Studies on Islet Amyloid Polypeptide Aggregation:
From Model Organism to Molecular Mechanisms
Sebastian W Schultz
Department of Clinical and Experimental Medicine Linköping University, Sweden
Linköping 2011
© Sebastian W Schultz
Cover: Drosophila brain; green: cell nuclei of ventral lateral neurons, red: neuropil
During the course of the research underlying this thesis, Sebastian W Schultz was enrolled in Forum Scientium, a multidisciplinary doctoral programme at Linköping University, Sweden.
Printed by LiU-‐Tryck, Linköping, Sweden, 2011
ISBN 978-‐91-‐7393-‐099-‐4 ISSN 0345-‐0082
Der Weg ist das Ziel
Department of Medical Cell Biology Uppsala University, Sweden Opponent
Anne Simonsen, Associate Professor Department of Biochemistry
University of Oslo, Norway
This thesis is based on the following papers, which are referred to in the text by their roman numerals:
I. Paulsson JF, Schultz SW, Kohler M, Leibiger I, Berggren PO, Westermark GT. Real-‐time monitoring of apoptosis by caspase-‐3-‐like protease induced FRET reduction triggered by amyloid aggregation. 2008, Exp Diabetes Res 2008: 865850.
A free, coloured version of this paper can be downloaded from:
www.hindawi.com/journals/edr/2008/865850/
II. Schultz SW, Nilsson KP, Westermark GT. Drosophila melanogaster as a model system for studies of islet amyloid polypeptide aggregation. 2011, PLoS One 6:e20221.
A free, coloured version of this paper can be downloaded from:
www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0020221
III. Schultz SW, Gu X, Rusten TE, Alenius M, Westermark GT. HIAPP and hproIAPP trigger selective autophagy and inhibit the neuro-‐ protective effect of autophagy. Manuscript.
The proper folding of a protein into its defined three-‐dimensional structure is one of the many fundamental challenges a cell encounters. A number of tightly controlled pathways have evolved to assist in the proper folding of a protein, but also to aid in the removal of misfolded proteins. Despite the presence of these pathways accumulation of misfolded proteins can still occur. Amyloid deposits consist of misfolded proteins with a characteristic highly ordered fibrillar structure that will exert affinity for the amyloid dye Congo red and has a unique X-‐ray diffraction pattern. Currently 27 different proteins have been identified as amyloid forming proteins in human, however the exact role of amyloid in the pathogenesis of the connected disease is most often unclear.
Islet amyloid is made up of the beta cell derived hormone islet amyloid polypeptide (IAPP) and is associated with the development of type 2 diabetes. Propagation of IAPP-‐fibrils is believed to be one important cause of the pancreatic beta cell death detected in patients with type 2 diabetes. IAPP is a naturally occurring polypeptide hormone stored and secreted together with insulin. IAPP and insulin arise from posttranslational processing of their biological inactive precursors proIAPP and proinsulin. In addition to human, cat and monkey IAPP will form amyloid deposits in conditions resembling human type 2 diabetes. However, IAPP from mouse and rat do not form amyloid as a result of the differences in amino acid sequence.
My main research goal was to establish a unique model system suitable to study the effects of proIAPP and IAPP aggregation. I selected Drosophila melanogaster due to its many suitable characteristics as a model organism and its superior genetic toolbox. I have demonstrated that over-‐expression of hproIAPP and hIAPP in the central nervous system (CNS) results in aggregate formation in the brain and neighbouring fat body. Consistent with previous studies, expression of mIAPP does not result in the formation of aggregates. To investigate the intracellular effects of hproIAPP and hIAPP aggregation on a specific population of neurons, we targeted the expression of these peptides specifically to 16 neurons in the brain, the pdf-‐ neurons. These pdf-‐neurons are divided into 2 clusters of 8 cells per brain hemisphere. First I showed that expression of aggregation prone hIAPP and hproIAPP resulted in significant death of the 8 cells, whereas expression of mIAPP had no such effect. In efforts to pinpoint the mechanisms behind the observed cell death I demonstrated that hproIAPP and hIAPP both pass the ERs quality control for protein folding and that the initiated cell death does not occur through classical apoptosis. Instead, selective autophagy is activated by hIAPP and hproIAPP. This activation counteracts the usually neuro-‐protective effects of autophagy and contributes to cell death. Strikingly, I also showed that Aβ, the amyloid protein implicated in Alzheimer’s disease, does not exhibit any intracellular toxicity when expressed in pdf-‐cells. This supports the existence of separate toxic pathways for different amyloid proteins.
Proteins are one of the building blocks of life. They are important for almost every process in the cell, e.g. forming a framework involved in cellular structure, activation of chemical reactions and mediating cell signals and cell interactions. However, proteins have to adopt a pre-‐defined three-‐dimensional fold, referred to as its native confirmation, in order to function. Because proteins are so important, cells have developed highly sophisticated and tightly controlled pathways used to assist their proper folding and to remove misfolded proteins. Despite quality control, accumulation of misfolded proteins can occur. Amyloidosis is a group of protein misfolding diseases. Hitherto, 27 different proteins have been identified as amyloid forming in man. Each amyloid protein is associated with a specific disease, but the exact role for amyloid in the pathogenesis of the illness is unclear. All amyloid deposits share certain characteristics, they have all affinity for amyloid specific dyes and methods providing high-‐resolution information reveal a highly ordered fibrillar structure.
