Since the first reports on the use ofsite directed mutagenesis in 1982! protein engineering or - when rationally aimed - protein design has been Tecognized as a promising and fascinating field of research in many countries. In Japan (PERI) and the U.S.A. (CARB)researchinstitutes have been founded with the focus on protein design. More and more researchinstitutes in the United States, Canada, Japan and Europe have been Starting broad research projects on protein design (UK: SERC, W. Germany: GBF, EMBL). Possible prospects for applications of designed proteins with modified activities or other new properties are very high,in the areas of pharmacology, enzyme applications in food industry’, waste treatment and chemical synthesis, vaccine design, biosensors etc.*". This conception wasvery clearly lined out in an excellent article by Kevin Ulmerin 19835, Encouraging results have so far been obtained only for a small number ofcases including insulin, proteases and peptidic protease inhibitors, and some others®’. On the other hand many unpredicted andsurprising results of site directed mutagenesis experiments are reported onscientific meetings and in the literature*’, That demonstrates that our methods andtools in thatarea arestill rather crude and urgently require improvement!", Simplecalculations show that the random approach to protein-engineeringis a very slow one. There are 10°* ways to arrange aminoacids in a medium sized protein chain of 250 amino acids. Ca. 10” molecules would form the whole estimated mass ofour universe. But even whenthe information about the seven most important aminoacids is available and only five changes should be tested for each of the positions, about 80,000 different protein-mutants have to be prepared andtested. This implies that a knowledge of the 3D-protein-structure and a good understandingofthe functionactivity relationship is absolutely essential in order to do rational protein-design. Research projects in protein-design require a close cooperation between groupsspecialising in protein-isolation and purification, in fermentation techniques, in genetic-engineering, in DNAsynthesis and protein-crystallography (protein-NMR techniques are being established). This interdisciplinary connection between protein chemists, molecular biologists and stereo-chemists is essential for the protein-design cycle (Fig. 1) consisting of design, cloning, expression andtesting new proteins starting from known ones.

The crystal structures of 4 variants of subtilisin, each one containing an engineered disulfide crosslink have been determined. The geometries of the engineered disulfide groups are atypical. For the Cys24-Cys87 and Cys22-Cys87 disulfides there is a relationship between their measured redox potentials and their calculated dihedral energies. Disulfide introduction produced cavities in the protein structures. The cavity produced by removal of Met119 in A29C/M119C (Ala29 to Cys, Met119 to Cys) was partially filled by a disordering of nearby Asn117. The cavities were often filled with ordered water molecules that replaced interactions of the removed groups. Molecular modelling provided insight into the location where a disulfide could be incorporated, and into its resulting geometry. The structures of A29C/M119C and of V26C/A232C showed that introduction of disulfides into buried hydrophobic regions resulted in long range concerted rearrangements.

Knowledge-based modelling can be envisaged as a number of steps concerned with the establishment and use of rules to generate a model of a protein. One of the most powerful procedures in learning rules is comparison of related Structures either through alignment of sequences to identify conserved residues or superposition of three dimensionalstructures to identify conserved conformations or motifs. Thus the first step in a knowledge-based modelling procedure is the systematic comparison of families of topologically similar structures. This step will lead to the establishment of "equivalences" between the structures compared and to their clustering based on measures of similarity. The second step involves the projection of the results of the comparisons of three dimensional structures down onto the level of sequence. This step establishes rules relating sequence to structure. These can be expressed as consensus sequences - templates - for topologically equivalenced residues, or as key residues in canonical structures, which are then used to align the sequence of the protein of unknown tertiary Structure. The third step uses the rules established in the second step to generate a three-dimensional model.

Conventional refinementof biological macromolecules involvesa series of steps, each of which consists ofa few cyclesofrestrained least-squares refinement with stereochemical and internal packingconstraints orrestraints that are followed by rebuilding the modelstructure with interactive computer graphics. Duringthefinal stages of refinement solvent molecules are usually included and alternative conformations for some atomsor residues in the protein may be introduced.

We have developed a novel method for mutating DNA sequences, based on site-specific, in vivo, recombination, known as the crossover linker method. A typical crossover linker contains; (i) a single-stranded overhang for an initial cohesive-end ligation with one terminus of a linearized plasmid, (ii) a mid-section carrying modified sequence information, and (iii) a "homology-searching" sequence at the other end, that is similar to a specific region in the opposite terminus of the plasmid. Following transformation of an E. coli host with a plasmid/linker complex, intramolecular recombination between the homologous regions of the resultant intermediate completes the circularization of the plasmid, with concomitant integration of the linker. Crossover linking is performed on double-stranded DNA and can be used to create deletions andinsertions, as well as to perform site-specific mutagenesis. Both single- and double-stranded linkers with "homology searching" region as short as 5 nucleotides can be used for gene modification. Deletions of over 1000 bp have been achieved using "homology searching" regions of approx. 20 nucleotides in length. In this article, the effectiveness, limitations and mechanism of this process are discussed with emphasis on the application of the crossoverlinker to the manipulation of protein-encoding sequences.

