Former research projects
Folding and three-dimensional structure of the recombinant murine prion protein
Mammalian prion diseases are believed to be caused by a single protein, the prion protein, which may exist in two different conformations, the cellular form, and the infectious "scrapie form". We found that the recombinant carboxy-terminal domain of the mouse prion protein (residues 121-231) is an autonomous folding unit. In contrast with model predictions, its three-dimensional structure, which was solved in collaboration with the group of Prof. K. Wüthrich in our institute, contains a two-stranded beta-sheet and three alpha-helices.
In collaboration with Prof. K. Wüthrich, we have also solved the three-dimensional solution structure of the complete recombinant cellular prion protein from the mouse, PrP(23-231). Its entire N-terminal segment 23-125 proved to be flexibly disordered in solution, while the C-terminal residues 126-231 adopt the same structure as in the isolated C-terminal domain PrP(121-231). As the segment 90-231 is protease-resistant in the oligomeric scrapie form PrPSc, we postulate that the minimal structural change that occurs during the conversion of the cellular prion protein to the scrapie form is that the segment 90-125 becomes structured or buried in PrPSc.
We are currently investigating the role of point mutations linked with inherited human prion diseases on the stability and aggregation behaviour of PrP(23-231) and PrP(121-231). We are also investigating the role of a scrapie-like unfolding intermediate of the prion protein that is exclusively populated at acidic pH values and possibly constitutes a percursor form of PrPSc. Moreover, we have found that the C-terminal domain of the prion protein undergoes one of the fastest protein folding reactions known so far and folds without kinetic intermediates. This excludes kinetic folding intermediates as source of PrPSc.
Catalysis of disulfide bond formation during protein folding
Disulfide bond formation constitutes the rate-limiting step during folding of secretory proteins. The reason is that a disulfide bond cannot form automatically when two cysteine residues come close during the folding process because disulfide bond formation is an oxidation process which requires an external oxidant. Formation of wrong disulfide bonds may be another kinetic trap during folding, especially for proteins with multiple disulfide bonds where the number of possible disulfide conformers increases by approximately one order of magnitude per additional disulfide bond. In the living cell, the efficient formation disulfide bonds and the rapid isomerization of wrong disulfide bonds is guaranteed by enzymes belonging to the disulfide oxidoreductase family. These enzymes share the thioredoxin fold and possess a catalytic disulfide bond with the consensus sequence Cys-Xaa-Xaa-Cys. During the last 6 years, we have worked on the characterization of the bacterial members of this enzyme family, in particular DsbA from the periplasm of Escherichia coli. DsbA is a monomer of 189 residues with extraordinary biophysical properties. The enzyme is the strongest oxidant in the family, undergoes the fastest disulfide exchange reactions known so far and transfers its catalytic disulfide bonds to folding polypeptides with rate constants that are close to the diffusion limit. The reason for the extremely reactive catalytic disulfide is a very low pKa of 3.5 for the more N-terminal active-site cysteine. In collaboration with Dr. Tad Holak ( Max-Planck-Institute of Biochemistry , Martinsried) we solved the NMR structure of reduced DsbA. We have also performed extensive mutagenesis experiments on DsbA to investigate the molecular basis of its extreme reactivity. Specifically, we found that the Xaa-Xaa dipeptide within the active-site sequence strongly influences the redox potential of the enzyme. Replacement of this sequence by the Xaa-Xaa sequences of more reducing members of the thioredoxin family yields more reducing DsbA variants. Analogous results were obtained with Xaa-Xaa variants of thioredoxin, the most reducing member of the enzyme family: Introduction of Xaa-Xaa sequences of other disulfide oxidoreductases exclusively yielded more oxidizing variants. By using these variants of thioredoxin and DsbA we are presently investigating the importance of the redox potential of these enzymes for their function in vivo. Moreover, we are investigating the interactions of these enzymes with their natural substrates and work on bacterial expression systems that provide high yields of correctly folded proteins with disulfide bonds in the periplasm.
General aspects of protein folding
We have started a new project where we apply random mutagensis experiments in concombination with screening or selection procedures at the bacterial colony level to obtain novel information on important problems in protein folding.
As an example, we have recently performed a random circular permutation analysis on the DsbA protein from E. coli to answer the question whether unique nucleation sites exist in a polypeptide chain that are essential for protein folding and stability. The disulfide oxidoreductase DsbA is a monomeric two-domain protein of 189 residues with known three-dimensional structure. After linkage of the natural terrmini we have randomly disrupted each regular secondary structure and each loop region of the protein by introduction of new termini and asked the question whether the circularly permuted proteins are still capable to fold to a functional tertiary structure. Surprisingly, we found that only 4 alpha-helices in the protein may not be disrupted by new termini without loss of folding competence. Random circular permutation may thus be used in the future to identify the segments in a protein that are essential for folding. As some of the circularly permuted DsbA variants that we have generated are more active than the wild type protein, circular permutation may also become a useful tool to generate proteins with improved functions. We have now extended this random circular permutation approach to other model proteins.
The bifunctional alpha-amylase/trypsin inhibitor from Ragi (RBI): Folding and three-dimensional structure of RBI and its interactions with the target enzymes alpha-amylase and trypsin
The bifunctional inhibitor (RBI) from Ragi (Indian finger millet) belongs to a relatively new class of homologous alpha-amylase and trypsin inhibitors from plants. The monomeric protein consists of 122 residues, contains 5 disulfide bridges and simultaneously binds a-amylase and trypsin.
We have determined the three-dimensional NMR structure of recombinant RBI expressed in E. coli. Although the overall fold of RBI is completely different from that of other trypsin inhibitors (four alpha-helices, one two-stranded, antiparallel beta-sheet), the conformation of the trypsin binding loop is virtually identical with the loops of trypsin inhibitors from other families. Like most other proteinacous, competitive trypsin inhibitors, RBI it is also a weak substrate of trypsin.
To elucidate the binding mode to alpha-amylases, we have focused on the X-ray structure determination of the complex between RBI and an alpha-amylase from insects, namely the enzyme from the larvae of the yellow meal worm Tenebrio molitor. We have solved the crystal structures of the free T. molitor alpha-amylase (TMA) and the complex between TMA and RBI. The inhibitor follows an entirely novel inhibition mode for alpha-amylases. Specifically, the N-terminal segment of RBI which is disordered in the solution structure of free RBI adopts a helical conformation in the TMA/RBI complex, and the free N-terminus of the inhibitor interacts electrostatically with the acidic active-site residues of TMA. The structure determination of the TMA/RBI complex in conjunction with biochemical experiments is consistent with fully independent binding sites of RBI for its target enzymes. As the inhibitor is not only found in seeds but also in the leaves of Ragi and inhibits two important digestive enzymes, it may consitute an important plant defense protein against predatory insects.
We have also investigated the oxidative folding pathway of RBI and the catalysis of its folding by the disulfide oxidoreductases DsbA and DsbC in vivo and in vitro. We found that the isomerization of disulfide bonds is rate-limiting for RBI folding under all redox conditions. DsbC is a particularly efficient catalyst of RBI folding, both in vivo and in vitro.