Over the past twenty-five years, genetic methods have generated a wealth of information on the regulation and the structure-function relationship of bacterial genes.These methods are based on the introduction of random mutations in a gene to alter its function. Subsequently, genetic techniques cure applied to localize the mutation, while the nature of the impairedfunction could be determined using biochemical methods. Classic examples of this approach is now considered to be the elucidation of the structure and function of genes, constituting the Escherichia coli lactose (lac) and tryptophan (trp) operons,and the detailed establishment of the structure and function of the repressor (lacl) of the lac operon. Recombinant DNA techniques and the development of appropriate expression systems have provided the means both to study structure and functionof eukaryotic (glyco-) proteins and to create defined mutations with a predestinedposition. The rationale for the construction of mutant genes should preferentiallyrely on detailed knowledge of the three-dimensional structure of the gene product.Elegant examples are the application of in vitro mutagenesis techniques to substitute amino-acid residues near the catalytic centre of subtilisin, a serine proteasefrom Bacillus species and to substituteanamino acid in the reactive site (i.e. Pi residue; methionine) of α-antitrypsin, a serine protease inhibitor. Such substitutions have resulted into mutant proteins which are less susceptible to oxidation and, in some cases, into mutant proteins with a higher specific activity than the wild-type protein.
If no data are available on the ternary structure of a protein, other strategies have to be developed to construct intelligent mutants to study the relation between the structure and the function of a eukaryotic protein. At least for a number of gene families, the gene structure is thought to be created by "exon shuffling", an evolutionary recombinational process to insert an exon or a set of exons which specify an additional structural and/or functional domain into a pre-existing gene. Both the structure of the tissue-type plasminogen activator protein(t-PA) and the t-PA gene suggest that this gene has evolved as a result of exon shuffling. As put forward by Gilbert (Science 228 (1985) 823), the "acid test"to prove the validity of the exon shuffling theory is either to delete, insert or to substitute exon(s) (i.e. in the corresponding cDNA) and toassay the properties of the mutant proteins to demonstrate that an exon or a set of adjacent exons encode (s) an autonomousfunction. Indeed, by the construction of specific deletions in full-length t-PA cDNA and expression of mutant proteins intissue-culture cells, we have shown by this approach that exon 2 of thet-PA gene encodes the function required forsecretion, exon 4 encodes the "finger" domain involved in fibrin binding(presumably on undegraded fibrin)
and the set of exons 8 and 9 specifies kringle 2, containing a lysine-binding sit(LBS) which interacts with carboxy-terminal lysines, generated in fibrin after plasmic digestion. Exons 10 through 14 encode the carboxy-ter-minal light chain of t-PA and harbor the catalytic centre of the molecule and represents the predominant "target site" for the fast-acting endothelial plasminogen activator inhibitor (PAI-1).
As a follow-up of this genetic approach to construct deletion mutants of t-PA, we also created substitution mutants of t-PA. Different mutants were constructed to substitute cDNA encoding thelight chain of t-PA by cDNA encoding the B-chain of urokinase (u-PA), in order to demonstrate that autonomous structural and functional domains of eitherone of the separate molecules are able toexert their intrinsic properties in a different context (C.J.M. de Vries et al., this volume). The possibilities and the limitations of this approach to study the structure and the function of t-PA and of other components of the fibrinolytic process will be outlined.
