|Home | About | Journals | Submit | Contact Us | Français|
Alzheimer’s disease (AD) process is understood to involve the accumulation of amyloid plaques and tau tangles in the brain. However, attempts at targeting the main culprits, neurotoxic Aβ peptides, have thus far proven unsuccessful for improving cognitive function. Recent clinical trials with passively administrated anti-Aβ antibodies failed to slow cognitive decline in mild-moderate AD patients, but suggest that an immunotherapeutic approach could be effective in patients with mild AD. In an AD mouse model (Tg2576) we tested the immunogenicity (cellular and humoral immune responses) and efficacy (AD-like pathology) of clinical grade Lu AF20513 vaccine. Lu AF20513 induces robust “non-self” T cell responses and production of anti-Aβ antibodies that reduce AD-like pathology in the brains of Tg2576 mice without inducing microglial activation and enhancing astrocytosis or CAA. Importantly, a single immunization with Lu AF20513 induces strong humoral immunity in mice with pre-existing memory Th cells. In addition, Lu AF20513 induces strong humoral responses in guinea pigs and monkeys. Collectively, these data suggest translation of Lu AF20513 to clinical setting with aims to (i) induce therapeutically potent anti-Aβ antibody responses in patients with mild AD, particularly if they have memory Th cells generated after immunizations with conventional Tetanus Toxoid vaccine; (ii) exclude likely pathological autoreactive T cell responses.
Neuropathological features of AD include deposition of the amyloid-β (Aβ) fragment of amyloid precursor protein (APP) in senile plaques, accumulation of neurofibrillary tangles (NFT) composed of tau protein, and death of neurons (Hardy and Allsop, 1991; Pike et al., 1991; Price and Sisodia, 1994; Hardy and Selkoe, 2002; Nikolaev et al., 2009). For over two decades, Aβ peptides have been thought to play a central role in the onset and progression of AD (Selkoe, 1991, 1994), and it was this proposal from which the ‘amyloid cascade hypothesis’ emerged (Hardy and Higgins, 1992; Golde et al., 2006; Hardy, 2006). According to this hypothesis, in AD, the normally soluble Aβ molecule undergoes conformational changes and is deposited as insoluble fibrils, and soluble oligomers and protofibrils. Importantly, the amyloid cascade hypothesis has evolved to focus mainly on soluble oligomers and protofibrils of Aβ, which are now considered to be the most toxic forms of Aβ, responsible for causing synaptic destruction (Harper et al., 1997; Walsh et al., 1997; Lambert et al., 1998; Yong et al., 2002; Klein et al., 2004; Cleary et al., 2005; Haass and Selkoe, 2007). Accordingly, therapeutic interventions for AD have been directed toward decreasing Aβ production using β- and γ-secretase inhibitors/modulators, or by immunotherapeutic strategies to enhance Aβ clearance and to block tau aggregation (Sigurdsson et al., 2004; Rafii and Aisen, 2009; Holtzman et al., 2011). The first AN1792 vaccine clinical trial in AD patients was unsuccessful due to a small but statistically significant incidence of meningoencephalitis (Orgogozo et al., 2003). However, vaccine strategies for AD treatment will remain highly promising if new-generation vaccines can avoid anti-Aβ T cell responses that may underlie the incidences of meningoencephalitis (Orgogozo et al., 2003), as well as the issue of T cell tolerance, which might have accounted for the low antibody titers in many patients in the AN1792 trial (Nicoll et al., 2003; Ferrer et al., 2004; Gilman et al., 2005; Boche et al., 2008; Holmes et al., 2008). In this translational study we have devised and validated a novel AD epitope vaccine, Lu AF20513, in which the Th cell epitopes of Aβ42 were replaced by two foreign Th epitopes from Tetanus Toxoid (TT), P2 and P30, and the immunodominant B cell epitope of amyloid, Aβ1–12. Our data reveal that the Lu AF20513 (i) overcomes T cell tolerance induced by self-antigen; (ii) greatly reduces the possibility of inducing harmful autoreactive T cell responses that may explain the failure of the AN1792 vaccine; and (iii) may improve the ability of the elderly to mount an effective immune response by stimulation of pre-existing memory Th cells. In this report on the immunogenicity, efficacy, and safety of GMP grade Lu AF20513 in the APP/Tg mouse model of AD, in guinea pigs and in cynomolgus monkeys, our data support the translation to Phase I/IIa clinical trials.
Female and male, 4–6 mo old Tg2576 mice (H-2bxs haplotype) were bred at the animal facility of the University of California at Irvine. Female, 6–8 weeks old B6SJL mice (H-2bxs haplotype) were obtained from the Jackson Laboratory. All animals were housed in a temperature and light-cycle controlled facility, and their care was under the guidelines of the National Institutes of Health and an approved IACUC protocol at University of California, Irvine.
Female and male guinea pigs (albino, Dunkin Hartley) were obtained from Charles River, Germany and were approximately 8–10 weeks of age at commencement of treatment. All animals were housed in a temperature and light-cycle controlled facility at TNO Triskelion B.V., the Netherlands, and their care was under the guidelines of the European Communities (Directive 86/609/EEC) and Dutch legislation (The experiments on Animals Act, 1997).
