|
 |
1 of 5 |  |
 |
 |
| Pharmacology & Therapeutics | ||
| Volume 86, Issue 2 | SummaryPlus | |
| May 2000 |  |
Article |
| Pages 111-144 | Journal Format-PDF (370 K) |
PII: S0163-7258(00)00036-X
Copyright (c)
2000 Elsevier Science Inc. All rights reserved.
Review article
Factors influencing the processing and function of the amyloid Christine M. Coughlana
and Kieran C. Breen
precursor proteinøøa potential
therapeutic target in Alzheimer's disease?
, 
, b
a Department of Pathology
and Laboratory Medicine, University of Pennsylvania Medical School, 34th and
Civic Center Boulevard, Philadelphia, PA 19104, USA
b Dundee Alzheimer's Disease Research Centre,
Department of Pharmacology and Neuroscience, University of Dundee, Ninewells
Hospital Medical School, Dundee DD1 9SY, UK
Available online 2
May 2000.
The amyloid precursor protein
(APP), which plays a pivotal role
in Alzheimer's disease (AD), can exist as either a membrane-bound or soluble
protein. The former is cleaved at the level of the plasma membrane to generate
the soluble form of the protein (APPs). An alternative pathway exists,
however, for the cleavage of APP
to generate a 40ø42 amino acid peptide termed amyloid (A), either within the lysosomal or the
endoplasmic reticulum/Golgi compartments of the cell. In AD, there is an
increase in the ratio of the 42 amino acid form of the A peptide (A42) to A40. The A42 form is the more amyloidogenic
form and has an increased potential to form the insoluble amyloid deposits
characteristic of AD pathology. Studies on the familial form of the disease,
with mutations in APP or in the
presenilin proteins, have confirmed an increase in A42 generation associated with the
early stages of the disease. This review will examine the factors that influence
APP processing, how they may act
to modulate the biological effects of APPs and A, and if they provide a viable target for
therapeutic intervention to modify the rate of progression of the disease.
Author Keywords: Alzheimer's disease; Amyloid; Presenilins; Apolipoprotein E; Inflammation; Calcium
Abbreviations: 
2M, 2-macroglobulin protein; APP, amyloid precursor protein; AD, Alzheimer's disease;
APLP, amyloid precursor-like protein; ApoE, apolipoprotein E; CAM, cell adhesion
molecule; cAMP, cyclic AMP; CCV, clathrin-coated vesicle; cGMP, cyclic GMP; CTF,
C-terminal fragment; DAG, diacylglycerol; DS, Down's syndrome; EGF, epidermal
growth factor; E/L, endosomal/lysosomal; ER, endoplasmic reticulum; ERAB,
endoplasmic reticulum amyloid
peptide-binding protein; ERK, extracellular signal-regulated kinase; FAD,
familial Alzheimer's disease; HSPG, heparan sulfate proteoglycan; IL,
interleukin; KPI, Kunitz protease inhibitor; LDL, low-density lipoprotein; LRP,
low-density lipoprotein receptor-related protein; LTP, long-term potentiation;
MAPK, mitogen-activated protein kinase; mAChR, muscarinic acetylcholine
receptor; NF
B, nuclear factor
B; NGF, nerve growth factor; PI,
phosphatidylinositol; PKA, protein kinase A; PKC, protein kinase C; PL,
phospholipase; PS, presenilin; RAGE, receptor for advanced glycation end
products; ROS, reactive oxygen species; TGF, transforming growth factor; TGN,
trans-Golgi network; UV-DDP, UV-damaged DNA-binding protein
precursor protein
precursor protein
interaction with membrane-bound proteins precursor protein
processing
precursor protein
precursor protein processing
function
aggregation
toxicity
and the
cytoskeleton
toxicity
2-Macroglobulin
Contents
Introduction 112
Amyloid precursor protein function 112
Cell adhesion 113
Neurotrophic actions of amyloid precursor protein 115
Amyloid precursor protein
interaction with membrane-bound proteins 116
Amyloid precursor protein
processing 117
Intracellular transport of amyloid precursor protein 118
Caveolae and cholesterol 118
The role of membrane-bound proteins in amyloid precursor protein processing 120
Presenilins 121
Protein kinase C 121
Muscarinic receptors 122
Growth factors 123
Thrombin 123
The inflammatory response 124
Other agents 124
Amyloid function 125
Amyloid aggregation 125
Calcium and amyloid toxicity
125
Amyloid and the cytoskeleton
127
Neurotransmitter modulation of amyloid toxicity 127
Presenilins 127
Apolipoprotein E 128
2-Macroglobulin
128
The inflammatory response 128
The C100 protein 129
Conclusion 130
Acknowledgments 130
Alzheimer's disease (AD) is a progressive neurodegenerative dementing illness
that affects up to 5% of those over 65 years and rising to >20% of those over
80 (Launer
et al., 1999). While the disease itself is not fatal, medical complications
associated with AD result in it being termed the fourth largest cause of death
in the Western world (after coronary heart disease, cancer, and stroke). AD is
characterised by an initial loss of short-term memory, gradually leading to
total mental incapacity and full dependence (Bouchard
& Rossor, 1996). It was initially reported in 1904 as a form of
presenile dementia in a 51-year-old patient, Auguste D. AD is responsible for
over 50% of all cases of dementia, although other types of dementia with similar
clinical symptoms also exist (Launer
et al., 1999). Currently, there are ~350,000 people in the United Kingdom
alone with AD, with a cost to the health service, both direct and indirect, of
over 2 billion pounds. Because AD initially was defined by its unique pathology
of neurofibrillary tangles and neuritic (senile) plaques, the disease can only
be diagnosed definitively at the post-mortem stage (Dickson
and Goedert).
While a majority of AD cases (>95%) are termed sporadic (i.e., have no direct
genetic basis), the remainder have been associated with specific gene mutations.
These mutations have been identified in both the amyloid precursor protein (APP) (Levy
and Goate)
and in the presenilin (PS) proteins (3.4
and 4.5).
In addition, a number of genetic risk factors have been identified. While these
are not an absolute requirement, they act to increase the likelihood of a person
developing the disease (Section
4.6).
While there are many components contained within the neuritic plaques of AD
brains (which are a general accumulation of cell debris following neuronal cell
death), early studies identified a key role for a 40ø42 amino acid
peptide (Glenner
and Masters).
This was termed the amyloid
(A) peptide because of the -pleated sheet amyloidogenic structure
of the aggregated peptide, as detected by Congo red and thioflavin-S staining.
The A peptides, which are
3ø4 kDa in length, are the building blocks of both the amyloid
fibrils in neuritic plaques, in addition to the vascular deposits that
accumulate in the brains of patients with AD. A is itself derived from the proteolytic
processing of one or more isoforms of the APP (Kang
et al., 1987). The isoforms of APP are Type I transmembrane sialoglycoproteins
that are encoded for by a single gene on chromosome 21 that contains 19 exons
(Goate
et al., 1991). The protein exists as 3 primary isoforms: APP695, APP751, and APP770 (the numbers referring to the
number of amino acids in the protein). Both APP751 and
APP770 contain the 56 amino acid Kunitz protease inhibitor (KPI)
domain, which is encoded on exon 7. In addition, APP770 contains a 19 amino acid
insert that codes for an OX-2 domain (Fig.
1; Kitaguchi;
Ponte
and Tanzi).
In addition, there is a splice variant, which lacks exon 15, termed L-APP (Pangalos
et al., 1995a). There are also two amyloid precursor-like proteins (APLP1
and 2), which have similar extracellular and intracellular domains to APP, but do not contain the 42 amino
acid A region (Wasco
and Wasco).
Fig. 1. Potential A
-secretase pathways. The locations of the KPI and OX-2 domains are indicated
APP has been proposed to serve
as a multifunctional protein that mediates both cell adhesion and neurite
outgrowth. Although these two functions are intricately linked and a large
number of cell adhesion molecules (CAMs) have been reported to exhibit
neuritogenic properties (Doherty
et al., 1991), a distinct role for APP as a neurotrophic agent that activates
cell-surface receptors to stimulate neurite outgrowth has also been proposed.
