Links between plasma apoE and glucose metabolism, brain insulin signaling, and synaptic integrity: Relevance to Alzheimer’s disease pathophysiology

(1)

Links between plasma apoE and glucose metabolism, brain insulin signaling, and synaptic integrity

Relevance to Alzheimer’s disease pathophysiology

Anna Edlund

Anna Edlund Links between plasma apoE and glucose metabolism, brain insulin signaling, and synaptic integrity

Doctoral Thesis in Neurochemistry with Molecular Neurobiology at Stockholm University, Sweden 2021

Department of Biochemistry and Biophysics

ISBN 978-91-7911-422-0

(2)

(3)

Links between plasma apoE and glucose

metabolism, brain insulin signaling, and synaptic integrity

Anna Edlund

Academic dissertation for the Degree of Doctor of Philosophy in Neurochemistry with Molecular Neurobiology at Stockholm University to be publicly defended on Monday 24 May 2021 at 14.00 in Magnélisalen, Kemiska övningslaboratoriet, Svante Arrhenius väg 16 B, and online via Zoom, public link is available at the department website.

Abstract

Human apolipoprotein E (apoE) exists as three main isoforms called apoE2, apoE3, and apoE4, of which the E4 isoform is associated with increased Alzheimer’s disease (AD) risk. Brain glucose hypometabolism, linked to synaptic dysfunction, occurs years before symptom onset in AD, especially in APOEε4-carriers. An association between a higher ratio of plasma apoE4 to apoE3 levels and cerebral glucose hypometabolism was recently discovered in cognitively healthy APOEε3/ε4 subjects. A lower plasma apoE level, regardless of isoform, is linked to increased AD risk. How the plasma apoE level affects neurodegenerative processes in the brain is poorly understood, given that apoE doesn’t cross the blood-brain barrier (BBB). The main aim of this thesis was therefore to investigate a relationship between plasma apoE and features of AD pathophysiology. We explored plasma apoE levels and dimer/monomer formation in APOEε3 and APOEε4 homozygous controls, in patients with mild cognitive impairment (MCI) and AD. In APOEε4-carriers versus non-carriers, plasma apoE levels were lower and significantly correlated with AD biomarkers. ApoE3 homodimers were less in AD patients than in controls. We next examined potential links between plasma apoE, glucose, and insulin levels in the previously examined cognitively healthy APOEε3/ε4 subjects. Lower plasma apoE3 was associated with higher glucose levels in males and subjects with body max index above 25. Plasma glucose levels were negatively correlated with the cerebral metabolic rate of glucose and neuropsychological test scores. To explore the potential effects of a hepatic APOEε4 phenotype on the brain, we compared liver humanized mice with an APOEε4/ε4 versus an APOEε2/ε3 genotype. Mice with an APOEε4/ε4 liver exhibited reduced endogenous mouse apoE in the brain, accompanied by changes in markers of synaptic integrity and insulin signaling. Plasma apoE4 levels were negatively associated with some of the assessed markers. We further explored the effects of a high-fat diet (HFD) in mice with livers humanized with the AD risk-neutral APOEε3/ε3 genotype.

Endogenous mouse apoE was increased in the hippocampus following an HFD, with concomitant effects on levels of synaptic markers. In the cortex, we found altered levels of insulin signaling and synaptic markers. Together, our findings indicate that alterations in apoE levels or distribution, hepatic APOEε4 phenotype, and HFD contribute to AD-related pathological processes.

Amyloidogenic processing of the amyloid precursor protein (APP) gives rise to Aβ peptides that assemble into the Aβ plaques found in AD. The binding of the adaptor protein Fe65, through its PTB2, to APP might enhance amyloidogenic APP processing. Fe65 is localized both in the cytoplasm and in the nucleus, with compartment-specific biological functions.

Mechanisms affecting Fe65 subcellular localization are poorly understood. We explored the impact of the different Fe65 interaction domains WW and PTB2 and APP processing on Fe65 cellular localization. By transfecting Fe65-domain deletion constructs into neuroblastoma cell lines, we found that deleting the PTB2 domain almost abolished nuclear localization. Upon pharmacological inhibition of APP secretases, we found decreased Fe65 localization to the nucleus. To conclude, Fe65-APP interaction and APP processing may be important factors governing the Fe65 cellular localization.

Keywords: Alzheimer’s disease, apolipoprotein E, insulin, metabolism, Amyloid precursor protein (APP), Fe65.

Stockholm 2021

http://urn.kb.se/resolve?urn=urn:nbn:se:su:diva-191938

ISBN 978-91-7911-422-0 ISBN 978-91-7911-423-7

Department of Biochemistry and Biophysics

Stockholm University, 106 91 Stockholm

(4)

(5)

LINKS BETWEEN PLASMA APOE AND GLUCOSE METABOLISM, BRAIN INSULIN SIGNALING, AND SYNAPTIC INTEGRITY

Anna Edlund

(6)

(7)

Links between plasma apoE and glucose metabolism, brain insulin signaling, and synaptic integrity

Anna Edlund

(8)

©Anna Edlund, Stockholm University 2021

ISBN print 978-91-7911-422-0 ISBN PDF 978-91-7911-423-7

Printed in Sweden by Universitetsservice US-AB, Stockholm 2021

(9)

"Forgive yourself for not knowing what you didn’t know before you learned it"

- Maya Angelou

(10)

(11)

i

List of Publications

This thesis is based on the five following papers /manuscrifts, referred to as paper I-V in the text.

I. Patra K, Giannisis A, Edlund AK, Botne Sando S, Lauridsen C, Berge G, Rolfseng Grøntvedt G, Bråthen G, White LR, and Nielsen HM. Plasma Apolipoprotein E Monomer and Dimer Proﬁle and Relevance to Alz- heimer’s Disease. Journal of Alzheimer’s Disease 71(4):1217-1231 (2019)

II. Edlund AK, Chen K, Lee W, Protas H, Su Y, Reiman E, Caselli R, and Nielsen HM. Plasma apolipoprotein E3 and glucose levels are associ- ated in APOEε3/ε4 carriers. Journal of Alzheimer’s Disease, Pre-press, pp. 1-16 (2021)

III. Giannisis A, Patra K, Edlund AK, Agirrezabala Nieto L, Benedicto Gras J, Moussaud S, de la Rosa A, Twohig D, Bengtsson T, Fu Y, Bu G, Bial G, Foquet L, Raber J, Hammarstedt C, Strom S, Kannisto K, Ellis E, and Nielsen HM. Brain integrity is altered by hepatic APOEe4 in humanized- liver mice. (Submitted to Molecular Psychiatry)

IV. Edlund AK, Giannisis A, Patra K, Morrema THJ, Bial G, Foquet L, Hoozeman J, Bengtsson T, and Nielsen HM. Impact of high-fat diet on brain integrity in APOEε3 humanized liver mice. (Submitted to Transla- tional Neurodegeneration)

V. Koistinen NA, Edlund AK, Menon PK, Ivanova EV, Bacanu S, and Iver- feldt K. Nuclear localization of amyloid-β precursor protein-binding protein Fe65 is dependent on regulated intramembrane proteolysis.

