Targeting amyloid clearance in Alzheimer's disease as a therapeutic strategy

Abstract

Targeting the amyloid‐β (Aβ) peptide cascade has been at the heart of therapeutic developments in Alzheimer's disease (AD) research for more than 25 years, yet no successful drugs have reached the marketplace based on this hypothesis. Nevertheless, the genetic and other evidence remains strong, if not overwhelming, that Aβ is central to the disease process. Most attention has focused on the biosynthesis of Aβ from its precursor protein through the successive actions of the β‐ and γ‐secretases leading to the development of inhibitors of these membrane proteases. However, the levels of Aβ are maintained through a balance of its biosynthesis and clearance, which occurs both through further proteolysis by a family of amyloid‐degrading enzymes (ADEs) and by a variety of transport processes. The development of late‐onset AD appears to arise from a failure of these clearance mechanisms rather than by overproduction of the peptide. This review focuses on the nature of these clearance mechanisms, particularly the various proteases known to be involved, and their regulation and potential as therapeutic targets in AD drug development. The majority of the ADEs are zinc metalloproteases [e.g., the neprilysin (NEP) family, insulin‐degrading enzyme, and angiotensin converting enzymes (ACE)]. Strategies for up‐regulating the expression and activity of these enzymes, such as genetic, epigenetic, stem cell technology, and other pharmacological approaches, will be highlighted. Modifiable physiological mechanisms affecting the efficiency of Aβ clearance, including brain perfusion, obesity, diabetes, and sleep, will also be outlined. These new insights provide optimism for future therapeutic developments in AD research.

Linked Articles

This article is part of a themed section on Therapeutics for Dementia and Alzheimer's Disease: New Directions for Precision Medicine. To view the other articles in this section visit http://onlinelibrary.wiley.com/doi/10.1111/bph.v176.18/issuetoc

Abbreviations

AD

Alzheimer's disease

ADE

amyloid‐degrading enzyme

AF

atrial fibrillation

AICD

APP intracellular domain

ApoE

apolipoprotein E

APP

amyloid precursor protein

ARB

angiotensin receptor blocker

Aβ

amyloid‐β peptide

ECE

endothelin‐converting enzyme

EE

enriched environment

EGCG

epigallocatechin 3‐gallate

FAD

familial Alzheimer's disease

HDAC

histone deacetylase

HIV

human immunodeficiency virus

IDE

insulin‐degrading enzyme

LRP1

lipoprotein receptor‐related‐protein‐1

NDAN

non‐demented individuals with Alzheimer's neuropathology

NEP

neprilysin

PS1

presenilin‐1

TTR

transthyretin

1. INTRODUCTION

The amyloid cascade hypothesis has been the predominant concept driving therapeutic development in Alzheimer's disease (AD) research since its formulation (Hardy & Higgins, 1992). In the original hypothesis, it is the excess production, aggregation, and deposition of the amyloid‐β peptide (Aβ), particularly Aβ42, derived from enhanced amyloidogenic processing of the amyloid precursor protein (APP) that precipitates the pathogenic process. Other hallmarks of the disease, such as the intracellular neurofibrillary tangles composed of hyper‐phosphorylated tau protein, are considered as downstream events in the original cascade, although both processes may develop independently. The simplicity of this hypothesis led to the numerous strategies to target the key enzymes in the biosynthesis of Aβ, namely, the β‐secretase (β‐APP‐cleaving enzyme 1) and the γ‐secretase/presenilin complex, both aspartic proteases, or to activate the alternative nonamyloidogenic pathway via α‐secretase (a disintegrin and metalloprotease). Over the years, this strategy has been broadened to pursue downstream targets of Aβ, peptide aggregation, Aβ immunotherapy, tau phosphorylation, neuroinflammation, and mechanisms of neuronal cell death. The hypothesis itself has been modified over the years to accommodate current knowledge but remains the key driver for much therapeutic development (Hardy, 2009).

The challenge to bring clinically successful therapeutics to the market for the treatment of dementia and AD has, however, now been an intensive and unfulfilled pursuit for almost three decades (Karran & de Strooper, 2016). In this review, we shall focus specifically on the mechanisms for brain Aβ clearance, in particular by the cohort of proteases that cleave the peptide, its variants and its aggregates (amyloid‐degrading enzymes, ADEs), as well as their regulation and potential as therapeutic targets.

2. PATHWAYS TO AMYLOID, ITS CLEARANCE, AND NEURONAL TOXICITY

Overwhelming genetic and biochemical evidence now implicates APP and its metabolite Aβ in the early and late stages of AD development. Perhaps most compelling is the realization that mutations in the human APP gene can be either neurotoxic or neuroprotective. For example, the A673T variant of APP, seen in 0.5% of Icelandic people, is significantly more common in those >85‐year olds without AD or major cognitive decline (Jonsson et al., 2012). Furthermore, duplication of the APP gene itself leads to early‐onset AD (Rovelet‐Lecrux et al., 2006). Nevertheless, a common argument to dismiss the connection between amyloid accumulation and subsequent dementia is that some individuals remain cognitively normal yet at post‐mortem are found to have significant accumulations of amyloid plaques and tau tangles (nondemented individuals with Alzheimer's neuropathology [NDAN]). This may well be a misleading argument and one which reflects an inherent resistance in these individuals to the initiating actions of Aβ in the amyloid cascade of disease progression or may suggest the involvement of an APP‐related species distinct from Aβ. An examination of the postsynaptic proteome of such NDAN individuals provides some veracity to this hypothesis identifying a unique proteomic signature for the NDAN postsynaptic densities (Zolochevska, Bjorklund, Woltjer, Wiktorowicz, & Taglialatela, 2018). Enhanced neurogenesis may also play a part, and these dementia‐resistant subjects also show an increased number of neural stem cells and new neurons in the hippocampal dentate gyrus as well as a unique expression of neurogenesis‐regulating miRNAs (Briley et al., 2016). These observations in themselves may provide new neuroprotective strategies and therapeutic targets in the search for novel treatments for AD.

