Active full-length DNA Aβ42 immunization in 3xTg-AD mice reduces not only amyloid deposition but also tau pathology

Background Alzheimer’s disease (AD) is the most well-known and most common type of age-related dementia. Amyloid deposition and hyperphosphorylation of tau protein are both pathological hallmarks of AD. Using a triple-transgenic mouse model (3xTg-AD) that develops plaques and tangles in the brain similar to human AD, we provide evidence that active full-length DNA amyloid-β peptide 1–42 (Aβ42) trimer immunization leads to reduction of both amyloid and tau aggregation and accumulation. Methods Immune responses were monitored by enzyme-linked immunosorbent assay (ELISA) (antibody production) and enzyme-linked immunospot (cellular activation, cytokine production). Brains from 20-month-old 3x Tg-AD mice that had received DNA Aβ42 immunotherapy were compared with brains from age- and gender-matched transgenic Aβ42 peptide-immunized and control mice by histology, Western blot analysis, and ELISA. Protein kinase activation and kinase levels were studied in Western blots from mouse hemibrain lysates. Results Quantitative ELISA showed a 40% reduction of Aβ42 peptide and a 25–50% reduction of total tau and different phosphorylated tau molecules in the DNA Aβ42 trimer-immunized 3xTg-AD mice compared with nonimmunized 3xTg-AD control animals. Plaque and Aβ peptide reductions in the brain were due to the anti-Aβ antibodies generated following the immunizations. Reductions of tau were likely due to indirect actions such as less Aβ in the brain resulting in less tau kinase activation. Conclusions The significance of these findings is that DNA Aβ42 trimer immunotherapy targets two major pathologies in AD—amyloid plaques and neurofibrillary tangles—in one vaccine without inducing inflammatory T-cell responses, which carry the danger of autoimmune inflammation, as found in a clinical trial using active Aβ42 peptide immunization in patients with AD (AN1792).


Introduction
Immunotherapeutic approaches have high potential for successful treatment interventions in Alzheimer's disease (AD). Following the lessons learned from the first anti-amyloid-β peptide 1-42 (anti-Aβ 42 ) clinical trial (AN1792), in which patients with AD received an Aβ 42 vaccine and QS-21 adjuvant, which led to encephalitis in 6% of the treated patients, a major focus is now on avoiding autoimmune inflammation [1][2][3]. Ongoing clinical trials are pursuing passive vaccination with mouse monoclonal antibodies (mAbs) or fully human antibodies against Aβ 42 peptide epitopes to avoid complications from autoimmunity [4][5][6][7]. A recent study in which patients received passive immunotherapy with an mAb targeting oligomeric or prefibrillar Aβ 42 reported positive results regarding amyloid reduction in the brain as well as improved cognitive measurements [8].
Besides amyloid accumulation, tau aggregation and spreading have been associated with progression of AD. In fact, increased tau levels showed high correlation with cognitive decline in patients with AD [9]. Tau immunotherapy is being evaluated in various preclinical and clinical trials as well, using active immunizations with peptides from different parts of the tau protein or passive immunizations using polyclonal or mAbs [10][11][12][13][14][15]. Antitau antibodies have been shown to act inside and outside of neurons and to reduce tau hyperphosphorylation as well as pathogenic tau seeding [16][17][18][19][20].
We report, for the first time in an AD mouse model, that active DNA Aβ 42 immunization into the skin targets two pathologies: amyloid-containing plaques and tau. DNA vaccination, in which not the antigen (peptide or protein) but the DNA encoding this peptide is administered, is an alternative route of vaccination. Genes encoded by the DNA are expressed within the skin, and the peptides are taken up by dendritic cells traveling to the regional lymph nodes and presenting the antigen to B and T cells [21]. Immune responses to DNA or peptide immunization differ qualitatively. We have shown previously that full-length DNA Aβ 42 trimer immunization is noninflammatory and induces a regulatory immune response [22][23][24][25]. DNA Aβ 42 trimer immunization has been shown to be effective in removing amyloid from the brain in immunized double-transgenic mice (APPswe/PS1 [26][27][28]). In the present study, we used a triple-transgenic AD mouse model (3xTg-AD) that exhibits Aβ and tau pathologies characteristic of human AD [29,30]. We found that immunotherapy with DNA Aβ 42 trimer leads to reduction of Aβ 40 /Aβ 42 peptides and amyloid plaques, and we show for the first time that DNA Aβ 42 trimer immunization leads also to significant reduction of tau from the mouse brain.

