tends to increase (Brookmeyer et al., 1998) if no effective therapy is developed in a near ...... Lehninger Principles of Biochemistry, Fourth Edition, New York, WH.
Ana Catarina Ribeiro da Graça Fonseca
Neuroprotective effects of statins in an in vitro model of Alzheimer’s disease
Faculdade de Ciências e Tecnologia Universidade de Coimbra 2008
Dissertação apresentada à Faculdade de Ciências e Tecnologias da Universidade de Coimbra para cumprimento dos requisitos necessários à obtenção do grau de Mestre em Biologia Celular, realizada sob a orientação científica da Doutora Cláudia Maria Fragão Pereira (Universidade de Coimbra).
Agradecimentos
Agradecimentos
2
Agradecimentos Os meus sinceros agradecimentos à Doutora Cláudia Pereira pela orientação, disponibilidade e incentivo prestados ao longo do Mestrado.
À Doutora Catarina Resende de Oliveira pela oportunidade que me deu de realizar um estágio extra-curricular no Centro de Neurociências e Biologia Celular da Universidade de Coimbra e posteriormente o trabalho experimental que conduziu a esta Dissertação de Mestrado.
À Doutora Sandra Cardoso pelo apoio e ensino de técnicas de laboratório, durante o estágio extra-curricular, muito úteis à realização do Mestrado.
Os meus agradecimentos à Rosa Resende pelo apoio constante, ensinamentos e pela sua verdadeira amizade.
À Doutora Teresa Proença, do Laboratório de Neuroquímica dos Hospitais da Universidade de Coimbra, pelo contributo que deu na obtenção de alguns dos resultados apresentados.
Obrigado aos bolseiros, investigadores, docentes e técnicos do Instituto de Bioquímica da Faculdade de Medicina e Centro de Neurociências de Coimbra que de alguma forma contribuíram para a realização deste trabalho, em particular, à Elisabete Ferreiro, Teresa Oliveira, Rui Costa, João Pedro Gomes, Daniela Arduino, Raquel Esteves, Isabel Nunes e Isabel Dantas.
3
Agradecimentos
4
Table of contents
Table of contents
6
Table of contents RESUMO............................................................................................................................................................... 9 ABSTRACT ........................................................................................................................................................ 13 ABBREVIATIONS............................................................................................................................................. 17 INTRODUCTION .............................................................................................................................................. 25 ALZHEIMER’S DISEASE ..................................................................................................................................... 27 DIAGNOSIS........................................................................................................................................................ 27 RISK FACTORS .................................................................................................................................................. 28 PROCESSING OF APP AND AΒ GENERATION ...................................................................................................... 29 CLEARANCE OF AΒ PEPTIDES ............................................................................................................................ 33 AMYLOID CASCADE HYPOTHESIS ...................................................................................................................... 33 Mitochondria and endoplasmic reticulum dysfunction ............................................................................... 35 Mitochondrial impairment .......................................................................................................................................35 Endoplasmic reticulum stress...................................................................................................................................39 Relationship between mitochondria and ER ............................................................................................................42
Oxidative stress ........................................................................................................................................... 43 Calcium dyshomeostasis ............................................................................................................................. 45 Cell death and survival pathways ............................................................................................................... 47 Mechanisms of apoptosis .........................................................................................................................................47 Survival pathways ....................................................................................................................................................50 PI3K/Akt pathway ..............................................................................................................................................51 Consequences of activation of PI3K/Akt pathway..............................................................................................52 ERK pathway......................................................................................................................................................54 Effects of activation of ERK...............................................................................................................................55
CHOLESTEROL .................................................................................................................................................. 57 Biosynthesis of cholesterol.......................................................................................................................... 57 Regulation of cholesterol levels .................................................................................................................. 60 Regulation of cholesterol synthesis..........................................................................................................................60 Elimination of cholesterol in the brain .....................................................................................................................61
Lipoproteins and AD................................................................................................................................... 62 Low density lipoprotein receptors............................................................................................................................63 Apolipoprotein E......................................................................................................................................................64
Lipid rafts.................................................................................................................................................... 66 Cholesterol and AD..................................................................................................................................... 67 Cholesterol regulation in AD ...................................................................................................................................73
CHOLESTEROL-BASED THERAPIES .................................................................................................................... 74 Statins.......................................................................................................................................................... 76 Pleiotropic effects of statins .....................................................................................................................................79 Small GTPases....................................................................................................................................................80 Statins and inflammation ....................................................................................................................................82 Statins and improvements in intellectual functions.............................................................................................83 Effects of statins in endothelium cells ................................................................................................................84 Anti-oxidant effects of statins.............................................................................................................................84 Other pleiotropic effects of statins ......................................................................................................................85 Statins and APP processing......................................................................................................................................86 Statins and the PI3K/Akt pathway ...........................................................................................................................89 Statins and ERK1/2 ..................................................................................................................................................89
CONCLUSIONS .................................................................................................................................................. 90 OBJECTIVES..................................................................................................................................................... 93 MATERIALS AND METHODS ....................................................................................................................... 97 MATERIALS ...................................................................................................................................................... 99 PRIMARY CORTICAL NEURONAL CULTURES .................................................................................................... 100 NEURONAL TREATMENTS ............................................................................................................................... 100 ASSESSMENT OF NEURONAL INJURY ............................................................................................................... 102 MTT assay ................................................................................................................................................. 102 Lactate dehydrogenase (LDH) measurements .......................................................................................... 102 Hoechst 33342 staining and TUNEL assay............................................................................................... 103 MEASUREMENT OF INTRACELLULAR CA2+ CONCENTRATION .......................................................................... 103 MEASUREMENT OF INTRACELLULAR REACTIVE OXYGEN SPECIES................................................................... 103 MEASUREMENT OF CASPASE-3-LIKE ACTIVITY ............................................................................................... 104 MEASUREMENT OF CHOLESTEROL LEVELS ..................................................................................................... 104 Total cholesterol ....................................................................................................................................... 104
7
Table of contents WESTERN BLOTTING ANALYSIS ...................................................................................................................... 105 DATA ANALYSIS ............................................................................................................................................. 106 RESULTS.......................................................................................................................................................... 107 DOSE-DEPENDENT EFFECT OF STATINS ON CELL VIABILITY ............................................................................ 109 STATINS AND AΒ PRESERVE THE INTEGRITY OF NEURONAL PLASMA MEMBRANE ........................................... 110 PRE-TREATMENT WITH STATINS PROTECT CORTICAL NEURONS AGAINST AΒ-INDUCED TOXICITY .................. 111 STATINS PREVENT THE INCREASE IN CYTOSOLIC CA2+ LEVELS AND ROS ACCUMULATION TRIGGERED BY Aβ. ....................................................................................................................................................................... 114 STATINS DECREASE CASPASE-3 ACTIVATION AND ABOLISH APOPTOSIS INDUCED BY Aβ ................................ 116 STATINS DO NOT PROTECT AGAINST AΒ-INDUCED TOXICITY THROUGH CHANGES IN CHOLESTEROL LEVELS .. 118 GERANYL PYROPHOSPHATE, BUT NOT GERANYLGERANYL PYROPHOSPHATE, ABOLISHES THE NEUROPROTECTIVE EFFECTS OF STATINS ........................................................................................................ 120 EFFECT OF SIMVASTATIN IN AKT AND ERK2 ACTIVATION ............................................................................. 121 DISCUSSION.................................................................................................................................................... 123 CONCLUSIONS............................................................................................................................................... 131 REFERENCES ................................................................................................................................................. 135
8
Resumo
Resumo
10
Resumo A Doença de Alzheimer (DA) é a causa de demência mais frequente depois dos 65 anos de idade. Os doentes de Alzheimer apresentam placas senis extracelulares (formadas principalmente por agregados de peptídeo β-amilóide, Aβ) e tranças neurofibrilhares intracelulares (compostas por fibras de proteína tau hiperfosforilada), no hipocampo e córtex. Na forma familiar da doença, mutações nos genes das presenilinas 1 ou 2 ou da proteína precursora da β-amilóide (APP) causam a doença. O envelhecimento é o principal factor de risco para a DA esporádica e a presença do alelo ε4 para a apolipoproteína E aumenta a probabilidade de desenvolver a doença e diminui a idade de início dos sintomas. Os factores ambientais (como a hipercolesterolemia e as doenças associadas) também desenpenham um papel importante nos casos esporádicos da DA. O peptídeo Aβ tem sido descrito como tendo uma função chave no processo neurodegenerativo associado à DA. Este peptídeo amiloidogénico causa hiperfosforilação da tau, disfunção das mitocôndrias e do retículo endoplasmático, stress oxidativo e desregulação da homeostase do cálcio, hiper-inflamação e consequente perda de sinapses e neurónios. A Aβ é gerada a partir do processamento da APP através da via amiloidogénica que ocorre nos locais da membrana ricos em colesterol denominados “lipid rafts”. O colesterol é essencial à sobrevivência celular. Por essa razão, os níveis de colesterol são finamente controlados por transporte, síntese, armazenamento e degradação. Como o colesterol é um dos principais componentes dos microdomínios lipídicos (“lipid rafts”) e as β- e γ-secretases estão activas nesses locais, a produção de Aβ aumenta quando os níveis de colesterol membranar aumentam. Contudo, uma pequena diminuição dos níveis de colesterol na membrana também potencia a produção de Aβ porque aumenta a fuidez dos “rafts” favorecendo o contacto entre a APP e as secretases (β- e γ-secretases) aí residentes. Apenas uma diminuição acentuada do colesterol reduz a produção de Aβ, causando no entanto danos irreversíveis nas células. Actualmente, as terapias disponíveis para a DA são terapias sintomáticas não estando disponíveis terapias que modifiquem a progressão da doença. As estatinas (fármacos usados para reduzir a hipercolesterolemia) têm demonstrado diminuir a probabilidade de desenvolver a DA mesmo em doses reduzidas que não afectam os níveis de colesterol. Estes fármacos inibem a síntese de colesterol e, desse modo, reduzem a produção dos intermediários isoprenóides. A prenilação das pequenas GTPases leva à sua activação. Assim, através da inibição das GTPases, as estatinas apresentam numerosos efeitos pleiotrópicos, incluindo acção anti-oxidante e anti-inflamatória, aumento da neurogénese e diminuição a progressão da arterosclerose, que podem ser vantajosos para os doentes de Alzheimer.
11
Resumo O objectivo deste estudo foi investigar os efeitos neuroprotectores da simvastatina e da lovastatina, estatinas lipofílicas que conseguem atravessar a barreira hemato-encefálica, contra a toxicidade induzida pelos peptídeos Aβ e analizar se essa protecção é independente dde colesterol. Usando culturas primárias de neurónios corticais tratados com o peptídeo Aβ, demonstrámos que a pré-incubação com estatinas previne o aumento na concentração de cálcio citosólico e a acumulação de espécies reactivas de oxigénio induzidos por Aβ através de mecanismos independentes da redução de colesterol. Pelo menos parte das acções neuroprotectoras das estatinas foram atribuídas à sua capacidade para diminuir os níveis de intermediários isoprenóides na via de biossíntese de colesterol, uma vez que os seus efeitos foram revertidos por geranil pirofosfato enquanto que a adição de colesterol não teve efeito. Consequentemente, ambas as estatinas mostraram proteger os neurónios corticais da apoptose dependente da caspase-3 induzida por Aβ. Além disso, os nossos resultados revelaram que a simvastatina, em concentrações neuroprotectoras contra a toxicidade induzida por Aβ, não conseguiu activar as proteínas Akt e ERK2, duas cinases que protegem os neurónios da morte celular por apoptose.
12
Abstract
Abstract
14
Abstract Alzheimer’s disease (AD) is the most frequent cause of dementia in people above 65 years old. AD patients present extracellular senile plaques (mainly formed by aggregates of amyloid-β peptide, Aβ) and intracellular neurofibrillary tangles (composed by fibers of hyperphosphorylated tau protein), in the hippocampus and cortex. In familiar forms of the disease, mutations in the genes for presenilins 1 or 2 or for the amyloid precursor protein (APP) cause the disease. Aging is the principal risk factor for sporadic AD and the presence of the allele ε4 for apolipoprotein E enhances the probability to develop the disease and decreases the age of onset. Environmental factors (like hypercholesterolemia and associated diseases) also play an important role in sporadic cases of AD. Aβ peptide has been described to play a key role in the neurodegenerative process associated with AD. Aβ causes hyperphosphorylation of tau, mitochondria and endoplasmic reticulum dysfunction, oxidative stress, impairment of calcium homeostasis and hyperinflammation, and consequent loss of synapses and neurons. Aβ is generated from APP processing through the amyloidogenic pathway that occurs inside the membrane cholesterolenriched lipid rafts. Cholesterol is essential for cellular survival. For that reason, cholesterol levels are finely controlled by transport, synthesis, storage and degradation. Since cholesterol is one of the principal components of membrane rafts and β- and γ-secretases are active in these regions, the production of Aβ augments when the membrane cholesterol levels increase. However, a slight decrease of membrane cholesterol levels also enhances Aβ generation, because rafts fluidity increases leading to the elevation of the contact between APP and raft resident secretases (β- and γ-secretases). Only a substantial depletion of cholesterol reduces Aβ production, but this also causes cell damage. Presently, no disease progressing-modifying therapies are available for AD. Statins (drugs used to reduce hypercholesterolemia) have been demonstrated to decrease the probability to develop AD even in low doses, which have no effect in cholesterol levels. These drugs inhibit cholesterol synthesis and, by that way, reduce the production of isoprenoid intermediates. Prenylation of small GTPases leads to their activation. Thus, through the inhibition of GTPases, statins have numerous pleiotropic effects, including antioxidant and anti-inflammatory actions, enhance neurogenesis and decrease the progression of atherosclerosis that could all be advantageous to AD patients. This study was aimed to investigate the neuroprotective effect of simvastatin and lovastatin, lipophilic statins that can cross the blood brain barrier, against the toxicity triggered by Aβ peptides and to analyze if such protection is cholesterol-independent. Using primary cultures of cortical neurons treated with Aβ peptide, we have demonstrated that pre15
Abstract incubation with statins prevents the rise in cytosolic Ca2+ concentration and the accumulation of reactive oxygen species induced by Aβ through mechanisms independent of cholesterol reduction. The neuroprotective actions of statins were rather attributable to their ability to reduce isoprenyl intermediates levels in the cholesterol biosynthetic pathway, since their effect was reversed by geranyl pyrophosphate while cholesterol addition was ineffective. Consequently, both statins were shown to rescue cortical neurons from Aβ-induced caspase3-dependent apoptosis. Moreover, our results revealed that simvastatin, at neuroprotective concentrations against Aβ-induced toxicity, is not able to activate Akt or ERK2 proteins, two signalling kinases with neuroprotective roles against apoptosis.