The identification of β‐ and γ‐secretases as the key enzymes involved in Aβ synthesis from its precursor protein, APP, highlighted their potential strategic importance in AD therapeutic development, particularly the β‐secretase as the primary and rate‐limiting step in the pathway. Among the constraints, however, on effective inhibitor development is that of enzyme specificity, or rather lack of it. Both β‐ and γ‐secretases, like the majority of proteases, have multiple physiological substrates. Furthermore, the Aβ peptide itself may have a variety of physiological roles, including by acting as a putative transcription factor, haemostatic agent, ion channel regulator, neurotrophic agent, or an antimicrobial peptide (Gosztyla, Brothers, & Robinson, 2018; Hardy, 2007; Maloney & Lahiri, 2011; Pearson & Peers, 2006; Yankner, Duffy, & Kirschner, 1990). Endogenous Aβ may even be necessary for synaptic plasticity (Puzzo et al., 2011), and complete elimination of Aβ production in primary neurons in culture can result in cell death (Plant, Boyle, Smith, Peers, & Pearson, 2003). Furthermore, it has been shown that protecting Aβ from exopeptidase attack by aminopeptidase A can protect cells from caspase activation and cell death (Sevalle et al., 2009). Hence, Aβ should be considered as a regulatory neuropeptide in its own right and total abolition of its production may not be a desirable strategy. It is the soluble oligomers of Aβ, rather than the monomers themselves or the amyloid plaques, that act as synaptotoxins and contribute to the cognitive impairment and cell death in AD, a concept for neurotoxicity originally proposed 20 years ago (Lambert et al., 1998). A variety of mechanisms likely contribute to the toxicity induced by oligomers. The pathways of Aβ production and regulation are shown in Figure 1.

Figure 1.

Cellular pathways to Aβ production and regulation of its clearance. In neuronal cells, proteolytic processing of the APP occurs via the major nonamyloidogenic pathway involving an α‐secretase. It cleaves APP within the Aβ sequence and produces an N‐terminal soluble ectodomain sAPPα and a membrane‐bound fragment, which is, in turn, cleaved by the γ‐secretase complex. This cleavage generates the AICD and a short peptide p3 of unknown function. A minor amyloidogenic pathway begins with APP cleavage by β‐secretase (BACE), which releases an N‐terminal soluble ectodomain sAPPβ and a C‐terminal fragment, which after cleavage by γ‐secretase generates Aβ and AICD. sAPPα and sAPPβ have neuroprotective properties, and Aβ has physiological roles and its action is terminated mainly by NEP or other ADEs. Due to its propensity to aggregate, Aβ can also form oligomers, which are toxic and cause neuronal cell death. Amyloid plaques formed from Aβ in a complex with other proteins are a hallmark of AD. The AICD produced in the α‐secretase pathway is mainly cytosolic and can be cleaved by IDE or caspases. AICD produced in the amyloidogenic pathway binds a stabilizing factor Fe65 and translocates to the nucleus where, together with a histone acetyl transferase Tip60 and other regulatory factors, binds to the promoters of a variety of genes, including NEP and TTR, up‐regulating their expression. Both NEP and TTR are involved in amyloid clearance

The levels of Aβ in its various forms represent an equilibrium between its biosynthesis and its removal or clearance. While overproduction of Aβ is primarily responsible for the development of familial early‐onset forms of AD, it is decreased clearance of Aβ that appears to be the predominant driving force leading to accumulation of the peptide in late‐onset (sporadic) forms of the disease (Mawuenyega et al., 2010). Hence, an understanding of the mechanisms regulating Aβ clearance and its deficits in later life can provide alternative avenues for therapeutic intervention.

Aβ clearance is principally mediated by two distinct mechanisms, namely, through its hydrolysis by cerebral proteases, both intra and extracellular, and independently by transport from the brain and subsequent proteolytic removal in the periphery. As much as 50% of the Aβ load may be removed peripherally emphasizing that interactions between the brain and the periphery may play a substantial role in the development and progression of AD (Wang, Gu, Masters, & Wang, 2017). The first identification of a physiologically relevant Aβ‐degrading protease came with the demonstration that the intracellular, cytosolic insulin‐degrading enzyme (IDE) could bind and hydrolyse Aβ at multiple sites (Kurochkin & Goto, 1994). Shortly afterwards, the plasma membrane zinc metallopeptidase neprilysin (neutral endopeptidase, NEP) was also shown to hydrolyse Aβ efficiently at several sites (Howell, Nalbantoglu, & Crine, 1995), and these two enzymes, and their homologues, have largely dominated the field in terms of physiologically relevant Aβ‐degrading proteases in the brain. In addition to extracellular and neuronal metabolism of Aβ, its uptake and clearance by glial cells, particularly microglia and astrocytes, also constitutes a significant pathway for removal of the peptide (Ries & Sastre, 2016).

Autophagic, proteasomal, and lysosomal pathways may modulate the levels of some AD‐associated proteins with significant crosstalk between these pathways. For example, the proteasome contributes to presenilin‐1 (PS1) degradation, and hence, proteasome inhibitors directly affect PS1‐mediated Aβ production (Marambaud, Ancolio, Lopez‐Perez, & Checler, 1998), although these mechanisms do not appear to play a substantial direct role in Aβ clearance (discussed in Baranello et al., 2015). However, Aβ40 in oligomeric form, while not itself a substrate for the proteasome, can block ubiquitin‐dependent protein turnover contributing to the observed accumulation of ubiquitinated proteins in AD (Tseng, Green, Chan, Blurton‐Jones, & LaFerla, 2008). Aβ42 oligomers do, however, appear to be cleaved, being competitive substrates for the chymotrypsin‐like activity of the human 20S proteasome (Zhao & Yang, 2010). The triggering receptor expressed on myeloid cells 2, which has been associated with AD risk from genome‐wide association studies, is another protein involved in Aβ clearance through its potent binding of Aβ oligomers while its deficiency impairs Aβ metabolism via proteasomal and/or lysosomal pathways in microglial cultures (Zhao et al., 2018).

Alternatively, extrusion of Aβ from the brain can occur by nonenzymic pathways. For example, both Aβ40 and Aβ42 are removed across the blood–brain barrier through their interaction with LDL receptor‐related protein‐1 (LRP1; Bell et al., 2007) which, in its soluble form, is primarily responsible for Aβ transport peripherally. Silencing of astrocytic LRP1 additionally down‐regulates the levels of several ADEs (Liu et al., 2017). Another protein important in the binding and transport of Aβ is the soluble form of the receptor of advanced glycation end products (Cai et al., 2016). After transport of Aβ into plasma, it is then cleared mainly in the kidney, which is highly enriched in the Aβ‐degrading peptidase NEP, and the liver (Ghiso et al., 2004).

A recently identified nonenzymic pathway for Aβ clearance from the brain is the so‐called glymphatic system, a paravascular pathway that facilitates CSF flow through the brain parenchyma and the removal of solutes from the interstitial fluid (Iliff et al., 2012). Other proteins which function in Aβ clearance from the brain are apolipoprotein E (ApoE) and α2‐macroglobulin which facilitate the transport of Aβ across the blood–brain barrier into the blood circulation (reviewed in Ries & Sastre, 2016). The ApoE‐mediated removal of Aβ is isoform‐dependent with its rate of clearance from the interstitial fluid in the order ApoE2 > ApoE3 > ApoE4 (Deane et al., 2008).