Female and male, purpose-bred cynomolgus monkeys (Macaca fascicularis), approximately 3 years of age, were obtained from Biodia, Mauritius. All animals were housed in a temperature and light-cycle controlled facility at Ricerca Biosciences LCC, France, and their care was in compliance with the following guidelines: Guide for the care and use of laboratory animals, NRC, 1996, Decree no 2001-464 regarding the experiments with laboratory animals, described in the Journal Officiel de la République Francaise on 29 May 2001, and Decree no 2001-486 relating to the protection of animals used in scientific experiments, described in the Journal Officiel de la République Francaise on 06 June 2001.
GMP grade recombinant protein vaccine (Lu AF20513) was used in this study, which composed of two foreign T helper cell epitopes from TT, P30 and P2, and 3 copies of the B cell epitopes of Aβ42, Aβ1–12 (Fig. 1A) avoiding known Th cell epitopes in Aβ. Lu AF20513 is expressed in E. coli and primarily found in the inclusion bodies. The inclusion bodies were solubilized and the protein was purified through a series of chromatographic methods and the buffer exchanges to the final formulation.
All Tg2576 mice were from the same breeding cohort and were allocated randomly to experimental and control groups. 4–6 mo old mice were immunized with Lu AF20513 as previously described (Cribbs et al., 2003; Petrushina et al., 2003; Agadjanyan et al., 2005). Briefly, Lu AF20513 protein was formulated in CFA/IFA or Quil-A adjuvants, as described previously (Cribbs et al., 2003), and mice were injected subcutaneously (SC) with indicated concentrations of antigen (Fig. 1B–D). Control mice were injected with an adjuvant only. All mice were boosted at monthly intervals. More detailed experimental protocols are described in Fig. 1B–D. The guinea pigs and cynomolgus monkeys were immunized monthly (SC) with 20μg and 50μg of Lu AF20513 formulated in Alhydrogel® per animal, respectively. Control groups of guinea pigs and cynomolgus monkeys were injected with adjuvant only. Blood was collected before injections (pre-bleed) or 10–11 days after the second injection with Lu AF20513 formulated in Alhydrogel®.
Analysis of T cell proliferation was performed in splenocyte cultures from individual animals using a [3H]-thymidine incorporation assay, as previously described (Cribbs et al., 2003; Agadjanyan et al., 2005; Davtyan et al., 2010). The same splenocytes used to assess T cell proliferation were utilized for detection of T cells producing IFNγ (Th1) or IL-4 (Th2) cytokines by ELISPOT (BD Pharmingen), as previously described (Cribbs et al., 2003; Agadjanyan et al., 2005; Petrushina et al., 2007). The cultures of splenocytes from experimental and control mice were re-stimulated in vitro with P30, P2, Lu AF20513, Aβ40 or an irrelevant peptide (25 μg/ml of each peptide) for 18 hours.
Total anti-Aβ antibodies were detected using ELISA as described previously(Ghochikyan et al., 2006; Petrushina et al., 2007; Davtyan et al., 2010). Anti-Aβ antibody concentrations in mice were calculated using a calibration curve generated with 6E10 monoclonal antibody (Signet). HRP-conjugated anti-IgG1, IgG2ab, IgG2b and IgM specific antibodies (Bethyl Laboratories, Inc.) were used to characterize the isotype profiles of antibodies in pooled sera at dilution 1:200. Anti-Aβ antibodies in guinea pigs and monkeys were detected using 96 wells ELISA plates coated with Aβ1–28 (American Peptide). In case of guinea pigs we used sheep anti-guinea pig IgG-biotin (Nordic Immunology, The Netherlands) coupled to streptavidin-europium (Perkin Elmer) as a secondary antibody, while in case of monkeys we used goat anti-monkey IgG antibodies conjugated with HRP (Nordic Immunology, The Netherlands). The individual samples were diluted 1:100, 1:200, 1:400, 1:800, 1:2400, 1:7200, 1:21600 and the titer was determined by back-calculating the OD of the assay cut point on a 4-parameter logistic curve using the Prism software.
Anti-Aβ1–12 antibodies were purified from sera of mice immunized with Lu AF20513 by an affinity column (SulfoLink, Pierce) as described previously(Mamikonyan et al., 2007). Column was immobilized with Aβ18-C peptide (GenScript). Purified antibodies were analyzed via 10% Bis-Tris gel (Invitrogen), and the concentration was determined using BCA protein assay kit (Pierce).
Binding studies were performed on the BIAcore 3000 SPR platform (Biacore) as described previously (Mamikonyan et al., 2007). Monomeric, oligomeric and fibrillar forms of Aβ42 peptides were prepared as described previously and confirmed by Electron Microscopy (EM) and binding of these peptides to 6E10, A11, and OC antibodies by dot blot (Kayed et al., 2007; Mamikonyan et al., 2007; Kayed et al., 2010). These peptides were immobilized to the surface of biosensor chip CM5 (GE Healthcare) via an amine coupling of primary amino groups of the appropriate peptide to carboxyl groups in the dextran matrix of the chip. Serial dilutions of anti-Aβ1–12 antibody, irrelevant mouse IgG (SouthernBiotech), or 6E10 antibody (666.6, 222.2, 74.1, 24.7, 8.23, 2.74 nM) in the running buffer containing 10 mM HEPES, 150 mM NaCl, 0.05% surfactant P20, pH 7.4, were injected at 20 μl/min over each immobilized form of peptide, and the kinetics of binding/dissociation was measured as change of the SPR signal [in resonance units (RU)]. Each injection was followed by a regeneration step of a 25-s pulse of 1 M NaCl, 50 mM NaOH. Fitting of experimental data was done with BIAevaluation 4.1.1 software using 1:1 interaction model to determine apparent binding constants.