16
Abbreviations
Abbreviations
18
Abbreviations ABCA
ATP-binding cassette transporter
ACAT
Acyl-coenzyme A cholesteryl acyltransferase
ACE
Angiotensin-converting enzyme
Ach
Acetylcholine
AChE
Acetylcholinesterase
AD
Alzheimer’s disease
ADAM
A disintegrin and metalloprotease
ADP
Adenosine diphosphate
AICD
APP intracellular carboxyl-terminal domain
AIF
Apoptosis-inducing factor
ALDH
Aldehyde dehydrogenase
AMP
Adenosine monophosphate
APAF-1
Apoptotic protease activating factor-1
APH
Anterior pharynx-defective-1
APOE
APOE gene
ApoE
Apolipoprotein E
ApoER
ApoE receptor
APP
Amyloid precursor protein
APP395
APP with 695 aminoacids
APP751
APP with 751 aminoacids
APP770
APP with 770 aminoacids
Asn
Asparagine aminoacid
ATF
Activating transcription factor
ATP
Adenosine triphosphate
Aβ
Amyloid beta peptide
Aβ1-40
Aβ with 40 aminoacids
Aβ1-42
Aβ with 42 aminoacids
Aβ25-35
Aβ peptide fragment with 25 to 35 aminoacids
α-sAPP
Soluble APP ectodomain derived from the clivage of α-secretase
BACE
β-site APP cleaving enzyme
BBB
Blood brain barrier
BiP
Binding immunoglobulin protein
β-sAPP
Soluble APP ectodomain derived from the clivage of β-secretase
CAD
Caspase activated DNase
Ca2+
Calcium ion 19
Abbreviations CaMK
Calcium- and calmodulin-dependent protein kinase
CaMKK
CaMK kinase
CBP
CREB binding protein
cdk5
Cyclin-dependent kinase 5
CHD
Coronary heart disease
CNS
Central nervous system
COX
Cytochrome c oxidase
CO2
Carbon dioxide
CREB
Cyclic AMP response element binding protein
CRP
C-reactive protein
CSF
Cerebrospinal fluid
C-terminal
Carboxyl-terminal
Cu+
Cupper ion
Cys
Cysteine aminoacid
C83
Fragment with 83 aminoacids of the APP C-terminal
C99
Fragment with 99 aminoacids of the APP C-terminal of
DD
Death domains
DISC
Death inducing signaling complex
DNA
Deoxyribonucleic acid
DRM
Detergent-resistant membrane
e-
Electron
EGF
Epidermal growth factor
eNOS
Endothelial nitric oxide synthase
ER
Endoplasmic reticulum
ERAD
ER associated proteolytic degradation
ERK
Extracellular signal-regulated kinase
FADD
Fas-associated death domains
FAS
Fatty acid synthase
Fe2+ or Fe3+
Iron ions
FPP
Farnesyl pyrophosphate
GGPP
Geranylgeranyl pyrophosphate
GPP
Geranyl pyrophosphate
Grp78
Glucose regulate protein 78
GSH
Glutathione
GSK-3β
Glycogen synthase kinase-3 beta 20
Abbreviations GST
Glutathione S-transferase
GTP
Guanine triphosphate
+
H
Hidrogen ion or proton
HDL
High density lipoprotein
HEK 293
Human embryonic kidney cells
HHE
4-hydroxy-2-hexenal
HMG-CoA
3-hydroxy-3-methylglutaryl coenzyme A
HNE
4-hydroxy-2-nonenal
H2O2
Hydrogen peroxide
IDE
Insulin-degrading enzyme
IDL
Intermediate density lipoprotein
IL
Interleukin
iNOS
Inducible nitric oxide synthase
IP3
Inositol 1,4,5-trisphosphate
IRE
Inositol requiring enzyme
JNK
c-Jun N-terminal kinase
K+
Potassium ion
kDa
Kilodaltons
LCAT
Lecithin cholesterol acyltransferase
LDL
Low density lipoprotein
LDLR
LDL receptor
Leu
Leucine aminoacid
LRP
Low density lipoprotein receptor-related protein
LTD
Long-term depression
LTP
Long-term potentiation
Lys
Lysine aminoacid
MAPK
Mitogen-activated protein kinase
MβCD
Methyl β-cyclodextrin
MDA
Malondialdehyde
MEGF
Multiple EGF repeat containing protein
MEK
MAPK/ERK kinase
MEKK
MEK kinase
Met
Methionine aminoacid
MPT
Mitochondrial permeability transition
MPTP
Mitochondrial permeability transition pore 21
Abbreviations mRNA
Messenger RNA
mtDNA
Mitochondrial DNA
+
Na
Sodium ion
nAChR
Nicotinic acetylcholine receptor
NADH
Reduced nicotinamide adenine dinucleotide
NADPH
NADH phosphate
NEP
Neprilysin
NF-κB
Nuclear factor-kappa B
NFT
Neurofibrillary tangles
NMDAR
N-methyl-D-aspartate receptor
NO
Nitric oxide
NPC
Neimann-Pick disease type C
NPC1
NPC protein 1
N-terminal
Amino-terminal
•OH
Hydroxyl radical
-
OH
Hydroxyl group
•O2
Superoxide anion
O2
Molecular oxygen
PERK
Protein kinase-like endoplasmic reticulum kinase
PDI
Protein disulfide isomerase
PEN-2
Presenilin enhancer 2
PH
Pleckstrin homology
Pi
Inorganic phosphate
PIP3
Phosphatidylinositol-3,4,5-triphosphate
PI3K
Phosphatidylinositol 3-kinase
PK
Protein kinase
PP
Protein phosphatase
PS
Presenilin
RAGE
Receptor for advanced glycation end products
RNA
Ribonucleic acid
RNS
Reactive nitrogen species
ROCK
Rho-associated protein kinase
ROS
Reactive oxygen species
SCAP
SREBP cleavage activating protein
Ser
Serine aminoacid 22
Abbreviations Smac
Second mitochondrial-derived activator of caspase
SRE
Sterol regulatory element
SREBP
Sterol regulatory element binding protein
TGN
Trans-Golgi network
Thr
Threonine aminoacid
TNF
Tumor necrosis factor
tRNA
Transfer ribonucleic acid
Tyr
Tyrosine aminoacid
UPR
Unfolding protein response
VLDL
Very low-density lipoprotein
VLDLR
VLDL receptor
VSCC
Voltage-sensitive calcium channel
Zn2+
Zinc ion
23
Abbreviations
24
Introduction
Introduction
26
Introduction
Alzheimer’s disease Alzheimer's disease (AD) is a progressive and fatal neurodegenerative disease that constitutes the principal cause of dementia in people over 65 years old (Kawas et al., 2000) affecting about 5% of those individuals and more than 35% of people over 80 years old (Brookmeyer et al., 1998). Currently there is no cure or prevention for AD. With the increased life expectancy and consequently population aging, the number of people with AD tends to increase (Brookmeyer et al., 1998) if no effective therapy is developed in a near future. Therefore, it is imperative to continue efforts to further elucidate the underlying pathogenic molecular mechanisms in order to reduce its frequency and progression. AD was initially described by the german physician Doctor Alois Alzheimer in 1906 after analysis of a patient’s brain that presented severe dementia. AD patients generally preserve motor functions, but exibit cognitive impairment that begins with memory loss without significant alterations in other cognitive spheres (Petersen et al., 1999). As the disease progresses, the loss of cognitive capacities are intensified resulting in failure of linguistic abilities, impaired visuospatial skills, poor judgment and indifferent attitude, alterations of personality and disorientation that converge to dementia and total dependence of patients by others. The patients may present noncognitive behavioural symptoms such as wandering, sleep disturbance and physical aggression. These clinical manifestations are due to progressive and irreversible synaptic and neuronal loss in hippocampus and neocortex, brain regions involved in learning and memory processes, and also in entorhinal cortex, amygdala and others subcortical regions such as basal forebrain cholinergic neurons, serotonergic neurons of the dorsal raphe and noradrenergic neurons of the locus coeruleus (Davies et al., 1987; DeKosky et al., 1996; Auld et al., 2002). Damage in white matter also occurs in AD (Bronge, 2002). The duration of the illness may vary from 3 to 20 years after the diagnosis but the degenerative process possibly begins 20 to 30 years before the first clinical symptoms become visible (Davies et al., 1988).
Diagnosis Until few years ago, only after patient’s death was possible to confirm the diagnosis of AD through the analysis of the post mortem brain and detection of the pathological hallmarks: intracellular neurofibrillary tangles (NFTs, composed by aggregates of hyperphosphorylated tau protein); extracellular senile plaques (formed mainly by aggregated amyloid beta peptide, Aβ, surrounded by reactive astrocytes and microglia); and 27
Introduction accumulation vascular Aβ (Davies et al., 1987; Selkoe, 2001, Figure 1.1). With the tools now available, physicians can be reasonably confident about making an accurate diagnosis of AD in a living person. The current tools for diagnosing AD include detailed patient history and information from family members, physical and neurological examinations and laboratory tests, neuropsychological testing. They may also do a computed tomography scan or a magnetic resonance imaging test.
Figure 1.1 – Plaques and tangles in Alzheimer’s disease. Extracellular senile plaques formed mainly by aggregated amyloid beta peptide surrounded by reactive astrocytes and microglia (a) and tangles composed by aggregates of hyperphosphorylated tau protein (b).
Recently, several methods of diagnosis in living subjects have emerged through imagiology with diverse markers (Sair et al., 2004) and through the analysis of the cerebrospinal fluid (CSF) (Montine et al., 2001; Simonsen et al., 2008), plasma (Irizarry, 2004) and urine. Due to the risks derived from the method of CSF attainment and the risks associated to radiations in imagiology and lack of specificity and difficulty in crossing the blood brain barrier (BBB) of the used markers for imagiology, the analyses of blood and urine samples are preferred. The main peripheral biomarkers yet in study are the Aβ peptide and others fragments of the amyloid precursor protein (APP), phosphorylated tau protein and 24S-hydroxycholesterol, a cholesterol metabolite (Sjögren et al., 2001; Blennow and Hampel, 2003; Simonsen et al., 2008).
Risk factors AD has two clinically and neuropathologically indistinguishable forms: the sporadic late-onset (90-95% of all cases), where the first symptoms frequently appear after 65 years of 28
Introduction age and the familiar early-onset form that starts earlier (after 40 years of age) and is associated with genetic mutations (Żekanowski et al., 2004). In all familiar cases of AD, the identified mutated genes increase the production, aggregation or stability of Aβ (Selkoe, 2001; Tanzi and Bertram, 2005). On the other hand, in the sporadic cases, the cause of the disease is unknown and probably develops when genetic, lifestyle and environmental factors work together to start a disease process. Nonetheless, it is know that sporadic AD is a multifactorial illness where the major risk factor is aging (Solfrizzi et al., 2004). Mutations in the genes of APP, present in chromosome 21 and of presenilins 1 and 2 (PS1 and PS2) present in chromosomes 14 and 1, respectively, are a direct cause of the disease in the families behaving them (Sherrington et al., 1995; Janicki and Monteiro, 1997; Janssen et al., 2001). However, familiar forms associated with these mutations correspond to less than 1% of all the cases (Blennow and Skoog, 1999). The presence of the allele ε4 in the gene of apolipoprotein E (ApoE) is the main factor of genetic risk in sporadic AD, increasing the probability to develop the disease (Corder et al., 1993). However, the presence of allele ε4 is not, by itself, sufficient to cause AD (Skoog et al., 1998). This allele is also correlated with increased risk for atherosclerosis and formation of amyloid plaques (Hofman et al., 1997). There are more than 50 other mutations identified as being probably associated with AD (St George-Hyslop, 2000; Rocchi et al., 2003). Environmental factors have a major role in the development of the sporadic form of the disease. The most frequent are vascular problems like atherosclerosis (Hofman et al., 1997; Casserly and Topol, 2004), stroke (Honig et al., 2003) and hypertension (Qiu et al., 2003), and hypercholesterolemia (Kivipelto et al., 2001, Marx, 2001; Wolozin, 2001; Simons et al., 2001), diabetes (Peila et al., 2002), obesity (Gustafson et al., 2003) and low levels of education at childhood (Katzman, 1993). The disruption of the BBB also seems to occur in AD patients (Skoog et al., 1998). Some authors found that steroidogenic pathways may play a role in AD (Webber et al., 2006).
Processing of APP and Aβ generation Different types of cells, like neurons and astrocytes, produce Aβ under normal conditions (Busciglio et al., 1993). The Aβ peptide has 4 kDa, is hydrophobic and results from the processing of APP by the amyloidogenic pathway (Reiss et al., 2004). APP is a transmembranar type I glycoprotein with a large extracellular domain and a short cytoplasmic C-terminal domain, expressed in all cells and with the gene in chromosome 21. APP is
29
Introduction mainly expressed in three isoforms: with 695 aminoacids (APP695, the major source of Aβ in the brain), with 751 aminoacids (APP751) and with 770 aminoacids (APP770). APP751 and APP770 anchorage a Kunitz-type protease inhibitor domain in the extracellular region. In the nervous system, APP695 is expressed predominantly in neurons, whereas APP770 and APP751 are found in neurons and non-neuronal cells (Koo et al., 1990; Selkoe, 2001).
Figure 1.2 – Sites of Aβ production, degradation and interaction. Amyloid beta (Aβ) peptide is produced from amyloid precursor protein (APP) along the secretory pathway (endoplasmic reticulum, Golgi apparatus, vesicles and cell membrane) and in endosomes after endocytosis. In Golgi and endosomes APP is cleaved predominantly by the β-secretase and in plasma membrane by the α-secretase. Aβ can be degraded in proteosomes and lysosomes. Aβ may enter to cells through the binding to low density lipoprotein receptorrelated protein (LRP). Aβ can also interact with N-methyl D-aspartate receptor (NMDA) recptors, the subunit α7 of the nicotinic acetylcholine receptors (nAChR) and receptor for advanced glycation end products (RAGE).
30
Introduction The processing of APP takes place along the secretory pathway and in endosomes after endocytosis (Figure 1.2). The APP can be cleaved by α-secretase, probably a member of a disintegrin and metalloprotease (ADAM) family, like ADAM9, ADAM10 or ADAM17 (Buxbaum et al., 1998; Koike et al., 1999; Postina et al., 2004) within the Aβ domain, resulting the amino-terminal (N-terminal) soluble extracellular α-secretase-cleaved APP (αsAPP or sAPPα) and the transmembranar C83 fragment. This is cleaved by γ-secretase, a transmembranar protein complex that includes PS1 or PS2, nicastrin, anterior pharynxdefective (APH)-1 and presenilin enhancer 2 (PEN-2) (De Strooper, 2003), originating the extracellular p3 and the APP intracellular carboxyl-terminal (C-terminal) domain (AICD) fragment. This fragment can migrate to the nucleus leading to the transcription of some genes (Kaether and Haass, 2004) and thus activating signal transduction pathways (Leissring et al., 2002). This pathway is called non-amyloidogenic, because it precludes the formation of Aβ. APP can also be cleaved by β-secretase, such as the β-site APP-cleaving enzyme (BACE, also known as memapsin 2, a type I transmembrane aspartic protease) 1 (Repetto et al., 2004) or BACE2 (Farzan et al., 2000), originating the N-terminal soluble extracellular betasecretase-cleaved APP (β-sAPP or sAPPβ) and the transmembranar C99 fragment. Clevage of C99 by γ-secretase, results in generation of the Aβ peptide and the AICD (Figure 1.3). Several evidences support the hypothesis that the amyloidogenic processing of APP by β- and γ-secretases occurs predominantly within lipid rafts (see “Lipid rafts”) or in its periphery in the membrane of trans-Golgi network (TGN) and endosomes while the nonamyloidogenic processing by α-secretase occurs outside lipid rafts in plasma membrane (Stephens and Austen, 1996; Hartmann et al., 1997; Kojro et al., 2001; Wahrle et al., 2002; Ehehalt et al., 2003; Vetrivel et al., 2004). The membranes of endoplasmic reticulum (ER), like of the others intracellular organelles, have low levels of cholesterol (Schroeder et al., 2001). So, increasing the membrane cholesterol levels of ER, augment the percentage of rafts that may lead to enhance the production of Aβ. Accordingly, the alteration in distribution or levels of cholesterol within membranes modifies the localization of APP molecules and their accessibility to different secretases (Ehehalt et al., 2003; Abad-Rodriguez et al., 2004). The cleavage by α- or β-secretases is a prerequisite for γ-secretase cleavage. Some studies show that the α- and β-secretase may compete for APP substrate and increased activity of one pathway leads to decreased APP processing in the other (Skovronsky et al. 2000). In normal cells, the non-amyloidogenic pathway predominates. BACE1 is very homologous to BACE2, although they cleave APP at different sites (Basi et al., 2003). γsecretase cuts mainly in two places leading to the formation, in the amiloidogenic pathway, of 31
Introduction Aβ with 40 amino acids (about 90%) or with 42 amino acids (about 10%). However, γsecretase can also cleave APP in other places originating Aβ with different lenghts (~38-43 amino acids).