Another transport protein which can facilitate Aβ clearance is transthyretin (TTR). TTR is the carrier protein of thyroxine and retinol but is also a major Aβ‐binding protein in the choroid plexus and CSF binding Aβ42 with high affinity. It is also a neuronal gene product, which is neuroprotective, and it may also represent a novel therapeutic target for AD particularly as we have shown its expression level is transcriptionally regulated by APP (Kerridge, Belyaev, Nalivaeva, & Turner, 2014). TTR facilitates Aβ transport and clearance preventing its deposition and toxicity, partially by acting as a metallopeptidase like NEP (Liz et al., 2012). Enhancing TTR interaction with Aβ can reduce cognitive deficits in vivo (Ribeiro et al., 2014). A novel mouse model has also shown that TTR attenuates the neuronal loss and memory deficits induced by Aβ oligomers (Brouillette et al., 2012), whereas deletion of TTR in mice reveals it as a key gene involved in the maintenance of memory during ageing (Brouillette & Quirion, 2008). TTR can suppress the processing of APP (Li, Song, Sanders, & Buxbaum, 2016) and is also a potential disease biomarker being significantly decreased in AD (Velayudhan et al., 2012).

TTR itself can undergo amyloidosis and is a recognized cause of death among centenarians who have resisted other pathologies of the ageing process (Coles & Young, 2012). TTR amyloidosis represents a disorder that is now becoming treatable by use of RNAi therapeutics. In particular, the agent patisiran has just received US Food and Drug Administration approval for treatment of the polyneuropathy caused by hereditary TTR‐mediated amyloidosis in adult patients (Adams et al., 2018). Gene‐based therapies for AD are likewise also being explored at various other intervention points in the amyloid cascade processes (reviewed in Loera‐Valencia et al., 2018). The pathways to amyloid clearance in the brain and periphery are summarized in Figure 2. Multiple mechanisms of clearance of neurotoxic proteins, including tau, huntingtin, and TDP43, also occur in other neurodegenerative disorders of ageing providing potential new therapeutic opportunities (Boland et al., 2018).

Figure 2.

Amyloid clearance in the brain and periphery. Schematic presentation of brain cellular and subcellular localization of the key ADEs as described in the text. Part of Aβ which is not cleaved by brain ADEs is transported via the BBB and the glymphatic system into the periphery where it is mainly degraded by the kidney and liver. Levels of Aβ in the periphery and brain are in equilibrium and enhancing Aβ cleavage peripherally may result in reduced brain Aβ levels (“peripheral sink” hypothesis)

3. THE KEY ADEs

3.1. The NEP family

The zinc membrane metallopeptidase now called NEP has been rediscovered over the decades with multiple substrates, physiological roles, and cellular locations. It is a cell‐surface, integral membrane ectoenzyme with its active site facing the extracellular space where it can degrade circulating regulatory peptides both peripherally and centrally (Turner & Tanzawa, 1997). It was originally discovered as a membrane metalloproteinase at the renal brush border but was subsequently identified as a key enzyme terminating the action of neuropeptides in the brain, especially enkephalins and substance P and hence serving as the “cholinesterase” of peptidergic synapses (Malfroy, Swerts, Llorens, & Schwartz, 1979; Matsas, Fulcher, Kenny, & Turner, 1983). It is also the physiological regulator of natriuretic peptide and endothelin levels and hence a modulator of cardiovascular function. NEP acts as an endopeptidase preferentially, but not exclusively, cleaving small (<50 amino acids) peptides on the N‐terminal side of hydrophobic residues, and hence, the highly hydrophobic Aβ peptides represent an ideal substrate. NEP gained major attention again in 2000 when it was shown to be the major enzyme activity degrading Aβ in brain and that its inhibition led to the accumulation and deposition of the peptide (Iwata et al., 2000). Aβ degradation was substantially, but not completely, abolished in NEP‐deficient mice suggesting that other proteolytic enzymes may also contribute to its catabolism but that NEP was rate‐limiting in vivo (Iwata et al., 2001).

A number of studies have shown an association between NEP polymorphisms and sporadic AD, as well as with the severity of cerebral amyloid angiopathy (Wood, Pickering, McHale, & Dechairo, 2007; Yamada, 2004). A recent meta‐analysis of NEP levels in AD concluded that NEP expression and activity were decreased in the cortex of elderly AD patients adding to the accumulation of evidence that NEP could be a valuable therapeutic target to combat the disease (Zhang et al., 2017). Furthermore, we have reported reduced NEP activity in the blood plasma of early AD patients (Zhuravin et al., 2015) in line with earlier studies showing that CSF NEP is reduced in prodromal AD (Maruyama et al., 2005).

Aβ levels can also contribute to certain degenerative diseases of the retina. In animal models, these raised Aβ levels can be significantly reduced by administration of NEP intravitreally, potentially providing a therapeutic option in such degenerative diseases (Parthasarathy et al., 2015). Other pathological processes can also modulate both Aβ and NEP levels. For example, human immunodeficiency virus (HIV)‐1 infection can increase extracellular Aβ levels in primary cultures of astrocytes while reducing the expression and activity of NEP (Martinez‐Bonet, Muñoz‐Fernández, & Álvarez, 2018). Hence, HIV‐1 presence in the brain might elevate Aβ accumulation, which could be a contributory factor in the detected increases in neurocognitive disorders in HIV‐affected individuals and accumulation of brain Aβ (Achim et al., 2009).

Long before the identification of NEP as a major ADE, it was seen as a potential, key cardiovascular target in pharmacology alongside ACE inhibitors and angiotensin receptor blockers (ARBs). It provided an additional pharmacological approach to blood pressure management through blocking its catabolism of atrial natriuretic peptides rather than acting primarily on the renin‐angiotensin system. Much financial investment was dedicated to the development and trials of dual NEP‐ACE vasopeptidase inhibitors such as omapatrilat (BMS‐186716; Robl et al., 1997). This investment rapidly ground to a halt following the reports of NEP actively degrading Aβ (Iwata et al., 2000) and hence providing the potential of NEP inhibitors to promote amyloid deposition and neuronal toxicity. Despite these concerns, in recent years, interest in this strategy has been revived leading to the development of sacubitril‐valsartan (LCZ696), a combined NEP inhibitor and ARB in the treatment of heart failure first approved in 2015. The dual cardioprotective action of the drug appears to have significant advantages in the management of heart failure (the PARADIGM‐HF Phase III trial; McMurray et al., 2014). NEP inhibition may also provide novel therapeutic benefits in corneal injury by promoting corneal wound healing (Genova, Meyer, Anderson, Harper, & Pieper, 2018). However, long‐term side effects of chronic NEP inhibition, such as a potential for promoting amyloid accumulation in the brain, and consequent dementia, as well as angioedema because of other physiological substrates of NEP, particularly bradykinin, still remain to be fully assessed (Campbell, 2017). Conversely, targeting the up‐regulation of NEP (or some other ADEs) to enhance Aβ metabolism nonselectively could lead to deleterious consequences in peripheral tissues, particularly in relation to cardiovascular events.