Methods
Animals 3xTg-AD [B6;129-Tg(APPSwe,tauP301L)1Lfa Psen1 tm1Mpm / Mmjax, MMRRC Stock No: 34830-JAX] mice had been purchased from the Mutant Mouse Research and Resource Center at The Jackson Laboratory and were bred and housed at the UT Southwestern Medical Center animal facility under conventional conditions. This mouse model had been developed by Oddo and colleagues [29,30]. Animal use was approved by the UT Southwestern Medical Center Animal Research Committee, and animal research was conducted under the Animal Research: Reporting of In Vivo Experiments guidelines [31].

Study design
Cohorts of 3xTg-AD mice were immunized with a DNA Aβ 42 trimer vaccine, Aβ 42 peptide (rPeptide, Watskinville, GA, luciferase (Luc) control DNA, or left untreated as controls. This mouse model that had been developed by Oddo and colleagues develops plaque and tangle pathology [29,30]. Cohort 1 consisted of 16 Tg female mice and 8 wild-type controls, and cohort 2 consisted of 34 Tg female mice. Parallel immunized groups of 3xTg-AD males (16 males in cohort 3, 15 males in cohort 4) showed no plaque pathologies at 18 and 20 months of age, so this study used females only. The mice were vaccinated at a total of 13 time points until 20 months of age for final analyses. Collected brains were cut in a sagittal plane. One hemibrain was frozen and used in enzyme-linked immunosorbent assays (ELI-SAs) and Western blots, and the other half was fixed for immunostaining with anti-Aβ and antitau antibodies.

Immunizations and collection of blood samples
Immunizations were started in 4-month-old mice in groups of four to eight mice (3xTg-AD and 129/SvJ wild-type controls) with three initial immunizations in biweekly intervals ( Fig. 1) with a Gal4/DNA Aβ 42 trimer double-plasmid system (4 μg of DNA/immunization, ratio of 3:1 DNA Aβ 42 trimer responder plasmid/Gal4 activator plasmid) via intradermal injection using a Helios gene gun (Bio-Rad Laboratories, Hercules, CA, USA) or via intraperitoneal injections of Aβ 42 peptide (100 μg of peptide/immunization) with Quil-A (Sigma-Aldrich, St. Louis, MO, USA) as adjuvant as previously described [22][23][24][25]. The immunizations were boosted in 6-week intervals until the mice were 18 or 20 months old (up to 13 immunizations) (Fig. 1a). Control mice received Luc DNA immunizations (group 1) or no treatment (naïve controls, group 2). Blood samples were collected at different time points throughout the study 10 days following the respective immunization time points.

Positive antibody staining area quantification
The Aβ and tau immunoreactive areas were quantified using the "area measure" tool in ImageJ software (National Institutes of Health, Bethesda, MD, USA [32]).
Immunostained sections (sagittal sections of mouse brain) were imaged with a 20× objective and were converted into 8-bit grayscale. The Analyze > Measure tool was used to measure the total area occupied by positive staining in each image. The total area was averaged for the sections per mouse group. Values are arbitrary units expressed as mean ± SEM per area.
Anti-Aβ 42 antibody ELISA and cytokine enzyme-linked immunospot assays ELISAs for antibody levels in mouse plasma were performed according to standard procedures. Cytokine concentrations from cell culture supernatants and enzymelinked immunospot (ELISPOT) assays to determine frequencies of cytokine-secreting cells were performed Timeline of the experimental procedures, anti-amyloid-β peptide 1-42 (anti-Aβ 42 ) antibody production upon immunization with DNA Aβ 42 trimer and Aβ 42 peptide in triple-transgenic Alzheimer's disease (3xTg-AD) and wild-type mice and cytokine secretion in restimulated splenocyte cultures. a Immunizations, blood draws, and final analyses are shown along the experimental timeline of 20 months. b High levels of anti-Aβ 42 antibodies (micrograms per milliliter of plasma) were found in all of the immunized mouse groups following the last immunization (wild-type mice and 3xTg-AD mice). Blue symbols indicate mice that had received DNA Aβ 42 trimer immunizations; yellow symbols indicate mice that had received Aβ 42 peptide immunizations. Antibody levels of two groups of 20-month-old 3xTg-AD mice are shown as group 1 (G1) and group 2 (G2). Plasma samples had been used in a 1:1000 dilution. Samples were run in triplicates, and the assay was repeated twice. Antibody isotype analyses from DNA Aβ 42 trimer-immunized 3xTg-AD mice (c) and Aβ 42 peptide-immunized 3xTg-AD mice (d). White bars show levels of anti-Aβ 42 antibodies of the immunoglobulin G1 (IgG1) isotype; gray bars show IgG2a antibody levels; hatched bars show IgG2b antibody levels; and black bars show IgM antibody levels. Differences in the amount of IgG1 (Th2) and IgG2a/c (Th1) antibody levels are statistically significant (p = 0.0068). Levels were measured as optical density at 450 nm (OD450). Plasma samples had been used in a 1:500 dilution, analyzed in triplicates, and the assay was repeated twice. e Antibody isotype profile of plasma samples from peptide-immunized mice in a 1:20,000 dilution. Interferon (IFN)-γ (f) and interleukin (IL)-17 (g) enzyme-linked immunospot analysis of splenocytes from 20-month-old 3xTg-AD mice (n = 4/group) and 129/SvJ wild-type mice (n = 4/group) that had received 13 Aβ 42 peptide or 13 DNA Aβ 42 trimer immunizations, respectively. No IFN-γ-or IL-17-secreting cells were found in DNA Aβ 42 -immunized mice, whereas high numbers of cells secreting IFN-γ and IL-17 were found in splenocytes from peptide-immunized mice upon Aβ 1-42 peptide or Aβ 10-26/17-31 peptide mix restimulation in vitro. **, and **** indicate p values of ≤ 0.01 and ≤ 0.001, respectively (unpaired Student's t test) according to standard procedures and as previously described using commercially available antibody sets for mouse interferon (IFN)-γ, interleukin (IL)-17, and IL-4 (eBioscience, San Diego, CA, USA) [23][24][25].

