Rationale for Peptide and DNA Based Epitope Vaccines for Alzheimer’s Disease Immunotherapy

Abstract

Amyloid-beta (Aβ) immunotherapy has received considerable attention as a promising approach for reducing the level of Aβ in the CNS of Alzheimer’s disease patients. However, the first Phase II clinical trial, for the immune therapy AN1792, was halted when a subset of those immunized with Aβ₄₂ developed adverse events in the central nervous system. In addition, data from the trial indicated that there was a low percentage of responders and generally low to moderate titers in the patients mat received the vaccine. Generated antibodies reduced β-amyloid deposits in the parenchyma of patients’ brains, but no reduction in soluble Aβ or significant improvements in cognitive function of patients were observed. These data and data from pre-clinical studies suggest that reduction in the most toxic oligomeric forms of Aβ is important for prevention or slowing down of the progression of cognitive decline, and that vaccination should be started prior to irreversible accumulation of the oligomeric Aβ, at the early stages of AD. Protective immunotherapy requires a development of safe and effective strategy for Aβ immunotherapy. In this review, the rationale for developing epitope vaccines for the treatment of AD will be discussed. We believe that an epitope vaccine will induce an adequate anti-Aβ antibody response in the absence of potentially adverse self T cell-mediated events, making it possible to start immunization at the early stages of AD.

OVERVIEW OF ALZHEIMER’S DISEASE

Alzheimer’s disease (AD) is one of the most devastating diseases of the century, and the risk of this disease increases drastically along with advancing age. AD is characterized clinically by an insidious onset and progressive cognitive decline that impacts memory, language, judgment and orientation in time and space. According to a study led by researchers at the Johns Hopkins Bloomberg School of Public Health, more than 26 million people worldwide were estimated to be living with Alzheimer’s disease in 2006 [1]. The researchers concluded that the global prevalence of Alzheimer’s disease will grow to more than 106 million by 2050. By that time, 43 percent of those with Alzheimer’s disease will need high-level care, equivalent to that of a nursing home. Even modest improvement in prevention rates or in a delay of the onset and progression of the disease will have enormous economical impact on public health care system.

The neuropathological features of the disease include neurofibrillary tangles (NFT), deposition of Aβ in senile plaques, and neuronal loss in affected brain regions [2]. These pathological changes result in a profound loss of synapses over the course of the disease, thereby contributing to a progressive reduction in the functional capacity of the patient. The origins of these neuropathological changes are largely unknown and likely heterogeneous. It is known that approximately 1% of the AD cases are associated with mutations in amyloid-β precursor protein (APP), presenilin-1 and presenilin-2[3]. However, 99% of AD cases are not associated with known mutations and the nature of the etiological agent(s) is not yet understood [4]. Nevertheless, independent of these causative factors, the consequences are very common and lead to the same neuropathological changes.

A critical aim in developing therapeutic interventions for AD is the identification of suitable targets. According to the Amyloid Cascade Hypothesis adopted by the majority of AD researchers, an over-production of Aβ is the primary pathological event, which precedes tau accumulation and leads to neurofibrillary degeneration and dementia [5,6]. The Aβ peptide is cleaved from the integral membrane amyloid precursor protein (APP) by β-secretase to yield a secreted fragment of APPβ and a C-terminal fragment (CTF) of APP (CTFβ). CTFβ is then cleaved by γ-secretase within the membrane to yield a smaller C-terminal fragment (CTFγ) and Aβ fragments of various lengths [7–9]. Over-production of these peptides, as well as the failure of their degradation by enzymes such as neprylisin and insulin degrading enzyme, lead to their oligomerization and aggregation over time to produce senile plaques. Therefore, therapeutic interventions have been aimed at preventing Aβ over-production or accelerating its degradation. Scientists have long considered Aβ peptides to be byproducts of APP cleavage and investigated mostly the toxic features of these peptides. Recent studies clearly demonstrated that this proteolytic pathway is a physiological process and Aβ becomes toxic when its levels are excessive. It has been shown that neuronal activity increases the formation of Aβ and that increased Aβ leads to depression of excitatory synaptic transmission [10]. These two findings have led to a hypothesis that Aβ may normally serve as a negative feedback signal that maintains neuronal activity within a normal dynamic range: too much neuronal activity leads to formation of more Aβ, which depresses excitatory synapses and reduces neuronal activity.

Currently it is known that genetically-derived familial AD is a consequence of over-production of Aβ. In contrast, sporadic AD may be a result of age-related impairment of Aβ elimination from the brain. Several pathways for elimination of Aβ have been identified: microglia-mediated clearance [11, 12], degradation by neprilisin and insulin degrading enzyme [13,14], efflux from the brain by active transport across the blood brain barrier (BBB) as a complex with low density lipoprotein receptor-related protein 1 (LRP)[15], or drainage with interstitial fluid along capillary and arterial blood vessel walls [16,17]. When age-related accumulation and aggregation of Aβ occur due to failure of these pathways, the immune system recognizes these aggregated proteins as “foreign” and attempts to remove them. Studies have demonstrated the presence of anti-Aβ antibodies in the blood and CSF of healthy humans and AD patients. However the titers of these antibodies are lower in AD patients compared with healthy individuals, suggesting that the immune system may be capable of protecting against AD [18–20].