The protein I have been working on is the hormone islet amyloid polypeptide (IAPP) that together with insulin and glucagon participates in the regulation of blood glucose. IAPP can form amyloid in pancreas and this is associated with type 2 diabetes. After food intake the blood glucose concentration raises, which leads to release of insulin from beta cells in the pancreas. Insulin facilitates cellular uptake of sugar and thereby lowers the blood glucose concentration. Patients that suffer from type 2 diabetes cannot produce sufficient amounts of insulin and they develop chronic elevated blood sugar level. One reason for the decreased insulin secretion is the replacement of beta cells by IAPP-‐amyloid, and it is believed that islet amyloid is responsible for this cell reduction and contributes to insulin deficiency.
One question that still remains to be answered is -‐ how does IAPP-‐amyloid mediate cell death? Since IAPP and insulin are produced by the same cells, death can be initiated from the inside or from the outside of the cell. For my work I have set up a new Drosophila melanogaster (fruit fly) model to study effects of aggregation of human IAPP and its precursor proIAPP. I have produced transgenic flies that secrete human IAPP or proIAPP and shown that expression of these proteins in the fly head results in aggregation (paper II). In paper III, I limited IAPP and proIAPP expression to a subset of 16 neurons, and showed that this caused cell death. The mechanism behind intracellular cell death was studied in detail and I was able to show that the autophagy (self-‐eating) pathway was selectively triggered by human IAPP and human proIAPP. Gained evidence indicates that activation of this self-‐eating (autophagy) pathway decreases the normal protective mechanism of this pathway and thereby contributes to cell death. I have included studies on Aβ, the protein that forms amyloid in patients with Alzheimer’s disease. Aβ expression in the 16 cells did not result in cell death. Instead, comparison of Aβ and IAPP/proIAPP expression revealed that amyloid proteins use different pathways to exhibit their toxicity.
INTRODUCTION ... 3
PROTEIN FOLDING AND MISFOLDING ... 4
AMYLOID AND AMYLOIDOSIS ... 5
History and definitions ... 5
Amyloid and diseases ... 6
Structure of amyloid ... 8
Non-‐fibrillar components in amyloid deposits ... 9
Amyloid formation ... 10
Toxic effects ... 11
Functional amyloid ... 12
ISLET AMYLOID POLYPEPTIDE (IAPP) ... 13
General introduction ... 13
Prohormone processing ... 15
IAPP and type 2 diabetes ... 17
IAPP fibril formation ... 18
Transgenic animal models with hIAPP ... 21
Aβ ... 22
Alzheimer’s disease ... 22
Aβ and IAPP ... 23
DROSOPHILA MELANOGASTER AS MODEL SYSTEM ... 25
History of Drosophila as model system ... 25
Huge genetic toolbox: Gal4/UAS system ... 26
Drosophila models for protein aggregation ... 28
MOLECULAR PATHWAYS CONNECTED TO PROTEIN MISFOLDING ... 31
ER-‐stress and Unfolded protein response (UPR) ... 31
Apoptosis ... 37
Autophagy ... 41
MATERIAL AND METHODS ... 53
WORKING WITH DROSOPHILA ... 54
P-‐element insertion ... 54
Survival assay ... 54
DETECTION METHODS ... 55
Immunofluorescence – tissue preparation ... 55
Congo Red or pFTAA ... 55
Image processing ... 56
RESULTS AND DISCUSSION ... 57
EXTRACELLULAR AMYLOID FORMATION INDUCES APOPTOSIS (PAPER I) ... 58
CHARACTERISATION OF A NEW DROSOPHILA MODEL FOR STUDIES OF IAPP AGGREGATION (PAPER II) ... 60
HPROIAPP AND HIAPP TRIGGER SELECTIVE AUTOPHAGY (PAPER III) ... 64
GENERAL DISCUSSION AND FUTURE PERSPECTIVES ... 69
ACKNOWLEDGEMENTS ... 73 REFERENCES ... 77
Abbreviations
Aβ amyloid-‐β peptide
AD Alzheimer’s disease
AGE advanced glycation end-‐products
Alfy PI3P-‐binding autophagy-‐linked FYVE domain protein
ApoE apolipoprotein E
APP Aβ precursor protein
ASK1 apoptosis signal regulation kinase-‐1
ATG autophagy-‐related genes
ATF6 activating transcription factor-‐6
Bchs blue cheese
Bcl-‐2 B cell lymphoma-‐2
BiP binding immunoglobulin protein
CGRP calcitonin gene-‐related peptide
CHOP C/EBP homologous protein
CMA chaperone mediated autophagy
CPE Carboxypeptidase E
CRLR calcitonin-‐receptor-‐like-‐receptor
CSF cerebrospinal fluid
CT calcitonin
CTR-‐2 calcitonin receptor 2 CVT cytosol-‐to-‐vacuole targeting
EDEM ER degradation-‐enhancing α1,2-‐mannosidase like protein
EM electron microscopy
EOFAD early-‐onset FAD
ER endoplasmic reticulum
ERAD ER associated degradation ERAF ER associated folding ERdj ER-‐resident J-‐domains
