蛋白質摺疊與錯誤摺疊－普立昂蛋白與泛素之研究

Abstract

Protein folding and misfolding has been a popular issue both in basic science and clinical research due to its promising application in protein design and combatting misfold-ing-associated diseases. In the Part I, we focus on the causing agent of transmissible spongiform encepha-lopathy (TSE), prion protein (PrP). We investigated the feasibilities of various fluores-cence resonance energy transfer (FRET) strategies to study PrP fibril structures. We found that PrP molecules labeled with both biotins and dyes are very aggregation-prone and frustrate specific immobilization of single proteins to the streptavidin-coated co-verslips. Moreover, pure fluorescence-labeled PrP, at least for the five types tested, can-not form fibrils but amorphous aggregates, suggesting that dye addition diverts the as-sociation pathway from forming amyloid fibrils to amorphous aggregates. However, fi-brils are allowed to form if PrP-Atto532 and PrP-Atto647N are mixed with unlabeled PrP. The FRET populations in this kind of fibrils depend on the ratio of fluorescence dilution, so it is not possible to extract useful distance information from this ensemble strategy. The smFRET strategy II, which uses PrP seed as the template for fluores-cence-labeled PrP to attach, seems promising to get the signals of a single fluores-cence-labeled PrP from fibrils. The fibrils attached with S132C-Q217C-Atto532-Atto647N showed FRET efficiency about 0.2-0.5, suggesting a huge separation of these two residues in fibrils relative to the native structures. In the fibril labeling experiment, Atto647N can be labeled on S132C, E146C, D147C and N181C fibrils but not on D144C, R151C, N174C and Q212C fibrils, implying that the sidechains of D144C, R151C, N174C and Q212C might be buried in the fibril core. This surface reactivity-based fibril labeling experiment can unveil the surface residues of fibrils and is useful for selecting suitable labeling sites. In the Part II, our study material is ubiquitin, a common target in both experimental and computational research of protein folding. In previous experimental work on a caged ubiquitin, V5C-DMC, our group measured the folding kinetics, volume change and the secondary structure contents of this partially misfolded structure, however sev-eral questions regarding the great solubility, the location of disrupted β-strands and the photolysis yield of V5C-DMC remain unclear. Here, we answer these questions through a combinatorial approach of all-atom molecular dynamics (MD) simulation and homolo-gy modeling. First, the simulated structure of V5C-DMC validated by experimental secondary structure contents show that the burial of hydrophobic cage inside the com-pact hydrophobic core of ubiquitin renders the hydrophobicity of V5C-DMC almost unchanged and thus a good solubility. Second, the β-strands disrupted by the cage are mainly in the C-terminal β-sheet, especially for the two hydrogen bonds between S3 and S5. The hydrogen bond loss is attributed to the rearrangement of residues, especially for residue 43, 50 and 67, in or near the C-terminal β-sheet and the subsequent swelling around C-terminal part of the β-barrel. Third, through calculating the ratio of experi-mentally observed volume change of a ubiquitin molecule to the volume of a DMC molecule, we estimate the photolysis yield 10-times lower than the previous theoretical study and to be only 1.3%. The results further support the low solvent accessibility of DMC. These findings prompt us to propose a new hypothesis that the long but kinet-ically downhill refolding process of the partially misfolded V5C observed in the previ-ous experiments could arise from the instant collapse of the DMC-free V5C upon pho-tolysis into a distorted hydrophobic core structure and constant trapping of this mis-folded structure by the recovered C-terminal hydrogen bonding networks along the re-folding trajectories.