Figure 1.3 – APP processing. APP may be cleaved by α- or β-secretases in non-amyloidogenic or amyloidogenic pathways, respectively. After that, the remained fragment of APP is cleaved by γ-secratase originating p3, in the non-amyloidogenic pathway, or Aβ, in the amyloidogenic pathway.
Although Aβ peptide is produced during the normal metabolism of cells, mutations that lead to the familiar form of AD increase the amyloidogenic APP processing and thus enhance the levels of Aβ1-42 and Aβ1-40. In AD, the Aβ1-42/Aβ1-40 rate is increased (Tomita et al., 1997), being Aβ with 42 amino acids related with all the forms of AD (Scheuner et al., 1996; St George-Hyslop, 2000). Furthermore, Aβ1-42 has a greater tendency to aggregate and is more neurotoxic to neurons than Aβ1-40 (Mattson, 1997). However, it is the sequence of amino acids 25 to 35 of Aβ (GSNKGAIIGLM) that seems to be responsible for the toxicity of the native full-length peptide (Kaneko et al., 2001) being the Aβ25-35 peptide associated to the increased rate and extent of aggregation of Aβ (Liu et al., 2004), induction of neuronal cell death (Resende et al., 2007), neuritic atrophy and synaptic loss (Grace et al., 2002) and inhibition of neurogenesis (Li and Zuo, 2005).
32
Introduction
Clearance of Aβ peptides Aβ can be eliminated from the brain by proteolytic degradation or receptor-mediated transport into the bloodstream (Shibata et al., 2000). Aβ may be degraded by several proteases including the zinc metalloendopeptidases insulin-degrading enzyme (IDE) that degrades monomers of Aβ and AICD (Farris et al., 2003) and pre-synaptic neprilysin (NEP) that degrades soluble forms of Aβ (Iwata et al., 2001), angiotensin-converting enzyme (ACE) (Oba et al., 2005), and the serine protease plasmin, reduced in AD brains (Ledesma et al., 2000). The majority of IDE is localized in the cytosol and only a small fraction exists in the plasma membrane. However, NEP is localized in the plasma membrane with its catalytic site outside of the cell. Thus, IDE cleaves Aβ in the cytosol and NEP in the periphery of the cells (Figure 1.4). Aβ peptides can also be degraded in proteosomes through the ER-associated proteolytic degradation system (ERAD) (Schmitz et al., 2004). These peptides can also be degraded by microglia or can be exported to the blood through low density lipoprotein receptor-related protein (LRP, see “Lipoproteins and AD”). These receptors are involved in Aβ influx to the brain and internalization to cells. Aged brains become less able to destroy or eject the toxic Aβ peptide produced during normal cellular activity.
Amyloid cascade hypothesis When occurs disequilibrium between the production and the degradation/clearance of Aβ, its concentration is inceased. Because the carboxyl terminal region is very hydrophobic, what leads to the predisposition of Aβ to aggregate at neutral pH, the peptide can form dimers, small soluble oligomers, filaments, protofibrils, fibrils, diffuse plaques and associate with many other proteins forming the senile plaques surrounded by astrocytes and microglia. All these aggregates of Aβ seem to be dominated by a β-sheet structure. The presence of some metals, as copper, iron and zinc, accelerate the aggregation of Aβ (Lynch et al., 2000; Cherny et al., 2001). If aggregated Aβ peptide is not eliminated, it can be toxic. The soluble forms (oligomers and protofibrils) are more neurotoxic and well correlated with cognitive deficits (Kayed et al., 2003; Resende et al., 2008), in particular the peptide with 56 kDa, named Aβ star (Cole and Frautschy, 2006). Nonetheless, monomers of Aβ in low concentrations appear to be protective, for example against oxidative stress (Zou et al., 2002). Despite the exact origin of the disease in the most cases is unknown, the pathogenesis of AD is strongly related with Aβ, as indicated by the following evidences: 1) mutations in APP and PS1/2 genes, that cause the familiar AD, increase the production of Aβ peptides 33
Introduction (Scheuner et al., 1996; Oakley et al., 2006); 2) people with trisomy 21 (chromosome 21 has the gene that codifies APP, see “Processing of APP and Aβ generation”) develop AD pathology early in life (Iwatsubo et al., 1995); 3) Aβ is toxic to neurons in culture (Resende et al., 2008); 4) APP processing is altered in AD, being the β-secretase, BACE, increased in cortex of AD patients (Evin et al., 2003); and 5) ApoE binds Aβ (Strittmatter et al., 1993) and may be involved in the formation of senile plaques and its ε4 allele is a risk factor for AD.
Figure 1.4 – Aβ life cycle. Mutations in amyloid precursor protein (APP) or proteins involved in APP processing genes (presenilin 1 or 2, PSEN1&2, β-site APP-cleaving enzyme, BACE1, nicastrin, NCSTN, anterior pharynx-defective 1A, APH1A, and presenilin enhancer 2, PEN-2) may result in increased generation of Aβ. This increase can be reduced by APP processing inhibitors and cholesterol lowering drugs. The presence of some metals like zinc and copper and apolipoprotein E (ApoE) catalize the oligomerization of Aβ. Metal complexing agents, aggregation blockers and Aβ vaccine can inhibit this process. On the other hand, Aβ may be degraded in lysosomes or by enzymes like insulin-degrading enzyme (IDE), neprilysin (NEP) and plasminogen (PLG). Aβ may also be imported into cells binding to ApoE or α2-macroglobulin (A2M) through low density lipoprotein receptor-related protein (LRP).
The overgeneration and accumulation of the Aβ peptide in the brain appears to be significant for the initiation and progression of AD (Hsiao et al., 1996; Parvathy et al., 2001). Thus, the Amyloid Hypothesis for AD sustains that the primary event involved in the disease 34
Introduction pathogenesis is the production and accumulation of Aβ peptides subsequent to abnormal proteolytic cleavage of APP (Figure 1.5). Aβ instigates cytotoxic effects through diverse mechanisms that lead first to synaptic loss and then to neuronal cell death. Some of these neurotoxic effects induced direct or indirectly by Aβ are: hyperphosphorylation of the tau protein, dysfunction of the homeostasis of Ca2+, oxidative stress, mitochondrial and ER dysfunction, neuroinflammation and alteration of synaptic transmission (Figure 1.6). It is described that Aβ can increase the phosphorylation of tau protein by enhancing the activity of some kinases, inducing the formation of neurofibrillary tangles, a neuropathological hallmark of AD. Hyperphosphorylated tau dissociates from microtubules disrupting the axonal transport and thus enhancing the neurodegenerative process. During AD, neuroinflammation mechanisms are upregulated. Microglia and astrocytes
can
have
both
neuroprotective
and
neurodestructive
functions.
The
immunocompetent cells are abundantly recruited to the affected areas. This seems to increase reactive oxygen and nitrogen species (ROS and RNS) and release proteases and cytokines that exacerbate AD pathology. The hyper-inflammation attenuates long-term potentiation (LTP) and the activation of the α7 nicotinic acetylcholine receptor (nAChR) reduces the inflammatory response. When Aβ binds to this receptor it can augment the inflammatory response and the intracellular Ca2+ levels. The dysfunction of Ca2+ homeostasis contributes to the alterations in LTP and longterm depression (LTD) in AD patients. Aβ can change the synaptic transmission acting in many locals and through several pathways. The cholinergic system is the first and the most affected, followed by the glutamatergic system.
Mitochondria and endoplasmic reticulum dysfunction
Mitochondrial impairment
The main function of mitochondria is ATP production through the coupling between oxidative phosphorylation and respiration. For that, the membrane potential due to the proton (H+) gradient plays a central role, since the ATP synthase (transmembranar enzyme that transfers a phosphate group to ADP originating ATP) uses the generated force of this gradient to produce ATP. Another function of mitochondria is the storage of Ca2+ in the matrix. Ca2+ is taken up by a uniporter on the inner mitochondrial membrane driven by the mitochondrial membrane 35
Introduction potential. Ca2+ is released from mitochondrial matrix via a Na2+-Ca2+ exchange protein in the inner membrane or via "Ca2+-induced-Ca2+-release" pathways. If intracellular Ca2+ concentration increases, the uptake of Ca2+ by mitochondria is activated. Mitochondrial Ca2+ uptake is beneficial at certain levels, since mitochondrial matrix Ca2+ activates various dehydrogenase enzymes of the citric acid cycle. However, elevated Ca2+ concentration can disrupt the membrane potential of mitochondria causing the mitochondrial permeability transition (MPT). The prolonged augment or the very elevated increase of Ca2+ levels, may lead to the opening of the MPT pore (MPTP) that disrupts the gradients of molecules with less than 1.5 kDa and through which pass some molecules to cytosol.
Figure 1.5 – Amyloid cascade hypothesis. According to the amyloid cascade hypothesis, the accumulation and oligomerization of amyloid-beta peptide (Aβ) is due to mutations in amyloid precursor protein (APP) and presenilin genes in familial Alzheimer’s disease (AD) and due to aging, genetic factors, like the presence of APOEε4 allele and many environmental factors in sporadic AD. This conducts to neuronal dysfunction that results in dementia.
36
Introduction When mitochondrial membranes are damaged, all the gradients, including the gradient of H+, are disrupted. In that way, the ATP synthase does not receive the electrons that cross throughout the respiratory chain and then they escape for oxygen, leading to the formation of ROS (Eckert et al., 2003) and diminution of the levels of ATP. On the other hand, ATP depletion leads to the reduction of the anti-oxidant glutathione (GSH) in the cell and thus, the anti-oxidant defences diminish. Moreover, ROS may destroy the mitochondrial membrane, and consequently its membrane potential, by oxidation of lipids and proteins and may create relatively large pores by where some molecules are released to the cytosol. It is the case of cytochrome c (which leads to caspase dependent apoptosis, see “Mechanisms of apoptosis”) and the apoptosis-inducing factor (AIF, which induces apoptosis by a caspases independent pathway). Furthermore, mitochondria are the major source of ROS (Polster and Fiskum, 2004), but also the main target for ROS damage. Mitochondrial dysfunction is an early event in AD (Castellani et al., 2002). Mitochondrial defects are upstream of selective neuronal loss (Davis et al., 1997), what indicates that this may be one cause of neuronal loss. Aging decreases the mitochondrial membrane potential (Hagen et al., 1997) maybe due to the disruption of the complexes in the respiratory chain (Kwong and Sohal, 2000). AD patients exibit reduced production of ATP (Eckert et al., 2003) and Aβ also depletes ATP levels (Keil et al., 2004) through inhibition of mitochondrial respiration (Pereira et al., 1998) as a result of the decrease in the activity of all five electron transport chain complexes (Gibson et al., 1998; Canevari et al., 1999; Swerdlow and Kish, 2002; Keil et al., 2004). Aβ-induced mitochondrial impairment has been shown to be related with abnormal APP processing (Casley et al., 2002) and to occur before the formation of the NFT. This reduction in ATP levels is due to the ROS formation (which modifies proteins of the mitochondrial complexes) induced by Aβ and due to the direct action of this peptide on the enzymatic activity of the mitochondrial complexes (Casley et al., 2002). Aβ can also induce the irreversible opening of MPTP (Abramova et al., 2002), which abolishes the H+ gradient. In that way, the functions of the mitochondria are modified and ROS production is amplified, which potentiates the effects of AD. During aging, mitochondrial DNA (mtDNA) suffers mutations, what may conduct to the disruption or decreased activity of the complexes from the mitochondrial respiratory chain. The components of cytochrome c oxidase (COX), for example, are codified by mtDNA and nuclear DNA. The reduction of the activity of this complex may indicate disruption in mtDNA, in nuclear DNA or in the recognition or import of proteins through the mitochondrial membrane. The decreased activity of COX is negatively correlated with the increase of mtDNA mutations (Lin et al., 2002). 37
Introduction
38
Introduction Figure 1.6 – Mechanisms of Ab-induced neurotoxicity in Alzheimer’s disease. Amyloid precursor protein (APP) is processed along the secretory pathway. After its complete translation to the endoplasmic reticulum (ER), APP is cleaved by α- or β-secretases and by γ-secretase originating diverse fragments as the AICD (which migrates for the nucleus and participates in the transcription of some genes) and the Aβ peptide that can be secreted into the extracellular space. Aβ can aggregate and form senile plaques that lead to the activation of astrocytes and microglia (inflammatory response). These liberate, between other substances, proteases, reactive oxygen species (ROS) and reactive nitrogen species (RNS) that can be toxic for the sorrounding cells. The binding of Aβ to nicotinic acetylcholine receptors (nAChR) also activates the inflammatory response and enhances the permeability of these receptors to Ca2+ resulting in the increase of intracellular Ca2+ levels. Consequently, the membrane potential remains elevated and opens the glutamate NMDA receptors that are also permeable to Ca2+. The large increase of Ca2+ activates diverse enzymes conducting to the elevated levels of ROS. These cause the peroxidation of membranes, forming 4-HNE that is toxic and opens pores in the mitochondrial membrane by which cytochrome c is released (that actives apoptosis-related caspases), oxidates proteins and induces the cleavage of DNA. Aβ can bind apolipoproteins and, by that way, enter into the cells when the lipoproteins are internalized. Inside the cells, Aβ can act in diverse pathways. For example, it can activate Cdk5 (a kinase that can directly phosphorylate tau and also activate GSK-3β (glycogen synthase kinase-3 beta) and the inhibitors of phosphatase 1, iPP1) and GSK-3β causing tau phosphorylation. Aβ also induces the formation of IP3 from the membrane phospholipids that, after binding to specific receptors, leads to the release of Ca2+ from the ER. Aβ increases the degradation of cholesterol in 7α-hydroxycholesterol that is pro-apoptotic. As cholesterol is used to the production of lipids and it is an essential part of the membranes, its levels in the cell influence several mechanisms. The intermediates of cholesterol synthesis also have a very important role in diverse pathways as geranylation and farnesylation of proteins, ubiquinone synthesis (co-factor in the respiratory mitochondrial chain) and isopentenylation of tRNA, between several others.
