You are on page 1of 6

PERSPECTIVES

The properties of attractor neural net-


OPINION works can provide significant clues to poten-
tial sites of action for Aβ. For example, loss
of both neurons and synapses are prominent
Alzheimer’s disease and Aβ toxicity: features in post-mortem studies of AD12–14.
However, analyses of attractor networks that
from top to bottom feature a loss of neurons and the consequent
loss of synapses indicate that neuronal loss
does not account, by itself, for the properties
David H. Small, Su San Mok and Joel C. Bornstein of the amnesia characteristic of AD15,16. By
contrast, a loss of synapses, without an asso-
Evidence implicating the β-amyloid protein In this article, we argue that a more inte- ciated loss of neurons, can account for all the
(Aβ) in the pathogenesis of Alzheimer’s grated approach to brain function is needed features of AD, assuming that a mechanism
disease has steadily accumulated. However, to assess the role of Aβ in AD. Specifically, that acts to keep the average firing of each
the mechanism by which Aβ causes we suggest that a ‘top–down’ approach neuron constant over time is also operat-
dementia is unknown. Here we argue that a derived from computational neuroscience5 ing15,16. Recent physiological studies have
more integrated, top–down approach to can provide clues to identify biochemical indicated that such a mechanism (synaptic
brain function is needed to assess the role of pathways that are central to AD and worth scaling, see below) is widespread in the ner-
Aβ in Alzheimer’s disease, and that more targeting for therapy. vous system17–19. The loss of synapses in AD
attention should be paid to the effects of Aβ is substantially greater than could be
on synaptic function rather than on cell death. A top–down approach explained by the loss of neurons12, which
A key to understanding the actions of Aβ in indicates that a deficit of synapses might be a
Alzheimer’s disease (AD) is a progressive AD is to determine how this protein affects primary problem in the early phases of AD.
neurological disorder that is characterized the neuronal networks in which memories The implication of these studies is that the
by memory loss and confusion. There is are stored, thereby causing dementia. This biochemical targets of Aβ leading to cell
good evidence indicating that the accumula- depends on an understanding of how stable death (FIG. 1) could be less important for
tion of the β-amyloid protein (Aβ) is a pri- memories form within a network, and how early treatment than those that produce
mary event in the pathogenesis of AD1. Aβ is specific types of damage might lead to the selective toxic effects on synapses.
a 4-kDa polypeptide derived by the prote- amnesia that is typical of AD. This type of However, studies of attractor networks
olytic cleavage of the Aβ precursor protein amnesia is characterized by difficulty in indicate that the dementia that is character-
(APP)2. The main form of Aβ contains 40 acquiring new memories, the sparing of istic of AD might actually precede any overt
amino-acid residues. However, carboxy-ter- established memories early in the disease, and loss of synapses or neurons. For example,
minally extended species that contain 42 or the later, gradual loss of these memories6. studies of this sort indicate that synapse-spe-
43 residues (‘long Aβ’) are also produced2. The pioneering work of Hopfield7,8 showed cific changes in synaptic strength due to pre-
These long forms of Aβ aggregate more that interconnected networks of model neu- or postsynaptic activity — such as long-term
readily than the 40-amino-acid peptide. It is rons can form associative memories if the potentiation (LTP) or long-term depression
generally believed that aggregated Aβ is the strength of synaptic connections between (LTD) — cannot, by themselves, account for
agent that is responsible for disease progres- these neurons changes as a result of prior the formation of several stable memories
sion, as the accumulation of aggregates is activity. In these computational models, asso- within one network19–22. This is because
toxic to the brain1. The fact that many famil- ciative memories are coded by the strength of inputs that excite overlapping parts of the
ial AD mutations increase the production of individual synaptic connections — the ampli- network will tend to be attracted to the
long Aβ (REF. 3) provides strong support for tudes of synaptic potentials — between differ- ‘strongest’ pattern of synapses, strengthening
this idea. ent neurons, and are relatively insensitive to it and swamping newer or weaker memo-
One of the central unanswered questions in the loss of individual cells. Furthermore, indi- ries. Studies of attractor networks also indi-
AD is how Aβ accumulation in the brain caus- vidual neurons can be involved in several cate that synaptic scaling can compensate for
es cognitive decline. This question has generally memories, a finding that has been confirmed this phenomenon, thereby allowing a net-
been addressed by using the ‘bottom–up’ strat- in electrophysiological studies9,10. work to store many stable memories17–22.
egy. In other words, the direct biochemical Networks of this type have been termed Synaptic scaling is a slower homeostatic
effects of Aβ are examined in vitro with the attractor neural networks11. They learn mechanism by which the strength of all
hope that they will help us to elucidate the because any given input pattern leads to a synaptic input to a neuron is regulated by
mechanism of neurotoxicity. This approach strengthening of specific synapses on some the overall activity of that neuron over time.
has led to the identification of numerous bio- neurons in the network. If an input is repeat- High levels of firing lead to all synapses on a
chemical pathways and mechanisms that are ed, the synaptic strengthening will ultimately neuron becoming weaker, whereas low levels
modified by Aβ (FIG. 1). However, it has been impose a pattern of synaptic strengths onto of firing increase the strength of these
difficult to determine which of these pathways the network, which ensures that inputs similar synapses. These changes are independent of
are the most important in the onset of AD, and to the training input will produce the same the synapse-specific mechanisms of LTP and
which should be targeted for the development output pattern. The network will be ‘attracted’ LTD, and act by preventing already strong
of new therapeutic agents. It seems unlikely to a common output pattern for related synapses from becoming even stronger.
that all the biochemical effects of Aβ are associ- inputs, so that part of a memory can be used Synaptic scaling seems to result, at least in
ated with dementia and, in fact, it is not clear to recall the whole, a characteristic of explicit part, from the insertion and removal of the
that cell death is the most relevant factor4. memory that is specifically vulnerable in AD. α-amino-3-hydroxy-5-methyl-4-isoxazole

