Nonviral Aβ DNA vaccine therapy against Alzheimer’s disease: Long-term effects and safety

Preparation and Characterization of Nonviral Aβ DNA Vaccines.

We prepared three types of nonviral Aβ DNA vaccines using a mammalian expression vector, pTarget. The sequence of Aβ1-42 and additional sequences were inserted in the plasmid, as shown in Fig. 1A. The first one contains only the Aβ1-42 sequence with the Kozak sequence at the 5′ end (referred to as K-Aβ vaccine) (Fig. 1A1). To the second, the Igκ signal sequence of mouse Ig was added to improve the secretion ability (IgL-Aβ vaccine) (Fig. 1A2), and the third possesses the Fc portion of human Ig at the 3′ end to maintain stability (Aβ-Fc vaccine) (Fig. 1A3). Before in vivo administration, these DNA vaccines were transfected to HEK295T cells, and the secretion of Aβ1-42 peptide into the culture supernatant was assayed with Western blotting (Fig. 1B). The production of intracellular Aβ1-42 peptide was confirmed in all three vaccines by ELISA (data not shown). It was clearly demonstrated that the supernatants of cultured cells that had been transfected with IgL-Aβ and Aβ-Fc vaccines contained translated proteins (4.5 and 35 kDa, respectively), whereas K-Aβ-transfected cells did not secrete the peptide into the extracellular space. These findings indicate that the addition of the leader sequence is important for transportation of the protein to the extracellular space as reported in ref. 18 and that this event is critical for the effects of DNA vaccines (see below).

Reduction of Amyloid Burden by Aβ DNA Vaccination.

We used two types of regimens to examine the effect of Aβ DNA vaccination, i.e., prophylactic and therapeutic. For the prophylactic protocol, vaccine administration was started at 3–4 months of age, before the appearance of amyloid deposition. APP23 mice received 6 weekly and subsequent biweekly injections of the vaccines and were examined at 7, 9, 12, 15, and 18 months of age (Fig. 1C1). The paraffin-embedded sections of the brain were stained immunohistochemically with 6F∕3D against Aβ8-17, and the area of amyloid depositions was quantitated as the total sum of the pixels with nih image software.

In the first series of the prophylactic treatment, mice were analyzed at 7, 9, and 12 months of age (Table 1, Exp. 1). At 7 months of age, granular amyloid depositions were recognized in the frontal cortex in the control groups (empty-vector-administered and untreated mice) (Fig. 2B). At this stage, Aβ plaques were not detected in the hippocampus (data not shown). In sharp contrast, cortical Aβ depositions in mice treated with Aβ-Fc (Fig. 2A), IgL-Aβ, and Aβ (data not shown) vaccines were significantly reduced (P < 0.01). The Aβ burden was reduced to ≈15–30% of the untreated groups (Fig. 2E). At 12 months of age, amyloid depositions in untreated mice were increased, and some of them became large (>50 μg) in the frontal cortices of the untreated mice (Fig. 4C, which is published as supporting information on the PNAS web site). Cortical Aβ depositions were significantly reduced (P < 0.01) to ≈30–50% of the untreated group (Fig. 4D) after Aβ-Fc (Fig. 4A) and IgL-Aβ (Fig. 4B) vaccine treatment. Aβ depositions in the hippocampus were also decreased equally (P < 0.01) (Table 1, Exp. 1, Group 3). It was shown that the suppressive effect of Aβ-Fc vaccine was almost equal to IgL-Aβ vaccine. However, K-Aβ vaccine was less effective than the former two (Figs. 2E and 4D) and was not used in subsequent experiments. The second part of the prophylactic treatment analyzed the mice at 15 and 18 months of age (Table 1, Exp. 2). At these time points, the plaques in untreated groups had rapidly increased. Untreated APP23 mice showed an age-dependent increase of amyloid plaques in the cerebral cortex (Fig. 2G, open squares) and hippocampus (data not shown). The prophylactic protocol, using Aβ-Fc vaccine, revealed that the final reduction rate of Aβ burden in the cerebral cortex at 18 months of age was ≈38.5% of untreated groups (Table 1, Exp. 2 and Fig. 2G, closed triangles). These results demonstrated that two of the three vaccines produced in this study were effective in prophylactic treatment.