Endoplasmic reticulum stress ER plays an important role in the maintenance of intracellular Ca2+ homeostasis (being the major intracellular organelle involved in Ca2+ storage, with about 0.2 to 1 mM of Ca2+) and in folding and processing of proteins (Chevet et al., 2001). The ER Ca2+ content is controlled by membrane Ca2+ channels associated with the ryanodine and the inositol 1,4,5triphosphate (IP3) receptor and by the SERCA-ATPase. ER is the first compartment of the secretory pathway. After entry into the ER lumen, proteins become folded and modified by ER resident chaperones and enzymes. If membrane or secretory proteins are not properly folded or assembled they accumulate in the ER lumen. The fluctuations of the concentration of Ca2+ in the ER lumen regulate protein synthesis and processing through secondary signals between the ER and the nucleus. Increased intracellular Ca2+ concentration (which may cause disruption of Ca2+ homeostasis in the ER lumen, Paschen, 2001), ROS, blocked ER-to-Golgi transport, inhibited protein degradation, extreme pH and other insults may lead to the accumulation of unfolding and misfolded proteins within the ER that induces the unfolded protein response (UPR) (Kaufman, 1999). In the UPR pathway, the existence of unfolded proteins is detected by protein sensors in the ER and a signal is sent to the nucleus that results in an augmented expression of transcription factors and ER chaperones (Kaufman, 1999; Katayama et al., 2004), in order to 39
Introduction enhance the protein-folding capacity of the ER and/or increase the degradation of misfolded proteins (Kaufman, 2002). Inositol requiring enzyme (IRE)1α and IRE1β (transmembrane Ser/Thr protein kinases that have intrinsic endoribonuclease, RNase, activity), activating transcription factor (ATF)6 and protein kinase-like endoplasmic reticulum kinase (PERK) are sensors for the UPR in mammals that activate three different pathways (Forman et al., 2003; Figure 1.7). When UPR is induced the transcription of some key enzymes of cholesterol biosynthesis is also upregulated (see “Biosynthesis of cholesterol”). Normally, the chaperone protein glucose regulate protein 78 (Grp78, also called binding immunoglobulin protein, BiP) binds and inhibits these three sensors. However, during the UPR the sensors are released because Grp78 binds to unfolded proteins (Forman et al., 2003). Grp78 expression is increased through the activation of these three pathways, leading to a negative feedback that downregulate the UPR. If the misfolded protein cannot be corrected, is guided through ERAD. EDEM, a class of mannosidase-like chaperone, conducts the translocation of the misfolded protein to the cytosol in temporary complexes with protein disulfide isomerase (PDI) and Grp78 (Ni and Lee, 2007). In the cytosol, the protein enters the ubiquitinproteasome pathway, as it is marked for proteosomes degradation by multiple ubiquitin molecules (Kopito, 1997). Activation of UPR leads to reduction of ER stress, if the unfolded proteins are reduced, or to death, if is not possible to eliminate the unfolded proteins (Mori, 2000). Therefore, prolonged or excessive UPR activation triggers the ER-stress-induced apoptosis pathway (Kaufman, 1999; Katayama et al., 2004) causing neuronal death through a mechanism dependent of caspases (Nakagawa and Yuan, 2000; Morishima et al., 2002). Apoptosis may also be induced by ER through alterations in Ca2+ levels and activation of pro-apoptotic Bcl-2 family proteins (Rutkowski and Kaufman, 2004; Xu et al., 2005; Boyce and Yuan, 2006). Aβ may contribute to the great increase of intracellular Ca2+ that happens in the brain of AD patients by inducing the release of Ca2+ from ER mediated by ryanodine receptors (RyR) and IP3 receptors (IP3R) (Ferreiro et al., 2004). ER stress has also been implicated in AD (Lindholm et al., 2006). In addition, Aβ inhibits the UPR, and thus reduces the capacity of the ER to rectify or to eliminate these unfolded proteins, contributing for the ER stress that modifies its functions. Moreover, Cazzaniga and colleages (2007) reported that the incubation of cultured neurons with the toxic sequence of Aβ (Aβ25-35) strongly enhances protein ubiquitination. Presenilins are transmembrane proteins of the ER and Aβ synthesis may occur in the ER. Mutations in PS1 and PS2, which can occur in familiar AD, reduce the UPR (Katayama et al., 1999, 2001). This may be because presenilin mutations increase the ER Ca2+ levels or because PS1 might be required for the activation and nuclear localization 40
Introduction of IRE1 (Niwa et al., 1999). Mutations related to familial AD in PS1 or PS2 genes or overexpression of presenilins result in an elevated Ca2+ levels in the ER lumen (Chan et al., 2000; LaFerla, 2002; Mattson and Chan, 2003). Presenilins knockout result in a decrease of ER Ca2+ levels (LaFerla, 2002). So, it seems that presenilins regulate the Ca2+ levels in the ER lumen that is important to the activity of some ER proteins (Michalak et al., 2002) and thus, the alteration of ER Ca2+ levels may induce ER stress and may lead to the accumulation of unfolded proteins in the ER and to apoptosis.
Figure 1.7 – ER stress. The presence of unfolded or misfolded proteins in endoplasmic reticulum (ER) induces the activation of inositol requiring enzyme (Ire)1, activating transcription factor (ATF)6 and protein kinase-like ER kinase (PERK). These proteins lead to the transcription of genes that codify chaperones and others proteins in order to repair the unfolded and misfolded proteins. If ER homeostasis can not be maintained, apoptosis is activated. During ER stress Ca2+ is released from ER and may induce the release of cytochrome c from mitochondria through the alteration of its membrane potential or through activation of Bak/Bax dimmer. Once in the cytosol, cytochrome c binds to Apaf-1 and procaspase-9 forming the apoptotic apoptosome. Moreover, cytosolic Ca2+ may activate caspases, like caspase-12, through the activation of calpain, also leading to apoptosis, and inhibit the anti-apoptotic Bcl-2 protein.
The mutant PS1 binds and inhibit IRE1 (Paschen and Frandsen, 2001) and reduce Grp78 levels in neuroblastoma cell lines (Katayama et al., 2001). PS1 mutations affect the 41
Introduction UPR and decrease the expression of Grp78 and the levels of Grp78 are also reduced in the brains of familial AD patients (Katayama et al., 1999). In healthy cells, Grp78 binds APP inhibiting Aβ generation (Yang Y et al., 1998). So, mutations in PS1, besides conduct to ER stress, may increase the generation of Aβ by reducing the levels of Grp78 available to bind APP. An aberrant spliced form of PS2 found in sporadic AD has similar effects on the UPR (Sato et al., 2001). However, Hoozemans and collaborators (2005) showed that the protein levels of Grp78 are increased in early AD temporal cortex and hippocampus and phosphorylated PERK is present in early AD brain but not in non-demented people. These results suggest that UPR is increased in the early stages of AD maybe as a neuroprotective strategy, but the mutations in presenilins inhibit the UPR leading to the increased amount of unfolded or misfolded proteins.
Relationship between mitochondria and ER
As mitochondria and ER are physically close and have linked functions, mitochondria impairment might compromise ER function and vice versa. The crosstalk between the ER and mitochondria is important, for instance, in the execution of cell death (Figure 1.7). Bcl-2 family proteins (see “Mechanisms of apoptosis”) play a crucial role in ER/mitochondria interactions and, for example, Bcl-2 and Bcl-xL proteins associate with mitochondria and ER membranes influencing their homeostasis (Breckenridge et al., 2003; Rao et al., 2004). Bcl-2, Bax and Bak can bind to ER Ca2+ channels, such as to IP3R, and regulate ER Ca2+ levels and release into the cytosol (Chen et al., 2004). Moreover, the presence of Aβ peptide promotes IP3 production from the plasma membrane, which binds to IP3Rs in the ER, leading to the release of Ca2+, what contributes for the increase of Ca2+ in the cytosol. As ER and mitochondria may be adjacent, the Ca2+ released from ER may be taken up by mitochondria. Indeed, ER stress may trigger MPT (Deniaud et al., 2008). When the levels of absorbed Ca2+ are elevated, the mitochondrial membrane potential changes and the ATP production diminishes, leading to increased ROS production, low levels of ATP and further deleterious consequences. If the membrane potential of mitochondria is maintained low and the levels of Ca2+ of the matrix are maintained high, the MPTP can open permanently what result in mitochondria impairment and, consequently, cell death.
42
Introduction Oxidative stress
ROS have important roles in signal transduction pathways as second messengers. For instance, various studies indicate that ROS play a critical role in the propagation of signals of growth factors, hormones and cytokines (Lo et al., 1996; Bae et al., 1997; Mahadev et al., 2001). Nevertheless, ROS can damage lipids, proteins, DNA and consequently cause cell death when present in high quantities. The brain is an organ very susceptible to oxidative damage: it has a high glucose metabolism and oxygen consumption; is very dependent of the oxidative phosphorylation reactions to produce ATP; is rich in polyunsaturated fatty acids (potential substrates for peroxidation); has high levels of iron; and has low doses of antioxidant defences (Halliwell et al., 1992). Oxidative stress has been reported as one of the first events in AD (Nunomura et al., 2001) and the toxicity of Aβ can be mediated by ROS accumulation (Behl et al., 1994). During the process of Aβ aggregation ROS are produced (Pereira et al., 2005). H2O2 and hydryoxyl radical (•OH) are generated via chemical reactions requiring Zn2+, Cu+ or Fe2+ (Lynch et al., 2000). After Aβ binds these metals, they can stimulate the aggregation of Aβ. Metals catalyze the conversion of H2O2 to •OH through the Fenton reaction (H2O2 + Fe2+ => •OH + -OH + Fe3+) and Haber-Weiss reaction (O2• + H2O2 => O2 + -OH + •OH). So, with the aggregation of Aβ, these metals also accumulate in neuritic plaques enhancing the production of ROS. There are many sources of ROS but, as mitochondrial respiratory chain consumes the majority of the oxygen utilized in the cell, the major intracellular source is the mitochondrial respiratory chain (Polster and Fiskum, 2004). During the respiratory pathway are formed ROS like the free radicals •OH, superoxide anion (•O2-) and non-radical derivatives, such as hydrogen peroxide (H2O2). Mitochondrion is also the most affected intracellular organelle by ROS. The increase in the concentration of intracellular Ca2+ levels induced by Aβ leads to the activation of diverse pathways, including the impairment of the complexes of the mitochondrial respiratory chain increasing ROS production. The ubiquinone site of the complex III, for example, turns molecular oxygen to •O2-, which can induce the formation of other potent oxygen derived free radicals. The binding of Aβ to RAGE (receptor for advanced glycation end products) can also induce the production of ROS (Yan et al., 1996). Aβ through RAGE induces caspasedependent apoptosis, activates the pathways of p38 MAPK, c-Jun N-terminal kinase (JNK, that may lead to apoptosis) (Pereira et al., 2005), ERK (extracellular signal-regulated kinase) and phosphatidylinositol 3-kinase (PI3K) and induces the translocation to the nucleus of NF43
Introduction κB (nuclear factor-kappa B), which leads to the expression of anti-apoptotic genes (Onyango et al., 2005, 2005b). In vitro, RAGE has been found on the surface of microglia and neurons and RAGE activation by Aβ is correlated with oxidant stress in neurons and activation of microglia (Yan et al., 1996). In addition, astrocytes and microglia surrounding senile plaques release ROS (Pereira et al., 2005) such as nitric oxide, elevated in AD's brain. ROS can reduce mitochondrial membrane potential, damage the mitochondrial respiratory chain and, consequently, diminish the ATP production (Mattson, 2000). The reduction of ATP leads to the decrease of the anti-oxidant GSH, which enhances oxidative stress and contributes to mitochondrial dysfunction and apoptosis. ROS create pores in the mitochondrial membrane releasing cytochrome c and, consequently, activating the pathway of caspases that leads to apoptosis (Pereira et al., 2005). For example, ROS increase the activity of caspase 2, involved in mitochondrial dysfunction (Marques et al., 2002; Eckert et al., 2003b), caspase 8 that activates the pro-apoptotic caspase 3 (Eckert et al., 2003b) and caspase 3 (Fadeel and Orrenius, 2005). ROS also induce the formation of IP3 from the plasma membrane leading to Ca2+ release from the ER to cytosol. The function of the ER may also be altered by the presence of oxidants (Hayashi et al., 2005). ROS increase the activity of β-secretase contributing for the augment of Aβ production (Gabuzda et al., 1994). ROS cause protein and lipid oxidation (Cardoso et al., 1998) leading to modifications of their structures and functions. AD brains have high levels of oxidative stress, and consequently, lipid peroxidation and protein oxidation, also associated with Aβ accumulation (Smith et al., 1991; Lovell et al., 1995; Mark et al., 1996). Lipid peroxidation changes the biophysical properties of membranes that can have profound effects on the activity of membrane-bound proteins. The peroxidation of membranar lipids (that increases the membrane fluidity and permeability), leads to the fragmentation of fatty acids, promotes the lipid-lipid and lipid-protein interactions and produces new fatty acids by endocyclization (Farber, 1994). Consequently, the activity of enzymes, ion channels and transporters in the membrane and the activity of second messenger systems are altered. For instance, oxidative stress induces the dysfunction of Na+/K+- and Ca2+-ATPases and the alteration of glucose and glutamate transport (Mattson, 2000). As a result, membranes of mitochondria, for example, depolarize decreasing ATP levels. The products of the fragmentation of the membranar fatty acids are predominantly chemically reactive aldehydes like malondialdehyde (MDA), acrolein, 4-hydroxy-2-nonenal (HNE) and 4-hydroxy-2-hexenal (HHE), that react with proteins, nucleic acids and lipids (Esterbauer et al., 1991). HNE is increased in brain (Markesbery and Lovell, 1998) and CSF 44
Introduction (Lovell et al., 1997) from AD patients. Proteins are covalently modify by HNE on cysteine, lysine and histidine residues by a process called Michael addition. This modification on ionmotive ATPases, neuronal glucose transporter GLUT3 and astrocyte glutamate transporter GLT-1 impairs their functions (Keller et al., 1997; Blanc et al., 1998). HNE enhances the phosphorylation of tau and destabilizes microtubules by modification of tubulin (Mattson et al., 1997; Neely et al., 2000), affecting the microtubule organization and neurite outgrowth (Neely et al., 1999). It also causes damages in membranar Ca2+ pumps contributing for the increase of intracellular Ca2+ levels (Pereira et al., 2005), diminishes the levels of GSH, modifies the protein structure, reduces the enzymatic activity and induces cellular death (Farber, 1994). The cells have oxidoreductases that detoxify HNE including glutathione Stransferase (GST) and aldehyde dehydrogenase (ALDH). The activity of cytosolic GST present in neurons is diminished (Lovell et al., 1998) and the activity of class 2 ALDH present in the glial cells is augmented (Picklo et al., 2001) in brain regions affected in AD.