PLoS One 21;12(3) (2017)

(12)

ii

Populärvetenskaplig sammanfattning

Alzheimers sjukdom

Demenssjukdomar är en grupp åldersrelaterade sjukdomar utan botemedel som påverkar minnet och tankeförmågan. Alzheimers sjukdom (AS) är den vanligaste formen av demenssjukdom som årligen drabbar kring 20 000 svenskar. Totalt lever ca 100 000 svenskar med AS. Mindre än 1% av AS fallen är ärfliga och kallas familjär AS. Genetiska mutationer leder då till insjunknande före 65 års ålder och ett snabbare sjukdomsförlopp. Den vanligaste formen av AS uppstår dock utan känd orsak och kallas för sporadisk AS, där insjuknandet oftast sker efter 65 års ålder. Risken att drabbas för sporadisk AS påverkas bland annat av ålder, högt kolesterolvärde, fetma, diabetes och apolipoprotein E4 (apoE4). Apolipoprotein E4 finns hos 15-20% av den skandinaviska befolkningen som bär på APOE4 genen. Gemensamt för båda formerna av AS är ansamlingen av restproduken amyloid-beta från klyvning av Amyloid- betaprekursorproteinet (APP) i så kallade beta-amyloida plack mellan hjärnans celler samt ansamling av tau-proteinet i neurofibrillnystan inuti nervcellerna. Beta-amyloida plack och neurofibrillnystan är kopplade till försämrad kommunikation mellan celler och leder till att nervceller dör. Vissa nervceller och delar i hjärnan är särskilt känsliga i AS, framförallt nervceller i tinningloben och hjässloberna, regioner som är viktiga för minnet.

Alzheimers sjukdom och apolipoprotein E

Apolipoprotein E (apoE) utsöndras både av leverceller och av celler i hjärnan och är involverad i transport av fett mellan organen i kroppen och mellan olika typer av celler. Proteinet finns i tre former, apoE2, apoE3 och apoE4 varav främst apoE4 är kopplat till en ökad risk att drabbas av AS. Hur apoE4 som produceras i hjärnan påverkar AS är relativt väl kartlagt och man har bland annat visat att personer med apoE4 har störningar i hur cellerna kommunicerar med varandra och att beta-amyloida plack ansamlas i hjärnan redan i yngre ålder. Individer med apoE4 har också störningar i hjärnans energiförsörjning då cellernas upptag av socker, hjärnans viktigaste energikälla, är försämrat redan innan individerna med apoE4 drabbas av

(13)

iii

AS. En färsk studie tyder på att förhöjda nivåer av apoE4 relativt till apoE3 i blodet kan påverka hjärnans upptag av socker negativt. I en stor dansk studie obeserverade man också ett samband mellan låga nivåer av apoE i blodet och förhöjd risk att drabbas av demens, inklusive AS.

Studie I-IV. Apolipoprotein E i blodet och förhöjda blodfetter är kopplade till Alzheimers sjukdom

Det är fortfarade okänt hur låga apoE-nivåer i blodet är kopplade till risken att drabbas av AS då apoE i blodet, främst utsöndrat av levern, inte kan ta sig in i hjärnan. Syftet med denna doktorsavhandling var därför att undersöka hur apoE och fetter i blodet bidrar till processer i hjärnan som kan leda till AS. Vi har undersökt nivåerna och sammansättningen av apoE i blodet samt hur proteinet relaterar till blodsockerhalterna. Våra resultat visar att lägre nivåer av apoE i blodet eventuellt bidrar till utvecklingen av AS då låga nivåer av apoE hade ett starkt samband med AS biomarkörer (beta-amyloid och tau) mätta i ryggmärgsvätska. Låga nivåer av apoE3 var också länkat till förhöjda blodsockernivåer och en negativ påverkan på processer i hjärnan. I vår studie av möss vars leverceller ersatts med männskliga leverceller som utsöndrar apoE4 eller apoe2+apoE3, såg vi en minskning av proteiner involverade i signalöverföring mellan nervceller och inflammation i hjärnan. Dessa resultat tyder på att en lever som utsöndrar apoE4, eller lägre nivå av apoE3, i blodet bidrar till att hjärnan blir känsligare och mer mottaglig för AS. Vi har även undersökt hur intaget av fett påverkar olika processer i hjärnan då fetma och ändrade nivåer av fetter i blodet ökar risken att drabbas av AS. Hos möss med apoE3-levrar påverkade en diet med högt fettinnehåll hjärnan negativt. Det återstår att undersöka hur en diet med högt fettinnehåll samverkar med en apoE4-utsöndrande lever.

Med denna doktorsavhandling har jag haft för avsikt att bidra till en ökad förståelse för hur apoE och fetter i blodet enskilt eller tillsammans påverkar processer i hjärnan som kan bidra till utvecklingen av AS. Sammanfattningsvis så verkar högre apoE3- nivåer i blodet vara bra för hjärnans hälsa medan apoE4 i blodet bidrar till försämrad hjärnhälsa, vilket kan hänga samman med regleringen av nivåerna av socker och fetter i blodet. Tidigare studier har visat att genom att hålla koll på och vid behov modifiera nivåer av fetter och socker i blodet så kan man sänka risken för utvecklingen av AS.

(14)

iv

Resultaten av våra studier stödjer ett sådant antagande samtidigt som våra studier visar att nivåerna av apoE i blodet kan utgöra ett nytt mål för framtida studier som syftar till att reducera eller eliminera risken för AS, framför allt hos individer som producerar apoE4.

