APP/A! pathway
C. L. Masters, K. Beyreuther
INTRODUCTION
The ultimate goal of current research into Alzheimer’s disease (AD) and related disorders of the ageing nervous system is to develop a disease-modifying therapeutic strategy. In numer- ical terms, the current emphasis is on the biochemical and molecular genetic pathways sur- rounding the amyloid !(A!) and the amyloid precursor protein (APP). In 2005 alone, more than 780 papers appeared on APP/A!, many of which deal with pre-clinical and clinical ther- apeutic strategies directed at the APP/A!pathway (Figure 8.1). In this chapter, we survey the 2005/2006 literature, which includes several useful reviews [1–8]. We have also reviewed the two public databases of clinical trials which enumerate the current status of clinical activity in this area (Novatis International, maintained by Kwon and Herrling [9] and the clinical trials registry sponsored by the National Institutes of Health (NIH) [www.clinicaltrials.gov]).
Why is there so much interest in the APP/A!pathway? The theory which underlies this pathway as the principal and proximal causal mechanism in AD is pinned to two critical series of observations: first, mutations in the gene encoding APP (and the presenilin [PS]
genes as components of the "-secretase machinery) are causally linked to early onset famil- ial AD; second, genetically engineered mice with these mutations recapitulate the human disease. More recently, a very tight association between the mean age at onset of pedigrees with PS mutation-related familial AD and the ratio of secreted A!40to A!42 has emerged [10]. This, together with the development of a robust A!-neuroimaging ligand (a thioflovin T analogue) which as a biomarker clearly differentiates AD and mild cognitive impairment (MCI) from normal controls and other neurologic diseases [11, 12], adds much more strength to the A!theory. But the single most important challenge to test the theory remains the demonstration that a drug targeting the APP/A!pathway actually modifies the natural his- tory of the disease. To this end, the criteria set out by Cummings [13], and listed in Table 8.1, have clarified the standards to be met when we come to assess this test of the A!theory of AD. The first criterion (a plausible mechanism of action in a validated model) has been achieved by many of the therapeutic strategies reviewed below. But no drug has yet met any of the other four criteria, although we remain optimistic that the current pace of activ- ity will deliver a result in the not too distant future.
Colin L. Masters, MD, Department of Pathology, The University of Melbourne and The Mental Health Research Institute of Victoria, Australia
Konrad Beyreuther, PhD, Centre for Molecular Biology Heidelberg, Zentrum für Molekulare Biologie Heidelberg, Heidelberg, Germany
©Atlas Medical Publishing Ltd 2007
UPSTREAM EVENTS IN THE APP/A!PATHWAY
The targets derived from the APP/A!pathway outlined in Figure 8.1 are listed in more detail in Table 8.2. While it is not a comprehensive or exhaustive listing, it does present a novel and logical way of classifying the wide range of current research activity being under- taken in this area.
AGE AND ENVIRONMENTAL FACTORS
Of all the external variables which determine risk of getting AD, age and the environment stand out as factors which demand explanations. Yet for all their obviousness, no reasonable
Upstream events Environment
(diet, exercise)
Neurotransmitter systems
Other neuronal functions
APP synthesis, transport, interaction
A! conformations, interactions, modifications
A! toxicity Microglial reaction (inflammation)
A! clearance and degradation Tau aggregation
and neuritic damage
ApoE
Downstream events A! biogenesis from APP
Central steps
Figure 8.1 Schematic outline of the upstream and downstream events which surround the central APP/A!pathway.
1. Plausible mechanism of action in a validated model.
2. Clinical trial evidence based on the Lerber staggered start design.
3. Difference in survival to a meaningful clinical outcome.
4. Change in rate (slope) of decline.
5. Demonstrable drug–placebo difference on an accepted biomarker of disease progression.
Table 8.1 Alzheimer’s disease modification by drug intervention: criteria (modified with permission from Cummings [13]).
explanations have been forthcoming. While many of the biochemical events listed in the APP/A!pathway are known to be developmentally regulated, very little information is yet available on what happens under normal ageing conditions. Partial loss of function of a crit- ical biochemical reaction would seem to be a good starting point for investigation, either as an upstream or downstream event; or a ‘double hit’ phenomenon could be invoked, as seen in the early development of ideas on oncogenesis. Whichever, the incontrovertible link between ageing and AD remains obscure in mechanistic terms.
