X-ray and solution structures of human beta-2 glycoprotein I reveal a new mechanism of autoantibody recognition

Venous and arterial thromboses in patients suffering from the autoimmune disorder Antiphospholipid Syndrome (APS) are caused by the presence of antiphospholipid antibodies (aPL). Emerging evidence indicates that autoantibodies targeting the epitope R39-R43 in the N-terminal domain, Domain I (DI), of β2-glycoprotein I (β2GPI) are among the most pathogenic aPL in patients with APS. How such autoantibodies engage β2GPI at the molecular level remains incompletely understood. Here, we have used X-ray crystallography, single-molecule FRET, and small-angle X-ray scattering to demonstrate that, in the free form, under physiological pH and salt concentrations, human recombinant β2GPI adopts an elongated, flexible conformation in which DI is exposed to the solvent, thus available for autoantibody recognition. Consistent with this structural model, binding and mutagenesis studies revealed that the elongated form interacts with a pathogenic anti-DI antibody in solution, without the need of phospholipids. Furthermore, complex formation was affected neither by the neighboring domains, nor by the presence of the linkers, nor by the glycosylations. Since the pathogenic autoantibody requires residues R39 and R43 for optimal binding, these findings challenge longstanding postulates in the field envisioning β2GPI adopting immunologic inert conformations featuring inaccessibility of the epitope R39-R43 in DI and support an alternative model whereby the preferential binding of anti-DI antibodies towards phospholipid-bound β2GPI arises from the ability of the pre-existing elongated form to bind to the membranes and then oligomerize, processes that are likely to be supported by protein conformational changes. Interfering with these steps may limit the pathogenic effects of anti-DI antibodies in APS patients. Significance In the autoimmune disorder called Antiphospholipid Syndrome (APS), the presence of autoantibodies targeting the plasma glycoprotein beta-2 glycoprotein I (β2GPI) is associated with arterial and venous thrombosis as well as pregnancy complications. Understanding how β2GPI becomes immunogenic and how autoantibodies in complex with β2GPI cause the blood to clot remains a top priority in the field. By elucidating the structural architecture of β2GPI free in solution, our studies challenge longstanding postulates in the field and shed new light on the pathogenic mechanisms of APS that may help the development of new diagnostics and therapeutic approaches.


51
In the autoimmune disorder called Antiphospholipid Syndrome (APS), the presence of autoantibodies 52 targeting the plasma glycoprotein beta-2 glycoprotein I (b2GPI) is associated with arterial and venous 53 thrombosis as well as pregnancy complications. Understanding how b2GPI becomes immunogenic and 54 how autoantibodies in complex with b2GPI cause the blood to clot remains a top priority in the field. By 55 elucidating the structural architecture of b2GPI free in solution, our studies challenge longstanding 5 and SAXS. Consequently, it remains unclear under what circumstances and how these forms 96 interconvert, what is their physiological role, and how they participate in the mechanism of autoantibody 97 recognition. Encouraged by our recent results with prothrombin(35-37), the second most common antigen 98 of aPL in APS, this work was initiated to investigate the structural and conformational properties of β2GPI 99 under conditions relevant to physiology and provide new insights into the mechanism of autoantibody 100 recognition. Our results based on X-ray crystallography, single-molecule FRET (smFRET), SAXS, 101 binding kinetics, and mutational studies unexpectedly reveal that human recombinant b2GPI adopts an 102 elongated, flexible conformation, not circular, in which DI is exposed to the solvent and therefore available 103 for autoantibody recognition. Based on this new evidence and previous findings, an alternative 104 mechanism to explain how negatively charged phospholipids may enhance the affinity toward anti-DI 105 autoantibodies without requiring opening of the protein structure or relocation of the glycosylations away 106 from DI is proposed, and its implication to our understanding of APS discussed.