ERManI ER degradation α1,2-‐mannosidase I
ESCRT endosomal sorting complex required for transport FAD familial form of Alzheimer’s disease
FADD Fas-‐associated death domain
GAGs Glycosaminoglycans
GFP green fluorescent protein
GS glycogen synthase
GSK3α glycogen synthase 3α
HDAC histone deacteylase
HFNs human fetal neurons
HS heparin sulphate Hsc heat shock cognate Hsf1 heat shock factor-‐1 Hsp heat shock protein
HSPG heparan sulphate proteoglycan
HSR heat shock response
Htt Huntingtin
IAPP islet amyloid polypeptide
IDE insulin degrading enzyme
IRE1 inositol-‐requiring protein-‐1 JNK c-‐Jun N-‐terminal kinase
LAMP lysosome-‐associated membrane type protein LC3 microtubule associated protein 1 light chain 3
mIAPP murine IAPP
MVBs multivesicular bodies
NEFA non-‐esterified fatty acids
NFT neurofibrillary tangles
NMR nuclear magnetic resonance
OST oligosaccharyltransferase
PAM peptidyl amidating monooxygenase
PC prohormone convertase
PD Parkinson’s disease
PE phosphatidylethanolamine
PERK protein kinase RNA-‐like ER kinase PI3K phosphatidylinositol 3-‐kinase
PI3P phosphatidylinositol (3,4,5)-‐trisphosphate
Poly-‐Q polyglutamine
PS1 presenilin-‐1
RAMP receptor activity-‐modifying protein ROS reactive oxygen species
SAP serum amyloid P
SDS sodium dodecyl sulphate
TNFR1 tumor necrosis factor receptor 1
TTR transthyretin
TUNEL terminal deoxynucleotidyl transferase dUTP nick labelling UAS upstream activating sequence
ULK Unc-‐51-‐like kinase
UGGT UDP-‐glucose:glycoprotein glucosyltransferase
UPR unfolded protein response
UPRE unfolded protein response element
UPS ubiquitin-‐proteasome system
Xbp1 X-‐box binding protein-‐1 YFP yellow fluorescent protein
Introduction
Protein folding and misfolding
One of the most fundamental processes in biology is the ability of a protein to fold into its defined three-‐dimensional structure. The function of a protein is tightly coupled to this defined conformation. Already in the 1950’s Anfinsen pointed out the relationship between the amino acid sequence of the enzyme ribonuclease and its functional conformation. This functional conformation could be destroyed by the addition of 8 M urea and the reducing agent β-‐mercaptoethanol but as soon as urea was removed and the protein re-‐oxidized, it reassembled into its native structure. The free energy gained in this assembly drives the refolding process [1]. As tribute to his work on ribonuclease Anfinsen was awarded the Nobel Prize in 1972.
The native state of a protein is thought to be the most stable structure under physiological conditions. However it was for long not clear how this structure could be adopted and there was no reasonable explanation for the Levinthal paradox [2]. The basic concept introduced by Levinthal is that the search for the proper three-‐ dimensional structure is a random “trial and error” event. If a protein of 100 amino acids had to try all of its putative conformations (each taking 10-‐11 seconds to find)
the calculated time for this exceeds the age of our universe. However, from experiments we now know that folding occurs in the order of milliseconds to seconds. This time discrepancy is known as the Levinthal paradox [3]. Today, the current concept is that a polypeptides search for its native structure is following a “folding funnel” or “folding landscape” with the native structure as the lowest accessible point. Because, on average native-‐like interactions are more stable than non-‐native ones, not all possible conformations have to be tested, instead it is sufficient to test a small number of possible conformations. The shape of this energy landscape is encoded in the amino-‐acid sequence [4]. The crowded intracellular milieu with a protein concentration of 300-‐400 mg/ml complicates protein folding, since it increases the risk for undesirable interactions with other molecules [4,5]. A way to circumvent this problem is the engagement of folding catalysts and chaperones. They function either by accelerating slow folding steps or by protecting partially folded proteins from misfolding [6,7]. Despite all cellular efforts to optimize folding can protein misfolding occur.
In fact, accumulation of misfolded proteins can have detrimental effects on the organism, and is indeed linked to many diseases, including amyloidosis. This dissertation deals with various aspects of misfolded proteins with focus on the amyloid forming islet amyloid polypeptide (IAPP), and the consequences that arise when cells are exposed to misfolded IAPP.
