehicle-treated control mice values. Moreover, the N-terminal and C-terminal fractions were virtually identical in their capacity to prevent Aß deposition. For ELISA measurement of Aß ending at amino acid 42, the vaccines were again effective in significantly reducing the soluble Aß fractions. In the SDS and FA extractable fractions, there was a trend for lower Aß42 levels, but the statistical power of the study was inadequate to determine if the reductions were statistically significant (we could not reject H₀).

Fig. 8.

ELISA assay shows both Qβ-Aβ1-9 and Qβ-Aβ28-40 vaccines reduced Aβ in the brain. Aβ 40 (8A) and Aβ 42 (8B) were measured in sequential extractions of Tris-buffer saline (TBS), 2% sodium dodecyl sulfate (SDS), and 70% formic acid. 6E10 was used as the capture antibody; 11A50-B10 and 12F4 were used as detection antibodies for Aβx-40 and Aβx-42, respectively. The mice received treatment with phosphate buffer solution (PBS) vehicle (hatched bars), Qβ-Aβ1-9 (open bars), and Qβ-Aβ28-40 (solid bars). For purposes of comparison, results are normalized to the appropriate PBS control treatment value. Control values were Aßx-40, TBS=410 pg/mg prot.; SDS=152 ng/mg prot.; FA=535 ng/mg prot. For Aßx-42, the PBS control values are TBS=260 pg/mg prot.; SDS=13 ng/mg prot.; FA=44 ng/mg prot. Data are presented as mean±SEM. Sample size is four to five per group. *P<0.05 compared to PBS control

Discussion

Previously, we reported that using bacteriophage-conjugated short Aβ peptides without using adjuvant as a vaccine elicited anti-Aβ immune responses in C57BL/6 mice (Chackerian et al. 2006). Recently, these data were replicated by an independent research group (Bach et al. 2009) using a different VLP method. However, none of these studies compared use of N-terminal- or C-terminal-specific vaccines. Critically, antibodies raised against the full-length Aß peptide are largely if not exclusively against epitopes in the N-terminal portion of the Aß molecule (Dickey et al. 2001; McLaurin et al. 2002). Because these residues are exposed when the APP molecule is at the cell surface, these antibodies may interact with not only Aß but also APP, potentially interfering with its functions and/or processing. However, the C-terminal residues of the Aß peptide are largely buried within the membrane when APP is at the cell surface, occluding them from potential interactions with C-terminal-specific antibodies. Thus, it is conceivable that the C-terminal specific antisera will have less impairment of APP function than N-terminal specific antibodies. However, a critical question is whether they can be effective in reducing Aß deposits. When combined with earlier work using C-terminal-specific monoclonal antibodies (Levites et al. 2006; Wilcock et al. 2006), there is little doubt that C-terminal-specific antibodies can effectively reduce amyloid loads in APP mice. Given that the mice in this study were started on the vaccine at an age when minimal deposits are present, yet measureable deposits were found at the end of the study, these data suggest that the vaccine was able to slow amyloid deposition, but provides no evidence regarding the removal of pre-existing deposits. However, to our knowledge, this is the first demonstration that active immunization with a C-terminal-specific vaccine can successfully prevent deposition of Aß in APP transgenic mice.

The production of IgG1 and IgG2b is indicative of a T-helper cell type 2 (Th2) response, which drives a largely humoral immune responses; whereas, the production of IgG2a and IgG2c is indicative of a T-helper cell type 1 (Th1) response, which drives cytotoxic T cell response. Previously, we found that Aβ N-terminal 1-9 peptide elicited anti-Aβ IgG which was predominately IgG2c, characteristic of a Th1-type immune response in C57BL/6 and SJL mice (Chackerian et al. 2006). Nevertheless, in this study, the isotypic profiles of IgG induced by the N-terminal peptide were dominated by IgG1 and IgG2b, a Th2 type immune response. The IgM response is also likely a Th2 type (Spooner et al. 2002), T cell independent immunoglobulin response.

To determine whether the antibodies induced by both N-terminal and C-terminal vaccines specifically bind their immunogen and cross-react with other domains of Aβ peptide, a series of competitive inhibition ELISAs were performed by incubating the antisera with various domains of Aβ peptide before incubation on microplates coated with Aβ1-40 peptide. Data showed that the vaccines did produce antisera with the expected epitope specificities, with N-terminal containing peptides inhibiting the N-terminal vaccine antisera and C-terminal containing peptides effective at competing with the C-terminal vaccine antisera.