In summary, when developing therapeutic drugs aimed to decrease Aβ production, one should have in mind that Aβ has a role to play in the normal function of the nervous system. The presence of anti-Aβ antibodies in healthy, elderly people supports the concept that immunotherapy may be a way to decrease the level of overproduced Aβ peptide and prevent its oligomerization and aggregation without complete inhibition of its constitutive production and physiological function.

Aβ IMMUNOTHERAPY

Pre-Clinical Studies

In 1999, Schenk and colleagues reported that a vaccine that contained fibrillar Aβ₄₂ when injected into APP-transgenic mice prevented the deposition of Aβ, as well as the development of dystrophic neurites and astrogliosis in their brains. Other researchers have reproduced and extended the original findings to include studies showing that active Aβ immunization with different peptide immunogens and routes of administration can generate anti-Aβ antibodies capable of lowering the cerebral Aβ burden and blocking the loss of cognitive abilities in APP/Tg mice as they accumulate Aβ with aging [21,22]. Studies of peripheral administration of anti-Aβ antibodies showed the presence of these antibodies in the brain [22,23] and the reduction of cerebral Aβ load if given before robust plaque deposition. Passive immunization improves behavior in APP/Tg mice, even in aged animals having cerebral Aβ loads unchanged via immunization [24–27]. These results clearly demonstrated that anti-Aβ-specific that anti-Aβ-specific antibodies are sufficient for clearing amyloid deposits from the brains of APP/Tg mice. Importantly, the reductions in neuropathology were associated with an improvement of behavioral functions in the immunized animals. At the time when Elan and Wyeth, encouraged by these data, started clinical trials for AN-1792, the only adverse responses observed in mouse AD models following Aβ immunotherapy were the three to four-fold increase of cerebral amyloid angiopathy (CAA) severity and the occurrence of microhemorrhages within the brain vasculature of very aged APP/Tg mice injected weekly with high doses of anti-Aβ monoclonal antibodies [28]. The sites of microhemorrhage have been co-localized with cerebral vascular Aβ deposits. Similar results were obtained in different strains of APP Tg mice [27,29,30]. Only later was it reported that active immunizations of APP/Tg mice with fibrillar Aβ₄₂ formulated in a strong conventional adjuvant (CFA/IFA) could also increase microhemorrhages in the brains of vaccinated animals [31].

Outcomes from the AN-1792 Clinical Trials

In the Phase I clinical trial initiated by Elan Pharmaceuticals in 2000, patients with mild and moderate AD were injected with fibrillar Aβ₄₂ formulated in strong Th1-type adjuvant QS21. 53% of patients generated positive antibody response without any obvious adverse events. A Phase II trial started in 2001, which was unexpectedly terminated before the subjects received all of the planned vaccinations when 6% of the subjects developed meningoencephalitis.

Positive outcome

The most important positive outcome of this trial is the evidence that anti-Aβ antibodies are capable of changing AD pathology. Post-mortem analysis of the brains of eight immunized patients has shown lower Aβ load compared with unimmunized controls. There was no evidence of a relationship between the degree of Aβ plaque removal and AN-1792 dose, whereas the mean antibody response during the treatment period was inversely correlated with post-mortem Aβ load [32]. In the subset of antibody responders, CSF tau was reduced compared with baseline level and the change was greater than in the placebo group. There was an interesting observation that in regions where plaques have been removed, tau-containing plaque-associated dystrophic neuritis was absent. Nevertheless, no effect was seen on CSF levels of Aβ₄₂. Importantly, a correlational trend in regression of the neuropsychological test battery (NTB) composite Z-scores and the geometric mean antibody titers was observed. Although this correlation was not significant, the direction of the trend favored higher IgG antibody titers [33]. Follow-up studies of a Zurich cohort of 30 patients with mild and moderate AD showed significantly slower rates of decline of cognitive functions in responders compared to patients without anti-Aβ plaque recognizing antibodies. Importantly, there was a dose-response relationship between the increase in serum antibodies against Aβ plaques and the clinical outcome [34]. Overall, even low/moderate titers of anti-Aβ antibodies generated in a small subset of immunized patients (19.7%) were capable of reducing parenchymal amyloid pathology [35–41] and slowing the progressive cognitive decline associated with the disease [33,34].