Amyloid and amyloidosis
History and definitionsIn 1854 the German physician Rudolph Virchow was the first to use the term amyloid (from Latin amylum = starch) to describe the macroscopic changes he found in some human organs after they had been treated with iodine and sulphuric acid [8]. At this time, this staining method was widely used by botanists to demonstrate cellulose [9]. Already five years later, Friedreich and Kekulé were able to show that amyloid isolated from the spleen was not “starch-‐like” material but instead it was mainly made up by protein [10]. With time, new staining methods evolved and in 1922 Bennhold introduced the cotton dye Congo red as a histological dye for amyloid [11]. In 1927 Divry and Florkin showed that Congo red emits green birefringence when observed in cross-‐polarized light [12]. A standardized Congo red staining protocol was introduced in 1962 and this is still in use [13,14]. The property of amyloid to emit green birefringence when stained with Congo red suggested a highly ordered structure, which was confirmed by Cohens and Calkins electron microscopy studies on amyloid fibrils. They showed that amyloid is made up of unbranched fibrils with a diameter of approximately 10 nm and undetermined length [15]. Further research revealed that all amyloid fibrils are made up of smaller sub-‐elements, named protofibrils, a finding that proved to be independent on the protein constituent of the amyloid [16]. X-‐ray diffraction analysis was used by Eanes at al. to define the well-‐ordered cross-‐β-‐sheet pattern of amyloid fibrils [17].
In order to be defined as amyloid, following criteria have to be fulfilled: 1. In vivo deposited material
2. Affinity for Congo red and presentation of green birefringence when viewed in polarized light
3. The characteristic fibrillar structure when investigated with an electron microscope
4. A specific X-‐ray diffraction pattern of the fibril
All stated criteria follow the consensus reached at the meeting of the Nomenclature Committee of the International Society of Amyloidosis in November 2006. During this meeting one previous characteristic of amyloid was actually revised. Due to the increasing evidence of intracellular amyloid, the definition of amyloid is no longer limited to extracellular material [18].
Amyloid and diseases
Today, at least 27 different proteins have been identified to form amyloid in humans and the heterogeneous group of diseases associated with such deposits is referred to as amyloidosis [19]. Each type of amyloidosis is characterised by a distinct fibril protein [18]. Despite the common structural features of amyloid fibrils exhibit amyloid proteins only modest primary, secondary and tertiary structure homology [20,21]. Dependant on the amyloid distribution the disease is divided into localized and systemic amyloidosis.
Amyloid that appears at a single site or in one tissue type is called localized amyloidoses. Typically, these deposits occur in close proximity of the amyloid protein expression site. Localized amyloidosis are often linked to ageing, e.g. Aβ deposition in Alzheimer’s disease or IAPP in type 2 diabetes.
Amyloid diseases with deposits that affect several organs are referred to as systemic amyloidoses. The amyloid precursor in systemic amyloidosis is a plasma protein. Examples of systemic amyloidosis are reactive amyloidosis or secondary amyloidosis
Table 1: Amyloid fibril proteins and their precursors in human [19]. Amyloid
protein Precursor
Systemic (S), or localized
(L) Syndrome or involved tissue AL Immunoglobulin light
chain S, L Primary Myeloma-‐associated
AH Immunoglobulin heavy
chain S, L Primary Myeloma-‐associated
Aβ2M β2-‐microglobulin S
L? Hemodialysis-‐associated Joints
ATTR Transthyretin S Familial
Senile systemic
AA (Apo)serum AA S Secondary, reactive
AApoAI Apolipoprotein AI S
L Familial Aorta, meniscus
AApoAII Apolipoprotein AII S Familial
AApoAIV Apolipoprotein AIV S Sporadic, associated with ageing
AGel Gelsolin S Familial (Finnish)
ALys Lysozyme S Familial
AFib Fibrinogen α-‐chain S Familial
ACys Cystatin C S Familial
ABri ABriPP S Familial dementia, British
ALect2 Leukocyte chemotactic
factor 2 S Mainly kidney
ADan ADanPP L Familial dementia, Danish
Aβ Aβ protein precursor
(AβPP) L Alzheimer’s disease, ageing
APrP Prion protein L Spongiform encephalopathies
ACal (Pro)calcitonin L C-‐cell thyroid tumors
AIAPP Islet amyloid
polypeptide (also called: amylin)
L Islets of Langerhans (type 2 diabetes)
Insulinomas AANF Atrial natriuretic factor L Cardiac atria
APro Prolactin L Ageing pituitary
Prolactinomas
AIns Insulin L Iatrogenic
AMed Lactadherin L Senile aortic, arterial media
AKer Kerato-‐epithelin L Cornea, familial
ALac Lactoferrin L Cornea
AOaap Odontogenic
ameloblast-‐associated protein
L Odontogenic tumors
ASemI Semenogelin I L Vesicula seminalis
Structure of amyloid
The high-‐resolution structures of different in vitro assembled amyloid-‐like fibrils have been solved. The primary building block of the fibrils, the actual protein, gives rise to two, or more, β-‐strands that run perpendicular to the fiber axis. Amyloid fibrils are easily identified when viewed in an electron microscope [22]. The highly ordered, repetitive composition of the fibrils give rise to a characteristic X-‐ray diffraction pattern with an inter-‐β-‐strand distance of 4.7Å and a distance of 6-‐11Å between stacked β-‐sheets. Association of 2-‐6 protofilaments, each 2.5-‐3.5 nm in diameter, forms fibrils (see Figure 1). By twisting around one another along the fiber axis, these protofilaments contribute to the rigidity of the amyloid fibril [23]. Amyloid fibrils from the same protein are able to form different morphologies, depending on the surrounding conditions [24]. Solid-‐state NMR and EM images have supported the idea of structural polymorphism in amyloids [25,26]. Different local minima in the energy landscape of the unfolded amyloid protein are accounted for this diversity in vivo [27]. The structural heterogeneity of fibrils includes degree of twisting, the number of filaments per fibril, and the diameter or mass per length of the fibrils [25,26].