The expression of APP as a
membrane-bound glycoprotein led to the initial suggestion that it may play a
potential role in the mediation of cell adhesion. This suggestion was further
fuelled by the irregular punctate-staining pattern of APP on the neuronal cell surface, which is
similar to the staining patterns of other CAMs (Shivers
et al., 1988), and also by its synaptic localisation (Schubert
and Storey).
There are reports of two pools of APP within the cell membrane, one with a rapid
half-life that gives rise to APPs and the other with a much slower
turnover rate that may be responsible for the adhesive properties of the protein
(Storey
et al., 1999). Initial studies demonstrated that a nerve growth factor
(NGF)- or fibroblast growth factor-induced increase in APP expression increases the adhesive strength
of PC12 cells to a collagen substrate (Schubert
et al., 1989). In addition, the treatment of neuronal cells with heparin,
which reduces the strength of their substrate interaction, results in a
significant reflex increase in APP mRNA, which has been proposed as a potential
compensatory mechanism (Octave
et al., 1989). Subsequent studies demonstrated that Fab' fragments of
anti-APP antisera can inhibit
both cell-cell and cell-substrate adhesion (Breen
and Chen),
and the down-regulation of cellular APP expression using an antisense strategy also
reduces the strength of cell-substrate adhesion (Coulson
et al., 1997). Furthermore, an overexpression of the protein in
Drosophila results in a blistered wing phenotype, which is suggestive of
an alteration in cell-cell adhesion in the dorsal and ventral epithelial cell
layers during development (Fossgreen
et al., 1998).
In keeping with its proposed role as a mediator of cell-substrate adhesion,
APP has been demonstrated to
interact with a number of elements of the extracellular matrix. These include
collagens I and IV (Breen
and Beher),
laminin (Multhaup
and Kibbey),
fibronectin (Narindrasorasak
et al., 1995), heparan sulfate proteoglycans (HSPGs) (Narindrasorasak
and Coulson),
and glycosaminoglycans (Multhaup,
1994). It has also been proposed to play a role as an integral component of
the extracellular matrix (Klier
et al., 1990). In astrocytes, the cell surface expression of APP co-localises with the 11 integrin at distinct points of contact with
the extracellular matrix (Yamazaki
et al., 1997). It is not known, however, if APP binds directly with the integrins or is
co-localised in focal adhesion points in which the cell interacts with elements
of the extracellular matrix. The secreted APPs can also bind to specific
proteins on the cell surface, with this interaction, at least in part, being
responsible for the adhesion-mediated neuritotropic properties of the protein (Ninomiya
et al., 1994).
The individual isoforms of APP
exhibit different adhesive characteristics. Using recombinant protein in a
binding assay, APP695
was demonstrated to have a lower affinity for both fibronectin and entactin than
the KPI-containing forms, and also had a significantly lower capacity for
binding to collagen IV (Narindrasorasak
et al., 1995). A similar trend was observed with cells overexpressing the
APP751 isoform, which
have a greater adhesive strength to collagen IV than cells expressing APP695 (Gillian
et al., 1997) and which have a greater potential to stimulate neurite
outgrowth in a co-culture system (Qiu
et al., 1995).
The heparin-binding capacity of APP has been examined in detail, as this may
play a key role in the adhesion-mediated neuritogenic properties of the protein
(Williamson
et al., 1995). At least four binding domains have been identified on the
APP695 protein at
amino acids 96ø119 (H-I), 131ø166 (H-II),
316ø346 (H-III), and 382ø447 (H-IV) (Small;
Clarris
and Mok;
Fig.
2). While the heparin-binding properties of APP provide a mechanism by which the protein
interacts with the HSPG elements of the extracellular matrix (Small
et al., 1996a), it may also mediate its binding to other components, such as
collagen (Breen
and Caceres)
and glypican (Williamson
et al., 1996). A potential heparin-binding site has also been identified
within the A region of the
protein. This may play a key role in the stimulation of amyloid deposition and
the accumulation of HSPG, which is an early event in plaque formation (Snow
and Maresh).
HSPG can potentiate the neurotoxic effect of the A25ø35 fragment, either by
altering its aggregation state (Woods
and Castillo)
or by serving as a potential cell surface receptor for the binding of A fibrils (Schulz
et al., 1998). Furthermore, the heparin-binding domain of the A peptide is identical to the
microglial-activating domain, suggesting that A stimulation of microglial cells may be a
heparin-dependent process (Giulian
et al., 1996).
Fig. 2. Schematic diagram of the A
In addition to the parent protein, the A peptide may also exhibit adhesive properties.
Both A and the C-terminal
fragment (CTF; CT105/C100) have been demonstrated to interact with elements of
the extracellular matrix, including laminin and fibronectin, to provide a
permissive substrate along which neurite outgrowth from spinal cord explants may
occur (Ghiso
and Koo).
This may help to provide an explanation for the neurotrophic actions of low
concentrations of A that were
originally reported (Whitson
and Yankner).
However, the stimulation of the activities of the tyrosine kinase,
phosphatidylinositol (PI)3 kinase, protein kinase C (PKC), and
phospholipase (PL)A2 enzymes by the peptide may also explain some of
the physiological activities of low concentrations of A (Luo;
Luo;
Luo
and Lehtonen).
The post-translational modifications of proteins play a key role in
determining both their processing and their adhesive properties (Breen
et al., 1998). APP can form
the core protein of a chondroitin sulfate proteoglycan (Shioi
et al., 1992), which is termed appican (Shioi
et al., 1996), and it may also exist as an HSPG (Schubert
et al., 1988). The attachment of chondroitin sulfate chains is restricted to
the Ser619 amino acid residue of the L-form of the protein (that lacks exon 15)
(Pangalos
et al., 1995a). The expression of appican, which can exist in either a
membrane-bound or a secreted protein (Shioi
et al., 1993), is restricted to neural cells (Pangalos
and Shioi),
although the secreted appican may contain the complete amyloidogenic sequence,
suggesting an altered processing pathway of the proteoglycan form of the APP (Salinero
et al., 1998). The presence of the chondroitin sulfate side chain is
important for the adhesive function of APP, as appican demonstrates an increased
adhesive potential when compared with the non-chondroitin sulfate proteoglycan
form of APP (Pangalos
et al., 1996), and the secreted form of appican can also interact with
elements of the extracellular matrix in order to promote the adhesion of neural
cells (Wu
et al., 1997).
While the properties of adhesion molecules have been well-characterised using
in vitro systems, their role in the CNS in vivo is less clear. It is proposed,
however, that they may play a key role in synaptic plasticity associated with
memory and learning (Breen
et al., 1998). By altering the adhesive potential of cells, the strength of
synaptic connections may be modified, resulting in the potential to alter the
existing synaptic connections or even generate new attachments. Because of its
adhesive properties, as well as its neuritogenic properties, APP has been proposed to play a role in the
modulation of synaptic plasticity (Sisodia
& Gallagher, 1998). Indirect evidence for such a role has been provided
by reports of a marked increase in APP levels in rats raised in an enriched
environment when compared with those in impoverished surroundings (Huber
et al., 1997). Early experiments demonstrated that the intraventricular
infusion of antibodies directed against the extracellular domain of APP results in an impairment of the
acquisition of a passive-avoidance memory task (Doyle
et al., 1990). This effect is similar to that reported for other CAMs (Doyle;
Scholey
and Arami).
Although antibody-intervention studies have provided an indication of the
potential functional role of a protein, the possibility exists that the
antibody-APP complex may actually
act to prevent the action(s) of adjacent cell-surface glycoproteins. Therefore,
studies have been carried out to investigate a positive role for APP (rather than a negative role for anti-APP antibodies). Initial experiments
were carried out using a 17-mer peptide (containing the RERMS pentapeptide),
corresponding to a proposed "active site" of APP that is within a small domain just
C-terminal to the KPI insertion site (Fig.
2). Intraventricular infusion of this peptide increases the synaptic density
in the frontoparietal cortex, with an associated increase in memory retention (Roch
et al., 1994). More recent studies have characterised the role of APP in memory in more detail by the
intraventricular infusion of recombinant protein. Both APP695 and APP751 can display nootropic
properties, as well as overcoming scopolamine-induced amnesia. In particular,
low doses of the protein (0.5ø5.0 pg) improved performance in both a
go/no-go visual discrimination task and an object recognition task, although
higher doses (>5.0 pg) are required to overcome the scopolamine-induced
amnesia (Meziane
et al., 1998).