NATURE REVIEWS | NEUROSCIENCE VOLUME 2 | AUGUST 2001 | 5 9 5


© 2001 Macmillan Magazines Ltd
PERSPECTIVES

Neurotoxicity
Astrocyte Peroxynitrites Neuron
85

iNOS NO 58
Na+/K+ ATPase
Aβ + Cu2+ p35/cdk5
NFκB Ion channel 56,57
59
Calpain
55
39,40
Ca2+ Ca2+
86 L-VGCC 41–43 p25/cdk5

ROS 54

Other 65
TNFα IL1β ATP
cytokines
62,63
p75NTR
60
APP
86 68
ER stress Caspase 12
NFκB RAGE
77
66,67
FAS/TNFR/etc FADD Caspase 8
79 87
Tyrosine ERAB
kinases 76 Apoptosis/
Aβ necrosis
64
Endosomal lysosomal
Class A 82–84 modification
scavenger 51–53
receptor 54
77 78 ROS
RAGE
78
cfms m-CSF NFκB
61
Aβ Binding to membrane

80,81
74,75
Other proteins?
cdk4/6 Rb Bax
E2F1
70–72
c-jun
73 Caspase 3
JNK Caspase 2 Other caspases?
69
NO
ROS Peroxynitrite
cytokines TNFR
Microglia RNS
e.g. TNFα
Neurotoxicity
Figure 1 | Pathways in neurons, astrocytes and microglia that are implicated in β-amyloid toxicity. The numbers shown in the figure refer to the
appropriate references in the bibliography. APP, Aβ precursor protein; Aβ, β-amyloid; Bax, B-cell-leukemia/lymphoma-2-associated X protein; cdk, cyclin-
dependent kinase; cfms, m-CSF receptor; E2F1, E2F transcription factor 1; ER, endoplasmic reticulum; ERAB, endoplasmic-reticulum-associated binding
protein; FADD, FAS-associated protein with death domain; FAS/TNFR, FAS/tumour necrosis factor receptor; IL1β, interleukin 1β; iNOS, inducible nitric oxide
synthase; JNK, c-Jun N-terminal kinase; L-VGCC, L-type voltage-gated Ca2+ channel; m-CSF, macrophage-colony-stimulating factor; NFκB, nuclear factor κB;
NO, nitric oxide; p25, calpain cleavage product of p35; p35, cyclin-dependent kinase 5 regulatory sumit; p75NTR, 75 kDa neurotrophin receptor; RAGE, receptor
for advanced glycation end products; Rb, retinoblastoma protein; RNS, radical nitrogen species; ROS, radical oxygen species; TNFα, tumour necrosis factor α.