Figure 1: Structure of the amyloid fibril. The β-‐strands of the amyloid protein are stacked perpendicular to the fiber axis. The intermolecular distance of β-‐strands of neighbouring units is 4.7Å. Two to six protofilaments twist around each other and give rise to the mature amyloid fibril.
Non-‐fibrillar components in amyloid deposits
The major amyloid constituent is the disease-‐specific fibril protein. In addition to this fibril protein other, non-‐fibrillar components are present, such as Glycosaminoglycans, Serum amyloid P (SAP) component and Apolipoprotein E (ApoE).
Glycosaminoglycans (GAGs) are negatively charged heteropolysaccharides composed of repeating disaccharide units. The structure of the repeating disaccharide unit defines the five GAG classes, namely heparin/heparin sulphate (HS), chondroitin sulphate, dermatan sulphate, hyaluronan, and keratan sulphate. All GAGs except for hyaluronan are usually found covalently linked to a protein backbone and this complex is then called proteoglycan. In the light of amyloidogenesis are heparan sulphate and the heparan sulphate proteoglycan (HSPG) perlecan the best studied GAG and proteoglycan. Numerous in vitro experiments showed the potential of GAGs and HSPGs to promote fibril formation by increasing the β-‐sheet content of the amyloidogenic protein. It is also reported that HS is involved in processing of the amyloid precursor proteins and thereby influencing fibril formation kinetics and/or toxicity [28,29]. Experiments in animal models affirm an active role for HS in amyloidogenesis [30,31]. The interaction of GAGs and amyloid is a target for drug therapy [32,33,34].
Serum amyloid P component belongs to the pentraxin superfamily and binds amyloid fibrils in an calcium-‐dependent manner [35]. The binding of SAP to amyloid fibrils is suggested to prevent proteolysis of amyloid fibrils [36]. Due to its high and specific affinity, radiolabelled SAP is used to monitor amyloid deposits in a non-‐ invasive manner [37].
Apolipoprotein E has been detected in association to numerous amyloid deposits, including IAPP derived islet amyloid and amyloid deposits of Alzheimer’s disease [38]. However, the exact role of ApoE in amyloidogenesis is unclear. Polymorphisms in the APOE gene, ε2, ε3, and ε4 strongly alter the likelihood of developing Alzheimer’s disease and cerebral amyloid angiopathy. It has been suggested that ApoE modulates Aβ metabolism and accumulation, although there are contradictive results on plaque density or number depending on the APOE genotype. Differential effects of APOE isoforms on lipid metabolism have been assigned a role in synaptic plasticity and neurodegeneration, independent of interactions with Aβ [39].
Amyloid formation
In vitro, many proteins are capable of forming amyloid-‐like fibrils if exposed to low pH, high temperature, high pressure, and/or presence of co-‐solvents that all reflect unphysiological circumstances [40]. In case of some globular proteins, such as lysozyme, superoxide dismutase 1, and transthyretin, denaturing conditions are close to physiological, but despite this can amyloid-‐like fibrils form in vitro. It is thought that aggregation in these cases is a direct consequence of fluctuations from the native state or other local unfolding events, and does not require global unfolding [41]. Amyloid-‐like fibril formation is in general thought to occur via a nucleation-‐dependant mechanism, resembling crystallisation kinetics [42,43]. A typical feature of a nucleation-‐dependant mechanism is the presence of a lag time before bigger aggregates are detectable. During the lag phase monomers self-‐ assemble and form oligomers that can act as nuclei for further fibrillization. The self-‐ assembly of monomers requires partially unfolding of the protein and is thermodynamically unfavourable [44]. This step only occurs if a critical concentration is exceeded. The lag phase is followed by an elongation phase. During this period protofibrils are formed that rapidly assemble into fibrils and grow as long as the concentration of available monomers/oligomers is sufficient. Equilibrium of monomers and fibrils characterises the final plateau phase. The time span of the lag phase can be significantly reduced by addition of nuclei in form of preformed oligomers and/or fibrils, a mechanism referred to as “seeding” [43,45] (see Figure 2). Seeding is also an in vivo finding [46,47,48,49].