APP may also play a role in
long-term potentiation (LTP), an electrophysiological correlate of memory and
learning (Moser
et al., 1998). The induction of LTP in vivo results in an increase in the
secretion of APPs in a
time-dependent manner, and this can be blocked by pretreatment with either a
protease inhibitor or the N-methyl-
-aspartate
antagonist AP5 (Fazeli
et al., 1994). Pretreatment of hippocampal slices with APPs produces an alteration in the
electrophysiological characteristics of both LTP and long-term depression,
further supporting a role for APP
in synaptic plasticity (Ishida
et al., 1997). Transgenic mice expressing the Swedish mutation of APP (K670N, M671L) display a
significant impairment of LTP. This effect is attenuated with age and is
accompanied by deficits in spatial memory (Chapman
et al., 1999). These effects, however, were not associated with any marked
reduction in synapse number or evidence of cell death. This suggests that the
mutations in APP, with the
associated increase in A
generation and decrease in APPs, may impair its ability to
modulate synaptic plasticity. These studies have been supported by in vitro
studies in which both the A
peptide and the CTF (CT105) shorten the duration of LTP induction and the
amplitude of excitatory postsynaptic potentials, and therefore, may contribute
to the cognitive deficits associated with AD (Cullen;
Cullen
and Lambert).
The effects of alterations in the expression levels of APP on memory and learning have also been
investigated using transgenic mouse models. A moderate increase in APP expression in transgenic mice
results in an increase in synaptophysin-immunoreactive synaptic connections (Mucke
et al., 1994), but a significant increase in expression of either the
full-length form of the protein or the C100 region actually inhibits learning
performance at 10 months of age (Hsiao
and BergerSweeney).
Consistent with this finding are the results from APP knock-out mice or mice containing a
shortened form of the protein APP
(
/), which do not exhibit marked changes in
behaviour, although there are subtle changes in locomotor activity, cognitive
ability, and LTP (Zheng;
Perez;
Tremml
and Dawson).
It is not clear whether these effects are associated with developmental changes
due to APP deficiency or to the
absence of the protein in the mature brain. There is no evidence, however, of
any neuronal loss in the brains of adult APP knock-out mice (Phinney
et al., 1999), although there are some changes in neuronal morphology and
synaptic efficiency (Seabrook
et al., 1999). These studies have been complicated by the potential
compensatory effects of other glycoproteins that exhibit similar functional
properties to those of APP. In
particular, the APP-like proteins
APLP1 and APLP2 may counterbalance the changes in APP expression (Wasco
and Sandbrink).
However, over 80% of animals that are null for both APP and APLP-2 died within the first week after
birth, suggesting that while APLP-2 and APP may have similar functions and can
substitute for each other, the proteins play a key role in the developmental
process (VonKoch
et al., 1997). The generation of inducible knock-outs, when the animals are
allowed to develop normally and the APP then turned off, will provide a useful model
system with which to separate out the developmental from the plastic functions
of the protein.
precursor proteinAlthough both soluble and membrane-bound APP influence adhesion-mediated neurite
outgrowth, the soluble form of the protein (APPs) has distinct neurotrophic and
neuroprotective properties. The generation of APPs and A are mutually exclusive (Fig.
1), and an increase in A
associated with AD with a parallel decrease in APPs generation may render the cells
more vulnerable to the actions of other neurotoxins. This may ultimately result
in the precipitous nerve cell death that accompanies the later stages of the
disease.
A primary role for APPs is as a neuroprotective agent
that acts to stabilise intracellular Ca2+ levels. The treatment of
cells with APPs
results in a rapid and prolonged decrease in cellular Ca2+ (Mattson
et al., 1993). Thus, the protein is considered to attenuate the toxic
actions of the A peptide by
overcoming the A-associated
elevation of free intracellular Ca2+ levels with the associated
induction of free radical species (Goodman
& Mattson, 1994). The neuroprotective properties of APPs, however, are not limited to
modifying the neurotoxic actions of A, and the protein can act as a neuroprotective
factor in response to a number of diverse stressful conditions associated with
elevations in intracellular Ca2+ levels. Treatment of cerebellar
granule cells with glutamate results in an increase in free intracellular
Ca2+ with subsequent cell death (Budd
& Nicholls, 1996). These actions are significantly attenuated following
co-treatment with APPs
(Mattson,
1994). Ischaemic damage in vivo, which is largely mediated by chronic
glutamate release and receptor stimulation, as well as the elevation of
intracellular Ca2+ associated with hypoglycemic damage, is also
protected against by the co-infusion of APPs (Mattson
and Smithswintosky).
A similar neuroprotective effect can be observed in vivo, as APP-deficient mice are more susceptible to
kainate-induced seizures (Steinbach
et al., 1998). The neuroprotective effect of APPs, which is N-methyl--aspartate
receptor-specific and not observed following activation of -amino-3-hydroxy-5-methyl-isoxazole-4-propionate
or kainate receptors, appears to be associated with the activation of
charybdotoxin-sensitive K+ channels, leading to a hyperpolarisation
of the neurons (Furukawa
et al., 1996a). This effect is mediated by the activation of a cyclic GMP
(cGMP)-mediated pathway and can be blocked by inhibitors of a cGMP-dependent
kinase and mimicked by nonhydrolysable cGMP analogues (Barger
and Furukawa).
The characteristics of the induction of long-term depression in hippocampal
slices and the frequency of spontaneous postsynaptic currents in the
Xenopus neuromuscular junction can also be influenced by the APP-mediated stimulation of cGMP
levels. These data provide further evidence for an interaction between APPs and guanylate cyclase
(Ishida
and Morimoto).
APPs also plays a
key role as a neurotrophic agent to stimulate neuronal cell differentiation and
neurite outgrowth. Transfection of an APP-null neural cell with APP cDNA results in an increase in neurite
outgrowth, and this property has been localised to an RERMS peptapeptide (amino
acids 328ø332). Furthermore, the binding of a larger peptide
(Ala319 to Met335) to cell membrane preparations is both
saturable and specific, suggesting that APPs mediates its effects by a
receptor-mediated mechanism (Jin
and Ohsawa).
An active domain has also been reported within amino acids 591ø612
within the A peptide region. This
domain is present in APPs, but not in APPs, suggesting that the alternative processing of
APP by specific secretases may
influence APPs
function (Furukawa
et al., 1996b). At low concentrations, APPs can also potentiate the actions
of NGF by the activation of the extracellular signal-regulated kinases ERK-1 and
-2 via a PI3 kinase pathway (Wallace
and Wallace).
In addition, there is an increase in the affinity of the p75 low-affinity NGF
receptor and a potentiation of the trkA tyrosine kinase enzyme (Akar
& Wallace, 1998). As these agents have also been demonstrated to
modulate APP expression and APPs generation, there
exists a complex relationship between the expression and function of the
protein. Furthermore, although the exact mechanisms underlying the physiological
activities of APPs are
not clearly understood, the evidence available suggests that the protein may
interact with a cell-surface receptor to mediate its biological effects.
precursor
protein interaction with membrane-bound proteinsThe membrane-bound form of APP
has also been proposed to modulate cellular activity, potentially by an
interaction with elements of a second messenger system pathway. Initial studies
suggested an interaction between the C-terminal domain of APP and the Go protein, with the
subsequent binding of GTP-S to
Go (Nishimoto
et al., 1993). This binding could be inhibited by the pretreatment of
Go with GTP-S. The
binding of the 22C11 antibody, which recognises the N-terminal of APP, resulted in an increase in GTP-S binding and an increase in GTPase
turnover. This effect could be blocked by an antibody directed against the
intracellular C-terminal of APP,
which may mediate APP-Go protein interaction (Okamoto
et al., 1995). More recent studies, however, have reported that the
interaction of APP with
Go results in an
inhibition of the high-affinity Go GTPase activity, with this effect
being potentiated by treatment of cell membranes with the 22C11 antibody (Brouillet
et al., 1999). The apparent anomaly between these results may reflect the
differences between the two experimental systems used. The first experiments
were carried out using reconstituted phospholipid vesicles, while the more
recent studies were carried out using whole cell preparations. This suggests
that other membrane-bound proteins may also influence APP-Go interaction. Subsequent events
in the APP-Go pathway
include the activation of mitogen-activated kinase enzymes and the
phosphorylation of specific proteins (Murayama
et al., 1996).