propionic acid (AMPA) subtype of gluta- in a narrow range, the number of memories potentials and synaptic potentials.
mate receptor into subsynaptic mem- that can be stored within a network is severely The timing of pre- and postsynaptic
branes17–19. The expression of this class of limited and independent of the size of the activity depends on dendritic action poten-
receptor is indeed altered in AD23, indicating network20. So, these theoretical studies indi- tials evoked in the postsynaptic neuron by
that the mechanisms regulating synaptic cate that a selective change in either LTP or the generation of action potentials at the
scaling might be a target of Aβ in AD. LTD, resulting from Aβ accumulation, would axon hillock29. These have been identified in
Computational studies have also pointed prevent the conversion of short-term memo- both in vitro29 and in vivo studies30.
to a second mechanism by which the forma- ries into stable long-term memories — that is, Electrophysiological analyses in neocortical
tion of memories within a network can be anterograde amnesia. This would be seen, slices indicate that, if a presynaptic action
stabilized. This mechanism depends on the regardless of whether the process affected is potential precedes a dendritic action poten-
balance between LTP and LTD induced at dif- enhanced or depressed by Aβ. tial by a few milliseconds, then LTP is
ferent synapses on the same neuron by specif- In addition to the biochemical pathways induced at that synapse27. If it trails a den-
ic patterns of pre- and postsynaptic firing. that mediate LTP and LTD24–26, the balance dritic action potential by the same margin,
Each time a neuron fires, there will be an of LTP and LTD in neocortical and hip- then LTD is induced27. This spike-timing-
increment of LTP at some synapses and LTD pocampal pyramidal cells depends on two dependent plasticity can form a stable
at others, keeping the overall strength of the factors. First, the relative timing of pre- and memory system28,31, but anything that pro-
inputs to the neuron relatively constant. postsynaptic activity at individual synaps- duces a change in dendritic length would
However, computational studies indicate that, es27,28, and second, the entry of Ca2+ during upset this stability by altering timing. The
unless the ratio of these two processes is with- temporally associated dendritic action Aβ deposits that are characteristic of AD

