Site-directed cross-linking with KLK proteases from prostate

Many biological processes depend on proteins that bind other molecules, which are often proteins as well. Some of these complexes are very stable and can be analyzed in structural biology. However, there are also short-lived complexes, which are much more difficult to investigate. The current procedures for cross-linking molecular components of unstable complexes are rather unspecific. Thus, new approaches for site-directed linking are desirable. So-called click chemistry reactions result in the formation of covalent bonds at defined positions in proteins, among them the well-established 1,3-dipolar cycloaddition, also known as azide-alkyne click reaction, and the newly developed thiol-ene coupling for proteins. Both approaches are based on non-natural or non-canonical amino acids (nnAA), which can be incorporated into selected proteins by manipulating the genetic code and the protein synthesis in cell cultures. A suitable model system for protein-ligand interactions are the human kallikrein-related peptidases (KLKs) from prostate, in particular KLKs 2, 3, 4, 5, and 11. These serine proteases have important physiological functions in fertilization and altered roles under pathological conditions as prostate cancer. Knowledge of several KLK crystal structures is a good basis for placing nnAAs at selected positions of the proteases for linking them to other molecules. To this end, mainly inactive variants of these KLKs will be employed, which do not cleave the linked molecules after the click reactions. The two approaches of the click reactions will be applied in two steps: 1. Inactive KLK variants containing reactive nnAAs are linked with peptides containing an azide and an alkyne as nnAA reaction partner, or a cysteine and an olefinic side-chain, respectively. These peptides are derived from natural substrates, for which the prostatic KLKs are specific. 2. The nnAA-KLKs will be linked to natural substrates, with corresponding modifications by recombinant production. After purification of the stabilized complexes, they are biochemically characterized and crystallized. If this procedure yields suitable crystals for X-ray diffraction, most likely the structures of the protein complexes can be determined. Site-directed cross-linking of proteins with nnAAs could become a general method, which is applicable to protein complexes with other biomolecules, such as nucleic acids, lipids, and sugars. Especially in this project information can be gained on the three-dimensional structural elements of the KLKs, which determine their substrate specificity, which is highly valuable for elucidating biological and disease-related processes, e.g. in prostate cancer. Knowledge of this tertiary specificity of the KLK proteases will enable the preparation of novel drugs with unprecedented efficacy.

Biological systems exhibit many short-lived protein complexes, which rarely can be captured in vitro for structural and functional studies. Therefore, we chose a model system of human proteases, the prostatic kallikrein-related peptidases (KLKs 2, 3, 4, 11), which form such complexes with their substrates as transition states. The major goal of the project was the stabilization of such complexes by establishing a system that allows to covalently link these proteases with specific, preferentially physiological substrates. Two so-called bioorthogonal reactions of the click chemistry, the copper(I) catalyzed 1,3-dipolar cycloaddition (CuAAC) and the thiol-ene coupling, seemed suitable for this approach. In the CuAAC, non-canonical amino acids (ncAAs) such as azido-phenylalanine (AzF) and homopropargyl glycine (Hpg) form stable triazole links, in the thiol-ene coupling an olefinic ncAA, such as S-allylcysteine (Sac) forms a thioether with cysteine (Cys). First, orthogonal translation systems (OTS) for the expression of KLKs, which contain ncAA), had to be generated. Based on molecular modeling calculations with structural coordinates optimal positions for ncAAs with reactive side-chains were determined. In a cooperation with the TU Berlin, corresponding OTS were developed for both mentioned click reactions. A significant scientific advancement was made with the development of the stop codon suppression method for the CuAAC, which allows to incorporate ncAAs with either azide or alkyne side-chains. By engineering and generating amino acyl-tRNA synthetases and tRNAs several OTS for the incorporation of a large set of ncAAs often with unnatural side-chains was facilitated. Test expressions of KLKs with these OTS for several ncAAs suitable for click reactions were successful. Protein expression media were optimized for the KLKs, containing buffer at neutral pH, Mg2+, and glucose or glycerol, with enhanced aeration, resulting in very high densities of E. coli cultures. This approach worked for the inactive Ser195Ala mutant of KLK2 by incorporating AzF in position 215 belonging to substrate recognition region of the active site. Optimized purification protocols for the native KLKs and some zymogens or pro-KLKs from inclusion bodies had already been developed. The KLK proteases were purified under denaturing conditions and then folded in buffer gradients on size exclusion chromatography (SEC) columns. A subsequent SEC ensured the purity and homogeneity of the protease samples. Molecular modeling and molecular dynamics calculations demonstrated that some of the native KLK proteases with inactivated catalytic residues in complexes with full-length protein substrates appeared to be relatively stable. In particular the complex of KLK2 with pro-KLK3/PSA showed this stability in simulations. Both proteins are a physiological pair of activator and activation target. Such complexes and their covalently linked click products with ncAA links are the first candidates for crystallization screening, which is a prerequisite for structure determination and related studies.