While the function of the APP-Go complex has not been
clarified, the induction of apoptosis and cell death in cells expressing
familial AD (FAD) mutations of APP at amino acid V642 (Yamatsuji
et al., 1996b) may occur via a G-protein-mediated mechanism. Go
appears to be constitutively activated in cells expressing mutant APP, and treatment with pertussis toxin, an
inhibitor of G-protein function, or co-transfection of the cells with an
inactive form of Go greatly reduced the occurrence of apoptotic cell
death (Yamatsuji
et al., 1996a). The Go-subunit involved appears to be the G component, as overexpression of cDNA coding
for G22 also induced DNA fragmentation
and cell death, thus also implicating this protein in the apoptotic pathway (Giambarella
et al., 1997b). FAD mutations of APP also inhibit the activity of the cyclic AMP
(cAMP) response element by a Go-dependent, but adenylate
cyclase-independent, mechanism following the binding of the mutant APP to the C-terminus of Go. This suggests that the APP mutants actually may demonstrate a
gain-of-function (Ikezu
and Giambarella).
In addition to Go, a number of other cellular proteins have been
reported to interact with the C-terminal of APP. APP-BP1 is a 59 kDa protein homologous to the
Arabidopsis AXR1 gene that plays a role in signal transduction and is
related to the ubiquitin-activating enzyme E1 (Chow
et al., 1996). APP also
interacts with the neuronal protein Fe65, which is a homologue of protein X11
(Fig.
3).
Fig. 3. Schematic diagram of the interaction of the cytoplasmic tail of A
Recent studies have identified a 130 kDa protein that binds to residues
645ø694 in the intracellular portion of APP and that co-immunoprecipitates with APP (Watanabe
et al., 1999). This protein is homologous to the UV-damaged DNA-binding
protein (UV-DDP) that plays a role in DNA repair. Although the role of the APP/UV-DDP interaction remains unclear,
this may serve as a signal transduction mechanism responsible for the regulation
of gene transcription, as UV-DDP has been proposed to translocate between the
cytoplasm and the nucleus. APP
has also been demonstrated to bind to the PS proteins within the membrane (Xia
et al., 1997) (Section
3.4). Although the PS proteins generally have been identified as elements of
the endoplasmic reticulum (ER) and cis-Golgi (Culvenor
et al., 1997), there is some evidence that they may also exist at the level
of the cell surface and play a role in the processes of secretion and
endocytosis (Efthimiopoulos
et al., 1998). They may also bind to APP on an opposing cell in order to mediate cell
adhesion (Dewji
& Singer, 1997). This trans-binding of the PS proteins and APP results in a transient increase in
protein tyrosine kinase activity and protein phosphorylation, and this PS:APP interaction may play a role in
intracellular signaling (Dewji
& Singer, 1998). Intriguingly, recent studies have proposed that PS1, in
fact, may exhibit -secretase
activity, and this may also explain the PS-APP interaction at the level of the Golgi (DeStrooper
and Wolfe).
Because of both its adhesive and neurotrophic properties, APP is likely to play a role in the process of
neuronal development. While there are no overt changes in APP knock-out animals, hippocampal cells
cultured from these animals exhibit reduced viability and defective neurite
outgrowth (Perez
et al., 1997). This deficit, however, can be rescued by co-culturing the
cells with astrocytes derived from control animals, suggesting that APP plays a key role in the
developmental process. APP
knock-out mice also demonstrate an increased sensitivity to the actions of the
excitotoxin glutamate. However, although the animals are more prone to kainic
acid-induced seizures, there is no evidence of increased cell death or apoptosis
(Steinbach
et al., 1998). However, further studies using inducible knock-out animals
are required to fully understand the function(s) of APP.
precursor protein
processingTwo alternative pathways that are mutually exclusive serve to process APP. The primary pathway results from
the cleavage at Lys16 within the A region of the protein by the -secretase protease enzyme. This results in
the generation of the soluble APPs that displays neurotrophic and
neuroprotective properties. Alternatively, it can be cleaved by the - and -secretases to generate the 40/42 amino acid
A peptide (Busciglio
and Mills).
The major species of A generated
under normal circumstances is the 40 amino acid A40 form (Haass
and Seubert).
However, in AD, there is an increase in the A42 peptide that is more
amyloidogenic and acts as a seed for amyloid deposition (Pike;
Borchelt
and Lemere).
A42 is also the major
A component of the amyloid
plaques (Roher
and Iwatsubo).
The choice of the pathway by which APP is processed is influenced by a large number
of factors that are both extra- and intracellular. Furthermore, the processing
mechanisms may be cell-, tissue-, or species-specific.
APP trafficking in
nonpolarised cells indicates that the protein is delivered to the cell surface
via a constitutive biosynthetic pathway, where it can be cleaved by -secretase and secreted as APPs. Alternatively, it can
be internalised by endocytosis and eventually degraded (Haass;
Nordstedt
and Koo).
In polarised cells, however, alternative splicing of the protein is particularly
important in determining the fate of the protein, with exon 15 (L-APP) playing a pivotal role in this process. In
canine kidney cells, APP is
sorted basolaterally, while cells transfected with L-APP677 expressed this isoform both
apically and basolaterally (Hartmann
et al., 1996).
In both mouse teratocarcinoma cells (P19) and human embryonic kidney cells
(HEK293), the generation of APPs by -secretase cleavage is reduced by the
expression of the KPI containing isoforms of APP, with a concomitant increase in A. Thus, in human brain and in animal
models of AD, the amount of KPI-containing APP produced may be a key factor in influencing
the deposition of A (Ho
et al., 1996). The KPI-containing isoforms are expressed both in the CNS and
in the peripheral tissues, whereas APP695 is expressed almost
exclusively in the brain. During normal aging, APP751 levels increase in parallel
with the age-associated loss of neurons (Tanaka
and Tanaka)
and the ratio of APP751/APP695 mRNA is correlated
positively with plaque density in the brains of AD and control patients (Johnson
et al., 1990).
While trace amounts of A
peptide have been detected as part of the normal cellular metabolism of APP (Haass
and Shoji),
an increase in the production of the peptide and its subsequent deposition as
insoluble amyloid plaques may represent the key pathological event that triggers
the disease process (Hardy
and Selkoe).
As such, any manipulation that diminishes the production of A may be a potential therapeutic target (Mills
and Schenck).
In non-neuronal cells, the secreted APPs fragments are generated
primarily via the -secretase
pathway, with a minority being produced by the /-secretase route (Fig.
1), primarily in the endosomal/lysosomal (E/L) pathway (Cole
and Golde).
There is some evidence, however, that some full-length APP can also be secreted (Vassilacopoulou
and Tezapsidis).
While the -secretase has not
been identified, it is thought to be localised either in the late Golgi or in
caveolae within the plasma membrane ( Mills
& Reiner, 1999; Section
3.2).
APP has also been associated
with ubiquitin in the E/L compartments in AD. While ubiquitin normally plays a
role in the targeting of proteins to the proteasome, in stress-related
conditions where there is an elevation of APP and its fragments, ubiquitin may start to
play a role in the processing or trafficking of APP to the E/L compartment. Temperature-block
studies of metabolically labelled proteins suggested that vesicle budding and
trafficking of APP to the cell
surface are not an absolute requirement for the generation of A in Neuro 2A rat neuroblastoma cells (Xu
et al., 1997). This implies that the - and -secretase cleavage can occur within the ER or
trans-Golgi network (TGN) (Cook
and Hartmann).
Although the -secretase pathway
appears to play a key role in the processing of APP in neuronal cells (Turner
and Tienari),
there is no conclusive evidence that deciphers the exact location of these
secretase pathways in any cell type. While there is evidence that the cleavage occurs in the late endosomal
compartment, with the cleavage
occurring in the early endosomes (Peraus
et al., 1997), other studies suggest that the cleavage occurs prior to the cleavage (Anderson
and Paganetti).
