Aβ is produced by the sequential cleavage of its precursor APP by two enzymes, namely β-secretase and γ-secretase (Fig. 9.2). β-secretase first cleaves APP at the extracellular cleavage site, and then γ-secretase cleaves the resultant C-terminal APP fragment within the transmembrane domain to generate several Aβ species of 36-43 amino acids. Among the Aβ peptides, the 40-amino-acid peptide (Aβ40) is the most abundant species, whereas the 42-amino-acid peptide (Aβ42) is less abundant but has a higher aggregation propensity. APP is also known to be cleaved by a third enzyme, α-secretase, within the Aβ sequence, resulting in no production of Aβ. Therefore, β-secretase and γ-secretase are considered as rational therapeutic targets to reduce Aβ production (Bignante et al. 2013).

Various immunological therapeutic approaches against Aβ, including both active and passive immunization, have been actively investigated (Lannfelt et al. 2014). Antibodies against Aβ are thought to enter the brain to some extent, and to remove Aβ from preexisting amyloid plaques via microglial activation and subsequent phagocytosis (Bard et al. 2000). Aβ immunotherapy is also expected to act in the peripheral circulation to reduce Aβ levels, leading to the excretion of monomeric Aβ from the brain according to the equilibrium of Aβ between the peripheral blood and CNS, which is called the “peripheral sink hypothesis” (DeMattos et al. 2001). Aβ antibodies are also expected to directly interfere with the aggregation and further accumulation of Aβ in the CNS. The active immunization approach offers many advantages over passive immunization, in particular, a sustained antibody response by a limited number of vaccinations, and its lower cost. Its potential disadvantage is the variability in the antibody response among patients, including the occurrence of unexpected responses. On the other hand, the potential advantages of passive immunization are the reproducible and reliable delivery of therapeutic antibodies to the patient, and their rapid clearance. This also results in the disadvantage of the requirement for repeated antibody infusions over extended periods.

According to the “amyloid hypothesis”, the aggregation and subsequent deposition of Aβ peptides as amyloid plaques in the brain are thought to trigger the pathogenic cascade in AD (Hardy and Selkoe 2002; Hardy and Higgins 1992). Conformational changes and aggregation of Aβ are facilitated by its interaction with various biomolecules such as other proteins, lipids, and metals. Therefore, various therapeutic approaches to prevent Aβ aggregation, either by inhibiting its self-assembly, or its interaction with other biomolecules have been investigated as described below (Dasilva et al. 2010).

Soto and colleagues first developed the idea of using Aβ aggregation inhibitors as potential therapies for AD. The core amyloidogenic sequences of Aβ required for its self-assembly are reported to be the hydrophobic residues 16-20 (KLVFF) and residues 25-35. They designed Aβ-based peptide inhibitors, the so-called “β-sheet breaker peptides”, by introducing a single proline substitution in the former core amyloidogenic region to abolish the β-sheet propensity of Aβ (Soto et al. 1996). Τhese β-sheet breaker peptides were indeed shown to inhibit Aβ fibril formation and to disassemble preformed Aβ fibrils in vitro, and furthermore, were shown to reduce amyloid deposition, neuronal loss, and accompanying brain inflammation in Aβ-injected amyloidosis rats, and APP transgenic mice (Soto et al. 1998; Permanne et al. 2002). These studies clearly indicated that Aβ aggregation is one of the causes of neurodegeneration in the pathogenesis of AD, and that the inhibition of Aβ aggregation leads to the suppression of downstream pathogenic events.

Tramiprosate (3-amino-1-propanesulfonic acid/Alzhemed; Neurochem Inc.) is a glycosaminoglycan (GAG) mimetic that interferes with the binding of GAGs with the Aβ peptide, thereby preventing Aβ fibril formation. Tramiprosate treatment was reported to reduce the amyloid plaque load, as well as soluble and insoluble Aβ levels in the brain of APP transgenic mice (Gervais et al. 2007). Although tramiprosate was shown in a Phase II study to be safe and well tolerated, to cross the blood-brain barrier (BBB), and to dose-dependently reduce CSF Aβ levels, a large Phase III clinical trial in mild-to-moderate AD patients failed to meet the predefined primary endpoints. However, post-hoc analyses revealed its positive and significant effects on secondary endpoints, including the attenuation of hippocampal volume loss, as well as a trend toward the slowing of cognitive decline (Aisen et al. 2011), suggesting the potential of tramiprosate as a disease-modifying drug for AD.

