A Novel ELISA Assay for the Detection of Anti-Prothrombin Antibodies in Antiphospholipid Syndrome Patients at High Risk of Thrombosis

Autoantibodies targeting prothrombin (aPT) can be found in antiphospholipid syndrome (APS) patients. However, their detection has proven difficult to standardize. Here, we developed a new ELISA assay to improve the identification of aPT and compared its performance with currently available anti-phosphatidylserine/prothrombin antibodies (aPS/PT) and autoantibodies targeting prothrombin bound to the plastic plate (aPT-A) assays using a cohort of 27 APS patients at high risk of thrombosis. We generated a novel prothrombin variant, ProTS525A-Biot, carrying an artificial tag at the C-terminus suitable for site-specific biotinylation and added the mutation S525A to improve stability. ProTS525A-Biot was immobilized to neutravidin-coated plates at the desired density and with a defined orientation, i.e., pointing the N-terminal fragment-1 toward the solvent. Antibodies against ProTS525A-Biot (aPT-Bio) were found in 24 out of 27 triple-positive APS patients (88%). When compared to aPS/PT and aPT-A, aPT-Bio showed an excellent linear correlation with aPS/PT (R2 = 0.85) but not with aPT-A (R2 = 0.40). Since aPS/PT but not aPT-A are an emerging biomarker of thrombosis in APS, this method may find utility for detecting pathogenic aPT in APS but also other prothrombotic conditions such as COVID-19.


INTRODUCTION
Antiphospholipid syndrome is a debilitating condition characterized by vascular thrombosis in the presence of antiphospholipid antibodies (aPL), which persist >12 weeks in patients' plasma (1). Among aPL, anti-prothrombin antibodies (aPT) are believed to be clinically relevant since they are linked to thrombosis (2). Currently, aPT are detected using two methods (3). First, prothrombin is directly immobilized onto a hydrophilic plastic plate, in which case aPT are named aPT-A. Alternatively, prothrombin is bound, in the presence of calcium ions, to plastic wells pre-coated with phosphatidylserine (PS), in which case aPT are referred as to anti-phosphatidylserine/ prothrombin antibodies (aPS/PT). Beyond the technical aspects, this distinction has strong clinical value since aPS/PT but not aPT-A are the ones often found in APS patients at high risk of thrombosis (3,4), arguing for a potential pathogenic role in the onset and progression of APS disease.
Despite recent advances in analytical methods, detection of aPT, and especially aPS/PT, has proven difficult to standardize because of the transient nature of the phospholipid-bound complex, which requires calcium ions, and the variable source/ purity of phospholipids and antigen. Furthermore, even though it is assumed that aPS/PT interact with prothrombin, they may react against PS and other plasma proteins capable of interacting with PS (5). The aim of this study was to develop an ELISA assay to improve the identification of aPT in correlation with thrombosis.

Protein Expression and Purification
All prothrombin variants used in this study were produced in Expi293 cells (Thermo Fisher Scientific, USA) in the presence of vitamin K and purified as described before (6). Briefly, the cDNA of human prothrombin (ProTWT, UniProtKB P00734) modified to include an epitope for the HPC4 antibody (7) at the C-terminus was cloned into a pDEST40 expression vector (Life Technologies, Inc.) and sequenced verified by Genewiz. Differently from ProTWT, ProTS525A-Biot carries 1) an extended C-terminal peptide sequence 1 GGGSGLNDIFEAQKIEWHE 19 inserted after the HPC4-purification tag that can be specifically biotinylated in vitro by the biotin ligase enzyme (Avidity, USA) at the lysine residue 14 and 2) a single point mutation, namely the catalytic serine (S) 525 was substituted to alanine (A). Genetic engineering of the original cDNA was attained by PCR using the Quickchange Lighting kit (Agilent) and appropriate primers (Integrated DNA Technologies). Transfection of Expi293 cells was performed using Lipofectamine 3000 (ThermoFisher, USA) following manufacturers' instructions. Selection of stably expressing clones was initiated 24 hours after transfection using the antibiotic geneticin (G-418, GoldBio, USA). Prothrombin secreted in the media was purified in three sequential steps: immunoaffinity, ionexchange chromatography and size exclusion chromatography, as detailed elsewhere (6). A key step for obtaining highly pure recombinant protein was to load the immunopurified material into a heparin column (HiTrap Heparin HP 1 ml, Cytiva) before passing the solution into a Q-column (HiTrap Q HP 1 ml, Cytiva). This is because prothrombin fragments generated during protein production, storage, and purification (namely thrombin and prethrombin-2), but not intact prothrombin, bind to heparin. In vitro biotinylation was performed using BirA500 biotin-protein ligase reaction kit (Avidity). Incorporation of biotin was verified using QuantTag Biotin Quantitation Kit (Vector Laboratories).

