Structural studies on B2-glycoprotein I and von Willebrand factor A3 domain

Abstract

The integrity of blood circulation is a prerequisite for life; its malfunctioning is a leading cause of morbidity and mortality in developed countries. For that reason the haemostatic system is a critical component of homeostasis. In Chapter I an overview is given of the biophysical and biochemical characterization of two human plasma proteins that serve important functions in haemostasis: ?2-glycoprotein I (?2gpI) and von Willebrand factor (vWf). ?2gpI has a proposed role in clearance of thrombogenic apoptotic cells and in coagulation. Moreover, ?2gpI is the key target of auto-antibodies in the auto-immune disease anti-phospholipid syndrome (APS), resulting in thrombotic complications. ?2gpI binds with its fifth C-terminal short concensus repeat (SCR) domain to anionic lipids. This affinity is increased by complexation of ?2gpI in divalent ?2gpI-antibody complexes, which results in shielding of anionic surfaces that are necessary in coagulation. Binding to anionic lipids is possibly the primary involvement of ?2gpI in coagulation. Multimeric vWf has a key function in the haemostatic system when an arterial vessel wall is damaged. Multi-domain vWf contains the binding site for collagen located in its A3 domain. When an arterial vessel wall is ruptured, binding of the A3 domain to exposed collagen is a first step in haemostasis, that facilitates recruitment of platelets for formation of a blood clot. The function of vWf can be disturbed by quantitative and qualitative alterations of the protein and by non-self compounds that induce spontaneous platelet aggregation. Disturbed vWF function results in bleeding tendencies or thrombotic complications. Structural insights into adhesion of ?2gpI to lipids and of vWf-A3 to collagen are of valuable help in understanding their role in homeostasis and in yielding insight into the molecular origin of, for example, coagulation disorders.

?2-Glycoprotein I In Chapter II we describe the crystal-structure elucidation of ?2gpI from human plasma. The structure is solved at 2.7 Å resolution with the multiple isomorphous replacement method and reveals that ?2gpI has an elongated fish-hook like arrangement of its five globular SCR domains. Half of the fifth domain has a unique face created by a six-residue insertion and a 19-residue C-terminal extension, together with some rearrangement of parts of the SCR-like core. Based on this structure and on biochemical data we propose a model for adhesion to anionic lipids. A large patch of 14 cationic residues provides favorable electrostatic interactions with anionic surfaces, and an exposed membrane-insertion loop containing a typical Trp-Lys sequence yields specificity for lipids. The lipid-adhesion site is located at the tip and outer curve of the ?2gpI fish-hook. From the crystal structure we infer that in the lipid-bound state, domains one to four point away from the lipid layer. Domains three and four are partly shielded for protein-protein interactions by the four N-glycans attached to these domains. Domains one and two are fully exposed and may provide binding sites in apoptosis, APS and coagulation. In Chapter III we address ?2gpI-protein complex formation in solution using gel-filtration chromatography. ?2gpI forms a stable complex with IgG 22F6 which has properties common to auto-antibodies that are active in APS. This indicates that lipid adhesion is not always a prerequisite for formation of active ?2gpI-antibody complexes. ?2gpI does not form complexes in solution with protein C and protein S which are anticoagulants. Complex formation between protein S and its inhibitor C4B binding protein lowers the concentration of free protein S, which results in inhibition of the anticoagulant function of protein S. The complex of protein S and C4B binding protein remains stable, even in the presence of ?2gpI. Therefore, our experiments do not reveal further evidence that supports an anticoagulant function of ?2gpI by maintaining the level of free protein S. With our binding experiments we cannot establish previously reported high-affinity complexation between ?2gpI and calmodulin. Stable complexes putatively only form in vivo when ?2gpI is bound to lipids. In Chapter IV we describe structural differences between intact ?2gpI and an isoform of ?2gpI (denoted ?2gpI *1) that has lost its ability to adhere to anionic lipids and that is present in plasma during oxidative stress. With gas chromatography-mass spectrometry and thin-layer chromatography we show that ?2gpI *1 has bound phosphatidyl-choline (PC) or choline plasmalogen, and, to a lesser extent, sphingomyelin. The crystal structure at 3.0 Å resolution of ?2gpI *1 indicates a binding pocket for a glycerophosphorylcholine moiety located within the adhesion site for anionic lipids. The choline moiety is buried deeply inside the pocket and is surrounded by an anionic residue and two aromatic residues; these are aspects common to several PC-binding proteins. With C-terminal sequencing, carbohydrate analysis and mass spectrometry we show that, neither enzymatic cleavage at Ala314-Phe315, nor mutation Trp316Ser, cleavage of sialic acid residues, or differences in N-glycan content, account for the observed difference in lipid-adhesive properties of ?2gpI *1. Therefore, specific binding of a neutral lipid seems to be fully responsible for blocking adhesion to anionic membranes. Based on these findings we discuss a possible role for ?2gpI *1 in defense mechanisms during oxidative stress. The crystal structure of ?2gpI *1, together with the structures of ?2gpI, which was partially cleaved at Lys317-Thr318 (Ref. 250), and of intact ?2gpI251, also provides new insights that contribute to a more detailed anionic lipid-adhesion model. In the Appendix to Chapter IV we address recently discerned specific structural radiation damage of proteins upon exposure of crystals to very intense X-ray beams of a third-generation synchrotron. We describe a new method for refinement of partially broken disulfide bonds which appear to be present in ?2gpI *1. Effects of specific structural radiation damage include complete cleavage of the carboxyl group of one Glu residue and partial cleavage of all eleven disulfide bonds, observed in the most severely affected crystal of ?2gpI *1. After refinement differences among disulfide bonds in position of the S?-atom and in electron density accounting for the partially broken bonds suggest that solvent-exposed bonds are more susceptible to radiation than buried bonds. Comparison of the severely affected structure of ?2gpI *1 with that of virtually not damaged ?2gpI *1 (Chapter IV) shows that specific structural radiation damage is localized to acidic residues and cystines. No changes in the global conformation of ?2gpI *1 are observed. This implies that the structural implications presented in Chapter IV are not compromised by radiation damage.

Von Willebrand factor A3-domain In Chapter V we describe the crystal-structure determination of the vWf (Se-Met) A3-domain, in its complex with a Fab fragment of antibody RU5, which inhibits binding of vWf to collagen. With this structure we aimed at locating the collagen-binding site in the A3 domain of vWf. The structure is solved at 2.0 Å resolution by molecular replacement using the anomalous signal of the seleno methionines. The epitope of RU5 involves residues that are part of an ?-helix and three solvent-exposed loops, located at the ‘bottom’ side of A3. Comparison with other structures of A3 shows that RU5 binding does not induce long-range conformational changes. Instead, RU5 most likely blocks collagen binding by steric hindrance, which implies that the collagen-binding site is located at or close to the epitope of RU5 near the bottom side of A3. Surprisingly, the collagen-binding site of vWf-A3 is located distant from the top face of the domain where collagen-binding sites are found in homologous integrin I-domains. Apparently, vWf-A3 and integrin I-domains bind collagen in fundamentally different ways. The structure of A3-RU5 provides an excellent starting point for further exploration of the collagen-binding site by means of site-directed mutagenesis.