The exact subcellular location of APP cleavage may differ between species, and
this factor may help to explain why AD is a human-specific disease (LeBlanc
& Goodyer, 1999). While the -secretase enzyme has not been identified yet,
its actions, which may be associated with multiple proteases, are determined by
the location of the APP within
the plasma membrane (Murphy
et al., 1999). Furthermore, the -secretases, which generate A40 and A42 may be distinct enzymes, with the
-40 secretase being a cysteine
protease and -42 being a serine
protease (Klafki;
Citron
and FigueiredoPereira),
with the generation of A40 and A42 occurring within different
compartments of the cell. It has been proposed that A40 is generated in the TGN, while
A42 is produced both
within the ER and the TGN with the A generated in the TGN being localised in two
distinct pools (Greenfield
et al., 1999).
precursor proteinIn an attempt to decipher the pathways and organelles involved in APP catabolism, studies have examined
the motifs present on APP that
appear to play important roles in the localisation of the protein. These studies
have confirmed that APP uses a
specific intracellular route that targets proteins to the E/L system, although,
to date, the full role of this organellar system is unclear (Nordstedt
et al., 1993).
There are a number of classes of transport vesicles, each of which plays a
specific role in intracellular protein transport. These lipid vesicles contain a
specialised protein coat and bud-off from membranes to transport proteins
between organelles within the cell. There are three main classes of coated
vesicles. Coatomer-coated vesicles mediate transport between the ER and the
Golgi. Clathrin-coated vesicles (CCVs) mediate the transport of transmembrane
proteins, such as APP and the
low-density lipoprotein (LDL) receptor, between the Golgi and the plasma
membrane. The third class of vesicles was originally identified as invaginations
at the level of the plasma membrane. These caveolae are thought to play a major
role in the recycling of proteins from the cell membrane, probably to the E/L
compartments.
The presence of both full-length APP and associated peptide fragments in the CCVs
suggests that the secretory cleavage of APP occurs in a pre-CCV compartment (Liu
and Wu).
In fact, it is possible that internalised caveolae and CCVs may be targeted to a
common endosomal pool (Simons
& Ikonen, 1997). In this model of APP processing, since cytoplasmic APP has been shown to be enriched in lysosomes
(Haass
et al., 1992a), it is likely that the -secretase cleavage occurs first in the
caveolae, with the remaining intact APP being cleared from the cell surface by the
clathrin-coated pits and targeted to the endosomes and lysosomes for proteolytic
processing (Koo
and Yamazaki).
In agreement with this hypothesis, the activation of PKC, a well-established
component of caveolae, which is known to stimulate the secretion of APP, inhibits caveolae internalisation
(potocytosis) (Lisanti
and Smart).
This suggests that PKC activation may prolong the time spent by APP at the level of the cell surface, thus
maintaining APP in close contact
with the plasma membrane form of -secretase, and may play a key role in the PKC
activation of APPs
secretion (Section
3.5).
Caveolae are sphingomyelin/cholesterol-rich membrane domains in which the cholesterol component plays a key structural role (Sargiacomo and Liu). Cholesterol-depleted cells have reduced numbers of caveolae (Anderson; Chang and Anderson) and cholesterol-binding drugs, such as nystatin and filipin, disassemble the membrane coat of caveolae in fibroblast cells (Rothberg et al., 1992).
The homo-oligomeric caveolin proteins that bind cholesterol and
glycosphingolipids serve as marker proteins for caveolae (Rothberg
and Ikezu).
These protein-protein and protein-lipid interactions are thought to be essential
for the formation of caveolae (Sargiacomo
et al., 1995). Caveolin contains a modular protein domain that directly
participates in the recognition of specific consensus motifs within its
interacting partners (Couet
et al., 1997). Caveolins play a key role in G-protein-coupled signalling by
interacting directly with the Gs-subunit (Li
et al., 1995) or indirectly via ligand-binding proteins (e.g., muscarinic
acetylcholine, -adrenergic, and
bradykinin receptors) (Raposo;
Raposo;
deWeerd
and Feron).
APP contains a caveolin-binding
motif within its cytoplasmic domain, which may explain the
internalisation/sequestration of APP into caveolae (Ikezu
et al., 1998). The -secretase cleavage product of APPs has been shown to be enriched
within caveolae (Bouillot
et al., 1996), and the co-expression of APP and caveolin-1 results in the -cleavage of APP at the plasma membrane. These results
suggest that caveolae, therefore, play a significant role in the -secretase cleavage pathway of APP (Fig.
4) (Ikezu
et al., 1998).
Fig. 4. The proposed role of caveolae in the processing of A
Treatment of cells with cholesterol oxidase results in the appearance of caveolin within the Golgi, suggesting that oxidized or otherwise damaged cholesterol may stimulate the movement of caveolin from the cell surface to the Golgi through the ER/Golgi pathway and that this translocation requires a functional cytoskeleton. Movement in the other direction from the Golgi to the plasma membrane, however, is a microtubule-independent process (Conrad et al., 1995).
While the -secretase-mediated
cleavage of APP may occur at the
level of the cell surface, suggesting that the protease involved is a
membrane-anchored protein (Haass
et al., 1992a), the cleavage can also occur intracellularly (Destrooper
et al., 1993), and a disintegrin and metalloprotease (ADAM 10) has been
proposed as a potential candidate enzyme for this cleavage (Lammich
et al., 1999). The activation of caspase enzymes associated with apoptosis
has also been proposed to generate A (Gervais
et al., 1999). However, the cleavage to generate A occurs only after APP has been modified by O-glycosylation,
suggesting that the process takes place within or after the Golgi complex (Tomita
et al., 1998). Although this may represent a caveolar-independent pathway,
and there are some reports that a specific pool of APP may not be associated with caveolae (Parkin
et al., 1997), it may also reflect the intracellular localisation of certain
forms of caveolae. No -secretase
cleavage could be detected at the level of plasma membrane in the absence of
caveolin-1, thus suggesting that caveolae may play a pivotal role in the cycling
of APP between the cell surface
and intracellular populations of caveolae (Ikezu
et al., 1998).
There are several links between cholesterol metabolism and the pathogenesis
of AD. In particular, specific allelic forms of apolipoprotein E (ApoE) serve as
a genetic risk factor for the development of the disease (Strittmatter
et al., 1993). There are three ApoE isoforms (termed E2, E3, and E4), and
people who are either homozygous or heterozygous for the E4 allele have an
increased susceptibility to AD. Cholesterol modulates certain physical
properties of membranes, and the internalisation of ApoE-lipoprotein complexes
serves as the primary cellular mechanism for cholesterol uptake (Guillaume
et al., 1996). This uptake is mediated by the LDL receptor pathway (Dehouck
et al., 1997). ApoE-enriched lipoproteins can bind the A peptide, with the resulting complexes being
taken up into primary neurons and astrocytes by the LDL receptor (Beffert
et al., 1998). The ApoE2 and ApoE3 alleles appear to bind A with a higher affinity than ApoE4, and
therefore, they may play an important role in clearance of the peptide prior to
its aggregation into insoluble deposits (Yang
et al., 1999). In addition, the LDL receptor-related protein (LRP) receptor
(Fig.
3) mediates the uptake of the secreted APPs from the extracellular space (Kounnas
et al., 1995). Cholesterol modulates APP processing, with a reduction in the levels
of APPs associated with increased cellular
cholesterol levels (Bodovitz
& Klein, 1996). This has been associated with an increase in the level
of both the mature and the immature forms of the APP holoprotein, while there was no apparent
effect on A production (Bodovitz
& Klein, 1996). Other studies report, however, that a 70% reduction in
the cellular cholesterol by lovastatin and methyl--cyclodextrin eliminates the production of
A without affecting the secretion
of APPs (Simons
et al., 1998), and increased cholesterol may be associated with the
"seeding" of amyloid deposits by A (Mizuno
et al., 1999). This effect may be due to a preferential interference with
the -cleavage pathway over the
-cleavage. Dietary cholesterol
may also modulate the processing of APP, with an increased cholesterol intake
leading to a decreased secretion of APPs, A40, and A42. This only occurs in the presence
of ApoE, suggesting that the effect of cholesterol on APP processing is an ApoE-dependent event (Howland
et al., 1998). Altered dietary cholesterol did not appear to have any effect
on the levels of the full-length holoprotein (Howland
et al., 1998). These links between cholesterol and APP, however, may reflect the caveolar
localisation of APP (Ikezu
et al., 1998), and a neuronal-specific form of caveolin may interact with
neuronally expressed APP.