Although the mechanisms by which immunotherapy acts remain unclear, several mechanisms were proposed within a decade. One of them is known as the peripheral sink hypothesis, in which circulating antibody removes Aβ from the brain by binding Aβ in plasma, resulting in a concentration gradient from brain to plasma and increased Aß efflux (or reduced uptake; DeMattos et al. 2001). In this study, the C-terminal vaccine increased the level of Aβ in the circulation fourfold greater than the N-terminal vaccine or control transgenic mice, suggesting the peripheral sink mechanisms may be active in this condition. In an independent passive immunotherapy study, we also found that the levels of Aβ in plasma were enhanced by a signal dose administration of antibody against Aβ C-terminal domain at least tenfold more than that found in the present study (Wilcock et al. 2006).

IgM is the first immunoglobulin produced following immunization as an immunological defense against infections. However, reports of an IgM response against Aβ with Aβ vaccination in an APP animal model have been rare (Sigurdsson et al. 2004; Maier et al. 2006). Eliciting an IgM response commonly requires an antigen with a repetitive structure. Bacteriophage can provide this repetitive structure to a single copy antigen, so in this study, bacteriophage-conjugated Aβ 1-9 peptide plus the gly-gly-cys three amino acid tag successfully induced a robust and prolonged IgM response against Aβ in Tg2576 mice. In contrast, the phage-conjugated Aβ 28-40 plus the tag failed to lead the IgM production, suggesting that the Aβ N-terminal is more prone to production of IgM than is the C-terminal. It is intriguing that the IgM titers were much greater than the IgG induced by the same immunogen from the first serum collection (after second inoculation) through the last one after the tenth inoculation in the absence of the immunoglobulin shift in class response. A plausible explanation for this finding is the phage, with its repetitive arrays, enhances the ability of the N-terminal Aβ peptide to directly trigger the B cell to differentiate into long-lasting memory plasma cells, which synthesize this IgM with high affinity or avidity, without the need for T cell help.

Importantly, even though the primary antibody response was IgM, the N-terminal vaccine was just as effective as the C-terminal vaccine in lowering amyloid deposition in the APP mouse brain. Although earlier work suggested that this might involve the peripheral sink hypothesis (Sigurdsson et al. 2004), this is unlikely in this case as no elevation of plasma Aß was detected in mice treated with the N-terminal vaccine (Fig. 4). Recently, another group (Banks et al. 2007) observed that anti-human Aβ monoclonal IgM antibody can cross the blood–brain barrier in spite of its large molecular weight. Coupled with the ability of the IgM produced by the vaccine to stain amyloid deposits in APP transgenic mouse brain sections (Fig. 5a), it seems reasonable to suggest that the IgM antibodies blocked Aß by an action in the central nervous system, instead of only working through the peripheral sink mechanism. Importantly, we cannot rule out that the low amounts of IgG produced by the N-terminal vaccine were solely responsible for the reductions in amyloid deposition. However, a significant contribution by IgM seems likely, given the similar extent of reduction caused by the two vaccines.

In sum, in this study, a bacteriophage was used as a vaccine platform instead of using an exogenous adjuvant to overcome the poor immunogenicity of short Aβ fragments. Only 10 μg of peptide per inoculation were used in these experiments which was tenfold less than our standard vaccine protocol with full-length Aβ 40/42 (Morgan et al. 2000). Previously, we reported that the Qβ bacterophage-Aβ (1-9) vaccine predominately elicited IgG 2c, a Th1 type isotype, response in C57BL/6 and SJL mice (Chackerian et al. 2006). However, in this study, the same immunogen induced a predominant IgG1 and IgG 2b immune response with only barely detectable levels of IgG 2c in the Tg2576 mice. It seems plausible that either the presence of self antigen in the APP mice or minor differences in genetic background strains can shift the generation of immunoglobulin isotypes to the vaccine and adjuvant (Spooner et al. 2002). As Aβ is a self protein in AD patients, the use of a VLP vaccine formulation may overcome the self-tolerance in AD patients to the Aβ vaccine and induce higher antibody titers response in AD patients. Therefore, it is likely that Qβ phage-Aβ vaccines will be an excellent candidate for active immunization studies in human AD.