The NEP gene family comprises seven members, a number of which have also been implicated in the clearance of brain Aβ and therefore provide additional targets to reduce Aβ levels. These include NEP2, and the endothelin‐converting enzymes ECE‐1 and ECE‐2, although NEP is the primary rate‐limiting enzyme in Aβ metabolism among these and other metallopeptidases (Shirotani et al., 2001; for review, see Nalivaeva, Beckett, Belyaev, & Turner, 2012). The differential cellular and subcellular localizations of NEP, ECE‐1, and ECE‐2 suggest that they regulate distinct pools of Aβ with NEP able to degrade extracellular Aβ and ECE‐1 and ECE‐2 controlling intracellular pools in early endosomes and/or the endosomal/lysosomal compartments (Eckman et al., 2006; Pacheco‐Quinto & Eckman, 2013). At the cellular level, ECE‐1 is more broadly distributed in the brain than NEP or ECE‐2 (Barnes, Walkden, Wilkinson, & Turner, 1997; Matsas, Kenny, & Turner, 1986; Pacheco‐Quinto, Eckman, & Eckman, 2016), which are enriched in GABAergic neurons in the hippocampus and neocortex (Pacheco‐Quinto et al., 2016). NEP2 shows a different subcellular localization from NEP (Whyteside & Turner, 2008) and does appear to have a significant contribution to Aβ metabolism in vivo, at least from transgenic studies in AD mouse models (Marr & Hafez, 2014). Furthermore, NEP2 activity levels are reduced in mild cognitively impaired individuals and AD patients relative to nonimpaired subjects in AD‐susceptible brain regions (Huang, Hafez, James, Bennett, & Marr, 2012). NEP2 may even prove to be a more selective therapeutic target for up‐regulation of Aβ clearance than NEP itself since it has a more restricted substrate specificity (Whyteside & Turner, 2008). The contributions of ECE‐1 and ECE‐2 to amyloid clearance both in vitro and in animal models have been documented elsewhere, and, like NEP, they remain viable targets for up‐regulation genetically or pharmacologically to enhance amyloid clearance and improve cognition (Nalivaeva, Beckett, et al., 2012; Nalivaeva & Turner, 2017).

3.2. Insulin‐degrading enzyme (insulysin)

Like NEP, IDE is a zinc metallopeptidase but is distinct structurally and catalytically from the NEP family. As implied by its name, its main physiological role is in the metabolism of insulin but it has a broad substrate specificity. It exists principally in the cytosol but is also found in mitochondria and peroxisomes. The original observation that IDE was able to hydrolyse Aβ in vitro (Kurochkin & Goto, 1994; for review, see Kurochkin, Guarnera, & Berezovsky, 2018) was extended by Selkoe and colleagues to demonstrate that a microglial cell line secreted a metalloprotease immunologically identical to IDE, which was able to regulate extracellular levels of Aβ (Qiu et al., 1998). IDE also protects mitochondria against Aβ accumulation and dysfunction (Leal et al., 2013). Since that time, IDE has been regarded as a significant contributor to Aβ degradation both in vitro and in vivo. Given the growing links between AD development and diabetes, IDE provides a metabolic connection between the two diseases.

The lack of any apparent secretory signal in the IDE protein sequence has led to the suggestion that it may be secreted from cells through an unconventional secretory pathway (Zhao, Li, & Leissring, 2009), possibly via an exosomal or autophagy‐based route (Bulloj, Leal, Xu, Castaño, & Morelli, 2010). This process is stimulated by statins which might be a contributory factor in the proposed protection against AD development by statin treatment (Glebov & Walter, 2012). A recent study has, however, now called into question the significance of IDE secretion from cells suggesting that the presence of extracellular IDE is largely a reflection of a loss in cell integrity (Song, Rodgers, & Hersh, 2018), and hence, the relevance of IDE to extracellular peptide degradation in vivo may require some reevaluation. Nevertheless, the evidence for a significant involvement of IDE in Aβ clearance in vivo remains strong as does its potential as a therapeutic target in AD prevention. Indeed, IDE, even in a proteolytically inactive form, appears to inhibit Aβ fibrillogenesis through a chaperone‐like role (de Tullio et al., 2013).

3.3. The ACE family

The first genetic links between the ACE gene and AD came from reports that a common polymorphism in the ACE gene was associated with sporadic cases of AD (Kehoe et al., 1999). Further exploring the links between ACE and Aβ pathology, it was revealed that ACE could hydrolyse Aβ40 at several sites (Hu, Igarashi, Kamata, & Nakagawa, 2001) and convert Aβ42 to the less neurotoxic Aβ40, consistent with its predominant carboxydipeptidase specificity (Zou et al., 2007). However, the effects of treatments with an ACE inhibitor on mouse models of AD generally have not revealed any decreased clearance of Aβ in vivo, whereas the use of inhibitors of either NEP or ECE do show an accumulation of Aβ (Eckman et al., 2006). This is consistent with the much lower efficiency of hydrolysis of Aβ by ACE compared with NEP, ECE, or IDE.

Nevertheless, further evidence for an association of ACE with AD pathology has come from the demonstration of decreased levels of ACE and elevated Aβ42 in the CSF of Alzheimer patients compared with controls (Rocha et al., 2018). This supports a number of other studies implicating the renin‐angiotensin system in AD pathology, although the study represented a relatively small patient sample size. Nevertheless, it adds credence to the close links between vascular pathology and AD. There are no reports that chronic use of ACE inhibitors in patients for hypertensive treatment predisposes to cognitive deficits through accumulation of brain Aβ, although a recent study has shown increased plasma Aβ42 in cognitively‐impaired individuals taking ACE inhibitors (Regenold et al., 2017). Whether this reflects enhanced clearance of the peptide from the brain or also increased levels of brain Aβ is unclear. Indeed, a recent meta‐analysis has shown that blockade of the renin‐angiotensin system, whether with ACE inhibitors or ARBs, decreased the risks of cognitive decline on ageing or AD development (Zhuang, Wang, Wang, Li, & Xing, 2016). At present therefore, ACE remains an equivocal target as an AD therapeutic and further work is needed to resolve these issues.