596 | AUGUST 2001 | VOLUME 2 www.nature.com/reviews/neuro


© 2001 Macmillan Magazines Ltd
PERSPECTIVES

AChE inhibitors
modify the shape and length of pyramidal AD, and not all will be worthwhile targeting
cell dendrites32, and biophysically realistic AChE Choline + acetate for therapy. The results of neural network
simulations indicate that these changes are studies provide insights into how memories
sufficient to modify the relative timing of ACh Aβ can be stored, and give us important clues to
pre- and postsynaptic action potentials32. Ca2+
the possible ways in which Aβ could act to
45–49 41–43 Ca2+
So, computational studies indicate that the deteriorate memory storage. Future studies
effects of Aβ on dendritic remodelling — a should be aimed at testing the predictions
continuous process in vivo33 — could be of generated by computational studies. Even
considerable importance in AD. though the discipline of computational neu-
α7 L-VGCC
The entry of Ca2+ during temporally asso- Ca2+
roscience is still in its infancy, certain concepts
nAChR
ciated dendritic action potentials and synap- are already clear. For example, theoretical
43
tic potentials is a second key factor. This is studies show that the effects of Aβ on cogni-
crucial for the development of both LTP and LTP tion are unlikely to be associated with cell
AChE
LTD, and depends, in large part, on L-type loss. The theory also indicates that the best
Ca2+ channels29,34, which are presumed to be MAP kinase 25 targets for therapeutic intervention are mech-
located on dendritic spines35. Aβ has been 44 anisms that alter synaptic plasticity.
shown to stimulate this type of Ca2+ channel Ultimately, however, the proof of any hypoth-
Cytoskeletal remodelling
(FIGS 1 and 2), either by a direct action on the esis concerning the role of Aβ in AD will rest
and tau phosphorylation
channel itself or through another receptor on the therapeutic efficacy of drugs that
(such as the α7 nicotinic acetylcholine recep- Dendritic spine specifically target the proposed mechanisms.
tor (α7 nAChR) — see below). So, distur- David H. Small*, Su San Mok* and Joel C.
Figure 2 | Effects of β-amyloid on pathways
bances in Ca2+ handling in dendrites and that control synaptic plasticity in dendrites. Bornstein‡ are at the Departments of *Pathology
spines could also be an important locus of β-Amyloid (Aβ) can alter Ca2+ concentrations in and ‡Physiology, The University of Melbourne,
Aβ action. dendrites, either by binding directly to α7 nicotinic Victoria 3010, Australia.
acetylcholine receptors (α7 nAChRs), or by
Correspondence to D.H.S.
e-mail: davidhs@unimelb.edu.au
How could Aβ influence plasticity? interacting directly or indirectly with L-type
As discussed extensively by Arendt36, there is voltage-gated Ca2+ channels (L-VGCC). Increased
intracellular Ca2+ could influence long-term
substantial evidence that Aβ has selective Links
potentiation (LTP) or long-term depression (LTD)
actions on synapses. Furthermore, Phinney directly; alternatively, by activating extracellular- DATABASE LINKS Alzheimer’s disease | APP |
et al.37 have documented aberrant axonal signal-regulated kinase/mitogen-activated protein α7 nAChR | ERK/MAP kinase | tau | Bax | cdk |
sprouting and ectopic terminal formation in kinase (ERK/MAP kinase), it might affect LTP and
E2F1 | ERAB | FADD | FAS/TNFR | IL1β |
APP transgenic mice. Nevertheless, the LTD, and alter dendritic architecture through
iNOS | JNK | L-VGCC | m-CSF | NFκB | p35 |
mechanism by which Aβ affects synaptic cytoskeletal remodelling. The numbers shown in
the figure refer to the corresponding references in p75NTR | RAGE | Rb | TNFα
plasticity remains a matter of conjecture. MIT ENCYCLOPEDIA OF COGNITIVE SCIENCES
the bibliography. ACh, acetylcholine; AChE,
Neural computation studies indicate the acetylcholinesterase. Computational neuroscience
type of physiological change that could
cause AD-type amnesia, but they do not
1. Small, D. H. & McLean, C. A. Alzheimer’s disease and the
indicate which of these are most important tein kinase (ERK/MAP kinase) cascade is also amyloid β protein: what is the role of amyloid?
in vivo. However, computational studies do known to be an important signalling pathway J. Neurochem. 73, 443–449 (1999).
2. Nunan, J. & Small, D. H. Regulation of APP cleavage by α-,
point to a number of questions that can be for synaptic plasticity25, and increased β- and γ-secretases. FEBS Lett. 483, 6–10 (2000).
tested experimentally. For example, what are ERK/MAP kinase is associated with tau phos- 3. Scheuner, D. et al. Secreted amyloid β-protein similar to
that in the senile plaques of Alzheimer’s disease is
the effects of Aβ on the balance between LTP phorylation and the cytoskeletal abnormali- increased in vivo by the presenilin 1 and 2 and APP
mutations linked to familial Alzheimer’s disease. Nature
and LTD in a single neuron? What effect ties found in dystrophic neurites in AD44. Med. 2, 864–870 (1996).
does Aβ have on dendritic remodelling in Recently, Aβ has been shown to activate the 4. Terry, R. D. Cell death or synaptic loss in Alzheimer
disease. J. Neuropathol. Exp. Neurol. 59, 1118–1119
vivo? What is the effect of Aβ on AMPA MAP kinase cascade in hippocampal neurons (2000).
receptor recycling? These questions remain through the α7 nAChR45. Aβ binds very 5. National Institute on Drug Abuse. Computational
neuroscience at the NIH. Nature Neurosci. 3, 1161–1164
unanswered, and should be the focus of strongly to this receptor46–48, although the (2000).
future research. precise effect of the interaction between Aβ 6. McClelland, J. L. & Goddard, N. H. Considerations arising

Not only do neural computation studies and α7 nAChRs might be complex49. For from a complementary learning systems perspective on
hippocampus and neocortex. Hippocampus 6, 654–665
point to promising experiments, but they also example, Aβ could partially block receptor (1996).
7. Hopfield, J. J. Neural networks and physical systems with
indicate that certain known actions of Aβ activation, but it might also inhibit receptor emergent collective computational abilities. Proc. Natl
might be important for the cognitive impair- desensitization49 (FIG. 2). As nicotinic agonists Acad. Sci. USA 79, 2554–2558 (1982).
8. Hopfield, J. J. Neurons with graded response have
ments in AD. For example, altering Ca2+ levels protect neurons from Aβ-induced neurotoxi- collective computational properties like those of two-state
through L-type Ca2+ channels on dendritic city50, the therapeutic effects of cholinesterase neurons. Proc. Natl Acad. Sci. USA 81, 3088–3092 (1984).
9. Sakurai, Y. Involvement of auditory cortical and
spines34,35 would be expected to influence inhibitors in AD patients could be due to hippocampal neurons in auditory working memory and
LTP29,34 and dendritic remodelling38 (FIG. 2). increased nicotinic receptor activation (FIG. 2). reference memory in the rat. J. Neurosci. 14, 2606–2623
(1994).
Several studies have shown that Aβ can pro- 10. Sakurai, Y. How do cell assemblies encode information in
mote Ca2+ entry through L-type channels39–43, Testing the role of Aβ in AD the brain? Neurosci. Biobehav. Rev. 23, 785–796 (1999).
11. Amit, D. J. & Treves, A. Associative memory neural network
although the mechanism by which this Aβ undoubtedly has multiple effects on neu- with low temporal spiking rates. Proc. Natl Acad. Sci. USA
86, 7871–7875 (1989).
occurs is still unknown. The extracellular-sig- rons. However, not all of these effects will 12. Davies, C. A., Mann, D. M., Sumpter, P. Q. & Yates, P. O.
nal-regulated kinase/mitogen-activated pro- contribute to the cognitive changes seen in A quantitative morphometric analysis of the neuronal and