Calcium dyshomeostasis A moderate increase of intracellular Ca2+ levels maintains neuronal survival. However, a fast, strong or prolonged increase of intracellular Ca2+ levels may induce neuronal cell death (Figure 1.6). In AD, the homeostasis of ions, including Ca2+, is altered. The ATP depletion found in AD (Hoyer, 1992) can reduce the activity of membrane ATPase pumps that maintain the equilibrium distribution of Na+, K+ and Ca2+. So, the intracellular concentrations of Na+ and Ca2+ may augment inside the cell, whereas K+ is released from the cell. For example, incubation of synaptosomes from adult human hippocampus with Aβ results in impairment of the plasma membrane calcium-ATPase and sodium pump (Mark et al., 1995). During aging the capacity to control the homeostasis of Ca2+ diminishes (Toescu and Verkhratsky, 2003). In AD patients, this capacity is even lower. These patients and transgenic mice for mutant APP (found in familiar AD cases) have the intracellular Ca2+ concentration increased in neurons of the cortex and hippocampus (Mattson et al., 1992; LaFerla, 2002) being synapses the first to suffer deregulation of Ca2+ (Mattson and Chan, 2003). The cytosolic Ca2+ signalling and synaptic plasticity dependent of Ca2+ are changed before the manifestation of AD-related morphological modifications (Larson et al., 1999). Compounds that affect the intracellular Ca2+ levels induce apoptosis and the induction of apoptosis by Aβ is due, at least in part, to the perturbation of intracellular Ca2+ homeostasis in cortical and hippocampal neurons (Agostinho and Oliveira, 2003; Ferreiro et al., 2004; Resende et al., 45
Introduction 2007). Deregulation of intraneuronal Ca2+ homeostasis by Aβ, particully through the release of Ca2+ through ER receptors (Ferreiro et al., 2004), increase the vulnerability of neurons to subsequent insults, like excitotoxicity (Mattson et al., 1992). The oligomeric forms of Aβ1-42 are more neurotoxic than the fibril forms in this process (Resende et al., 2008). Furthermore, oligomers of Aβ can form channels in the cell membrane, maybe with four or six Aβ subunits in each leaflet of the bilayer, allow a Ca2+ current, blocked by zinc, that results in loss of intracellular Ca2+ homeostasis (Durell et al., 1994; Kawahara and Kuroda, 2000). Moreover, Aβ damage membrane Ca2+ pumps (Mark et al., 1995) and enhance the influx of Ca2+ through voltage-sensitive Ca2+ channels (VSCC) (Silei et al., 1999). Aβ, through its binding to nAChRs and the induction of the formation of IP3, ROS and HNE (Mattson and Chan, 2003) also leads to an increase of the concentration of intracellular Ca2+. Mutations in PS1 and PS2 seem to augment the pool of ER Ca2+ available for release, making the cells more vulnerable to death induced by a variety of stimuli (Mattson and Chan, 2003). The variations of intraluminal Ca2+ concentration of ER also control various intra-ER chaperones responsible for correct protein folding (Michalak et al., 2002). Increased Ca2+ concentration may induce phosphorylation of APP (that leads to accumulation of intraneuronal Aβ) and tau by activation of Ca2+ dependent kinases (like transglutaminase increased in AD brain, Johnson et al., 1997) and cyclin-dependent kinase 5 (cdk5) and glycogen synthase kinase-3 beta (GSK-3β), similar to those found in AD (Pierrot et al., 2006), increase Aβ production (Querfurth and Selkoe, 1994), enhance the activity of Ca2+- and calmodulin-dependent protein kinase (CaMK)II, ras-mitogen activated kinase and transcription factors (Vanhoutte et al., 1999), induce loss of mitochondrial membrane potential that leads to elevation of ROS by mitochondria and consequent release of cytochrome c, caspase-3 activation and apoptotic cell death (Mattson and Chan, 2003; Ferreiro et al., 2006; Resende et al., 2007) and cause plasma membrane depolarization that activate some membrane receptors (Mattson, 2000) by which may enter more Ca2+. The increase in the concentration of intracellular Ca2+ that results in the activation of proteases (for example, calpains, cystein proteases involved in the regulation of cytoskeletal remodelling) and kinases (such as protein kinase, PK,C and Akt) (Mattson and Chan, 2003) may activate several other pathways. For instance, elevated intracellular Ca2+ concentration activates calpains, like m-calpain which mediate caspase-12 activation and Bcl-xL inactivation, leading to apoptosis (Nakagawa and Yan, 2000). Therefore, continuous alterations of Ca2+ concentration following excitotoxicity and oxidative stress can result in neuronal apoptosis in AD. The sAPP fragment protects neurons from the toxicity of Aβ, 46
Introduction excitotoxicity and mutations in the PS1 by stabilizing the concentration of intracellular Ca2+ (Mattson and Chan, 2003).
Cell death and survival pathways
In the brain of AD patients both signs of necrosis and apoptosis have been detected, being the rate of apoptotic cell death augmented 30 to 50 fold above that observed in age matched controls (Yang F et al., 1998). In addition, Aβ can induce necrosis as well as apoptosis (Behl et al., 1994b). These two types of cell death are morphologically and biochemically distinct. Necrosis is characterized by cell swelling, rapid cellular membrane lysis and expulsion of the cytoplasm. The resultant cellular debris can damage the involving cells. Contrarily, apoptosis is a genetically controlled cellular response to developmental or environmental stimuli, where alterations in the chromatin and nucleus occur by a series of well organized cellular events that require ATP, RNA and protein synthesis. The morphological changes of apoptotic cells comprise condensation and fragmentation of heterochromatin, loss of the nuclear envelope, membrane blebbing and cellular fragmentation into apoptotic bodies, while most of the organelles are maintained intact. The resultant cellular debris cannot damage the organism. Apoptosis is crucial to the normal physiology of multicellular organisms, because is an important mechanism of negative selection that removes damaged or excessive cells to sculpt the body during development. Even so, excessive apoptosis or survival (during the development) may conduct to a large number of diseases.
Mechanisms of apoptosis
Apoptosis may be subdivided in 3 stages: initiation, through the delivery of death signals (e.g. growth factor deprivation, toxins, ultra-violet light, anticancer drugs); regulation, by death modulators (e.g. Bcl-2 family, p53, kinases, phosphatases); and execution, through death effectors (such as nuclear receptors and caspases). Some groups of molecules involved in the apoptotic cascade are activated by diverse signalling pathways influencing each other in a positive or negative manner. Some members of the Bcl-2 family (like Bcl-2, Bcl-XL, Bcl-w, BAG-1, Mcl-1 and A1) promote cell survival whereas other members (like Bax, Bak, Bid, Bik, Bcl-xL, Bim, Bad and Bak) promote cell death. These family members are mainly localized in the outer mitochondrial membrane and are critical in the regulation of the permeability of the 47
Introduction mitochondrial outer membrane and mitochondrial membrane potential (Festjens et al., 2004). Bcl-2 family proteins homo- and heterodimerize and the balance between specific homo- and heterodimers are important for the maintenance of cell survival or the induction of cell death (Cory et al., 2003). They act through the regulation of MPTP formation in the outer mitochondrial membrane. When MPTP is formed, the mitochondrial membrane permeability is disrupted and cytochrome c, AIF and second mitochondria-derived activator of caspase (Smac) are released from the intermembrane space. Smac interacts with some inhibitors of apoptosis reducing its inhibitory effect on caspases. By other way, some Bcl-2 family members, like Bax, may itself form pores in the outer mitochondrial membrane allowing the release of these proteins to the cytosol. Caspases constitute a mammalian member family of specific cysteine proteases and they are central elements in regulating apoptosis. They can activate DNase and induce the cleavage of enzymes, cytoskeletal proteins and ion channels. All caspases are produced in cells as catalytically inactive zymogens that require proteolytic activation. Caspase-2, -8, -9 and -10 are called initiator caspases, since initiate the cascade of caspases. Caspases-3, -6 and -7 are named effector caspases because are activated by the initiator caspases and perform several cleavages that lead to apoptosis. At least a portion of caspase-9 is mitochondrial while caspase-1 and caspase-3 are principally cytosolic (Nakagawa et al., 2000). Procaspase-12 (caspase-12 in murine is homologous to caspase-4 in humans) is localized at the ER and is specifically activated by perturbations to ER homeostasis such as ER stress and mobilization of intracellular Ca2+ ion store. Caspase-12-/- cortical neurons are resistant to Aβ peptidemediated neurotoxicity (Nakagawa et al., 2000). This suggests that Aβ induces apoptosis, at least in part, through the perturbation of ER homeostasis. Caspase-3, the major effector protease in apoptosis, is produced as a pro-enzyme composed of a short N-terminal pro-domain followed by p17 and p21 subunits. It may be cleaved to generate the p17 active fragment after a cell death stimulus. Once activated, caspase-3 induces a series of mechanisms that conduct to cell death. One of them is the activation of caspase-activated DNase (CAD), which is responsible for DNA fragmentation (Widlak, 2000). The pro-caspase-3 can be activated by intrinsic and extrinsic pathways (Yan and Shi, 2005; Figure 1.8). Intrinsic pathways are activated by some stress stimuli (like increased ROS and Ca2+ induced by Aβ and DNA damage) and are regulated by Bcl-2/Bax proteins. For example, translocation of Bax homodimers to mitochondria leads to the opening of MPTP and consequent release of cytochrome c and others from mitochondria to the cytosol (Fadeel and Orrenius, 2005). The binding of dephosphorylated Bad to Bcl-2 can also lead to the 48
Introduction opening of MPTP. In the cytosol, cytochrome c binds to APAF-1 (apoptotic protease activating factor-1) and induces its oligomerization, which recruits procaspase-9 to form the apoptosome with subsequent caspase-9 activation. Active caspase-9 cleaves procaspase-3 to produce active caspase-3 and procaspase-7 in caspase-7. Active caspase-3 cleaves numerous important cellular proteins to perform cell death and activate additional downstream caspases like caspase-2 and -6 that trigger the release of mitochondrial cytochrome c and further activation of caspase-9.
Figure 1.8 – Some cellular proteins involved in apoptosis regulation. Apoptosis may be induced by two pathways: intrinsic and extrinsic. In the intrinsic pathway, apoptosis is activated by stress stimuli and it is regulated by Bcl-2 family proteins. The extrinsic pathway is activated by the binding of a ligand to death receptors, inducing the formation of the death-inducing signal complex.
In extrinsic pathways a ligand, like Fas, binds to death domains (DD) receptor on the plasma membrane that recruits an adaptor molecule, Fas-associated death domain (FADD), and procaspase-8 or -10 to form the death inducing signalling complex (DISC). Procaspase-8 49
Introduction and -10 are cleaved to form the active caspase-8 or -10. Caspase-8 activates the caspase-3, -6 and -7 that conduct to apoptosis. Caspase-8 can also cleave Bid and the truncated Bid translocates to the outer membrane of mitochondria and provokes the release of more proapoptotic factors that induce the intrinsic apoptotic pathway. Eckert and collaborators (2003) proposed that high levels of Aβ activate death receptors and, subsequent, the extrinsic pathway that leads to apoptosis and low levels of Aβ, through oxidative stress, activate the intrinsic pathway of apoptosis. Mature neurons remains in the G0 phase of the cell cycle, but neurons of AD brain present active cell cycle proteins (Yang Y et al., 2003; Herrup et al., 2004). In neurons, the reactivation of the cell cycle is an early step of apoptosis and may be responsible for the abnormal tau phosphorylation and aggregation, leading to neurofibrillary tangles in AD (Hamdane et al., 2003). So, apoptosis in AD neurons can begin from a failure of degenerating neurons to re-entry the cell cycle.
Survival pathways
The cellular decision to undergo apoptosis is determined by the interaction of multiple survival and death signals. Numerous factors promote cell survival, such as neurotrophins and growth factors. Survival factors inhibit the intrinsic cell death mechanisms and in that way prevent apoptosis. Neurons can express a diversity of anti-apoptotic proteins that prevent cell degeneration caused by both environmental and genetic insults (Mattson, 2000; Yuan et al., 2003). Familial AD mutations in presenilin genes increase the vulnerability of neurons to apoptosis and accelerate neurodegeneration (Guo et al., 1997). In the study of Passer and coworkers (1999), presenilins were shown to interact with Bcl-XL, promoting cell survival through the C-terminal. Alg-3, a fragment from the C-terminal of PS2, is neuroprotective abolishing Fas- and tumor necrosis factor (TNF)α-induced apoptosis (Vito et al., 1997). The C-terminal portion of PS1 homologous to Alg-3 can also be protective in Jurkat cells (Vézina et al., 1999). Moreover, PS1 and PS2, during apoptosis, may be substrates of caspases, for instance, caspase-3 produces Alg-3 from PS2 (Selkoe, 2001; Breckenridge et al., 2003), what suggests that the cleavage of presenilins may be an anti-apoptotic strategy. Exposure of cells to oxidant stress may induce activation of the members of the mitogen activated protein kinases (MAPK), like ERK1/2, c-Jun N-terminal kinase/stressactivated protein kinase (JNK/SAPK), and p38 MAPK (Ueda et al., 2002; Torres and Forman, 2003), and the PI3K/Akt pathway (Yang JY et al., 2004). The relative contribution 50
Introduction to neuronal survival of ERK1/2 and PI3K/Akt pathways depends on the particular type of cell damage (Hetman and Xia, 2000). In HeLa cells, inhibition of ERK activity is sufficient to trigger p38 MAPK activation and apoptosis dependent of caspases and activation of PI3kinase and Akt inhibits p38 MAPK activation and apoptosis (Berra et a., 1998). Activation of PI3K and ERK pathways and inhibition of p38 and JNK protect cybrids from caspasedependent apoptosis induced by oxidative stress (Onyango et al., 2005, 2005b). Statins activate Akt and ERK (Eto et al., 2002; Skaletz-Rorowski et al., 2003) and induce direct neuroprotection through the induction of PI3K/Akt and Ras/Erk pathway (Urbich et al., 2002).