Similarly, the interactions between the environment and the risk for AD have attracted many epidemiological studies. Diet and exercise remain as the two most interesting vari- ables. General caloric restriction has often been associated with longevity in rodent models of ageing, and recent studies in transgenic AD models [14, 15] and normal rodents [16] sug- gest an effect on A!plaque load or #-secretase processing of APP. The effects of exercise [17]
Pathway step/event Target/drug
Upstream events Ageing
Environment Exercise, diet
Neurotransmitter systems modulation Cholinergic: AChE/BuChE inhibition Glutamatergic: NMDA antagonism Serotonergic: Anti-depressants Other cerebral or general systemic factors
Central steps: A!biogenesis from APP APP gene target
APP interactions, transport X11 gene target, growth factors, metal homeostasis, ZnT3, oestrogen, cholesterol APP proteolytic processing
"-secretase (PS, Aph1, PEN2, Nct) Inhibitor, modulator (NSAID),
gene target and associated $- and %-cleavages
!-secretase (BACE 1) Inhibitor, modulator/interaction (PAR 4),
immunomodulations of !-cleavage site, gene target
#-secretase Stimulation (PKC activation)
A!and its varied conformations Monomers/dimer/trimer (A4, A8, A12)
Metal binding sites Metal–protein attenuating compounds (MPAC) GAG binding sites
Oxidative modifications Di-tyrosine, methionine oxidation
A!–lipid interactions Lipid–protein attenuating compounds (LPAC) A!–protein interactions Protein–protein attenuating compounds (PPAC)
!-oligomers/protofibrils Anti-aggregants/dis-aggregants Polymers/fibrils Anti-fibrillogenics/de-fibrillants Downstream events
A!-induced ‘toxicity’ through oxidative Anti-oxidants, natural product screens, oestrogen damage (protein, mitochondria, lipid,
sterols, nucleic acid, etc.)
Anti-inflammatory (anti-microglia) Sigma 1 receptor, PPAR-", PGE2, NSAID, iNOS inhibition
A!–tau direct/indirect interactions Microtubule stabilizers, kinase, inhibitors, anti-aggregants, etc.
A!–ApoE interactions Statins/cholesterol
A!–clearance/neutralization Immunization, immunomodulation
A!-degradation IDE, NPE, ACE
Table 8.2 Alzheimer’s disease therapeutic targets derived from the APP/A!pathway
and environmental enrichment [18, 19] have also been examined in transgenic AD models with encouraging results. One group [19] found a change in a downstream event, an increase in the enzymatic activity of neprilysin, an A!-degrading protease, in response to environmental enrichment. These downstream events are discussed in the section
‘Modulating the A!Degradation Pathway’.
Specific dietary intakes, especially naturally occurring anti-oxidants (see section:
‘Ameliorating the toxic gain-of-function of A!: anti-oxidants, neuroprotectants, and other products of natural origin’) or metal ions (see section: ‘Monomers (A4), dimers (A8) and trimers (A12)’) remain largely under-investigated as AD risk factors. As methods for diagno- sis and population-based screening improve (using plasma biomarkers or specific ligands of A!for neuroimaging), it will become more feasible to examine analytically the dietary risk profiles of discrete populations, overcoming current limitations on sensitivity and specificity of case-ascertainment. A surprising study has already pre-empted dietary modulation of AD through the consumption of transgenic A!-expressing potatoes [20]! The proposed mech- anism involves low-level immune-mediated clearance of A!deposits (see section: ‘Using immunization and immunomodulation of A!to promote clearance and inhibit toxicity (neu- tralization)’). One wonders where this approach might lead – perhaps the production of transgenic A!-over expressing beef or lamb stock will appear on future menus?