Plasma Samples and IgGs
Collection of citrated plasma and purification of IgGs was described earlier (8). Briefly, venous blood was collected at the University of Padua in 3.8% sodium citrate (9:1) and centrifuged twice at 2000xg for 15 min at 4°C. Plasma was stored in 25 µL aliquots at -80°C ready for individual use. All the patients in this study participated in the TRAPS trial and gave their written informed consent to utilize their stored residual plasma in this study. APS was diagnosed according to the Sydney Criteria (15). Total IgG extracts were purified using Protein G spin columns ( T h e r m o F i s h e r , U S A ) fo l l o w e d b y s i z e e x c l u s i o n chromatography to obtain pure IgG monomers. Each preparation was >98% pure as judged by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The IgG concentration was determined by reading at 280 nm with a molar extinction coefficient of 1.00 M -1 cm 2 .

ELISA Assays
Neutravidin (3100, Thermo Fisher Scientific, USA) was resuspended in water and its concentration determined by reading absorbance at 280 nm using a molar extinction coefficient of 1.66 M -1 cm 2 . For the detection of aPT-Biot, one hundred microliters of a solution of neutravidin solubilized in 0.1M sodium bicarbonate pH 9.6 were added to a Nunc MAXISORP plate (MilliporeSigma, USA) and incubated overnight at 4°C. After washing three times with 200 µl/well of 20 mM Tris pH 7.4, 145 mM NaCl, 5 mM CaCl 2 , Tween 20 0.02% (TBS-T), wells were blocked with 200 µl of 20 mM Tris pH 7.4, 145 mM NaCl, 5 mM CaCl2, 1% BSA (TBS-B) for 60 min at room temperature. One hundred microliters of ProTS525A-Biot or ProTS525A at the desired concentration were incubated for 60 min at room temperature before repeating the washing step. Next, one hundred microliters of plasma (1:100 v/v dilution) or IgGs prepared in TBS-B were added to each well and incubated for 60 minutes at room temperature (18-22°C). Plates were washed 3 times TBS-T before adding 100 ml of 1:10,000 dilution of peroxidase conjugated anti-human IgG, g-chain specific (A6029, MilliporeSigma, USA) for 60 minutes at room temperature. Finally, plates were washed three more times with TBS-T and then incubated with 100 ml of 1-Step 3,3',5,5'-Tetramethylbenzidine (TMB) Liquid Substrate (34028, ThermoFisher, USA). After 30 minutes, the colorimetric reaction was quenched with 100 ml of TMB-stop solution. The optical density at 450 nm was recorded using a SPARK microplate reader (TECAN, Austria). Statistical analysis between the groups was performed using t-tests in Prism 9.0 (*** < 0.001). aPT-A and aPS/PT were detected using procedures published previously (8).