Anti-Aβ1–12 antibodies (1.25μg/ml) were screened for the ability to bind to human Aβ plaques using 50μm brain sections of formalin-fixed cortical tissue from a severe AD case (received from Brain Bank and Tissue Repository, MIND, UC Irvine) using immunohistochemistry as described previously (Ghochikyan et al., 2003; Agadjanyan et al., 2005; Davtyan et al., 2011). A digital camera (Olympus) was used to capture images of the plaques at a 10x magnification.
Cell culture MTT assay was performed as described previously with minor modifications (Davtyan et al., 2011). Aβ42 oligomers and fibrils were incubated alone or with purified anti-Aβ1–12 antibodies or irrelevant antibodies (BD, Biosciences) for 2h at room temperature with occasional mixing to ensure maximal interaction. After incubation, the peptide/immune sera mixtures were diluted into culture media so that the final concentration of peptide and antibodies was 2μM and 0.2μM, respectively. This media was then added (100μl) to SH-SY5Y cells (ATCC). The treatment time was 48h. Untreated controls were run in parallel. Following incubation, neurotoxicity was assayed using the MTT assay according to the manufacturer’s instructions (Promega Corp.). Cell viability was calculated by dividing the absorbance of wells containing samples by the absorbance of wells containing medium alone.
At 15–17 mo of age, vaccinated and control mice were sacrificed for further neuropathological analysis as described previously (Movsesyan et al., 2008b). The left hemisphere was snap frozen and reserved for measurement both soluble and insoluble Aβ levels by ELISA. The right hemisphere was fixed in 4% paraformaldehyde in PBS at +4°C for 24 hrs for further sectioning.
50 μm-thick coronal sections of fixed hemi-brains were cut using vibratome. To assess the extent of neuropathology, several sets of free-floating equally spaced sections have been selected for each mouse brain, and immunostained using the antibodies described below. Aβ cored and defused deposits were detected with 6E10 (1:1000, Signet), as described in (Petrushina et al., 2007; Movsesyan et al., 2008b). Activated microglia were stained with the anti-I-A/I-E (marker of MHC II alloantigens, 1:100, BD Pharmingen), and astrocytes were labeled with anti-GFAP (glial fibrillary acidic protein, 1:500, Dako) antibody, both as we described (Petrushina et al., 2007; Movsesyan et al., 2008b)..
Fibrillar Aβ deposits were visualized using Thioflavin S (ThS; Sigma-Aldrich), and hemosiderin deposits were detected via Prussian blue staining. Both assays were performed as described in (Petrushina et al., 2007; Movsesyan et al., 2008b).
The number of 6E10- or ThS-positive plaques was analyzed through the whole hemisphere, as well as in cortical and hippocampal regions. Measurement of number of blood vessels showing amyloid depositions characteristic of cerebral amyloid angiopathy (CAA) was done in 6E10-stained brain sections; activated microglial and hemosiderin-positive profiles were counted through the whole hemisphere as well. The mean semi-quantitative scores per hemisphere or per neuroanatomical region in those assays were determined based on visual microscopic inspection of sets of 5 coronal sections equally spaced between −0.80 to −2.92 mm with respect to Bregma, performed by three independent observers blinded to the treatment conditions.
NIH Image J 1.45s software was used to analyze the number of GFAP-positive astrocytes. For every animal, total of 19 images (802 x 650 μm each) of cortex, striatum and hippocampus were selected at approximately the same plane (3 sections between −0.82 to −2.75 mm with respect to Bregma). The images were captured using a MD700 video camera (AmScope) and 10x objective.
To determine the levels of both soluble Aβ40 and Aβ42 and insoluble total Aβ, a 10% (w/v) of brain homogenate was prepared from each mouse brain at Amorfix (Canada) using in PBS/2% NP-40 containing proprietary protease inhibitors. Each 10% brain homogenate was centrifuged at 100,000xg for 1 hr at 4°C. The supernatant was collected as the source of soluble aggregates for analysis of soluble Aβ40 and Aβ42 using human β-amyloid ELISA kits (Invitrogen), according to manufacturer’s recommendations. The final values of soluble Aβ were expressed as nanograms (ng) per gram wet weight of hemi-brain. The remaining pellet was solubilized in 70% formic acid by sonication and centrifuged at 100,000xg for 1 hr at 4°C. Avoiding upper lipid layer, the lower aqueous layer was collected and stored at −70°C. Quantification of insoluble Aβ level was done using human β-amyloid ELISA kit (Invitrogen), according to manufacturer’s recommendations. The final values of insoluble Aβ were expressed as micrograms (μg) per gram wet weight of hemi-brain.