THE EFFECT OF MODULATION OF NEUROTRANSMITTER SYSTEMS ON APP PROCESSING
Acetylcholinesterase (AChE) was discovered to be present in AD amyloid plaques 40 years ago, and the activity of choline acetyl transferase (CAT) was found to be decreased in the AD brain 30 years ago. From these observations the cholinergic hypothesis/theory of AD arose, which led to the development of AChE inhibitors (AChEI) as a therapeutic strategy, with apparent success, despite the lack of any plausible explanation for the presence of AChE in plaques and the underlying loss of CAT. A paradox then emerged: subjects treated with AChEI responded with a compensatory increase in AChE levels. This might have been expected to negate the intended effect of the AChEI on the availability of ACh for choliner- gic transmission. At the same time, clinical trials of AChEIs and their meta-analyses contin- ued to show favourable, albeit mild, effects on cognitive parameters, at least during the first 6–12 months of treatment. Against this background, basic and clinical investigators have recently turned their attention towards other possible mechanisms of action of the AChEIs, especially on the APP/A!pathway, and have begun to ask whether these drugs might have any disease-modifying effects [21].
Various aspects of AChEI actions on the upstream and downstream APP/A!pathway have been reported: attenuating the effects of A!-induced neuronal cytoxicity [22], promot- ing #-secretase or decreasing !-secretase activity [21, 23], inhibiting A!aggregation [24, 25]
or inhibiting GSK 3!activity and tau phosphorylation [21]. One group found no effect on A!amyloid plaque load while still improving behavioural deficits in a transgenic mouse model [26], while another group found that inhibitors of butyrylcholinesterase had a lower- ing effect on cellular APP and A!and brain A!in transgenic mice [27].
The modulation of glutamatergic transmission in AD has also received increasing attention with the results of the memantine clinical trials aimed at blocking (non-competitively) the action of N-methyl-D-aspartase (NMDA) receptors. With the growing awareness that the toxic soluble oligomers of A!may inhibit long-term potentiation (LTP) at the pre-synaptic level and that A!promotes the endocytosis of the NMDA receptor (mediated in part through #7 nico- tinic receptor, protein phosphatase PP2B, and tyrosine phosphase STEP [28]), the finding that memantine has beneficial behavioural effects in both A!toxicity models [29] and APP trans- genic mouse models [30] requires further work which might tie all these observations together.
Finally, behavioural intervention with antidepressants has also been explored in relation to in vitroAPP processing [31]. It would seem less likely that the tricyclics or serotonin reuptake
inhibitors will ever be subjected to AD-modification trials, but if further pre-clinical studies emerge showing effects on APP processing, then an argument could be made for additional clinical studies in the early phases of AD.
OTHER CEREBRAL OR GENERAL SYSTEMIC FACTORS
One suspects that there will be many other upstream factors which play into the APP/A!
pathway, but few have been identified to date. A particularly contentious area has been the role of the vascular supply to the brain and the effects of ischaemia (atherosclerosis) and hypertension. Historically, this has deep roots, going back to the days when ‘arteriosclero- sis’ was thought to cause all forms of dementia. Similarly, head trauma has been considered as a risk factor for AD, and APP has been identified as a sensitive marker of axonal damage following traumatic brain injury. But neither hypoxia nor trauma has yet been shown to be major risk factors for AD, and neither has been shown to promote the long-term amyloido- genic processing of APP.
CENTRAL STEPS IN THE APP/A!PATHWAY
TARGETING THE APP GENE OR GENES WITH PRODUCTS INTERACTING DIRECTLY WITH APP With the advent of RNA interference (RNAi) silencing, it is to be expected that attempts at direct APP gene regulation will emerge. As a forerunner to this, models in which the over- expressed human APP transgene in mice can be downregulated with doxycline provide a proof-of-principle that rapid control over A!expression and deposition can be obtained without gross adverse side-effects [32]. Unexpectedly, A!deposits formed before the onset of downregulation seemed to be remarkably stable, indicating that any treatment of this type in isolation might have to be administered early in the natural history of AD. Using RNAi tech- niques in transfected cell lines [33], targeting the X11 gene (APAB) successfully increased APP C-terminal fragments and lowered A!levels; X11 is a known interactor with the cytoplasmic domain of APP, and presents a novel method of possibly modulating "-secretase cleavage.