Aβ and tau ELISAs
For semiquantitative analyses of total Aβ 42 , Aβ 40 , and tau (total tau, pT231, pS396, pT181, and pS199) levels in the brain, standard ELISAs were used (Thermo Fisher Scientific). Frozen mouse hemibrains of female mice were homogenized with a Dounce homogenizer in 10 volumes (wet brain weight) of extraction buffer [

Statistics
For statistical analysis (unpaired Student's t test with two-tailed p values, nonparametric Mann-Whitney U test, parametric multiple comparisons one-way analysis of variance [ANOVA] and column statistics), Prism software version 6 for Windows (GraphPad Software, La Jolla, CA, USA) was used. p ≤ 0.05 was considered significant.

Histology showing amyloid reduction from brain
In the initial studies, we used male and female mice and found large differences in the pathology between sexes. In 20-month-old mice, large numbers of Aβ plaques were found in the subiculum of the hippocampus in female mice ( Fig. 2a), whereas no plaques were found in the 20-month-old males (Fig. 2b). Also, for tau antibody staining (HT7, AT180) in parallel sections, much less pathology was found in male mice (data not shown), and therefore we continued immunotherapy in the following groups only in females. In 18-month-old mice, amyloid plaques were abundant in the female mice (Fig. 2c). In age-matched male mice, only a few neurons with intracellular Aβ 42 staining were found, but no plaques (Fig. 2d).
Aβ 42 immunotherapy led to a reduction of the number of amyloid plaques in the hippocampus of treated mice. In Fig. 2e-h, staining for NeuN, which stains neurons (red color), and an Aβ antibody (McSA1), which stains amyloid plaques (brown color) are shown for the hippocampal area for representative examples of the different mouse groups in one experimental cohort. The mAb McSA1 recognizes the N-terminal region of the human Aβ peptide (Aβ [1][2][3][4][5][6][7][8][9][10][11][12] ). This epitope is present in β-C-terminal fragment and amyloid precursor protein (APP) as well, but McSA1 has been reported as highly specific for Aβ as opposed to APP or soluble APP following competition studies with these antigens [33,34]. Figure 2e shows staining of the hippocampus subiculum of a 20-month-old 3xTg-AD control female mouse. Figure 2f shows this area stained for neurons and amyloid in a 20-month-old wild-type control mouse. A reduction of amyloid plaques was seen in all mice that had received Aβ immunotherapy. Representative sections are shown for one Aβ 42 peptide-immunized mouse (Fig. 2g) and one DNA Aβ 42 -immunized 3xTg-AD mouse (Fig. 2h).
Immunohistological staining of plaques in the brains of these mice was subjected to the counting of plaques > 10 μm in corresponding 1-mm 2 areas (subiculum/ CA1) of 15 control mice (7 DNA Aβ 42 -immunized mice and 8 Aβ 42 peptide-immunized mice) by two blinded experimenters. These analyses showed significantly reduced plaque numbers in the DNA Aβ 42 -immunized mice (p = 0.0238 by Student's t test compared with control mice) and a nonsignificant reduction in the Aβ 42 peptide-immunized mice (p = 0.6809). Also, the difference in plaque numbers between the DNA Aβ 42 -and peptide-immunized mice was significant (p = 0.0487) (Fig. 2i).