Studie V. Interaktionen med amyloidprekursorproteinet (APP) är viktig för transport av Fe65 till cellens kärna

Proteinet Fe65 funkar som en kopplingsdosa i cellen och olika delar i proteinet är viktiga för att sammanlänka Fe65 med andra protein. Interaktioner med andra proteiner, däribland APP, är viktiga för att avgöra var i cellen Fe65 befinner sig. I våra studier har vi undersökt hur sönderdelning av APP och hur Fe65s olika delar påverkar var i cellen adaptorproteinet Fe65 befinner sig. Detta är viktigt då Fe65 tillsammans med en del av APP och ett annat protein (Tip60) kan färdas till cellens kärna och där leda till att mer APP produceras och sönderdelas till amyloid-beta. Våra studier visade att klyvningen av APP samt interaktionen mellan APPs intracellulära fragment och Fe65 är viktiga för att Fe65 ska kunna färdas till cellkärnan.

Att förhindra Fe65 transport till cellens kärna skulle kunna vara ett möjligt sätt att minska mängden APP som tillverkas och klyvs till amyloid-beta, vilket tillsammans kan minska mängden amyloid-beta plack som ansamlas i hjärnvävnaden hos patienter med AS.

(15)

v

Abbrevations

A Amyloid-

ABCA1 ATP-binding cassette transporter sub-family A member 1 AD Alzheimer’s disease

ADAM a disintegrin and metalloprotease AFT APP/Fe65/Tip60

ALS Amyotrophic lateral sclerosis

AMPAR α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor AICD Amyloid- precursor protein intracellular domain

ApoA-I Apolipoprotein A-I ApoA-II Apolipoprotein A-II ApoB Apolipoprotein B ApoE Apolipoprotein E

ApoER2 Apolipoprotein E receptor 2

APOE-TR Apolipoprotein E targeted replacement mice APP Amyloid- precursor protein

ARC Arcuate nucleus ATP Adenosine triphosphate BACE site APP cleaving enzyme BCA Bicinchoninic acid

BDNF Brain-derived neurotropic factor BEC Brain endothelial cell

BSA Bovine serum albumin CCA Cerebral amyloid angiopathy CETP Cholesteryl ester transfer protein CMRgl Cerebral metabolic rate of glucose CNS Central nervous system

CSF Cerebrospinal fluid

(16)

vi

DAB 3,3′-Diaminobenzidine DLB Dementia with Lewy bodies

EAAT2 Excitatory Amino Acid Transporter-2 ELISA Enzyme-linked immunosorbent assay EOAD Early-onset Alzheimer’s disease

FAD-tg Familial Alzheimer’s disease transgenic mice

FDG-PET Fluorodeoxyglucose (FDG)-positron emission tomography FRGN Fah -/-, Rag 2 -/-, IL2rg -/-, non-obese diabetic mouse FTD Frontotemporal dementia

GLUT Glucose transporter GMV Gray matter volume

GFAP Glial fibrillary acidic protein GWAS Genome-wide association studies HDL High-density lipoprotein

HOMA-IR Homeostatic model assessment for insulin resistance HRP Horse radish peroxidase

ICC Immunocytochemistry IF Immunofluorescence IHC Immunohistochemistry IR Insulin receptor

IRS-1 Insulin receptor substrate-1 LDL Low-density lipoprotein

LDLR Low-density lipoprotein receptor LOAD Late-onset Alzheimer’s disease

LRP1 Low-density lipoprotein receptor-related protein 1 LTP Long-term potentiation

LXR Liver X receptor LXRE LXR response element MCI Mild cognitive impairment MMSE Mini mental state examination

(17)

vii

MRI Magnetic resonance imaging MS Multiple sclerosis

NIA-AA National institute on Aging – Alzheimer’s association NIRKO Neuron-specific knock-out of insulin receptors NFT Neurofibrillary tangles

NMDAR N-methyl-D-aspartate receptor OxPhos Oxidative phosphorylation PD Parkinson’s disease

PDD Parkinson’s disease dementia PET Positron emission tomography PI3K Phosphoinositide 3-kinases PiB Pittsburg compound B

PSD-95 Postsynaptic density protein 95 p-tau Phosphorylated tau

PTB Phosphotyrosine-binding PVDF Polyvinyl fluoride

RCT Reverse cholesterol transport SCD Subjective cognitive decline

SDS-PAGE Sodium dodecyl sulphate polyacrylamide gel electrophoresis SGK1 Serum- and glucocorticoid-induced kinase 1

SGLT Sodium-glucose linked transporter SOD1 Superoxide dismutase 1

T2DM Type 2 diabetes mellitus

TAP-tag Tandem-affinity purification tag TrkB Tropomyosin-related kinase receptor B TNF-   umor necrosis factor alpha

VaD Vascular dementia

VLDL Very low density lipoprotein

VLDLR Very low density lipoprotein receptor

(18)

viii

1. Introduction

1.1 Neurodegenerative disorders and dementia

Neurodegenerative disorders are a heterogeneous group of age-related diseases such as Alzheimer’s disease (AD), Parkinson’s disease (PD), dementia with Lewy bodies (DLB), frontotemporal dementia (FTD), and amyotrophic lateral sclerosis (ALS). A common pathological mechanism associated with neurodegenerative diseases is the build-up of different toxic protein aggre- gates, composed of tau, amyloid-β (Aa peptide metabolite formed following sequential proteolytic processing of the amyloid-beta precursor protein, APP), -synuclein, and superoxide dismutase 1 (SOD1). The protein aggre- gates are linked to ”selective vulnerability” in specific parts of the nervous system, leading to a loss of synapses and neurons which causes a progressive loss of function, including cognitive functions and/or motor skills, reviewed in [1, 2]. As of yet, many of the disorders have no effective cures and the incidence of neurodegenerative disorders is expected to increase with the aging of populations worldwide [3].

Dementia, a common consequence of neurodegenerative disease, is characterized by memory loss and cognitive impairment driven by the loss of neuronal function. A recent systematic analysis found that the number of people living with dementia increased from 20.2 million in 1990 to 43.8 million in 2016 and that dementia was the 5th leading cause of death in the world [4].

Alzheimer’s disease is the leading cause of neurodegenerative dementia worldwide, accounting for between 60-80% of all cases. Other types of dementia include vascular dementia (VaD) FTD, DLB, and Parkinson’s disease dementia (PDD). Dementia is often preceded by subjective cognitive decline (SCD), where the affected person experiences self-reported cognitive decline that can’t be objectively detected using cognitive tests [5]. Thereafter, the per-

(22)

2

son may progress to mild cognitive impairment (MCI) considered to be a tran- sitional stage between normal aging and early dementia. At this stage, patients suffer from objective memory impairment but are still not demented [6].