NATURE REVIEWS | NEUROSCIENCE VOLUME 2 | AUGUST 2001 | 5 9 7


© 2001 Macmillan Magazines Ltd
PERSPECTIVES

synaptic content of the frontal and temporal cortex in homeostasis and render human cortical neurons vulnerable 64. Kane, M. D. et al. Inhibitors of V-type ATPases, bafilomycin
patients with Alzheimer’s disease. J. Neurol. Sci. 78, to excitotoxicity. J. Neurosci. 12, 376–389 (1992). A1 and concanamycin A, protect against β-amyloid-
151–164 (1987). 41. Weiss, J. H., Pike, C. J. & Cotman, C. W. Ca2+ channel mediated effects on 3-(4,5-dimethylthiazol-2-yl)-2,5-
13. Esiri, M. M., Pearson, R. C. A., Steele, J. E., Bowen, D. M. & blockers attenuate β-amyloid peptide toxicity to cortical diphenyltetrazolium bromide (MTT) reduction.
Powell, T. P. A quantitative study of the neurofibrillary tangles neurons in culture. J. Neurochem. 62, 372–375 (1994). J. Neurochem. 72, 1939–1947 (1999).
and the choline acetyltransferase activity in the cerebral 42. Ueda, K., Shinohara, S., Yagami, T., Asakura, K. & 65. Zhang, Z. et al. Amyloid β-mediated oxidative and
cortex and the amygdala in Alzheimer’s disease. Kawasaki, K. Amyloid β protein potentiates Ca2+ influx metabolic stress in rat cortical neurons: no direct evidence
J. Neurol. Neurosurg. Psychiatry 53, 161–165 (1990). through L-type voltage-sensitive Ca2+ channels: a possible for a role for H2O2 generation. J. Neurochem. 67,
14. Pearson, R. C. A., Esiri, M. M., Hiorns, R. W., Wilcock, G. K. involvement of free radicals. J. Neurochem. 68, 265–271 1595–1606 (1996).
& Powell, T. P. Anatomical correlates of the distribution of the (1997). 66. Ivins, K. J., Ivins, J. K., Sharp, J. P. & Cotman, C. W.
pathological changes in the neocortex in Alzheimer disease. 43. Sberna, G., Sáez-Valero, J., Beyreuther, K., Masters, C. L. & Multiple pathways of apoptosis in PC12 cells. CrmA inhibits
Proc. Natl Acad. Sci. USA 82, 4531–4534 (1985). Small, D. H. The amyloid-β protein of Alzheimer’s disease apoptosis induced by β-amyloid. J. Biol. Chem. 274,
15. Horn, D., Levy, N. & Ruppin, E. Neuronal-based synaptic increases acetylcholinesterase expression by increasing 2107–2112 (1999).
compensation: a computational study in Alzheimer’s intracellular calcium in embryonal carcinoma P19 cells. 67. Ivins, K. J., Thornton, P. L., Rohn, T. T. & Cotman, C. W.
disease. Neural Comput. 8, 1227–1243 (1996). J. Neurochem. 69, 1177–1184 (1997). Neuronal apoptosis induced by β-amyloid is mediated by
16. Ruppin, E. & Reggia, J. A. A neural model of memory 44. Ferrer, I. et al. Phosphorylated Map kinase (ERK1, ERK2) caspase-8. Neurobiol. Dis. 6, 440–449 (1999).
impairment in diffuse cerebral atrophy. Br. J. Psychiatry 166, expression is associated with early tau deposition in 68. Nakagawa, T. et al. Caspase-12 mediates endoplasmic-
19–28 (1995). neurones and glial cells, but not with increased nuclear DNA reticulum-specific apoptosis and cytotoxicity by amyloid-β.
17. Abbott, L. F. & Nelson, S. B. Synaptic plasticity: taming the vulnerability and cell death, in Alzheimer disease, Pick’s Nature 403, 98–103 (2000).
beast. Nature Neurosci. 3, 1178–1183 (2000). disease, progressive supranuclear palsy and corticobasal 69. Troy, C. M. et al. Caspase-2 mediates neuronal cell death
18. Turrigiano, G. G. Homeostatic plasticity in neuronal degeneration. Brain Pathol. 11, 144–158 (2001). induced by β-amyloid. J. Neurosci. 20, 1386–1392 (2000).
networks: the more things change, the more they stay the 45. Dineley, K. T. et al. β-Amyloid activates the mitogen- 70. Anderson, A. J., Pike, C. J. & Cotman, C. W. Differential