PBT2, an 8-hydroxy quinolone derivative, works as a chelator of copper, zinc and iron, and thereby disrupts the aberrant interactions of these metals with Aβ. Treatment of APP/PS1 transgenic mice with PBT2 resulted in a decrease in the brain Aβ levels and cognitive improvements (Adlard et al. 2008). In a Phase IIa clinical trial in early AD patients, the cognitive efficacy of PBT2 was limited to only two measures of executive function, although a dose-dependent and significant reduction in CSF Aβ42 levels was observed (Lannfelt et al. 2008).

Scyllo-inositol (cyclohexanehexol/AZD-103), a derivative of the natural glycolipid phosphatidylinositols, has been shown to interfere with Aβ aggregation by competing with phosphatidylinositol lipids for binding to Aβ, which is known to facilitate Aβ fibril formation. McLaurin and colleagues showed that oral administration of scyllo-inositol reduces brain Aβ levels and amyloid plaques, and improves cognitive deficits in APP transgenic mice (McLaurin et al. 2006). These therapeutic effects were still observed even when mice were treated after the onset of AD-like symptoms. However, a Phase II clinical trial of scyllo-inositol resulted in no evidence of clinical benefit, and only a decrease in the CSF Aβ levels was observed (Salloway et al. 2011). Therefore, the clinical efficacy of scyllo-inositol awaits validation in a larger Phase III study.

Epigallocatechin-3-gallate (EGCG) is a major polyphenolic component of green tea. Although much attention was paid to its anti-oxidant activity against Aβ-induced toxicity in earlier studies, EGCG was subsequently shown to bind directly to Aβ, thereby inhibiting its fibrillization, and instead, redirecting Aβ to assemble into non-toxic large spherical aggregates in vitro (Ehrnhoefer et al. 2008). EGCG also targets other aggregation-prone proteins, such as mutant huntingtin with a polyglutamine expansion, tau, α-Syn, prion, transthyretin (TTR), and islet amyloid polypeptide (IAPP) to inhibit their amyloid fibrillization. Administration of EGCG to APP transgenic mice was reported to reduce Aβ levels and amyloid plaque load, as well as cognitive impairment, indicating its therapeutic potential in vivo (Rezai-Zadeh et al. 2005).

Curcumin is a major constituent of the dietary spice turmeric, which has been reported to exert anti-oxidant and anti-inflammatory properties (Monroy et al. 2013). Curcumin was found to inhibit Aβ fibril formation and also to destabilize preformed Aβ fibrils in vitro (Ono et al. 2004). It was shown to bind directly to the fibrillar conformation of Aβ, and to also inhibit the aggregation of other amyloidogenic proteins including Aβ, α-Syn, prion, TTR and IAPP. In vivo studies showed that peripherally-injected curcumin crosses the BBB and labels amyloid plaques in APP transgenic mice, and its administration reduced amyloid plaque burden (Yang et al. 2005). In a Phase II clinical trial in mild-to-moderate AD patients, oral administration of curcumin was shown to be generally safe and well tolerated, but it failed to show any clinical or biomarker efficacy (Ringman et al. 2012), and the plasma curcumin levels were found to be very low. The major obstacle to the clinical application of curcumin is its poor bioavailability, mainly due to its insolubility, poor absorption, and rapid metabolism (Anand et al. 2007). Therefore, derivatives of curcumin with improved bioavailability and delivery, are highly anticipated due their role as potential therapeutic agents against these incurable diseases in the near future.

Since the aggregation and subsequent deposition of Aβ in the brain are regarded as the primary culprits in the pathogenesis of AD (Hardy and Selkoe 2002), the degradation of Aβ, and not only its production, is thought to be one of the more attractive therapeutic targets for AD. In particular, a recent report showing that the clearance of Aβ, rather than its production, is altered in AD patients, has highlighted the importance of Aβ degradation (Mawuenyega et al. 2010). Although there are many enzymes capable of cleaving Aβ in vitro, only a few have been confirmed to be involved in degrading Aβ in vivo.