Figure 2: Illustration of kinetics of amyloid formation. Addition of preformed fibrils and protein aggregates can shorten the lag phase (seeding effect).
Events that can lead to nucleation in vivo are interactions between the amyloid protein and cell membranes, increased protein synthesis and deficiencies in protein clearance [50] (see Toxic effects).
Toxic effects
In general, diseases associated with amyloid are of late onset and actual deposits have degenerative effects [45]. The role of amyloid in different diseases has been subject of discussion over a long period and during the last decade many new insights into structural properties of amyloid fibril precursor species have shed a new light on how to think about amyloid cytotoxicity. In 2006, the year this PhD thesis was initiated, it was believed that amyloid cytotoxicity is coupled to common mechanism independent of protein or peptide. Until then, several in vitro studies had shown that oligomeric species and/or protofibrils of several amyloid proteins were able to permeabilize cell membranes, resulting in cell dysfunction [51,52,53,54,55]. In the same year Cohen et al. were able to demonstrate in a C. elegans model that protofibrils of Aβ were toxic, whereas high molecular weight Aβ aggregates were not [56]. Today, oligomers are still seen as the major cause for cytotoxicity. Over the last few years there has been growing evidence for the concept that the same amyloidogenic peptide/protein can give rise to structurally different oligomers and structural distinct fibrils. This led to the proposal of an aggregation energy landscape with several local energy minima corresponding to distinguishable oligomeric states [50]. But toxicity is not only thought to be dependent on the structure of the oligomeric species but also on the biophysical and biochemical properties of the interacting membrane. Anionic surfaces (e.g. anionic phospholipid-‐rich liposomes, glycosaminoglycans) seem to play an important role as potent triggers for protein fibrillization. Also mature fibrils can be ascribed certain toxicity since the deposited amyloid can be massive and affect exchange of oxygen and nutrients. Moreover, mature fibrils might contribute to cytotoxicity by leakage of toxic oligomers [50]. When it comes to IAPP it is still unclear if toxic oligomers exist in vivo. In vitro, beta cell toxicity has been shown in the presence of freshly solubilized IAPP and this leads to activation of apoptosis [55,57,58]. On the other hand have different studies shown that even pre-‐formed IAPP fibrils induce beta cell death [59,60]. A recent study could show that there exists a significant relation between the amount of deposited islet amyloid and measured beta cell apoptosis. This latter study strongly suggests that islet amyloid deposition contributes to beta cell death [61]. The inhibitory effect of amyloid inhibitors on beta cell death further challenges the concept of toxic oligomers (reviewed in [62]). The oligomeric state might be transient, and this complicates the interpretation of the in vitro assays where cells are incubated with oligomers. If cells are incubated for longer times with oligomers, these oligomers might alter their structure and start fibrillization. So in order to be able to ascribe toxicity to oligomers it is crucial to make sure that these oligomers are stable. An alternative pathway for IAPP toxicity has been suggested by Engel et al.. In this model, IAPP binds to membranes, which results in fibril growth, significant changes of membrane curvature and will over time lead to physical breakage of the membrane. Notably, the kinetic profile of hIAPP fibril formation matched that of membrane leakage [63]. The model of membrane interaction as crucial step in
mediating toxicity might be of general nature. Membranes can serve as a template that allow orientation of monomers in a way that favour aggregation [64]. In addition membrane interaction of amyloidogenic proteins can lead to increased local protein concentration and thereby catalyse aggregation [65]. Finally, it has been shown that membranes have the ability to alter the conformation of a protein and in this way induce aggregation [66,67].
Taken together results from different studies that all tried to identify toxic species of amyloidogenic proteins, it becomes clear that aggregation pathways have a major influence on how toxicity is mediated. Since these aggregation pathways not necessarily are the same for different amyloid-‐related peptides, we have to reconsider the concept that there exists a general mechanism that accounts for toxicity.
In parallel to the attempt of identifying a toxic amyloid species, several groups have started to look at molecular pathways that might be altered upon protein aggregation and subsequent amyloid formation. Several pathways, such as autophagy, endoplasmic reticulum associated degradation (ERAD) and unfolded protein response (UPR), have been identified to be triggered upon protein aggregation (intra-‐ and extracellular) and a more detailed overview of our current knowledge how these pathways influence cell survival is given in a separate section of this introduction (see Molecular pathways connected to protein misfolding).