RESULTS
Recent studies from our laboratory have documented that aPS/ PT, despite being heterogeneous, preferentially bind to the N-terminal portion of prothrombin, also known as fragment-1 (8). Because of this finding, we hypothesized that detection of aPT in correlation with thrombosis might be possible by using a prothrombin derivative that can be immobilized to ELISA plates at the desired density and with a defined orientation, i.e., pointing the N-terminal fragment-1 toward the solvent. To this goal, we created ProTS525A-Biot. ProTS525A-Biot is a catalytically inactive version of prothrombin, which is specifically biotinylated at the C-terminus and can therefore be immobilized onto neutravidin coated plates at the desired density and orientation. Since prothrombin spontaneously converts to thrombin over time, thus resulting in fragment-1 and prethrombin-1 (16), the S->A substitution was chosen to improve the long-term storage and stability of the protein at physiological pH and ionic strength concentrations.
ProTS525A-Biot was expressed in Expi293 cells in the presence of vitamin K. Size exclusion chromatography (SEC) after in vitro biotinylation reaction revealed that that the protein is highly pure and monomeric ( Figure 1A). To confirm functional integrity and the presence of the S525A mutation, the conversion ProTS525A-Biot to thrombin by the prothrombinase complex was monitored by gel electrophoresis ( Figure 1B) and by a chromogenic assay (6) ( Figure 1C), respectively. As expected for a structurally intact yet inactive enzyme, ProTS525A-Biot was properly converted to thrombin by the prothrombinase complex ( Figure 1B), but the resulting enzyme displayed no catalytic activity ( Figure 1C). Given the importance of the N-terminal region of prothrombin in the context of aPS/PT binding, we performed additional SPR binding experiments using immobilized negatively charged liposomes to assess whether glutamic acids underwent proper g-carboxylation. As shown in Figure 1D, binding of ProTS525A-Biot to liposomes was concentration dependent and saturable. The affinity constant calculated by plotting the response unit at equilibrium versus [ProTS525A-Biot] yielded a K d =0.23 ± 0.10 µM (inset, Figure 1D). This value is close to previously reported values obtained for ProTWT under the same experimental conditions (8), confirming structural and functional integrity of the Gla-domain.
We have previously shown that prothrombin is a very dynamic molecule, adopting closed and open forms in solution (6,10). To elucidate whether ProTS525A-Biot retains the same dynamic properties of ProTWT when free and bound to neutravidin, we performed confocal ( Figure 1E) and TIRFbased smFRET experiments ( Figure 1F), respectively. Based on previous studies, we labeled positions 120 in kringle-1 and 478 in the serine protease domain with the FRET pair AlexaFluor 555/ 647. We found that ProT120/478/S525A-Biot, like ProT120/478, interconverted between closed (high FRET) and open (low FRET) forms in solution ( Figure 1E). Remarkably, a similar behavior was observed when ProT120/478/S525A-Biot was bound to neutravidin, documenting that the engineered tag, active site mutation and immobilization strategy do not affect the structural and dynamic properties of the antigen ( Figure 1F).
After having established that ProTS525A-Biot retained identical structural and functional properties compared to ProTWT, we tested whether aPT could recognize ProTS525A-Biot bound to neutravidin plates ( Figure 2). We initially used total IgG extracts that were purified from 5 APS patients with high titers of aPT-A and aPS/PT. The laboratory profile of these patients is shown in Table 1. Neutravidin at 1 µg/well was immobilized onto hydrophilic plates and, after extensive washing, 100 µl of ProTS525A-Biot at a concentration of 10 µg/ml was added. Non-biotinylated ProTS525A was used as a negative control. The results in Figure 2B show that aPT strongly reacted against ProTS525A-Biot. To differentiate these autoantibodies from aPT-A and aPS/PT, we called them aPT-Biot. Importantly, we observed no reactivity against neutravidin that was incubated with non-biotinylated ProTS525A. This result validates our design and documents no cross-reactivity of aPT against neutravidin. Intra and inter-assay variability were assessed by independently repeating the ELISA experiment three times, using IgG samples from 17 patients ( Table 2). Intra-assay variability, calculated on triplicates of the same sample, was < 5%; inter-assay variability was, on average, 10% (min 3%, max 17%).
Given that a single molecule of neutravidin can theoretically bind up to 4 molecules of biotin, we next varied the density of ProTS525A-Biot (10, 1, 0.1, 0.01 µg/well) while keeping neutravidin constant (1 µg/well). Using P27 as a source of autoantibodies, we found that the signal increased hyperbolically reaching saturation at 1 µg/well of ProTS525A-Biot ( Figure 2C). This datapoint defines the lowest concentration of ProTS525A-Biot to achieve maximum signal. Thus, the combination 1 mg/well of neutravidin and 1 mg/well of ProTS525A-Biot, unless otherwise specified, will be used from now on in our assays.
Given that the molecular weight of prothrombin and neutravidin are comparable (72 kDa vs 60 kDa), the results in Figure 2C suggest that clustering of prothrombin driven by multivalent neutravidin, not solely antigen density, stimulates antigen-antibody complex formation. To test this hypothesis, we systematically lowered the concentration of neutravidin, from 5 µg/well to 0.25 µg/well, while keeping ProTS525A-Biot constant (1 µg/well). We observed significantly lower signal at higher neutravidin concentration ( Figure 2D). This effect is in line with our stated hypothesis, which predicts loss of binding at high concentrations of neutravidin due to antigen spreading. Finally, to test for stability, we stored plates with bound ProTS525A-Biot for up to 2 months at 4°C. No significant differences compared to freshly prepared plates was observed. Given the promising results with total IgG extracts, aPT-Biot were researched in a cohort of 27 triple-positive APS patients available from our previous studies (8) (P01-P27) and 5 healthy controls (C01-C05) ( Table 1), using 1:100 v/v diluted plasma and not total IgG extracts. This design was chosen to match more closely what happens in clinical laboratories. Importantly, since all the patients fulfill the definition of triple positive APS, they are considered at high risk of developing thrombosis (17,18). Given a cutoff of 0.16 (OD 450 nm), defined as mean OD value for healthy controls +/-3 STD, we found 24 out of 27 (88%) APS patients positive for aPT-Biot ( Figure 3A). When compared to aPS/PT detected using the kit manufactured by INOVA ( Figure 3B) and aPT-A ( Figure 3C), aPT-Bio showed an excellent linear correlation with aPS/PT (R 2 = 0.85) but not with aPT-A (R 2 = 0.40). Interestingly similar considerations applied when we compared aPS/PT with aPT-A ( Figure 3D), underscoring the similarity aPT-Biot and aPS/PT and confirming the difference between aPS/PT and aPT-A discovered in earlier studies (3).