Histology showing reduction in levels of phospho-tau
The use of the 3xTg-AD mouse model allowed us to analyze a second pathology of human AD, which is the hyperphosphorylation of tau and development of neurofibrillary tangles. IHC of 3xTg-AD brain sections with different antibodies specific for tau molecules phosphorylated at specific residues (AT180, AT8, AT270, pT404, pS212, Tyr18) showed that Aβ 42 immunotherapy also led to a significant reduction in the levels of tau phosphorylation. In Fig. 3a, the age progression for tau phosphorylation in the 3xTg-AD mouse model is shown. Brains from 2-, 4-, 7-, 9-, 12-, and 18-month-old mice (n = 4/group) were harvested, and PFA-fixed, paraffin-embedded sections were analyzed with the mAb AT180, which detects tau phosphorylated at residue T231. In the comparison of the staining pattern with brains from 18-month-old 3xTg-AD mice, which had received DNA Aβ 42 immunizations, we observed that the AT180 staining intensity of the immunized 18-month-old mice appeared more like the staining intensity in brains from 7-or 9-month-old mice (Fig. 3b). Sections from four 18-month-old Luc immunized control mice, five 18-month-old DNA Aβ 42 -immunized mice, and six 18-month-old Aβ 42 peptide-immunized mice were semiquantitatively analyzed with the area measure tool in Ima-geJ software. The results showed an about 40% reduction after DNA Aβ 42 immunization and an approximately 20% reduction after Aβ 42 peptide immunization (Fig. 3c). However, owing to high SDs and the small number of control animals, the results were not statistically significant.
Staining with the AT8 antibody specific for pS201/ pT205, which is a late tau phosphorylation site [35], was less prominent in 18-month-old mice, but good staining was observed in 20-month-old mice, which showed reduction of AT8 staining in DNA Aβ 42 trimer-immunized mice. Figure 3d shows that AT8-positive neurons were detected in the hippocampus of 20-month-old 3xTg-AD control mice (sections from two mice). Two representative sections from the Aβ 42 peptide-immunized 20-month-old 3xTg-AD mice are shown in Fig. 3e. The brains showed fewer AT8-positive neurons than in the control animals. Much less staining was found in DNA Aβ 42 trimer-immunized mice. Figure 3 shows the respective brain sections of the hippocampus from two mice (insets show higher magnification of subiculum in Fig. 3f ). The histological data indicating a possible reduction of tau in the Aβ 42 -immunized mice led to further substantiation of this finding by Western blot analysis and a panel of commercially available tau ELISAs that allowed testing for statistical significance of reduction of different tau phosphorylation patterns.

Western blot analysis of total tau
The reduction of tau in mice that had received Aβ 42 immunotherapy was further analyzed using Western blotting of the brain lysates. In the comparison of total tau detected with the mAb Tau12, it was found that both immunotherapies led to a reduction in tau. The reduction was not significant in Aβ 42 peptide-immunized mice and was higher and significant in DNA Aβ 42 -immunized mice (p values of 0.0302, 0.0142, and 0.0023 from three independently performed Western blot analyses with detergent-soluble brain lysates). Figure 3g and h shows the results of one of these experiments (Western blot and ImageJ analysis of gray-level intensities of the bands, respectively). Total tau and phosphorylated tau were further analyzed by Western blotting, and the results are shown in Fig. 4. All band intensities were normalized to band intensities found in the reprobing of the Western blots with antibodies to housekeeping proteins. In the detergent-soluble fractions, tau detected with the mAb Tau12 was significantly reduced in brain lysates from DNA Aβ 42 -immunized mice (p = 0.0059 by Student's unpaired t test) (Fig. 4a). The intensity of the Western blot band reactive to the mAb AT8 was only slightly reduced in DNA Aβ 42 -immunized mice (p = 0.3224, nonsignificant). The AT8-reactive protein band was found at higher molecular weight (about 65 kDa), which might correspond to the 64 kDa tau, a Tris-buffered saline-extractable hyperphosphorylated tau species described in the rTg4510 mouse brain (Fig. 4a, middle panel) [36]. In Fig. 4b, two different total human tau antibodies, 43D and HT7, were directly compared in parallel-run SDS-PAGE. Significant reductions of tau were found in the brain samples from DNA A B Fig. 4 Western blot analyses for total and phosphorylated tau. Equal amounts of proteins from detergent-soluble brain lysates of 20-month-old tripletransgenic Alzheimer's disease (3xTg-AD) mice (D1-D5 = DNA Aβ 42 -immunized mice, P1-P4 = amyloid-β 1-42 [Aβ 42 ] peptide-immunized mice, C1-C5 = 3xTg-AD control mice, wt = wild-type controls) were separated by SDS-PAGE, blotted onto nitrocellulose filters, and probed using antibodies specific for total human tau (a, upper panel), and phosphorylated tau AT8 (a, middle panel), and β-tubulin as a loading control (a, bottom panel). The graph on the right-hand side of the SDS-PAGE pictures shows analyses of the band intensities performed with ImageJ software. All gray-level intensities of tau protein bands were normalized to the gray-level intensities of protein bands of the housekeeping proteins β-tubulin or β-actin, respectively. The reduction of total tau in the DNA-immunized mice compared with the 3xTg-AD control animals was highly significant (p = 0.0059). Of note, gray-level intensities for sample D2 were not included in thess calculations, because the loading control for this sample indicated a much lower protein content (a, bottom panel). b A comparison of total tau levels in DNA-immunized mice, 3xTg-AD control mice, and wt control mice in Western blots is shown using two different antibodies. In the upper panel, 43D (Tau1-100) was used for detection; in the middle panel, antibody HT7 was used; and in the lower panel, the same membrane was probed with a β-actin antibody as a protein loading control. The graph on the right-hand side of the panels shows the analyses of graylevel intensities for the protein bands with ImageJ software normalized to gray-level intensities of the housekeeping protein β-actin. Differences were statistically significant with p values of 0.0152 (HT7) and 0.0138 (43D). * and ** indicate p values of ≤ 0.05 and ≤ 0.01, respectively (Mann-Whitney U test) Aβ 42 -immunized mice (HT7 antibody, p = 0.0152; 43D antibody, p = 0.0138).
These results are consistent with the ELISA results described in the "Quantification of tau in ELISAs" section below. Although the reductions in the detergent-soluble brain lysate fractions were obvious but statistically not significant, reductions in the nonsoluble brain lysates were highly significant. However, the nonsoluble fractions could not be tested, owing to the extraction method used with 5 M guanidine for the solubilization of the pellet. These samples are not compatible with SDS-PAGE. In future mouse cohorts, we will use a different extraction protocol allowing the nonsoluble brain lysate fractions to be analyzed in Western blots (SDS-PAGE).