Negative outcome

Although the results from the Phase I trial showed good tolerability, in the phase IIa portion of the AN-1792 immunotherapy a subset of individuals developed adverse events in the central nervous system [34–38,41–43]. Importantly, only vaccinated participants (n=18) developed meningoencephalitis, whereas none of the control patients (n=72) injected with placebo developed adverse events [42]. Postmortem examination of the brains from two vaccinated AD patients with neuroinflammation [35,36] showed an infiltration of T cells in the leptomeninges, densest in the areas with amyloid angiopathy. There was also sparse T cell involvement in the cerebral cortex, perivascular spaces and within the parenchyma. In the first case report, infiltrated T cells (CD3⁺, CD45RO⁺) were primarily CD4⁺, although a few CD8⁺ cells were also detected. B cells (CD79a⁺ and CD20⁺) were not present in the brain. In addition, diffuse abnormalities in the cerebral white matter were observed, with a marked reduction in the density of myelinated fibers and extensive macrophage infiltration [35]. The second case report generally agreed with the first one, although there were some distinct differences, such as the presence of mostly CD8⁺ instead of CD4⁺ mature T cells, multi-nucleated giant cells filled with dense deposits of Aβ, and multiple cortical hemorrhages [36]. Another case report on an AD patient without clinical presentation of encephalitis also demonstrated some lymphocytic infiltration in the leptomeninges [37]. These results indicate that anti-Aβ-specific T cells may induce significant side effects in AD patients vaccinated with full-length Aβ_42. The immunized cases had severe CAA compared with unimmunized cases. Although the pathophysiology of these side effects is not completely clear, the main feature was T cell infiltration associated with the meninges being affected by CAA.

Importantly, the same case reports indicated that anti-Aβ antibodies were not responsible for the observed adverse effects after active vaccination, and that the AN1792 vaccine was beneficial for AD patients because it significantly reduced amyloid β-plaque load. However a recent assessment of the relationship between Aβ₄₂ immune response, degree of plaque removal and long-term clinical outcomes demonstrated that clearance of amyloid plaques in the brains of AD patients is not accompanied by the prevention of progressive neurodegeneration [32]. These data suggest that plaque removal is not enough to stop the progression of neurodegeneration and to improve cognitive function. It is possible that amyloid plaques may represent a defense mechanism that removes potentially neurotoxic soluble/oligomeric forms of Aβ peptide. If this is true, the disruption of amyloid plaques without the efficient removal of soluble Aβ, as was observed in the AN1792 trial [32,38] could increase adverse events such as CAA and microhemorrhages without improvement in cognitive functions.

Conclusions from the AN-1792 clinical trials

The results of the first vaccination of elderly AD patients with the AN-1792 vaccine confirmed that anti-Aβ antibodies are beneficial for AD patients and may at least slow the progression of disease. However this trial raised concerns about the safety and the efficacy of the active immunization strategy with Aβ₄₂ self-peptide. The low number of vaccine responders, the relatively low antibody titers generated even after multiple immunizations suggest that alternative immunotherapeutic strategies should be pursued. New generation vaccines should be able to induce high titers of anti-Aβ antibodies without the generation of autoreactive T cell activation. Th2-type rather than Th1-type adjuvant should be used to avoid inflammation in the brain. In addition, lessons from the AN-1792 trial showed that even anti-Aβ antibodies may reduce the amyloid plaques in the brain of vaccinated subjects, the increase of soluble Aβ in result of the solubilization of these plaques may have more toxic effect on neurons [38]. Therefore AD immunotherapy may be most effective if started before the accumulation of toxic forms of Aβ peptide in people with early pre-clinical stages of AD or even in healthy people with family history or genetic susceptibility to AD. Hopefully, the diagnosis of AD at very early stages will be possible in the nearest future by measuring biomarkers such as the Tau/Aβ and pTau/Aβ ratios in CSF [44–46] and/or detecting accumulation of Aβ in the brains using PIB-PET scans [47]. The availability of a safe and effective vaccine at that time would allow the prevention or at least a delay in the onset and progression of AD.

NEW GENERATION AD VACCINES

As discussed above, one challenge associated with the clinical use of the self Aβ₄₂ peptide as a vaccine is the potential for the development of unwanted anti-Aβ/APP Th1 immune responses [35,36,42]. The molecular basis for this reactivity is the uptake of fibrillar Aβ₄₂ (fAβ₄₂) by antigen-presenting cells (APC) and presentation of self-epitopes to T cells through MHC class molecules. If these epitopes are recognized by specific Th1 lymphocytes, the T cells will become activated and produce proinflammatory cytokines. In addition, activated microglia that are a major source of inflammatory factors and are associated with Aβ plaques in AD brain samples, can mediate the permeability of the BBB and recruit immune cells from the periphery. Subsequently, activated microglia expressing MHC class II molecules may process and present Aβ peptide to Aβ-specific Th1 cells that have crossed the BBB, and induce additional inflammation. Although the activation of microglia as a response to signals from injured neurons is a protective process [48], chronic inflammation may be harmful and lead to neurodegeneration [48]. Neurons have only a limited capacity for repair and cannot tolerate long-term inflammation, so inflammatory processes must be tightly regulated [49].

Different types of T cells may have different effects on microglial cell activation and amyloid clearance. Th1 cells can promote microglia activation and inflammation through secretion of IFNγ and up-regulation of MHC class II molecules by microglial cells. The anti-inflammatory cytokines produced by Th2 cells have been shown to exhibit neuroprotective properties [49] and appear to support eventual microglia apoptosis [50] following an inflammatory stimulus, and have been demonstrated to act as an important feedback mechanism to turn off an inflammatory response in vivo [51]. Glial cells, in turn, can facilitate apoptosis of infiltrating T cells in order to reduce brain inflammation. Interestingly, one report demonstrated that Th1, but not Th2 cells underwent apoptosis after such activation [52]. Recently, interesting studies have been performed to demonstrate the effect of T cells and related cytokines on Aβ degradation in human primary cultures of macrophages and microglia. These studies demonstrated that proinflammatory cytokines inhibited, whereas anti-inflammatory and regulatory cytokines (IL-4, IL-10 and TGFβ) enhanced Aβ degradation in macrophages. IFNγ and TNFα directly suppressed the expression of Aβ degradation enzymes in monocyte-derived macrophages [53]. These data also support the suggestion that generation of Th2 response in AD patients is more beneficial and less harmful than Th1 response.