DISCUSSION
Developing novel, robust, and easy to interpret immunogenic assays to detect autoantibodies in patients' plasma is a challenging yet essential task to achieve. In this study, we report the design and technical validation of a new ELISA assay for the detection of potentially pathogenic aPT in human plasma and provide initial evidence regarding its potential utility in APS patients.
One of the main problems faced by investigators when developing an ELISA assay is how to immobilize the antigen without perturbing its structural and functional properties, which are relevant for physiology and pathology. This problem needs to be considered even more carefully when antigens are inherently flexible and autoantibodies target conformational, and not linear epitopes, as is often the case for aPL (8,19,20). Here, we solved this issue by generating a recombinant version of prothrombin, which is site-specifically biotinylated and can be immobilized with proper orientation and desired density. Importantly, this construct retains its native structural, functional, and dynamic features in solution and when immobilized on a surface coated with neutravidin ( Figure 1). To our knowledge, this approach is novel and has never been applied before for the detection of aPT.
In addition to successfully detecting aPT in patients' plasma, a fascinating observation was that aPT-Biot correlated well with aPS/PT but not aPT-A. Since aPS/PT are an emerging biomarker of thrombosis in APS (4,21) and are also found in other prothrombotic diseases such as COVID-19 (22,23), this result is quite remarkable; it indicates that aPT-Biot may, too, find utility in the clinical practice. Rigorously designed prospective studies on a larger cohort of patients and matching controls will prove whether this idea has any merit. Despite significant efforts and some progress, detection of aPL, and especially aPS/PT, still suffers from platform-dependent variability due to different protocols for immobilizing the antigens -which are often undisclosed due to the use of proprietary technologies-as well as the intrinsic time-dependent instability of the reagents that are immobilized onto the plastic plate (24,25). In this context, detection of aPT-Biot provides several technical advantages compared to aPS/PT. First, provided availability of ProTS525A-Biot, this new ELISA format is very straightforward to prepare and perform, favoring transparency and reproducibility. Second, since the interaction between biotin and neutravidin is practically irreversible and calciumindependent, ELISA plates can be stored for months without loss of activity and, in principle, ELISA assays could be performed with plasma collected with a variety of anticoagulants, such as sodium citrate but also ethylenediaminetetraacetic acid (EDTA)an interesting possibility whose validation, however, requires more experimental work. Third, since a subgroup of pathogenic aPL are known to recognize lipids and not proteins interacting with lipids (5), elimination of PS should eliminate confusions regarding which aPL are being detected.
From a biochemical standpoint, the fact that aPT-Biot correlates with aPS/PT is an intriguing finding since aPS/PT are believed to recognize cryptic epitopes resulting from the interaction of prothrombin with PS (3). While epitope mapping studies are necessary to compare similarities between these two groups of autoantibodies, because of our design and results in Figure 3B, aPT-Biot, like aPS/PT (8), may engage conformational epitopes that are, at least in part, contained in fragment-1. If this hypothesis is confirmed, the main role of PS could be to properly orient and concentrate the antigen on the lipid surface, thus facilitating autoantibody binding. This antigen orientation    hypothesis could also help rationalize the remarkable difference between aPT-Biot and aPT-A, indicating that part of the prothrombin molecule recognized by aPT-Biot and aPS/PT is hidden or damaged when bound to hydrophilic plastic plates. However, it is also possible that aPT-Biot are different from aPS/ PT and that, for example, aPS/PT might prefer epitopes that are unique in that they form when prothrombin is bound to PS, a situation that is reminiscent of recently discovered antiphospholipid antibodies targeting the endothelial protein C receptor in complex with lysobisphosphatidic acid (26).
In conclusion, we believe the new ELISA assay reported here represents an encouraging step towards a more specific detection method for aPT. Given the universality of the neutravidin-biotin system, it is also easy to envision how ProTS525A-Biot can be immobilized to other types of surfaces to enable highly reproducible and cost-effective detection of aPT in single and multiplex automated systems and to facilitate purification and characterization of aPT. Finally, a similar strategy could be applied to other antigens of aPL, such as b 2 GPI, that, despite being primed for autoantibody binding in solution (20), is currently being absorbed onto plastic plates, a harsh method resulting in unpredictable effects on its native structure. In this context, it is worth noting that encouraging results towards a more personalized diagnosis of APS have been obtained using a fragment of b 2 GPI, domain I (DI), which was either tagged at the N-terminus (27) or the C-terminus (28).

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/supplementary material. Further inquiries can be directed to the corresponding author.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by University of Padua (VP). The patients/participants provided their written informed consent to participate in this study.