Quantification of Aβ x-42 and Aβ x-40 in ELISAs
After analysis of brain histology as shown in Fig. 2, ELI-SAs were used for semiquantitative analyses of reduction of Aβ x- 40 and Aβ x-42 peptides in DNA Aβ 42 trimer-and Aβ 42 peptide-immunized female 3xTg-AD mice. An increase of Aβ 42 and Aβ 40 peptides in brains from 3xTg-AD mice with age is shown in Fig. 5a. ELISAs were also used to quantify the reduction of Aβ x- 40 and Aβ x-42 peptides due to DNA Aβ 42 trimer and Aβ 42 peptide immunization ( Fig. 5b and c). Statistical significance for reduction of Aβ 42 and Aβ 40 was reached in the comparison of DNA Aβ 42 trimer-immunized mice (n = 7, blue bars) compared with control animals (n = 14, black bars) in the nonsoluble fractions (p = 0.0461, Mann-Whitney U test, for Aβ x-42 ; p = 0.0125 for Aβ x-40 ). These reductions were nonsignificant in the one-way ANOVA (Fig. 5b). In the soluble brain lysate fractions, a reduction of both Aβ peptides was highly significant (p < 0.0008, Mann-Whitney U test; p = 0.0123, one-way-ANOVA, for Aβ x-42 ; p = 0.0017, Mann-Whitney U test; p = 0.0028, one-way ANOVA for Aβ x-40 ) (Fig. 5c) in DNA Aβ 42 -immunized mice. Aβ x-42 peptides were also reduced in brains from Aβ 42 peptide-immunized 3xTg-AD mice in the nonsoluble lysate and detergent-soluble lysates, but levels did not reach statistical significance (p = 0.2766 for nonsoluble Aβ x-42 , p = 0.0815 for soluble Aβ x-42 ). Much less removal was found for Aβ x-40 peptides in brains from Aβ 42 peptide-immunized mice (Fig. 5b and c, yellow bars, right-hand graphs).

Quantification of tau in ELISAs
Histological analyses of the mouse brains with tau antibodies AT180 and AT8 (early and late tau phosphorylation) showed reduced staining in the immunized mice (Fig. 3). ELISAs were used for detection of total tau, pT231 tau, p396 tau, pT181 tau, and pS199 tau in the semiquantitative analyses of tau reduction in DNA Aβ 42 trimer-and Aβ 42 peptide-immunized 3xTg-AD mice. Tau was reduced in both mouse groups, which had received Aβ 42 immunotherapy or DNA or peptide vaccine (Fig. 6a-e, Table 1). However, statistical significance was reached only in the DNA Aβ 42 trimer-immunized mice (Table 1).
In comparison of the two Aβ immunotherapies, a better reduction with high significance for phosphorylated tau molecules was found in DNA Aβ 42 trimer-immunized mice. Percentages of reduction were calculated for the groups, and the results are shown in Table 1. A greater than 20% higher reduction was found in the detergent-soluble brain fractions of DNA-immunized mice for pT181 and pS396. This was statistically significant in the comparison of DNA-and peptide-immunized mice for pS396 (p < 0.0311) ( Table 1). For the nonsoluble brain fractions, 12-25% higher reductions were found in lysates from DNA Aβ 42 trimer-immunized mice for total tau, pT231, pT181, and pS396. These values were statistically significant for the comparisons with the age-and gender-matched control mice (Fig. 6) and also in the comparison between the differently immunized groups of mice ( Table 1).

