Engineering of tetrameric heme catalases to analyze the sequence of their ´in vivo´ assembly, and to prepare stable dimers and monomers

Typical catalases are tetrameric heme-containing enzymes of outstanding catalytic efficiency. This function strictly depends on the native quaternary structure, which itself is characterized by pronounced exchange of arms between adjacent monomers (domain swapping). Upon tetramerization the extended N-terminal arm of each subunit hooks into a narrow loop formed by the so-called wrapping domain of a second subunit. We suggest that this pseudo- knot can not be formed in the course of association of already folded monomers. In vivo dimerization and eventually tetramerization of only partially folded polypeptide chains will occur, followed by an ordered docking of the exchanged arms, and finally the wrapping loops holding them in place. Four alternative pathways are suggested, depending on which type of intermediate dimer will form. We plan to distinguish between these possibilities, and to prove this mechanism of pseudo-knot formation by engineering a typical representative, catalase A from S. cerevisiae. Based on the known crystal structure of this enzyme we will introduce inter-subunit disulfide bonds in different contact areas. The respective single-site replacements will, if necessary, be followed by directed molecular evolution (DNA shuffling) to correct for any unwanted effects of these modifications. In a second step, the presumptive later steps of catalase assembly will be intentionally disturbed by further modifications (mainly of the regions directly involved in knot formation). The in vivo-assembly will then stop at the respective stage. The preceding folding/assembly intermediate will only accumulate if sufficiently stabilized by covalent linkage. Performing this type of experiment with catalase A species capable of disulfide-formation at different areas, this will occur only if the respective contact actually occured in the assembly intermediate, and thus the nature of this intermediate can be identified. Introducing different covalent inter-subunit linkages will furthermore lead to greatly enhanced stability of the engineered catalase. This may be important with respect to a number of potential applications, and it should also be possible to prepare dimeric catalase of sufficient stability to demonstrate the effects of altered quaternary structure of the enzyme on its catalytic functions. An even more radical engineering approach aims at preparing monomeric variants of the yeast catalase. Following removal of severe conformational constraints (replacement of several proline residues, and insertion of a flexible loop at the hinge connecting the N-terminal arm and the compactly folded part of the catalase molecule) a monomer should emerge with a high tendency to form an intra-molecular rather than an inter-molecular knot. Eventually a stable monomeric, though otherwise still typical catalase will be obtained. This variant will be extremely interesting for pharmaceutical applications, may allow in vitro refolding, and will offer answers to the long-standing questions why and in what way the performance of tetrameric catalases depends on the native quaternary structure.

By means of site-directed mutagenesis cystein residues were introduced into 3 different regions of inter-subunit contact in catalase A from bakers yeast. In each region at least one combination of these residues led to successful formation of covalent disulfide bonds, connecting pairs of subunits within the native tetrameric structure. After introducing further modifications (mostly truncations) we started to unveal the sequence of steps which lead to the formation of a complex "pseudo-knot" in the native enzyme structure. Typical catalases consist of 4 identical subunits, pairs of which are interwoven in a very intricate pseudo-knot like structure. We suggested a model of the formation of this complex structural feature under conditions of de novo synthesis of the protein, which we test on a typical member of this group, catalase A from bakers yeast. The basic idea of our experimental outset is to "freeze" intermediates of this assembly process. To allow this, cysteine residues were introduced by site-directed mutagenesis on different zones of potential contact between subunits. When pairs of such residues come into close contact, covalent disulfide bridges would form and such stabilise ("freeze") the respective transient interaction. A large number of single residue exchanges were performed to introduce cysteines at 4 different regions in the protein monomer. Unfortunately, the unavoidable side-effects of these modifications were too large in one of these regions, whereas in each of the other 3 regions we obtained at least one pair of engineered cysteines which would interact in the planned manner, without reflecting serious side-effects. However, under normal conditions of catalase expression, the rate of disulfide formation turned out to be too low. This could not be significantly improved under conditions of increased oxidative stress, nor by heterologous expression in the methanotrophic yeast Pichia pastoris. So finally we direct catalase A through the secretory pathway by fusion with the respective signal peptide. This includes passage through the endoplasmic reticulum which strongly favours formation of disulfide bonds, and leads to secretion of the correctly folded and assembled protein into the medium. These experimental problems, which at least to this extent were not expected, severely slowed down the progress of this project, so that we only started with the experiments which should reveal the pathway of the in vivo assembly of the enzyme; so far only preliminary results are available. The more general implications of this project could be the (to our knowledge for the first time) successful formation of stable, covalent bonds between the subunits of a soluble protein species. In addition to their potential to solve the problem of the formation of the pseudo-knot, these modifications also dramatically increase the stability of the enzyme (catalases show irreversible dissociation upon dilution; since this is accompanied by complete loss of activity, so far the suitability of catalases in any kind of application is very limited). One of the enzyme mutants revealed an unexpected behaviour: the species L47C unfolds spontaneously. The stable product of this partial unfolding shows close similarity to members of the cytochrome P450 family of major pharmaceutical interest, without, however being so complex and unstable as typical members of this group. L47C thus could serve as a convenient model for this class.