6–8 weeks old B6SJL mice were immunized 3 times with P30 peptides formulated in Quil-A (50μg/mouse and n=30). Control mice were injected with adjuvant only (n=25). After resting period for 6 months, 10 mice from P30 primed and control groups were terminated and memory T cell response was analyzed before and after a single booster injection using ELISPOT assay. Recall responses of CD4+ T cells were detected after isolation of this subpopulation of cells from splenocytes using specific isolation kit and as suggested by the manufacturer (Militenyi Biotec). Isolated CD4+ T cells from splenocytes of primed (n=4) and non-immune mice (n=4) were used for enrichment of naive splenocytes (5x104 CD4+ cells were added into 15x104 splenocytes) in indicated experiments. The purity of isolated CD4+ cells and the efficacy of depletion were analyzed by flow cytometry (Militenyi Biotec). For booster injection we used 100μg/mouse Lu AF20513 and both cellular and humoral immune responses were analyzed in experimental and control mice as indicated in Figure S6.
All statistical parameters [mean, standard deviation (SD), significant difference, etc.] used in experiments were calculated using Prism 6 software (GraphPad Software, Inc.). Statistically significant differences were examined using an analysis of variance (ANOVA) and post hoc comparisons were done using Tukey’s test (P value less than 0.05 was considered as statistically different).
One challenge associated with the clinical use of the Aβ “self” T cell epitopes as part of a vaccine to treat AD patients are the development of undesirable anti-Aβ or anti-APP specific T cell responses (Cribbs et al., 2003; Orgogozo et al., 2003; Ferrer et al., 2004; Agadjanyan et al., 2005; Lemere, 2009). With this in mind, we synthesized Lu AF20513 vaccine, which was designed to activate non-self CD4+ T lymphocytes (Th) specific to TT, pre-existent in most individuals due to prior tetanus immunization, in order to enhance antibody production by Aβ-specific B cells (Fig. 1A).
We first tested the cellular responses to Lu AF20513 vaccination in Tg2576 mice (Fig. 2A,B), an APP over-expressing model of AD (Hsiao et al., 1996). Control groups of mice were injected with adjuvant only (Fig. 1B). Vaccinations of Tg2576 mice with Lu AF20513 led to production of IFNγ and IL-4 after re-stimulation with P30, but not Aβ40 (Fig. 2A). In proliferation assays, Th cells were activated with Lu AF20513 and P30, but not Aβ40 peptide (Fig. 2B). T cell responses were not detected in mice injected with adjuvant only (Fig. 2C,D). Of note, here we tested Th cell responses to only one TT epitope, P30, based on our previous experiments performed in wild-type B6SJL mice, which demonstrated that two immunizations with Lu AF20513 induced equally effective Th cell responses specific to both P2 and P30 epitopes of TT and Lu AF20513 (Fig. 2E,F). It should also be mentioned that re-stimulation of immune splenocytes with Lu AF20513 induced significantly stronger Th cell proliferation (P≤0.001) than when stimulated by P2 and P30 peptides individually (Fig. 2E,F). Thus, splenocytes from Lu AF20513-immunized wild-type and Tg2576 micedid not induce autoreactive anti-Aβspecific Th cell responses (Fig. 2).
To demonstrate that activation of P30 and P2 specific Th cells culminates in activation of anti-Aβ specific B cells, we measured the anti-Aβ42 antibody concentrations in Tg2576 mice vaccinated with Lu AF20513 formulated in two different adjuvants, CFA/IFA or Quil-A (Fig. 1C). Of note these animals have been in very early stage of AD-like pathology at the start of immunization (Lesne et al., 2006). Control mice injected with CFA/IFA and Quil-A adjuvants did not generate anti-Aβ42 antibodies, while Lu AF20513 formulated in either adjuvant induced robust antibody production after 2–3 immunizations. Of note, no differences were seen in the level and kinetics of antibody production in mice immunized with Lu AF20513 formulated either in CFA/IFA or in Quil-A (Fig. 3A). To characterize the type of humoral immune response, we measured the production of IgG1, IgG2ab, IgG2b, and IgM isotypes of anti-Aβ antibodies in Tg2576 mice immunized with Lu AF20513 formulated in CFA/IFA or Quil-A. We did not observe any differences in the magnitude and the type of antibody responses: mice from both vaccinated groups that received either CFA/IFA or Quil-A generated mostly IgG antibodies (IgG2b>IgG2ab>IgG1) (Fig. 3B). Since CFA/IFA adjuvant is used only for animal studies, while the equivalent of Quil-A for human use, QS21, is already used in many clinical trials, we combined mice from groups immunized with Lu AF20513/Quil-A and Lu AF20513/CFA/IFA as indicated in Fig. 1D and performed subsequent immunizations with Lu AF20513 formulated in Quil-A only. The data demonstrated that the anti-Aβ antibody concentration in vaccinated mice reached the maximum titers after 3 immunizations (Fig. 3A), declined slightly after that and remained at steady state during subsequent immunizations until the end of the experiment (Fig. 3C). The levels of IgG1, IgG2ab, IgG2b immune responses were robust and stable, while the level of IgM was low during the whole vaccination process (Fig. 3B,D). In sum, these results demonstrate that Th cells specific to P30 and P2 epitopes from TT are providing the support necessary for a therapeutically potent anti-Aβ antibody response in Tg2576 mice vaccinated with the Lu AF20513.