Studies show that cognitively normal elderly and AD patients have an increase in auto-reactive T cells specific for the Aβ peptide [52]. It is possible that AD patients developing meningoencephalitis in the AN-1792 clinical trial had stronger autoimmune T cell responses and inflammation in the brain that was elicited by immunization with fAβ₄₂ formulated in the strong Th1-type adjuvant, QS21. Thus, it may be important to induce anti-inflammatory Th2 cell responses instead of triggering additional activation of Th1 cells that induce an increased inflammatory environment in the brain. Another way to avoid additional activation of autoreactive T cells is the replacement of self Aβ Th cell epitope with a foreign Th cell epitope while keeping the self-B cell epitope intact. Fortunately, studies mapping B and T cells epitopes of Aβ₄₂ demonstrated that these two epitopes are distinct (see next section).

B cell epitope/s of Aβ₄₂

Several groups have endeavored to map murine Aβ B cell epitopes using small peptides derived from Aβ₄₂ and antisera from mice immunized with fibrillar Aβ₄₂ [54–58]. For example, Lemere et al., [56, 57] described intranasal vaccination of PDAPP mice and determined that among six overlapping peptides, only Aβ_1–15 abolished immunoreactivity of experimental sera with AD plaques in tissue sections. Using four peptides spanning aa 1–16, 10–20, 20–29, and 29–40 of Aβ₄₀, another group [55] demonstrated that antisera from PS1/Tg mice recognize Aβ_1–16 along with Aβ₄₀ and Aβ₄₂ peptides. We also demonstrated that Aβ_1–15 was the major epitope recognized by antibodies raised in wild type and APP/Tg mice immunized with fibrillar Aβ₄₂ formulated in different adjuvants [58]. These results concur with other reports, and support that the Aβ_1–15 peptide represents the major B cell epitope of Aβ₄₂ in mice of different haplotypes and in APP/Tg 2576 animals. In addition, these data have been confirmed using proteolytic cleavage of intact, immobilized immune complexes between Aβ₄₂ and polyclonal antibodies from APP/Tg mice immunized with protofibrillar aggregates of Aβ₄₂ followed by high-resolution Fourier-transform ion cyclotron resonance mass spectrometry (FT-ICR-MS)[59]. This analysis revealed AP4–10 as the smallest region of the Aβ peptide that was recognized with high affinity by these antibodies. Part of this antigenic epitope (₃EFRH₆) was found in amyloid plaques in situ and antibodies to this epitope had therapeutic effects in vitro [60–63]. It has also been demonstrated that antisera against aa 1–11, 1–7, and 1–5 of Aβ₄₂ are more potent in binding to β-amyloid plaques and triggering ex vivo phagocytosis than antibodies to aa 3–7, 5–11, and 11–26 [64]. In addition, these experiments were confirmed by using polyclonal rabbit antibodies to the fibrillar Aβ₄₂ [65]. Such antibodies bound to the linear Aβ_1–8 sequence with relatively high affinity. Importantly, epitope mapping of antibodies generated after AN-1792 vaccination are also primarily N-terminal specific [66].

T cell epitopes of Aβ₄₂

Two human immunodominant cytotoxic and T helper epitopes predicted by computer models are located at the central and C-terminal regions, respectively [67–69]. Monsonego et al. mapped the major T cell epitope in the Aβ_15–42 peptide, which is segregated from the dominant B cell epitopes identified in Aβ_1–15 is [70]. Fine mapping using five overlapped peptides within Aβ_15–42 peptide showed that T cell epitopes were most frequently localized in the Aβ_16–30 region.

We mapped T helper responses to Aβ₄₂ in immune BALB/c mice and demonstrated in vitro T cell activation after re-stimulation with peptides Aβ_6–20, Aβ_1–28, or Aβ_1–40, but not with Aβ_1–15 [58]. Although these data with wild type animals of H2^d haplotype indicate that smaller overlapping peptides might be used to map the T cell epitope of Aβ₄₂ more precisely, it is clear that the Aβ_1–15 peptide does not possess the T cell epitope of Aβ₄₂ in these mice. However, Das et al. demonstrated that the major T cell epitope in APP/Tg mice, as well as HLA-DR3 and HLA-DQ8 Tg mice, is Aβ_1–16. T cells from HLA-DR4 mice responded to both Aβ_1–16 and Aβ_11–28 whereas HLA-DQ6 Tg mice failed to mount significant B and T cell responses to Aβ_1–42 [71]. These data indicated that T cell epitopes recognized by different HLA haplotypes may be localized in different regions of the Aβ peptide. Therefore, it is reasonable to minimize the size of the B cell epitope to avoid the activation of autoreactive T cells in humans bearing specific MHC class II haplotypes. One way of achieving this is to design a vaccine that includes a self Aβ B cell epitope plus a foreign MHC class II T cell epitope that would provide T help for the generation of an effective anti-Aβ antibody response. This approach might avoid the generation of T cell immunity against Aβ and preclude the concerns raised in the AN-1792 clinical trial.