The progressive clinical changes in AD is divided into three broad phases within the Alzheimer’s disease continuum, namely preclinical AD, MCI due to AD and dementia due to AD (figure 1) [7].

Figure 1. Schematic representation of the Alzheimer’s disease clinical continuum. Adapted from [7] with permission from John Wiley & Sons, Inc.

1.2 Alzheimer’s disease

Alzheimer’s disease is histopathologically characterized by the accumulation of intracellular neurofibrillary tangles (NFTs) composed of aggregated hyperphosphorylated tau and extracellular senile plaques composed of aggregated A [8]. The disease is divided into early-onset AD (EOAD) and late- onset AD (LOAD), depending on whether the disease debut occurs before or after the age of 65 years. In addition to age, which is considered the largest LOAD risk factor, females have a higher risk of developing AD than males [9]. Genome-wide association studies (GWAS) have revealed various genetic risk factors, for example variants of APOE, TREM2, and CLU, that increase the risk of LOAD (reviewed in [10]). Of these genetic risk factors, the APOE

4 allele encoding the apolipoprotein E4 (apoE4) isoform is considered the strongest [11].

Modifiable risk factors such as: poor diet, physical inactivity, obesity, diabetes mellitus, hypertension, depression, and smoking have also been linked to an increased risk of LOAD [12, 13]. Obese individuals often suffer from

(23)

3

aberrant lipid metabolism, leading to dyslipidemia, with elevated levels of triglycerides, very-low density lipoprotein (VLDL), low-density lipoprotein (LDL)-cholesterol, apolipoprotein B (apoB), concomitant with decreased levels of high-density lipoprotein (HDL)-cholesterol [14]. The role of dyslipidemia in the pathogenesis of AD remains poorly understood [15]. Obe- sity is also heavily associated with type2-diabetes mellitus (T2DM), occurring in 80% of cases. Dysregulation of plasma lipid levels in obese individuals contributes to decreased sensitivity to circulating insulin, referred to as insulin resistance, which precedes T2DM [16]. The increased incidence of AD in individuals with T2DM and obesity has been attributed to insulin resistance, a common trait shared among the disorders, reviewed in [17].

In contrast, less than 1% of AD cases are caused by rare familial autosomal dominant mutations, in the PSEN1 and PSEN2 genes, coding the -secretase proteases presenilin-1 and presenilin-2, and in the APP gene [18]. These mu- tations lead to increased production of Aand/or a shift in production from A1-40 to the more neurotoxic A1-42 form, causing earlier onset of AD symptoms, as early as between 30-50 years of age [19]. Based on this, the amyloid cascade hypothesis was formulated – stating deposition of A into senile plaques as the causative factor promoting downstream AD pathology includ- ing tau pathology, inflammation and neurodegeneration [20]. However, this hypothesis has been challenged following the failure of numerous clinical tri- als aimed at reducing A production and plaque load, reviewed in [21]. Dis- ease progression and severity seem to be more closely connected to the accumulation of NFTs in different AD brain areas [21].

Progress in identification and development of fluid and imaging biomarkers have lead to a biological definition of AD, proposed by National Institute on Aging – Alzheimer’s Association (NIA-AA) research framework, based on presence of A deposition, pathologic tau and biomarkers of neurodegeneration [AT(N)] [8]. Alzheimer’s disease biomarkers, indicative of disease progression, have been proposed to be ”temporally ordered” [22-24],

(24)

4

summarized in figure 2. Reduced levels of A1-42 in cerebrospinal fluid (CSF) occurs early in AD disease progression [25] and has been suggested to reflect A accumulation into plaques. The levels of plasma A1-42 have been found to negatively correlate with CSF A1-42 in AD subjects [26] and A levels in plasma may show promise as an AD biomarker, reviewed in [27]. Amyloid-

plaque load can be monitored with positron emission tomography (PET) imaging combined with radiolabeled PET-tracers such as Pittsburg compound B (PiB) that bind to to Aβ plaques, reviewed in [28]. Accumulation of A

plaques seems to begin in several of the core regions in the default mode net- work, including precuneus, medial orbitofrontal and posterior cingulate corti- ces [29]. The deposition of A into plaques precedes cognitive impairment and can be detected two decades before clinical symptoms, especially in car- riers of the APOE4 allele [27, 28]. Amyloid- plaques have also been found in cognitively healthy individuals [30, 31]. Instead, the disease progression in AD seems to follow patterns of NFTs formation and accumulation of phosphorylated-tau (p-tau) and total tau in CSF [24]. Levels of total tau and p-tau correlate with NFT formation observed at autopsy and have been proposed to reflect neuronal damage and degeneration [32]. The appearance of NFT fol- lows a pattern refered to as Braak stages, first described by Braak and Braak [33]. Briefly, NFTs are first found in the enthorinal cortex (stage I), followed by the CA1 region of the hippocampus (stage II) whereafter NFT accumulation occurs in limbic structures (stage III) followed by accumulation in the amygdala, thalamus (stage IV), and claustrum before the NFT reaches all iso- cortical areas (stage V), lastly, the primary sensory, motor and visual areas (stage VI) are affected, reviewed in [34]. Tau-specific PET tracers have been developed in order to detect and monitor the progression of tau pathology in AD and other tau-related dementias (tauopathies), and to distinguish AD from other tauopathies [35].

(25)

5

Neuropathological events caused by NFTs and Aplaques lead to brain atrophy, however the loss of dendritic spines and synapses might correlate better with cognitive decline than the loss of neurons [36], especially at an early stage of the disease [36]. Synapse numbers have also been reported to be decreased in the hippocampus of individuals with early AD compared to cognitively unimpaired or MCI subjects. The majority of the MCI subjects (75%) also had a lower synapse count than unimpaired subjects [35]. Reduc- tions in cerebral glucose metabolism, assessed with [18F]-fluorodeoxyglucose (FDG)-PET, is an another early event in AD disease pathophysiology which precedes the clinical onset of symptoms and has been suggested to be closely related to decreased synaptic activity [37]. The role of glucose hypometabolism in AD is further described in section 1.7.2.