Analyses of kinase variations
Western blot analyses were performed to detect whether different enzymatic kinase patterns could be found in brains from immunized mice. Significantly reduced levels of phosphorylated MEK (MAP2K), and phosphorylated ERK1/2 (p44/p42 mitogen-activated protein kinase [MAPK]), as well as reduced levels for the activated form of GSK3β (Y216), were found in brains from DNA-immunized mice. Figure 7 shows the detection MEK1/2 and phospho-MEK1/2 (Fig. 7a), as well as ERK1/2 and phosphorylated ERK1/2 (Fig. 7b), in brain lysates from seven DNA Aβ 42 -immunized mice compared with seven age-and gender-matched 3xTg-AD control mice and two 20-month-old wild-type mice.
Results from the semiquantitative analysis of gray-level intensities (ImageJ software) are depicted in Fig. 8. Reductions in protein levels of phospho-MEK1/2 (Fig. 8a), total ERK1/2, and phospho-ERK1/2 (Fig. 8b) in DNA Aβ 42 -immunized mice were significant (p values of 0.0379, 0.0006, and 0.0087, respectively, by Mann-Whitney U test). Significant reductions were also found for protein levels of activated GSK3β (p = 0.0006) (Fig. 8c). These data were further normalized against the protein levels of total MEK1/2, total ERK1/2, and total GSKα/β for each of the bands individually to compensate for the possibility of different overall protein levels for the tested enzymes in the brain lysates and shown as a percentage of protein (percentage of phosphorylated MEK, ERK, and GSKα/β). The percentage difference for phospho-MEK1/2 was highly significant between control and DNA Aβ 42 -immunized mice (p = 0.0031). In the comparison of phosho-ERK1/2 with total ERK1/2 in the DNA Aβ 42 -immunized mice, the reduction was not significant, because these mice already had less total ERK1/ 2. The percentage reduction of phospho-GSK3β (Y216) in the DNA Aβ 42 -immunized mice was highly significant (p = 0.006). No differences in protein levels between the mouse groups were observed for the proteins MEK1/2 (Fig. 7a), GSK3α/β (Fig. 7c), and the housekeeping protein β-tubulin (Fig. 7d). Of note, a blot with GSK3α/β is shown in the comparison for activated GSK3β because it appears that there is weak cross-reactivity of this specific antibody with both GSK3 bands (Fig. 7c, lower panel), but differences were seen only for the strong reactivity with GSK3β phosphorylated at residue Y216 (46 kD band). Only this activated form of GSK3β is described and discussed. No significant differences were found for total GSK3β protein levels in brain lysates from control and immunized mice (data not shown).

