Unless otherwise stated, coagulation factors were from Haematologic Technologies, Essex Junction, Vermont, USA. Porcine gelatin, bovine serum albumin (BSA) and conjugated antibodies were from Sigma-Aldrich, Suffolk, UK. Chromogenic substrates for ELISA were from KPL, Gaithersburg, Maryland, USA.
Patients and healthy controls
Serum was obtained from 228 patients (University College London Hospital) with APS, n = 59; SLE and no APS (SLE/APS-), n = 106; rheumatoid arthritis (RA), n = 12; Sjögren’s syndrome (SS), n = 13; myositis (Myo), n = 23; systemic sclerosis (SSc), n = 15; and 40 healthy controls (HC). All patients satisfied relevant disease classification criteria for APS , SLE , RA , SS , Myo  and SSc . Informed consent and full ethical approval from the local ethics board were obtained (National Research Ethics Committee- London Hampstead, reference number 12/LO/0373). Patients in the SLE cohort had their disease activity recorded using the Classic British Isles Lupus Assessment Group (BILAG) index . We did not use the BILAG 2004 index  as many of the samples were obtained prior to its routine use.
Immunologic characterisation and purification of IgG
IgG was protein G-purified (Fisher Scientific, Loughborough, UK) dialysed in PBS and the concentration determined by spectrophotometry. Serum and purified IgG aCL and anti-β2GPI titers were measured as previously described . The presence of IgG directed against Thr; FXa; FVIIa; prothrombin (PT); phosphatidylserine (PS)/FXa complex; and AT-III was measured by ELISA. All samples were tested in duplicate and considered positive when the test optical density (OD) minus the background OD exceeded the mean OD + 3 SD of HC. Results were expressed as percentage binding of a positive control and number and percentage of positive patients for the relevant IgG was calculated accordingly.
Unless otherwise stated, in all ELISAs half of the plate was coated with antigen (test side) and buffer alone to the other half (background side); the coating step was performed overnight at 4°C; and all other steps were at room temperature (RT). Anti-Thr antibodies were detected as described previously (16). Costar plates (Costar, UK) were coated with 10 μg/ml human alpha-Thr in Tris-buffered saline (TBS). Plates were blocked with TBS/0.3% gelatin for 1 hour. Serum 1:50 in TBS/0.1% gelatin was incubated for 1.5 hours at RT and bound IgG detected by addition of alkaline phosphatase (ALP)-conjugated anti-human IgG in TBS/0.1% gelatin for 1 hour followed by addition of substrate and absorbance read at 405 nm.
Anti-PT antibodies were detected as described previously . MaxiSorp plates (Nunc, UK) were coated with 10 μg/ml human alpha-PT in TBS. Plates were blocked with TBS/1% BSA for 2 hours. Serum 1:50 in TBS/1% BSA was incubated for 1.5 hours and bound IgG detected by addition of ALP conjugate in TBS/1% BSA for 1 hour followed by addition of substrate and absorbance read at 405 nm.
Anti-FVIIa antibodies were detected by modifications of the method of Bidot et al. . MaxiSorp plates were coated with 1.5 μg/ml recombinant human FVIIa (Novo Nordisk) in PBS. Plates were blocked with PBS/2% BSA for 1.5 hours at 37°C. Serum diluted in PBS/1% BSA was incubated for 1 hour. Bound IgG was detected by addition of 50 μl horseradish peroxidase (HRP)-conjugated goat anti-human IgG in PBS/1% BSA for 1 hour followed by addition of substrate and absorbance read at 450 nm.
Anti-FXa antibodies were detected according to the method described by Yang et al. . Costar plates were coated with 5 μg/ml FXa in TBS. Plates were blocked with 150 μl TBS/0.3% gelatin for 1.5 hours. Serum 1:50 or IgG (200 μg/ml) diluted in TBS/0.3% gelatin were incubated for 1.5 hours. Bound IgG was detected by the addition of HRP conjugate in TBS/0.3% gelatin for 1 hour followed by addition of substrate and absorbance read at 450 nm.
IgG against PS/FXa complexes were detected by modification of the PS/PT method of Atsumi et al . All wells of polysorp plates (Nunc, UK) were coated with PS (50 μg/ml) (Sigma-Aldrich, UK) in methanol:chloroform 4:1 and incubated uncovered overnight. Plates were blocked with TBS/1% BSA/5 mM CaCl2 (TBSA-Ca) for 1 hour. FXa (5 μg/ml) in TBSA-Ca was added to the test half of the plate and TBSA-Ca alone to the control half for 1 hr. Serum or IgG in TBSA-Ca was incubated for 1 hr, followed by ALP conjugate for 45 minutes. Bound IgG was detected by addition of substrate and absorbance read at 405 nm.