Figure 2. Alzheimer’s disease pathological already changes occur during the preclinical phase of the disease. Deposition of A into amyloid plaques is paralleled by reduced CSF A levels in the earliest phase of the disease. Simultaneously, the cerebral glucose metabolism become lower. Tau pathology, as mirrored in elevated CSF tau levels, occurs somewhat later in the disease development and co-incides with reductions in hippocampal volume. These pathological processes accumulate and eventually result in cognitive impairment. [22]. (Permission not required to reprint figure from [22])

1.3 The amyloid precursor protein (APP)

One of the main pathological hallmarks of AD known as senile or Aβ plaques, was first reported by Dr. Alois Alzheimer in 1907 [38]. In 1984,

(26)

6

Glenner and Wong isolated and purified A plaques and discovered that they contained of 4.2 kDa peptides of 40 or 42 amino acids in length [39]. Shortly thereafter, these peptides were shown to be fragments from a longer protein [40] now known as APP. The full-length protein is a transmembrane protein with a large, glycosylated extracellular N-terminal domain and a short cyto- plasmic C-terminal domain [41]. The APP is most commonly processed in two ways, either through the non-amyloidogenic pathway or via the amyloidogenic pathway, reviewed in [42]. In the non-amyloidogenic pathway, the extracellular N-terminal is shed by an α-secretase, such as ADAM10, which cleaves APP close to the plasma membrane within the A sequence [43], thereby preventing subsequent A generation. In the amyloidogenic pathway, the shedding is instead carried out by the aspartyl protease -site APP cleaving enzyme (BACE1) within acidic intracellular compartments, like the endosomes [44]. The first APP cleavage events release either sAPP (non-amy- loiodgenic processing) or sAPP (amyloidogenic processing) [45]. The membrane tethered truncated C-terminals, C83 or C99, are thereafter proteolyti- cally processed within the transmembrane domain by presenillin-1 (or -2), the catalytical cores of the multimeric-secretase complex [46, 47]. The second cleavage event leads to the release of p3 (non-amyloidogenic processing) or A (amyloidogenic processing) and the APP intracellular domain, AICD, reviewed in [48]. The release of an intracellular domain from a transmembrane protein is refered to as regulated membrane proteolysis (RIP) which is an important signaling mechanism involved in cell-to-cell interactions [49]. Other RIP substrates includes Notch, LRP-1, and tumor necrosis factor alpha (TNF-

, to name a few [50]. The AICD is rapidly degraded in the cytosol and re- quires interaction partners, such as the adaptor protein Fe65, in order to translocate to the nucleus [51].

(27)

7

1.3.1 APP and the brain-enriched adaptor protein Fe65

The adaptor protein Fe65 is predominantly expressed in neurons and is composed of three protein-protein interaction domains [52]. The C-terminal WW-domain is named after two conserved tryptophans within the domain and mediates interactions with proline-rich proteins containing the motifs PPXP (X=any aminoacid) or PPLP [53]. The WW domain interaction partners include the actin-binding protein Mena (mammalian homolog of enabled) [54]

and c-Abl tyrosine kinase [55]. The structure of Fe65 also contains two con- secutive phosphotyrosine-binding (PTB) domains called PTB1 and PTB2, reviewed in [56]. Binding partners of PTB1 include Tau [57] and LRP1, of which the latter has been proposed to mediate of APP processing [58]. In addition to LRP-1, Fe65 also interacts with other liporotein receptors through its PTB1 domain, including the apoE receptor 2 (apoER2) and VLDL receptor (VLDLR), affecting their trafficking and processing [59, 60]. Fe65 binds to the YENPTY endocytosis sorting domain within the C-terminal domain of APP through its PTB2 domain [61]. This interaction is suggested to promote trafficking of APP to endosomes, thereby enhancing amyloidogenic processing of APP [62].

Full-length APP may function as an anchor of Fe65 in the cytoplasm, therefore, proteolytic processing of APP may be required for the release of Fe65 [63]. Due to APP cleavage within the endosomes, placing AICD in closer proximity to the cell nuclei than non-amyloidogenic processing occurring at the plasma membrane, the amyloidogenic processing pathway has been suggested to be the main APP cleavage pathway involved in nuclear signaling [64]. Fe65 stabilizes AICD following release by -secretase [51] and then re- cruits the histone acetylase Tip60 through its PTB1, resulting in the formation of a trimeric complex called the AFT-complex [65]. This complex has been detected in the nucleus where it is suggested to play a role in gene transcription [66]. Target genes suggested to be upregulated by AFT-signaling include APP, BACE1, KAT5 (encoding Tip60), and GSK3B, reviewed in [67], whereas

(28)

8

LRP1 and CLU has been found to be down-regulated [68, 69]. The effect of APP processing on LRP1 suggests that APP, through AFT-signaling, may be involved in regulation of brain apoE and cholesterol metabolism [68].

Figure 3. Proteolytic processing of APP leads to the relase of AICD. The AICD fragment is either quickly degraded or forms a complex together with Fe65 and TIP60 (AFT-complex) that translocates to the nucleus and promotes transcription of APP and BACE or hinders transcrip- tion of LRPI and CLU, reviwed in [67]. Amyloidogenic processing occuring within the endo- somes has been proposed to be the main APP cleavage pathway involved in AFT-complex formation and nuclear translocation [64], however the non-amyloidogenic processing of APP may also contribute to the translocation. The interaction of Fe65 with LRP-1 and APP at the plasma membrane may increase APP localization to the endosomes and promote amyloidogenic APP processing [58].

Fe65 may also be able to translocate to the nucleus independently of AICD, how this occurs is however poorly understood. Recent evidence suggests that predicted nuclear localization signal (NLS) in Fe65 may not functional for nuclear import [70]. Instead, the adaptor protein may be dependent on interaction partners, such as the acetyltransferase Tip60, for nuclear translocation [70]. In addition, phosphorylation of Fe65 may serve as an important regulator of subcellular localization. The kinase c-Abl, which interacts with Fe65 through the WW-domain has been found to phosphorylate Fe65 at Tyr547 within PTB2 and stimulate APP/Fe65 mediated gene transcription [55]. Phos- phorylation of Fe65 also regulates binding to interaction partners. GSK3-beta

Cytosol

Endosome Extracellular

Nucleus

(29)

9

phosphorylate Fe65 at Thr579 within the PTB2 domain which enhance Fe65- APP interaction and stimulate APP processing [71]. Phosphorylation of Fe65 at Ser610 by serum- and glucocorticoid-induced kinase 1 (SGK1) has the op- posite effect and weaken the bindig of Fe65 to APP, with a concomitant decrease in amyloidogenic APP processing [72].