The homologue of ACE, ACE2 (Tipnis et al., 2000), is also present in the brain where it plays a distinct role in CNS regulation of the cardiovascular system via the angiotensin‐(1‐7)/mas receptor pathway and protects the brain from ischaemic injury (Xia & Lazartigues, 2008). However, ACE2 displays a different specificity from ACE acting as a monocarboxypeptidase (Turner, Tipnis, Guy, Rice, & Hooper, 2002), but it does have activity on some amyloid peptides but not Aβ40 or Aβ42. In particular, it can convert Aβ43 to Aβ42, which can then be further metabolized by ACE to the less amyloidogenic Aβ40 (Liu et al., 2014). More recently, the ACE2 activator diminazene was shown to induce stimulation of the ACE2/angiotensin(1‐7)/mas axis reducing cognitive deficits in AD, most probably through activation of the PI3K/Akt pathway (Kamel et al., 2018). Furthermore, ACE2 appears to be reduced in the post‐mortem AD brain correlating with amyloid and tau pathology (Kehoe, Wong, Al Mulhim, Palmer, & Miners, 2016). The bifunctional roles of the ACE and NEP/ECE families as regulators of the angiotensin and endothelin cardiovascular axes and in amyloid clearance emphasize the interdependence of these two pathways and the close links between cerebral blood flow and neurodegeneration.

3.4. Other ADEs

In total, the number of reported ADEs is approaching 20, although for many of these, while they may be able to hydrolyse Aβ peptides in vitro, their physiological or pathological role in Aβ clearance or AD is much less strong than for those described above. However, they may well come into play in specific cellular or subcellular locations, or when the activity of other ADEs is compromised. The properties of many of those have been tabulated in (Nalivaeva, Beckett, et al., 2012). A large family of zinc metalloproteinases, some of which are able to degrade Aβ peptides, are the MMPs. In particular, MMP‐2 (gelatinase A) and MMP‐9 are able to degrade fibrillary forms of Aβ and may thereby contribute to plaque clearance from the brain (Roher et al., 1994; Yan et al., 2006). Levels of MMP‐9 are raised in the plasma of AD patients, perhaps as a compensatory mechanism (Lorenzl, Albers, Relkin, et al., 2003), and the enzyme may represent a therapeutic target in AD. Levels of the natural tissue inhibitors of metalloproteinases are reported to be increased in the AD brain, which might contribute to the decreased Aβ clearance that is seen in disease progression (Lorenzl, Albers, LeWitt, et al., 2003). Other classes of proteases may also make some contribution to Aβ metabolism, including the serine proteases plasmin and acyl peptide hydrolase, the lysosomal cysteine protease, cathepsin B, the zinc proteases aminopeptidase A, and glutamate carboxypeptidase II (see Nalivaeva, Beckett, et al., 2012 for critical review).

4. MANIPULATION OF ADEs: FROM GENE TRANSFER TO EPIGENETICS AND PHARMACOLOGY

Since up‐regulation of the expression and activity of ADEs, particularly NEP, represents an important therapeutic avenue in amyloid clearance, significant efforts have been directed over the years into a search for genetic, pharmacological, and other approaches to achieve these goals. Various in vivo animal models have shown the efficacy of NEP, and of some other ADEs, in lowering Aβ burden. For example, lentiviral gene transfer of NEP reduced amyloid pathology in mouse models of AD (Marr et al., 2003). More detailed studies using adeno‐associated virus vectors for intracranial gene transfer of either NEP or IDE, also in an AD mouse model (APP+PS1), revealed that expression of NEP, but not IDE, was able to reduce cerebral amyloid load over a prolonged period (Carty et al., 2013). Previous studies had also shown, using a similar strategy, that intracranial gene transfer of the NEP homologue, ECE‐1, produced significant reductions in amyloid load in the cortex and hippocampus (Carty et al., 2008). These and other studies suggest that such viral vectors encoding certain ADEs may be appropriate candidates for gene therapy trials in AD.

Since brain and peripheral Aβ levels maintain an equilibrium, and there is a gradient of Aβ from the brain into blood, attempts have been made to regulate brain Aβ levels by enhancing peripheral rather than central NEP activity. The feasibility of this “peripheral sink” hypothesis (Matsuoka et al., 2003) was demonstrated with NEP bound to erythrocytes or leukocytes (Guan et al., 2009). Subsequently, virally mediated up‐regulation of NEP in skeletal muscle was shown to reduce cerebral amyloid levels and associated brain pathology (Liu et al., 2009). Similarly, by using an adeno‐associated virus (AAV) vector expressing a soluble form of the enzyme, NEP secreted into plasma was able to bring about very significant reductions in plasma and brain Aβ, including of Aβ oligomers (Liu et al., 2010). It did not appear to affect levels of other physiologically important NEP substrates such as bradykinin and substance P. Although technically quite demanding, this approach, with further refinement, might provide an efficient strategy for controlling central Aβ levels. However, any such up‐regulation needs to be established from an early stage of disease development. Not all studies support the concept of the amyloid sink hypothesis. Henderson et al. (2014), for example, failed to detect any change in brain Aβ levels in Tg2576 AD transgenic mice, as well as in rats and monkeys, after i.v. dosing with an albumin‐fused form of NEP, although peripheral Aβ levels were significantly decreased in these animals. Translation of these virally mediated or other delivery strategies into practical clinical approaches remains problematic but not insuperable. Nevertheless, pharmacological methods for up‐regulation of Aβ‐degrading activity perhaps offer more feasible alternatives (see below).