PI3K/Akt pathway
PI3K has several isoforms, the best characterized is the class IA heterodimer composed of a regulatory subunit with 85 kDa (p85) and a catalytic subunit with 110 kDa (p110) that regulates a variety of cell responses including cell division and survival. PI3K may be activated by the direct binding to GTPases, such as Ras (Anderson and Jackson, 2003) or by growth factors through the association of the p85 subunit with specific phosphotyrosines on cytoplasmic domain of growth factor receptor or on receptor-associated adapter protein. By that way, PI3K is recruited to the plasma membrane and the p110 subunit phosphorylates phosphoinositides at the D3 position. Produced phospholipids can activate several other pathways, including Akt. The phospholipids phosphatidylinositol-3,4bisphosphate (PI(3,4)P2) and phosphatidylinositol-3,4,5-trisphosphate (PIP3) at the plasma membrane resulted from PI3K phosphorylation of phosphoinositides, can attach the pleckstrin homology domain of Akt, which, by this way, is recruited to the membrane and may be phosphorylated at Thr308 on the kinase domain and at Ser473 on the tail domain (Shiojima and Walsh, 2002; Song et al., 2005). Phosphorylation of Akt at both residues is needed for the full activation of Akt. Akt, the major serine-threonine kinase, may also be activated by a PI3K-independent pathway. One example is the activation of Akt by CaMK kinase (CaMKK). This kinase is activated by increased Ca2+ levels in the cytoplasm and it activates CaMKI and IV by phosphorylation. Additionally, CaMKK can phosphorylate Akt on the residue Thr308 resulting in its activation and protection against apoptosis (Yano et al., 1998). The small GTPase Ras can also activate Akt (Cardone et al., 1998). Akt (or protein kinase B, PKB, or related to the A and C protein kinases, RAC-PK) family has various members: the human PKBα, PKBβ1 and 2 and PKBγ in rats. It is a protein 51
Introduction kinase with homology to protein kinases A and C within the catalytic domain. Akt contains the pleckstrin homology (PH) domain in the amino terminus, a central kinase domain and a carboxy-terminal regulatory domain. This kinase has many protein ligands and, thus, may be regulated by and act in several pathways (Brazil et al., 2002). Akt is implicated in synaptic plasticity (Sanna et al., 2002) and spatial memory formation (Mizuno et al., 2003). Akt is a crucial mediator of cell survival in response to growth factor stimulation, Ca2+ influx and apoptotic stimuli. For example, it is the major regulator of insulin signalling and glucose metabolism, regulates cell growth, cell cycle and cell proliferation and promotes cell survival by inhibiting apoptosis through phosphorylation and inactivation of several targets (Cardone et al., 1998; Brunet et al., 1999; Tu et al., 2000). These targets may be proteins in the cytosol or Akt can translocate into the nucleus and regulate transcription factors (Brunet et al., 2001). The consensus site phosphorylated by Akt is RXRYYS/T (R is arginine, X is any amino acid, Y is preferably small residues other than glycine, S is serine and T is threonine). Akt may be inactivated through dephosphorylation by protein phosphatese (PP)2A.
Consequences of activation of PI3K/Akt pathway
The first direct in vivo substrate of Akt identified was GSK-3β. Negatively regulating GSK-3β, through the phosphorylation at Ser9 at an N-terminal non-catalytic residue, Akt induces activation of glycogen synthesis and many other pathways and protects neurons from apoptosis (Crowder and Freeman, 2000). Akt prevents the JNK activation (Cerezo et al., 1998; Okubo et al., 1998), indirectly antagonizes p38 MAPK activation through caspase inhibition (Berra et al., 1998) and inhibits apoptosis signal-regulating kinase 1 (ASK1) (Kim AH et al., 2001), which is a MAPK kinase kinase family member that phosphorylates MAPK kinases upstream of p38 MAPK and JNK kinases. Furthermore, Akt inhibits Raf-1 (which may be activated by Ras) that can activate MEK (which activates ERK1/2) and consequently, inhibits the activation of ERK1/2. PI3K/Akt is a survival pathway in many cell types including neurons (Philpott et al., 1997). Akt has numerous substrates in the cytosol and in the nucleus (Figure 1.9, Laulor and Alessi, 2001), including the pro-apoptotic proteins: GSK-3β, Bad, caspase-9 and Forkhead transcription factor, which are suppressed upon phosphorylation by Akt (Datta et al., 1997; Cardone et al., 1998; Pap and Cooper, 1998; Brunet et al., 1999). Blocking the activity of PI3K, the ability of trophic factors to promote survival is suppressed. In addition, in some cell types the inhibition of PI3K is sufficient to induce apoptosis and the expression of permanently active mutants of PI3K or Akt is enough to inhibit apoptosis. Although in some 52
Introduction cells the inhibition of PI3K/Akt pathway is not sufficient to induce apoptosis (Philpott et al., 1997), the activation of this pathway is anti-apoptotic.
Figure 1.9– Ras can activate PI3K/Akt and MEK/ERK pathways. The small GTPase Ras associates and activates Raf-1 kinase that activates MEK and subsequently ERK. ERK has many substrates in the cytosol that may be phosphorylated in order to prevent apoptosis, and its p90 subunit may translocate to the nucleus where induces the transcription of many anti-apoptotic genes. Ras can also induce PI3K-mediated Akt activation that phosphorylates many substrates in the cytosol, such as GSK-3β, caspase-9 and the Bcl-2 family member Bad, in order to prevent apoptosis and can also induce the transcription of some anti-apoptotic genes.
Akt promotes cell survival by inhibiting the action of proteins that mediate apoptosis (Laulor and Alessi, 2001). For example, Bad may translocate from the cytosol into mitochondria and promote apoptotic cell death by inhibiting Bcl-2 or Bcl-xL through protein53
Introduction protein interactions. Akt phosphorylates Bad at Ser136 decreasing the binding of Bad to BclxL or Bcl-2 at the mitochondrial membrane and increasing its binding to the molecular chaperone 14-3-3 in the cytosol, allowing Bcl-2 or Bcl-xL to function as inhibitors of apoptosis (Zha et al., 1996; Datta et al., 1997). On the other hand, Zhou and collaborators (2000) report that addition of cytochrome c to lysates from cells expressing active Akt does not activate caspase-9 or -3 and that Akt blocks cytochrome c-induced activation of the caspase cascade. The explanation for this may be that activated Akt phosphorylates procaspase-9 on Ser196 (Cardone et al., 1998) and, by that way, the apoptosome can not process the pro-caspase-9 in caspase-9 inducing apoptosis. Akt can also regulate the expression of anti-apoptotic transcription factors and negatively regulate the expression of pro-apoptotic molecules. For instance, Akt phosphorylates the transcription factor CREB (cyclic AMP response element binding protein) increasing binding to CREB binding protein (CBP) and enhancing CREB transcriptional activity (Pugazhenthi et al., 1999). Another transcription factor regulated by Akt is NF-κB. Phosphorylation of IKK results in the release of the inhibitory factor IκB and nuclear translocation of active NF-κB (Romashkova and Makarov, 1999) which induces expression of numerous pro-survival genes. The forkhead transcription factors (family of transcription factors important in regulating the expression of genes involved in cell growth, proliferation and differentiation) may be required for the expression of pro-apoptotic molecules like the Fas ligand. Its sequestration in the cytoplasm subsequent to phosphorylation by Akt may result in diminished expression of these proteins and thus, promotes cell survival (Brunet et al., 1999). Inhibitors of PI3K, in addition to the inhibition of Akt activation (Onyango et al., 2005b), also inhibit sterol regulatory element binding protein (SREBP)-1 and -2 (transcription factors of genes involved in lipid synthesis), by inhibiting the transport of SCAP (SREBP cleavage activating protein) from the ER to the Golgi, in response to growth factors and insulin (Goldstein et al., 2006; Du et al., 2006) suggesting that the PI3K pathway is involved in lipogenesis. Porstmann and co-workers (2005) reported that Akt increases the expression of SREBP-1 and -2 and enhances the translocation of SREBP-1 to the nucleus increasing, in that way, the synthesis of FAS (fatty acid synthase), fatty acids and phosphoglycerides.
ERK pathway
MAPK family is composed by many serine/threonine protein kinases including ERK, p-38 and JNK. MAPK pathways are the principal mediators that propagate signals from the 54
Introduction membrane to the nucleus. ERK is the first identified and best characterized and has some isoforms, like ERK1 (p44, with 44 kDa) and ERK2 (p42, with 42 kDa). The JNK and the p38 MAPK are regularly associated with induction of apoptosis while ERK pathway seems to protect cells from apoptosis (Onyango et al., 2005, 2005b). ERK1 and ERK2 are about 83% identical with a greater similarity in the core regions involved in binding substrates. They are ubiquitously expressed, but with different abundances. In post-mitotic neurons they are highly expressed and are involved in adaptive responses like LTP. ERK is activated by phosphorylation on Thr202 and Tyr204 within the motif ThrGlu-Tyr in subdomain VIII of the catalytic domain. The exact mechanisms by which ERK1/2 signalling pathways promote neuronal survival is under study. Nonetheless, it is known that ERK1/2 pathway is activated by a wide variety of mitogens and survival factors, including growth factors, ligands of the heterotrimeric G protein-coupled receptors, cytokines and microtubule disorganization, but in general all the signals involve small GTP binding proteins that activate MAPK/ERK kinase (MEK)1/MEK2 that, in turn, phosphorylate and activate ERK1 and/or ERK2 (Repasky et al., 2004). For example, once activated, the GTPase Ras interacts with and activates Raf-1 kinase, which in turn activates MEK1/2 that activates ERK1/2 through phosphorylation on Tyr and Thr residues (Repasky et al., 2004, Figure 1.9). ERK1/2 phosphorylates cytoplasmic or nuclear targets regulating gene expression through phosphorylation of a variety of transcription factors. The kinase cascade permits to detect inputs from other signalling pathways to amplify or suppress the signal. For instance, many MEK family members can be phosphorylated by kinases in other pathways influencing their ability to interact in complexes. Some MEK kinases (MEKKs) may also regulate more than one MAPK cascade and some cascades may be controlled by numerous unrelated MEKKs (Roux and Blenis, 2004).
Effects of activation of ERK
In some cell types and in the presence of certain insults, activation of ERK induces cell degeneration and may function as an inducer of apoptosis. For example, in cerebellar granule neurons ERK promotes plasma membrane damage and nuclear condensation independently of caspase-3 (Subramaniam et al., 2004). Late and continued ERK activation has been associated with neuronal cell death and specific ERK inhibitors diminish neuronal damages (Noshita et al., 2002; de Bernardo et al., 2004). Additionally, in primary cultures of cortical neurons, glutamate-induced neuronal injury was shown to require ERK activation (Stanciu et al., 2000). Furthermore, the pathway Rac-ERK stimulates the production of ROS 55
Introduction (Woo et al., 2002). Moreover, traumatic brain injury-induced cortical neuron death was mediated by the interleukin (IL)-1 receptor through ERK phosphorylation (Lu et al., 2005). However, ERK plays a protective role against apoptosis in several cases (Onyango et al., 2005) through inhibition of caspases activation. Activation of ERK under some conditions of stress or by growth factors has been shown to confer a survival advantage to cells. In rat PC12 pheochromocytoma cells, for example, inhibition of ERK is critical for induction of apoptosis and overexpression of constitutively active mutants of MEK1 prevents the apoptosis induced by nerve growth factor (Xia et al., 1995). Activated ERK1/2 phosphorylates many substrates in all cellular compartments (Roux and Blenis, 2004). Harris FM et al. (2004) reported that the expression of ApoE in neurons is regulated by a diffusible factor or factors released from astrocytes and that this regulation depends on the activity of the ERK kinase pathway in neurons. CREB has a neuroprotective role in brain injury models (Mabuchi et al., 2001) and ERK is one of the multiple kinases that can phosphorylate CREB via ribosomal S6 kinase (Walton and Dragunow, 2000). ERK1/2 may inhibit apoptosis through the phosphorylation of p90RSK. This inactivates the proapoptotic Bad and stimulates several transcription factors, such as CREB, that increase the expression of the mitochondrial membrane pro-survival proteins Bcl-xL and Bcl-2. In the nervous system, the ERK1/2 signalling cascade is an important regulator of neuronal differentiation, plasticity and survival (Hetman and Gozdz, 2004). For instance, in myeloma cell lines, IL-6 induces cell growth through the Ras-ERK pathway (Ogata et al., 1997). In lymphoblasts from AD patients, apoptosis is prevented through the modulation of ERK1/2 activity by Ca2+/calmodulin (CaM) (Bartolomé et al., 2007). ERK1/2 pathway activation protects neurons from GSK-3β-induced apoptosis through phosphorylation at Ser-9 (Eldar-Finkelman et al., 1995) and by a Ser9 and Tyr216 phosphorylation-independent mechanism (Hetman et al., 2002). In the brain, ERK has been reported to mediate long-term potentiation (Sweatt, 2001). Addictionally, Ras/ERK pathway might protect sympathetic neurons against apoptosis (Xue et al., 2000). ERK is activated in neurons and glia, in regions that exhibit cell damage, such as after neonatal rat cerebral hypoxia-ischemia (Wang et al., 2003) and through ROS (Yamakawa et al., 2002) and/or death stimuli (Toborek et al., 2007). The activation of Ras, Raf-1, MEK1/2 and ERK1/2 in perinuclear and nuclear regions in neurons is enhanced in AD (Gärtner et al., 1999; Pei et al., 2002; Mei et al., 2006). Activation of ERK dramatically stimulates APP processing at the cell surface by α-secretase and inhibits Aβ secretion (Gandy et al., 2004). In AD neurons, ERK activation occurs at the same site of oxidative damage and hyperphosphorylation of tau (Perry et al., 1999; Ferrer et al., 2001). 56
Introduction
Cholesterol The cellular membrane is enriched in different types of lipids being phospholipids, glycolipids and cholesterol the three most important asymmetrically distributed between the two leaflets of lipid bilayers. Cholesterol is a major component of neural membranes. It plays an essential role in membrane organization, dynamics and function. Cholesterol is a crucial component for the formation and maintenance of the permeability of the cellular membrane and several studies indicate that membrane cholesterol has a direct role in cell signalling and in the regulation of the activity of membrane-bound enzymes, ion channels and receptors (Simons and Ikonen, 2000; Breusegem et al., 2005). In the plasma membrane, cholesterol is strictly organized into structural and kinetic pools like leaflet domains, lipid rafts (see “Lipid rafts”) or caveolae (small membrane invaginations, rich in the protein caveolin, seen in many cell types). The brain is the richer human organ in cholesterol containing a quarter of the total cholesterol body content, despite representing only 2% of body weight. Therefore cholesterol is present abundantly in the membranes of the glial cells and neurons and in the membranes of myelin. Cholesterol turnover in the central nervous system (CNS) is slow (in humans the cholesterol turnover rate is about 0.02%, Dietschy and Turley, 2002), being the half life estimated in around 5 years in humans (Björkhem et al., 1998). However, the CNS is a site of high lipid turnover compared with other organs. The presence of cholesterol in neuronal membranes is known to induce large changes in membrane fluidity. In the blood, cholesterol is carried by diverse types of lipoproteins with different densities in agreement with its composition in lipids and proteins. Cholesterol can be internalized through lipoproteins, which carry cholesterol to the membrane receptors.
Biosynthesis of cholesterol
All eukaryotic cells, except the mature red blood cells lacking nucleus and organelles, are able to synthesize cholesterol from acetate. The cholesterol in circulation comes from the liver or can be obtained from the diet. Nevertheless, about 95% of the cholesterol in the brain is synthesized in situ by oligodendrocytes, astrocytes and neurons. Neurons can produce enough cholesterol to survive, but the formation of new synapses needs more quantity of cholesterol and other lipids that are provided by glial cells such as astrocytes (Pfrieger, 2003).