112
To get a better grasp of the structural architecture of b2GPI under conditions relevant to physiology, we 113 set out to perform structural and biophysical studies of fully glycosylated human recombinant b2GPI. Two 114 versions of the proteins were successfully expressed and purified under native conditions at high yield 115 and purity. The first version, called LT-b2GPI, contained a long multifunctional cleavable tag at the N-116 terminus, located right before the natural N-terminal sequence 1 GRTC 4 ( Fig. 2A). The tag was then 117 cleaved with enterokinase to generate the intact, mature protein (hrβ2GPI). Removal of the tag was 118 confirmed by N-terminal sequencing (Fig. 2B). The second version, called ST-b2GPI, contains a shorter, 119 non-cleavable purification tag at the N-terminus that, based on our previous work(36), is expected not to 120 affect the conformational properties of the protein (Fig. 2B). ST-b2GPI was made to eliminate the 121 enterokinase cleavage step that was very laborious and not as efficient as expected. The presence of 122 the short tag was confirmed by N-terminal sequencing and accounted for the different electrophoretic 123 mobility observed between recombinant and plasma purified protein before and after enzymatic removal 124 of the N-glycosylations (Fig. 2B).

125
To evaluate the functional integrity of the recombinant proteins, LT-b2GPI, hrβ2GPI and ST-b2GPI 126 were tested side by side with plasma purified β2GPI (pβ2GPI) in several biochemical assays. Using 127 surface plasmon resonance (SPR), we found that all variants interacted avidly with liposomes containing 128 negatively charged phospholipids such as phosphatidylserine, yet they failed to interact with 129 phospholipids entirely made of phosphatidylcholine ( Fig. 2C-D). Importantly, the values of the affinity 130 constants were similar for all the constructs and consistent with published data(39), and so was the 6 inhibitory effect of physiological concentrations of calcium chloride. These results document structural 132 integrity of the hydrophobic loop in DV and also prove that the phospholipid binding activity of β2GPI is 133 not perturbed by the presence or removal of the purification tags.

134
In addition to properly interacting with phospholipids, the recombinant proteins were also successfully 135 recognized in ELISA assays by aPL isolated from four triple positive APS patients, which contain anti-DI 136 antibodies(22, 35, 40) (Fig. 2E)

148
The crystals belong to the orthorhombic space group C2221 (Table 1)

153
Given that purification of the recombinant proteins occurs under native conditions, this result indicates 154 that, right after cell secretion, b2GPI does not contain free thiols.

155
All three independently solved X-ray crystal structures depicted β2GPI featuring an elongated 156 conformation spanning ~140 Å in length, from the N-to the C-terminus. The first three domains, DI-DIII,

157
are aligned along the vertical axis of the molecule, whereas DIV and DV bend, forcing the molecule to 158 adopt a characteristic J-shaped elongated form resembling a hockey stick. DI and DV are located >100

159
Å apart and both of them are exposed to the solvent. Interestingly, however, the side chain of residue 160 R43, which is part of the cryptic epitope recognized by anti-DI antibodies, is not exposed to the solvent 161 and is part of a hydrogen bond network made up by residues R39, G41 and T57 (Fig. 3E).

162
Overall, the three new structures are superimposable and similar to the published ones(24, 25), yet 163 they are not identical (Fig. 4A). One significant difference regards the conformation the hydrophobic loop 164 in DV (residues 308-319), which, given its flexibility and exposure to the solvent, varies in every structure.

165
Another difference is the significant extra electron density after molecular replacement in the datasets at 166 a higher resolution (i.e., 2.4 and 2.6Å), suggesting that the input structural model used to solve the 7 structures (1C1Z(25)) was incomplete (Fig. 4B). We attributed this density to the N-linked glycosylations  conformation(2). To address this concern, we applied smFRET to β2GPI (37,41,42

188
Guided by our new structures, we generated four FRET pairs in the ST background, S13C/S112C, 189 S13C/S312C, S112C/S312C and S190C/S312C (Fig. 5A- couple, the crystal structure predicts no FRET for the mutants S112C/S312C and S13C/S312C. In 197 contrast, high FRET and low but measurable FRET values are expected for the FRET pair S13C/S112C 198 and S190C/S312C, respectively. This is because residues 13 and 112 are located ~24 Å apart while the 8 reported a negligible FRET signal, whereas probes attached to the S13C/S112C and S190C/S312C

209
suggesting that, in contrast to previous findings, variation of the ionic strength and pH produces minor 210 conformational changes that could not be detected by our FRET pairs.