1.4 Apolipoproteins and lipoprotein particles

Lipoproteins are particles composed of lipids and proteins, enabling the transport of water-insoluble lipids in aqueous extracellular fluids such as blood [73]. The particles contain a hydrophobic core that carries triglycerides and cholesteryl esters (CE) [74]. The core is surrounded by unesterified cholesterol and a monolayer with amphipathic phospholipids, most commonly phosphatidylcholine (PC) and sphingomyelin (SM), with their polar head- groups facing the aqueous surrounding of the blood [74]. The lipoproteins are divided into seven different classes characterized by their size, lipid composi- tion and the types of apolipoproteins they contain [73], summarized in table 1. Apolipoproteins, such as apoB, apolipoprotein AI-II (apoA-I and apoA-II), apolipoprotein CI-III (apoC-I, apoC-II, and apoC-III), and apoE, are bound to the surface of the lipoprotein and contribute with structural support [74], promote interaction with lipoprotein lipases (LPL) [75], as well as mediate lipoprotein uptake and clearance by lipoprotein receptors [76].

Table 1. Summary of lipoproteins with their associated apolipoprotein and lipid cargo.

(Adapted after Feingold [73])

Class Density

/diameter

Core structural apolipoproteins

Associated proteins

Associated Lipids Chylomicron <0.94 g/mL

75-1200 nm apoB48 apoC, apoE,

apoA-II Triglycerides Chylomicron

remnants 0.93-1.006 g/mL

30-80 nm apoB48 apoE Triglycerides

Cholesterol VLDL 0.94- 1.006 g/mL

30-80 nm apoB100 apoE, apoC Triglycerides IDL (VLDL

remnants)

1.006-1.019 g/mL

25-35 nm apoB100 Triglycerides

Cholesterol

(30)

10

LDL 1.019-1.063 g/mL

25-35 apoB100 Cholesterol

Lp(a) 1.063-1.085 g/mL

~30 nm apoB100, apo(a) Cholesterol

HDL 1.063-1.210 g/mL

5-12 nm apoA-I +/-

apoA-II apoC, apoE Cholesterol Phospholipids Abbrevations: very-low density lipoprotein (VLDL), intermediate density lipoprotein (IDL), low-density lipoprotein (LDL), liporprotein (a), high-density lipoprotein (HDL)

Apolipoprotein B exists as two different isoforms, the full-length protein apoB100 and the C-terminally truncated version apoB48 [77]. The two different isoforms of apoB assemble in different lipoprotein particles. apoB100 is produced mainly by the liver in humans and in mice and serves as the core apolipoprotein in LDL, VLDL, IDL, and Lp(a) particles [78]. Apoliprotein B48 is secreted by cells of the small intestine in humans and function as the core apolipoprotein in chylomicrons [79], which are important in the distribution of dietary lipids, such as triacylglycerols, phospholipids and cholesterol esters, reviewed in [80]. Mice apoB48 on the other hand, is produced both in the mice intestine and by hepatocytes [81]. The turnover rates of apoB100 as and apoB48 containing lipoproteins have been found to be higher in mice than in humans [82].

High-density lipoprotein (HDL) particles are a heterogeneous group of lipoproteins that play an important role in reverse cholesterol transport (RCT), the transport of cholesterol back to the liver from for example macrophages [83]. The protein apolipoprotein A-I (apo A-I), synthesized and secreted by the liver and intestines, function as the core protein in the majority of the HDL particles and constitutes about 70% of the total protein content of HDL particles. Following secretion, apoA-I interacts with ATP-binding cassette transporters sub-family A member 1 (ABCA1), who mediate lipidation of apoA-I [84]. ABCA1 is a transmembrane protein composed of two transmembrane domains and two intracellular ABC domains belonging to a family of active transporters requiring energy gained from hydrolysis of adenosine triphosphate (ATP) to carry out transport of cargo such as cholesterol and phospho-

(31)

11

lipids across the plasma membrane [85]. Apolipoprotein A-II (apoA-II), synthesized and secreted exclusively by the liver, is present on two- thirds of the HDL particles and constitutes about 20% of the HDL protein [86]. The HDL particles can be divided into two subclasses, dependent on the presence of apoA-II, namely: LPAI, which contains apoA-I but lack apoA-II, and LPAI:AII that contains both apoA-I and apoA-II, respectively [87].

Serum cholesterol is mainly found in HDL particles in mice while in LDL in humans [88]. Human and mice HDL particles display different size patterns. In mice, the HDL particles have a unimodal distribution, which shifts to a human-like, polymodal profile upon the introduction of human apoA-I in transgenic mice [89]. Repopulation of mice with human hepatocytes promotes a shift from a HDL mouse phenotype to a human-like plasma lipoprotein profile [90]. Mice lack cholesterol ester transfer protein (CETP) which exchanges cholesterol in HDL particles for triglycerides in VLDL and LDL[91]. How- ever, CETP is not expressed by hepatocytes and is not required to establish a human-like cholesterol lipoprotein profile in humanized-liver mice [92]. The differences in lipid metabolism between humans and mice may complicate translating findings from studies on mice to the human situation.

1.4.1 APOE and the gene product apolipoprotein E

Apolipoprotein E is a 299 amino acid long glycoprotein, existing in three common isoforms in humans, which arise from polymorphism in the gene on chromosome 19 coding for apoE. The APOE 2 allele gives rise to an isoform with two cysteines at position 112 and 158 whereas the 3 variant has an arginine residue at position 158. In 4 carriers, both sites have arginines [93].

Of the three different main variants, the 4 allele of APOE is considered the ancestral genotype from which the other variants have evolved [94]. Why APOE4 evolved is not clear, but shifts in diet, inflammatory response, and vitamin D reuptake have been suggested to be involved, reviewed in [95].