57
Introduction Astrocytes can synthesize two or three fold more cholesterol than neurons and secrete cholesterol as lipoprotein particles (Vance et al., 2005). The pathway that conducts to the synthesis of cholesterol is a rich source of hydrophobic molecules important for several inter and intracellular functions. The cholesterol synthesis is divided in three phases (Berg et al., 2001; Figure 1.10). First, in cytosol, the condensation of acetyl-coenzyme A and acetoacetyl-coenzyme A forms 3-hydroxy-3methylglutaryl-coenzyme A (HMG-CoA) catalyzed by HMG-CoA synthase. HMG-CoA is reduced to form mevalonate by HMG-CoA reductase, a limitant enzyme that is highly regulated, found mainly in the ER membrane. Second, mevalonate is phosphorylated by two sequential phosphate transfers from ATP, leading to a pyrophosphate derivative. Pyrophosphomevolanate decarboxylase catalyzes ATP-dependent decarboxylation, with dehydration, to yield isopentenyl pyrophosphate (which is needed to the isopentenylation of tRNA and many others biomolecules). Isopentenyl pyrophosphate is the first of several compounds in the pathway that are named isoprenoids, because contain a structure similar to isoprene. Isopentenyl pyrophosphate isomerase converts isopentenyl pyrophosphate in dimethylallyl pyrophosphate and vice versa by a mechanism that involves protonation and deprotonation. Prenyl transferase catalyzes the condensation of dimethylallyl pyrophosphate with isopentenyl pyrophosphate to form geranyl pyrophosphate (that is implied in the phosphorylation of tau protein) and this with another isopentenyl pyrophosphate to form farnesyl pyrophosphate. These two pyrophosphates may be used for geranylation and farnesylation of proteins that anchor these proteins to membranes and in that way regulating its activity. It is the case of many proteins involved in cell signalling like Ras and other GTPases (see “Small GTPases”). Farnesyl pyrophosphate may also be used in the production of many molecules, such as ubiquinones including the coenzyme Q10, a co-factor in the mitochondrial respiratory chain with an important antioxidant role in both mitochondria and lipid membranes in its reduced form (ubiquinol or CoQ10 H2). Farnesyl pyrophosphate serves also as precursor for synthesis of various non-steroidal isoprenoids. Squalene synthase catalyzes the condensation of two molecules of farnesyl pyrophosphate, with reduction by NADPH (reduced nicotinamide adenine dinucleotide phosphate), to yield squalene which migrates to the ER. Finally, inside the ER, squalene is converted, after a two step cyclization, into lanosterol that is transformed into cholesterol after 19 reactions. The mevalonate pathway is very important to ensure normal growth, differentiation and maintenance of neuronal tissues, since the activity of HMG-CoA reductase and cholesterol synthesis are very high in the early phases of brain development.
58
Introduction
59
Introduction Figure 1.10 – Cellular cholesterol synthesis and internalization. Cholesterol synthesis can be divided in 3 steps. The first begins with the production of acetoacyl CoA in the cytosol and terminates with the generation of mevalonate from 3-hydroxy-3-methylglutayl coenzyme A (HMG-CoA) by HMG-CoA reductase (limitant enzyme present in the membrane of endoplasmic reticulum (ER). The second begins with the phosphorylation of mevalonate in the cytosol, passes by the generation of isoprenoids (for instance, farnesyl pyrophosphate) and terminates with the production of squalene. Finally, squalene enters to ER where is used to produce cholesterol by a series of reactions. Statins inhibit HMG-CoA reductase reducing, by that way, the synthesis of cholesterol and all the intermediates in cholesterol synthesis, like geranyl pyrophosphate and farnesyl pyrophosphate (which is converted in geranylgeranyl pyrophosphate). Cholesterol can be converted in cholesterol esters by acetyl-CoA cholesteryl acetyltransferase (ACAT). Cholesterol in lipoproteins can be internalized through the low density lipoprotein receptors (LDLR). When Aβ binds to apolipoproteins it can be internalized by the cells.
Regulation of cholesterol levels
As the cholesterol homeostasis is very important for many mechanisms within neuronal cells, it is finely controlled in the brain through various interdependent processes, like synthesis, storage, degradation and transport. The cholesterol flux in AD individuals is enhanced (Lütjohann et al., 2000). Cholesterol binding to lipoproteins can cross BBB (in small quantities) and plasma membranes through binding to lipoprotein receptors. The synthesis of cholesterol decreases with the increase of internalized cholesterol. Internalized cholesterol passes through the endosome via a mechanism involving NPC1 (Neimann-Pick disease type C protein 1) and then is transported to the plasma membrane in vesicles. HMGCoA reductase and acyl-coenzyme A cholesteryl acyltransferase (ACAT) play a strict game of feedback, synthesizing and eliminating cholesterol from the cytosol, respectively.
Regulation of cholesterol synthesis
Cholesterol synthesis can be regulated in short and long-term, being the HMG-CoA reductase the rate-determining step and the major control point. When cellular ATP is low, cellular AMP is high and activates AMP-dependent protein kinase (a heterotrimeric complex, consisting of the catalytic α-subunit and β- and γ-subunits, found in all eukaryotes) which catalyzes the inhibition of HMG-CoA reductase by phosphorylation. Thus, the cell does not use ATP energy in synthesizing cholesterol when ATP levels are low. Mevalonate, farnesol (dephosphorylated farnesyl pyrophosphate), cholesterol and oxidized derivatives of cholesterol stimulate the degradation of HMG-CoA reductase. HMG-CoA reductase has a transmembrane sterol-sensing domain with an important role in activating the degradation of the enzyme via the proteosome. Cholesterol also decreases the activity of HMG-CoA reductase, by a feedback mechanism. 60
Introduction The synthesis of cholesterol can be regulated by SREBP, a family of transcription factors that regulate intracellular cholesterol and lipid metabolism (Sundqvist and Ericsson, 2003; Goldstein et al., 2006). When the levels of sterols are low, the cholesterol sensor SCAP in the ER conducts SREBP via COPII vesicles to the Golgi. At this point, SREBP is cleaved by proteases (Site-1/2 proteases) to release the transcriptionally active N-terminal of the protein. This enters the cell nucleus (Yang et al., 2002) and binds to the sterol-regulatory elements (SRE, the DNA sequence TCACNCCAC) activating the transcription of genes that have this sequence in their promoters, including genes of enzymes of cholesterol synthesis pathway (Sakakura et al., 2001), like the genes for the HMG-CoA reductase, FAS, the key regulatory enzyme in lipid biosynthesis (Bennett et al., 2004; Porstmann et al., 2005) and low density lipoprotein (LDL) receptor, the major determinant of plasma LDL cholesterol concentrations (Horton et al., 2003). When the levels of cholesterol are sensed to be normal, the SCAP/SREBP complex is retained in the ER by the proteins Insig-1/2 (Yang et al., 2002). SREBP family contains three members: SREBP-1a, SREBP-1c and SREBP-2 (Eberlé et al., 2004). SREBP gene promoters also contain the SRE sequence and thus, may be regulated by positive feedback. SREBP-1c has been shown to be a factor involved in adipocyte differentiation, SREBP-1a and SREBP-1c prefer genes involved in the synthesis of fatty acids and SREBP-2 principally regulates genes of cholesterol metabolism (Amemiya-Kudo et al., 2002; Horton et al., 2003). Furthermore 24- and 25-hydroxycholesterol are potent inhibitors of cholesterol biosynthesis and LDL receptor activity through inhibition of SREBP.
Elimination of cholesterol in the brain
When cholesterol exists in excess in the brain, it must be eliminated. For that, there are various mechanisms, for example the cholesterol associated with lipoproteins can cross the BBB to the blood. However, only a small amount of brain cholesterol passes through the BBB by this mechanism and other mechanisms are necessary. Cholesterol may be eliminated by acetylation or oxidation or released through a mechanism involving the ATP-binding cassette transporter (ABCA)1. ACAT is the enzyme responsible for acetylation. Cholesterol can be transformed into cholesterol esters in the ER by ACAT for storage as lipid droplets thus regulating free cholesterol levels. ACAT is also responsible for the elimination of membranar cholesterol. Cholesterol may be degraded by monooxygenases of the family of cytochrome P450 that use NADPH and O2 (Berg et al., 2001; Reiss et al., 2004). The principal oxysterol resulting from the action of these enzymes is 24S-hydroxycholesterol. The enzyme 24S61
Introduction hydroxylase is only expressed in the brain (Björkhem et al., 1998) and so, 24Shydroxycholesterol is a marker for cholesterol degradation in the brain. Others oxysterols may be produced, like 27-hydroxycholesterol, 25- hydroxycholesterol or, in low quantities, the 7α-hydroxycholesterol (this molecule in cells is pro-apoptotic in nanomolar concentrations). Oxysterols are more soluble and cross the BBB more rapidly than cholesterol and are subsequently converted into normal bile acids by the liver (Berg et al., 2001; Reiss et al., 2004) or excreted in bile in its sulphated and glucuronidated form. These oxysterols are potent regulatory molecules that act in a variety of processes, including the modulation of cholesterol homeostasis (Reiss et al., 2004). For instance, 25hydroxycholesterol is one of the most potent suppressors of cholesterol synthesis in cultured cells. 24S-hydroxycholesterol produced by neurons is taken by astrocytes and stimulates cholesterol efflux (Pfrieger, 2003; Björkhem and Meaney 2004). The levels of 24Shydroxycholesterol decrease as a consequence of severe neurological diseases affecting the number of neurons in the brain (Bretillon et al., 2000). In the brain of AD patients, there is a strong correlation between 24S-hydroxycholesterol and lathosterol, not detected in controls (Heverin et al., 2004), suggesting that cholesterol is degraded mostly in 24Shydroxycholesterol. In the presence of ROS, cholesterol may be oxidized in 7ketocholesterol, 7α-hydroxycholesterol and others. Brain cholesterol may also be transported to the blood associated, for example, to ApoE. This process involves ABCA1 (Wahrle et al., 2004). ABCA1 transports cholesterol and phospholipids from cells to high density lipoproteins (HDL). This complex seems to stimulate cholesterol efflux from CNS preventing the accumulation of cholesterol in neurons and glial cells (Koldamova et al., 2003). Cultured mice astrocytes without ABCA1 secrete lipoproteins with significantly reduced ApoE and cholesterol. These mice have considerably reduced ApoE levels in cortex (80% less than normal) and CSF (98% less than normal) (Wahrle et al., 2004). People with a polymorphism (R219K) in the ABCA1 gene have less cholesterol in CSF and this polymorphism seems to decrease in 1.7 years the manifestation of AD (Wollmer et al., 2003).
Lipoproteins and AD
Lipoproteins are spherical particles containing proteins and lipids. There are many types of lipoproteins such as enzymes, transporters, structural proteins, antigens, adhesins and toxins. Lipoproteins may be classified (in order from larger and less dense - more lipid than protein - to smaller and more dense - more protein than lipid) as: chylomicrons, VLDL, 62
Introduction intermediate density lipoprotein (IDL), LDL (that carry cholesterol from the liver to cells) and HDL (which carry cholesterol from the body’s tissues to the liver that excretes it in the form of bile salts). The proteins in lipoproteins are called apolipoproteins and act as enzyme co-factors, receptor ligands and lipid transfer carriers. There are many classes of apolipoproteins such as ApoE which is the principal apolipoprotein responsible for the transport, uptake and redistribution of lipids in the brain (Berg et al., 2001; Reiss et al., 2004). Other apolipoproteins in the CNS are ApoD, ApoAI and ApoJ, although in very low amounts. AD patients present increased LDL levels and reduced HDL levels in comparison with control subjects (Kuo et al., 1998). LDL are more dangerous particles than HDL, because may lead to the development of diverse cardiovascular diseases like atherosclerosis, also related to AD (Hofman et al., 1997). Hypercholesterolemia, in particular LDL, augments the accumulation of Aβ and accelerates AD pathology (Kuo et al., 1998; Refolo et al., 2000). On the other hand, elevated HDL-cholesterol was associated with a significantly decreased risk of dementia (Bonarek et al., 2000). Burgess and collaborators (2006) reported that in early AD stages, the presence of Aβ in plasma may affect peripheral lipid metabolism leading to a significant increase (about 30%) in VLDL-triglyceride secretion rate without affecting total cholesterol levels. In addition, some studies showed that AD patients have elevated serum TG levels (Sabbagh et al., 2004; Cankurtaran et al., 2005), but others have found no association (Romas et al., 1999).
Low density lipoprotein receptors
The uptake of exogenous cholesterol in the brain requires the internalization of the lipoprotein-cholesterol through the binding of apolipoprotein to its surface receptor. Low density lipoprotein receptor (LDLR) family is a group of lipoprotein receptors expressed in the CNS of several animals. In mammalians, they act as clearance receptors for various ligands like lipoproteins, proteases, protease inhibitors and vitamin carriers. This family contains more than ten identified mammalian members that function in receptor-mediated endocytosis and cellular signalling (May et al., 2005): LDLR, LRP1, LRP1b, VLDL receptor (VLDLR), megalin/LRP2, multiple epidermal growth factor (EGF) repeat-containing protein (MEGF)7/LRP4, LRP5, LRP6, ApoE receptor (ApoER)2/LRP8
and SorLA-1/LRP11.
Neurons express LDLR, LRP, ApoER2 and VLDLR, astrocytes express LDLR and LRP and microglia express VLDLR and LRP. Lipoproteins bind to these receptors through the apolipoproteins and can then cross the BBB or can be internalized into the cells. 63
Introduction Subsequent to the binding of the lipoprotein complex (e.g. ApoE and cholesterol) to a receptor, the lipoprotein–LDL receptor complex is internalized to late endosomes and can be delivered to lysosomes for subsequent degradation, releasing lipids that are transported to the ER through a vesicle that contains the NPC1 protein (this protein is inactive in Neimann-Pick disease). Alternatively, lipoproteins can be targeted for transcytosis across the BBB into the plasma (Herz, 2003). In AD, the receptors that bind ApoE are expressed in brain regions where amyloid deposition and neuronal loss occur (Andersen and Willnow, 2006). These receptors are important to supply cholesterol to neurons, but may bind ApoE with Aβ from the extra-neuronal space contributing to the internalization of this peptide. The LRP can also directly bind Aβ and promote its transport across BBB (Deane et al., 2004). The transport of Aβ across the BBB to the blood has been shown to be mediated by LRP1 and the uptake into the brain can be mediated by RAGE and the macrophage scavenger receptor (Zlokovic, 2004). The activation of ApoE receptors can promote the activation of several signal pathways including the activation of PI3K/Akt, GSK-3β and ERK and the inhibition of JNK (Beffert et al., 2002; Hoe et al., 2005). LRP1 interacts with the Kunitz-type protease inhibitor domain in APP770 and APP751 mediating its degradation (Rebeck et al., 2001). However, this receptor can also bind the 3 forms of APP by the adaptor protein Fe65 in the cytoplasm, which binds the cytoplasmic tails of APP and of LRP1 (Pietrzik et al., 2004). This accelerates endocytosis of APP and consequently increases the cleavage by β- and γ-secretases enhancing Aβ formation (Cam et al., 2005). LRP1B may also bind APP, but the rate of endocytosis is slower permitting the cleavage by α-secretase and diminishing Aβ production (Cam et al., 2004). Other LDLRs can also bind APP and affect its processing (Andersen et al., 2005; Hoe et al., 2005b; Offe et al., 2006). SorLA/LR11 alters APP trafficking in endosomal compartments diminishing APP processing by β- and γ-secretases (Offe et al., 2006; Spoelgen et al., 2006).