Epitope Vaccines

To estimate the impact of antibodies against different parts of the Aβ peptide in reduction of plaque burden and neuronal protection, Bard et al. immunized PDAPP mice with different N-terminal fragments (Aβ_1–5, Aβ_3–9, Aβ_5–11) or a fragment derived from an internal region of the peptide (Aβ_15–24) conjugated with a T cell epitope derived form ovalbumin. This study demonstrated that antibodies specific to the N-terminal region of Aβ are important for plaque clearance and neuronal protection via an Fc-mediated mechanism [64]. Based on these data, different groups started to work on the design and further refinement of epitope vaccines composed of various N-terminal fragments of Aβ and non-self T cell epitopes.

First generation epitope vaccines tested in mice were based on the Aβ_1–15 peptide [81]. We decided to incorporate a foreign Th cell epitope into the Aβ_1–15 vaccine to activate non-self antigen-specific T lymphocytes which would initiate and direct the formation of anti-Aβ antibodies. Incorporation of a promiscuous foreign T cell epitope into the Aβ-immunogen provides a number of significant advantages to Aβ-immunotherapy design. We selected PADRE (aK-Cha-VAAWTLKAAa, where “a” is D alanine, “Cha” is L-cyclohexylalanine), a synthetic, non-natural Pan HLA DR-binding Epitope [72,73], as our candidate foreign T cell epitope. The potency of this molecule to generate Th cells that induce strong antibody responses against many different antigens is well established [74–78]. When combined with disease-specific antigens, PADRE induces important co-stimulatory signals that amplify the antigen-specific immune response, which generates high titer and high affinity IgG antibodies [72,73,75]. PADRE tested against human T cells in a proliferation assay was found to be 100-fold more potent on a molar basis than a tetanus-derived universal epitope [79]. In addition, PADRE binds with high affinity to 15 of 16 of the most common HLA-DR types tested to date and with high/intermediate affinity to mouse I-A^b/d and I-E^b/d MHC haplotypes [72,73,75]. Because of its binding promiscuity, PADRE is likely to overcome the problems posed by the extreme polymorphism of HLA-DR molecules in the human population. Several other advantages of PADRE made it preferable as a T cell epitope over other carrier-proteins (OVA, KHL, Tetanus Toxoid-derived universal epitope, or MTB) traditionally used to make vaccines: (i) PADRE can be easily synthesized; (ii) the antibody responses are generated not to PADRE, but to the B cell epitope only; (iii) it has been used in human clinical studies and has been shown to be safe and well tolerated [78]. Thus, we coupled the foreign T cell epitope PADRE with an Aβ B cell epitope and predicted that the vaccine would: (1) activate non-self PADRE-specific Th lymphocytes; (2) induce therapeutic anti-Aβ antibody formation; and (3) help to generate a more potent antibody production than with the self-T cell epitope of Aβ.

PADRE-Aβ_1–15 was synthesized on a 4-branched multiple antigenic peptide (MAP) backbone to generate PADRE-Aβ_1–15-MAP. MAPs are capable of inducing high titer antibody responses specific to many different antigens because multiple copies of immunogen attached to the core matrix are significantly more potent [80]. BALB/c mice, known as moderate responders to the PADRE T cell epitope [77], generated robust anti-Aβ₄₂ humoral immune responses after immunization with the PADRE-Aβ_1–15-MAP epitope vaccine formulated in the Th2- type adjuvant, alum. The antibody titers were equivalent to titers in mice immunized with Aβ_1–33, which contains both the B and T cell epitopes of Aβ₄₂. Thus, even though BALB/c mice are low responders to PADRE, they still generated a strong anti-Aβ₄₂ response to the prototype epitope AD vaccine. The antibodies induced by PADRE-Aβ_1–15-MAP were specific to Aβ, because they did not bind to the MAP backbone or PADRE in ELISA assays [81]. As we expected, PADRE-Aβ_1–15-MAP formulated in the Th2 adjuvant, alum, induced anti-Aβ antibodies of mainly the IgG1 isotype in H-2^d mice. Finally, anti-Aβ antibodies generated after immunization of mice with PADRE-Aβ_1–15-MAP were potentially therapeutic because they bound to amyloid plaques in brain tissue from an AD case [81]. Analysis of T cell proliferation in splenocytes from animals immunized with our epitope vaccine demonstrated that mice generated robust T cell responses only to the PADRE but not to the Aβ peptide or MAP backbone. The predominant cytokine produced by splenocytes from immunized mice was IL-4. These results were not surprising, since antigen was formulated in alum, a Th2-type adjuvant, and we immunized BALB/c mice that are normally Th2-dominant [75,82]. Thus, this Aβ epitope vaccine was capable of activating PADRE-specific T cells that stimulated B cells to produce anti-Aβ_1–15 antibodies.