Neprilysin, a zinc-dependent metalloprotease, is the most extensively studied Aβ-degrading enzyme in the brain (Iwata et al. 2000). Genetic disruption of the neprilysin gene in mice confirmed its importance in the degradation of endogenous Aβ, as well as exogenously administered Aβ (Iwata et al. 2001). On the other hand, expression of neprilysin in APP transgenic mice, either by crossing them with neprilysin transgenic mice or by viral vector-mediated gene transfer, has been shown to significantly reduce brain Aβ levels and amyloid plaques, and to improve their life span (Leissring et al. 2003; Marr et al. 2003), indicating its therapeutic potential. It is noteworthy that the effect of environmental enrichment on reducing Aβ load in APP mice is closely correlated with the upregulation of neprilysin levels (Lazarov et al. 2005).

Insulin-degrading enzyme (IDE), a thiol metalloendopeptidase that degrades small peptides such as insulin and glucagon, is another protease that plays an important role in degrading Aβ in the brain (Qiu et al. 1998). Indeed, IDE knockout mice exhibited increased levels of endogenous Aβ in the brain, and developed hyperinsulinemia and glucose intolerance, which have also been proven to be risk factors for AD (Farris et al. 2003). The overexpression of IDE as well as neprilysin, has been shown to reduce the Aβ load and associated brain changes in APP transgenic mice (Leissring et al. 2003). Therefore, the discovery of these Aβ-degrading enzymes has revealed novel therapeutic opportunities for AD through the upregulation of their activities.

From the failures of most Aβ-targeted disease-modifying therapies in demonstrating clinical efficacy for AD patients as described above, two major concerns have emerged. First, the time point of treatment initiation might be too late in the course of the disease. In most preclinical studies, therapeutic interventions were started before disease onset in genetically-engineered animals destined to develop the disease. On the other hand, AD-related pathological changes are considered to begin many years before the clinical onset of symptoms (Price and Morris 1999; Bateman et al. 2012), which might prevent the translation of preclinical results obtained from animal models, to clinical efficacy in human AD patients. Another concern is the validity of the therapeutic target. Notably, although Aβ immunotherapies have succeeded in the prevention and removal of amyloid deposition in AD patients, they did not lead to an improvement of cognitive decline. While both amyloid plaques and NFTs are the characteristic pathological features of AD, NFT pathology is known to be more closely correlated with the severity of clinical symptoms of AD than amyloid plaques (Arriagada et al. 1992). Furthermore, the reduction of endogenous tau levels was reported to ameliorate cognitive deficits without affecting the Aβ pathology in APP transgenic mice (Roberson et al. 2007). These facts challenge the current amyloid cascade hypothesis for AD pathogenesis. In addition, the discovery of tau mutations in frontotemporal dementia and parkinsonism linked to chromosome 17 (FTDP-17), another form of degenerative dementia, highlighted the significant role of tau in the pathogenesis, which is solely sufficient to cause neurodegeneration in the absence of amyloid pathology (Hutton et al. 1998; Spillantini et al. 1998). Therefore, an alternative therapeutic target, namely tau, the key component of NFTs, is currently being considered and intensively investigated (Giacobini and Gold 2013).

Tau is a member of the microtubule-associated protein family that is involved in the assembly and stabilization of microtubules. It is predominantly expressed in neurons, particularly in their axons, to maintain neuronal morphology and axonal transport. Six tau isoforms are produced via alternative mRNA splicing, which mainly differ in their number of three or four microtubule binding repeats, and all are expressed in the adult human brain. In the AD brain, tau is abnormally hyperphosphorylated and aggregated into NFTs. Hyperphosphorylation of tau is known to reduce the binding ability of tau to microtubules, similar to its FTDP-linked mutations (Hong et al. 1998). However, whereas the hyperphosphorylation of tau is reported to promote its self-assembly into NFTs (Alonso et al. 2001), tau itself can assemble into β-sheet-rich filaments through its microtubule binding repeat. Among the various aggregated species, tau oligomers are believed to be toxic, leading to synaptic dysfunction (Lasagna-Reeves et al. 2011). Thus, therapeutic approaches to inhibit tau aggregation are currently being investigated.