Parathyroid hormone (PTH) is the main regulator of calcium homeostasis and controls resorption as well as formation of bone. We assayed overlapping PTH fragments for their capability to stimulate either adenylate cyclase activity or DNA synthesis in in appropriate renal or skeletal cell systems. From these studies we concluded that PTH-induced enhancement of cellular cAMP levels and of cell proliferation are exerted by different functional domains of the hormone molecule. Their conincidence with structural domains will be discussed. After establishment of an efficient expression system for human PTH in E. coli both functional and structural considerations lead to a series of hormone variants. Their biological characterization showed that individual PTH effects could selectively be affected by these mutations. Therefore, site-directed mutagenesis conceivably opens up the possibility to select for either anabolic or catabolic in vivo activities of PTH. Functional design of a potent inducer-of bone growth might be of considerable clinical interest.

Artificial mutagenesis of milk-clotting aspartic proteases, chymosin and a fungal aspartic protease from Mucor pusillus (MPR), was carried out by recombinant DNA techniques. The native and the modified chymosins were prepared by using the expression system of Escherichia coli and their activities were measured with aciddenatured haemoglobin and synthetic peptides. A marked change of substrate specificity with a change of Km or kcat was observed with the mutation of 17077) on’ the flap structure to Phe. Involvement of Lys(221) in determining the pH-activity profile was also suggested. Correctly processed but highly glycosylated MPR was secreted from yeast cells carrying the fungal gene. The decreased milk-clotting activity of the yeast MPR was improved by treatment with endoglycosidase H. Exchange of the Tyr residue on the flap and Trp(45) suggested that the hydrogen bonding between these residues is required for correct arrangement of the S1 subsite. X-ray crystallographic analysis of several aspartic proteases has revealed that 3-dimensional structures of these enzymes are very similar to each other (1). The molecules are bilobal, composing of two topologically similar domains rich in 6 -sheet structures. Their junction forms an extended substrate binding cleft and the two essential aspartyl residues reside at the bottom of the Cleft . A flexible flap region is located at the entrance of the cleft and a tyrosine residue on the flap as well as several hydrophobic residues in the adjacent region are involved to form the Sl subsite for substrate binding. In spite of these high similarity in tertiary structures, marked diversity of the catalytic activity and substrate specificity in these enzymes suggest that different sets of amino acid residues may be involved in their catalytic functions. Calf chymosin and a fungal aspartic protease produced by Mucor pusillus are characterized by their relatively high milk clotting activity and low proteolytic activity, which allow them to be used as milk coagulants in cheese industry. Mutagenesis of these two enzymes was performed by using the recombinant DNA technique to obtain information on the structure-function relationship in this characteristic group of the aspartic proteases.

The nucleotide sequence of xylanase gene(xynA) of Bacillus pumilus IPO, a hyperproducer of xylanase, was determined and the amino acid sequence was deduced from it. Xylanase is produced as a preenzyme consisting of the mature enzyme of 201 amino acid residues and a signal peptide of 27 residues. Xylanase was analyzed by X-ray crystallography at the level of 2.2 A resolution. It is consisted with two domains, smaller and larger, between two a crevasse suitable to accept xylan molecule was observed. The mutant xylanases obtained by site directed mutagenesis having amino acid alteration; Glu93-Ser93 and Glul82_Asp182 had no catalytic activity (less than 1/10,000).

The construction of new proteins is a challenging goal in present peptide and protein chemistry (2). Success in the chemical synthesis of de novo designed small proteins has been limited until now because of our still limited knowledge of the factors that determine the folding of a polypeptide chain. We have proposed a new strategy which aims at avoiding this problem by the use of the specific possibilities of peptide chemistry, for the synthesis of template— assembled synthetic proteins (TASPs).

Many changes have been made by protein engineering in lactate dehydrogenase from B.stearothermophilus. The role of groups involved in susbtrate and coenzyme binding, catalysis and effector molecule binding has been deduced and variants with improved thermal stability developed. The native substrate catalysis patern of lactate dehydrogenase has been modified by many orders of magnitude, to convert this protein into essentially a highly active malate dehydrogenase. Work is Currently on-going to modify this further and to convert malate dehydrogenase, Similarly by protein engineering, into a functional lactate dehydrogenase.