Intranasal immunization of mice with other epitope vaccines based on Aβ_1–15 generated high concentrations of therapeutically potent anti-Aβ antibodies of IgG1 and IgG2b isotypes without the activation of Aβ-specific T cells [83]. Two principally different vaccines have been designed: a tandem repeat of Aβ_1–15 peptide linked by two lysines (2×Aβ_1–15) or Aβ_1–15 peptide conjugated with a T helper cell epitope from the HIV IIIB gp120 glycoprotein (T1-Aβ_1–15). Surprisingly, 2×Aβ_1–15 is formulated in mutant Escherichia coli heat-labile enterotoxin LT adjuvant was more potent at generating anti-Aβ antibodies than T1-Aβ_1–15+LT. Both vaccines generated robust T cell responses specific to the individual vaccines and no responses were generated against full length Aβ₄₂ or Aβ_1–15. The authors speculated that the absence of T cell responses against Aβ_1–15 after immunization with the 2×Aβ_1–15 immunogen may be attributable to the formation of a new, unidentified T cell epitope that is not present in a single sequence of Aβ_1–15. Such new epitopes that work very well in mice would not necessarily work in humans, therefore epitope vaccines should be carefully designed in order to avoid unanticipated responses. Four copies of the Aβ_1–15 epitope (4×Aβ_1–15) were used for generating an AD vaccine based on an adenoviral vector [84]. Immunization of 12 month old Tg 2576 mice induced high titers of anti-Aβ antibodies, reduced cerebral insoluble Aβ levels and behavioral impairment in old Tg 2576 mice, however did not change the level of soluble Aβ in the brain. Like the epitope vaccine 2×Aβ_1–15 described by Maier et al., splenocytes from immunized mice were activated in vitro by 4×Aβ_1–15, but not with Aβ40. Although the authors did not use Aβ_1–15 for re-stimulation of splenocytes, it is likely that again in this case, a new T cell epitope was generated due to the use of multiple copies of the Aβ peptide. DNA prime and adenovirus boost regimens have been used to induce responses against the small epitope Aβ_1–6. Even the presence of 11 copies of this epitope induced only very low titers of anti-Aβ antibodies, probably due to the absence of a potent T cell epitope [85]. Smaller parts of the N-terminal region of the Aβ peptide were used in the design of vaccines based on viral-like particles (VLP). Incorporation of the Aβ B cell epitope into a viral capsid protein allows the expression of this epitope on the surface of VLP in a repetitive and ordered array. Such organization of the epitope may induce T cell-independent B cell activation and production of high titers of anti-Aβ antibodies. On the other hand, T cell epitopes from the viral proteins may help B cells to induce T cell-dependent humoral response. Vaccination of mice and rabbits with papillomavirus VLP that displayed Aβ_1–9 on the viral capsid surface induced measurable titers of anti-Aβ antibodies in rabbits and moderate titers in APP/PS1 Tg mice [86]. Generated antibodies were therapeutically potent and recognized amyloid plaques in brain tissue sections from an AD case. Even the moderate titers of anti-N-terminal antibodies reduced the Aβ load in the cortex and hippocampus of immunized mice, yet the total Aβ in brain homogenates were largely unchanged, although a trend towards lower levels was observed. Another papillomavirus VLP containing Aβ_1–9 was generated through the streptavidin bridge technique [87]. Immunization of C57BL/6 mice with VLPs formulated in IFA induced high titers of anti-Aβ antibodies of predominantly IgG1 isotype without the generation of Aβ-specific T cell responses. The same antigen incorporated into the bacteriophage Qp induced even higher antibody response [87]. Another Qp-based vaccine comprising the Aβ_1–6 epitope (CAD106) covalently linked to VLPs was tested in various species [88,89]. CAD106 injected without any adjuvant induced high titers of antibodies in mice, rabbits and monkeys. The therapeutic potency of generated antibodies was shown in vitro and in vivo. Antibodies bound to amyloid deposits in human and APP Tg mouse brain sections, and protected neurons from oligomeric and fibrillar toxicity in vitro. Immunization of APP23 and APP24 Tg mice with CAD106 prevented cortical and vascular plaque deposition. Immunization of aged mice with advanced amyloid pathology reduced the amyloid deposition without inducing microhemorrhages. Currently, Novartis is investigating the safety, tolerability and Aβ-specific antibody responses in a group of patients with mild to moderate AD following three subcutaneous injections of CAD106[89]. Although the report from this trial is not available yet, a VLP-based vaccine is already approved for human papillomavirus, if the safety of QP-VLP based vaccines is established, this platform may serve as an effective advanced delivery system for Aβ epitopes.

All of the epitope based vaccines described above, with the exception of our first generation epitope vaccine based on the Aβ_1–15 B cell epitope and synthetic, non-natural epitope PADRE, either do not contain any well characterized T cell epitope or are based on multiple T cell epitopes from viral proteins in VLP. Although we did not observe anti-Aβ_1–15 specific T cells in our experiments with the APP/Tg mice, in our second generation epitope vaccine we decided to shorten the B cell epitope to aa 1–11, based on the report that APP/Tg and HLA class II Tg mice immunized with fibrillar Aβ₄₂ recognized the Aβ_1–16 peptide [71]. In addition, our new generation epitope vaccines contain 2 or 3 copies of the Aβ B cell epitope, significantly increasing the antibody response and affinity of antibody-antigen binding [63, 99, 100 and unpublished data].

