The misfolding and aggregation of the beta cell hormone islet amyloid polypeptide (IAPP) into amyloid fibrils is the main pathological finding in islets of Langerhans in type 2 diabetes. Pathological assemblies of IAPP are cytotoxic and believed to contribute to the loss of insulin-producing beta cells. Changes in the microenvironment that could trigger the aggregation of IAPP are largely unknown. So is the possibility that islet amyloid can spread within or between tissues. The present thesis have explored the roles of glycosaminoglycan heparan sulfate (HS) and the novel anti-amyloid chaperone Bri2 BRICHOS domain in the assembly of IAPP amyloid and cytotoxic IAPP aggregates. Furthermore, cross-seeding as a molecular interaction between the observed connection of type 2 diabetes and Alzheimer disease has been examined. The N-terminal region of IAPP was required for binding to HS structures and induction of binding promoted amyloid formation. Interference in the HS-IAPP interaction by heparanase degradation of HS or by introducing short, soluble HS-structure fragments reduced amyloid deposition in cultured islets. Cytotoxicity induced by extracellular, aggregating IAPP was mediated via interactions with cell-surface HS. This suggests that HS plays an important role in islet amyloid deposition and associated toxicity. BRICHOS domain containing protein Bri2 was highly expressed in human beta cells and colocalized with IAPP intracellularly and in islet amyloid deposits. The BRICHOS domain effectively attenuated both IAPP amyloid formation and IAPP-induced cytotoxicity. These results propose Bri2 BRICHOS as a novel chaperone preventing IAPP aggregation in beta cells. The intravenous injection of IAPP, proIAPP or amyloid-β (Aβ) fibrils enhanced islet amyloidosis in transgenic human IAPP mice, demonstrating that both homologous- and heterologous seeding of islet amyloid can occur in vivo. IAPP colocalized with Aβ in brain amyloid from AD patients, and AD patients diagnosed with T2D displayed increased proportions of neuritic plaques, the more pathogenic plaque subtype. In conclusion, both IAPP amyloid formation and the cytotoxic effects of IAPP is dependent on interactions with HS whereas interactions with Bri2 BRICHOS is protective. Cross-seeding between Aβ and IAPP can occur in vivo and the two peptides colocalize in brain amyloid in AD patients.

Alzheimer’s disease (AD), the most common form of dementia is associated with fibril formation of amyloid-ß peptides (Aß). Aß, proteolytically derived from Aß precursor protein (AßPP), is the major component of amyloid plaques in AD brains. Familial British and Danish dementias (FBD and FDD) share pathological and clinical characteristics with AD, and the underlying mechanisms are associated with amyloid formation of mutant peptides released from the Bri2 protein. Bri2 interacts with AßPP and its BRICHOS domain has been shown to delay Aß40 and Aß42 fibril formation and toxicity in vitro and in vivo. This makes Bri2 BRICHOS a promising anti-amyloid chaperone and a potential treatment strategy for AD. Furthermore, Bri2 BRICHOS possesses a general chaperone activity as it suppresses non-fibrillar aggregation of destabilized citrate synthase (CS). Recent findings show that Bri2 BRICHOS produced in E.coli can form different molecular weight assemblies, ranging from monomers to dimers and poly-disperse oligomers. The oligomers inhibit CS aggregation, whereas the monomers and dimers are more efficient against Aß42 fibrillation and neurotoxicity, respectively. The work in this thesis shows that similar Bri2 BRICHOS quaternary structures are formed in eukaryotic cells as in E.coli. Larger BRICHOS oligomers were found in cell media, derived from proteolytically processed endogenous Bri2 in SH-SY5Y cells, as well as in human embryonic kidney (HEK293) cells transfected with a Bri2 BRICHOS construct. Recombinant human Bri2 BRICHOS mutants with one or none of the two strictly conserved cysteine residues were studied. All mutant monomers become proteolytically degraded during purification, but form stable oligomers. Single Cys to Ser mutants form stable disulfide-dependent dimers that differ in ability to prevent Aß42 fibrillation, the most stable mutant (C164S) being even more efficient than the wildtype Bri2 BRICHOS dimer. This result suggests that intra or intermolecular disulfide(s) and oligomerization affect Bri2 BRICHOS stability and activity towards Aß42 fibril formation.

Biochemical processes all involve associations and dissociations of chemical entities. Understanding these is of substantial importance for many modern pharmaceutical applications. In this thesis, longstanding problems with regard to ligand binding are treated with computational methods, applied to proteins of key pharmaceutical importance. Homology modeling, docking, molecular dynamics simulations and free-energy calculations are used here for quantitative characterization of ligand binding to proteins. By combining computational tools, valuable contributions have been made for pharmaceutically relevant areas: a neglected tropical disease, an ion channel anti-drug-target, and GPCR drug-targets. We report three compounds inhibiting cruzain, the main cysteine protease of the protozoa causing Chagas’ disease. The compounds were found through an extensive virtual screening study and validated with experimental enzymatic assays. The compounds inhibit the enzyme in the μM-range and are therefore valuable in further lead optimization studies. A high-resolution crystal structure of the BRICHOS domain is reported, together with molecular dynamics simulations and hydrogen-deuterium exchange mass spectrometry studies. This work revealed a plausible mechanism for how the chaperone activity of the domain may operate. Rationalization of structure-activity relationships for a set of analogous blockers of the hERG potassium channel is given. A homology model of the ion channel was used for docking compounds and molecular dynamics simulations together with the linear interaction energy method employed for calculating the binding free-energies. The three-dimensional coordinates of two GPCRs, 5HT1B and 5HT2B, were derived from homology modeling and evaluated in the GPCR Dock 2013 assessment. Our models were in good correlation with the experimental structures and all of them placed among the top quarter of all models assessed.  Finally, a computational method, based on molecular dynamics free-energy calculations, for performing alanine scanning was validated with the A2A adenosine receptor bound to either agonist or antagonist. The calculated binding free-energies were found to be in good agreement with experimental data and the method was subsequently extended to non-alanine mutations. With extensive experimental mutation data, this scheme is a valuable tool for quantitative understanding of ligand binding and can ultimately be used for structure-based drug design.