Discussion
DNA Aβ 42 immunotherapy results in significant reductions in Aβ 42 peptide and plaque load in brains of the 3xTg-AD mouse model at 20 months of age, consistent with our previous results in double-transgenic mice [26,27]. New findings shown with this vaccine for the first time were significant reductions of total tau and phosphorylated tau in brains of mice that had received active DNA Aβ 42 trimer immunizations. This finding was confirmed by histology, Western blot analysis, and ELISA. Despite the 10× levels of anti-Aβ antibodies in peptideimmunized mice, peptide immunization was less efficacious, which is indicative of different Aβ species detected and removed by the antibodies generated following DNA Aβ 42 immunization (e.g., Aβ oligomers). In fact, this was highly consistent throughout the study with more Aβ and more tau removed in DNA Aβ 42 trimer-immunized mice than in Aβ 42 peptide-immunized mice in all three assay systems used (immunohistology, Western blotting, ELISA). We had previously shown that the expression of the DNA Aβ 42 trimer vaccine in skin shows production of Aβ oligomers [37]. We had also previously shown that the epitope specificity of antibodies produced after DNA Aβ 42 immunization differs from the Aβ 1-15 B-cell epitope specificity and shows a wide reactivity with epitopes across the Aβ 1-42 peptide [38][39][40]. Aβ oligomers in particular activate tau kinases, leading to hyperphosphorylation, and Aβ oligomers are also strong activators for cellular caspases, leading to tau cleavage and tau aggregation. Hyperphosphorylated tau and truncated tau are both prone to self-aggregation and tau accumulation in neurons, and phosphorylation of d Analysis of tau phosphorylated at residue T181 (pT181). e Analysis of tau phosphorylated at residue S199 (pS199). All data are based on the analyses and comparison of 7 DNA Aβ 42 trimer-immunized mice, 9 Aβ 42 peptide-immunized mice, and 14 age-and gender-matched 3xTg-AD control mice. All samples were run in duplicates, and the assay was repeated twice. * p ≤ 0.05, ** p ≤ 0.01, and *** p ≤ 0.005 (Mann-Whitney U test) specific residues in tau (e.g., S422) are important for caspase-mediated cleavage. Thus, less tau phosphorylation leads to less tau truncation via caspase-mediated cleavage and therefore reduces tau aggregation and total tau levels, explaining why Aβ 42 immunization and reduction of Aβ 42 peptides in brain led also to reduction of total tau [41][42][43][44][45][46][47][48][49]. DNA Aβ 42 immunotherapy led to a noninflammatory immune response with no T-cell proliferation and no inflammatory cytokines produced during the cellular immune responses in the 3xTg-AD mouse model, similar to the immune responses we had found in the wild-type mouse model [22][23][24][25]. Although in the Balb/c wild-type mouse strain IgG1 was a dominant IgG antibody isotype in the humoral immune response [37], other mouse strains showed also a strong anti-Aβ 42 IgG2b antibody production similar to the one found in the 3xTg-AD mouse used in the present study (unpublished data). Aβ 42 peptide immunization led to a mixed immune response with high levels of all antibody isotypes, including IgG2a/c, and high levels of inflammatory cytokines in AD mouse models and wild-type mice [22][23][24][25]28]. In a prime boost study, in which the immune response was first primed with Aβ 42 peptide immunizations and then boosted with DNA Aβ 42 immunizations in wild-type mice, we found that even though the anti-Aβ 42 antibody isotype profile had high levels of "inflammatory" IgG2a/c antibodies, no inflammatory cytokines were detected in the cellular in vitro assays, providing evidence that the DNA immunization resulted in the downregulation of inflammatory cellular responses [50]. Antibody isotypes strongly influence the therapeutic effect of a treatment or vaccine because the different antibody isotypes have different effector functions (complement binding, Fc receptor binding). In AD immunotherapy, microglial activation is thought to help remove excess Aβ from the brain, so that FcR binding is not a negative feature of the antibody per se. Furthermore, the epitope detected A B C D Fig. 7 Significant changes in enzymes of the Ras/mitogen-activated protein kinase kinase (MEK)/extracellular signal-regulated kinase (ERK) signaling pathway and glycogen synthase kinase 3β (GSK3β) following DNA amyloid-β 1-42 peptide (Aβ 42 ) immunization. Equal amounts of proteins from soluble brain lysates of 20-month-old triple-transgenic Alzheimer's disease (3xTg-AD) mice (D1-D7 = DNA Aβ 42 -immunized mice, C1-C7 = 3xTg-AD control mice, wt = wild-type control mice) were separated by SDS-PAGE, blotted onto nitrocellulose filters, and probed using antibodies specific for MEK (a, upper panel) and its active form phosphorylated MEK (a lower panel), total ERK1/2 (b, upper panel) and the phosphorylated forms of ERK1/2 (b lower panel), and GSK3α/β (c, upper panel) and activated GSK3β (c, lower panel). Of note, a blot with GSK3α/β is shown in the comparison for activated GSK3β because it appears that there was weak crossreactivity of this specific antibody with both GSK3 bands (c, lower panel), but differences were seen only for the strong reactivity with GSK3β phosphorylated at residue Y216 (46 kD band). As a loading control, the blots were reprobed with the housekeeping protein β-tubulin (d). All assays were performed three times in independent experiments. Shown are representative results from one of these assays by the respective antibody pool is crucial for removal of amyloid from the brain [51][52][53]. Thus, the multivalent nature of the humoral immune response following DNA Aβ 42 immunization is beneficial in many aspects.