An alternative expression strategy for up‐regulating NEP levels has been developed following the observation that the APP intracellular domain (AICD), which is produced alongside Aβ in the APP processing pathway, acts as a transcriptional regulator, one of whose targets is NEP. Hence, enhanced APP processing provides, via AICD, a feedback mechanism for regulating Aβ levels through increased expression and activity of NEP (Pardossi‐Piquard et al., 2005; see Figure 1). Our further studies have confirmed this phenomenon and demonstrated that AICD, produced via the amyloidogenic pathway, binds to the NEP gene promoter displacing repressor histone deacetylases (HDACs) and hence increasing enzyme expression and activity in an epigenetic mechanism that is neuron‐specific (Belyaev et al., 2010; Belyaev, Nalivaeva, Makova, & Turner, 2009; see Figure 1). Treatment of SH‐SY5Y cells expressing low NEP levels with the HDAC inhibitors trichostatin A or valproic acid likewise increased NEP mRNA and protein levels, and its activity (Belyaev et al., 2009). An independent study has also confirmed that valproic acid could reduce amyloid burden and improve cognitive deficits in transgenic AD mice although the authors did not analyse NEP expression in that study (Qing et al., 2008). However, valproic acid treatment has been shown to increase NEP expression and activity in hypoxic rats (Nalivaeva, Belyaev, et al., 2012) and AD model mice (Wang et al., 2014). These studies have suggested that HDAC inhibitors such as valproic acid, which is a widely used antiepileptic drug, might be repurposed for increasing levels of NEP and enhancing Aβ clearance in AD patients. Further impetus to this strategy has come from our observation that APP, via AICD, also up‐regulates expression of the Aβ transport protein TTR providing a dual regulation of Aβ clearance (Kerridge et al., 2014). Although valproic acid has clear benefits in treatment of epilepsy, its application in AD is not established from clinical studies to date. Given that valproic acid, being a nonselective HDAC inhibitor, regulates several hundred genes in the human, it might be that a more selective HDAC inhibitor is required for future clinical evaluation in AD requiring identification of which of the numerous HDAC activities are specifically responsible for NEP and TTR regulation. Such selectivity of action is important since HDACs can potentially have both neuroprotective and neurotoxic effects (Thomas & D'Mello, 2018).

Several independent studies have shown regulatory links between somatostatin (SST) receptor expression and NEP levels. Such a possible connection was first proposed by Saito and colleagues (Saito et al., 2005) based on the knowledge that NEP is the major degrading enzyme for somatostatin (Barnes, Doherty, & Turner, 1995) and whose content decreases in the brain with age, particularly in AD (Saito et al., 2005). Furthermore, the somatostatin degradation rate was previously shown to be altered in a region‐specific manner in AD‐affected brains (Ichai et al., 1994). Somatostatin agonists have been shown to up‐regulate NEP expression via a receptor‐mediated mechanism, and a synthetic agonist of SST₄ receptors, NNC 26‐9100, was shown to be effective in up‐regulating NEP expression and activity, reducing Aβ42 levels and improving learning and memory in an AD transgenic mouse model (Sandoval et al., 2013). Manipulation of somatostatin receptor levels can also be achieved by overexpression of postsynaptic density protein‐93 which up‐regulates the SST₄ receptor, and hence NEP, with subsequent reduction in brain Aβ levels and improvement of cognitive functions in the APP/PS1 mouse model of AD (Yu et al., 2017). Somatostatin is also a substrate for the dimeric IDE and can directly bind to one active site thereby enhancing the Aβ‐degrading capacity of the other subunit through allosteric regulation of its enzyme activity (Ciaccio et al., 2009).

The observations that certain receptor agonists can indirectly regulate expression of ADEs have led to exploring similar approaches for enzyme up‐regulation. For example, an agonist of PPAR‐δ (GW742) was found to up‐regulate the expression both of NEP and IDE in 5XFAD mice reducing amyloid plaque burden and inflammation (Kalinin, Richardson, & Feinstein, 2009). Treatment of SH‐SY5Y neuroblastoma cells overexpressing APP with the TK inhibitor Gleevec (imatinib) also resulted in AICD‐dependent up‐regulation of both NEP and TTR genes leading to decreased levels of total cellular Aβ (Kerridge et al., 2014), consistent with the prior observation of Bauer, Pardossi‐Piquard, Dunys, Roy, and Checler (2011) that augmentation of NEP upon Gleevec treatment was abolished by APP depletion. Following the report that the nuclear retinoid X receptor agonist and anticancer drug, bexarotene, could clear Aβ and reduce plaque levels in an AD model mouse (Cramer et al., 2012), we have demonstrated that it was also able to up‐regulate NEP and IDE expression in SH‐SY5Y cells overexpressing APP, although the mechanisms of their gene regulation were different (Nalivaeva, Belyaev, & Turner, 2016). Among endogenously produced compounds in the brain, several have been proposed to be involved in activation of ADEs. For example, some of the neuroprotective effects of kynurenic acid originating from tryptophan metabolism were shown to be related to NEP activation (Klein et al., 2013). The 5‐HT metabolite, 5‐hydroxyindoleacetic acid, also induces NEP in vivo and in neuronal cell lines and improves disease symptoms in an AD mouse model (Klein et al., 2018). There is also a report that NEP activity might be modulated by cAMP since, in human endothelial cells, activation of adenylate cyclase by forskolin or PGE₁ induced NEP protein expression and increased its enzyme activity (Graf et al., 1995), an observation that surprisingly has not subsequently been followed up.

Stem cell technology has also been applied to develop possible delivery methods for ADEs, especially NEP, following the report that human adipose tissue‐derived mesenchymal stem cells secrete exosomes that contain catalytically active NEP which could degrade Aβ when transferred into neuronal cell lines (Katsuda et al., 2013). This observation has been exploited to show that genetically modified neuronal stem cells stably expressing and secreting NEP could significantly decrease Aβ levels and related pathology both in Thy1‐APP and 3xTg‐AD transgenic mice. Not only was this effective in the regions where the stem cells were grafted (amygdala and septum) (Blurton‐Jones et al., 2014). Variants of these procedures have confirmed the viability of this stem cell approach in various animal models (Sharma et al., 2018).

The age‐related depletion in androgen and oestrogen levels has also been implicated in AD development and progression, which in turn has been correlated with an increase in Aβ levels. Testosterone and oestrogen supplementation also improve cognitive deficits in animal models of AD. This is consistent with increased expression and activity of NEP in the brain as a result of the presence of androgen‐ and oestrogen‐response elements in the NEP gene (Shen et al., 2000; Xiao, Sun, Liu, Zhang, & Huang, 2009). Nonsteroid modulators of the androgen receptor, as well as testosterone treatment, can also up‐regulate IDE levels, as can oestrogen receptor agonists (George, Petit, Gouras, Brundin, & Olsson, 2013). Hence, an up‐regulation of sex steroid levels in the brain may slow or ameliorate some of the symptoms of AD through enhanced clearance of Aβ peptides.