When considering the clinical applications, it is critical to know the effects of the vaccines in therapeutic application. For this purpose, the vaccination was started at 12 months of age, 6 months after the start of Aβ deposition, and the brains were examined at 15 (Fig. 2) and 18 (Fig. 5, which is published as supporting information on the PNAS web site) months of age. In therapeutic treatment, amyloid plaques in the cortex were significantly decreased (P < 0.01) (Fig. 2F) by Aβ-Fc and IgL-Aβ vaccination (Fig. 2D) compared with the controls (Fig. 2C). Aβ depositions in the hippocampus were also decreased (P < 0.01) (Table 1, Exp. 3). Although the therapeutic protocol (Fig. 2G, open circles) seemed to be less effective than the prophylactic one (Fig. 2G closed triangles), the difference was not significant. It should be noted that APP23 mice treated with the therapeutic protocol received DNA vaccines for only 3 and 6 months, respectively (Table 1, Exp. 3). Thus, Aβ DNA vaccines had sufficient effects, even if the vaccines were administrated after amyloid depositions appeared.

Recently, it was reported that the intracellular Aβ deposition in cortical pyramidal neurons is the first neurodegenerative event in Alzheimer’s disease development (19). Therefore, we counted the number of neurons containing intracellular Aβ depositions in the cortices of Aβ-Fc-vaccine-administered and control mice. Aβ-deposited neurons were significantly decreased with both the prophylactic (50.2% of untreated control, P < 0.01) and therapeutic (59.54%, P < 0.05) treatments at 15 months of age (Fig. 6, which is published as supporting information on the PNAS web site).

Change in the Serum Aβ Antibody Titer After Vaccine Administration.

The titers of serum anti-Aβ antibodies after the prophylactic treatment (Table 1, Exps. 1 and 2) were determined by ELISA. The levels of anti-Aβ antibodies were significantly increased (∗∗, P < 0.01; ∗, P < 0.05) 2- to 4-fold compared with the untreated and empty-vector-vaccinated mice (Fig. 3A). The titers showed an age-dependent increase in both treated and untreated mice, because the antibody production was induced in untreated APP23 mice by high levels of Aβ in the sera of aged mice. In contrast, the anti-Aβ antibodies in the sera of wild-type B6 mice were below the detection limit (data not shown). We also analyzed the relationships between the amounts of amyloid depositions and anti-Aβ titers (Fig. 3B). The amounts of amyloid depositions were significantly smaller in mice with high antibody titers. A significant correlation between the antibody titer and the reduction of Aβ burden was observed at 7 months (r = −0.642) (Fig. 3B, 7 months) and 9 months (r = −0.38965) (Fig. 3B, 9 months) of age by the CORREL function. The difference became less clear at a later stage (Fig. 3B, 12 months).

T Cell-Proliferation Assay After Vaccine Administration.

To determine whether DNA vaccination induces the T cell activation and proliferation that are key steps for the development of neuroinflammation, APP23 and B6 mice were injected with DNA vaccines. A group of mice were also immunized with Aβ peptide∕CFA. Three weeks after the first injection, lymphocytes were isolated and cultured with Aβ peptide (0–10 μM) for 3 days. Incorporation of [³H]thymidine was measured by using liquid scintillation spectrometry. T cells from APP23 mice did not react with Aβ peptide after both Aβ DNA vaccine and Aβ peptide administration (Fig. 3C). In sharp contrast, T cells from B6 mice responded significantly to Aβ peptide after immunization with Aβ peptide, but not after Aβ DNA vaccination (Fig. 3D). Notably, Aβ DNA vaccination did not induce T cell proliferation in either APP or B6 mice after three injections (Fig. 3C) and 5-month treatment (data not shown). These findings suggest that T helper (Th)1 cells in APP23 mice are in an immunologically tolerant state against Aβ1-42, probably because of long-term exposure to a high level of Aβ1-42.

Pathological Examination of the Brains of Vaccinated Mice.