The first tau aggregation inhibitor reported was the phenothiazine methylene blue (MB), which blocks tau-tau interactions through the microtubule-binding repeat domain (Wischik et al. 1996). Administration of MB was shown to reduce the soluble tau levels, but not pre-existing NFTs, resulting in an improvement of cognitive deficits in tau transgenic mice (O’Leary et al. 2010). A Phase II clinical trial of MB in mild-to-moderate AD patients was conducted and reported to have beneficial effects (Schirmer et al. 2011). Detailed reports on this Phase II study are anticipated, and Phase III trials with a new form of this compound are now underway (Wischik et al. 2014).

Recent studies have indicated that tau aggregates can be secreted from neurons, and extracellular tau aggregates contribute to the neuron-to-neuron propagation of tau pathology via a prion-like transmission mechanism (Clavaguera et al. 2009; Frost et al. 2009). Therefore, therapeutic approaches to remove extracellular tau aggregates by immunotherapy are under investigation.

The first active immunization study using recombinant human tau with no phosphorylation resulted in the induction of tauopathy-like pathologies and encephalomyelitis with neurological deficits in mice, indicating the potential risk of using unphosphorylated tau to elicit an autoimmune response (Rosenmann et al. 2006). However, subsequent active immunization studies using phosphorylated tau have succeeded in reducing tau pathology and improving cognitive deficits in mutant tau transgenic mice (Asuni et al. 2007). Passive immunization studies using either phosphorylated tau antibodies or a conformation-dependent tau antibody, have also demonstrated significant improvements in motor deficits and tau pathology in mutant tau transgenic mice (Boutajangout et al. 2011; Chai et al. 2011). Taken together, these results indicate the potential of tau immunotherapy as a therapeutic approach for AD and related tauopathies.

The hyperphosphorylation of tau is believed to be one of the most critical steps in tau-mediated neurodegeneration, although it is still controversial as to whether phosphorylation of tau triggers its aggregation, or is just a consequence of its deposition. Therefore, kinases and phosphatases involved in tau phosphorylation and dephosphorylation, respectively, are regarded as therapeutic targets for AD and other tauopathies. Although several kinases, such as glycogen synthase kinase 3 β (GSK-3β), cyclin dependent kinase 5 (CDK5), casein kinase 1 (CK1), and mitogen activated protein kinases (MAPKs) have been reported to phosphorylate tau in vitro, GSK-3β is believed to be the most promising candidate tau kinase because it is activated by Aβ and is essential for Aβ-induced neurotoxicity, thus linking Aβ to tau phosphorylation.

Administration of the GSK-3 inhibitor lithium chloride has been shown to reduce the levels of tau phosphorylation and insoluble aggregation, as well as the degree of axonal degeneration in mutant tau transgenic mice (Noble et al. 2005). Accordingly, small short-term clinical trials of lithium chloride have been conducted in mild-to-moderate AD patients, resulting in no beneficial effect on cognitive impairment, but a significant decrease in CSF phosphorylated tau levels (Forlenza et al. 2011; Hampel et al. 2009). Tideglusib (NP-12), a novel specific GSK-3 inhibitor, was tested in double-transgenic mice co-expressing human mutant APP and tau, resulting in decreased tau phosphorylation and Aβ deposition, as well as the prevention of memory deficits (Sereno et al. 2009). Currently, phase II clinical trials of tideglusib in patients with AD and progressive supranuclear palsy (PSP), one of the tauopathies, are ongoing.

Another approach to inhibit tau hyperphosphorylation is to activate the protein phosphatases that dephosphorylate tau. Protein phosphatase 2A (PP2A) was reported to be the main phosphatase involved in tau phosphorylation in the brain (Goedert et al. 1992; Gong et al. 2000). Sodium selenate was shown to increase PP2A activity, to prevent tau hyperphosphorylation and neurodegeneration, and to improve motor performance in mouse models of tauopathy (van Eersel et al. 2010).