The provision of an enriched environment (EE) to ameliorate the symptoms of cognitive impairment and dementia has many advocates (see, e.g., Costa et al., 2007), and this procedure has also been linked with enhanced amyloid clearance. For example, EE was reported to reduce Aβ levels and amyloid deposition in an AD transgenic mouse model (Hu et al., 2010) and, in a separate study, NEP was among the numerous genes up‐regulated in the brains of wild type and AD mice kept under EE (Lazarov et al., 2005). EE has also shown beneficial effects in an ischaemia model in rats, which led to a significant increase in NEP (but not IDE) expression and reduced Aβ burden in the affected animal brains (Briones, Rogozinska, & Woods, 2009). Such effects were not replicated, however, in the more aggressive form of pathology seen in the 5XFAD mouse model (Hüttenrauch, Walter, Kaufmann, Weggen, & Wirths, 2017) suggesting, perhaps not surprisingly, that early intervention is important and EE is likely to be more beneficial in initial or milder forms of AD. This concept is backed up by a study comparing preventive (before disease onset) with therapeutic (after onset) EE in which different target genes were activated under the two protocols and only preventive EE up‐regulated IDE and somatostatin levels and hence NEP activity (Herring et al., 2011). Interpretation of these studies in terms of the efficacy of EE is no doubt complicated by the variety of animal models and EE protocols applied. However, this does not diminish the potential significance of this approach for maintaining and improving cognitive functions in ageing animals, elderly people, and AD patients.

Along with genetic and epigenetic approaches for up‐regulation of ADE gene expression, numerous attempts are being made to search for compounds in nature able to enhance the activity of ADEs. A number of natural products have been shown to elevate expression of NEP, of which the most explored has been the green tea flavonoid epigallocatechin 3‐gallate (EGCG; Melzig & Janka, 2003). In an animal model of AD, EGCG administration decreased Aβ accumulation and rescued cognitive decline in a mechanism that was abolished on silencing of NEP (Chang et al., 2015). Independently, EGCG has also been shown to increase NEP secretion from astrocytes, facilitating extracellular Aβ metabolism, in a mechanism involving activation of ERK and phosphoinositide 3‐kinase (Yamamoto et al., 2017). NEP secretion from astrocytes and extracellular Aβ clearance can also be enhanced by statin treatment again through an ERK‐regulated mechanism (Yamamoto et al., 2016). Although the mechanisms of EGCG action are not yet fully understood, its activity as an HDAC inhibitor might explain the up‐regulatory effect on NEP gene expression, which is down‐regulated by HDACs (Belyaev et al., 2009). Moreover, green tea supplementation was also shown to up‐regulate NEP protein expression and activity in the kidney and small intestine of genetically obese mice alongside body fat and weight loss. This supports an independent role for the enzyme in maintenance of healthy body weight since the effect was not seen in NEP‐deficient mice (Muenzner et al., 2016). In a screen of animal venoms for compounds activating the amyloid‐degrading ECE‐1, a peptide isolated from the pit viper (Bothrops asper) provided a novel and promising AD drug lead showing dual activity, binding to and stimulating allosterically both the ADEs, ECE‐1, and NEP (Smith et al., 2016).

Another class of compounds with some neuroprotective and antiamyloidogenic activities are the tetracycline antibiotics, such as minocycline (Ryu, Franciosi, Sattayaprasert, Kim, & Mc Larnon, 2004). Injections of minocycline were able to reverse the cognitive deficit seen in an AD transgenic mouse model (Seabrook, Jiang, Maier, & Lewere, 2006). Treatment with minocycline was also found to be protective against AD‐related pathology induced by i.c.v. injections to rats of a synthetic Aβ(25‐35) peptide. The mechanism appeared to involve, in part, up‐regulation of the levels of somatostatin and hence of NEP levels (Burgos‐Ramos, Puebla‐Jiménez, & Arilla‐Ferreiro, 2009). One cautionary note, however, in strategies to up‐regulate or directly enhance the activity of ADEs is that the broad specificity of some of these enzymes needs to be taken into account to minimize potential off‐target effects, especially on the cardiovascular system.

5. FACTORS PREDISPOSING TO AD

Apart from genetic background and advanced age, various clinical conditions and life‐style factors can predispose individuals to development of dementia and AD at earlier stages in life. There are well documented links between ischaemic stroke and development of late‐onset AD pathology in surviving individuals (Zhou et al., 2015). Both chronic ischaemia and stroke lead to increased Aβ levels in brain vasculature and in blood plasma (Lee et al., 2005). Accumulation of Aβ in brain vasculature, in turn, can further impair cerebral blood flow leading to microvascular damage and accelerating AD pathology (Stopa et al., 2008).

The data on ADEs in the brain affected by ischaemia and stroke to date are rather limited and contradictory. Using an experimental model of ischaemia in rats, we have shown a reduction of ECE and NEP protein levels in the cortical tissue of the affected hemisphere (Nalivaeva et al., 2004). Reduced IDE activity was also found in patients with cerebral amyloid angiopathy (Morelli et al., 2004). Cellular and animal data confirm that acute and chronic hypoxia significantly affect NEP expression and activity. Our earlier data clearly demonstrate that NEP expression is down‐regulated by hypoxia in neurons while activated in astrocytes (Fisk, Nalivaeva, Boyle, Peers, & Turner, 2007). This reduction of NEP expression was found to be related to increased degradation of AICD by hypoxia‐activated caspases (Kerridge, Kozlova, Nalivaeva, & Turner, 2015). Decreased levels of NEP in the cortex and hippocampus were found in rats subjected to prenatal hypoxia, which correlated with their cognitive deficits (Nalivaeva, Belyaev, et al., 2012). Treatment of hypoxia in NB7 neuroblastoma cells or in rats with the caspase inhibitors Z‐DEVD‐FMK or Ac‐DEVD‐CHO restored AICD and NEP levels and also resulted in improved animal performance of cognitive tasks (Kerridge et al., 2015; Kozlova et al., 2015). In AD transgenic mice, prenatal hypoxia was also shown to accelerate accumulation of amyloid plaques and reduce NEP expression via increasing levels of HDAC1 and decreasing histone H3 acetylation in the NEP promoter region (Wang et al., 2014). One of the reasons why ADEs are affected by ischaemia and hypoxia is that they are very susceptible to oxidative stress producing a downward AD spiral (Nalivaeva & Turner, 2017; Shinall, Song, & Hersh, 2005).

Diabetes is also closely linked to AD pathogenesis leading to the terminology of type 3 diabetes because of their interactions (Steen et al., 2005) not least because of the key role played by IDE in both diseases, hydrolyzing insulin and Aβ peptides. The expression and activity of various ADEs are modified in the diabetic state. For example, increased Aβ levels combined with reduced NEP expression were found in the hippocampus of streptozotocin‐induced type 1 diabetic monkeys while no changes in other ADEs were reported (Morales‐Corraliza et al., 2016). However, in streptozotocin‐treated rats, we have also demonstrated a decrease in IDE expression in the cortex and hippocampus (Kochkina et al., 2015).

NEP activity has been linked with obesity, insulin resistance, and the metabolic syndrome in a human patient study and in genetically obese mice (Standeven et al., 2011). Epidemiological studies also suggest that obesity might predispose to late‐onset AD, which correlates with the observation that leptin significantly decreases NEP (but not IDE) expression in cultures of rat astrocytes, possibly via activation of the ERK cascade, leading to decreased Aβ cleavage (Yamamoto et al., 2014).