APP-INTERACTING SYSTEMS
As a presumptive cell surface receptor, APP probably has ligands and effector mechanisms for signal transduction. Nearly 200 proteins have been reported as having direct inter- actions with APP. Suspected ligands in the extracellular domain include growth factors (nerve growth factor [NGF] in particular), heparin-containing extracellular matrix, metals (through the extracellular Cu/Zn binding domain) and APP itself through hetero- and homo-dimerization. Small compounds such as propentofyline [34] can affect NGF release, and through this modulate the amyloidogenic pathway. Other small compounds may bind directly to APP [35] and affect its processing.
A controversial area involves the effects of hormones (oestrogens and testosterone espe- cially) and how they may affect APP metabolism. Conflicting results in experimental mod- els have appeared, in which oestrogen deficiency exacerbates A!in the APP23 transgenic model [36] and neither oestrogen deprivation nor replacement affected A!deposition in the PDAPP transgenic model [37]. Further studies are clearly required to unravel this important area where there is an epidemiological impression that females have a higher incidence of AD than males (this impression does not appear to have ever been subjected to a prospec- tive analytical epidemiological study). The mechanisms through which oestrogen/testos- terone might act remain obscure, but include oestrogen-dependent regulation of metal homeostasis in the brain through the expression of the neuronal zinc-transporter, ZnT3 (see also chapter 12).
Cholesterol and inhibitors of cholesterol synthesis (statins) have been shown to signifi- cantly alter APP processing in vitro, with a reduction in !-secretase cleavage and lessened A!production. While some early phase clinical trials with stains have shown encouraging results [38], others have not [39, 40]. Cholesterol-independent effects have also been noted for statins acting on isoprenyl intermediates in the cholesterol biosynthetic pathways, with a putative anti-inflammatory effect induced by reactive microglia [41, 42]. This might con- flict with the current theory that microglia are involved in the beneficial process of clearing A!deposits (see sections: ‘Suppressing Brain “inflammation” ’ and ‘Using Immunization and Immunomodulation of A!to Promote Clearance and Inhibit Toxicity (Neutralization)’).
Statins also have been implicated in the toxic gain-of-function of A!interacting with
#7-nicotinic AChR [43], although the mechanism for this remains unclear.
If eventually cholesterol does prove to be a risk factor for AD, then the observations [44]
of an association between AD and the expression levels and haplotypes of the 5®ion of the cholesterol 25-hydroxylase (CH25H) gene on chromosome 10 may provide a plausible explanation: one in which cerebral cholesterol metabolism (as distinct from systemic chol- esterol and its association with atherosclerosis) directly plays into the APP processing and transport pathways.
APP PROTEOLYTIC PROCESSING
The biogenesis of A!has been the prime validated drug target for AD since the discovery of the proteolytic processing of APP in 1987 (providing the fertile ground for nearly 20 years of intensive research). Molecular details of the C-terminal cleavage ("-secretase) were the first to emerge, followed by the #- and !-cleavage mechanisms. Subsequent elucidation of '-, $-, and %-cleavages has added another layer of complexity. Drug discovery programs reflect this sequence of events: many large pharmaceutical companies have "-secretase inhibitors or modulators in clinical development, while the !-secretase inhibitors are several years behind, largely in pre-clinical discovery.
g-secretase inhibitors and modulators
During 2005, the first publications of in vivo"-secretase inhibition/modulation of A!42bio- genesis appeared. One of the first known inhibitors (DAPT) was shown to be effective in acute experiments in behavioural tests (contextual fear conditioning) in the Tg 2576 AD mouse model [45]. Modifications to the chemical structure of DAPT has now improved its delivery to the brain [46], as with other compounds [47], in the hope of achieving lower effective dosages minimizing the risk of adverse peripheral effects. Many diverse classes of inhibitors and mod- ulators are showing very favourable acute pharmacokinetics, with rapid lowering of plasma and CSF A!levels [48–53]. Importantly, there is now strong evidence linking plasma and cere- brospinal fluid (CSF) A!levels, indicating that the brain/CSF pool of A!is at least in part a significant proportion of the plasma A!pool. There are still methodological issues in measur- ing A!, using either enzyme-linked immunosorbent assay (ELISA) or Western blotting tech- niques (which soluble oligomeric species are being measured, and what forms of A!: total, A!40, A!42?). Nevertheless, these preliminary data offer some hope that plasma A!species may eventually prove to be a reliable marker of cerebral A!turnover. Further explorations of the properties of "-secretase inhibitors are revealing unanticipated effects on synaptic function [54]. New classes of "-secretase inhibitors/modulators continue to be disclosed [55–57], as part of the effect to develop compounds devoid of side-effects. The major concern is the inhi- bition of signalling in the Notch pathway, which affects cellular differentiation [58, 59].