same. Trends Neurosci. 22, 221–227 (1999). activated protein kinase cascade via hippocampal α7 induction of immediate early gene proteins in cultured
19. Turrigiano, G. G. & Nelson, S. G. Hebb and homeostasis in nicotinic acetylcholine receptors: in vitro and in vivo neurons by β-amyloid (Aβ): association of c-Jun with Aβ-
neuronal plasticity. Curr. Opin. Neurobiol. 10, 358–364 mechanisms related to Alzheimer’s disease. J. Neurosci. 21, induced apoptosis. J. Neurochem. 65, 1487–1498 (1995).
(2000). 4125–4133 (2001). 71. Estus, S. et al. Aggregated amyloid-β protein induces
20. Chechik, G., Meilijson, I. & Ruppin, E. Effective neuronal 46. Wang, H. Y. et al. β-Amyloid1–42 binds to α7 nicotinic cortical neuronal apoptosis and concomitant ‘apoptotic’
learning with ineffective Hebbian learning rules. Neural acetylcholine receptor with high affinity. J. Biol. Chem. 275, pattern of gene induction. J. Neurosci. 17, 7736–7745
Comput. 13, 817–840 (2001). 5626–5632 (2000). (1997).
21. Horn, D., Levy, N. & Ruppin, E. Memory maintenance via 47. Liu, Q. S., Kawai, H. & Berg, D. K. β-Amyloid peptide blocks 72. Kihiko, M. E., Tucker, H. M., Rydel, R. E. & Estus, S. c-Jun
neuronal regulation. Neural Comput. 10, 1–18 (1998). the response of α7-containing nicotinic receptors on contributes to amyloid β-induced neuronal apoptosis but is
22. Horn, D., Levy, N. & Ruppin, E. Neuronal regulation versus hippocampal neurons. Proc. Natl Acad. Sci. USA 98, not necessary for amyloid β-induced c-jun induction.
synaptic unlearning in memory maintenance mechanisms. 4734–4739 (2001). J. Neurochem. 73, 2609–2612 (1999).
Netw. Comput. Neural Syst. 9, 577–586 (1998). 48. Pettit, D. L., Shao, Z. & Yakel, J. L. β-Amyloid1–42 peptide 73. Troy, C. M. et al. β-Amyloid-induced neuronal apoptosis
23. Harrison, P. J., Barton, A. J., Najlerahim, A. & Pearson, R. C. directly modulates nicotinic receptors in the rat hippocampal requires c-Jun N-terminal kinase activation. J. Neurochem.
A. Distribution of a kainate/AMPA receptor mRNA in normal slice. J. Neurosci. 21, RC120, 1–5 (2001). 77, 157–164 (2001).
and Alzheimer brain. Neuroreport 1, 149–152 (1990). 49. Cheung, N. S., Small, D. H. & Livett, B. G. An amyloid 74. Giovanni, A., Wirtz-Brugger, F., Keramaris, E., Slack, R. &
24. Bear, M. F. Mechanism for a sliding synaptic modification peptide, βA4 25–35, mimics the function of substance P on Park, D. S. Involvement of cell cycle elements, cyclin-
threshold. Neuron 15, 1–4 (1995). modulation of nicotine-evoked secretion and desensitization dependent kinases, pRb, and E2F·DP, in β-amyloid-induced
25. Impey, S., Obrietan, K. & Storm, D. R. Making new in cultured bovine adrenal chromaffin cells. J. Neurochem. neuronal death. J. Biol. Chem. 274, 19011–19016 (1999).
connections: role of ERK/MAP kinase signaling in neuronal 60, 1163–1166 (1993). 75. Giovanni, A. et al. E2F1 mediates death of B-amyloid-
plasticity. Neuron 23, 11–14 (1999). 50. Kihara, T. et al. α7 nicotinic receptor transduces signals to treated cortical neurons in a manner independent of p53
26. Di Cristo, G. et al. Requirement of ERK activation for visual phosphatidylinositol 3-kinase to block a β-amyloid-induced and dependent on Bax and caspase 3. J. Biol. Chem. 275,
cortical plasticity. Science 292, 2337–2340 (2001). neurotoxicity. J. Biol. Chem. 276, 13541–13546 (2001). 11553–11560 (2000).
27. Markram, H., Lübke, J., Frotscher, M. & Sakmann, B. 51. Behl, C., Davis, J. B., Lesley, R. & Schubert, D. Hydrogen 76. Loo, D. T. et al. Apoptosis is induced by β-amyloid in