In order to functionally characterize the anti-Aβ1–12 antibodies generated in response to the Lu AF20513, we measured the binding of these antibodies to various forms of β-amyloid including plaques in the brains from a clinical AD case as well as their neuroprotective ability (Fig. 4).
One of the important features of anti-Aβ antibodies is their capacity to bind the toxic forms of β-amyloid (Glabe, 2008). Our results from an SPR assay showed that polyclonal anti-Aβ antibodies bound with high average affinity to immobilized monomers (KD=2.83x10−8M), oligomers (KD=9.9x10−9M) and fibrils (KD=2.8x10−8M) (Fig. 4A,B), while irrelevant mouse IgG antibody purified from normal mouse serum bound to all these forms of Aβ42 on the base-line level. The binding affinity of purified antibodies to oligomers was higher than binding affinity to fibrils and monomers. Notably, the affinity of binding for control monoclonal antibody, 6E10, is higher for monomeric form of Aβ42 (KD=1.21x10−10M) compared with that of oligomeric (KD=9.37x10−9M) and fibrillar (KD=2.97x10−9M) forms (Fig. 4B). Nevertheless, the affinity of binding to oligomeric forms is similar for anti-Aβ1–12 and 6E10 antibodies (KD=9.9x10−9M vs KD=9.37x10−9M). Additionally, the anti-Aβ antibody generated in mice immunized with Lu AF20513 also bound to Aβ plaques in the brain tissues from an AD case, whereas irrelevant mouse IgG did not bind to AD plaques (Fig. 4C).
Next, we tested the protective effect of anti-Aβ antibodies against neurotoxicity induced by Aβ42 oligomers and fibrils in an in vitro model. Culture of SH-SY5Y neuroblastoma cells with Aβ42 oligomers and fibrils had cytotoxic effects, reducing cell viability to 50% and 63.1%, respectively, whereas pre-incubation with anti-Aβ antibodies rescued the cell viability to 79.5% and 91.6% (Fig. 4D). Binding to amyloid plaques in the brain sections along with the neuroprotective effect of anti-Aβ1–12 antibodies suggest that the Lu AF20513 vaccination strategy is functionally active.
As a pre-clinical test of the therapeutic efficacy of Lu AF20513 vaccination, we studied neurological changes in the brains of aged Tg2576 mice (15–17 mo old) vaccinated with Lu AF20513 (Fig. 1D). Fig. 5A shows a significant decrease in hemi-brain plaque burden in mice immunized with Lu AF20513 as compared to the control adjuvant-only injected group. We detected significantly less 6E10-positive diffuse and cored plaques in both cortical and hippocampal regions of brains of Lu AF20513 vaccinated mice as compared to the same brain regions of control animals (Fig. 5B). Additionally, the results showed a significant reduction of cored Thioflavin S (ThS)-positive Aβ plaques in the hemi-brains as well as in cortical and hippocampal regions of these brains obtained from vaccinated mice versus brains isolated from control Tg2576 mice (Fig. 5C,D). These data were confirmed by detection of insoluble Aβ in the brains of vaccinated and control mice. As demonstrated in Fig. 5E, vaccination with Lu AF20513 induced a significant reduction (P<0.001) of insoluble total Aβ in the brains of immunized mice vs control animals.
Increasing evidence suggests that soluble (or diffusible) Aβ oligomers mediate different toxic pathways in AD such as tau hyperphosphorylation (De Felice et al., 2008), impairment of memory, and neuronal death (Lesne et al., 2006; Haass and Selkoe, 2007; Shankar et al., 2008). A potential problem of immunotherapy is that a reduction of insoluble Aβ may lead to increased levels of soluble forms of this peptide (Patton et al., 2006). Importantly, we observed that vaccination significantly reduced (P<0.001) the levels of potentially toxic forms of amyloid, soluble Aβ42 and Aβ40 peptides in the brains of Lu AF20513 immunized Tg2576 mice compared to control (adjuvant injected) animals (Fig. 5F). Hence, Lu AF20513 vaccination reduces the levels of pathologic forms of Aβ in the brains of immunized mice.
To detect inflammation-related pathology in the brains of animals immunized with the Lu AF20513 vaccine, microglial (MHC class II) and astrocyte (glial fibrillary acidic protein, GFAP) activation were examined. Semi-quantitative image analysis of MHC class II-positive cells (I-Ab/I-Eb) demonstrated significantly fewer activated (MHC class II+) microglial cells in Tg2576 mice vaccinated with the Lu AF20513 (Fig. 6A, including representative MHC class II staining). Astrocyte-specific marker in the brains of Lu AF20513 immunized mice revealed a significant reduction in GFAP+ cells via quantitative image analysis indicative of the presence of fewer astrocytes in comparison to the control group (Fig. 6B, including representative GFAP staining).