Caveolar dysfunction, therefore, would lead to a down-regulation of the -secretase pathway in AD, which may
result in an overproduction of the toxic A peptide (Ikezu
et al., 1998).
precursor protein processingAPP has two internalisation
signals on its cytoplasmic tail, Gly-Tyr-Glu-Asn-Pro-Thr-Tyr (GYENPTY) and
Tyr-Thr-Ser-Ile (YTSI), both of which appear to play a role in the lysosomal
targeting of the protein (Lai
et al., 1995). The cytoplasmic region of the GYENPTY motif interacts with
the phosphotyrosine-binding/protein interaction domain of X11 (Borg
et al., 1996) and Fe65 (McLoughlin
& Miller, 1996) (Fig.
3). Non-neuronal cells require the GYENPTY motif for the endocytosis of cell
surface-expressed APP by
clathrin-coated pits (Haass
and Koo),
and this motif is sufficient for subsequent A production. The use of mutation/substitution
studies of various motifs and regions of the APP protein, and the resultant pattern of
peptide formation, have confirmed that internalisation results in a reduction of
the secretory cleavage of APP.
This further emphasises the importance of APP re-internalisation in the processing pathway
of this protein.
The efficient targeting of APP
to lysosomes requires both the YTSI and the GYENPTY motifs, with the
Tyr653 residue of the YTSI motif being particularly important for its
basolateral expression in polarised cells (Lai
et al., 1998). The interaction of X11 or X11 with APP has a significant effect on APP processing, with a decrease in the quantity
of APPs, A40, and A42 being recovered from the
conditioned medium of HEK 293 cells expressing X11 (Borg
et al., 1998). Co-expression of X11 with APP containing the Swedish mutation (APPSwe) resulted in a
decrease in the secretion of both A40 and A42 compared with APPSwe expression alone (Borg
et al., 1998). Transfection of APPSwe lacking the N-terminal
cytoplasmic tail showed no such reduction by X11.
Fe65 interacts with the LRP via its N-terminal tail and also interacts
through its carboxyl terminal domain, with the NpxY motif in the cytoplasmic
tail of APP (Fiore
et al., 1995). Fe65 is also capable of interacting through its WW domain
with Mena (the mammalian homologue of enabled), which, in turn, interacts with
the cytoskeleton and the CP2/LSF/LBP1 transcription factors to produce a
macromolecular complex (see Fig.
3; Russo
and Zambrano).
APP and LRP can also interact
with the mammalian homologue of disabled via the carboxyl terminal NPxY
motif (Duilio
and Howell),
with mammalian homologue of disabled knock-out mice showing abnormal
neuroanatomical organisation of the cortical cell layer (Howell
et al., 1997b). The KPI-containing isoforms of APP can also interact directly with the
extracellular domain of LRP (Kounnas
et al., 1995), although these isoforms usually are expressed in very low
levels in neurons. The multivalent complex comprising LRP, Fe65, and Mena can
modulate the intracellular trafficking of APP. Mutations in the NpxY motif of the
cytoplasmic domain result in an increased secretion of APPs, with no effect on A production, suggesting that the two
APP processing pathways may exist
independently (Jacobsen
et al., 1994). This deletion of the NPTY motif, however, may not only impair
endocytosis of cell surface APP,
but it may also alter its intracellular trafficking by increasing its direct
targeting into the endocytic pathway without first reaching the cell surface (Guenette
et al., 1996). Preliminary studies have also proposed the possibility of
genetic polymorphisms in the Fe65 gene being associated with sporadic AD, with
these changes potentially modifying the interaction of the protein with APP (Hu
et al., 1998).
A detailed investigation of neuronal APP transport and processing in the rat CNS has
revealed that the protein is transported in an anterograde fashion from the cell
body to the synaptic sites, with the majority of the APP existing as the APP695 isoform. Only mature, fully
glycosylated APP can be
transported by this fast axonal transport (McFarlane
and Buxbaum),
after which it is inserted into the plasma membrane, where it has a rapid
turnover rate (Morin
et al., 1993). The A region
is also essential for the axonal sorting of the protein (Tienari
et al., 1996). Synaptosomal studies have revealed that full-length APP in the nerve terminals is included
in structures associated with the endocytic pathway, particularly in the
multilamellar organelles and the CCVs, with APP being excluded from small synaptic vesicles
(Marquez-Sterling
et al., 1997). In this dentate gyrus model, full-length APP has been shown to exist inside
clathrin-coated synaptic vesicles that are recycled to the early endosomal
compartments, where sorting takes place. While a portion of the APP may recycle back to the presynaptic plasma
membrane, APP is definitely
transported retrogradely to the somatodendritic domain. A portion of this is
subsequently directed transcytotically to the somatic surface (Simons
and Yamazaki),
but delivery to the late E/L compartments where degradation takes place probably
also occurs (Marquez-Sterling
et al., 1997). However, whether this APP is in the presynaptic terminal or associated
with the extracellular space cannot be determined in this model. It has been
proposed that it may serve as a ligand that interacts with a postsynaptic
receptor (Breen;
Furukawa
and Gillian).
There are two key pieces of evidence to suggest that the CTFs of APP appear to be the penultimate
precursors of the A peptides (Estus
and Simons).
Firstly, cells and transgenic animals that express the Swedish APP variant secrete high levels of A in parallel with an increase in APP CTFs that contain an N-terminus at
the -secretase cleavage site (Thinakaran;
Lamb
and McPhie).
Secondly, neurons from mice that lack the PS1 protein and that fail to secrete
A accumulate large quantities of
the CTF within the cell (DeStrooper
and Naruse).
These findings suggest that the accumulation of CTF may itself be toxic (Section
4.9). This hypothesis is backed up by the observation that accumulated CTFs
are found in cells that have internalised A42 peptide aggregates. Indeed,
several reports point to the toxicity of CTF both in vitro (Kim
& Suh, 1996) and in vivo (OsterGranite
et al., 1996).
Mutations in the genes coding for the PS proteins PS1 and PS2 are responsible for a large number of cases of FAD (Sherrington et al., 1995). These are membrane-spanning proteins with 7 transmembrane domains that are located within the ER and the cis Golgi (Culvenor and Nakai). While their exact function remains to be elucidated, they have been proposed to play a role in protein trafficking within the cell (Naruse et al., 1998). In addition, PS may facilitate Notch-mediated target signalling during mammalian embryogenesis (Ray et al., 1999). Indeed, PS1 knock-out mice display abnormalities in body segmentation, in addition to cerebral haemorrhage (Conlon and Wong).
PS1 and 2 mutations serve to modify the processing of APP in a manner that leads to the increased
generation of the A42
peptide with a parallel decrease in APPs (Borchelt;
Citron
and Murayama).
These mutations are thought to potentially modulate the trafficking of APP (Weidemann
et al., 1997). However, it is still unclear as to whether the effects of
mutant PS proteins are due to a direct effect on the proteolytic processing of
APP or an indirect effect on its
trafficking. Studies with other ER chaperone proteins have demonstrated that
mutations of these proteins result in altered APP processing and A production (Yang
et al., 1998). Furthermore, it is not clear whether the C- and N-terminal
fragments of PS are the active components or whether the full-length PS is the
functional moiety. Intriguingly, a recent report has proposed that PS1 itself
may be the -secretase enzyme (DeStrooper
and Wolfe).
The PKC family is a heterogeneous family of phospholipid-dependent kinases
that can be divided into three categories, based on their cofactor requirements.
The conventional PKC isoforms (,
, and ) require Ca2+, phosphatidylserine,
and diacylglycerol (DAG)/phorbol esters as cofactors. Novel PKCs , 
, 
, and 
require only phosphatidylserine and
DAG/phorbol esters, while the atypical PKC enzymes 
and 
, whose regulation has not been fully
deciphered yet, require phosphatidylserine for maximal activity. PKC enzymes
contain four functional domains. The C1 domain binds DAG and phorbol esters in
all but the atypical PKC enzymes. The C2 domain binds to the acidic
phospholipids (and Ca2+ in the conventional PKC enzymes). These C1
and C2 interactions stimulate the transport of PKC to the plasma membrane, where
the phosphorylation of cellular proteins occurs following the binding of ATP to
the C3 domain, the substrate protein being bound to the C4 domain.