(32)

12

Apolipoprotein E is produced primarily by liver hepatocytes in the periph- eral compartment and it has been estimated that 75-90% of the body’s total apoE originates from the liver [96, 97]. Apolipoprotein E is also produced in the brain, further described in section 1.4.2 below. Apolipoprotein E is involved in cholesterol transport both in the periphery and in the central nervous system (CNS) [98] through association with lipoprotein particles and uptake through lipoprotein receptors [99]. Structural differences between the three main apoE isoforms, apoE2, apoE3, and apoE4, give rise to differences in their interaction with lipoprotein particles and lipoprotein receptors [97]. An interaction between the C-terminal and N-terminal domains in the apoE4 isoform has been suggested to contribute to decreased phospholipid-binding capacity [100] and causes a preference for large (30-40 nm), triglyceride-rich VLDL particles [101, 102]. The association of apoE4 with VLDL leads to increased receptor-mediated endocytosis of VLDL particles by LDL-receptors and LDL receptor-related proteins expressed in the liver [103]. As a consequence, LDL- receptors are down-regulated, leading to increased plasma levels of LDL [104, 105] and an increased risk for atherosclerosis and cardiovascular disease in apoE4 carriers, in addition to the increased risk of AD [97]. In contrast to apoE4, apoE2 and apoE3 contain cysteines which allows these isoforms to form disulfide bonds. Apolipoprotein E3 has been found to exist as homodimers [106] as well as heterodimers with apoA-II [107]. ApoE2 and apoE3 bind to small HDL particles, in particular HDL1 and HDLc [108]. Apolipopro- tein E3 and apoE4 have an arginine at position 158 that forms a salt bridge with aspartic acid-154. This salt bridge is absent in the apoE2 isoform due to the cysteine residue at position 158, leading to a different conformation of the side chains of the basic amino acids between position 136-150. This confor- mational difference in apoE2 reduces the binding to LDL receptors to approximately 2% of apoE3 and apoE4 receptor-binding capability [109]. Defective binding of apoE2 to LDL receptors reduces the ability of apoE2 to clear triglyceride- and cholesterol-rich VLDL and predisposes apoE2 homozygotes to

(33)

13

type III hyperlipoproteinemia if the lipid clearance capability of apoE2 is overwhelmed [97].

1.4.2 Brain lipid and lipoprotein metabolism

Most plasma lipids, including cholesterol and lipoproteins, are restricted from entering the central nervous system by the BBB, which also limits CNS lipids and lipoprotein from exiting the CNS [110]. Central nervous system cholesterol is instead synthesized in situ from acetate mainly by astrocytes, oligodendrocytes, and microglia cells. Cholesterol is synthesized in neurons as well but to a much smaller extent [111]. Neurons on the other hand play an important role in cholesterol homeostasis by producing 24S-hydroxycholes- terol. This metabolite is more hydrophilic than cholesterol and diffuses across the BBB and is thereafter transported by plasma lipoproteins to the liver, me- tabolized to bile acids, and excreted [112]. Proper redistribution of lipids and cholesterol within the CNS is vital to support neuronal plasticity, organelle biogenesis, and synaptogenesis [113]. Regulation of CNS lipid metabolism is carried out mainly by apoE and apoA-I which form different kinds of HDL- like particles [114]. Apolipoprotein E is produced mainly by astrocytes in the CNS under normal conditions [115] but can also be produced by stressed or injured neurons [116]. Unlike plasma apoE, the CNS apoE is heavily glycosylated and sialylated, modifications which alter the isoelectric point of the apoE isoforms and might decrease the lipid binding-ability of apoE in an isoform-dependent manner [117]. Apolipoprotein AI on the other hand is produced in the periphery and cross from the blood to CSF [118], primarily at the choroid plexus [119]. Lipidation of CNS apoE and apoA-I is promoted by ABCA1 [85, 120]. Uptake of lipids are thereafter mediated through binding of apoE /apo-A1 on HDL-like particles to LDL and LRP-1 receptors [121].

(34)

14

1.5 Apolipoprotein E in neurodegenerative disorders

APOE3 is the most commonly occurring allele of the apoE isoforms and is found in 60-70% of the general population. APOE3 is considered risk-neutral when it comes to AD, whereas APOE2, found in 5-10% of the population, is protective. APOE4, found in 15-20% of the population is the isoform linked to increased AD risk [122]. It has been estimated that 40-50% of all LOAD cases occur in carriers of the APOE4 allele [123]. One 4 allele increases AD risk 4-fold and two 4 alleles increase AD risk 12-fold, relative to risk neutral

3/3 carriers [124]. In persons of an African American or Hispanic ancestry, the presence of APOE4 allele is not associated with the same risk as for non- Hispanic whites, while the overall frequency of AD appears to be be higher [125].

Apolipoprotein E4 plays multiple roles linked to the development of AD pathology, previously reviewed in [122] and summarized in figure 4. Apolipo- protein E4 increases Aoligomerization compared to apoE2 and apoE3 [126].

In agreement, cognitively normal APOE4-carriers have a higher Aβ plaque load in the brain prior to cognitive decline than APOE3-carriers [127].

Apolipoprotein E4 may enhance APP expression through a MAP-kinase mediate signaling cascade [128]. Several other mechanisms behind apoE4-promoted A accumulation have also been proposed including: reduced A clear- ance by microglia cells, astrocytes, neurons, and via the BBB in APOE4 car- riers, reviewed in [129]. Cerebrovascular dysfunction may be greater in APOE4-carriers than in non-carriers, reviewed in [130] andncreased BBB permeability has recently been reported in APOE4 vs. APOE3 knockin-mice [131]. Interestingly, APOE4-linked BBB dysfunction may also predict cog- nitive decline independent of Aβ and tau accumulation in the brain [132].

Apoliprotein E4 was suggested to worsen tau pathology compared to apoE3 [133] and APOE4-carriers were recently shown to have increased CSF tau levels compard to non-carriers, and the increase in CSF tau levels correlated

(35)

15

with impaired long-term potentiation (LTP)-like cortical plasticity [134]. In- terestingly, tau hyperphosphorylation was recently shown to lead to accumulation of oligomeric insulin species inside neurons [135], but whether this is promoted by different apoE isoforms is unclear. The apolipoprotein E4 isoform has been implicated in brain insulin resistance and glucose hypometabolism [136], further described in section 1.7.2. Glucose hypometabolism may be linked to apoE4-induced mitochondrial dysfunction. Mitochondrial respi- ration was recently shown to be decreased in the brains of APOE targeted re- placement mice (APOE-TR) with human APOE4 compared to human APOE3 [137]. In agreement, Neuro-2a neuronal cells expressing apoE4 had lower reserve capacity to generate ATP compared to cells expressing apoE3 [138]. The production of ATP is tightly connected to synaptic activity and is vital for synaptic function [139]. In accordance, a recent proteomics study of post-mortem AD brain tissue identified decreased levels of proteins involved in mitochondrial and synaptic function in APOE4-carriers [140]. In the same study, higher levels of proteins involved in neuroinflammation were recorded in APOE4-carriers [140], supporting a role of apoE4 in neuroinflammation, reviewed in [141].