DNA and peptide epitope vaccines were designed based on Aβ_1–11 and PADRE epitopes. The peptide epitope vaccine was synthesized on a 4-branched MAP backbone to generate 2Aβ_1–11 PADRE-MAP immunogen and was formulated in Quil A adjuvant. MAPs provide multiple copies of immunogen attached to the core matrix and significantly enhance antibody responses [80,90]. 2Aβ_1–11-PADRE-MAP formulated in QuilA induced robust anti-Aβ antibody responses in wild-type and two different APP Tg mice: 3×Tg-AD and Tg 2576, without activation of autoreactive T cells. However, some problems associated with the development of safe and effective epitope vaccines still exist. For example, a peptide epitope vaccine is immunogenic when it is synthesized on a MAP backbone but cannot be easily scaled up for clinical studies. Another issue associated with a peptide epitope vaccine strategy is the requirement for a potent conventional adjuvant. The only adjuvant currently approved for use in humans is alum, which may not be capable of inducing strong anti-Aβ immunity in elderly patients as was previously shown in mice [91].

An alternative to the peptide vaccination approach is DNA immunization. DNA-based vaccination provides a unique method of vaccination [92], exhibiting properties that may be advantageous for the development of vaccines against a variety of pathogens as well as for human diseases including cancer, autoimmune disorders, and Alzheimer’s disease. A unique property of DNA-based vaccination is the ability to induce prolonged, endogenous antigen synthesis and processing within the subject’s own cells. DNA immunization has been shown to generate protective humoral and cellular immune responses against multiple viral, bacterial and tumor antigens [93–98]. This approach also permits rational inactivation or removal of sequences encoding potentially toxic or oncogenic protein domains as well as the inclusion of molecular adjuvants, such as cytokines, which can direct T helper cell responses toward the desired pathway.

Thus, we designed and constructed a new generation DNA epitope vaccine composed of macrophage derived chemokine (MDC) as a molecular adjuvant, 3 copies of the Aβ_1–11 B cell epitope, and the promiscuous T cell epitope, PADRE. Both wildtype and 3xTg-AD mice immunized with the DNA epitope vaccine induced robust anti-Aβ humoral and Th2-polarized immune responses. Importantly, the DNA epitope vaccine, like the peptide epitope vaccine, did not generate autoimmune cellular responses specific to Aβ [63,99,100]. MDC molecular adjuvant incorporated into the vaccine design not only facilitates the internalization of antigen into endo-lysosomal compartments of target CCR4⁺ APC, and increases the efficiency of processing and presentation of antigens, but also plays a critical role in the antigen-induced recruitment of Th2 ceils via chemotaxis and activation of CCR4-expressing Th2-type CD4⁺ T cells [101–103], followed by B-cell activation [104].

In vitro characterization of therapeutic features of anti-Aβ_1–11 antibodies generated after immunization of 3xTg-AD mice with the DNA epitope vaccine demonstrated that the antibodies bound to different forms of Aβ₄₂, although the affinity of binding to oligomers is higher than to fibrils. The antibodies bound to amyloid plaques in the brain sections of an AD case, reduced amyloid plaque burden after intracranial injection of APP Tg mice, and inhibited the fibril-mediated cytotoxicity of Aβ₄₂. Substoichiometric concentrations of purified anti-Aβ_1–11 antibody prevented aggregation of Aβ₄₂, and induced disaggregation of preformed Aβ₄₂ fibrils down to a non-filamentous and nontoxic species. Although reduced oligomer-mediated cytotoxicity was observed upon preincubation of Aβ oligomers with the anti-Aβ_1–11 antibody, antibodies could not disaggregate oligomers though they did delay Aβ₄₂ oligomer formation. These in vitro observations suggest that therapeutic vaccination cannot disrupt in vivo toxic oligomers and may only minimally inhibit preexisting AD pathology and support the hypothesis that immunotherapy should be started before Aβ accumulation in the vasculature and parenchyma of brains induces an unalterable process. In other words, if AD pathology is associated with accumulation of oligomeric forms of Aβ₄₂ in the brain, the therapeutic vaccination strategy will likely not be effective. To verify this concept, we compared data generated after immunization of the APP Tg mice with or without pre-existing AD-like pathology at the start of immunization [99,100]. In both cases, vaccination significantly decreased the Aβ depositions in the brains of mice. A potential problem of immunotherapy could be the fact that a reduction of insoluble Aβ may lead to increased levels of soluble forms of this peptide [38], primarily oligomers, the most toxic form for neurons, and impair cognitive function [105]. In these studies we demonstrated that early vaccination prevented the accumulation of both insoluble and soluble/oligomeric Aβ and prevented cognitive dysfunction in immunized animals, whereas vaccination started in mice with pre-existing AD-like pathology reduced insoluble but not soluble/oligomeric Aβ. In both cases vaccination with the epitope vaccine stimulated less activation of microglia compared with non-immunized controls and did not induce microhemorrhages. These data and the fact that the AN-1792 vaccine was not clinically effective in AD patients, suggest that AD immunotherapy may be most effective not as a therapeutic measure but as a prophylactic measure for the treatment of people with early pre-clinical stages of AD.