To address how either DNA Aβ 42 or Aβ peptide vaccinations can cause both Aβ and tau reduction, we investigated a number of kinases involved in tau phosphorylation that had also been shown to be activated by Aβ 42 peptides and in particular Aβ 42 oligomers [54][55][56]. We were able to show differences for the activated/phosphorylated kinases MEK1/2, p40/p42 MAPK1 and MAPK2 (ERK1/2), and GSK-3β in brain protein lysates from female DNA Aβ 42 -immunized mice compared with the age-and gender (female)-matched 3xTg-AD control mice, supportive of the assumption that a higher removal of Aβ oligomers after DNA Aβ 42 trimer immunization has significant effects on tau pathology via changes on cellular kinases. ERK1 and ERK2 are both highly expressed in the brain, and it had been shown in vitro that ERK2 is capable of phosphorylating a large number of residues in tau. Activation of the RAS-RAF-MEK-ERK signaling pathway by APP and A B C Fig. 8 Changes in enzymes of the Ras/mitogen-activated protein kinase kinase (MEK)/extracellular signal-regulated kinase (ERK) signaling pathway and glycogen synthase kinase 3β (GSK3β) following DNA amyloid-β 1-42 peptide (Aβ 42 ) immunization. Gray-level intensities (arbitrary units) of the protein bands from the Western blots shown in Fig. 7 were semiquantitatively analyzed using the ImageJ software package (National Institutes of Health). Black bars represent levels found in triple-transgenic Alzheimer's disease (3xTg-AD) control mice (n = 7); blue bars represent levels found in DNA Aβ 42 trimer-immunized mice (n = 7); and gray bars represent levels in wild-type mice (n = 2). a Analyses for MEK. b Analyses for ERK. c Analyses for GSK3β. The first graph in each row shows gray-level intensities for total enzyme; the second graph shows gray-level intensities for the active (phosphorylated) forms of the respective kinases; and the third graph shows the normalized data in which the levels of the phosphorylated kinases were calculated as a percentage of the total enzyme levels for each of the mouse brain lysates used. *, **, and *** indicate p values of ≤ 0.05, ≤ 0.01 and ≤ 0.005, respectively (Mann-Whitney U test) Aβ 42 oligomers in a cell culture system as well as in postmortem human AD brains indicated a pathologic link between Aβ and this particular MAPK pathway [57][58][59]. It has been suggested by others that the two main pathologies of AD, amyloid and tau aggregation, affect the aging brain and cause changes in large-scale neuronal circuits [60]. We show in the present study that DNA Aβ 42 immunization led to significant changes in several pathways. Further analyses of the mechanism of action on tau reduction and changes in cellular signaling pathways in the DNA Aβ 42 trimer-immunized mice are goals for future research.
It had been shown before in the 3xTg-AD mouse model that antitau immunotherapy or passive anti-Aβ immunotherapy led to removal of tau or Aβ or both [19,43,55,[61][62][63][64]. Of note, these studies were of passive immunotherapy using preformed mAbs or the intracranial injection of anti-Aβ or antitau antibodies, which is different from the active immunization done in the present study. It had also been shown before that immunization with a DNA vaccine encoding Aβ 1-11 or a short tau epitope led to the production of antibodies against Aβ or tau, respectively [65,66]. We show for the first time a different mechanism in which active DNA Aβ 42 trimer immunization in the 3xTg-AD mouse model results in reduction of both pathologies with one vaccine: Aβ reduction due to antibodies generated against Aβ and tau reduction due to an indirect mechanism in which less Aβ led to less tau kinase activation and therefore to less tau phosphorylation. Of note, also in AN-1792, a clinical trial using Aβ 42 immunotherapy in patients with AD, a trend toward reduction in cerebrospinal fluid phospho-tau concentrations was reported, and analysis of postmortem brain tissue showed a reduction of aggregated tau in neuronal processes [1][2][3]67].
Current assessments of Aβ immunotherapy for the prevention of AD in several completed and ongoing trials show divergent responses [68]. Positive results in patients treated with the mAb aducanumab support the effectiveness of Aβ immunotherapy in patients with early AD [8]. Aducanumab is a fully human IgG1 antibody that corresponds to the IgG2a antibody isotype in the mouse. Of note, aducanumab has been characterized as an antibody that binds to soluble Aβ 42 oligomers and insoluble Aβ 42 fibrils prepared in vitro, but not Aβ 42 monomers, consistent with the detection of conformational but not linear epitopes. This antibody reactivity might be similar to antibodies generated in response to immunization with DNA Aβ 42 trimer as shown in the present study.

Conclusions
We present data showing for the first time that active immunization with a DNA plasmid coding for an Aβ trimer (3xAβ 1-42 ) designed to induce an anti-Aβ humoral immune response in the 3xTg-AD mouse model significantly reduced both of the main AD pathologies, amyloid and tau. We show a significant reduction in activated protein levels for p44/p44 MAPK (ERK1/2), the upstream MEK, and GSK3β. Our data support significant changes in the Ras-Raf-MEK-ERK signaling pathway in AD mouse model brain due to DNA Aβ 42 immunotherapy, and this is a goal of further studies. DNA Aβ 42 immunization in patients with AD has the potential to modify early and late changes in this disease. It is expected that DNA Aβ 42 trimer immunotherapy in a clinical trial will reduce both plaques and tangles in patients with AD.