One possible side effect of the disruption of amyloid plaques could be the increase of cerebral amyloid angiopathy (CAA), as was observed in pre-clinical trials (Wilcock et al., 2007) and the AN1792 trial (Nicoll et al., 2003; Ferrer et al., 2004; Masliah et al., 2005). To assess the effect of vaccination with Lu AF20513 on vascular deposition of Aβ, we calculated 6E10-positive vessels in the brain sections of immunized and control Tg2576 mice. No differences in the number of amyloid-containing blood vessels were observed in brains of vaccinated and control mice (Fig. 6C).
However, the ratio of amyloid-containing blood vessels to parenchymal plaques was increased in vaccinated mice vs. control animals (5.07±5.6 vs. 0.24±0.15, respectively) suggesting that amyloid or amyloid-antibody immune complexes may be redistributed to the vessels for clearance from the brain.
Next, we analyzed microhemorrhages in brains of Lu AF20513 immunized and control mice. Previously, it was reported that active immunizations with Aβ42 formulated in strong adjuvants, CFA/IFA (Wilcock et al., 2007) may induce cerebral microhemorrhages in ~20 months old APP/Tg mice. However, microhemorrhages have not been reported in 15–17 months old mice. Characteristic blue hemosiderin-positive profiles were not increased in Lu AF20513 immunized mice compared with control animals (Fig. 6D).
Thus, immunization with Lu AF20513 did not exacerbate cerebral amyloid angiopathy or increase microhemorrhages, and had the positive impact of lowering microglial activation and astrocytosis.
Encouraged by the results generated in APP/Tg mice, we tested the Lu AF20513 vaccine in guinea pigs and cynomolgus monkeys after formulation in the widely used mild adjuvant Alhydrogel® to strengthen the link to a clinical situation. Fifteen out of sixteen guinea pigs (94%) responded to two immunizations with Lu AF20513 by producing anti-Aβ antibodies with the average titer equal to 1:7954 ± 7676 (Table 1). Of note, data from the first clinical trial with the AN1792 vaccine demonstrated that anti-Aβ antibody titers ≥1:2200 showed some slowing of cognitive decline and localized reduction of plaques (Hock et al., 2003; Gilman et al., 2005; Masliah et al., 2005; Nicoll et al., 2006). Lu AF20513 vaccine generated therapeutically relevant titers of anti-Aβ antibody (>1:2200) in 80% of responding guinea pigs (Table 1). More importantly, all 14 monkeys immunized twice with Lu AF20513 formulated in Alhydrogel® responded to the vaccine (average titer equal to 1:6491 ± 5338) and 71.5% of these animals generated therapeutically relevant titers of anti-Aβ IgG antibody (Table 2). Of note, in pre-bleed sera (Table 1 and and2)2) we detected only a background level of anti-Aβ antibodies.
One important aspect of the design of the Lu AF20513 is that immunization with this vaccine should quickly induce a robust anti-Aβ antibody production in elderly individuals with pre-existing TT-reactive memory Th cells. To test this hypothesis, we immunized two groups of B6SJL mice with either P30 peptide formulated in Quil-A, or Quil-A, only, and first analyzed the Th memory cells specific to P30 after 6 mo of resting period (Fig. 7A). After in vitro re-stimulation of immune, but not control splenocytes with P30, we detected high numbers of cells producing IFNγ (Fig. 7B). Importantly, depletion of almost 95% of CD4+ cells from primed splenocytes completely abrogated detection of IFNγ-producing cells (Fig. 7C). Moreover, after enrichment of splenocytes isolated from naïve (non-immunized) mice with CD4+ T cells isolated from P30 primed splenocytes (~75% non-immune splenocytes mixed with ~25% of P30 primed CD4+ T cells), we restored immune responses to P30 (Fig. 7D). Hence, these data reveal that after a 6 month resting period, mice have functional memory CD4+ Th cells specific to the P30 epitope of TT vaccine.
Next we tested whether a single booster injection of Lu AF20513 formulated in Quil-A adjuvant could activate pre-existing anti-P30 specific memory Th cells and lead to a quicker and stronger anti-TT cellular responses and anti-Aβ antibody responses. Thus, after a 6 month resting period, we boosted the P30-primed mice with Lu AF20513 formulated in Quil-A and analyzed both cellular and humoral immune responses (Fig. 8). Boosting of experimental mice with epitope vaccine induced strong Th cell responses specific to P30: very large number of cells producing IFNγ was detected in this group of mice with pre-existing memory Th cells vs control mice (Fig. 8A). Most importantly, the single injection with Lu AF20513 formulated in the strong Th1 adjuvant Quil-A led to induction of robust anti-Aβ antibody responses only in mice with pre-existing memory Th cells: concentrations of anti-Aβ antibodies were significantly higher (P≤0.001) than that in control mice (Fig. 8B). Thus, a single immunization with Lu AF20513 strongly activated pre-existing memory CD4+ T cells specific to the Th epitopes of this vaccine and rapidly led to the robust production of antibodies specific to the B cell epitope (Aβ1–12) of the same vaccine.