Regulation of synaptic efficacy by coincidence of peroxide mediates amyloid β protein toxicity. Cell 77, cultured central nervous system neurons. Proc. Natl Acad.
postsynaptic APs and EPSPs. Science 275, 213–215 817–827 (1994). Sci. USA 90, 7951–7955 (1993).
(1997). 52. Schubert, D. et al. Amyloid peptides are toxic via a common 77. Yan, S. D. et al. RAGE and amyloid-β peptide neurotoxicity
28. Van Rossum, M. C. W., Bi, G. Q. & Turrigiano, G. G. Stable oxidative mechanism. Proc. Natl Acad. Sci. USA 92, in Alzheimer’s disease. Nature 382, 685–691 (1996).
hebbian learning from spike timing-dependent plasticity. 1989–1993 (1995). 78. Yan, S. D. et al. Amyloid-β peptide-receptor for advanced
J. Neurosci. 20, 8812–8821 (2000). 53. Pike, C. J., Ramezan-Arab, N. & Cotman, C. W. β-Amyloid glycation endproduct interaction elicits neuronal expression
29. Magee, J. C. & Johnston, D. Synaptic activation of voltage- neurotoxicity in vitro: evidence of oxidative stress but not of macrophage-colony stimulating factor: a proinflammatory
gated channels in the dendrites of hippocampal pyramidal protection by antioxidants. J. Neurochem. 69, 1601–1611 pathway in Alzheimer disease. Proc. Natl Acad. Sci. USA
neurons. Science 275, 209–213 (1997). (1997). 94, 5296–5301 (1997).
30. Kamondi, A., Acsady, L. & Buzsaki, G. Dendritic spikes are 54. Hensley, K. et al. A model for β-amyloid aggregation and 79. Combs, C. K., Karlo, J. C., Kao, S. C. & Landreth, G. E. β-
enhanced by cooperative network activity in the intact neurotoxicity based on free radical generation by the Amyloid stimulation of microglia and monocytes results in
hippocampus. J. Neurosci. 18, 3919–3928 (1998). peptide: relevance to Alzheimer disease. Proc. Natl Acad. TNFα-dependent expression of inducible nitric oxide
31. Kistler, W. M. & Van Hemmen, J. L. Modeling synaptic Sci. USA 91, 3270–3274 (1994). synthase and neuronal apoptosis. J. Neurosci. 21,
plasticity in conjunction with the timing of pre- and 55. White, A. R., Bush, A. I., Beyreuther, K., Masters, C. L. & 1179–1188 (2001).
postsynaptic action potentials. Neural Comput. 12, 385–405 Cappai, R. Exacerbation of copper toxicity in primary 80. McDonald, D. R., Bamberger, M. E., Combs, C. K. &
(2000). neuronal cultures depleted of cellular glutathione. Landreth, G. E. β-Amyloid fibrils activate parallel mitogen-
32. Knowles, R. B. et al. Plaque-induced neurite abnormalities: J. Neurochem. 72, 2092–2098 (1999). activated protein kinase pathways in microglia and THP1
implication for disruption of neural networks in Alzheimer’s 56. Kawahara, M. & Kuroda, Y. Molecular mechanism of monocytes. J. Neurosci. 18, 4451–4460 (1998).
disease. Proc. Natl Acad. Sci. USA 96, 5274–5279 (1999). neurodegeneration induced by Alzheimer’s β-amyloid 81. Combs, C. K., Johnson, D. E., Cannady, S. B., Lehman, T.
33. Purves, D., Voyvodic, J. T., Magrassi, L. & Yawo, H. Nerve protein: channel formation and disruption of calcium M. & Landreth, G. E. Identification of microglial signal
terminal remodeling visualized in living mice by repeated homeostasis. Brain Res. Bull. 53, 389–397 (2000). transduction pathways mediating a neurotoxic response to
examination of the same neuron. Science 238, 1122–1126 57. Kawahara, M., Kuroda, Y., Arispe, N. & Rojas, E. Alzheimer’s amyloidogenic fragments of β-amyloid and prion proteins.
(1987). β-amyloid, human islet amylin, and prion protein fragment J. Neurosci. 19, 928–939 (1999).
34. Kapur, A., Yeckel, M. F., Gray, R. & Johnston, D. L-type evoke intracellular free calcium elevations by a common 82. Meda, L. et al. Activation of microglial cells by β-amyloid