Figure 4. Schematic overview of pathological changes promoted by apoE4. Processes outlined in bold have been studied in this thesis.

(36)

16

In addition to an increased risk of AD, carrying the APOEε4 allele has also been implicated in the development or disease progression of other disorders of the nervous system. For example, apoE4 is associated with an increased risk of cerebral amyloid angiopathy (CAA) [142], a condition characterized by the accumulation of Aβ in brain arterioles with an increased risk of brain hemorrhage [143]. Carrying the APOEε4 allele has also been linked to poorer outcome following traumatic brain injury [144]. Additionally, APOEε4 is a risk factor for DLB and Parkinson disease dementia (PDD) [145] and may also be a risk factor in VaD, however, the findings are less conclusive with regards to VaD, reviewed in [146]. In persons with Downs syndrome, who have an extra copy of chromosome 21 (trisomy-21) where APP is located, presence of APOEε4 allele may increase the risk of Downs syndrome-associated dementia, reviewed in [147], while APOEε2 appears protective [148]. In other neurological disorders, such as multiple sclerosis (MS) and ALS, it's less clear whether or not APOEε4 is associated with an increased risk of disease, reviewed in [149]. Nonetheless, APOEε4 may modulate disease progression, especially in ALS subjects, where the disease progress more rapidly in AP- OEε4 carriers than in non-carriers [150]. Also, APOEε4 may lower the age of onset of ALS [151] whereas the APOEε2 allele may delay the onset of ALS [152]. APOEε2 on the other hand, despite offering protection against some neurological disorders, may be associated with an increased risk of PD [153]

and CAA [142].

1.5.1 Insights from in vivo apoE models

Mouse models of AD with familial AD mutations fail to fully recapitulate the effects of human apoE on AD pathology due to differences between murine and human apoE [154]. Apolipoprotein E targeted replacement mice, originally developed to explore the effect of different apoE isoforms on cardiovascular disease [155], are now also used to explore brain pathology linked

(37)

17

to AD due to the increased risk of AD in carriers of the APOEε4 allele. In APOE-TR mice the murine gene coding for apoE is targeted for replacement with human APOE [155, 156]. The inserted human APOE gene is under the control of the mouse promoter, without altering the regulatory sequences. This yields human apoE mRNA expression, tissue distribution and levels similar to the mouse apoE mRNA [155]. However, concern has been raised consider- ing transcriptional regulation differences between h-APOE and m-APOE [157], in particular distal elements like the multi-enhancer region containing the liver X receptor (LXR) response element (LXRE), reviewed in [154]. This doesn’t seem to affect expression, since protein levels of human apoE in APOE3-TR and APOE4-TR mice has been reported to be expressed at physi- ological levels, reviewed in [158]. Regional differences of apoE expression levels has been reported in the brain of APOE-TR mice, with highest level in the cerebellum, and lowest in the cortex and hippocampus [159]. Plasma lipid levels are differently affected in APOE-TR mice depending on genotype. Mice with two copies of the human 4-allele exhibits approximately twice the amount of plasma cholesterol and apoB48 in VLDL particles than mice with two copies of the human 3-allele [156], whereas APOE2-TR mice suffer from type III hyperliproteinemia, reviewed in [160].

APOE targeted-replacement is a useful model system to investigate how apoE4 affects brain function, providing clues to mechanisms that make the brains of APOEε4 carriers more vulnerable to AD [161]. Targeted-replace- ment mice with ε4 exhibit disturbances in the pre-synaptic compartment compared to ε3 TR, such as fewer dendritic spines and reduced branching in cortical neurons [162] and in the medial enthorinal cortex [163], as well as decreased production of glutamate [164] and reduced excitatory synaptic trans- mission [165]. Introducing h-apoE in familial AD (FAD) transgenic (Tg) (FAD-Tg) mice delays accumulation/deposition of A in an apoE-isoform dependent manner (apoE4>apoE3>apoE2) [166, 167] compared to m-apoE FAD-Tg or FAD/tg m-apoE^-/- mice [168]. In addition to effects on A, age-

(38)

18

dependent deficits in cognition and down-regulation of N-methyl-D-aspartate receptor (NMDAR) signaling has been shown to be greater in E4FAD mice compared to E2FAD and E3FAD [169].

APOE deficient mice (e.g. APOE-/- due to homozygous deletion) [170]

have also been used to study the function of apoE in the brain [171, 172]. Mice deficient in apoE display age-dependent synaptic alterations with decreased synaptophysin immunoreactivity in nerve terminals and microtubule-associated protein 2 (MAP2) immunoreactivity in dendrites in the cortex and hippocampus. These observations implicate that apoE plays an important role in synaptic maintenance and stability [171]. Functionally, apoE deficiency might be linked to age-related alterations in behavior, learning, and memory observed in apoE knock-out mice compared to wild-type mice [172]. Apolipo- protein E deficiency negatively impacts BBB function [173, 174], potentially due to vascular dysfunction and atherosclerosis caused by markedly altered plasma lipid levels [175, 176]. Apolipoprotein E may also provide some protection against oxidative stress, as apoE deficiency renders mice vulnerable to oxidative insults [177].

1.6 Altered plasma lipid and lipoprotein levels are linked to AD risk

Epidemiological studies have revealed that elevated cholesterol levels at midlife increase the risk of AD and vascular dementia [178]. Supporting these findings, higher serum levels of cholesterol in cognitively normal late middle age individuals were associated with glucose hypometabolism in brain regions vulnerable to AD. The association between plasma cholesterol and lower cerebral metabolic rate of glucose (CMRgl) was more pronounced in carriers of APOEε4 [179]. In contrast, increased AD risk from elevated cholesterol and LDL was found in non-carriers but not in APOE4 carriers in the Nigerian Yoruba cohort [180]. Cholesterol may also affect brain health through its me-

Links between plasma apoE and glucose metabolism, brain insulin signaling, and synaptic integrity: Relevance to Alzheimer’s disease pathophysiology

Links between plasma apoE and glucose metabolism, brain insulin signaling, and synaptic integrity

Anna Edlund

Links between plasma apoE and glucose

metabolism, brain insulin signaling, and synaptic integrity

Links between plasma apoE and glucose metabolism, brain insulin signaling, and synaptic integrity

Anna Edlund

List of Publications

Populärvetenskaplig sammanfattning

Abbrevations

Table of Contents

1. Introduction