According to the Amyloid Cascade Hypothesis (Hardy and Higgins, 1992) the accumulation of Aβ peptide is a primary pathological event in the development of AD, that precedes tau accumulation and leads to neurodegeneration and dementia. The identification of the Aβ peptide as a target for therapeutic interventions for AD has led to many preclinical and clinical studies. Despite the fact that the recent Pfizer’s and Eli-Lilly’s clinical data with passively administrated anti-Aβ antibodies have been disappointing, there is a consensus in the field that the removal or lowering of Aβ in patients with very early AD pathology or even in pre-symptomatic subjects could be an effective measure. Obviously, for such preventive treatments passive vaccination is not practical while a safe active vaccine might be beneficial. Although the first clinical trial with AN1792 was unsuccessful due to meningoencephalitis (possibly caused by the activation of autoreactive T cells) (Nicoll et al., 2003; Ferrer et al., 2004; Masliah et al., 2005), currently various epitope-based vaccines, similar to one that we suggested earlier (Agadjanyan et al., 2005; Petrushina et al., 2007; Movsesyan et al., 2008b), are in clinical trials (for examples UBITh®, V950, CAD106, ACC-001) (Wang et al., 2007; Lemere and Masliah, 2010; Wiessner et al., 2011; Winblad et al., 2012; http://clinicaltrials.gov). Designs of these vaccines aim to circumvent the problem of T cell autoreactivity and overcome tolerance induction to self-antigen (Nicoll et al., 2003; Ferrer et al., 2004; Masliah et al., 2005). Currently, the published data are available only on clinical trial with CAD106 composed of bacteriophage Qβ expressing Aβ1–6 B cell epitope. Although, the authors reported that anti-Aβ antibodies were detected in 62% of low dose and 82% of high dose subjects, the exact titers of antibodies are not presented, so actual magnitude of the humoral immune responses is not clear (Winblad et al., 2012). Importantly, no data have been presented in this report on antigen-specific cellular immune responses generated after vaccinations of AD patients. Likewise, humoral and cellular immune responses to CAD106 have not been properly reported in pre-clinical mouse and monkey studies (Wiessner et al., 2011).
In this study, we tested a novel clinical grade epitope vaccine, Lu AF20513 composed of the same immunodominant B cell epitope of Aβ42, (Aβ1–12) fused with two Th epitopes from a conventional TT vaccine that are recognized by different human MHC-class II molecules (James et al., 2007). Importantly, an active immunization approach has been tested in Tg2576 mice with very early-stage AD-like pathology. In our previous studies we tested several protein, peptide, and DNA-based AD epitope vaccines containing a synthetic Th cell epitope PADRE and showed that they are immunogenic in mice and do not induce T cell responses against self-Aβ molecules (Mamikonyan et al., 2007; Petrushina et al., 2007; Movsesyan et al., 2008b; Movsesyan et al., 2008a; Ghochikyan, 2009; Davtyan et al., 2010; Davtyan et al., 2012). In the design of the Lu AF20513 vaccine, we replaced PADRE with P2 and P30 epitopes (Fig. 1A) with the goal of activating non-self pre-existing memory Th cells in the general human population, which is immune to TT due to the public health vaccination program. Data from this study demonstrated that Lu AF20513 induced therapeutically potent Aβ-specific humoral immune responses in mice (Fig. 3), and strong anti-P2 and anti-P30 Th cell responses without the activation of autoreactive Th cells (Fig. 2). These anti-Aβ antibodies bound to oligomeric and fibrillar forms of Aβ (Fig. 4A,B) and pathological plaques in the brain sections from an AD patient (Fig. 4C), while also protecting neuronal cells from Aβ42 oligomer- and fibril-mediated toxicity (Fig. 4D). The most striking is our observation that Lu AF20513 vaccination offers protection from the development of cored and diffuse plaques in the brains of Tg2576 mice (Fig. 5A–D).
One potential problem with immunotherapy is the fact that a reduction of insoluble Aβ plaques may lead to increased levels of soluble forms of Aβ, which are toxic to neurons. Two major mechanisms for antibody-mediated clearance of Aβ have been suggested: sequestration of Aβ from the CNS into the periphery (‘peripheral sink’) (DeMattos et al., 2001) and entry of anti-Aβ antibodies into the CNS and clearance of antigen-antibody complexes by microglial cells (Bard et al., 2000). Regardless of the mechanism of action, we observed that anti-Aβ antibodies generated by Lu AF20513 vaccinations inhibited the accumulation of not only insoluble, but also of soluble forms of Aβ42 and Aβ40 peptides in the brains of experimental vs control Tg2576 mice (Fig. 5E,F). Importantly, despite the significant reduction in plaque burden and soluble Aβ42, no differences in CAA was observed in brains of vaccinated mice vs control animals (Fig. 6C). Increased ratio of amyloid-containing blood vessels to the numbers of parenchymal plaques in vaccinated mice suggests that anti-Aβ antibodies were not as effective at inhibiting cerebral vascular deposition of Aβ as they were at blocking amyloid plaque formation. In addition, we demonstrated that the inhibition of Aβ42 and Aβ40 depositions in the brains of Lu AF20513 vaccinated Tg2576 mice is associated with a reduction in the frequencies of activated microglia and astrocytes (Fig. 6A,B).