Apolipoprotein E
ApoE is synthesized in the human brain mainly by astrocytes but also in oligodendrocytes, ependymal layer cells and neurons. In vivo, ApoE is important for the translocation of cholesterol from astrocytes to neurons and in the regulation and homeostasis of this process (Levi et al., 2005). The APOE gene that codifies this 34 kDa secretory protein with 299 amino acids is localized in the chromosome 19 and has 3 alleles: ε2, ε3 e ε4 that codify to ApoE2, ApoE3 and ApoE4 proteins, respectively. The more frequent allele is the ε3 64
Introduction followed by the ε2. ApoE levels in the plasma, cortex and hippocampus diminish from the ε2 to ε3 alleles and are lower when the ε4 allele is present (Figure 1.11). The ε4 variant is the most frequent genetic risk factor for late-onset AD, being strongly correlated with the increase of the prevalence of the disease (Strittmatter et al., 1993; Bonarek et al., 2000; Herz and Beffert, 2000) and with the decrease of the age of onset (Corder et al., 1993) and morerapid cognitive decline (Packard et al., 2007). However, the presence of the ε4 allele is neither necessary nor sufficient to cause AD (Puglielli et al., 2003). In studies conducted in transgenic animal models, ApoE has been shown to be involved in the accumulation of Aβ in an allele dependent manner, being the ApoE4 the most potent (Holtzman et al., 2000; Fryer et al., 2003). ApoE3 and ApoE2 bind Aβ better than ApoE4 (Tokuda et al., 2000). Consequently, ApoE4 isoform decreases the internalization of Aβ and subsequent degradation, and thus increases its oligomerization. On the other hand, the binding of Aβ to ApoE avoids by itself the formation of aggregates (Beffert et al., 1999). For this reason, the aggregation process is more efficient in the presence of ApoE4. Therefore, altering the cholesterol content of ApoE-containing lipid particles may decrease the interaction between ApoE and Aβ and thereby affects Aβ aggregation and clearance, enhancing
Aβ
accumulation
and
formation
of
senile
plaques.
Hence,
during
hypercholesterolemia, more lipids bind to ApoE reducing its ability to inhibit Aβ aggregation (Beffert and Poirier, 1998). In that way, the increase in lipids in the brain contributes to enhance Aβ toxicity. Nevertheless, other hypothesis was proposed. Some studies reported that lipidated ApoE binds aggregated Aβ in an isoform-specific manner, being ApoE4 much more efficient than the other forms, and may cause enhanced deposition of the peptide (Stratman et al., 2005). Once internalized to lysosomes, part of ApoE-Aβ complex is degraded. The fibrillization of the reminiscent Aβ peptide bound to ApoE is promoted and fibrils are then secreted, a process that is accelerated by ApoE4. On the other hand, the internalization of lipoproteins increases the cholesterol content of the cell and, consequently, may increase the levels of membrane cholesterol which enhances the production of Aβ (see “Cholesterol and AD”). As lipoproteins with ApoE4 tend to be enriched in cholesterol, ApoE4 could thus increase the production and secretion of Aβ (Puglielli et al., 2003). Additionally, hypercholesterolemia seems to increase ApoE levels in the brain exacerbating the accumulation of Aβ (Wu et al., 2003).
65
Introduction
Figure 1.11 – The presence of APOEε4 allele may induce the activation of GTPases. In the presence of the APOEε4 allele lower levels of apolipoprotein E (ApoE) are present in the brain and, as a result, cholesterol transport into cells decreases. In order to compensate this depletion, and because neurons need high cholesterol levels, the biosynthesis of cholesterol increases. Consequently, the production of cholesterol biosynthesis intermediates geranyl pyrophosphate (GPP) and farnesyl pyrophosphate (FPP) is also enhanced, activating GTPases. As statins inhibit cholesterol synthesis, they may normalize the levels of active GTPases in people carrying the APOEε4 allele.
ApoE4 was also shown to enhance mitochondrial damage (Gibson et al., 2000), affect the extent of neuronal cell loss and cholinergic activity and is associated with a poor synaptic remodelling and defective compensatory plasticity in vulnerable brain areas in AD (Arendt et al., 1997; Beffert et al., 1998). Furthermore, ApoE4 and Aβ1-42 act synergistically to reduce neuronal viability (Manelli et al., 2006). ApoE4 has a decreased antioxidant activity (Miyata and Smith, 1996), deregulates neuronal signalling pathways (Herz and Beffert, 2000) and disrupts cytoskeletal structure and function (Nathan et al., 1995) compared with the other ApoE isoforms. ApoE can modulate the effects of sAPP on Ca2+ homeostasis in cultured rat hippocampal neurons, enhancing the ability of sAPP to lower intracellular Ca2+ levels, with ApoE3 being more effective than ApoE4 (Barger and Mattson, 1997).
Lipid rafts
To better understand the relationship between membrane cholesterol and Aβ production is necessary to understand the concept of lipid rafts (Figure 1.12). Lipid rafts or 66
Introduction detergent-resistant membrane (DRM's) domains of the membrane enriched in sphingolipids and cholesterol. The rafts are floating on the exofacial side of the membrane bilayer (Simons and Ehehalt, 2002) and have important roles in neural cell functions including signal transduction, adhesion, sorting, trafficking, and organization of bilayer constituents like enzymes, receptors and ion channels (Simons and Ikonen, 2000). Raft formation in neural membranes arises by self association of sphingolipids via their long saturated hydrocarbon chains. Cholesterol locates between these hydrocarbon chains under the large head groups of the sphingolipids compacting them (Simons and Ikonen, 2000). In these microdomains the forces between the membrane lipids are stronger despite membrane fluidity is allowed (Laude and Prior, 2004). This condition results in the “preference” of some proteins for these regions. These proteins suffer lipid modifications that allow them to intercalate into the lipid raft structure and, consequently, the proteins tend to be constitutively associated with raft domains (Simons and Ehehalt, 2002). Gα subunits of heterotrimeric G proteins and endothelial nitric oxide synthase NOS (eNOS) are two constitutive raft resident proteins. Other proteins can move in and out of the rafts controlled by factors such as ligand-binding or oligomerization (Zacharias et al., 2002). These regions occupy around half of the plasma membranes and each have about twenty or thirty protein molecules (Simons and Ehehalt, 2002), what leads to suppose that some enzymes in the rafts rarely interact with the corresponding substrate. Under these conditions, many rafts cluster together permitting the interaction between the enzymes and corresponding substrates. After endocytosis, the rafts can cluster and become redistributed and some enzymes can interact with the substrates. This may be the case of BACE and APP (Simons and Ehehalt, 2002). If cholesterol is eliminated from the rafts, the majority of the proteins dissociates from the rafts and loses their normal functions.
Cholesterol and AD
Many of the mechanisms of neurotoxicity of AD depend on the brain cholesterol homeostasis. To some authors the cholesterol homeostasis dysfunction is the primary cause and the major target for therapy of various diseases like AD (Koudinov and Koudinova, 2005). The first evidence that shows the involvement of cholesterol metabolism in AD emerged through the studies of ApoE, which is the major cholesterol carrier protein in the brain and the major genetic risk factor for late-onset AD. Some other genes have been related to AD, for example the gene CYP46 that encodes 24-hydroxycholesterol (Papassotiropoulos et al. 2003). Other evidence that correlates cholesterol and AD is the link between vascular 67
Introduction risk factors and dementia (Hofman et al., 1997). HDL is critical for the maturation of synapses and the maintenance of synaptic plasticity (Koudinov and Koudinova, 2001; Mauch et al., 2001). Epidemiological studies suggest that diet may influence the risk for AD. People with similar pool of genes in different environments have distinct risks for AD. For instance, Nigerians in Africa and Japanese people living in Japan have a much lower incidence of AD when compared with African and Japanese Americans, respectively, living in the United States of America (Graves et al., 1999; Hendrie et al., 2001). When the lifestyle of populations screened is compared, the diet and fat intake seems to play an important role (Hendrie et al., 2001).
Figure 1.12 – Lipid rafts. (A) Lipid rafts are regions of cell membranes enriched with sphingolipids and cholesterol in the outer leaflet. These regions are thicker and have less fluidity than the rest of the membrane. Certain proteins suffer modifications and anchor to lipid rafts. Proteins with covalently attached long-chain acyl groups are normally found in the inner leaflet whereas GPI-liked proteins are mainly found in the outer leaflet. Prenylated proteins, like GTPases as Ras, tend to be excluded from rafts. (B) Atomic force microscopy shows the great thickness of lipid rafts. Sharp peaks represent GPI-linked proteins and are found almost exclusively in rafts.
The studies concerning plasma cholesterol levels in AD patients and whether high levels of plasma lipids could be considered a risk factor for AD are conflicting. In some reports, increased levels of midlife total cholesterol are associated with a 2-3-fold increase in the risk to develop dementia and AD later in life (Kivipelto et al., 2002; Pappolla et al., 2003; Panza et al., 2006), and this risk is higher in people with elevated systolic blood pressure and hypercholesterolemia (Kivipelto et al., 2001). Other studies reported that AD subjects may 68
Introduction have elevated total cholesterol and triglyceride levels (Sabbagh et al., 2004). In other studies, an opposite relationship was found (Mielke et al., 2005). Still, some studies described that there isn’t any significant relationship between the levels of cholesterol and the risk to develop dementia and AD (Romas et al., 1999; Solfrizzi et al., 2004). The consumption of fat seems to be important to increase AD risk but is not the only environment factor that influences the incidence of the disease. Other nutrients in diet, the level of exercise and other aspects in the lifestyle can also be important and may mask the influence of cholesterol in diet. So, all these variables must be considered in epidemiological studies. Although epidemiological studies are inconclusive, genetic and biochemical studies have strengthened the hypothesis that elevated levels of cholesterol are a risk factor for AD and decreased neuronal cholesterol levels inhibit Aβ formation (Marx, 2001; Simons et al., 2001; Wolozin, 2001; Buxbaum et al., 2002). Low levels of HDL-cholesterol were associated with an elevated risk of dementia, whereas high levels of HDL-cholesterol were associated with a larger hippocampal volume and protection against dementia and AD (Wolf et al., 2004). HDL, in majority in human brain, diminishes Aβ toxicity (Farhangrazi et al., 1997) by reducing Aβ aggregation (Olesen and Dago, 2000). High levels of LDL and total cholesterol are associated with cognitive impairment (Yaffe et al., 2002) and had been found in postmortem analysis of AD individuals compared to controls (Kuo et al., 1998). In animals, hypercholesterolemia induced through diet significantly intensifies AD neuropathology (Shie et al., 2002). For example, animals with elevated consumption of cholesterol present increased intraneuronal and extracellular deposition of Aβ and a reduction of α-sAPP in the hippocampus and frontal cortex (Refolo et al. 2000), as occurs in the brain of AD patients. αsAPP is also down-regulated in cultured cells incubated with cholesterol (Bodovitz and Klein, 1996). On the other hand, depletion of cellular cholesterol results in increased levels of α-sAPP, maybe because this enhances the expression of ADAM10 (an α-secretase) (Kojro et al., 2001). In addition, the cholesterol-lowering drug BM15.766 reduces brain Aβ peptides and Aβ-load by more than twofold (Refolo et al., 2001). The activity of lecithin cholesterol acyltransferase (LCAT) is lower in AD patients comparatively with non-demented people (Puglielli et al., 2003). This enzyme is responsible for the reverse transport of cholesterol in humans, that is, eliminates the cholesterol of the peripheral cells (HDL-mediated removal of surplus cholesterol). Consequently, it is expected that patients with AD have levels of cholesterol in the brain more elevated than people without the disease. Down’s syndrome patients which have 3 copies of APP gene also have the activity of this enzyme decreased (Lacko et al., 1983) what indicates that this effect is related with the etiology and/or vascular pathology present in both diseases. 69
Introduction Homeostasis and distribution of cholesterol seems to regulate the formation of Aβ in studies with cellular and animal models of AD (Puglielli et al., 2003). BM15.766, an inhibitor of the dehydrocholesterol reductase enzyme that catalyses the last step in cholesterol biosynthesis reduces Aβ levels in a transgenic mouse model of AD (Refolo et al., 2001). According to Simons and collaborators (1998), cholesterol is necessary for the formation of Aβ. Subasinghe and co-workers (2003) demonstrated that the toxicity of Aβ in vascular smooth muscle cells is a direct consequence of its interaction with lipids in the cellular membrane. In another study in hippocampal neurons, low cholesterol content was shown to reduce the ability of Aβ to form oligomeric aggregates (the most toxic) without disturbing the production of Aβ (Schneider et al., 2006). High levels of intracellular cholesterol affect the production of Aβ (Puglielli et al., 2003) enhancing the release of Aβ in vitro and in vivo as a result in the alteration of the composition of the membrane lipids that change the activity of secretases, since they are transmembrane proteases, facilitating the processing of APP by β- and γ-secretase (Wahrle et al., 2002; Burns et al., 2003). On the other hand, low intracellular cholesterol favours the processing of APP by α-secretase leading to reduced Aβ formation, because the lipid-rafts where β-secretase is active are disrupted (Kojro et al., 2001; Ehehalt et al., 2003). That is, increasing the concentration of cholesterol, the activity of α-secretase slightly diminishes and the activity of β-secretase extremely increases (Wolozin, 2002b) what leads to an augment of Aβ production and consequently, enhances its deposition and neurotoxicity. By the contrary, cholesterol depletion in cultured hippocampal neurons, for example, inhibits the formation of Aβ (Simons et al., 1998). In addition, membrane cholesterol levels can modulate the association of Aβ with GM1 ganglioside (a raft component) that seems to be involved in the aggregation of Aβ in AD brain (Kakio et al., 2002). Cholesterol depletion was also shown to reduce peptide fibrillogenesis promoted by the binding of soluble Aβ with rafts (Mizuno et al., 1999). Mutations and pharmacological inhibitors of the Niemann-Pick complex cholesterol transport pathway change the localization of presenilin and enhance Aβ production, indicating that the subcellular distribution of cholesterol also influences APP cleavage (Runz et al., 2002; Burns et al., 2003; Jin et al., 2004). Cholesterol-derived aldehydes can change Aβ, initially promoting Schiff base (functional group that contains a carbon-nitrogen double bond with the nitrogen atom connected to an aryl or alkyl group but not hydrogen) formation and then accelerating the early steps of amyloidogenesis (Hayashi et al., 2004; Bieschke et al., 2005). In addition, intracellular cholesterol may also play a crucial role in the modulation of tau phosphorylation and the maintenance of microtubule stability (Fan et al., 2001). 70
Introduction Furthermore, free fatty acids stimulate the assembly of both amyloid and tau filaments (Wilson and Binder, 1997). Despite the evidence supporting that high cholesterol levels increase the amyloidogenic processing of APP and Aβ release, several results have been published that are in contradiction with this hypothesis. PCl2 cells and cortical neurons with plasma membranes enriched with cholesterol were shown to be more resistant to the cytotoxic action of Aβ (Arispe and Doh, 2002; Sponne et al., 2004). In addition, methyl β-cyclodextrin (MβCD) that selectively removes cholesterol from membranes (and consequently disrupts membrane raft domains), decreases α-sAPP levels in human embryonic kidney cells (HEK 293 cells) stably transfected with human APP751 (Bodovitz and Klein, 1996) indicating that cholesterol reduces the fluidity of membranes impeding the interaction of APP and αsecretase. Moreover, elevated dietary cholesterol in animals led to significant reduction in brain levels of secreted APP derivatives (Howland et al., 1998). In the study of AbadRodriguez and colleagues (2004) a moderate decrease in cholesterol levels (