The presence or absence of neuroinflammation in the brain was examined immunohistochemically after the long-term administration of DNA vaccines. Because the T cell assay demonstrated the significant differences between APP23 and B6 mice, pathological examinations were performed in both strains. Brain sections were stained with anti-CD5 mAb against T cells and with Mac-3 against macrophages. In the thymus (Fig. 7A, which is published as supporting information on the PNAS web site) and spleen (Fig. 7B), a large number of lymphocytes and macrophages were stained positively. In sharp contrast, inflammatory foci in the brain parenchyma and meninges were not detected in either APP23 (Fig. 7 C and D) or B6 (Fig. 7 E and F) mice.

Animals.

The APP23 transgenic mice used in this study express human APP751 cDNA with Swedish double mutation under the control of the neuron-specific mouse Thy-1 promoter (35, 36). The Thy-1 expression cassette lacks intron 3, which contains the elements required for expression in the thymus. RT-PCR and Western blot analysis of APP23 mice have confirmed the lack of detectable expression of the transgene in the thymus and spleen (D. Abramowski, C. Sturchler-Pierrat, and M.S., unpublished data). This finding is in agreement with a report by Moechars et al. (37), who used the same expression cassette and also found that the exogenous transgene with this promoter sequence is expressed only in the brain but not in the thymus. APP23 mice were initially established on a B6D2 background and have been continuously backcrossed to C57BL∕6J (B6). In APP23 mice, amyloid depositions appear from 6 months of age, predominantly in the neocortex and hippocampus. Aβ plaques have most of the characteristics of human Alzheimer’s disease plaques, including fibrillary Aβ cores, and are surrounded by dystrophic neuritis and activated glial cells. Region-specific amyloid-associated neurodegeneration, including neuron loss, synapse deficits, and cholinergic alterations, are present in these mice (38). Wild-type B6 mice were purchased from Charles River Breeding Laboratories (Kanagawa, Japan). All animal experiments were approved by the institute committee and performed in accordance with institutional guidelines.

The Development of DNA Vaccine.

We prepared three Aβ DNA vaccines using a pTarget mammalian expression system (Promega, Tokyo) (Fig. 1A). The DNA fragment, encoding Aβ1-42, was made to anneal two oligonucleotides covering the entire Aβ1-42 sequence. The Kozak sequence was inserted at the 5′ end of the Aβ1-42 sequence (referred to as K-Aβ vaccine). In the second vaccine, the signal sequence of mouse Igκ was added to the 5′ end of the Aβ1-42 sequence to improve the secretory efficiency of Aβ peptide (IgL-Aβ vaccine). The third vaccine was made by adding the Fc portion of human immunoglobulins to the 3′ end of the Aβ1-42 sequence to stabilize the secreted protein (Aβ-Fc vaccine). To prevent unwanted disulfide bonds, three cystines in the sequence were substituted with serine.

Transfection of DNA Vaccines and Western Blot Analysis.

Three DNA vaccines were transfected to HEK293T cells and the amounts of produced Aβ peptide were measured by using ELISA. HEK293T cells in 60–70% confluence were prepared on 6-well plates (Costar, Cambridge, MA). The cells were first cultured in serum-free RPMI medium 1640 for 2 h with one of three DNA vaccines (K-Aβ, IgL-Aβ, or Aβ-Fc) and Lipofectamine PLUS reagent (Invitrogen). Then, the cells were cultured overnight in RPMI medium 1640 supplemented with 5% FBS for cellular stabilization and growth. For the Western blot analysis, the cells were again cultured in serum-free medium for an additional 8 h to remove unnecessary proteins.

Culture supernatants and cell pellets were harvested and run on NuPAGE 12% Bis-Tris gel (Invitrogen) and transferred to the PVDF membrane (Immobilon-P; Millipore). After blocking with 10% nonfat milk, the blots were incubated with 6E10 (anti-human Aβ1–17 antibody, 1:100; Abcam, Cambridge, U.K.) at 4°C for 1 h, followed by incubation with biotin-conjugated anti-mouse IgG (1:1,000; Vector Laboratories) for 1 h and with ABC-HRP (VectorLaboratories) for 1 h. The blots were developed by enhanced chemiluminescence reagents (Immunostar kit; Wako Biochemicals) according to the manufacturer’s instructions.