Because our data on glial activation was collected at the end of 11 months of active immunotherapy we cannot exclude the possibility that glial cells were activated acutely following active immunization protocol due to Fc-mediated microglial activation in response to antibody-Aβ immune complex formation in the CNS as it was observed in old (19–20 mo old) APP Tg mice, with substantial pre-existing parenchymal as well as CAA, in response to immunotherapy (Wilcock et al., 2001). However, we should mention that because immunization was initiated prior to the onset of amyloid deposition (Tg2576 mice were 4–6 mo old) we believe that any acute inflammatory response would be rather mild under these experimental conditions. Moreover, we believe that early immunotherapy intervention may be critical to reduce the incidence of adverse cerebrovascular events, such as microhemorrhages and vasogenic edema.
Collectively, data generated in Tg2576 mice vaccinated with Lu AF20513 suggest that this epitope vaccine may be safe for translation to human clinical studies. Of note, in this protective study the behavioral changes in the cohort of vaccinated and control Tg2576 mice were not analyzed since there are many reports demonstrating that immunization with Aβ42 generates antibodies specific to N-terminus of Aβ (Janus et al., 2000; Lemere et al., 2000; Morgan et al., 2000) that improve cognitive functions in various APP mice including Tg2576, especially if vaccination is started prior to onset of amyloid-like pathology. In addition, results from previous active and passive vaccination clinical trials suggest that early intervention with immunotherapeutic approaches are likely to be most clinically effective (Lemere and Masliah, 2010; Delrieu et al., 2012; Eli Lilly and Company Announcement, 2012; Johnson & Johnson Announcement, 2012).
We have also included analysis of anti-Aβ antibody titers in guinea pigs and monkeys, which demonstrated the immunogenic potential of the Lu AF20513 vaccine in animals that are naturally immune tolerant to Aβ42 and exhibit diverse genetic backgrounds for MHC class II molecules as compared to mice. Two immunizations of guinea pigs where sufficient for the generation of anti-Aβ IgG antibody responses in 15 out of 16 immunized animals, and 80% of responders generated anti-Aβ antibody titers over 1:2,200 (Table 1). These represent therapeutically relevant titers, as was previously reported in the AN1792 trials (Gilman et al., 2005). Perhaps most importantly, Lu AF20513 vaccinations induced robust humoral immune responses in 100% of cynomolgus monkeys and in 71.5% of monkeys the anti-Aβ antibody titers were therapeutically relevant (Table 2). These findings may translate into adequate clinical responses to vaccination in the elderly, where the generation of robust immunity is more difficult due to immunological senescence.
It is important to note that while several different epitope vaccines are currently in clinical trials (Wang et al., 2007; Lemere and Masliah, 2010; Wiessner et al., 2011; Ghochikyan and Agadjanyan, 2012) and (http://clinicaltrials.gov), the immune responses generated by these vaccines have not been fully characterized: to our knowledge, there are no published data on the exact specificity of Th cells that could be activated by these vaccines either in animal models of AD or in humans. Since the antigen-specific T cell responses were never defined, our challenge was to design an active vaccine against AD in humans, with polymorphic MHC genes, that overcomes the safety and efficacy barriers of previous approaches. Here, we revealed the immunological mechanism of action of Lu AF20513 by demonstrating that this vaccine does indeed activate Th cells specific to TT, leading to enhanced anti-Aβ antibody production. The most remarkable finding in this study is that a single injection with Lu AF20513 activates pre-existing memory CD4+ T cells specific to foreign Th epitopes and quickly induces strong anti-Aβ antibody responses (Fig. 8). Therefore, this Lu AF20513 may represent an effective and safe form of active immunotherapy that may overcome the limited ability of the elderly to respond to vaccinations by activating pre-existing anti-P30/P2 memory Th cells (Fig 9) in the general human population vaccinated with a conventional TT vaccine.
This work was supported by funding from NIH (NS-50895, NS-065518, AG-20241 and NS 057395) and Alzheimer’s Association (IIRG-0728314). H.D. and N.M. were supported by NIA T32 training grant (AG000096). Additional support for AD case tissues was provided by University of California, Irvine Alzheimer’s Disease Research Center Grant P50 AG16573. We thank Jasja Wolthoorn (TNO Triskelion), Robert Pels Rijcken (TNO Triskelion), and Hervé Giorgi (Ricerca Biosciences SAS) for their services in conducting the guinea pig and monkey studies.
Conflict of Interest: A.K. Larsen, P.J. Madsen, D.K. Ditlevsen are employees of H. Lundbeck A/S; K.M. Wegener and L.O. Pedersen are employees and shareholders of H. Lundbeck A/S.
Author ContributionsH.D., A.G., A.K.L., D.H.C., L.O.P. and M.G.A. designed research; H.D., A.G., I.P., A.H., A.D., A.P., N.M., A.K., S.R., P.J.M., K.M.W. and D.K.D. performed research; H.D., A.G., I.P., A.M.M, A.K., A.K.L., K.M.W., D.K.D, D.H.C., L.O.P. and M.G.A analyzed data; H.D., A.G., I.P., A.M.M., L.O.P and M.G.A. wrote the paper.