Anti -AT-III ELISA
To detect anti-AT-III IgG, maxisorp plates were coated with 10 μg/ml human AT-III in PBS. Plates were blocked with PBS/4% gelatin for 2 hours. Serum or IgG in PBS/1% gelatin were incubated for 1 hour at 37°C, followed by HRP conjugate in PBS/1% gelatin for 1 hour at 37°C. Bound IgG was detected by addition of substrate and absorbance read at 450 nm.
Chaotropic ELISA for determination of avidity of anti-FXa antibodies
This ELISA was adapted from that described previously to measure the avidity of IgG-Thr  and IgG-β2GPI  interactions. Briefly, Costar plates were coated and blocked as per the anti-FXa ELISA. IgG, purified from patient sera-positive for anti-FXa (absorbance units (AU) > mean + 3SD of HC), was diluted in TBS/0.3% gelatin containing increasing concentrations of NaCl (0.15 M, 0.25 M, 0.35 M, 0.5 M, 1 M, 2 M, 3 M, and 4.5 M) and incubated for 1.5 hours. Bound IgG was detected as per the anti-FXa ELISA and avidity determined by calculating the percentage of maximum binding (100%, at 0.15 M NaCl) maintained with each concentration of NaCl.
Clotting assay for FXa activity
FXa-activated clotting time (ACT) was measured using a KC4 coagulometer (Amelung, Lemgo, Germany). Equal volumes of FXa (final concentration 3.9 nM) in PBS and purified IgGs (final concentration 200 μg/ml) diluted in normal human plasma (NHP) (Sekisui Diagnostics, LLC) were mixed and incubated for 10 minutes at 37°C. Reaction mixture was re-calcified by the addition of equal volume of Recalmix (Hep-test clotting assay, Sekisui Diagnostics, LLC) containing CaCl2, PL, FV and fibrinogen. The time it took for the mixture to clot (in seconds) was recorded and a normal reference range was determined by incubating FXa (0.12 to 250 nM) with only NHP.
Functional assay for FXa activity and AT-III inactivation of FXa
The effects of anti-FXa IgG on FXa activity were studied in 0.1 M Tris, 0.1 M CaCl2 and 1 mg/ml BSA (pH7.8) buffer at RT. Briefly, human FXa (final concentration 2 nM) was mixed with IgG (final concentration 200 μg/ml) and incubated for 10 minutes at 37°C. Subsequently, FXa chromogenic substrate S-2765 (Chromogenix; DiaPharma) was added, and generation of P-nitroaniline was monitored at 405 nm. The standard curve was generated by 1:1 mixing FXa with PBS, in which the IgGs were diluted. The activity of FXa was determined based on the rate of hydrolysis of S-2765 from the linear range of absorbance at 405 nm over time. Results were expressed as % inhibition of IgG with FXa compared to FXa alone. To examine whether APS and SLE IgG have a differential inhibitory effect on FXa activity, equal amounts of IgG from different APS and SLE samples with similar FXa binding were mixed (to a final concentration of 200 μg/ml).
The effects of FXa-reactive IgG on FXa inactivation by AT-III were studied in the presence of AT-III and heparin, modified from Bock et al. . Human AT-III was used at a concentration that was 10 times that of human FXa. Assay buffer was the same as FXa activity assay buffer. Briefly, FXa (final concentration 2 nM) was incubated with IgG (final concentration 200 μg/ml) for 10 minutes at 37°C. AT-III (final concentration 20 nM) in assay buffer containing heparin (0.05 IU/ml) was added, followed immediately by the addition of S-2765 (660 μM) and OD measured at 405 nm. The percentage of FXa inactivation by AT-III was calculated as:
(1 - (residual FXa activity with AT-III)/(initial FXa activity without AT-III)) × 100%.
Data analysis was performed using GraphPad Prism. Normality of distribution was assessed using the Kolmogorov-Smirnov test. Differences in antibody titers between patient groups were compared by one-way analysis of variance (ANOVA), followed by Tukey honest significant difference (HSD) post hoc analysis. Frequencies of anti-SP antibodies in patient groups were compared by non-parametric chi-square test. The association of anti-FXa antibody titers with aCL, anti-β2GPI, and AT-III binding with inhibition of AT-III inactivation were assessed using Spearman’s rank correlation coefficient. Differences between the APS-patient and SLE-patient groups in the avidity of anti-FXa were ascertained by two-tailed t-test. The effects of polyclonal anti-FXa from patients with APS, SLE, and HC on FXa activity, AT-III-mediated inactivation of FXa and on FXa-activated clotting time were compared using the Kruskal-Wallis test followed by the Mann-Whitney test.