Ironically, "-secretase inhibitor compounds originally developed for AD are now being trialled in phase II studies of acute lymphoblastic leukaemia (NCT00100152-Clinical Trials.gov [60]) and advanced breast cancer (NCT00106145-Clinical Trials.gov [61]).
The first in-human phase I results to be published [62, 63] have shown that the Lilly com- pound LY450139 achieved a significant lowering of plasma A!, but not CSF A!, in normal volunteers (up to 50 mg/day for 14 days) or subjects with AD (up to 40 mg/day for 6 weeks). The drug was well-tolerated. Higher dosages may be required to achieve a reduction in CSF levels. The results of phase II studies with readouts on cognitive variables are eagerly awaited. In the meantime, further research on the mechanistic operations of the
"-secretase complex [64] may lead to new paths of drug discovery, as might gene target- ing of PS, PEN-2, APH-1, and nicastrin lead to selective regulation of "-secretase activity [65, 66].
b-secretase (BACE) inhibitors
Although approximately 5 years behind the development of the "-secretase inhibitors, much progress has been made in the discovery and design of compounds which target the active site of BACE-1. Improved assays [67] and structural-based in silico designs [68–72] have added to the existing pipe-line of drugs in early pre-clinical development [73–76] or early discovery programs [77, 78]. Other proteins interacting with BACE-1 may become drug tar- gets [79], and gene targeting of BACE-1 mRNA using siRNA is also producing encourage- ing preliminary results [80]. As with "-secretase, unanticipated side-effects on other BACE-1 substrates or downstream consequences of BACE-1 inhibition may prove difficult to cir- cumvent. As a consolation, inhibitors of BACE-1 may also turn out to have anti-angiogenic and anti-neoplastic activities [81].
DRUGS TARGETING AbAND ITS VARIED CONFORMATIONS Monomers (A4), dimers (A8) and trimers (A12)
In contrast to the inhibition of A!biogenesis, therapeutic strategies which directly target A!
itself should inherently have a lower risk of throwing up unanticipated side-effects, as the accumulated A!molecule is restricted to AD. If the A!fragment (or its domain within APP) does, however, subserve some critical normal function, then targeting A!itself might inter- fere with this function and thereby lead to adverse side-effects, but to date, a normal func- tion for A!has not been identified. APP knockout mice are viable and healthy, providing some support for this idea.
Current models of the physical state of A!are evolving. Whilst resident in the mem- brane, A!is assumed to be in an #-helical conformation. Following sequential !- and
"-cleavages, A!as a monomer (A4), dimer (A8) [or perhaps even as a trimer (A12)] is trans- located into the extra-cytosolic space, and may transition there into a !-strand enriched structure. These structures may then progress towards !-oligomers/protofibrils through to polymers/fibrils of amyloid filaments.
The mechanisms through which A!causes damage to neurones (‘the toxic gain of func- tion’) are slowly emerging. There are many theories: the two most favoured include the abil- ity of A!to generate oxidative stress and the hydrophobic interaction of A!with lipid membranes, particularly the synaptic plasma membrane. Our current working model incorporates both theories: we have defined a metal binding domain near the N-terminus of A!which is capable of binding Zn2((which causes A!to precipitate) or redox-active Cu2(. When Cu2(binds A!, it not only causes a significant increase in insolubility, but induces a series of electron transfers which result in histidine bridge formation, tyrosine 10 radical- ization, di-tyrosine cross-linking and oxidation of methionine 35. Ultimately, in the presence of reductants, this results in the production of H2O2and hydroxyl radicals, capable of inflict- ing short-range oxidative damage to proteins, lipids, sterols, nucleic acids, etc. Our studies show that toxicity to neurones in culture is associated with the ability of A!to associate with the lipid head-group on the outer surface of the plasma membrane.