calcium channels are required for one form of hippocampal mechanism in a hypothalamic GnRH neuronal cell line. J. protein and interferon-γ. Nature 374, 647–650 (1995).
mossy fiber LTP. J. Neurophysiol. 79, 2181–2190 (1998). Biol. Chem. 275, 14077–14083 (2000). 83. El Khoury, J. et al. Scavenger receptor-mediated adhesion
35. Segal, M. Imaging of calcium variations in living dendritic 58. Mark, R. J., Hensley, K., Butterfield, D. A. & Mattson, M. P. of microglia to β-amyloid fibrils. Nature 382, 716–719
spines of cultured rat hippocampal neurons. J. Physiol. 486, Amyloid β-peptide impairs ion-motive ATPase activities: (1996).
283–295 (1995). evidence for a role in loss of neuronal Ca2+ homeostasis and 84. Paresce, D. M., Ghosh, R. N. & Maxfield, F. R. Microglial
36. Arendt, T. Alzheimer’s disease as a disorder of mechanisms cell death. J. Neurosci. 15, 6239–6249 (1995). cells internalize aggregates of the Alzheimer’s disease
underlying structural brain self-organization. Neuroscience 59. Lee, M. S. et al. Neurotoxicity induces cleavage of p35 to amyloid β-protein via a scavenger receptor. Neuron 17,
102, 723–765 (2001). p25 by calpain. Nature 405, 360–364 (2000). 553–565 (1996).
37. Phinney, A. L. et al. Cerebral amyloid induces aberrant 60. Lorenzo, A. et al. Amyloid β interacts with the amyloid 85. Akama, K. T., Albanese, C., Pestell, R. G. & Van Eldik, L. J.
axonal sprouting and ectopic terminal formation in amyloid precursor protein: a potential toxic mechanism in Alzheimer’s Amyloid β-peptide stimulates nitric oxide production in
precursor protein transgenic mice. J. Neurosci. 19, disease. Nature Neurosci. 3, 460–464 (2000). astrocytes through an NFκB-dependent mechanism. Proc.
8552–8559 (1999). 61. Hertel, C. et al. Inhibition of the electrostatic interaction Natl Acad. Sci. USA 95, 5795–5800 (1998).
38. Meberg, P. J., Kossel, A. H., Williams, C. B. & Kater, S. B. between β-amyloid peptide and membranes prevents 86. Akama, K. T. & Van Eldik, L. J. β-Amyloid stimulation of
Calcium-dependent alterations in dendritic architecture of β-amyloid-induced toxicity. Proc. Natl Acad. Sci. USA 94, inducible nitric-oxide synthase in astrocytes is interleukin-
hippocampal pyramidal neurons. Neuroreport 10, 639–644 9412–9416 (1997). 1β- and tumor necrosis factor-α (TNFα)-dependent, and
(1999). 62. Yaar, M. et al. Binding of β-amyloid to the p75 neurotrophin involves a TNFα receptor-associated factor- and NFκB-
39. Mattson, M. P., Tomaselli, K. J. & Rydel, R. E. Calcium- receptor induces apoptosis. A possible mechanism for inducing kinase-dependent signaling mechanism. J. Biol.
destabilizing and neurodegenerative effects of aggregated β- Alzheimer’s disease. J. Clin. Invest. 100, 2333–2340 (1997). Chem. 275, 7918–7924 (2000).
amyloid peptide are attenuated by basic FGF. Brain Res. 63. Kuner, P., Schubenel, R. & Hertel, C. β-Amyloid binds to 87. Yan, S. D. et al. An intracellular protein that binds amyloid–β
621, 35–49 (1993). p75NTR and activates NFκB in human neuroblastoma cells. peptide and mediates neurotoxicity in Alzheimer’s disease.
40. Mattson, M. P. et al. β-Amyloid peptides destabilize calcium J. Neurosci. Res. 54, 798–804 (1998). Nature 389, 689–695 (1997).

598 | AUGUST 2001 | VOLUME 2 www.nature.com/reviews/neuro


© 2001 Macmillan Magazines Ltd
Copyright of Nature Reviews Neuroscience is the property of Nature Publishing Group and its content may not
be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written
permission. However, users may print, download, or email articles for individual use.

You might also like