The central theme of this thesis is the role of the aggregation state of A b in AD. The amyloid cascade hypothesis in Alzheimer's disease predicts a central role for A b in the pathogenesis of the disease. A b is a highly fibrillogenic peptide and during the aggregation process many different aggregation intermediates are formed. Initial studies have shown that aggregation of A b into fibrils is a prerequisite for its toxicity (111,122,125) . Later studies demonstrated neurotoxicity of different prefibrillar intermediates in the A b aggregation pathway (22,46,63,73,79,80,143) . These different aggregation species activate different molecular pathways in the cell. We have studied one of the molecular mechanisms underlying A b toxicity, the unfolded protein response. In this thesis the complex relation between the aggregation state, the subcellular localization and the toxicity and ER stress induction by A b is studied. Since the A b aggregation state is an important determinant of its toxicity, prevention of A b aggregation may inhibit A b induced neurotoxicity. To this end, we investigated intervention in A b 1-42 aggregation by using a multivalent aggregation inhibitor as a potential therapeutic treatment. The data presented in this thesis (and summarized below) indicate the important role of the aggregation state of A b in Alzheimer's disease.

A b induced toxicity and ER stress: role of aggregation state and localization

In a previous study our group showed the activation of the UPR in neurons in AD brain (53) . This activation was shown to correlate well with the Braak score for amyloid deposits, suggesting that the increased levels of A b found in AD may be responsible for the UPR activation. Moreover, the data indicate that activation of the UPR occurs as early as Braak stage III in the temporal cortex (53) and precedes tangle formation, suggesting it is a relatively early response, likely to be mediated by oligomeric or low fibrillar Aß. We tested this hypothesis by studying the role of the aggregation state of A b on UPR induction ( chapter 2 ). We found a mild induction of ER stress by oligomeric A b, when applied extracellularly. Fibrillar A b had no effect on UPR activation. This mild induction does have functional implications, since we demonstrate that ER stress-mediated toxicity, contributes to oligomer-specific Aß 1-42 toxicity via an apoptotic mechanism, indicating that UPR activation might be one of the molecular mechanisms responsible for oligomer-specific A b toxicity.

Next we investigated the mechanism by which A b can induce ER stress. It is possible that the induction of the UPR is only found with oligomeric Aß 1-42 because the more bulky fibrillar Aß 1-42 is not as easily internalized. The experiments discussed in chapter 3 confirm this hypothesis. When fluorescent A b oligomers or fibrils are presented to the cell, we clearly find that oligomers are internalized and fibrils remain outside. Moreover, we find that inhibition of uptake of A b oligomers, also partly inhibits oligomer toxicity, indicating a direct relation between the aggregation state, subcellular localization and the toxicity of A b . Our study thus indicates that selective uptake of oligomers is a determinant of oligomer specific toxicity.

Selective uptake of oligomers by mammalian cells may explain the effects observed on toxicity and ER stress. Therefore, w e studied the interaction of A b with the endoplasmic reticulum, by analyzing the subsequent trafficking of extracellular applied A b oligomers ( chapter 3) . We do not find any presence of fluorescent A b 42 in compartments of the early secretory pathway. Our data shows that oligomers are internalized by mammalian cells by endocytosis, and subsequently accumulate in lysosomes. This suggests that ER stress induction by extracellular applied A b 42 is not a direct effect, but rather a secondary toxic effect of A b accumulation in lysosomes.

To study whether intracellularly produced A b can interact with the ER stress response we employed APP overexpressing cell lines ( chapter 4 ). We find increased UPR induction in both APP cell lines compared to the parental cell line; moreover we find a higher UPR induction in cells overexpressing mutAPP compared to cells overexpressing wtAPP. Thus, with increasing A b 1-42 levels we find increasing UPR induction in these cells when treated with tunicamycin. In addition we find increased sensitivity for tunicamycin in the APPmut cells compared to the other cell lines, which is prevented when A b production was inhibited using a g -secretase inhibitor. Since the only difference between the mutAPP and wtAPP cell lines is the level of A b 1-42 production, the differences observed in UPR induction and ER stress induced toxicity can be mostly attributed to this A b species.

Although we can not show that extracellular A b interacts directly with the ER, i t is possible that intracellular produced A b can have a direct effect on the ER. The presence of aggregated A b in the ER may directly interact with the quality control systems in the ER, as has been shown in sporadic Inclusion Body Myositis (sIBM), a degenerative muscle disease characterized by the formation of intracellular A b inclusion bodies (28) . Several ER chaperones (including BiP) are upregulated in sIBM, suggesting activation of the UPR (135) . Whether intracellularly produced A b accumulates in the ER in our cell model and thus directly interacts with the quality control systems in the ER remains to be shown.

In SUMMARY, our data indicate that extracellular oligomeric A b activates the ER stress response, probably due to its specific uptake into the cell. This, in combination with our finding that intracellular produced A b also interacts with the ER stress response, suggests that A b has to be present intracellularly to confer its toxicity via the unfolded protein response. Although we can not rule out that these are two unrelated mechanisms. These data suggest that A b is most likely responsible for the increased UPR activation we previously found in AD brains. ER stress may be an important mechanism of A b induced neuronal loss in AD pathogenesis. Better understanding of the role of ER stress in AD pathology will create the possibility for targeted therapeutic intervention.

The nature of the A b species causing toxicity and eventually neurodegeneration in vivo is not known, but previous studies do indicate that A b must be in an aggregated state to be toxic. This suggests that prevention of A b aggregation may inhibit A b neurotoxicity. This was studied in chapter 5 . The use of multiple simultaneous interactions to enhance the affinity and specificity of binding is an important concept in biology, known as multivalency. We have used this concept and designed a more potent aggregation inhibitor by coupling 4 KLVFF subunits to a first generation dendrimer. In chapter 5 we show that this dendrimeric KLVFF is indeed more potent than a monomeric KLVFF molecule, both in inhibiting further aggregation of LMW and PF A b aggregates as well as breaking down existing PF and fibrillar A b aggregates. We added a novel approach to targeting of the A b aggregation process, by designing a multivalent KLVFF-based aggregation inhibitor that is highly potent in inhibiting further aggregation and in breaking down existing aggregates.