Structure-Based Design of a Blood Coagulation Factor XIII Blocker

FXIII belongs to the family of transglutaminases, mainly catalyzing the cross-linking of proteins by formation of an isopetide bond between a glutamine and a lysine side chain. FXIII represents the last enzyme of the blood coagulation cascade providing final stability to the blood clot by cross-linking of fibrin fibers. Consequently, inhibition of FXIII would still allow the formation of a weak blood clot, most likely resulting in a lower bleeding risk by using a FXIII blocker compared to other anticoagulants. The primary goal of the thesis was to obtain a crystal structure of factor XIII in its active conformation (FXIIIa) and subsequently to use the gained information about the chemical composition of the active site for the development of FXIIIa blockers. In collaboration with the biotech company Zedira GmbH FXIIIa° could be crystallized in complex with the covalently-attached inhibitor ZED1301. The crystal structure shows that an enormous conformational change occurs during the transition from the inactive to the active state, whereby two β-barrel domains move out of space exposing the active site for inhibitors or substrates. The inhibitor ZED1301 binds covalently by its warhead (α,β-unsaturated carboxylic ester) to Cys 314 of the active site. The binding of the ligand induces the formation of a hydrophobic tunnel by rotation of the indole ring of Trp 370. The natural function of this tunnel is most likely the shielding of the intermediately formed thioester from hydrolytic cleavage. The substrate binding site can be subdivided into three areas: The catalytic site, an area located N-terminal named α-space as well as an area located C-terminal named β-space. In the β-space a hydrophobic pocket is formed, presumably induced by the tryptophan indole ring of the inhibitor. Apart from the usage for the structure-based development of FXIIIa-blockers, the crystal structure of FXIIIa° provides mechanistic insights at atomic level that in turn are of fundamental importance for the development of inhibitors for FXIIIa and other human transglutaminases. In the crystal structure of transglutaminase 2 in complex with a covalent inhibitor (named TG2a*) published in 2007, the enzyme adopts an active state, as in case of FXIIIa°, however not with a globular but linear conformation. Based on the new structural and mechanistic insights obtained by the crystal structure of FXIIIa° and further experimental data from the literature, it has been attempted to explain the occurrence of two different active conformations adopted by FXIII and TG2. First, at closer inspection of the crystal structure of TG2a*, strand III of the three-stranded β-sheet of the β-barrel 1 domain is displaced, forming a five-stranded β-sheet out of the four-stranded β-sheet of the β-barrel 1 domain and strand II of the original three-stranded β-sheet. The displaced strand III now adopts an α-helical structural element in the region of the β-space which thus is enormously reduced in size. Remarkably, amino acids forming the α-helix belong to calcium binding site 2. Consequently, occupancy of calcium binding site 2 does not allow a linear domain arrangement because the β-barrel 1 domain would clash with the three-stranded β-sheet. Since only the occupancy of calcium binding site 2 enables the formation of the catalytic dyad, the linear state should not exhibit transamidase activity. By use of a homology model based on the crystal structure of FXIIIa°, it could be shown that TG2 can adopt a conformation equivalent to FXIII in the active state (FXIIIa°). This is in accordance with the observation that some FXIII inhibitors of the α(wh)xxxPW-type have the same affinity against FXIII and TG2. However, due to their size these inhibitors should not be able to bind to the active site of the linear state of TG2 (TG2a*) since they would clash with the α-helix in β-space. Inhibitors of the α(wh)xxxPW-type bind most likely only to the globular active conformation of TG2 as suggested in the present thesis. Consequently, the inhibitor seems to determine if TG2 adopts the linear or globular active state. This is also indicated by FRET measurements performed by the Keillor lab in Canada showing that, depending on the inhibitor, the enzyme adapts to different conformational states. One state corresponds to the linear conformation, the other state to a more compact conformation. The measurement of the distances between the N- and the C-terminus of the inactive and the active globular state shows that the distances between the N- and the C-terminus decrease during the transformation to the active conformation. Thus, the compact state found by the FRET experiment corresponds most likely the globular active state. Transferring these findings to the endogenous biochemical function of transglutaminase 2, the complementarity of substrates binding to the linear or the globular active state determines whether the linear active state is formed and the glutamine residue is hydrolyzed in the catalytic center (deamidation) or the globular active state is formed and the glutamine residue is cross-linked with the lysine residue of a co-substrate by formation of an iso-peptide bond. The hypothesis of the substrate-induced chemoselectivity could give an explanation that simultaneously particular glutamine residues of a protein become deamidated whereas other glutamine residues of the same protein become transamidated. Whether a substrate is deamidated or transamidated can also be affected by the chemical environment (i.e. pH value). Analysis of the crystal structures also shows that the calcium concentration might impair the ratio of the linear (TG2a*) to the globular (TG2a) state, since the occupancy of calcium binding sites presumably does not allow the formation of the linear state. Consequently, with increasing calcium concentration the adoption of the globular state should be preferred. However, at this point it cannot be determined which influencing factor (substrate complementary or calcium concentration) plays the prominent role concerning the chemoselectivity. Nevertheless, an investigation of the transamidation of FXIII substrates of different sizes by Siebenlist and co-workers shows that the higher the molecular weight the lower the calcium concentration which is required for substrate turnover. It appears that the calcium concentration is less mandatory in case of binding of macromolecular substrates to FXIII. This indicates that calcium shifts the equilibrium from the inactive state to a pre-active species where the catalytic center becomes exposed subsequently to a conformational change of the two β-barrel domains. Thus, at high calcium concentrations even low molecular weight substrates with minor affinity might bind to FXIII. At this point it should be mentioned that FXIII exists in the inactive state as a dimer and calcium might also shift the equilibrium between the dimeric and the monomeric state. As already described, the substrate induces depending on its structural and chemical constitution the formation of the linear or globular active state. Reversely, also the present conformation of the active TG2 (TG2a* or TG2a) might determine which inhibitors are potently bound. If the hypothesis that the enzyme adopts the globular active state depending on certain influencing factors turns true, a potential inhibitor of the linear active state could exhibit a strikingly lower affinity in the organism under certain conditions as the in-vitro experiment suggested. Therefore, it might be useful to develop both, TG2-blockers active on the linear and the globular state. Consequently, drugs might result for diseases where the deamidase or the transamidase activity is disease relevant.
FXIII gehört zur Proteinklasse der Transglutaminasen, deren Hauptfunktion die Quervernetzung von Proteinen durch die Ausbildung einer Isopeptidbindung zwischen Glutamin- und Lysinseitenketten darstellt. FXIII ist das letzte Enzym der Blutgerinnungskaskade und verleiht durch Quervernetzung der Fibrinfasern dem Blutgerinnsel seine finale Stabilität. Dies bedeutet letztendlich, dass sich bei der Inhibierung von FXIII noch ein schwaches Blutgerinnsel ausbilden könnte, wodurch das Blutungsrisiko bei einem FXIII-Blocker geringer im Vergleich zu allen anderen Antikoagulantien sein sollte. Primäres Ziel der vorliegenden Arbeit war es, die Kristallstruktur des Faktors XIII in der aktiven Konformation (FXIIIa) zu erhalten, um anschließend die Information bezüglich der chemischen Beschaffenheit des aktiven Zentrums für die Entwicklung von FXIIIa-Blockern nutzen zu können.