Administration of the Vaccines.

DNA vaccines (100 μg) and bupivacaine (0.25 mg) in 100 μl was administered i.m. on a weekly basis for 6 weeks (39). The vaccines without bupivacaine were injected biweekly thereafter. To examine the prophylactic effect, vaccination was started at 3–4 months of age, before amyloid plaque appearance. The therapeutic treatment was started at 12 months of age, after amyloid plaque appearance.

Immunohistochemistry.

Mice were killed under deep anesthesia, and the brains were removed and immersion-fixed in 4% paraformaldehyde. Paraffin-embedded sections (6 μm) were stained immunohistochemically with mAb (6F∕3D) against Aβ8–17 (DAKO), anti-CD5 mAb against T lymphocytes (BD Biosciences Pharmingen) and Mac-3 against mononuclear phagocytes (BD Biosciences Pharmingen). For 6F∕3D staining, the sections were pretreated in formic acid for 3 min. The sections were then incubated in the primary antibody at a 1:200 dilution. After washing, the sections were incubated with biotinylated horse anti-mouse IgG (Vector Laboratories), followed by a horseradish peroxidase (HRP)-labeled Vectastain Elite ABC kit (Vector Laboratories). HRP-binding sites were detected in 0.005% diaminobenzidine and 0.01% hydrogen peroxide. CD5 (1:25) and Mac-3 (1:25) stainings were performed similarly, with overnight incubation of the primary antibodies.

Quantitative Analysis of Aβ Burden.

Aβ burdens were quantitated in the cerebral cortex and hippocampus, according to the method described in ref. 40. All of the procedures were performed by an individual blind to the experimental condition of the study. The images under an Olympus Vanox microscope were captured with a 3 charge-coupled device Olympus digital camera. The amyloid load was measured in 10 fields from the cingulated to retrosplenial cortex in the left hemispheres of the mice (600 × 400 μm each), chosen randomly. Analysis in the hippocampus was performed on the entire hippocampus in a similar manner. Aβ depositions that occupied the field were expressed as pixels by using the nih image software.

ELISA.

Microtiter plates were coated with 2 μg∕ml human Aβ1-40 (Peptide Institute, Osaka) in 0.1 M sodium carbonate buffer (pH 9.5) at 4°C overnight. After washing three times, plates were incubated for 2 h with serial dilutions of plasma samples in PBS in 12 rows of wells starting with 4-fold-diluted plasma (the greatest dilution tested was 1:213). The plates were washed and incubated with a 1,000-fold dilution of biotinylated anti-mouse IgG (Vector Laboratories), followed by incubation with 2-fold dilutions of Vectastain ABC-kit solution (Vector Laboratories). Bound antibodies were detected by using SIGMA FAST (Sigma-Aldrich), and the absorbance at 450 nm was read on an automated plate reader (Model 550; Bio-Rad). The antibody titer was defined as the reciprocal of the greatest dilution of plasma that gives half-maximal binding to Aβ, which was determined by dividing the highest OD₄₅₀ value in the dilution range of each sample by 2.

T Cell-Proliferation Assay.

The proliferative responses of draining lymph node cells were assayed in microtiter plates (Costar, Cambridge, MA) by the uptake of [³H]thymidine. Aβ1-42 peptide (50 μg) emulsified with CFA (twice) and Fc-Aβ vaccine (three times) was injected into APP23 mice or B6 mice, and then the drainage lymph nodes were taken 3 weeks after the first injection. Lymph node cells (2 × 10⁵ cells per well) were cultured with 0.3–10 μM Aβ1-42 peptide for 3 days and subsequently pulsed for 18 h with 0.5 μCi (1 Ci = 37 GBq) of [³H]thymidine (Amersham Pharmacia Biotech). Incorporation of [³H]thymidine was measured by liquid scintillation spectrometry.

Statistical Analysis.

Student’s t test or Mann–Whitney’s U test was used for the statistical analysis. Correlations between the antibody titer and the reduction of Aβ burden was estimated by the CORREL function.