Currently Elan Corporation is conducting a Phase 2 clinical trial (ACC-001) using as an immunogen the Aβ_1–7 epitope conjugated with the carrier protein CRM197 and formulated in QS21 adjuvant. The ACC-001 trial was temporarily suspended due to vasculitis (inflammation of the blood vessels) detected in one patient. It is not yet clear if the skin lesions were related to the antigen, the alternative adjuvant QS-21, or to something else entirely. The results of this trial will provide important perspective regarding the potential efficacy of an epitope vaccine approach for AD. However, some safety concerns may still remain due to the use of the strong inflammatory QS21 adjuvant.

FUTURE DIRECTIONS

One important obstacle for active vaccination of AD patients is hypo-responsiveness to vaccines in the elderly. It is known that immunosenescence in the elderly contributes to the poor outcome of vaccination and increased susceptibility of the elderly to infectious disease. The T cell population shifts to a lower ratio of naïve to memory cells, and fewer naïve T cells are produced by the thymus with age [106,107]. The involution of the thymus is almost complete at the age of 60 years and new T cells can no longer be generated. In addition, the in vivo production of proinflammatory cytokines is upregulated in the elderly [107,108]. This problem could potentially be overcome by regular vaccination of people with very early-stage AD, which can be diagnosed by measuring tau/Aβ and ptau/Aβ ratio in the CSF and/or by screening for accumulation of Aβ in the brains using PIB-PET scan [44–47], and subsequently providing yearly booster vaccinations. Another possible strategy to counteract immunosenescence is to recruit previously generated memory T cells produced during childhood vaccination or during prior exposure to human pathogens. The majority of elderly people already possess a broad panel of specific memory T cells. Vaccination of these people with a recombinant vaccine composed of a B cell epitope of the desired antigen (Aβ peptide in the case of AD vaccination) and Th epitopes from conventional vaccines or common pathogens, may boost the pre-existing memory T cells to expand rapidly and differentiate into effector T cells, which would lead to a faster and stronger antibody response specific to the fusion protein antigen. Tetanus toxin (TT), diphtheria toxin (DT), influenza virus, HBV, measles are examples of such conventional vaccines/pathogens. As a result of mandatory vaccination against TT, DT, measles, yearly vaccination and/or infection with influenza, and vaccination against HBV, almost everyone should have memory CD4⁺ T cells specific to epitopes in these vaccines. Thus, a booster injection with an epitope vaccine containing these conventional Th epitopes along with a strong Th2 molecular adjuvant should be enough for the rapid activation of pre-existing anti-TT, anti-DT, anti-HBV and anti-flu memory Th cells, resulting in the induction of strong and potent anti-Aβ antibody production in the elderly.

Nevertheless, even if effective strategies are developed to increase antibody responses in elderly people, analysis of pre-clinical and clinical data accumulated over the past decade suggest that vaccination will be effective if started before the accumulation of toxic forms of β-amyloid or at least at the early stages of AD. Antibodies may prevent or delay the oligomerization of Aβ₄₂, but it is very difficult to disrupt preformed oligomers [63]. Of course, a prophylactic vaccine must be very safe. Incorporation of universal foreign T cell epitopes, as well as a molecular adjuvant that could be much safer than currently available conventional adjuvants, into a DNA epitope vaccine provides such a possibility. Another effective way to develop safe AD vaccines is with an epitope vaccine composed of a B cell epitope mimicking the structure of toxic forms of Aβ₄₂ and a foreign T cell epitope. Such a vaccine has the potential to be safe even for vaccination of young, healthy individuals and could prevent the accumulation of toxic β-amyloid peptides. Thus, the generation of peptides mimicking toxic forms of Aβ peptide (mimotopes) is one the most promising strategies for the development of a safe and effective AD vaccine.

CONCLUSIONS

In summary, we believe that AD progression could be safely prevented or delayed if (i) the vaccine induces high titers of antibodies specific to toxic forms of Aβ without activation of harmful autoreactive T cells and induction of inflammation; (ii) vaccination starts before or at least at the early stages of the accumulation of toxic forms of β-amyloid peptide. Present-day achievements in development of diagnostic and prognostic methodologies make this vaccination scheme possible.

ABBREVIATIONS

Aβ

Amyloid-beta

AD

Alzheimer’s disease

APC

Antigen-presenting cells

APP

Amyloid precursor protein

BBB

Blood brain barrier

CAA

Cerebral amyloid angiopathy

CSF

Cerebral spinal fluid

CTF

C-terminal fragment

LRP

Lipoprotein receptor-related protein 1

MAP

Multiple antigenic peptide

MDC

Macrophage derived chemokine

NFT

Neurofibrillary tangles

PADRE

Pan DR epitope

TT

Tetanus toxin

VLP

Vital like particles