Colchicine sensitivity studies (Nitsch
et al., 1993) and the effects of cholera toxin on constitutive versus
PKC-induced APP secretion (Efthimiopoulos
et al., 1994) suggest that the receptor-activated and constitutive -secretase processing may occur by
different mechanisms. This statement is further supported by the observation
that whilst both PKC and - were involved in phorbol
ester-regulated secretion of APPs from the rat fibroblast 3Y1
cells that were stably overexpressing these isoforms only, PKC was involved in the stimulation of basal
APP secretion (Slack
and Kinouchi).
In the same studies, the PKC
isoform showed no such effects. Ionophore studies demonstrated that monensin
decreases both the regulated and constitutive secretion of APP, which, in turn, increased the intracellular
levels of APP, possibly by
upsetting the normal trafficking of the protein within the cell. Whilst ammonium
chloride exhibited similar properties, nigericin showed little or no effect on
APP processing (Lahiri
and Lahiri).
Along the same lines, it is interesting that the basal levels of APPs are higher in synaptosomes
derived from the cortex and hippocampus of methylazoxymethanol-treated rats, in
which PKC is permanently hyperactivated in these brain areas, while synaptosomes
derived from other areas in the same animals show similar levels to control rats
(Caputi
et al., 1997). In A431 cells, activation of the epidermal growth factor
(EGF) increases the secretion of APPs through a joint PKC/tyrosine
kinase mechanism (Slack
et al., 1997). Elevations of intracellular Ca2+ levels in turn
may be responsible for the tyrosine kinase-dependent release of APPs (Petryniak
et al., 1996). Intracellular cAMP inhibits both the constitutive and the
phorbol ester-mediated secretory cleavage of the amyloid precursor protein and,
therefore, serves to modulate the PKC pathway (Efthimiopoulos
et al., 1996).
The observation that skin fibroblasts from controls, people with Down's
syndrome (DS), and patients with a deletion in chromosome 21 show that PKC
stimulation of APPs
release is inversely proportional to the expression level of the APP holoprotein (Govoni
et al., 1996). This PKC-dependent secretory mechanism has been shown to be
defective in skin fibroblasts derived from individuals with sporadic AD due to
altered levels of PKC (Bergamaschi
and Ibarreta),
which is one of the major PKC isoforms present in skin fibroblasts, as well as
in the brain (Ibarreta
et al., 1999). This, in turn, implies a role for these isoforms in the PKC
stimulation of APPs
release. Studies with benzolactam, a novel PKC activator with improved
selectivity for the PKC, -, and - isoforms, demonstrated an enhanced secretion
of APP from fibroblasts of AD
patients and in PC12 cells (Ibarreta
et al., 1999), thus implying an additional role for PKC- (in addition to those previously reported for
PKC and PKC). Use of a specific inhibitor of PKC, Go-6976, suggests a partial
dependence of PKC on constitutive APPs secretion from human
fibroblasts, but a complete dependence of the phorbol ester-mediated PKC
stimulation of APPs
release (Benussi
et al., 1998). The role of PKC is further confirmed by the observation that
the proliferative state of the cells also affects their secretion of APPs, with differentiation
of rat primary neurons resulting in an increase in basal levels of secretion
coincident with the appearance of PKC expression (Salinetti
et al., 1996).
Treatment of cells or synaptosomal membrane preparations with phorbol esters
has been shown to increase the secretion of APPs (Ibarreta
and McLaughlin)
and to inhibit the production of A (Koo,
1997). This process is not associated with the direct phosphorylation of
APP, as the removal of the
cytoplasmic tail of this protein, which is the region of APP likely to be phosphorylated (Ando
et al., 1999), had no effect on this phorbol ester-mediated induction of
protein secretion. The deletion, however, did result in a significant increase
in the basal level of APPs secretion (Koo
et al., 1996), possibly by reducing the rate of APP internalisation, thus emphasising the
importance of the endocytic pathway in the processing of APP to A (Section
3.3). The PKC-associated increase in APPs generation is not due to an
increase in the resident time of APP at the cell surface (Koo,
1997). There are, therefore, two hypotheses that may explain the effects of
phorbol esters on APPs
secretion: (1) PKC enhances the activity of the -secretase enzyme(s) and/or (2) PKC shuttles
APP into the resident compartment
of the -secretase.
The observation that the activation of PKC prevents potocytosis
(internalisation of caveolae) (Section
3.2), which in turn leads to an increase in the secretion of APPs, would seem to suggest the
caveolae as being the potential resident compartment of the -secretase enzyme (Section
3.2). The effect of PKC on the secretion of APPs is also associated with the
budding of secretory vesicles from the Golgi network (Xu
et al., 1995). Furthermore, there is evidence that PKC may also directly
activate the -secretase activity
present within the proteasome multicatalytic complex (Marambaud
and Marambaud)
and that PKC-stimulated APPs secretion may be attenuated by
the inhibition of a hydroxamic acid-based inhibitor-sensitive protease (Racchi
et al., 1999b). The effect of PKC activation on APPs generation is also influenced by
other factors, including the cellular differentiation state, as well as the
quantity of APP within the cell
(Loffler
and Racchi).
Decreased cellular APP associated
with a deletion in chromosome 21 resulted in an increased response to PKC
activation when compared with control fibroblasts. In contrast, there was an
attenuation of the response in fibroblasts from DS patients, who have an
increased level of APP (Racchi
et al., 1999a). An activation of mitogen-activated protein kinase (MAPK) has
also been implicated as a PKC effector (DesdouitsMagnen
et al., 1998).
M1 muscarinic acetylcholine receptors (mAChRs) are thought to be predominantly localised postsynaptically, whereas a population of the M2 mAChR autoreceptors exist presynaptically and regulate acetylcholine release (Hulme et al., 1990). Reduced numbers of M2 receptors in AD brains have been reported in numerous studies (Svensson and Greenamyre). This is consistent with these receptors being present on degenerating cholinergic nerve terminals. However, M2 receptors have also been reported to exist postsynaptically (Dawson et al., 1990), with this expression potentially masking even greater reductions in the presynaptic M2 receptor levels. The levels of M1 receptor seem to remain unchanged in AD, although there is some evidence to suggest that the interactions between the M1 receptor with its associated G-proteins, and the subsequent activation of phosphoinositide signaling, may be impaired in AD (Ferraridileo and Jope).
Muscarinic receptor activation has both PKC-associated and tyrosine
phosphorylation-dependent components (Slack
et al., 1995). The M1, M3, and M5 mAChRs are associated with the
phosphoinositide cascade and stimulate PI hydrolysis through the activation of
PLC, which in turn can stimulate PKC, and PLA2, with a subsequent
increase in APPs
secretion (Nitsch;
Emmerling
and Wolf).
The stimulation of other PKC-coupled neurotransmitter receptors (glutamatergic
and serotonergic subtypes) also increases the secretion of APPs (Lee
and Nitsch).
These observations point to the existence of a system that should be capable of
regulating APP catabolism by both
PKC-dependent and PKC-independent mechanisms, with the MAPK signalling pathway
satisfying both of these conditions (Cobb
and Malarkey).
The tyrosine phosphorylation component of the muscarinic effect is mediated
via an elevation of intracellular Ca2+, suggesting that this increase
may mediate the tyrosine phosphorylation-dependent release of APPs in response to carbachol (Petryniak
et al., 1996). MAPK enzymes, also described as ERKs, are the terminal
enzymes in a three-level kinase cascade involving the sequential activation of
raf, MAPK, and ERK (Pelech
& Charest, 1996). It is worth noting that PS1 (Section
4.5) has also been shown to have a consensus sequence for ERK-dependent
phosphorylation, and may serve as a substrate for PKC (Seeger
and Walter).
Activation of the MAPK pathway was found to be necessary for the regulation of
the secretory processing of APP
(Mills
et al., 1997).