Intracellular antibodies, or so-called intrabodies that recognize the expanded polyQ protein have been applied to inhibit its aggregation. The first reported intrabody, C4, was identified from a single-chain Fv phage-display library by Messer and colleagues in 2001, which recognizes the N-terminal region adjacent to the polyQ stretch of huntingtin (Htt). Indeed, they showed that expression of C4 in cell culture potently inhibits the formation of polyQ-expanded mutant Htt inclusions (Lecerf et al. 2001). Furthermore, they successfully demonstrated that the expression of C4 suppresses neurodegeneration in a Drosophila model of HD (Wolfgang et al. 2005), and its viral vector-mediated expression delays Htt aggregation in the brain of HD model mice (Snyder-Keller et al. 2010).

Other intrabodies include V_L 12.3, which recognizes the N-terminal region of Htt (Colby et al. 2004), and MW7 and Happ1, both of which recognize the proline-rich region (PRR) of Htt (Khoshnan et al. 2002; Southwell et al. 2008). V_L 12.3 was shown to suppress Htt aggregation and cytotoxicity in yeast, although it failed to show any beneficial effects on HD mouse models upon viral vector-mediated expression (Southwell et al. 2009). On the other hand, Happ1 was shown to promote the turnover of mutant Htt, and Happ1 expression successfully suppressed Htt aggregation, and improved motor and cognitive deficits in HD mouse models (Southwell et al. 2008, 2009). Viral vector-mediated expression of another intrabody EM48, which is thought to recognize a C-terminal epitope of human Htt exon1, was also shown to suppress Htt neuropil aggregation and to ameliorate neurological symptoms in a mouse model of HD (Wang et al. 2008).

The application of intrabodies appears to be an attractive approach with regard to their high binding affinity to the disease-causative proteins. However, most of these intrabodies cannot be applied to polyQ diseases other than HD, and may possibly exert unfavorable effects on the normal Htt protein, because intrabodies that specifically recognize the expanded polyQ stretch itself have not been identified so far. The most critical issue is their delivery into the human brain, as they currently require to be expressed via gene delivery due to their large size.

Peptides have potential as therapeutic molecules since they are of relatively low molecular weight, and are therefore expected to be delivered to the target organ by their administration (Takeuchi et al. 2014). Indeed, various peptides-based therapeutic approaches have been applied to inhibit protein aggregation, and have been reported to exert therapeutic effects in cell culture and animal models of other neurodegenerative diseases, such as AD and PD (El-Agnaf et al. 2004; Soto et al. 1998).

Based on the hypothesis that peptides with selective binding affinity to the expanded polyQ stretch are expected to interfere with the conformational changes and aggregation of polyQ proteins, we screened combinatorial peptide phage display libraries. We successfully identified polyQ binding peptide 1 (QBP1: SNWKWWPGIFD), which selectively binds to the abnormally expanded polyQ stretch with a dissociation constant of 5.7 μM (Nagai et al. 2000; Okamoto et al. 2009; Popiel et al. 2013), and showed that QBP1 indeed inhibits the toxic β-sheet conformational transition and aggregation of the expanded polyQ protein in vitro (Nagai et al. 2007, 2000). We also found that expression of QBP1 significantly suppresses polyQ oligomer formation and polyQ-induced cytotoxicity in cell culture (Nagai et al. 2000; Takahashi et al. 2008, 2007), and suppresses polyQ-induced neurodegeneration in Drosophila models of the polyQ diseases (Nagai et al. 2003). We further demonstrated that QBP1 conjugated with protein transduction domains (PTD-QBP1) is efficiently delivered into cultured cells, and its oral administration successfully suppresses polyQ-induced premature death in Drosophila, indicating the potential of PTD-QBP1 as a therapeutic molecule (Popiel et al. 2007). We further investigated the therapeutic potential of PTD-QBP1 on a polyQ diseases mouse model; however, repeated intraperitoneal injection of PTD-QBP1 resulted in only a modest improvement of body weight loss, with no improvement in the motor phenotypes, nor suppression of inclusions in the brains, probably due to its low BBB permeability (Popiel et al. 2009). Therefore, one of the future directions is obviously to design low molecular weight chemical analogs of QBP1 with high BBB permeability (Popiel et al. 2013; Tomita et al. 2009).