Functional amyloid
Since many, structurally unrelated proteins are capable of forming amyloid-‐like fibrils in vitro, it has been speculated that amyloid structures have been a prominent fold in early life [68]. In coherence with this speculation, the field of functional amyloid has evolved over the last decade. Originally it was hypothesised that some organisms have during evolution taken advantage of the widespread potential of proteins to fold in a stable, amyloid-‐like manner [69]. Today, several functional amyloid structures are reported in lower organisms, including curly and chaplins in bacteria [70,71], Sup32p and Ure2p in fungi [72,73], and chorion in insects [74]. In aplysia (sea slug) conversion of CBEP to an amyloid-‐like structure has been suggested to play a functional role in memory storage [75]. In humans Mα, a component of Pmel17, has been described to play a role as functional amyloid as it serves as template for melanin and thereby is involved in melanin polymerisation [76]. Maji et al. suggested in 2009 that peptide and protein hormones are stored in secretory granules in an amyloid like aggregation state [77]. Their hypothesis is based on different in vitro experiments in which they showed how 31 of 42 investigated protein hormones formed amyloid-‐like structures at pH 5.5 in the presence of heparin -‐ conditions that mimic the environment of secretory granules. In addition, they also investigated mouse pituitary tissue and were able to detect
amyloid like structures. The proposed working model is that either a critical concentration in the Golgi per se and/or processing of prohormones can trigger amyloid formation. As a result, hormones can be packed in secretory granules at a highest density possible and even be stored over long periods due to high stability of the amyloid entity. At the same time, the secretory granules could serve as an “inert” membrane container protecting the cell of putative toxic effects of the formed amyloid. In this model, amyloid fibrils will be destabilized once they are released from the secretory granules and are exposed to pH 7.4 [77]. Unfortunately, neither insulin nor IAPP were part of the investigated protein hormones though. It is known however, that IAPP fibrils are extremely stable and generally need harsh conditions for depolymerisation [78]. It is questionable if secreted IAPP fibrils are able to dissolve once they are secreted from β-‐cells.
Islet amyloid polypeptide (IAPP)
General introductionEugene Opie reported in 1901 a hyaline substance to replace areas of the islets of Langerhans in autopsy material from a patient with type 2 diabetes [79]. Already in 1973 the characteristic interaction of extracellular amyloid fibrils with β-‐cell membranes was described [80]. But it was not until 1986 the amyloid protein was sequenced and for the first time fully characterised as 37 amino acid residue polypeptide [81,82]. This peptide was initially being called islet amyloid peptide (IAP) and later islet amyloid polypeptide (IAPP). Short after the very first description of IAPP, a second report was published describing the same polypeptide naming it diabetes associated peptide (DAP) and later amylin [83].
The gene for IAPP consists of 3 exons of which exon one is non-‐coding. It is situated on the short arm of chromosome 12 and has a promoter region similar to the promoter region of insulin [84,85,86,87]. IAPP belongs to the calcitonin gene peptide family together with calcitonin (CT), calcitonin gene-‐related peptide (CGRP), intermedin and adrenomedullin [88]. Sequence homology of hIAPP with CGRP-‐I and II is 43-‐46% and with human CT 20% [89,90].
IAPP is mainly expressed in the beta cells in the islets of Langerhans. Here, IAPP is stored in secretory granules together with insulin and those hormones are co-‐
secreted upon stimulation [91,92,93]. The intra-‐granular concentration of IAPP is 1-‐ 4 mM and the insulin concentration is 10-‐40 times higher [94,95]. The plasma concentration of IAPP ranges between 2-‐10 pM [96]. Expression of IAPP has been found in mammals, avian, and the bony fish [94,97,98,99,100,101,102]. In rodents, expression of IAPP was also reported in delta cells in the islets of Langerhans, the gastrointestinal tract, in sensory neurons and in the central nervous system [103,104,105].
Over the years several different biological functions have been ascribed to IAPP. These functions include auto-‐ and paracrine effects in the islets of Langerhans, actions as a satiety peptide in the brain, antagonising insulin action in skeletal muscles and also a role in calcium homeostasis in regard to bone mass. Each of these different functions is briefly highlighted below.
Auto-‐ and paracrine effects of IAPP are reported to regulate insulin secretion. Autocrine actions include a dual role for IAPP on insulin secretion. Transgenic mice that are deficient for IAPP show normal basal levels of circulating insulin and glucose. However, these knock-‐out mice have increased insulin responses and blood glucose elimination upon glucose administration when compared to wild type controls. It can be concluded that usually IAPP limits the degree of glucose-‐induced insulin secretion [106]. Studies about 5 years later gave a more differentiated picture of IAPPs role in insulin secretion. Akesson et al. detected a modest increase of basal insulin secretion in the presence of low IAPP concentrations (10-‐10 – 10-‐6 M) and
physiological glucose concentrations (7 mM). In contrast, high IAPP concentrations (10-‐6 – 10-‐5 M) inhibited glucose stimulated (10 mM & 16.7 mM) insulin secretion
[107]. In addition it has been shown that IAPP acts in a paracrine manner on alpha-‐ and delta-‐cells and suppresses glucagon and somatostatin release, respectively [107,108]. The observed inhibitory effect of IAPP on glucagon release was already seen at low concentrations (10-‐10 and 10-‐8 M) [107].
Today, IAPP has also been identified as a satiety hormone. This action was a matter of discussion but the identification of receptor activity-‐modifying proteins (RAMPs) was a major break-‐through [109,110]. McLatchie et al. showed that RAMPs, single-‐ transmembrane-‐domain proteins, can bind to the Calcitonin-‐receptor-‐like receptor (CRLR). This binding and hence newly formed RAMP:CRLR complex has high affinities for substrates that do not bind CRLR alone. If any of the three RAMPs (RAMP-‐1, -‐2, or-‐3) binds to calcitonin receptor 2 (CTR-‐2), a class of receptors with affinity for IAPP is formed [109,111,112,113]. It is not clear if effects of IAPP in the brain are due to local expression in neurons or if IAPP crosses the blood-‐brain barrier [114].