The activation of G-protein-coupled receptors can result in the increased
tyrosine phosphorylation of the paxillin protein. This protein is involved in
the formation of focal adhesions, which are complexes of cytoskeletal-associated
proteins that are localised to the integrin-extracellular matrix-binding sites
(Jockusch
et al., 1995). APPs release may occur in conjunction
with a cytoskeletal reorganisation associated with the tyrosine phosphorylation
of paxillin (Zachary
and Rankin).
This hypothesis is supported by a direct relationship between the strength of
cell adherence and the release of APPs (Monning
et al., 1995), and is consistent with the function of APP as a mediator of cell adhesion (Breen
and Gillian).
Since muscarinic receptor-associated activation of PKC has been shown to
reduce the production of A, a
decrease in the density of the mAChR may also attenuate the release of APPs. It is also interesting
that the mAChR subtypes, which stimulate the secretion of APPs, are enriched in those brain
areas that are most affected by AD. This has led to the hypothesis that impaired
signalling through cortical/hippocampal M1/M3 mAChR may play a role in the
pathological processing of APP (Robner
et al., 1998). The increase in APPs secretion by phorbol esters, and
the associated decrease in A
generation, can be observed with all of the APP isoforms (Buxbaum
and Caporaso).
However, these effects were seen to be both species- and cell type-specific,
indicating that the use of PKC activators alone may not be a viable strategy to
reduce human brain A levels (LeBlanc
et al., 1998). However, the treatment of AD with M1-specific agonists may be
beneficial, although to date, selective agonists have been difficult to develop.
This lack of specificity may result in a stimulation of the presynaptic M2
receptors, with a consequent decrease in neocortical acetylcholine release.
NGF has been shown to increase the level of total APP mRNA, with a particular elevation of the
APP695 isoform and a
parallel up-regulation of APPs secretion (Rossner
et al., 1998). This effect was originally reported in PC12 cells (Fukuyama
et al., 1993), in mouse neuronal cells in culture (Ohyagi
& Tabira, 1993), and in hippocampal explant cultures (Clarris
et al., 1994). While antibodies against NGF block this effect, the addition
of exogenous APP mimics it, but
only in the presence of heparin sulfate proteoglycans (Ohyagi
& Tabira, 1993). In NGF-deprived rat superior cervical ganglion cells
and primary dissociated cultures of sympathetic neurons, neuronal degeneration
has been associated with a decrease in APP695 and an increase in APP751/APP770 (Smith
et al., 1993). The effect of NGF on A generation has not been clarified, although
maintenance of PC12 cells in serum-free medium in the absence of NGF results in
an increase in the generation of an amyloidogenic CTF of APP (Baskin
et al., 1991). NGF is believed to stimulate APPs production via the
p75NTR receptor subunit, the activation of sphingomyelinase, and the
subsequent generation of ceramide and the activation of nuclear factor B (NFB). Activation of the TrkA receptor results in
a suppression of APP mRNA
expression via the p21Ras-activated signalling pathways involving MAPKs and
PI3 kinase, whilst there is a concomitant increase in the secretory
processing via the PLC
transduction cascade (Robner
et al., 1998). Other growth factors, such as EGF, also stimulate APPs release, although the
actions of EGF are mediated via PKC and phosphoinositide-specific PLC pathways (Slack
et al., 1997).
Thrombin is a multifunctional serine protease that has been detected in
amyloid plaques and has also been reported to be capable of cleaving APP at a site just upstream of the
amino-terminus of the A domain (Davis-Salinas
et al., 1994). While this led to the suggestion that thrombin may play a
role in the generation of the A
peptide, subsequent studies reported that high concentrations of thrombin
actually increased the generation of APPs, although at low concentrations,
it actually acts to break down APPs. The altered proteolysis of
APPs may result in the
abolition of the neurotrophic/neuroprotective properties of APPs, and therefore, thrombin may be
indirectly involved in the pathogenesis of AD through this proteolytic activity
(Davis-Salinas
et al., 1994).
There is increasing evidence that plaque-associated activated microglia and
microglial-derived cytokines are important pathogenic factors in the progression
of the neuropathological changes associated with AD. In particular, they are
thought to act to potentiate the cell damage in the early stages of the disease,
and this has led to the suggestion that the AD pathology may have a
multi-factorial origin. In particular, the activated microglial cells
surrounding the plaque core may play a pivotal role in the neurodegenerative
process. While under normal conditions microglial cells are dormant, in AD, they
express cell-surface antigens that are characteristic of activated cells in the
chronic inflammatory state (Akiyama
et al., 1994). This activation may be due in part to the binding or
internalisation of microaggregates of the A peptide in an effort to clear the initial
depositions of the aggregated peptide (Giulian
and Paresce).
Furthermore, A has been
discovered to bind to the receptor for advanced glycation end products (RAGE)
present on the surface of microglial cells (Yan
et al., 1996). The presence of other glycated proteins in the AD
pathological lesions may also play a role in microglial activation by RAGE (Ledesma
and Vitek),
and inhibitors of advanced glycation (e.g., tenilsetam) have demonstrated some
evidence of cognitive enhancement in patients with AD (Munch
et al., 1994).
The microglial cells, however, are associated with the mature plaques seen in
the later stages of neurodegeneration rather than with the earlier diffuse
plaques, suggesting that their activation is not an early event in the
development of the disease pathology (Mackenzie
et al., 1995). While activated microglia may also play a key role in the
increased generation of the A
peptide from APP (Schaffer
et al., 1993), these cells have also been reported to take up A by endocytosis and to degrade the peptide by a
metalloprotease, which degrades soluble peptide prior to polymerisation (Paresce
and Mentlein).
This appears to be a saturable process, which may be overwhelmed by large
quantities of A, similar to those
produced in AD (Paresce
et al., 1997).
Astrocytes can also be activated by the microglial-derived interleukin (IL)-1
to generate the acute-phase protein 1-antichymotrypsin (Kanemaru
et al., 1996), in addition to additional cytokines and complement proteins
(Mrak
et al., 1995). Furthermore, the buildup of proteoglycans that is observed in
AD brains (Small
et al., 1996b) may play a role in the development of the inflammatory
response (Leveugle
& Fillit, 1994).
IL-1, which is produced by activated microglial cells, has numerous actions
within the CNS that may contribute to the neurodegeneration associated with AD.
It up-regulates the expression of APP in neuronal cells both in vivo (Brugg
et al., 1995) and in vitro (Buxbaum
and Rogers),
and this may give rise to an increase in the generation of the A peptide. There is also a positive feedback
loop, with the A peptide further
stimulating the production of both IL-1 and IL-6 by activated astroglial cells
(Delbo
and Lorton).
IL production is probably an early event in the disease progression, as raised
IL-6 levels have been detected in the immature diffuse senile plaques (Hull
et al., 1996). This acts to increase neuronal intracellular calcium levels
(Barger
& Van Eldik, 1992) and therefore, may potentiate the neurotoxic effects
of the A peptide, which also
raises intracellular calcium (Cotman
et al., 1992).
Transforming growth factor (TGF)- is a member of the cytokine family, produced
primarily by microglia in the CNS, that plays a role in the control of the
inflammatory response. TGF- is
expressed following a variety of insults in the brain, including AD-associated
cell death (Flanders
et al., 1995). It acts to increase both APP expression (Gray
& Patel, 1993) and A
peptide generation and accumulation, and it may play a role in the development
of AD pathology (Harris-White
et al., 1998). A similar effect can be observed with prostaglandin
E2, which exerts its actions by the activation of protein kinase A
(PKA) and an increase in cAMP levels (Lee
et al., 1999).
Other signalling systems also play a key role in the regulation of APP secretion. Activation of PKA, with
the subsequent generation cAMP, blocks phorbol ester-stimulated cleavage of
APPs in C6 glioma
cells (Efthimiopoulos
and Xu).
However, stimulation of the PKA pathway alone increases both the expression of
the APP holoprotein and the
subsequent release of APPs and A in human HEK293 cells (Lee
and Marambaud).
Treatment of these cells with the PKA inhibitors H89 or PKI results in a
decrease in basal production of A
without any effect on the generation of APPs. The apparent contradiction in
these results may be due to a differential effect of cAMP in individual cell
lines.
A significant number of diverse cell surface neurotransmitter receptors have
also been reported to regulate APP secretion, including the G-protein- and
tyrosine kinase-coupled receptors (Beyreuth
(6K)
(5K)
(9K)
(29K)