211
To rule out potential artifacts arising from the substitution of natural amino acids with cysteine and 212 incorporation of fluorescent dyes, we collected SAXS data for the ST-b2GPI and pb2GPI, under 213 physiological conditions (Fig. 6A-B). SAXS is a biophysical method that is particularly useful to assess 214 the overall shape of biological macromolecules in solution (45) aPL. The binding of β2GPI to MBB2 was followed using SPR (Fig. 7), a technique that allows to measure 227 association (on) and dissociation (off) rate constants in real-time thus enabling a deeper understanding 228 of the chemical nature of the molecular interaction.

229
To retain the native conformation of β2GPI in solution, we immobilized MBB2 to the chip's surface 230 and injected β2GPI in the fluid phase. Binding between MBB2 and β2GPI should occur only if DI is 231 exposed to the solvent. This experimental setup is different from previously reported interaction data 232 between MBB2 and β2GPI in which β2GPI was covalently immobilized on a dextran-based chip and the 233 antibody was used in the fluid-phase to mimic binding of MBB2 to β2GPI bound to negatively charged 234 phospholipids(13). The results of the experiments shown in Fig. 7A  at 300 mM NaCl to 3.2*10 7 M -1 at 15 mM NaCl (Fig. 7C). Importantly, 246 the higher affinity of MBBS for β2GPI at low salt originated from a substantial reduction of the off-rate 247 whereas the on-rate remained mostly unaffected. A significant ~30-fold reduction of the affinity under 248 physiological conditions was also detected after mutating the positively charged residues R39, R43 and 249 K44 in DI with the neutral amino acid alanine (Fig. 7D-E), confirming the electrostatic nature of such 250 interaction and the ability of MBB2 to interact with an epitope of DI that is targeted by pathogenic aPL.

251
Previous studies have proposed that the epitope R39-R43 is cryptic because it is buried by DV in the 252 O-circular form (27)

263
Owing to its flexibility, the structural architecture of b2GPI has remained controversial, and so is the 264 mechanism of autoantibody recognition. A first major conclusion emerging from this study is that human 265 recombinant b2GPI expressed in mammalian cells and purified under native conditions adopts an 266 elongated, flexible conformation in which DV and DI are exposed to the solvent. In the free form, under 267 physiological pH and salt concentrations, b2GPI is therefore primed for phospholipid binding and 268 autoantibody recognition. A second major conclusion is that the recombinant protein is structurally and 269 functionally identical to b2GPI purified from plasma using the perchloric acid method. Hence, the 270 elongated conformation of β2GPI is not an artifact caused by the harsh purification methods or 271 crystallization conditions but a genuine conformation of the protein in solution.

272
The recognition that the elongated form of b2GPI exists and, according to our smFRET experiments,

273
perhaps predominates in human plasma bears important implications in our understanding of the APS 10 pathology. It also provides new ideas for the development of APS-focused diagnostics and therapeutics.

275
Regarding the mechanism of anti-DI antibody recognition, our structural and binding data indicates that

316
It is important to acknowledge that, even though all the structural and biochemical experiments were 317 performed using highly purified proteins solubilized in physiological buffers, b2GPI has never been 318 exposed to endothelial or circulating blood cells. Hence, it remains possible, although unlikely, that the before (57). Plasma-derived β2GPI (pβ2GPI) was purified using the perchloric acid method, as described 332 previously(57). MBB2 was produced as described before (13