Effects of IAPP on glycaemic control have also led to the development of pramlintide (symlin). Pramlintide is a hIAPP analogue with proline substitutions at position 25, 28, and 29. The proline substiutions abrogate the capacity to form amyloid fibrils.
This hIAPP analogue is today an approved drug for use in conjugation with insulin therapy in patients with type 1 or type 2 diabetes. Furthermore reveal preliminary data a weight loss in obese patients with and without diabetes upon symlin intake [115,116,117].
A recent study in rats suggests a role for IAPP in maternal regulations and IAPP mRNA was up-‐regulated in the preoptic area of the hypothalamus of lactating dams [118]. IAPP also has been attributed a role in reducing pain [119,120].
In skeletal muscles IAPP has been found to inhibit insulin-‐stimulated incorporation of glucose into glycogen. The effect is described to occur via inhibition of glycogen synthase (GS) and activation of glycogen phosphorylase (GP), [121,122]. Insulin on the other hands stimulates dephosphorylation of GS thereby promoting glycogen synthesis. These effects of IAPP that are contrary to insulin action on skeletal muscles, are accounted for playing a role in developing insulin resistance [123].
Finally, I want to mention IAPPs effect on calcium homeostasis. Infusion of IAPP decreases circulating levels of calcium in humans [124]. Mice deficient for IAPP show a 50% reduction in bone mass when compared to wild-‐type littermates; an effect due to increased bone resorption mediated by IAPP [125].
Prohormone processing
Biological mature human IAPP derives from proteolytic cleavage of the 89 amino acid hormone preproIAPP. The first 22 amino acids account for the signal peptide and are cleaved off after entrance into the endoplasmic reticulum (ER) [126]. The remaining, 67 amino acid long, proIAPP enters the secretory pathway and there it is cleaved at its C-‐terminal and N-‐terminal site, giving rise to mature IAPP (see Figure 3) [126,127,128,129].
Processing of proIAPP is sequential and occurs first at the C-‐terminal site where prohormone convertase (PC) 1/3 cleaves at di-‐basic amino acid residues K50-‐R51
[128,130]. In the secretory granules PC2 removes the N-‐terminal flanking peptide processing after di-‐basic residues K10-‐R11 [129,130]. Notably, in absence of PC 1/3 is
PC 2 capable to cleave at the C-‐terminal processing site. This redundancy does not work the other way round. Removal of the N-‐terminal flanking can solely be achieved by PC 2 [128]. Carboxypeptidase E (CPE) removes the dibasic residues lysine and arginine at the C-‐terminus of processed proIAPP. The exposed glycine is carboxyamidated by the peptidyl amidating monooxygenase (PAM) complex. Presence of active CPE is also necessary in order to facilitate processing at the N-‐ terminal site by PC 2 [131].
Both prohormone convertase are produced as precursor molecules themselves and have to undergo cleavage events in order to become fully active. PC 1/3 is first auto-‐ catalytically cleaved at its N-‐terminus. This occurs already in the ER. In mature secretory granules PC 1/3 is additionally cleaved at the C-‐terminal. This site-‐specific maturation of PC 1/3 may explain the observed granule-‐specific processing by PC 1/3. At the same time, a partial activation of PC 1/3 in the late TGN was demonstrated and it was shown that C-‐terminal cleavage of proIAPP is already initiated in the TGN before entering secretory granules [130,132,133,134]. Sorting of PC2 starts in the ER where 7B2 binds to proPC 2 and enables relocalization of proPC 2 to the TGN. The binding of proPC 2 to 7B2 requires proper folding of proPC 2 [135,136]. ProPC 2 is finally cleaved in the secretory granule, a prerequisite to gain enzymatic function [134,137].
In mature IAPP, a disulphide bridge is present between cysteine 2 and 7 [94,138]. Several studies show that impaired processing of proIAPP influences fibril formation of IAPP [139,140,141,142,143]. The implications of incomplete processing of proIAPP on fibril formation are discussed below (see IAPP fibril formation).
Figure 3: Schematic drawing of prohormone processing. Both proIAPP and proinsulin are sequentially processed by the prohormone convertases 2 and 1/3.
The prohormone convertases that process proIAPP also sequentially cleave proinsulin into insulin. Initially PC 1/3 cleaves at two arginines at position 31 and 32 (R31-‐R32), separating the B-‐chain from the C-‐peptide [144,145]. PC 1/3 cleavage gives
rise to the transient intermediate des-‐31,32 proinsulin. Thereafter, PC 2 removes the C-‐peptide from the A-‐chain after residues lysine 64 and arginine 65 [146]. Dibasic residues at the cleavage sites are removed by CPE [147].