X-Ray Crystallographic and Single-Molecule Fluorescence Studies of Beta-2 Glycoprotein I Reveal an Alternative Mechanism of Autoantibody Recognition

Background. Antiphospholipid antibodies (aPL) recognizing an epitope comprising residues R39-R43 in the N-terminal domain, Domain I (DI), of beta-2 glycoprotein I (b2GPI) are considered among the most pathogenic in patients with Antiphospholipid Syndrome (APS). How such autoantibodies engage b2GPI at the molecular level remains incompletely understood.

Aim. To better understand how pathogenic anti-DI antibodies engage b2GPI at the molecular level.

Results. Under physiological conditions, b2GPI is believed to adopt a closed conformation featuring an intramolecular interaction between DI and DV with amino acids R39 and R43 in DI being masked by DV. This conformation is therefore predicted to be immunologically inert, incapable of reacting against pathogenic anti-DI antibodies. Once bound to the membranes, however, b2GPI is believed to undergo a dramatic conformational change which liberates DI to the solvent. To get a better grasp of the molecular flexibility of b2GPI under conditions relevant to physiology, we expressed and purified fully-glycosylated human recombinant b2GPI (hr-b2GPI) from HEK293 cells at high yield and purity suitable for structural biology and biophysical studies. After native purification, we found that the recombinant protein bound to heparin and negatively charged phospholipids with affinities comparable to those obtained for b2GPI that was purified from plasma using the perchloric acid method (p-b2GPI); hr-b2GPI also displayed similar reactivity against anti-b2GPI immunoglobulin G antibodies that were isolated from 5 APS patients. Surprisingly, hr-b2GPI and p-b2GPI were structurally similar, too. The X-ray crystal structures of hr-b2GPI and p-b2GPI solved at 2.6 and 2.4 Å resolution were superimposable documenting a J-shaped elongated conformation of the molecule in which DI was located > 90 Å away from the C-terminal DV. Both structures were characterized by 22 oxidized cysteine residues forming 11 disulfide bonds, 4 N-glycosylations, and an intact yet flexible phospholipid-binding loop in DV. Since crystallization occurred at high salt concentrations, validation of the crystal structure of hr-b2GPI in solution was obtained by single-molecule Förster Resonance Energy Transfer (smFRET) and small-angle X-ray scattering (SAXS), while surface plasmon resonance (SPR) was used to probe the binding of a recently developed monoclonal anti-DI antibody, i.e., MBBS, to hr-b2GPI and p-b2GPI in solution. In keeping with the X-ray structural data, donor and acceptor fluorophores incorporated at positions 13/312 in DI and DV and 112/312 in DII and DV reported no measurable energy transfer whereas probes located at positions 13/112 in DI and DII displayed very high energy transfer. Likewise, the scattering profiles of the recombinant and plasma purified proteins returned similar hydrodynamic radii characteristic of elongated, flexible protein structures, and not circular. Notably, both hr-b2GPI and p-b2GPI in the elongated conformation were capable of interacting with MBBS without the need of phospholipids, even though addition of negatively charged phospholipids decreased the apparent dissociation affinity constant due to a reduction of the dissociation rate constant and a remarkable time-dependent accumulation of b2GPI onto the lipid surface, suggestive of a phospholipid-induced oligomerization mechanism.

Conclusions. This study demonstrates that human b2GPI can adopt an elongated conformation in solution that is primed for phospholipid, heparin, and autoantibodies binding with DI constitutively exposed to the solvent. The fact that phospholipid-bound b2GPI is a better antigen for anti-DI autoantibody under physiological conditions as compared to the elongated form in solution can be explained by the relatively low affinity and bivalency of such autoantibodies that likely recognize a peptide motif pattern rather than a specific sequence of residues. Whether other context-dependent conformational changes occur after binding of the protein to the lipid surface, thus facilitating aPL binding, remain to be established. Since our studies failed to detect the closed form of b2GPI previously documented by electron and atomic force microscopy studies, it is possible that this conformation may arise from chemical and/or posttranslational modifications that occur in vivo while the protein circulates in the plasma.