Given that impaired brain circulation and AD often coexist, it is important to understand how these two pathological processes may interact. By examining hypoperfusion of the brain in an AD animal model (Yang et al., 2018), an elevated content of Aβ and its oligomeric forms has been observed, which correlated with increased β‐secretase levels and a decrease of the α‐secretase. In this model, decreased expression and levels of NEP and of LRP‐1 were also seen. Hence, brain hypoperfusion can amplify the effects of AD on steady‐state levels of Aβ through its combined effects on Aβ synthesis, transport, and clearance (Yang et al., 2018). Chronic cerebral hypoperfusion can result from atrial fibrillation (AF) which is a major risk factor for stroke and is, in turn, strongly linked with the incidence of vascular dementia. Uncontrolled AF also roughly doubles the risk of developing subsequent AD (Dublin et al., 2011). These correlations can be explained through the observed effects of hypoperfusion on Aβ levels and emphasize the importance to brain health of maintaining normal cardiac rhythm and protecting against the effects of AF through the use of anticoagulant therapy. However, some of the newer classes of anticoagulants may themselves partially inhibit the clearance of plasma Aβ, through their inhibition of coagulation proteases (Yang, Bhattacharya, Li, & Zhang, 2017), complicating the interpretation of these data.

6. SLEEP, AMYLOID CLEARANCE, AND NEURODEGENERATION

Disturbances of sleep patterns are commonly seen in AD patients even at early stages of disease progression (Moran et al., 2005). This includes patterns of diminished rapid eye movement (REM) sleep as well as reduced quality and quantity of sleep (Ju et al., 2013). In particular, highly fragmented sleep increases the risk of developing AD (Lim, Kowgier, Yu, Buchman, & Bennett, 2013). While such changes might be indicators of disease progression, they may in themselves be risk factors for AD development. A number of studies have now, for example, convincingly shown that Aβ dynamics reveal distinct variations during the sleep–wake cycle. In both mouse and human studies, Aβ levels can show diurnal variations of up to one‐third with lowest levels during sleep and rising during the awake period, in part regulated by the neuropeptide orexin (Kang et al., 2009). In a seminal study examining metabolite, especially Aβ, clearance from the adult mouse brain, it was concluded that sleep actively promotes this process, in part through activation of the glymphatic system (Xie et al., 2013). This led to the conclusion that a function of sleep physiology is to stimulate removal of metabolite “waste products” such as amyloid accumulating during the wake phase. In this study, radio‐iodinated Aβ40 was removed twice as quickly in sleeping compared with awake mice, as was an inert tracer, inulin. Short‐term total sleep deprivation in humans also leads to an increase in peripheral levels of Aβ which correlates with decreases in the levels of the plasma Aβ clearance proteins soluble LRP‐1 and soluble advanced glycation end product specific receptor (RAGE; Wei et al., 2017). As little as one night of sleep deprivation can lead to significant Aβ accumulation (Shokri‐Kojori et al., 2018). Even in a Drosophila AD model, deprivation of sleep leads to accumulation of Aβ via changes in neuronal hyperexcitability, and overexpression of Aβ causes reduction and fragmentation of sleep in the flies (Tabuchi et al., 2015).

While sleep length may provide some neuroprotective effect through promoting clearance of amyloid, a very recent meta‐analysis of 74 studies has revealed that too much sleep (>8 hr) also can have adverse health implications, particularly in relation to cardiovascular events (Kwok et al., 2018). These studies support the recommendations of the US National Sleep Foundation that 7–8 hr of sleep is optimal for those 65 years and over. For both neurodegeneration and cardiovascular disease, it is perhaps sleep quality that is the primary factor. Moderation in all things is the key here and provides support to the concept that a variety of environmental factors, for example, mental activity, dietary regime, sleep and physical exercise, and particularly general promotion of cardiovascular health, is also protective to some degree in relation to the progression of neurodegenerative disease (Tariq & Barber, 2018). In most cases, underlying molecular mechanisms for these possible protective measures are not known, although exercise may be protective against AD by up‐regulating brain‐derived neurotrophic factor (BDNF) and sAPPα, hence reducing Aβ production and thereby promoting neuronal survival, plasticity, and cognitive function (Nigam et al., 2017).

Tau pathology is also likely to influence the sleep abnormalities observed in dementias since mouse models of such pathologies also show sleep disturbances not unlike those seen in AD models (Koss et al., 2016), and some tauopathies also elicit similar changes in sleep patterns (Bonakis et al., 2014). Sleep behaviour disorders are also an early feature of synucleinopathies (Högl, Stefani, & Videnovic, 2018).

7. CONCLUSIONS AND FUTURE PERSPECTIVES

More than a quarter of a century since the first identification of specific mutations in the APP gene leading to early‐onset AD (Chartier‐Harlin et al., 1991), there is still no targeted therapeutic based on the amyloid cascade hypothesis in the clinics despite much initial optimism. While some have questioned the validity of the hypothesis as a consequence, the overwhelming bulk of evidence links Aβ accumulation, as a result either of its overproduction or its impaired clearance, or both, in the early and late phases of the disease and in the subsequent cell death and cognitive impairment. However, AD is a heterogeneous disease with multiple interacting factors and timescales. Hence, the need to have a diverse and balanced strategy in the management of AD of which controlling the steady‐state level of Aβ is a requirement. But we need to look beyond amyloid and also address phospho‐tau levels, inflammatory processes, and synaptic loss as well as the normal physiological relevance of APP and Aβ. There will be no single magic bullet to slow or reverse the devastating neuronal loss seen in AD. A critical element of any strategy will be, however, to enable early diagnosis through validation of biomarkers, to enhance clearance of the accumulating toxic proteins, and to identify pharmacological agents that are able to block the neurotoxic effects of Aβ replicating the neuroprotection afforded to NDAN individuals.

7.1. Nomenclature of targets and ligands

Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to Pharmacology (Harding et al., 2018), and are permanently archived in the Concise Guide to PHARMACOLOGY 2017/18 (Alexander, Christopoulos et al., 2017; Alexander, Cidlowski et al., 2017; Alexander, Fabbro et al., 2017; Alexander, Kelly et al., 2017).

CONFLICT OF INTEREST

The authors declare no conflicts of interest.

Nalivaeva NN, Turner AJ. Targeting amyloid clearance in Alzheimer's disease as a therapeutic strategy. Br J Pharmacol. 2019;176:3447–3463. 10.1111/bph.14593